{ "best_metric": null, "best_model_checkpoint": null, "epoch": 5.0, "global_step": 4460, "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": 23.989, "step": 1 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 23.6178, "step": 2 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 24.0653, "step": 3 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 23.7042, "step": 4 }, { "epoch": 0.01, "learning_rate": 0.0, "loss": 23.3349, "step": 5 }, { "epoch": 0.01, "learning_rate": 0.0, "loss": 24.5588, "step": 6 }, { "epoch": 0.01, "learning_rate": 6.000000000000001e-08, "loss": 23.0751, "step": 7 }, { "epoch": 0.01, "learning_rate": 1.2000000000000002e-07, "loss": 23.8228, "step": 8 }, { "epoch": 0.01, "learning_rate": 1.8e-07, "loss": 24.0497, "step": 9 }, { "epoch": 0.01, "learning_rate": 2.4000000000000003e-07, "loss": 23.2444, "step": 10 }, { "epoch": 0.01, "learning_rate": 3.0000000000000004e-07, "loss": 24.0266, "step": 11 }, { "epoch": 0.01, "learning_rate": 3.6e-07, "loss": 23.3999, "step": 12 }, { "epoch": 0.01, "learning_rate": 4.2e-07, "loss": 23.4658, "step": 13 }, { "epoch": 0.02, "learning_rate": 4.800000000000001e-07, "loss": 23.2779, "step": 14 }, { "epoch": 0.02, "learning_rate": 5.4e-07, "loss": 23.3397, "step": 15 }, { "epoch": 0.02, "learning_rate": 6.000000000000001e-07, "loss": 23.0846, "step": 16 }, { "epoch": 0.02, "learning_rate": 6.599999999999999e-07, "loss": 22.6688, "step": 17 }, { "epoch": 0.02, "learning_rate": 7.2e-07, "loss": 22.7822, "step": 18 }, { "epoch": 0.02, "learning_rate": 7.799999999999999e-07, "loss": 21.1833, "step": 19 }, { "epoch": 0.02, "learning_rate": 8.4e-07, "loss": 20.6991, "step": 20 }, { "epoch": 0.02, "learning_rate": 9e-07, "loss": 20.3698, "step": 21 }, { "epoch": 0.02, "learning_rate": 9.600000000000001e-07, "loss": 19.7831, "step": 22 }, { "epoch": 0.03, "learning_rate": 1.0200000000000002e-06, "loss": 19.6545, "step": 23 }, { "epoch": 0.03, "learning_rate": 1.08e-06, "loss": 15.7066, "step": 24 }, { "epoch": 0.03, "learning_rate": 1.14e-06, "loss": 14.7721, "step": 25 }, { "epoch": 0.03, "learning_rate": 1.2000000000000002e-06, "loss": 14.3433, "step": 26 }, { "epoch": 0.03, "learning_rate": 1.26e-06, "loss": 14.6628, "step": 27 }, { "epoch": 0.03, "learning_rate": 1.3199999999999999e-06, "loss": 13.749, "step": 28 }, { "epoch": 0.03, "learning_rate": 1.38e-06, "loss": 13.1664, "step": 29 }, { "epoch": 0.03, "learning_rate": 1.44e-06, "loss": 12.8713, "step": 30 }, { "epoch": 0.03, "learning_rate": 1.5e-06, "loss": 12.7932, "step": 31 }, { "epoch": 0.04, "learning_rate": 1.5599999999999999e-06, "loss": 12.4594, "step": 32 }, { "epoch": 0.04, "learning_rate": 1.62e-06, "loss": 12.4, "step": 33 }, { "epoch": 0.04, "learning_rate": 1.68e-06, "loss": 12.1008, "step": 34 }, { "epoch": 0.04, "learning_rate": 1.74e-06, "loss": 12.3833, "step": 35 }, { "epoch": 0.04, "learning_rate": 1.8e-06, "loss": 12.252, "step": 36 }, { "epoch": 0.04, "learning_rate": 1.86e-06, "loss": 11.9621, "step": 37 }, { "epoch": 0.04, "learning_rate": 1.9200000000000003e-06, "loss": 11.736, "step": 38 }, { "epoch": 0.04, "learning_rate": 1.98e-06, "loss": 11.9301, "step": 39 }, { "epoch": 0.04, "learning_rate": 2.0400000000000004e-06, "loss": 11.658, "step": 40 }, { "epoch": 0.05, "learning_rate": 2.1000000000000002e-06, "loss": 11.267, "step": 41 }, { "epoch": 0.05, "learning_rate": 2.16e-06, "loss": 11.4205, "step": 42 }, { "epoch": 0.05, "learning_rate": 2.22e-06, "loss": 11.1984, "step": 43 }, { "epoch": 0.05, "learning_rate": 2.28e-06, "loss": 11.3742, "step": 44 }, { "epoch": 0.05, "learning_rate": 2.34e-06, "loss": 11.2985, "step": 45 }, { "epoch": 0.05, "learning_rate": 2.4000000000000003e-06, "loss": 11.0919, "step": 46 }, { "epoch": 0.05, "learning_rate": 2.46e-06, "loss": 10.6774, "step": 47 }, { "epoch": 0.05, "learning_rate": 2.52e-06, "loss": 10.8074, "step": 48 }, { "epoch": 0.05, "learning_rate": 2.58e-06, "loss": 10.8503, "step": 49 }, { "epoch": 0.06, "learning_rate": 2.6399999999999997e-06, "loss": 10.7582, "step": 50 }, { "epoch": 0.06, "learning_rate": 2.7e-06, "loss": 10.2194, "step": 51 }, { "epoch": 0.06, "learning_rate": 2.76e-06, "loss": 10.1342, "step": 52 }, { "epoch": 0.06, "learning_rate": 2.82e-06, "loss": 10.0029, "step": 53 }, { "epoch": 0.06, "learning_rate": 2.88e-06, "loss": 9.983, "step": 54 }, { "epoch": 0.06, "learning_rate": 2.9400000000000002e-06, "loss": 9.7462, "step": 55 }, { "epoch": 0.06, "learning_rate": 3e-06, "loss": 9.9687, "step": 56 }, { "epoch": 0.06, "learning_rate": 3.06e-06, "loss": 9.7789, "step": 57 }, { "epoch": 0.07, "learning_rate": 3.1199999999999998e-06, "loss": 9.6711, "step": 58 }, { "epoch": 0.07, "learning_rate": 3.18e-06, "loss": 9.2986, "step": 59 }, { "epoch": 0.07, "learning_rate": 3.24e-06, "loss": 9.3476, "step": 60 }, { "epoch": 0.07, "learning_rate": 3.3e-06, "loss": 9.2774, "step": 61 }, { "epoch": 0.07, "learning_rate": 3.36e-06, "loss": 9.1055, "step": 62 }, { "epoch": 0.07, "learning_rate": 3.4200000000000003e-06, "loss": 9.0546, "step": 63 }, { "epoch": 0.07, "learning_rate": 3.48e-06, "loss": 9.0394, "step": 64 }, { "epoch": 0.07, "learning_rate": 3.54e-06, "loss": 8.6973, "step": 65 }, { "epoch": 0.07, "learning_rate": 3.6e-06, "loss": 8.8709, "step": 66 }, { "epoch": 0.08, "learning_rate": 3.66e-06, "loss": 8.7381, "step": 67 }, { "epoch": 0.08, "learning_rate": 3.72e-06, "loss": 8.5203, "step": 68 }, { "epoch": 0.08, "learning_rate": 3.7800000000000002e-06, "loss": 8.5401, "step": 69 }, { "epoch": 0.08, "learning_rate": 3.8400000000000005e-06, "loss": 8.3454, "step": 70 }, { "epoch": 0.08, "learning_rate": 3.9e-06, "loss": 8.5086, "step": 71 }, { "epoch": 0.08, "learning_rate": 3.96e-06, "loss": 8.3971, "step": 72 }, { "epoch": 0.08, "learning_rate": 4.0200000000000005e-06, "loss": 8.1215, "step": 73 }, { "epoch": 0.08, "learning_rate": 4.080000000000001e-06, "loss": 8.3456, "step": 74 }, { "epoch": 0.08, "learning_rate": 4.14e-06, "loss": 8.0819, "step": 75 }, { "epoch": 0.09, "learning_rate": 4.2000000000000004e-06, "loss": 8.1952, "step": 76 }, { "epoch": 0.09, "learning_rate": 4.26e-06, "loss": 8.0351, "step": 77 }, { "epoch": 0.09, "learning_rate": 4.32e-06, "loss": 7.9785, "step": 78 }, { "epoch": 0.09, "learning_rate": 4.3799999999999996e-06, "loss": 7.8219, "step": 79 }, { "epoch": 0.09, "learning_rate": 4.44e-06, "loss": 7.8892, "step": 80 }, { "epoch": 0.09, "learning_rate": 4.5e-06, "loss": 7.8939, "step": 81 }, { "epoch": 0.09, "learning_rate": 4.56e-06, "loss": 7.7575, "step": 82 }, { "epoch": 0.09, "learning_rate": 4.62e-06, "loss": 7.8758, "step": 83 }, { "epoch": 0.09, "learning_rate": 4.68e-06, "loss": 7.6336, "step": 84 }, { "epoch": 0.1, "learning_rate": 4.74e-06, "loss": 7.6044, "step": 85 }, { "epoch": 0.1, "learning_rate": 4.800000000000001e-06, "loss": 7.5865, "step": 86 }, { "epoch": 0.1, "learning_rate": 4.86e-06, "loss": 7.4691, "step": 87 }, { "epoch": 0.1, "learning_rate": 4.92e-06, "loss": 7.6055, "step": 88 }, { "epoch": 0.1, "learning_rate": 4.980000000000001e-06, "loss": 7.6453, "step": 89 }, { "epoch": 0.1, "learning_rate": 5.04e-06, "loss": 7.6395, "step": 90 }, { "epoch": 0.1, "learning_rate": 5.1e-06, "loss": 7.729, "step": 91 }, { "epoch": 0.1, "learning_rate": 5.16e-06, "loss": 7.6319, "step": 92 }, { "epoch": 0.1, "learning_rate": 5.22e-06, "loss": 7.4062, "step": 93 }, { "epoch": 0.11, "learning_rate": 5.279999999999999e-06, "loss": 7.4348, "step": 94 }, { "epoch": 0.11, "learning_rate": 5.34e-06, "loss": 7.6356, "step": 95 }, { "epoch": 0.11, "learning_rate": 5.4e-06, "loss": 7.6285, "step": 96 }, { "epoch": 0.11, "learning_rate": 5.46e-06, "loss": 7.4937, "step": 97 }, { "epoch": 0.11, "learning_rate": 5.52e-06, "loss": 7.5724, "step": 98 }, { "epoch": 0.11, "learning_rate": 5.58e-06, "loss": 7.3202, "step": 99 }, { "epoch": 0.11, "learning_rate": 5.64e-06, "loss": 7.1796, "step": 100 }, { "epoch": 0.11, "learning_rate": 5.7000000000000005e-06, "loss": 7.5936, "step": 101 }, { "epoch": 0.11, "learning_rate": 5.76e-06, "loss": 7.5819, "step": 102 }, { "epoch": 0.12, "learning_rate": 5.82e-06, "loss": 7.502, "step": 103 }, { "epoch": 0.12, "learning_rate": 5.8800000000000005e-06, "loss": 7.4363, "step": 104 }, { "epoch": 0.12, "learning_rate": 5.940000000000001e-06, "loss": 7.3253, "step": 105 }, { "epoch": 0.12, "learning_rate": 6e-06, "loss": 7.3953, "step": 106 }, { "epoch": 0.12, "learning_rate": 6.0600000000000004e-06, "loss": 7.4679, "step": 107 }, { "epoch": 0.12, "learning_rate": 6.12e-06, "loss": 7.4803, "step": 108 }, { "epoch": 0.12, "learning_rate": 6.18e-06, "loss": 7.1671, "step": 109 }, { "epoch": 0.12, "learning_rate": 6.2399999999999995e-06, "loss": 7.2751, "step": 110 }, { "epoch": 0.12, "learning_rate": 6.3e-06, "loss": 7.3579, "step": 111 }, { "epoch": 0.13, "learning_rate": 6.36e-06, "loss": 7.4779, "step": 112 }, { "epoch": 0.13, "learning_rate": 6.42e-06, "loss": 7.3051, "step": 113 }, { "epoch": 0.13, "learning_rate": 6.48e-06, "loss": 7.2543, "step": 114 }, { "epoch": 0.13, "learning_rate": 6.54e-06, "loss": 7.3323, "step": 115 }, { "epoch": 0.13, "learning_rate": 6.6e-06, "loss": 7.1692, "step": 116 }, { "epoch": 0.13, "learning_rate": 6.660000000000001e-06, "loss": 7.2783, "step": 117 }, { "epoch": 0.13, "learning_rate": 6.72e-06, "loss": 7.1964, "step": 118 }, { "epoch": 0.13, "learning_rate": 6.78e-06, "loss": 7.2585, "step": 119 }, { "epoch": 0.13, "learning_rate": 6.840000000000001e-06, "loss": 7.2869, "step": 120 }, { "epoch": 0.14, "learning_rate": 6.900000000000001e-06, "loss": 7.1767, "step": 121 }, { "epoch": 0.14, "learning_rate": 6.96e-06, "loss": 7.3241, "step": 122 }, { "epoch": 0.14, "learning_rate": 7.0200000000000006e-06, "loss": 7.1308, "step": 123 }, { "epoch": 0.14, "learning_rate": 7.08e-06, "loss": 7.2816, "step": 124 }, { "epoch": 0.14, "learning_rate": 7.14e-06, "loss": 7.3685, "step": 125 }, { "epoch": 0.14, "learning_rate": 7.2e-06, "loss": 7.4656, "step": 126 }, { "epoch": 0.14, "learning_rate": 7.26e-06, "loss": 7.4408, "step": 127 }, { "epoch": 0.14, "learning_rate": 7.32e-06, "loss": 7.3301, "step": 128 }, { "epoch": 0.14, "learning_rate": 7.3800000000000005e-06, "loss": 7.1445, "step": 129 }, { "epoch": 0.15, "learning_rate": 7.44e-06, "loss": 7.2517, "step": 130 }, { "epoch": 0.15, "learning_rate": 7.5e-06, "loss": 7.2445, "step": 131 }, { "epoch": 0.15, "learning_rate": 7.5600000000000005e-06, "loss": 7.3213, "step": 132 }, { "epoch": 0.15, "learning_rate": 7.62e-06, "loss": 7.2867, "step": 133 }, { "epoch": 0.15, "learning_rate": 7.680000000000001e-06, "loss": 7.3281, "step": 134 }, { "epoch": 0.15, "learning_rate": 7.74e-06, "loss": 7.1007, "step": 135 }, { "epoch": 0.15, "learning_rate": 7.8e-06, "loss": 7.1761, "step": 136 }, { "epoch": 0.15, "learning_rate": 7.860000000000001e-06, "loss": 7.458, "step": 137 }, { "epoch": 0.15, "learning_rate": 7.92e-06, "loss": 7.1824, "step": 138 }, { "epoch": 0.16, "learning_rate": 7.98e-06, "loss": 7.179, "step": 139 }, { "epoch": 0.16, "learning_rate": 8.040000000000001e-06, "loss": 7.1635, "step": 140 }, { "epoch": 0.16, "learning_rate": 8.1e-06, "loss": 7.3679, "step": 141 }, { "epoch": 0.16, "learning_rate": 8.160000000000001e-06, "loss": 7.3065, "step": 142 }, { "epoch": 0.16, "learning_rate": 8.220000000000001e-06, "loss": 7.241, "step": 143 }, { "epoch": 0.16, "learning_rate": 8.28e-06, "loss": 7.0796, "step": 144 }, { "epoch": 0.16, "learning_rate": 8.340000000000001e-06, "loss": 7.1336, "step": 145 }, { "epoch": 0.16, "learning_rate": 8.400000000000001e-06, "loss": 7.1218, "step": 146 }, { "epoch": 0.16, "learning_rate": 8.459999999999999e-06, "loss": 7.4148, "step": 147 }, { "epoch": 0.17, "learning_rate": 8.52e-06, "loss": 6.8795, "step": 148 }, { "epoch": 0.17, "learning_rate": 8.58e-06, "loss": 7.2323, "step": 149 }, { "epoch": 0.17, "learning_rate": 8.64e-06, "loss": 6.7981, "step": 150 }, { "epoch": 0.17, "learning_rate": 8.7e-06, "loss": 7.3171, "step": 151 }, { "epoch": 0.17, "learning_rate": 8.759999999999999e-06, "loss": 7.2228, "step": 152 }, { "epoch": 0.17, "learning_rate": 8.82e-06, "loss": 7.2525, "step": 153 }, { "epoch": 0.17, "learning_rate": 8.88e-06, "loss": 7.1167, "step": 154 }, { "epoch": 0.17, "learning_rate": 8.939999999999999e-06, "loss": 7.105, "step": 155 }, { "epoch": 0.17, "learning_rate": 9e-06, "loss": 7.1635, "step": 156 }, { "epoch": 0.18, "learning_rate": 9.06e-06, "loss": 7.3667, "step": 157 }, { "epoch": 0.18, "learning_rate": 9.12e-06, "loss": 7.3148, "step": 158 }, { "epoch": 0.18, "learning_rate": 9.18e-06, "loss": 7.2948, "step": 159 }, { "epoch": 0.18, "learning_rate": 9.24e-06, "loss": 7.2752, "step": 160 }, { "epoch": 0.18, "learning_rate": 9.3e-06, "loss": 7.0993, "step": 161 }, { "epoch": 0.18, "learning_rate": 9.36e-06, "loss": 7.13, "step": 162 }, { "epoch": 