diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,30125 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 98.0004, + "global_step": 5000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 14.0549, + "step": 1 + }, + { + "epoch": 0.0, + "learning_rate": 1e-06, + "loss": 13.5798, + "step": 2 + }, + { + "epoch": 0.0, + "learning_rate": 2e-06, + "loss": 13.4432, + "step": 3 + }, + { + "epoch": 0.0, + "learning_rate": 3e-06, + "loss": 14.0655, + "step": 4 + }, + { + "epoch": 0.0, + "learning_rate": 4e-06, + "loss": 13.5395, + "step": 5 + }, + { + "epoch": 0.0, + "learning_rate": 5e-06, + "loss": 12.7124, + "step": 6 + }, + { + "epoch": 0.0, + "learning_rate": 6e-06, + "loss": 13.1198, + "step": 7 + }, + { + "epoch": 0.0, + "learning_rate": 7e-06, + "loss": 13.6739, + "step": 8 + }, + { + "epoch": 0.0, + "learning_rate": 8e-06, + "loss": 13.7482, + "step": 9 + }, + { + "epoch": 0.0, + "learning_rate": 9e-06, + "loss": 13.8056, + "step": 10 + }, + { + "epoch": 0.0, + "learning_rate": 1e-05, + "loss": 13.8331, + "step": 11 + }, + { + "epoch": 0.0, + "learning_rate": 1.1e-05, + "loss": 14.3513, + "step": 12 + }, + { + "epoch": 0.0, + "learning_rate": 1.2e-05, + "loss": 13.6611, + "step": 13 + }, + { + "epoch": 0.0, + "learning_rate": 1.3e-05, + "loss": 13.9781, + "step": 14 + }, + { + "epoch": 0.0, + "learning_rate": 1.4e-05, + "loss": 15.9339, + "step": 15 + }, + { + "epoch": 0.0, + "learning_rate": 1.5e-05, + "loss": 13.6124, + "step": 16 + }, + { + "epoch": 0.0, + "learning_rate": 1.6e-05, + "loss": 13.6459, + "step": 17 + }, + { + "epoch": 0.0, + "learning_rate": 1.7000000000000003e-05, + "loss": 13.6403, + "step": 18 + }, + { + "epoch": 0.0, + "learning_rate": 1.7000000000000003e-05, + "loss": 13.9373, + "step": 19 + }, + { + "epoch": 0.0, + "learning_rate": 1.8e-05, + "loss": 13.7237, + "step": 20 + }, + { + "epoch": 0.0, + "learning_rate": 1.9e-05, + "loss": 12.9863, + "step": 21 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 13.9444, + "step": 22 + }, + { + "epoch": 0.0, + "learning_rate": 2.1000000000000002e-05, + "loss": 13.0741, + "step": 23 + }, + { + "epoch": 0.0, + "learning_rate": 2.2e-05, + "loss": 13.1559, + "step": 24 + }, + { + "epoch": 0.01, + "learning_rate": 2.3e-05, + "loss": 12.5891, + "step": 25 + }, + { + "epoch": 0.01, + "learning_rate": 2.4e-05, + "loss": 12.6598, + "step": 26 + }, + { + "epoch": 0.01, + "learning_rate": 2.5e-05, + "loss": 13.0467, + "step": 27 + }, + { + "epoch": 0.01, + "learning_rate": 2.6e-05, + "loss": 11.3803, + "step": 28 + }, + { + "epoch": 0.01, + "learning_rate": 2.7e-05, + "loss": 11.5069, + "step": 29 + }, + { + "epoch": 0.01, + "learning_rate": 2.8e-05, + "loss": 10.1899, + "step": 30 + }, + { + "epoch": 0.01, + "learning_rate": 2.9000000000000004e-05, + "loss": 9.8524, + "step": 31 + }, + { + "epoch": 0.01, + "learning_rate": 3e-05, + "loss": 8.671, + "step": 32 + }, + { + "epoch": 0.01, + "learning_rate": 3.1e-05, + "loss": 9.0877, + "step": 33 + }, + { + "epoch": 0.01, + "learning_rate": 3.2e-05, + "loss": 8.3881, + "step": 34 + }, + { + "epoch": 0.01, + "learning_rate": 3.3e-05, + "loss": 8.277, + "step": 35 + }, + { + "epoch": 0.01, + "learning_rate": 3.4000000000000007e-05, + "loss": 8.0669, + "step": 36 + }, + { + "epoch": 0.01, + "learning_rate": 3.5000000000000004e-05, + "loss": 7.3799, + "step": 37 + }, + { + "epoch": 0.01, + "learning_rate": 3.6e-05, + "loss": 7.1627, + "step": 38 + }, + { + "epoch": 0.01, + "learning_rate": 3.7e-05, + "loss": 6.9334, + "step": 39 + }, + { + "epoch": 0.01, + "learning_rate": 3.8e-05, + "loss": 6.3581, + "step": 40 + }, + { + "epoch": 0.01, + "learning_rate": 3.9e-05, + "loss": 6.1673, + "step": 41 + }, + { + "epoch": 0.01, + "learning_rate": 4e-05, + "loss": 6.411, + "step": 42 + }, + { + "epoch": 0.01, + "learning_rate": 4.1e-05, + "loss": 6.2606, + "step": 43 + }, + { + "epoch": 0.01, + "learning_rate": 4.2000000000000004e-05, + "loss": 5.9232, + "step": 44 + }, + { + "epoch": 0.01, + "learning_rate": 4.2999999999999995e-05, + "loss": 5.6804, + "step": 45 + }, + { + "epoch": 0.01, + "learning_rate": 4.4e-05, + "loss": 5.7039, + "step": 46 + }, + { + "epoch": 0.01, + "learning_rate": 4.4999999999999996e-05, + "loss": 5.3503, + "step": 47 + }, + { + "epoch": 0.01, + "learning_rate": 4.6e-05, + "loss": 5.3466, + "step": 48 + }, + { + "epoch": 0.01, + "learning_rate": 4.7000000000000004e-05, + "loss": 5.2484, + "step": 49 + }, + { + "epoch": 0.01, + "learning_rate": 4.8e-05, + "loss": 5.3019, + "step": 50 + }, + { + "epoch": 0.01, + "learning_rate": 4.9000000000000005e-05, + "loss": 4.9601, + "step": 51 + }, + { + "epoch": 1.0, + "learning_rate": 5e-05, + "loss": 7.2562, + "step": 52 + }, + { + "epoch": 1.0, + "learning_rate": 5.1e-05, + "loss": 4.8344, + "step": 53 + }, + { + "epoch": 1.0, + "learning_rate": 5.2e-05, + "loss": 4.526, + "step": 54 + }, + { + "epoch": 1.0, + "learning_rate": 5.3e-05, + "loss": 4.4054, + "step": 55 + }, + { + "epoch": 1.0, + "learning_rate": 5.4e-05, + "loss": 4.3966, + "step": 56 + }, + { + "epoch": 1.0, + "learning_rate": 5.5e-05, + "loss": 4.4109, + "step": 57 + }, + { + "epoch": 1.0, + "learning_rate": 5.6e-05, + "loss": 4.4884, + "step": 58 + }, + { + "epoch": 1.0, + "learning_rate": 5.7e-05, + "loss": 4.4897, + "step": 59 + }, + { + "epoch": 1.0, + "learning_rate": 5.800000000000001e-05, + "loss": 4.2576, + "step": 60 + }, + { + "epoch": 1.0, + "learning_rate": 5.9e-05, + "loss": 4.2444, + "step": 61 + }, + { + "epoch": 1.0, + "learning_rate": 6e-05, + "loss": 4.2013, + "step": 62 + }, + { + "epoch": 1.0, + "learning_rate": 6.1e-05, + "loss": 4.276, + "step": 63 + }, + { + "epoch": 1.0, + "learning_rate": 6.2e-05, + "loss": 4.0932, + "step": 64 + }, + { + "epoch": 1.0, + "learning_rate": 6.3e-05, + "loss": 4.1363, + "step": 65 + }, + { + "epoch": 1.0, + "learning_rate": 6.4e-05, + "loss": 4.2697, + "step": 66 + }, + { + "epoch": 1.0, + "learning_rate": 6.500000000000001e-05, + "loss": 4.2277, + "step": 67 + }, + { + "epoch": 1.0, + "learning_rate": 6.6e-05, + "loss": 4.1792, + "step": 68 + }, + { + "epoch": 1.0, + "learning_rate": 6.7e-05, + "loss": 4.2452, + "step": 69 + }, + { + "epoch": 1.0, + "learning_rate": 6.800000000000001e-05, + "loss": 3.9818, + "step": 70 + }, + { + "epoch": 1.0, + "learning_rate": 6.900000000000001e-05, + "loss": 4.06, + "step": 71 + }, + { + "epoch": 1.0, + "learning_rate": 7.000000000000001e-05, + "loss": 3.8792, + "step": 72 + }, + { + "epoch": 1.0, + "learning_rate": 7.099999999999999e-05, + "loss": 3.959, + "step": 73 + }, + { + "epoch": 1.0, + "learning_rate": 7.2e-05, + "loss": 3.8818, + "step": 74 + }, + { + "epoch": 1.0, + "learning_rate": 7.3e-05, + "loss": 3.8928, + "step": 75 + }, + { + "epoch": 1.0, + "learning_rate": 7.4e-05, + "loss": 3.8781, + "step": 76 + }, + { + "epoch": 1.01, + "learning_rate": 7.5e-05, + "loss": 3.8462, + "step": 77 + }, + { + "epoch": 1.01, + "learning_rate": 7.6e-05, + "loss": 3.9276, + "step": 78 + }, + { + "epoch": 1.01, + "learning_rate": 7.7e-05, + "loss": 3.6924, + "step": 79 + }, + { + "epoch": 1.01, + "learning_rate": 7.8e-05, + "loss": 3.7754, + "step": 80 + }, + { + "epoch": 1.01, + "learning_rate": 7.9e-05, + "loss": 3.8196, + "step": 81 + }, + { + "epoch": 1.01, + "learning_rate": 8e-05, + "loss": 3.8313, + "step": 82 + }, + { + "epoch": 1.01, + "learning_rate": 8.1e-05, + "loss": 3.7371, + "step": 83 + }, + { + "epoch": 1.01, + "learning_rate": 8.2e-05, + "loss": 3.678, + "step": 84 + }, + { + "epoch": 1.01, + "learning_rate": 8.300000000000001e-05, + "loss": 3.6476, + "step": 85 + }, + { + "epoch": 1.01, + "learning_rate": 8.400000000000001e-05, + "loss": 3.7103, + "step": 86 + }, + { + "epoch": 1.01, + "learning_rate": 8.5e-05, + "loss": 3.6186, + "step": 87 + }, + { + "epoch": 1.01, + "learning_rate": 8.599999999999999e-05, + "loss": 3.6316, + "step": 88 + }, + { + "epoch": 1.01, + "learning_rate": 8.7e-05, + "loss": 3.6121, + "step": 89 + }, + { + "epoch": 1.01, + "learning_rate": 8.8e-05, + "loss": 3.632, + "step": 90 + }, + { + "epoch": 1.01, + "learning_rate": 8.9e-05, + "loss": 3.5848, + "step": 91 + }, + { + "epoch": 1.01, + "learning_rate": 8.999999999999999e-05, + "loss": 3.524, + "step": 92 + }, + { + "epoch": 1.01, + "learning_rate": 9.1e-05, + "loss": 3.4952, + "step": 93 + }, + { + "epoch": 1.01, + "learning_rate": 9.2e-05, + "loss": 3.4631, + "step": 94 + }, + { + "epoch": 1.01, + "learning_rate": 9.3e-05, + "loss": 3.4995, + "step": 95 + }, + { + "epoch": 1.01, + "learning_rate": 9.400000000000001e-05, + "loss": 3.4561, + "step": 96 + }, + { + "epoch": 1.01, + "learning_rate": 9.5e-05, + "loss": 3.4721, + "step": 97 + }, + { + "epoch": 1.01, + "learning_rate": 9.6e-05, + "loss": 3.4174, + "step": 98 + }, + { + "epoch": 1.01, + "learning_rate": 9.7e-05, + "loss": 3.4683, + "step": 99 + }, + { + "epoch": 1.01, + "learning_rate": 9.800000000000001e-05, + "loss": 3.3924, + "step": 100 + }, + { + "epoch": 1.01, + "learning_rate": 9.900000000000001e-05, + "loss": 3.3729, + "step": 101 + }, + { + "epoch": 1.01, + "learning_rate": 0.0001, + "loss": 3.3934, + "step": 102 + }, + { + "epoch": 2.0, + "learning_rate": 0.000101, + "loss": 5.041, + "step": 103 + }, + { + "epoch": 2.0, + "learning_rate": 0.000102, + "loss": 3.3011, + "step": 104 + }, + { + "epoch": 2.0, + "learning_rate": 0.000103, + "loss": 3.3318, + "step": 105 + }, + { + "epoch": 2.0, + "learning_rate": 0.000104, + "loss": 3.2764, + "step": 106 + }, + { + "epoch": 2.0, + "learning_rate": 0.000105, + "loss": 3.2957, + "step": 107 + }, + { + "epoch": 2.0, + "learning_rate": 0.000106, + "loss": 3.2698, + "step": 108 + }, + { + "epoch": 2.0, + "learning_rate": 0.000107, + "loss": 3.2745, + "step": 109 + }, + { + "epoch": 2.0, + "learning_rate": 0.000108, + "loss": 3.2799, + "step": 110 + }, + { + "epoch": 2.0, + "learning_rate": 0.000109, + "loss": 3.2881, + "step": 111 + }, + { + "epoch": 2.0, + "learning_rate": 0.00011, + "loss": 3.2733, + "step": 112 + }, + { + "epoch": 2.0, + "learning_rate": 0.000111, + "loss": 3.2953, + "step": 113 + }, + { + "epoch": 2.0, + "learning_rate": 0.000112, + "loss": 3.2228, + "step": 114 + }, + { + "epoch": 2.0, + "learning_rate": 0.00011300000000000001, + "loss": 3.2584, + "step": 115 + }, + { + "epoch": 2.0, + "learning_rate": 0.000114, + "loss": 3.1881, + "step": 116 + }, + { + "epoch": 2.0, + "learning_rate": 0.000115, + "loss": 3.2122, + "step": 117 + }, + { + "epoch": 2.0, + "learning_rate": 0.00011600000000000001, + "loss": 3.2166, + "step": 118 + }, + { + "epoch": 2.0, + "learning_rate": 0.00011700000000000001, + "loss": 3.2568, + "step": 119 + }, + { + "epoch": 2.0, + "learning_rate": 0.000118, + "loss": 3.1737, + "step": 120 + }, + { + "epoch": 2.0, + "learning_rate": 0.00011899999999999999, + "loss": 3.1706, + "step": 121 + }, + { + "epoch": 2.0, + "learning_rate": 0.00012, + "loss": 3.1903, + "step": 122 + }, + { + "epoch": 2.0, + "learning_rate": 0.000121, + "loss": 3.1808, + "step": 123 + }, + { + "epoch": 2.0, + "learning_rate": 0.000122, + "loss": 3.1488, + "step": 124 + }, + { + "epoch": 2.0, + "learning_rate": 0.000123, + "loss": 3.1552, + "step": 125 + }, + { + "epoch": 2.0, + "learning_rate": 0.000124, + "loss": 3.1774, + "step": 126 + }, + { + "epoch": 2.0, + "learning_rate": 0.000125, + "loss": 3.1511, + "step": 127 + }, + { + "epoch": 2.01, + "learning_rate": 0.000126, + "loss": 3.1858, + "step": 128 + }, + { + "epoch": 2.01, + "learning_rate": 0.000127, + "loss": 3.1378, + "step": 129 + }, + { + "epoch": 2.01, + "learning_rate": 0.000128, + "loss": 3.1646, + "step": 130 + }, + { + "epoch": 2.01, + "learning_rate": 0.00012900000000000002, + "loss": 3.1315, + "step": 131 + }, + { + "epoch": 2.01, + "learning_rate": 0.00013000000000000002, + "loss": 3.1225, + "step": 132 + }, + { + "epoch": 2.01, + "learning_rate": 0.000131, + "loss": 3.1483, + "step": 133 + }, + { + "epoch": 2.01, + "learning_rate": 0.000132, + "loss": 3.1401, + "step": 134 + }, + { + "epoch": 2.01, + "learning_rate": 0.000133, + "loss": 3.1392, + "step": 135 + }, + { + "epoch": 2.01, + "learning_rate": 0.000134, + "loss": 3.1153, + "step": 136 + }, + { + "epoch": 2.01, + "learning_rate": 0.000135, + "loss": 3.1529, + "step": 137 + }, + { + "epoch": 2.01, + "learning_rate": 0.00013600000000000003, + "loss": 3.1382, + "step": 138 + }, + { + "epoch": 2.01, + "learning_rate": 0.00013700000000000002, + "loss": 3.1573, + "step": 139 + }, + { + "epoch": 2.01, + "learning_rate": 0.00013800000000000002, + "loss": 3.1495, + "step": 140 + }, + { + "epoch": 2.01, + "learning_rate": 0.00013900000000000002, + "loss": 3.1046, + "step": 141 + }, + { + "epoch": 2.01, + "learning_rate": 0.00014000000000000001, + "loss": 3.1063, + "step": 142 + }, + { + "epoch": 2.01, + "learning_rate": 0.00014099999999999998, + "loss": 3.1169, + "step": 143 + }, + { + "epoch": 2.01, + "learning_rate": 0.00014199999999999998, + "loss": 3.1046, + "step": 144 + }, + { + "epoch": 2.01, + "learning_rate": 0.00014299999999999998, + "loss": 3.0821, + "step": 145 + }, + { + "epoch": 2.01, + "learning_rate": 0.000144, + "loss": 3.0905, + "step": 146 + }, + { + "epoch": 2.01, + "learning_rate": 0.000145, + "loss": 3.0985, + "step": 147 + }, + { + "epoch": 2.01, + "learning_rate": 0.000146, + "loss": 3.1296, + "step": 148 + }, + { + "epoch": 2.01, + "learning_rate": 0.000147, + "loss": 3.0747, + "step": 149 + }, + { + "epoch": 2.01, + "learning_rate": 0.000148, + "loss": 3.0875, + "step": 150 + }, + { + "epoch": 2.01, + "learning_rate": 0.000149, + "loss": 3.0827, + "step": 151 + }, + { + "epoch": 2.01, + "learning_rate": 0.00015, + "loss": 3.0893, + "step": 152 + }, + { + "epoch": 2.01, + "learning_rate": 0.000151, + "loss": 3.0591, + "step": 153 + }, + { + "epoch": 3.0, + "learning_rate": 0.000152, + "loss": 4.597, + "step": 154 + }, + { + "epoch": 3.0, + "learning_rate": 0.000153, + "loss": 3.0374, + "step": 155 + }, + { + "epoch": 3.0, + "learning_rate": 0.000154, + "loss": 3.0576, + "step": 156 + }, + { + "epoch": 3.0, + "learning_rate": 0.000155, + "loss": 3.0427, + "step": 157 + }, + { + "epoch": 3.0, + "learning_rate": 0.000156, + "loss": 3.0601, + "step": 158 + }, + { + "epoch": 3.0, + "learning_rate": 0.000157, + "loss": 3.0391, + "step": 159 + }, + { + "epoch": 3.0, + "learning_rate": 0.000158, + "loss": 3.0592, + "step": 160 + }, + { + "epoch": 3.0, + "learning_rate": 0.00015900000000000002, + "loss": 3.0594, + "step": 161 + }, + { + "epoch": 3.0, + "learning_rate": 0.00016, + "loss": 3.0543, + "step": 162 + }, + { + "epoch": 3.0, + "learning_rate": 0.000161, + "loss": 3.0814, + "step": 163 + }, + { + "epoch": 3.0, + "learning_rate": 0.000162, + "loss": 3.0239, + "step": 164 + }, + { + "epoch": 3.0, + "learning_rate": 0.000163, + "loss": 3.0469, + "step": 165 + }, + { + "epoch": 3.0, + "learning_rate": 0.000164, + "loss": 3.0379, + "step": 166 + }, + { + "epoch": 3.0, + "learning_rate": 0.000165, + "loss": 3.0567, + "step": 167 + }, + { + "epoch": 3.0, + "learning_rate": 0.00016600000000000002, + "loss": 3.0685, + "step": 168 + }, + { + "epoch": 3.0, + "learning_rate": 0.00016700000000000002, + "loss": 3.0361, + "step": 169 + }, + { + "epoch": 3.0, + "learning_rate": 0.00016800000000000002, + "loss": 3.0228, + "step": 170 + }, + { + "epoch": 3.0, + "learning_rate": 0.00016900000000000002, + "loss": 3.0479, + "step": 171 + }, + { + "epoch": 3.0, + "learning_rate": 0.00017, + "loss": 3.0459, + "step": 172 + }, + { + "epoch": 3.0, + "learning_rate": 0.000171, + "loss": 3.0544, + "step": 173 + }, + { + "epoch": 3.0, + "learning_rate": 0.00017199999999999998, + "loss": 3.0512, + "step": 174 + }, + { + "epoch": 3.0, + "learning_rate": 0.000173, + "loss": 3.0668, + "step": 175 + }, + { + "epoch": 3.0, + "learning_rate": 0.000174, + "loss": 3.0597, + "step": 176 + }, + { + "epoch": 3.0, + "learning_rate": 0.000175, + "loss": 3.0594, + "step": 177 + }, + { + "epoch": 3.0, + "learning_rate": 0.000176, + "loss": 3.0396, + "step": 178 + }, + { + "epoch": 3.01, + "learning_rate": 0.000177, + "loss": 3.0311, + "step": 179 + }, + { + "epoch": 3.01, + "learning_rate": 0.000178, + "loss": 3.0772, + "step": 180 + }, + { + "epoch": 3.01, + "learning_rate": 0.000179, + "loss": 3.0254, + "step": 181 + }, + { + "epoch": 3.01, + "learning_rate": 0.00017999999999999998, + "loss": 3.0163, + "step": 182 + }, + { + "epoch": 3.01, + "learning_rate": 0.000181, + "loss": 3.0319, + "step": 183 + }, + { + "epoch": 3.01, + "learning_rate": 0.000182, + "loss": 3.0668, + "step": 184 + }, + { + "epoch": 3.01, + "learning_rate": 0.000183, + "loss": 3.0287, + "step": 185 + }, + { + "epoch": 3.01, + "learning_rate": 0.000184, + "loss": 3.008, + "step": 186 + }, + { + "epoch": 3.01, + "learning_rate": 0.000185, + "loss": 3.037, + "step": 187 + }, + { + "epoch": 3.01, + "learning_rate": 0.000186, + "loss": 3.0334, + "step": 188 + }, + { + "epoch": 3.01, + "learning_rate": 0.000187, + "loss": 3.0187, + "step": 189 + }, + { + "epoch": 3.01, + "learning_rate": 0.00018800000000000002, + "loss": 3.0028, + "step": 190 + }, + { + "epoch": 3.01, + "learning_rate": 0.000189, + "loss": 3.0039, + "step": 191 + }, + { + "epoch": 3.01, + "learning_rate": 0.00019, + "loss": 3.0362, + "step": 192 + }, + { + "epoch": 3.01, + "learning_rate": 0.000191, + "loss": 3.0287, + "step": 193 + }, + { + "epoch": 3.01, + "learning_rate": 0.000192, + "loss": 3.0107, + "step": 194 + }, + { + "epoch": 3.01, + "learning_rate": 0.000193, + "loss": 2.996, + "step": 195 + }, + { + "epoch": 3.01, + "learning_rate": 0.000194, + "loss": 3.0047, + "step": 196 + }, + { + "epoch": 3.01, + "learning_rate": 0.00019500000000000002, + "loss": 3.0124, + "step": 197 + }, + { + "epoch": 3.01, + "learning_rate": 0.00019600000000000002, + "loss": 3.0256, + "step": 198 + }, + { + "epoch": 3.01, + "learning_rate": 0.00019700000000000002, + "loss": 3.0131, + "step": 199 + }, + { + "epoch": 3.01, + "learning_rate": 0.00019800000000000002, + "loss": 2.9806, + "step": 200 + }, + { + "epoch": 3.01, + "learning_rate": 0.000199, + "loss": 2.9948, + "step": 201 + }, + { + "epoch": 3.01, + "learning_rate": 0.0002, + "loss": 2.9833, + "step": 202 + }, + { + "epoch": 3.01, + "learning_rate": 0.000201, + "loss": 2.9532, + "step": 203 + }, + { + "epoch": 3.01, + "learning_rate": 0.000202, + "loss": 2.9771, + "step": 204 + }, + { + "epoch": 4.0, + "learning_rate": 0.00020300000000000003, + "loss": 4.4334, + "step": 205 + }, + { + "epoch": 4.0, + "learning_rate": 0.000204, + "loss": 2.9473, + "step": 206 + }, + { + "epoch": 4.0, + "learning_rate": 0.000205, + "loss": 2.9579, + "step": 207 + }, + { + "epoch": 4.0, + "learning_rate": 0.000206, + "loss": 2.9469, + "step": 208 + }, + { + "epoch": 4.0, + "learning_rate": 0.000207, + "loss": 2.9705, + "step": 209 + }, + { + "epoch": 4.0, + "learning_rate": 0.000208, + "loss": 2.9611, + "step": 210 + }, + { + "epoch": 4.0, + "learning_rate": 0.00020899999999999998, + "loss": 3.0024, + "step": 211 + }, + { + "epoch": 4.0, + "learning_rate": 0.00021, + "loss": 2.9354, + "step": 212 + }, + { + "epoch": 4.0, + "learning_rate": 0.000211, + "loss": 2.9459, + "step": 213 + }, + { + "epoch": 4.0, + "learning_rate": 0.000212, + "loss": 2.9608, + "step": 214 + }, + { + "epoch": 4.0, + "learning_rate": 0.000213, + "loss": 2.9406, + "step": 215 + }, + { + "epoch": 4.0, + "learning_rate": 0.000214, + "loss": 2.9443, + "step": 216 + }, + { + "epoch": 4.0, + "learning_rate": 0.000215, + "loss": 2.9521, + "step": 217 + }, + { + "epoch": 4.0, + "learning_rate": 0.000216, + "loss": 2.9137, + "step": 218 + }, + { + "epoch": 4.0, + "learning_rate": 0.00021700000000000002, + "loss": 2.9236, + "step": 219 + }, + { + "epoch": 4.0, + "learning_rate": 0.000218, + "loss": 2.9117, + "step": 220 + }, + { + "epoch": 4.0, + "learning_rate": 0.000219, + "loss": 2.8929, + "step": 221 + }, + { + "epoch": 4.0, + "learning_rate": 0.00022, + "loss": 2.9047, + "step": 222 + }, + { + "epoch": 4.0, + "learning_rate": 0.000221, + "loss": 2.9151, + "step": 223 + }, + { + "epoch": 4.0, + "learning_rate": 0.000222, + "loss": 2.8896, + "step": 224 + }, + { + "epoch": 4.0, + "learning_rate": 0.000223, + "loss": 2.9011, + "step": 225 + }, + { + "epoch": 4.0, + "learning_rate": 0.000224, + "loss": 2.892, + "step": 226 + }, + { + "epoch": 4.0, + "learning_rate": 0.00022500000000000002, + "loss": 2.8629, + "step": 227 + }, + { + "epoch": 4.0, + "learning_rate": 0.00022600000000000002, + "loss": 2.8839, + "step": 228 + }, + { + "epoch": 4.0, + "learning_rate": 0.00022700000000000002, + "loss": 2.8943, + "step": 229 + }, + { + "epoch": 4.01, + "learning_rate": 0.000228, + "loss": 2.8849, + "step": 230 + }, + { + "epoch": 4.01, + "learning_rate": 0.000229, + "loss": 2.8109, + "step": 231 + }, + { + "epoch": 4.01, + "learning_rate": 0.00023, + "loss": 2.8471, + "step": 232 + }, + { + "epoch": 4.01, + "learning_rate": 0.000231, + "loss": 2.8288, + "step": 233 + }, + { + "epoch": 4.01, + "learning_rate": 0.00023200000000000003, + "loss": 2.7901, + "step": 234 + }, + { + "epoch": 4.01, + "learning_rate": 0.00023300000000000003, + "loss": 2.8227, + "step": 235 + }, + { + "epoch": 4.01, + "learning_rate": 0.00023400000000000002, + "loss": 2.7824, + "step": 236 + }, + { + "epoch": 4.01, + "learning_rate": 0.000235, + "loss": 2.7338, + "step": 237 + }, + { + "epoch": 4.01, + "learning_rate": 0.000236, + "loss": 2.789, + "step": 238 + }, + { + "epoch": 4.01, + "learning_rate": 0.000237, + "loss": 2.7613, + "step": 239 + }, + { + "epoch": 4.01, + "learning_rate": 0.00023799999999999998, + "loss": 2.7738, + "step": 240 + }, + { + "epoch": 4.01, + "learning_rate": 0.00023899999999999998, + "loss": 2.7113, + "step": 241 + }, + { + "epoch": 4.01, + "learning_rate": 0.00024, + "loss": 2.7146, + "step": 242 + }, + { + "epoch": 4.01, + "learning_rate": 0.000241, + "loss": 2.6654, + "step": 243 + }, + { + "epoch": 4.01, + "learning_rate": 0.000242, + "loss": 2.6511, + "step": 244 + }, + { + "epoch": 4.01, + "learning_rate": 0.000243, + "loss": 2.66, + "step": 245 + }, + { + "epoch": 4.01, + "learning_rate": 0.000244, + "loss": 2.7349, + "step": 246 + }, + { + "epoch": 4.01, + "learning_rate": 0.000245, + "loss": 2.5457, + "step": 247 + }, + { + "epoch": 4.01, + "learning_rate": 0.000246, + "loss": 2.5344, + "step": 248 + }, + { + "epoch": 4.01, + "learning_rate": 0.000247, + "loss": 2.5405, + "step": 249 + }, + { + "epoch": 4.01, + "learning_rate": 0.000248, + "loss": 2.5133, + "step": 250 + }, + { + "epoch": 4.01, + "learning_rate": 0.000249, + "loss": 2.4938, + "step": 251 + }, + { + "epoch": 4.01, + "learning_rate": 0.00025, + "loss": 2.4441, + "step": 252 + }, + { + "epoch": 4.01, + "learning_rate": 0.00025100000000000003, + "loss": 2.3799, + "step": 253 + }, + { + "epoch": 4.01, + "learning_rate": 0.000252, + "loss": 2.3894, + "step": 254 + }, + { + "epoch": 4.01, + "learning_rate": 0.000253, + "loss": 2.4055, + "step": 255 + }, + { + "epoch": 5.0, + "learning_rate": 0.000254, + "loss": 3.4286, + "step": 256 + }, + { + "epoch": 5.0, + "learning_rate": 0.000255, + "loss": 2.2523, + "step": 257 + }, + { + "epoch": 5.0, + "learning_rate": 0.000256, + "loss": 2.1968, + "step": 258 + }, + { + "epoch": 5.0, + "learning_rate": 0.000257, + "loss": 2.1651, + "step": 259 + }, + { + "epoch": 5.0, + "learning_rate": 0.00025800000000000004, + "loss": 2.1438, + "step": 260 + }, + { + "epoch": 5.0, + "learning_rate": 0.000259, + "loss": 2.1382, + "step": 261 + }, + { + "epoch": 5.0, + "learning_rate": 0.00026000000000000003, + "loss": 2.0954, + "step": 262 + }, + { + "epoch": 5.0, + "learning_rate": 0.000261, + "loss": 2.0832, + "step": 263 + }, + { + "epoch": 5.0, + "learning_rate": 0.000262, + "loss": 2.0374, + "step": 264 + }, + { + "epoch": 5.0, + "learning_rate": 0.000263, + "loss": 2.0102, + "step": 265 + }, + { + "epoch": 5.0, + "learning_rate": 0.000264, + "loss": 1.8697, + "step": 266 + }, + { + "epoch": 5.0, + "learning_rate": 0.00026500000000000004, + "loss": 1.9066, + "step": 267 + }, + { + "epoch": 5.0, + "learning_rate": 0.000266, + "loss": 1.8453, + "step": 268 + }, + { + "epoch": 5.0, + "learning_rate": 0.00026700000000000004, + "loss": 1.8282, + "step": 269 + }, + { + "epoch": 5.0, + "learning_rate": 0.000268, + "loss": 1.8173, + "step": 270 + }, + { + "epoch": 5.0, + "learning_rate": 0.00026900000000000003, + "loss": 1.7595, + "step": 271 + }, + { + "epoch": 5.0, + "learning_rate": 0.00027, + "loss": 1.7467, + "step": 272 + }, + { + "epoch": 5.0, + "learning_rate": 0.00027100000000000003, + "loss": 1.6821, + "step": 273 + }, + { + "epoch": 5.0, + "learning_rate": 0.00027200000000000005, + "loss": 1.6618, + "step": 274 + }, + { + "epoch": 5.0, + "learning_rate": 0.000273, + "loss": 1.5253, + "step": 275 + }, + { + "epoch": 5.0, + "learning_rate": 0.00027400000000000005, + "loss": 1.5638, + "step": 276 + }, + { + "epoch": 5.0, + "learning_rate": 0.000275, + "loss": 1.5509, + "step": 277 + }, + { + "epoch": 5.0, + "learning_rate": 0.00027600000000000004, + "loss": 1.5943, + "step": 278 + }, + { + "epoch": 5.0, + "learning_rate": 0.000277, + "loss": 1.4793, + "step": 279 + }, + { + "epoch": 5.0, + "learning_rate": 0.00027800000000000004, + "loss": 1.552, + "step": 280 + }, + { + "epoch": 5.01, + "learning_rate": 0.000279, + "loss": 1.455, + "step": 281 + }, + { + "epoch": 5.01, + "learning_rate": 0.00028000000000000003, + "loss": 1.4579, + "step": 282 + }, + { + "epoch": 5.01, + "learning_rate": 0.00028100000000000005, + "loss": 1.3605, + "step": 283 + }, + { + "epoch": 5.01, + "learning_rate": 0.00028199999999999997, + "loss": 1.3283, + "step": 284 + }, + { + "epoch": 5.01, + "learning_rate": 0.000283, + "loss": 1.257, + "step": 285 + }, + { + "epoch": 5.01, + "learning_rate": 0.00028399999999999996, + "loss": 1.4056, + "step": 286 + }, + { + "epoch": 5.01, + "learning_rate": 0.000285, + "loss": 1.2556, + "step": 287 + }, + { + "epoch": 5.01, + "learning_rate": 0.00028599999999999996, + "loss": 1.2564, + "step": 288 + }, + { + "epoch": 5.01, + "learning_rate": 0.000287, + "loss": 1.3543, + "step": 289 + }, + { + "epoch": 5.01, + "learning_rate": 0.000288, + "loss": 1.2534, + "step": 290 + }, + { + "epoch": 5.01, + "learning_rate": 0.000289, + "loss": 1.3025, + "step": 291 + }, + { + "epoch": 5.01, + "learning_rate": 0.00029, + "loss": 1.2839, + "step": 292 + }, + { + "epoch": 5.01, + "learning_rate": 0.00029099999999999997, + "loss": 1.2912, + "step": 293 + }, + { + "epoch": 5.01, + "learning_rate": 0.000292, + "loss": 1.1856, + "step": 294 + }, + { + "epoch": 5.01, + "learning_rate": 0.00029299999999999997, + "loss": 1.2636, + "step": 295 + }, + { + "epoch": 5.01, + "learning_rate": 0.000294, + "loss": 1.2823, + "step": 296 + }, + { + "epoch": 5.01, + "learning_rate": 0.000295, + "loss": 1.1815, + "step": 297 + }, + { + "epoch": 5.01, + "learning_rate": 0.000296, + "loss": 1.291, + "step": 298 + }, + { + "epoch": 5.01, + "learning_rate": 0.000297, + "loss": 1.1361, + "step": 299 + }, + { + "epoch": 5.01, + "learning_rate": 0.000298, + "loss": 1.1664, + "step": 300 + }, + { + "epoch": 5.01, + "learning_rate": 0.000299, + "loss": 1.2074, + "step": 301 + }, + { + "epoch": 5.01, + "learning_rate": 0.0003, + "loss": 1.1134, + "step": 302 + }, + { + "epoch": 5.01, + "learning_rate": 0.000301, + "loss": 1.0702, + "step": 303 + }, + { + "epoch": 5.01, + "learning_rate": 0.000302, + "loss": 1.1094, + "step": 304 + }, + { + "epoch": 5.01, + "learning_rate": 0.000303, + "loss": 1.0676, + "step": 305 + }, + { + "epoch": 5.01, + "learning_rate": 0.000304, + "loss": 1.1341, + "step": 306 + }, + { + "epoch": 6.0, + "learning_rate": 0.000305, + "loss": 1.6572, + "step": 307 + }, + { + "epoch": 6.0, + "learning_rate": 0.000306, + "loss": 1.0313, + "step": 308 + }, + { + "epoch": 6.0, + "learning_rate": 0.000307, + "loss": 1.0713, + "step": 309 + }, + { + "epoch": 6.0, + "learning_rate": 0.000308, + "loss": 1.0569, + "step": 310 + }, + { + "epoch": 6.0, + "learning_rate": 0.00030900000000000003, + "loss": 1.0765, + "step": 311 + }, + { + "epoch": 6.0, + "learning_rate": 0.00031, + "loss": 1.1707, + "step": 312 + }, + { + "epoch": 6.0, + "learning_rate": 0.000311, + "loss": 1.0867, + "step": 313 + }, + { + "epoch": 6.0, + "learning_rate": 0.000312, + "loss": 1.0459, + "step": 314 + }, + { + "epoch": 6.0, + "learning_rate": 0.000313, + "loss": 1.0016, + "step": 315 + }, + { + "epoch": 6.0, + "learning_rate": 0.000314, + "loss": 1.0112, + "step": 316 + }, + { + "epoch": 6.0, + "learning_rate": 0.000315, + "loss": 1.0838, + "step": 317 + }, + { + "epoch": 6.0, + "learning_rate": 0.000316, + "loss": 1.0675, + "step": 318 + }, + { + "epoch": 6.0, + "learning_rate": 0.000317, + "loss": 0.9579, + "step": 319 + }, + { + "epoch": 6.0, + "learning_rate": 0.00031800000000000003, + "loss": 1.0439, + "step": 320 + }, + { + "epoch": 6.0, + "learning_rate": 0.000319, + "loss": 1.1176, + "step": 321 + }, + { + "epoch": 6.0, + "learning_rate": 0.00032, + "loss": 1.1274, + "step": 322 + }, + { + "epoch": 6.0, + "learning_rate": 0.000321, + "loss": 1.0202, + "step": 323 + }, + { + "epoch": 6.0, + "learning_rate": 0.000322, + "loss": 0.9822, + "step": 324 + }, + { + "epoch": 6.0, + "learning_rate": 0.000323, + "loss": 1.0668, + "step": 325 + }, + { + "epoch": 6.0, + "learning_rate": 0.000324, + "loss": 1.0563, + "step": 326 + }, + { + "epoch": 6.0, + "learning_rate": 0.00032500000000000004, + "loss": 0.9538, + "step": 327 + }, + { + "epoch": 6.0, + "learning_rate": 0.000326, + "loss": 0.9723, + "step": 328 + }, + { + "epoch": 6.0, + "learning_rate": 0.00032700000000000003, + "loss": 1.0416, + "step": 329 + }, + { + "epoch": 6.0, + "learning_rate": 0.000328, + "loss": 1.014, + "step": 330 + }, + { + "epoch": 6.0, + "learning_rate": 0.00032900000000000003, + "loss": 0.9332, + "step": 331 + }, + { + "epoch": 6.01, + "learning_rate": 0.00033, + "loss": 0.9775, + "step": 332 + }, + { + "epoch": 6.01, + "learning_rate": 0.000331, + "loss": 0.9065, + "step": 333 + }, + { + "epoch": 6.01, + "learning_rate": 0.00033200000000000005, + "loss": 0.9182, + "step": 334 + }, + { + "epoch": 6.01, + "learning_rate": 0.000333, + "loss": 0.8738, + "step": 335 + }, + { + "epoch": 6.01, + "learning_rate": 0.00033400000000000004, + "loss": 0.9793, + "step": 336 + }, + { + "epoch": 6.01, + "learning_rate": 0.000335, + "loss": 1.0159, + "step": 337 + }, + { + "epoch": 6.01, + "learning_rate": 0.00033600000000000004, + "loss": 0.9988, + "step": 338 + }, + { + "epoch": 6.01, + "learning_rate": 0.000337, + "loss": 0.9081, + "step": 339 + }, + { + "epoch": 6.01, + "learning_rate": 0.00033800000000000003, + "loss": 1.0337, + "step": 340 + }, + { + "epoch": 6.01, + "learning_rate": 0.00033900000000000005, + "loss": 0.9922, + "step": 341 + }, + { + "epoch": 6.01, + "learning_rate": 0.00034, + "loss": 0.9477, + "step": 342 + }, + { + "epoch": 6.01, + "learning_rate": 0.00034100000000000005, + "loss": 0.9709, + "step": 343 + }, + { + "epoch": 6.01, + "learning_rate": 0.000342, + "loss": 0.9871, + "step": 344 + }, + { + "epoch": 6.01, + "learning_rate": 0.00034300000000000004, + "loss": 0.9441, + "step": 345 + }, + { + "epoch": 6.01, + "learning_rate": 0.00034399999999999996, + "loss": 0.9492, + "step": 346 + }, + { + "epoch": 6.01, + "learning_rate": 0.000345, + "loss": 0.8762, + "step": 347 + }, + { + "epoch": 6.01, + "learning_rate": 0.000346, + "loss": 0.8677, + "step": 348 + }, + { + "epoch": 6.01, + "learning_rate": 0.000347, + "loss": 0.9295, + "step": 349 + }, + { + "epoch": 6.01, + "learning_rate": 0.000348, + "loss": 0.9432, + "step": 350 + }, + { + "epoch": 6.01, + "learning_rate": 0.00034899999999999997, + "loss": 0.9466, + "step": 351 + }, + { + "epoch": 6.01, + "learning_rate": 0.00035, + "loss": 0.9067, + "step": 352 + }, + { + "epoch": 6.01, + "learning_rate": 0.00035099999999999997, + "loss": 0.9241, + "step": 353 + }, + { + "epoch": 6.01, + "learning_rate": 0.000352, + "loss": 0.9323, + "step": 354 + }, + { + "epoch": 6.01, + "learning_rate": 0.00035299999999999996, + "loss": 0.9355, + "step": 355 + }, + { + "epoch": 6.01, + "learning_rate": 0.000354, + "loss": 0.9181, + "step": 356 + }, + { + "epoch": 6.01, + "learning_rate": 0.000355, + "loss": 0.9289, + "step": 357 + }, + { + "epoch": 7.0, + "learning_rate": 0.000356, + "loss": 1.2843, + "step": 358 + }, + { + "epoch": 7.0, + "learning_rate": 0.000357, + "loss": 0.8914, + "step": 359 + }, + { + "epoch": 7.0, + "learning_rate": 0.000358, + "loss": 0.8758, + "step": 360 + }, + { + "epoch": 7.0, + "learning_rate": 0.000359, + "loss": 0.7892, + "step": 361 + }, + { + "epoch": 7.0, + "learning_rate": 0.00035999999999999997, + "loss": 0.7409, + "step": 362 + }, + { + "epoch": 7.0, + "learning_rate": 0.000361, + "loss": 0.8115, + "step": 363 + }, + { + "epoch": 7.0, + "learning_rate": 0.000362, + "loss": 0.7966, + "step": 364 + }, + { + "epoch": 7.0, + "learning_rate": 0.000363, + "loss": 0.8079, + "step": 365 + }, + { + "epoch": 7.0, + "learning_rate": 0.000364, + "loss": 0.8488, + "step": 366 + }, + { + "epoch": 7.0, + "learning_rate": 0.000365, + "loss": 0.8192, + "step": 367 + }, + { + "epoch": 7.0, + "learning_rate": 0.000366, + "loss": 0.9027, + "step": 368 + }, + { + "epoch": 7.0, + "learning_rate": 0.000367, + "loss": 0.8116, + "step": 369 + }, + { + "epoch": 7.0, + "learning_rate": 0.000368, + "loss": 0.8974, + "step": 370 + }, + { + "epoch": 7.0, + "learning_rate": 0.000369, + "loss": 0.9091, + "step": 371 + }, + { + "epoch": 7.0, + "learning_rate": 0.00037, + "loss": 0.8979, + "step": 372 + }, + { + "epoch": 7.0, + "learning_rate": 0.000371, + "loss": 0.8377, + "step": 373 + }, + { + "epoch": 7.0, + "learning_rate": 0.000372, + "loss": 0.8826, + "step": 374 + }, + { + "epoch": 7.0, + "learning_rate": 0.000373, + "loss": 0.8026, + "step": 375 + }, + { + "epoch": 7.0, + "learning_rate": 0.000374, + "loss": 0.7074, + "step": 376 + }, + { + "epoch": 7.0, + "learning_rate": 0.000375, + "loss": 0.8376, + "step": 377 + }, + { + "epoch": 7.0, + "learning_rate": 0.00037600000000000003, + "loss": 0.9013, + "step": 378 + }, + { + "epoch": 7.0, + "learning_rate": 0.000377, + "loss": 0.8975, + "step": 379 + }, + { + "epoch": 7.0, + "learning_rate": 0.000378, + "loss": 0.8073, + "step": 380 + }, + { + "epoch": 7.0, + "learning_rate": 0.000379, + "loss": 0.9192, + "step": 381 + }, + { + "epoch": 7.0, + "learning_rate": 0.00038, + "loss": 0.8539, + "step": 382 + }, + { + "epoch": 7.01, + "learning_rate": 0.000381, + "loss": 0.8854, + "step": 383 + }, + { + "epoch": 7.01, + "learning_rate": 0.000382, + "loss": 0.8367, + "step": 384 + }, + { + "epoch": 7.01, + "learning_rate": 0.00038300000000000004, + "loss": 0.8858, + "step": 385 + }, + { + "epoch": 7.01, + "learning_rate": 0.000384, + "loss": 0.7788, + "step": 386 + }, + { + "epoch": 7.01, + "learning_rate": 0.00038500000000000003, + "loss": 0.73, + "step": 387 + }, + { + "epoch": 7.01, + "learning_rate": 0.000386, + "loss": 0.8452, + "step": 388 + }, + { + "epoch": 7.01, + "learning_rate": 0.00038700000000000003, + "loss": 0.8089, + "step": 389 + }, + { + "epoch": 7.01, + "learning_rate": 0.000388, + "loss": 0.7731, + "step": 390 + }, + { + "epoch": 7.01, + "learning_rate": 0.000389, + "loss": 0.818, + "step": 391 + }, + { + "epoch": 7.01, + "learning_rate": 0.00039000000000000005, + "loss": 0.7598, + "step": 392 + }, + { + "epoch": 7.01, + "learning_rate": 0.000391, + "loss": 0.8162, + "step": 393 + }, + { + "epoch": 7.01, + "learning_rate": 0.00039200000000000004, + "loss": 0.8072, + "step": 394 + }, + { + "epoch": 7.01, + "learning_rate": 0.000393, + "loss": 0.8518, + "step": 395 + }, + { + "epoch": 7.01, + "learning_rate": 0.00039400000000000004, + "loss": 0.8983, + "step": 396 + }, + { + "epoch": 7.01, + "learning_rate": 0.000395, + "loss": 0.8349, + "step": 397 + }, + { + "epoch": 7.01, + "learning_rate": 0.00039600000000000003, + "loss": 0.8727, + "step": 398 + }, + { + "epoch": 7.01, + "learning_rate": 0.00039700000000000005, + "loss": 0.677, + "step": 399 + }, + { + "epoch": 7.01, + "learning_rate": 0.000398, + "loss": 0.9066, + "step": 400 + }, + { + "epoch": 7.01, + "learning_rate": 0.00039900000000000005, + "loss": 0.7787, + "step": 401 + }, + { + "epoch": 7.01, + "learning_rate": 0.0004, + "loss": 0.7617, + "step": 402 + }, + { + "epoch": 7.01, + "learning_rate": 0.00040100000000000004, + "loss": 0.7957, + "step": 403 + }, + { + "epoch": 7.01, + "learning_rate": 0.000402, + "loss": 0.7772, + "step": 404 + }, + { + "epoch": 7.01, + "learning_rate": 0.00040300000000000004, + "loss": 0.7372, + "step": 405 + }, + { + "epoch": 7.01, + "learning_rate": 0.000404, + "loss": 0.7805, + "step": 406 + }, + { + "epoch": 7.01, + "learning_rate": 0.00040500000000000003, + "loss": 0.7119, + "step": 407 + }, + { + "epoch": 7.01, + "learning_rate": 0.00040600000000000006, + "loss": 0.7632, + "step": 408 + }, + { + "epoch": 8.0, + "learning_rate": 0.00040699999999999997, + "loss": 1.1222, + "step": 409 + }, + { + "epoch": 8.0, + "learning_rate": 0.000408, + "loss": 0.7341, + "step": 410 + }, + { + "epoch": 8.0, + "learning_rate": 0.00040899999999999997, + "loss": 0.7642, + "step": 411 + }, + { + "epoch": 8.0, + "learning_rate": 0.00041, + "loss": 0.7111, + "step": 412 + }, + { + "epoch": 8.0, + "learning_rate": 0.00041099999999999996, + "loss": 0.7078, + "step": 413 + }, + { + "epoch": 8.0, + "learning_rate": 0.000412, + "loss": 0.694, + "step": 414 + }, + { + "epoch": 8.0, + "learning_rate": 0.000413, + "loss": 0.758, + "step": 415 + }, + { + "epoch": 8.0, + "learning_rate": 0.000414, + "loss": 0.7754, + "step": 416 + }, + { + "epoch": 8.0, + "learning_rate": 0.000415, + "loss": 0.7304, + "step": 417 + }, + { + "epoch": 8.0, + "learning_rate": 0.000416, + "loss": 0.7399, + "step": 418 + }, + { + "epoch": 8.0, + "learning_rate": 0.000417, + "loss": 0.6947, + "step": 419 + }, + { + "epoch": 8.0, + "learning_rate": 0.00041799999999999997, + "loss": 0.8267, + "step": 420 + }, + { + "epoch": 8.0, + "learning_rate": 0.000419, + "loss": 0.7688, + "step": 421 + }, + { + "epoch": 8.0, + "learning_rate": 0.00042, + "loss": 0.742, + "step": 422 + }, + { + "epoch": 8.0, + "learning_rate": 0.000421, + "loss": 0.7814, + "step": 423 + }, + { + "epoch": 8.0, + "learning_rate": 0.000422, + "loss": 0.7282, + "step": 424 + }, + { + "epoch": 8.0, + "learning_rate": 0.000423, + "loss": 0.7311, + "step": 425 + }, + { + "epoch": 8.0, + "learning_rate": 0.000424, + "loss": 0.7463, + "step": 426 + }, + { + "epoch": 8.0, + "learning_rate": 0.000425, + "loss": 0.6976, + "step": 427 + }, + { + "epoch": 8.0, + "learning_rate": 0.000426, + "loss": 0.6953, + "step": 428 + }, + { + "epoch": 8.0, + "learning_rate": 0.000427, + "loss": 0.7492, + "step": 429 + }, + { + "epoch": 8.0, + "learning_rate": 0.000428, + "loss": 0.7183, + "step": 430 + }, + { + "epoch": 8.0, + "learning_rate": 0.000429, + "loss": 0.7461, + "step": 431 + }, + { + "epoch": 8.0, + "learning_rate": 0.00043, + "loss": 0.7882, + "step": 432 + }, + { + "epoch": 8.01, + "learning_rate": 0.000431, + "loss": 0.7483, + "step": 433 + }, + { + "epoch": 8.01, + "learning_rate": 0.000432, + "loss": 0.6394, + "step": 434 + }, + { + "epoch": 8.01, + "learning_rate": 0.000433, + "loss": 0.6884, + "step": 435 + }, + { + "epoch": 8.01, + "learning_rate": 0.00043400000000000003, + "loss": 0.726, + "step": 436 + }, + { + "epoch": 8.01, + "learning_rate": 0.000435, + "loss": 0.738, + "step": 437 + }, + { + "epoch": 8.01, + "learning_rate": 0.000436, + "loss": 0.7224, + "step": 438 + }, + { + "epoch": 8.01, + "learning_rate": 0.000437, + "loss": 0.726, + "step": 439 + }, + { + "epoch": 8.01, + "learning_rate": 0.000438, + "loss": 0.7532, + "step": 440 + }, + { + "epoch": 8.01, + "learning_rate": 0.000439, + "loss": 0.7457, + "step": 441 + }, + { + "epoch": 8.01, + "learning_rate": 0.00044, + "loss": 0.7415, + "step": 442 + }, + { + "epoch": 8.01, + "learning_rate": 0.000441, + "loss": 0.745, + "step": 443 + }, + { + "epoch": 8.01, + "learning_rate": 0.000442, + "loss": 0.806, + "step": 444 + }, + { + "epoch": 8.01, + "learning_rate": 0.00044300000000000003, + "loss": 0.6942, + "step": 445 + }, + { + "epoch": 8.01, + "learning_rate": 0.000444, + "loss": 0.7294, + "step": 446 + }, + { + "epoch": 8.01, + "learning_rate": 0.00044500000000000003, + "loss": 0.6722, + "step": 447 + }, + { + "epoch": 8.01, + "learning_rate": 0.000446, + "loss": 0.8172, + "step": 448 + }, + { + "epoch": 8.01, + "learning_rate": 0.000447, + "loss": 0.7054, + "step": 449 + }, + { + "epoch": 8.01, + "learning_rate": 0.000448, + "loss": 0.7239, + "step": 450 + }, + { + "epoch": 8.01, + "learning_rate": 0.000449, + "loss": 0.6902, + "step": 451 + }, + { + "epoch": 8.01, + "learning_rate": 0.00045000000000000004, + "loss": 0.6845, + "step": 452 + }, + { + "epoch": 8.01, + "learning_rate": 0.000451, + "loss": 0.802, + "step": 453 + }, + { + "epoch": 8.01, + "learning_rate": 0.00045200000000000004, + "loss": 0.7052, + "step": 454 + }, + { + "epoch": 8.01, + "learning_rate": 0.000453, + "loss": 0.7022, + "step": 455 + }, + { + "epoch": 8.01, + "learning_rate": 0.00045400000000000003, + "loss": 0.7216, + "step": 456 + }, + { + "epoch": 8.01, + "learning_rate": 0.000455, + "loss": 0.7299, + "step": 457 + }, + { + "epoch": 8.01, + "learning_rate": 0.000456, + "loss": 0.742, + "step": 458 + }, + { + "epoch": 8.01, + "learning_rate": 0.00045700000000000005, + "loss": 0.7343, + "step": 459 + }, + { + "epoch": 9.0, + "learning_rate": 0.000458, + "loss": 1.0297, + "step": 460 + }, + { + "epoch": 9.0, + "learning_rate": 0.00045900000000000004, + "loss": 0.7289, + "step": 461 + }, + { + "epoch": 9.0, + "learning_rate": 0.00046, + "loss": 0.7411, + "step": 462 + }, + { + "epoch": 9.0, + "learning_rate": 0.00046100000000000004, + "loss": 0.5639, + "step": 463 + }, + { + "epoch": 9.0, + "learning_rate": 0.000462, + "loss": 0.6847, + "step": 464 + }, + { + "epoch": 9.0, + "learning_rate": 0.00046300000000000003, + "loss": 0.7322, + "step": 465 + }, + { + "epoch": 9.0, + "learning_rate": 0.00046400000000000006, + "loss": 0.6765, + "step": 466 + }, + { + "epoch": 9.0, + "learning_rate": 0.000465, + "loss": 0.6043, + "step": 467 + }, + { + "epoch": 9.0, + "learning_rate": 0.00046600000000000005, + "loss": 0.6472, + "step": 468 + }, + { + "epoch": 9.0, + "learning_rate": 0.000467, + "loss": 0.7528, + "step": 469 + }, + { + "epoch": 9.0, + "learning_rate": 0.00046800000000000005, + "loss": 0.7047, + "step": 470 + }, + { + "epoch": 9.0, + "learning_rate": 0.00046899999999999996, + "loss": 0.6477, + "step": 471 + }, + { + "epoch": 9.0, + "learning_rate": 0.00047, + "loss": 0.6668, + "step": 472 + }, + { + "epoch": 9.0, + "learning_rate": 0.000471, + "loss": 0.6465, + "step": 473 + }, + { + "epoch": 9.0, + "learning_rate": 0.000472, + "loss": 0.7854, + "step": 474 + }, + { + "epoch": 9.0, + "learning_rate": 0.000473, + "loss": 0.68, + "step": 475 + }, + { + "epoch": 9.0, + "learning_rate": 0.000474, + "loss": 0.7194, + "step": 476 + }, + { + "epoch": 9.0, + "learning_rate": 0.000475, + "loss": 0.6752, + "step": 477 + }, + { + "epoch": 9.0, + "learning_rate": 0.00047599999999999997, + "loss": 0.6185, + "step": 478 + }, + { + "epoch": 9.0, + "learning_rate": 0.000477, + "loss": 0.7177, + "step": 479 + }, + { + "epoch": 9.0, + "learning_rate": 0.00047799999999999996, + "loss": 0.6221, + "step": 480 + }, + { + "epoch": 9.0, + "learning_rate": 0.000479, + "loss": 0.6638, + "step": 481 + }, + { + "epoch": 9.0, + "learning_rate": 0.00048, + "loss": 0.7152, + "step": 482 + }, + { + "epoch": 9.0, + "learning_rate": 0.000481, + "loss": 0.816, + "step": 483 + }, + { + "epoch": 9.01, + "learning_rate": 0.000482, + "loss": 0.731, + "step": 484 + }, + { + "epoch": 9.01, + "learning_rate": 0.000483, + "loss": 0.7709, + "step": 485 + }, + { + "epoch": 9.01, + "learning_rate": 0.000484, + "loss": 0.6906, + "step": 486 + }, + { + "epoch": 9.01, + "learning_rate": 0.00048499999999999997, + "loss": 0.6604, + "step": 487 + }, + { + "epoch": 9.01, + "learning_rate": 0.000486, + "loss": 0.6435, + "step": 488 + }, + { + "epoch": 9.01, + "learning_rate": 0.000487, + "loss": 0.632, + "step": 489 + }, + { + "epoch": 9.01, + "learning_rate": 0.000488, + "loss": 0.7089, + "step": 490 + }, + { + "epoch": 9.01, + "learning_rate": 0.000489, + "loss": 0.6635, + "step": 491 + }, + { + "epoch": 9.01, + "learning_rate": 0.00049, + "loss": 0.6246, + "step": 492 + }, + { + "epoch": 9.01, + "learning_rate": 0.000491, + "loss": 0.6404, + "step": 493 + }, + { + "epoch": 9.01, + "learning_rate": 0.000492, + "loss": 0.695, + "step": 494 + }, + { + "epoch": 9.01, + "learning_rate": 0.0004930000000000001, + "loss": 0.7437, + "step": 495 + }, + { + "epoch": 9.01, + "learning_rate": 0.000494, + "loss": 0.6726, + "step": 496 + }, + { + "epoch": 9.01, + "learning_rate": 0.000495, + "loss": 0.6414, + "step": 497 + }, + { + "epoch": 9.01, + "learning_rate": 0.000496, + "loss": 0.735, + "step": 498 + }, + { + "epoch": 9.01, + "learning_rate": 0.000497, + "loss": 0.6662, + "step": 499 + }, + { + "epoch": 9.01, + "learning_rate": 0.000498, + "loss": 0.6768, + "step": 500 + }, + { + "epoch": 9.01, + "eval_cer": 0.12347378501316734, + "eval_loss": 0.42200666666030884, + "eval_runtime": 49.6408, + "eval_samples_per_second": 8.38, + "eval_steps_per_second": 0.262, + "eval_wer": 0.5143090760425184, + "step": 500 + }, + { + "epoch": 9.01, + "learning_rate": 0.000499, + "loss": 0.7422, + "step": 501 + }, + { + "epoch": 9.01, + "learning_rate": 0.0005, + "loss": 0.6348, + "step": 502 + }, + { + "epoch": 9.01, + "learning_rate": 0.0004998888888888889, + "loss": 0.5971, + "step": 503 + }, + { + "epoch": 9.01, + "learning_rate": 0.0004997777777777778, + "loss": 0.648, + "step": 504 + }, + { + "epoch": 9.01, + "learning_rate": 0.0004996666666666667, + "loss": 0.646, + "step": 505 + }, + { + "epoch": 9.01, + "learning_rate": 0.0004995555555555555, + "loss": 0.6606, + "step": 506 + }, + { + "epoch": 9.01, + "learning_rate": 0.0004994444444444445, + "loss": 0.6227, + "step": 507 + }, + { + "epoch": 9.01, + "learning_rate": 0.0004993333333333334, + "loss": 0.6629, + "step": 508 + }, + { + "epoch": 9.01, + "learning_rate": 0.0004992222222222223, + "loss": 0.7011, + "step": 509 + }, + { + "epoch": 9.01, + "learning_rate": 0.0004991111111111111, + "loss": 0.678, + "step": 510 + }, + { + "epoch": 10.0, + "learning_rate": 0.000499, + "loss": 0.909, + "step": 511 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004988888888888889, + "loss": 0.6337, + "step": 512 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004987777777777778, + "loss": 0.5849, + "step": 513 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004986666666666667, + "loss": 0.6565, + "step": 514 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004985555555555556, + "loss": 0.5688, + "step": 515 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004984444444444445, + "loss": 0.626, + "step": 516 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004983333333333334, + "loss": 0.6137, + "step": 517 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004982222222222223, + "loss": 0.6327, + "step": 518 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004981111111111112, + "loss": 0.5867, + "step": 519 + }, + { + "epoch": 10.0, + "learning_rate": 0.000498, + "loss": 0.6565, + "step": 520 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004978888888888889, + "loss": 0.6581, + "step": 521 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004977777777777778, + "loss": 0.6414, + "step": 522 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004976666666666667, + "loss": 0.6249, + "step": 523 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004975555555555556, + "loss": 0.5574, + "step": 524 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004974444444444445, + "loss": 0.6284, + "step": 525 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004973333333333334, + "loss": 0.5809, + "step": 526 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004972222222222222, + "loss": 0.71, + "step": 527 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004971111111111111, + "loss": 0.5489, + "step": 528 + }, + { + "epoch": 10.0, + "learning_rate": 0.000497, + "loss": 0.6269, + "step": 529 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004968888888888889, + "loss": 0.7125, + "step": 530 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004967777777777777, + "loss": 0.5815, + "step": 531 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004966666666666666, + "loss": 0.7194, + "step": 532 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004965555555555555, + "loss": 0.609, + "step": 533 + }, + { + "epoch": 10.0, + "learning_rate": 0.0004964444444444444, + "loss": 0.6414, + "step": 534 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004963333333333333, + "loss": 0.6962, + "step": 535 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004962222222222222, + "loss": 0.697, + "step": 536 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004961111111111111, + "loss": 0.5743, + "step": 537 + }, + { + "epoch": 10.01, + "learning_rate": 0.000496, + "loss": 0.6423, + "step": 538 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004958888888888889, + "loss": 0.6151, + "step": 539 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004957777777777778, + "loss": 0.611, + "step": 540 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004956666666666666, + "loss": 0.5369, + "step": 541 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004955555555555556, + "loss": 0.618, + "step": 542 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004954444444444445, + "loss": 0.6782, + "step": 543 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004953333333333334, + "loss": 0.5876, + "step": 544 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004952222222222222, + "loss": 0.6653, + "step": 545 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004951111111111111, + "loss": 0.5834, + "step": 546 + }, + { + "epoch": 10.01, + "learning_rate": 0.000495, + "loss": 0.6019, + "step": 547 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004948888888888889, + "loss": 0.6462, + "step": 548 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004947777777777778, + "loss": 0.6748, + "step": 549 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004946666666666667, + "loss": 0.5948, + "step": 550 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004945555555555556, + "loss": 0.5838, + "step": 551 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004944444444444445, + "loss": 0.5598, + "step": 552 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004943333333333334, + "loss": 0.6443, + "step": 553 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004942222222222223, + "loss": 0.5588, + "step": 554 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004941111111111111, + "loss": 0.6111, + "step": 555 + }, + { + "epoch": 10.01, + "learning_rate": 0.000494, + "loss": 0.5721, + "step": 556 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004938888888888889, + "loss": 0.5927, + "step": 557 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004937777777777778, + "loss": 0.6765, + "step": 558 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004936666666666667, + "loss": 0.5785, + "step": 559 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004935555555555556, + "loss": 0.5902, + "step": 560 + }, + { + "epoch": 10.01, + "learning_rate": 0.0004934444444444445, + "loss": 0.6125, + "step": 561 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004933333333333334, + "loss": 0.822, + "step": 562 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004932222222222223, + "loss": 0.5337, + "step": 563 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004931111111111112, + "loss": 0.5106, + "step": 564 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004930000000000001, + "loss": 0.6029, + "step": 565 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004928888888888888, + "loss": 0.5844, + "step": 566 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004927777777777777, + "loss": 0.5993, + "step": 567 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004926666666666666, + "loss": 0.5912, + "step": 568 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004925555555555555, + "loss": 0.5606, + "step": 569 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004924444444444444, + "loss": 0.6435, + "step": 570 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004923333333333333, + "loss": 0.6116, + "step": 571 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004922222222222222, + "loss": 0.5487, + "step": 572 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004921111111111111, + "loss": 0.6272, + "step": 573 + }, + { + "epoch": 11.0, + "learning_rate": 0.000492, + "loss": 0.5587, + "step": 574 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004918888888888889, + "loss": 0.5725, + "step": 575 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004917777777777777, + "loss": 0.5835, + "step": 576 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004916666666666666, + "loss": 0.569, + "step": 577 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004915555555555556, + "loss": 0.5931, + "step": 578 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004914444444444445, + "loss": 0.5497, + "step": 579 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004913333333333333, + "loss": 0.5836, + "step": 580 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004912222222222222, + "loss": 0.6033, + "step": 581 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004911111111111111, + "loss": 0.5017, + "step": 582 + }, + { + "epoch": 11.0, + "learning_rate": 0.000491, + "loss": 0.5434, + "step": 583 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004908888888888889, + "loss": 0.5072, + "step": 584 + }, + { + "epoch": 11.0, + "learning_rate": 0.0004907777777777778, + "loss": 0.5064, + "step": 585 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004906666666666667, + "loss": 0.5239, + "step": 586 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004905555555555556, + "loss": 0.5862, + "step": 587 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004904444444444445, + "loss": 0.5847, + "step": 588 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004903333333333334, + "loss": 0.5729, + "step": 589 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004902222222222222, + "loss": 0.5543, + "step": 590 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004901111111111111, + "loss": 0.5255, + "step": 591 + }, + { + "epoch": 11.01, + "learning_rate": 0.00049, + "loss": 0.5912, + "step": 592 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004898888888888889, + "loss": 0.5237, + "step": 593 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004897777777777778, + "loss": 0.5548, + "step": 594 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004896666666666667, + "loss": 0.6288, + "step": 595 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004895555555555556, + "loss": 0.5595, + "step": 596 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004894444444444445, + "loss": 0.5944, + "step": 597 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004893333333333334, + "loss": 0.5753, + "step": 598 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004892222222222223, + "loss": 0.5821, + "step": 599 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004891111111111112, + "loss": 0.5641, + "step": 600 + }, + { + "epoch": 11.01, + "learning_rate": 0.000489, + "loss": 0.5103, + "step": 601 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004888888888888889, + "loss": 0.5494, + "step": 602 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004887777777777778, + "loss": 0.571, + "step": 603 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004886666666666667, + "loss": 0.5394, + "step": 604 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004885555555555556, + "loss": 0.6423, + "step": 605 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004884444444444445, + "loss": 0.5114, + "step": 606 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004883333333333333, + "loss": 0.5468, + "step": 607 + }, + { + "epoch": 11.01, + "learning_rate": 0.00048822222222222225, + "loss": 0.5545, + "step": 608 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004881111111111111, + "loss": 0.5689, + "step": 609 + }, + { + "epoch": 11.01, + "learning_rate": 0.000488, + "loss": 0.5773, + "step": 610 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004878888888888889, + "loss": 0.6025, + "step": 611 + }, + { + "epoch": 11.01, + "learning_rate": 0.0004877777777777778, + "loss": 0.5587, + "step": 612 + }, + { + "epoch": 12.0, + "learning_rate": 0.0004876666666666667, + "loss": 0.8583, + "step": 613 + }, + { + "epoch": 12.0, + "learning_rate": 0.0004875555555555556, + "loss": 0.517, + "step": 614 + }, + { + "epoch": 12.0, + "learning_rate": 0.0004874444444444445, + "loss": 0.536, + "step": 615 + }, + { + "epoch": 12.0, + "learning_rate": 0.0004873333333333333, + "loss": 0.5524, + "step": 616 + }, + { + "epoch": 12.0, + "learning_rate": 0.0004872222222222222, + "loss": 0.44, + "step": 617 + }, + { + "epoch": 12.0, + "learning_rate": 0.0004871111111111111, + "loss": 0.5401, + "step": 618 + }, + { + "epoch": 12.0, + "learning_rate": 0.000487, + "loss": 0.5004, + "step": 619 + }, + { + "epoch": 12.0, + "learning_rate": 0.00048688888888888886, + "loss": 0.5457, + "step": 620 + }, + { + "epoch": 12.0, + "learning_rate": 0.00048677777777777776, + "loss": 0.5847, + "step": 621 + }, + { + "epoch": 12.0, + "learning_rate": 0.0004866666666666667, + "loss": 0.5089, + "step": 622 + }, + { + "epoch": 12.0, + "learning_rate": 0.00048655555555555556, + "loss": 0.5133, + "step": 623 + }, + { + "epoch": 12.0, + "learning_rate": 0.00048644444444444446, + "loss": 0.4999, + "step": 624 + }, + { + "epoch": 12.0, + "learning_rate": 0.00048633333333333335, + "loss": 0.5041, + "step": 625 + }, + { + "epoch": 12.0, + "learning_rate": 0.00048622222222222225, + "loss": 0.554, + "step": 626 + }, + { + "epoch": 12.0, + "learning_rate": 0.0004861111111111111, + "loss": 0.5011, + "step": 627 + }, + { + "epoch": 12.0, + "learning_rate": 0.000486, + "loss": 0.5099, + "step": 628 + }, + { + "epoch": 12.0, + "learning_rate": 0.0004858888888888889, + "loss": 0.5313, + "step": 629 + }, + { + "epoch": 12.0, + "learning_rate": 0.00048577777777777774, + "loss": 0.5215, + "step": 630 + }, + { + "epoch": 12.0, + "learning_rate": 0.0004856666666666667, + "loss": 0.4955, + "step": 631 + }, + { + "epoch": 12.0, + "learning_rate": 0.0004855555555555556, + "loss": 0.5023, + "step": 632 + }, + { + "epoch": 12.0, + "learning_rate": 0.0004854444444444445, + "loss": 0.4876, + "step": 633 + }, + { + "epoch": 12.0, + "learning_rate": 0.00048533333333333333, + "loss": 0.6148, + "step": 634 + }, + { + "epoch": 12.0, + "learning_rate": 0.00048522222222222223, + "loss": 0.5286, + "step": 635 + }, + { + "epoch": 12.0, + "learning_rate": 0.0004851111111111111, + "loss": 0.5259, + "step": 636 + }, + { + "epoch": 12.01, + "learning_rate": 0.00048499999999999997, + "loss": 0.5316, + "step": 637 + }, + { + "epoch": 12.01, + "learning_rate": 0.00048488888888888887, + "loss": 0.6368, + "step": 638 + }, + { + "epoch": 12.01, + "learning_rate": 0.00048477777777777777, + "loss": 0.5529, + "step": 639 + }, + { + "epoch": 12.01, + "learning_rate": 0.0004846666666666667, + "loss": 0.5147, + "step": 640 + }, + { + "epoch": 12.01, + "learning_rate": 0.00048455555555555556, + "loss": 0.5989, + "step": 641 + }, + { + "epoch": 12.01, + "learning_rate": 0.00048444444444444446, + "loss": 0.5232, + "step": 642 + }, + { + "epoch": 12.01, + "learning_rate": 0.00048433333333333336, + "loss": 0.5394, + "step": 643 + }, + { + "epoch": 12.01, + "learning_rate": 0.0004842222222222222, + "loss": 0.5042, + "step": 644 + }, + { + "epoch": 12.01, + "learning_rate": 0.0004841111111111111, + "loss": 0.4697, + "step": 645 + }, + { + "epoch": 12.01, + "learning_rate": 0.000484, + "loss": 0.503, + "step": 646 + }, + { + "epoch": 12.01, + "learning_rate": 0.0004838888888888889, + "loss": 0.5823, + "step": 647 + }, + { + "epoch": 12.01, + "learning_rate": 0.00048377777777777774, + "loss": 0.4907, + "step": 648 + }, + { + "epoch": 12.01, + "learning_rate": 0.0004836666666666667, + "loss": 0.5391, + "step": 649 + }, + { + "epoch": 12.01, + "learning_rate": 0.0004835555555555556, + "loss": 0.4983, + "step": 650 + }, + { + "epoch": 12.01, + "learning_rate": 0.00048344444444444444, + "loss": 0.4825, + "step": 651 + }, + { + "epoch": 12.01, + "learning_rate": 0.00048333333333333334, + "loss": 0.4844, + "step": 652 + }, + { + "epoch": 12.01, + "learning_rate": 0.00048322222222222223, + "loss": 0.5254, + "step": 653 + }, + { + "epoch": 12.01, + "learning_rate": 0.00048311111111111113, + "loss": 0.4195, + "step": 654 + }, + { + "epoch": 12.01, + "learning_rate": 0.000483, + "loss": 0.5406, + "step": 655 + }, + { + "epoch": 12.01, + "learning_rate": 0.0004828888888888889, + "loss": 0.5281, + "step": 656 + }, + { + "epoch": 12.01, + "learning_rate": 0.0004827777777777778, + "loss": 0.5053, + "step": 657 + }, + { + "epoch": 12.01, + "learning_rate": 0.00048266666666666667, + "loss": 0.5827, + "step": 658 + }, + { + "epoch": 12.01, + "learning_rate": 0.00048255555555555557, + "loss": 0.5556, + "step": 659 + }, + { + "epoch": 12.01, + "learning_rate": 0.00048244444444444447, + "loss": 0.547, + "step": 660 + }, + { + "epoch": 12.01, + "learning_rate": 0.00048233333333333337, + "loss": 0.4965, + "step": 661 + }, + { + "epoch": 12.01, + "learning_rate": 0.0004822222222222222, + "loss": 0.5752, + "step": 662 + }, + { + "epoch": 12.01, + "learning_rate": 0.0004821111111111111, + "loss": 0.4804, + "step": 663 + }, + { + "epoch": 13.0, + "learning_rate": 0.000482, + "loss": 0.7943, + "step": 664 + }, + { + "epoch": 13.0, + "learning_rate": 0.00048188888888888885, + "loss": 0.4982, + "step": 665 + }, + { + "epoch": 13.0, + "learning_rate": 0.0004817777777777778, + "loss": 0.4763, + "step": 666 + }, + { + "epoch": 13.0, + "learning_rate": 0.0004816666666666667, + "loss": 0.4385, + "step": 667 + }, + { + "epoch": 13.0, + "learning_rate": 0.0004815555555555556, + "loss": 0.473, + "step": 668 + }, + { + "epoch": 13.0, + "learning_rate": 0.00048144444444444444, + "loss": 0.4966, + "step": 669 + }, + { + "epoch": 13.0, + "learning_rate": 0.00048133333333333334, + "loss": 0.5147, + "step": 670 + }, + { + "epoch": 13.0, + "learning_rate": 0.00048122222222222224, + "loss": 0.4766, + "step": 671 + }, + { + "epoch": 13.0, + "learning_rate": 0.0004811111111111111, + "loss": 0.4806, + "step": 672 + }, + { + "epoch": 13.0, + "learning_rate": 0.000481, + "loss": 0.4702, + "step": 673 + }, + { + "epoch": 13.0, + "learning_rate": 0.0004808888888888889, + "loss": 0.4875, + "step": 674 + }, + { + "epoch": 13.0, + "learning_rate": 0.00048077777777777783, + "loss": 0.456, + "step": 675 + }, + { + "epoch": 13.0, + "learning_rate": 0.0004806666666666667, + "loss": 0.5206, + "step": 676 + }, + { + "epoch": 13.0, + "learning_rate": 0.0004805555555555556, + "loss": 0.5342, + "step": 677 + }, + { + "epoch": 13.0, + "learning_rate": 0.0004804444444444445, + "loss": 0.443, + "step": 678 + }, + { + "epoch": 13.0, + "learning_rate": 0.0004803333333333333, + "loss": 0.4426, + "step": 679 + }, + { + "epoch": 13.0, + "learning_rate": 0.0004802222222222222, + "loss": 0.4946, + "step": 680 + }, + { + "epoch": 13.0, + "learning_rate": 0.0004801111111111111, + "loss": 0.4828, + "step": 681 + }, + { + "epoch": 13.0, + "learning_rate": 0.00048, + "loss": 0.396, + "step": 682 + }, + { + "epoch": 13.0, + "learning_rate": 0.00047988888888888886, + "loss": 0.4399, + "step": 683 + }, + { + "epoch": 13.0, + "learning_rate": 0.0004797777777777778, + "loss": 0.4634, + "step": 684 + }, + { + "epoch": 13.0, + "learning_rate": 0.0004796666666666667, + "loss": 0.4919, + "step": 685 + }, + { + "epoch": 13.0, + "learning_rate": 0.0004795555555555556, + "loss": 0.5211, + "step": 686 + }, + { + "epoch": 13.0, + "learning_rate": 0.00047944444444444445, + "loss": 0.4588, + "step": 687 + }, + { + "epoch": 13.01, + "learning_rate": 0.00047933333333333335, + "loss": 0.4518, + "step": 688 + }, + { + "epoch": 13.01, + "learning_rate": 0.00047922222222222225, + "loss": 0.4562, + "step": 689 + }, + { + "epoch": 13.01, + "learning_rate": 0.0004791111111111111, + "loss": 0.4454, + "step": 690 + }, + { + "epoch": 13.01, + "learning_rate": 0.000479, + "loss": 0.5109, + "step": 691 + }, + { + "epoch": 13.01, + "learning_rate": 0.0004788888888888889, + "loss": 0.4628, + "step": 692 + }, + { + "epoch": 13.01, + "learning_rate": 0.00047877777777777784, + "loss": 0.4737, + "step": 693 + }, + { + "epoch": 13.01, + "learning_rate": 0.0004786666666666667, + "loss": 0.5267, + "step": 694 + }, + { + "epoch": 13.01, + "learning_rate": 0.0004785555555555556, + "loss": 0.5315, + "step": 695 + }, + { + "epoch": 13.01, + "learning_rate": 0.0004784444444444445, + "loss": 0.5021, + "step": 696 + }, + { + "epoch": 13.01, + "learning_rate": 0.0004783333333333333, + "loss": 0.474, + "step": 697 + }, + { + "epoch": 13.01, + "learning_rate": 0.0004782222222222222, + "loss": 0.5011, + "step": 698 + }, + { + "epoch": 13.01, + "learning_rate": 0.0004781111111111111, + "loss": 0.4293, + "step": 699 + }, + { + "epoch": 13.01, + "learning_rate": 0.00047799999999999996, + "loss": 0.4892, + "step": 700 + }, + { + "epoch": 13.01, + "learning_rate": 0.00047788888888888886, + "loss": 0.4883, + "step": 701 + }, + { + "epoch": 13.01, + "learning_rate": 0.0004777777777777778, + "loss": 0.4378, + "step": 702 + }, + { + "epoch": 13.01, + "learning_rate": 0.0004776666666666667, + "loss": 0.4903, + "step": 703 + }, + { + "epoch": 13.01, + "learning_rate": 0.00047755555555555556, + "loss": 0.5362, + "step": 704 + }, + { + "epoch": 13.01, + "learning_rate": 0.00047744444444444445, + "loss": 0.4896, + "step": 705 + }, + { + "epoch": 13.01, + "learning_rate": 0.00047733333333333335, + "loss": 0.5074, + "step": 706 + }, + { + "epoch": 13.01, + "learning_rate": 0.00047722222222222225, + "loss": 0.4652, + "step": 707 + }, + { + "epoch": 13.01, + "learning_rate": 0.0004771111111111111, + "loss": 0.5394, + "step": 708 + }, + { + "epoch": 13.01, + "learning_rate": 0.000477, + "loss": 0.3936, + "step": 709 + }, + { + "epoch": 13.01, + "learning_rate": 0.0004768888888888889, + "loss": 0.4794, + "step": 710 + }, + { + "epoch": 13.01, + "learning_rate": 0.0004767777777777778, + "loss": 0.4271, + "step": 711 + }, + { + "epoch": 13.01, + "learning_rate": 0.0004766666666666667, + "loss": 0.4637, + "step": 712 + }, + { + "epoch": 13.01, + "learning_rate": 0.0004765555555555556, + "loss": 0.4115, + "step": 713 + }, + { + "epoch": 13.01, + "learning_rate": 0.0004764444444444445, + "loss": 0.5076, + "step": 714 + }, + { + "epoch": 14.0, + "learning_rate": 0.00047633333333333333, + "loss": 0.687, + "step": 715 + }, + { + "epoch": 14.0, + "learning_rate": 0.0004762222222222222, + "loss": 0.4929, + "step": 716 + }, + { + "epoch": 14.0, + "learning_rate": 0.0004761111111111111, + "loss": 0.405, + "step": 717 + }, + { + "epoch": 14.0, + "learning_rate": 0.00047599999999999997, + "loss": 0.4908, + "step": 718 + }, + { + "epoch": 14.0, + "learning_rate": 0.00047588888888888887, + "loss": 0.4359, + "step": 719 + }, + { + "epoch": 14.0, + "learning_rate": 0.0004757777777777778, + "loss": 0.4373, + "step": 720 + }, + { + "epoch": 14.0, + "learning_rate": 0.0004756666666666667, + "loss": 0.4575, + "step": 721 + }, + { + "epoch": 14.0, + "learning_rate": 0.00047555555555555556, + "loss": 0.4227, + "step": 722 + }, + { + "epoch": 14.0, + "learning_rate": 0.00047544444444444446, + "loss": 0.4934, + "step": 723 + }, + { + "epoch": 14.0, + "learning_rate": 0.00047533333333333336, + "loss": 0.4231, + "step": 724 + }, + { + "epoch": 14.0, + "learning_rate": 0.0004752222222222222, + "loss": 0.5097, + "step": 725 + }, + { + "epoch": 14.0, + "learning_rate": 0.0004751111111111111, + "loss": 0.4392, + "step": 726 + }, + { + "epoch": 14.0, + "learning_rate": 0.000475, + "loss": 0.4537, + "step": 727 + }, + { + "epoch": 14.0, + "learning_rate": 0.00047488888888888895, + "loss": 0.5366, + "step": 728 + }, + { + "epoch": 14.0, + "learning_rate": 0.0004747777777777778, + "loss": 0.4613, + "step": 729 + }, + { + "epoch": 14.0, + "learning_rate": 0.0004746666666666667, + "loss": 0.429, + "step": 730 + }, + { + "epoch": 14.0, + "learning_rate": 0.0004745555555555556, + "loss": 0.4585, + "step": 731 + }, + { + "epoch": 14.0, + "learning_rate": 0.00047444444444444444, + "loss": 0.4444, + "step": 732 + }, + { + "epoch": 14.0, + "learning_rate": 0.00047433333333333333, + "loss": 0.3779, + "step": 733 + }, + { + "epoch": 14.0, + "learning_rate": 0.00047422222222222223, + "loss": 0.5084, + "step": 734 + }, + { + "epoch": 14.0, + "learning_rate": 0.00047411111111111113, + "loss": 0.4803, + "step": 735 + }, + { + "epoch": 14.0, + "learning_rate": 0.000474, + "loss": 0.4817, + "step": 736 + }, + { + "epoch": 14.0, + "learning_rate": 0.00047388888888888893, + "loss": 0.4398, + "step": 737 + }, + { + "epoch": 14.0, + "learning_rate": 0.0004737777777777778, + "loss": 0.4912, + "step": 738 + }, + { + "epoch": 14.01, + "learning_rate": 0.00047366666666666667, + "loss": 0.4657, + "step": 739 + }, + { + "epoch": 14.01, + "learning_rate": 0.00047355555555555557, + "loss": 0.4324, + "step": 740 + }, + { + "epoch": 14.01, + "learning_rate": 0.00047344444444444447, + "loss": 0.4911, + "step": 741 + }, + { + "epoch": 14.01, + "learning_rate": 0.00047333333333333336, + "loss": 0.4558, + "step": 742 + }, + { + "epoch": 14.01, + "learning_rate": 0.0004732222222222222, + "loss": 0.4461, + "step": 743 + }, + { + "epoch": 14.01, + "learning_rate": 0.0004731111111111111, + "loss": 0.4176, + "step": 744 + }, + { + "epoch": 14.01, + "learning_rate": 0.000473, + "loss": 0.4614, + "step": 745 + }, + { + "epoch": 14.01, + "learning_rate": 0.0004728888888888889, + "loss": 0.4151, + "step": 746 + }, + { + "epoch": 14.01, + "learning_rate": 0.0004727777777777778, + "loss": 0.4474, + "step": 747 + }, + { + "epoch": 14.01, + "learning_rate": 0.0004726666666666667, + "loss": 0.4731, + "step": 748 + }, + { + "epoch": 14.01, + "learning_rate": 0.0004725555555555556, + "loss": 0.4675, + "step": 749 + }, + { + "epoch": 14.01, + "learning_rate": 0.00047244444444444444, + "loss": 0.4354, + "step": 750 + }, + { + "epoch": 14.01, + "learning_rate": 0.00047233333333333334, + "loss": 0.4534, + "step": 751 + }, + { + "epoch": 14.01, + "learning_rate": 0.00047222222222222224, + "loss": 0.4852, + "step": 752 + }, + { + "epoch": 14.01, + "learning_rate": 0.0004721111111111111, + "loss": 0.448, + "step": 753 + }, + { + "epoch": 14.01, + "learning_rate": 0.000472, + "loss": 0.4025, + "step": 754 + }, + { + "epoch": 14.01, + "learning_rate": 0.00047188888888888893, + "loss": 0.3624, + "step": 755 + }, + { + "epoch": 14.01, + "learning_rate": 0.00047177777777777783, + "loss": 0.5024, + "step": 756 + }, + { + "epoch": 14.01, + "learning_rate": 0.0004716666666666667, + "loss": 0.5091, + "step": 757 + }, + { + "epoch": 14.01, + "learning_rate": 0.0004715555555555556, + "loss": 0.447, + "step": 758 + }, + { + "epoch": 14.01, + "learning_rate": 0.00047144444444444447, + "loss": 0.4492, + "step": 759 + }, + { + "epoch": 14.01, + "learning_rate": 0.0004713333333333333, + "loss": 0.4008, + "step": 760 + }, + { + "epoch": 14.01, + "learning_rate": 0.0004712222222222222, + "loss": 0.4168, + "step": 761 + }, + { + "epoch": 14.01, + "learning_rate": 0.0004711111111111111, + "loss": 0.4363, + "step": 762 + }, + { + "epoch": 14.01, + "learning_rate": 0.000471, + "loss": 0.4741, + "step": 763 + }, + { + "epoch": 14.01, + "learning_rate": 0.0004708888888888889, + "loss": 0.4363, + "step": 764 + }, + { + "epoch": 14.01, + "learning_rate": 0.0004707777777777778, + "loss": 0.4208, + "step": 765 + }, + { + "epoch": 15.0, + "learning_rate": 0.0004706666666666667, + "loss": 0.738, + "step": 766 + }, + { + "epoch": 15.0, + "learning_rate": 0.00047055555555555555, + "loss": 0.4492, + "step": 767 + }, + { + "epoch": 15.0, + "learning_rate": 0.00047044444444444445, + "loss": 0.3879, + "step": 768 + }, + { + "epoch": 15.0, + "learning_rate": 0.00047033333333333335, + "loss": 0.446, + "step": 769 + }, + { + "epoch": 15.0, + "learning_rate": 0.00047022222222222224, + "loss": 0.4483, + "step": 770 + }, + { + "epoch": 15.0, + "learning_rate": 0.0004701111111111111, + "loss": 0.4282, + "step": 771 + }, + { + "epoch": 15.0, + "learning_rate": 0.00047, + "loss": 0.4804, + "step": 772 + }, + { + "epoch": 15.0, + "learning_rate": 0.00046988888888888894, + "loss": 0.4342, + "step": 773 + }, + { + "epoch": 15.0, + "learning_rate": 0.0004697777777777778, + "loss": 0.3937, + "step": 774 + }, + { + "epoch": 15.0, + "learning_rate": 0.0004696666666666667, + "loss": 0.418, + "step": 775 + }, + { + "epoch": 15.0, + "learning_rate": 0.0004695555555555556, + "loss": 0.4393, + "step": 776 + }, + { + "epoch": 15.0, + "learning_rate": 0.0004694444444444445, + "loss": 0.4799, + "step": 777 + }, + { + "epoch": 15.0, + "learning_rate": 0.0004693333333333333, + "loss": 0.4578, + "step": 778 + }, + { + "epoch": 15.0, + "learning_rate": 0.0004692222222222222, + "loss": 0.4534, + "step": 779 + }, + { + "epoch": 15.0, + "learning_rate": 0.0004691111111111111, + "loss": 0.4711, + "step": 780 + }, + { + "epoch": 15.0, + "learning_rate": 0.00046899999999999996, + "loss": 0.4806, + "step": 781 + }, + { + "epoch": 15.0, + "learning_rate": 0.0004688888888888889, + "loss": 0.4673, + "step": 782 + }, + { + "epoch": 15.0, + "learning_rate": 0.0004687777777777778, + "loss": 0.4315, + "step": 783 + }, + { + "epoch": 15.0, + "learning_rate": 0.0004686666666666667, + "loss": 0.412, + "step": 784 + }, + { + "epoch": 15.0, + "learning_rate": 0.00046855555555555555, + "loss": 0.4747, + "step": 785 + }, + { + "epoch": 15.0, + "learning_rate": 0.00046844444444444445, + "loss": 0.4315, + "step": 786 + }, + { + "epoch": 15.0, + "learning_rate": 0.00046833333333333335, + "loss": 0.4224, + "step": 787 + }, + { + "epoch": 15.0, + "learning_rate": 0.0004682222222222222, + "loss": 0.4018, + "step": 788 + }, + { + "epoch": 15.0, + "learning_rate": 0.0004681111111111111, + "loss": 0.4367, + "step": 789 + }, + { + "epoch": 15.01, + "learning_rate": 0.00046800000000000005, + "loss": 0.4286, + "step": 790 + }, + { + "epoch": 15.01, + "learning_rate": 0.00046788888888888894, + "loss": 0.4425, + "step": 791 + }, + { + "epoch": 15.01, + "learning_rate": 0.0004677777777777778, + "loss": 0.4043, + "step": 792 + }, + { + "epoch": 15.01, + "learning_rate": 0.0004676666666666667, + "loss": 0.4541, + "step": 793 + }, + { + "epoch": 15.01, + "learning_rate": 0.0004675555555555556, + "loss": 0.4432, + "step": 794 + }, + { + "epoch": 15.01, + "learning_rate": 0.00046744444444444443, + "loss": 0.4252, + "step": 795 + }, + { + "epoch": 15.01, + "learning_rate": 0.0004673333333333333, + "loss": 0.3998, + "step": 796 + }, + { + "epoch": 15.01, + "learning_rate": 0.0004672222222222222, + "loss": 0.4617, + "step": 797 + }, + { + "epoch": 15.01, + "learning_rate": 0.0004671111111111111, + "loss": 0.4479, + "step": 798 + }, + { + "epoch": 15.01, + "learning_rate": 0.000467, + "loss": 0.3705, + "step": 799 + }, + { + "epoch": 15.01, + "learning_rate": 0.0004668888888888889, + "loss": 0.4537, + "step": 800 + }, + { + "epoch": 15.01, + "learning_rate": 0.0004667777777777778, + "loss": 0.3808, + "step": 801 + }, + { + "epoch": 15.01, + "learning_rate": 0.00046666666666666666, + "loss": 0.4345, + "step": 802 + }, + { + "epoch": 15.01, + "learning_rate": 0.00046655555555555556, + "loss": 0.3754, + "step": 803 + }, + { + "epoch": 15.01, + "learning_rate": 0.00046644444444444446, + "loss": 0.3613, + "step": 804 + }, + { + "epoch": 15.01, + "learning_rate": 0.00046633333333333336, + "loss": 0.4107, + "step": 805 + }, + { + "epoch": 15.01, + "learning_rate": 0.0004662222222222222, + "loss": 0.4263, + "step": 806 + }, + { + "epoch": 15.01, + "learning_rate": 0.0004661111111111111, + "loss": 0.3806, + "step": 807 + }, + { + "epoch": 15.01, + "learning_rate": 0.00046600000000000005, + "loss": 0.3881, + "step": 808 + }, + { + "epoch": 15.01, + "learning_rate": 0.0004658888888888889, + "loss": 0.4287, + "step": 809 + }, + { + "epoch": 15.01, + "learning_rate": 0.0004657777777777778, + "loss": 0.4351, + "step": 810 + }, + { + "epoch": 15.01, + "learning_rate": 0.0004656666666666667, + "loss": 0.4161, + "step": 811 + }, + { + "epoch": 15.01, + "learning_rate": 0.0004655555555555556, + "loss": 0.4461, + "step": 812 + }, + { + "epoch": 15.01, + "learning_rate": 0.00046544444444444443, + "loss": 0.4156, + "step": 813 + }, + { + "epoch": 15.01, + "learning_rate": 0.00046533333333333333, + "loss": 0.4045, + "step": 814 + }, + { + "epoch": 15.01, + "learning_rate": 0.00046522222222222223, + "loss": 0.4112, + "step": 815 + }, + { + "epoch": 15.01, + "learning_rate": 0.0004651111111111111, + "loss": 0.4183, + "step": 816 + }, + { + "epoch": 16.0, + "learning_rate": 0.000465, + "loss": 0.6889, + "step": 817 + }, + { + "epoch": 16.0, + "learning_rate": 0.0004648888888888889, + "loss": 0.371, + "step": 818 + }, + { + "epoch": 16.0, + "learning_rate": 0.0004647777777777778, + "loss": 0.4362, + "step": 819 + }, + { + "epoch": 16.0, + "learning_rate": 0.00046466666666666667, + "loss": 0.4056, + "step": 820 + }, + { + "epoch": 16.0, + "learning_rate": 0.00046455555555555557, + "loss": 0.4161, + "step": 821 + }, + { + "epoch": 16.0, + "learning_rate": 0.00046444444444444446, + "loss": 0.4016, + "step": 822 + }, + { + "epoch": 16.0, + "learning_rate": 0.0004643333333333333, + "loss": 0.3923, + "step": 823 + }, + { + "epoch": 16.0, + "learning_rate": 0.0004642222222222222, + "loss": 0.3582, + "step": 824 + }, + { + "epoch": 16.0, + "learning_rate": 0.0004641111111111111, + "loss": 0.4301, + "step": 825 + }, + { + "epoch": 16.0, + "learning_rate": 0.00046400000000000006, + "loss": 0.4284, + "step": 826 + }, + { + "epoch": 16.0, + "learning_rate": 0.0004638888888888889, + "loss": 0.3569, + "step": 827 + }, + { + "epoch": 16.0, + "learning_rate": 0.0004637777777777778, + "loss": 0.3291, + "step": 828 + }, + { + "epoch": 16.0, + "learning_rate": 0.0004636666666666667, + "loss": 0.3972, + "step": 829 + }, + { + "epoch": 16.0, + "learning_rate": 0.00046355555555555554, + "loss": 0.4429, + "step": 830 + }, + { + "epoch": 16.0, + "learning_rate": 0.00046344444444444444, + "loss": 0.4388, + "step": 831 + }, + { + "epoch": 16.0, + "learning_rate": 0.00046333333333333334, + "loss": 0.3789, + "step": 832 + }, + { + "epoch": 16.0, + "learning_rate": 0.00046322222222222224, + "loss": 0.4336, + "step": 833 + }, + { + "epoch": 16.0, + "learning_rate": 0.0004631111111111111, + "loss": 0.4487, + "step": 834 + }, + { + "epoch": 16.0, + "learning_rate": 0.00046300000000000003, + "loss": 0.3349, + "step": 835 + }, + { + "epoch": 16.0, + "learning_rate": 0.00046288888888888893, + "loss": 0.3465, + "step": 836 + }, + { + "epoch": 16.0, + "learning_rate": 0.0004627777777777778, + "loss": 0.3542, + "step": 837 + }, + { + "epoch": 16.0, + "learning_rate": 0.0004626666666666667, + "loss": 0.4248, + "step": 838 + }, + { + "epoch": 16.0, + "learning_rate": 0.00046255555555555557, + "loss": 0.3619, + "step": 839 + }, + { + "epoch": 16.0, + "learning_rate": 0.00046244444444444447, + "loss": 0.3525, + "step": 840 + }, + { + "epoch": 16.0, + "learning_rate": 0.0004623333333333333, + "loss": 0.3679, + "step": 841 + }, + { + "epoch": 16.01, + "learning_rate": 0.0004622222222222222, + "loss": 0.3674, + "step": 842 + }, + { + "epoch": 16.01, + "learning_rate": 0.0004621111111111111, + "loss": 0.3684, + "step": 843 + }, + { + "epoch": 16.01, + "learning_rate": 0.000462, + "loss": 0.4082, + "step": 844 + }, + { + "epoch": 16.01, + "learning_rate": 0.0004618888888888889, + "loss": 0.4207, + "step": 845 + }, + { + "epoch": 16.01, + "learning_rate": 0.0004617777777777778, + "loss": 0.4122, + "step": 846 + }, + { + "epoch": 16.01, + "learning_rate": 0.0004616666666666667, + "loss": 0.4088, + "step": 847 + }, + { + "epoch": 16.01, + "learning_rate": 0.00046155555555555555, + "loss": 0.3583, + "step": 848 + }, + { + "epoch": 16.01, + "learning_rate": 0.00046144444444444445, + "loss": 0.4077, + "step": 849 + }, + { + "epoch": 16.01, + "learning_rate": 0.00046133333333333334, + "loss": 0.4847, + "step": 850 + }, + { + "epoch": 16.01, + "learning_rate": 0.0004612222222222222, + "loss": 0.42, + "step": 851 + }, + { + "epoch": 16.01, + "learning_rate": 0.00046111111111111114, + "loss": 0.3991, + "step": 852 + }, + { + "epoch": 16.01, + "learning_rate": 0.00046100000000000004, + "loss": 0.4226, + "step": 853 + }, + { + "epoch": 16.01, + "learning_rate": 0.00046088888888888894, + "loss": 0.3714, + "step": 854 + }, + { + "epoch": 16.01, + "learning_rate": 0.0004607777777777778, + "loss": 0.4245, + "step": 855 + }, + { + "epoch": 16.01, + "learning_rate": 0.0004606666666666667, + "loss": 0.3678, + "step": 856 + }, + { + "epoch": 16.01, + "learning_rate": 0.0004605555555555556, + "loss": 0.4636, + "step": 857 + }, + { + "epoch": 16.01, + "learning_rate": 0.0004604444444444444, + "loss": 0.3877, + "step": 858 + }, + { + "epoch": 16.01, + "learning_rate": 0.0004603333333333333, + "loss": 0.4074, + "step": 859 + }, + { + "epoch": 16.01, + "learning_rate": 0.0004602222222222222, + "loss": 0.3992, + "step": 860 + }, + { + "epoch": 16.01, + "learning_rate": 0.00046011111111111117, + "loss": 0.3991, + "step": 861 + }, + { + "epoch": 16.01, + "learning_rate": 0.00046, + "loss": 0.4669, + "step": 862 + }, + { + "epoch": 16.01, + "learning_rate": 0.0004598888888888889, + "loss": 0.4139, + "step": 863 + }, + { + "epoch": 16.01, + "learning_rate": 0.0004597777777777778, + "loss": 0.3974, + "step": 864 + }, + { + "epoch": 16.01, + "learning_rate": 0.00045966666666666665, + "loss": 0.4313, + "step": 865 + }, + { + "epoch": 16.01, + "learning_rate": 0.00045955555555555555, + "loss": 0.358, + "step": 866 + }, + { + "epoch": 16.01, + "learning_rate": 0.00045944444444444445, + "loss": 0.4176, + "step": 867 + }, + { + "epoch": 17.0, + "learning_rate": 0.00045933333333333335, + "loss": 0.629, + "step": 868 + }, + { + "epoch": 17.0, + "learning_rate": 0.0004592222222222222, + "loss": 0.36, + "step": 869 + }, + { + "epoch": 17.0, + "learning_rate": 0.00045911111111111115, + "loss": 0.3707, + "step": 870 + }, + { + "epoch": 17.0, + "learning_rate": 0.00045900000000000004, + "loss": 0.4055, + "step": 871 + }, + { + "epoch": 17.0, + "learning_rate": 0.0004588888888888889, + "loss": 0.3701, + "step": 872 + }, + { + "epoch": 17.0, + "learning_rate": 0.0004587777777777778, + "loss": 0.4102, + "step": 873 + }, + { + "epoch": 17.0, + "learning_rate": 0.0004586666666666667, + "loss": 0.3549, + "step": 874 + }, + { + "epoch": 17.0, + "learning_rate": 0.0004585555555555556, + "loss": 0.4124, + "step": 875 + }, + { + "epoch": 17.0, + "learning_rate": 0.0004584444444444444, + "loss": 0.3941, + "step": 876 + }, + { + "epoch": 17.0, + "learning_rate": 0.0004583333333333333, + "loss": 0.4097, + "step": 877 + }, + { + "epoch": 17.0, + "learning_rate": 0.0004582222222222222, + "loss": 0.3804, + "step": 878 + }, + { + "epoch": 17.0, + "learning_rate": 0.0004581111111111111, + "loss": 0.4229, + "step": 879 + }, + { + "epoch": 17.0, + "learning_rate": 0.000458, + "loss": 0.3646, + "step": 880 + }, + { + "epoch": 17.0, + "learning_rate": 0.0004578888888888889, + "loss": 0.3988, + "step": 881 + }, + { + "epoch": 17.0, + "learning_rate": 0.0004577777777777778, + "loss": 0.4306, + "step": 882 + }, + { + "epoch": 17.0, + "learning_rate": 0.00045766666666666666, + "loss": 0.4229, + "step": 883 + }, + { + "epoch": 17.0, + "learning_rate": 0.00045755555555555556, + "loss": 0.3477, + "step": 884 + }, + { + "epoch": 17.0, + "learning_rate": 0.00045744444444444446, + "loss": 0.4653, + "step": 885 + }, + { + "epoch": 17.0, + "learning_rate": 0.0004573333333333333, + "loss": 0.3931, + "step": 886 + }, + { + "epoch": 17.0, + "learning_rate": 0.0004572222222222222, + "loss": 0.3903, + "step": 887 + }, + { + "epoch": 17.0, + "learning_rate": 0.00045711111111111115, + "loss": 0.3892, + "step": 888 + }, + { + "epoch": 17.0, + "learning_rate": 0.00045700000000000005, + "loss": 0.3988, + "step": 889 + }, + { + "epoch": 17.0, + "learning_rate": 0.0004568888888888889, + "loss": 0.4121, + "step": 890 + }, + { + "epoch": 17.0, + "learning_rate": 0.0004567777777777778, + "loss": 0.4281, + "step": 891 + }, + { + "epoch": 17.0, + "learning_rate": 0.0004566666666666667, + "loss": 0.401, + "step": 892 + }, + { + "epoch": 17.01, + "learning_rate": 0.00045655555555555553, + "loss": 0.3818, + "step": 893 + }, + { + "epoch": 17.01, + "learning_rate": 0.00045644444444444443, + "loss": 0.4, + "step": 894 + }, + { + "epoch": 17.01, + "learning_rate": 0.00045633333333333333, + "loss": 0.4181, + "step": 895 + }, + { + "epoch": 17.01, + "learning_rate": 0.00045622222222222223, + "loss": 0.4034, + "step": 896 + }, + { + "epoch": 17.01, + "learning_rate": 0.0004561111111111111, + "loss": 0.3664, + "step": 897 + }, + { + "epoch": 17.01, + "learning_rate": 0.000456, + "loss": 0.4268, + "step": 898 + }, + { + "epoch": 17.01, + "learning_rate": 0.0004558888888888889, + "loss": 0.4194, + "step": 899 + }, + { + "epoch": 17.01, + "learning_rate": 0.00045577777777777777, + "loss": 0.369, + "step": 900 + }, + { + "epoch": 17.01, + "learning_rate": 0.00045566666666666667, + "loss": 0.4, + "step": 901 + }, + { + "epoch": 17.01, + "learning_rate": 0.00045555555555555556, + "loss": 0.445, + "step": 902 + }, + { + "epoch": 17.01, + "learning_rate": 0.00045544444444444446, + "loss": 0.3936, + "step": 903 + }, + { + "epoch": 17.01, + "learning_rate": 0.0004553333333333333, + "loss": 0.4184, + "step": 904 + }, + { + "epoch": 17.01, + "learning_rate": 0.0004552222222222222, + "loss": 0.4241, + "step": 905 + }, + { + "epoch": 17.01, + "learning_rate": 0.00045511111111111116, + "loss": 0.489, + "step": 906 + }, + { + "epoch": 17.01, + "learning_rate": 0.000455, + "loss": 0.365, + "step": 907 + }, + { + "epoch": 17.01, + "learning_rate": 0.0004548888888888889, + "loss": 0.346, + "step": 908 + }, + { + "epoch": 17.01, + "learning_rate": 0.0004547777777777778, + "loss": 0.3636, + "step": 909 + }, + { + "epoch": 17.01, + "learning_rate": 0.0004546666666666667, + "loss": 0.4008, + "step": 910 + }, + { + "epoch": 17.01, + "learning_rate": 0.00045455555555555554, + "loss": 0.3425, + "step": 911 + }, + { + "epoch": 17.01, + "learning_rate": 0.00045444444444444444, + "loss": 0.4242, + "step": 912 + }, + { + "epoch": 17.01, + "learning_rate": 0.00045433333333333334, + "loss": 0.4168, + "step": 913 + }, + { + "epoch": 17.01, + "learning_rate": 0.00045422222222222223, + "loss": 0.3946, + "step": 914 + }, + { + "epoch": 17.01, + "learning_rate": 0.00045411111111111113, + "loss": 0.3502, + "step": 915 + }, + { + "epoch": 17.01, + "learning_rate": 0.00045400000000000003, + "loss": 0.4116, + "step": 916 + }, + { + "epoch": 17.01, + "learning_rate": 0.00045388888888888893, + "loss": 0.3625, + "step": 917 + }, + { + "epoch": 17.01, + "learning_rate": 0.0004537777777777778, + "loss": 0.3877, + "step": 918 + }, + { + "epoch": 18.0, + "learning_rate": 0.00045366666666666667, + "loss": 0.5991, + "step": 919 + }, + { + "epoch": 18.0, + "learning_rate": 0.00045355555555555557, + "loss": 0.3676, + "step": 920 + }, + { + "epoch": 18.0, + "learning_rate": 0.0004534444444444444, + "loss": 0.3616, + "step": 921 + }, + { + "epoch": 18.0, + "learning_rate": 0.0004533333333333333, + "loss": 0.3781, + "step": 922 + }, + { + "epoch": 18.0, + "learning_rate": 0.00045322222222222226, + "loss": 0.4162, + "step": 923 + }, + { + "epoch": 18.0, + "learning_rate": 0.00045311111111111116, + "loss": 0.4122, + "step": 924 + }, + { + "epoch": 18.0, + "learning_rate": 0.000453, + "loss": 0.3049, + "step": 925 + }, + { + "epoch": 18.0, + "learning_rate": 0.0004528888888888889, + "loss": 0.3764, + "step": 926 + }, + { + "epoch": 18.0, + "learning_rate": 0.0004527777777777778, + "loss": 0.3625, + "step": 927 + }, + { + "epoch": 18.0, + "learning_rate": 0.00045266666666666665, + "loss": 0.3809, + "step": 928 + }, + { + "epoch": 18.0, + "learning_rate": 0.00045255555555555555, + "loss": 0.3671, + "step": 929 + }, + { + "epoch": 18.0, + "learning_rate": 0.00045244444444444444, + "loss": 0.497, + "step": 930 + }, + { + "epoch": 18.0, + "learning_rate": 0.00045233333333333334, + "loss": 0.4079, + "step": 931 + }, + { + "epoch": 18.0, + "learning_rate": 0.00045222222222222224, + "loss": 0.3961, + "step": 932 + }, + { + "epoch": 18.0, + "learning_rate": 0.00045211111111111114, + "loss": 0.3851, + "step": 933 + }, + { + "epoch": 18.0, + "learning_rate": 0.00045200000000000004, + "loss": 0.3669, + "step": 934 + }, + { + "epoch": 18.0, + "learning_rate": 0.0004518888888888889, + "loss": 0.4124, + "step": 935 + }, + { + "epoch": 18.0, + "learning_rate": 0.0004517777777777778, + "loss": 0.3505, + "step": 936 + }, + { + "epoch": 18.0, + "learning_rate": 0.0004516666666666667, + "loss": 0.3555, + "step": 937 + }, + { + "epoch": 18.0, + "learning_rate": 0.0004515555555555556, + "loss": 0.3953, + "step": 938 + }, + { + "epoch": 18.0, + "learning_rate": 0.0004514444444444444, + "loss": 0.3475, + "step": 939 + }, + { + "epoch": 18.0, + "learning_rate": 0.0004513333333333333, + "loss": 0.3258, + "step": 940 + }, + { + "epoch": 18.0, + "learning_rate": 0.00045122222222222227, + "loss": 0.3878, + "step": 941 + }, + { + "epoch": 18.0, + "learning_rate": 0.0004511111111111111, + "loss": 0.3946, + "step": 942 + }, + { + "epoch": 18.0, + "learning_rate": 0.000451, + "loss": 0.388, + "step": 943 + }, + { + "epoch": 18.01, + "learning_rate": 0.0004508888888888889, + "loss": 0.359, + "step": 944 + }, + { + "epoch": 18.01, + "learning_rate": 0.0004507777777777778, + "loss": 0.4321, + "step": 945 + }, + { + "epoch": 18.01, + "learning_rate": 0.00045066666666666665, + "loss": 0.3748, + "step": 946 + }, + { + "epoch": 18.01, + "learning_rate": 0.00045055555555555555, + "loss": 0.3645, + "step": 947 + }, + { + "epoch": 18.01, + "learning_rate": 0.00045044444444444445, + "loss": 0.3935, + "step": 948 + }, + { + "epoch": 18.01, + "learning_rate": 0.0004503333333333333, + "loss": 0.3539, + "step": 949 + }, + { + "epoch": 18.01, + "learning_rate": 0.00045022222222222225, + "loss": 0.3577, + "step": 950 + }, + { + "epoch": 18.01, + "learning_rate": 0.00045011111111111114, + "loss": 0.4002, + "step": 951 + }, + { + "epoch": 18.01, + "learning_rate": 0.00045000000000000004, + "loss": 0.4279, + "step": 952 + }, + { + "epoch": 18.01, + "learning_rate": 0.0004498888888888889, + "loss": 0.2846, + "step": 953 + }, + { + "epoch": 18.01, + "learning_rate": 0.0004497777777777778, + "loss": 0.423, + "step": 954 + }, + { + "epoch": 18.01, + "learning_rate": 0.0004496666666666667, + "loss": 0.3819, + "step": 955 + }, + { + "epoch": 18.01, + "learning_rate": 0.0004495555555555555, + "loss": 0.4093, + "step": 956 + }, + { + "epoch": 18.01, + "learning_rate": 0.0004494444444444444, + "loss": 0.3513, + "step": 957 + }, + { + "epoch": 18.01, + "learning_rate": 0.0004493333333333333, + "loss": 0.3757, + "step": 958 + }, + { + "epoch": 18.01, + "learning_rate": 0.0004492222222222223, + "loss": 0.3483, + "step": 959 + }, + { + "epoch": 18.01, + "learning_rate": 0.0004491111111111111, + "loss": 0.3496, + "step": 960 + }, + { + "epoch": 18.01, + "learning_rate": 0.000449, + "loss": 0.3448, + "step": 961 + }, + { + "epoch": 18.01, + "learning_rate": 0.0004488888888888889, + "loss": 0.4007, + "step": 962 + }, + { + "epoch": 18.01, + "learning_rate": 0.00044877777777777776, + "loss": 0.3756, + "step": 963 + }, + { + "epoch": 18.01, + "learning_rate": 0.00044866666666666666, + "loss": 0.3636, + "step": 964 + }, + { + "epoch": 18.01, + "learning_rate": 0.00044855555555555556, + "loss": 0.3591, + "step": 965 + }, + { + "epoch": 18.01, + "learning_rate": 0.00044844444444444445, + "loss": 0.3705, + "step": 966 + }, + { + "epoch": 18.01, + "learning_rate": 0.0004483333333333333, + "loss": 0.4406, + "step": 967 + }, + { + "epoch": 18.01, + "learning_rate": 0.00044822222222222225, + "loss": 0.3806, + "step": 968 + }, + { + "epoch": 18.01, + "learning_rate": 0.00044811111111111115, + "loss": 0.383, + "step": 969 + }, + { + "epoch": 19.0, + "learning_rate": 0.000448, + "loss": 0.5715, + "step": 970 + }, + { + "epoch": 19.0, + "learning_rate": 0.0004478888888888889, + "loss": 0.338, + "step": 971 + }, + { + "epoch": 19.0, + "learning_rate": 0.0004477777777777778, + "loss": 0.372, + "step": 972 + }, + { + "epoch": 19.0, + "learning_rate": 0.0004476666666666667, + "loss": 0.4413, + "step": 973 + }, + { + "epoch": 19.0, + "learning_rate": 0.00044755555555555553, + "loss": 0.3865, + "step": 974 + }, + { + "epoch": 19.0, + "learning_rate": 0.00044744444444444443, + "loss": 0.3816, + "step": 975 + }, + { + "epoch": 19.0, + "learning_rate": 0.0004473333333333334, + "loss": 0.3452, + "step": 976 + }, + { + "epoch": 19.0, + "learning_rate": 0.0004472222222222222, + "loss": 0.3755, + "step": 977 + }, + { + "epoch": 19.0, + "learning_rate": 0.0004471111111111111, + "loss": 0.3986, + "step": 978 + }, + { + "epoch": 19.0, + "learning_rate": 0.000447, + "loss": 0.4021, + "step": 979 + }, + { + "epoch": 19.0, + "learning_rate": 0.0004468888888888889, + "loss": 0.3557, + "step": 980 + }, + { + "epoch": 19.0, + "learning_rate": 0.00044677777777777777, + "loss": 0.3256, + "step": 981 + }, + { + "epoch": 19.0, + "learning_rate": 0.00044666666666666666, + "loss": 0.3681, + "step": 982 + }, + { + "epoch": 19.0, + "learning_rate": 0.00044655555555555556, + "loss": 0.3822, + "step": 983 + }, + { + "epoch": 19.0, + "learning_rate": 0.0004464444444444444, + "loss": 0.4128, + "step": 984 + }, + { + "epoch": 19.0, + "learning_rate": 0.00044633333333333336, + "loss": 0.4464, + "step": 985 + }, + { + "epoch": 19.0, + "learning_rate": 0.00044622222222222226, + "loss": 0.3516, + "step": 986 + }, + { + "epoch": 19.0, + "learning_rate": 0.00044611111111111115, + "loss": 0.3417, + "step": 987 + }, + { + "epoch": 19.0, + "learning_rate": 0.000446, + "loss": 0.396, + "step": 988 + }, + { + "epoch": 19.0, + "learning_rate": 0.0004458888888888889, + "loss": 0.3416, + "step": 989 + }, + { + "epoch": 19.0, + "learning_rate": 0.0004457777777777778, + "loss": 0.3419, + "step": 990 + }, + { + "epoch": 19.0, + "learning_rate": 0.00044566666666666664, + "loss": 0.359, + "step": 991 + }, + { + "epoch": 19.0, + "learning_rate": 0.00044555555555555554, + "loss": 0.3933, + "step": 992 + }, + { + "epoch": 19.0, + "learning_rate": 0.00044544444444444444, + "loss": 0.4562, + "step": 993 + }, + { + "epoch": 19.0, + "learning_rate": 0.0004453333333333334, + "loss": 0.3864, + "step": 994 + }, + { + "epoch": 19.01, + "learning_rate": 0.00044522222222222223, + "loss": 0.3568, + "step": 995 + }, + { + "epoch": 19.01, + "learning_rate": 0.00044511111111111113, + "loss": 0.2815, + "step": 996 + }, + { + "epoch": 19.01, + "learning_rate": 0.00044500000000000003, + "loss": 0.3929, + "step": 997 + }, + { + "epoch": 19.01, + "learning_rate": 0.00044488888888888887, + "loss": 0.3477, + "step": 998 + }, + { + "epoch": 19.01, + "learning_rate": 0.00044477777777777777, + "loss": 0.3635, + "step": 999 + }, + { + "epoch": 19.01, + "learning_rate": 0.00044466666666666667, + "loss": 0.3801, + "step": 1000 + }, + { + "epoch": 19.01, + "eval_cer": 0.105518314579842, + "eval_loss": 0.3303411602973938, + "eval_runtime": 47.8253, + "eval_samples_per_second": 8.698, + "eval_steps_per_second": 0.272, + "eval_wer": 0.4403107113654947, + "step": 1000 + }, + { + "epoch": 19.01, + "learning_rate": 0.00044455555555555557, + "loss": 0.3283, + "step": 1001 + }, + { + "epoch": 19.01, + "learning_rate": 0.0004444444444444444, + "loss": 0.3547, + "step": 1002 + }, + { + "epoch": 19.01, + "learning_rate": 0.00044433333333333336, + "loss": 0.3921, + "step": 1003 + }, + { + "epoch": 19.01, + "learning_rate": 0.00044422222222222226, + "loss": 0.3663, + "step": 1004 + }, + { + "epoch": 19.01, + "learning_rate": 0.0004441111111111111, + "loss": 0.3922, + "step": 1005 + }, + { + "epoch": 19.01, + "learning_rate": 0.000444, + "loss": 0.3732, + "step": 1006 + }, + { + "epoch": 19.01, + "learning_rate": 0.0004438888888888889, + "loss": 0.4092, + "step": 1007 + }, + { + "epoch": 19.01, + "learning_rate": 0.0004437777777777778, + "loss": 0.3555, + "step": 1008 + }, + { + "epoch": 19.01, + "learning_rate": 0.00044366666666666664, + "loss": 0.3691, + "step": 1009 + }, + { + "epoch": 19.01, + "learning_rate": 0.00044355555555555554, + "loss": 0.3574, + "step": 1010 + }, + { + "epoch": 19.01, + "learning_rate": 0.00044344444444444444, + "loss": 0.3213, + "step": 1011 + }, + { + "epoch": 19.01, + "learning_rate": 0.00044333333333333334, + "loss": 0.2858, + "step": 1012 + }, + { + "epoch": 19.01, + "learning_rate": 0.00044322222222222224, + "loss": 0.3286, + "step": 1013 + }, + { + "epoch": 19.01, + "learning_rate": 0.00044311111111111114, + "loss": 0.4018, + "step": 1014 + }, + { + "epoch": 19.01, + "learning_rate": 0.00044300000000000003, + "loss": 0.3712, + "step": 1015 + }, + { + "epoch": 19.01, + "learning_rate": 0.0004428888888888889, + "loss": 0.3921, + "step": 1016 + }, + { + "epoch": 19.01, + "learning_rate": 0.0004427777777777778, + "loss": 0.3404, + "step": 1017 + }, + { + "epoch": 19.01, + "learning_rate": 0.0004426666666666667, + "loss": 0.3175, + "step": 1018 + }, + { + "epoch": 19.01, + "learning_rate": 0.0004425555555555555, + "loss": 0.3105, + "step": 1019 + }, + { + "epoch": 19.01, + "learning_rate": 0.0004424444444444444, + "loss": 0.3392, + "step": 1020 + }, + { + "epoch": 20.0, + "learning_rate": 0.00044233333333333337, + "loss": 0.5401, + "step": 1021 + }, + { + "epoch": 20.0, + "learning_rate": 0.00044222222222222227, + "loss": 0.3543, + "step": 1022 + }, + { + "epoch": 20.0, + "learning_rate": 0.0004421111111111111, + "loss": 0.3293, + "step": 1023 + }, + { + "epoch": 20.0, + "learning_rate": 0.000442, + "loss": 0.3105, + "step": 1024 + }, + { + "epoch": 20.0, + "learning_rate": 0.0004418888888888889, + "loss": 0.3435, + "step": 1025 + }, + { + "epoch": 20.0, + "learning_rate": 0.00044177777777777775, + "loss": 0.29, + "step": 1026 + }, + { + "epoch": 20.0, + "learning_rate": 0.00044166666666666665, + "loss": 0.3427, + "step": 1027 + }, + { + "epoch": 20.0, + "learning_rate": 0.00044155555555555555, + "loss": 0.3279, + "step": 1028 + }, + { + "epoch": 20.0, + "learning_rate": 0.00044144444444444445, + "loss": 0.3672, + "step": 1029 + }, + { + "epoch": 20.0, + "learning_rate": 0.00044133333333333335, + "loss": 0.3999, + "step": 1030 + }, + { + "epoch": 20.0, + "learning_rate": 0.00044122222222222224, + "loss": 0.2961, + "step": 1031 + }, + { + "epoch": 20.0, + "learning_rate": 0.00044111111111111114, + "loss": 0.3739, + "step": 1032 + }, + { + "epoch": 20.0, + "learning_rate": 0.000441, + "loss": 0.3728, + "step": 1033 + }, + { + "epoch": 20.0, + "learning_rate": 0.0004408888888888889, + "loss": 0.3866, + "step": 1034 + }, + { + "epoch": 20.0, + "learning_rate": 0.0004407777777777778, + "loss": 0.372, + "step": 1035 + }, + { + "epoch": 20.0, + "learning_rate": 0.0004406666666666667, + "loss": 0.2854, + "step": 1036 + }, + { + "epoch": 20.0, + "learning_rate": 0.0004405555555555555, + "loss": 0.2334, + "step": 1037 + }, + { + "epoch": 20.0, + "learning_rate": 0.0004404444444444445, + "loss": 0.4069, + "step": 1038 + }, + { + "epoch": 20.0, + "learning_rate": 0.0004403333333333334, + "loss": 0.3783, + "step": 1039 + }, + { + "epoch": 20.0, + "learning_rate": 0.0004402222222222222, + "loss": 0.3534, + "step": 1040 + }, + { + "epoch": 20.0, + "learning_rate": 0.0004401111111111111, + "loss": 0.3333, + "step": 1041 + }, + { + "epoch": 20.0, + "learning_rate": 0.00044, + "loss": 0.3367, + "step": 1042 + }, + { + "epoch": 20.0, + "learning_rate": 0.0004398888888888889, + "loss": 0.4625, + "step": 1043 + }, + { + "epoch": 20.0, + "learning_rate": 0.00043977777777777776, + "loss": 0.4111, + "step": 1044 + }, + { + "epoch": 20.0, + "learning_rate": 0.00043966666666666666, + "loss": 0.397, + "step": 1045 + }, + { + "epoch": 20.01, + "learning_rate": 0.00043955555555555555, + "loss": 0.3579, + "step": 1046 + }, + { + "epoch": 20.01, + "learning_rate": 0.0004394444444444445, + "loss": 0.3687, + "step": 1047 + }, + { + "epoch": 20.01, + "learning_rate": 0.00043933333333333335, + "loss": 0.3256, + "step": 1048 + }, + { + "epoch": 20.01, + "learning_rate": 0.00043922222222222225, + "loss": 0.3161, + "step": 1049 + }, + { + "epoch": 20.01, + "learning_rate": 0.00043911111111111115, + "loss": 0.3676, + "step": 1050 + }, + { + "epoch": 20.01, + "learning_rate": 0.000439, + "loss": 0.3885, + "step": 1051 + }, + { + "epoch": 20.01, + "learning_rate": 0.0004388888888888889, + "loss": 0.3114, + "step": 1052 + }, + { + "epoch": 20.01, + "learning_rate": 0.0004387777777777778, + "loss": 0.3167, + "step": 1053 + }, + { + "epoch": 20.01, + "learning_rate": 0.00043866666666666663, + "loss": 0.3397, + "step": 1054 + }, + { + "epoch": 20.01, + "learning_rate": 0.00043855555555555553, + "loss": 0.3843, + "step": 1055 + }, + { + "epoch": 20.01, + "learning_rate": 0.0004384444444444445, + "loss": 0.3303, + "step": 1056 + }, + { + "epoch": 20.01, + "learning_rate": 0.0004383333333333334, + "loss": 0.3731, + "step": 1057 + }, + { + "epoch": 20.01, + "learning_rate": 0.0004382222222222222, + "loss": 0.3528, + "step": 1058 + }, + { + "epoch": 20.01, + "learning_rate": 0.0004381111111111111, + "loss": 0.3259, + "step": 1059 + }, + { + "epoch": 20.01, + "learning_rate": 0.000438, + "loss": 0.3299, + "step": 1060 + }, + { + "epoch": 20.01, + "learning_rate": 0.00043788888888888887, + "loss": 0.3814, + "step": 1061 + }, + { + "epoch": 20.01, + "learning_rate": 0.00043777777777777776, + "loss": 0.3495, + "step": 1062 + }, + { + "epoch": 20.01, + "learning_rate": 0.00043766666666666666, + "loss": 0.3359, + "step": 1063 + }, + { + "epoch": 20.01, + "learning_rate": 0.00043755555555555556, + "loss": 0.3433, + "step": 1064 + }, + { + "epoch": 20.01, + "learning_rate": 0.00043744444444444446, + "loss": 0.3399, + "step": 1065 + }, + { + "epoch": 20.01, + "learning_rate": 0.00043733333333333336, + "loss": 0.3702, + "step": 1066 + }, + { + "epoch": 20.01, + "learning_rate": 0.00043722222222222225, + "loss": 0.3689, + "step": 1067 + }, + { + "epoch": 20.01, + "learning_rate": 0.00043711111111111115, + "loss": 0.3547, + "step": 1068 + }, + { + "epoch": 20.01, + "learning_rate": 0.000437, + "loss": 0.3525, + "step": 1069 + }, + { + "epoch": 20.01, + "learning_rate": 0.0004368888888888889, + "loss": 0.3254, + "step": 1070 + }, + { + "epoch": 20.01, + "learning_rate": 0.0004367777777777778, + "loss": 0.3539, + "step": 1071 + }, + { + "epoch": 21.0, + "learning_rate": 0.00043666666666666664, + "loss": 0.5189, + "step": 1072 + }, + { + "epoch": 21.0, + "learning_rate": 0.00043655555555555554, + "loss": 0.3132, + "step": 1073 + }, + { + "epoch": 21.0, + "learning_rate": 0.0004364444444444445, + "loss": 0.3048, + "step": 1074 + }, + { + "epoch": 21.0, + "learning_rate": 0.0004363333333333334, + "loss": 0.3152, + "step": 1075 + }, + { + "epoch": 21.0, + "learning_rate": 0.00043622222222222223, + "loss": 0.334, + "step": 1076 + }, + { + "epoch": 21.0, + "learning_rate": 0.00043611111111111113, + "loss": 0.2771, + "step": 1077 + }, + { + "epoch": 21.0, + "learning_rate": 0.000436, + "loss": 0.3129, + "step": 1078 + }, + { + "epoch": 21.0, + "learning_rate": 0.00043588888888888887, + "loss": 0.3499, + "step": 1079 + }, + { + "epoch": 21.0, + "learning_rate": 0.00043577777777777777, + "loss": 0.3598, + "step": 1080 + }, + { + "epoch": 21.0, + "learning_rate": 0.00043566666666666667, + "loss": 0.3132, + "step": 1081 + }, + { + "epoch": 21.0, + "learning_rate": 0.0004355555555555555, + "loss": 0.3295, + "step": 1082 + }, + { + "epoch": 21.0, + "learning_rate": 0.00043544444444444446, + "loss": 0.3301, + "step": 1083 + }, + { + "epoch": 21.0, + "learning_rate": 0.00043533333333333336, + "loss": 0.3497, + "step": 1084 + }, + { + "epoch": 21.0, + "learning_rate": 0.00043522222222222226, + "loss": 0.3957, + "step": 1085 + }, + { + "epoch": 21.0, + "learning_rate": 0.0004351111111111111, + "loss": 0.3562, + "step": 1086 + }, + { + "epoch": 21.0, + "learning_rate": 0.000435, + "loss": 0.3721, + "step": 1087 + }, + { + "epoch": 21.0, + "learning_rate": 0.0004348888888888889, + "loss": 0.3548, + "step": 1088 + }, + { + "epoch": 21.0, + "learning_rate": 0.0004347777777777778, + "loss": 0.3079, + "step": 1089 + }, + { + "epoch": 21.0, + "learning_rate": 0.00043466666666666664, + "loss": 0.3387, + "step": 1090 + }, + { + "epoch": 21.0, + "learning_rate": 0.00043455555555555554, + "loss": 0.334, + "step": 1091 + }, + { + "epoch": 21.0, + "learning_rate": 0.0004344444444444445, + "loss": 0.3069, + "step": 1092 + }, + { + "epoch": 21.0, + "learning_rate": 0.00043433333333333334, + "loss": 0.3257, + "step": 1093 + }, + { + "epoch": 21.0, + "learning_rate": 0.00043422222222222224, + "loss": 0.3083, + "step": 1094 + }, + { + "epoch": 21.0, + "learning_rate": 0.00043411111111111113, + "loss": 0.3157, + "step": 1095 + }, + { + "epoch": 21.0, + "learning_rate": 0.00043400000000000003, + "loss": 0.3467, + "step": 1096 + }, + { + "epoch": 21.01, + "learning_rate": 0.0004338888888888889, + "loss": 0.3464, + "step": 1097 + }, + { + "epoch": 21.01, + "learning_rate": 0.0004337777777777778, + "loss": 0.3186, + "step": 1098 + }, + { + "epoch": 21.01, + "learning_rate": 0.0004336666666666667, + "loss": 0.3125, + "step": 1099 + }, + { + "epoch": 21.01, + "learning_rate": 0.00043355555555555557, + "loss": 0.3277, + "step": 1100 + }, + { + "epoch": 21.01, + "learning_rate": 0.00043344444444444447, + "loss": 0.3567, + "step": 1101 + }, + { + "epoch": 21.01, + "learning_rate": 0.00043333333333333337, + "loss": 0.3927, + "step": 1102 + }, + { + "epoch": 21.01, + "learning_rate": 0.00043322222222222227, + "loss": 0.3413, + "step": 1103 + }, + { + "epoch": 21.01, + "learning_rate": 0.0004331111111111111, + "loss": 0.3558, + "step": 1104 + }, + { + "epoch": 21.01, + "learning_rate": 0.000433, + "loss": 0.3704, + "step": 1105 + }, + { + "epoch": 21.01, + "learning_rate": 0.0004328888888888889, + "loss": 0.343, + "step": 1106 + }, + { + "epoch": 21.01, + "learning_rate": 0.00043277777777777775, + "loss": 0.3459, + "step": 1107 + }, + { + "epoch": 21.01, + "learning_rate": 0.00043266666666666665, + "loss": 0.3606, + "step": 1108 + }, + { + "epoch": 21.01, + "learning_rate": 0.0004325555555555556, + "loss": 0.2966, + "step": 1109 + }, + { + "epoch": 21.01, + "learning_rate": 0.0004324444444444445, + "loss": 0.3693, + "step": 1110 + }, + { + "epoch": 21.01, + "learning_rate": 0.00043233333333333334, + "loss": 0.3304, + "step": 1111 + }, + { + "epoch": 21.01, + "learning_rate": 0.00043222222222222224, + "loss": 0.337, + "step": 1112 + }, + { + "epoch": 21.01, + "learning_rate": 0.00043211111111111114, + "loss": 0.3435, + "step": 1113 + }, + { + "epoch": 21.01, + "learning_rate": 0.000432, + "loss": 0.3386, + "step": 1114 + }, + { + "epoch": 21.01, + "learning_rate": 0.0004318888888888889, + "loss": 0.3767, + "step": 1115 + }, + { + "epoch": 21.01, + "learning_rate": 0.0004317777777777778, + "loss": 0.2989, + "step": 1116 + }, + { + "epoch": 21.01, + "learning_rate": 0.0004316666666666667, + "loss": 0.338, + "step": 1117 + }, + { + "epoch": 21.01, + "learning_rate": 0.0004315555555555556, + "loss": 0.3685, + "step": 1118 + }, + { + "epoch": 21.01, + "learning_rate": 0.0004314444444444445, + "loss": 0.322, + "step": 1119 + }, + { + "epoch": 21.01, + "learning_rate": 0.0004313333333333334, + "loss": 0.3229, + "step": 1120 + }, + { + "epoch": 21.01, + "learning_rate": 0.0004312222222222222, + "loss": 0.3318, + "step": 1121 + }, + { + "epoch": 21.01, + "learning_rate": 0.0004311111111111111, + "loss": 0.3365, + "step": 1122 + }, + { + "epoch": 22.0, + "learning_rate": 0.000431, + "loss": 0.5018, + "step": 1123 + }, + { + "epoch": 22.0, + "learning_rate": 0.0004308888888888889, + "loss": 0.375, + "step": 1124 + }, + { + "epoch": 22.0, + "learning_rate": 0.00043077777777777776, + "loss": 0.3348, + "step": 1125 + }, + { + "epoch": 22.0, + "learning_rate": 0.00043066666666666665, + "loss": 0.2645, + "step": 1126 + }, + { + "epoch": 22.0, + "learning_rate": 0.0004305555555555556, + "loss": 0.2949, + "step": 1127 + }, + { + "epoch": 22.0, + "learning_rate": 0.00043044444444444445, + "loss": 0.3147, + "step": 1128 + }, + { + "epoch": 22.0, + "learning_rate": 0.00043033333333333335, + "loss": 0.3123, + "step": 1129 + }, + { + "epoch": 22.0, + "learning_rate": 0.00043022222222222225, + "loss": 0.3631, + "step": 1130 + }, + { + "epoch": 22.0, + "learning_rate": 0.00043011111111111115, + "loss": 0.3369, + "step": 1131 + }, + { + "epoch": 22.0, + "learning_rate": 0.00043, + "loss": 0.3543, + "step": 1132 + }, + { + "epoch": 22.0, + "learning_rate": 0.0004298888888888889, + "loss": 0.4244, + "step": 1133 + }, + { + "epoch": 22.0, + "learning_rate": 0.0004297777777777778, + "loss": 0.3306, + "step": 1134 + }, + { + "epoch": 22.0, + "learning_rate": 0.00042966666666666663, + "loss": 0.3385, + "step": 1135 + }, + { + "epoch": 22.0, + "learning_rate": 0.0004295555555555556, + "loss": 0.3558, + "step": 1136 + }, + { + "epoch": 22.0, + "learning_rate": 0.0004294444444444445, + "loss": 0.4108, + "step": 1137 + }, + { + "epoch": 22.0, + "learning_rate": 0.0004293333333333334, + "loss": 0.3578, + "step": 1138 + }, + { + "epoch": 22.0, + "learning_rate": 0.0004292222222222222, + "loss": 0.3154, + "step": 1139 + }, + { + "epoch": 22.0, + "learning_rate": 0.0004291111111111111, + "loss": 0.3338, + "step": 1140 + }, + { + "epoch": 22.0, + "learning_rate": 0.000429, + "loss": 0.3709, + "step": 1141 + }, + { + "epoch": 22.0, + "learning_rate": 0.00042888888888888886, + "loss": 0.3472, + "step": 1142 + }, + { + "epoch": 22.0, + "learning_rate": 0.00042877777777777776, + "loss": 0.3719, + "step": 1143 + }, + { + "epoch": 22.0, + "learning_rate": 0.00042866666666666666, + "loss": 0.4556, + "step": 1144 + }, + { + "epoch": 22.0, + "learning_rate": 0.0004285555555555556, + "loss": 0.3626, + "step": 1145 + }, + { + "epoch": 22.0, + "learning_rate": 0.00042844444444444446, + "loss": 0.3524, + "step": 1146 + }, + { + "epoch": 22.0, + "learning_rate": 0.00042833333333333335, + "loss": 0.3113, + "step": 1147 + }, + { + "epoch": 22.01, + "learning_rate": 0.00042822222222222225, + "loss": 0.2908, + "step": 1148 + }, + { + "epoch": 22.01, + "learning_rate": 0.0004281111111111111, + "loss": 0.3079, + "step": 1149 + }, + { + "epoch": 22.01, + "learning_rate": 0.000428, + "loss": 0.3061, + "step": 1150 + }, + { + "epoch": 22.01, + "learning_rate": 0.0004278888888888889, + "loss": 0.3383, + "step": 1151 + }, + { + "epoch": 22.01, + "learning_rate": 0.0004277777777777778, + "loss": 0.3237, + "step": 1152 + }, + { + "epoch": 22.01, + "learning_rate": 0.00042766666666666664, + "loss": 0.3282, + "step": 1153 + }, + { + "epoch": 22.01, + "learning_rate": 0.0004275555555555556, + "loss": 0.3531, + "step": 1154 + }, + { + "epoch": 22.01, + "learning_rate": 0.0004274444444444445, + "loss": 0.369, + "step": 1155 + }, + { + "epoch": 22.01, + "learning_rate": 0.00042733333333333333, + "loss": 0.3427, + "step": 1156 + }, + { + "epoch": 22.01, + "learning_rate": 0.00042722222222222223, + "loss": 0.3798, + "step": 1157 + }, + { + "epoch": 22.01, + "learning_rate": 0.0004271111111111111, + "loss": 0.3541, + "step": 1158 + }, + { + "epoch": 22.01, + "learning_rate": 0.000427, + "loss": 0.3489, + "step": 1159 + }, + { + "epoch": 22.01, + "learning_rate": 0.00042688888888888887, + "loss": 0.3351, + "step": 1160 + }, + { + "epoch": 22.01, + "learning_rate": 0.00042677777777777777, + "loss": 0.3243, + "step": 1161 + }, + { + "epoch": 22.01, + "learning_rate": 0.0004266666666666667, + "loss": 0.3855, + "step": 1162 + }, + { + "epoch": 22.01, + "learning_rate": 0.00042655555555555556, + "loss": 0.4335, + "step": 1163 + }, + { + "epoch": 22.01, + "learning_rate": 0.00042644444444444446, + "loss": 0.356, + "step": 1164 + }, + { + "epoch": 22.01, + "learning_rate": 0.00042633333333333336, + "loss": 0.3004, + "step": 1165 + }, + { + "epoch": 22.01, + "learning_rate": 0.00042622222222222226, + "loss": 0.405, + "step": 1166 + }, + { + "epoch": 22.01, + "learning_rate": 0.0004261111111111111, + "loss": 0.4094, + "step": 1167 + }, + { + "epoch": 22.01, + "learning_rate": 0.000426, + "loss": 0.3307, + "step": 1168 + }, + { + "epoch": 22.01, + "learning_rate": 0.0004258888888888889, + "loss": 0.3767, + "step": 1169 + }, + { + "epoch": 22.01, + "learning_rate": 0.00042577777777777774, + "loss": 0.309, + "step": 1170 + }, + { + "epoch": 22.01, + "learning_rate": 0.0004256666666666667, + "loss": 0.3473, + "step": 1171 + }, + { + "epoch": 22.01, + "learning_rate": 0.0004255555555555556, + "loss": 0.3165, + "step": 1172 + }, + { + "epoch": 22.01, + "learning_rate": 0.0004254444444444445, + "loss": 0.3547, + "step": 1173 + }, + { + "epoch": 23.0, + "learning_rate": 0.00042533333333333334, + "loss": 0.505, + "step": 1174 + }, + { + "epoch": 23.0, + "learning_rate": 0.00042522222222222223, + "loss": 0.3357, + "step": 1175 + }, + { + "epoch": 23.0, + "learning_rate": 0.00042511111111111113, + "loss": 0.2983, + "step": 1176 + }, + { + "epoch": 23.0, + "learning_rate": 0.000425, + "loss": 0.2934, + "step": 1177 + }, + { + "epoch": 23.0, + "learning_rate": 0.0004248888888888889, + "loss": 0.3392, + "step": 1178 + }, + { + "epoch": 23.0, + "learning_rate": 0.00042477777777777777, + "loss": 0.3059, + "step": 1179 + }, + { + "epoch": 23.0, + "learning_rate": 0.0004246666666666667, + "loss": 0.3373, + "step": 1180 + }, + { + "epoch": 23.0, + "learning_rate": 0.00042455555555555557, + "loss": 0.333, + "step": 1181 + }, + { + "epoch": 23.0, + "learning_rate": 0.00042444444444444447, + "loss": 0.3195, + "step": 1182 + }, + { + "epoch": 23.0, + "learning_rate": 0.00042433333333333337, + "loss": 0.2638, + "step": 1183 + }, + { + "epoch": 23.0, + "learning_rate": 0.0004242222222222222, + "loss": 0.3351, + "step": 1184 + }, + { + "epoch": 23.0, + "learning_rate": 0.0004241111111111111, + "loss": 0.3373, + "step": 1185 + }, + { + "epoch": 23.0, + "learning_rate": 0.000424, + "loss": 0.3611, + "step": 1186 + }, + { + "epoch": 23.0, + "learning_rate": 0.0004238888888888889, + "loss": 0.3338, + "step": 1187 + }, + { + "epoch": 23.0, + "learning_rate": 0.00042377777777777775, + "loss": 0.3538, + "step": 1188 + }, + { + "epoch": 23.0, + "learning_rate": 0.0004236666666666667, + "loss": 0.3771, + "step": 1189 + }, + { + "epoch": 23.0, + "learning_rate": 0.0004235555555555556, + "loss": 0.4264, + "step": 1190 + }, + { + "epoch": 23.0, + "learning_rate": 0.00042344444444444444, + "loss": 0.3723, + "step": 1191 + }, + { + "epoch": 23.0, + "learning_rate": 0.00042333333333333334, + "loss": 0.2974, + "step": 1192 + }, + { + "epoch": 23.0, + "learning_rate": 0.00042322222222222224, + "loss": 0.3669, + "step": 1193 + }, + { + "epoch": 23.0, + "learning_rate": 0.00042311111111111114, + "loss": 0.3651, + "step": 1194 + }, + { + "epoch": 23.0, + "learning_rate": 0.000423, + "loss": 0.4192, + "step": 1195 + }, + { + "epoch": 23.0, + "learning_rate": 0.0004228888888888889, + "loss": 0.3001, + "step": 1196 + }, + { + "epoch": 23.0, + "learning_rate": 0.0004227777777777778, + "loss": 0.3528, + "step": 1197 + }, + { + "epoch": 23.0, + "learning_rate": 0.0004226666666666667, + "loss": 0.3091, + "step": 1198 + }, + { + "epoch": 23.01, + "learning_rate": 0.0004225555555555556, + "loss": 0.3468, + "step": 1199 + }, + { + "epoch": 23.01, + "learning_rate": 0.0004224444444444445, + "loss": 0.3695, + "step": 1200 + }, + { + "epoch": 23.01, + "learning_rate": 0.00042233333333333337, + "loss": 0.2957, + "step": 1201 + }, + { + "epoch": 23.01, + "learning_rate": 0.0004222222222222222, + "loss": 0.4423, + "step": 1202 + }, + { + "epoch": 23.01, + "learning_rate": 0.0004221111111111111, + "loss": 0.3086, + "step": 1203 + }, + { + "epoch": 23.01, + "learning_rate": 0.000422, + "loss": 0.3064, + "step": 1204 + }, + { + "epoch": 23.01, + "learning_rate": 0.00042188888888888886, + "loss": 0.3497, + "step": 1205 + }, + { + "epoch": 23.01, + "learning_rate": 0.00042177777777777775, + "loss": 0.3411, + "step": 1206 + }, + { + "epoch": 23.01, + "learning_rate": 0.0004216666666666667, + "loss": 0.3139, + "step": 1207 + }, + { + "epoch": 23.01, + "learning_rate": 0.0004215555555555556, + "loss": 0.3738, + "step": 1208 + }, + { + "epoch": 23.01, + "learning_rate": 0.00042144444444444445, + "loss": 0.3555, + "step": 1209 + }, + { + "epoch": 23.01, + "learning_rate": 0.00042133333333333335, + "loss": 0.3459, + "step": 1210 + }, + { + "epoch": 23.01, + "learning_rate": 0.00042122222222222225, + "loss": 0.3529, + "step": 1211 + }, + { + "epoch": 23.01, + "learning_rate": 0.0004211111111111111, + "loss": 0.3356, + "step": 1212 + }, + { + "epoch": 23.01, + "learning_rate": 0.000421, + "loss": 0.3589, + "step": 1213 + }, + { + "epoch": 23.01, + "learning_rate": 0.0004208888888888889, + "loss": 0.3443, + "step": 1214 + }, + { + "epoch": 23.01, + "learning_rate": 0.0004207777777777778, + "loss": 0.3658, + "step": 1215 + }, + { + "epoch": 23.01, + "learning_rate": 0.0004206666666666667, + "loss": 0.4464, + "step": 1216 + }, + { + "epoch": 23.01, + "learning_rate": 0.0004205555555555556, + "loss": 0.363, + "step": 1217 + }, + { + "epoch": 23.01, + "learning_rate": 0.0004204444444444445, + "loss": 0.3646, + "step": 1218 + }, + { + "epoch": 23.01, + "learning_rate": 0.0004203333333333333, + "loss": 0.3261, + "step": 1219 + }, + { + "epoch": 23.01, + "learning_rate": 0.0004202222222222222, + "loss": 0.2989, + "step": 1220 + }, + { + "epoch": 23.01, + "learning_rate": 0.0004201111111111111, + "loss": 0.358, + "step": 1221 + }, + { + "epoch": 23.01, + "learning_rate": 0.00042, + "loss": 0.3368, + "step": 1222 + }, + { + "epoch": 23.01, + "learning_rate": 0.00041988888888888886, + "loss": 0.2883, + "step": 1223 + }, + { + "epoch": 23.01, + "learning_rate": 0.0004197777777777778, + "loss": 0.3227, + "step": 1224 + }, + { + "epoch": 24.0, + "learning_rate": 0.0004196666666666667, + "loss": 0.4307, + "step": 1225 + }, + { + "epoch": 24.0, + "learning_rate": 0.00041955555555555556, + "loss": 0.3015, + "step": 1226 + }, + { + "epoch": 24.0, + "learning_rate": 0.00041944444444444445, + "loss": 0.267, + "step": 1227 + }, + { + "epoch": 24.0, + "learning_rate": 0.00041933333333333335, + "loss": 0.3778, + "step": 1228 + }, + { + "epoch": 24.0, + "learning_rate": 0.00041922222222222225, + "loss": 0.3239, + "step": 1229 + }, + { + "epoch": 24.0, + "learning_rate": 0.0004191111111111111, + "loss": 0.3161, + "step": 1230 + }, + { + "epoch": 24.0, + "learning_rate": 0.000419, + "loss": 0.3783, + "step": 1231 + }, + { + "epoch": 24.0, + "learning_rate": 0.0004188888888888889, + "loss": 0.3967, + "step": 1232 + }, + { + "epoch": 24.0, + "learning_rate": 0.0004187777777777778, + "loss": 0.325, + "step": 1233 + }, + { + "epoch": 24.0, + "learning_rate": 0.0004186666666666667, + "loss": 0.333, + "step": 1234 + }, + { + "epoch": 24.0, + "learning_rate": 0.0004185555555555556, + "loss": 0.3444, + "step": 1235 + }, + { + "epoch": 24.0, + "learning_rate": 0.0004184444444444445, + "loss": 0.3204, + "step": 1236 + }, + { + "epoch": 24.0, + "learning_rate": 0.00041833333333333333, + "loss": 0.3611, + "step": 1237 + }, + { + "epoch": 24.0, + "learning_rate": 0.0004182222222222222, + "loss": 0.3315, + "step": 1238 + }, + { + "epoch": 24.0, + "learning_rate": 0.0004181111111111111, + "loss": 0.3399, + "step": 1239 + }, + { + "epoch": 24.0, + "learning_rate": 0.00041799999999999997, + "loss": 0.2728, + "step": 1240 + }, + { + "epoch": 24.0, + "learning_rate": 0.00041788888888888887, + "loss": 0.3199, + "step": 1241 + }, + { + "epoch": 24.0, + "learning_rate": 0.0004177777777777778, + "loss": 0.3139, + "step": 1242 + }, + { + "epoch": 24.0, + "learning_rate": 0.0004176666666666667, + "loss": 0.3589, + "step": 1243 + }, + { + "epoch": 24.0, + "learning_rate": 0.00041755555555555556, + "loss": 0.3486, + "step": 1244 + }, + { + "epoch": 24.0, + "learning_rate": 0.00041744444444444446, + "loss": 0.3364, + "step": 1245 + }, + { + "epoch": 24.0, + "learning_rate": 0.00041733333333333336, + "loss": 0.3471, + "step": 1246 + }, + { + "epoch": 24.0, + "learning_rate": 0.0004172222222222222, + "loss": 0.3147, + "step": 1247 + }, + { + "epoch": 24.0, + "learning_rate": 0.0004171111111111111, + "loss": 0.2806, + "step": 1248 + }, + { + "epoch": 24.0, + "learning_rate": 0.000417, + "loss": 0.2728, + "step": 1249 + }, + { + "epoch": 24.01, + "learning_rate": 0.0004168888888888889, + "loss": 0.3041, + "step": 1250 + }, + { + "epoch": 24.01, + "learning_rate": 0.0004167777777777778, + "loss": 0.2777, + "step": 1251 + }, + { + "epoch": 24.01, + "learning_rate": 0.0004166666666666667, + "loss": 0.3836, + "step": 1252 + }, + { + "epoch": 24.01, + "learning_rate": 0.0004165555555555556, + "loss": 0.2869, + "step": 1253 + }, + { + "epoch": 24.01, + "learning_rate": 0.00041644444444444444, + "loss": 0.3493, + "step": 1254 + }, + { + "epoch": 24.01, + "learning_rate": 0.00041633333333333333, + "loss": 0.3152, + "step": 1255 + }, + { + "epoch": 24.01, + "learning_rate": 0.00041622222222222223, + "loss": 0.3145, + "step": 1256 + }, + { + "epoch": 24.01, + "learning_rate": 0.00041611111111111113, + "loss": 0.2717, + "step": 1257 + }, + { + "epoch": 24.01, + "learning_rate": 0.000416, + "loss": 0.3219, + "step": 1258 + }, + { + "epoch": 24.01, + "learning_rate": 0.00041588888888888887, + "loss": 0.3461, + "step": 1259 + }, + { + "epoch": 24.01, + "learning_rate": 0.0004157777777777778, + "loss": 0.3475, + "step": 1260 + }, + { + "epoch": 24.01, + "learning_rate": 0.00041566666666666667, + "loss": 0.3495, + "step": 1261 + }, + { + "epoch": 24.01, + "learning_rate": 0.00041555555555555557, + "loss": 0.3464, + "step": 1262 + }, + { + "epoch": 24.01, + "learning_rate": 0.00041544444444444447, + "loss": 0.2677, + "step": 1263 + }, + { + "epoch": 24.01, + "learning_rate": 0.00041533333333333336, + "loss": 0.35, + "step": 1264 + }, + { + "epoch": 24.01, + "learning_rate": 0.0004152222222222222, + "loss": 0.3555, + "step": 1265 + }, + { + "epoch": 24.01, + "learning_rate": 0.0004151111111111111, + "loss": 0.3466, + "step": 1266 + }, + { + "epoch": 24.01, + "learning_rate": 0.000415, + "loss": 0.4103, + "step": 1267 + }, + { + "epoch": 24.01, + "learning_rate": 0.00041488888888888885, + "loss": 0.382, + "step": 1268 + }, + { + "epoch": 24.01, + "learning_rate": 0.0004147777777777778, + "loss": 0.3651, + "step": 1269 + }, + { + "epoch": 24.01, + "learning_rate": 0.0004146666666666667, + "loss": 0.2906, + "step": 1270 + }, + { + "epoch": 24.01, + "learning_rate": 0.0004145555555555556, + "loss": 0.3196, + "step": 1271 + }, + { + "epoch": 24.01, + "learning_rate": 0.00041444444444444444, + "loss": 0.3361, + "step": 1272 + }, + { + "epoch": 24.01, + "learning_rate": 0.00041433333333333334, + "loss": 0.3428, + "step": 1273 + }, + { + "epoch": 24.01, + "learning_rate": 0.00041422222222222224, + "loss": 0.3287, + "step": 1274 + }, + { + "epoch": 24.01, + "learning_rate": 0.0004141111111111111, + "loss": 0.2994, + "step": 1275 + }, + { + "epoch": 25.0, + "learning_rate": 0.000414, + "loss": 0.4692, + "step": 1276 + }, + { + "epoch": 25.0, + "learning_rate": 0.0004138888888888889, + "loss": 0.2804, + "step": 1277 + }, + { + "epoch": 25.0, + "learning_rate": 0.00041377777777777783, + "loss": 0.3512, + "step": 1278 + }, + { + "epoch": 25.0, + "learning_rate": 0.0004136666666666667, + "loss": 0.3078, + "step": 1279 + }, + { + "epoch": 25.0, + "learning_rate": 0.0004135555555555556, + "loss": 0.3087, + "step": 1280 + }, + { + "epoch": 25.0, + "learning_rate": 0.00041344444444444447, + "loss": 0.3006, + "step": 1281 + }, + { + "epoch": 25.0, + "learning_rate": 0.0004133333333333333, + "loss": 0.2581, + "step": 1282 + }, + { + "epoch": 25.0, + "learning_rate": 0.0004132222222222222, + "loss": 0.294, + "step": 1283 + }, + { + "epoch": 25.0, + "learning_rate": 0.0004131111111111111, + "loss": 0.2831, + "step": 1284 + }, + { + "epoch": 25.0, + "learning_rate": 0.000413, + "loss": 0.3107, + "step": 1285 + }, + { + "epoch": 25.0, + "learning_rate": 0.0004128888888888889, + "loss": 0.3066, + "step": 1286 + }, + { + "epoch": 25.0, + "learning_rate": 0.0004127777777777778, + "loss": 0.3222, + "step": 1287 + }, + { + "epoch": 25.0, + "learning_rate": 0.0004126666666666667, + "loss": 0.3931, + "step": 1288 + }, + { + "epoch": 25.0, + "learning_rate": 0.00041255555555555555, + "loss": 0.3605, + "step": 1289 + }, + { + "epoch": 25.0, + "learning_rate": 0.00041244444444444445, + "loss": 0.315, + "step": 1290 + }, + { + "epoch": 25.0, + "learning_rate": 0.00041233333333333335, + "loss": 0.3104, + "step": 1291 + }, + { + "epoch": 25.0, + "learning_rate": 0.00041222222222222224, + "loss": 0.4154, + "step": 1292 + }, + { + "epoch": 25.0, + "learning_rate": 0.0004121111111111111, + "loss": 0.3194, + "step": 1293 + }, + { + "epoch": 25.0, + "learning_rate": 0.000412, + "loss": 0.3029, + "step": 1294 + }, + { + "epoch": 25.0, + "learning_rate": 0.00041188888888888894, + "loss": 0.2988, + "step": 1295 + }, + { + "epoch": 25.0, + "learning_rate": 0.0004117777777777778, + "loss": 0.3048, + "step": 1296 + }, + { + "epoch": 25.0, + "learning_rate": 0.0004116666666666667, + "loss": 0.3554, + "step": 1297 + }, + { + "epoch": 