0.18, "learning_rate": 9.42e-06, "loss": 7.1973, "step": 163 }, { "epoch": 0.18, "learning_rate": 9.48e-06, "loss": 7.1612, "step": 164 }, { "epoch": 0.18, "learning_rate": 9.54e-06, "loss": 7.1636, "step": 165 }, { "epoch": 0.19, "learning_rate": 9.600000000000001e-06, "loss": 7.1414, "step": 166 }, { "epoch": 0.19, "learning_rate": 9.66e-06, "loss": 7.1084, "step": 167 }, { "epoch": 0.19, "learning_rate": 9.72e-06, "loss": 7.195, "step": 168 }, { "epoch": 0.19, "learning_rate": 9.780000000000001e-06, "loss": 7.2064, "step": 169 }, { "epoch": 0.19, "learning_rate": 9.84e-06, "loss": 7.1028, "step": 170 }, { "epoch": 0.19, "learning_rate": 9.9e-06, "loss": 7.2801, "step": 171 }, { "epoch": 0.19, "learning_rate": 9.960000000000001e-06, "loss": 7.3017, "step": 172 }, { "epoch": 0.19, "learning_rate": 1.002e-05, "loss": 7.3084, "step": 173 }, { "epoch": 0.2, "learning_rate": 1.008e-05, "loss": 7.1552, "step": 174 }, { "epoch": 0.2, "learning_rate": 1.0140000000000001e-05, "loss": 7.0637, "step": 175 }, { "epoch": 0.2, "learning_rate": 1.02e-05, "loss": 7.1764, "step": 176 }, { "epoch": 0.2, "learning_rate": 1.0260000000000002e-05, "loss": 7.1514, "step": 177 }, { "epoch": 0.2, "learning_rate": 1.032e-05, "loss": 7.1016, "step": 178 }, { "epoch": 0.2, "learning_rate": 1.0379999999999999e-05, "loss": 7.2535, "step": 179 }, { "epoch": 0.2, "learning_rate": 1.044e-05, "loss": 7.2312, "step": 180 }, { "epoch": 0.2, "learning_rate": 1.05e-05, "loss": 7.233, "step": 181 }, { "epoch": 0.2, "learning_rate": 1.0559999999999999e-05, "loss": 7.0902, "step": 182 }, { "epoch": 0.21, "learning_rate": 1.062e-05, "loss": 7.1947, "step": 183 }, { "epoch": 0.21, "learning_rate": 1.068e-05, "loss": 7.2737, "step": 184 }, { "epoch": 0.21, "learning_rate": 1.074e-05, "loss": 7.2071, "step": 185 }, { "epoch": 0.21, "learning_rate": 1.08e-05, "loss": 7.3035, "step": 186 }, { "epoch": 0.21, "learning_rate": 1.086e-05, "loss": 7.2594, "step": 187 }, { "epoch": 0.21, "learning_rate": 1.092e-05, "loss": 7.0527, "step": 188 }, { "epoch": 0.21, "learning_rate": 1.098e-05, "loss": 7.0849, "step": 189 }, { "epoch": 0.21, "learning_rate": 1.104e-05, "loss": 7.1792, "step": 190 }, { "epoch": 0.21, "learning_rate": 1.11e-05, "loss": 7.1796, "step": 191 }, { "epoch": 0.22, "learning_rate": 1.116e-05, "loss": 7.2825, "step": 192 }, { "epoch": 0.22, "learning_rate": 1.1220000000000001e-05, "loss": 7.2917, "step": 193 }, { "epoch": 0.22, "learning_rate": 1.128e-05, "loss": 7.1955, "step": 194 }, { "epoch": 0.22, "learning_rate": 1.134e-05, "loss": 6.9729, "step": 195 }, { "epoch": 0.22, "learning_rate": 1.1400000000000001e-05, "loss": 7.2888, "step": 196 }, { "epoch": 0.22, "learning_rate": 1.146e-05, "loss": 7.0468, "step": 197 }, { "epoch": 0.22, "learning_rate": 1.152e-05, "loss": 7.0792, "step": 198 }, { "epoch": 0.22, "learning_rate": 1.1580000000000001e-05, "loss": 6.932, "step": 199 }, { "epoch": 0.22, "learning_rate": 1.164e-05, "loss": 6.8468, "step": 200 }, { "epoch": 0.23, "learning_rate": 1.1700000000000001e-05, "loss": 7.292, "step": 201 }, { "epoch": 0.23, "learning_rate": 1.1760000000000001e-05, "loss": 7.342, "step": 202 }, { "epoch": 0.23, "learning_rate": 1.182e-05, "loss": 7.1952, "step": 203 }, { "epoch": 0.23, "learning_rate": 1.1880000000000001e-05, "loss": 7.2021, "step": 204 }, { "epoch": 0.23, "learning_rate": 1.1940000000000001e-05, "loss": 7.1782, "step": 205 }, { "epoch": 0.23, "learning_rate": 1.2e-05, "loss": 7.1298, "step": 206 }, { "epoch": 0.23, "learning_rate": 1.2060000000000001e-05, "loss": 7.1787, "step": 207 }, { "epoch": 0.23, "learning_rate": 1.2120000000000001e-05, "loss": 7.0922, "step": 208 }, { "epoch": 0.23, "learning_rate": 1.2180000000000002e-05, "loss": 7.1139, "step": 209 }, { "epoch": 0.24, "learning_rate": 1.224e-05, "loss": 7.1447, "step": 210 }, { "epoch": 0.24, "learning_rate": 1.2299999999999999e-05, "loss": 6.9867, "step": 211 }, { "epoch": 0.24, "learning_rate": 1.236e-05, "loss": 7.0632, "step": 212 }, { "epoch": 0.24, "learning_rate": 1.242e-05, "loss": 7.1346, "step": 213 }, { "epoch": 0.24, "learning_rate": 1.2479999999999999e-05, "loss": 7.1375, "step": 214 }, { "epoch": 0.24, "learning_rate": 1.254e-05, "loss": 7.1059, "step": 215 }, { "epoch": 0.24, "learning_rate": 1.26e-05, "loss": 7.2099, "step": 216 }, { "epoch": 0.24, "learning_rate": 1.2659999999999999e-05, "loss": 7.1114, "step": 217 }, { "epoch": 0.24, "learning_rate": 1.272e-05, "loss": 7.1167, "step": 218 }, { "epoch": 0.25, "learning_rate": 1.278e-05, "loss": 7.0278, "step": 219 }, { "epoch": 0.25, "learning_rate": 1.284e-05, "loss": 7.1962, "step": 220 }, { "epoch": 0.25, "learning_rate": 1.29e-05, "loss": 7.1355, "step": 221 }, { "epoch": 0.25, "learning_rate": 1.296e-05, "loss": 7.0232, "step": 222 }, { "epoch": 0.25, "learning_rate": 1.302e-05, "loss": 7.2442, "step": 223 }, { "epoch": 0.25, "learning_rate": 1.308e-05, "loss": 7.1077, "step": 224 }, { "epoch": 0.25, "learning_rate": 1.314e-05, "loss": 7.0455, "step": 225 }, { "epoch": 0.25, "learning_rate": 1.32e-05, "loss": 7.2867, "step": 226 }, { "epoch": 0.25, "learning_rate": 1.326e-05, "loss": 7.0915, "step": 227 }, { "epoch": 0.26, "learning_rate": 1.3320000000000001e-05, "loss": 7.0863, "step": 228 }, { "epoch": 0.26, "learning_rate": 1.338e-05, "loss": 7.1873, "step": 229 }, { "epoch": 0.26, "learning_rate": 1.344e-05, "loss": 7.1986, "step": 230 }, { "epoch": 0.26, "learning_rate": 1.3500000000000001e-05, "loss": 7.038, "step": 231 }, { "epoch": 0.26, "learning_rate": 1.356e-05, "loss": 7.2411, "step": 232 }, { "epoch": 0.26, "learning_rate": 1.362e-05, "loss": 7.1234, "step": 233 }, { "epoch": 0.26, "learning_rate": 1.3680000000000001e-05, "loss": 7.1433, "step": 234 }, { "epoch": 0.26, "learning_rate": 1.374e-05, "loss": 7.1708, "step": 235 }, { "epoch": 0.26, "learning_rate": 1.3800000000000002e-05, "loss": 6.9796, "step": 236 }, { "epoch": 0.27, "learning_rate": 1.3860000000000001e-05, "loss": 7.0731, "step": 237 }, { "epoch": 0.27, "learning_rate": 1.392e-05, "loss": 7.1961, "step": 238 }, { "epoch": 0.27, "learning_rate": 1.3980000000000002e-05, "loss": 6.9977, "step": 239 }, { "epoch": 0.27, "learning_rate": 1.4040000000000001e-05, "loss": 7.037, "step": 240 }, { "epoch": 0.27, "learning_rate": 1.4099999999999999e-05, "loss": 7.075, "step": 241 }, { "epoch": 0.27, "learning_rate": 1.416e-05, "loss": 7.1406, "step": 242 }, { "epoch": 0.27, "learning_rate": 1.422e-05, "loss": 6.9958, "step": 243 }, { "epoch": 0.27, "learning_rate": 1.428e-05, "loss": 6.9093, "step": 244 }, { "epoch": 0.27, "learning_rate": 1.434e-05, "loss": 7.0448, "step": 245 }, { "epoch": 0.28, "learning_rate": 1.44e-05, "loss": 6.9749, "step": 246 }, { "epoch": 0.28, "learning_rate": 1.446e-05, "loss": 6.9813, "step": 247 }, { "epoch": 0.28, "learning_rate": 1.452e-05, "loss": 7.0664, "step": 248 }, { "epoch": 0.28, "learning_rate": 1.458e-05, "loss": 6.7864, "step": 249 }, { "epoch": 0.28, "learning_rate": 1.464e-05, "loss": 6.4862, "step": 250 }, { "epoch": 0.28, "learning_rate": 1.47e-05, "loss": 7.704, "step": 251 }, { "epoch": 0.28, "learning_rate": 1.4760000000000001e-05, "loss": 7.6232, "step": 252 }, { "epoch": 0.28, "learning_rate": 1.482e-05, "loss": 7.3528, "step": 253 }, { "epoch": 0.28, "learning_rate": 1.488e-05, "loss": 7.3366, "step": 254 }, { "epoch": 0.29, "learning_rate": 1.4940000000000001e-05, "loss": 7.129, "step": 255 }, { "epoch": 0.29, "learning_rate": 1.5e-05, "loss": 7.2649, "step": 256 }, { "epoch": 0.29, "learning_rate": 1.506e-05, "loss": 7.1178, "step": 257 }, { "epoch": 0.29, "learning_rate": 1.5120000000000001e-05, "loss": 7.0816, "step": 258 }, { "epoch": 0.29, "learning_rate": 1.518e-05, "loss": 7.0115, "step": 259 }, { "epoch": 0.29, "learning_rate": 1.524e-05, "loss": 7.1326, "step": 260 }, { "epoch": 0.29, "learning_rate": 1.53e-05, "loss": 7.11, "step": 261 }, { "epoch": 0.29, "learning_rate": 1.5360000000000002e-05, "loss": 7.0056, "step": 262 }, { "epoch": 0.29, "learning_rate": 1.542e-05, "loss": 7.1262, "step": 263 }, { "epoch": 0.3, "learning_rate": 1.548e-05, "loss": 7.0983, "step": 264 }, { "epoch": 0.3, "learning_rate": 1.554e-05, "loss": 7.0463, "step": 265 }, { "epoch": 0.3, "learning_rate": 1.56e-05, "loss": 7.1723, "step": 266 }, { "epoch": 0.3, "learning_rate": 1.5660000000000003e-05, "loss": 7.192, "step": 267 }, { "epoch": 0.3, "learning_rate": 1.5720000000000002e-05, "loss": 7.1711, "step": 268 }, { "epoch": 0.3, "learning_rate": 1.578e-05, "loss": 7.0282, "step": 269 }, { "epoch": 0.3, "learning_rate": 1.584e-05, "loss": 7.088, "step": 270 }, { "epoch": 0.3, "learning_rate": 1.59e-05, "loss": 6.9821, "step": 271 }, { "epoch": 0.3, "learning_rate": 1.596e-05, "loss": 6.9921, "step": 272 }, { "epoch": 0.31, "learning_rate": 1.6020000000000002e-05, "loss": 7.0025, "step": 273 }, { "epoch": 0.31, "learning_rate": 1.6080000000000002e-05, "loss": 6.8786, "step": 274 }, { "epoch": 0.31, "learning_rate": 1.614e-05, "loss": 7.1619, "step": 275 }, { "epoch": 0.31, "learning_rate": 1.62e-05, "loss": 6.9629, "step": 276 }, { "epoch": 0.31, "learning_rate": 1.626e-05, "loss": 7.0722, "step": 277 }, { "epoch": 0.31, "learning_rate": 1.6320000000000003e-05, "loss": 7.0543, "step": 278 }, { "epoch": 0.31, "learning_rate": 1.6380000000000002e-05, "loss": 7.0624, "step": 279 }, { "epoch": 0.31, "learning_rate": 1.6440000000000002e-05, "loss": 7.0812, "step": 280 }, { "epoch": 0.32, "learning_rate": 1.65e-05, "loss": 6.9212, "step": 281 }, { "epoch": 0.32, "learning_rate": 1.656e-05, "loss": 7.0685, "step": 282 }, { "epoch": 0.32, "learning_rate": 1.6620000000000004e-05, "loss": 6.8209, "step": 283 }, { "epoch": 0.32, "learning_rate": 1.6680000000000003e-05, "loss": 7.1441, "step": 284 }, { "epoch": 0.32, "learning_rate": 1.6740000000000002e-05, "loss": 7.0629, "step": 285 }, { "epoch": 0.32, "learning_rate": 1.6800000000000002e-05, "loss": 6.8677, "step": 286 }, { "epoch": 0.32, "learning_rate": 1.686e-05, "loss": 7.1085, "step": 287 }, { "epoch": 0.32, "learning_rate": 1.6919999999999997e-05, "loss": 6.9752, "step": 288 }, { "epoch": 0.32, "learning_rate": 1.698e-05, "loss": 7.1591, "step": 289 }, { "epoch": 0.33, "learning_rate": 1.704e-05, "loss": 7.2093, "step": 290 }, { "epoch": 0.33, "learning_rate": 1.71e-05, "loss": 7.1283, "step": 291 }, { "epoch": 0.33, "learning_rate": 1.716e-05, "loss": 6.9561, "step": 292 }, { "epoch": 0.33, "learning_rate": 1.7219999999999998e-05, "loss": 6.9242, "step": 293 }, { "epoch": 0.33, "learning_rate": 1.728e-05, "loss": 6.9559, "step": 294 }, { "epoch": 0.33, "learning_rate": 1.734e-05, "loss": 6.8141, "step": 295 }, { "epoch": 0.33, "learning_rate": 1.74e-05, "loss": 6.8798, "step": 296 }, { "epoch": 0.33, "learning_rate": 1.746e-05, "loss": 6.9039, "step": 297 }, { "epoch": 0.33, "learning_rate": 1.7519999999999998e-05, "loss": 6.6345, "step": 298 }, { "epoch": 0.34, "learning_rate": 1.758e-05, "loss": 6.6949, "step": 299 }, { "epoch": 0.34, "learning_rate": 1.764e-05, "loss": 6.4176, "step": 300 }, { "epoch": 0.34, "learning_rate": 1.77e-05, "loss": 7.3819, "step": 301 }, { "epoch": 0.34, "learning_rate": 1.776e-05, "loss": 7.3948, "step": 302 }, { "epoch": 0.34, "learning_rate": 1.782e-05, "loss": 7.2879, "step": 303 }, { "epoch": 0.34, "learning_rate": 1.7879999999999998e-05, "loss": 7.3106, "step": 304 }, { "epoch": 0.34, "learning_rate": 1.794e-05, "loss": 7.0604, "step": 305 }, { "epoch": 0.34, "learning_rate": 1.8e-05, "loss": 7.1098, "step": 306 }, { "epoch": 0.34, "learning_rate": 1.806e-05, "loss": 7.0389, "step": 307 }, { "epoch": 0.35, "learning_rate": 1.812e-05, "loss": 7.0779, "step": 308 }, { "epoch": 0.35, "learning_rate": 1.818e-05, "loss": 7.0045, "step": 309 }, { "epoch": 0.35, "learning_rate": 1.824e-05, "loss": 7.1571, "step": 310 }, { "epoch": 0.35, "learning_rate": 1.83e-05, "loss": 7.1784, "step": 311 }, { "epoch": 0.35, "learning_rate": 1.836e-05, "loss": 7.085, "step": 312 }, { "epoch": 0.35, "learning_rate": 1.842e-05, "loss": 7.1189, "step": 313 }, { "epoch": 0.35, "learning_rate": 1.848e-05, "loss": 7.0881, "step": 314 }, { "epoch": 0.35, "learning_rate": 1.854e-05, "loss": 6.9862, "step": 315 }, { "epoch": 0.35, "learning_rate": 1.86e-05, "loss": 7.1703, "step": 316 }, { "epoch": 0.36, "learning_rate": 1.866e-05, "loss": 6.9885, "step": 317 }, { "epoch": 0.36, "learning_rate": 1.872e-05, "loss": 7.0322, "step": 318 }, { "epoch": 0.36, "learning_rate": 1.878e-05, "loss": 6.8968, "step": 319 }, { "epoch": 0.36, "learning_rate": 1.884e-05, "loss": 7.1231, "step": 320 }, { "epoch": 0.36, "learning_rate": 1.8900000000000002e-05, "loss": 7.0431, "step": 321 }, { "epoch": 0.36, "learning_rate": 1.896e-05, "loss": 6.9539, "step": 322 }, { "epoch": 0.36, "learning_rate": 1.902e-05, "loss": 7.0294, "step": 323 }, { "epoch": 0.36, "learning_rate": 