25.0, + "learning_rate": 0.0004115555555555556, + "loss": 0.3231, + "step": 1298 + }, + { + "epoch": 25.0, + "learning_rate": 0.0004114444444444445, + "loss": 0.3362, + "step": 1299 + }, + { + "epoch": 25.0, + "learning_rate": 0.0004113333333333333, + "loss": 0.3107, + "step": 1300 + }, + { + "epoch": 25.01, + "learning_rate": 0.0004112222222222222, + "loss": 0.3409, + "step": 1301 + }, + { + "epoch": 25.01, + "learning_rate": 0.0004111111111111111, + "loss": 0.3026, + "step": 1302 + }, + { + "epoch": 25.01, + "learning_rate": 0.00041099999999999996, + "loss": 0.3, + "step": 1303 + }, + { + "epoch": 25.01, + "learning_rate": 0.0004108888888888889, + "loss": 0.2638, + "step": 1304 + }, + { + "epoch": 25.01, + "learning_rate": 0.0004107777777777778, + "loss": 0.2824, + "step": 1305 + }, + { + "epoch": 25.01, + "learning_rate": 0.0004106666666666667, + "loss": 0.2755, + "step": 1306 + }, + { + "epoch": 25.01, + "learning_rate": 0.00041055555555555555, + "loss": 0.3392, + "step": 1307 + }, + { + "epoch": 25.01, + "learning_rate": 0.00041044444444444445, + "loss": 0.2961, + "step": 1308 + }, + { + "epoch": 25.01, + "learning_rate": 0.00041033333333333335, + "loss": 0.3296, + "step": 1309 + }, + { + "epoch": 25.01, + "learning_rate": 0.0004102222222222222, + "loss": 0.3204, + "step": 1310 + }, + { + "epoch": 25.01, + "learning_rate": 0.0004101111111111111, + "loss": 0.3503, + "step": 1311 + }, + { + "epoch": 25.01, + "learning_rate": 0.00041, + "loss": 0.3036, + "step": 1312 + }, + { + "epoch": 25.01, + "learning_rate": 0.00040988888888888894, + "loss": 0.2924, + "step": 1313 + }, + { + "epoch": 25.01, + "learning_rate": 0.0004097777777777778, + "loss": 0.314, + "step": 1314 + }, + { + "epoch": 25.01, + "learning_rate": 0.0004096666666666667, + "loss": 0.334, + "step": 1315 + }, + { + "epoch": 25.01, + "learning_rate": 0.0004095555555555556, + "loss": 0.3072, + "step": 1316 + }, + { + "epoch": 25.01, + "learning_rate": 0.00040944444444444443, + "loss": 0.2924, + "step": 1317 + }, + { + "epoch": 25.01, + "learning_rate": 0.0004093333333333333, + "loss": 0.3342, + "step": 1318 + }, + { + "epoch": 25.01, + "learning_rate": 0.0004092222222222222, + "loss": 0.3278, + "step": 1319 + }, + { + "epoch": 25.01, + "learning_rate": 0.0004091111111111111, + "loss": 0.2716, + "step": 1320 + }, + { + "epoch": 25.01, + "learning_rate": 0.00040899999999999997, + "loss": 0.3363, + "step": 1321 + }, + { + "epoch": 25.01, + "learning_rate": 0.0004088888888888889, + "loss": 0.3413, + "step": 1322 + }, + { + "epoch": 25.01, + "learning_rate": 0.0004087777777777778, + "loss": 0.2512, + "step": 1323 + }, + { + "epoch": 25.01, + "learning_rate": 0.00040866666666666666, + "loss": 0.33, + "step": 1324 + }, + { + "epoch": 25.01, + "learning_rate": 0.00040855555555555556, + "loss": 0.2832, + "step": 1325 + }, + { + "epoch": 25.01, + "learning_rate": 0.00040844444444444446, + "loss": 0.3187, + "step": 1326 + }, + { + "epoch": 26.0, + "learning_rate": 0.00040833333333333336, + "loss": 0.4953, + "step": 1327 + }, + { + "epoch": 26.0, + "learning_rate": 0.0004082222222222222, + "loss": 0.2973, + "step": 1328 + }, + { + "epoch": 26.0, + "learning_rate": 0.0004081111111111111, + "loss": 0.3111, + "step": 1329 + }, + { + "epoch": 26.0, + "learning_rate": 0.000408, + "loss": 0.2901, + "step": 1330 + }, + { + "epoch": 26.0, + "learning_rate": 0.0004078888888888889, + "loss": 0.322, + "step": 1331 + }, + { + "epoch": 26.0, + "learning_rate": 0.0004077777777777778, + "loss": 0.2894, + "step": 1332 + }, + { + "epoch": 26.0, + "learning_rate": 0.0004076666666666667, + "loss": 0.2967, + "step": 1333 + }, + { + "epoch": 26.0, + "learning_rate": 0.0004075555555555556, + "loss": 0.3624, + "step": 1334 + }, + { + "epoch": 26.0, + "learning_rate": 0.00040744444444444443, + "loss": 0.3028, + "step": 1335 + }, + { + "epoch": 26.0, + "learning_rate": 0.00040733333333333333, + "loss": 0.364, + "step": 1336 + }, + { + "epoch": 26.0, + "learning_rate": 0.00040722222222222223, + "loss": 0.3516, + "step": 1337 + }, + { + "epoch": 26.0, + "learning_rate": 0.0004071111111111111, + "loss": 0.3153, + "step": 1338 + }, + { + "epoch": 26.0, + "learning_rate": 0.00040699999999999997, + "loss": 0.4031, + "step": 1339 + }, + { + "epoch": 26.0, + "learning_rate": 0.0004068888888888889, + "loss": 0.34, + "step": 1340 + }, + { + "epoch": 26.0, + "learning_rate": 0.0004067777777777778, + "loss": 0.3122, + "step": 1341 + }, + { + "epoch": 26.0, + "learning_rate": 0.00040666666666666667, + "loss": 0.2679, + "step": 1342 + }, + { + "epoch": 26.0, + "learning_rate": 0.00040655555555555557, + "loss": 0.3192, + "step": 1343 + }, + { + "epoch": 26.0, + "learning_rate": 0.00040644444444444446, + "loss": 0.28, + "step": 1344 + }, + { + "epoch": 26.0, + "learning_rate": 0.0004063333333333333, + "loss": 0.3233, + "step": 1345 + }, + { + "epoch": 26.0, + "learning_rate": 0.0004062222222222222, + "loss": 0.3133, + "step": 1346 + }, + { + "epoch": 26.0, + "learning_rate": 0.0004061111111111111, + "loss": 0.3252, + "step": 1347 + }, + { + "epoch": 26.0, + "learning_rate": 0.00040600000000000006, + "loss": 0.2983, + "step": 1348 + }, + { + "epoch": 26.0, + "learning_rate": 0.0004058888888888889, + "loss": 0.3175, + "step": 1349 + }, + { + "epoch": 26.0, + "learning_rate": 0.0004057777777777778, + "loss": 0.3301, + "step": 1350 + }, + { + "epoch": 26.0, + "learning_rate": 0.0004056666666666667, + "loss": 0.3224, + "step": 1351 + }, + { + "epoch": 26.01, + "learning_rate": 0.00040555555555555554, + "loss": 0.3116, + "step": 1352 + }, + { + "epoch": 26.01, + "learning_rate": 0.00040544444444444444, + "loss": 0.2953, + "step": 1353 + }, + { + "epoch": 26.01, + "learning_rate": 0.00040533333333333334, + "loss": 0.2823, + "step": 1354 + }, + { + "epoch": 26.01, + "learning_rate": 0.00040522222222222224, + "loss": 0.2657, + "step": 1355 + }, + { + "epoch": 26.01, + "learning_rate": 0.0004051111111111111, + "loss": 0.2809, + "step": 1356 + }, + { + "epoch": 26.01, + "learning_rate": 0.00040500000000000003, + "loss": 0.2746, + "step": 1357 + }, + { + "epoch": 26.01, + "learning_rate": 0.00040488888888888893, + "loss": 0.2933, + "step": 1358 + }, + { + "epoch": 26.01, + "learning_rate": 0.0004047777777777778, + "loss": 0.2824, + "step": 1359 + }, + { + "epoch": 26.01, + "learning_rate": 0.00040466666666666667, + "loss": 0.3312, + "step": 1360 + }, + { + "epoch": 26.01, + "learning_rate": 0.00040455555555555557, + "loss": 0.3433, + "step": 1361 + }, + { + "epoch": 26.01, + "learning_rate": 0.00040444444444444447, + "loss": 0.2686, + "step": 1362 + }, + { + "epoch": 26.01, + "learning_rate": 0.0004043333333333333, + "loss": 0.2977, + "step": 1363 + }, + { + "epoch": 26.01, + "learning_rate": 0.0004042222222222222, + "loss": 0.2601, + "step": 1364 + }, + { + "epoch": 26.01, + "learning_rate": 0.0004041111111111111, + "loss": 0.2815, + "step": 1365 + }, + { + "epoch": 26.01, + "learning_rate": 0.000404, + "loss": 0.2867, + "step": 1366 + }, + { + "epoch": 26.01, + "learning_rate": 0.0004038888888888889, + "loss": 0.3091, + "step": 1367 + }, + { + "epoch": 26.01, + "learning_rate": 0.0004037777777777778, + "loss": 0.3034, + "step": 1368 + }, + { + "epoch": 26.01, + "learning_rate": 0.0004036666666666667, + "loss": 0.3146, + "step": 1369 + }, + { + "epoch": 26.01, + "learning_rate": 0.00040355555555555555, + "loss": 0.2704, + "step": 1370 + }, + { + "epoch": 26.01, + "learning_rate": 0.00040344444444444444, + "loss": 0.298, + "step": 1371 + }, + { + "epoch": 26.01, + "learning_rate": 0.00040333333333333334, + "loss": 0.3748, + "step": 1372 + }, + { + "epoch": 26.01, + "learning_rate": 0.0004032222222222222, + "loss": 0.3644, + "step": 1373 + }, + { + "epoch": 26.01, + "learning_rate": 0.0004031111111111111, + "loss": 0.3219, + "step": 1374 + }, + { + "epoch": 26.01, + "learning_rate": 0.00040300000000000004, + "loss": 0.2631, + "step": 1375 + }, + { + "epoch": 26.01, + "learning_rate": 0.00040288888888888894, + "loss": 0.323, + "step": 1376 + }, + { + "epoch": 26.01, + "learning_rate": 0.0004027777777777778, + "loss": 0.3198, + "step": 1377 + }, + { + "epoch": 27.0, + "learning_rate": 0.0004026666666666667, + "loss": 0.4203, + "step": 1378 + }, + { + "epoch": 27.0, + "learning_rate": 0.0004025555555555556, + "loss": 0.2473, + "step": 1379 + }, + { + "epoch": 27.0, + "learning_rate": 0.0004024444444444444, + "loss": 0.2581, + "step": 1380 + }, + { + "epoch": 27.0, + "learning_rate": 0.0004023333333333333, + "loss": 0.284, + "step": 1381 + }, + { + "epoch": 27.0, + "learning_rate": 0.0004022222222222222, + "loss": 0.2687, + "step": 1382 + }, + { + "epoch": 27.0, + "learning_rate": 0.0004021111111111111, + "loss": 0.2615, + "step": 1383 + }, + { + "epoch": 27.0, + "learning_rate": 0.000402, + "loss": 0.3111, + "step": 1384 + }, + { + "epoch": 27.0, + "learning_rate": 0.0004018888888888889, + "loss": 0.3482, + "step": 1385 + }, + { + "epoch": 27.0, + "learning_rate": 0.0004017777777777778, + "loss": 0.3305, + "step": 1386 + }, + { + "epoch": 27.0, + "learning_rate": 0.00040166666666666665, + "loss": 0.2681, + "step": 1387 + }, + { + "epoch": 27.0, + "learning_rate": 0.00040155555555555555, + "loss": 0.3401, + "step": 1388 + }, + { + "epoch": 27.0, + "learning_rate": 0.00040144444444444445, + "loss": 0.2396, + "step": 1389 + }, + { + "epoch": 27.0, + "learning_rate": 0.00040133333333333335, + "loss": 0.3481, + "step": 1390 + }, + { + "epoch": 27.0, + "learning_rate": 0.0004012222222222222, + "loss": 0.3336, + "step": 1391 + }, + { + "epoch": 27.0, + "learning_rate": 0.0004011111111111111, + "loss": 0.3235, + "step": 1392 + }, + { + "epoch": 27.0, + "learning_rate": 0.00040100000000000004, + "loss": 0.3371, + "step": 1393 + }, + { + "epoch": 27.0, + "learning_rate": 0.0004008888888888889, + "loss": 0.2974, + "step": 1394 + }, + { + "epoch": 27.0, + "learning_rate": 0.0004007777777777778, + "loss": 0.3051, + "step": 1395 + }, + { + "epoch": 27.0, + "learning_rate": 0.0004006666666666667, + "loss": 0.3352, + "step": 1396 + }, + { + "epoch": 27.0, + "learning_rate": 0.0004005555555555556, + "loss": 0.3166, + "step": 1397 + }, + { + "epoch": 27.0, + "learning_rate": 0.0004004444444444444, + "loss": 0.2934, + "step": 1398 + }, + { + "epoch": 27.0, + "learning_rate": 0.0004003333333333333, + "loss": 0.2926, + "step": 1399 + }, + { + "epoch": 27.0, + "learning_rate": 0.0004002222222222222, + "loss": 0.2819, + "step": 1400 + }, + { + "epoch": 27.0, + "learning_rate": 0.00040011111111111107, + "loss": 0.3152, + "step": 1401 + }, + { + "epoch": 27.0, + "learning_rate": 0.0004, + "loss": 0.316, + "step": 1402 + }, + { + "epoch": 27.01, + "learning_rate": 0.0003998888888888889, + "loss": 0.2905, + "step": 1403 + }, + { + "epoch": 27.01, + "learning_rate": 0.0003997777777777778, + "loss": 0.2878, + "step": 1404 + }, + { + "epoch": 27.01, + "learning_rate": 0.00039966666666666666, + "loss": 0.2772, + "step": 1405 + }, + { + "epoch": 27.01, + "learning_rate": 0.00039955555555555556, + "loss": 0.2565, + "step": 1406 + }, + { + "epoch": 27.01, + "learning_rate": 0.00039944444444444446, + "loss": 0.2174, + "step": 1407 + }, + { + "epoch": 27.01, + "learning_rate": 0.0003993333333333333, + "loss": 0.3737, + "step": 1408 + }, + { + "epoch": 27.01, + "learning_rate": 0.0003992222222222222, + "loss": 0.2655, + "step": 1409 + }, + { + "epoch": 27.01, + "learning_rate": 0.00039911111111111115, + "loss": 0.2908, + "step": 1410 + }, + { + "epoch": 27.01, + "learning_rate": 0.00039900000000000005, + "loss": 0.311, + "step": 1411 + }, + { + "epoch": 27.01, + "learning_rate": 0.0003988888888888889, + "loss": 0.3075, + "step": 1412 + }, + { + "epoch": 27.01, + "learning_rate": 0.0003987777777777778, + "loss": 0.2811, + "step": 1413 + }, + { + "epoch": 27.01, + "learning_rate": 0.0003986666666666667, + "loss": 0.3558, + "step": 1414 + }, + { + "epoch": 27.01, + "learning_rate": 0.00039855555555555553, + "loss": 0.3028, + "step": 1415 + }, + { + "epoch": 27.01, + "learning_rate": 0.00039844444444444443, + "loss": 0.3077, + "step": 1416 + }, + { + "epoch": 27.01, + "learning_rate": 0.00039833333333333333, + "loss": 0.2863, + "step": 1417 + }, + { + "epoch": 27.01, + "learning_rate": 0.00039822222222222223, + "loss": 0.3295, + "step": 1418 + }, + { + "epoch": 27.01, + "learning_rate": 0.0003981111111111111, + "loss": 0.3551, + "step": 1419 + }, + { + "epoch": 27.01, + "learning_rate": 0.000398, + "loss": 0.308, + "step": 1420 + }, + { + "epoch": 27.01, + "learning_rate": 0.0003978888888888889, + "loss": 0.2906, + "step": 1421 + }, + { + "epoch": 27.01, + "learning_rate": 0.00039777777777777777, + "loss": 0.2678, + "step": 1422 + }, + { + "epoch": 27.01, + "learning_rate": 0.00039766666666666667, + "loss": 0.2795, + "step": 1423 + }, + { + "epoch": 27.01, + "learning_rate": 0.00039755555555555556, + "loss": 0.2558, + "step": 1424 + }, + { + "epoch": 27.01, + "learning_rate": 0.00039744444444444446, + "loss": 0.2926, + "step": 1425 + }, + { + "epoch": 27.01, + "learning_rate": 0.0003973333333333333, + "loss": 0.3115, + "step": 1426 + }, + { + "epoch": 27.01, + "learning_rate": 0.0003972222222222222, + "loss": 0.2812, + "step": 1427 + }, + { + "epoch": 27.01, + "learning_rate": 0.00039711111111111116, + "loss": 0.2598, + "step": 1428 + }, + { + "epoch": 28.0, + "learning_rate": 0.00039700000000000005, + "loss": 0.511, + "step": 1429 + }, + { + "epoch": 28.0, + "learning_rate": 0.0003968888888888889, + "loss": 0.3392, + "step": 1430 + }, + { + "epoch": 28.0, + "learning_rate": 0.0003967777777777778, + "loss": 0.2014, + "step": 1431 + }, + { + "epoch": 28.0, + "learning_rate": 0.0003966666666666667, + "loss": 0.2765, + "step": 1432 + }, + { + "epoch": 28.0, + "learning_rate": 0.00039655555555555554, + "loss": 0.3197, + "step": 1433 + }, + { + "epoch": 28.0, + "learning_rate": 0.00039644444444444444, + "loss": 0.2527, + "step": 1434 + }, + { + "epoch": 28.0, + "learning_rate": 0.00039633333333333334, + "loss": 0.31, + "step": 1435 + }, + { + "epoch": 28.0, + "learning_rate": 0.0003962222222222222, + "loss": 0.2814, + "step": 1436 + }, + { + "epoch": 28.0, + "learning_rate": 0.00039611111111111113, + "loss": 0.24, + "step": 1437 + }, + { + "epoch": 28.0, + "learning_rate": 0.00039600000000000003, + "loss": 0.3211, + "step": 1438 + }, + { + "epoch": 28.0, + "learning_rate": 0.00039588888888888893, + "loss": 0.2957, + "step": 1439 + }, + { + "epoch": 28.0, + "learning_rate": 0.00039577777777777777, + "loss": 0.2532, + "step": 1440 + }, + { + "epoch": 28.0, + "learning_rate": 0.00039566666666666667, + "loss": 0.3334, + "step": 1441 + }, + { + "epoch": 28.0, + "learning_rate": 0.00039555555555555557, + "loss": 0.2976, + "step": 1442 + }, + { + "epoch": 28.0, + "learning_rate": 0.0003954444444444444, + "loss": 0.394, + "step": 1443 + }, + { + "epoch": 28.0, + "learning_rate": 0.0003953333333333333, + "loss": 0.3018, + "step": 1444 + }, + { + "epoch": 28.0, + "learning_rate": 0.0003952222222222222, + "loss": 0.268, + "step": 1445 + }, + { + "epoch": 28.0, + "learning_rate": 0.00039511111111111116, + "loss": 0.2677, + "step": 1446 + }, + { + "epoch": 28.0, + "learning_rate": 0.000395, + "loss": 0.241, + "step": 1447 + }, + { + "epoch": 28.0, + "learning_rate": 0.0003948888888888889, + "loss": 0.3369, + "step": 1448 + }, + { + "epoch": 28.0, + "learning_rate": 0.0003947777777777778, + "loss": 0.28, + "step": 1449 + }, + { + "epoch": 28.0, + "learning_rate": 0.0003946666666666667, + "loss": 0.2817, + "step": 1450 + }, + { + "epoch": 28.0, + "learning_rate": 0.00039455555555555554, + "loss": 0.2857, + "step": 1451 + }, + { + "epoch": 28.0, + "learning_rate": 0.00039444444444444444, + "loss": 0.269, + "step": 1452 + }, + { + "epoch": 28.0, + "learning_rate": 0.00039433333333333334, + "loss": 0.3343, + "step": 1453 + }, + { + "epoch": 28.01, + "learning_rate": 0.0003942222222222222, + "loss": 0.3308, + "step": 1454 + }, + { + "epoch": 28.01, + "learning_rate": 0.00039411111111111114, + "loss": 0.2328, + "step": 1455 + }, + { + "epoch": 28.01, + "learning_rate": 0.00039400000000000004, + "loss": 0.2802, + "step": 1456 + }, + { + "epoch": 28.01, + "learning_rate": 0.00039388888888888893, + "loss": 0.345, + "step": 1457 + }, + { + "epoch": 28.01, + "learning_rate": 0.0003937777777777778, + "loss": 0.2812, + "step": 1458 + }, + { + "epoch": 28.01, + "learning_rate": 0.0003936666666666667, + "loss": 0.287, + "step": 1459 + }, + { + "epoch": 28.01, + "learning_rate": 0.0003935555555555556, + "loss": 0.2965, + "step": 1460 + }, + { + "epoch": 28.01, + "learning_rate": 0.0003934444444444444, + "loss": 0.2636, + "step": 1461 + }, + { + "epoch": 28.01, + "learning_rate": 0.0003933333333333333, + "loss": 0.2831, + "step": 1462 + }, + { + "epoch": 28.01, + "learning_rate": 0.0003932222222222222, + "loss": 0.3653, + "step": 1463 + }, + { + "epoch": 28.01, + "learning_rate": 0.00039311111111111117, + "loss": 0.2618, + "step": 1464 + }, + { + "epoch": 28.01, + "learning_rate": 0.000393, + "loss": 0.3284, + "step": 1465 + }, + { + "epoch": 28.01, + "learning_rate": 0.0003928888888888889, + "loss": 0.3122, + "step": 1466 + }, + { + "epoch": 28.01, + "learning_rate": 0.0003927777777777778, + "loss": 0.353, + "step": 1467 + }, + { + "epoch": 28.01, + "learning_rate": 0.00039266666666666665, + "loss": 0.2916, + "step": 1468 + }, + { + "epoch": 28.01, + "learning_rate": 0.00039255555555555555, + "loss": 0.3033, + "step": 1469 + }, + { + "epoch": 28.01, + "learning_rate": 0.00039244444444444445, + "loss": 0.2843, + "step": 1470 + }, + { + "epoch": 28.01, + "learning_rate": 0.00039233333333333335, + "loss": 0.2676, + "step": 1471 + }, + { + "epoch": 28.01, + "learning_rate": 0.00039222222222222225, + "loss": 0.2933, + "step": 1472 + }, + { + "epoch": 28.01, + "learning_rate": 0.00039211111111111114, + "loss": 0.2522, + "step": 1473 + }, + { + "epoch": 28.01, + "learning_rate": 0.00039200000000000004, + "loss": 0.3366, + "step": 1474 + }, + { + "epoch": 28.01, + "learning_rate": 0.0003918888888888889, + "loss": 0.3289, + "step": 1475 + }, + { + "epoch": 28.01, + "learning_rate": 0.0003917777777777778, + "loss": 0.2894, + "step": 1476 + }, + { + "epoch": 28.01, + "learning_rate": 0.0003916666666666667, + "loss": 0.3175, + "step": 1477 + }, + { + "epoch": 28.01, + "learning_rate": 0.0003915555555555556, + "loss": 0.2504, + "step": 1478 + }, + { + "epoch": 28.01, + "learning_rate": 0.0003914444444444444, + "loss": 0.2951, + "step": 1479 + }, + { + "epoch": 29.0, + "learning_rate": 0.0003913333333333333, + "loss": 0.4458, + "step": 1480 + }, + { + "epoch": 29.0, + "learning_rate": 0.0003912222222222223, + "loss": 0.2694, + "step": 1481 + }, + { + "epoch": 29.0, + "learning_rate": 0.0003911111111111111, + "loss": 0.3026, + "step": 1482 + }, + { + "epoch": 29.0, + "learning_rate": 0.000391, + "loss": 0.2957, + "step": 1483 + }, + { + "epoch": 29.0, + "learning_rate": 0.0003908888888888889, + "loss": 0.2965, + "step": 1484 + }, + { + "epoch": 29.0, + "learning_rate": 0.0003907777777777778, + "loss": 0.2937, + "step": 1485 + }, + { + "epoch": 29.0, + "learning_rate": 0.00039066666666666666, + "loss": 0.263, + "step": 1486 + }, + { + "epoch": 29.0, + "learning_rate": 0.00039055555555555556, + "loss": 0.2802, + "step": 1487 + }, + { + "epoch": 29.0, + "learning_rate": 0.00039044444444444445, + "loss": 0.2604, + "step": 1488 + }, + { + "epoch": 29.0, + "learning_rate": 0.0003903333333333333, + "loss": 0.2273, + "step": 1489 + }, + { + "epoch": 29.0, + "learning_rate": 0.00039022222222222225, + "loss": 0.274, + "step": 1490 + }, + { + "epoch": 29.0, + "learning_rate": 0.00039011111111111115, + "loss": 0.2338, + "step": 1491 + }, + { + "epoch": 29.0, + "learning_rate": 0.00039000000000000005, + "loss": 0.2802, + "step": 1492 + }, + { + "epoch": 29.0, + "learning_rate": 0.0003898888888888889, + "loss": 0.3652, + "step": 1493 + }, + { + "epoch": 29.0, + "learning_rate": 0.0003897777777777778, + "loss": 0.2882, + "step": 1494 + }, + { + "epoch": 29.0, + "learning_rate": 0.0003896666666666667, + "loss": 0.3232, + "step": 1495 + }, + { + "epoch": 29.0, + "learning_rate": 0.00038955555555555553, + "loss": 0.2425, + "step": 1496 + }, + { + "epoch": 29.0, + "learning_rate": 0.00038944444444444443, + "loss": 0.2548, + "step": 1497 + }, + { + "epoch": 29.0, + "learning_rate": 0.00038933333333333333, + "loss": 0.2668, + "step": 1498 + }, + { + "epoch": 29.0, + "learning_rate": 0.0003892222222222223, + "loss": 0.248, + "step": 1499 + }, + { + "epoch": 29.0, + "learning_rate": 0.0003891111111111111, + "loss": 0.3616, + "step": 1500 + }, + { + "epoch": 29.0, + "eval_cer": 0.08780225041896098, + "eval_loss": 0.3539605140686035, + "eval_runtime": 56.193, + "eval_samples_per_second": 7.403, + "eval_steps_per_second": 0.231, + "eval_wer": 0.37162714636140637, + "step": 1500 + }, + { + "epoch": 29.0, + "learning_rate": 0.000389, + "loss": 0.3331, + "step": 1501 + }, + { + "epoch": 29.0, + "learning_rate": 0.0003888888888888889, + "loss": 0.2593, + "step": 1502 + }, + { + "epoch": 29.0, + "learning_rate": 0.00038877777777777777, + "loss": 0.3251, + "step": 1503 + }, + { + "epoch": 29.0, + "learning_rate": 0.00038866666666666666, + "loss": 0.2898, + "step": 1504 + }, + { + "epoch": 29.01, + "learning_rate": 0.00038855555555555556, + "loss": 0.2235, + "step": 1505 + }, + { + "epoch": 29.01, + "learning_rate": 0.00038844444444444446, + "loss": 0.3117, + "step": 1506 + }, + { + "epoch": 29.01, + "learning_rate": 0.0003883333333333333, + "loss": 0.2717, + "step": 1507 + }, + { + "epoch": 29.01, + "learning_rate": 0.00038822222222222226, + "loss": 0.245, + "step": 1508 + }, + { + "epoch": 29.01, + "learning_rate": 0.00038811111111111115, + "loss": 0.2402, + "step": 1509 + }, + { + "epoch": 29.01, + "learning_rate": 0.000388, + "loss": 0.2457, + "step": 1510 + }, + { + "epoch": 29.01, + "learning_rate": 0.0003878888888888889, + "loss": 0.3059, + "step": 1511 + }, + { + "epoch": 29.01, + "learning_rate": 0.0003877777777777778, + "loss": 0.2898, + "step": 1512 + }, + { + "epoch": 29.01, + "learning_rate": 0.0003876666666666667, + "loss": 0.2808, + "step": 1513 + }, + { + "epoch": 29.01, + "learning_rate": 0.00038755555555555554, + "loss": 0.2807, + "step": 1514 + }, + { + "epoch": 29.01, + "learning_rate": 0.00038744444444444444, + "loss": 0.2955, + "step": 1515 + }, + { + "epoch": 29.01, + "learning_rate": 0.00038733333333333333, + "loss": 0.3607, + "step": 1516 + }, + { + "epoch": 29.01, + "learning_rate": 0.00038722222222222223, + "loss": 0.298, + "step": 1517 + }, + { + "epoch": 29.01, + "learning_rate": 0.00038711111111111113, + "loss": 0.3424, + "step": 1518 + }, + { + "epoch": 29.01, + "learning_rate": 0.00038700000000000003, + "loss": 0.2639, + "step": 1519 + }, + { + "epoch": 29.01, + "learning_rate": 0.0003868888888888889, + "loss": 0.254, + "step": 1520 + }, + { + "epoch": 29.01, + "learning_rate": 0.00038677777777777777, + "loss": 0.255, + "step": 1521 + }, + { + "epoch": 29.01, + "learning_rate": 0.00038666666666666667, + "loss": 0.2619, + "step": 1522 + }, + { + "epoch": 29.01, + "learning_rate": 0.00038655555555555557, + "loss": 0.286, + "step": 1523 + }, + { + "epoch": 29.01, + "learning_rate": 0.0003864444444444444, + "loss": 0.2506, + "step": 1524 + }, + { + "epoch": 29.01, + "learning_rate": 0.0003863333333333333, + "loss": 0.3152, + "step": 1525 + }, + { + "epoch": 29.01, + "learning_rate": 0.00038622222222222226, + "loss": 0.2988, + "step": 1526 + }, + { + "epoch": 29.01, + "learning_rate": 0.00038611111111111116, + "loss": 0.2896, + "step": 1527 + }, + { + "epoch": 29.01, + "learning_rate": 0.000386, + "loss": 0.2559, + "step": 1528 + }, + { + "epoch": 29.01, + "learning_rate": 0.0003858888888888889, + "loss": 0.2547, + "step": 1529 + }, + { + "epoch": 29.01, + "learning_rate": 0.0003857777777777778, + "loss": 0.2862, + "step": 1530 + }, + { + "epoch": 30.0, + "learning_rate": 0.00038566666666666664, + "loss": 0.4272, + "step": 1531 + }, + { + "epoch": 30.0, + "learning_rate": 0.00038555555555555554, + "loss": 0.299, + "step": 1532 + }, + { + "epoch": 30.0, + "learning_rate": 0.00038544444444444444, + "loss": 0.3202, + "step": 1533 + }, + { + "epoch": 30.0, + "learning_rate": 0.0003853333333333334, + "loss": 0.252, + "step": 1534 + }, + { + "epoch": 30.0, + "learning_rate": 0.00038522222222222224, + "loss": 0.2552, + "step": 1535 + }, + { + "epoch": 30.0, + "learning_rate": 0.00038511111111111114, + "loss": 0.2688, + "step": 1536 + }, + { + "epoch": 30.0, + "learning_rate": 0.00038500000000000003, + "loss": 0.2642, + "step": 1537 + }, + { + "epoch": 30.0, + "learning_rate": 0.0003848888888888889, + "loss": 0.2991, + "step": 1538 + }, + { + "epoch": 30.0, + "learning_rate": 0.0003847777777777778, + "loss": 0.3036, + "step": 1539 + }, + { + "epoch": 30.0, + "learning_rate": 0.0003846666666666667, + "loss": 0.2327, + "step": 1540 + }, + { + "epoch": 30.0, + "learning_rate": 0.00038455555555555557, + "loss": 0.2873, + "step": 1541 + }, + { + "epoch": 30.0, + "learning_rate": 0.0003844444444444444, + "loss": 0.3185, + "step": 1542 + }, + { + "epoch": 30.0, + "learning_rate": 0.00038433333333333337, + "loss": 0.2627, + "step": 1543 + }, + { + "epoch": 30.0, + "learning_rate": 0.00038422222222222227, + "loss": 0.2513, + "step": 1544 + }, + { + "epoch": 30.0, + "learning_rate": 0.0003841111111111111, + "loss": 0.3016, + "step": 1545 + }, + { + "epoch": 30.0, + "learning_rate": 0.000384, + "loss": 0.3254, + "step": 1546 + }, + { + "epoch": 30.0, + "learning_rate": 0.0003838888888888889, + "loss": 0.3413, + "step": 1547 + }, + { + "epoch": 30.0, + "learning_rate": 0.0003837777777777778, + "loss": 0.3514, + "step": 1548 + }, + { + "epoch": 30.0, + "learning_rate": 0.00038366666666666665, + "loss": 0.2796, + "step": 1549 + }, + { + "epoch": 30.0, + "learning_rate": 0.00038355555555555555, + "loss": 0.2924, + "step": 1550 + }, + { + "epoch": 30.0, + "learning_rate": 0.00038344444444444445, + "loss": 0.2926, + "step": 1551 + }, + { + "epoch": 30.0, + "learning_rate": 0.00038333333333333334, + "loss": 0.3254, + "step": 1552 + }, + { + "epoch": 30.0, + "learning_rate": 0.00038322222222222224, + "loss": 0.3199, + "step": 1553 + }, + { + "epoch": 30.0, + "learning_rate": 0.00038311111111111114, + "loss": 0.2917, + "step": 1554 + }, + { + "epoch": 30.0, + "learning_rate": 0.00038300000000000004, + "loss": 0.2836, + "step": 1555 + }, + { + "epoch": 30.01, + "learning_rate": 0.0003828888888888889, + "loss": 0.2254, + "step": 1556 + }, + { + "epoch": 30.01, + "learning_rate": 0.0003827777777777778, + "loss": 0.2636, + "step": 1557 + }, + { + "epoch": 30.01, + "learning_rate": 0.0003826666666666667, + "loss": 0.2555, + "step": 1558 + }, + { + "epoch": 30.01, + "learning_rate": 0.0003825555555555555, + "loss": 0.3178, + "step": 1559 + }, + { + "epoch": 30.01, + "learning_rate": 0.0003824444444444444, + "loss": 0.2931, + "step": 1560 + }, + { + "epoch": 30.01, + "learning_rate": 0.0003823333333333334, + "loss": 0.2331, + "step": 1561 + }, + { + "epoch": 30.01, + "learning_rate": 0.0003822222222222223, + "loss": 0.3121, + "step": 1562 + }, + { + "epoch": 30.01, + "learning_rate": 0.0003821111111111111, + "loss": 0.2904, + "step": 1563 + }, + { + "epoch": 30.01, + "learning_rate": 0.000382, + "loss": 0.2753, + "step": 1564 + }, + { + "epoch": 30.01, + "learning_rate": 0.0003818888888888889, + "loss": 0.2573, + "step": 1565 + }, + { + "epoch": 30.01, + "learning_rate": 0.00038177777777777776, + "loss": 0.2587, + "step": 1566 + }, + { + "epoch": 30.01, + "learning_rate": 0.00038166666666666666, + "loss": 0.3236, + "step": 1567 + }, + { + "epoch": 30.01, + "learning_rate": 0.00038155555555555555, + "loss": 0.2797, + "step": 1568 + }, + { + "epoch": 30.01, + "learning_rate": 0.00038144444444444445, + "loss": 0.2735, + "step": 1569 + }, + { + "epoch": 30.01, + "learning_rate": 0.00038133333333333335, + "loss": 0.2546, + "step": 1570 + }, + { + "epoch": 30.01, + "learning_rate": 0.00038122222222222225, + "loss": 0.3075, + "step": 1571 + }, + { + "epoch": 30.01, + "learning_rate": 0.00038111111111111115, + "loss": 0.2621, + "step": 1572 + }, + { + "epoch": 30.01, + "learning_rate": 0.000381, + "loss": 0.3496, + "step": 1573 + }, + { + "epoch": 30.01, + "learning_rate": 0.0003808888888888889, + "loss": 0.3317, + "step": 1574 + }, + { + "epoch": 30.01, + "learning_rate": 0.0003807777777777778, + "loss": 0.2669, + "step": 1575 + }, + { + "epoch": 30.01, + "learning_rate": 0.0003806666666666667, + "loss": 0.3364, + "step": 1576 + }, + { + "epoch": 30.01, + "learning_rate": 0.00038055555555555553, + "loss": 0.3241, + "step": 1577 + }, + { + "epoch": 30.01, + "learning_rate": 0.00038044444444444443, + "loss": 0.2806, + "step": 1578 + }, + { + "epoch": 30.01, + "learning_rate": 0.0003803333333333334, + "loss": 0.2567, + "step": 1579 + }, + { + "epoch": 30.01, + "learning_rate": 0.0003802222222222222, + "loss": 0.3124, + "step": 1580 + }, + { + "epoch": 30.01, + "learning_rate": 0.0003801111111111111, + "loss": 0.3158, + "step": 1581 + }, + { + "epoch": 31.0, + "learning_rate": 0.00038, + "loss": 0.4234, + "step": 1582 + }, + { + "epoch": 31.0, + "learning_rate": 0.0003798888888888889, + "loss": 0.2911, + "step": 1583 + }, + { + "epoch": 31.0, + "learning_rate": 0.00037977777777777776, + "loss": 0.2165, + "step": 1584 + }, + { + "epoch": 31.0, + "learning_rate": 0.00037966666666666666, + "loss": 0.2572, + "step": 1585 + }, + { + "epoch": 31.0, + "learning_rate": 0.00037955555555555556, + "loss": 0.2573, + "step": 1586 + }, + { + "epoch": 31.0, + "learning_rate": 0.0003794444444444444, + "loss": 0.3187, + "step": 1587 + }, + { + "epoch": 31.0, + "learning_rate": 0.00037933333333333336, + "loss": 0.2278, + "step": 1588 + }, + { + "epoch": 31.0, + "learning_rate": 0.00037922222222222225, + "loss": 0.2363, + "step": 1589 + }, + { + "epoch": 31.0, + "learning_rate": 0.00037911111111111115, + "loss": 0.2682, + "step": 1590 + }, + { + "epoch": 31.0, + "learning_rate": 0.000379, + "loss": 0.2507, + "step": 1591 + }, + { + "epoch": 31.0, + "learning_rate": 0.0003788888888888889, + "loss": 0.316, + "step": 1592 + }, + { + "epoch": 31.0, + "learning_rate": 0.0003787777777777778, + "loss": 0.2765, + "step": 1593 + }, + { + "epoch": 31.0, + "learning_rate": 0.00037866666666666664, + "loss": 0.2734, + "step": 1594 + }, + { + "epoch": 31.0, + "learning_rate": 0.00037855555555555554, + "loss": 0.2657, + "step": 1595 + }, + { + "epoch": 31.0, + "learning_rate": 0.0003784444444444445, + "loss": 0.2905, + "step": 1596 + }, + { + "epoch": 31.0, + "learning_rate": 0.0003783333333333334, + "loss": 0.3635, + "step": 1597 + }, + { + "epoch": 31.0, + "learning_rate": 0.00037822222222222223, + "loss": 0.2941, + "step": 1598 + }, + { + "epoch": 31.0, + "learning_rate": 0.00037811111111111113, + "loss": 0.3079, + "step": 1599 + }, + { + "epoch": 31.0, + "learning_rate": 0.000378, + "loss": 0.2776, + "step": 1600 + }, + { + "epoch": 31.0, + "learning_rate": 0.00037788888888888887, + "loss": 0.2561, + "step": 1601 + }, + { + "epoch": 31.0, + "learning_rate": 0.00037777777777777777, + "loss": 0.268, + "step": 1602 + }, + { + "epoch": 31.0, + "learning_rate": 0.00037766666666666667, + "loss": 0.2403, + "step": 1603 + }, + { + "epoch": 31.0, + "learning_rate": 0.00037755555555555557, + "loss": 0.2798, + "step": 1604 + }, + { + "epoch": 31.0, + "learning_rate": 0.00037744444444444446, + "loss": 0.3099, + "step": 1605 + }, + { + "epoch": 31.0, + "learning_rate": 0.00037733333333333336, + "loss": 0.2758, + "step": 1606 + }, + { + "epoch": 31.01, + "learning_rate": 0.00037722222222222226, + "loss": 0.2738, + "step": 1607 + }, + { + "epoch": 31.01, + "learning_rate": 0.0003771111111111111, + "loss": 0.3022, + "step": 1608 + }, + { + "epoch": 31.01, + "learning_rate": 0.000377, + "loss": 0.24, + "step": 1609 + }, + { + "epoch": 31.01, + "learning_rate": 0.0003768888888888889, + "loss": 0.2427, + "step": 1610 + }, + { + "epoch": 31.01, + "learning_rate": 0.0003767777777777778, + "loss": 0.2326, + "step": 1611 + }, + { + "epoch": 31.01, + "learning_rate": 0.00037666666666666664, + "loss": 0.3168, + "step": 1612 + }, + { + "epoch": 31.01, + "learning_rate": 0.00037655555555555554, + "loss": 0.2794, + "step": 1613 + }, + { + "epoch": 31.01, + "learning_rate": 0.0003764444444444445, + "loss": 0.328, + "step": 1614 + }, + { + "epoch": 31.01, + "learning_rate": 0.00037633333333333334, + "loss": 0.306, + "step": 1615 + }, + { + "epoch": 31.01, + "learning_rate": 0.00037622222222222224, + "loss": 0.2812, + "step": 1616 + }, + { + "epoch": 31.01, + "learning_rate": 0.00037611111111111113, + "loss": 0.3422, + "step": 1617 + }, + { + "epoch": 31.01, + "learning_rate": 0.00037600000000000003, + "loss": 0.3062, + "step": 1618 + }, + { + "epoch": 31.01, + "learning_rate": 0.0003758888888888889, + "loss": 0.2847, + "step": 1619 + }, + { + "epoch": 31.01, + "learning_rate": 0.0003757777777777778, + "loss": 0.3376, + "step": 1620 + }, + { + "epoch": 31.01, + "learning_rate": 0.00037566666666666667, + "loss": 0.2964, + "step": 1621 + }, + { + "epoch": 31.01, + "learning_rate": 0.0003755555555555555, + "loss": 0.2941, + "step": 1622 + }, + { + "epoch": 31.01, + "learning_rate": 0.00037544444444444447, + "loss": 0.2483, + "step": 1623 + }, + { + "epoch": 31.01, + "learning_rate": 0.00037533333333333337, + "loss": 0.2645, + "step": 1624 + }, + { + "epoch": 31.01, + "learning_rate": 0.00037522222222222227, + "loss": 0.3321, + "step": 1625 + }, + { + "epoch": 31.01, + "learning_rate": 0.0003751111111111111, + "loss": 0.2652, + "step": 1626 + }, + { + "epoch": 31.01, + "learning_rate": 0.000375, + "loss": 0.2554, + "step": 1627 + }, + { + "epoch": 31.01, + "learning_rate": 0.0003748888888888889, + "loss": 0.2665, + "step": 1628 + }, + { + "epoch": 31.01, + "learning_rate": 0.00037477777777777775, + "loss": 0.3248, + "step": 1629 + }, + { + "epoch": 31.01, + "learning_rate": 0.00037466666666666665, + "loss": 0.2951, + "step": 1630 + }, + { + "epoch": 31.01, + "learning_rate": 0.00037455555555555555, + "loss": 0.307, + "step": 1631 + }, + { + "epoch": 31.01, + "learning_rate": 0.0003744444444444445, + "loss": 0.2848, + "step": 1632 + }, + { + "epoch": 32.0, + "learning_rate": 0.00037433333333333334, + "loss": 0.4679, + "step": 1633 + }, + { + "epoch": 32.0, + "learning_rate": 0.00037422222222222224, + "loss": 0.2618, + "step": 1634 + }, + { + "epoch": 32.0, + "learning_rate": 0.00037411111111111114, + "loss": 0.2807, + "step": 1635 + }, + { + "epoch": 32.0, + "learning_rate": 0.000374, + "loss": 0.3315, + "step": 1636 + }, + { + "epoch": 32.0, + "learning_rate": 0.0003738888888888889, + "loss": 0.2653, + "step": 1637 + }, + { + "epoch": 32.0, + "learning_rate": 0.0003737777777777778, + "loss": 0.2968, + "step": 1638 + }, + { + "epoch": 32.0, + "learning_rate": 0.0003736666666666667, + "loss": 0.3254, + "step": 1639 + }, + { + "epoch": 32.0, + "learning_rate": 0.0003735555555555555, + "loss": 0.2597, + "step": 1640 + }, + { + "epoch": 32.0, + "learning_rate": 0.0003734444444444445, + "loss": 0.2577, + "step": 1641 + }, + { + "epoch": 32.0, + "learning_rate": 0.0003733333333333334, + "loss": 0.2671, + "step": 1642 + }, + { + "epoch": 32.0, + "learning_rate": 0.0003732222222222222, + "loss": 0.2761, + "step": 1643 + }, + { + "epoch": 32.0, + "learning_rate": 0.0003731111111111111, + "loss": 0.2565, + "step": 1644 + }, + { + "epoch": 32.0, + "learning_rate": 0.000373, + "loss": 0.3047, + "step": 1645 + }, + { + "epoch": 32.0, + "learning_rate": 0.0003728888888888889, + "loss": 0.3494, + "step": 1646 + }, + { + "epoch": 32.0, + "learning_rate": 0.00037277777777777776, + "loss": 0.3747, + "step": 1647 + }, + { + "epoch": 32.0, + "learning_rate": 0.00037266666666666665, + "loss": 0.3437, + "step": 1648 + }, + { + "epoch": 32.0, + "learning_rate": 0.00037255555555555555, + "loss": 0.2927, + "step": 1649 + }, + { + "epoch": 32.0, + "learning_rate": 0.00037244444444444445, + "loss": 0.2777, + "step": 1650 + }, + { + "epoch": 32.0, + "learning_rate": 0.00037233333333333335, + "loss": 0.3515, + "step": 1651 + }, + { + "epoch": 32.0, + "learning_rate": 0.00037222222222222225, + "loss": 0.2968, + "step": 1652 + }, + { + "epoch": 32.0, + "learning_rate": 0.00037211111111111115, + "loss": 0.2961, + "step": 1653 + }, + { + "epoch": 32.0, + "learning_rate": 0.000372, + "loss": 0.2821, + "step": 1654 + }, + { + "epoch": 32.0, + "learning_rate": 0.0003718888888888889, + "loss": 0.2559, + "step": 1655 + }, + { + "epoch": 32.0, + "learning_rate": 0.0003717777777777778, + "loss": 0.3718, + "step": 1656 + }, + { + "epoch": 32.01, + "learning_rate": 0.00037166666666666663, + "loss": 0.2835, + "step": 1657 + }, + { + "epoch": 32.01, + "learning_rate": 0.00037155555555555553, + "loss": 0.2935, + "step": 1658 + }, + { + "epoch": 32.01, + "learning_rate": 0.0003714444444444445, + "loss": 0.2574, + "step": 1659 + }, + { + "epoch": 32.01, + "learning_rate": 0.0003713333333333334, + "loss": 0.3039, + "step": 1660 + }, + { + "epoch": 32.01, + "learning_rate": 0.0003712222222222222, + "loss": 0.2568, + "step": 1661 + }, + { + "epoch": 32.01, + "learning_rate": 0.0003711111111111111, + "loss": 0.2708, + "step": 1662 + }, + { + "epoch": 32.01, + "learning_rate": 0.000371, + "loss": 0.2464, + "step": 1663 + }, + { + "epoch": 32.01, + "learning_rate": 0.00037088888888888886, + "loss": 0.2795, + "step": 1664 + }, + { + "epoch": 32.01, + "learning_rate": 0.00037077777777777776, + "loss": 0.2985, + "step": 1665 + }, + { + "epoch": 32.01, + "learning_rate": 0.00037066666666666666, + "loss": 0.293, + "step": 1666 + }, + { + "epoch": 32.01, + "learning_rate": 0.0003705555555555556, + "loss": 0.306, + "step": 1667 + }, + { + "epoch": 32.01, + "learning_rate": 0.00037044444444444446, + "loss": 0.2913, + "step": 1668 + }, + { + "epoch": 32.01, + "learning_rate": 0.00037033333333333335, + "loss": 0.2817, + "step": 1669 + }, + { + "epoch": 32.01, + "learning_rate": 0.00037022222222222225, + "loss": 0.2994, + "step": 1670 + }, + { + "epoch": 32.01, + "learning_rate": 0.0003701111111111111, + "loss": 0.2795, + "step": 1671 + }, + { + "epoch": 32.01, + "learning_rate": 0.00037, + "loss": 0.2653, + "step": 1672 + }, + { + "epoch": 32.01, + "learning_rate": 0.0003698888888888889, + "loss": 0.2503, + "step": 1673 + }, + { + "epoch": 32.01, + "learning_rate": 0.0003697777777777778, + "loss": 0.2675, + "step": 1674 + }, + { + "epoch": 32.01, + "learning_rate": 0.00036966666666666664, + "loss": 0.3405, + "step": 1675 + }, + { + "epoch": 32.01, + "learning_rate": 0.0003695555555555556, + "loss": 0.2064, + "step": 1676 + }, + { + "epoch": 32.01, + "learning_rate": 0.0003694444444444445, + "loss": 0.2855, + "step": 1677 + }, + { + "epoch": 32.01, + "learning_rate": 0.00036933333333333333, + "loss": 0.2987, + "step": 1678 + }, + { + "epoch": 32.01, + "learning_rate": 0.00036922222222222223, + "loss": 0.2353, + "step": 1679 + }, + { + "epoch": 32.01, + "learning_rate": 0.0003691111111111111, + "loss": 0.2876, + "step": 1680 + }, + { + "epoch": 32.01, + "learning_rate": 0.000369, + "loss": 0.2887, + "step": 1681 + }, + { + "epoch": 32.01, + "learning_rate": 0.00036888888888888887, + "loss": 0.2893, + "step": 1682 + }, + { + "epoch": 32.01, + "learning_rate": 0.00036877777777777777, + "loss": 0.258, + "step": 1683 + }, + { + "epoch": 33.0, + "learning_rate": 0.00036866666666666667, + "loss": 0.4462, + "step": 1684 + }, + { + "epoch": 33.0, + "learning_rate": 0.00036855555555555556, + "loss": 0.2899, + "step": 1685 + }, + { + "epoch": 33.0, + "learning_rate": 0.00036844444444444446, + "loss": 0.2577, + "step": 1686 + }, + { + "epoch": 33.0, + "learning_rate": 0.00036833333333333336, + "loss": 0.2966, + "step": 1687 + }, + { + "epoch": 33.0, + "learning_rate": 0.00036822222222222226, + "loss": 0.2832, + "step": 1688 + }, + { + "epoch": 33.0, + "learning_rate": 0.0003681111111111111, + "loss": 0.2612, + "step": 1689 + }, + { + "epoch": 33.0, + "learning_rate": 0.000368, + "loss": 0.2788, + "step": 1690 + }, + { + "epoch": 33.0, + "learning_rate": 0.0003678888888888889, + "loss": 0.3099, + "step": 1691 + }, + { + "epoch": 33.0, + "learning_rate": 0.00036777777777777774, + "loss": 0.2742, + "step": 1692 + }, + { + "epoch": 33.0, + "learning_rate": 0.00036766666666666664, + "loss": 0.2849, + "step": 1693 + }, + { + "epoch": 33.0, + "learning_rate": 0.0003675555555555556, + "loss": 0.2413, + "step": 1694 + }, + { + "epoch": 33.0, + "learning_rate": 0.0003674444444444445, + "loss": 0.2904, + "step": 1695 + }, + { + "epoch": 33.0, + "learning_rate": 0.00036733333333333334, + "loss": 0.2356, + "step": 1696 + }, + { + "epoch": 33.0, + "learning_rate": 0.00036722222222222223, + "loss": 0.3092, + "step": 1697 + }, + { + "epoch": 33.0, + "learning_rate": 0.00036711111111111113, + "loss": 0.2869, + "step": 1698 + }, + { + "epoch": 33.0, + "learning_rate": 0.000367, + "loss": 0.286, + "step": 1699 + }, + { + "epoch": 33.0, + "learning_rate": 0.0003668888888888889, + "loss": 0.3003, + "step": 1700 + }, + { + "epoch": 33.0, + "learning_rate": 0.00036677777777777777, + "loss": 0.2698, + "step": 1701 + }, + { + "epoch": 33.0, + "learning_rate": 0.00036666666666666667, + "loss": 0.2141, + "step": 1702 + }, + { + "epoch": 33.0, + "learning_rate": 0.00036655555555555557, + "loss": 0.3046, + "step": 1703 + }, + { + "epoch": 33.0, + "learning_rate": 0.00036644444444444447, + "loss": 0.2944, + "step": 1704 + }, + { + "epoch": 33.0, + "learning_rate": 0.00036633333333333337, + "loss": 0.2724, + "step": 1705 + }, + { + "epoch": 33.0, + "learning_rate": 0.0003662222222222222, + "loss": 0.2316, + "step": 1706 + }, + { + "epoch": 33.0, + "learning_rate": 0.0003661111111111111, + "loss": 0.2838, + "step": 1707 + }, + { + "epoch": 33.01, + "learning_rate": 0.000366, + "loss": 0.3034, + "step": 1708 + }, + { + "epoch": 33.01, + "learning_rate": 0.0003658888888888889, + "loss": 0.2921, + "step": 1709 + }, + { + "epoch": 33.01, + "learning_rate": 0.00036577777777777775, + "loss": 0.2825, + "step": 1710 + }, + { + "epoch": 33.01, + "learning_rate": 0.00036566666666666665, + "loss": 0.3122, + "step": 1711 + }, + { + "epoch": 33.01, + "learning_rate": 0.0003655555555555556, + "loss": 0.2767, + "step": 1712 + }, + { + "epoch": 33.01, + "learning_rate": 0.00036544444444444444, + "loss": 0.2736, + "step": 1713 + }, + { + "epoch": 33.01, + "learning_rate": 0.00036533333333333334, + "loss": 0.2509, + "step": 1714 + }, + { + "epoch": 33.01, + "learning_rate": 0.00036522222222222224, + "loss": 0.267, + "step": 1715 + }, + { + "epoch": 33.01, + "learning_rate": 0.00036511111111111114, + "loss": 0.2401, + "step": 1716 + }, + { + "epoch": 33.01, + "learning_rate": 0.000365, + "loss": 0.3056, + "step": 1717 + }, + { + "epoch": 33.01, + "learning_rate": 0.0003648888888888889, + "loss": 0.2375, + "step": 1718 + }, + { + "epoch": 33.01, + "learning_rate": 0.0003647777777777778, + "loss": 0.2877, + "step": 1719 + }, + { + "epoch": 33.01, + "learning_rate": 0.0003646666666666666, + "loss": 0.2754, + "step": 1720 + }, + { + "epoch": 33.01, + "learning_rate": 0.0003645555555555556, + "loss": 0.2794, + "step": 1721 + }, + { + "epoch": 33.01, + "learning_rate": 0.00036444444444444447, + "loss": 0.293, + "step": 1722 + }, + { + "epoch": 33.01, + "learning_rate": 0.00036433333333333337, + "loss": 0.3155, + "step": 1723 + }, + { + "epoch": 33.01, + "learning_rate": 0.0003642222222222222, + "loss": 0.2088, + "step": 1724 + }, + { + "epoch": 33.01, + "learning_rate": 0.0003641111111111111, + "loss": 0.2922, + "step": 1725 + }, + { + "epoch": 33.01, + "learning_rate": 0.000364, + "loss": 0.2838, + "step": 1726 + }, + { + "epoch": 33.01, + "learning_rate": 0.00036388888888888886, + "loss": 0.2825, + "step": 1727 + }, + { + "epoch": 33.01, + "learning_rate": 0.00036377777777777775, + "loss": 0.3155, + "step": 1728 + }, + { + "epoch": 33.01, + "learning_rate": 0.0003636666666666667, + "loss": 0.2461, + "step": 1729 + }, + { + "epoch": 33.01, + "learning_rate": 0.0003635555555555556, + "loss": 0.2769, + "step": 1730 + }, + { + "epoch": 33.01, + "learning_rate": 0.00036344444444444445, + "loss": 0.247, + "step": 1731 + }, + { + "epoch": 33.01, + "learning_rate": 0.00036333333333333335, + "loss": 0.3079, + "step": 1732 + }, + { + "epoch": 33.01, + "learning_rate": 0.00036322222222222224, + "loss": 0.2939, + "step": 1733 + }, + { + "epoch": 33.01, + "learning_rate": 0.0003631111111111111, + "loss": 0.2664, + "step": 1734 + }, + { + "epoch": 34.0, + "learning_rate": 0.000363, + "loss": 0.4253, + "step": 1735 + }, + { + "epoch": 34.0, + "learning_rate": 0.0003628888888888889, + "loss": 0.2974, + "step": 1736 + }, + { + "epoch": 34.0, + "learning_rate": 0.0003627777777777778, + "loss": 0.2869, + "step": 1737 + }, + { + "epoch": 34.0, + "learning_rate": 0.0003626666666666667, + "loss": 0.2545, + "step": 1738 + }, + { + "epoch": 34.0, + "learning_rate": 0.0003625555555555556, + "loss": 0.246, + "step": 1739 + }, + { + "epoch": 34.0, + "learning_rate": 0.0003624444444444445, + "loss": 0.2291, + "step": 1740 + }, + { + "epoch": 34.0, + "learning_rate": 0.0003623333333333333, + "loss": 0.2306, + "step": 1741 + }, + { + "epoch": 34.0, + "learning_rate": 0.0003622222222222222, + "loss": 0.2162, + "step": 1742 + }, + { + "epoch": 34.0, + "learning_rate": 0.0003621111111111111, + "loss": 0.2337, + "step": 1743 + }, + { + "epoch": 34.0, + "learning_rate": 0.000362, + "loss": 0.2457, + "step": 1744 + }, + { + "epoch": 34.0, + "learning_rate": 0.00036188888888888886, + "loss": 0.228, + "step": 1745 + }, + { + "epoch": 34.0, + "learning_rate": 0.00036177777777777776, + "loss": 0.2001, + "step": 1746 + }, + { + "epoch": 34.0, + "learning_rate": 0.0003616666666666667, + "loss": 0.2204, + "step": 1747 + }, + { + "epoch": 34.0, + "learning_rate": 0.00036155555555555556, + "loss": 0.2678, + "step": 1748 + }, + { + "epoch": 34.0, + "learning_rate": 0.00036144444444444445, + "loss": 0.2388, + "step": 1749 + }, + { + "epoch": 34.0, + "learning_rate": 0.00036133333333333335, + "loss": 0.2536, + "step": 1750 + }, + { + "epoch": 34.0, + "learning_rate": 0.00036122222222222225, + "loss": 0.3096, + "step": 1751 + }, + { + "epoch": 34.0, + "learning_rate": 0.0003611111111111111, + "loss": 0.2554, + "step": 1752 + }, + { + "epoch": 34.0, + "learning_rate": 0.000361, + "loss": 0.2678, + "step": 1753 + }, + { + "epoch": 34.0, + "learning_rate": 0.0003608888888888889, + "loss": 0.3093, + "step": 1754 + }, + { + "epoch": 34.0, + "learning_rate": 0.00036077777777777774, + "loss": 0.2985, + "step": 1755 + }, + { + "epoch": 34.0, + "learning_rate": 0.0003606666666666667, + "loss": 0.2576, + "step": 1756 + }, + { + "epoch": 34.0, + "learning_rate": 0.0003605555555555556, + "loss": 0.2887, + "step": 1757 + }, + { + "epoch": 34.0, + "learning_rate": 0.0003604444444444445, + "loss": 0.2883, + "step": 1758 + }, + { + "epoch": 34.01, + "learning_rate": 0.00036033333333333333, + "loss": 0.2378, + "step": 1759 + }, + { + "epoch": 34.01, + "learning_rate": 0.0003602222222222222, + "loss": 0.2974, + "step": 1760 + }, + { + "epoch": 34.01, + "learning_rate": 0.0003601111111111111, + "loss": 0.269, + "step": 1761 + }, + { + "epoch": 34.01, + "learning_rate": 0.00035999999999999997, + "loss": 0.2334, + "step": 1762 + }, + { + "epoch": 34.01, + "learning_rate": 0.00035988888888888887, + "loss": 0.3041, + "step": 1763 + }, + { + "epoch": 34.01, + "learning_rate": 0.00035977777777777776, + "loss": 0.2648, + "step": 1764 + }, + { + "epoch": 34.01, + "learning_rate": 0.0003596666666666667, + "loss": 0.216, + "step": 1765 + }, + { + "epoch": 34.01, + "learning_rate": 0.00035955555555555556, + "loss": 0.2639, + "step": 1766 + }, + { + "epoch": 34.01, + "learning_rate": 0.00035944444444444446, + "loss": 0.3071, + "step": 1767 + }, + { + "epoch": 34.01, + "learning_rate": 0.00035933333333333336, + "loss": 0.3064, + "step": 1768 + }, + { + "epoch": 34.01, + "learning_rate": 0.0003592222222222222, + "loss": 0.2734, + "step": 1769 + }, + { + "epoch": 34.01, + "learning_rate": 0.0003591111111111111, + "loss": 0.2434, + "step": 1770 + }, + { + "epoch": 34.01, + "learning_rate": 0.000359, + "loss": 0.2224, + "step": 1771 + }, + { + "epoch": 34.01, + "learning_rate": 0.0003588888888888889, + "loss": 0.3342, + "step": 1772 + }, + { + "epoch": 34.01, + "learning_rate": 0.00035877777777777774, + "loss": 0.3045, + "step": 1773 + }, + { + "epoch": 34.01, + "learning_rate": 0.0003586666666666667, + "loss": 0.2804, + "step": 1774 + }, + { + "epoch": 34.01, + "learning_rate": 0.0003585555555555556, + "loss": 0.2564, + "step": 1775 + }, + { + "epoch": 34.01, + "learning_rate": 0.00035844444444444444, + "loss": 0.2956, + "step": 1776 + }, + { + "epoch": 34.01, + "learning_rate": 0.00035833333333333333, + "loss": 0.2425, + "step": 1777 + }, + { + "epoch": 34.01, + "learning_rate": 0.00035822222222222223, + "loss": 0.2196, + "step": 1778 + }, + { + "epoch": 34.01, + "learning_rate": 0.00035811111111111113, + "loss": 0.2438, + "step": 1779 + }, + { + "epoch": 34.01, + "learning_rate": 0.000358, + "loss": 0.2414, + "step": 1780 + }, + { + "epoch": 34.01, + "learning_rate": 0.00035788888888888887, + "loss": 0.3138, + "step": 1781 + }, + { + "epoch": 34.01, + "learning_rate": 0.00035777777777777777, + "loss": 0.3188, + "step": 1782 + }, + { + "epoch": 34.01, + "learning_rate": 0.00035766666666666667, + "loss": 0.2352, + "step": 1783 + }, + { + "epoch": 34.01, + "learning_rate": 0.00035755555555555557, + "loss": 0.2903, + "step": 1784 + }, + { + "epoch": 34.01, + "learning_rate": 0.00035744444444444447, + "loss": 0.301, + "step": 1785 + }, + { + "epoch": 35.0, + "learning_rate": 0.00035733333333333336, + "loss": 0.4598, + "step": 1786 + }, + { + "epoch": 35.0, + "learning_rate": 0.0003572222222222222, + "loss": 0.2352, + "step": 1787 + }, + { + "epoch": 35.0, + "learning_rate": 0.0003571111111111111, + "loss": 0.2986, + "step": 1788 + }, + { + "epoch": 35.0, + "learning_rate": 0.000357, + "loss": 0.2466, + "step": 1789 + }, + { + "epoch": 35.0, + "learning_rate": 0.00035688888888888885, + "loss": 0.2983, + "step": 1790 + }, + { + "epoch": 35.0, + "learning_rate": 0.0003567777777777778, + "loss": 0.2547, + "step": 1791 + }, + { + "epoch": 35.0, + "learning_rate": 0.0003566666666666667, + "loss": 0.2193, + "step": 1792 + }, + { + "epoch": 35.0, + "learning_rate": 0.0003565555555555556, + "loss": 0.294, + "step": 1793 + }, + { + "epoch": 35.0, + "learning_rate": 0.00035644444444444444, + "loss": 0.3256, + "step": 1794 + }, + { + "epoch": 35.0, + "learning_rate": 0.00035633333333333334, + "loss": 0.2105, + "step": 1795 + }, + { + "epoch": 35.0, + "learning_rate": 0.00035622222222222224, + "loss": 0.207, + "step": 1796 + }, + { + "epoch": 35.0, + "learning_rate": 0.0003561111111111111, + "loss": 0.2676, + "step": 1797 + }, + { + "epoch": 35.0, + "learning_rate": 0.000356, + "loss": 0.2717, + "step": 1798 + }, + { + "epoch": 35.0, + "learning_rate": 0.0003558888888888889, + "loss": 0.2616, + "step": 1799 + }, + { + "epoch": 35.0, + "learning_rate": 0.00035577777777777783, + "loss": 0.2833, + "step": 1800 + }, + { + "epoch": 35.0, + "learning_rate": 0.0003556666666666667, + "loss": 0.215, + "step": 1801 + }, + { + "epoch": 35.0, + "learning_rate": 0.00035555555555555557, + "loss": 0.232, + "step": 1802 + }, + { + "epoch": 35.0, + "learning_rate": 0.00035544444444444447, + "loss": 0.3455, + "step": 1803 + }, + { + "epoch": 35.0, + "learning_rate": 0.0003553333333333333, + "loss": 0.2389, + "step": 1804 + }, + { + "epoch": 35.0, + "learning_rate": 0.0003552222222222222, + "loss": 0.262, + "step": 1805 + }, + { + "epoch": 35.0, + "learning_rate": 0.0003551111111111111, + "loss": 0.3, + "step": 1806 + }, + { + "epoch": 35.0, + "learning_rate": 0.000355, + "loss": 0.2608, + "step": 1807 + }, + { + "epoch": 35.0, + "learning_rate": 0.00035488888888888885, + "loss": 0.3314, + "step": 1808 + }, + { + "epoch": 35.0, + "learning_rate": 0.0003547777777777778, + "loss": 0.2841, + "step": 1809 + }, + { + "epoch": 35.01, + "learning_rate": 0.0003546666666666667, + "loss": 0.2789, + "step": 1810 + }, + { + "epoch": 35.01, + "learning_rate": 0.0003545555555555556, + "loss": 0.265, + "step": 1811 + }, + { + "epoch": 35.01, + "learning_rate": 0.00035444444444444445, + "loss": 0.2366, + "step": 1812 + }, + { + "epoch": 35.01, + "learning_rate": 0.00035433333333333334, + "loss": 0.2742, + "step": 1813 + }, + { + "epoch": 35.01, + "learning_rate": 0.00035422222222222224, + "loss": 0.2246, + "step": 1814 + }, + { + "epoch": 35.01, + "learning_rate": 0.0003541111111111111, + "loss": 0.2528, + "step": 1815 + }, + { + "epoch": 35.01, + "learning_rate": 0.000354, + "loss": 0.2504, + "step": 1816 + }, + { + "epoch": 35.01, + "learning_rate": 0.0003538888888888889, + "loss": 0.2477, + "step": 1817 + }, + { + "epoch": 35.01, + "learning_rate": 0.00035377777777777784, + "loss": 0.2592, + "step": 1818 + }, + { + "epoch": 35.01, + "learning_rate": 0.0003536666666666667, + "loss": 0.269, + "step": 1819 + }, + { + "epoch": 35.01, + "learning_rate": 0.0003535555555555556, + "loss": 0.2633, + "step": 1820 + }, + { + "epoch": 35.01, + "learning_rate": 0.0003534444444444445, + "loss": 0.2618, + "step": 1821 + }, + { + "epoch": 35.01, + "learning_rate": 0.0003533333333333333, + "loss": 0.2436, + "step": 1822 + }, + { + "epoch": 35.01, + "learning_rate": 0.0003532222222222222, + "loss": 0.2968, + "step": 1823 + }, + { + "epoch": 35.01, + "learning_rate": 0.0003531111111111111, + "loss": 0.3047, + "step": 1824 + }, + { + "epoch": 35.01, + "learning_rate": 0.00035299999999999996, + "loss": 0.2651, + "step": 1825 + }, + { + "epoch": 35.01, + "learning_rate": 0.00035288888888888886, + "loss": 0.2806, + "step": 1826 + }, + { + "epoch": 35.01, + "learning_rate": 0.0003527777777777778, + "loss": 0.2672, + "step": 1827 + }, + { + "epoch": 35.01, + "learning_rate": 0.0003526666666666667, + "loss": 0.262, + "step": 1828 + }, + { + "epoch": 35.01, + "learning_rate": 0.00035255555555555555, + "loss": 0.2482, + "step": 1829 + }, + { + "epoch": 35.01, + "learning_rate": 0.00035244444444444445, + "loss": 0.2571, + "step": 1830 + }, + { + "epoch": 35.01, + "learning_rate": 0.00035233333333333335, + "loss": 0.2545, + "step": 1831 + }, + { + "epoch": 35.01, + "learning_rate": 0.00035222222222222225, + "loss": 0.2792, + "step": 1832 + }, + { + "epoch": 35.01, + "learning_rate": 0.0003521111111111111, + "loss": 0.2851, + "step": 1833 + }, + { + "epoch": 35.01, + "learning_rate": 0.000352, + "loss": 0.2992, + "step": 1834 + }, + { + "epoch": 35.01, + "learning_rate": 0.0003518888888888889, + "loss": 0.1938, + "step": 1835 + }, + { + "epoch": 35.01, + "learning_rate": 0.0003517777777777778, + "loss": 0.3246, + "step": 1836 + }, + { + "epoch": 36.0, + "learning_rate": 0.0003516666666666667, + "loss": 0.3748, + "step": 1837 + }, + { + "epoch": 36.0, + "learning_rate": 0.0003515555555555556, + "loss": 0.2467, + "step": 1838 + }, + { + "epoch": 36.0, + "learning_rate": 0.0003514444444444445, + "loss": 0.2752, + "step": 1839 + }, + { + "epoch": 36.0, + "learning_rate": 0.0003513333333333333, + "loss": 0.2282, + "step": 1840 + }, + { + "epoch": 36.0, + "learning_rate": 0.0003512222222222222, + "loss": 0.249, + "step": 1841 + }, + { + "epoch": 36.0, + "learning_rate": 0.0003511111111111111, + "loss": 0.22, + "step": 1842 + }, + { + "epoch": 36.0, + "learning_rate": 0.00035099999999999997, + "loss": 0.2437, + "step": 1843 + }, + { + "epoch": 36.0, + "learning_rate": 0.00035088888888888886, + "loss": 0.2794, + "step": 1844 + }, + { + "epoch": 36.0, + "learning_rate": 0.0003507777777777778, + "loss": 0.2654, + "step": 1845 + }, + { + "epoch": 36.0, + "learning_rate": 0.0003506666666666667, + "loss": 0.2849, + "step": 1846 + }, + { + "epoch": 36.0, + "learning_rate": 0.00035055555555555556, + "loss": 0.2537, + "step": 1847 + }, + { + "epoch": 36.0, + "learning_rate": 0.00035044444444444446, + "loss": 0.2767, + "step": 1848 + }, + { + "epoch": 36.0, + "learning_rate": 0.00035033333333333336, + "loss": 0.2463, + "step": 1849 + }, + { + "epoch": 36.0, + "learning_rate": 0.0003502222222222222, + "loss": 0.238, + "step": 1850 + }, + { + "epoch": 36.0, + "learning_rate": 0.0003501111111111111, + "loss": 0.2748, + "step": 1851 + }, + { + "epoch": 36.0, + "learning_rate": 0.00035, + "loss": 0.271, + "step": 1852 + }, + { + "epoch": 36.0, + "learning_rate": 0.00034988888888888895, + "loss": 0.2563, + "step": 1853 + }, + { + "epoch": 36.0, + "learning_rate": 0.0003497777777777778, + "loss": 0.2796, + "step": 1854 + }, + { + "epoch": 36.0, + "learning_rate": 0.0003496666666666667, + "loss": 0.2228, + "step": 1855 + }, + { + "epoch": 36.0, + "learning_rate": 0.0003495555555555556, + "loss": 0.258, + "step": 1856 + }, + { + "epoch": 36.0, + "learning_rate": 0.00034944444444444443, + "loss": 0.2995, + "step": 1857 + }, + { + "epoch": 36.0, + "learning_rate": 0.00034933333333333333, + "loss": 0.2602, + "step": 1858 + }, + { + "epoch": 36.0, + "learning_rate": 0.00034922222222222223, + "loss": 0.2229, + "step": 1859 + }, + { + "epoch": 36.0, + "learning_rate": 0.00034911111111111113, + "loss": 0.2856, + "step": 1860 + }, + { + "epoch": 36.01, + "learning_rate": 0.00034899999999999997, + "loss": 0.2932, + "step": 1861 + }, + { + "epoch": 36.01, + "learning_rate": 0.0003488888888888889, + "loss": 0.2503, + "step": 1862 + }, + { + "epoch": 36.01, + "learning_rate": 0.0003487777777777778, + "loss": 0.2494, + "step": 1863 + }, + { + "epoch": 36.01, + "learning_rate": 0.00034866666666666667, + "loss": 0.2242, + "step": 1864 + }, + { + "epoch": 36.01, + "learning_rate": 0.00034855555555555557, + "loss": 0.2549, + "step": 1865 + }, + { + "epoch": 36.01, + "learning_rate": 0.00034844444444444446, + "loss": 0.2627, + "step": 1866 + }, + { + "epoch": 36.01, + "learning_rate": 0.00034833333333333336, + "loss": 0.2807, + "step": 1867 + }, + { + "epoch": 36.01, + "learning_rate": 0.0003482222222222222, + "loss": 0.2413, + "step": 1868 + }, + { + "epoch": 36.01, + "learning_rate": 0.0003481111111111111, + "loss": 0.2538, + "step": 1869 + }, + { + "epoch": 36.01, + "learning_rate": 0.000348, + "loss": 0.3039, + "step": 1870 + }, + { + "epoch": 36.01, + "learning_rate": 0.0003478888888888889, + "loss": 0.2572, + "step": 1871 + }, + { + "epoch": 36.01, + "learning_rate": 0.0003477777777777778, + "loss": 0.2684, + "step": 1872 + }, + { + "epoch": 36.01, + "learning_rate": 0.0003476666666666667, + "loss": 0.2943, + "step": 1873 + }, + { + "epoch": 36.01, + "learning_rate": 0.0003475555555555556, + "loss": 0.3052, + "step": 1874 + }, + { + "epoch": 36.01, + "learning_rate": 0.00034744444444444444, + "loss": 0.3038, + "step": 1875 + }, + { + "epoch": 36.01, + "learning_rate": 0.00034733333333333334, + "loss": 0.2628, + "step": 1876 + }, + { + "epoch": 36.01, + "learning_rate": 0.00034722222222222224, + "loss": 0.2899, + "step": 1877 + }, + { + "epoch": 36.01, + "learning_rate": 0.0003471111111111111, + "loss": 0.2493, + "step": 1878 + }, + { + "epoch": 36.01, + "learning_rate": 0.000347, + "loss": 0.2198, + "step": 1879 + }, + { + "epoch": 36.01, + "learning_rate": 0.00034688888888888893, + "loss": 0.2253, + "step": 1880 + }, + { + "epoch": 36.01, + "learning_rate": 0.00034677777777777783, + "loss": 0.2645, + "step": 1881 + }, + { + "epoch": 36.01, + "learning_rate": 0.00034666666666666667, + "loss": 0.2826, + "step": 1882 + }, + { + "epoch": 36.01, + "learning_rate": 0.00034655555555555557, + "loss": 0.2711, + "step": 1883 + }, + { + "epoch": 36.01, + "learning_rate": 0.00034644444444444447, + "loss": 0.2213, + "step": 1884 + }, + { + "epoch": 36.01, + "learning_rate": 0.0003463333333333333, + "loss": 0.3046, + "step": 1885 + }, + { + "epoch": 36.01, + "learning_rate": 0.0003462222222222222, + "loss": 0.2738, + "step": 1886 + }, + { + "epoch": 36.01, + "learning_rate": 0.0003461111111111111, + "loss": 0.2823, + "step": 1887 + }, + { + "epoch": 37.0, + "learning_rate": 0.000346, + "loss": 0.3858, + "step": 1888 + }, + { + "epoch": 37.0, + "learning_rate": 0.0003458888888888889, + "loss": 0.2639, + "step": 1889 + }, + { + "epoch": 37.0, + "learning_rate": 0.0003457777777777778, + "loss": 0.244, + "step": 1890 + }, + { + "epoch": 37.0, + "learning_rate": 0.0003456666666666667, + "loss": 0.2248, + "step": 1891 + }, + { + "epoch": 37.0, + "learning_rate": 0.00034555555555555555, + "loss": 0.2391, + "step": 1892 + }, + { + "epoch": 37.0, + "learning_rate": 0.00034544444444444444, + "loss": 0.2631, + "step": 1893 + }, + { + "epoch": 37.0, + "learning_rate": 0.00034533333333333334, + "loss": 0.2032, + "step": 1894 + }, + { + "epoch": 37.0, + "learning_rate": 0.00034522222222222224, + "loss": 0.297, + "step": 1895 + }, + { + "epoch": 37.0, + "learning_rate": 0.0003451111111111111, + "loss": 0.248, + "step": 1896 + }, + { + "epoch": 37.0, + "learning_rate": 0.000345, + "loss": 0.2908, + "step": 1897 + }, + { + "epoch": 37.0, + "learning_rate": 0.00034488888888888894, + "loss": 0.3049, + "step": 1898 + }, + { + "epoch": 37.0, + "learning_rate": 0.0003447777777777778, + "loss": 0.2103, + "step": 1899 + }, + { + "epoch": 37.0, + "learning_rate": 0.0003446666666666667, + "loss": 0.281, + "step": 1900 + }, + { + "epoch": 37.0, + "learning_rate": 0.0003445555555555556, + "loss": 0.2244, + "step": 1901 + }, + { + "epoch": 37.0, + "learning_rate": 0.0003444444444444445, + "loss": 0.2975, + "step": 1902 + }, + { + "epoch": 37.0, + "learning_rate": 0.0003443333333333333, + "loss": 0.2805, + "step": 1903 + }, + { + "epoch": 37.0, + "learning_rate": 0.0003442222222222222, + "loss": 0.2801, + "step": 1904 + }, + { + "epoch": 37.0, + "learning_rate": 0.0003441111111111111, + "loss": 0.2772, + "step": 1905 + }, + { + "epoch": 37.0, + "learning_rate": 0.00034399999999999996, + "loss": 0.2331, + "step": 1906 + }, + { + "epoch": 37.0, + "learning_rate": 0.0003438888888888889, + "loss": 0.2366, + "step": 1907 + }, + { + "epoch": 37.0, + "learning_rate": 0.0003437777777777778, + "loss": 0.2668, + "step": 1908 + }, + { + "epoch": 37.0, + "learning_rate": 0.0003436666666666667, + "loss": 0.2734, + "step": 1909 + }, + { + "epoch": 37.0, + "learning_rate": 0.00034355555555555555, + "loss": 0.2585, + "step": 1910 + }, + { + "epoch": 37.0, + "learning_rate": 0.00034344444444444445, + "loss": 0.3503, + "step": 1911 + }, + { + "epoch": 37.01, + "learning_rate": 0.00034333333333333335, + "loss": 0.2165, + "step": 1912 + }, + { + "epoch": 37.01, + "learning_rate": 0.0003432222222222222, + "loss": 0.2649, + "step": 1913 + }, + { + "epoch": 37.01, + "learning_rate": 0.0003431111111111111, + "loss": 0.2439, + "step": 1914 + }, + { + "epoch": 37.01, + "learning_rate": 0.00034300000000000004, + "loss": 0.2154, + "step": 1915 + }, + { + "epoch": 37.01, + "learning_rate": 0.00034288888888888894, + "loss": 0.2057, + "step": 1916 + }, + { + "epoch": 37.01, + "learning_rate": 0.0003427777777777778, + "loss": 