1.908e-05, "loss": 7.1256, "step": 324 }, { "epoch": 0.36, "learning_rate": 1.914e-05, "loss": 7.1073, "step": 325 }, { "epoch": 0.37, "learning_rate": 1.9200000000000003e-05, "loss": 7.0236, "step": 326 }, { "epoch": 0.37, "learning_rate": 1.9260000000000002e-05, "loss": 6.9379, "step": 327 }, { "epoch": 0.37, "learning_rate": 1.932e-05, "loss": 7.0584, "step": 328 }, { "epoch": 0.37, "learning_rate": 1.938e-05, "loss": 7.0735, "step": 329 }, { "epoch": 0.37, "learning_rate": 1.944e-05, "loss": 7.0698, "step": 330 }, { "epoch": 0.37, "learning_rate": 1.95e-05, "loss": 7.0466, "step": 331 }, { "epoch": 0.37, "learning_rate": 1.9560000000000002e-05, "loss": 6.8519, "step": 332 }, { "epoch": 0.37, "learning_rate": 1.9620000000000002e-05, "loss": 7.039, "step": 333 }, { "epoch": 0.37, "learning_rate": 1.968e-05, "loss": 7.0502, "step": 334 }, { "epoch": 0.38, "learning_rate": 1.974e-05, "loss": 7.1384, "step": 335 }, { "epoch": 0.38, "learning_rate": 1.98e-05, "loss": 6.9223, "step": 336 }, { "epoch": 0.38, "learning_rate": 1.9860000000000003e-05, "loss": 6.9143, "step": 337 }, { "epoch": 0.38, "learning_rate": 1.9920000000000002e-05, "loss": 7.0827, "step": 338 }, { "epoch": 0.38, "learning_rate": 1.9980000000000002e-05, "loss": 7.0512, "step": 339 }, { "epoch": 0.38, "learning_rate": 2.004e-05, "loss": 7.0057, "step": 340 }, { "epoch": 0.38, "learning_rate": 2.01e-05, "loss": 7.0963, "step": 341 }, { "epoch": 0.38, "learning_rate": 2.016e-05, "loss": 7.0157, "step": 342 }, { "epoch": 0.38, "learning_rate": 2.0220000000000003e-05, "loss": 6.9806, "step": 343 }, { "epoch": 0.39, "learning_rate": 2.0280000000000002e-05, "loss": 6.9626, "step": 344 }, { "epoch": 0.39, "learning_rate": 2.0340000000000002e-05, "loss": 6.7833, "step": 345 }, { "epoch": 0.39, "learning_rate": 2.04e-05, "loss": 6.8474, "step": 346 }, { "epoch": 0.39, "learning_rate": 2.046e-05, "loss": 6.8274, "step": 347 }, { "epoch": 0.39, "learning_rate": 2.0520000000000003e-05, "loss": 6.9157, "step": 348 }, { "epoch": 0.39, "learning_rate": 2.0580000000000003e-05, "loss": 6.8351, "step": 349 }, { "epoch": 0.39, "learning_rate": 2.064e-05, "loss": 6.5647, "step": 350 }, { "epoch": 0.39, "learning_rate": 2.07e-05, "loss": 7.1992, "step": 351 }, { "epoch": 0.39, "learning_rate": 2.0759999999999998e-05, "loss": 7.2511, "step": 352 }, { "epoch": 0.4, "learning_rate": 2.082e-05, "loss": 7.2099, "step": 353 }, { "epoch": 0.4, "learning_rate": 2.088e-05, "loss": 7.1572, "step": 354 }, { "epoch": 0.4, "learning_rate": 2.094e-05, "loss": 7.1885, "step": 355 }, { "epoch": 0.4, "learning_rate": 2.1e-05, "loss": 7.0708, "step": 356 }, { "epoch": 0.4, "learning_rate": 2.1059999999999998e-05, "loss": 7.2303, "step": 357 }, { "epoch": 0.4, "learning_rate": 2.1119999999999998e-05, "loss": 7.0022, "step": 358 }, { "epoch": 0.4, "learning_rate": 2.118e-05, "loss": 7.0732, "step": 359 }, { "epoch": 0.4, "learning_rate": 2.124e-05, "loss": 6.8536, "step": 360 }, { "epoch": 0.4, "learning_rate": 2.13e-05, "loss": 6.9712, "step": 361 }, { "epoch": 0.41, "learning_rate": 2.136e-05, "loss": 7.0666, "step": 362 }, { "epoch": 0.41, "learning_rate": 2.1419999999999998e-05, "loss": 7.0119, "step": 363 }, { "epoch": 0.41, "learning_rate": 2.148e-05, "loss": 7.1132, "step": 364 }, { "epoch": 0.41, "learning_rate": 2.154e-05, "loss": 7.0009, "step": 365 }, { "epoch": 0.41, "learning_rate": 2.16e-05, "loss": 6.9363, "step": 366 }, { "epoch": 0.41, "learning_rate": 2.166e-05, "loss": 6.9805, "step": 367 }, { "epoch": 0.41, "learning_rate": 2.172e-05, "loss": 6.9944, "step": 368 }, { "epoch": 0.41, "learning_rate": 2.178e-05, "loss": 6.9386, "step": 369 }, { "epoch": 0.41, "learning_rate": 2.184e-05, "loss": 7.0549, "step": 370 }, { "epoch": 0.42, "learning_rate": 2.19e-05, "loss": 6.9824, "step": 371 }, { "epoch": 0.42, "learning_rate": 2.196e-05, "loss": 7.1437, "step": 372 }, { "epoch": 0.42, "learning_rate": 2.202e-05, "loss": 6.9763, "step": 373 }, { "epoch": 0.42, "learning_rate": 2.208e-05, "loss": 7.0529, "step": 374 }, { "epoch": 0.42, "learning_rate": 2.214e-05, "loss": 6.8479, "step": 375 }, { "epoch": 0.42, "learning_rate": 2.22e-05, "loss": 6.9973, "step": 376 }, { "epoch": 0.42, "learning_rate": 2.226e-05, "loss": 6.8522, "step": 377 }, { "epoch": 0.42, "learning_rate": 2.232e-05, "loss": 6.9552, "step": 378 }, { "epoch": 0.42, "learning_rate": 2.238e-05, "loss": 6.9154, "step": 379 }, { "epoch": 0.43, "learning_rate": 2.2440000000000002e-05, "loss": 7.0564, "step": 380 }, { "epoch": 0.43, "learning_rate": 2.25e-05, "loss": 7.0364, "step": 381 }, { "epoch": 0.43, "learning_rate": 2.256e-05, "loss": 6.9092, "step": 382 }, { "epoch": 0.43, "learning_rate": 2.262e-05, "loss": 7.0595, "step": 383 }, { "epoch": 0.43, "learning_rate": 2.268e-05, "loss": 7.0184, "step": 384 }, { "epoch": 0.43, "learning_rate": 2.274e-05, "loss": 6.9381, "step": 385 }, { "epoch": 0.43, "learning_rate": 2.2800000000000002e-05, "loss": 6.9469, "step": 386 }, { "epoch": 0.43, "learning_rate": 2.286e-05, "loss": 6.9595, "step": 387 }, { "epoch": 0.43, "learning_rate": 2.292e-05, "loss": 6.9449, "step": 388 }, { "epoch": 0.44, "learning_rate": 2.298e-05, "loss": 6.9221, "step": 389 }, { "epoch": 0.44, "learning_rate": 2.304e-05, "loss": 7.1336, "step": 390 }, { "epoch": 0.44, "learning_rate": 2.3100000000000002e-05, "loss": 7.0099, "step": 391 }, { "epoch": 0.44, "learning_rate": 2.3160000000000002e-05, "loss": 6.9662, "step": 392 }, { "epoch": 0.44, "learning_rate": 2.322e-05, "loss": 6.96, "step": 393 }, { "epoch": 0.44, "learning_rate": 2.328e-05, "loss": 7.1539, "step": 394 }, { "epoch": 0.44, "learning_rate": 2.334e-05, "loss": 6.9282, "step": 395 }, { "epoch": 0.44, "learning_rate": 2.3400000000000003e-05, "loss": 6.7545, "step": 396 }, { "epoch": 0.45, "learning_rate": 2.3460000000000002e-05, "loss": 6.6451, "step": 397 }, { "epoch": 0.45, "learning_rate": 2.3520000000000002e-05, "loss": 6.722, "step": 398 }, { "epoch": 0.45, "learning_rate": 2.358e-05, "loss": 6.7318, "step": 399 }, { "epoch": 0.45, "learning_rate": 2.364e-05, "loss": 6.4948, "step": 400 }, { "epoch": 0.45, "learning_rate": 2.37e-05, "loss": 7.3706, "step": 401 }, { "epoch": 0.45, "learning_rate": 2.3760000000000003e-05, "loss": 7.3375, "step": 402 }, { "epoch": 0.45, "learning_rate": 2.3820000000000002e-05, "loss": 7.3005, "step": 403 }, { "epoch": 0.45, "learning_rate": 2.3880000000000002e-05, "loss": 7.1926, "step": 404 }, { "epoch": 0.45, "learning_rate": 2.394e-05, "loss": 7.2948, "step": 405 }, { "epoch": 0.46, "learning_rate": 2.4e-05, "loss": 7.0679, "step": 406 }, { "epoch": 0.46, "learning_rate": 2.4060000000000003e-05, "loss": 7.0573, "step": 407 }, { "epoch": 0.46, "learning_rate": 2.4120000000000003e-05, "loss": 6.9789, "step": 408 }, { "epoch": 0.46, "learning_rate": 2.4180000000000002e-05, "loss": 7.2078, "step": 409 }, { "epoch": 0.46, "learning_rate": 2.4240000000000002e-05, "loss": 7.0875, "step": 410 }, { "epoch": 0.46, "learning_rate": 2.43e-05, "loss": 7.0323, "step": 411 }, { "epoch": 0.46, "learning_rate": 2.4360000000000004e-05, "loss": 7.019, "step": 412 }, { "epoch": 0.46, "learning_rate": 2.442e-05, "loss": 7.0001, "step": 413 }, { "epoch": 0.46, "learning_rate": 2.448e-05, "loss": 6.9005, "step": 414 }, { "epoch": 0.47, "learning_rate": 2.454e-05, "loss": 6.9795, "step": 415 }, { "epoch": 0.47, "learning_rate": 2.4599999999999998e-05, "loss": 7.0655, "step": 416 }, { "epoch": 0.47, "learning_rate": 2.4659999999999998e-05, "loss": 6.9416, "step": 417 }, { "epoch": 0.47, "learning_rate": 2.472e-05, "loss": 7.0493, "step": 418 }, { "epoch": 0.47, "learning_rate": 2.478e-05, "loss": 7.0227, "step": 419 }, { "epoch": 0.47, "learning_rate": 2.484e-05, "loss": 7.0334, "step": 420 }, { "epoch": 0.47, "learning_rate": 2.49e-05, "loss": 7.0272, "step": 421 }, { "epoch": 0.47, "learning_rate": 2.4959999999999998e-05, "loss": 7.0854, "step": 422 }, { "epoch": 0.47, "learning_rate": 2.502e-05, "loss": 7.0001, "step": 423 }, { "epoch": 0.48, "learning_rate": 2.508e-05, "loss": 6.9993, "step": 424 }, { "epoch": 0.48, "learning_rate": 2.514e-05, "loss": 7.0321, "step": 425 }, { "epoch": 0.48, "learning_rate": 2.52e-05, "loss": 7.1149, "step": 426 }, { "epoch": 0.48, "learning_rate": 2.526e-05, "loss": 6.8887, "step": 427 }, { "epoch": 0.48, "learning_rate": 2.5319999999999998e-05, "loss": 7.1664, "step": 428 }, { "epoch": 0.48, "learning_rate": 2.538e-05, "loss": 7.0092, "step": 429 }, { "epoch": 0.48, "learning_rate": 2.544e-05, "loss": 6.9772, "step": 430 }, { "epoch": 0.48, "learning_rate": 2.55e-05, "loss": 7.1404, "step": 431 }, { "epoch": 0.48, "learning_rate": 2.556e-05, "loss": 7.0523, "step": 432 }, { "epoch": 0.49, "learning_rate": 2.562e-05, "loss": 6.762, "step": 433 }, { "epoch": 0.49, "learning_rate": 2.568e-05, "loss": 6.9687, "step": 434 }, { "epoch": 0.49, "learning_rate": 2.574e-05, "loss": 6.8621, "step": 435 }, { "epoch": 0.49, "learning_rate": 2.58e-05, "loss": 7.0696, "step": 436 }, { "epoch": 0.49, "learning_rate": 2.586e-05, "loss": 7.0226, "step": 437 }, { "epoch": 0.49, "learning_rate": 2.592e-05, "loss": 7.0763, "step": 438 }, { "epoch": 0.49, "learning_rate": 2.5980000000000002e-05, "loss": 6.8442, "step": 439 }, { "epoch": 0.49, "learning_rate": 2.604e-05, "loss": 7.0126, "step": 440 }, { "epoch": 0.49, "learning_rate": 2.61e-05, "loss": 7.0358, "step": 441 }, { "epoch": 0.5, "learning_rate": 2.616e-05, "loss": 6.954, "step": 442 }, { "epoch": 0.5, "learning_rate": 2.622e-05, "loss": 6.8816, "step": 443 }, { "epoch": 0.5, "learning_rate": 2.628e-05, "loss": 7.0431, "step": 444 }, { "epoch": 0.5, "learning_rate": 2.6340000000000002e-05, "loss": 6.7244, "step": 445 }, { "epoch": 0.5, "learning_rate": 2.64e-05, "loss": 6.772, "step": 446 }, { "epoch": 0.5, "learning_rate": 2.646e-05, "loss": 6.822, "step": 447 }, { "epoch": 0.5, "learning_rate": 2.652e-05, "loss": 6.8692, "step": 448 }, { "epoch": 0.5, "learning_rate": 2.658e-05, "loss": 6.6106, "step": 449 }, { "epoch": 0.5, "learning_rate": 2.6640000000000002e-05, "loss": 6.8333, "step": 450 }, { "epoch": 0.51, "learning_rate": 2.6700000000000002e-05, "loss": 7.2419, "step": 451 }, { "epoch": 0.51, "learning_rate": 2.676e-05, "loss": 7.3483, "step": 452 }, { "epoch": 0.51, "learning_rate": 2.682e-05, "loss": 7.3776, "step": 453 }, { "epoch": 0.51, "learning_rate": 2.688e-05, "loss": 7.1304, "step": 454 }, { "epoch": 0.51, "learning_rate": 2.6940000000000003e-05, "loss": 7.1138, "step": 455 }, { "epoch": 0.51, "learning_rate": 2.7000000000000002e-05, "loss": 7.16, "step": 456 }, { "epoch": 0.51, "learning_rate": 2.7060000000000002e-05, "loss": 6.9576, "step": 457 }, { "epoch": 0.51, "learning_rate": 2.712e-05, "loss": 6.9716, "step": 458 }, { "epoch": 0.51, "learning_rate": 2.718e-05, "loss": 7.0103, "step": 459 }, { "epoch": 0.52, "learning_rate": 2.724e-05, "loss": 7.0957, "step": 460 }, { "epoch": 0.52, "learning_rate": 2.7300000000000003e-05, "loss": 6.9116, "step": 461 }, { "epoch": 0.52, "learning_rate": 2.7360000000000002e-05, "loss": 7.0365, "step": 462 }, { "epoch": 0.52, "learning_rate": 2.7420000000000002e-05, "loss": 6.9483, "step": 463 }, { "epoch": 0.52, "learning_rate": 2.748e-05, "loss": 6.9044, "step": 464 }, { "epoch": 0.52, "learning_rate": 2.754e-05, "loss": 7.0571, "step": 465 }, { "epoch": 0.52, "learning_rate": 2.7600000000000003e-05, "loss": 6.9853, "step": 466 }, { "epoch": 0.52, "learning_rate": 2.7660000000000003e-05, "loss": 6.9274, "step": 467 }, { "epoch": 0.52, "learning_rate": 2.7720000000000002e-05, "loss": 7.0054, "step": 468 }, { "epoch": 0.53, "learning_rate": 2.778e-05, "loss": 6.9376, "step": 469 }, { "epoch": 0.53, "learning_rate": 2.784e-05, "loss": 7.0067, "step": 470 }, { "epoch": 0.53, "learning_rate": 2.79e-05, "loss": 6.9132, "step": 471 }, { "epoch": 0.53, "learning_rate": 2.7960000000000003e-05, "loss": 6.9248, "step": 472 }, { "epoch": 0.53, "learning_rate": 2.8020000000000003e-05, "loss": 6.9471, "step": 473 }, { "epoch": 0.53, "learning_rate": 2.8080000000000002e-05, "loss": 6.8781, "step": 474 }, { "epoch": 0.53, "learning_rate": 2.8139999999999998e-05, "loss": 7.0458, "step": 475 }, { "epoch": 0.53, "learning_rate": 2.8199999999999998e-05, "loss": 7.0387, "step": 476 }, { "epoch": 0.53, "learning_rate": 2.826e-05, "loss": 7.1225, "step": 477 }, { "epoch": 0.54, "learning_rate": 2.832e-05, "loss": 6.859, "step": 478 }, { "epoch": 0.54, "learning_rate": 2.838e-05, "loss": 6.8874, "step": 479 }, { "epoch": 0.54, "learning_rate": 2.844e-05, "loss": 6.8669, "step": 480 }, { "epoch": 0.54, "learning_rate": 2.8499999999999998e-05, "loss": 6.9159, "step": 481 }, { "epoch": 0.54, "learning_rate": 2.856e-05, "loss": 6.9117, "step": 482 }, { "epoch": 0.54, "learning_rate": 2.862e-05, "loss": 6.9011, "step": 483 }, { "epoch": 0.54, "learning_rate": 2.868e-05, "loss": 6.9893, "step": 484 }, { "epoch": 0.54, "learning_rate": 2.874e-05, "loss": 6.9789, "step": 485 }, { "epoch": 0.54, "learning_rate": 2.88e-05, "loss": 6.9288, "step": 486 }, { "epoch": 0.55, "learning_rate": 2.8859999999999998e-05, "loss": 6.9514, "step": 487 }, { "epoch": 0.55, "learning_rate": 2.892e-05, "loss": 6.8562, "step": 488 }, { "epoch": 0.55, "learning_rate": 2.898e-05, "loss": 7.046, "step": 489 }, { "epoch": 0.55, "learning_rate": 2.904e-05, "loss": 7.0476, "step": 490 }, { "epoch": 0.55, "learning_rate": 2.91e-05, "loss": 6.8383, "step": 491 }, { "epoch": 0.55, "learning_rate": 2.916e-05, "loss": 6.9706, "step": 492 }, { "epoch": 0.55, "learning_rate": 2.922e-05, "loss": 6.9968, "step": 493 }, { "epoch": 0.55, "learning_rate": 2.928e-05, "loss": 7.0871, "step": 494 }, { "epoch": 0.55, "learning_rate": 2.934e-05, "loss": 7.2337, "step": 495 }, { "epoch": 0.56, "learning_rate": 2.94e-05, "loss": 7.0516, "step": 496 }, { "epoch": 0.56, "learning_rate": 2.946e-05, "loss": 6.902, "step": 497 }, { "epoch": 0.56, "learning_rate": 2.9520000000000002e-05, "loss": 7.4479, "step": 498 }, { "epoch": 0.56, "learning_rate": 2.958e-05, "loss": 7.2373, "step": 499 }, { "epoch": 0.56, "learning_rate": 2.964e-05, "loss": 6.7189, "step": 500 }, { "epoch": 0.56, "eval_loss": 6.97957706451416, "eval_runtime": 418.4828, "eval_samples_per_second": 6.313, "eval_steps_per_second": 0.791, "eval_wer": 0.9350059500198334, "step": 500 }, { "epoch": 0.56, "learning_rate": 2.97e-05, "loss": 7.3256, "step": 501 }, { "epoch": 0.56, "learning_rate": 2.976e-05, "loss": 7.0178, "step": 502 }, { "epoch": 0.56, "learning_rate": 2.982e-05, "loss": 7.1042, "step": 503 }, { "epoch": 0.57, "learning_rate": 2.9880000000000002e-05, "loss": 7.1199, "step": 504 }, { "epoch": 0.57, "learning_rate": 2.994e-05, "loss": 7.0949, "step": 505 }, { "epoch": 0.57, "learning_rate": 3e-05, "loss": 7.055, "step": 506 }, { "epoch": 0.57, "learning_rate": 2.999242424242424e-05, "loss": 7.0444, "step": 507 }, { "epoch": 0.57, "learning_rate": 2.998484848484849e-05, "loss": 6.9305, "step": 508 }, { "epoch": 0.57, "learning_rate": 2.997727272727273e-05, "loss": 7.0002, "step": 509 }, { "epoch": 0.57, "learning_rate": 2.996969696969697e-05, "loss": 6.9943, "step": 510 }, { "epoch": 0.57, "learning_rate": 2.996212121212121e-05, "loss": 6.9715, "step": 511 }, { "epoch": 0.57, "learning_rate": 2.9954545454545458e-05, "loss": 7.0021, "step": 512 }, { "epoch": 0.58, "learning_rate": 2.99469696969697e-05, "loss": 7.1063, "step": 513 }, { "epoch": 0.58, "learning_rate": 2.993939393939394e-05, "loss": 7.0164, "step": 514 }, { "epoch": 0.58, "learning_rate": 2.993181818181818e-05, "loss": 6.9971, "step": 515 }, { "epoch": 0.58, "learning_rate": 2.9924242424242427e-05, "loss": 6.8436, "step": 516 }, { "epoch": 0.58, "learning_rate": 2.9916666666666668e-05, "loss": 7.0411, "step": 517 }, { "epoch": 0.58, "learning_rate": 2.9909090909090908e-05, "loss": 6.9065, "step": 518 }, { "epoch": 0.58, "learning_rate": 2.9901515151515152e-05, "loss": 7.0454, "step": 519 }, { "epoch": 0.58, "learning_rate": 2.9893939393939396e-05, "loss": 7.0887, "step": 520 }, { "epoch": 0.58, "learning_rate": 2.9886363636363637e-05, "loss": 6.8294, "step": 521 }, { "epoch": 0.59, "learning_rate": 2.9878787878787877e-05, "loss": 6.8443, "step": 522 }, { "epoch": 0.59, "learning_rate": 2.987121212121212e-05, "loss": 6.8003, "step": 523 }, { "epoch": 0.59, "learning_rate": 2.9863636363636365e-05, "loss": 6.8755, "step": 524 }, { "epoch": 0.59, "learning_rate": 2.9856060606060606e-05, "loss": 7.0167, "step": 525 }, { "epoch": 0.59, "learning_rate": 2.9848484848484847e-05, "loss": 6.8139, "step": 526 }, { "epoch": 0.59, "learning_rate": 2.9840909090909094e-05, "loss": 6.9642, "step": 527 }, { "epoch": 0.59, "learning_rate": 2.9833333333333335e-05, "loss": 6.9764, "step": 528 }, { "epoch": 0.59, "learning_rate": 2.9825757575757575e-05, "loss": 7.0131, "step": 529 }, { "epoch": 0.59, "learning_rate": 2.981818181818182e-05, "loss": 7.048, "step": 530 }, { "epoch": 0.6, "learning_rate": 2.9810606060606063e-05, "loss": 6.9529, "step": 531 }, { "epoch": 0.6, "learning_rate": 2.9803030303030304e-05, "loss": 6.8581, "step": 532 }, { "epoch": 0.6, "learning_rate": 2.9795454545454544e-05, "loss": 6.8582, "step": 533 }, { "epoch": 0.6, "learning_rate": 2.978787878787879e-05, "loss": 6.8489, "step": 534 }, { "epoch": 0.6, "learning_rate": 2.9780303030303032e-05, "loss": 6.8839, "step": 535 }, { "epoch": 0.6, "learning_rate": 2.9772727272727273e-05, "loss": 7.0189, "step": 536 }, { "epoch": 0.6, "learning_rate": 2.9765151515151517e-05, "loss": 6.9644, "step": 537 }, { "epoch": 0.6, "learning_rate": 2.9757575757575757e-05, "loss": 6.9962, "step": 538 }, { "epoch": 0.6, "learning_rate": 2.975e-05, "loss": 6.9847, "step": 539 }, { "epoch": 0.61, "learning_rate": 2.9742424242424242e-05, "loss": 6.9312, "step": 540 }, { "epoch": 0.61, "learning_rate": 2.9734848484848486e-05, "loss": 7.0639, "step": 541 }, { "epoch": 0.61, "learning_rate": 2.972727272727273e-05, "loss": 6.9054, "step": 542 }, { "epoch": 0.61, "learning_rate": 2.971969696969697e-05, "loss": 6.8389, "step": 543 }, { "epoch": 0.61, "learning_rate": 2.971212121212121e-05, "loss": 6.7012, "step": 544 }, { "epoch": 0.61, "learning_rate": 2.9704545454545455e-05, "loss": 6.7442, "step": 545 }, { "epoch": 0.61, "learning_rate": 2.96969696969697e-05, "loss": 6.7798, "step": 546 }, { "epoch": 0.61, "learning_rate": 2.968939393939394e-05, "loss": 6.9363, "step": 547 }, { "epoch": 0.61, "learning_rate": 2.9681818181818184e-05, "loss": 6.8192, "step": 548 }, { "epoch": 0.62, "learning_rate": 2.9674242424242424e-05, "loss": 6.4543, "step": 549 }, { "epoch": 0.62, "learning_rate": 2.966666666666667e-05, "loss": 6.6238, "step": 550 }, { "epoch": 0.62, "learning_rate": 2.965909090909091e-05, "loss": 7.3363, "step": 551 }, { "epoch": 0.62, "learning_rate": 2.9651515151515153e-05, "loss": 7.6111, "step": 552 }, { "epoch": 0.62, "learning_rate": 2.9643939393939394e-05, "loss": 7.3625, "step": 553 }, { "epoch": 0.62, "learning_rate": 2.9636363636363638e-05, "loss": 7.3178, "step": 554 }, { "epoch": 0.62, "learning_rate": 2.9628787878787878e-05, "loss": 7.149, "step": 555 }, { "epoch": 0.62, "learning_rate": 2.9621212121212122e-05, "loss": 7.0668, "step": 556 }, { "epoch": 0.62, "learning_rate": 2.9613636363636363e-05, "loss": 6.9875, "step": 557 }, { "epoch": 0.63, "learning_rate": 2.9606060606060607e-05, "loss": 7.0268, "step": 558 }, { "epoch": 0.63, "learning_rate": 2.959848484848485e-05, "loss": 6.9423, "step": 559 }, { "epoch": 0.63, "learning_rate": 2.959090909090909e-05, "loss": 6.8895, "step": 560 }, { "epoch": 0.63, "learning_rate": 2.9583333333333335e-05, "loss": 7.0109, "step": 561 }, { "epoch": 0.63, "learning_rate": 2.9575757575757576e-05, "loss": 6.8688, "step": 562 }, { "epoch": 0.63, "learning_rate": 2.956818181818182e-05, "loss": 6.9204, "step": 563 }, { "epoch": 0.63, "learning_rate": 2.956060606060606e-05, "loss": 6.9138, "step": 564 }, { "epoch": 0.63, "learning_rate": 2.9553030303030304e-05, "loss": 7.0982, "step": 565 }, { "epoch": 0.63, "learning_rate": 2.9545454545454545e-05, "loss": 6.8102, "step": 566 }, { "epoch": 0.64, "learning_rate": 2.953787878787879e-05, "loss": 6.9336, "step": 567 }, { "epoch": 0.64, "learning_rate": 2.953030303030303e-05, "loss": 6.8465, "step": 568 }, { "epoch": 0.64, "learning_rate": 2.9522727272727274e-05, "loss": 6.9016, "step": 569 }, { "epoch": 0.64, "learning_rate": 2.9515151515151518e-05, "loss": 6.9977, "step": 570 }, { "epoch": 0.64, "learning_rate": 2.9507575757575758e-05, "loss": 6.8656, "step": 571 }, { "epoch": 0.64, "learning_rate": 2.95e-05, "loss": 7.0155, "step": 572 }, { "epoch": 0.64, "learning_rate": 2.9492424242424243e-05, "loss": 6.8246, "step": 573 }, { "epoch": 0.64, "learning_rate": 2.9484848484848487e-05, "loss": 6.7715, "step": 574 }, { "epoch": 0.64, "learning_rate": 2.9477272727272727e-05, "loss": 6.9652, "step": 575 }, { "epoch": 0.65, "learning_rate": 2.946969696969697e-05, "loss": 6.7729, "step": 576 }, { "epoch": 0.65, "learning_rate": 2.9462121212121212e-05, "loss": 6.9424, "step": 577 }, { "epoch": 0.65, "learning_rate": 2.9454545454545456e-05, "loss": 6.9762, "step": 578 }, { "epoch": 0.65, "learning_rate": 2.9446969696969697e-05, "loss": 6.8195, "step": 579 }, { "epoch": 0.65, "learning_rate": 2.943939393939394e-05, "loss": 7.0203, "step": 580 }, { "epoch": 0.65, "learning_rate": 2.9431818181818184e-05, "loss": 7.0054, "step": 581 }, { "epoch": 0.65, "learning_rate": 2.9424242424242425e-05, "loss": 6.8635, "step": 582 }, { "epoch": 0.65, "learning_rate": 2.9416666666666666e-05, "loss": 6.8314, "step": 583 }, { "epoch": 0.65, "learning_rate": 2.940909090909091e-05, "loss": 6.866, "step": 584 }, { "epoch": 0.66, "learning_rate": 2.9401515151515154e-05, "loss": 6.8752, "step": 585 }, { "epoch": 0.66, "learning_rate": 2.9393939393939394e-05, "loss": 6.9335, "step": 586 }, { "epoch": 0.66, "learning_rate": 2.9386363636363635e-05, "loss": 6.9919, "step": 587 }, { "epoch": 0.66, "learning_rate": 2.9378787878787882e-05, "loss": 6.8794, "step": 588 }, { "epoch": 0.66, "learning_rate": 2.9371212121212123e-05, "loss": 6.7843, "step": 589 }, { "epoch": 0.66, "learning_rate": 2.9363636363636363e-05, "loss": 6.8796, "step": 590 }, { "epoch": 0.66, "learning_rate": 2.9356060606060604e-05, "loss": 6.8827, "step": 591 }, { "epoch": 0.66, "learning_rate": 2.934848484848485e-05, "loss": 6.893, "step": 592 }, { "epoch": 0.66, "learning_rate": 2.9340909090909092e-05, "loss": 6.9547, "step": 593 }, { "epoch": 0.67, "learning_rate": 2.9333333333333333e-05, "loss": 6.6969, "step": 594 }, { "epoch": 0.67, "learning_rate": 2.9325757575757577e-05, "loss": 6.7273, "step": 595 }, { "epoch": 0.67, "learning_rate": 2.931818181818182e-05, "loss": 6.6028, "step": 596 }, { "epoch": 0.67, "learning_rate": 2.931060606060606e-05, "loss": 6.6261, "step": 597 }, { "epoch": 0.67, "learning_rate": 2.9303030303030302e-05, "loss": 6.576, "step": 598 }, { "epoch": 0.67, "learning_rate": 2.929545454545455e-05, "loss": 6.445, "step": 599 }, { "epoch": 0.67, "learning_rate": 2.928787878787879e-05, "loss": 6.1639, "step": 600 }, { "epoch": 0.67, "learning_rate": 2.928030303030303e-05, "loss": 7.295, "step": 601 }, { "epoch": 0.67, "learning_rate": 2.927272727272727e-05, "loss": 7.1209, "step": 602 }, { "epoch": 0.68, "learning_rate": 2.9265151515151518e-05, "loss": 7.3114, "step": 603 }, { "epoch": 0.68, "learning_rate": 2.925757575757576e-05, "loss": 7.2098, "step": 604 }, { "epoch": 0.68, "learning_rate": 2.925e-05, "loss": 7.2582, "step": 605 }, { "epoch": 0.68, "learning_rate": 2.924242424242424e-05, "loss": 6.8832, "step": 606 }, { "epoch": 0.68, "learning_rate": 2.9234848484848487e-05, "loss": 6.9865, "step": 607 }, { "epoch": 0.68, "learning_rate": 2.9227272727272728e-05, "loss": 6.9672, "step": 608 }, { "epoch": 0.68, "learning_rate": 2.921969696969697e-05, "loss": 6.9289, "step": 609 }, { "epoch": 0.68, "learning_rate": 2.9212121212121216e-05, "loss": 6.9261, "step": 610 }, { "epoch": 0.68, "learning_rate": 2.9204545454545457e-05, "loss": 6.8803, "step": 611 }, { "epoch": 0.69, "learning_rate": 2.9196969696969697e-05, "loss": 6.9343, "step": 612 }, { "epoch": 0.69, "learning_rate": 2.9189393939393938e-05, "loss": 6.8361, "step": 613 }, { "epoch": 0.69, "learning_rate": 2.9181818181818185e-05, "loss": 6.8748, "step": 614 }, { "epoch": 0.69, "learning_rate": 2.9174242424242426e-05, "loss": 7.0698, "step": 615 }, { "epoch": 0.69, "learning_rate": 2.9166666666666666e-05, "loss": 6.7753, "step": 616 }, { "epoch": 0.69, "learning_rate": 2.9159090909090907e-05, "loss": 6.8795, "step": 617 }, { "epoch": 0.69, "learning_rate": 2.9151515151515154e-05, "loss": 6.7372, "step": 618 }, { "epoch": 0.69, "learning_rate": 2.9143939393939395e-05, "loss": 6.8502, "step": 619 }, { "epoch": 0.7, "learning_rate": 2.9136363636363636e-05, "loss": 6.8756, "step": 620 }, { "epoch": 0.7, "learning_rate": 2.912878787878788e-05, "loss": 6.9272, "step": 621 }, { "epoch": 0.7, "learning_rate": 2.9121212121212124e-05, "loss": 6.8317, "step": 622 }, { "epoch": 0.7, "learning_rate": 2.9113636363636364e-05, "loss": 6.9235, "step": 623 }, { "epoch": 0.7, "learning_rate": 2.9106060606060605e-05, "loss": 6.8002, "step": 624 }, { "epoch": 0.7, "learning_rate": 2.909848484848485e-05, "loss": 6.9922, "step": 625 }, { "epoch": 0.7, "learning_rate": 2.9090909090909093e-05, "loss": 6.8456, "step": 626 }, { "epoch": 0.7, "learning_rate": 2.9083333333333333e-05, "loss": 6.9406, "step": 627 }, { "epoch": 0.7, "learning_rate": 2.9075757575757574e-05, "loss": 6.8633, "step": 628 }, { "epoch": 0.71, "learning_rate": 2.906818181818182e-05, "loss": 6.8614, "step": 629 }, { "epoch": 0.71, "learning_rate": 