0.2079, + "step": 1917 + }, + { + "epoch": 37.01, + "learning_rate": 0.0003426666666666667, + "loss": 0.2063, + "step": 1918 + }, + { + "epoch": 37.01, + "learning_rate": 0.0003425555555555556, + "loss": 0.22, + "step": 1919 + }, + { + "epoch": 37.01, + "learning_rate": 0.0003424444444444444, + "loss": 0.2446, + "step": 1920 + }, + { + "epoch": 37.01, + "learning_rate": 0.0003423333333333333, + "loss": 0.2281, + "step": 1921 + }, + { + "epoch": 37.01, + "learning_rate": 0.0003422222222222222, + "loss": 0.257, + "step": 1922 + }, + { + "epoch": 37.01, + "learning_rate": 0.0003421111111111111, + "loss": 0.2637, + "step": 1923 + }, + { + "epoch": 37.01, + "learning_rate": 0.000342, + "loss": 0.2356, + "step": 1924 + }, + { + "epoch": 37.01, + "learning_rate": 0.0003418888888888889, + "loss": 0.1976, + "step": 1925 + }, + { + "epoch": 37.01, + "learning_rate": 0.0003417777777777778, + "loss": 0.2164, + "step": 1926 + }, + { + "epoch": 37.01, + "learning_rate": 0.00034166666666666666, + "loss": 0.2555, + "step": 1927 + }, + { + "epoch": 37.01, + "learning_rate": 0.00034155555555555556, + "loss": 0.2266, + "step": 1928 + }, + { + "epoch": 37.01, + "learning_rate": 0.00034144444444444446, + "loss": 0.2377, + "step": 1929 + }, + { + "epoch": 37.01, + "learning_rate": 0.00034133333333333335, + "loss": 0.2275, + "step": 1930 + }, + { + "epoch": 37.01, + "learning_rate": 0.0003412222222222222, + "loss": 0.2957, + "step": 1931 + }, + { + "epoch": 37.01, + "learning_rate": 0.0003411111111111111, + "loss": 0.2526, + "step": 1932 + }, + { + "epoch": 37.01, + "learning_rate": 0.00034100000000000005, + "loss": 0.2862, + "step": 1933 + }, + { + "epoch": 37.01, + "learning_rate": 0.0003408888888888889, + "loss": 0.2977, + "step": 1934 + }, + { + "epoch": 37.01, + "learning_rate": 0.0003407777777777778, + "loss": 0.2194, + "step": 1935 + }, + { + "epoch": 37.01, + "learning_rate": 0.0003406666666666667, + "loss": 0.2134, + "step": 1936 + }, + { + "epoch": 37.01, + "learning_rate": 0.0003405555555555556, + "loss": 0.2169, + "step": 1937 + }, + { + "epoch": 37.01, + "learning_rate": 0.00034044444444444443, + "loss": 0.2045, + "step": 1938 + }, + { + "epoch": 38.0, + "learning_rate": 0.00034033333333333333, + "loss": 0.336, + "step": 1939 + }, + { + "epoch": 38.0, + "learning_rate": 0.00034022222222222223, + "loss": 0.234, + "step": 1940 + }, + { + "epoch": 38.0, + "learning_rate": 0.00034011111111111107, + "loss": 0.2171, + "step": 1941 + }, + { + "epoch": 38.0, + "learning_rate": 0.00034, + "loss": 0.2653, + "step": 1942 + }, + { + "epoch": 38.0, + "learning_rate": 0.0003398888888888889, + "loss": 0.2483, + "step": 1943 + }, + { + "epoch": 38.0, + "learning_rate": 0.0003397777777777778, + "loss": 0.2205, + "step": 1944 + }, + { + "epoch": 38.0, + "learning_rate": 0.00033966666666666666, + "loss": 0.2958, + "step": 1945 + }, + { + "epoch": 38.0, + "learning_rate": 0.00033955555555555556, + "loss": 0.2266, + "step": 1946 + }, + { + "epoch": 38.0, + "learning_rate": 0.00033944444444444446, + "loss": 0.2863, + "step": 1947 + }, + { + "epoch": 38.0, + "learning_rate": 0.0003393333333333333, + "loss": 0.2369, + "step": 1948 + }, + { + "epoch": 38.0, + "learning_rate": 0.0003392222222222222, + "loss": 0.2119, + "step": 1949 + }, + { + "epoch": 38.0, + "learning_rate": 0.0003391111111111111, + "loss": 0.2206, + "step": 1950 + }, + { + "epoch": 38.0, + "learning_rate": 0.00033900000000000005, + "loss": 0.2527, + "step": 1951 + }, + { + "epoch": 38.0, + "learning_rate": 0.0003388888888888889, + "loss": 0.2222, + "step": 1952 + }, + { + "epoch": 38.0, + "learning_rate": 0.0003387777777777778, + "loss": 0.2958, + "step": 1953 + }, + { + "epoch": 38.0, + "learning_rate": 0.0003386666666666667, + "loss": 0.2511, + "step": 1954 + }, + { + "epoch": 38.0, + "learning_rate": 0.00033855555555555554, + "loss": 0.2571, + "step": 1955 + }, + { + "epoch": 38.0, + "learning_rate": 0.00033844444444444444, + "loss": 0.2217, + "step": 1956 + }, + { + "epoch": 38.0, + "learning_rate": 0.00033833333333333334, + "loss": 0.2646, + "step": 1957 + }, + { + "epoch": 38.0, + "learning_rate": 0.00033822222222222223, + "loss": 0.2536, + "step": 1958 + }, + { + "epoch": 38.0, + "learning_rate": 0.0003381111111111111, + "loss": 0.2224, + "step": 1959 + }, + { + "epoch": 38.0, + "learning_rate": 0.00033800000000000003, + "loss": 0.28, + "step": 1960 + }, + { + "epoch": 38.0, + "learning_rate": 0.00033788888888888893, + "loss": 0.2726, + "step": 1961 + }, + { + "epoch": 38.0, + "learning_rate": 0.00033777777777777777, + "loss": 0.2293, + "step": 1962 + }, + { + "epoch": 38.01, + "learning_rate": 0.00033766666666666667, + "loss": 0.2514, + "step": 1963 + }, + { + "epoch": 38.01, + "learning_rate": 0.00033755555555555557, + "loss": 0.2918, + "step": 1964 + }, + { + "epoch": 38.01, + "learning_rate": 0.00033744444444444447, + "loss": 0.2624, + "step": 1965 + }, + { + "epoch": 38.01, + "learning_rate": 0.0003373333333333333, + "loss": 0.2613, + "step": 1966 + }, + { + "epoch": 38.01, + "learning_rate": 0.0003372222222222222, + "loss": 0.2023, + "step": 1967 + }, + { + "epoch": 38.01, + "learning_rate": 0.0003371111111111111, + "loss": 0.2672, + "step": 1968 + }, + { + "epoch": 38.01, + "learning_rate": 0.000337, + "loss": 0.2383, + "step": 1969 + }, + { + "epoch": 38.01, + "learning_rate": 0.0003368888888888889, + "loss": 0.2577, + "step": 1970 + }, + { + "epoch": 38.01, + "learning_rate": 0.0003367777777777778, + "loss": 0.2693, + "step": 1971 + }, + { + "epoch": 38.01, + "learning_rate": 0.0003366666666666667, + "loss": 0.2129, + "step": 1972 + }, + { + "epoch": 38.01, + "learning_rate": 0.00033655555555555554, + "loss": 0.2282, + "step": 1973 + }, + { + "epoch": 38.01, + "learning_rate": 0.00033644444444444444, + "loss": 0.2231, + "step": 1974 + }, + { + "epoch": 38.01, + "learning_rate": 0.00033633333333333334, + "loss": 0.2255, + "step": 1975 + }, + { + "epoch": 38.01, + "learning_rate": 0.0003362222222222222, + "loss": 0.2325, + "step": 1976 + }, + { + "epoch": 38.01, + "learning_rate": 0.00033611111111111114, + "loss": 0.2754, + "step": 1977 + }, + { + "epoch": 38.01, + "learning_rate": 0.00033600000000000004, + "loss": 0.2522, + "step": 1978 + }, + { + "epoch": 38.01, + "learning_rate": 0.00033588888888888893, + "loss": 0.2191, + "step": 1979 + }, + { + "epoch": 38.01, + "learning_rate": 0.0003357777777777778, + "loss": 0.2775, + "step": 1980 + }, + { + "epoch": 38.01, + "learning_rate": 0.0003356666666666667, + "loss": 0.2815, + "step": 1981 + }, + { + "epoch": 38.01, + "learning_rate": 0.0003355555555555556, + "loss": 0.217, + "step": 1982 + }, + { + "epoch": 38.01, + "learning_rate": 0.0003354444444444444, + "loss": 0.2379, + "step": 1983 + }, + { + "epoch": 38.01, + "learning_rate": 0.0003353333333333333, + "loss": 0.2972, + "step": 1984 + }, + { + "epoch": 38.01, + "learning_rate": 0.0003352222222222222, + "loss": 0.2451, + "step": 1985 + }, + { + "epoch": 38.01, + "learning_rate": 0.00033511111111111117, + "loss": 0.2307, + "step": 1986 + }, + { + "epoch": 38.01, + "learning_rate": 0.000335, + "loss": 0.2258, + "step": 1987 + }, + { + "epoch": 38.01, + "learning_rate": 0.0003348888888888889, + "loss": 0.2145, + "step": 1988 + }, + { + "epoch": 38.01, + "learning_rate": 0.0003347777777777778, + "loss": 0.2586, + "step": 1989 + }, + { + "epoch": 39.0, + "learning_rate": 0.00033466666666666665, + "loss": 0.3355, + "step": 1990 + }, + { + "epoch": 39.0, + "learning_rate": 0.00033455555555555555, + "loss": 0.2299, + "step": 1991 + }, + { + "epoch": 39.0, + "learning_rate": 0.00033444444444444445, + "loss": 0.2316, + "step": 1992 + }, + { + "epoch": 39.0, + "learning_rate": 0.00033433333333333335, + "loss": 0.2387, + "step": 1993 + }, + { + "epoch": 39.0, + "learning_rate": 0.0003342222222222222, + "loss": 0.1774, + "step": 1994 + }, + { + "epoch": 39.0, + "learning_rate": 0.00033411111111111114, + "loss": 0.1815, + "step": 1995 + }, + { + "epoch": 39.0, + "learning_rate": 0.00033400000000000004, + "loss": 0.2222, + "step": 1996 + }, + { + "epoch": 39.0, + "learning_rate": 0.0003338888888888889, + "loss": 0.252, + "step": 1997 + }, + { + "epoch": 39.0, + "learning_rate": 0.0003337777777777778, + "loss": 0.2328, + "step": 1998 + }, + { + "epoch": 39.0, + "learning_rate": 0.0003336666666666667, + "loss": 0.2238, + "step": 1999 + }, + { + "epoch": 39.0, + "learning_rate": 0.0003335555555555556, + "loss": 0.2334, + "step": 2000 + }, + { + "epoch": 39.0, + "eval_cer": 0.08415130476418482, + "eval_loss": 0.3666168451309204, + "eval_runtime": 49.9921, + "eval_samples_per_second": 8.321, + "eval_steps_per_second": 0.26, + "eval_wer": 0.3671300081766149, + "step": 2000 + }, + { + "epoch": 39.0, + "learning_rate": 0.0003334444444444444, + "loss": 0.2276, + "step": 2001 + }, + { + "epoch": 39.0, + "learning_rate": 0.0003333333333333333, + "loss": 0.1751, + "step": 2002 + }, + { + "epoch": 39.0, + "learning_rate": 0.0003332222222222222, + "loss": 0.3156, + "step": 2003 + }, + { + "epoch": 39.0, + "learning_rate": 0.0003331111111111111, + "loss": 0.2238, + "step": 2004 + }, + { + "epoch": 39.0, + "learning_rate": 0.000333, + "loss": 0.2501, + "step": 2005 + }, + { + "epoch": 39.0, + "learning_rate": 0.0003328888888888889, + "loss": 0.2143, + "step": 2006 + }, + { + "epoch": 39.0, + "learning_rate": 0.0003327777777777778, + "loss": 0.2249, + "step": 2007 + }, + { + "epoch": 39.0, + "learning_rate": 0.00033266666666666666, + "loss": 0.2485, + "step": 2008 + }, + { + "epoch": 39.0, + "learning_rate": 0.00033255555555555556, + "loss": 0.2393, + "step": 2009 + }, + { + "epoch": 39.0, + "learning_rate": 0.00033244444444444445, + "loss": 0.2363, + "step": 2010 + }, + { + "epoch": 39.0, + "learning_rate": 0.0003323333333333333, + "loss": 0.2535, + "step": 2011 + }, + { + "epoch": 39.0, + "learning_rate": 0.0003322222222222222, + "loss": 0.2891, + "step": 2012 + }, + { + "epoch": 39.0, + "learning_rate": 0.00033211111111111115, + "loss": 0.2383, + "step": 2013 + }, + { + "epoch": 39.01, + "learning_rate": 0.00033200000000000005, + "loss": 0.2436, + "step": 2014 + }, + { + "epoch": 39.01, + "learning_rate": 0.0003318888888888889, + "loss": 0.285, + "step": 2015 + }, + { + "epoch": 39.01, + "learning_rate": 0.0003317777777777778, + "loss": 0.2363, + "step": 2016 + }, + { + "epoch": 39.01, + "learning_rate": 0.0003316666666666667, + "loss": 0.2128, + "step": 2017 + }, + { + "epoch": 39.01, + "learning_rate": 0.00033155555555555553, + "loss": 0.2635, + "step": 2018 + }, + { + "epoch": 39.01, + "learning_rate": 0.00033144444444444443, + "loss": 0.3404, + "step": 2019 + }, + { + "epoch": 39.01, + "learning_rate": 0.00033133333333333333, + "loss": 0.2342, + "step": 2020 + }, + { + "epoch": 39.01, + "learning_rate": 0.0003312222222222222, + "loss": 0.2161, + "step": 2021 + }, + { + "epoch": 39.01, + "learning_rate": 0.0003311111111111111, + "loss": 0.2465, + "step": 2022 + }, + { + "epoch": 39.01, + "learning_rate": 0.000331, + "loss": 0.2596, + "step": 2023 + }, + { + "epoch": 39.01, + "learning_rate": 0.0003308888888888889, + "loss": 0.2269, + "step": 2024 + }, + { + "epoch": 39.01, + "learning_rate": 0.00033077777777777776, + "loss": 0.2512, + "step": 2025 + }, + { + "epoch": 39.01, + "learning_rate": 0.00033066666666666666, + "loss": 0.2541, + "step": 2026 + }, + { + "epoch": 39.01, + "learning_rate": 0.00033055555555555556, + "loss": 0.287, + "step": 2027 + }, + { + "epoch": 39.01, + "learning_rate": 0.00033044444444444446, + "loss": 0.3296, + "step": 2028 + }, + { + "epoch": 39.01, + "learning_rate": 0.0003303333333333333, + "loss": 0.241, + "step": 2029 + }, + { + "epoch": 39.01, + "learning_rate": 0.0003302222222222222, + "loss": 0.2501, + "step": 2030 + }, + { + "epoch": 39.01, + "learning_rate": 0.00033011111111111115, + "loss": 0.2419, + "step": 2031 + }, + { + "epoch": 39.01, + "learning_rate": 0.00033, + "loss": 0.2342, + "step": 2032 + }, + { + "epoch": 39.01, + "learning_rate": 0.0003298888888888889, + "loss": 0.2364, + "step": 2033 + }, + { + "epoch": 39.01, + "learning_rate": 0.0003297777777777778, + "loss": 0.2078, + "step": 2034 + }, + { + "epoch": 39.01, + "learning_rate": 0.0003296666666666667, + "loss": 0.2478, + "step": 2035 + }, + { + "epoch": 39.01, + "learning_rate": 0.00032955555555555554, + "loss": 0.2132, + "step": 2036 + }, + { + "epoch": 39.01, + "learning_rate": 0.00032944444444444444, + "loss": 0.2108, + "step": 2037 + }, + { + "epoch": 39.01, + "learning_rate": 0.00032933333333333333, + "loss": 0.264, + "step": 2038 + }, + { + "epoch": 39.01, + "learning_rate": 0.00032922222222222223, + "loss": 0.2106, + "step": 2039 + }, + { + "epoch": 39.01, + "learning_rate": 0.00032911111111111113, + "loss": 0.2524, + "step": 2040 + }, + { + "epoch": 40.0, + "learning_rate": 0.00032900000000000003, + "loss": 0.3488, + "step": 2041 + }, + { + "epoch": 40.0, + "learning_rate": 0.0003288888888888889, + "loss": 0.2352, + "step": 2042 + }, + { + "epoch": 40.0, + "learning_rate": 0.00032877777777777777, + "loss": 0.2214, + "step": 2043 + }, + { + "epoch": 40.0, + "learning_rate": 0.00032866666666666667, + "loss": 0.2008, + "step": 2044 + }, + { + "epoch": 40.0, + "learning_rate": 0.00032855555555555557, + "loss": 0.224, + "step": 2045 + }, + { + "epoch": 40.0, + "learning_rate": 0.0003284444444444444, + "loss": 0.1842, + "step": 2046 + }, + { + "epoch": 40.0, + "learning_rate": 0.0003283333333333333, + "loss": 0.2589, + "step": 2047 + }, + { + "epoch": 40.0, + "learning_rate": 0.00032822222222222226, + "loss": 0.2389, + "step": 2048 + }, + { + "epoch": 40.0, + "learning_rate": 0.00032811111111111116, + "loss": 0.2564, + "step": 2049 + }, + { + "epoch": 40.0, + "learning_rate": 0.000328, + "loss": 0.202, + "step": 2050 + }, + { + "epoch": 40.0, + "learning_rate": 0.0003278888888888889, + "loss": 0.2214, + "step": 2051 + }, + { + "epoch": 40.0, + "learning_rate": 0.0003277777777777778, + "loss": 0.2659, + "step": 2052 + }, + { + "epoch": 40.0, + "learning_rate": 0.00032766666666666664, + "loss": 0.2569, + "step": 2053 + }, + { + "epoch": 40.0, + "learning_rate": 0.00032755555555555554, + "loss": 0.2839, + "step": 2054 + }, + { + "epoch": 40.0, + "learning_rate": 0.00032744444444444444, + "loss": 0.2157, + "step": 2055 + }, + { + "epoch": 40.0, + "learning_rate": 0.00032733333333333334, + "loss": 0.2518, + "step": 2056 + }, + { + "epoch": 40.0, + "learning_rate": 0.00032722222222222224, + "loss": 0.2222, + "step": 2057 + }, + { + "epoch": 40.0, + "learning_rate": 0.00032711111111111114, + "loss": 0.1849, + "step": 2058 + }, + { + "epoch": 40.0, + "learning_rate": 0.00032700000000000003, + "loss": 0.207, + "step": 2059 + }, + { + "epoch": 40.0, + "learning_rate": 0.0003268888888888889, + "loss": 0.2272, + "step": 2060 + }, + { + "epoch": 40.0, + "learning_rate": 0.0003267777777777778, + "loss": 0.2041, + "step": 2061 + }, + { + "epoch": 40.0, + "learning_rate": 0.0003266666666666667, + "loss": 0.1874, + "step": 2062 + }, + { + "epoch": 40.0, + "learning_rate": 0.00032655555555555557, + "loss": 0.2343, + "step": 2063 + }, + { + "epoch": 40.0, + "learning_rate": 0.0003264444444444444, + "loss": 0.2457, + "step": 2064 + }, + { + "epoch": 40.01, + "learning_rate": 0.0003263333333333333, + "loss": 0.2768, + "step": 2065 + }, + { + "epoch": 40.01, + "learning_rate": 0.00032622222222222227, + "loss": 0.2174, + "step": 2066 + }, + { + "epoch": 40.01, + "learning_rate": 0.0003261111111111111, + "loss": 0.2358, + "step": 2067 + }, + { + "epoch": 40.01, + "learning_rate": 0.000326, + "loss": 0.2149, + "step": 2068 + }, + { + "epoch": 40.01, + "learning_rate": 0.0003258888888888889, + "loss": 0.2386, + "step": 2069 + }, + { + "epoch": 40.01, + "learning_rate": 0.0003257777777777778, + "loss": 0.1967, + "step": 2070 + }, + { + "epoch": 40.01, + "learning_rate": 0.00032566666666666665, + "loss": 0.2174, + "step": 2071 + }, + { + "epoch": 40.01, + "learning_rate": 0.00032555555555555555, + "loss": 0.2497, + "step": 2072 + }, + { + "epoch": 40.01, + "learning_rate": 0.00032544444444444445, + "loss": 0.2539, + "step": 2073 + }, + { + "epoch": 40.01, + "learning_rate": 0.0003253333333333333, + "loss": 0.2169, + "step": 2074 + }, + { + "epoch": 40.01, + "learning_rate": 0.00032522222222222224, + "loss": 0.1507, + "step": 2075 + }, + { + "epoch": 40.01, + "learning_rate": 0.00032511111111111114, + "loss": 0.2536, + "step": 2076 + }, + { + "epoch": 40.01, + "learning_rate": 0.00032500000000000004, + "loss": 0.2063, + "step": 2077 + }, + { + "epoch": 40.01, + "learning_rate": 0.0003248888888888889, + "loss": 0.2397, + "step": 2078 + }, + { + "epoch": 40.01, + "learning_rate": 0.0003247777777777778, + "loss": 0.2443, + "step": 2079 + }, + { + "epoch": 40.01, + "learning_rate": 0.0003246666666666667, + "loss": 0.2366, + "step": 2080 + }, + { + "epoch": 40.01, + "learning_rate": 0.0003245555555555555, + "loss": 0.23, + "step": 2081 + }, + { + "epoch": 40.01, + "learning_rate": 0.0003244444444444444, + "loss": 0.2085, + "step": 2082 + }, + { + "epoch": 40.01, + "learning_rate": 0.0003243333333333333, + "loss": 0.2389, + "step": 2083 + }, + { + "epoch": 40.01, + "learning_rate": 0.00032422222222222227, + "loss": 0.2438, + "step": 2084 + }, + { + "epoch": 40.01, + "learning_rate": 0.0003241111111111111, + "loss": 0.2254, + "step": 2085 + }, + { + "epoch": 40.01, + "learning_rate": 0.000324, + "loss": 0.2254, + "step": 2086 + }, + { + "epoch": 40.01, + "learning_rate": 0.0003238888888888889, + "loss": 0.2306, + "step": 2087 + }, + { + "epoch": 40.01, + "learning_rate": 0.00032377777777777776, + "loss": 0.2441, + "step": 2088 + }, + { + "epoch": 40.01, + "learning_rate": 0.00032366666666666666, + "loss": 0.2552, + "step": 2089 + }, + { + "epoch": 40.01, + "learning_rate": 0.00032355555555555555, + "loss": 0.2589, + "step": 2090 + }, + { + "epoch": 40.01, + "learning_rate": 0.00032344444444444445, + "loss": 0.2482, + "step": 2091 + }, + { + "epoch": 41.0, + "learning_rate": 0.0003233333333333333, + "loss": 0.3307, + "step": 2092 + }, + { + "epoch": 41.0, + "learning_rate": 0.00032322222222222225, + "loss": 0.2017, + "step": 2093 + }, + { + "epoch": 41.0, + "learning_rate": 0.00032311111111111115, + "loss": 0.2083, + "step": 2094 + }, + { + "epoch": 41.0, + "learning_rate": 0.000323, + "loss": 0.2186, + "step": 2095 + }, + { + "epoch": 41.0, + "learning_rate": 0.0003228888888888889, + "loss": 0.2304, + "step": 2096 + }, + { + "epoch": 41.0, + "learning_rate": 0.0003227777777777778, + "loss": 0.2143, + "step": 2097 + }, + { + "epoch": 41.0, + "learning_rate": 0.0003226666666666667, + "loss": 0.2302, + "step": 2098 + }, + { + "epoch": 41.0, + "learning_rate": 0.00032255555555555553, + "loss": 0.179, + "step": 2099 + }, + { + "epoch": 41.0, + "learning_rate": 0.00032244444444444443, + "loss": 0.2266, + "step": 2100 + }, + { + "epoch": 41.0, + "learning_rate": 0.0003223333333333334, + "loss": 0.2145, + "step": 2101 + }, + { + "epoch": 41.0, + "learning_rate": 0.0003222222222222222, + "loss": 0.1731, + "step": 2102 + }, + { + "epoch": 41.0, + "learning_rate": 0.0003221111111111111, + "loss": 0.1999, + "step": 2103 + }, + { + "epoch": 41.0, + "learning_rate": 0.000322, + "loss": 0.2582, + "step": 2104 + }, + { + "epoch": 41.0, + "learning_rate": 0.0003218888888888889, + "loss": 0.2071, + "step": 2105 + }, + { + "epoch": 41.0, + "learning_rate": 0.00032177777777777776, + "loss": 0.2751, + "step": 2106 + }, + { + "epoch": 41.0, + "learning_rate": 0.00032166666666666666, + "loss": 0.2707, + "step": 2107 + }, + { + "epoch": 41.0, + "learning_rate": 0.00032155555555555556, + "loss": 0.2164, + "step": 2108 + }, + { + "epoch": 41.0, + "learning_rate": 0.0003214444444444444, + "loss": 0.2408, + "step": 2109 + }, + { + "epoch": 41.0, + "learning_rate": 0.00032133333333333336, + "loss": 0.2152, + "step": 2110 + }, + { + "epoch": 41.0, + "learning_rate": 0.00032122222222222225, + "loss": 0.2702, + "step": 2111 + }, + { + "epoch": 41.0, + "learning_rate": 0.00032111111111111115, + "loss": 0.2222, + "step": 2112 + }, + { + "epoch": 41.0, + "learning_rate": 0.000321, + "loss": 0.2037, + "step": 2113 + }, + { + "epoch": 41.0, + "learning_rate": 0.0003208888888888889, + "loss": 0.2041, + "step": 2114 + }, + { + "epoch": 41.0, + "learning_rate": 0.0003207777777777778, + "loss": 0.2019, + "step": 2115 + }, + { + "epoch": 41.01, + "learning_rate": 0.00032066666666666664, + "loss": 0.1949, + "step": 2116 + }, + { + "epoch": 41.01, + "learning_rate": 0.00032055555555555554, + "loss": 0.2581, + "step": 2117 + }, + { + "epoch": 41.01, + "learning_rate": 0.00032044444444444443, + "loss": 0.1955, + "step": 2118 + }, + { + "epoch": 41.01, + "learning_rate": 0.0003203333333333334, + "loss": 0.2218, + "step": 2119 + }, + { + "epoch": 41.01, + "learning_rate": 0.00032022222222222223, + "loss": 0.208, + "step": 2120 + }, + { + "epoch": 41.01, + "learning_rate": 0.00032011111111111113, + "loss": 0.2174, + "step": 2121 + }, + { + "epoch": 41.01, + "learning_rate": 0.00032, + "loss": 0.205, + "step": 2122 + }, + { + "epoch": 41.01, + "learning_rate": 0.00031988888888888887, + "loss": 0.2156, + "step": 2123 + }, + { + "epoch": 41.01, + "learning_rate": 0.00031977777777777777, + "loss": 0.287, + "step": 2124 + }, + { + "epoch": 41.01, + "learning_rate": 0.00031966666666666667, + "loss": 0.1867, + "step": 2125 + }, + { + "epoch": 41.01, + "learning_rate": 0.00031955555555555556, + "loss": 0.1849, + "step": 2126 + }, + { + "epoch": 41.01, + "learning_rate": 0.0003194444444444444, + "loss": 0.2157, + "step": 2127 + }, + { + "epoch": 41.01, + "learning_rate": 0.00031933333333333336, + "loss": 0.2303, + "step": 2128 + }, + { + "epoch": 41.01, + "learning_rate": 0.00031922222222222226, + "loss": 0.279, + "step": 2129 + }, + { + "epoch": 41.01, + "learning_rate": 0.0003191111111111111, + "loss": 0.1816, + "step": 2130 + }, + { + "epoch": 41.01, + "learning_rate": 0.000319, + "loss": 0.2396, + "step": 2131 + }, + { + "epoch": 41.01, + "learning_rate": 0.0003188888888888889, + "loss": 0.1983, + "step": 2132 + }, + { + "epoch": 41.01, + "learning_rate": 0.0003187777777777778, + "loss": 0.2044, + "step": 2133 + }, + { + "epoch": 41.01, + "learning_rate": 0.00031866666666666664, + "loss": 0.215, + "step": 2134 + }, + { + "epoch": 41.01, + "learning_rate": 0.00031855555555555554, + "loss": 0.2597, + "step": 2135 + }, + { + "epoch": 41.01, + "learning_rate": 0.00031844444444444444, + "loss": 0.229, + "step": 2136 + }, + { + "epoch": 41.01, + "learning_rate": 0.00031833333333333334, + "loss": 0.2159, + "step": 2137 + }, + { + "epoch": 41.01, + "learning_rate": 0.00031822222222222224, + "loss": 0.2552, + "step": 2138 + }, + { + "epoch": 41.01, + "learning_rate": 0.00031811111111111113, + "loss": 0.1992, + "step": 2139 + }, + { + "epoch": 41.01, + "learning_rate": 0.00031800000000000003, + "loss": 0.2372, + "step": 2140 + }, + { + "epoch": 41.01, + "learning_rate": 0.0003178888888888889, + "loss": 0.2185, + "step": 2141 + }, + { + "epoch": 41.01, + "learning_rate": 0.0003177777777777778, + "loss": 0.2023, + "step": 2142 + }, + { + "epoch": 42.0, + "learning_rate": 0.00031766666666666667, + "loss": 0.3373, + "step": 2143 + }, + { + "epoch": 42.0, + "learning_rate": 0.0003175555555555555, + "loss": 0.2282, + "step": 2144 + }, + { + "epoch": 42.0, + "learning_rate": 0.0003174444444444444, + "loss": 0.1786, + "step": 2145 + }, + { + "epoch": 42.0, + "learning_rate": 0.00031733333333333337, + "loss": 0.1953, + "step": 2146 + }, + { + "epoch": 42.0, + "learning_rate": 0.00031722222222222227, + "loss": 0.2145, + "step": 2147 + }, + { + "epoch": 42.0, + "learning_rate": 0.0003171111111111111, + "loss": 0.2285, + "step": 2148 + }, + { + "epoch": 42.0, + "learning_rate": 0.000317, + "loss": 0.1828, + "step": 2149 + }, + { + "epoch": 42.0, + "learning_rate": 0.000317, + "loss": 0.1878, + "step": 2150 + }, + { + "epoch": 42.0, + "learning_rate": 0.0003168888888888889, + "loss": 0.1551, + "step": 2151 + }, + { + "epoch": 42.0, + "learning_rate": 0.00031677777777777775, + "loss": 0.1947, + "step": 2152 + }, + { + "epoch": 42.0, + "learning_rate": 0.00031666666666666665, + "loss": 0.2597, + "step": 2153 + }, + { + "epoch": 42.0, + "learning_rate": 0.00031655555555555555, + "loss": 0.2511, + "step": 2154 + }, + { + "epoch": 42.0, + "learning_rate": 0.00031644444444444444, + "loss": 0.2133, + "step": 2155 + }, + { + "epoch": 42.0, + "learning_rate": 0.00031633333333333334, + "loss": 0.1814, + "step": 2156 + }, + { + "epoch": 42.0, + "learning_rate": 0.00031622222222222224, + "loss": 0.2495, + "step": 2157 + }, + { + "epoch": 42.0, + "learning_rate": 0.00031611111111111114, + "loss": 0.243, + "step": 2158 + }, + { + "epoch": 42.0, + "learning_rate": 0.000316, + "loss": 0.1678, + "step": 2159 + }, + { + "epoch": 42.0, + "learning_rate": 0.0003158888888888889, + "loss": 0.2746, + "step": 2160 + }, + { + "epoch": 42.0, + "learning_rate": 0.0003157777777777778, + "loss": 0.2407, + "step": 2161 + }, + { + "epoch": 42.0, + "learning_rate": 0.0003156666666666667, + "loss": 0.2587, + "step": 2162 + }, + { + "epoch": 42.0, + "learning_rate": 0.0003155555555555555, + "loss": 0.2535, + "step": 2163 + }, + { + "epoch": 42.0, + "learning_rate": 0.0003154444444444445, + "loss": 0.2359, + "step": 2164 + }, + { + "epoch": 42.0, + "learning_rate": 0.00031533333333333337, + "loss": 0.2146, + "step": 2165 + }, + { + "epoch": 42.0, + "learning_rate": 0.0003152222222222222, + "loss": 0.2212, + "step": 2166 + }, + { + "epoch": 42.01, + "learning_rate": 0.0003151111111111111, + "loss": 0.2833, + "step": 2167 + }, + { + "epoch": 42.01, + "learning_rate": 0.000315, + "loss": 0.2377, + "step": 2168 + }, + { + "epoch": 42.01, + "learning_rate": 0.0003148888888888889, + "loss": 0.168, + "step": 2169 + }, + { + "epoch": 42.01, + "learning_rate": 0.00031477777777777776, + "loss": 0.2445, + "step": 2170 + }, + { + "epoch": 42.01, + "learning_rate": 0.00031466666666666665, + "loss": 0.2124, + "step": 2171 + }, + { + "epoch": 42.01, + "learning_rate": 0.00031455555555555555, + "loss": 0.1877, + "step": 2172 + }, + { + "epoch": 42.01, + "learning_rate": 0.0003144444444444445, + "loss": 0.2371, + "step": 2173 + }, + { + "epoch": 42.01, + "learning_rate": 0.00031433333333333335, + "loss": 0.1713, + "step": 2174 + }, + { + "epoch": 42.01, + "learning_rate": 0.00031422222222222225, + "loss": 0.234, + "step": 2175 + }, + { + "epoch": 42.01, + "learning_rate": 0.00031411111111111114, + "loss": 0.253, + "step": 2176 + }, + { + "epoch": 42.01, + "learning_rate": 0.000314, + "loss": 0.2057, + "step": 2177 + }, + { + "epoch": 42.01, + "learning_rate": 0.0003138888888888889, + "loss": 0.2656, + "step": 2178 + }, + { + "epoch": 42.01, + "learning_rate": 0.0003137777777777778, + "loss": 0.2233, + "step": 2179 + }, + { + "epoch": 42.01, + "learning_rate": 0.00031366666666666663, + "loss": 0.2736, + "step": 2180 + }, + { + "epoch": 42.01, + "learning_rate": 0.00031355555555555553, + "loss": 0.1797, + "step": 2181 + }, + { + "epoch": 42.01, + "learning_rate": 0.0003134444444444445, + "loss": 0.2076, + "step": 2182 + }, + { + "epoch": 42.01, + "learning_rate": 0.0003133333333333334, + "loss": 0.2312, + "step": 2183 + }, + { + "epoch": 42.01, + "learning_rate": 0.0003132222222222222, + "loss": 0.2, + "step": 2184 + }, + { + "epoch": 42.01, + "learning_rate": 0.0003131111111111111, + "loss": 0.1956, + "step": 2185 + }, + { + "epoch": 42.01, + "learning_rate": 0.000313, + "loss": 0.2415, + "step": 2186 + }, + { + "epoch": 42.01, + "learning_rate": 0.00031288888888888886, + "loss": 0.2237, + "step": 2187 + }, + { + "epoch": 42.01, + "learning_rate": 0.00031277777777777776, + "loss": 0.2184, + "step": 2188 + }, + { + "epoch": 42.01, + "learning_rate": 0.00031266666666666666, + "loss": 0.2491, + "step": 2189 + }, + { + "epoch": 42.01, + "learning_rate": 0.00031255555555555556, + "loss": 0.1751, + "step": 2190 + }, + { + "epoch": 42.01, + "learning_rate": 0.00031244444444444446, + "loss": 0.1979, + "step": 2191 + }, + { + "epoch": 42.01, + "learning_rate": 0.00031233333333333335, + "loss": 0.2458, + "step": 2192 + }, + { + "epoch": 42.01, + "learning_rate": 0.00031222222222222225, + "loss": 0.2284, + "step": 2193 + }, + { + "epoch": 43.0, + "learning_rate": 0.00031211111111111115, + "loss": 0.3352, + "step": 2194 + }, + { + "epoch": 43.0, + "learning_rate": 0.000312, + "loss": 0.2321, + "step": 2195 + }, + { + "epoch": 43.0, + "learning_rate": 0.0003118888888888889, + "loss": 0.211, + "step": 2196 + }, + { + "epoch": 43.0, + "learning_rate": 0.0003117777777777778, + "loss": 0.2421, + "step": 2197 + }, + { + "epoch": 43.0, + "learning_rate": 0.00031166666666666663, + "loss": 0.2668, + "step": 2198 + }, + { + "epoch": 43.0, + "learning_rate": 0.00031155555555555553, + "loss": 0.1999, + "step": 2199 + }, + { + "epoch": 43.0, + "learning_rate": 0.0003114444444444445, + "loss": 0.1948, + "step": 2200 + }, + { + "epoch": 43.0, + "learning_rate": 0.0003113333333333334, + "loss": 0.2426, + "step": 2201 + }, + { + "epoch": 43.0, + "learning_rate": 0.00031122222222222223, + "loss": 0.1674, + "step": 2202 + }, + { + "epoch": 43.0, + "learning_rate": 0.0003111111111111111, + "loss": 0.2415, + "step": 2203 + }, + { + "epoch": 43.0, + "learning_rate": 0.000311, + "loss": 0.2051, + "step": 2204 + }, + { + "epoch": 43.0, + "learning_rate": 0.00031088888888888887, + "loss": 0.191, + "step": 2205 + }, + { + "epoch": 43.0, + "learning_rate": 0.00031077777777777777, + "loss": 0.1914, + "step": 2206 + }, + { + "epoch": 43.0, + "learning_rate": 0.00031066666666666666, + "loss": 0.1974, + "step": 2207 + }, + { + "epoch": 43.0, + "learning_rate": 0.0003105555555555555, + "loss": 0.2332, + "step": 2208 + }, + { + "epoch": 43.0, + "learning_rate": 0.00031044444444444446, + "loss": 0.1893, + "step": 2209 + }, + { + "epoch": 43.0, + "learning_rate": 0.00031033333333333336, + "loss": 0.2332, + "step": 2210 + }, + { + "epoch": 43.0, + "learning_rate": 0.00031022222222222226, + "loss": 0.2077, + "step": 2211 + }, + { + "epoch": 43.0, + "learning_rate": 0.0003101111111111111, + "loss": 0.2587, + "step": 2212 + }, + { + "epoch": 43.0, + "learning_rate": 0.00031, + "loss": 0.2082, + "step": 2213 + }, + { + "epoch": 43.0, + "learning_rate": 0.0003098888888888889, + "loss": 0.2337, + "step": 2214 + }, + { + "epoch": 43.0, + "learning_rate": 0.0003097777777777778, + "loss": 0.2031, + "step": 2215 + }, + { + "epoch": 43.0, + "learning_rate": 0.00030966666666666664, + "loss": 0.256, + "step": 2216 + }, + { + "epoch": 43.0, + "learning_rate": 0.00030955555555555554, + "loss": 0.2315, + "step": 2217 + }, + { + "epoch": 43.01, + "learning_rate": 0.0003094444444444445, + "loss": 0.2073, + "step": 2218 + }, + { + "epoch": 43.01, + "learning_rate": 0.00030933333333333334, + "loss": 0.1772, + "step": 2219 + }, + { + "epoch": 43.01, + "learning_rate": 0.00030922222222222223, + "loss": 0.2021, + "step": 2220 + }, + { + "epoch": 43.01, + "learning_rate": 0.00030911111111111113, + "loss": 0.2203, + "step": 2221 + }, + { + "epoch": 43.01, + "learning_rate": 0.00030900000000000003, + "loss": 0.1847, + "step": 2222 + }, + { + "epoch": 43.01, + "learning_rate": 0.0003088888888888889, + "loss": 0.2153, + "step": 2223 + }, + { + "epoch": 43.01, + "learning_rate": 0.00030877777777777777, + "loss": 0.211, + "step": 2224 + }, + { + "epoch": 43.01, + "learning_rate": 0.00030866666666666667, + "loss": 0.2603, + "step": 2225 + }, + { + "epoch": 43.01, + "learning_rate": 0.00030855555555555557, + "loss": 0.2368, + "step": 2226 + }, + { + "epoch": 43.01, + "learning_rate": 0.00030844444444444447, + "loss": 0.2113, + "step": 2227 + }, + { + "epoch": 43.01, + "learning_rate": 0.00030833333333333337, + "loss": 0.2101, + "step": 2228 + }, + { + "epoch": 43.01, + "learning_rate": 0.00030822222222222226, + "loss": 0.2816, + "step": 2229 + }, + { + "epoch": 43.01, + "learning_rate": 0.0003081111111111111, + "loss": 0.2355, + "step": 2230 + }, + { + "epoch": 43.01, + "learning_rate": 0.000308, + "loss": 0.2189, + "step": 2231 + }, + { + "epoch": 43.01, + "learning_rate": 0.0003078888888888889, + "loss": 0.2414, + "step": 2232 + }, + { + "epoch": 43.01, + "learning_rate": 0.00030777777777777775, + "loss": 0.2235, + "step": 2233 + }, + { + "epoch": 43.01, + "learning_rate": 0.00030766666666666665, + "loss": 0.2197, + "step": 2234 + }, + { + "epoch": 43.01, + "learning_rate": 0.0003075555555555556, + "loss": 0.2512, + "step": 2235 + }, + { + "epoch": 43.01, + "learning_rate": 0.0003074444444444445, + "loss": 0.1833, + "step": 2236 + }, + { + "epoch": 43.01, + "learning_rate": 0.00030733333333333334, + "loss": 0.2183, + "step": 2237 + }, + { + "epoch": 43.01, + "learning_rate": 0.00030722222222222224, + "loss": 0.2154, + "step": 2238 + }, + { + "epoch": 43.01, + "learning_rate": 0.00030711111111111114, + "loss": 0.1924, + "step": 2239 + }, + { + "epoch": 43.01, + "learning_rate": 0.000307, + "loss": 0.1909, + "step": 2240 + }, + { + "epoch": 43.01, + "learning_rate": 0.0003068888888888889, + "loss": 0.2215, + "step": 2241 + }, + { + "epoch": 43.01, + "learning_rate": 0.0003067777777777778, + "loss": 0.2122, + "step": 2242 + }, + { + "epoch": 43.01, + "learning_rate": 0.0003066666666666667, + "loss": 0.2761, + "step": 2243 + }, + { + "epoch": 43.01, + "learning_rate": 0.0003065555555555556, + "loss": 0.1763, + "step": 2244 + }, + { + "epoch": 44.0, + "learning_rate": 0.00030644444444444447, + "loss": 0.3352, + "step": 2245 + }, + { + "epoch": 44.0, + "learning_rate": 0.00030633333333333337, + "loss": 0.1956, + "step": 2246 + }, + { + "epoch": 44.0, + "learning_rate": 0.0003062222222222222, + "loss": 0.2076, + "step": 2247 + }, + { + "epoch": 44.0, + "learning_rate": 0.0003061111111111111, + "loss": 0.2137, + "step": 2248 + }, + { + "epoch": 44.0, + "learning_rate": 0.000306, + "loss": 0.2376, + "step": 2249 + }, + { + "epoch": 44.0, + "learning_rate": 0.0003058888888888889, + "loss": 0.2116, + "step": 2250 + }, + { + "epoch": 44.0, + "learning_rate": 0.00030577777777777775, + "loss": 0.1984, + "step": 2251 + }, + { + "epoch": 44.0, + "learning_rate": 0.00030566666666666665, + "loss": 0.2369, + "step": 2252 + }, + { + "epoch": 44.0, + "learning_rate": 0.0003055555555555556, + "loss": 0.228, + "step": 2253 + }, + { + "epoch": 44.0, + "learning_rate": 0.00030544444444444445, + "loss": 0.2074, + "step": 2254 + }, + { + "epoch": 44.0, + "learning_rate": 0.00030533333333333335, + "loss": 0.2131, + "step": 2255 + }, + { + "epoch": 44.0, + "learning_rate": 0.00030522222222222224, + "loss": 0.2037, + "step": 2256 + }, + { + "epoch": 44.0, + "learning_rate": 0.00030511111111111114, + "loss": 0.2024, + "step": 2257 + }, + { + "epoch": 44.0, + "learning_rate": 0.000305, + "loss": 0.242, + "step": 2258 + }, + { + "epoch": 44.0, + "learning_rate": 0.0003048888888888889, + "loss": 0.2247, + "step": 2259 + }, + { + "epoch": 44.0, + "learning_rate": 0.0003047777777777778, + "loss": 0.2701, + "step": 2260 + }, + { + "epoch": 44.0, + "learning_rate": 0.00030466666666666663, + "loss": 0.241, + "step": 2261 + }, + { + "epoch": 44.0, + "learning_rate": 0.0003045555555555556, + "loss": 0.1699, + "step": 2262 + }, + { + "epoch": 44.0, + "learning_rate": 0.0003044444444444445, + "loss": 0.2624, + "step": 2263 + }, + { + "epoch": 44.0, + "learning_rate": 0.0003043333333333334, + "loss": 0.261, + "step": 2264 + }, + { + "epoch": 44.0, + "learning_rate": 0.0003042222222222222, + "loss": 0.2428, + "step": 2265 + }, + { + "epoch": 44.0, + "learning_rate": 0.0003041111111111111, + "loss": 0.2693, + "step": 2266 + }, + { + "epoch": 44.0, + "learning_rate": 0.000304, + "loss": 0.2372, + "step": 2267 + }, + { + "epoch": 44.0, + "learning_rate": 0.00030388888888888886, + "loss": 0.2388, + "step": 2268 + }, + { + "epoch": 44.01, + "learning_rate": 0.00030377777777777776, + "loss": 0.2447, + "step": 2269 + }, + { + "epoch": 44.01, + "learning_rate": 0.00030366666666666666, + "loss": 0.2499, + "step": 2270 + }, + { + "epoch": 44.01, + "learning_rate": 0.0003035555555555556, + "loss": 0.2715, + "step": 2271 + }, + { + "epoch": 44.01, + "learning_rate": 0.00030344444444444445, + "loss": 0.2401, + "step": 2272 + }, + { + "epoch": 44.01, + "learning_rate": 0.00030333333333333335, + "loss": 0.2201, + "step": 2273 + }, + { + "epoch": 44.01, + "learning_rate": 0.00030322222222222225, + "loss": 0.1854, + "step": 2274 + }, + { + "epoch": 44.01, + "learning_rate": 0.0003031111111111111, + "loss": 0.2124, + "step": 2275 + }, + { + "epoch": 44.01, + "learning_rate": 0.000303, + "loss": 0.2237, + "step": 2276 + }, + { + "epoch": 44.01, + "learning_rate": 0.0003028888888888889, + "loss": 0.2361, + "step": 2277 + }, + { + "epoch": 44.01, + "learning_rate": 0.0003027777777777778, + "loss": 0.2211, + "step": 2278 + }, + { + "epoch": 44.01, + "learning_rate": 0.00030266666666666663, + "loss": 0.2446, + "step": 2279 + }, + { + "epoch": 44.01, + "learning_rate": 0.0003025555555555556, + "loss": 0.1822, + "step": 2280 + }, + { + "epoch": 44.01, + "learning_rate": 0.0003024444444444445, + "loss": 0.1917, + "step": 2281 + }, + { + "epoch": 44.01, + "learning_rate": 0.00030233333333333333, + "loss": 0.1948, + "step": 2282 + }, + { + "epoch": 44.01, + "learning_rate": 0.0003022222222222222, + "loss": 0.2132, + "step": 2283 + }, + { + "epoch": 44.01, + "learning_rate": 0.0003021111111111111, + "loss": 0.1832, + "step": 2284 + }, + { + "epoch": 44.01, + "learning_rate": 0.000302, + "loss": 0.1768, + "step": 2285 + }, + { + "epoch": 44.01, + "learning_rate": 0.00030188888888888887, + "loss": 0.2431, + "step": 2286 + }, + { + "epoch": 44.01, + "learning_rate": 0.00030177777777777776, + "loss": 0.2094, + "step": 2287 + }, + { + "epoch": 44.01, + "learning_rate": 0.0003016666666666667, + "loss": 0.2488, + "step": 2288 + }, + { + "epoch": 44.01, + "learning_rate": 0.00030155555555555556, + "loss": 0.2445, + "step": 2289 + }, + { + "epoch": 44.01, + "learning_rate": 0.00030144444444444446, + "loss": 0.2253, + "step": 2290 + }, + { + "epoch": 44.01, + "learning_rate": 0.00030133333333333336, + "loss": 0.2004, + "step": 2291 + }, + { + "epoch": 44.01, + "learning_rate": 0.00030122222222222226, + "loss": 0.2123, + "step": 2292 + }, + { + "epoch": 44.01, + "learning_rate": 0.0003011111111111111, + "loss": 0.216, + "step": 2293 + }, + { + "epoch": 44.01, + "learning_rate": 0.000301, + "loss": 0.219, + "step": 2294 + }, + { + "epoch": 44.01, + "learning_rate": 0.0003008888888888889, + "loss": 0.1893, + "step": 2295 + }, + { + "epoch": 45.0, + "learning_rate": 0.00030077777777777774, + "loss": 0.2989, + "step": 2296 + }, + { + "epoch": 45.0, + "learning_rate": 0.0003006666666666667, + "loss": 0.2035, + "step": 2297 + }, + { + "epoch": 45.0, + "learning_rate": 0.0003005555555555556, + "loss": 0.2402, + "step": 2298 + }, + { + "epoch": 45.0, + "learning_rate": 0.0003004444444444445, + "loss": 0.1938, + "step": 2299 + }, + { + "epoch": 45.0, + "learning_rate": 0.00030033333333333333, + "loss": 0.1832, + "step": 2300 + }, + { + "epoch": 45.0, + "learning_rate": 0.00030022222222222223, + "loss": 0.1904, + "step": 2301 + }, + { + "epoch": 45.0, + "learning_rate": 0.00030011111111111113, + "loss": 0.1929, + "step": 2302 + }, + { + "epoch": 45.0, + "learning_rate": 0.0003, + "loss": 0.201, + "step": 2303 + }, + { + "epoch": 45.0, + "learning_rate": 0.00029988888888888887, + "loss": 0.2075, + "step": 2304 + }, + { + "epoch": 45.0, + "learning_rate": 0.00029977777777777777, + "loss": 0.1811, + "step": 2305 + }, + { + "epoch": 45.0, + "learning_rate": 0.0002996666666666667, + "loss": 0.2105, + "step": 2306 + }, + { + "epoch": 45.0, + "learning_rate": 0.00029955555555555557, + "loss": 0.2045, + "step": 2307 + }, + { + "epoch": 45.0, + "learning_rate": 0.00029944444444444446, + "loss": 0.2306, + "step": 2308 + }, + { + "epoch": 45.0, + "learning_rate": 0.00029933333333333336, + "loss": 0.1857, + "step": 2309 + }, + { + "epoch": 45.0, + "learning_rate": 0.0002992222222222222, + "loss": 0.1886, + "step": 2310 + }, + { + "epoch": 45.0, + "learning_rate": 0.0002991111111111111, + "loss": 0.247, + "step": 2311 + }, + { + "epoch": 45.0, + "learning_rate": 0.000299, + "loss": 0.1921, + "step": 2312 + }, + { + "epoch": 45.0, + "learning_rate": 0.0002988888888888889, + "loss": 0.1942, + "step": 2313 + }, + { + "epoch": 45.0, + "learning_rate": 0.00029877777777777775, + "loss": 0.2293, + "step": 2314 + }, + { + "epoch": 45.0, + "learning_rate": 0.0002986666666666667, + "loss": 0.2615, + "step": 2315 + }, + { + "epoch": 45.0, + "learning_rate": 0.0002985555555555556, + "loss": 0.2367, + "step": 2316 + }, + { + "epoch": 45.0, + "learning_rate": 0.00029844444444444444, + "loss": 0.2238, + "step": 2317 + }, + { + "epoch": 45.0, + "learning_rate": 0.00029833333333333334, + "loss": 0.2147, + "step": 2318 + }, + { + "epoch": 45.0, + "learning_rate": 0.00029822222222222224, + "loss": 0.2342, + "step": 2319 + }, + { + "epoch": 45.01, + "learning_rate": 0.00029811111111111114, + "loss": 0.1853, + "step": 2320 + }, + { + "epoch": 45.01, + "learning_rate": 0.000298, + "loss": 0.2038, + "step": 2321 + }, + { + "epoch": 45.01, + "learning_rate": 0.0002978888888888889, + "loss": 0.2334, + "step": 2322 + }, + { + "epoch": 45.01, + "learning_rate": 0.0002977777777777778, + "loss": 0.2267, + "step": 2323 + }, + { + "epoch": 45.01, + "learning_rate": 0.0002976666666666667, + "loss": 0.1821, + "step": 2324 + }, + { + "epoch": 45.01, + "learning_rate": 0.00029755555555555557, + "loss": 0.2622, + "step": 2325 + }, + { + "epoch": 45.01, + "learning_rate": 0.00029744444444444447, + "loss": 0.1842, + "step": 2326 + }, + { + "epoch": 45.01, + "learning_rate": 0.00029733333333333337, + "loss": 0.204, + "step": 2327 + }, + { + "epoch": 45.01, + "learning_rate": 0.0002972222222222222, + "loss": 0.1745, + "step": 2328 + }, + { + "epoch": 45.01, + "learning_rate": 0.0002971111111111111, + "loss": 0.2412, + "step": 2329 + }, + { + "epoch": 45.01, + "learning_rate": 0.000297, + "loss": 0.2207, + "step": 2330 + }, + { + "epoch": 45.01, + "learning_rate": 0.00029688888888888885, + "loss": 0.1964, + "step": 2331 + }, + { + "epoch": 45.01, + "learning_rate": 0.00029677777777777775, + "loss": 0.2225, + "step": 2332 + }, + { + "epoch": 45.01, + "learning_rate": 0.0002966666666666667, + "loss": 0.2475, + "step": 2333 + }, + { + "epoch": 45.01, + "learning_rate": 0.0002965555555555556, + "loss": 0.2056, + "step": 2334 + }, + { + "epoch": 45.01, + "learning_rate": 0.00029644444444444445, + "loss": 0.2122, + "step": 2335 + }, + { + "epoch": 45.01, + "learning_rate": 0.00029633333333333334, + "loss": 0.1709, + "step": 2336 + }, + { + "epoch": 45.01, + "learning_rate": 0.00029622222222222224, + "loss": 0.2096, + "step": 2337 + }, + { + "epoch": 45.01, + "learning_rate": 0.0002961111111111111, + "loss": 0.2108, + "step": 2338 + }, + { + "epoch": 45.01, + "learning_rate": 0.000296, + "loss": 0.1974, + "step": 2339 + }, + { + "epoch": 45.01, + "learning_rate": 0.0002958888888888889, + "loss": 0.2325, + "step": 2340 + }, + { + "epoch": 45.01, + "learning_rate": 0.0002957777777777778, + "loss": 0.2067, + "step": 2341 + }, + { + "epoch": 45.01, + "learning_rate": 0.0002956666666666667, + "loss": 0.2023, + "step": 2342 + }, + { + "epoch": 45.01, + "learning_rate": 0.0002955555555555556, + "loss": 0.2389, + "step": 2343 + }, + { + "epoch": 45.01, + "learning_rate": 0.0002954444444444445, + "loss": 0.2293, + "step": 2344 + }, + { + "epoch": 45.01, + "learning_rate": 0.0002953333333333333, + "loss": 0.206, + "step": 2345 + }, + { + "epoch": 45.01, + "learning_rate": 0.0002952222222222222, + "loss": 0.2146, + "step": 2346 + }, + { + "epoch": 46.0, + "learning_rate": 0.0002951111111111111, + "loss": 0.3222, + "step": 2347 + }, + { + "epoch": 46.0, + "learning_rate": 0.000295, + "loss": 0.2191, + "step": 2348 + }, + { + "epoch": 46.0, + "learning_rate": 0.00029488888888888886, + "loss": 0.2067, + "step": 2349 + }, + { + "epoch": 46.0, + "learning_rate": 0.0002947777777777778, + "loss": 0.1903, + "step": 2350 + }, + { + "epoch": 46.0, + "learning_rate": 0.0002946666666666667, + "loss": 0.1481, + "step": 2351 + }, + { + "epoch": 46.0, + "learning_rate": 0.00029455555555555555, + "loss": 0.1945, + "step": 2352 + }, + { + "epoch": 46.0, + "learning_rate": 0.00029444444444444445, + "loss": 0.1961, + "step": 2353 + }, + { + "epoch": 46.0, + "learning_rate": 0.00029433333333333335, + "loss": 0.1831, + "step": 2354 + }, + { + "epoch": 46.0, + "learning_rate": 0.00029422222222222225, + "loss": 0.1857, + "step": 2355 + }, + { + "epoch": 46.0, + "learning_rate": 0.0002941111111111111, + "loss": 0.2372, + "step": 2356 + }, + { + "epoch": 46.0, + "learning_rate": 0.000294, + "loss": 0.1984, + "step": 2357 + }, + { + "epoch": 46.0, + "learning_rate": 0.0002938888888888889, + "loss": 0.1683, + "step": 2358 + }, + { + "epoch": 46.0, + "learning_rate": 0.0002937777777777778, + "loss": 0.2101, + "step": 2359 + }, + { + "epoch": 46.0, + "learning_rate": 0.0002936666666666667, + "loss": 0.1899, + "step": 2360 + }, + { + "epoch": 46.0, + "learning_rate": 0.0002935555555555556, + "loss": 0.2381, + "step": 2361 + }, + { + "epoch": 46.0, + "learning_rate": 0.0002934444444444445, + "loss": 0.1649, + "step": 2362 + }, + { + "epoch": 46.0, + "learning_rate": 0.0002933333333333333, + "loss": 0.1889, + "step": 2363 + }, + { + "epoch": 46.0, + "learning_rate": 0.0002932222222222222, + "loss": 0.1513, + "step": 2364 + }, + { + "epoch": 46.0, + "learning_rate": 0.0002931111111111111, + "loss": 0.1927, + "step": 2365 + }, + { + "epoch": 46.0, + "learning_rate": 0.00029299999999999997, + "loss": 0.1972, + "step": 2366 + }, + { + "epoch": 46.0, + "learning_rate": 0.00029288888888888886, + "loss": 0.2027, + "step": 2367 + }, + { + "epoch": 46.0, + "learning_rate": 0.0002927777777777778, + "loss": 0.1623, + "step": 2368 + }, + { + "epoch": 46.0, + "learning_rate": 0.0002926666666666667, + "loss": 0.1999, + "step": 2369 + }, + { + "epoch": 46.0, + "learning_rate": 0.00029255555555555556, + "loss": 0.2522, + "step": 2370 + }, + { + "epoch": 46.01, + "learning_rate": 0.00029244444444444446, + "loss": 0.2011, + "step": 2371 + }, + { + "epoch": 46.01, + "learning_rate": 0.00029233333333333336, + "loss": 0.1876, + "step": 2372 + }, + { + "epoch": 46.01, + "learning_rate": 0.0002922222222222222, + "loss": 0.2484, + "step": 2373 + }, + { + "epoch": 46.01, + "learning_rate": 0.0002921111111111111, + "loss": 0.2034, + "step": 2374 + }, + { + "epoch": 46.01, + "learning_rate": 0.000292, + "loss": 0.2361, + "step": 2375 + }, + { + "epoch": 46.01, + "learning_rate": 0.0002918888888888889, + "loss": 0.2144, + "step": 2376 + }, + { + "epoch": 46.01, + "learning_rate": 0.0002917777777777778, + "loss": 0.2373, + "step": 2377 + }, + { + "epoch": 46.01, + "learning_rate": 0.0002916666666666667, + "loss": 0.2131, + "step": 2378 + }, + { + "epoch": 46.01, + "learning_rate": 0.0002915555555555556, + "loss": 0.2104, + "step": 2379 + }, + { + "epoch": 46.01, + "learning_rate": 0.00029144444444444443, + "loss": 0.2041, + "step": 2380 + }, + { + "epoch": 46.01, + "learning_rate": 0.00029133333333333333, + "loss": 0.1835, + "step": 2381 + }, + { + "epoch": 46.01, + "learning_rate": 0.00029122222222222223, + "loss": 0.2301, + "step": 2382 + }, + { + "epoch": 46.01, + "learning_rate": 0.00029111111111111113, + "loss": 0.2299, + "step": 2383 + }, + { + "epoch": 46.01, + "learning_rate": 0.00029099999999999997, + "loss": 0.2568, + "step": 2384 + }, + { + "epoch": 46.01, + "learning_rate": 0.00029088888888888887, + "loss": 0.2933, + "step": 2385 + }, + { + "epoch": 46.01, + "learning_rate": 0.0002907777777777778, + "loss": 0.1907, + "step": 2386 + }, + { + "epoch": 46.01, + "learning_rate": 0.00029066666666666667, + "loss": 0.203, + "step": 2387 + }, + { + "epoch": 46.01, + "learning_rate": 0.00029055555555555556, + "loss": 0.2496, + "step": 2388 + }, + { + "epoch": 46.01, + "learning_rate": 0.00029044444444444446, + "loss": 0.24, + "step": 2389 + }, + { + "epoch": 46.01, + "learning_rate": 0.00029033333333333336, + "loss": 0.2572, + "step": 2390 + }, + { + "epoch": 46.01, + "learning_rate": 0.0002902222222222222, + "loss": 0.272, + "step": 2391 + }, + { + "epoch": 46.01, + "learning_rate": 0.0002901111111111111, + "loss": 0.2162, + "step": 2392 + }, + { + "epoch": 46.01, + "learning_rate": 0.00029, + "loss": 0.2211, + "step": 2393 + }, + { + "epoch": 46.01, + "learning_rate": 0.00028988888888888885, + "loss": 0.2408, + "step": 2394 + }, + { + "epoch": 46.01, + "learning_rate": 0.0002897777777777778, + "loss": 0.222, + "step": 2395 + }, + { + "epoch": 46.01, + "learning_rate": 0.0002896666666666667, + "loss": 0.2153, + "step": 2396 + }, + { + "epoch": 46.01, + "learning_rate": 0.0002895555555555556, + "loss": 0.1865, + "step": 2397 + }, + { + "epoch": 47.0, + "learning_rate": 0.00028944444444444444, + "loss": 0.2872, + "step": 2398 + }, + { + "epoch": 47.0, + "learning_rate": 0.00028933333333333334, + "loss": 0.226, + "step": 2399 + }, + { + "epoch": 47.0, + "learning_rate": 0.00028922222222222224, + "loss": 0.2, + "step": 2400 + }, + { + "epoch": 47.0, + "learning_rate": 0.0002891111111111111, + "loss": 0.1924, + "step": 2401 + }, + { + "epoch": 47.0, + "learning_rate": 0.000289, + "loss": 0.1837, + "step": 2402 + }, + { + "epoch": 47.0, + "learning_rate": 0.0002888888888888889, + "loss": 0.2453, + "step": 2403 + }, + { + "epoch": 47.0, + "learning_rate": 0.00028877777777777783, + "loss": 0.2274, + "step": 2404 + }, + { + "epoch": 47.0, + "learning_rate": 0.00028866666666666667, + "loss": 0.2174, + "step": 2405 + }, + { + "epoch": 47.0, + "learning_rate": 0.00028855555555555557, + "loss": 0.1701, + "step": 2406 + }, + { + "epoch": 47.0, + "learning_rate": 0.00028844444444444447, + "loss": 0.2438, + "step": 2407 + }, + { + "epoch": 47.0, + "learning_rate": 0.0002883333333333333, + "loss": 0.2115, + "step": 2408 + }, + { + "epoch": 47.0, + "learning_rate": 0.0002882222222222222, + "loss": 0.1875, + "step": 2409 + }, + { + "epoch": 47.0, + "learning_rate": 0.0002881111111111111, + "loss": 0.2299, + "step": 2410 + }, + { + "epoch": 47.0, + "learning_rate": 0.000288, + "loss": 0.1886, + "step": 2411 + }, + { + "epoch": 47.0, + "learning_rate": 0.0002878888888888889, + "loss": 0.2515, + "step": 2412 + }, + { + "epoch": 47.0, + "learning_rate": 0.0002877777777777778, + "loss": 0.1958, + "step": 2413 + }, + { + "epoch": 47.0, + "learning_rate": 0.0002876666666666667, + "loss": 0.2204, + "step": 2414 + }, + { + "epoch": 47.0, + "learning_rate": 0.00028755555555555555, + "loss": 0.1952, + "step": 2415 + }, + { + "epoch": 47.0, + "learning_rate": 0.00028744444444444444, + "loss": 0.1909, + "step": 2416 + }, + { + "epoch": 47.0, + "learning_rate": 0.00028733333333333334, + "loss": 0.22, + "step": 2417 + }, + { + "epoch": 47.0, + "learning_rate": 0.00028722222222222224, + "loss": 0.214, + "step": 2418 + }, + { + "epoch": 47.0, + "learning_rate": 0.0002871111111111111, + "loss": 0.2881, + "step": 2419 + }, + { + "epoch": 47.0, + "learning_rate": 0.000287, + "loss": 0.2318, + "step": 2420 + }, + { + "epoch": 47.0, + "learning_rate": 0.00028688888888888894, + "loss": 0.2153, + "step": 2421 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002867777777777778, + "loss": 0.2018, + "step": 2422 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002866666666666667, + "loss": 0.2105, + "step": 2423 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002865555555555556, + "loss": 0.2101, + "step": 2424 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002864444444444445, + "loss": 0.1874, + "step": 2425 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002863333333333333, + "loss": 0.2071, + "step": 2426 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002862222222222222, + "loss": 0.2138, + "step": 2427 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002861111111111111, + "loss": 0.2528, + "step": 2428 + }, + { + "epoch": 47.01, + "learning_rate": 0.00028599999999999996, + "loss": 0.201, + "step": 2429 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002858888888888889, + "loss": 0.2262, + "step": 2430 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002857777777777778, + "loss": 0.2371, + "step": 2431 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002856666666666667, + "loss": 0.2338, + "step": 2432 + }, + { + "epoch": 47.01, + "learning_rate": 0.00028555555555555555, + "loss": 0.2078, + "step": 2433 + }, + { + "epoch": 47.01, + "learning_rate": 0.00028544444444444445, + "loss": 0.2145, + "step": 2434 + }, + { + "epoch": 47.01, + "learning_rate": 0.00028533333333333335, + "loss": 0.1973, + "step": 2435 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002852222222222222, + "loss": 0.2542, + "step": 2436 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002851111111111111, + "loss": 0.2357, + "step": 2437 + }, + { + "epoch": 47.01, + "learning_rate": 0.000285, + "loss": 0.1863, + "step": 2438 + }, + { + "epoch": 47.01, + "learning_rate": 0.00028488888888888894, + "loss": 0.2518, + "step": 2439 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002847777777777778, + "loss": 0.1948, + "step": 2440 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002846666666666667, + "loss": 0.1907, + "step": 2441 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002845555555555556, + "loss": 0.2575, + "step": 2442 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002844444444444444, + "loss": 0.2217, + "step": 2443 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002843333333333333, + "loss": 0.2394, + "step": 2444 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002842222222222222, + "loss": 0.2045, + "step": 2445 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002841111111111111, + "loss": 0.2512, + "step": 2446 + }, + { + "epoch": 47.01, + "learning_rate": 0.00028399999999999996, + "loss": 0.2393, + "step": 2447 + }, + { + "epoch": 47.01, + "learning_rate": 0.0002838888888888889, + "loss": 0.1732, + "step": 2448 + }, + { + "epoch": 48.0, + "learning_rate": 0.0002837777777777778, + "loss": 0.3277, + "step": 2449 + }, + { + "epoch": 48.0, + "learning_rate": 0.00028366666666666666, + "loss": 0.1987, + "step": 2450 + }, + { + "epoch": 48.0, + "learning_rate": 0.00028355555555555556, + "loss": 0.1967, + "step": 2451 + }, + { + "epoch": 48.0, + "learning_rate": 0.00028344444444444446, + "loss": 0.2124, + "step": 2452 + }, + { + "epoch": 48.0, + "learning_rate": 0.00028333333333333335, + "loss": 0.2361, + "step": 2453 + }, + { + "epoch": 48.0, + "learning_rate": 0.0002832222222222222, + "loss": 0.1794, + "step": 2454 + }, + { + "epoch": 48.0, + "learning_rate": 0.0002831111111111111, + "loss": 0.2301, + "step": 2455 + }, + { + "epoch": 48.0, + "learning_rate": 0.000283, + "loss": 0.195, + "step": 2456 + }, + { + "epoch": 48.0, + "learning_rate": 0.0002828888888888889, + "loss": 0.1848, + "step": 2457 + }, + { + "epoch": 48.0, + "learning_rate": 0.0002827777777777778, + "loss": 0.184, + "step": 2458 + }, + { + "epoch": 48.0, + "learning_rate": 0.0002826666666666667, + "loss": 0.2239, + "step": 2459 + }, + { + "epoch": 48.0, + "learning_rate": 0.0002825555555555556, + "loss": 0.2604, + "step": 2460 + }, + { + "epoch": 48.0, + "learning_rate": 0.00028244444444444443, + "loss": 0.1909, + "step": 2461 + }, + { + "epoch": 48.0, + "learning_rate": 0.00028233333333333333, + "loss": 0.1726, + "step": 2462 + }, + { + "epoch": 48.0, + "learning_rate": 0.00028222222222222223, + "loss": 0.1813, + "step": 2463 + }, + { + "epoch": 48.0, + "learning_rate": 0.00028211111111111107, + "loss": 0.1766, + "step": 2464 + }, + { + "epoch": 48.0, + "learning_rate": 0.00028199999999999997, + "loss": 0.1888, + "step": 2465 + }, + { + "epoch": 48.0, + "learning_rate": 0.0002818888888888889, + "loss": 0.1982, + "step": 2466 + }, + { + "epoch": 48.0, + "learning_rate": 0.0002817777777777778, + "loss": 0.2061, + "step": 2467 + }, + { + "epoch": 48.0, + "learning_rate": 0.00028166666666666666, + "loss": 0.1665, + "step": 2468 + }, + { + "epoch": 48.0, + "learning_rate": 0.00028155555555555556, + "loss": 0.1866, + "step": 2469 + }, + { + "epoch": 48.0, + "learning_rate": 0.00028144444444444446, + "loss": 0.1814, + "step": 2470 + }, + { + "epoch": 48.0, + "learning_rate": 0.0002813333333333333, + "loss": 0.1929, + "step": 2471 + }, + { + "epoch": 48.0, + "learning_rate": 0.0002812222222222222, + "loss": 0.2072, + "step": 2472 + }, + { + "epoch": 48.01, + "learning_rate": 0.0002811111111111111, + "loss": 0.2001, + "step": 2473 + }, + { + "epoch": 48.01, + "learning_rate": 0.00028100000000000005, + "loss": 0.2228, + "step": 2474 + }, + { + "epoch": 48.01, + "learning_rate": 0.0002808888888888889, + "loss": 0.1922, + "step": 2475 + }, + { + "epoch": 48.01, + "learning_rate": 0.0002807777777777778, + "loss": 0.1851, + "step": 2476 + }, + { + "epoch": 48.01, + "learning_rate": 0.0002806666666666667, + "loss": 0.162, + "step": 2477 + }, + { + "epoch": 48.01, + "learning_rate": 0.00028055555555555554, + "loss": 0.2399, + "step": 2478 + }, + { + "epoch": 48.01, + "learning_rate": 0.00028044444444444444, + "loss": 0.2029, + "step": 2479 + }, + { + "epoch": 48.01, + "learning_rate": 0.00028033333333333333, + "loss": 0.2213, + "step": 2480 + }, + { + "epoch": 48.01, + "learning_rate": 0.00028022222222222223, + "loss": 0.2299, + "step": 2481 + }, + { + "epoch": 48.01, + "learning_rate": 0.0002801111111111111, + "loss": 0.2163, + "step": 2482 + }, + { + "epoch": 48.01, + "learning_rate": 0.00028000000000000003, + "loss": 0.1841, + "step": 2483 + }, + { + "epoch": 48.01, + "learning_rate": 0.00027988888888888893, + "loss": 0.2447, + "step": 2484 + }, + { + "epoch": 48.01, + "learning_rate": 0.00027977777777777777, + "loss": 0.1764, + "step": 2485 + }, + { + "epoch": 48.01, + "learning_rate": 0.00027966666666666667, + "loss": 0.1857, + "step": 2486 + }, + { + "epoch": 48.01, + "learning_rate": 0.00027955555555555557, + "loss": 0.1843, + "step": 2487 + }, + { + "epoch": 48.01, + "learning_rate": 0.00027944444444444447, + "loss": 0.2019, + "step": 2488 + }, + { + "epoch": 48.01, + "learning_rate": 0.0002793333333333333, + "loss": 0.2394, + "step": 2489 + }, + { + "epoch": 48.01, + "learning_rate": 0.0002792222222222222, + "loss": 0.1753, + "step": 2490 + }, + { + "epoch": 48.01, + "learning_rate": 0.0002791111111111111, + "loss": 0.1803, + "step": 2491 + }, + { + "epoch": 48.01, + "learning_rate": 0.000279, + "loss": 0.2293, + "step": 2492 + }, + { + "epoch": 48.01, + "learning_rate": 0.0002788888888888889, + "loss": 0.1874, + "step": 2493 + }, + { + "epoch": 48.01, + "learning_rate": 0.0002787777777777778, + "loss": 0.2199, + "step": 2494 + }, + { + "epoch": 48.01, + "learning_rate": 0.0002786666666666667, + "loss": 0.1815, + "step": 2495 + }, + { + "epoch": 48.01, + "learning_rate": 0.00027855555555555554, + "loss": 0.1996, + "step": 2496 + }, + { + "epoch": 48.01, + "learning_rate": 0.00027844444444444444, + "loss": 0.1701, + "step": 2497 + }, + { + "epoch": 48.01, + "learning_rate": 0.00027833333333333334, + "loss": 0.2176, + "step": 2498 + }, + { + "epoch": 48.01, + "learning_rate": 0.0002782222222222222, + "loss": 0.2068, + "step": 2499 + }, + { + "epoch": 49.0, + "learning_rate": 0.0002781111111111111, + "loss": 0.3141, + "step": 2500 + }, + { + "epoch": 49.0, + "eval_cer": 0.0819367967440747, + "eval_loss": 0.34071585536003113, + "eval_runtime": 51.3852, + "eval_samples_per_second": 8.096, + "eval_steps_per_second": 0.253, + "eval_wer": 0.3372853638593622, + "step": 2500 + }, + { + "epoch": 49.0, + "learning_rate": 0.00027800000000000004, + "loss": 0.1769, + "step": 2501 + }, + { + "epoch": 49.0, + "learning_rate": 0.00027788888888888893, + "loss": 0.2132, + "step": 2502 + }, + { + "epoch": 49.0, + "learning_rate": 0.0002777777777777778, + "loss": 0.2043, + "step": 2503 + }, + { + "epoch": 49.0, + "learning_rate": 0.0002776666666666667, + "loss": 0.1839, + "step": 2504 + }, + { + "epoch": 49.0, + "learning_rate": 0.0002775555555555556, + "loss": 0.1429, + "step": 2505 + }, + { + "epoch": 49.0, + "learning_rate": 0.0002774444444444444, + "loss": 0.2099, + "step": 2506 + }, + { + "epoch": 49.0, + "learning_rate": 0.0002773333333333333, + "loss": 0.2261, + "step": 2507 + }, + { + "epoch": 49.0, + "learning_rate": 0.0002772222222222222, + "loss": 0.1903, + "step": 2508 + }, + { + "epoch": 49.0, + "learning_rate": 0.0002771111111111111, + "loss": 0.2151, + "step": 2509 + }, + { + "epoch": 49.0, + "learning_rate": 0.000277, + "loss": 0.2188, + "step": 2510 + }, + { + "epoch": 49.0, + "learning_rate": 0.0002768888888888889, + "loss": 0.2305, + "step": 2511 + }, + { + "epoch": 49.0, + "learning_rate": 0.0002767777777777778, + "loss": 0.2045, + "step": 2512 + }, + { + "epoch": 49.0, + "learning_rate": 0.00027666666666666665, + "loss": 0.2561, + "step": 2513 + }, + { + "epoch": 49.0, + "learning_rate": 0.00027655555555555555, + "loss": 0.1949, + "step": 2514 + }, + { + "epoch": 49.0, + "learning_rate": 0.00027644444444444445, + "loss": 0.1815, + "step": 2515 + }, + { + "epoch": 49.0, + "learning_rate": 0.00027633333333333335, + "loss": 0.175, + "step": 2516 + }, + { + "epoch": 49.0, + "learning_rate": 0.0002762222222222222, + "loss": 0.1862, + "step": 2517 + }, + { + "epoch": 49.0, + "learning_rate": 0.0002761111111111111, + "loss": 0.2129, + "step": 2518 + }, + { + "epoch": 49.0, + "learning_rate": 0.00027600000000000004, + "loss": 0.2203, + "step": 2519 + }, + { + "epoch": 49.0, + "learning_rate": 0.0002758888888888889, + "loss": 0.1817, + "step": 2520 + }, + { + "epoch": 49.0, + "learning_rate": 0.0002757777777777778, + "loss": 0.2078, + "step": 2521 + }, + { + "epoch": 49.0, + "learning_rate": 0.0002756666666666667, + "loss": 0.1985, + "step": 2522 + }, + { + "epoch": 49.0, + "learning_rate": 0.0002755555555555556, + "loss": 0.199, + "step": 2523 + }, + { + "epoch": 49.01, + "learning_rate": 0.0002754444444444444, + "loss": 0.226, + "step": 2524 + }, + { + "epoch": 49.01, + "learning_rate": 0.0002753333333333333, + "loss": 0.2697, + "step": 2525 + }, + { + "epoch": 49.01, + "learning_rate": 0.0002752222222222222, + "loss": 0.1678, + "step": 2526 + }, + { + "epoch": 49.01, + "learning_rate": 0.00027511111111111106, + "loss": 0.2266, + "step": 2527 + }, + { + "epoch": 49.01, + "learning_rate": 0.000275, + "loss": 0.2072, + "step": 2528 + }, + { + "epoch": 49.01, + "learning_rate": 0.0002748888888888889, + "loss": 0.2266, + "step": 2529 + }, + { + "epoch": 49.01, + "learning_rate": 0.0002747777777777778, + "loss": 0.2038, + "step": 2530 + }, + { + "epoch": 49.01, + "learning_rate": 0.00027466666666666666, + "loss": 0.1697, + "step": 2531 + }, + { + "epoch": 49.01, + "learning_rate": 0.00027455555555555556, + "loss": 0.2307, + "step": 2532 + }, + { + "epoch": 49.01, + "learning_rate": 0.00027444444444444445, + "loss": 0.1772, + "step": 2533 + }, + { + "epoch": 49.01, + "learning_rate": 