2.9060606060606062e-05, "loss": 7.0274, "step": 630 }, { "epoch": 0.71, "learning_rate": 2.9053030303030302e-05, "loss": 6.8269, "step": 631 }, { "epoch": 0.71, "learning_rate": 2.9045454545454546e-05, "loss": 6.6792, "step": 632 }, { "epoch": 0.71, "learning_rate": 2.903787878787879e-05, "loss": 6.894, "step": 633 }, { "epoch": 0.71, "learning_rate": 2.903030303030303e-05, "loss": 6.8096, "step": 634 }, { "epoch": 0.71, "learning_rate": 2.902272727272727e-05, "loss": 6.882, "step": 635 }, { "epoch": 0.71, "learning_rate": 2.9015151515151516e-05, "loss": 6.9043, "step": 636 }, { "epoch": 0.71, "learning_rate": 2.900757575757576e-05, "loss": 6.8804, "step": 637 }, { "epoch": 0.72, "learning_rate": 2.9e-05, "loss": 6.6925, "step": 638 }, { "epoch": 0.72, "learning_rate": 2.8992424242424244e-05, "loss": 6.9852, "step": 639 }, { "epoch": 0.72, "learning_rate": 2.8984848484848485e-05, "loss": 6.6962, "step": 640 }, { "epoch": 0.72, "learning_rate": 2.897727272727273e-05, "loss": 6.7846, "step": 641 }, { "epoch": 0.72, "learning_rate": 2.896969696969697e-05, "loss": 6.9276, "step": 642 }, { "epoch": 0.72, "learning_rate": 2.8962121212121213e-05, "loss": 6.8053, "step": 643 }, { "epoch": 0.72, "learning_rate": 2.8954545454545457e-05, "loss": 6.6757, "step": 644 }, { "epoch": 0.72, "learning_rate": 2.8946969696969698e-05, "loss": 6.6482, "step": 645 }, { "epoch": 0.72, "learning_rate": 2.893939393939394e-05, "loss": 6.7305, "step": 646 }, { "epoch": 0.73, "learning_rate": 2.8931818181818182e-05, "loss": 6.5578, "step": 647 }, { "epoch": 0.73, "learning_rate": 2.8924242424242426e-05, "loss": 6.3727, "step": 648 }, { "epoch": 0.73, "learning_rate": 2.8916666666666667e-05, "loss": 6.4598, "step": 649 }, { "epoch": 0.73, "learning_rate": 2.890909090909091e-05, "loss": 6.2114, "step": 650 }, { "epoch": 0.73, "learning_rate": 2.890151515151515e-05, "loss": 7.4471, "step": 651 }, { "epoch": 0.73, "learning_rate": 2.8893939393939396e-05, "loss": 7.3689, "step": 652 }, { "epoch": 0.73, "learning_rate": 2.8886363636363636e-05, "loss": 7.422, "step": 653 }, { "epoch": 0.73, "learning_rate": 2.887878787878788e-05, "loss": 7.2048, "step": 654 }, { "epoch": 0.73, "learning_rate": 2.887121212121212e-05, "loss": 7.1455, "step": 655 }, { "epoch": 0.74, "learning_rate": 2.8863636363636365e-05, "loss": 7.2232, "step": 656 }, { "epoch": 0.74, "learning_rate": 2.8856060606060605e-05, "loss": 6.8327, "step": 657 }, { "epoch": 0.74, "learning_rate": 2.884848484848485e-05, "loss": 7.0654, "step": 658 }, { "epoch": 0.74, "learning_rate": 2.884090909090909e-05, "loss": 6.8101, "step": 659 }, { "epoch": 0.74, "learning_rate": 2.8833333333333334e-05, "loss": 6.7933, "step": 660 }, { "epoch": 0.74, "learning_rate": 2.8825757575757578e-05, "loss": 6.8043, "step": 661 }, { "epoch": 0.74, "learning_rate": 2.881818181818182e-05, "loss": 6.8815, "step": 662 }, { "epoch": 0.74, "learning_rate": 2.8810606060606063e-05, "loss": 7.0172, "step": 663 }, { "epoch": 0.74, "learning_rate": 2.8803030303030303e-05, "loss": 6.7709, "step": 664 }, { "epoch": 0.75, "learning_rate": 2.8795454545454547e-05, "loss": 6.8094, "step": 665 }, { "epoch": 0.75, "learning_rate": 2.8787878787878788e-05, "loss": 6.9672, "step": 666 }, { "epoch": 0.75, "learning_rate": 2.8780303030303032e-05, "loss": 6.8542, "step": 667 }, { "epoch": 0.75, "learning_rate": 2.8772727272727272e-05, "loss": 6.7564, "step": 668 }, { "epoch": 0.75, "learning_rate": 2.8765151515151516e-05, "loss": 6.8855, "step": 669 }, { "epoch": 0.75, "learning_rate": 2.8757575757575757e-05, "loss": 6.8109, "step": 670 }, { "epoch": 0.75, "learning_rate": 2.875e-05, "loss": 6.7459, "step": 671 }, { "epoch": 0.75, "learning_rate": 2.8742424242424245e-05, "loss": 6.8477, "step": 672 }, { "epoch": 0.75, "learning_rate": 2.8734848484848485e-05, "loss": 6.7863, "step": 673 }, { "epoch": 0.76, "learning_rate": 2.8727272727272726e-05, "loss": 6.7913, "step": 674 }, { "epoch": 0.76, "learning_rate": 2.871969696969697e-05, "loss": 6.8215, "step": 675 }, { "epoch": 0.76, "learning_rate": 2.8712121212121214e-05, "loss": 6.8796, "step": 676 }, { "epoch": 0.76, "learning_rate": 2.8704545454545455e-05, "loss": 6.7825, "step": 677 }, { "epoch": 0.76, "learning_rate": 2.86969696969697e-05, "loss": 6.651, "step": 678 }, { "epoch": 0.76, "learning_rate": 2.868939393939394e-05, "loss": 6.8619, "step": 679 }, { "epoch": 0.76, "learning_rate": 2.8681818181818183e-05, "loss": 6.7986, "step": 680 }, { "epoch": 0.76, "learning_rate": 2.8674242424242424e-05, "loss": 6.9059, "step": 681 }, { "epoch": 0.76, "learning_rate": 2.8666666666666668e-05, "loss": 6.7744, "step": 682 }, { "epoch": 0.77, "learning_rate": 2.8659090909090912e-05, "loss": 6.7859, "step": 683 }, { "epoch": 0.77, "learning_rate": 2.8651515151515152e-05, "loss": 6.87, "step": 684 }, { "epoch": 0.77, "learning_rate": 2.8643939393939393e-05, "loss": 6.9059, "step": 685 }, { "epoch": 0.77, "learning_rate": 2.8636363636363637e-05, "loss": 6.7264, "step": 686 }, { "epoch": 0.77, "learning_rate": 2.862878787878788e-05, "loss": 6.7419, "step": 687 }, { "epoch": 0.77, "learning_rate": 2.862121212121212e-05, "loss": 6.9621, "step": 688 }, { "epoch": 0.77, "learning_rate": 2.8613636363636362e-05, "loss": 6.8194, "step": 689 }, { "epoch": 0.77, "learning_rate": 2.860606060606061e-05, "loss": 6.798, "step": 690 }, { "epoch": 0.77, "learning_rate": 2.859848484848485e-05, "loss": 6.6721, "step": 691 }, { "epoch": 0.78, "learning_rate": 2.859090909090909e-05, "loss": 6.7275, "step": 692 }, { "epoch": 0.78, "learning_rate": 2.858333333333333e-05, "loss": 6.5806, "step": 693 }, { "epoch": 0.78, "learning_rate": 2.857575757575758e-05, "loss": 6.7685, "step": 694 }, { "epoch": 0.78, "learning_rate": 2.856818181818182e-05, "loss": 6.7368, "step": 695 }, { "epoch": 0.78, "learning_rate": 2.856060606060606e-05, "loss": 6.7751, "step": 696 }, { "epoch": 0.78, "learning_rate": 2.8553030303030304e-05, "loss": 6.638, "step": 697 }, { "epoch": 0.78, "learning_rate": 2.8545454545454548e-05, "loss": 6.4661, "step": 698 }, { "epoch": 0.78, "learning_rate": 2.853787878787879e-05, "loss": 6.3249, "step": 699 }, { "epoch": 0.78, "learning_rate": 2.853030303030303e-05, "loss": 6.2262, "step": 700 }, { "epoch": 0.79, "learning_rate": 2.8522727272727276e-05, "loss": 7.306, "step": 701 }, { "epoch": 0.79, "learning_rate": 2.8515151515151517e-05, "loss": 7.2799, "step": 702 }, { "epoch": 0.79, "learning_rate": 2.8507575757575758e-05, "loss": 7.2649, "step": 703 }, { "epoch": 0.79, "learning_rate": 2.8499999999999998e-05, "loss": 7.2053, "step": 704 }, { "epoch": 0.79, "learning_rate": 2.8492424242424246e-05, "loss": 7.199, "step": 705 }, { "epoch": 0.79, "learning_rate": 2.8484848484848486e-05, "loss": 7.1839, "step": 706 }, { "epoch": 0.79, "learning_rate": 2.8477272727272727e-05, "loss": 7.0621, "step": 707 }, { "epoch": 0.79, "learning_rate": 2.8469696969696967e-05, "loss": 7.0646, "step": 708 }, { "epoch": 0.79, "learning_rate": 2.8462121212121215e-05, "loss": 6.9529, "step": 709 }, { "epoch": 0.8, "learning_rate": 2.8454545454545455e-05, "loss": 6.843, "step": 710 }, { "epoch": 0.8, "learning_rate": 2.8446969696969696e-05, "loss": 6.8182, "step": 711 }, { "epoch": 0.8, "learning_rate": 2.8439393939393943e-05, "loss": 6.7262, "step": 712 }, { "epoch": 0.8, "learning_rate": 2.8431818181818184e-05, "loss": 6.7078, "step": 713 }, { "epoch": 0.8, "learning_rate": 2.8424242424242424e-05, "loss": 6.8404, "step": 714 }, { "epoch": 0.8, "learning_rate": 2.8416666666666665e-05, "loss": 6.8755, "step": 715 }, { "epoch": 0.8, "learning_rate": 2.8409090909090912e-05, "loss": 6.9271, "step": 716 }, { "epoch": 0.8, "learning_rate": 2.8401515151515153e-05, "loss": 7.0749, "step": 717 }, { "epoch": 0.8, "learning_rate": 2.8393939393939394e-05, "loss": 6.8647, "step": 718 }, { "epoch": 0.81, "learning_rate": 2.8386363636363634e-05, "loss": 6.6871, "step": 719 }, { "epoch": 0.81, "learning_rate": 2.837878787878788e-05, "loss": 6.7701, "step": 720 }, { "epoch": 0.81, "learning_rate": 2.8371212121212122e-05, "loss": 6.7868, "step": 721 }, { "epoch": 0.81, "learning_rate": 2.8363636363636363e-05, "loss": 6.74, "step": 722 }, { "epoch": 0.81, "learning_rate": 2.8356060606060607e-05, "loss": 6.688, "step": 723 }, { "epoch": 0.81, "learning_rate": 2.834848484848485e-05, "loss": 6.9508, "step": 724 }, { "epoch": 0.81, "learning_rate": 2.834090909090909e-05, "loss": 6.8659, "step": 725 }, { "epoch": 0.81, "learning_rate": 2.8333333333333332e-05, "loss": 6.9148, "step": 726 }, { "epoch": 0.82, "learning_rate": 2.8325757575757576e-05, "loss": 6.7847, "step": 727 }, { "epoch": 0.82, "learning_rate": 2.831818181818182e-05, "loss": 6.7245, "step": 728 }, { "epoch": 0.82, "learning_rate": 2.831060606060606e-05, "loss": 6.7244, "step": 729 }, { "epoch": 0.82, "learning_rate": 2.83030303030303e-05, "loss": 6.8705, "step": 730 }, { "epoch": 0.82, "learning_rate": 2.829545454545455e-05, "loss": 6.9539, "step": 731 }, { "epoch": 0.82, "learning_rate": 2.828787878787879e-05, "loss": 6.844, "step": 732 }, { "epoch": 0.82, "learning_rate": 2.828030303030303e-05, "loss": 6.8117, "step": 733 }, { "epoch": 0.82, "learning_rate": 2.8272727272727274e-05, "loss": 6.9797, "step": 734 }, { "epoch": 0.82, "learning_rate": 2.8265151515151518e-05, "loss": 6.6998, "step": 735 }, { "epoch": 0.83, "learning_rate": 2.8257575757575758e-05, "loss": 6.7844, "step": 736 }, { "epoch": 0.83, "learning_rate": 2.825e-05, "loss": 6.8331, "step": 737 }, { "epoch": 0.83, "learning_rate": 2.8242424242424243e-05, "loss": 6.8229, "step": 738 }, { "epoch": 0.83, "learning_rate": 2.8234848484848487e-05, "loss": 6.708, "step": 739 }, { "epoch": 0.83, "learning_rate": 2.8227272727272727e-05, "loss": 6.7854, "step": 740 }, { "epoch": 0.83, "learning_rate": 2.821969696969697e-05, "loss": 6.7564, "step": 741 }, { "epoch": 0.83, "learning_rate": 2.8212121212121212e-05, "loss": 6.817, "step": 742 }, { "epoch": 0.83, "learning_rate": 2.8204545454545456e-05, "loss": 6.5686, "step": 743 }, { "epoch": 0.83, "learning_rate": 2.8196969696969697e-05, "loss": 6.7598, "step": 744 }, { "epoch": 0.84, "learning_rate": 2.818939393939394e-05, "loss": 6.6466, "step": 745 }, { "epoch": 0.84, "learning_rate": 2.8181818181818185e-05, "loss": 6.7893, "step": 746 }, { "epoch": 0.84, "learning_rate": 2.8174242424242425e-05, "loss": 6.4812, "step": 747 }, { "epoch": 0.84, "learning_rate": 2.8166666666666666e-05, "loss": 6.3893, "step": 748 }, { "epoch": 0.84, "learning_rate": 2.815909090909091e-05, "loss": 6.3558, "step": 749 }, { "epoch": 0.84, "learning_rate": 2.8151515151515154e-05, "loss": 6.2476, "step": 750 }, { "epoch": 0.84, "learning_rate": 2.8143939393939394e-05, "loss": 7.1348, "step": 751 }, { "epoch": 0.84, "learning_rate": 2.813636363636364e-05, "loss": 7.2034, "step": 752 }, { "epoch": 0.84, "learning_rate": 2.812878787878788e-05, "loss": 7.246, "step": 753 }, { "epoch": 0.85, "learning_rate": 2.8121212121212123e-05, "loss": 7.1175, "step": 754 }, { "epoch": 0.85, "learning_rate": 2.8113636363636364e-05, "loss": 7.0929, "step": 755 }, { "epoch": 0.85, "learning_rate": 2.8106060606060607e-05, "loss": 7.0721, "step": 756 }, { "epoch": 0.85, "learning_rate": 2.8098484848484848e-05, "loss": 7.0036, "step": 757 }, { "epoch": 0.85, "learning_rate": 2.8090909090909092e-05, "loss": 6.8068, "step": 758 }, { "epoch": 0.85, "learning_rate": 2.8083333333333333e-05, "loss": 6.8836, "step": 759 }, { "epoch": 0.85, "learning_rate": 2.8075757575757577e-05, "loss": 6.8197, "step": 760 }, { "epoch": 0.85, "learning_rate": 2.8068181818181817e-05, "loss": 6.7779, "step": 761 }, { "epoch": 0.85, "learning_rate": 2.806060606060606e-05, "loss": 6.776, "step": 762 }, { "epoch": 0.86, "learning_rate": 2.8053030303030305e-05, "loss": 6.621, "step": 763 }, { "epoch": 0.86, "learning_rate": 2.8045454545454546e-05, "loss": 6.5803, "step": 764 }, { "epoch": 0.86, "learning_rate": 2.803787878787879e-05, "loss": 6.7886, "step": 765 }, { "epoch": 0.86, "learning_rate": 2.803030303030303e-05, "loss": 6.7141, "step": 766 }, { "epoch": 0.86, "learning_rate": 2.8022727272727274e-05, "loss": 6.7378, "step": 767 }, { "epoch": 0.86, "learning_rate": 2.8015151515151515e-05, "loss": 6.7547, "step": 768 }, { "epoch": 0.86, "learning_rate": 2.800757575757576e-05, "loss": 6.8721, "step": 769 }, { "epoch": 0.86, "learning_rate": 2.8e-05, "loss": 6.6842, "step": 770 }, { "epoch": 0.86, "learning_rate": 2.7992424242424244e-05, "loss": 6.8605, "step": 771 }, { "epoch": 0.87, "learning_rate": 2.7984848484848484e-05, "loss": 6.8976, "step": 772 }, { "epoch": 0.87, "learning_rate": 2.7977272727272728e-05, "loss": 6.8286, "step": 773 }, { "epoch": 0.87, "learning_rate": 2.7969696969696972e-05, "loss": 6.6636, "step": 774 }, { "epoch": 0.87, "learning_rate": 2.7962121212121213e-05, "loss": 6.5441, "step": 775 }, { "epoch": 0.87, "learning_rate": 2.7954545454545453e-05, "loss": 6.6575, "step": 776 }, { "epoch": 0.87, "learning_rate": 2.7946969696969697e-05, "loss": 6.6192, "step": 777 }, { "epoch": 0.87, "learning_rate": 2.793939393939394e-05, "loss": 6.7865, "step": 778 }, { "epoch": 0.87, "learning_rate": 2.7931818181818182e-05, "loss": 6.7239, "step": 779 }, { "epoch": 0.87, "learning_rate": 2.7924242424242426e-05, "loss": 6.7283, "step": 780 }, { "epoch": 0.88, "learning_rate": 2.7916666666666666e-05, "loss": 6.7262, "step": 781 }, { "epoch": 0.88, "learning_rate": 2.790909090909091e-05, "loss": 6.7623, "step": 782 }, { "epoch": 0.88, "learning_rate": 2.790151515151515e-05, "loss": 6.8005, "step": 783 }, { "epoch": 0.88, "learning_rate": 2.7893939393939395e-05, "loss": 6.6546, "step": 784 }, { "epoch": 0.88, "learning_rate": 2.788636363636364e-05, "loss": 6.7347, "step": 785 }, { "epoch": 0.88, "learning_rate": 2.787878787878788e-05, "loss": 6.6833, "step": 786 }, { "epoch": 0.88, "learning_rate": 2.787121212121212e-05, "loss": 6.7398, "step": 787 }, { "epoch": 0.88, "learning_rate": 2.7863636363636364e-05, "loss": 6.7808, "step": 788 }, { "epoch": 0.88, "learning_rate": 2.7856060606060608e-05, "loss": 6.7221, "step": 789 }, { "epoch": 0.89, "learning_rate": 2.784848484848485e-05, "loss": 6.6886, "step": 790 }, { "epoch": 0.89, "learning_rate": 2.784090909090909e-05, "loss": 6.6355, "step": 791 }, { "epoch": 0.89, "learning_rate": 2.7833333333333337e-05, "loss": 6.5175, "step": 792 }, { "epoch": 0.89, "learning_rate": 2.7825757575757577e-05, "loss": 6.622, "step": 793 }, { "epoch": 0.89, "learning_rate": 2.7818181818181818e-05, "loss": 6.5694, "step": 794 }, { "epoch": 0.89, "learning_rate": 2.781060606060606e-05, "loss": 6.6603, "step": 795 }, { "epoch": 0.89, "learning_rate": 2.7803030303030306e-05, "loss": 6.4916, "step": 796 }, { "epoch": 0.89, "learning_rate": 2.7795454545454547e-05, "loss": 6.5314, "step": 797 }, { "epoch": 0.89, "learning_rate": 2.7787878787878787e-05, "loss": 6.4234, "step": 798 }, { "epoch": 0.9, "learning_rate": 2.778030303030303e-05, "loss": 6.485, "step": 799 }, { "epoch": 0.9, "learning_rate": 2.7772727272727275e-05, "loss": 6.0093, "step": 800 }, { "epoch": 0.9, "learning_rate": 2.7765151515151516e-05, "loss": 7.0658, "step": 801 }, { "epoch": 0.9, "learning_rate": 2.7757575757575756e-05, "loss": 7.0436, "step": 802 }, { "epoch": 0.9, "learning_rate": 2.7750000000000004e-05, "loss": 7.1325, "step": 803 }, { "epoch": 0.9, "learning_rate": 2.7742424242424244e-05, "loss": 7.0685, "step": 804 }, { "epoch": 0.9, "learning_rate": 2.7734848484848485e-05, "loss": 6.8217, "step": 805 }, { "epoch": 0.9, "learning_rate": 2.7727272727272725e-05, "loss": 6.9695, "step": 806 }, { "epoch": 0.9, "learning_rate": 2.7719696969696973e-05, "loss": 7.0004, "step": 807 }, { "epoch": 0.91, "learning_rate": 2.7712121212121213e-05, "loss": 6.6772, "step": 808 }, { "epoch": 0.91, "learning_rate": 2.7704545454545454e-05, "loss": 6.8208, "step": 809 }, { "epoch": 0.91, "learning_rate": 2.7696969696969695e-05, "loss": 6.7359, "step": 810 }, { "epoch": 0.91, "learning_rate": 2.7689393939393942e-05, "loss": 6.8195, "step": 811 }, { "epoch": 0.91, "learning_rate": 2.7681818181818183e-05, "loss": 6.8647, "step": 812 }, { "epoch": 0.91, "learning_rate": 2.7674242424242423e-05, "loss": 6.6528, "step": 813 }, { "epoch": 0.91, "learning_rate": 2.766666666666667e-05, "loss": 6.7528, "step": 814 }, { "epoch": 0.91, "learning_rate": 2.765909090909091e-05, "loss": 6.7955, "step": 815 }, { "epoch": 0.91, "learning_rate": 2.7651515151515152e-05, "loss": 6.6911, "step": 816 }, { "epoch": 0.92, "learning_rate": 2.7643939393939392e-05, "loss": 6.8393, "step": 817 }, { "epoch": 0.92, "learning_rate": 2.763636363636364e-05, "loss": 6.7553, "step": 818 }, { "epoch": 0.92, "learning_rate": 2.762878787878788e-05, "loss": 6.6556, "step": 819 }, { "epoch": 0.92, "learning_rate": 2.762121212121212e-05, "loss": 6.8734, "step": 820 }, { "epoch": 0.92, "learning_rate": 2.761363636363636e-05, "loss": 6.766, "step": 821 }, { "epoch": 0.92, "learning_rate": 2.760606060606061e-05, "loss": 6.6906, "step": 822 }, { "epoch": 0.92, "learning_rate": 2.759848484848485e-05, "loss": 6.8203, "step": 823 }, { "epoch": 0.92, "learning_rate": 2.759090909090909e-05, "loss": 6.7373, "step": 824 }, { "epoch": 0.92, "learning_rate": 2.7583333333333334e-05, "loss": 6.9338, "step": 825 }, { "epoch": 0.93, "learning_rate": 2.7575757575757578e-05, "loss": 6.8177, "step": 826 }, { "epoch": 0.93, "learning_rate": 2.756818181818182e-05, "loss": 6.6787, "step": 827 }, { "epoch": 0.93, "learning_rate": 2.756060606060606e-05, "loss": 6.6088, "step": 828 }, { "epoch": 0.93, "learning_rate": 2.7553030303030303e-05, "loss": 6.6691, "step": 829 }, { "epoch": 0.93, "learning_rate": 2.7545454545454547e-05, "loss": 6.6751, "step": 830 }, { "epoch": 0.93, "learning_rate": 2.7537878787878788e-05, "loss": 6.6552, "step": 831 }, { "epoch": 0.93, "learning_rate": 2.753030303030303e-05, "loss": 6.6665, "step": 832 }, { "epoch": 0.93, "learning_rate": 2.7522727272727276e-05, "loss": 6.7171, "step": 833 }, { "epoch": 0.93, "learning_rate": 2.7515151515151516e-05, "loss": 6.7268, "step": 834 }, { "epoch": 0.94, "learning_rate": 2.7507575757575757e-05, "loss": 6.6984, "step": 835 }, { "epoch": 0.94, "learning_rate": 2.75e-05, "loss": 6.8598, "step": 836 }, { "epoch": 0.94, "learning_rate": 2.7492424242424245e-05, "loss": 6.7034, "step": 837 }, { "epoch": 0.94, "learning_rate": 2.7484848484848486e-05, "loss": 6.7362, "step": 838 }, { "epoch": 0.94, "learning_rate": 2.7477272727272726e-05, "loss": 6.6249, "step": 839 }, { "epoch": 0.94, "learning_rate": 2.746969696969697e-05, "loss": 6.5884, "step": 840 }, { "epoch": 0.94, "learning_rate": 2.7462121212121214e-05, "loss": 6.7509, "step": 841 }, { "epoch": 0.94, "learning_rate": 2.7454545454545455e-05, "loss": 6.6089, "step": 842 }, { "epoch": 0.95, "learning_rate": 2.74469696969697e-05, "loss": 6.8511, "step": 843 }, { "epoch": 0.95, "learning_rate": 2.743939393939394e-05, "loss": 6.6099, "step": 844 }, { "epoch": 0.95, "learning_rate": 2.7431818181818183e-05, "loss": 6.5816, "step": 845 }, { "epoch": 0.95, "learning_rate": 2.7424242424242424e-05, "loss": 6.5209, "step": 846 }, { "epoch": 0.95, "learning_rate": 2.7416666666666668e-05, "loss": 6.3091, "step": 847 }, { "epoch": 0.95, "learning_rate": 2.7409090909090912e-05, "loss": 6.3801, "step": 848 }, { "epoch": 0.95, "learning_rate": 2.7401515151515152e-05, "loss": 6.333, "step": 849 }, { "epoch": 0.95, "learning_rate": 2.7393939393939393e-05, "loss": 5.8617, "step": 850 }, { "epoch": 0.95, "learning_rate": 2.7386363636363637e-05, "loss": 7.1265, "step": 851 }, { "epoch": 0.96, "learning_rate": 2.737878787878788e-05, "loss": 7.029, "step": 852 }, { "epoch": 0.96, "learning_rate": 2.737121212121212e-05, "loss": 7.2721, "step": 853 }, { "epoch": 0.96, "learning_rate": 2.7363636363636366e-05, "loss": 7.2135, "step": 854 }, { "epoch": 0.96, "learning_rate": 2.7356060606060606e-05, "loss": 7.0014, "step": 855 }, { "epoch": 0.96, "learning_rate": 2.734848484848485e-05, "loss": 7.0637, "step": 856 }, { "epoch": 0.96, "learning_rate": 2.734090909090909e-05, "loss": 7.0073, "step": 857 }, { "epoch": 0.96, "learning_rate": 2.7333333333333335e-05, "loss": 6.8891, "step": 858 }, { "epoch": 0.96, "learning_rate": 2.7325757575757575e-05, "loss": 6.6923, "step": 859 }, { "epoch": 0.96, "learning_rate": 2.731818181818182e-05, "loss": 6.7991, "step": 860 }, { "epoch": 0.97, "learning_rate": 2.731060606060606e-05, "loss": 6.9208, "step": 861 }, { "epoch": 0.97, "learning_rate": 2.7303030303030304e-05, "loss": 6.784, "step": 862 }, { "epoch": 0.97, "learning_rate": 2.7295454545454545e-05, "loss": 6.7121, "step": 863 }, { "epoch": 0.97, "learning_rate": 2.728787878787879e-05, "loss": 6.9096, "step": 864 }, { "epoch": 0.97, "learning_rate": 2.7280303030303033e-05, "loss": 6.8366, "step": 865 }, { "epoch": 0.97, "learning_rate": 2.7272727272727273e-05, "loss": 6.6548, "step": 866 }, { "epoch": 0.97, "learning_rate": 2.7265151515151517e-05, "loss": 6.8529, "step": 867 }, { "epoch": 0.97, "learning_rate": 2.7257575757575758e-05, "loss": 6.6905, "step": 868 }, { "epoch": 0.97, "learning_rate": 2.725e-05, "loss": 6.7369, "step": 869 }, { "epoch": 0.98, "learning_rate": 2.7242424242424242e-05, "loss": 6.7093, "step": 870 }, { "epoch": 0.98, "learning_rate": 2.7234848484848486e-05, "loss": 6.6294, "step": 871 }, { "epoch": 0.98, "learning_rate": 2.7227272727272727e-05, "loss": 6.7178, "step": 872 }, { "epoch": 0.98, "learning_rate": 2.721969696969697e-05, "loss": 6.7482, "step": 873 }, { "epoch": 0.98, "learning_rate": 2.721212121212121e-05, "loss": 6.7552, "step": 874 }, { "epoch": 0.98, "learning_rate": 2.7204545454545455e-05, "loss": 6.8323, "step": 875 }, { "epoch": 0.98, "learning_rate": 2.71969696969697e-05, "loss": 6.6095, "step": 876 }, { "epoch": 0.98, "learning_rate": 2.718939393939394e-05, "loss": 6.6004, "step": 877 }, { "epoch": 0.98, "learning_rate": 2.718181818181818e-05, "loss": 6.6428, "step": 878 }, { "epoch": 0.99, "learning_rate": 2.7174242424242425e-05, "loss": 6.7186, "step": 879 }, { "epoch": 0.99, "learning_rate": 2.716666666666667e-05, "loss": 6.7986, "step": 880 }, { "epoch": 0.99, "learning_rate": 2.715909090909091e-05, "loss": 6.5193, "step": 881 }, { "epoch": 0.99, "learning_rate": 2.7151515151515153e-05, "loss": 6.639, "step": 882 }, { "epoch": 0.99, "learning_rate": 2.7143939393939394e-05, "loss": 6.5754, "step": 883 }, { "epoch": 0.99, "learning_rate": 2.7136363636363638e-05, "loss": 6.6453, "step": 884 }, { "epoch": 0.99, "learning_rate": 2.712878787878788e-05, "loss": 6.6327, "step": 885 }, { "epoch": 0.99, "learning_rate": 2.7121212121212122e-05, "loss": 6.7297, "step": 886 }, { "epoch": 0.99, "learning_rate": 2.7113636363636366e-05, "loss": 6.3966, "step": 887 }, { "epoch": 1.0, "learning_rate": 2.7106060606060607e-05, "loss": 6.5328, "step": 888 }, { "epoch": 1.0, "learning_rate": 2.7098484848484848e-05, "loss": 6.5383, "step": 889 }, { "epoch": 1.0, "learning_rate": 2.709090909090909e-05, "loss": 6.2863, "step": 890 }, { "epoch": 1.0, "learning_rate": 2.7083333333333335e-05, "loss": 6.1991, "step": 891 }, { "epoch": 1.0, "learning_rate": 2.7075757575757576e-05, "loss": 6.1719, "step": 892 }, { "epoch": 1.0, "learning_rate": 2.7068181818181817e-05, "loss": 6.9742, "step": 893 }, { "epoch": 1.0, "learning_rate": 2.706060606060606e-05, "loss": 7.0636, "step": 894 }, { "epoch": 1.0, "learning_rate": 2.7053030303030305e-05, "loss": 6.7303, "step": 895 }, { "epoch": 1.0, "learning_rate": 2.7045454545454545e-05, "loss": 6.7463, "step": 896 }, { "epoch": 1.01, "learning_rate": 2.7037878787878786e-05, "loss": 6.6953, "step": 897 }, { "epoch": 1.01, "learning_rate": 2.7030303030303033e-05, "loss": 6.8204, "step": 898 }, { "epoch": 1.01, "learning_rate": 2.7022727272727274e-05, "loss": 6.7014, "step": 899 }, { "epoch": 1.01, "learning_rate": 2.7015151515151514e-05, "loss": 6.774, "step": 900 }, { "epoch": 1.01, "learning_rate": 2.700757575757576e-05, "loss": 6.7822, "step": 901 }, { "epoch": 1.01, "learning_rate": 2.7000000000000002e-05, "loss": 6.8193, "step": 902 }, { "epoch": 1.01, "learning_rate": 2.6992424242424243e-05, "loss": 6.5925, "step": 903 }, { "epoch": 1.01, "learning_rate": 2.6984848484848484e-05, "loss": 6.7828, "step": 904 }, { "epoch": 1.01, "learning_rate": 2.697727272727273e-05, "loss": 6.5804, "step": 905 }, { "epoch": 1.02, "learning_rate": 2.696969696969697e-05, "loss": 6.6881, "step": 906 }, { "epoch": 1.02, "learning_rate": 2.6962121212121212e-05, "loss": 6.8107, "step": 907 }, { "epoch": 1.02, "learning_rate": 2.6954545454545453e-05, "loss": 6.8302, "step": 908 }, { "epoch": 1.02, "learning_rate": 2.69469696969697e-05, "loss": 6.6885, "step": 909 }, { "epoch": 1.02, "learning_rate": 2.693939393939394e-05, "loss": 6.5457, "step": 910 }, { "epoch": 1.02, "learning_rate": 2.693181818181818e-05, "loss": 6.6074, "step": 911 }, { "epoch": 1.02, "learning_rate": 2.6924242424242422e-05, "loss": 6.5991, "step": 912 }, { "epoch": 1.02, "learning_rate": 2.691666666666667e-05, "loss": 6.6272, "step": 913 }, { "epoch": 1.02, "learning_rate": 2.690909090909091e-05, "loss": 6.679, "step": 914 }, { "epoch": 1.03, "learning_rate": 2.690151515151515e-05, "loss": 6.4529, "step": 915 }, { "epoch": 1.03, "learning_rate": 2.6893939393939398e-05, "loss": 6.6102, "step": 916 }, { "epoch": 1.03, "learning_rate": 2.688636363636364e-05, "loss": 6.5549, "step": 917 }, { "epoch": 1.03, "learning_rate": 2.687878787878788e-05, "loss": 6.6926, "step": 918 }, { "epoch": 1.03, "learning_rate": 2.687121212121212e-05, "loss": 6.7019, "step": 919 }, { "epoch": 1.03, "learning_rate": 2.6863636363636367e-05, "loss": 6.6112, "step": 