0.0002743333333333333, + "loss": 0.205, + "step": 2534 + }, + { + "epoch": 49.01, + "learning_rate": 0.0002742222222222222, + "loss": 0.2526, + "step": 2535 + }, + { + "epoch": 49.01, + "learning_rate": 0.00027411111111111115, + "loss": 0.2074, + "step": 2536 + }, + { + "epoch": 49.01, + "learning_rate": 0.00027400000000000005, + "loss": 0.2029, + "step": 2537 + }, + { + "epoch": 49.01, + "learning_rate": 0.0002738888888888889, + "loss": 0.168, + "step": 2538 + }, + { + "epoch": 49.01, + "learning_rate": 0.0002737777777777778, + "loss": 0.208, + "step": 2539 + }, + { + "epoch": 49.01, + "learning_rate": 0.0002736666666666667, + "loss": 0.1541, + "step": 2540 + }, + { + "epoch": 49.01, + "learning_rate": 0.00027355555555555553, + "loss": 0.1841, + "step": 2541 + }, + { + "epoch": 49.01, + "learning_rate": 0.00027344444444444443, + "loss": 0.2026, + "step": 2542 + }, + { + "epoch": 49.01, + "learning_rate": 0.00027333333333333333, + "loss": 0.2016, + "step": 2543 + }, + { + "epoch": 49.01, + "learning_rate": 0.0002732222222222222, + "loss": 0.1665, + "step": 2544 + }, + { + "epoch": 49.01, + "learning_rate": 0.0002731111111111111, + "loss": 0.2146, + "step": 2545 + }, + { + "epoch": 49.01, + "learning_rate": 0.000273, + "loss": 0.1539, + "step": 2546 + }, + { + "epoch": 49.01, + "learning_rate": 0.0002728888888888889, + "loss": 0.1931, + "step": 2547 + }, + { + "epoch": 49.01, + "learning_rate": 0.00027277777777777776, + "loss": 0.1979, + "step": 2548 + }, + { + "epoch": 49.01, + "learning_rate": 0.00027266666666666666, + "loss": 0.2134, + "step": 2549 + }, + { + "epoch": 49.01, + "learning_rate": 0.00027255555555555556, + "loss": 0.1752, + "step": 2550 + }, + { + "epoch": 50.0, + "learning_rate": 0.00027244444444444446, + "loss": 0.2986, + "step": 2551 + }, + { + "epoch": 50.0, + "learning_rate": 0.0002723333333333333, + "loss": 0.1762, + "step": 2552 + }, + { + "epoch": 50.0, + "learning_rate": 0.0002722222222222222, + "loss": 0.1369, + "step": 2553 + }, + { + "epoch": 50.0, + "learning_rate": 0.00027211111111111115, + "loss": 0.1935, + "step": 2554 + }, + { + "epoch": 50.0, + "learning_rate": 0.00027200000000000005, + "loss": 0.1753, + "step": 2555 + }, + { + "epoch": 50.0, + "learning_rate": 0.0002718888888888889, + "loss": 0.1898, + "step": 2556 + }, + { + "epoch": 50.0, + "learning_rate": 0.0002717777777777778, + "loss": 0.2051, + "step": 2557 + }, + { + "epoch": 50.0, + "learning_rate": 0.0002716666666666667, + "loss": 0.1871, + "step": 2558 + }, + { + "epoch": 50.0, + "learning_rate": 0.00027155555555555554, + "loss": 0.1587, + "step": 2559 + }, + { + "epoch": 50.0, + "learning_rate": 0.00027144444444444443, + "loss": 0.2331, + "step": 2560 + }, + { + "epoch": 50.0, + "learning_rate": 0.00027133333333333333, + "loss": 0.1976, + "step": 2561 + }, + { + "epoch": 50.0, + "learning_rate": 0.0002712222222222222, + "loss": 0.1779, + "step": 2562 + }, + { + "epoch": 50.0, + "learning_rate": 0.00027111111111111113, + "loss": 0.2577, + "step": 2563 + }, + { + "epoch": 50.0, + "learning_rate": 0.00027100000000000003, + "loss": 0.1831, + "step": 2564 + }, + { + "epoch": 50.0, + "learning_rate": 0.0002708888888888889, + "loss": 0.2031, + "step": 2565 + }, + { + "epoch": 50.0, + "learning_rate": 0.00027077777777777777, + "loss": 0.2171, + "step": 2566 + }, + { + "epoch": 50.0, + "learning_rate": 0.00027066666666666667, + "loss": 0.1956, + "step": 2567 + }, + { + "epoch": 50.0, + "learning_rate": 0.00027055555555555557, + "loss": 0.1942, + "step": 2568 + }, + { + "epoch": 50.0, + "learning_rate": 0.0002704444444444444, + "loss": 0.2098, + "step": 2569 + }, + { + "epoch": 50.0, + "learning_rate": 0.0002703333333333333, + "loss": 0.2307, + "step": 2570 + }, + { + "epoch": 50.0, + "learning_rate": 0.0002702222222222222, + "loss": 0.229, + "step": 2571 + }, + { + "epoch": 50.0, + "learning_rate": 0.00027011111111111116, + "loss": 0.2049, + "step": 2572 + }, + { + "epoch": 50.0, + "learning_rate": 0.00027, + "loss": 0.1943, + "step": 2573 + }, + { + "epoch": 50.0, + "learning_rate": 0.0002698888888888889, + "loss": 0.2032, + "step": 2574 + }, + { + "epoch": 50.01, + "learning_rate": 0.0002697777777777778, + "loss": 0.2067, + "step": 2575 + }, + { + "epoch": 50.01, + "learning_rate": 0.0002696666666666667, + "loss": 0.1543, + "step": 2576 + }, + { + "epoch": 50.01, + "learning_rate": 0.00026955555555555554, + "loss": 0.1993, + "step": 2577 + }, + { + "epoch": 50.01, + "learning_rate": 0.00026944444444444444, + "loss": 0.2261, + "step": 2578 + }, + { + "epoch": 50.01, + "learning_rate": 0.00026933333333333334, + "loss": 0.1952, + "step": 2579 + }, + { + "epoch": 50.01, + "learning_rate": 0.0002692222222222222, + "loss": 0.2051, + "step": 2580 + }, + { + "epoch": 50.01, + "learning_rate": 0.00026911111111111114, + "loss": 0.1862, + "step": 2581 + }, + { + "epoch": 50.01, + "learning_rate": 0.00026900000000000003, + "loss": 0.2016, + "step": 2582 + }, + { + "epoch": 50.01, + "learning_rate": 0.00026888888888888893, + "loss": 0.1517, + "step": 2583 + }, + { + "epoch": 50.01, + "learning_rate": 0.0002687777777777778, + "loss": 0.1852, + "step": 2584 + }, + { + "epoch": 50.01, + "learning_rate": 0.0002686666666666667, + "loss": 0.2102, + "step": 2585 + }, + { + "epoch": 50.01, + "learning_rate": 0.00026855555555555557, + "loss": 0.1681, + "step": 2586 + }, + { + "epoch": 50.01, + "learning_rate": 0.0002684444444444444, + "loss": 0.1894, + "step": 2587 + }, + { + "epoch": 50.01, + "learning_rate": 0.0002683333333333333, + "loss": 0.204, + "step": 2588 + }, + { + "epoch": 50.01, + "learning_rate": 0.0002682222222222222, + "loss": 0.2399, + "step": 2589 + }, + { + "epoch": 50.01, + "learning_rate": 0.00026811111111111117, + "loss": 0.1902, + "step": 2590 + }, + { + "epoch": 50.01, + "learning_rate": 0.000268, + "loss": 0.2339, + "step": 2591 + }, + { + "epoch": 50.01, + "learning_rate": 0.0002678888888888889, + "loss": 0.1611, + "step": 2592 + }, + { + "epoch": 50.01, + "learning_rate": 0.0002677777777777778, + "loss": 0.2161, + "step": 2593 + }, + { + "epoch": 50.01, + "learning_rate": 0.00026766666666666665, + "loss": 0.1853, + "step": 2594 + }, + { + "epoch": 50.01, + "learning_rate": 0.00026755555555555555, + "loss": 0.1648, + "step": 2595 + }, + { + "epoch": 50.01, + "learning_rate": 0.00026744444444444445, + "loss": 0.1952, + "step": 2596 + }, + { + "epoch": 50.01, + "learning_rate": 0.00026733333333333334, + "loss": 0.2308, + "step": 2597 + }, + { + "epoch": 50.01, + "learning_rate": 0.00026722222222222224, + "loss": 0.1888, + "step": 2598 + }, + { + "epoch": 50.01, + "learning_rate": 0.00026711111111111114, + "loss": 0.2142, + "step": 2599 + }, + { + "epoch": 50.01, + "learning_rate": 0.00026700000000000004, + "loss": 0.2008, + "step": 2600 + }, + { + "epoch": 50.01, + "learning_rate": 0.0002668888888888889, + "loss": 0.2228, + "step": 2601 + }, + { + "epoch": 51.0, + "learning_rate": 0.0002667777777777778, + "loss": 0.2747, + "step": 2602 + }, + { + "epoch": 51.0, + "learning_rate": 0.0002666666666666667, + "loss": 0.1755, + "step": 2603 + }, + { + "epoch": 51.0, + "learning_rate": 0.0002665555555555556, + "loss": 0.1509, + "step": 2604 + }, + { + "epoch": 51.0, + "learning_rate": 0.0002664444444444444, + "loss": 0.1947, + "step": 2605 + }, + { + "epoch": 51.0, + "learning_rate": 0.0002663333333333333, + "loss": 0.1446, + "step": 2606 + }, + { + "epoch": 51.0, + "learning_rate": 0.00026622222222222227, + "loss": 0.1667, + "step": 2607 + }, + { + "epoch": 51.0, + "learning_rate": 0.0002661111111111111, + "loss": 0.2015, + "step": 2608 + }, + { + "epoch": 51.0, + "learning_rate": 0.000266, + "loss": 0.2031, + "step": 2609 + }, + { + "epoch": 51.0, + "learning_rate": 0.0002658888888888889, + "loss": 0.1806, + "step": 2610 + }, + { + "epoch": 51.0, + "learning_rate": 0.0002657777777777778, + "loss": 0.1892, + "step": 2611 + }, + { + "epoch": 51.0, + "learning_rate": 0.00026566666666666666, + "loss": 0.1904, + "step": 2612 + }, + { + "epoch": 51.0, + "learning_rate": 0.00026555555555555555, + "loss": 0.189, + "step": 2613 + }, + { + "epoch": 51.0, + "learning_rate": 0.00026544444444444445, + "loss": 0.2002, + "step": 2614 + }, + { + "epoch": 51.0, + "learning_rate": 0.0002653333333333333, + "loss": 0.1776, + "step": 2615 + }, + { + "epoch": 51.0, + "learning_rate": 0.00026522222222222225, + "loss": 0.2237, + "step": 2616 + }, + { + "epoch": 51.0, + "learning_rate": 0.00026511111111111115, + "loss": 0.212, + "step": 2617 + }, + { + "epoch": 51.0, + "learning_rate": 0.00026500000000000004, + "loss": 0.1699, + "step": 2618 + }, + { + "epoch": 51.0, + "learning_rate": 0.0002648888888888889, + "loss": 0.1712, + "step": 2619 + }, + { + "epoch": 51.0, + "learning_rate": 0.0002647777777777778, + "loss": 0.1777, + "step": 2620 + }, + { + "epoch": 51.0, + "learning_rate": 0.0002646666666666667, + "loss": 0.1633, + "step": 2621 + }, + { + "epoch": 51.0, + "learning_rate": 0.00026455555555555553, + "loss": 0.1955, + "step": 2622 + }, + { + "epoch": 51.0, + "learning_rate": 0.00026444444444444443, + "loss": 0.1941, + "step": 2623 + }, + { + "epoch": 51.0, + "learning_rate": 0.0002643333333333333, + "loss": 0.2093, + "step": 2624 + }, + { + "epoch": 51.0, + "learning_rate": 0.0002642222222222223, + "loss": 0.2077, + "step": 2625 + }, + { + "epoch": 51.01, + "learning_rate": 0.0002641111111111111, + "loss": 0.1772, + "step": 2626 + }, + { + "epoch": 51.01, + "learning_rate": 0.000264, + "loss": 0.2027, + "step": 2627 + }, + { + "epoch": 51.01, + "learning_rate": 0.0002638888888888889, + "loss": 0.2188, + "step": 2628 + }, + { + "epoch": 51.01, + "learning_rate": 0.00026377777777777776, + "loss": 0.2479, + "step": 2629 + }, + { + "epoch": 51.01, + "learning_rate": 0.00026366666666666666, + "loss": 0.2022, + "step": 2630 + }, + { + "epoch": 51.01, + "learning_rate": 0.00026355555555555556, + "loss": 0.1454, + "step": 2631 + }, + { + "epoch": 51.01, + "learning_rate": 0.00026344444444444446, + "loss": 0.1927, + "step": 2632 + }, + { + "epoch": 51.01, + "learning_rate": 0.0002633333333333333, + "loss": 0.1884, + "step": 2633 + }, + { + "epoch": 51.01, + "learning_rate": 0.00026322222222222225, + "loss": 0.1642, + "step": 2634 + }, + { + "epoch": 51.01, + "learning_rate": 0.00026311111111111115, + "loss": 0.2028, + "step": 2635 + }, + { + "epoch": 51.01, + "learning_rate": 0.000263, + "loss": 0.1238, + "step": 2636 + }, + { + "epoch": 51.01, + "learning_rate": 0.0002628888888888889, + "loss": 0.2002, + "step": 2637 + }, + { + "epoch": 51.01, + "learning_rate": 0.0002627777777777778, + "loss": 0.2132, + "step": 2638 + }, + { + "epoch": 51.01, + "learning_rate": 0.0002626666666666667, + "loss": 0.1932, + "step": 2639 + }, + { + "epoch": 51.01, + "learning_rate": 0.00026255555555555553, + "loss": 0.229, + "step": 2640 + }, + { + "epoch": 51.01, + "learning_rate": 0.00026244444444444443, + "loss": 0.1781, + "step": 2641 + }, + { + "epoch": 51.01, + "learning_rate": 0.00026233333333333333, + "loss": 0.2256, + "step": 2642 + }, + { + "epoch": 51.01, + "learning_rate": 0.00026222222222222223, + "loss": 0.1531, + "step": 2643 + }, + { + "epoch": 51.01, + "learning_rate": 0.00026211111111111113, + "loss": 0.1813, + "step": 2644 + }, + { + "epoch": 51.01, + "learning_rate": 0.000262, + "loss": 0.2269, + "step": 2645 + }, + { + "epoch": 51.01, + "learning_rate": 0.0002618888888888889, + "loss": 0.1797, + "step": 2646 + }, + { + "epoch": 51.01, + "learning_rate": 0.00026177777777777777, + "loss": 0.1826, + "step": 2647 + }, + { + "epoch": 51.01, + "learning_rate": 0.00026166666666666667, + "loss": 0.1307, + "step": 2648 + }, + { + "epoch": 51.01, + "learning_rate": 0.00026155555555555556, + "loss": 0.1983, + "step": 2649 + }, + { + "epoch": 51.01, + "learning_rate": 0.0002614444444444444, + "loss": 0.1805, + "step": 2650 + }, + { + "epoch": 51.01, + "learning_rate": 0.0002613333333333333, + "loss": 0.1822, + "step": 2651 + }, + { + "epoch": 51.01, + "learning_rate": 0.00026122222222222226, + "loss": 0.165, + "step": 2652 + }, + { + "epoch": 52.0, + "learning_rate": 0.00026111111111111116, + "loss": 0.2894, + "step": 2653 + }, + { + "epoch": 52.0, + "learning_rate": 0.000261, + "loss": 0.1535, + "step": 2654 + }, + { + "epoch": 52.0, + "learning_rate": 0.0002608888888888889, + "loss": 0.1807, + "step": 2655 + }, + { + "epoch": 52.0, + "learning_rate": 0.0002607777777777778, + "loss": 0.172, + "step": 2656 + }, + { + "epoch": 52.0, + "learning_rate": 0.00026066666666666664, + "loss": 0.1473, + "step": 2657 + }, + { + "epoch": 52.0, + "learning_rate": 0.00026055555555555554, + "loss": 0.1676, + "step": 2658 + }, + { + "epoch": 52.0, + "learning_rate": 0.00026044444444444444, + "loss": 0.1976, + "step": 2659 + }, + { + "epoch": 52.0, + "learning_rate": 0.0002603333333333334, + "loss": 0.1664, + "step": 2660 + }, + { + "epoch": 52.0, + "learning_rate": 0.00026022222222222223, + "loss": 0.1688, + "step": 2661 + }, + { + "epoch": 52.0, + "learning_rate": 0.00026011111111111113, + "loss": 0.1785, + "step": 2662 + }, + { + "epoch": 52.0, + "learning_rate": 0.00026000000000000003, + "loss": 0.1607, + "step": 2663 + }, + { + "epoch": 52.0, + "learning_rate": 0.0002598888888888889, + "loss": 0.1477, + "step": 2664 + }, + { + "epoch": 52.0, + "learning_rate": 0.0002597777777777778, + "loss": 0.1671, + "step": 2665 + }, + { + "epoch": 52.0, + "learning_rate": 0.00025966666666666667, + "loss": 0.1862, + "step": 2666 + }, + { + "epoch": 52.0, + "learning_rate": 0.00025955555555555557, + "loss": 0.2077, + "step": 2667 + }, + { + "epoch": 52.0, + "learning_rate": 0.0002594444444444444, + "loss": 0.2021, + "step": 2668 + }, + { + "epoch": 52.0, + "learning_rate": 0.00025933333333333337, + "loss": 0.2429, + "step": 2669 + }, + { + "epoch": 52.0, + "learning_rate": 0.00025922222222222226, + "loss": 0.1298, + "step": 2670 + }, + { + "epoch": 52.0, + "learning_rate": 0.0002591111111111111, + "loss": 0.1705, + "step": 2671 + }, + { + "epoch": 52.0, + "learning_rate": 0.000259, + "loss": 0.176, + "step": 2672 + }, + { + "epoch": 52.0, + "learning_rate": 0.0002588888888888889, + "loss": 0.1488, + "step": 2673 + }, + { + "epoch": 52.0, + "learning_rate": 0.0002587777777777778, + "loss": 0.1805, + "step": 2674 + }, + { + "epoch": 52.0, + "learning_rate": 0.00025866666666666665, + "loss": 0.1685, + "step": 2675 + }, + { + "epoch": 52.0, + "learning_rate": 0.00025855555555555555, + "loss": 0.1531, + "step": 2676 + }, + { + "epoch": 52.01, + "learning_rate": 0.00025844444444444444, + "loss": 0.2198, + "step": 2677 + }, + { + "epoch": 52.01, + "learning_rate": 0.00025833333333333334, + "loss": 0.1717, + "step": 2678 + }, + { + "epoch": 52.01, + "learning_rate": 0.00025822222222222224, + "loss": 0.1851, + "step": 2679 + }, + { + "epoch": 52.01, + "learning_rate": 0.00025811111111111114, + "loss": 0.1965, + "step": 2680 + }, + { + "epoch": 52.01, + "learning_rate": 0.00025800000000000004, + "loss": 0.1598, + "step": 2681 + }, + { + "epoch": 52.01, + "learning_rate": 0.0002578888888888889, + "loss": 0.1476, + "step": 2682 + }, + { + "epoch": 52.01, + "learning_rate": 0.0002577777777777778, + "loss": 0.1563, + "step": 2683 + }, + { + "epoch": 52.01, + "learning_rate": 0.0002576666666666667, + "loss": 0.1997, + "step": 2684 + }, + { + "epoch": 52.01, + "learning_rate": 0.0002575555555555555, + "loss": 0.1877, + "step": 2685 + }, + { + "epoch": 52.01, + "learning_rate": 0.0002574444444444444, + "loss": 0.2184, + "step": 2686 + }, + { + "epoch": 52.01, + "learning_rate": 0.00025733333333333337, + "loss": 0.1859, + "step": 2687 + }, + { + "epoch": 52.01, + "learning_rate": 0.00025722222222222227, + "loss": 0.1906, + "step": 2688 + }, + { + "epoch": 52.01, + "learning_rate": 0.0002571111111111111, + "loss": 0.2006, + "step": 2689 + }, + { + "epoch": 52.01, + "learning_rate": 0.000257, + "loss": 0.1894, + "step": 2690 + }, + { + "epoch": 52.01, + "learning_rate": 0.0002568888888888889, + "loss": 0.1681, + "step": 2691 + }, + { + "epoch": 52.01, + "learning_rate": 0.00025677777777777775, + "loss": 0.216, + "step": 2692 + }, + { + "epoch": 52.01, + "learning_rate": 0.00025666666666666665, + "loss": 0.1933, + "step": 2693 + }, + { + "epoch": 52.01, + "learning_rate": 0.00025655555555555555, + "loss": 0.1744, + "step": 2694 + }, + { + "epoch": 52.01, + "learning_rate": 0.00025644444444444445, + "loss": 0.173, + "step": 2695 + }, + { + "epoch": 52.01, + "learning_rate": 0.00025633333333333335, + "loss": 0.1914, + "step": 2696 + }, + { + "epoch": 52.01, + "learning_rate": 0.00025622222222222225, + "loss": 0.1814, + "step": 2697 + }, + { + "epoch": 52.01, + "learning_rate": 0.00025611111111111114, + "loss": 0.2071, + "step": 2698 + }, + { + "epoch": 52.01, + "learning_rate": 0.000256, + "loss": 0.181, + "step": 2699 + }, + { + "epoch": 52.01, + "learning_rate": 0.0002558888888888889, + "loss": 0.126, + "step": 2700 + }, + { + "epoch": 52.01, + "learning_rate": 0.0002557777777777778, + "loss": 0.129, + "step": 2701 + }, + { + "epoch": 52.01, + "learning_rate": 0.0002556666666666667, + "loss": 0.1788, + "step": 2702 + }, + { + "epoch": 52.01, + "learning_rate": 0.00025555555555555553, + "loss": 0.1941, + "step": 2703 + }, + { + "epoch": 53.0, + "learning_rate": 0.0002554444444444444, + "loss": 0.2944, + "step": 2704 + }, + { + "epoch": 53.0, + "learning_rate": 0.0002553333333333334, + "loss": 0.1934, + "step": 2705 + }, + { + "epoch": 53.0, + "learning_rate": 0.0002552222222222222, + "loss": 0.1715, + "step": 2706 + }, + { + "epoch": 53.0, + "learning_rate": 0.0002551111111111111, + "loss": 0.1451, + "step": 2707 + }, + { + "epoch": 53.0, + "learning_rate": 0.000255, + "loss": 0.1755, + "step": 2708 + }, + { + "epoch": 53.0, + "learning_rate": 0.0002548888888888889, + "loss": 0.1454, + "step": 2709 + }, + { + "epoch": 53.0, + "learning_rate": 0.00025477777777777776, + "loss": 0.186, + "step": 2710 + }, + { + "epoch": 53.0, + "learning_rate": 0.00025466666666666666, + "loss": 0.1718, + "step": 2711 + }, + { + "epoch": 53.0, + "learning_rate": 0.00025455555555555556, + "loss": 0.1771, + "step": 2712 + }, + { + "epoch": 53.0, + "learning_rate": 0.0002544444444444444, + "loss": 0.1733, + "step": 2713 + }, + { + "epoch": 53.0, + "learning_rate": 0.00025433333333333335, + "loss": 0.1079, + "step": 2714 + }, + { + "epoch": 53.0, + "learning_rate": 0.00025422222222222225, + "loss": 0.1586, + "step": 2715 + }, + { + "epoch": 53.0, + "learning_rate": 0.00025411111111111115, + "loss": 0.1739, + "step": 2716 + }, + { + "epoch": 53.0, + "learning_rate": 0.000254, + "loss": 0.201, + "step": 2717 + }, + { + "epoch": 53.0, + "learning_rate": 0.0002538888888888889, + "loss": 0.1648, + "step": 2718 + }, + { + "epoch": 53.0, + "learning_rate": 0.0002537777777777778, + "loss": 0.1702, + "step": 2719 + }, + { + "epoch": 53.0, + "learning_rate": 0.00025366666666666663, + "loss": 0.1686, + "step": 2720 + }, + { + "epoch": 53.0, + "learning_rate": 0.00025355555555555553, + "loss": 0.1818, + "step": 2721 + }, + { + "epoch": 53.0, + "learning_rate": 0.0002534444444444445, + "loss": 0.1409, + "step": 2722 + }, + { + "epoch": 53.0, + "learning_rate": 0.0002533333333333334, + "loss": 0.1742, + "step": 2723 + }, + { + "epoch": 53.0, + "learning_rate": 0.00025322222222222223, + "loss": 0.1829, + "step": 2724 + }, + { + "epoch": 53.0, + "learning_rate": 0.0002531111111111111, + "loss": 0.1711, + "step": 2725 + }, + { + "epoch": 53.0, + "learning_rate": 0.000253, + "loss": 0.1873, + "step": 2726 + }, + { + "epoch": 53.0, + "learning_rate": 0.00025288888888888887, + "loss": 0.1792, + "step": 2727 + }, + { + "epoch": 53.01, + "learning_rate": 0.00025277777777777777, + "loss": 0.1959, + "step": 2728 + }, + { + "epoch": 53.01, + "learning_rate": 0.00025266666666666666, + "loss": 0.2402, + "step": 2729 + }, + { + "epoch": 53.01, + "learning_rate": 0.00025255555555555556, + "loss": 0.1574, + "step": 2730 + }, + { + "epoch": 53.01, + "learning_rate": 0.00025244444444444446, + "loss": 0.1729, + "step": 2731 + }, + { + "epoch": 53.01, + "learning_rate": 0.00025233333333333336, + "loss": 0.1813, + "step": 2732 + }, + { + "epoch": 53.01, + "learning_rate": 0.00025222222222222226, + "loss": 0.1608, + "step": 2733 + }, + { + "epoch": 53.01, + "learning_rate": 0.0002521111111111111, + "loss": 0.1805, + "step": 2734 + }, + { + "epoch": 53.01, + "learning_rate": 0.000252, + "loss": 0.1824, + "step": 2735 + }, + { + "epoch": 53.01, + "learning_rate": 0.0002518888888888889, + "loss": 0.166, + "step": 2736 + }, + { + "epoch": 53.01, + "learning_rate": 0.0002517777777777778, + "loss": 0.1488, + "step": 2737 + }, + { + "epoch": 53.01, + "learning_rate": 0.00025166666666666664, + "loss": 0.1407, + "step": 2738 + }, + { + "epoch": 53.01, + "learning_rate": 0.00025155555555555554, + "loss": 0.1549, + "step": 2739 + }, + { + "epoch": 53.01, + "learning_rate": 0.0002514444444444445, + "loss": 0.2257, + "step": 2740 + }, + { + "epoch": 53.01, + "learning_rate": 0.00025133333333333333, + "loss": 0.1959, + "step": 2741 + }, + { + "epoch": 53.01, + "learning_rate": 0.00025122222222222223, + "loss": 0.1683, + "step": 2742 + }, + { + "epoch": 53.01, + "learning_rate": 0.00025111111111111113, + "loss": 0.1776, + "step": 2743 + }, + { + "epoch": 53.01, + "learning_rate": 0.00025100000000000003, + "loss": 0.1931, + "step": 2744 + }, + { + "epoch": 53.01, + "learning_rate": 0.0002508888888888889, + "loss": 0.1694, + "step": 2745 + }, + { + "epoch": 53.01, + "learning_rate": 0.00025077777777777777, + "loss": 0.1803, + "step": 2746 + }, + { + "epoch": 53.01, + "learning_rate": 0.00025066666666666667, + "loss": 0.1778, + "step": 2747 + }, + { + "epoch": 53.01, + "learning_rate": 0.0002505555555555555, + "loss": 0.1799, + "step": 2748 + }, + { + "epoch": 53.01, + "learning_rate": 0.00025044444444444447, + "loss": 0.1891, + "step": 2749 + }, + { + "epoch": 53.01, + "learning_rate": 0.00025033333333333336, + "loss": 0.1469, + "step": 2750 + }, + { + "epoch": 53.01, + "learning_rate": 0.00025022222222222226, + "loss": 0.1382, + "step": 2751 + }, + { + "epoch": 53.01, + "learning_rate": 0.0002501111111111111, + "loss": 0.1717, + "step": 2752 + }, + { + "epoch": 53.01, + "learning_rate": 0.00025, + "loss": 0.1826, + "step": 2753 + }, + { + "epoch": 53.01, + "learning_rate": 0.0002498888888888889, + "loss": 0.1748, + "step": 2754 + }, + { + "epoch": 54.0, + "learning_rate": 0.00024977777777777775, + "loss": 0.2711, + "step": 2755 + }, + { + "epoch": 54.0, + "learning_rate": 0.0002496666666666667, + "loss": 0.1851, + "step": 2756 + }, + { + "epoch": 54.0, + "learning_rate": 0.00024955555555555554, + "loss": 0.1602, + "step": 2757 + }, + { + "epoch": 54.0, + "learning_rate": 0.00024944444444444444, + "loss": 0.2076, + "step": 2758 + }, + { + "epoch": 54.0, + "learning_rate": 0.00024933333333333334, + "loss": 0.1225, + "step": 2759 + }, + { + "epoch": 54.0, + "learning_rate": 0.00024922222222222224, + "loss": 0.1644, + "step": 2760 + }, + { + "epoch": 54.0, + "learning_rate": 0.00024911111111111114, + "loss": 0.1526, + "step": 2761 + }, + { + "epoch": 54.0, + "learning_rate": 0.000249, + "loss": 0.1856, + "step": 2762 + }, + { + "epoch": 54.0, + "learning_rate": 0.0002488888888888889, + "loss": 0.1968, + "step": 2763 + }, + { + "epoch": 54.0, + "learning_rate": 0.0002487777777777778, + "loss": 0.1562, + "step": 2764 + }, + { + "epoch": 54.0, + "learning_rate": 0.0002486666666666667, + "loss": 0.1796, + "step": 2765 + }, + { + "epoch": 54.0, + "learning_rate": 0.0002485555555555556, + "loss": 0.1721, + "step": 2766 + }, + { + "epoch": 54.0, + "learning_rate": 0.00024844444444444447, + "loss": 0.1207, + "step": 2767 + }, + { + "epoch": 54.0, + "learning_rate": 0.0002483333333333333, + "loss": 0.1121, + "step": 2768 + }, + { + "epoch": 54.0, + "learning_rate": 0.0002482222222222222, + "loss": 0.1244, + "step": 2769 + }, + { + "epoch": 54.0, + "learning_rate": 0.0002481111111111111, + "loss": 0.1804, + "step": 2770 + }, + { + "epoch": 54.0, + "learning_rate": 0.000248, + "loss": 0.1492, + "step": 2771 + }, + { + "epoch": 54.0, + "learning_rate": 0.0002478888888888889, + "loss": 0.1975, + "step": 2772 + }, + { + "epoch": 54.0, + "learning_rate": 0.0002477777777777778, + "loss": 0.1647, + "step": 2773 + }, + { + "epoch": 54.0, + "learning_rate": 0.0002476666666666667, + "loss": 0.1852, + "step": 2774 + }, + { + "epoch": 54.0, + "learning_rate": 0.00024755555555555555, + "loss": 0.1665, + "step": 2775 + }, + { + "epoch": 54.0, + "learning_rate": 0.00024744444444444445, + "loss": 0.166, + "step": 2776 + }, + { + "epoch": 54.0, + "learning_rate": 0.00024733333333333335, + "loss": 0.155, + "step": 2777 + }, + { + "epoch": 54.0, + "learning_rate": 0.00024722222222222224, + "loss": 0.1348, + "step": 2778 + }, + { + "epoch": 54.01, + "learning_rate": 0.00024711111111111114, + "loss": 0.1595, + "step": 2779 + }, + { + "epoch": 54.01, + "learning_rate": 0.000247, + "loss": 0.1546, + "step": 2780 + }, + { + "epoch": 54.01, + "learning_rate": 0.0002468888888888889, + "loss": 0.1613, + "step": 2781 + }, + { + "epoch": 54.01, + "learning_rate": 0.0002467777777777778, + "loss": 0.1937, + "step": 2782 + }, + { + "epoch": 54.01, + "learning_rate": 0.0002466666666666667, + "loss": 0.1894, + "step": 2783 + }, + { + "epoch": 54.01, + "learning_rate": 0.0002465555555555556, + "loss": 0.1891, + "step": 2784 + }, + { + "epoch": 54.01, + "learning_rate": 0.0002464444444444444, + "loss": 0.1881, + "step": 2785 + }, + { + "epoch": 54.01, + "learning_rate": 0.0002463333333333333, + "loss": 0.1757, + "step": 2786 + }, + { + "epoch": 54.01, + "learning_rate": 0.0002462222222222222, + "loss": 0.1724, + "step": 2787 + }, + { + "epoch": 54.01, + "learning_rate": 0.0002461111111111111, + "loss": 0.1482, + "step": 2788 + }, + { + "epoch": 54.01, + "learning_rate": 0.000246, + "loss": 0.2089, + "step": 2789 + }, + { + "epoch": 54.01, + "learning_rate": 0.00024588888888888886, + "loss": 0.1558, + "step": 2790 + }, + { + "epoch": 54.01, + "learning_rate": 0.0002457777777777778, + "loss": 0.2, + "step": 2791 + }, + { + "epoch": 54.01, + "learning_rate": 0.00024566666666666666, + "loss": 0.1294, + "step": 2792 + }, + { + "epoch": 54.01, + "learning_rate": 0.00024555555555555556, + "loss": 0.1569, + "step": 2793 + }, + { + "epoch": 54.01, + "learning_rate": 0.00024544444444444445, + "loss": 0.208, + "step": 2794 + }, + { + "epoch": 54.01, + "learning_rate": 0.00024533333333333335, + "loss": 0.141, + "step": 2795 + }, + { + "epoch": 54.01, + "learning_rate": 0.00024522222222222225, + "loss": 0.1738, + "step": 2796 + }, + { + "epoch": 54.01, + "learning_rate": 0.0002451111111111111, + "loss": 0.1663, + "step": 2797 + }, + { + "epoch": 54.01, + "learning_rate": 0.000245, + "loss": 0.1079, + "step": 2798 + }, + { + "epoch": 54.01, + "learning_rate": 0.0002448888888888889, + "loss": 0.1842, + "step": 2799 + }, + { + "epoch": 54.01, + "learning_rate": 0.0002447777777777778, + "loss": 0.1595, + "step": 2800 + }, + { + "epoch": 54.01, + "learning_rate": 0.0002446666666666667, + "loss": 0.1857, + "step": 2801 + }, + { + "epoch": 54.01, + "learning_rate": 0.0002445555555555556, + "loss": 0.1781, + "step": 2802 + }, + { + "epoch": 54.01, + "learning_rate": 0.00024444444444444443, + "loss": 0.1499, + "step": 2803 + }, + { + "epoch": 54.01, + "learning_rate": 0.00024433333333333333, + "loss": 0.2196, + "step": 2804 + }, + { + "epoch": 54.01, + "learning_rate": 0.0002442222222222222, + "loss": 0.2036, + "step": 2805 + }, + { + "epoch": 55.0, + "learning_rate": 0.00024411111111111112, + "loss": 0.2372, + "step": 2806 + }, + { + "epoch": 55.0, + "learning_rate": 0.000244, + "loss": 0.1747, + "step": 2807 + }, + { + "epoch": 55.0, + "learning_rate": 0.0002438888888888889, + "loss": 0.152, + "step": 2808 + }, + { + "epoch": 55.0, + "learning_rate": 0.0002437777777777778, + "loss": 0.176, + "step": 2809 + }, + { + "epoch": 55.0, + "learning_rate": 0.00024366666666666666, + "loss": 0.1861, + "step": 2810 + }, + { + "epoch": 55.0, + "learning_rate": 0.00024355555555555556, + "loss": 0.1629, + "step": 2811 + }, + { + "epoch": 55.0, + "learning_rate": 0.00024344444444444443, + "loss": 0.1535, + "step": 2812 + }, + { + "epoch": 55.0, + "learning_rate": 0.00024333333333333336, + "loss": 0.1599, + "step": 2813 + }, + { + "epoch": 55.0, + "learning_rate": 0.00024322222222222223, + "loss": 0.1663, + "step": 2814 + }, + { + "epoch": 55.0, + "learning_rate": 0.00024311111111111113, + "loss": 0.15, + "step": 2815 + }, + { + "epoch": 55.0, + "learning_rate": 0.000243, + "loss": 0.1507, + "step": 2816 + }, + { + "epoch": 55.0, + "learning_rate": 0.00024288888888888887, + "loss": 0.1527, + "step": 2817 + }, + { + "epoch": 55.0, + "learning_rate": 0.0002427777777777778, + "loss": 0.1355, + "step": 2818 + }, + { + "epoch": 55.0, + "learning_rate": 0.00024266666666666667, + "loss": 0.1616, + "step": 2819 + }, + { + "epoch": 55.0, + "learning_rate": 0.00024255555555555556, + "loss": 0.1609, + "step": 2820 + }, + { + "epoch": 55.0, + "learning_rate": 0.00024244444444444443, + "loss": 0.1898, + "step": 2821 + }, + { + "epoch": 55.0, + "learning_rate": 0.00024233333333333336, + "loss": 0.1721, + "step": 2822 + }, + { + "epoch": 55.0, + "learning_rate": 0.00024222222222222223, + "loss": 0.1791, + "step": 2823 + }, + { + "epoch": 55.0, + "learning_rate": 0.0002421111111111111, + "loss": 0.132, + "step": 2824 + }, + { + "epoch": 55.0, + "learning_rate": 0.000242, + "loss": 0.1573, + "step": 2825 + }, + { + "epoch": 55.0, + "learning_rate": 0.00024188888888888887, + "loss": 0.1415, + "step": 2826 + }, + { + "epoch": 55.0, + "learning_rate": 0.0002417777777777778, + "loss": 0.1997, + "step": 2827 + }, + { + "epoch": 55.0, + "learning_rate": 0.00024166666666666667, + "loss": 0.1661, + "step": 2828 + }, + { + "epoch": 55.0, + "learning_rate": 0.00024155555555555557, + "loss": 0.1593, + "step": 2829 + }, + { + "epoch": 55.01, + "learning_rate": 0.00024144444444444444, + "loss": 0.1549, + "step": 2830 + }, + { + "epoch": 55.01, + "learning_rate": 0.00024133333333333334, + "loss": 0.1862, + "step": 2831 + }, + { + "epoch": 55.01, + "learning_rate": 0.00024122222222222223, + "loss": 0.1456, + "step": 2832 + }, + { + "epoch": 55.01, + "learning_rate": 0.0002411111111111111, + "loss": 0.1717, + "step": 2833 + }, + { + "epoch": 55.01, + "learning_rate": 0.000241, + "loss": 0.1527, + "step": 2834 + }, + { + "epoch": 55.01, + "learning_rate": 0.0002408888888888889, + "loss": 0.1649, + "step": 2835 + }, + { + "epoch": 55.01, + "learning_rate": 0.0002407777777777778, + "loss": 0.1947, + "step": 2836 + }, + { + "epoch": 55.01, + "learning_rate": 0.00024066666666666667, + "loss": 0.1555, + "step": 2837 + }, + { + "epoch": 55.01, + "learning_rate": 0.00024055555555555554, + "loss": 0.1352, + "step": 2838 + }, + { + "epoch": 55.01, + "learning_rate": 0.00024044444444444444, + "loss": 0.2074, + "step": 2839 + }, + { + "epoch": 55.01, + "learning_rate": 0.00024033333333333334, + "loss": 0.1581, + "step": 2840 + }, + { + "epoch": 55.01, + "learning_rate": 0.00024022222222222224, + "loss": 0.1672, + "step": 2841 + }, + { + "epoch": 55.01, + "learning_rate": 0.0002401111111111111, + "loss": 0.2196, + "step": 2842 + }, + { + "epoch": 55.01, + "learning_rate": 0.00024, + "loss": 0.2091, + "step": 2843 + }, + { + "epoch": 55.01, + "learning_rate": 0.0002398888888888889, + "loss": 0.1798, + "step": 2844 + }, + { + "epoch": 55.01, + "learning_rate": 0.0002397777777777778, + "loss": 0.196, + "step": 2845 + }, + { + "epoch": 55.01, + "learning_rate": 0.00023966666666666667, + "loss": 0.1636, + "step": 2846 + }, + { + "epoch": 55.01, + "learning_rate": 0.00023955555555555554, + "loss": 0.1976, + "step": 2847 + }, + { + "epoch": 55.01, + "learning_rate": 0.00023944444444444444, + "loss": 0.186, + "step": 2848 + }, + { + "epoch": 55.01, + "learning_rate": 0.00023933333333333334, + "loss": 0.2136, + "step": 2849 + }, + { + "epoch": 55.01, + "learning_rate": 0.00023922222222222224, + "loss": 0.2088, + "step": 2850 + }, + { + "epoch": 55.01, + "learning_rate": 0.0002391111111111111, + "loss": 0.1507, + "step": 2851 + }, + { + "epoch": 55.01, + "learning_rate": 0.00023899999999999998, + "loss": 0.1641, + "step": 2852 + }, + { + "epoch": 55.01, + "learning_rate": 0.0002388888888888889, + "loss": 0.1249, + "step": 2853 + }, + { + "epoch": 55.01, + "learning_rate": 0.00023877777777777778, + "loss": 0.1605, + "step": 2854 + }, + { + "epoch": 55.01, + "learning_rate": 0.00023866666666666668, + "loss": 0.1642, + "step": 2855 + }, + { + "epoch": 55.01, + "learning_rate": 0.00023855555555555555, + "loss": 0.1874, + "step": 2856 + }, + { + "epoch": 56.0, + "learning_rate": 0.00023844444444444445, + "loss": 0.2094, + "step": 2857 + }, + { + "epoch": 56.0, + "learning_rate": 0.00023833333333333334, + "loss": 0.1597, + "step": 2858 + }, + { + "epoch": 56.0, + "learning_rate": 0.00023822222222222224, + "loss": 0.1374, + "step": 2859 + }, + { + "epoch": 56.0, + "learning_rate": 0.0002381111111111111, + "loss": 0.1926, + "step": 2860 + }, + { + "epoch": 56.0, + "learning_rate": 0.00023799999999999998, + "loss": 0.1174, + "step": 2861 + }, + { + "epoch": 56.0, + "learning_rate": 0.0002378888888888889, + "loss": 0.1884, + "step": 2862 + }, + { + "epoch": 56.0, + "learning_rate": 0.00023777777777777778, + "loss": 0.1487, + "step": 2863 + }, + { + "epoch": 56.0, + "learning_rate": 0.00023766666666666668, + "loss": 0.2002, + "step": 2864 + }, + { + "epoch": 56.0, + "learning_rate": 0.00023755555555555555, + "loss": 0.1673, + "step": 2865 + }, + { + "epoch": 56.0, + "learning_rate": 0.00023744444444444448, + "loss": 0.1579, + "step": 2866 + }, + { + "epoch": 56.0, + "learning_rate": 0.00023733333333333335, + "loss": 0.1479, + "step": 2867 + }, + { + "epoch": 56.0, + "learning_rate": 0.00023722222222222222, + "loss": 0.2284, + "step": 2868 + }, + { + "epoch": 56.0, + "learning_rate": 0.00023711111111111112, + "loss": 0.1358, + "step": 2869 + }, + { + "epoch": 56.0, + "learning_rate": 0.000237, + "loss": 0.1587, + "step": 2870 + }, + { + "epoch": 56.0, + "learning_rate": 0.0002368888888888889, + "loss": 0.1669, + "step": 2871 + }, + { + "epoch": 56.0, + "learning_rate": 0.00023677777777777778, + "loss": 0.2064, + "step": 2872 + }, + { + "epoch": 56.0, + "learning_rate": 0.00023666666666666668, + "loss": 0.1656, + "step": 2873 + }, + { + "epoch": 56.0, + "learning_rate": 0.00023655555555555555, + "loss": 0.1738, + "step": 2874 + }, + { + "epoch": 56.0, + "learning_rate": 0.00023644444444444445, + "loss": 0.1423, + "step": 2875 + }, + { + "epoch": 56.0, + "learning_rate": 0.00023633333333333335, + "loss": 0.1375, + "step": 2876 + }, + { + "epoch": 56.0, + "learning_rate": 0.00023622222222222222, + "loss": 0.2032, + "step": 2877 + }, + { + "epoch": 56.0, + "learning_rate": 0.00023611111111111112, + "loss": 0.1554, + "step": 2878 + }, + { + "epoch": 56.0, + "learning_rate": 0.000236, + "loss": 0.151, + "step": 2879 + }, + { + "epoch": 56.0, + "learning_rate": 0.00023588888888888892, + "loss": 0.1697, + "step": 2880 + }, + { + "epoch": 56.01, + "learning_rate": 0.0002357777777777778, + "loss": 0.1841, + "step": 2881 + }, + { + "epoch": 56.01, + "learning_rate": 0.00023566666666666666, + "loss": 0.139, + "step": 2882 + }, + { + "epoch": 56.01, + "learning_rate": 0.00023555555555555556, + "loss": 0.1662, + "step": 2883 + }, + { + "epoch": 56.01, + "learning_rate": 0.00023544444444444445, + "loss": 0.1616, + "step": 2884 + }, + { + "epoch": 56.01, + "learning_rate": 0.00023533333333333335, + "loss": 0.1754, + "step": 2885 + }, + { + "epoch": 56.01, + "learning_rate": 0.00023522222222222222, + "loss": 0.1589, + "step": 2886 + }, + { + "epoch": 56.01, + "learning_rate": 0.00023511111111111112, + "loss": 0.1651, + "step": 2887 + }, + { + "epoch": 56.01, + "learning_rate": 0.000235, + "loss": 0.1198, + "step": 2888 + }, + { + "epoch": 56.01, + "learning_rate": 0.0002348888888888889, + "loss": 0.1724, + "step": 2889 + }, + { + "epoch": 56.01, + "learning_rate": 0.0002347777777777778, + "loss": 0.1929, + "step": 2890 + }, + { + "epoch": 56.01, + "learning_rate": 0.00023466666666666666, + "loss": 0.1423, + "step": 2891 + }, + { + "epoch": 56.01, + "learning_rate": 0.00023455555555555556, + "loss": 0.1516, + "step": 2892 + }, + { + "epoch": 56.01, + "learning_rate": 0.00023444444444444446, + "loss": 0.1854, + "step": 2893 + }, + { + "epoch": 56.01, + "learning_rate": 0.00023433333333333336, + "loss": 0.1784, + "step": 2894 + }, + { + "epoch": 56.01, + "learning_rate": 0.00023422222222222223, + "loss": 0.1587, + "step": 2895 + }, + { + "epoch": 56.01, + "learning_rate": 0.0002341111111111111, + "loss": 0.1564, + "step": 2896 + }, + { + "epoch": 56.01, + "learning_rate": 0.00023400000000000002, + "loss": 0.1448, + "step": 2897 + }, + { + "epoch": 56.01, + "learning_rate": 0.0002338888888888889, + "loss": 0.1591, + "step": 2898 + }, + { + "epoch": 56.01, + "learning_rate": 0.0002337777777777778, + "loss": 0.1631, + "step": 2899 + }, + { + "epoch": 56.01, + "learning_rate": 0.00023366666666666666, + "loss": 0.1465, + "step": 2900 + }, + { + "epoch": 56.01, + "learning_rate": 0.00023355555555555556, + "loss": 0.1694, + "step": 2901 + }, + { + "epoch": 56.01, + "learning_rate": 0.00023344444444444446, + "loss": 0.1543, + "step": 2902 + }, + { + "epoch": 56.01, + "learning_rate": 0.00023333333333333333, + "loss": 0.1826, + "step": 2903 + }, + { + "epoch": 56.01, + "learning_rate": 0.00023322222222222223, + "loss": 0.1772, + "step": 2904 + }, + { + "epoch": 56.01, + "learning_rate": 0.0002331111111111111, + "loss": 0.1414, + "step": 2905 + }, + { + "epoch": 56.01, + "learning_rate": 0.00023300000000000003, + "loss": 0.1865, + "step": 2906 + }, + { + "epoch": 56.01, + "learning_rate": 0.0002328888888888889, + "loss": 0.1309, + "step": 2907 + }, + { + "epoch": 57.0, + "learning_rate": 0.0002327777777777778, + "loss": 0.2884, + "step": 2908 + }, + { + "epoch": 57.0, + "learning_rate": 0.00023266666666666667, + "loss": 0.1432, + "step": 2909 + }, + { + "epoch": 57.0, + "learning_rate": 0.00023255555555555554, + "loss": 0.1477, + "step": 2910 + }, + { + "epoch": 57.0, + "learning_rate": 0.00023244444444444446, + "loss": 0.1836, + "step": 2911 + }, + { + "epoch": 57.0, + "learning_rate": 0.00023233333333333333, + "loss": 0.1841, + "step": 2912 + }, + { + "epoch": 57.0, + "learning_rate": 0.00023222222222222223, + "loss": 0.1621, + "step": 2913 + }, + { + "epoch": 57.0, + "learning_rate": 0.0002321111111111111, + "loss": 0.1323, + "step": 2914 + }, + { + "epoch": 57.0, + "learning_rate": 0.00023200000000000003, + "loss": 0.157, + "step": 2915 + }, + { + "epoch": 57.0, + "learning_rate": 0.0002318888888888889, + "loss": 0.1422, + "step": 2916 + }, + { + "epoch": 57.0, + "learning_rate": 0.00023177777777777777, + "loss": 0.1598, + "step": 2917 + }, + { + "epoch": 57.0, + "learning_rate": 0.00023166666666666667, + "loss": 0.2168, + "step": 2918 + }, + { + "epoch": 57.0, + "learning_rate": 0.00023155555555555554, + "loss": 0.166, + "step": 2919 + }, + { + "epoch": 57.0, + "learning_rate": 0.00023144444444444447, + "loss": 0.1544, + "step": 2920 + }, + { + "epoch": 57.0, + "learning_rate": 0.00023133333333333334, + "loss": 0.1518, + "step": 2921 + }, + { + "epoch": 57.0, + "learning_rate": 0.00023122222222222223, + "loss": 0.2032, + "step": 2922 + }, + { + "epoch": 57.0, + "learning_rate": 0.0002311111111111111, + "loss": 0.1476, + "step": 2923 + }, + { + "epoch": 57.0, + "learning_rate": 0.000231, + "loss": 0.1255, + "step": 2924 + }, + { + "epoch": 57.0, + "learning_rate": 0.0002308888888888889, + "loss": 0.2182, + "step": 2925 + }, + { + "epoch": 57.0, + "learning_rate": 0.00023077777777777777, + "loss": 0.165, + "step": 2926 + }, + { + "epoch": 57.0, + "learning_rate": 0.00023066666666666667, + "loss": 0.2115, + "step": 2927 + }, + { + "epoch": 57.0, + "learning_rate": 0.00023055555555555557, + "loss": 0.144, + "step": 2928 + }, + { + "epoch": 57.0, + "learning_rate": 0.00023044444444444447, + "loss": 0.1429, + "step": 2929 + }, + { + "epoch": 57.0, + "learning_rate": 0.00023033333333333334, + "loss": 0.1858, + "step": 2930 + }, + { + "epoch": 57.0, + "learning_rate": 0.0002302222222222222, + "loss": 0.1541, + "step": 2931 + }, + { + "epoch": 57.01, + "learning_rate": 0.0002301111111111111, + "loss": 0.1431, + "step": 2932 + }, + { + "epoch": 57.01, + "learning_rate": 0.00023, + "loss": 0.1703, + "step": 2933 + }, + { + "epoch": 57.01, + "learning_rate": 0.0002298888888888889, + "loss": 0.1547, + "step": 2934 + }, + { + "epoch": 57.01, + "learning_rate": 0.00022977777777777778, + "loss": 0.1868, + "step": 2935 + }, + { + "epoch": 57.01, + "learning_rate": 0.00022966666666666667, + "loss": 0.141, + "step": 2936 + }, + { + "epoch": 57.01, + "learning_rate": 0.00022955555555555557, + "loss": 0.1879, + "step": 2937 + }, + { + "epoch": 57.01, + "learning_rate": 0.00022944444444444444, + "loss": 0.1455, + "step": 2938 + }, + { + "epoch": 57.01, + "learning_rate": 0.00022933333333333334, + "loss": 0.2008, + "step": 2939 + }, + { + "epoch": 57.01, + "learning_rate": 0.0002292222222222222, + "loss": 0.1817, + "step": 2940 + }, + { + "epoch": 57.01, + "learning_rate": 0.0002291111111111111, + "loss": 0.1478, + "step": 2941 + }, + { + "epoch": 57.01, + "learning_rate": 0.000229, + "loss": 0.1686, + "step": 2942 + }, + { + "epoch": 57.01, + "learning_rate": 0.0002288888888888889, + "loss": 0.1796, + "step": 2943 + }, + { + "epoch": 57.01, + "learning_rate": 0.00022877777777777778, + "loss": 0.1755, + "step": 2944 + }, + { + "epoch": 57.01, + "learning_rate": 0.00022866666666666665, + "loss": 0.1347, + "step": 2945 + }, + { + "epoch": 57.01, + "learning_rate": 0.00022855555555555558, + "loss": 0.2266, + "step": 2946 + }, + { + "epoch": 57.01, + "learning_rate": 0.00022844444444444445, + "loss": 0.2007, + "step": 2947 + }, + { + "epoch": 57.01, + "learning_rate": 0.00022833333333333334, + "loss": 0.1585, + "step": 2948 + }, + { + "epoch": 57.01, + "learning_rate": 0.00022822222222222222, + "loss": 0.1637, + "step": 2949 + }, + { + "epoch": 57.01, + "learning_rate": 0.00022811111111111111, + "loss": 0.1529, + "step": 2950 + }, + { + "epoch": 57.01, + "learning_rate": 0.000228, + "loss": 0.168, + "step": 2951 + }, + { + "epoch": 57.01, + "learning_rate": 0.00022788888888888888, + "loss": 0.1852, + "step": 2952 + }, + { + "epoch": 57.01, + "learning_rate": 0.00022777777777777778, + "loss": 0.1675, + "step": 2953 + }, + { + "epoch": 57.01, + "learning_rate": 0.00022766666666666665, + "loss": 0.1734, + "step": 2954 + }, + { + "epoch": 57.01, + "learning_rate": 0.00022755555555555558, + "loss": 0.1843, + "step": 2955 + }, + { + "epoch": 57.01, + "learning_rate": 0.00022744444444444445, + "loss": 0.1723, + "step": 2956 + }, + { + "epoch": 57.01, + "learning_rate": 0.00022733333333333335, + "loss": 0.1777, + "step": 2957 + }, + { + "epoch": 57.01, + "learning_rate": 0.00022722222222222222, + "loss": 0.1808, + "step": 2958 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022711111111111112, + "loss": 0.2716, + "step": 2959 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022700000000000002, + "loss": 0.204, + "step": 2960 + }, + { + "epoch": 58.0, + "learning_rate": 0.0002268888888888889, + "loss": 0.1627, + "step": 2961 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022677777777777778, + "loss": 0.1537, + "step": 2962 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022666666666666666, + "loss": 0.2019, + "step": 2963 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022655555555555558, + "loss": 0.1453, + "step": 2964 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022644444444444445, + "loss": 0.1354, + "step": 2965 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022633333333333332, + "loss": 0.1461, + "step": 2966 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022622222222222222, + "loss": 0.1646, + "step": 2967 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022611111111111112, + "loss": 0.1484, + "step": 2968 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022600000000000002, + "loss": 0.1417, + "step": 2969 + }, + { + "epoch": 58.0, + "learning_rate": 0.0002258888888888889, + "loss": 0.1584, + "step": 2970 + }, + { + "epoch": 58.0, + "learning_rate": 0.0002257777777777778, + "loss": 0.1697, + "step": 2971 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022566666666666666, + "loss": 0.1698, + "step": 2972 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022555555555555556, + "loss": 0.1636, + "step": 2973 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022544444444444446, + "loss": 0.2266, + "step": 2974 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022533333333333333, + "loss": 0.1956, + "step": 2975 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022522222222222222, + "loss": 0.1834, + "step": 2976 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022511111111111112, + "loss": 0.1399, + "step": 2977 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022500000000000002, + "loss": 0.164, + "step": 2978 + }, + { + "epoch": 58.0, + "learning_rate": 0.0002248888888888889, + "loss": 0.1684, + "step": 2979 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022477777777777776, + "loss": 0.1627, + "step": 2980 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022466666666666666, + "loss": 0.1744, + "step": 2981 + }, + { + "epoch": 58.0, + "learning_rate": 0.00022455555555555556, + "loss": 0.1595, + "step": 2982 + }, + { + "epoch": 58.01, + "learning_rate": 0.00022444444444444446, + "loss": 0.1848, + "step": 2983 + }, + { + "epoch": 58.01, + "learning_rate": 0.00022433333333333333, + "loss": 0.1564, + "step": 2984 + }, + { + "epoch": 58.01, + "learning_rate": 0.00022422222222222223, + "loss": 0.171, + "step": 2985 + }, + { + "epoch": 58.01, + "learning_rate": 0.00022411111111111113, + "loss": 0.1558, + "step": 2986 + }, + { + "epoch": 58.01, + "learning_rate": 0.000224, + "loss": 0.151, + "step": 2987 + }, + { + "epoch": 58.01, + "learning_rate": 0.0002238888888888889, + "loss": 0.1753, + "step": 2988 + }, + { + "epoch": 58.01, + "learning_rate": 0.00022377777777777777, + "loss": 0.1697, + "step": 2989 + }, + { + "epoch": 58.01, + "learning_rate": 0.0002236666666666667, + "loss": 0.1594, + "step": 2990 + }, + { + "epoch": 58.01, + "learning_rate": 0.00022355555555555556, + "loss": 0.1366, + "step": 2991 + }, + { + "epoch": 58.01, + "learning_rate": 0.00022344444444444446, + "loss": 0.1649, + "step": 2992 + }, + { + "epoch": 58.01, + "learning_rate": 0.00022333333333333333, + "loss": 0.1697, + "step": 2993 + }, + { + "epoch": 58.01, + "learning_rate": 0.0002232222222222222, + "loss": 0.1732, + "step": 2994 + }, + { + "epoch": 58.01, + "learning_rate": 0.00022311111111111113, + "loss": 0.1214, + "step": 2995 + }, + { + "epoch": 58.01, + "learning_rate": 0.000223, + "loss": 0.1946, + "step": 2996 + }, + { + "epoch": 58.01, + "learning_rate": 0.0002228888888888889, + "loss": 0.1523, + "step": 2997 + }, + { + "epoch": 58.01, + "learning_rate": 0.00022277777777777777, + "loss": 0.155, + "step": 2998 + }, + { + "epoch": 58.01, + "learning_rate": 0.0002226666666666667, + "loss": 0.1589, + "step": 2999 + }, + { + "epoch": 58.01, + "learning_rate": 0.00022255555555555557, + "loss": 0.1926, + "step": 3000 + }, + { + "epoch": 58.01, + "eval_cer": 0.08666507062485038, + "eval_loss": 0.3885515034198761, + "eval_runtime": 47.8257, + "eval_samples_per_second": 8.698, + "eval_steps_per_second": 0.272, + "eval_wer": 0.35200327064595255, + "step": 3000 + }, + { + "epoch": 58.01, + "learning_rate": 0.00022244444444444444, + "loss": 0.1611, + "step": 3001 + }, + { + "epoch": 58.01, + "learning_rate": 0.00022233333333333333, + "loss": 0.1705, + "step": 3002 + }, + { + "epoch": 58.01, + "learning_rate": 0.0002222222222222222, + "loss": 0.1769, + "step": 3003 + }, + { + "epoch": 58.01, + "learning_rate": 0.00022211111111111113, + "loss": 0.1941, + "step": 3004 + }, + { + "epoch": 58.01, + "learning_rate": 0.000222, + "loss": 0.1812, + "step": 3005 + }, + { + "epoch": 58.01, + "learning_rate": 0.0002218888888888889, + "loss": 0.1882, + "step": 3006 + }, + { + "epoch": 58.01, + "learning_rate": 0.00022177777777777777, + "loss": 0.1748, + "step": 3007 + }, + { + "epoch": 58.01, + "learning_rate": 0.00022166666666666667, + "loss": 0.1813, + "step": 3008 + }, + { + "epoch": 58.01, + "learning_rate": 0.00022155555555555557, + "loss": 0.1609, + "step": 3009 + }, + { + "epoch": 59.0, + "learning_rate": 0.00022144444444444444, + "loss": 0.2406, + "step": 3010 + }, + { + "epoch": 59.0, + "learning_rate": 0.00022133333333333334, + "loss": 0.1815, + "step": 3011 + }, + { + "epoch": 59.0, + "learning_rate": 0.0002212222222222222, + "loss": 0.1423, + "step": 3012 + }, + { + "epoch": 59.0, + "learning_rate": 0.00022111111111111113, + "loss": 0.2101, + "step": 3013 + }, + { + "epoch": 59.0, + "learning_rate": 0.000221, + "loss": 0.18, + "step": 3014 + }, + { + "epoch": 59.0, + "learning_rate": 0.00022088888888888888, + "loss": 0.1455, + "step": 3015 + }, + { + "epoch": 59.0, + "learning_rate": 0.00022077777777777777, + "loss": 0.1778, + "step": 3016 + }, + { + "epoch": 59.0, + "learning_rate": 0.00022066666666666667, + "loss": 0.1805, + "step": 3017 + }, + { + "epoch": 59.0, + "learning_rate": 0.00022055555555555557, + "loss": 0.1185, + "step": 3018 + }, + { + "epoch": 59.0, + "learning_rate": 0.00022044444444444444, + "loss": 0.1408, + "step": 3019 + }, + { + "epoch": 59.0, + "learning_rate": 0.00022033333333333334, + "loss": 0.1295, + "step": 3020 + }, + { + "epoch": 59.0, + "learning_rate": 0.00022022222222222224, + "loss": 0.1263, + "step": 3021 + }, + { + "epoch": 59.0, + "learning_rate": 0.0002201111111111111, + "loss": 0.1803, + "step": 3022 + }, + { + "epoch": 59.0, + "learning_rate": 0.00022, + "loss": 0.1938, + "step": 3023 + }, + { + "epoch": 59.0, + "learning_rate": 0.00021988888888888888, + "loss": 0.152, + "step": 3024 + }, + { + "epoch": 59.0, + "learning_rate": 0.00021977777777777778, + "loss": 0.1195, + "step": 3025 + }, + { + "epoch": 59.0, + "learning_rate": 0.00021966666666666668, + "loss": 0.1746, + "step": 3026 + }, + { + "epoch": 59.0, + "learning_rate": 0.00021955555555555557, + "loss": 0.0963, + "step": 3027 + }, + { + "epoch": 59.0, + "learning_rate": 0.00021944444444444444, + "loss": 0.1624, + "step": 3028 + }, + { + "epoch": 59.0, + "learning_rate": 0.00021933333333333332, + "loss": 0.2004, + "step": 3029 + }, + { + "epoch": 59.0, + "learning_rate": 0.00021922222222222224, + "loss": 0.1638, + "step": 3030 + }, + { + "epoch": 59.0, + "learning_rate": 0.0002191111111111111, + "loss": 0.1269, + "step": 3031 + }, + { + "epoch": 59.0, + "learning_rate": 0.000219, + "loss": 0.1571, + "step": 3032 + }, + { + "epoch": 59.0, + "learning_rate": 0.00021888888888888888, + "loss": 0.1532, + "step": 3033 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021877777777777778, + "loss": 0.1287, + "step": 3034 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021866666666666668, + "loss": 0.154, + "step": 3035 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021855555555555558, + "loss": 0.125, + "step": 3036 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021844444444444445, + "loss": 0.1696, + "step": 3037 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021833333333333332, + "loss": 0.1718, + "step": 3038 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021822222222222224, + "loss": 0.1519, + "step": 3039 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021811111111111112, + "loss": 0.1975, + "step": 3040 + }, + { + "epoch": 59.01, + "learning_rate": 0.000218, + "loss": 0.1612, + "step": 3041 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021788888888888888, + "loss": 0.1194, + "step": 3042 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021777777777777776, + "loss": 0.2027, + "step": 3043 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021766666666666668, + "loss": 0.1689, + "step": 3044 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021755555555555555, + "loss": 0.1639, + "step": 3045 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021744444444444445, + "loss": 0.1641, + "step": 3046 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021733333333333332, + "loss": 0.1615, + "step": 3047 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021722222222222225, + "loss": 0.1362, + "step": 3048 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021711111111111112, + "loss": 0.1628, + "step": 3049 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021700000000000002, + "loss": 0.1299, + "step": 3050 + }, + { + "epoch": 59.01, + "learning_rate": 0.0002168888888888889, + "loss": 0.1207, + "step": 3051 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021677777777777779, + "loss": 0.1668, + "step": 3052 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021666666666666668, + "loss": 0.1699, + "step": 3053 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021655555555555555, + "loss": 0.1554, + "step": 3054 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021644444444444445, + "loss": 0.1926, + "step": 3055 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021633333333333332, + "loss": 0.1729, + "step": 3056 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021622222222222225, + "loss": 0.1759, + "step": 3057 + }, + { + "epoch": 59.01, + "learning_rate": 0.00021611111111111112, + "loss": 0.1489, + "step": 3058 + }, + { + "epoch": 59.01, + "learning_rate": 0.000216, + "loss": 0.1507, + "step": 3059 + }, + { + "epoch": 59.01, + "learning_rate": 0.0002158888888888889, + "loss": 0.1842, + "step": 3060 + }, + { + "epoch": 60.0, + "learning_rate": 0.0002157777777777778, + "loss": 0.2454, + "step": 3061 + }, + { + "epoch": 60.0, + "learning_rate": 0.0002156666666666667, + "loss": 0.1565, + "step": 3062 + }, + { + "epoch": 60.0, + "learning_rate": 0.00021555555555555556, + "loss": 0.1671, + "step": 3063 + }, + { + "epoch": 60.0, + "learning_rate": 0.00021544444444444446, + "loss": 0.1563, + "step": 3064 + }, + { + "epoch": 60.0, + "learning_rate": 0.00021533333333333333, + "loss": 0.1171, + "step": 3065 + }, + { + "epoch": 60.0, + "learning_rate": 0.00021522222222222223, + "loss": 0.1601, + "step": 3066 + }, + { + "epoch": 60.0, + "learning_rate": 0.00021511111111111112, + "loss": 0.1915, + "step": 3067 + }, + { + "epoch": 60.0, + "learning_rate": 0.000215, + "loss": 0.1398, + "step": 3068 + }, + { + "epoch": 60.0, + "learning_rate": 0.0002148888888888889, + "loss": 0.1625, + "step": 3069 + }, + { + "epoch": 60.0, + "learning_rate": 0.0002147777777777778, + "loss": 0.1573, + "step": 3070 + }, + { + "epoch": 60.0, + "learning_rate": 0.0002146666666666667, + "loss": 0.1702, + "step": 3071 + }, + { + "epoch": 60.0, + "learning_rate": 0.00021455555555555556, + "loss": 0.1782, + "step": 3072 + }, + { + "epoch": 60.0, + "learning_rate": 0.00021444444444444443, + "loss": 0.0999, + "step": 3073 + }, + { + "epoch": 60.0, + "learning_rate": 0.00021433333333333333, + "loss": 0.2056, + "step": 3074 + }, + { + "epoch": 60.0, + "learning_rate": 0.00021422222222222223, + "loss": 0.1537, + "step": 3075 + }, + { + "epoch": 60.0, + "learning_rate": 0.00021411111111111113, + "loss": 0.1711, + "step": 3076 + }, + { + "epoch": 60.0, + "learning_rate": 0.000214, + "loss": 0.1925, + "step": 3077 + }, + { + "epoch": 60.0, + "learning_rate": 0.0002138888888888889, + "loss": 0.1884, + "step": 3078 + }, + { + "epoch": 60.0, + "learning_rate": 0.0002137777777777778, + "loss": 0.1335, + "step": 3079 + }, + { + "epoch": 60.0, + "learning_rate": 0.00021366666666666667, + "loss": 0.1861, + "step": 3080 + }, + { + "epoch": 60.0, + "learning_rate": 0.00021355555555555556, + "loss": 0.1338, + "step": 3081 + }, + { + "epoch": 60.0, + "learning_rate": 0.00021344444444444443, + "loss": 0.1543, + "step": 3082 + }, + { + "epoch": 60.0, + "learning_rate": 0.00021333333333333336, + "loss": 0.1402, + "step": 3083 + }, + { + "epoch": 60.0, + "learning_rate": 0.00021322222222222223, + "loss": 0.1948, + "step": 3084 + }, + { + "epoch": 60.01, + "learning_rate": 0.00021311111111111113, + "loss": 0.1941, + "step": 3085 + }, + { + "epoch": 60.01, + "learning_rate": 0.000213, + "loss": 0.1867, + "step": 3086 + }, + { + "epoch": 60.01, + "learning_rate": 0.00021288888888888887, + "loss": 0.1008, + "step": 3087 + }, + { + "epoch": 60.01, + "learning_rate": 0.0002127777777777778, + "loss": 0.171, + "step": 3088 + }, + { + "epoch": 60.01, + "learning_rate": 0.00021266666666666667, + "loss": 0.1136, + "step": 3089 + }, + { + "epoch": 60.01, + "learning_rate": 0.00021255555555555557, + "loss": 0.1323, + "step": 3090 + }, + { + "epoch": 60.01, + "learning_rate": 0.00021244444444444444, + "loss": 0.1675, + "step": 3091 + }, + { + "epoch": 60.01, + "learning_rate": 0.00021233333333333336, + "loss": 0.1524, + "step": 3092 + }, + { + "epoch": 60.01, + "learning_rate": 0.00021222222222222223, + "loss": 0.1726, + "step": 3093 + }, + { + "epoch": 60.01, + "learning_rate": 0.0002121111111111111, + "loss": 0.1354, + "step": 3094 + }, + { + "epoch": 60.01, + "learning_rate": 0.000212, + "loss": 0.1347, + "step": 3095 + }, + { + "epoch": 60.01, + "learning_rate": 0.00021188888888888887, + "loss": 0.1487, + "step": 3096 + }, + { + "epoch": 60.01, + "learning_rate": 0.0002117777777777778, + "loss": 0.1673, + "step": 3097 + }, + { + "epoch": 60.01, + "learning_rate": 0.00021166666666666667, + "loss": 0.1544, + "step": 3098 + }, + { + "epoch": 60.01, + "learning_rate": 0.00021155555555555557, + "loss": 0.1269, + "step": 3099 + }, + { + "epoch": 60.01, + "learning_rate": 0.00021144444444444444, + "loss": 0.128, + "step": 3100 + }, + { + "epoch": 60.01, + "learning_rate": 0.00021133333333333334, + "loss": 0.1316, + "step": 3101 + }, + { + "epoch": 60.01, + "learning_rate": 0.00021122222222222224, + "loss": 0.1529, + "step": 3102 + }, + { + "epoch": 60.01, + "learning_rate": 0.0002111111111111111, + "loss": 0.1695, + "step": 3103 + }, + { + "epoch": 60.01, + "learning_rate": 0.000211, + "loss": 0.1231, + "step": 3104 + }, + { + "epoch": 60.01, + "learning_rate": 0.00021088888888888888, + "loss": 0.1576, + "step": 3105 + }, + { + "epoch": 60.01, + "learning_rate": 0.0002107777777777778, + "loss": 0.1671, + "step": 3106 + }, + { + "epoch": 60.01, + "learning_rate": 0.00021066666666666667, + "loss": 0.149, + "step": 3107 + }, + { + "epoch": 60.01, + "learning_rate": 0.00021055555555555554, + "loss": 0.1645, + "step": 3108 + }, + { + "epoch": 60.01, + "learning_rate": 0.00021044444444444444, + "loss": 0.2023, + "step": 3109 + }, + { + "epoch": 60.01, + "learning_rate": 0.00021033333333333334, + "loss": 0.1454, + "step": 3110 + }, + { + "epoch": 60.01, + "learning_rate": 0.00021022222222222224, + "loss": 0.1414, + "step": 3111 + }, + { + "epoch": 61.0, + "learning_rate": 0.0002101111111111111, + "loss": 0.2538, + "step": 3112 + }, + { + "epoch": 61.0, + "learning_rate": 0.00021, + "loss": 0.1298, + "step": 3113 + }, + { + "epoch": 61.0, + "learning_rate": 0.0002098888888888889, + "loss": 0.142, + "step": 3114 + }, + { + "epoch": 61.0, + "learning_rate": 0.00020977777777777778, + "loss": 0.1673, + "step": 3115 + }, + { + "epoch": 61.0, + "learning_rate": 0.00020966666666666668, + "loss": 0.167, + "step": 3116 + }, + { + "epoch": 61.0, + "learning_rate": 0.00020955555555555555, + "loss": 0.1655, + "step": 3117 + }, + { + "epoch": 61.0, + "learning_rate": 0.00020944444444444445, + "loss": 0.1486, + "step": 3118 + }, + { + "epoch": 61.0, + "learning_rate": 0.00020933333333333334, + "loss": 0.146, + "step": 3119 + }, + { + "epoch": 61.0, + "learning_rate": 0.00020922222222222224, + "loss": 0.1775, + "step": 3120 + }, + { + "epoch": 61.0, + "learning_rate": 0.0002091111111111111, + "loss": 0.1603, + "step": 3121 + }, + { + "epoch": 61.0, + "learning_rate": 0.00020899999999999998, + "loss": 0.1609, + "step": 3122 + }, + { + "epoch": 61.0, + "learning_rate": 0.0002088888888888889, + "loss": 0.1549, + "step": 3123 + }, + { + "epoch": 61.0, + "learning_rate": 0.00020877777777777778, + "loss": 0.163, + "step": 3124 + }, + { + "epoch": 61.0, + "learning_rate": 0.00020866666666666668, + "loss": 0.176, + "step": 3125 + }, + { + "epoch": 61.0, + "learning_rate": 0.00020855555555555555, + "loss": 0.1261, + "step": 3126 + }, + { + "epoch": 61.0, + "learning_rate": 0.00020844444444444445, + "loss": 0.1669, + "step": 3127 + }, + { + "epoch": 61.0, + "learning_rate": 0.00020833333333333335, + "loss": 0.2127, + "step": 3128 + }, + { + "epoch": 61.0, + "learning_rate": 0.00020822222222222222, + "loss": 0.1406, + "step": 3129 + }, + { + "epoch": 61.0, + "learning_rate": 0.00020811111111111112, + "loss": 0.2388, + "step": 3130 + }, + { + "epoch": 61.0, + "learning_rate": 0.000208, + "loss": 0.1259, + "step": 3131 + }, + { + "epoch": 61.0, + "learning_rate": 0.0002078888888888889, + "loss": 0.1617, + "step": 3132 + }, + { + "epoch": 61.0, + "learning_rate": 0.00020777777777777778, + "loss": 0.1705, + "step": 3133 + }, + { + "epoch": 61.0, + "learning_rate": 0.00020766666666666668, + "loss": 0.1431, + "step": 3134 + }, + { + "epoch": 61.0, + "learning_rate": 0.00020755555555555555, + "loss": 0.1205, + "step": 3135 + }, + { + "epoch": 61.01, + "learning_rate": 0.00020744444444444442, + "loss": 0.1722, + "step": 3136 + }, + { + "epoch": 61.01, + "learning_rate": 0.00020733333333333335, + "loss": 0.1527, + "step": 3137 + }, + { + "epoch": 61.01, + "learning_rate": 0.00020722222222222222, + "loss": 0.1769, + "step": 3138 + }, + { + "epoch": 61.01, + "learning_rate": 0.00020711111111111112, + "loss": 0.1598, + "step": 3139 + }, + { + "epoch": 61.01, + "learning_rate": 0.000207, + "loss": 0.1962, + "step": 3140 + }, + { + "epoch": 61.01, + "learning_rate": 0.00020688888888888892, + "loss": 0.1543, + "step": 3141 + }, + { + "epoch": 61.01, + "learning_rate": 0.0002067777777777778, + "loss": 0.1443, + "step": 3142 + }, + { + "epoch": 61.01, + "learning_rate": 0.00020666666666666666, + "loss": 0.1975, + "step": 3143 + }, + { + "epoch": 61.01, + "learning_rate": 0.00020655555555555556, + "loss": 0.1898, + "step": 3144 + }, + { + "epoch": 61.01, + "learning_rate": 0.00020644444444444445, + "loss": 0.1299, + "step": 3145 + }, + { + "epoch": 61.01, + "learning_rate": 0.00020633333333333335, + "loss": 0.1621, + "step": 3146 + }, + { + "epoch": 61.01, + "learning_rate": 0.00020622222222222222, + "loss": 0.1524, + "step": 3147 + }, + { + "epoch": 61.01, + "learning_rate": 0.00020611111111111112, + "loss": 0.1614, + "step": 3148 + }, + { + "epoch": 61.01, + "learning_rate": 0.000206, + "loss": 0.1391, + "step": 3149 + }, + { + "epoch": 61.01, + "learning_rate": 0.0002058888888888889, + "loss": 0.1659, + "step": 3150 + }, + { + "epoch": 61.01, + "learning_rate": 0.0002057777777777778, + "loss": 0.1315, + "step": 3151 + }, + { + "epoch": 61.01, + "learning_rate": 0.00020566666666666666, + "loss": 0.1612, + "step": 3152 + }, + { + "epoch": 61.01, + "learning_rate": 0.00020555555555555556, + "loss": 0.126, + "step": 3153 + }, + { + "epoch": 61.01, + "learning_rate": 0.00020544444444444446, + "loss": 0.2065, + "step": 3154 + }, + { + "epoch": 61.01, + "learning_rate": 0.00020533333333333336, + "loss": 0.1487, + "step": 3155 + }, + { + "epoch": 61.01, + "learning_rate": 0.00020522222222222223, + "loss": 0.1731, + "step": 3156 + }, + { + "epoch": 61.01, + "learning_rate": 0.0002051111111111111, + "loss": 0.1573, + "step": 3157 + }, + { + "epoch": 61.01, + "learning_rate": 0.000205, + "loss": 0.1464, + "step": 3158 + }, + { + "epoch": 61.01, + "learning_rate": 0.0002048888888888889, + "loss": 0.1469, + "step": 3159 + }, + { + "epoch": 61.01, + "learning_rate": 0.0002047777777777778, + "loss": 0.1199, + "step": 3160 + }, + { + "epoch": 61.01, + "learning_rate": 0.00020466666666666666, + "loss": 0.1797, + "step": 3161 + }, + { + "epoch": 61.01, + "learning_rate": 0.00020455555555555556, + "loss": 0.1575, + "step": 3162 + }, + { + "epoch": 62.0, + "learning_rate": 0.00020444444444444446, + "loss": 0.2503, + "step": 3163 + }, + { + "epoch": 62.0, + "learning_rate": 0.00020433333333333333, + "loss": 0.1307, + "step": 3164 + }, + { + "epoch": 62.0, + "learning_rate": 0.00020422222222222223, + "loss": 0.1568, + "step": 3165 + }, + { + "epoch": 62.0, + "learning_rate": 0.0002041111111111111, + "loss": 0.1435, + "step": 3166 + }, + { + "epoch": 62.0, + "learning_rate": 0.000204, + "loss": 0.1845, + "step": 3167 + }, + { + "epoch": 62.0, + "learning_rate": 0.0002038888888888889, + "loss": 0.1752, + "step": 3168 + }, + { + "epoch": 62.0, + "learning_rate": 0.0002037777777777778, + "loss": 0.1487, + "step": 3169 + }, + { + "epoch": 62.0, + "learning_rate": 0.00020366666666666667, + "loss": 0.1657, + "step": 3170 + }, + { + "epoch": 62.0, + "learning_rate": 0.00020355555555555554, + "loss": 0.1319, + "step": 3171 + }, + { + "epoch": 62.0, + "learning_rate": 0.00020344444444444446, + "loss": 0.1397, + "step": 3172 + }, + { + "epoch": 62.0, + "learning_rate": 0.00020333333333333333, + "loss": 0.1514, + "step": 3173 + }, + { + "epoch": 62.0, + "learning_rate": 0.00020322222222222223, + "loss": 0.1689, + "step": 3174 + }, + { + "epoch": 62.0, + "learning_rate": 0.0002031111111111111, + "loss": 0.1234, + "step": 3175 + }, + { + "epoch": 62.0, + "learning_rate": 0.00020300000000000003, + "loss": 0.1585, + "step": 3176 + }, + { + "epoch": 62.0, + "learning_rate": 0.0002028888888888889, + "loss": 0.1559, + "step": 3177 + }, + { + "epoch": 62.0, + "learning_rate": 0.00020277777777777777, + "loss": 0.1537, + "step": 3178 + }, + { + "epoch": 62.0, + "learning_rate": 0.00020266666666666667, + "loss": 0.1764, + "step": 3179 + }, + { + "epoch": 62.0, + "learning_rate": 0.00020255555555555554, + "loss": 0.1486, + "step": 3180 + }, + { + "epoch": 62.0, + "learning_rate": 0.00020244444444444447, + "loss": 0.1371, + "step": 3181 + }, + { + "epoch": 62.0, + "learning_rate": 0.00020233333333333334, + "loss": 0.1218, + "step": 3182 + }, + { + "epoch": 62.0, + "learning_rate": 0.00020222222222222223, + "loss": 0.1499, + "step": 3183 + }, + { + "epoch": 62.0, + "learning_rate": 0.0002021111111111111, + "loss": 0.1369, + "step": 3184 + }, + { + "epoch": 62.0, + "learning_rate": 0.000202, + "loss": 0.1657, + "step": 3185 + }, + { + "epoch": 62.0, + "learning_rate": 0.0002018888888888889, + "loss": 0.1453, + "step": 3186 + }, + { + "epoch": 62.01, + "learning_rate": 0.00020177777777777777, + "loss": 0.1791, + "step": 3187 + }, + { + "epoch": 62.01, + "learning_rate": 0.00020166666666666667, + "loss": 0.1651, + "step": 3188 + }, + { + "epoch": 62.01, + "learning_rate": 0.00020155555555555554, + "loss": 0.1918, + "step": 3189 + }, + { + "epoch": 62.01, + "learning_rate": 0.00020144444444444447, + "loss": 0.1546, + "step": 3190 + }, + { + "epoch": 62.01, + "learning_rate": 0.00020133333333333334, + "loss": 0.1557, + "step": 3191 + }, + { + "epoch": 62.01, + "learning_rate": 0.0002012222222222222, + "loss": 0.1551, + "step": 3192 + }, + { + "epoch": 62.01, + "learning_rate": 0.0002011111111111111, + "loss": 0.1717, + "step": 3193 + }, + { + "epoch": 62.01, + "learning_rate": 0.000201, + "loss": 0.1419, + "step": 3194 + }, + { + "epoch": 62.01, + "learning_rate": 0.0002008888888888889, + "loss": 0.1462, + "step": 3195 + }, + { + "epoch": 62.01, + "learning_rate": 0.00020077777777777778, + "loss": 0.114, + "step": 3196 + }, + { + "epoch": 62.01, + "learning_rate": 0.00020066666666666667, + "loss": 0.1583, + "step": 3197 + }, + { + "epoch": 62.01, + "learning_rate": 0.00020055555555555555, + "loss": 0.1304, + "step": 3198 + }, + { + "epoch": 62.01, + "learning_rate": 0.00020044444444444444, + "loss": 0.1534, + "step": 3199 + }, + { + "epoch": 62.01, + "learning_rate": 0.00020033333333333334, + "loss": 0.1803, + "step": 3200 + }, + { + "epoch": 62.01, + "learning_rate": 0.0002002222222222222, + "loss": 0.1548, + "step": 3201 + }, + { + "epoch": 62.01, + "learning_rate": 0.0002001111111111111, + "loss": 0.1451, + "step": 3202 + }, + { + "epoch": 62.01, + "learning_rate": 0.0002, + "loss": 0.121, + "step": 3203 + }, + { + "epoch": 62.01, + "learning_rate": 0.0001998888888888889, + "loss": 0.1488, + "step": 3204 + }, + { + "epoch": 62.01, + "learning_rate": 0.00019977777777777778, + "loss": 0.1514, + "step": 3205 + }, + { + "epoch": 62.01, + "learning_rate": 0.00019966666666666665, + "loss": 0.1348, + "step": 3206 + }, + { + "epoch": 62.01, + "learning_rate": 0.00019955555555555558, + "loss": 0.1518, + "step": 3207 + }, + { + "epoch": 62.01, + "learning_rate": 0.00019944444444444445, + "loss": 0.117, + "step": 3208 + }, + { + "epoch": 62.01, + "learning_rate": 0.00019933333333333334, + "loss": 0.1534, + "step": 3209 + }, + { + "epoch": 62.01, + "learning_rate": 0.00019922222222222222, + "loss": 0.145, + "step": 3210 + }, + { + "epoch": 62.01, + "learning_rate": 0.00019911111111111111, + "loss": 0.1576, + "step": 3211 + }, + { + "epoch": 62.01, + "learning_rate": 0.000199, + "loss": 0.172, + "step": 3212 + }, + { + "epoch": 62.01, + "learning_rate": 0.00019888888888888888, + "loss": 0.1793, + "step": 3213 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019877777777777778, + "loss": 0.2088, + "step": 3214 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019866666666666665, + "loss": 0.1477, + "step": 3215 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019855555555555558, + "loss": 0.1715, + "step": 3216 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019844444444444445, + "loss": 0.1188, + "step": 3217 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019833333333333335, + "loss": 0.1081, + "step": 3218 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019822222222222222, + "loss": 0.1777, + "step": 3219 + }, + { + "epoch": 63.0, + "learning_rate": 0.0001981111111111111, + "loss": 0.1619, + "step": 3220 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019800000000000002, + "loss": 0.1178, + "step": 3221 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019788888888888889, + "loss": 0.165, + "step": 3222 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019777777777777778, + "loss": 0.102, + "step": 3223 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019766666666666666, + "loss": 0.1297, + "step": 3224 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019755555555555558, + "loss": 0.1372, + "step": 3225 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019744444444444445, + "loss": 0.164, + "step": 3226 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019733333333333335, + "loss": 0.225, + "step": 3227 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019722222222222222, + "loss": 0.1159, + "step": 3228 + }, + { + "epoch": 63.0, + "learning_rate": 0.0001971111111111111, + "loss": 0.2153, + "step": 3229 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019700000000000002, + "loss": 0.1279, + "step": 3230 + }, + { + "epoch": 63.0, + "learning_rate": 0.0001968888888888889, + "loss": 0.1406, + "step": 3231 + }, + { + "epoch": 63.0, + "learning_rate": 0.0001967777777777778, + "loss": 0.135, + "step": 3232 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019666666666666666, + "loss": 0.1369, + "step": 3233 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019655555555555558, + "loss": 0.1386, + "step": 3234 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019644444444444445, + "loss": 0.1476, + "step": 3235 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019633333333333333, + "loss": 0.1389, + "step": 3236 + }, + { + "epoch": 63.0, + "learning_rate": 0.00019622222222222222, + "loss": 0.1797, + "step": 3237 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019611111111111112, + "loss": 0.178, + "step": 3238 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019600000000000002, + "loss": 0.1281, + "step": 3239 + }, + { + "epoch": 63.01, + "learning_rate": 0.0001958888888888889, + "loss": 0.1663, + "step": 3240 + }, + { + "epoch": 63.01, + "learning_rate": 0.0001957777777777778, + "loss": 0.1187, + "step": 3241 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019566666666666666, + "loss": 0.1222, + "step": 3242 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019555555555555556, + "loss": 0.1406, + "step": 3243 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019544444444444446, + "loss": 0.1638, + "step": 3244 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019533333333333333, + "loss": 0.1236, + "step": 3245 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019522222222222223, + "loss": 0.1484, + "step": 3246 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019511111111111113, + "loss": 0.1193, + "step": 3247 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019500000000000002, + "loss": 0.1444, + "step": 3248 + }, + { + "epoch": 63.01, + "learning_rate": 0.0001948888888888889, + "loss": 0.1197, + "step": 3249 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019477777777777777, + "loss": 0.1638, + "step": 3250 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019466666666666666, + "loss": 0.1672, + "step": 3251 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019455555555555556, + "loss": 0.1247, + "step": 3252 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019444444444444446, + "loss": 0.1868, + "step": 3253 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019433333333333333, + "loss": 0.1594, + "step": 3254 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019422222222222223, + "loss": 0.141, + "step": 3255 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019411111111111113, + "loss": 0.1163, + "step": 3256 + }, + { + "epoch": 63.01, + "learning_rate": 0.000194, + "loss": 0.1292, + "step": 3257 + }, + { + "epoch": 63.01, + "learning_rate": 0.0001938888888888889, + "loss": 0.1954, + "step": 3258 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019377777777777777, + "loss": 0.133, + "step": 3259 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019366666666666667, + "loss": 0.1502, + "step": 3260 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019355555555555557, + "loss": 0.1151, + "step": 3261 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019344444444444446, + "loss": 0.1232, + "step": 3262 + }, + { + "epoch": 63.01, + "learning_rate": 0.00019333333333333333, + "loss": 0.1876, + "step": 3263 + }, + { + "epoch": 63.01, + "learning_rate": 0.0001932222222222222, + "loss": 0.1759, + "step": 3264 + }, + { + "epoch": 64.0, + "learning_rate": 0.00019311111111111113, + "loss": 0.2086, + "step": 3265 + }, + { + "epoch": 64.0, + "learning_rate": 0.000193, + "loss": 0.1785, + "step": 3266 + }, + { + "epoch": 64.0, + "learning_rate": 0.0001928888888888889, + "loss": 0.1228, + "step": 3267 + }, + { + "epoch": 64.0, + "learning_rate": 0.00019277777777777777, + "loss": 0.1247, + "step": 3268 + }, + { + "epoch": 64.0, + "learning_rate": 0.0001926666666666667, + "loss": 0.1225, + "step": 3269 + }, + { + "epoch": 64.0, + "learning_rate": 0.00019255555555555557, + "loss": 0.145, + "step": 3270 + }, + { + "epoch": 64.0, + "learning_rate": 0.00019244444444444444, + "loss": 0.1152, + "step": 3271 + }, + { + "epoch": 64.0, + "learning_rate": 0.00019233333333333334, + "loss": 0.1357, + "step": 3272 + }, + { + "epoch": 64.0, + "learning_rate": 0.0001922222222222222, + "loss": 0.1043, + "step": 3273 + }, + { + "epoch": 64.0, + "learning_rate": 0.00019211111111111113, + "loss": 0.1147, + "step": 3274 + }, + { + "epoch": 64.0, + "learning_rate": 0.000192, + "loss": 0.1188, + "step": 3275 + }, + { + "epoch": 64.0, + "learning_rate": 0.0001918888888888889, + "loss": 0.1231, + "step": 3276 + }, + { + "epoch": 64.0, + "learning_rate": 0.00019177777777777777, + "loss": 0.1631, + "step": 3277 + }, + { + "epoch": 64.0, + "learning_rate": 0.00019166666666666667, + "loss": 0.1395, + "step": 3278 + }, + { + "epoch": 64.0, + "learning_rate": 0.00019155555555555557, + "loss": 0.139, + "step": 3279 + }, + { + "epoch": 64.0, + "learning_rate": 0.00019144444444444444, + "loss": 0.1406, + "step": 3280 + }, + { + "epoch": 64.0, + "learning_rate": 0.00019133333333333334, + "loss": 0.153, + "step": 3281 + }, + { + "epoch": 64.0, + "learning_rate": 0.0001912222222222222, + "loss": 0.1569, + "step": 3282 + }, + { + "epoch": 64.0, + "learning_rate": 0.00019111111111111114, + "loss": 0.1286, + "step": 3283 + }, + { + "epoch": 64.0, + "learning_rate": 0.000191, + "loss": 0.1069, + "step": 3284 + }, + { + "epoch": 64.0, + "learning_rate": 0.00019088888888888888, + "loss": 0.1392, + "step": 3285 + }, + { + "epoch": 64.0, + "learning_rate": 0.00019077777777777778, + "loss": 0.1139, + "step": 3286 + }, + { + "epoch": 64.0, + "learning_rate": 0.00019066666666666668, + "loss": 0.1434, + "step": 3287 + }, + { + "epoch": 64.0, + "learning_rate": 0.00019055555555555557, + "loss": 0.1096, + "step": 3288 + }, + { + "epoch": 64.0, + "learning_rate": 0.00019044444444444444, + "loss": 0.151, + "step": 3289 + }, + { + "epoch": 64.01, + "learning_rate": 0.00019033333333333334, + "loss": 0.1403, + "step": 3290 + }, + { + "epoch": 64.01, + "learning_rate": 0.00019022222222222221, + "loss": 0.1536, + "step": 3291 + }, + { + "epoch": 64.01, + "learning_rate": 0.0001901111111111111, + "loss": 0.1554, + "step": 3292 + }, + { + "epoch": 64.01, + "learning_rate": 0.00019, + "loss": 0.1262, + "step": 3293 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018988888888888888, + "loss": 0.1591, + "step": 3294 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018977777777777778, + "loss": 0.1796, + "step": 3295 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018966666666666668, + "loss": 0.1216, + "step": 3296 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018955555555555558, + "loss": 0.1529, + "step": 3297 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018944444444444445, + "loss": 0.1595, + "step": 3298 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018933333333333332, + "loss": 0.1079, + "step": 3299 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018922222222222224, + "loss": 0.1491, + "step": 3300 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018911111111111112, + "loss": 0.1574, + "step": 3301 + }, + { + "epoch": 64.01, + "learning_rate": 0.000189, + "loss": 0.1294, + "step": 3302 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018888888888888888, + "loss": 0.0961, + "step": 3303 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018877777777777778, + "loss": 0.1682, + "step": 3304 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018866666666666668, + "loss": 0.1454, + "step": 3305 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018855555555555555, + "loss": 0.1646, + "step": 3306 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018844444444444445, + "loss": 0.1448, + "step": 3307 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018833333333333332, + "loss": 0.1681, + "step": 3308 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018822222222222225, + "loss": 0.1226, + "step": 3309 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018811111111111112, + "loss": 0.1615, + "step": 3310 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018800000000000002, + "loss": 0.1727, + "step": 3311 + }, + { + "epoch": 64.01, + "learning_rate": 0.0001878888888888889, + "loss": 0.12, + "step": 3312 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018777777777777776, + "loss": 0.1492, + "step": 3313 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018766666666666668, + "loss": 0.1434, + "step": 3314 + }, + { + "epoch": 64.01, + "learning_rate": 0.00018755555555555555, + "loss": 0.1348, + "step": 3315 + }, + { + "epoch": 65.0, + "learning_rate": 0.00018744444444444445, + "loss": 0.2311, + "step": 3316 + }, + { + "epoch": 65.0, + "learning_rate": 0.00018733333333333332, + "loss": 0.1184, + "step": 3317 + }, + { + "epoch": 65.0, + "learning_rate": 0.00018722222222222225, + "loss": 0.1358, + "step": 3318 + }, + { + "epoch": 65.0, + "learning_rate": 0.00018711111111111112, + "loss": 0.1392, + "step": 3319 + }, + { + "epoch": 65.0, + "learning_rate": 0.000187, + "loss": 0.1095, + "step": 3320 + }, + { + "epoch": 65.0, + "learning_rate": 0.0001868888888888889, + "loss": 0.1102, + "step": 3321 + }, + { + "epoch": 65.0, + "learning_rate": 0.00018677777777777776, + "loss": 0.1225, + "step": 3322 + }, + { + "epoch": 65.0, + "learning_rate": 0.0001866666666666667, + "loss": 0.1625, + "step": 3323 + }, + { + "epoch": 65.0, + "learning_rate": 0.00018655555555555556, + "loss": 0.149, + "step": 3324 + }, + { + "epoch": 65.0, + "learning_rate": 0.00018644444444444446, + "loss": 0.1227, + "step": 3325 + }, + { + "epoch": 65.0, + "learning_rate": 0.00018633333333333333, + "loss": 0.1093, + "step": 3326 + }, + { + "epoch": 65.0, + "learning_rate": 0.00018622222222222223, + "loss": 0.133, + "step": 3327 + }, + { + "epoch": 65.0, + "learning_rate": 0.00018611111111111112, + "loss": 0.1598, + "step": 3328 + }, + { + "epoch": 65.0, + "learning_rate": 0.000186, + "loss": 0.2028, + "step": 3329 + }, + { + "epoch": 65.0, + "learning_rate": 0.0001858888888888889, + "loss": 0.1407, + "step": 3330 + }, + { + "epoch": 65.0, + "learning_rate": 0.00018577777777777776, + "loss": 0.1793, + "step": 3331 + }, + { + "epoch": 65.0, + "learning_rate": 0.0001856666666666667, + "loss": 0.1602, + "step": 3332 + }, + { + "epoch": 65.0, + "learning_rate": 0.00018555555555555556, + "loss": 0.1488, + "step": 3333 + }, + { + "epoch": 65.0, + "learning_rate": 0.00018544444444444443, + "loss": 0.1611, + "step": 3334 + }, + { + "epoch": 65.0, + "learning_rate": 0.00018533333333333333, + "loss": 0.1552, + "step": 3335 + }, + { + "epoch": 65.0, + "learning_rate": 0.00018522222222222223, + "loss": 0.1773, + "step": 3336 + }, + { + "epoch": 65.0, + "learning_rate": 0.00018511111111111113, + "loss": 0.1399, + "step": 3337 + }, + { + "epoch": 65.0, + "learning_rate": 0.000185, + "loss": 0.1433, + "step": 3338 + }, + { + "epoch": 65.0, + "learning_rate": 0.0001848888888888889, + "loss": 0.1627, + "step": 3339 + }, + { + "epoch": 65.0, + "learning_rate": 0.0001847777777777778, + "loss": 0.1229, + "step": 3340 + }, + { + "epoch": 65.01, + "learning_rate": 0.00018466666666666666, + "loss": 0.1745, + "step": 3341 + }, + { + "epoch": 65.01, + "learning_rate": 0.00018455555555555556, + "loss": 0.1354, + "step": 3342 + }, + { + "epoch": 65.01, + "learning_rate": 0.00018444444444444443, + "loss": 0.142, + "step": 3343 + }, + { + "epoch": 65.01, + "learning_rate": 0.00018433333333333333, + "loss": 0.1241, + "step": 3344 + }, + { + "epoch": 65.01, + "learning_rate": 0.00018422222222222223, + "loss": 0.1668, + "step": 3345 + }, + { + "epoch": 65.01, + "learning_rate": 0.00018411111111111113, + "loss": 0.1353, + "step": 3346 + }, + { + "epoch": 65.01, + "learning_rate": 0.000184, + "loss": 0.1679, + "step": 3347 + }, + { + "epoch": 65.01, + "learning_rate": 0.00018388888888888887, + "loss": 0.1466, + "step": 3348 + }, + { + "epoch": 65.01, + "learning_rate": 0.0001837777777777778, + "loss": 0.1087, + "step": 3349 + }, + { + "epoch": 65.01, + "learning_rate": 0.00018366666666666667, + "loss": 0.1093, + "step": 3350 + }, + { + "epoch": 65.01, + "learning_rate": 0.00018355555555555557, + "loss": 0.1171, + "step": 3351 + }, + { + "epoch": 65.01, + "learning_rate": 0.00018344444444444444, + "loss": 0.1127, + "step": 3352 + }, + { + "epoch": 65.01, + "learning_rate": 0.00018333333333333334, + "loss": 0.0976, + "step": 3353 + }, + { + "epoch": 65.01, + "learning_rate": 0.00018322222222222223, + "loss": 0.1542, + "step": 3354 + }, + { + "epoch": 65.01, + "learning_rate": 0.0001831111111111111, + "loss": 0.1294, + "step": 3355 + }, + { + "epoch": 65.01, + "learning_rate": 0.000183, + "loss": 0.1141, + "step": 3356 + }, + { + "epoch": 65.01, + "learning_rate": 0.00018288888888888887, + "loss": 0.1098, + "step": 3357 + }, + { + "epoch": 65.01, + "learning_rate": 0.0001827777777777778, + "loss": 0.1697, + "step": 3358 + }, + { + "epoch": 65.01, + "learning_rate": 0.00018266666666666667, + "loss": 0.1071, + "step": 3359 + }, + { + "epoch": 65.01, + "learning_rate": 0.00018255555555555557, + "loss": 0.148, + "step": 3360 + }, + { + "epoch": 65.01, + "learning_rate": 0.00018244444444444444, + "loss": 0.158, + "step": 3361 + }, + { + "epoch": 65.01, + "learning_rate": 0.0001823333333333333, + "loss": 0.148, + "step": 3362 + }, + { + "epoch": 65.01, + "learning_rate": 0.00018222222222222224, + "loss": 0.1337, + "step": 3363 + }, + { + "epoch": 65.01, + "learning_rate": 0.0001821111111111111, + "loss": 0.16, + "step": 3364 + }, + { + "epoch": 65.01, + "learning_rate": 0.000182, + "loss": 0.1614, + "step": 3365 + }, + { + "epoch": 65.01, + "learning_rate": 0.00018188888888888888, + "loss": 0.1893, + "step": 3366 + }, + { + "epoch": 66.0, + "learning_rate": 0.0001817777777777778, + "loss": 0.2778, + "step": 3367 + }, + { + "epoch": 66.0, + "learning_rate": 0.00018166666666666667, + "loss": 0.1095, + "step": 3368 + }, + { + "epoch": 66.0, + "learning_rate": 0.00018155555555555554, + "loss": 0.1299, + "step": 3369 + }, + { + "epoch": 66.0, + "learning_rate": 0.00018144444444444444, + "loss": 0.1623, + "step": 3370 + }, + { + "epoch": 66.0, + "learning_rate": 0.00018133333333333334, + "loss": 0.1223, + "step": 3371 + }, + { + "epoch": 66.0, + "learning_rate": 0.00018122222222222224, + "loss": 0.1171, + "step": 3372 + }, + { + "epoch": 66.0, + "learning_rate": 0.0001811111111111111, + "loss": 0.1212, + "step": 3373 + }, + { + "epoch": 66.0, + "learning_rate": 0.000181, + "loss": 0.1236, + "step": 3374 + }, + { + "epoch": 66.0, + "learning_rate": 0.00018088888888888888, + "loss": 0.1451, + "step": 3375 + }, + { + "epoch": 66.0, + "learning_rate": 0.00018077777777777778, + "loss": 0.121, + "step": 3376 + }, + { + "epoch": 66.0, + "learning_rate": 0.00018066666666666668, + "loss": 0.1365, + "step": 3377 + }, + { + "epoch": 66.0, + "learning_rate": 0.00018055555555555555, + "loss": 0.131, + "step": 3378 + }, + { + "epoch": 66.0, + "learning_rate": 0.00018044444444444445, + "loss": 0.133, + "step": 3379 + }, + { + "epoch": 66.0, + "learning_rate": 0.00018033333333333334, + "loss": 0.1384, + "step": 3380 + }, + { + "epoch": 66.0, + "learning_rate": 0.00018022222222222224, + "loss": 0.1668, + "step": 3381 + }, + { + "epoch": 66.0, + "learning_rate": 0.0001801111111111111, + "loss": 0.138, + "step": 3382 + }, + { + "epoch": 66.0, + "learning_rate": 0.00017999999999999998, + "loss": 0.1224, + "step": 3383 + }, + { + "epoch": 66.0, + "learning_rate": 0.00017988888888888888, + "loss": 0.1621, + "step": 3384 + }, + { + "epoch": 66.0, + "learning_rate": 0.00017977777777777778, + "loss": 0.1306, + "step": 3385 + }, + { + "epoch": 66.0, + "learning_rate": 0.00017966666666666668, + "loss": 0.1318, + "step": 3386 + }, + { + "epoch": 66.0, + "learning_rate": 0.00017955555555555555, + "loss": 0.1116, + "step": 3387 + }, + { + "epoch": 66.0, + "learning_rate": 0.00017944444444444445, + "loss": 0.1205, + "step": 3388 + }, + { + "epoch": 66.0, + "learning_rate": 0.00017933333333333335, + "loss": 0.1612, + "step": 3389 + }, + { + "epoch": 66.0, + "learning_rate": 0.00017922222222222222, + "loss": 0.1146, + "step": 3390 + }, + { + "epoch": 66.0, + "learning_rate": 0.00017911111111111112, + "loss": 0.1571, + "step": 3391 + }, + { + "epoch": 66.01, + "learning_rate": 0.000179, + "loss": 0.1282, + "step": 3392 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017888888888888889, + "loss": 0.0978, + "step": 3393 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017877777777777778, + "loss": 0.1189, + "step": 3394 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017866666666666668, + "loss": 0.1293, + "step": 3395 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017855555555555555, + "loss": 0.1321, + "step": 3396 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017844444444444442, + "loss": 0.1547, + "step": 3397 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017833333333333335, + "loss": 0.1648, + "step": 3398 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017822222222222222, + "loss": 0.1113, + "step": 3399 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017811111111111112, + "loss": 0.1129, + "step": 3400 + }, + { + "epoch": 66.01, + "learning_rate": 0.000178, + "loss": 0.1154, + "step": 3401 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017788888888888892, + "loss": 0.136, + "step": 3402 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017777777777777779, + "loss": 0.1173, + "step": 3403 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017766666666666666, + "loss": 0.1529, + "step": 3404 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017755555555555556, + "loss": 0.1383, + "step": 3405 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017744444444444443, + "loss": 0.1252, + "step": 3406 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017733333333333335, + "loss": 0.1218, + "step": 3407 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017722222222222222, + "loss": 0.1464, + "step": 3408 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017711111111111112, + "loss": 0.1224, + "step": 3409 + }, + { + "epoch": 66.01, + "learning_rate": 0.000177, + "loss": 0.1231, + "step": 3410 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017688888888888892, + "loss": 0.1548, + "step": 3411 + }, + { + "epoch": 66.01, + "learning_rate": 0.0001767777777777778, + "loss": 0.1905, + "step": 3412 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017666666666666666, + "loss": 0.1426, + "step": 3413 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017655555555555556, + "loss": 0.1213, + "step": 3414 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017644444444444443, + "loss": 0.1231, + "step": 3415 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017633333333333335, + "loss": 0.1692, + "step": 3416 + }, + { + "epoch": 66.01, + "learning_rate": 0.00017622222222222223, + "loss": 0.1235, + "step": 3417 + }, + { + "epoch": 67.0, + "learning_rate": 0.00017611111111111112, + "loss": 0.1568, + "step": 3418 + }, + { + "epoch": 67.0, + "learning_rate": 0.000176, + "loss": 0.1339, + "step": 3419 + }, + { + "epoch": 67.0, + "learning_rate": 0.0001758888888888889, + "loss": 0.1013, + "step": 3420 + }, + { + "epoch": 67.0, + "learning_rate": 0.0001757777777777778, + "loss": 0.1303, + "step": 3421 + }, + { + "epoch": 67.0, + "learning_rate": 0.00017566666666666666, + "loss": 0.1333, + "step": 3422 + }, + { + "epoch": 67.0, + "learning_rate": 0.00017555555555555556, + "loss": 0.1454, + "step": 3423 + }, + { + "epoch": 67.0, + "learning_rate": 0.00017544444444444443, + "loss": 0.1469, + "step": 3424 + }, + { + "epoch": 67.0, + "learning_rate": 0.00017533333333333336, + "loss": 0.1483, + "step": 3425 + }, + { + "epoch": 67.0, + "learning_rate": 0.00017522222222222223, + "loss": 0.0967, + "step": 3426 + }, + { + "epoch": 67.0, + "learning_rate": 0.0001751111111111111, + "loss": 0.131, + "step": 3427 + }, + { + "epoch": 67.0, + "learning_rate": 0.000175, + "loss": 0.1199, + "step": 3428 + }, + { + "epoch": 67.0, + "learning_rate": 0.0001748888888888889, + "loss": 0.1315, + "step": 3429 + }, + { + "epoch": 67.0, + "learning_rate": 0.0001747777777777778, + "loss": 0.0965, + "step": 3430 + }, + { + "epoch": 67.0, + "learning_rate": 0.00017466666666666667, + "loss": 0.1143, + "step": 3431 + }, + { + "epoch": 67.0, + "learning_rate": 0.00017455555555555556, + "loss": 0.1397, + "step": 3432 + }, + { + "epoch": 67.0, + "learning_rate": 0.00017444444444444446, + "loss": 0.0867, + "step": 3433 + }, + { + "epoch": 67.0, + "learning_rate": 0.00017433333333333333, + "loss": 0.14, + "step": 3434 + }, + { + "epoch": 67.0, + "learning_rate": 0.00017422222222222223, + "loss": 0.1194, + "step": 3435 + }, + { + "epoch": 67.0, + "learning_rate": 0.0001741111111111111, + "loss": 0.1484, + "step": 3436 + }, + { + "epoch": 67.0, + "learning_rate": 0.000174, + "loss": 0.1742, + "step": 3437 + }, + { + "epoch": 67.0, + "learning_rate": 0.0001738888888888889, + "loss": 0.1508, + "step": 3438 + }, + { + "epoch": 67.0, + "learning_rate": 0.0001737777777777778, + "loss": 0.139, + "step": 3439 + }, + { + "epoch": 67.0, + "learning_rate": 0.00017366666666666667, + "loss": 0.1064, + "step": 3440 + }, + { + "epoch": 67.0, + "learning_rate": 0.00017355555555555554, + "loss": 0.1341, + "step": 3441 + }, + { + "epoch": 67.0, + "learning_rate": 0.00017344444444444447, + "loss": 0.1191, + "step": 3442 + }, + { + "epoch": 67.01, + "learning_rate": 0.00017333333333333334, + "loss": 0.1291, + "step": 3443 + }, + { + "epoch": 67.01, + "learning_rate": 0.00017322222222222223, + "loss": 0.1042, + "step": 3444 + }, + { + "epoch": 67.01, + "learning_rate": 0.0001731111111111111, + "loss": 0.1593, + "step": 3445 + }, + { + "epoch": 67.01, + "learning_rate": 0.000173, + "loss": 0.1537, + "step": 3446 + }, + { + "epoch": 67.01, + "learning_rate": 0.0001728888888888889, + "loss": 0.1579, + "step": 3447 + }, + { + "epoch": 67.01, + "learning_rate": 0.00017277777777777777, + "loss": 0.1555, + "step": 3448 + }, + { + "epoch": 67.01, + "learning_rate": 0.00017266666666666667, + "loss": 0.1283, + "step": 3449 + }, + { + "epoch": 67.01, + "learning_rate": 0.00017255555555555554, + "loss": 0.1111, + "step": 3450 + }, + { + "epoch": 67.01, + "learning_rate": 0.00017244444444444447, + "loss": 0.1396, + "step": 3451 + }, + { + "epoch": 67.01, + "learning_rate": 0.00017233333333333334, + "loss": 0.1506, + "step": 3452 + }, + { + "epoch": 67.01, + "learning_rate": 0.00017222222222222224, + "loss": 0.1264, + "step": 3453 + }, + { + "epoch": 67.01, + "learning_rate": 0.0001721111111111111, + "loss": 0.1122, + "step": 3454 + }, + { + "epoch": 67.01, + "learning_rate": 0.00017199999999999998, + "loss": 0.1262, + "step": 3455 + }, + { + "epoch": 67.01, + "learning_rate": 0.0001718888888888889, + "loss": 0.198, + "step": 3456 + }, + { + "epoch": 67.01, + "learning_rate": 0.00017177777777777778, + "loss": 0.1059, + "step": 3457 + }, + { + "epoch": 67.01, + "learning_rate": 0.00017166666666666667, + "loss": 0.1579, + "step": 3458 + }, + { + "epoch": 67.01, + "learning_rate": 0.00017155555555555555, + "loss": 0.1582, + "step": 3459 + }, + { + "epoch": 67.01, + "learning_rate": 0.00017144444444444447, + "loss": 0.1251, + "step": 3460 + }, + { + "epoch": 67.01, + "learning_rate": 0.00017133333333333334, + "loss": 0.1265, + "step": 3461 + }, + { + "epoch": 67.01, + "learning_rate": 0.0001712222222222222, + "loss": 0.1445, + "step": 3462 + }, + { + "epoch": 67.01, + "learning_rate": 0.0001711111111111111, + "loss": 0.1383, + "step": 3463 + }, + { + "epoch": 67.01, + "learning_rate": 0.000171, + "loss": 0.183, + "step": 3464 + }, + { + "epoch": 67.01, + "learning_rate": 0.0001708888888888889, + "loss": 0.1287, + "step": 3465 + }, + { + "epoch": 67.01, + "learning_rate": 0.00017077777777777778, + "loss": 0.1675, + "step": 3466 + }, + { + "epoch": 67.01, + "learning_rate": 0.00017066666666666668, + "loss": 0.1094, + "step": 3467 + }, + { + "epoch": 67.01, + "learning_rate": 0.00017055555555555555, + "loss": 0.1163, + "step": 3468 + }, + { + "epoch": 68.0, + "learning_rate": 0.00017044444444444445, + "loss": 0.194, + "step": 3469 + }, + { + "epoch": 68.0, + "learning_rate": 0.00017033333333333334, + "loss": 0.1162, + "step": 3470 + }, + { + "epoch": 68.0, + "learning_rate": 0.00017022222222222222, + "loss": 0.0886, + "step": 3471 + }, + { + "epoch": 68.0, + "learning_rate": 0.00017011111111111111, + "loss": 0.1401, + "step": 3472 + }, + { + "epoch": 68.0, + "learning_rate": 0.00017, + "loss": 0.1305, + "step": 3473 + }, + { + "epoch": 68.0, + "learning_rate": 0.0001698888888888889, + "loss": 0.1341, + "step": 3474 + }, + { + "epoch": 68.0, + "learning_rate": 0.00016977777777777778, + "loss": 0.1158, + "step": 3475 + }, + { + "epoch": 68.0, + "learning_rate": 0.00016966666666666665, + "loss": 0.1296, + "step": 3476 + }, + { + "epoch": 68.0, + "learning_rate": 0.00016955555555555555, + "loss": 0.1019, + "step": 3477 + }, + { + "epoch": 68.0, + "learning_rate": 0.00016944444444444445, + "loss": 0.1491, + "step": 3478 + }, + { + "epoch": 68.0, + "learning_rate": 0.00016933333333333335, + "loss": 0.1105, + "step": 3479 + }, + { + "epoch": 68.0, + "learning_rate": 0.00016922222222222222, + "loss": 0.125, + "step": 3480 + }, + { + "epoch": 68.0, + "learning_rate": 0.00016911111111111112, + "loss": 0.1314, + "step": 3481 + }, + { + "epoch": 68.0, + "learning_rate": 0.00016900000000000002, + "loss": 0.0939, + "step": 3482 + }, + { + "epoch": 68.0, + "learning_rate": 0.00016888888888888889, + "loss": 0.1579, + "step": 3483 + }, + { + "epoch": 68.0, + "learning_rate": 0.00016877777777777778, + "loss": 0.1675, + "step": 3484 + }, + { + "epoch": 68.0, + "learning_rate": 0.00016866666666666666, + "loss": 0.151, + "step": 3485 + }, + { + "epoch": 68.0, + "learning_rate": 0.00016855555555555555, + "loss": 0.1072, + "step": 3486 + }, + { + "epoch": 68.0, + "learning_rate": 0.00016844444444444445, + "loss": 0.164, + "step": 3487 + }, + { + "epoch": 68.0, + "learning_rate": 0.00016833333333333335, + "loss": 0.1277, + "step": 3488 + }, + { + "epoch": 68.0, + "learning_rate": 0.00016822222222222222, + "loss": 0.134, + "step": 3489 + }, + { + "epoch": 68.0, + "learning_rate": 0.0001681111111111111, + "loss": 0.1444, + "step": 3490 + }, + { + "epoch": 68.0, + "learning_rate": 0.00016800000000000002, + "loss": 0.122, + "step": 3491 + }, + { + "epoch": 68.0, + "learning_rate": 0.0001678888888888889, + "loss": 0.1572, + "step": 3492 + }, + { + "epoch": 68.0, + "learning_rate": 0.0001677777777777778, + "loss": 0.1128, + "step": 3493 + }, + { + "epoch": 68.01, + "learning_rate": 0.00016766666666666666, + "loss": 0.1366, + "step": 3494 + }, + { + "epoch": 68.01, + "learning_rate": 0.00016755555555555558, + "loss": 0.1274, + "step": 3495 + }, + { + "epoch": 68.01, + "learning_rate": 0.00016744444444444445, + "loss": 0.1341, + "step": 3496 + }, + { + "epoch": 68.01, + "learning_rate": 0.00016733333333333333, + "loss": 0.0907, + "step": 3497 + }, + { + "epoch": 68.01, + "learning_rate": 0.00016722222222222222, + "loss": 0.1229, + "step": 3498 + }, + { + "epoch": 68.01, + "learning_rate": 0.0001671111111111111, + "loss": 0.1188, + "step": 3499 + }, + { + "epoch": 68.01, + "learning_rate": 0.00016700000000000002, + "loss": 0.1372, + "step": 3500 + }, + { + "epoch": 68.01, + "eval_cer": 0.07433564759396696, + "eval_loss": 0.3414906859397888, + "eval_runtime": 48.7667, + "eval_samples_per_second": 8.53, + "eval_steps_per_second": 0.267, + "eval_wer": 0.31888798037612426, + "step": 3500 + }, + { + "epoch": 68.01, + "learning_rate": 0.0001668888888888889, + "loss": 0.1345, + "step": 3501 + }, + { + "epoch": 68.01, + "learning_rate": 0.0001667777777777778, + "loss": 0.11, + "step": 3502 + }, + { + "epoch": 68.01, + "learning_rate": 0.00016666666666666666, + "loss": 0.1616, + "step": 3503 + }, + { + "epoch": 68.01, + "learning_rate": 0.00016655555555555556, + "loss": 0.1707, + "step": 3504 + }, + { + "epoch": 68.01, + "learning_rate": 0.00016644444444444446, + "loss": 0.1286, + "step": 3505 + }, + { + "epoch": 68.01, + "learning_rate": 0.00016633333333333333, + "loss": 0.1204, + "step": 3506 + }, + { + "epoch": 68.01, + "learning_rate": 0.00016622222222222223, + "loss": 0.1416, + "step": 3507 + }, + { + "epoch": 68.01, + "learning_rate": 0.0001661111111111111, + "loss": 0.1508, + "step": 3508 + }, + { + "epoch": 68.01, + "learning_rate": 0.00016600000000000002, + "loss": 0.1255, + "step": 3509 + }, + { + "epoch": 68.01, + "learning_rate": 0.0001658888888888889, + "loss": 0.1226, + "step": 3510 + }, + { + "epoch": 68.01, + "learning_rate": 0.00016577777777777777, + "loss": 0.1329, + "step": 3511 + }, + { + "epoch": 68.01, + "learning_rate": 0.00016566666666666666, + "loss": 0.1319, + "step": 3512 + }, + { + "epoch": 68.01, + "learning_rate": 0.00016555555555555556, + "loss": 0.1431, + "step": 3513 + }, + { + "epoch": 68.01, + "learning_rate": 0.00016544444444444446, + "loss": 0.1476, + "step": 3514 + }, + { + "epoch": 68.01, + "learning_rate": 0.00016533333333333333, + "loss": 0.1307, + "step": 3515 + }, + { + "epoch": 68.01, + "learning_rate": 0.00016522222222222223, + "loss": 0.1305, + "step": 3516 + }, + { + "epoch": 68.01, + "learning_rate": 0.0001651111111111111, + "loss": 0.1387, + "step": 3517 + }, + { + "epoch": 68.01, + "learning_rate": 0.000165, + "loss": 0.1544, + "step": 3518 + }, + { + "epoch": 68.01, + "learning_rate": 0.0001648888888888889, + "loss": 0.0976, + "step": 3519 + }, + { + "epoch": 69.0, + "learning_rate": 0.00016477777777777777, + "loss": 0.2129, + "step": 3520 + }, + { + "epoch": 69.0, + "learning_rate": 0.00016466666666666667, + "loss": 0.1533, + "step": 3521 + }, + { + "epoch": 69.0, + "learning_rate": 0.00016455555555555556, + "loss": 0.12, + "step": 3522 + }, + { + "epoch": 69.0, + "learning_rate": 0.00016444444444444446, + "loss": 0.1205, + "step": 3523 + }, + { + "epoch": 69.0, + "learning_rate": 0.00016433333333333333, + "loss": 0.1232, + "step": 3524 + }, + { + "epoch": 69.0, + "learning_rate": 0.0001642222222222222, + "loss": 0.1311, + "step": 3525 + }, + { + "epoch": 69.0, + "learning_rate": 0.00016411111111111113, + "loss": 0.1274, + "step": 3526 + }, + { + "epoch": 69.0, + "learning_rate": 0.000164, + "loss": 0.1169, + "step": 3527 + }, + { + "epoch": 69.0, + "learning_rate": 0.0001638888888888889, + "loss": 0.121, + "step": 3528 + }, + { + "epoch": 69.0, + "learning_rate": 0.00016377777777777777, + "loss": 0.1498, + "step": 3529 + }, + { + "epoch": 69.0, + "learning_rate": 0.00016366666666666667, + "loss": 0.153, + "step": 3530 + }, + { + "epoch": 69.0, + "learning_rate": 0.00016355555555555557, + "loss": 0.1622, + "step": 3531 + }, + { + "epoch": 69.0, + "learning_rate": 0.00016344444444444444, + "loss": 0.106, + "step": 3532 + }, + { + "epoch": 69.0, + "learning_rate": 0.00016333333333333334, + "loss": 0.1527, + "step": 3533 + }, + { + "epoch": 69.0, + "learning_rate": 0.0001632222222222222, + "loss": 0.0935, + "step": 3534 + }, + { + "epoch": 69.0, + "learning_rate": 0.00016311111111111113, + "loss": 0.186, + "step": 3535 + }, + { + "epoch": 69.0, + "learning_rate": 0.000163, + "loss": 0.1552, + "step": 3536 + }, + { + "epoch": 69.0, + "learning_rate": 0.0001628888888888889, + "loss": 0.1629, + "step": 3537 + }, + { + "epoch": 69.0, + "learning_rate": 0.00016277777777777777, + "loss": 0.1235, + "step": 3538 + }, + { + "epoch": 69.0, + "learning_rate": 0.00016266666666666665, + "loss": 0.1479, + "step": 3539 + }, + { + "epoch": 69.0, + "learning_rate": 0.00016255555555555557, + "loss": 0.1118, + "step": 3540 + }, + { + "epoch": 69.0, + "learning_rate": 0.00016244444444444444, + "loss": 0.1442, + "step": 3541 + }, + { + "epoch": 69.0, + "learning_rate": 0.00016233333333333334, + "loss": 0.1352, + "step": 3542 + }, + { + "epoch": 69.0, + "learning_rate": 0.0001622222222222222, + "loss": 0.1347, + "step": 3543 + }, + { + "epoch": 69.0, + "learning_rate": 0.00016211111111111114, + "loss": 0.136, + "step": 3544 + }, + { + "epoch": 69.01, + "learning_rate": 0.000162, + "loss": 0.1727, + "step": 3545 + }, + { + "epoch": 69.01, + "learning_rate": 0.00016188888888888888, + "loss": 0.1225, + "step": 3546 + }, + { + "epoch": 69.01, + "learning_rate": 0.00016177777777777778, + "loss": 0.1475, + "step": 3547 + }, + { + "epoch": 69.01, + "learning_rate": 0.00016166666666666665, + "loss": 0.1418, + "step": 3548 + }, + { + "epoch": 69.01, + "learning_rate": 0.00016155555555555557, + "loss": 0.1009, + "step": 3549 + }, + { + "epoch": 69.01, + "learning_rate": 0.00016144444444444444, + "loss": 0.1301, + "step": 3550 + }, + { + "epoch": 69.01, + "learning_rate": 0.00016133333333333334, + "loss": 0.139, + "step": 3551 + }, + { + "epoch": 69.01, + "learning_rate": 0.00016122222222222221, + "loss": 0.1697, + "step": 3552 + }, + { + "epoch": 69.01, + "learning_rate": 0.0001611111111111111, + "loss": 0.099, + "step": 3553 + }, + { + "epoch": 69.01, + "learning_rate": 0.000161, + "loss": 0.1643, + "step": 3554 + }, + { + "epoch": 69.01, + "learning_rate": 0.00016088888888888888, + "loss": 0.1488, + "step": 3555 + }, + { + "epoch": 69.01, + "learning_rate": 0.00016077777777777778, + "loss": 0.132, + "step": 3556 + }, + { + "epoch": 69.01, + "learning_rate": 0.00016066666666666668, + "loss": 0.1319, + "step": 3557 + }, + { + "epoch": 69.01, + "learning_rate": 0.00016055555555555558, + "loss": 0.1416, + "step": 3558 + }, + { + "epoch": 69.01, + "learning_rate": 0.00016044444444444445, + "loss": 0.1492, + "step": 3559 + }, + { + "epoch": 69.01, + "learning_rate": 0.00016033333333333332, + "loss": 0.16, + "step": 3560 + }, + { + "epoch": 69.01, + "learning_rate": 0.00016022222222222222, + "loss": 0.143, + "step": 3561 + }, + { + "epoch": 69.01, + "learning_rate": 0.00016011111111111111, + "loss": 0.1013, + "step": 3562 + }, + { + "epoch": 69.01, + "learning_rate": 0.00016, + "loss": 0.1355, + "step": 3563 + }, + { + "epoch": 69.01, + "learning_rate": 0.00015988888888888888, + "loss": 0.0898, + "step": 3564 + }, + { + "epoch": 69.01, + "learning_rate": 0.00015977777777777778, + "loss": 0.1438, + "step": 3565 + }, + { + "epoch": 69.01, + "learning_rate": 0.00015966666666666668, + "loss": 0.1544, + "step": 3566 + }, + { + "epoch": 69.01, + "learning_rate": 0.00015955555555555555, + "loss": 0.1071, + "step": 3567 + }, + { + "epoch": 69.01, + "learning_rate": 0.00015944444444444445, + "loss": 0.1335, + "step": 3568 + }, + { + "epoch": 69.01, + "learning_rate": 0.00015933333333333332, + "loss": 0.1351, + "step": 3569 + }, + { + "epoch": 69.01, + "learning_rate": 0.00015922222222222222, + "loss": 0.1516, + "step": 3570 + }, + { + "epoch": 70.0, + "learning_rate": 0.00015911111111111112, + "loss": 0.1649, + "step": 3571 + }, + { + "epoch": 70.0, + "learning_rate": 0.00015900000000000002, + "loss": 0.1206, + "step": 3572 + }, + { + "epoch": 70.0, + "learning_rate": 0.0001588888888888889, + "loss": 0.1291, + "step": 3573 + }, + { + "epoch": 70.0, + "learning_rate": 0.00015877777777777776, + "loss": 0.1321, + "step": 3574 + }, + { + "epoch": 70.0, + "learning_rate": 0.00015866666666666668, + "loss": 0.1637, + "step": 3575 + }, + { + "epoch": 70.0, + "learning_rate": 0.00015855555555555555, + "loss": 0.1282, + "step": 3576 + }, + { + "epoch": 70.0, + "learning_rate": 0.00015844444444444445, + "loss": 0.107, + "step": 3577 + }, + { + "epoch": 70.0, + "learning_rate": 0.00015833333333333332, + "loss": 0.1246, + "step": 3578 + }, + { + "epoch": 70.0, + "learning_rate": 0.00015822222222222222, + "loss": 0.1027, + "step": 3579 + }, + { + "epoch": 70.0, + "learning_rate": 0.00015811111111111112, + "loss": 0.1305, + "step": 3580 + }, + { + "epoch": 70.0, + "learning_rate": 0.000158, + "loss": 0.1197, + "step": 3581 + }, + { + "epoch": 70.0, + "learning_rate": 0.0001578888888888889, + "loss": 0.1605, + "step": 3582 + }, + { + "epoch": 70.0, + "learning_rate": 0.00015777777777777776, + "loss": 0.1672, + "step": 3583 + }, + { + "epoch": 70.0, + "learning_rate": 0.00015766666666666669, + "loss": 0.1123, + "step": 3584 + }, + { + "epoch": 70.0, + "learning_rate": 0.00015755555555555556, + "loss": 0.1121, + "step": 3585 + }, + { + "epoch": 70.0, + "learning_rate": 0.00015744444444444446, + "loss": 0.1322, + "step": 3586 + }, + { + "epoch": 70.0, + "learning_rate": 0.00015733333333333333, + "loss": 0.1547, + "step": 3587 + }, + { + "epoch": 70.0, + "learning_rate": 0.00015722222222222225, + "loss": 0.134, + "step": 3588 + }, + { + "epoch": 70.0, + "learning_rate": 0.00015711111111111112, + "loss": 0.1703, + "step": 3589 + }, + { + "epoch": 70.0, + "learning_rate": 0.000157, + "loss": 0.151, + "step": 3590 + }, + { + "epoch": 70.0, + "learning_rate": 0.0001568888888888889, + "loss": 0.1191, + "step": 3591 + }, + { + "epoch": 70.0, + "learning_rate": 0.00015677777777777776, + "loss": 0.1252, + "step": 3592 + }, + { + "epoch": 70.0, + "learning_rate": 0.0001566666666666667, + "loss": 0.1603, + "step": 3593 + }, + { + "epoch": 70.0, + "learning_rate": 0.00015655555555555556, + "loss": 0.1473, + "step": 3594 + }, + { + "epoch": 70.0, + "learning_rate": 0.00015644444444444443, + "loss": 0.1262, + "step": 3595 + }, + { + "epoch": 70.01, + "learning_rate": 0.00015633333333333333, + "loss": 0.1578, + "step": 3596 + }, + { + "epoch": 70.01, + "learning_rate": 0.00015622222222222223, + "loss": 0.1664, + "step": 3597 + }, + { + "epoch": 70.01, + "learning_rate": 0.00015611111111111113, + "loss": 0.1348, + "step": 3598 + }, + { + "epoch": 70.01, + "learning_rate": 0.000156, + "loss": 0.1334, + "step": 3599 + }, + { + "epoch": 70.01, + "learning_rate": 0.0001558888888888889, + "loss": 0.116, + "step": 3600 + }, + { + "epoch": 70.01, + "learning_rate": 0.00015577777777777777, + "loss": 0.1144, + "step": 3601 + }, + { + "epoch": 70.01, + "learning_rate": 0.0001556666666666667, + "loss": 0.1056, + "step": 3602 + }, + { + "epoch": 70.01, + "learning_rate": 0.00015555555555555556, + "loss": 0.1212, + "step": 3603 + }, + { + "epoch": 70.01, + "learning_rate": 0.00015544444444444443, + "loss": 0.0982, + "step": 3604 + }, + { + "epoch": 70.01, + "learning_rate": 0.00015533333333333333, + "loss": 0.1326, + "step": 3605 + }, + { + "epoch": 70.01, + "learning_rate": 0.00015522222222222223, + "loss": 0.1537, + "step": 3606 + }, + { + "epoch": 70.01, + "learning_rate": 0.00015511111111111113, + "loss": 0.1358, + "step": 3607 + }, + { + "epoch": 70.01, + "learning_rate": 0.000155, + "loss": 0.1308, + "step": 3608 + }, + { + "epoch": 70.01, + "learning_rate": 0.0001548888888888889, + "loss": 0.1428, + "step": 3609 + }, + { + "epoch": 70.01, + "learning_rate": 0.00015477777777777777, + "loss": 0.1198, + "step": 3610 + }, + { + "epoch": 70.01, + "learning_rate": 0.00015466666666666667, + "loss": 0.1563, + "step": 3611 + }, + { + "epoch": 70.01, + "learning_rate": 0.00015455555555555557, + "loss": 0.0763, + "step": 3612 + }, + { + "epoch": 70.01, + "learning_rate": 0.00015444444444444444, + "loss": 0.141, + "step": 3613 + }, + { + "epoch": 70.01, + "learning_rate": 0.00015433333333333334, + "loss": 0.1312, + "step": 3614 + }, + { + "epoch": 70.01, + "learning_rate": 0.00015422222222222223, + "loss": 0.1283, + "step": 3615 + }, + { + "epoch": 70.01, + "learning_rate": 0.00015411111111111113, + "loss": 0.1458, + "step": 3616 + }, + { + "epoch": 70.01, + "learning_rate": 0.000154, + "loss": 0.1553, + "step": 3617 + }, + { + "epoch": 70.01, + "learning_rate": 0.00015388888888888887, + "loss": 0.1363, + "step": 3618 + }, + { + "epoch": 70.01, + "learning_rate": 0.0001537777777777778, + "loss": 0.142, + "step": 3619 + }, + { + "epoch": 70.01, + "learning_rate": 0.00015366666666666667, + "loss": 0.1053, + "step": 3620 + }, + { + "epoch": 70.01, + "learning_rate": 0.00015355555555555557, + "loss": 0.1406, + "step": 3621 + }, + { + "epoch": 71.0, + "learning_rate": 0.00015344444444444444, + "loss": 0.2024, + "step": 3622 + }, + { + "epoch": 71.0, + "learning_rate": 0.00015333333333333334, + "loss": 0.1225, + "step": 3623 + }, + { + "epoch": 71.0, + "learning_rate": 0.00015322222222222224, + "loss": 0.1087, + "step": 3624 + }, + { + "epoch": 71.0, + "learning_rate": 0.0001531111111111111, + "loss": 0.1176, + "step": 3625 + }, + { + "epoch": 71.0, + "learning_rate": 0.000153, + "loss": 0.1075, + "step": 3626 + }, + { + "epoch": 71.0, + "learning_rate": 0.00015288888888888888, + "loss": 0.1144, + "step": 3627 + }, + { + "epoch": 71.0, + "learning_rate": 0.0001527777777777778, + "loss": 0.13, + "step": 3628 + }, + { + "epoch": 71.0, + "learning_rate": 0.00015266666666666667, + "loss": 0.1054, + "step": 3629 + }, + { + "epoch": 71.0, + "learning_rate": 0.00015255555555555557, + "loss": 0.1479, + "step": 3630 + }, + { + "epoch": 71.0, + "learning_rate": 0.00015244444444444444, + "loss": 0.1159, + "step": 3631 + }, + { + "epoch": 71.0, + "learning_rate": 0.00015233333333333331, + "loss": 0.106, + "step": 3632 + }, + { + "epoch": 71.0, + "learning_rate": 0.00015222222222222224, + "loss": 0.13, + "step": 3633 + }, + { + "epoch": 71.0, + "learning_rate": 0.0001521111111111111, + "loss": 0.1265, + "step": 3634 + }, + { + "epoch": 71.0, + "learning_rate": 0.000152, + "loss": 0.1262, + "step": 3635 + }, + { + "epoch": 71.0, + "learning_rate": 0.00015188888888888888, + "loss": 0.1491, + "step": 3636 + }, + { + "epoch": 71.0, + "learning_rate": 0.0001517777777777778, + "loss": 0.1212, + "step": 3637 + }, + { + "epoch": 71.0, + "learning_rate": 0.00015166666666666668, + "loss": 0.1375, + "step": 3638 + }, + { + "epoch": 71.0, + "learning_rate": 0.00015155555555555555, + "loss": 0.114, + "step": 3639 + }, + { + "epoch": 71.0, + "learning_rate": 0.00015144444444444445, + "loss": 0.1658, + "step": 3640 + }, + { + "epoch": 71.0, + "learning_rate": 0.00015133333333333332, + "loss": 0.1225, + "step": 3641 + }, + { + "epoch": 71.0, + "learning_rate": 0.00015122222222222224, + "loss": 0.1385, + "step": 3642 + }, + { + "epoch": 71.0, + "learning_rate": 0.0001511111111111111, + "loss": 0.1164, + "step": 3643 + }, + { + "epoch": 71.0, + "learning_rate": 0.000151, + "loss": 0.1336, + "step": 3644 + }, + { + "epoch": 71.0, + "learning_rate": 0.00015088888888888888, + "loss": 0.1192, + "step": 3645 + }, + { + "epoch": 71.0, + "learning_rate": 0.00015077777777777778, + "loss": 0.1653, + "step": 3646 + }, + { + "epoch": 71.01, + "learning_rate": 0.00015066666666666668, + "loss": 0.1065, + "step": 3647 + }, + { + "epoch": 71.01, + "learning_rate": 0.00015055555555555555, + "loss": 0.1488, + "step": 3648 + }, + { + "epoch": 71.01, + "learning_rate": 0.00015044444444444445, + "loss": 0.1578, + "step": 3649 + }, + { + "epoch": 71.01, + "learning_rate": 0.00015033333333333335, + "loss": 0.1237, + "step": 3650 + }, + { + "epoch": 71.01, + "learning_rate": 0.00015022222222222224, + "loss": 0.1809, + "step": 3651 + }, + { + "epoch": 71.01, + "learning_rate": 0.00015011111111111112, + "loss": 0.111, + "step": 3652 + }, + { + "epoch": 71.01, + "learning_rate": 0.00015, + "loss": 0.1382, + "step": 3653 + }, + { + "epoch": 71.01, + "learning_rate": 0.00014988888888888889, + "loss": 0.1098, + "step": 3654 + }, + { + "epoch": 71.01, + "learning_rate": 0.00014977777777777778, + "loss": 0.1319, + "step": 3655 + }, + { + "epoch": 71.01, + "learning_rate": 0.00014966666666666668, + "loss": 0.1437, + "step": 3656 + }, + { + "epoch": 71.01, + "learning_rate": 0.00014955555555555555, + "loss": 0.1545, + "step": 3657 + }, + { + "epoch": 71.01, + "learning_rate": 0.00014944444444444445, + "loss": 0.1727, + "step": 3658 + }, + { + "epoch": 71.01, + "learning_rate": 0.00014933333333333335, + "loss": 0.1269, + "step": 3659 + }, + { + "epoch": 71.01, + "learning_rate": 0.00014922222222222222, + "loss": 0.1566, + "step": 3660 + }, + { + "epoch": 71.01, + "learning_rate": 0.00014911111111111112, + "loss": 0.1992, + "step": 3661 + }, + { + "epoch": 71.01, + "learning_rate": 0.000149, + "loss": 0.0992, + "step": 3662 + }, + { + "epoch": 71.01, + "learning_rate": 0.0001488888888888889, + "loss": 0.1466, + "step": 3663 + }, + { + "epoch": 71.01, + "learning_rate": 0.00014877777777777779, + "loss": 0.1572, + "step": 3664 + }, + { + "epoch": 71.01, + "learning_rate": 0.00014866666666666668, + "loss": 0.1125, + "step": 3665 + }, + { + "epoch": 71.01, + "learning_rate": 0.00014855555555555556, + "loss": 0.1285, + "step": 3666 + }, + { + "epoch": 71.01, + "learning_rate": 0.00014844444444444443, + "loss": 0.1272, + "step": 3667 + }, + { + "epoch": 71.01, + "learning_rate": 0.00014833333333333335, + "loss": 0.1462, + "step": 3668 + }, + { + "epoch": 71.01, + "learning_rate": 0.00014822222222222222, + "loss": 0.1357, + "step": 3669 + }, + { + "epoch": 71.01, + "learning_rate": 0.00014811111111111112, + "loss": 0.1452, + "step": 3670 + }, + { + "epoch": 71.01, + "learning_rate": 0.000148, + "loss": 0.1139, + "step": 3671 + }, + { + "epoch": 71.01, + "learning_rate": 0.0001478888888888889, + "loss": 0.1452, + "step": 3672 + }, + { + "epoch": 72.0, + "learning_rate": 0.0001477777777777778, + "loss": 0.2024, + "step": 3673 + }, + { + "epoch": 72.0, + "learning_rate": 0.00014766666666666666, + "loss": 0.1032, + "step": 3674 + }, + { + "epoch": 72.0, + "learning_rate": 0.00014755555555555556, + "loss": 0.1305, + "step": 3675 + }, + { + "epoch": 72.0, + "learning_rate": 0.00014744444444444443, + "loss": 0.1149, + "step": 3676 + }, + { + "epoch": 72.0, + "learning_rate": 0.00014733333333333335, + "loss": 0.1176, + "step": 3677 + }, + { + "epoch": 72.0, + "learning_rate": 0.00014722222222222223, + "loss": 0.1315, + "step": 3678 + }, + { + "epoch": 72.0, + "learning_rate": 0.00014711111111111112, + "loss": 0.1122, + "step": 3679 + }, + { + "epoch": 72.0, + "learning_rate": 0.000147, + "loss": 0.125, + "step": 3680 + }, + { + "epoch": 72.0, + "learning_rate": 0.0001468888888888889, + "loss": 0.1171, + "step": 3681 + }, + { + "epoch": 72.0, + "learning_rate": 0.0001467777777777778, + "loss": 0.1184, + "step": 3682 + }, + { + "epoch": 72.0, + "learning_rate": 0.00014666666666666666, + "loss": 0.1134, + "step": 3683 + }, + { + "epoch": 72.0, + "learning_rate": 0.00014655555555555556, + "loss": 0.1019, + "step": 3684 + }, + { + "epoch": 72.0, + "learning_rate": 0.00014644444444444443, + "loss": 0.1122, + "step": 3685 + }, + { + "epoch": 72.0, + "learning_rate": 0.00014633333333333336, + "loss": 0.1089, + "step": 3686 + }, + { + "epoch": 72.0, + "learning_rate": 0.00014622222222222223, + "loss": 0.1411, + "step": 3687 + }, + { + "epoch": 72.0, + "learning_rate": 0.0001461111111111111, + "loss": 0.1343, + "step": 3688 + }, + { + "epoch": 72.0, + "learning_rate": 0.000146, + "loss": 0.132, + "step": 3689 + }, + { + "epoch": 72.0, + "learning_rate": 0.0001458888888888889, + "loss": 0.1113, + "step": 3690 + }, + { + "epoch": 72.0, + "learning_rate": 0.0001457777777777778, + "loss": 0.1034, + "step": 3691 + }, + { + "epoch": 72.0, + "learning_rate": 0.00014566666666666667, + "loss": 0.1172, + "step": 3692 + }, + { + "epoch": 72.0, + "learning_rate": 0.00014555555555555556, + "loss": 0.1381, + "step": 3693 + }, + { + "epoch": 72.0, + "learning_rate": 0.00014544444444444444, + "loss": 0.134, + "step": 3694 + }, + { + "epoch": 72.0, + "learning_rate": 0.00014533333333333333, + "loss": 0.2234, + "step": 3695 + }, + { + "epoch": 72.0, + "learning_rate": 0.00014522222222222223, + "loss": 0.1631, + "step": 3696 + }, + { + "epoch": 72.0, + "learning_rate": 0.0001451111111111111, + "loss": 0.1957, + "step": 3697 + }, + { + "epoch": 72.01, + "learning_rate": 0.000145, + "loss": 0.1034, + "step": 3698 + }, + { + "epoch": 72.01, + "learning_rate": 0.0001448888888888889, + "loss": 0.1414, + "step": 3699 + }, + { + "epoch": 72.01, + "learning_rate": 0.0001447777777777778, + "loss": 0.1389, + "step": 3700 + }, + { + "epoch": 72.01, + "learning_rate": 0.00014466666666666667, + "loss": 0.1075, + "step": 3701 + }, + { + "epoch": 72.01, + "learning_rate": 0.00014455555555555554, + "loss": 0.1454, + "step": 3702 + }, + { + "epoch": 72.01, + "learning_rate": 0.00014444444444444444, + "loss": 0.1245, + "step": 3703 + }, + { + "epoch": 72.01, + "learning_rate": 0.00014433333333333334, + "loss": 0.1467, + "step": 3704 + }, + { + "epoch": 72.01, + "learning_rate": 0.00014422222222222223, + "loss": 0.1157, + "step": 3705 + }, + { + "epoch": 72.01, + "learning_rate": 0.0001441111111111111, + "loss": 0.1066, + "step": 3706 + }, + { + "epoch": 72.01, + "learning_rate": 0.000144, + "loss": 0.1049, + "step": 3707 + }, + { + "epoch": 72.01, + "learning_rate": 0.0001438888888888889, + "loss": 0.1317, + "step": 3708 + }, + { + "epoch": 72.01, + "learning_rate": 0.00014377777777777777, + "loss": 0.0954, + "step": 3709 + }, + { + "epoch": 72.01, + "learning_rate": 0.00014366666666666667, + "loss": 0.1274, + "step": 3710 + }, + { + "epoch": 72.01, + "learning_rate": 0.00014355555555555554, + "loss": 0.1388, + "step": 3711 + }, + { + "epoch": 72.01, + "learning_rate": 0.00014344444444444447, + "loss": 0.1218, + "step": 3712 + }, + { + "epoch": 72.01, + "learning_rate": 0.00014333333333333334, + "loss": 0.1451, + "step": 3713 + }, + { + "epoch": 72.01, + "learning_rate": 0.00014322222222222224, + "loss": 0.1233, + "step": 3714 + }, + { + "epoch": 72.01, + "learning_rate": 0.0001431111111111111, + "loss": 0.1115, + "step": 3715 + }, + { + "epoch": 72.01, + "learning_rate": 0.00014299999999999998, + "loss": 0.1356, + "step": 3716 + }, + { + "epoch": 72.01, + "learning_rate": 0.0001428888888888889, + "loss": 0.1153, + "step": 3717 + }, + { + "epoch": 72.01, + "learning_rate": 0.00014277777777777778, + "loss": 0.1099, + "step": 3718 + }, + { + "epoch": 72.01, + "learning_rate": 0.00014266666666666667, + "loss": 0.146, + "step": 3719 + }, + { + "epoch": 72.01, + "learning_rate": 0.00014255555555555555, + "loss": 0.1325, + "step": 3720 + }, + { + "epoch": 72.01, + "learning_rate": 0.00014244444444444447, + "loss": 0.1392, + "step": 3721 + }, + { + "epoch": 72.01, + "learning_rate": 0.00014233333333333334, + "loss": 0.1266, + "step": 3722 + }, + { + "epoch": 72.01, + "learning_rate": 0.0001422222222222222, + "loss": 0.1112, + "step": 3723 + }, + { + "epoch": 73.0, + "learning_rate": 0.0001421111111111111, + "loss": 0.174, + "step": 3724 + }, + { + "epoch": 73.0, + "learning_rate": 0.00014199999999999998, + "loss": 0.1503, + "step": 3725 + }, + { + "epoch": 73.0, + "learning_rate": 0.0001418888888888889, + "loss": 0.1142, + "step": 3726 + }, + { + "epoch": 73.0, + "learning_rate": 0.00014177777777777778, + "loss": 0.0885, + "step": 3727 + }, + { + "epoch": 73.0, + "learning_rate": 0.00014166666666666668, + "loss": 0.0988, + "step": 3728 + }, + { + "epoch": 73.0, + "learning_rate": 0.00014155555555555555, + "loss": 0.089, + "step": 3729 + }, + { + "epoch": 73.0, + "learning_rate": 0.00014144444444444445, + "loss": 0.1133, + "step": 3730 + }, + { + "epoch": 73.0, + "learning_rate": 0.00014133333333333334, + "loss": 0.1339, + "step": 3731 + }, + { + "epoch": 73.0, + "learning_rate": 0.00014122222222222222, + "loss": 0.1444, + "step": 3732 + }, + { + "epoch": 73.0, + "learning_rate": 0.00014111111111111111, + "loss": 0.1187, + "step": 3733 + }, + { + "epoch": 73.0, + "learning_rate": 0.00014099999999999998, + "loss": 0.1159, + "step": 3734 + }, + { + "epoch": 73.0, + "learning_rate": 0.0001408888888888889, + "loss": 0.1548, + "step": 3735 + }, + { + "epoch": 73.0, + "learning_rate": 0.00014077777777777778, + "loss": 0.1212, + "step": 3736 + }, + { + "epoch": 73.0, + "learning_rate": 0.00014066666666666665, + "loss": 0.0926, + "step": 3737 + }, + { + "epoch": 73.0, + "learning_rate": 0.00014055555555555555, + "loss": 0.1414, + "step": 3738 + }, + { + "epoch": 73.0, + "learning_rate": 0.00014044444444444445, + "loss": 0.1346, + "step": 3739 + }, + { + "epoch": 73.0, + "learning_rate": 0.00014033333333333335, + "loss": 0.1008, + "step": 3740 + }, + { + "epoch": 73.0, + "learning_rate": 0.00014022222222222222, + "loss": 0.132, + "step": 3741 + }, + { + "epoch": 73.0, + "learning_rate": 0.00014011111111111112, + "loss": 0.1744, + "step": 3742 + }, + { + "epoch": 73.0, + "learning_rate": 0.00014000000000000001, + "loss": 0.0966, + "step": 3743 + }, + { + "epoch": 73.0, + "learning_rate": 0.00013988888888888889, + "loss": 0.1106, + "step": 3744 + }, + { + "epoch": 73.0, + "learning_rate": 0.00013977777777777778, + "loss": 0.1241, + "step": 3745 + }, + { + "epoch": 73.0, + "learning_rate": 0.00013966666666666666, + "loss": 0.1335, + "step": 3746 + }, + { + "epoch": 73.0, + "learning_rate": 0.00013955555555555555, + "loss": 0.1151, + "step": 3747 + }, + { + "epoch": 73.0, + "learning_rate": 0.00013944444444444445, + "loss": 0.1175, + "step": 3748 + }, + { + "epoch": 73.01, + "learning_rate": 0.00013933333333333335, + "loss": 0.1181, + "step": 3749 + }, + { + "epoch": 73.01, + "learning_rate": 0.00013922222222222222, + "loss": 0.1024, + "step": 3750 + }, + { + "epoch": 73.01, + "learning_rate": 0.0001391111111111111, + "loss": 0.1414, + "step": 3751 + }, + { + "epoch": 73.01, + "learning_rate": 0.00013900000000000002, + "loss": 0.1661, + "step": 3752 + }, + { + "epoch": 73.01, + "learning_rate": 0.0001388888888888889, + "loss": 0.1275, + "step": 3753 + }, + { + "epoch": 73.01, + "learning_rate": 0.0001387777777777778, + "loss": 0.1294, + "step": 3754 + }, + { + "epoch": 73.01, + "learning_rate": 0.00013866666666666666, + "loss": 0.0911, + "step": 3755 + }, + { + "epoch": 73.01, + "learning_rate": 0.00013855555555555556, + "loss": 0.1484, + "step": 3756 + }, + { + "epoch": 73.01, + "learning_rate": 0.00013844444444444445, + "loss": 0.1402, + "step": 3757 + }, + { + "epoch": 73.01, + "learning_rate": 0.00013833333333333333, + "loss": 0.1045, + "step": 3758 + }, + { + "epoch": 73.01, + "learning_rate": 0.00013822222222222222, + "loss": 0.1319, + "step": 3759 + }, + { + "epoch": 73.01, + "learning_rate": 0.0001381111111111111, + "loss": 0.1493, + "step": 3760 + }, + { + "epoch": 73.01, + "learning_rate": 0.00013800000000000002, + "loss": 0.1078, + "step": 3761 + }, + { + "epoch": 73.01, + "learning_rate": 0.0001378888888888889, + "loss": 0.1054, + "step": 3762 + }, + { + "epoch": 73.01, + "learning_rate": 0.0001377777777777778, + "loss": 0.0956, + "step": 3763 + }, + { + "epoch": 73.01, + "learning_rate": 0.00013766666666666666, + "loss": 0.1203, + "step": 3764 + }, + { + "epoch": 73.01, + "learning_rate": 0.00013755555555555553, + "loss": 0.1186, + "step": 3765 + }, + { + "epoch": 73.01, + "learning_rate": 0.00013744444444444446, + "loss": 0.1316, + "step": 3766 + }, + { + "epoch": 73.01, + "learning_rate": 0.00013733333333333333, + "loss": 0.1019, + "step": 