920 }, { "epoch": 1.03, "learning_rate": 2.6856060606060608e-05, "loss": 6.5809, "step": 921 }, { "epoch": 1.03, "learning_rate": 2.6848484848484848e-05, "loss": 6.8535, "step": 922 }, { "epoch": 1.03, "learning_rate": 2.684090909090909e-05, "loss": 6.7082, "step": 923 }, { "epoch": 1.04, "learning_rate": 2.6833333333333336e-05, "loss": 6.7171, "step": 924 }, { "epoch": 1.04, "learning_rate": 2.6825757575757577e-05, "loss": 6.647, "step": 925 }, { "epoch": 1.04, "learning_rate": 2.6818181818181817e-05, "loss": 6.6273, "step": 926 }, { "epoch": 1.04, "learning_rate": 2.681060606060606e-05, "loss": 6.6784, "step": 927 }, { "epoch": 1.04, "learning_rate": 2.6803030303030305e-05, "loss": 6.6241, "step": 928 }, { "epoch": 1.04, "learning_rate": 2.6795454545454546e-05, "loss": 6.6371, "step": 929 }, { "epoch": 1.04, "learning_rate": 2.6787878787878787e-05, "loss": 6.7507, "step": 930 }, { "epoch": 1.04, "learning_rate": 2.678030303030303e-05, "loss": 6.5986, "step": 931 }, { "epoch": 1.04, "learning_rate": 2.6772727272727275e-05, "loss": 6.5318, "step": 932 }, { "epoch": 1.05, "learning_rate": 2.6765151515151515e-05, "loss": 6.5305, "step": 933 }, { "epoch": 1.05, "learning_rate": 2.6757575757575756e-05, "loss": 6.7093, "step": 934 }, { "epoch": 1.05, "learning_rate": 2.6750000000000003e-05, "loss": 6.4997, "step": 935 }, { "epoch": 1.05, "learning_rate": 2.6742424242424244e-05, "loss": 6.4198, "step": 936 }, { "epoch": 1.05, "learning_rate": 2.6734848484848484e-05, "loss": 6.5809, "step": 937 }, { "epoch": 1.05, "learning_rate": 2.6727272727272728e-05, "loss": 6.3639, "step": 938 }, { "epoch": 1.05, "learning_rate": 2.6719696969696972e-05, "loss": 6.1069, "step": 939 }, { "epoch": 1.05, "learning_rate": 2.6712121212121213e-05, "loss": 6.326, "step": 940 }, { "epoch": 1.05, "learning_rate": 2.6704545454545453e-05, "loss": 5.9753, "step": 941 }, { "epoch": 1.06, "learning_rate": 2.6696969696969697e-05, "loss": 6.0674, "step": 942 }, { "epoch": 1.06, "learning_rate": 2.668939393939394e-05, "loss": 6.8791, "step": 943 }, { "epoch": 1.06, "learning_rate": 2.6681818181818182e-05, "loss": 6.8847, "step": 944 }, { "epoch": 1.06, "learning_rate": 2.6674242424242426e-05, "loss": 7.1174, "step": 945 }, { "epoch": 1.06, "learning_rate": 2.6666666666666667e-05, "loss": 7.0578, "step": 946 }, { "epoch": 1.06, "learning_rate": 2.665909090909091e-05, "loss": 7.2107, "step": 947 }, { "epoch": 1.06, "learning_rate": 2.665151515151515e-05, "loss": 7.2078, "step": 948 }, { "epoch": 1.06, "learning_rate": 2.6643939393939395e-05, "loss": 6.9677, "step": 949 }, { "epoch": 1.07, "learning_rate": 2.663636363636364e-05, "loss": 6.9737, "step": 950 }, { "epoch": 1.07, "learning_rate": 2.662878787878788e-05, "loss": 6.6288, "step": 951 }, { "epoch": 1.07, "learning_rate": 2.662121212121212e-05, "loss": 6.846, "step": 952 }, { "epoch": 1.07, "learning_rate": 2.6613636363636364e-05, "loss": 6.7793, "step": 953 }, { "epoch": 1.07, "learning_rate": 2.660606060606061e-05, "loss": 6.6556, "step": 954 }, { "epoch": 1.07, "learning_rate": 2.659848484848485e-05, "loss": 6.892, "step": 955 }, { "epoch": 1.07, "learning_rate": 2.6590909090909093e-05, "loss": 6.6987, "step": 956 }, { "epoch": 1.07, "learning_rate": 2.6583333333333333e-05, "loss": 6.7238, "step": 957 }, { "epoch": 1.07, "learning_rate": 2.6575757575757577e-05, "loss": 6.7182, "step": 958 }, { "epoch": 1.08, "learning_rate": 2.6568181818181818e-05, "loss": 6.7265, "step": 959 }, { "epoch": 1.08, "learning_rate": 2.6560606060606062e-05, "loss": 6.5642, "step": 960 }, { "epoch": 1.08, "learning_rate": 2.6553030303030303e-05, "loss": 6.6598, "step": 961 }, { "epoch": 1.08, "learning_rate": 2.6545454545454547e-05, "loss": 6.7631, "step": 962 }, { "epoch": 1.08, "learning_rate": 2.6537878787878787e-05, "loss": 6.7984, "step": 963 }, { "epoch": 1.08, "learning_rate": 2.653030303030303e-05, "loss": 6.6623, "step": 964 }, { "epoch": 1.08, "learning_rate": 2.6522727272727272e-05, "loss": 6.6686, "step": 965 }, { "epoch": 1.08, "learning_rate": 2.6515151515151516e-05, "loss": 6.6619, "step": 966 }, { "epoch": 1.08, "learning_rate": 2.650757575757576e-05, "loss": 6.5895, "step": 967 }, { "epoch": 1.09, "learning_rate": 2.65e-05, "loss": 6.5362, "step": 968 }, { "epoch": 1.09, "learning_rate": 2.6492424242424244e-05, "loss": 6.7139, "step": 969 }, { "epoch": 1.09, "learning_rate": 2.6484848484848485e-05, "loss": 6.6055, "step": 970 }, { "epoch": 1.09, "learning_rate": 2.647727272727273e-05, "loss": 6.6339, "step": 971 }, { "epoch": 1.09, "learning_rate": 2.646969696969697e-05, "loss": 6.682, "step": 972 }, { "epoch": 1.09, "learning_rate": 2.6462121212121214e-05, "loss": 6.7208, "step": 973 }, { "epoch": 1.09, "learning_rate": 2.6454545454545454e-05, "loss": 6.7157, "step": 974 }, { "epoch": 1.09, "learning_rate": 2.6446969696969698e-05, "loss": 6.5989, "step": 975 }, { "epoch": 1.09, "learning_rate": 2.643939393939394e-05, "loss": 6.5975, "step": 976 }, { "epoch": 1.1, "learning_rate": 2.6431818181818183e-05, "loss": 6.8275, "step": 977 }, { "epoch": 1.1, "learning_rate": 2.6424242424242427e-05, "loss": 6.5399, "step": 978 }, { "epoch": 1.1, "learning_rate": 2.6416666666666667e-05, "loss": 6.5623, "step": 979 }, { "epoch": 1.1, "learning_rate": 2.6409090909090908e-05, "loss": 6.559, "step": 980 }, { "epoch": 1.1, "learning_rate": 2.6401515151515152e-05, "loss": 6.5813, "step": 981 }, { "epoch": 1.1, "learning_rate": 2.6393939393939396e-05, "loss": 6.75, "step": 982 }, { "epoch": 1.1, "learning_rate": 2.6386363636363636e-05, "loss": 6.7563, "step": 983 }, { "epoch": 1.1, "learning_rate": 2.637878787878788e-05, "loss": 6.5481, "step": 984 }, { "epoch": 1.1, "learning_rate": 2.637121212121212e-05, "loss": 6.414, "step": 985 }, { "epoch": 1.11, "learning_rate": 2.6363636363636365e-05, "loss": 6.5106, "step": 986 }, { "epoch": 1.11, "learning_rate": 2.6356060606060606e-05, "loss": 6.4137, "step": 987 }, { "epoch": 1.11, "learning_rate": 2.634848484848485e-05, "loss": 6.2457, "step": 988 }, { "epoch": 1.11, "learning_rate": 2.6340909090909094e-05, "loss": 6.1434, "step": 989 }, { "epoch": 1.11, "learning_rate": 2.6333333333333334e-05, "loss": 6.223, "step": 990 }, { "epoch": 1.11, "learning_rate": 2.6325757575757575e-05, "loss": 5.7618, "step": 991 }, { "epoch": 1.11, "learning_rate": 2.631818181818182e-05, "loss": 5.5582, "step": 992 }, { "epoch": 1.11, "learning_rate": 2.6310606060606063e-05, "loss": 6.6504, "step": 993 }, { "epoch": 1.11, "learning_rate": 2.6303030303030303e-05, "loss": 6.7755, "step": 994 }, { "epoch": 1.12, "learning_rate": 2.6295454545454544e-05, "loss": 6.7547, "step": 995 }, { "epoch": 1.12, "learning_rate": 2.6287878787878788e-05, "loss": 6.8227, "step": 996 }, { "epoch": 1.12, "learning_rate": 2.6280303030303032e-05, "loss": 6.6377, "step": 997 }, { "epoch": 1.12, "learning_rate": 2.6272727272727273e-05, "loss": 6.8144, "step": 998 }, { "epoch": 1.12, "learning_rate": 2.6265151515151513e-05, "loss": 6.6482, "step": 999 }, { "epoch": 1.12, "learning_rate": 2.625757575757576e-05, "loss": 6.5068, "step": 1000 }, { "epoch": 1.12, "eval_loss": 6.482268333435059, "eval_runtime": 859.3494, "eval_samples_per_second": 3.074, "eval_steps_per_second": 0.385, "eval_wer": 1.3923244744149148, "step": 1000 }, { "epoch": 1.12, "learning_rate": 2.625e-05, "loss": 6.6951, "step": 1001 }, { "epoch": 1.12, "learning_rate": 2.624242424242424e-05, "loss": 6.6909, "step": 1002 }, { "epoch": 1.12, "learning_rate": 2.6234848484848486e-05, "loss": 6.5886, "step": 1003 }, { "epoch": 1.13, "learning_rate": 2.622727272727273e-05, "loss": 6.7534, "step": 1004 }, { "epoch": 1.13, "learning_rate": 2.621969696969697e-05, "loss": 6.7156, "step": 1005 }, { "epoch": 1.13, "learning_rate": 2.621212121212121e-05, "loss": 6.6131, "step": 1006 }, { "epoch": 1.13, "learning_rate": 2.6204545454545458e-05, "loss": 6.5765, "step": 1007 }, { "epoch": 1.13, "learning_rate": 2.61969696969697e-05, "loss": 6.661, "step": 1008 }, { "epoch": 1.13, "learning_rate": 2.618939393939394e-05, "loss": 6.6822, "step": 1009 }, { "epoch": 1.13, "learning_rate": 2.618181818181818e-05, "loss": 6.6625, "step": 1010 }, { "epoch": 1.13, "learning_rate": 2.6174242424242427e-05, "loss": 6.6155, "step": 1011 }, { "epoch": 1.13, "learning_rate": 2.6166666666666668e-05, "loss": 6.7312, "step": 1012 }, { "epoch": 1.14, "learning_rate": 2.615909090909091e-05, "loss": 6.5661, "step": 1013 }, { "epoch": 1.14, "learning_rate": 2.615151515151515e-05, "loss": 6.5899, "step": 1014 }, { "epoch": 1.14, "learning_rate": 2.6143939393939397e-05, "loss": 6.5424, "step": 1015 }, { "epoch": 1.14, "learning_rate": 2.6136363636363637e-05, "loss": 6.5411, "step": 1016 }, { "epoch": 1.14, "learning_rate": 2.6128787878787878e-05, "loss": 6.5202, "step": 1017 }, { "epoch": 1.14, "learning_rate": 2.6121212121212122e-05, "loss": 6.6637, "step": 1018 }, { "epoch": 1.14, "learning_rate": 2.6113636363636366e-05, "loss": 6.6558, "step": 1019 }, { "epoch": 1.14, "learning_rate": 2.6106060606060606e-05, "loss": 6.6943, "step": 1020 }, { "epoch": 1.14, "learning_rate": 2.6098484848484847e-05, "loss": 6.6101, "step": 1021 }, { "epoch": 1.15, "learning_rate": 2.6090909090909094e-05, "loss": 6.5388, "step": 1022 }, { "epoch": 1.15, "learning_rate": 2.6083333333333335e-05, "loss": 6.5799, "step": 1023 }, { "epoch": 1.15, "learning_rate": 2.6075757575757575e-05, "loss": 6.681, "step": 1024 }, { "epoch": 1.15, "learning_rate": 2.6068181818181816e-05, "loss": 6.4326, "step": 1025 }, { "epoch": 1.15, "learning_rate": 2.6060606060606063e-05, "loss": 6.4913, "step": 1026 }, { "epoch": 1.15, "learning_rate": 2.6053030303030304e-05, "loss": 6.7244, "step": 1027 }, { "epoch": 1.15, "learning_rate": 2.6045454545454545e-05, "loss": 6.5176, "step": 1028 }, { "epoch": 1.15, "learning_rate": 2.603787878787879e-05, "loss": 6.4077, "step": 1029 }, { "epoch": 1.15, "learning_rate": 2.6030303030303033e-05, "loss": 6.5917, "step": 1030 }, { "epoch": 1.16, "learning_rate": 2.6022727272727273e-05, "loss": 6.4544, "step": 1031 }, { "epoch": 1.16, "learning_rate": 2.6015151515151514e-05, "loss": 6.5218, "step": 1032 }, { "epoch": 1.16, "learning_rate": 2.6007575757575758e-05, "loss": 6.5526, "step": 1033 }, { "epoch": 1.16, "learning_rate": 2.6000000000000002e-05, "loss": 6.4643, "step": 1034 }, { "epoch": 1.16, "learning_rate": 2.5992424242424242e-05, "loss": 6.5597, "step": 1035 }, { "epoch": 1.16, "learning_rate": 2.5984848484848483e-05, "loss": 6.4667, "step": 1036 }, { "epoch": 1.16, "learning_rate": 2.597727272727273e-05, "loss": 6.3499, "step": 1037 }, { "epoch": 1.16, "learning_rate": 2.596969696969697e-05, "loss": 6.4928, "step": 1038 }, { "epoch": 1.16, "learning_rate": 2.596212121212121e-05, "loss": 6.253, "step": 1039 }, { "epoch": 1.17, "learning_rate": 2.5954545454545456e-05, "loss": 6.206, "step": 1040 }, { "epoch": 1.17, "learning_rate": 2.59469696969697e-05, "loss": 5.8702, "step": 1041 }, { "epoch": 1.17, "learning_rate": 2.593939393939394e-05, "loss": 5.4373, "step": 1042 }, { "epoch": 1.17, "learning_rate": 2.593181818181818e-05, "loss": 7.062, "step": 1043 }, { "epoch": 1.17, "learning_rate": 2.5924242424242425e-05, "loss": 6.9461, "step": 1044 }, { "epoch": 1.17, "learning_rate": 2.591666666666667e-05, "loss": 7.1466, "step": 1045 }, { "epoch": 1.17, "learning_rate": 2.590909090909091e-05, "loss": 7.1245, "step": 1046 }, { "epoch": 1.17, "learning_rate": 2.5901515151515153e-05, "loss": 7.0992, "step": 1047 }, { "epoch": 1.17, "learning_rate": 2.5893939393939394e-05, "loss": 6.9411, "step": 1048 }, { "epoch": 1.18, "learning_rate": 2.5886363636363638e-05, "loss": 6.888, "step": 1049 }, { "epoch": 1.18, "learning_rate": 2.587878787878788e-05, "loss": 6.7803, "step": 1050 }, { "epoch": 1.18, "learning_rate": 2.5871212121212122e-05, "loss": 6.7519, "step": 1051 }, { "epoch": 1.18, "learning_rate": 2.5863636363636363e-05, "loss": 6.8667, "step": 1052 }, { "epoch": 1.18, "learning_rate": 2.5856060606060607e-05, "loss": 6.6916, "step": 1053 }, { "epoch": 1.18, "learning_rate": 2.5848484848484848e-05, "loss": 6.8212, "step": 1054 }, { "epoch": 1.18, "learning_rate": 2.584090909090909e-05, "loss": 6.562, "step": 1055 }, { "epoch": 1.18, "learning_rate": 2.5833333333333336e-05, "loss": 6.6254, "step": 1056 }, { "epoch": 1.18, "learning_rate": 2.5825757575757576e-05, "loss": 6.7135, "step": 1057 }, { "epoch": 1.19, "learning_rate": 2.581818181818182e-05, "loss": 6.7497, "step": 1058 }, { "epoch": 1.19, "learning_rate": 2.581060606060606e-05, "loss": 6.6196, "step": 1059 }, { "epoch": 1.19, "learning_rate": 2.5803030303030305e-05, "loss": 6.4851, "step": 1060 }, { "epoch": 1.19, "learning_rate": 2.5795454545454545e-05, "loss": 6.5653, "step": 1061 }, { "epoch": 1.19, "learning_rate": 2.578787878787879e-05, "loss": 6.6105, "step": 1062 }, { "epoch": 