3767 + }, + { + "epoch": 73.01, + "learning_rate": 0.00013722222222222223, + "loss": 0.0943, + "step": 3768 + }, + { + "epoch": 73.01, + "learning_rate": 0.0001371111111111111, + "loss": 0.0932, + "step": 3769 + }, + { + "epoch": 73.01, + "learning_rate": 0.00013700000000000002, + "loss": 0.0954, + "step": 3770 + }, + { + "epoch": 73.01, + "learning_rate": 0.0001368888888888889, + "loss": 0.1081, + "step": 3771 + }, + { + "epoch": 73.01, + "learning_rate": 0.00013677777777777777, + "loss": 0.144, + "step": 3772 + }, + { + "epoch": 73.01, + "learning_rate": 0.00013666666666666666, + "loss": 0.1472, + "step": 3773 + }, + { + "epoch": 73.01, + "learning_rate": 0.00013655555555555556, + "loss": 0.098, + "step": 3774 + }, + { + "epoch": 74.0, + "learning_rate": 0.00013644444444444446, + "loss": 0.1658, + "step": 3775 + }, + { + "epoch": 74.0, + "learning_rate": 0.00013633333333333333, + "loss": 0.0944, + "step": 3776 + }, + { + "epoch": 74.0, + "learning_rate": 0.00013622222222222223, + "loss": 0.09, + "step": 3777 + }, + { + "epoch": 74.0, + "learning_rate": 0.0001361111111111111, + "loss": 0.1152, + "step": 3778 + }, + { + "epoch": 74.0, + "learning_rate": 0.00013600000000000003, + "loss": 0.1058, + "step": 3779 + }, + { + "epoch": 74.0, + "learning_rate": 0.0001358888888888889, + "loss": 0.1029, + "step": 3780 + }, + { + "epoch": 74.0, + "learning_rate": 0.00013577777777777777, + "loss": 0.1031, + "step": 3781 + }, + { + "epoch": 74.0, + "learning_rate": 0.00013566666666666667, + "loss": 0.0884, + "step": 3782 + }, + { + "epoch": 74.0, + "learning_rate": 0.00013555555555555556, + "loss": 0.1066, + "step": 3783 + }, + { + "epoch": 74.0, + "learning_rate": 0.00013544444444444446, + "loss": 0.1198, + "step": 3784 + }, + { + "epoch": 74.0, + "learning_rate": 0.00013533333333333333, + "loss": 0.1262, + "step": 3785 + }, + { + "epoch": 74.0, + "learning_rate": 0.0001352222222222222, + "loss": 0.1261, + "step": 3786 + }, + { + "epoch": 74.0, + "learning_rate": 0.0001351111111111111, + "loss": 0.086, + "step": 3787 + }, + { + "epoch": 74.0, + "learning_rate": 0.000135, + "loss": 0.1039, + "step": 3788 + }, + { + "epoch": 74.0, + "learning_rate": 0.0001348888888888889, + "loss": 0.1186, + "step": 3789 + }, + { + "epoch": 74.0, + "learning_rate": 0.00013477777777777777, + "loss": 0.0977, + "step": 3790 + }, + { + "epoch": 74.0, + "learning_rate": 0.00013466666666666667, + "loss": 0.1019, + "step": 3791 + }, + { + "epoch": 74.0, + "learning_rate": 0.00013455555555555557, + "loss": 0.1468, + "step": 3792 + }, + { + "epoch": 74.0, + "learning_rate": 0.00013444444444444447, + "loss": 0.1236, + "step": 3793 + }, + { + "epoch": 74.0, + "learning_rate": 0.00013433333333333334, + "loss": 0.1275, + "step": 3794 + }, + { + "epoch": 74.0, + "learning_rate": 0.0001342222222222222, + "loss": 0.1117, + "step": 3795 + }, + { + "epoch": 74.0, + "learning_rate": 0.0001341111111111111, + "loss": 0.1041, + "step": 3796 + }, + { + "epoch": 74.0, + "learning_rate": 0.000134, + "loss": 0.0949, + "step": 3797 + }, + { + "epoch": 74.0, + "learning_rate": 0.0001338888888888889, + "loss": 0.1245, + "step": 3798 + }, + { + "epoch": 74.0, + "learning_rate": 0.00013377777777777777, + "loss": 0.13, + "step": 3799 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013366666666666667, + "loss": 0.087, + "step": 3800 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013355555555555557, + "loss": 0.1211, + "step": 3801 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013344444444444444, + "loss": 0.1176, + "step": 3802 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013333333333333334, + "loss": 0.1051, + "step": 3803 + }, + { + "epoch": 74.01, + "learning_rate": 0.0001332222222222222, + "loss": 0.1127, + "step": 3804 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013311111111111114, + "loss": 0.1157, + "step": 3805 + }, + { + "epoch": 74.01, + "learning_rate": 0.000133, + "loss": 0.1149, + "step": 3806 + }, + { + "epoch": 74.01, + "learning_rate": 0.0001328888888888889, + "loss": 0.097, + "step": 3807 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013277777777777778, + "loss": 0.1399, + "step": 3808 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013266666666666665, + "loss": 0.1291, + "step": 3809 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013255555555555557, + "loss": 0.1241, + "step": 3810 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013244444444444444, + "loss": 0.1221, + "step": 3811 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013233333333333334, + "loss": 0.121, + "step": 3812 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013222222222222221, + "loss": 0.0954, + "step": 3813 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013211111111111114, + "loss": 0.0947, + "step": 3814 + }, + { + "epoch": 74.01, + "learning_rate": 0.000132, + "loss": 0.1358, + "step": 3815 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013188888888888888, + "loss": 0.0849, + "step": 3816 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013177777777777778, + "loss": 0.0987, + "step": 3817 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013166666666666665, + "loss": 0.1067, + "step": 3818 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013155555555555558, + "loss": 0.0933, + "step": 3819 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013144444444444445, + "loss": 0.1511, + "step": 3820 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013133333333333335, + "loss": 0.1151, + "step": 3821 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013122222222222222, + "loss": 0.1185, + "step": 3822 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013111111111111111, + "loss": 0.1047, + "step": 3823 + }, + { + "epoch": 74.01, + "learning_rate": 0.000131, + "loss": 0.0973, + "step": 3824 + }, + { + "epoch": 74.01, + "learning_rate": 0.00013088888888888888, + "loss": 0.1151, + "step": 3825 + }, + { + "epoch": 75.0, + "learning_rate": 0.00013077777777777778, + "loss": 0.1439, + "step": 3826 + }, + { + "epoch": 75.0, + "learning_rate": 0.00013066666666666665, + "loss": 0.0843, + "step": 3827 + }, + { + "epoch": 75.0, + "learning_rate": 0.00013055555555555558, + "loss": 0.0952, + "step": 3828 + }, + { + "epoch": 75.0, + "learning_rate": 0.00013044444444444445, + "loss": 0.103, + "step": 3829 + }, + { + "epoch": 75.0, + "learning_rate": 0.00013033333333333332, + "loss": 0.1373, + "step": 3830 + }, + { + "epoch": 75.0, + "learning_rate": 0.00013022222222222222, + "loss": 0.0886, + "step": 3831 + }, + { + "epoch": 75.0, + "learning_rate": 0.00013011111111111112, + "loss": 0.1077, + "step": 3832 + }, + { + "epoch": 75.0, + "learning_rate": 0.00013000000000000002, + "loss": 0.1208, + "step": 3833 + }, + { + "epoch": 75.0, + "learning_rate": 0.0001298888888888889, + "loss": 0.0843, + "step": 3834 + }, + { + "epoch": 75.0, + "learning_rate": 0.00012977777777777779, + "loss": 0.0874, + "step": 3835 + }, + { + "epoch": 75.0, + "learning_rate": 0.00012966666666666668, + "loss": 0.1013, + "step": 3836 + }, + { + "epoch": 75.0, + "learning_rate": 0.00012955555555555555, + "loss": 0.1194, + "step": 3837 + }, + { + "epoch": 75.0, + "learning_rate": 0.00012944444444444445, + "loss": 0.104, + "step": 3838 + }, + { + "epoch": 75.0, + "learning_rate": 0.00012933333333333332, + "loss": 0.1182, + "step": 3839 + }, + { + "epoch": 75.0, + "learning_rate": 0.00012922222222222222, + "loss": 0.1034, + "step": 3840 + }, + { + "epoch": 75.0, + "learning_rate": 0.00012911111111111112, + "loss": 0.0845, + "step": 3841 + }, + { + "epoch": 75.0, + "learning_rate": 0.00012900000000000002, + "loss": 0.1292, + "step": 3842 + }, + { + "epoch": 75.0, + "learning_rate": 0.0001288888888888889, + "loss": 0.1318, + "step": 3843 + }, + { + "epoch": 75.0, + "learning_rate": 0.00012877777777777776, + "loss": 0.0822, + "step": 3844 + }, + { + "epoch": 75.0, + "learning_rate": 0.00012866666666666669, + "loss": 0.1329, + "step": 3845 + }, + { + "epoch": 75.0, + "learning_rate": 0.00012855555555555556, + "loss": 0.126, + "step": 3846 + }, + { + "epoch": 75.0, + "learning_rate": 0.00012844444444444446, + "loss": 0.1534, + "step": 3847 + }, + { + "epoch": 75.0, + "learning_rate": 0.00012833333333333333, + "loss": 0.1086, + "step": 3848 + }, + { + "epoch": 75.0, + "learning_rate": 0.00012822222222222222, + "loss": 0.1627, + "step": 3849 + }, + { + "epoch": 75.0, + "learning_rate": 0.00012811111111111112, + "loss": 0.143, + "step": 3850 + }, + { + "epoch": 75.01, + "learning_rate": 0.000128, + "loss": 0.1272, + "step": 3851 + }, + { + "epoch": 75.01, + "learning_rate": 0.0001278888888888889, + "loss": 0.0954, + "step": 3852 + }, + { + "epoch": 75.01, + "learning_rate": 0.00012777777777777776, + "loss": 0.0997, + "step": 3853 + }, + { + "epoch": 75.01, + "learning_rate": 0.0001276666666666667, + "loss": 0.1009, + "step": 3854 + }, + { + "epoch": 75.01, + "learning_rate": 0.00012755555555555556, + "loss": 0.164, + "step": 3855 + }, + { + "epoch": 75.01, + "learning_rate": 0.00012744444444444446, + "loss": 0.1245, + "step": 3856 + }, + { + "epoch": 75.01, + "learning_rate": 0.00012733333333333333, + "loss": 0.1253, + "step": 3857 + }, + { + "epoch": 75.01, + "learning_rate": 0.0001272222222222222, + "loss": 0.0863, + "step": 3858 + }, + { + "epoch": 75.01, + "learning_rate": 0.00012711111111111113, + "loss": 0.0998, + "step": 3859 + }, + { + "epoch": 75.01, + "learning_rate": 0.000127, + "loss": 0.1106, + "step": 3860 + }, + { + "epoch": 75.01, + "learning_rate": 0.0001268888888888889, + "loss": 0.1233, + "step": 3861 + }, + { + "epoch": 75.01, + "learning_rate": 0.00012677777777777777, + "loss": 0.0935, + "step": 3862 + }, + { + "epoch": 75.01, + "learning_rate": 0.0001266666666666667, + "loss": 0.0854, + "step": 3863 + }, + { + "epoch": 75.01, + "learning_rate": 0.00012655555555555556, + "loss": 0.1161, + "step": 3864 + }, + { + "epoch": 75.01, + "learning_rate": 0.00012644444444444443, + "loss": 0.1309, + "step": 3865 + }, + { + "epoch": 75.01, + "learning_rate": 0.00012633333333333333, + "loss": 0.1007, + "step": 3866 + }, + { + "epoch": 75.01, + "learning_rate": 0.00012622222222222223, + "loss": 0.1118, + "step": 3867 + }, + { + "epoch": 75.01, + "learning_rate": 0.00012611111111111113, + "loss": 0.1028, + "step": 3868 + }, + { + "epoch": 75.01, + "learning_rate": 0.000126, + "loss": 0.1101, + "step": 3869 + }, + { + "epoch": 75.01, + "learning_rate": 0.0001258888888888889, + "loss": 0.1042, + "step": 3870 + }, + { + "epoch": 75.01, + "learning_rate": 0.00012577777777777777, + "loss": 0.1147, + "step": 3871 + }, + { + "epoch": 75.01, + "learning_rate": 0.00012566666666666667, + "loss": 0.1016, + "step": 3872 + }, + { + "epoch": 75.01, + "learning_rate": 0.00012555555555555557, + "loss": 0.1435, + "step": 3873 + }, + { + "epoch": 75.01, + "learning_rate": 0.00012544444444444444, + "loss": 0.105, + "step": 3874 + }, + { + "epoch": 75.01, + "learning_rate": 0.00012533333333333334, + "loss": 0.1116, + "step": 3875 + }, + { + "epoch": 75.01, + "learning_rate": 0.00012522222222222223, + "loss": 0.1064, + "step": 3876 + }, + { + "epoch": 76.0, + "learning_rate": 0.00012511111111111113, + "loss": 0.1416, + "step": 3877 + }, + { + "epoch": 76.0, + "learning_rate": 0.000125, + "loss": 0.144, + "step": 3878 + }, + { + "epoch": 76.0, + "learning_rate": 0.00012488888888888887, + "loss": 0.0972, + "step": 3879 + }, + { + "epoch": 76.0, + "learning_rate": 0.00012477777777777777, + "loss": 0.1032, + "step": 3880 + }, + { + "epoch": 76.0, + "learning_rate": 0.00012466666666666667, + "loss": 0.1658, + "step": 3881 + }, + { + "epoch": 76.0, + "learning_rate": 0.00012455555555555557, + "loss": 0.1349, + "step": 3882 + }, + { + "epoch": 76.0, + "learning_rate": 0.00012444444444444444, + "loss": 0.0853, + "step": 3883 + }, + { + "epoch": 76.0, + "learning_rate": 0.00012433333333333334, + "loss": 0.1041, + "step": 3884 + }, + { + "epoch": 76.0, + "learning_rate": 0.00012422222222222224, + "loss": 0.1113, + "step": 3885 + }, + { + "epoch": 76.0, + "learning_rate": 0.0001241111111111111, + "loss": 0.1411, + "step": 3886 + }, + { + "epoch": 76.0, + "learning_rate": 0.000124, + "loss": 0.1218, + "step": 3887 + }, + { + "epoch": 76.0, + "learning_rate": 0.0001238888888888889, + "loss": 0.1584, + "step": 3888 + }, + { + "epoch": 76.0, + "learning_rate": 0.00012377777777777777, + "loss": 0.102, + "step": 3889 + }, + { + "epoch": 76.0, + "learning_rate": 0.00012366666666666667, + "loss": 0.1106, + "step": 3890 + }, + { + "epoch": 76.0, + "learning_rate": 0.00012355555555555557, + "loss": 0.0972, + "step": 3891 + }, + { + "epoch": 76.0, + "learning_rate": 0.00012344444444444444, + "loss": 0.0947, + "step": 3892 + }, + { + "epoch": 76.0, + "learning_rate": 0.00012333333333333334, + "loss": 0.1018, + "step": 3893 + }, + { + "epoch": 76.0, + "learning_rate": 0.0001232222222222222, + "loss": 0.1261, + "step": 3894 + }, + { + "epoch": 76.0, + "learning_rate": 0.0001231111111111111, + "loss": 0.1517, + "step": 3895 + }, + { + "epoch": 76.0, + "learning_rate": 0.000123, + "loss": 0.1234, + "step": 3896 + }, + { + "epoch": 76.0, + "learning_rate": 0.0001228888888888889, + "loss": 0.1046, + "step": 3897 + }, + { + "epoch": 76.0, + "learning_rate": 0.00012277777777777778, + "loss": 0.1068, + "step": 3898 + }, + { + "epoch": 76.0, + "learning_rate": 0.00012266666666666668, + "loss": 0.1866, + "step": 3899 + }, + { + "epoch": 76.0, + "learning_rate": 0.00012255555555555555, + "loss": 0.1641, + "step": 3900 + }, + { + "epoch": 76.0, + "learning_rate": 0.00012244444444444445, + "loss": 0.0908, + "step": 3901 + }, + { + "epoch": 76.01, + "learning_rate": 0.00012233333333333334, + "loss": 0.1211, + "step": 3902 + }, + { + "epoch": 76.01, + "learning_rate": 0.00012222222222222221, + "loss": 0.0888, + "step": 3903 + }, + { + "epoch": 76.01, + "learning_rate": 0.0001221111111111111, + "loss": 0.1059, + "step": 3904 + }, + { + "epoch": 76.01, + "learning_rate": 0.000122, + "loss": 0.0849, + "step": 3905 + }, + { + "epoch": 76.01, + "learning_rate": 0.0001218888888888889, + "loss": 0.1142, + "step": 3906 + }, + { + "epoch": 76.01, + "learning_rate": 0.00012177777777777778, + "loss": 0.1043, + "step": 3907 + }, + { + "epoch": 76.01, + "learning_rate": 0.00012166666666666668, + "loss": 0.1087, + "step": 3908 + }, + { + "epoch": 76.01, + "learning_rate": 0.00012155555555555556, + "loss": 0.1168, + "step": 3909 + }, + { + "epoch": 76.01, + "learning_rate": 0.00012144444444444443, + "loss": 0.1158, + "step": 3910 + }, + { + "epoch": 76.01, + "learning_rate": 0.00012133333333333333, + "loss": 0.1034, + "step": 3911 + }, + { + "epoch": 76.01, + "learning_rate": 0.00012122222222222222, + "loss": 0.1166, + "step": 3912 + }, + { + "epoch": 76.01, + "learning_rate": 0.00012111111111111112, + "loss": 0.1168, + "step": 3913 + }, + { + "epoch": 76.01, + "learning_rate": 0.000121, + "loss": 0.1131, + "step": 3914 + }, + { + "epoch": 76.01, + "learning_rate": 0.0001208888888888889, + "loss": 0.121, + "step": 3915 + }, + { + "epoch": 76.01, + "learning_rate": 0.00012077777777777778, + "loss": 0.112, + "step": 3916 + }, + { + "epoch": 76.01, + "learning_rate": 0.00012066666666666667, + "loss": 0.1345, + "step": 3917 + }, + { + "epoch": 76.01, + "learning_rate": 0.00012055555555555555, + "loss": 0.117, + "step": 3918 + }, + { + "epoch": 76.01, + "learning_rate": 0.00012044444444444445, + "loss": 0.1131, + "step": 3919 + }, + { + "epoch": 76.01, + "learning_rate": 0.00012033333333333334, + "loss": 0.0898, + "step": 3920 + }, + { + "epoch": 76.01, + "learning_rate": 0.00012022222222222222, + "loss": 0.104, + "step": 3921 + }, + { + "epoch": 76.01, + "learning_rate": 0.00012011111111111112, + "loss": 0.1492, + "step": 3922 + }, + { + "epoch": 76.01, + "learning_rate": 0.00012, + "loss": 0.0979, + "step": 3923 + }, + { + "epoch": 76.01, + "learning_rate": 0.0001198888888888889, + "loss": 0.1239, + "step": 3924 + }, + { + "epoch": 76.01, + "learning_rate": 0.00011977777777777777, + "loss": 0.145, + "step": 3925 + }, + { + "epoch": 76.01, + "learning_rate": 0.00011966666666666667, + "loss": 0.0952, + "step": 3926 + }, + { + "epoch": 76.01, + "learning_rate": 0.00011955555555555556, + "loss": 0.1215, + "step": 3927 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011944444444444445, + "loss": 0.1397, + "step": 3928 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011933333333333334, + "loss": 0.0767, + "step": 3929 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011922222222222222, + "loss": 0.1056, + "step": 3930 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011911111111111112, + "loss": 0.1228, + "step": 3931 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011899999999999999, + "loss": 0.1137, + "step": 3932 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011888888888888889, + "loss": 0.1477, + "step": 3933 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011877777777777778, + "loss": 0.1048, + "step": 3934 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011866666666666667, + "loss": 0.0865, + "step": 3935 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011855555555555556, + "loss": 0.1255, + "step": 3936 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011844444444444446, + "loss": 0.1259, + "step": 3937 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011833333333333334, + "loss": 0.1048, + "step": 3938 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011822222222222223, + "loss": 0.1165, + "step": 3939 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011811111111111111, + "loss": 0.1062, + "step": 3940 + }, + { + "epoch": 77.0, + "learning_rate": 0.000118, + "loss": 0.1227, + "step": 3941 + }, + { + "epoch": 77.0, + "learning_rate": 0.0001178888888888889, + "loss": 0.1056, + "step": 3942 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011777777777777778, + "loss": 0.1105, + "step": 3943 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011766666666666668, + "loss": 0.0694, + "step": 3944 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011755555555555556, + "loss": 0.1046, + "step": 3945 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011744444444444445, + "loss": 0.1059, + "step": 3946 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011733333333333333, + "loss": 0.1067, + "step": 3947 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011722222222222223, + "loss": 0.1014, + "step": 3948 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011711111111111111, + "loss": 0.1047, + "step": 3949 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011700000000000001, + "loss": 0.0934, + "step": 3950 + }, + { + "epoch": 77.0, + "learning_rate": 0.0001168888888888889, + "loss": 0.1012, + "step": 3951 + }, + { + "epoch": 77.0, + "learning_rate": 0.00011677777777777778, + "loss": 0.1367, + "step": 3952 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011666666666666667, + "loss": 0.1137, + "step": 3953 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011655555555555555, + "loss": 0.108, + "step": 3954 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011644444444444445, + "loss": 0.1066, + "step": 3955 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011633333333333333, + "loss": 0.1063, + "step": 3956 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011622222222222223, + "loss": 0.1016, + "step": 3957 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011611111111111112, + "loss": 0.1105, + "step": 3958 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011600000000000001, + "loss": 0.1106, + "step": 3959 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011588888888888889, + "loss": 0.1134, + "step": 3960 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011577777777777777, + "loss": 0.1021, + "step": 3961 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011566666666666667, + "loss": 0.1589, + "step": 3962 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011555555555555555, + "loss": 0.1008, + "step": 3963 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011544444444444445, + "loss": 0.1315, + "step": 3964 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011533333333333334, + "loss": 0.1373, + "step": 3965 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011522222222222223, + "loss": 0.1051, + "step": 3966 + }, + { + "epoch": 77.01, + "learning_rate": 0.0001151111111111111, + "loss": 0.0935, + "step": 3967 + }, + { + "epoch": 77.01, + "learning_rate": 0.000115, + "loss": 0.0971, + "step": 3968 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011488888888888889, + "loss": 0.128, + "step": 3969 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011477777777777779, + "loss": 0.1139, + "step": 3970 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011466666666666667, + "loss": 0.0873, + "step": 3971 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011455555555555556, + "loss": 0.0796, + "step": 3972 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011444444444444445, + "loss": 0.1397, + "step": 3973 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011433333333333333, + "loss": 0.0854, + "step": 3974 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011422222222222222, + "loss": 0.0942, + "step": 3975 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011411111111111111, + "loss": 0.1178, + "step": 3976 + }, + { + "epoch": 77.01, + "learning_rate": 0.000114, + "loss": 0.0779, + "step": 3977 + }, + { + "epoch": 77.01, + "learning_rate": 0.00011388888888888889, + "loss": 0.1229, + "step": 3978 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011377777777777779, + "loss": 0.1369, + "step": 3979 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011366666666666667, + "loss": 0.1211, + "step": 3980 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011355555555555556, + "loss": 0.0771, + "step": 3981 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011344444444444444, + "loss": 0.0657, + "step": 3982 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011333333333333333, + "loss": 0.1208, + "step": 3983 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011322222222222223, + "loss": 0.0916, + "step": 3984 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011311111111111111, + "loss": 0.0892, + "step": 3985 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011300000000000001, + "loss": 0.0782, + "step": 3986 + }, + { + "epoch": 78.0, + "learning_rate": 0.0001128888888888889, + "loss": 0.1247, + "step": 3987 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011277777777777778, + "loss": 0.1003, + "step": 3988 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011266666666666666, + "loss": 0.1027, + "step": 3989 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011255555555555556, + "loss": 0.1086, + "step": 3990 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011244444444444445, + "loss": 0.0822, + "step": 3991 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011233333333333333, + "loss": 0.1293, + "step": 3992 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011222222222222223, + "loss": 0.1412, + "step": 3993 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011211111111111111, + "loss": 0.0948, + "step": 3994 + }, + { + "epoch": 78.0, + "learning_rate": 0.000112, + "loss": 0.1259, + "step": 3995 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011188888888888888, + "loss": 0.0974, + "step": 3996 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011177777777777778, + "loss": 0.0886, + "step": 3997 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011166666666666667, + "loss": 0.0992, + "step": 3998 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011155555555555556, + "loss": 0.0947, + "step": 3999 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011144444444444445, + "loss": 0.091, + "step": 4000 + }, + { + "epoch": 78.0, + "eval_cer": 0.0757122336605219, + "eval_loss": 0.3749677836894989, + "eval_runtime": 51.1989, + "eval_samples_per_second": 8.125, + "eval_steps_per_second": 0.254, + "eval_wer": 0.31643499591169255, + "step": 4000 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011133333333333335, + "loss": 0.1147, + "step": 4001 + }, + { + "epoch": 78.0, + "learning_rate": 0.00011122222222222222, + "loss": 0.1162, + "step": 4002 + }, + { + "epoch": 78.0, + "learning_rate": 0.0001111111111111111, + "loss": 0.0566, + "step": 4003 + }, + { + "epoch": 78.01, + "learning_rate": 0.000111, + "loss": 0.1179, + "step": 4004 + }, + { + "epoch": 78.01, + "learning_rate": 0.00011088888888888889, + "loss": 0.091, + "step": 4005 + }, + { + "epoch": 78.01, + "learning_rate": 0.00011077777777777778, + "loss": 0.1194, + "step": 4006 + }, + { + "epoch": 78.01, + "learning_rate": 0.00011066666666666667, + "loss": 0.1267, + "step": 4007 + }, + { + "epoch": 78.01, + "learning_rate": 0.00011055555555555557, + "loss": 0.1104, + "step": 4008 + }, + { + "epoch": 78.01, + "learning_rate": 0.00011044444444444444, + "loss": 0.1014, + "step": 4009 + }, + { + "epoch": 78.01, + "learning_rate": 0.00011033333333333334, + "loss": 0.1265, + "step": 4010 + }, + { + "epoch": 78.01, + "learning_rate": 0.00011022222222222222, + "loss": 0.1206, + "step": 4011 + }, + { + "epoch": 78.01, + "learning_rate": 0.00011011111111111112, + "loss": 0.1269, + "step": 4012 + }, + { + "epoch": 78.01, + "learning_rate": 0.00011, + "loss": 0.0922, + "step": 4013 + }, + { + "epoch": 78.01, + "learning_rate": 0.00010988888888888889, + "loss": 0.1257, + "step": 4014 + }, + { + "epoch": 78.01, + "learning_rate": 0.00010977777777777779, + "loss": 0.0674, + "step": 4015 + }, + { + "epoch": 78.01, + "learning_rate": 0.00010966666666666666, + "loss": 0.0874, + "step": 4016 + }, + { + "epoch": 78.01, + "learning_rate": 0.00010955555555555556, + "loss": 0.1282, + "step": 4017 + }, + { + "epoch": 78.01, + "learning_rate": 0.00010944444444444444, + "loss": 0.1299, + "step": 4018 + }, + { + "epoch": 78.01, + "learning_rate": 0.00010933333333333334, + "loss": 0.1126, + "step": 4019 + }, + { + "epoch": 78.01, + "learning_rate": 0.00010922222222222222, + "loss": 0.1059, + "step": 4020 + }, + { + "epoch": 78.01, + "learning_rate": 0.00010911111111111112, + "loss": 0.079, + "step": 4021 + }, + { + "epoch": 78.01, + "learning_rate": 0.000109, + "loss": 0.1097, + "step": 4022 + }, + { + "epoch": 78.01, + "learning_rate": 0.00010888888888888888, + "loss": 0.1201, + "step": 4023 + }, + { + "epoch": 78.01, + "learning_rate": 0.00010877777777777778, + "loss": 0.1247, + "step": 4024 + }, + { + "epoch": 78.01, + "learning_rate": 0.00010866666666666666, + "loss": 0.087, + "step": 4025 + }, + { + "epoch": 78.01, + "learning_rate": 0.00010855555555555556, + "loss": 0.0769, + "step": 4026 + }, + { + "epoch": 78.01, + "learning_rate": 0.00010844444444444444, + "loss": 0.0743, + "step": 4027 + }, + { + "epoch": 78.01, + "learning_rate": 0.00010833333333333334, + "loss": 0.0721, + "step": 4028 + }, + { + "epoch": 78.01, + "learning_rate": 0.00010822222222222223, + "loss": 0.1041, + "step": 4029 + }, + { + "epoch": 79.0, + "learning_rate": 0.00010811111111111112, + "loss": 0.1888, + "step": 4030 + }, + { + "epoch": 79.0, + "learning_rate": 0.000108, + "loss": 0.0839, + "step": 4031 + }, + { + "epoch": 79.0, + "learning_rate": 0.0001078888888888889, + "loss": 0.1075, + "step": 4032 + }, + { + "epoch": 79.0, + "learning_rate": 0.00010777777777777778, + "loss": 0.1055, + "step": 4033 + }, + { + "epoch": 79.0, + "learning_rate": 0.00010766666666666666, + "loss": 0.0904, + "step": 4034 + }, + { + "epoch": 79.0, + "learning_rate": 0.00010755555555555556, + "loss": 0.1102, + "step": 4035 + }, + { + "epoch": 79.0, + "learning_rate": 0.00010744444444444445, + "loss": 0.1051, + "step": 4036 + }, + { + "epoch": 79.0, + "learning_rate": 0.00010733333333333334, + "loss": 0.0944, + "step": 4037 + }, + { + "epoch": 79.0, + "learning_rate": 0.00010722222222222222, + "loss": 0.0927, + "step": 4038 + }, + { + "epoch": 79.0, + "learning_rate": 0.00010711111111111111, + "loss": 0.1186, + "step": 4039 + }, + { + "epoch": 79.0, + "learning_rate": 0.000107, + "loss": 0.0915, + "step": 4040 + }, + { + "epoch": 79.0, + "learning_rate": 0.0001068888888888889, + "loss": 0.1224, + "step": 4041 + }, + { + "epoch": 79.0, + "learning_rate": 0.00010677777777777778, + "loss": 0.1301, + "step": 4042 + }, + { + "epoch": 79.0, + "learning_rate": 0.00010666666666666668, + "loss": 0.118, + "step": 4043 + }, + { + "epoch": 79.0, + "learning_rate": 0.00010655555555555556, + "loss": 0.1226, + "step": 4044 + }, + { + "epoch": 79.0, + "learning_rate": 0.00010644444444444444, + "loss": 0.1284, + "step": 4045 + }, + { + "epoch": 79.0, + "learning_rate": 0.00010633333333333333, + "loss": 0.0899, + "step": 4046 + }, + { + "epoch": 79.0, + "learning_rate": 0.00010622222222222222, + "loss": 0.0899, + "step": 4047 + }, + { + "epoch": 79.0, + "learning_rate": 0.00010611111111111112, + "loss": 0.0893, + "step": 4048 + }, + { + "epoch": 79.0, + "learning_rate": 0.000106, + "loss": 0.0841, + "step": 4049 + }, + { + "epoch": 79.0, + "learning_rate": 0.0001058888888888889, + "loss": 0.084, + "step": 4050 + }, + { + "epoch": 79.0, + "learning_rate": 0.00010577777777777778, + "loss": 0.1095, + "step": 4051 + }, + { + "epoch": 79.0, + "learning_rate": 0.00010566666666666667, + "loss": 0.0878, + "step": 4052 + }, + { + "epoch": 79.0, + "learning_rate": 0.00010555555555555555, + "loss": 0.1418, + "step": 4053 + }, + { + "epoch": 79.0, + "learning_rate": 0.00010544444444444444, + "loss": 0.1094, + "step": 4054 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010533333333333334, + "loss": 0.099, + "step": 4055 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010522222222222222, + "loss": 0.0711, + "step": 4056 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010511111111111112, + "loss": 0.0952, + "step": 4057 + }, + { + "epoch": 79.01, + "learning_rate": 0.000105, + "loss": 0.0704, + "step": 4058 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010488888888888889, + "loss": 0.1192, + "step": 4059 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010477777777777777, + "loss": 0.1179, + "step": 4060 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010466666666666667, + "loss": 0.1101, + "step": 4061 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010455555555555556, + "loss": 0.1152, + "step": 4062 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010444444444444445, + "loss": 0.0761, + "step": 4063 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010433333333333334, + "loss": 0.1117, + "step": 4064 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010422222222222222, + "loss": 0.1202, + "step": 4065 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010411111111111111, + "loss": 0.1038, + "step": 4066 + }, + { + "epoch": 79.01, + "learning_rate": 0.000104, + "loss": 0.1552, + "step": 4067 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010388888888888889, + "loss": 0.0757, + "step": 4068 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010377777777777778, + "loss": 0.0811, + "step": 4069 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010366666666666667, + "loss": 0.0707, + "step": 4070 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010355555555555556, + "loss": 0.0992, + "step": 4071 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010344444444444446, + "loss": 0.1111, + "step": 4072 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010333333333333333, + "loss": 0.0992, + "step": 4073 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010322222222222223, + "loss": 0.1013, + "step": 4074 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010311111111111111, + "loss": 0.1262, + "step": 4075 + }, + { + "epoch": 79.01, + "learning_rate": 0.000103, + "loss": 0.1004, + "step": 4076 + }, + { + "epoch": 79.01, + "learning_rate": 0.0001028888888888889, + "loss": 0.0849, + "step": 4077 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010277777777777778, + "loss": 0.1294, + "step": 4078 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010266666666666668, + "loss": 0.1151, + "step": 4079 + }, + { + "epoch": 79.01, + "learning_rate": 0.00010255555555555555, + "loss": 0.0836, + "step": 4080 + }, + { + "epoch": 80.0, + "learning_rate": 0.00010244444444444445, + "loss": 0.1385, + "step": 4081 + }, + { + "epoch": 80.0, + "learning_rate": 0.00010233333333333333, + "loss": 0.1105, + "step": 4082 + }, + { + "epoch": 80.0, + "learning_rate": 0.00010222222222222223, + "loss": 0.0747, + "step": 4083 + }, + { + "epoch": 80.0, + "learning_rate": 0.00010211111111111111, + "loss": 0.1543, + "step": 4084 + }, + { + "epoch": 80.0, + "learning_rate": 0.000102, + "loss": 0.0847, + "step": 4085 + }, + { + "epoch": 80.0, + "learning_rate": 0.0001018888888888889, + "loss": 0.1013, + "step": 4086 + }, + { + "epoch": 80.0, + "learning_rate": 0.00010177777777777777, + "loss": 0.0753, + "step": 4087 + }, + { + "epoch": 80.0, + "learning_rate": 0.00010166666666666667, + "loss": 0.1206, + "step": 4088 + }, + { + "epoch": 80.0, + "learning_rate": 0.00010155555555555555, + "loss": 0.0912, + "step": 4089 + }, + { + "epoch": 80.0, + "learning_rate": 0.00010144444444444445, + "loss": 0.0654, + "step": 4090 + }, + { + "epoch": 80.0, + "learning_rate": 0.00010133333333333333, + "loss": 0.073, + "step": 4091 + }, + { + "epoch": 80.0, + "learning_rate": 0.00010122222222222223, + "loss": 0.1456, + "step": 4092 + }, + { + "epoch": 80.0, + "learning_rate": 0.00010111111111111112, + "loss": 0.0958, + "step": 4093 + }, + { + "epoch": 80.0, + "learning_rate": 0.000101, + "loss": 0.1044, + "step": 4094 + }, + { + "epoch": 80.0, + "learning_rate": 0.00010088888888888889, + "loss": 0.0841, + "step": 4095 + }, + { + "epoch": 80.0, + "learning_rate": 0.00010077777777777777, + "loss": 0.0985, + "step": 4096 + }, + { + "epoch": 80.0, + "learning_rate": 0.00010066666666666667, + "loss": 0.1039, + "step": 4097 + }, + { + "epoch": 80.0, + "learning_rate": 0.00010055555555555555, + "loss": 0.076, + "step": 4098 + }, + { + "epoch": 80.0, + "learning_rate": 0.00010044444444444445, + "loss": 0.1035, + "step": 4099 + }, + { + "epoch": 80.0, + "learning_rate": 0.00010033333333333334, + "loss": 0.0842, + "step": 4100 + }, + { + "epoch": 80.0, + "learning_rate": 0.00010022222222222222, + "loss": 0.0927, + "step": 4101 + }, + { + "epoch": 80.0, + "learning_rate": 0.0001001111111111111, + "loss": 0.1278, + "step": 4102 + }, + { + "epoch": 80.0, + "learning_rate": 0.0001, + "loss": 0.0972, + "step": 4103 + }, + { + "epoch": 80.0, + "learning_rate": 9.988888888888889e-05, + "loss": 0.1254, + "step": 4104 + }, + { + "epoch": 80.0, + "learning_rate": 9.977777777777779e-05, + "loss": 0.1117, + "step": 4105 + }, + { + "epoch": 80.01, + "learning_rate": 9.966666666666667e-05, + "loss": 0.0948, + "step": 4106 + }, + { + "epoch": 80.01, + "learning_rate": 9.955555555555556e-05, + "loss": 0.0939, + "step": 4107 + }, + { + "epoch": 80.01, + "learning_rate": 9.944444444444444e-05, + "loss": 0.0943, + "step": 4108 + }, + { + "epoch": 80.01, + "learning_rate": 9.933333333333333e-05, + "loss": 0.1106, + "step": 4109 + }, + { + "epoch": 80.01, + "learning_rate": 9.922222222222222e-05, + "loss": 0.1086, + "step": 4110 + }, + { + "epoch": 80.01, + "learning_rate": 9.911111111111111e-05, + "loss": 0.1097, + "step": 4111 + }, + { + "epoch": 80.01, + "learning_rate": 9.900000000000001e-05, + "loss": 0.1365, + "step": 4112 + }, + { + "epoch": 80.01, + "learning_rate": 9.888888888888889e-05, + "loss": 0.1086, + "step": 4113 + }, + { + "epoch": 80.01, + "learning_rate": 9.877777777777779e-05, + "loss": 0.1048, + "step": 4114 + }, + { + "epoch": 80.01, + "learning_rate": 9.866666666666668e-05, + "loss": 0.1111, + "step": 4115 + }, + { + "epoch": 80.01, + "learning_rate": 9.855555555555555e-05, + "loss": 0.0833, + "step": 4116 + }, + { + "epoch": 80.01, + "learning_rate": 9.844444444444444e-05, + "loss": 0.1224, + "step": 4117 + }, + { + "epoch": 80.01, + "learning_rate": 9.833333333333333e-05, + "loss": 0.0946, + "step": 4118 + }, + { + "epoch": 80.01, + "learning_rate": 9.822222222222223e-05, + "loss": 0.1163, + "step": 4119 + }, + { + "epoch": 80.01, + "learning_rate": 9.811111111111111e-05, + "loss": 0.0966, + "step": 4120 + }, + { + "epoch": 80.01, + "learning_rate": 9.800000000000001e-05, + "loss": 0.1385, + "step": 4121 + }, + { + "epoch": 80.01, + "learning_rate": 9.78888888888889e-05, + "loss": 0.1244, + "step": 4122 + }, + { + "epoch": 80.01, + "learning_rate": 9.777777777777778e-05, + "loss": 0.0871, + "step": 4123 + }, + { + "epoch": 80.01, + "learning_rate": 9.766666666666666e-05, + "loss": 0.0936, + "step": 4124 + }, + { + "epoch": 80.01, + "learning_rate": 9.755555555555556e-05, + "loss": 0.0883, + "step": 4125 + }, + { + "epoch": 80.01, + "learning_rate": 9.744444444444445e-05, + "loss": 0.1169, + "step": 4126 + }, + { + "epoch": 80.01, + "learning_rate": 9.733333333333333e-05, + "loss": 0.1075, + "step": 4127 + }, + { + "epoch": 80.01, + "learning_rate": 9.722222222222223e-05, + "loss": 0.1328, + "step": 4128 + }, + { + "epoch": 80.01, + "learning_rate": 9.711111111111111e-05, + "loss": 0.0897, + "step": 4129 + }, + { + "epoch": 80.01, + "learning_rate": 9.7e-05, + "loss": 0.0699, + "step": 4130 + }, + { + "epoch": 80.01, + "learning_rate": 9.688888888888888e-05, + "loss": 0.1133, + "step": 4131 + }, + { + "epoch": 81.0, + "learning_rate": 9.677777777777778e-05, + "loss": 0.1603, + "step": 4132 + }, + { + "epoch": 81.0, + "learning_rate": 9.666666666666667e-05, + "loss": 0.1004, + "step": 4133 + }, + { + "epoch": 81.0, + "learning_rate": 9.655555555555557e-05, + "loss": 0.0953, + "step": 4134 + }, + { + "epoch": 81.0, + "learning_rate": 9.644444444444445e-05, + "loss": 0.0873, + "step": 4135 + }, + { + "epoch": 81.0, + "learning_rate": 9.633333333333335e-05, + "loss": 0.1366, + "step": 4136 + }, + { + "epoch": 81.0, + "learning_rate": 9.622222222222222e-05, + "loss": 0.0976, + "step": 4137 + }, + { + "epoch": 81.0, + "learning_rate": 9.61111111111111e-05, + "loss": 0.1167, + "step": 4138 + }, + { + "epoch": 81.0, + "learning_rate": 9.6e-05, + "loss": 0.072, + "step": 4139 + }, + { + "epoch": 81.0, + "learning_rate": 9.588888888888889e-05, + "loss": 0.1071, + "step": 4140 + }, + { + "epoch": 81.0, + "learning_rate": 9.577777777777779e-05, + "loss": 0.1068, + "step": 4141 + }, + { + "epoch": 81.0, + "learning_rate": 9.566666666666667e-05, + "loss": 0.0898, + "step": 4142 + }, + { + "epoch": 81.0, + "learning_rate": 9.555555555555557e-05, + "loss": 0.1036, + "step": 4143 + }, + { + "epoch": 81.0, + "learning_rate": 9.544444444444444e-05, + "loss": 0.1048, + "step": 4144 + }, + { + "epoch": 81.0, + "learning_rate": 9.533333333333334e-05, + "loss": 0.1225, + "step": 4145 + }, + { + "epoch": 81.0, + "learning_rate": 9.522222222222222e-05, + "loss": 0.1181, + "step": 4146 + }, + { + "epoch": 81.0, + "learning_rate": 9.511111111111111e-05, + "loss": 0.0613, + "step": 4147 + }, + { + "epoch": 81.0, + "learning_rate": 9.5e-05, + "loss": 0.1025, + "step": 4148 + }, + { + "epoch": 81.0, + "learning_rate": 9.488888888888889e-05, + "loss": 0.1179, + "step": 4149 + }, + { + "epoch": 81.0, + "learning_rate": 9.477777777777779e-05, + "loss": 0.1061, + "step": 4150 + }, + { + "epoch": 81.0, + "learning_rate": 9.466666666666666e-05, + "loss": 0.1031, + "step": 4151 + }, + { + "epoch": 81.0, + "learning_rate": 9.455555555555556e-05, + "loss": 0.0838, + "step": 4152 + }, + { + "epoch": 81.0, + "learning_rate": 9.444444444444444e-05, + "loss": 0.098, + "step": 4153 + }, + { + "epoch": 81.0, + "learning_rate": 9.433333333333334e-05, + "loss": 0.116, + "step": 4154 + }, + { + "epoch": 81.0, + "learning_rate": 9.422222222222223e-05, + "loss": 0.1207, + "step": 4155 + }, + { + "epoch": 81.0, + "learning_rate": 9.411111111111112e-05, + "loss": 0.0995, + "step": 4156 + }, + { + "epoch": 81.01, + "learning_rate": 9.400000000000001e-05, + "loss": 0.0963, + "step": 4157 + }, + { + "epoch": 81.01, + "learning_rate": 9.388888888888888e-05, + "loss": 0.1091, + "step": 4158 + }, + { + "epoch": 81.01, + "learning_rate": 9.377777777777778e-05, + "loss": 0.1639, + "step": 4159 + }, + { + "epoch": 81.01, + "learning_rate": 9.366666666666666e-05, + "loss": 0.1519, + "step": 4160 + }, + { + "epoch": 81.01, + "learning_rate": 9.355555555555556e-05, + "loss": 0.087, + "step": 4161 + }, + { + "epoch": 81.01, + "learning_rate": 9.344444444444444e-05, + "loss": 0.0831, + "step": 4162 + }, + { + "epoch": 81.01, + "learning_rate": 9.333333333333334e-05, + "loss": 0.1417, + "step": 4163 + }, + { + "epoch": 81.01, + "learning_rate": 9.322222222222223e-05, + "loss": 0.0747, + "step": 4164 + }, + { + "epoch": 81.01, + "learning_rate": 9.311111111111111e-05, + "loss": 0.0774, + "step": 4165 + }, + { + "epoch": 81.01, + "learning_rate": 9.3e-05, + "loss": 0.0858, + "step": 4166 + }, + { + "epoch": 81.01, + "learning_rate": 9.288888888888888e-05, + "loss": 0.0979, + "step": 4167 + }, + { + "epoch": 81.01, + "learning_rate": 9.277777777777778e-05, + "loss": 0.1072, + "step": 4168 + }, + { + "epoch": 81.01, + "learning_rate": 9.266666666666666e-05, + "loss": 0.0908, + "step": 4169 + }, + { + "epoch": 81.01, + "learning_rate": 9.255555555555556e-05, + "loss": 0.1357, + "step": 4170 + }, + { + "epoch": 81.01, + "learning_rate": 9.244444444444445e-05, + "loss": 0.1066, + "step": 4171 + }, + { + "epoch": 81.01, + "learning_rate": 9.233333333333333e-05, + "loss": 0.1369, + "step": 4172 + }, + { + "epoch": 81.01, + "learning_rate": 9.222222222222222e-05, + "loss": 0.0948, + "step": 4173 + }, + { + "epoch": 81.01, + "learning_rate": 9.211111111111112e-05, + "loss": 0.0942, + "step": 4174 + }, + { + "epoch": 81.01, + "learning_rate": 9.2e-05, + "loss": 0.0961, + "step": 4175 + }, + { + "epoch": 81.01, + "learning_rate": 9.18888888888889e-05, + "loss": 0.1145, + "step": 4176 + }, + { + "epoch": 81.01, + "learning_rate": 9.177777777777778e-05, + "loss": 0.1185, + "step": 4177 + }, + { + "epoch": 81.01, + "learning_rate": 9.166666666666667e-05, + "loss": 0.0831, + "step": 4178 + }, + { + "epoch": 81.01, + "learning_rate": 9.155555555555555e-05, + "loss": 0.079, + "step": 4179 + }, + { + "epoch": 81.01, + "learning_rate": 9.144444444444444e-05, + "loss": 0.1136, + "step": 4180 + }, + { + "epoch": 81.01, + "learning_rate": 9.133333333333334e-05, + "loss": 0.083, + "step": 4181 + }, + { + "epoch": 81.01, + "learning_rate": 9.122222222222222e-05, + "loss": 0.0822, + "step": 4182 + }, + { + "epoch": 82.0, + "learning_rate": 9.111111111111112e-05, + "loss": 0.1095, + "step": 4183 + }, + { + "epoch": 82.0, + "learning_rate": 9.1e-05, + "loss": 0.113, + "step": 4184 + }, + { + "epoch": 82.0, + "learning_rate": 9.08888888888889e-05, + "loss": 0.09, + "step": 4185 + }, + { + "epoch": 82.0, + "learning_rate": 9.077777777777777e-05, + "loss": 0.1172, + "step": 4186 + }, + { + "epoch": 82.0, + "learning_rate": 9.066666666666667e-05, + "loss": 0.0896, + "step": 4187 + }, + { + "epoch": 82.0, + "learning_rate": 9.055555555555556e-05, + "loss": 0.0754, + "step": 4188 + }, + { + "epoch": 82.0, + "learning_rate": 9.044444444444444e-05, + "loss": 0.0917, + "step": 4189 + }, + { + "epoch": 82.0, + "learning_rate": 9.033333333333334e-05, + "loss": 0.1143, + "step": 4190 + }, + { + "epoch": 82.0, + "learning_rate": 9.022222222222222e-05, + "loss": 0.0858, + "step": 4191 + }, + { + "epoch": 82.0, + "learning_rate": 9.011111111111112e-05, + "loss": 0.0963, + "step": 4192 + }, + { + "epoch": 82.0, + "learning_rate": 8.999999999999999e-05, + "loss": 0.0974, + "step": 4193 + }, + { + "epoch": 82.0, + "learning_rate": 8.988888888888889e-05, + "loss": 0.0705, + "step": 4194 + }, + { + "epoch": 82.0, + "learning_rate": 8.977777777777778e-05, + "loss": 0.1217, + "step": 4195 + }, + { + "epoch": 82.0, + "learning_rate": 8.966666666666667e-05, + "loss": 0.1218, + "step": 4196 + }, + { + "epoch": 82.0, + "learning_rate": 8.955555555555556e-05, + "loss": 0.1354, + "step": 4197 + }, + { + "epoch": 82.0, + "learning_rate": 8.944444444444444e-05, + "loss": 0.1043, + "step": 4198 + }, + { + "epoch": 82.0, + "learning_rate": 8.933333333333334e-05, + "loss": 0.0821, + "step": 4199 + }, + { + "epoch": 82.0, + "learning_rate": 8.922222222222221e-05, + "loss": 0.1076, + "step": 4200 + }, + { + "epoch": 82.0, + "learning_rate": 8.911111111111111e-05, + "loss": 0.127, + "step": 4201 + }, + { + "epoch": 82.0, + "learning_rate": 8.9e-05, + "loss": 0.0831, + "step": 4202 + }, + { + "epoch": 82.0, + "learning_rate": 8.888888888888889e-05, + "loss": 0.0803, + "step": 4203 + }, + { + "epoch": 82.0, + "learning_rate": 8.877777777777778e-05, + "loss": 0.0911, + "step": 4204 + }, + { + "epoch": 82.0, + "learning_rate": 8.866666666666668e-05, + "loss": 0.0801, + "step": 4205 + }, + { + "epoch": 82.0, + "learning_rate": 8.855555555555556e-05, + "loss": 0.1217, + "step": 4206 + }, + { + "epoch": 82.0, + "learning_rate": 8.844444444444446e-05, + "loss": 0.0784, + "step": 4207 + }, + { + "epoch": 82.01, + "learning_rate": 8.833333333333333e-05, + "loss": 0.1106, + "step": 4208 + }, + { + "epoch": 82.01, + "learning_rate": 8.822222222222221e-05, + "loss": 0.1072, + "step": 4209 + }, + { + "epoch": 82.01, + "learning_rate": 8.811111111111111e-05, + "loss": 0.0919, + "step": 4210 + }, + { + "epoch": 82.01, + "learning_rate": 8.8e-05, + "loss": 0.0748, + "step": 4211 + }, + { + "epoch": 82.01, + "learning_rate": 8.78888888888889e-05, + "loss": 0.1159, + "step": 4212 + }, + { + "epoch": 82.01, + "learning_rate": 8.777777777777778e-05, + "loss": 0.1204, + "step": 4213 + }, + { + "epoch": 82.01, + "learning_rate": 8.766666666666668e-05, + "loss": 0.0776, + "step": 4214 + }, + { + "epoch": 82.01, + "learning_rate": 8.755555555555555e-05, + "loss": 0.1255, + "step": 4215 + }, + { + "epoch": 82.01, + "learning_rate": 8.744444444444445e-05, + "loss": 0.1039, + "step": 4216 + }, + { + "epoch": 82.01, + "learning_rate": 8.733333333333333e-05, + "loss": 0.0821, + "step": 4217 + }, + { + "epoch": 82.01, + "learning_rate": 8.722222222222223e-05, + "loss": 0.1082, + "step": 4218 + }, + { + "epoch": 82.01, + "learning_rate": 8.711111111111112e-05, + "loss": 0.1147, + "step": 4219 + }, + { + "epoch": 82.01, + "learning_rate": 8.7e-05, + "loss": 0.1155, + "step": 4220 + }, + { + "epoch": 82.01, + "learning_rate": 8.68888888888889e-05, + "loss": 0.0887, + "step": 4221 + }, + { + "epoch": 82.01, + "learning_rate": 8.677777777777777e-05, + "loss": 0.0618, + "step": 4222 + }, + { + "epoch": 82.01, + "learning_rate": 8.666666666666667e-05, + "loss": 0.1245, + "step": 4223 + }, + { + "epoch": 82.01, + "learning_rate": 8.655555555555555e-05, + "loss": 0.0707, + "step": 4224 + }, + { + "epoch": 82.01, + "learning_rate": 8.644444444444445e-05, + "loss": 0.1121, + "step": 4225 + }, + { + "epoch": 82.01, + "learning_rate": 8.633333333333334e-05, + "loss": 0.1233, + "step": 4226 + }, + { + "epoch": 82.01, + "learning_rate": 8.622222222222223e-05, + "loss": 0.1164, + "step": 4227 + }, + { + "epoch": 82.01, + "learning_rate": 8.611111111111112e-05, + "loss": 0.1051, + "step": 4228 + }, + { + "epoch": 82.01, + "learning_rate": 8.599999999999999e-05, + "loss": 0.129, + "step": 4229 + }, + { + "epoch": 82.01, + "learning_rate": 8.588888888888889e-05, + "loss": 0.1339, + "step": 4230 + }, + { + "epoch": 82.01, + "learning_rate": 8.577777777777777e-05, + "loss": 0.1046, + "step": 4231 + }, + { + "epoch": 82.01, + "learning_rate": 8.566666666666667e-05, + "loss": 0.0687, + "step": 4232 + }, + { + "epoch": 82.01, + "learning_rate": 8.555555555555556e-05, + "loss": 0.136, + "step": 4233 + }, + { + "epoch": 83.0, + "learning_rate": 8.544444444444445e-05, + "loss": 0.1398, + "step": 4234 + }, + { + "epoch": 83.0, + "learning_rate": 8.533333333333334e-05, + "loss": 0.0684, + "step": 4235 + }, + { + "epoch": 83.0, + "learning_rate": 8.522222222222222e-05, + "loss": 0.0612, + "step": 4236 + }, + { + "epoch": 83.0, + "learning_rate": 8.511111111111111e-05, + "loss": 0.1086, + "step": 4237 + }, + { + "epoch": 83.0, + "learning_rate": 8.5e-05, + "loss": 0.0893, + "step": 4238 + }, + { + "epoch": 83.0, + "learning_rate": 8.488888888888889e-05, + "loss": 0.0866, + "step": 4239 + }, + { + "epoch": 83.0, + "learning_rate": 8.477777777777778e-05, + "loss": 0.0779, + "step": 4240 + }, + { + "epoch": 83.0, + "learning_rate": 8.466666666666667e-05, + "loss": 0.0671, + "step": 4241 + }, + { + "epoch": 83.0, + "learning_rate": 8.455555555555556e-05, + "loss": 0.0851, + "step": 4242 + }, + { + "epoch": 83.0, + "learning_rate": 8.444444444444444e-05, + "loss": 0.0843, + "step": 4243 + }, + { + "epoch": 83.0, + "learning_rate": 8.433333333333333e-05, + "loss": 0.0821, + "step": 4244 + }, + { + "epoch": 83.0, + "learning_rate": 8.422222222222223e-05, + "loss": 0.1034, + "step": 4245 + }, + { + "epoch": 83.0, + "learning_rate": 8.411111111111111e-05, + "loss": 0.1102, + "step": 4246 + }, + { + "epoch": 83.0, + "learning_rate": 8.400000000000001e-05, + "loss": 0.0745, + "step": 4247 + }, + { + "epoch": 83.0, + "learning_rate": 8.38888888888889e-05, + "loss": 0.0997, + "step": 4248 + }, + { + "epoch": 83.0, + "learning_rate": 8.377777777777779e-05, + "loss": 0.1165, + "step": 4249 + }, + { + "epoch": 83.0, + "learning_rate": 8.366666666666666e-05, + "loss": 0.0988, + "step": 4250 + }, + { + "epoch": 83.0, + "learning_rate": 8.355555555555555e-05, + "loss": 0.1135, + "step": 4251 + }, + { + "epoch": 83.0, + "learning_rate": 8.344444444444445e-05, + "loss": 0.0703, + "step": 4252 + }, + { + "epoch": 83.0, + "learning_rate": 8.333333333333333e-05, + "loss": 0.1268, + "step": 4253 + }, + { + "epoch": 83.0, + "learning_rate": 8.322222222222223e-05, + "loss": 0.0817, + "step": 4254 + }, + { + "epoch": 83.0, + "learning_rate": 8.311111111111111e-05, + "loss": 0.0842, + "step": 4255 + }, + { + "epoch": 83.0, + "learning_rate": 8.300000000000001e-05, + "loss": 0.132, + "step": 4256 + }, + { + "epoch": 83.0, + "learning_rate": 8.288888888888888e-05, + "loss": 0.1092, + "step": 4257 + }, + { + "epoch": 83.0, + "learning_rate": 8.277777777777778e-05, + "loss": 0.0918, + "step": 4258 + }, + { + "epoch": 83.01, + "learning_rate": 8.266666666666667e-05, + "loss": 0.102, + "step": 4259 + }, + { + "epoch": 83.01, + "learning_rate": 8.255555555555555e-05, + "loss": 0.0969, + "step": 4260 + }, + { + "epoch": 83.01, + "learning_rate": 8.244444444444445e-05, + "loss": 0.1021, + "step": 4261 + }, + { + "epoch": 83.01, + "learning_rate": 8.233333333333333e-05, + "loss": 0.119, + "step": 4262 + }, + { + "epoch": 83.01, + "learning_rate": 8.222222222222223e-05, + "loss": 0.0878, + "step": 4263 + }, + { + "epoch": 83.01, + "learning_rate": 8.21111111111111e-05, + "loss": 0.0738, + "step": 4264 + }, + { + "epoch": 83.01, + "learning_rate": 8.2e-05, + "loss": 0.082, + "step": 4265 + }, + { + "epoch": 83.01, + "learning_rate": 8.188888888888889e-05, + "loss": 0.1218, + "step": 4266 + }, + { + "epoch": 83.01, + "learning_rate": 8.177777777777778e-05, + "loss": 0.0968, + "step": 4267 + }, + { + "epoch": 83.01, + "learning_rate": 8.166666666666667e-05, + "loss": 0.0884, + "step": 4268 + }, + { + "epoch": 83.01, + "learning_rate": 8.155555555555557e-05, + "loss": 0.0708, + "step": 4269 + }, + { + "epoch": 83.01, + "learning_rate": 8.144444444444445e-05, + "loss": 0.088, + "step": 4270 + }, + { + "epoch": 83.01, + "learning_rate": 8.133333333333332e-05, + "loss": 0.0723, + "step": 4271 + }, + { + "epoch": 83.01, + "learning_rate": 8.122222222222222e-05, + "loss": 0.1018, + "step": 4272 + }, + { + "epoch": 83.01, + "learning_rate": 8.11111111111111e-05, + "loss": 0.1065, + "step": 4273 + }, + { + "epoch": 83.01, + "learning_rate": 8.1e-05, + "loss": 0.1047, + "step": 4274 + }, + { + "epoch": 83.01, + "learning_rate": 8.088888888888889e-05, + "loss": 0.0878, + "step": 4275 + }, + { + "epoch": 83.01, + "learning_rate": 8.077777777777779e-05, + "loss": 0.1077, + "step": 4276 + }, + { + "epoch": 83.01, + "learning_rate": 8.066666666666667e-05, + "loss": 0.1382, + "step": 4277 + }, + { + "epoch": 83.01, + "learning_rate": 8.055555555555556e-05, + "loss": 0.0927, + "step": 4278 + }, + { + "epoch": 83.01, + "learning_rate": 8.044444444444444e-05, + "loss": 0.1155, + "step": 4279 + }, + { + "epoch": 83.01, + "learning_rate": 8.033333333333334e-05, + "loss": 0.1056, + "step": 4280 + }, + { + "epoch": 83.01, + "learning_rate": 8.022222222222222e-05, + "loss": 0.145, + "step": 4281 + }, + { + "epoch": 83.01, + "learning_rate": 8.011111111111111e-05, + "loss": 0.0859, + "step": 4282 + }, + { + "epoch": 83.01, + "learning_rate": 8e-05, + "loss": 0.0711, + "step": 4283 + }, + { + "epoch": 83.01, + "learning_rate": 7.988888888888889e-05, + "loss": 0.1186, + "step": 4284 + }, + { + "epoch": 84.0, + "learning_rate": 7.977777777777778e-05, + "loss": 0.1691, + "step": 4285 + }, + { + "epoch": 84.0, + "learning_rate": 7.966666666666666e-05, + "loss": 0.0731, + "step": 4286 + }, + { + "epoch": 84.0, + "learning_rate": 7.955555555555556e-05, + "loss": 0.0852, + "step": 4287 + }, + { + "epoch": 84.0, + "learning_rate": 7.944444444444444e-05, + "loss": 0.0948, + "step": 4288 + }, + { + "epoch": 84.0, + "learning_rate": 7.933333333333334e-05, + "loss": 0.0901, + "step": 4289 + }, + { + "epoch": 84.0, + "learning_rate": 7.922222222222223e-05, + "loss": 0.0822, + "step": 4290 + }, + { + "epoch": 84.0, + "learning_rate": 7.911111111111111e-05, + "loss": 0.0844, + "step": 4291 + }, + { + "epoch": 84.0, + "learning_rate": 7.9e-05, + "loss": 0.0963, + "step": 4292 + }, + { + "epoch": 84.0, + "learning_rate": 7.888888888888888e-05, + "loss": 0.068, + "step": 4293 + }, + { + "epoch": 84.0, + "learning_rate": 7.877777777777778e-05, + "loss": 0.0996, + "step": 4294 + }, + { + "epoch": 84.0, + "learning_rate": 7.866666666666666e-05, + "loss": 0.1151, + "step": 4295 + }, + { + "epoch": 84.0, + "learning_rate": 7.855555555555556e-05, + "loss": 0.1002, + "step": 4296 + }, + { + "epoch": 84.0, + "learning_rate": 7.844444444444445e-05, + "loss": 0.1024, + "step": 4297 + }, + { + "epoch": 84.0, + "learning_rate": 7.833333333333334e-05, + "loss": 0.0963, + "step": 4298 + }, + { + "epoch": 84.0, + "learning_rate": 7.822222222222222e-05, + "loss": 0.1125, + "step": 4299 + }, + { + "epoch": 84.0, + "learning_rate": 7.811111111111111e-05, + "loss": 0.0818, + "step": 4300 + }, + { + "epoch": 84.0, + "learning_rate": 7.8e-05, + "loss": 0.1036, + "step": 4301 + }, + { + "epoch": 84.0, + "learning_rate": 7.788888888888888e-05, + "loss": 0.1079, + "step": 4302 + }, + { + "epoch": 84.0, + "learning_rate": 7.777777777777778e-05, + "loss": 0.1045, + "step": 4303 + }, + { + "epoch": 84.0, + "learning_rate": 7.766666666666667e-05, + "loss": 0.1253, + "step": 4304 + }, + { + "epoch": 84.0, + "learning_rate": 7.755555555555556e-05, + "loss": 0.0978, + "step": 4305 + }, + { + "epoch": 84.0, + "learning_rate": 7.744444444444445e-05, + "loss": 0.0756, + "step": 4306 + }, + { + "epoch": 84.0, + "learning_rate": 7.733333333333333e-05, + "loss": 0.103, + "step": 4307 + }, + { + "epoch": 84.0, + "learning_rate": 7.722222222222222e-05, + "loss": 0.1236, + "step": 4308 + }, + { + "epoch": 84.0, + "learning_rate": 7.711111111111112e-05, + "loss": 0.0936, + "step": 4309 + }, + { + "epoch": 84.01, + "learning_rate": 7.7e-05, + "loss": 0.0941, + "step": 4310 + }, + { + "epoch": 84.01, + "learning_rate": 7.68888888888889e-05, + "loss": 0.1367, + "step": 4311 + }, + { + "epoch": 84.01, + "learning_rate": 7.677777777777778e-05, + "loss": 0.121, + "step": 4312 + }, + { + "epoch": 84.01, + "learning_rate": 7.666666666666667e-05, + "loss": 0.0937, + "step": 4313 + }, + { + "epoch": 84.01, + "learning_rate": 7.655555555555555e-05, + "loss": 0.0898, + "step": 4314 + }, + { + "epoch": 84.01, + "learning_rate": 7.644444444444444e-05, + "loss": 0.0627, + "step": 4315 + }, + { + "epoch": 84.01, + "learning_rate": 7.633333333333334e-05, + "loss": 0.0702, + "step": 4316 + }, + { + "epoch": 84.01, + "learning_rate": 7.622222222222222e-05, + "loss": 0.0977, + "step": 4317 + }, + { + "epoch": 84.01, + "learning_rate": 7.611111111111112e-05, + "loss": 0.0901, + "step": 4318 + }, + { + "epoch": 84.01, + "learning_rate": 7.6e-05, + "loss": 0.1216, + "step": 4319 + }, + { + "epoch": 84.01, + "learning_rate": 7.58888888888889e-05, + "loss": 0.0882, + "step": 4320 + }, + { + "epoch": 84.01, + "learning_rate": 7.577777777777777e-05, + "loss": 0.1222, + "step": 4321 + }, + { + "epoch": 84.01, + "learning_rate": 7.566666666666666e-05, + "loss": 0.1085, + "step": 4322 + }, + { + "epoch": 84.01, + "learning_rate": 7.555555555555556e-05, + "loss": 0.1053, + "step": 4323 + }, + { + "epoch": 84.01, + "learning_rate": 7.544444444444444e-05, + "loss": 0.0867, + "step": 4324 + }, + { + "epoch": 84.01, + "learning_rate": 7.533333333333334e-05, + "loss": 0.0975, + "step": 4325 + }, + { + "epoch": 84.01, + "learning_rate": 7.522222222222222e-05, + "loss": 0.1242, + "step": 4326 + }, + { + "epoch": 84.01, + "learning_rate": 7.511111111111112e-05, + "loss": 0.1237, + "step": 4327 + }, + { + "epoch": 84.01, + "learning_rate": 7.5e-05, + "loss": 0.1198, + "step": 4328 + }, + { + "epoch": 84.01, + "learning_rate": 7.488888888888889e-05, + "loss": 0.1401, + "step": 4329 + }, + { + "epoch": 84.01, + "learning_rate": 7.477777777777778e-05, + "loss": 0.1236, + "step": 4330 + }, + { + "epoch": 84.01, + "learning_rate": 7.466666666666667e-05, + "loss": 0.1114, + "step": 4331 + }, + { + "epoch": 84.01, + "learning_rate": 7.455555555555556e-05, + "loss": 0.1344, + "step": 4332 + }, + { + "epoch": 84.01, + "learning_rate": 7.444444444444444e-05, + "loss": 0.0721, + "step": 4333 + }, + { + "epoch": 84.01, + "learning_rate": 7.433333333333334e-05, + "loss": 0.0995, + "step": 4334 + }, + { + "epoch": 84.01, + "learning_rate": 7.422222222222221e-05, + "loss": 0.0942, + "step": 4335 + }, + { + "epoch": 85.0, + "learning_rate": 7.411111111111111e-05, + "loss": 0.1209, + "step": 4336 + }, + { + "epoch": 85.0, + "learning_rate": 7.4e-05, + "loss": 0.1076, + "step": 4337 + }, + { + "epoch": 85.0, + "learning_rate": 7.38888888888889e-05, + "loss": 0.0773, + "step": 4338 + }, + { + "epoch": 85.0, + "learning_rate": 7.377777777777778e-05, + "loss": 0.0879, + "step": 4339 + }, + { + "epoch": 85.0, + "learning_rate": 7.366666666666668e-05, + "loss": 0.0763, + "step": 4340 + }, + { + "epoch": 85.0, + "learning_rate": 7.355555555555556e-05, + "loss": 0.1123, + "step": 4341 + }, + { + "epoch": 85.0, + "learning_rate": 7.344444444444445e-05, + "loss": 0.1161, + "step": 4342 + }, + { + "epoch": 85.0, + "learning_rate": 7.333333333333333e-05, + "loss": 0.087, + "step": 4343 + }, + { + "epoch": 85.0, + "learning_rate": 7.322222222222222e-05, + "loss": 0.1051, + "step": 4344 + }, + { + "epoch": 85.0, + "learning_rate": 7.311111111111111e-05, + "loss": 0.1018, + "step": 4345 + }, + { + "epoch": 85.0, + "learning_rate": 7.3e-05, + "loss": 0.0629, + "step": 4346 + }, + { + "epoch": 85.0, + "learning_rate": 7.28888888888889e-05, + "loss": 0.1017, + "step": 4347 + }, + { + "epoch": 85.0, + "learning_rate": 7.277777777777778e-05, + "loss": 0.1278, + "step": 4348 + }, + { + "epoch": 85.0, + "learning_rate": 7.266666666666667e-05, + "loss": 0.074, + "step": 4349 + }, + { + "epoch": 85.0, + "learning_rate": 7.255555555555555e-05, + "loss": 0.1007, + "step": 4350 + }, + { + "epoch": 85.0, + "learning_rate": 7.244444444444445e-05, + "loss": 0.0687, + "step": 4351 + }, + { + "epoch": 85.0, + "learning_rate": 7.233333333333333e-05, + "loss": 0.1077, + "step": 4352 + }, + { + "epoch": 85.0, + "learning_rate": 7.222222222222222e-05, + "loss": 0.103, + "step": 4353 + }, + { + "epoch": 85.0, + "learning_rate": 7.211111111111112e-05, + "loss": 0.0901, + "step": 4354 + }, + { + "epoch": 85.0, + "learning_rate": 7.2e-05, + "loss": 0.1283, + "step": 4355 + }, + { + "epoch": 85.0, + "learning_rate": 7.188888888888889e-05, + "loss": 0.0974, + "step": 4356 + }, + { + "epoch": 85.0, + "learning_rate": 7.177777777777777e-05, + "loss": 0.0854, + "step": 4357 + }, + { + "epoch": 85.0, + "learning_rate": 7.166666666666667e-05, + "loss": 0.1336, + "step": 4358 + }, + { + "epoch": 85.0, + "learning_rate": 7.155555555555555e-05, + "loss": 0.1069, + "step": 4359 + }, + { + "epoch": 85.0, + "learning_rate": 7.144444444444445e-05, + "loss": 0.1107, + "step": 4360 + }, + { + "epoch": 85.01, + "learning_rate": 7.133333333333334e-05, + "loss": 0.0549, + "step": 4361 + }, + { + "epoch": 85.01, + "learning_rate": 7.122222222222224e-05, + "loss": 0.0701, + "step": 4362 + }, + { + "epoch": 85.01, + "learning_rate": 7.11111111111111e-05, + "loss": 0.0909, + "step": 4363 + }, + { + "epoch": 85.01, + "learning_rate": 7.099999999999999e-05, + "loss": 0.0711, + "step": 4364 + }, + { + "epoch": 85.01, + "learning_rate": 7.088888888888889e-05, + "loss": 0.0843, + "step": 4365 + }, + { + "epoch": 85.01, + "learning_rate": 7.077777777777777e-05, + "loss": 0.095, + "step": 4366 + }, + { + "epoch": 85.01, + "learning_rate": 7.066666666666667e-05, + "loss": 0.0912, + "step": 4367 + }, + { + "epoch": 85.01, + "learning_rate": 7.055555555555556e-05, + "loss": 0.0716, + "step": 4368 + }, + { + "epoch": 85.01, + "learning_rate": 7.044444444444446e-05, + "loss": 0.0964, + "step": 4369 + }, + { + "epoch": 85.01, + "learning_rate": 7.033333333333333e-05, + "loss": 0.0813, + "step": 4370 + }, + { + "epoch": 85.01, + "learning_rate": 7.022222222222222e-05, + "loss": 0.0928, + "step": 4371 + }, + { + "epoch": 85.01, + "learning_rate": 7.011111111111111e-05, + "loss": 0.0779, + "step": 4372 + }, + { + "epoch": 85.01, + "learning_rate": 7.000000000000001e-05, + "loss": 0.0686, + "step": 4373 + }, + { + "epoch": 85.01, + "learning_rate": 6.988888888888889e-05, + "loss": 0.1097, + "step": 4374 + }, + { + "epoch": 85.01, + "learning_rate": 6.977777777777778e-05, + "loss": 0.1063, + "step": 4375 + }, + { + "epoch": 85.01, + "learning_rate": 6.966666666666668e-05, + "loss": 0.0951, + "step": 4376 + }, + { + "epoch": 85.01, + "learning_rate": 6.955555555555555e-05, + "loss": 0.0884, + "step": 4377 + }, + { + "epoch": 85.01, + "learning_rate": 6.944444444444444e-05, + "loss": 0.0768, + "step": 4378 + }, + { + "epoch": 85.01, + "learning_rate": 6.933333333333333e-05, + "loss": 0.0708, + "step": 4379 + }, + { + "epoch": 85.01, + "learning_rate": 6.922222222222223e-05, + "loss": 0.0838, + "step": 4380 + }, + { + "epoch": 85.01, + "learning_rate": 6.911111111111111e-05, + "loss": 0.0745, + "step": 4381 + }, + { + "epoch": 85.01, + "learning_rate": 6.900000000000001e-05, + "loss": 0.0851, + "step": 4382 + }, + { + "epoch": 85.01, + "learning_rate": 6.88888888888889e-05, + "loss": 0.0894, + "step": 4383 + }, + { + "epoch": 85.01, + "learning_rate": 6.877777777777777e-05, + "loss": 0.0798, + "step": 4384 + }, + { + "epoch": 85.01, + "learning_rate": 6.866666666666666e-05, + "loss": 0.0748, + "step": 4385 + }, + { + "epoch": 85.01, + "learning_rate": 6.855555555555555e-05, + "loss": 0.0872, + "step": 4386 + }, + { + "epoch": 86.0, + "learning_rate": 6.844444444444445e-05, + "loss": 0.0725, + "step": 4387 + }, + { + "epoch": 86.0, + "learning_rate": 6.833333333333333e-05, + "loss": 0.0664, + "step": 4388 + }, + { + "epoch": 86.0, + "learning_rate": 6.822222222222223e-05, + "loss": 0.0843, + "step": 4389 + }, + { + "epoch": 86.0, + "learning_rate": 6.811111111111111e-05, + "loss": 0.1098, + "step": 4390 + }, + { + "epoch": 86.0, + "learning_rate": 6.800000000000001e-05, + "loss": 0.0726, + "step": 4391 + }, + { + "epoch": 86.0, + "learning_rate": 6.788888888888888e-05, + "loss": 0.1124, + "step": 4392 + }, + { + "epoch": 86.0, + "learning_rate": 6.777777777777778e-05, + "loss": 0.0783, + "step": 4393 + }, + { + "epoch": 86.0, + "learning_rate": 6.766666666666667e-05, + "loss": 0.1179, + "step": 4394 + }, + { + "epoch": 86.0, + "learning_rate": 6.755555555555555e-05, + "loss": 0.0854, + "step": 4395 + }, + { + "epoch": 86.0, + "learning_rate": 6.744444444444445e-05, + "loss": 0.1243, + "step": 4396 + }, + { + "epoch": 86.0, + "learning_rate": 6.733333333333333e-05, + "loss": 0.0952, + "step": 4397 + }, + { + "epoch": 86.0, + "learning_rate": 6.722222222222223e-05, + "loss": 0.067, + "step": 4398 + }, + { + "epoch": 86.0, + "learning_rate": 6.71111111111111e-05, + "loss": 0.0732, + "step": 4399 + }, + { + "epoch": 86.0, + "learning_rate": 6.7e-05, + "loss": 0.127, + "step": 4400 + }, + { + "epoch": 86.0, + "learning_rate": 6.688888888888889e-05, + "loss": 0.1036, + "step": 4401 + }, + { + "epoch": 86.0, + "learning_rate": 6.677777777777779e-05, + "loss": 0.082, + "step": 4402 + }, + { + "epoch": 86.0, + "learning_rate": 6.666666666666667e-05, + "loss": 0.0916, + "step": 4403 + }, + { + "epoch": 86.0, + "learning_rate": 6.655555555555557e-05, + "loss": 0.0574, + "step": 4404 + }, + { + "epoch": 86.0, + "learning_rate": 6.644444444444445e-05, + "loss": 0.1069, + "step": 4405 + }, + { + "epoch": 86.0, + "learning_rate": 6.633333333333332e-05, + "loss": 0.1158, + "step": 4406 + }, + { + "epoch": 86.0, + "learning_rate": 6.622222222222222e-05, + "loss": 0.1104, + "step": 4407 + }, + { + "epoch": 86.0, + "learning_rate": 6.611111111111111e-05, + "loss": 0.1208, + "step": 4408 + }, + { + "epoch": 86.0, + "learning_rate": 6.6e-05, + "loss": 0.0883, + "step": 4409 + }, + { + "epoch": 86.0, + "learning_rate": 6.588888888888889e-05, + "loss": 0.0923, + "step": 4410 + }, + { + "epoch": 86.0, + "learning_rate": 6.577777777777779e-05, + "loss": 0.0911, + "step": 4411 + }, + { + "epoch": 86.01, + "learning_rate": 6.566666666666667e-05, + "loss": 0.1327, + "step": 4412 + }, + { + "epoch": 86.01, + "learning_rate": 6.555555555555556e-05, + "loss": 0.0766, + "step": 4413 + }, + { + "epoch": 86.01, + "learning_rate": 6.544444444444444e-05, + "loss": 0.1051, + "step": 4414 + }, + { + "epoch": 86.01, + "learning_rate": 6.533333333333333e-05, + "loss": 0.0749, + "step": 4415 + }, + { + "epoch": 86.01, + "learning_rate": 6.522222222222222e-05, + "loss": 0.065, + "step": 4416 + }, + { + "epoch": 86.01, + "learning_rate": 6.511111111111111e-05, + "loss": 0.0807, + "step": 4417 + }, + { + "epoch": 86.01, + "learning_rate": 6.500000000000001e-05, + "loss": 0.0901, + "step": 4418 + }, + { + "epoch": 86.01, + "learning_rate": 6.488888888888889e-05, + "loss": 0.0894, + "step": 4419 + }, + { + "epoch": 86.01, + "learning_rate": 6.477777777777778e-05, + "loss": 0.0922, + "step": 4420 + }, + { + "epoch": 86.01, + "learning_rate": 6.466666666666666e-05, + "loss": 0.0723, + "step": 4421 + }, + { + "epoch": 86.01, + "learning_rate": 6.455555555555556e-05, + "loss": 0.111, + "step": 4422 + }, + { + "epoch": 86.01, + "learning_rate": 6.444444444444444e-05, + "loss": 0.0924, + "step": 4423 + }, + { + "epoch": 86.01, + "learning_rate": 6.433333333333334e-05, + "loss": 0.1023, + "step": 4424 + }, + { + "epoch": 86.01, + "learning_rate": 6.422222222222223e-05, + "loss": 0.0922, + "step": 4425 + }, + { + "epoch": 86.01, + "learning_rate": 6.411111111111111e-05, + "loss": 0.1128, + "step": 4426 + }, + { + "epoch": 86.01, + "learning_rate": 6.4e-05, + "loss": 0.081, + "step": 4427 + }, + { + "epoch": 86.01, + "learning_rate": 6.388888888888888e-05, + "loss": 0.0905, + "step": 4428 + }, + { + "epoch": 86.01, + "learning_rate": 6.377777777777778e-05, + "loss": 0.1175, + "step": 4429 + }, + { + "epoch": 86.01, + "learning_rate": 6.366666666666666e-05, + "loss": 0.0812, + "step": 4430 + }, + { + "epoch": 86.01, + "learning_rate": 6.355555555555556e-05, + "loss": 0.096, + "step": 4431 + }, + { + "epoch": 86.01, + "learning_rate": 6.344444444444445e-05, + "loss": 0.1069, + "step": 4432 + }, + { + "epoch": 86.01, + "learning_rate": 6.333333333333335e-05, + "loss": 0.0929, + "step": 4433 + }, + { + "epoch": 86.01, + "learning_rate": 6.322222222222222e-05, + "loss": 0.1049, + "step": 4434 + }, + { + "epoch": 86.01, + "learning_rate": 6.311111111111112e-05, + "loss": 0.128, + "step": 4435 + }, + { + "epoch": 86.01, + "learning_rate": 6.3e-05, + "loss": 0.0522, + "step": 4436 + }, + { + "epoch": 86.01, + "learning_rate": 6.288888888888888e-05, + "loss": 0.1261, + "step": 4437 + }, + { + "epoch": 87.0, + "learning_rate": 6.277777777777778e-05, + "loss": 0.1221, + "step": 4438 + }, + { + "epoch": 87.0, + "learning_rate": 6.266666666666667e-05, + "loss": 0.0742, + "step": 4439 + }, + { + "epoch": 87.0, + "learning_rate": 6.255555555555557e-05, + "loss": 0.08, + "step": 4440 + }, + { + "epoch": 87.0, + "learning_rate": 6.244444444444444e-05, + "loss": 0.1018, + "step": 4441 + }, + { + "epoch": 87.0, + "learning_rate": 6.233333333333334e-05, + "loss": 0.0574, + "step": 4442 + }, + { + "epoch": 87.0, + "learning_rate": 6.222222222222222e-05, + "loss": 0.1576, + "step": 4443 + }, + { + "epoch": 87.0, + "learning_rate": 6.211111111111112e-05, + "loss": 0.0902, + "step": 4444 + }, + { + "epoch": 87.0, + "learning_rate": 6.2e-05, + "loss": 0.1029, + "step": 4445 + }, + { + "epoch": 87.0, + "learning_rate": 6.188888888888889e-05, + "loss": 0.068, + "step": 4446 + }, + { + "epoch": 87.0, + "learning_rate": 6.177777777777779e-05, + "loss": 0.0491, + "step": 4447 + }, + { + "epoch": 87.0, + "learning_rate": 6.166666666666667e-05, + "loss": 0.0855, + "step": 4448 + }, + { + "epoch": 87.0, + "learning_rate": 6.155555555555555e-05, + "loss": 0.0874, + "step": 4449 + }, + { + "epoch": 87.0, + "learning_rate": 6.144444444444445e-05, + "loss": 0.0862, + "step": 4450 + }, + { + "epoch": 87.0, + "learning_rate": 6.133333333333334e-05, + "loss": 0.092, + "step": 4451 + }, + { + "epoch": 87.0, + "learning_rate": 6.122222222222222e-05, + "loss": 0.0782, + "step": 4452 + }, + { + "epoch": 87.0, + "learning_rate": 6.111111111111111e-05, + "loss": 0.1327, + "step": 4453 + }, + { + "epoch": 87.0, + "learning_rate": 6.1e-05, + "loss": 0.0801, + "step": 4454 + }, + { + "epoch": 87.0, + "learning_rate": 6.088888888888889e-05, + "loss": 0.0748, + "step": 4455 + }, + { + "epoch": 87.0, + "learning_rate": 6.077777777777778e-05, + "loss": 0.0675, + "step": 4456 + }, + { + "epoch": 87.0, + "learning_rate": 6.0666666666666666e-05, + "loss": 0.0916, + "step": 4457 + }, + { + "epoch": 87.0, + "learning_rate": 6.055555555555556e-05, + "loss": 0.084, + "step": 4458 + }, + { + "epoch": 87.0, + "learning_rate": 6.044444444444445e-05, + "loss": 0.0802, + "step": 4459 + }, + { + "epoch": 87.0, + "learning_rate": 6.0333333333333334e-05, + "loss": 0.1481, + "step": 4460 + }, + { + "epoch": 87.0, + "learning_rate": 6.0222222222222225e-05, + "loss": 0.0893, + "step": 4461 + }, + { + "epoch": 87.0, + "learning_rate": 6.011111111111111e-05, + "loss": 0.0916, + "step": 4462 + }, + { + "epoch": 87.01, + "learning_rate": 6e-05, + "loss": 0.1056, + "step": 4463 + }, + { + "epoch": 87.01, + "learning_rate": 5.9888888888888886e-05, + "loss": 0.0844, + "step": 4464 + }, + { + "epoch": 87.01, + "learning_rate": 5.977777777777778e-05, + "loss": 0.1455, + "step": 4465 + }, + { + "epoch": 87.01, + "learning_rate": 5.966666666666667e-05, + "loss": 0.1019, + "step": 4466 + }, + { + "epoch": 87.01, + "learning_rate": 5.955555555555556e-05, + "loss": 0.1031, + "step": 4467 + }, + { + "epoch": 87.01, + "learning_rate": 5.9444444444444445e-05, + "loss": 0.0688, + "step": 4468 + }, + { + "epoch": 87.01, + "learning_rate": 5.933333333333334e-05, + "loss": 0.072, + "step": 4469 + }, + { + "epoch": 87.01, + "learning_rate": 5.922222222222223e-05, + "loss": 0.0931, + "step": 4470 + }, + { + "epoch": 87.01, + "learning_rate": 5.911111111111111e-05, + "loss": 0.0644, + "step": 4471 + }, + { + "epoch": 87.01, + "learning_rate": 5.9e-05, + "loss": 0.1122, + "step": 4472 + }, + { + "epoch": 87.01, + "learning_rate": 5.888888888888889e-05, + "loss": 0.0705, + "step": 4473 + }, + { + "epoch": 87.01, + "learning_rate": 5.877777777777778e-05, + "loss": 0.0823, + "step": 4474 + }, + { + "epoch": 87.01, + "learning_rate": 5.8666666666666665e-05, + "loss": 0.0676, + "step": 4475 + }, + { + "epoch": 87.01, + "learning_rate": 5.8555555555555557e-05, + "loss": 0.0862, + "step": 4476 + }, + { + "epoch": 87.01, + "learning_rate": 5.844444444444445e-05, + "loss": 0.1212, + "step": 4477 + }, + { + "epoch": 87.01, + "learning_rate": 5.833333333333333e-05, + "loss": 0.0723, + "step": 4478 + }, + { + "epoch": 87.01, + "learning_rate": 5.8222222222222224e-05, + "loss": 0.1139, + "step": 4479 + }, + { + "epoch": 87.01, + "learning_rate": 5.8111111111111116e-05, + "loss": 0.0901, + "step": 4480 + }, + { + "epoch": 87.01, + "learning_rate": 5.800000000000001e-05, + "loss": 0.074, + "step": 4481 + }, + { + "epoch": 87.01, + "learning_rate": 5.7888888888888885e-05, + "loss": 0.1164, + "step": 4482 + }, + { + "epoch": 87.01, + "learning_rate": 5.7777777777777776e-05, + "loss": 0.0974, + "step": 4483 + }, + { + "epoch": 87.01, + "learning_rate": 5.766666666666667e-05, + "loss": 0.0938, + "step": 4484 + }, + { + "epoch": 87.01, + "learning_rate": 5.755555555555555e-05, + "loss": 0.084, + "step": 4485 + }, + { + "epoch": 87.01, + "learning_rate": 5.7444444444444444e-05, + "loss": 0.0944, + "step": 4486 + }, + { + "epoch": 87.01, + "learning_rate": 5.7333333333333336e-05, + "loss": 0.0879, + "step": 4487 + }, + { + "epoch": 87.01, + "learning_rate": 5.722222222222223e-05, + "loss": 0.1039, + "step": 4488 + }, + { + "epoch": 88.0, + "learning_rate": 5.711111111111111e-05, + "loss": 0.1602, + "step": 4489 + }, + { + "epoch": 88.0, + "learning_rate": 5.7e-05, + "loss": 0.0638, + "step": 4490 + }, + { + "epoch": 88.0, + "learning_rate": 5.6888888888888895e-05, + "loss": 0.0757, + "step": 4491 + }, + { + "epoch": 88.0, + "learning_rate": 5.677777777777778e-05, + "loss": 0.0873, + "step": 4492 + }, + { + "epoch": 88.0, + "learning_rate": 5.6666666666666664e-05, + "loss": 0.0914, + "step": 4493 + }, + { + "epoch": 88.0, + "learning_rate": 5.6555555555555555e-05, + "loss": 0.0823, + "step": 4494 + }, + { + "epoch": 88.0, + "learning_rate": 5.644444444444445e-05, + "loss": 0.0692, + "step": 4495 + }, + { + "epoch": 88.0, + "learning_rate": 5.633333333333333e-05, + "loss": 0.0559, + "step": 4496 + }, + { + "epoch": 88.0, + "learning_rate": 5.622222222222222e-05, + "loss": 0.0843, + "step": 4497 + }, + { + "epoch": 88.0, + "learning_rate": 5.6111111111111114e-05, + "loss": 0.1046, + "step": 4498 + }, + { + "epoch": 88.0, + "learning_rate": 5.6e-05, + "loss": 0.0754, + "step": 4499 + }, + { + "epoch": 88.0, + "learning_rate": 5.588888888888889e-05, + "loss": 0.0893, + "step": 4500 + }, + { + "epoch": 88.0, + "eval_cer": 0.07122336605219057, + "eval_loss": 0.35592591762542725, + "eval_runtime": 53.0469, + "eval_samples_per_second": 7.842, + "eval_steps_per_second": 0.245, + "eval_wer": 0.29681112019623873, + "step": 4500 + }, + { + "epoch": 88.0, + "learning_rate": 5.577777777777778e-05, + "loss": 0.06, + "step": 4501 + }, + { + "epoch": 88.0, + "learning_rate": 5.5666666666666674e-05, + "loss": 0.0972, + "step": 4502 + }, + { + "epoch": 88.0, + "learning_rate": 5.555555555555555e-05, + "loss": 0.0855, + "step": 4503 + }, + { + "epoch": 88.0, + "learning_rate": 5.544444444444444e-05, + "loss": 0.0814, + "step": 4504 + }, + { + "epoch": 88.0, + "learning_rate": 5.5333333333333334e-05, + "loss": 0.1093, + "step": 4505 + }, + { + "epoch": 88.0, + "learning_rate": 5.522222222222222e-05, + "loss": 0.0881, + "step": 4506 + }, + { + "epoch": 88.0, + "learning_rate": 5.511111111111111e-05, + "loss": 0.0737, + "step": 4507 + }, + { + "epoch": 88.0, + "learning_rate": 5.5e-05, + "loss": 0.0966, + "step": 4508 + }, + { + "epoch": 88.0, + "learning_rate": 5.4888888888888893e-05, + "loss": 0.0861, + "step": 4509 + }, + { + "epoch": 88.0, + "learning_rate": 5.477777777777778e-05, + "loss": 0.0798, + "step": 4510 + }, + { + "epoch": 88.0, + "learning_rate": 5.466666666666667e-05, + "loss": 0.0701, + "step": 4511 + }, + { + "epoch": 88.0, + "learning_rate": 5.455555555555556e-05, + "loss": 0.0985, + "step": 4512 + }, + { + "epoch": 88.0, + "learning_rate": 5.444444444444444e-05, + "loss": 0.083, + "step": 4513 + }, + { + "epoch": 88.01, + "learning_rate": 5.433333333333333e-05, + "loss": 0.0724, + "step": 4514 + }, + { + "epoch": 88.01, + "learning_rate": 5.422222222222222e-05, + "loss": 0.1085, + "step": 4515 + }, + { + "epoch": 88.01, + "learning_rate": 5.411111111111111e-05, + "loss": 0.1018, + "step": 4516 + }, + { + "epoch": 88.01, + "learning_rate": 5.4e-05, + "loss": 0.0537, + "step": 4517 + }, + { + "epoch": 88.01, + "learning_rate": 5.388888888888889e-05, + "loss": 0.1052, + "step": 4518 + }, + { + "epoch": 88.01, + "learning_rate": 5.377777777777778e-05, + "loss": 0.1022, + "step": 4519 + }, + { + "epoch": 88.01, + "learning_rate": 5.366666666666667e-05, + "loss": 0.0781, + "step": 4520 + }, + { + "epoch": 88.01, + "learning_rate": 5.355555555555556e-05, + "loss": 0.1155, + "step": 4521 + }, + { + "epoch": 88.01, + "learning_rate": 5.344444444444445e-05, + "loss": 0.0993, + "step": 4522 + }, + { + "epoch": 88.01, + "learning_rate": 5.333333333333334e-05, + "loss": 0.0595, + "step": 4523 + }, + { + "epoch": 88.01, + "learning_rate": 5.322222222222222e-05, + "loss": 0.0918, + "step": 4524 + }, + { + "epoch": 88.01, + "learning_rate": 5.311111111111111e-05, + "loss": 0.0743, + "step": 4525 + }, + { + "epoch": 88.01, + "learning_rate": 5.3e-05, + "loss": 0.0968, + "step": 4526 + }, + { + "epoch": 88.01, + "learning_rate": 5.288888888888889e-05, + "loss": 0.0785, + "step": 4527 + }, + { + "epoch": 88.01, + "learning_rate": 5.277777777777778e-05, + "loss": 0.0826, + "step": 4528 + }, + { + "epoch": 88.01, + "learning_rate": 5.266666666666667e-05, + "loss": 0.0838, + "step": 4529 + }, + { + "epoch": 88.01, + "learning_rate": 5.255555555555556e-05, + "loss": 0.0902, + "step": 4530 + }, + { + "epoch": 88.01, + "learning_rate": 5.2444444444444445e-05, + "loss": 0.1029, + "step": 4531 + }, + { + "epoch": 88.01, + "learning_rate": 5.2333333333333336e-05, + "loss": 0.088, + "step": 4532 + }, + { + "epoch": 88.01, + "learning_rate": 5.222222222222223e-05, + "loss": 0.07, + "step": 4533 + }, + { + "epoch": 88.01, + "learning_rate": 5.211111111111111e-05, + "loss": 0.0612, + "step": 4534 + }, + { + "epoch": 88.01, + "learning_rate": 5.2e-05, + "loss": 0.0723, + "step": 4535 + }, + { + "epoch": 88.01, + "learning_rate": 5.188888888888889e-05, + "loss": 0.0939, + "step": 4536 + }, + { + "epoch": 88.01, + "learning_rate": 5.177777777777778e-05, + "loss": 0.1029, + "step": 4537 + }, + { + "epoch": 88.01, + "learning_rate": 5.1666666666666664e-05, + "loss": 0.0805, + "step": 4538 + }, + { + "epoch": 88.01, + "learning_rate": 5.1555555555555556e-05, + "loss": 0.0911, + "step": 4539 + }, + { + "epoch": 89.0, + "learning_rate": 5.144444444444445e-05, + "loss": 0.1325, + "step": 4540 + }, + { + "epoch": 89.0, + "learning_rate": 5.133333333333334e-05, + "loss": 0.0797, + "step": 4541 + }, + { + "epoch": 89.0, + "learning_rate": 5.1222222222222223e-05, + "loss": 0.0714, + "step": 4542 + }, + { + "epoch": 89.0, + "learning_rate": 5.1111111111111115e-05, + "loss": 0.0678, + "step": 4543 + }, + { + "epoch": 89.0, + "learning_rate": 5.1e-05, + "loss": 0.0711, + "step": 4544 + }, + { + "epoch": 89.0, + "learning_rate": 5.0888888888888884e-05, + "loss": 0.0629, + "step": 4545 + }, + { + "epoch": 89.0, + "learning_rate": 5.0777777777777776e-05, + "loss": 0.1026, + "step": 4546 + }, + { + "epoch": 89.0, + "learning_rate": 5.066666666666667e-05, + "loss": 0.0734, + "step": 4547 + }, + { + "epoch": 89.0, + "learning_rate": 5.055555555555556e-05, + "loss": 0.0597, + "step": 4548 + }, + { + "epoch": 89.0, + "learning_rate": 5.044444444444444e-05, + "loss": 0.099, + "step": 4549 + }, + { + "epoch": 89.0, + "learning_rate": 5.0333333333333335e-05, + "loss": 0.0664, + "step": 4550 + }, + { + "epoch": 89.0, + "learning_rate": 5.0222222222222226e-05, + "loss": 0.0752, + "step": 4551 + }, + { + "epoch": 89.0, + "learning_rate": 5.011111111111111e-05, + "loss": 0.0786, + "step": 4552 + }, + { + "epoch": 89.0, + "learning_rate": 5e-05, + "loss": 0.0596, + "step": 4553 + }, + { + "epoch": 89.0, + "learning_rate": 4.9888888888888894e-05, + "loss": 0.088, + "step": 4554 + }, + { + "epoch": 89.0, + "learning_rate": 4.977777777777778e-05, + "loss": 0.093, + "step": 4555 + }, + { + "epoch": 89.0, + "learning_rate": 4.966666666666666e-05, + "loss": 0.0941, + "step": 4556 + }, + { + "epoch": 89.0, + "learning_rate": 4.9555555555555555e-05, + "loss": 0.0579, + "step": 4557 + }, + { + "epoch": 89.0, + "learning_rate": 4.9444444444444446e-05, + "loss": 0.1103, + "step": 4558 + }, + { + "epoch": 89.0, + "learning_rate": 4.933333333333334e-05, + "loss": 0.0924, + "step": 4559 + }, + { + "epoch": 89.0, + "learning_rate": 4.922222222222222e-05, + "loss": 0.0914, + "step": 4560 + }, + { + "epoch": 89.0, + "learning_rate": 4.9111111111111114e-05, + "loss": 0.0753, + "step": 4561 + }, + { + "epoch": 89.0, + "learning_rate": 4.9000000000000005e-05, + "loss": 0.1089, + "step": 4562 + }, + { + "epoch": 89.0, + "learning_rate": 4.888888888888889e-05, + "loss": 0.0957, + "step": 4563 + }, + { + "epoch": 89.0, + "learning_rate": 4.877777777777778e-05, + "loss": 0.0652, + "step": 4564 + }, + { + "epoch": 89.01, + "learning_rate": 4.8666666666666666e-05, + "loss": 0.1079, + "step": 4565 + }, + { + "epoch": 89.01, + "learning_rate": 4.855555555555556e-05, + "loss": 0.0803, + "step": 4566 + }, + { + "epoch": 89.01, + "learning_rate": 4.844444444444444e-05, + "loss": 0.0565, + "step": 4567 + }, + { + "epoch": 89.01, + "learning_rate": 4.8333333333333334e-05, + "loss": 0.0844, + "step": 4568 + }, + { + "epoch": 89.01, + "learning_rate": 4.8222222222222225e-05, + "loss": 0.0867, + "step": 4569 + }, + { + "epoch": 89.01, + "learning_rate": 4.811111111111111e-05, + "loss": 0.0827, + "step": 4570 + }, + { + "epoch": 89.01, + "learning_rate": 4.8e-05, + "loss": 0.0795, + "step": 4571 + }, + { + "epoch": 89.01, + "learning_rate": 4.788888888888889e-05, + "loss": 0.1065, + "step": 4572 + }, + { + "epoch": 89.01, + "learning_rate": 4.7777777777777784e-05, + "loss": 0.0795, + "step": 4573 + }, + { + "epoch": 89.01, + "learning_rate": 4.766666666666667e-05, + "loss": 0.0871, + "step": 4574 + }, + { + "epoch": 89.01, + "learning_rate": 4.7555555555555553e-05, + "loss": 0.0895, + "step": 4575 + }, + { + "epoch": 89.01, + "learning_rate": 4.7444444444444445e-05, + "loss": 0.0682, + "step": 4576 + }, + { + "epoch": 89.01, + "learning_rate": 4.733333333333333e-05, + "loss": 0.0537, + "step": 4577 + }, + { + "epoch": 89.01, + "learning_rate": 4.722222222222222e-05, + "loss": 0.0724, + "step": 4578 + }, + { + "epoch": 89.01, + "learning_rate": 4.711111111111111e-05, + "loss": 0.0589, + "step": 4579 + }, + { + "epoch": 89.01, + "learning_rate": 4.7000000000000004e-05, + "loss": 0.0989, + "step": 4580 + }, + { + "epoch": 89.01, + "learning_rate": 4.688888888888889e-05, + "loss": 0.0759, + "step": 4581 + }, + { + "epoch": 89.01, + "learning_rate": 4.677777777777778e-05, + "loss": 0.0612, + "step": 4582 + }, + { + "epoch": 89.01, + "learning_rate": 4.666666666666667e-05, + "loss": 0.0968, + "step": 4583 + }, + { + "epoch": 89.01, + "learning_rate": 4.6555555555555556e-05, + "loss": 0.0957, + "step": 4584 + }, + { + "epoch": 89.01, + "learning_rate": 4.644444444444444e-05, + "loss": 0.0685, + "step": 4585 + }, + { + "epoch": 89.01, + "learning_rate": 4.633333333333333e-05, + "loss": 0.111, + "step": 4586 + }, + { + "epoch": 89.01, + "learning_rate": 4.6222222222222224e-05, + "loss": 0.0768, + "step": 4587 + }, + { + "epoch": 89.01, + "learning_rate": 4.611111111111111e-05, + "loss": 0.1044, + "step": 4588 + }, + { + "epoch": 89.01, + "learning_rate": 4.6e-05, + "loss": 0.1068, + "step": 4589 + }, + { + "epoch": 89.01, + "learning_rate": 4.588888888888889e-05, + "loss": 0.0666, + "step": 4590 + }, + { + "epoch": 90.0, + "learning_rate": 4.5777777777777776e-05, + "loss": 0.1635, + "step": 4591 + }, + { + "epoch": 90.0, + "learning_rate": 4.566666666666667e-05, + "loss": 0.0983, + "step": 4592 + }, + { + "epoch": 90.0, + "learning_rate": 4.555555555555556e-05, + "loss": 0.0775, + "step": 4593 + }, + { + "epoch": 90.0, + "learning_rate": 4.544444444444445e-05, + "loss": 0.0516, + "step": 4594 + }, + { + "epoch": 90.0, + "learning_rate": 4.5333333333333335e-05, + "loss": 0.0567, + "step": 4595 + }, + { + "epoch": 90.0, + "learning_rate": 4.522222222222222e-05, + "loss": 0.0696, + "step": 4596 + }, + { + "epoch": 90.0, + "learning_rate": 4.511111111111111e-05, + "loss": 0.0563, + "step": 4597 + }, + { + "epoch": 90.0, + "learning_rate": 4.4999999999999996e-05, + "loss": 0.0695, + "step": 4598 + }, + { + "epoch": 90.0, + "learning_rate": 4.488888888888889e-05, + "loss": 0.0813, + "step": 4599 + }, + { + "epoch": 90.0, + "learning_rate": 4.477777777777778e-05, + "loss": 0.0625, + "step": 4600 + }, + { + "epoch": 90.0, + "learning_rate": 4.466666666666667e-05, + "loss": 0.0679, + "step": 4601 + }, + { + "epoch": 90.0, + "learning_rate": 4.4555555555555555e-05, + "loss": 0.1048, + "step": 4602 + }, + { + "epoch": 90.0, + "learning_rate": 4.4444444444444447e-05, + "loss": 0.1014, + "step": 4603 + }, + { + "epoch": 90.0, + "learning_rate": 4.433333333333334e-05, + "loss": 0.0785, + "step": 4604 + }, + { + "epoch": 90.0, + "learning_rate": 4.422222222222223e-05, + "loss": 0.0624, + "step": 4605 + }, + { + "epoch": 90.0, + "learning_rate": 4.411111111111111e-05, + "loss": 0.1159, + "step": 4606 + }, + { + "epoch": 90.0, + "learning_rate": 4.4e-05, + "loss": 0.0751, + "step": 4607 + }, + { + "epoch": 90.0, + "learning_rate": 4.388888888888889e-05, + "loss": 0.0914, + "step": 4608 + }, + { + "epoch": 90.0, + "learning_rate": 4.3777777777777775e-05, + "loss": 0.084, + "step": 4609 + }, + { + "epoch": 90.0, + "learning_rate": 4.3666666666666666e-05, + "loss": 0.0919, + "step": 4610 + }, + { + "epoch": 90.0, + "learning_rate": 4.355555555555556e-05, + "loss": 0.0692, + "step": 4611 + }, + { + "epoch": 90.0, + "learning_rate": 4.344444444444445e-05, + "loss": 0.0784, + "step": 4612 + }, + { + "epoch": 90.0, + "learning_rate": 4.3333333333333334e-05, + "loss": 0.0922, + "step": 4613 + }, + { + "epoch": 90.0, + "learning_rate": 4.3222222222222226e-05, + "loss": 0.1021, + "step": 4614 + }, + { + "epoch": 90.0, + "learning_rate": 4.311111111111112e-05, + "loss": 0.0688, + "step": 4615 + }, + { + "epoch": 90.01, + "learning_rate": 4.2999999999999995e-05, + "loss": 0.0792, + "step": 4616 + }, + { + "epoch": 90.01, + "learning_rate": 4.2888888888888886e-05, + "loss": 0.0667, + "step": 4617 + }, + { + "epoch": 90.01, + "learning_rate": 4.277777777777778e-05, + "loss": 0.0742, + "step": 4618 + }, + { + "epoch": 90.01, + "learning_rate": 4.266666666666667e-05, + "loss": 0.0909, + "step": 4619 + }, + { + "epoch": 90.01, + "learning_rate": 4.2555555555555554e-05, + "loss": 0.1067, + "step": 4620 + }, + { + "epoch": 90.01, + "learning_rate": 4.2444444444444445e-05, + "loss": 0.1036, + "step": 4621 + }, + { + "epoch": 90.01, + "learning_rate": 4.233333333333334e-05, + "loss": 0.0884, + "step": 4622 + }, + { + "epoch": 90.01, + "learning_rate": 4.222222222222222e-05, + "loss": 0.0884, + "step": 4623 + }, + { + "epoch": 90.01, + "learning_rate": 4.211111111111111e-05, + "loss": 0.0789, + "step": 4624 + }, + { + "epoch": 90.01, + "learning_rate": 4.2000000000000004e-05, + "loss": 0.0596, + "step": 4625 + }, + { + "epoch": 90.01, + "learning_rate": 4.1888888888888896e-05, + "loss": 0.0783, + "step": 4626 + }, + { + "epoch": 90.01, + "learning_rate": 4.1777777777777774e-05, + "loss": 0.0895, + "step": 4627 + }, + { + "epoch": 90.01, + "learning_rate": 4.1666666666666665e-05, + "loss": 0.0825, + "step": 4628 + }, + { + "epoch": 90.01, + "learning_rate": 4.155555555555556e-05, + "loss": 0.082, + "step": 4629 + }, + { + "epoch": 90.01, + "learning_rate": 4.144444444444444e-05, + "loss": 0.111, + "step": 4630 + }, + { + "epoch": 90.01, + "learning_rate": 4.133333333333333e-05, + "loss": 0.1012, + "step": 4631 + }, + { + "epoch": 90.01, + "learning_rate": 4.1222222222222224e-05, + "loss": 0.0929, + "step": 4632 + }, + { + "epoch": 90.01, + "learning_rate": 4.1111111111111116e-05, + "loss": 0.1018, + "step": 4633 + }, + { + "epoch": 90.01, + "learning_rate": 4.1e-05, + "loss": 0.0723, + "step": 4634 + }, + { + "epoch": 90.01, + "learning_rate": 4.088888888888889e-05, + "loss": 0.0598, + "step": 4635 + }, + { + "epoch": 90.01, + "learning_rate": 4.0777777777777783e-05, + "loss": 0.0693, + "step": 4636 + }, + { + "epoch": 90.01, + "learning_rate": 4.066666666666666e-05, + "loss": 0.0657, + "step": 4637 + }, + { + "epoch": 90.01, + "learning_rate": 4.055555555555555e-05, + "loss": 0.071, + "step": 4638 + }, + { + "epoch": 90.01, + "learning_rate": 4.0444444444444444e-05, + "loss": 0.0972, + "step": 4639 + }, + { + "epoch": 90.01, + "learning_rate": 4.0333333333333336e-05, + "loss": 0.1024, + "step": 4640 + }, + { + "epoch": 90.01, + "learning_rate": 4.022222222222222e-05, + "loss": 0.1144, + "step": 4641 + }, + { + "epoch": 91.0, + "learning_rate": 4.011111111111111e-05, + "loss": 0.1468, + "step": 4642 + }, + { + "epoch": 91.0, + "learning_rate": 4e-05, + "loss": 0.0574, + "step": 4643 + }, + { + "epoch": 91.0, + "learning_rate": 3.988888888888889e-05, + "loss": 0.089, + "step": 4644 + }, + { + "epoch": 91.0, + "learning_rate": 3.977777777777778e-05, + "loss": 0.0691, + "step": 4645 + }, + { + "epoch": 91.0, + "learning_rate": 3.966666666666667e-05, + "loss": 0.0671, + "step": 4646 + }, + { + "epoch": 91.0, + "learning_rate": 3.9555555555555556e-05, + "loss": 0.0725, + "step": 4647 + }, + { + "epoch": 91.0, + "learning_rate": 3.944444444444444e-05, + "loss": 0.0653, + "step": 4648 + }, + { + "epoch": 91.0, + "learning_rate": 3.933333333333333e-05, + "loss": 0.0822, + "step": 4649 + }, + { + "epoch": 91.0, + "learning_rate": 3.922222222222222e-05, + "loss": 0.072, + "step": 4650 + }, + { + "epoch": 91.0, + "learning_rate": 3.911111111111111e-05, + "loss": 0.0639, + "step": 4651 + }, + { + "epoch": 91.0, + "learning_rate": 3.9e-05, + "loss": 0.0941, + "step": 4652 + }, + { + "epoch": 91.0, + "learning_rate": 3.888888888888889e-05, + "loss": 0.0759, + "step": 4653 + }, + { + "epoch": 91.0, + "learning_rate": 3.877777777777778e-05, + "loss": 0.0822, + "step": 4654 + }, + { + "epoch": 91.0, + "learning_rate": 3.866666666666667e-05, + "loss": 0.0652, + "step": 4655 + }, + { + "epoch": 91.0, + "learning_rate": 3.855555555555556e-05, + "loss": 0.1278, + "step": 4656 + }, + { + "epoch": 91.0, + "learning_rate": 3.844444444444445e-05, + "loss": 0.0644, + "step": 4657 + }, + { + "epoch": 91.0, + "learning_rate": 3.8333333333333334e-05, + "loss": 0.0714, + "step": 4658 + }, + { + "epoch": 91.0, + "learning_rate": 3.822222222222222e-05, + "loss": 0.0528, + "step": 4659 + }, + { + "epoch": 91.0, + "learning_rate": 3.811111111111111e-05, + "loss": 0.0645, + "step": 4660 + }, + { + "epoch": 91.0, + "learning_rate": 3.8e-05, + "loss": 0.0898, + "step": 4661 + }, + { + "epoch": 91.0, + "learning_rate": 3.788888888888889e-05, + "loss": 0.0721, + "step": 4662 + }, + { + "epoch": 91.0, + "learning_rate": 3.777777777777778e-05, + "loss": 0.0617, + "step": 4663 + }, + { + "epoch": 91.0, + "learning_rate": 3.766666666666667e-05, + "loss": 0.0956, + "step": 4664 + }, + { + "epoch": 91.0, + "learning_rate": 3.755555555555556e-05, + "loss": 0.0605, + "step": 4665 + }, + { + "epoch": 91.0, + "learning_rate": 3.7444444444444446e-05, + "loss": 0.1123, + "step": 4666 + }, + { + "epoch": 91.01, + "learning_rate": 3.733333333333334e-05, + "loss": 0.117, + "step": 4667 + }, + { + "epoch": 91.01, + "learning_rate": 3.722222222222222e-05, + "loss": 0.0715, + "step": 4668 + }, + { + "epoch": 91.01, + "learning_rate": 3.711111111111111e-05, + "loss": 0.0589, + "step": 4669 + }, + { + "epoch": 91.01, + "learning_rate": 3.7e-05, + "loss": 0.1025, + "step": 4670 + }, + { + "epoch": 91.01, + "learning_rate": 3.688888888888889e-05, + "loss": 0.072, + "step": 4671 + }, + { + "epoch": 91.01, + "learning_rate": 3.677777777777778e-05, + "loss": 0.0692, + "step": 4672 + }, + { + "epoch": 91.01, + "learning_rate": 3.6666666666666666e-05, + "loss": 0.0588, + "step": 4673 + }, + { + "epoch": 91.01, + "learning_rate": 3.655555555555556e-05, + "loss": 0.0459, + "step": 4674 + }, + { + "epoch": 91.01, + "learning_rate": 3.644444444444445e-05, + "loss": 0.1431, + "step": 4675 + }, + { + "epoch": 91.01, + "learning_rate": 3.633333333333333e-05, + "loss": 0.0654, + "step": 4676 + }, + { + "epoch": 91.01, + "learning_rate": 3.6222222222222225e-05, + "loss": 0.0841, + "step": 4677 + }, + { + "epoch": 91.01, + "learning_rate": 3.611111111111111e-05, + "loss": 0.0822, + "step": 4678 + }, + { + "epoch": 91.01, + "learning_rate": 3.6e-05, + "loss": 0.0807, + "step": 4679 + }, + { + "epoch": 91.01, + "learning_rate": 3.5888888888888886e-05, + "loss": 0.0717, + "step": 4680 + }, + { + "epoch": 91.01, + "learning_rate": 3.577777777777778e-05, + "loss": 0.0915, + "step": 4681 + }, + { + "epoch": 91.01, + "learning_rate": 3.566666666666667e-05, + "loss": 0.0724, + "step": 4682 + }, + { + "epoch": 91.01, + "learning_rate": 3.555555555555555e-05, + "loss": 0.1111, + "step": 4683 + }, + { + "epoch": 91.01, + "learning_rate": 3.5444444444444445e-05, + "loss": 0.0855, + "step": 4684 + }, + { + "epoch": 91.01, + "learning_rate": 3.5333333333333336e-05, + "loss": 0.0585, + "step": 4685 + }, + { + "epoch": 91.01, + "learning_rate": 3.522222222222223e-05, + "loss": 0.1013, + "step": 4686 + }, + { + "epoch": 91.01, + "learning_rate": 3.511111111111111e-05, + "loss": 0.0662, + "step": 4687 + }, + { + "epoch": 91.01, + "learning_rate": 3.5000000000000004e-05, + "loss": 0.1033, + "step": 4688 + }, + { + "epoch": 91.01, + "learning_rate": 3.488888888888889e-05, + "loss": 0.0879, + "step": 4689 + }, + { + "epoch": 91.01, + "learning_rate": 3.477777777777777e-05, + "loss": 0.0946, + "step": 4690 + }, + { + "epoch": 91.01, + "learning_rate": 3.4666666666666665e-05, + "loss": 0.1312, + "step": 4691 + }, + { + "epoch": 91.01, + "learning_rate": 3.4555555555555556e-05, + "loss": 0.1115, + "step": 4692 + }, + { + "epoch": 92.0, + "learning_rate": 3.444444444444445e-05, + "loss": 0.1256, + "step": 4693 + }, + { + "epoch": 92.0, + "learning_rate": 3.433333333333333e-05, + "loss": 0.0728, + "step": 4694 + }, + { + "epoch": 92.0, + "learning_rate": 3.4222222222222224e-05, + "loss": 0.0846, + "step": 4695 + }, + { + "epoch": 92.0, + "learning_rate": 3.4111111111111115e-05, + "loss": 0.0809, + "step": 4696 + }, + { + "epoch": 92.0, + "learning_rate": 3.4000000000000007e-05, + "loss": 0.0604, + "step": 4697 + }, + { + "epoch": 92.0, + "learning_rate": 3.388888888888889e-05, + "loss": 0.0732, + "step": 4698 + }, + { + "epoch": 92.0, + "learning_rate": 3.3777777777777776e-05, + "loss": 0.0797, + "step": 4699 + }, + { + "epoch": 92.0, + "learning_rate": 3.366666666666667e-05, + "loss": 0.0503, + "step": 4700 + }, + { + "epoch": 92.0, + "learning_rate": 3.355555555555555e-05, + "loss": 0.066, + "step": 4701 + }, + { + "epoch": 92.0, + "learning_rate": 3.3444444444444443e-05, + "loss": 0.1017, + "step": 4702 + }, + { + "epoch": 92.0, + "learning_rate": 3.3333333333333335e-05, + "loss": 0.0573, + "step": 4703 + }, + { + "epoch": 92.0, + "learning_rate": 3.3222222222222226e-05, + "loss": 0.0804, + "step": 4704 + }, + { + "epoch": 92.0, + "learning_rate": 3.311111111111111e-05, + "loss": 0.111, + "step": 4705 + }, + { + "epoch": 92.0, + "learning_rate": 3.3e-05, + "loss": 0.0644, + "step": 4706 + }, + { + "epoch": 92.0, + "learning_rate": 3.2888888888888894e-05, + "loss": 0.092, + "step": 4707 + }, + { + "epoch": 92.0, + "learning_rate": 3.277777777777778e-05, + "loss": 0.0593, + "step": 4708 + }, + { + "epoch": 92.0, + "learning_rate": 3.266666666666666e-05, + "loss": 0.0812, + "step": 4709 + }, + { + "epoch": 92.0, + "learning_rate": 3.2555555555555555e-05, + "loss": 0.071, + "step": 4710 + }, + { + "epoch": 92.0, + "learning_rate": 3.2444444444444446e-05, + "loss": 0.0716, + "step": 4711 + }, + { + "epoch": 92.0, + "learning_rate": 3.233333333333333e-05, + "loss": 0.0815, + "step": 4712 + }, + { + "epoch": 92.0, + "learning_rate": 3.222222222222222e-05, + "loss": 0.072, + "step": 4713 + }, + { + "epoch": 92.0, + "learning_rate": 3.2111111111111114e-05, + "loss": 0.0785, + "step": 4714 + }, + { + "epoch": 92.0, + "learning_rate": 3.2e-05, + "loss": 0.0712, + "step": 4715 + }, + { + "epoch": 92.0, + "learning_rate": 3.188888888888889e-05, + "loss": 0.094, + "step": 4716 + }, + { + "epoch": 92.0, + "learning_rate": 3.177777777777778e-05, + "loss": 0.0754, + "step": 4717 + }, + { + "epoch": 92.01, + "learning_rate": 3.166666666666667e-05, + "loss": 0.0906, + "step": 4718 + }, + { + "epoch": 92.01, + "learning_rate": 3.155555555555556e-05, + "loss": 0.0665, + "step": 4719 + }, + { + "epoch": 92.01, + "learning_rate": 3.144444444444444e-05, + "loss": 0.0563, + "step": 4720 + }, + { + "epoch": 92.01, + "learning_rate": 3.1333333333333334e-05, + "loss": 0.0926, + "step": 4721 + }, + { + "epoch": 92.01, + "learning_rate": 3.122222222222222e-05, + "loss": 0.1064, + "step": 4722 + }, + { + "epoch": 92.01, + "learning_rate": 3.111111111111111e-05, + "loss": 0.0949, + "step": 4723 + }, + { + "epoch": 92.01, + "learning_rate": 3.1e-05, + "loss": 0.0881, + "step": 4724 + }, + { + "epoch": 92.01, + "learning_rate": 3.088888888888889e-05, + "loss": 0.1142, + "step": 4725 + }, + { + "epoch": 92.01, + "learning_rate": 3.077777777777778e-05, + "loss": 0.0652, + "step": 4726 + }, + { + "epoch": 92.01, + "learning_rate": 3.066666666666667e-05, + "loss": 0.0792, + "step": 4727 + }, + { + "epoch": 92.01, + "learning_rate": 3.0555555555555554e-05, + "loss": 0.0843, + "step": 4728 + }, + { + "epoch": 92.01, + "learning_rate": 3.0444444444444445e-05, + "loss": 0.1007, + "step": 4729 + }, + { + "epoch": 92.01, + "learning_rate": 3.0333333333333333e-05, + "loss": 0.0643, + "step": 4730 + }, + { + "epoch": 92.01, + "learning_rate": 3.0222222222222225e-05, + "loss": 0.0607, + "step": 4731 + }, + { + "epoch": 92.01, + "learning_rate": 3.0111111111111113e-05, + "loss": 0.1044, + "step": 4732 + }, + { + "epoch": 92.01, + "learning_rate": 3e-05, + "loss": 0.0814, + "step": 4733 + }, + { + "epoch": 92.01, + "learning_rate": 2.988888888888889e-05, + "loss": 0.1056, + "step": 4734 + }, + { + "epoch": 92.01, + "learning_rate": 2.977777777777778e-05, + "loss": 0.0915, + "step": 4735 + }, + { + "epoch": 92.01, + "learning_rate": 2.966666666666667e-05, + "loss": 0.0707, + "step": 4736 + }, + { + "epoch": 92.01, + "learning_rate": 2.9555555555555556e-05, + "loss": 0.0802, + "step": 4737 + }, + { + "epoch": 92.01, + "learning_rate": 2.9444444444444445e-05, + "loss": 0.0789, + "step": 4738 + }, + { + "epoch": 92.01, + "learning_rate": 2.9333333333333333e-05, + "loss": 0.0738, + "step": 4739 + }, + { + "epoch": 92.01, + "learning_rate": 2.9222222222222224e-05, + "loss": 0.0872, + "step": 4740 + }, + { + "epoch": 92.01, + "learning_rate": 2.9111111111111112e-05, + "loss": 0.1047, + "step": 4741 + }, + { + "epoch": 92.01, + "learning_rate": 2.9000000000000004e-05, + "loss": 0.052, + "step": 4742 + }, + { + "epoch": 92.01, + "learning_rate": 2.8888888888888888e-05, + "loss": 0.0901, + "step": 4743 + }, + { + "epoch": 93.0, + "learning_rate": 2.8777777777777776e-05, + "loss": 0.1107, + "step": 4744 + }, + { + "epoch": 93.0, + "learning_rate": 2.8666666666666668e-05, + "loss": 0.0442, + "step": 4745 + }, + { + "epoch": 93.0, + "learning_rate": 2.8555555555555556e-05, + "loss": 0.0858, + "step": 4746 + }, + { + "epoch": 93.0, + "learning_rate": 2.8444444444444447e-05, + "loss": 0.0625, + "step": 4747 + }, + { + "epoch": 93.0, + "learning_rate": 2.8333333333333332e-05, + "loss": 0.047, + "step": 4748 + }, + { + "epoch": 93.0, + "learning_rate": 2.8222222222222223e-05, + "loss": 0.078, + "step": 4749 + }, + { + "epoch": 93.0, + "learning_rate": 2.811111111111111e-05, + "loss": 0.0927, + "step": 4750 + }, + { + "epoch": 93.0, + "learning_rate": 2.8e-05, + "loss": 0.0599, + "step": 4751 + }, + { + "epoch": 93.0, + "learning_rate": 2.788888888888889e-05, + "loss": 0.0748, + "step": 4752 + }, + { + "epoch": 93.0, + "learning_rate": 2.7777777777777776e-05, + "loss": 0.0862, + "step": 4753 + }, + { + "epoch": 93.0, + "learning_rate": 2.7666666666666667e-05, + "loss": 0.0834, + "step": 4754 + }, + { + "epoch": 93.0, + "learning_rate": 2.7555555555555555e-05, + "loss": 0.0696, + "step": 4755 + }, + { + "epoch": 93.0, + "learning_rate": 2.7444444444444447e-05, + "loss": 0.0718, + "step": 4756 + }, + { + "epoch": 93.0, + "learning_rate": 2.7333333333333335e-05, + "loss": 0.113, + "step": 4757 + }, + { + "epoch": 93.0, + "learning_rate": 2.722222222222222e-05, + "loss": 0.074, + "step": 4758 + }, + { + "epoch": 93.0, + "learning_rate": 2.711111111111111e-05, + "loss": 0.1124, + "step": 4759 + }, + { + "epoch": 93.0, + "learning_rate": 2.7e-05, + "loss": 0.0687, + "step": 4760 + }, + { + "epoch": 93.0, + "learning_rate": 2.688888888888889e-05, + "loss": 0.0703, + "step": 4761 + }, + { + "epoch": 93.0, + "learning_rate": 2.677777777777778e-05, + "loss": 0.0751, + "step": 4762 + }, + { + "epoch": 93.0, + "learning_rate": 2.666666666666667e-05, + "loss": 0.0986, + "step": 4763 + }, + { + "epoch": 93.0, + "learning_rate": 2.6555555555555555e-05, + "loss": 0.0697, + "step": 4764 + }, + { + "epoch": 93.0, + "learning_rate": 2.6444444444444446e-05, + "loss": 0.1148, + "step": 4765 + }, + { + "epoch": 93.0, + "learning_rate": 2.6333333333333334e-05, + "loss": 0.0703, + "step": 4766 + }, + { + "epoch": 93.0, + "learning_rate": 2.6222222222222222e-05, + "loss": 0.078, + "step": 4767 + }, + { + "epoch": 93.0, + "learning_rate": 2.6111111111111114e-05, + "loss": 0.0732, + "step": 4768 + }, + { + "epoch": 93.01, + "learning_rate": 2.6e-05, + "loss": 0.078, + "step": 4769 + }, + { + "epoch": 93.01, + "learning_rate": 2.588888888888889e-05, + "loss": 0.0675, + "step": 4770 + }, + { + "epoch": 93.01, + "learning_rate": 2.5777777777777778e-05, + "loss": 0.0799, + "step": 4771 + }, + { + "epoch": 93.01, + "learning_rate": 2.566666666666667e-05, + "loss": 0.0784, + "step": 4772 + }, + { + "epoch": 93.01, + "learning_rate": 2.5555555555555557e-05, + "loss": 0.0964, + "step": 4773 + }, + { + "epoch": 93.01, + "learning_rate": 2.5444444444444442e-05, + "loss": 0.0731, + "step": 4774 + }, + { + "epoch": 93.01, + "learning_rate": 2.5333333333333334e-05, + "loss": 0.0537, + "step": 4775 + }, + { + "epoch": 93.01, + "learning_rate": 2.522222222222222e-05, + "loss": 0.0567, + "step": 4776 + }, + { + "epoch": 93.01, + "learning_rate": 2.5111111111111113e-05, + "loss": 0.087, + "step": 4777 + }, + { + "epoch": 93.01, + "learning_rate": 2.5e-05, + "loss": 0.0529, + "step": 4778 + }, + { + "epoch": 93.01, + "learning_rate": 2.488888888888889e-05, + "loss": 0.0728, + "step": 4779 + }, + { + "epoch": 93.01, + "learning_rate": 2.4777777777777777e-05, + "loss": 0.0762, + "step": 4780 + }, + { + "epoch": 93.01, + "learning_rate": 2.466666666666667e-05, + "loss": 0.062, + "step": 4781 + }, + { + "epoch": 93.01, + "learning_rate": 2.4555555555555557e-05, + "loss": 0.1071, + "step": 4782 + }, + { + "epoch": 93.01, + "learning_rate": 2.4444444444444445e-05, + "loss": 0.1089, + "step": 4783 + }, + { + "epoch": 93.01, + "learning_rate": 2.4333333333333333e-05, + "loss": 0.0857, + "step": 4784 + }, + { + "epoch": 93.01, + "learning_rate": 2.422222222222222e-05, + "loss": 0.091, + "step": 4785 + }, + { + "epoch": 93.01, + "learning_rate": 2.4111111111111113e-05, + "loss": 0.0618, + "step": 4786 + }, + { + "epoch": 93.01, + "learning_rate": 2.4e-05, + "loss": 0.0754, + "step": 4787 + }, + { + "epoch": 93.01, + "learning_rate": 2.3888888888888892e-05, + "loss": 0.0948, + "step": 4788 + }, + { + "epoch": 93.01, + "learning_rate": 2.3777777777777777e-05, + "loss": 0.0707, + "step": 4789 + }, + { + "epoch": 93.01, + "learning_rate": 2.3666666666666665e-05, + "loss": 0.06, + "step": 4790 + }, + { + "epoch": 93.01, + "learning_rate": 2.3555555555555556e-05, + "loss": 0.0734, + "step": 4791 + }, + { + "epoch": 93.01, + "learning_rate": 2.3444444444444444e-05, + "loss": 0.0963, + "step": 4792 + }, + { + "epoch": 93.01, + "learning_rate": 2.3333333333333336e-05, + "loss": 0.0709, + "step": 4793 + }, + { + "epoch": 93.01, + "learning_rate": 2.322222222222222e-05, + "loss": 0.1193, + "step": 4794 + }, + { + "epoch": 94.0, + "learning_rate": 2.3111111111111112e-05, + "loss": 0.1127, + "step": 4795 + }, + { + "epoch": 94.0, + "learning_rate": 2.3e-05, + "loss": 0.0516, + "step": 4796 + }, + { + "epoch": 94.0, + "learning_rate": 2.2888888888888888e-05, + "loss": 0.068, + "step": 4797 + }, + { + "epoch": 94.0, + "learning_rate": 2.277777777777778e-05, + "loss": 0.0562, + "step": 4798 + }, + { + "epoch": 94.0, + "learning_rate": 2.2666666666666668e-05, + "loss": 0.0636, + "step": 4799 + }, + { + "epoch": 94.0, + "learning_rate": 2.2555555555555556e-05, + "loss": 0.0581, + "step": 4800 + }, + { + "epoch": 94.0, + "learning_rate": 2.2444444444444444e-05, + "loss": 0.0829, + "step": 4801 + }, + { + "epoch": 94.0, + "learning_rate": 2.2333333333333335e-05, + "loss": 0.0624, + "step": 4802 + }, + { + "epoch": 94.0, + "learning_rate": 2.2222222222222223e-05, + "loss": 0.0628, + "step": 4803 + }, + { + "epoch": 94.0, + "learning_rate": 2.2111111111111115e-05, + "loss": 0.0641, + "step": 4804 + }, + { + "epoch": 94.0, + "learning_rate": 2.2e-05, + "loss": 0.0718, + "step": 4805 + }, + { + "epoch": 94.0, + "learning_rate": 2.1888888888888887e-05, + "loss": 0.0433, + "step": 4806 + }, + { + "epoch": 94.0, + "learning_rate": 2.177777777777778e-05, + "loss": 0.0755, + "step": 4807 + }, + { + "epoch": 94.0, + "learning_rate": 2.1666666666666667e-05, + "loss": 0.1149, + "step": 4808 + }, + { + "epoch": 94.0, + "learning_rate": 2.155555555555556e-05, + "loss": 0.0829, + "step": 4809 + }, + { + "epoch": 94.0, + "learning_rate": 2.1444444444444443e-05, + "loss": 0.0804, + "step": 4810 + }, + { + "epoch": 94.0, + "learning_rate": 2.1333333333333335e-05, + "loss": 0.0672, + "step": 4811 + }, + { + "epoch": 94.0, + "learning_rate": 2.1222222222222223e-05, + "loss": 0.0631, + "step": 4812 + }, + { + "epoch": 94.0, + "learning_rate": 2.111111111111111e-05, + "loss": 0.1066, + "step": 4813 + }, + { + "epoch": 94.0, + "learning_rate": 2.1000000000000002e-05, + "loss": 0.0724, + "step": 4814 + }, + { + "epoch": 94.0, + "learning_rate": 2.0888888888888887e-05, + "loss": 0.0718, + "step": 4815 + }, + { + "epoch": 94.0, + "learning_rate": 2.077777777777778e-05, + "loss": 0.065, + "step": 4816 + }, + { + "epoch": 94.0, + "learning_rate": 2.0666666666666666e-05, + "loss": 0.0889, + "step": 4817 + }, + { + "epoch": 94.0, + "learning_rate": 2.0555555555555558e-05, + "loss": 0.0868, + "step": 4818 + }, + { + "epoch": 94.0, + "learning_rate": 2.0444444444444446e-05, + "loss": 0.1152, + "step": 4819 + }, + { + "epoch": 94.01, + "learning_rate": 2.033333333333333e-05, + "loss": 0.0951, + "step": 4820 + }, + { + "epoch": 94.01, + "learning_rate": 2.0222222222222222e-05, + "loss": 0.1048, + "step": 4821 + }, + { + "epoch": 94.01, + "learning_rate": 2.011111111111111e-05, + "loss": 0.0615, + "step": 4822 + }, + { + "epoch": 94.01, + "learning_rate": 2e-05, + "loss": 0.0626, + "step": 4823 + }, + { + "epoch": 94.01, + "learning_rate": 1.988888888888889e-05, + "loss": 0.0762, + "step": 4824 + }, + { + "epoch": 94.01, + "learning_rate": 1.9777777777777778e-05, + "loss": 0.0989, + "step": 4825 + }, + { + "epoch": 94.01, + "learning_rate": 1.9666666666666666e-05, + "loss": 0.0926, + "step": 4826 + }, + { + "epoch": 94.01, + "learning_rate": 1.9555555555555554e-05, + "loss": 0.066, + "step": 4827 + }, + { + "epoch": 94.01, + "learning_rate": 1.9444444444444445e-05, + "loss": 0.0862, + "step": 4828 + }, + { + "epoch": 94.01, + "learning_rate": 1.9333333333333333e-05, + "loss": 0.0571, + "step": 4829 + }, + { + "epoch": 94.01, + "learning_rate": 1.9222222222222225e-05, + "loss": 0.0724, + "step": 4830 + }, + { + "epoch": 94.01, + "learning_rate": 1.911111111111111e-05, + "loss": 0.0784, + "step": 4831 + }, + { + "epoch": 94.01, + "learning_rate": 1.9e-05, + "loss": 0.072, + "step": 4832 + }, + { + "epoch": 94.01, + "learning_rate": 1.888888888888889e-05, + "loss": 0.0481, + "step": 4833 + }, + { + "epoch": 94.01, + "learning_rate": 1.877777777777778e-05, + "loss": 0.0713, + "step": 4834 + }, + { + "epoch": 94.01, + "learning_rate": 1.866666666666667e-05, + "loss": 0.1223, + "step": 4835 + }, + { + "epoch": 94.01, + "learning_rate": 1.8555555555555553e-05, + "loss": 0.0996, + "step": 4836 + }, + { + "epoch": 94.01, + "learning_rate": 1.8444444444444445e-05, + "loss": 0.0994, + "step": 4837 + }, + { + "epoch": 94.01, + "learning_rate": 1.8333333333333333e-05, + "loss": 0.059, + "step": 4838 + }, + { + "epoch": 94.01, + "learning_rate": 1.8222222222222224e-05, + "loss": 0.0734, + "step": 4839 + }, + { + "epoch": 94.01, + "learning_rate": 1.8111111111111112e-05, + "loss": 0.0655, + "step": 4840 + }, + { + "epoch": 94.01, + "learning_rate": 1.8e-05, + "loss": 0.0754, + "step": 4841 + }, + { + "epoch": 94.01, + "learning_rate": 1.788888888888889e-05, + "loss": 0.0935, + "step": 4842 + }, + { + "epoch": 94.01, + "learning_rate": 1.7777777777777777e-05, + "loss": 0.0779, + "step": 4843 + }, + { + "epoch": 94.01, + "learning_rate": 1.7666666666666668e-05, + "loss": 0.0856, + "step": 4844 + }, + { + "epoch": 94.01, + "learning_rate": 1.7555555555555556e-05, + "loss": 0.0698, + "step": 4845 + }, + { + "epoch": 95.0, + "learning_rate": 1.7444444444444444e-05, + "loss": 0.0907, + "step": 4846 + }, + { + "epoch": 95.0, + "learning_rate": 1.7333333333333332e-05, + "loss": 0.0912, + "step": 4847 + }, + { + "epoch": 95.0, + "learning_rate": 1.7222222222222224e-05, + "loss": 0.0879, + "step": 4848 + }, + { + "epoch": 95.0, + "learning_rate": 1.7111111111111112e-05, + "loss": 0.0586, + "step": 4849 + }, + { + "epoch": 95.0, + "learning_rate": 1.7000000000000003e-05, + "loss": 0.0851, + "step": 4850 + }, + { + "epoch": 95.0, + "learning_rate": 1.6888888888888888e-05, + "loss": 0.0525, + "step": 4851 + }, + { + "epoch": 95.0, + "learning_rate": 1.6777777777777776e-05, + "loss": 0.0782, + "step": 4852 + }, + { + "epoch": 95.0, + "learning_rate": 1.6666666666666667e-05, + "loss": 0.0673, + "step": 4853 + }, + { + "epoch": 95.0, + "learning_rate": 1.6555555555555556e-05, + "loss": 0.0403, + "step": 4854 + }, + { + "epoch": 95.0, + "learning_rate": 1.6444444444444447e-05, + "loss": 0.1016, + "step": 4855 + }, + { + "epoch": 95.0, + "learning_rate": 1.633333333333333e-05, + "loss": 0.0744, + "step": 4856 + }, + { + "epoch": 95.0, + "learning_rate": 1.6222222222222223e-05, + "loss": 0.0886, + "step": 4857 + }, + { + "epoch": 95.0, + "learning_rate": 1.611111111111111e-05, + "loss": 0.0741, + "step": 4858 + }, + { + "epoch": 95.0, + "learning_rate": 1.6e-05, + "loss": 0.0927, + "step": 4859 + }, + { + "epoch": 95.0, + "learning_rate": 1.588888888888889e-05, + "loss": 0.0804, + "step": 4860 + }, + { + "epoch": 95.0, + "learning_rate": 1.577777777777778e-05, + "loss": 0.0698, + "step": 4861 + }, + { + "epoch": 95.0, + "learning_rate": 1.5666666666666667e-05, + "loss": 0.0803, + "step": 4862 + }, + { + "epoch": 95.0, + "learning_rate": 1.5555555555555555e-05, + "loss": 0.0889, + "step": 4863 + }, + { + "epoch": 95.0, + "learning_rate": 1.5444444444444446e-05, + "loss": 0.0926, + "step": 4864 + }, + { + "epoch": 95.0, + "learning_rate": 1.5333333333333334e-05, + "loss": 0.0761, + "step": 4865 + }, + { + "epoch": 95.0, + "learning_rate": 1.5222222222222223e-05, + "loss": 0.0883, + "step": 4866 + }, + { + "epoch": 95.0, + "learning_rate": 1.5111111111111112e-05, + "loss": 0.0859, + "step": 4867 + }, + { + "epoch": 95.0, + "learning_rate": 1.5e-05, + "loss": 0.0884, + "step": 4868 + }, + { + "epoch": 95.0, + "learning_rate": 1.488888888888889e-05, + "loss": 0.0635, + "step": 4869 + }, + { + "epoch": 95.0, + "learning_rate": 1.4777777777777778e-05, + "loss": 0.0515, + "step": 4870 + }, + { + "epoch": 95.01, + "learning_rate": 1.4666666666666666e-05, + "loss": 0.0606, + "step": 4871 + }, + { + "epoch": 95.01, + "learning_rate": 1.4555555555555556e-05, + "loss": 0.0775, + "step": 4872 + }, + { + "epoch": 95.01, + "learning_rate": 1.4444444444444444e-05, + "loss": 0.1044, + "step": 4873 + }, + { + "epoch": 95.01, + "learning_rate": 1.4333333333333334e-05, + "loss": 0.1136, + "step": 4874 + }, + { + "epoch": 95.01, + "learning_rate": 1.4222222222222224e-05, + "loss": 0.0979, + "step": 4875 + }, + { + "epoch": 95.01, + "learning_rate": 1.4111111111111112e-05, + "loss": 0.0907, + "step": 4876 + }, + { + "epoch": 95.01, + "learning_rate": 1.4e-05, + "loss": 0.0716, + "step": 4877 + }, + { + "epoch": 95.01, + "learning_rate": 1.3888888888888888e-05, + "loss": 0.0664, + "step": 4878 + }, + { + "epoch": 95.01, + "learning_rate": 1.3777777777777778e-05, + "loss": 0.0702, + "step": 4879 + }, + { + "epoch": 95.01, + "learning_rate": 1.3666666666666667e-05, + "loss": 0.0838, + "step": 4880 + }, + { + "epoch": 95.01, + "learning_rate": 1.3555555555555555e-05, + "loss": 0.0844, + "step": 4881 + }, + { + "epoch": 95.01, + "learning_rate": 1.3444444444444445e-05, + "loss": 0.0613, + "step": 4882 + }, + { + "epoch": 95.01, + "learning_rate": 1.3333333333333335e-05, + "loss": 0.0425, + "step": 4883 + }, + { + "epoch": 95.01, + "learning_rate": 1.3222222222222223e-05, + "loss": 0.0859, + "step": 4884 + }, + { + "epoch": 95.01, + "learning_rate": 1.3111111111111111e-05, + "loss": 0.0838, + "step": 4885 + }, + { + "epoch": 95.01, + "learning_rate": 1.3e-05, + "loss": 0.1051, + "step": 4886 + }, + { + "epoch": 95.01, + "learning_rate": 1.2888888888888889e-05, + "loss": 0.084, + "step": 4887 + }, + { + "epoch": 95.01, + "learning_rate": 1.2777777777777779e-05, + "loss": 0.0907, + "step": 4888 + }, + { + "epoch": 95.01, + "learning_rate": 1.2666666666666667e-05, + "loss": 0.0731, + "step": 4889 + }, + { + "epoch": 95.01, + "learning_rate": 1.2555555555555557e-05, + "loss": 0.0746, + "step": 4890 + }, + { + "epoch": 95.01, + "learning_rate": 1.2444444444444445e-05, + "loss": 0.0921, + "step": 4891 + }, + { + "epoch": 95.01, + "learning_rate": 1.2333333333333334e-05, + "loss": 0.0861, + "step": 4892 + }, + { + "epoch": 95.01, + "learning_rate": 1.2222222222222222e-05, + "loss": 0.0681, + "step": 4893 + }, + { + "epoch": 95.01, + "learning_rate": 1.211111111111111e-05, + "loss": 0.1279, + "step": 4894 + }, + { + "epoch": 95.01, + "learning_rate": 1.211111111111111e-05, + "loss": 0.0939, + "step": 4895 + }, + { + "epoch": 95.01, + "learning_rate": 1.2e-05, + "loss": 0.0543, + "step": 4896 + }, + { + "epoch": 96.0, + "learning_rate": 1.1888888888888888e-05, + "loss": 0.1326, + "step": 4897 + }, + { + "epoch": 96.0, + "learning_rate": 1.1777777777777778e-05, + "loss": 0.0818, + "step": 4898 + }, + { + "epoch": 96.0, + "learning_rate": 1.1666666666666668e-05, + "loss": 0.052, + "step": 4899 + }, + { + "epoch": 96.0, + "learning_rate": 1.1555555555555556e-05, + "loss": 0.0535, + "step": 4900 + }, + { + "epoch": 96.0, + "learning_rate": 1.1444444444444444e-05, + "loss": 0.0664, + "step": 4901 + }, + { + "epoch": 96.0, + "learning_rate": 1.1333333333333334e-05, + "loss": 0.09, + "step": 4902 + }, + { + "epoch": 96.0, + "learning_rate": 1.1222222222222222e-05, + "loss": 0.0727, + "step": 4903 + }, + { + "epoch": 96.0, + "learning_rate": 1.1111111111111112e-05, + "loss": 0.0946, + "step": 4904 + }, + { + "epoch": 96.0, + "learning_rate": 1.1e-05, + "loss": 0.0536, + "step": 4905 + }, + { + "epoch": 96.0, + "learning_rate": 1.088888888888889e-05, + "loss": 0.0657, + "step": 4906 + }, + { + "epoch": 96.0, + "learning_rate": 1.077777777777778e-05, + "loss": 0.0838, + "step": 4907 + }, + { + "epoch": 96.0, + "learning_rate": 1.0666666666666667e-05, + "loss": 0.0637, + "step": 4908 + }, + { + "epoch": 96.0, + "learning_rate": 1.0555555555555555e-05, + "loss": 0.0878, + "step": 4909 + }, + { + "epoch": 96.0, + "learning_rate": 1.0444444444444443e-05, + "loss": 0.0635, + "step": 4910 + }, + { + "epoch": 96.0, + "learning_rate": 1.0333333333333333e-05, + "loss": 0.0812, + "step": 4911 + }, + { + "epoch": 96.0, + "learning_rate": 1.0222222222222223e-05, + "loss": 0.0813, + "step": 4912 + }, + { + "epoch": 96.0, + "learning_rate": 1.0111111111111111e-05, + "loss": 0.064, + "step": 4913 + }, + { + "epoch": 96.0, + "learning_rate": 1e-05, + "loss": 0.0469, + "step": 4914 + }, + { + "epoch": 96.0, + "learning_rate": 9.888888888888889e-06, + "loss": 0.085, + "step": 4915 + }, + { + "epoch": 96.0, + "learning_rate": 9.777777777777777e-06, + "loss": 0.0882, + "step": 4916 + }, + { + "epoch": 96.0, + "learning_rate": 9.666666666666667e-06, + "loss": 0.0929, + "step": 4917 + }, + { + "epoch": 96.0, + "learning_rate": 9.555555555555555e-06, + "loss": 0.0589, + "step": 4918 + }, + { + "epoch": 96.0, + "learning_rate": 9.444444444444445e-06, + "loss": 0.0759, + "step": 4919 + }, + { + "epoch": 96.0, + "learning_rate": 9.333333333333334e-06, + "loss": 0.0918, + "step": 4920 + }, + { + "epoch": 96.0, + "learning_rate": 9.222222222222222e-06, + "loss": 0.0964, + "step": 4921 + }, + { + "epoch": 96.01, + "learning_rate": 9.111111111111112e-06, + "loss": 0.0679, + "step": 4922 + }, + { + "epoch": 96.01, + "learning_rate": 9e-06, + "loss": 0.0702, + "step": 4923 + }, + { + "epoch": 96.01, + "learning_rate": 8.888888888888888e-06, + "loss": 0.0666, + "step": 4924 + }, + { + "epoch": 96.01, + "learning_rate": 8.777777777777778e-06, + "loss": 0.0868, + "step": 4925 + }, + { + "epoch": 96.01, + "learning_rate": 8.666666666666666e-06, + "loss": 0.1193, + "step": 4926 + }, + { + "epoch": 96.01, + "learning_rate": 8.555555555555556e-06, + "loss": 0.0721, + "step": 4927 + }, + { + "epoch": 96.01, + "learning_rate": 8.444444444444444e-06, + "loss": 0.0564, + "step": 4928 + }, + { + "epoch": 96.01, + "learning_rate": 8.333333333333334e-06, + "loss": 0.0711, + "step": 4929 + }, + { + "epoch": 96.01, + "learning_rate": 8.222222222222223e-06, + "loss": 0.0499, + "step": 4930 + }, + { + "epoch": 96.01, + "learning_rate": 8.111111111111112e-06, + "loss": 0.0635, + "step": 4931 + }, + { + "epoch": 96.01, + "learning_rate": 8e-06, + "loss": 0.0702, + "step": 4932 + }, + { + "epoch": 96.01, + "learning_rate": 7.88888888888889e-06, + "loss": 0.0701, + "step": 4933 + }, + { + "epoch": 96.01, + "learning_rate": 7.777777777777777e-06, + "loss": 0.0773, + "step": 4934 + }, + { + "epoch": 96.01, + "learning_rate": 7.666666666666667e-06, + "loss": 0.0773, + "step": 4935 + }, + { + "epoch": 96.01, + "learning_rate": 7.555555555555556e-06, + "loss": 0.0605, + "step": 4936 + }, + { + "epoch": 96.01, + "learning_rate": 7.444444444444445e-06, + "loss": 0.0834, + "step": 4937 + }, + { + "epoch": 96.01, + "learning_rate": 7.333333333333333e-06, + "loss": 0.0577, + "step": 4938 + }, + { + "epoch": 96.01, + "learning_rate": 7.222222222222222e-06, + "loss": 0.1, + "step": 4939 + }, + { + "epoch": 96.01, + "learning_rate": 7.111111111111112e-06, + "loss": 0.0703, + "step": 4940 + }, + { + "epoch": 96.01, + "learning_rate": 7e-06, + "loss": 0.0836, + "step": 4941 + }, + { + "epoch": 96.01, + "learning_rate": 6.888888888888889e-06, + "loss": 0.0735, + "step": 4942 + }, + { + "epoch": 96.01, + "learning_rate": 6.777777777777778e-06, + "loss": 0.0917, + "step": 4943 + }, + { + "epoch": 96.01, + "learning_rate": 6.6666666666666675e-06, + "loss": 0.1039, + "step": 4944 + }, + { + "epoch": 96.01, + "learning_rate": 6.5555555555555556e-06, + "loss": 0.0763, + "step": 4945 + }, + { + "epoch": 96.01, + "learning_rate": 6.4444444444444445e-06, + "loss": 0.0725, + "step": 4946 + }, + { + "epoch": 96.01, + "learning_rate": 6.333333333333333e-06, + "loss": 0.0663, + "step": 4947 + }, + { + "epoch": 97.0, + "learning_rate": 6.222222222222222e-06, + "loss": 0.1206, + "step": 4948 + }, + { + "epoch": 97.0, + "learning_rate": 6.111111111111111e-06, + "loss": 0.0611, + "step": 4949 + }, + { + "epoch": 97.0, + "learning_rate": 6e-06, + "loss": 0.079, + "step": 4950 + }, + { + "epoch": 97.0, + "learning_rate": 5.888888888888889e-06, + "loss": 0.0601, + "step": 4951 + }, + { + "epoch": 97.0, + "learning_rate": 5.777777777777778e-06, + "loss": 0.0936, + "step": 4952 + }, + { + "epoch": 97.0, + "learning_rate": 5.666666666666667e-06, + "loss": 0.0737, + "step": 4953 + }, + { + "epoch": 97.0, + "learning_rate": 5.555555555555556e-06, + "loss": 0.0582, + "step": 4954 + }, + { + "epoch": 97.0, + "learning_rate": 5.444444444444445e-06, + "loss": 0.0508, + "step": 4955 + }, + { + "epoch": 97.0, + "learning_rate": 5.333333333333334e-06, + "loss": 0.0724, + "step": 4956 + }, + { + "epoch": 97.0, + "learning_rate": 5.222222222222222e-06, + "loss": 0.0764, + "step": 4957 + }, + { + "epoch": 97.0, + "learning_rate": 5.1111111111111115e-06, + "loss": 0.0472, + "step": 4958 + }, + { + "epoch": 97.0, + "learning_rate": 5e-06, + "loss": 0.0734, + "step": 4959 + }, + { + "epoch": 97.0, + "learning_rate": 4.8888888888888885e-06, + "loss": 0.0667, + "step": 4960 + }, + { + "epoch": 97.0, + "learning_rate": 4.777777777777777e-06, + "loss": 0.0627, + "step": 4961 + }, + { + "epoch": 97.0, + "learning_rate": 4.666666666666667e-06, + "loss": 0.0819, + "step": 4962 + }, + { + "epoch": 97.0, + "learning_rate": 4.555555555555556e-06, + "loss": 0.0514, + "step": 4963 + }, + { + "epoch": 97.0, + "learning_rate": 4.444444444444444e-06, + "loss": 0.0732, + "step": 4964 + }, + { + "epoch": 97.0, + "learning_rate": 4.333333333333333e-06, + "loss": 0.0561, + "step": 4965 + }, + { + "epoch": 97.0, + "learning_rate": 4.222222222222222e-06, + "loss": 0.0491, + "step": 4966 + }, + { + "epoch": 97.0, + "learning_rate": 4.111111111111112e-06, + "loss": 0.0933, + "step": 4967 + }, + { + "epoch": 97.0, + "learning_rate": 4e-06, + "loss": 0.0531, + "step": 4968 + }, + { + "epoch": 97.0, + "learning_rate": 3.888888888888889e-06, + "loss": 0.0875, + "step": 4969 + }, + { + "epoch": 97.0, + "learning_rate": 3.777777777777778e-06, + "loss": 0.0738, + "step": 4970 + }, + { + "epoch": 97.0, + "learning_rate": 3.6666666666666666e-06, + "loss": 0.0887, + "step": 4971 + }, + { + "epoch": 97.0, + "learning_rate": 3.555555555555556e-06, + "loss": 0.0865, + "step": 4972 + }, + { + "epoch": 97.01, + "learning_rate": 3.4444444444444444e-06, + "loss": 0.0936, + "step": 4973 + }, + { + "epoch": 97.01, + "learning_rate": 3.3333333333333337e-06, + "loss": 0.0828, + "step": 4974 + }, + { + "epoch": 97.01, + "learning_rate": 3.2222222222222222e-06, + "loss": 0.104, + "step": 4975 + }, + { + "epoch": 97.01, + "learning_rate": 3.111111111111111e-06, + "loss": 0.0846, + "step": 4976 + }, + { + "epoch": 97.01, + "learning_rate": 3e-06, + "loss": 0.0609, + "step": 4977 + }, + { + "epoch": 97.01, + "learning_rate": 2.888888888888889e-06, + "loss": 0.0681, + "step": 4978 + }, + { + "epoch": 97.01, + "learning_rate": 2.777777777777778e-06, + "loss": 0.0985, + "step": 4979 + }, + { + "epoch": 97.01, + "learning_rate": 2.666666666666667e-06, + "loss": 0.0816, + "step": 4980 + }, + { + "epoch": 97.01, + "learning_rate": 2.5555555555555557e-06, + "loss": 0.0771, + "step": 4981 + }, + { + "epoch": 97.01, + "learning_rate": 2.4444444444444442e-06, + "loss": 0.094, + "step": 4982 + }, + { + "epoch": 97.01, + "learning_rate": 2.3333333333333336e-06, + "loss": 0.0658, + "step": 4983 + }, + { + "epoch": 97.01, + "learning_rate": 2.222222222222222e-06, + "loss": 0.0832, + "step": 4984 + }, + { + "epoch": 97.01, + "learning_rate": 2.111111111111111e-06, + "loss": 0.0691, + "step": 4985 + }, + { + "epoch": 97.01, + "learning_rate": 2e-06, + "loss": 0.0687, + "step": 4986 + }, + { + "epoch": 97.01, + "learning_rate": 1.888888888888889e-06, + "loss": 0.0906, + "step": 4987 + }, + { + "epoch": 97.01, + "learning_rate": 1.777777777777778e-06, + "loss": 0.0708, + "step": 4988 + }, + { + "epoch": 97.01, + "learning_rate": 1.6666666666666669e-06, + "loss": 0.0714, + "step": 4989 + }, + { + "epoch": 97.01, + "learning_rate": 1.5555555555555556e-06, + "loss": 0.089, + "step": 4990 + }, + { + "epoch": 97.01, + "learning_rate": 1.4444444444444445e-06, + "loss": 0.095, + "step": 4991 + }, + { + "epoch": 97.01, + "learning_rate": 1.3333333333333334e-06, + "loss": 0.075, + "step": 4992 + }, + { + "epoch": 97.01, + "learning_rate": 1.2222222222222221e-06, + "loss": 0.0973, + "step": 4993 + }, + { + "epoch": 97.01, + "learning_rate": 1.111111111111111e-06, + "loss": 0.1056, + "step": 4994 + }, + { + "epoch": 97.01, + "learning_rate": 1e-06, + "loss": 0.0672, + "step": 4995 + }, + { + "epoch": 97.01, + "learning_rate": 8.88888888888889e-07, + "loss": 0.0633, + "step": 4996 + }, + { + "epoch": 97.01, + "learning_rate": 7.777777777777778e-07, + "loss": 0.0677, + "step": 4997 + }, + { + "epoch": 97.01, + "learning_rate": 6.666666666666667e-07, + "loss": 0.08, + "step": 4998 + }, + { + "epoch": 98.0, + "learning_rate": 5.555555555555555e-07, + "loss": 0.1245, + "step": 4999 + }, + { + "epoch": 98.0, + "learning_rate": 4.444444444444445e-07, + "loss": 0.095, + "step": 5000 + }, + { + "epoch": 98.0, + "eval_cer": 0.06942781900885803, + "eval_loss": 0.35193362832069397, + "eval_runtime": 51.7004, + "eval_samples_per_second": 8.046, + "eval_steps_per_second": 0.251, + "eval_wer": 0.2927228127555192, + "step": 5000 + }, + { + "epoch": 98.0, + "step": 5000, + "total_flos": 5.3874161547319706e+19, + "train_loss": 0.48236389900147914, + "train_runtime": 12714.8682, + "train_samples_per_second": 25.167, + "train_steps_per_second": 0.393 + } + ], + "max_steps": 5000, + "num_train_epochs": 9223372036854775807, + "total_flos": 5.3874161547319706e+19, + "trial_name": null, + "trial_params": null +}