1.19, "learning_rate": 2.578030303030303e-05, "loss": 6.5808, "step": 1063 }, { "epoch": 1.19, "learning_rate": 2.5772727272727274e-05, "loss": 6.6104, "step": 1064 }, { "epoch": 1.19, "learning_rate": 2.5765151515151515e-05, "loss": 6.6932, "step": 1065 }, { "epoch": 1.2, "learning_rate": 2.575757575757576e-05, "loss": 6.6468, "step": 1066 }, { "epoch": 1.2, "learning_rate": 2.575e-05, "loss": 6.5187, "step": 1067 }, { "epoch": 1.2, "learning_rate": 2.5742424242424243e-05, "loss": 6.5481, "step": 1068 }, { "epoch": 1.2, "learning_rate": 2.5734848484848487e-05, "loss": 6.7243, "step": 1069 }, { "epoch": 1.2, "learning_rate": 2.5727272727272728e-05, "loss": 6.6252, "step": 1070 }, { "epoch": 1.2, "learning_rate": 2.571969696969697e-05, "loss": 6.6215, "step": 1071 }, { "epoch": 1.2, "learning_rate": 2.5712121212121212e-05, "loss": 6.6989, "step": 1072 }, { "epoch": 1.2, "learning_rate": 2.5704545454545456e-05, "loss": 6.5251, "step": 1073 }, { "epoch": 1.2, "learning_rate": 2.5696969696969697e-05, "loss": 6.4698, "step": 1074 }, { "epoch": 1.21, "learning_rate": 2.568939393939394e-05, "loss": 6.4893, "step": 1075 }, { "epoch": 1.21, "learning_rate": 2.568181818181818e-05, "loss": 6.5649, "step": 1076 }, { "epoch": 1.21, "learning_rate": 2.5674242424242425e-05, "loss": 6.4526, "step": 1077 }, { "epoch": 1.21, "learning_rate": 2.5666666666666666e-05, "loss": 6.6422, "step": 1078 }, { "epoch": 1.21, "learning_rate": 2.565909090909091e-05, "loss": 6.6524, "step": 1079 }, { "epoch": 1.21, "learning_rate": 2.5651515151515154e-05, "loss": 6.526, "step": 1080 }, { "epoch": 1.21, "learning_rate": 2.5643939393939395e-05, "loss": 6.5175, "step": 1081 }, { "epoch": 1.21, "learning_rate": 2.5636363636363635e-05, "loss": 6.4897, "step": 1082 }, { "epoch": 1.21, "learning_rate": 2.562878787878788e-05, "loss": 6.5359, "step": 1083 }, { "epoch": 1.22, "learning_rate": 2.5621212121212123e-05, "loss": 6.462, "step": 1084 }, { "epoch": 1.22, "learning_rate": 2.5613636363636364e-05, "loss": 6.4337, "step": 1085 }, { "epoch": 1.22, "learning_rate": 2.5606060606060604e-05, "loss": 6.3802, "step": 1086 }, { "epoch": 1.22, "learning_rate": 2.559848484848485e-05, "loss": 6.3148, "step": 1087 }, { "epoch": 1.22, "learning_rate": 2.5590909090909092e-05, "loss": 6.4152, "step": 1088 }, { "epoch": 1.22, "learning_rate": 2.5583333333333333e-05, "loss": 6.2734, "step": 1089 }, { "epoch": 1.22, "learning_rate": 2.5575757575757577e-05, "loss": 6.042, "step": 1090 }, { "epoch": 1.22, "learning_rate": 2.556818181818182e-05, "loss": 6.0186, "step": 1091 }, { "epoch": 1.22, "learning_rate": 2.556060606060606e-05, "loss": 5.7229, "step": 1092 }, { "epoch": 1.23, "learning_rate": 2.5553030303030302e-05, "loss": 6.8215, "step": 1093 }, { "epoch": 1.23, "learning_rate": 2.5545454545454546e-05, "loss": 6.6024, "step": 1094 }, { "epoch": 1.23, "learning_rate": 2.553787878787879e-05, "loss": 6.7126, "step": 1095 }, { "epoch": 1.23, "learning_rate": 2.553030303030303e-05, "loss": 6.8177, "step": 1096 }, { "epoch": 1.23, "learning_rate": 2.552272727272727e-05, "loss": 6.8104, "step": 1097 }, { "epoch": 1.23, "learning_rate": 2.5515151515151515e-05, "loss": 6.693, "step": 1098 }, { "epoch": 1.23, "learning_rate": 2.550757575757576e-05, "loss": 6.6987, "step": 1099 }, { "epoch": 1.23, "learning_rate": 2.55e-05, "loss": 6.8174, "step": 1100 }, { "epoch": 1.23, "learning_rate": 2.549242424242424e-05, "loss": 6.6545, "step": 1101 }, { "epoch": 1.24, "learning_rate": 2.5484848484848488e-05, "loss": 6.5316, "step": 1102 }, { "epoch": 1.24, "learning_rate": 2.547727272727273e-05, "loss": 6.4843, "step": 1103 }, { "epoch": 1.24, "learning_rate": 2.546969696969697e-05, "loss": 6.6978, "step": 1104 }, { "epoch": 1.24, "learning_rate": 2.5462121212121213e-05, "loss": 6.6031, "step": 1105 }, { "epoch": 1.24, "learning_rate": 2.5454545454545457e-05, "loss": 6.569, "step": 1106 }, { "epoch": 1.24, "learning_rate": 2.5446969696969698e-05, "loss": 6.5567, "step": 1107 }, { "epoch": 1.24, "learning_rate": 2.5439393939393938e-05, "loss": 6.5549, "step": 1108 }, { "epoch": 1.24, "learning_rate": 2.5431818181818186e-05, "loss": 6.5183, "step": 1109 }, { "epoch": 1.24, "learning_rate": 2.5424242424242426e-05, "loss": 6.6444, "step": 1110 }, { "epoch": 1.25, "learning_rate": 2.5416666666666667e-05, "loss": 6.6752, "step": 1111 }, { "epoch": 1.25, "learning_rate": 2.5409090909090907e-05, "loss": 6.5876, "step": 1112 }, { "epoch": 1.25, "learning_rate": 2.5401515151515155e-05, "loss": 6.6476, "step": 1113 }, { "epoch": 1.25, "learning_rate": 2.5393939393939395e-05, "loss": 6.6023, "step": 1114 }, { "epoch": 1.25, "learning_rate": 2.5386363636363636e-05, "loss": 6.5756, "step": 1115 }, { "epoch": 1.25, "learning_rate": 2.5378787878787876e-05, "loss": 6.5043, "step": 1116 }, { "epoch": 1.25, "learning_rate": 2.5371212121212124e-05, "loss": 6.5081, "step": 1117 }, { "epoch": 1.25, "learning_rate": 2.5363636363636364e-05, "loss": 6.6765, "step": 1118 }, { "epoch": 1.25, "learning_rate": 2.5356060606060605e-05, "loss": 6.5346, "step": 1119 }, { "epoch": 1.26, "learning_rate": 2.534848484848485e-05, "loss": 6.4833, "step": 1120 }, { "epoch": 1.26, "learning_rate": 2.5340909090909093e-05, "loss": 6.596, "step": 1121 }, { "epoch": 1.26, "learning_rate": 2.5333333333333334e-05, "loss": 6.5829, "step": 1122 }, { "epoch": 1.26, "learning_rate": 2.5325757575757574e-05, "loss": 6.5067, "step": 1123 }, { "epoch": 1.26, "learning_rate": 2.531818181818182e-05, "loss": 6.5981, "step": 1124 }, { "epoch": 1.26, "learning_rate": 2.5310606060606062e-05, "loss": 6.5457, "step": 1125 }, { "epoch": 1.26, "learning_rate": 2.5303030303030303e-05, "loss": 6.6617, "step": 1126 }, { "epoch": 1.26, "learning_rate": 2.5295454545454543e-05, "loss": 6.5273, "step": 1127 }, { "epoch": 1.26, "learning_rate": 2.528787878787879e-05, "loss": 6.5045, "step": 1128 }, { "epoch": 1.27, "learning_rate": 2.528030303030303e-05, "loss": 6.6678, "step": 1129 }, { "epoch": 1.27, "learning_rate": 2.5272727272727272e-05, "loss": 6.4557, "step": 1130 }, { "epoch": 1.27, "learning_rate": 2.5265151515151516e-05, "loss": 6.5222, "step": 1131 }, { "epoch": 1.27, "learning_rate": 2.525757575757576e-05, "loss": 6.4297, "step": 1132 }, { "epoch": 1.27, "learning_rate": 2.525e-05, "loss": 6.56, "step": 1133 }, { "epoch": 1.27, "learning_rate": 2.524242424242424e-05, "loss": 6.5498, "step": 1134 }, { "epoch": 1.27, "learning_rate": 2.5234848484848485e-05, "loss": 6.1015, "step": 1135 }, { "epoch": 1.27, "learning_rate": 2.522727272727273e-05, "loss": 6.2317, "step": 1136 }, { "epoch": 1.27, "learning_rate": 2.521969696969697e-05, "loss": 6.4477, "step": 1137 }, { "epoch": 1.28, "learning_rate": 2.521212121212121e-05, "loss": 6.1762, "step": 1138 }, { "epoch": 1.28, "learning_rate": 2.5204545454545458e-05, "loss": 6.0463, "step": 1139 }, { "epoch": 1.28, "learning_rate": 2.5196969696969698e-05, "loss": 6.0327, "step": 1140 }, { "epoch": 1.28, "learning_rate": 2.518939393939394e-05, "loss": 6.0975, "step": 1141 }, { "epoch": 1.28, "learning_rate": 2.5181818181818183e-05, "loss": 5.9256, "step": 1142 }, { "epoch": 1.28, "learning_rate": 2.5174242424242427e-05, "loss": 7.1017, "step": 1143 }, { "epoch": 1.28, "learning_rate": 2.5166666666666667e-05, "loss": 6.9287, "step": 1144 }, { "epoch": 1.28, "learning_rate": 2.5159090909090908e-05, "loss": 6.9897, "step": 1145 }, { "epoch": 1.28, "learning_rate": 2.5151515151515152e-05, "loss": 7.0973, "step": 1146 }, { "epoch": 1.29, "learning_rate": 2.5143939393939396e-05, "loss": 6.9895, "step": 1147 }, { "epoch": 1.29, "learning_rate": 2.5136363636363637e-05, "loss": 6.9623, "step": 1148 }, { "epoch": 1.29, "learning_rate": 2.512878787878788e-05, "loss": 6.8364, "step": 1149 }, { "epoch": 1.29, "learning_rate": 2.512121212121212e-05, "loss": 6.7068, "step": 1150 }, { "epoch": 1.29, "learning_rate": 2.5113636363636365e-05, "loss": 6.6209, "step": 1151 }, { "epoch": 1.29, "learning_rate": 2.5106060606060606e-05, "loss": 6.528, "step": 1152 }, { "epoch": 1.29, "learning_rate": 2.509848484848485e-05, "loss": 6.631, "step": 1153 }, { "epoch": 1.29, "learning_rate": 2.509090909090909e-05, "loss": 6.7321, "step": 1154 }, { "epoch": 1.29, "learning_rate": 2.5083333333333334e-05, "loss": 6.5778, "step": 1155 }, { "epoch": 1.3, "learning_rate": 2.5075757575757575e-05, "loss": 6.5723, "step": 1156 }, { "epoch": 1.3, "learning_rate": 2.506818181818182e-05, "loss": 6.6036, "step": 1157 }, { "epoch": 1.3, "learning_rate": 2.5060606060606063e-05, "loss": 6.7276, "step": 1158 }, { "epoch": 1.3, "learning_rate": 2.5053030303030303e-05, "loss": 6.6307, "step": 1159 }, { "epoch": 1.3, "learning_rate": 2.5045454545454547e-05, "loss": 6.512, "step": 1160 }, { "epoch": 1.3, "learning_rate": 2.5037878787878788e-05, "loss": 6.5435, "step": 1161 }, { "epoch": 1.3, "learning_rate": 2.5030303030303032e-05, "loss": 6.6687, "step": 1162 }, { "epoch": 1.3, "learning_rate": 2.5022727272727273e-05, "loss": 6.5186, "step": 1163 }, { "epoch": 1.3, "learning_rate": 2.5015151515151517e-05, "loss": 6.4758, "step": 1164 }, { "epoch": 1.31, "learning_rate": 2.5007575757575757e-05, "loss": 6.5817, "step": 1165 }, { "epoch": 1.31, "learning_rate": 2.5e-05, "loss": 6.4284, "step": 1166 }, { "epoch": 1.31, "learning_rate": 2.4992424242424242e-05, "loss": 6.7649, "step": 1167 }, { "epoch": 1.31, "learning_rate": 2.4984848484848486e-05, "loss": 6.6287, "step": 1168 }, { "epoch": 1.31, "learning_rate": 2.4977272727272726e-05, "loss": 6.6955, "step": 1169 }, { "epoch": 1.31, "learning_rate": 2.496969696969697e-05, "loss": 6.6591, "step": 1170 }, { "epoch": 1.31, "learning_rate": 2.4962121212121214e-05, "loss": 6.6137, "step": 1171 }, { "epoch": 1.31, "learning_rate": 2.4954545454545455e-05, "loss": 6.6936, "step": 1172 }, { "epoch": 1.32, "learning_rate": 2.49469696969697e-05, "loss": 6.5402, "step": 1173 }, { "epoch": 1.32, "learning_rate": 2.493939393939394e-05, "loss": 6.5573, "step": 1174 }, { "epoch": 1.32, "learning_rate": 2.4931818181818184e-05, "loss": 6.5447, "step": 1175 }, { "epoch": 1.32, "learning_rate": 2.4924242424242424e-05, "loss": 6.6235, "step": 1176 }, { "epoch": 1.32, "learning_rate": 2.4916666666666668e-05, "loss": 6.6007, "step": 1177 }, { "epoch": 1.32, "learning_rate": 2.490909090909091e-05, "loss": 6.4877, "step": 1178 }, { "epoch": 1.32, "learning_rate": 2.4901515151515153e-05, "loss": 6.6202, "step": 1179 }, { "epoch": 1.32, "learning_rate": 2.4893939393939393e-05, "loss": 6.4485, "step": 1180 }, { "epoch": 1.32, "learning_rate": 2.4886363636363637e-05, "loss": 6.5057, "step": 1181 }, { "epoch": 1.33, "learning_rate": 2.487878787878788e-05, "loss": 6.5721, "step": 1182 }, { "epoch": 1.33, "learning_rate": 2.4871212121212122e-05, "loss": 6.533, "step": 1183 }, { "epoch": 1.33, "learning_rate": 2.4863636363636362e-05, "loss": 6.6082, "step": 1184 }, { "epoch": 1.33, "learning_rate": 2.4856060606060606e-05, "loss": 6.2381, "step": 1185 }, { "epoch": 1.33, "learning_rate": 2.484848484848485e-05, "loss": 6.2148, "step": 1186 }, { "epoch": 1.33, "learning_rate": 2.484090909090909e-05, "loss": 6.173, "step": 1187 }, { "epoch": 1.33, "learning_rate": 2.483333333333333e-05, "loss": 6.1948, "step": 1188 }, { "epoch": 1.33, "learning_rate": 2.4825757575757576e-05, "loss": 5.8943, "step": 1189 }, { "epoch": 1.33, "learning_rate": 2.481818181818182e-05, "loss": 5.7825, "step": 1190 }, { "epoch": 1.34, "learning_rate": 2.481060606060606e-05, "loss": 5.9438, "step": 1191 }, { "epoch": 1.34, "learning_rate": 2.4803030303030304e-05, "loss": 5.5114, "step": 1192 }, { "epoch": 1.34, "learning_rate": 2.4795454545454548e-05, "loss": 6.8513, "step": 1193 }, { "epoch": 1.34, "learning_rate": 2.478787878787879e-05, "loss": 6.9022, "step": 1194 }, { "epoch": 1.34, "learning_rate": 2.478030303030303e-05, "loss": 6.7898, "step": 1195 }, { "epoch": 1.34, "learning_rate": 2.4772727272727273e-05, "loss": 6.8622, "step": 1196 }, { "epoch": 1.34, "learning_rate": 2.4765151515151517e-05, "loss": 6.8559, "step": 1197 }, { "epoch": 1.34, "learning_rate": 2.4757575757575758e-05, "loss": 6.6843, "step": 1198 }, { "epoch": 1.34, "learning_rate": 2.475e-05, "loss": 6.8355, "step": 1199 }, { "epoch": 1.35, "learning_rate": 2.4742424242424242e-05, "loss": 6.7026, "step": 1200 }, { "epoch": 1.35, "learning_rate": 2.4734848484848486e-05, "loss": 6.6125, "step": 1201 }, { "epoch": 1.35, "learning_rate": 2.4727272727272727e-05, "loss": 6.6118, "step": 1202 }, { "epoch": 1.35, "learning_rate": 2.4719696969696968e-05, "loss": 6.4883, "step": 1203 }, { "epoch": 1.35, "learning_rate": 2.4712121212121215e-05, "loss": 6.5555, "step": 1204 }, { "epoch": 1.35, "learning_rate": 2.4704545454545456e-05, "loss": 6.5668, "step": 1205 }, { "epoch": 1.35, "learning_rate": 2.4696969696969696e-05, "loss": 6.421, "step": 1206 }, { "epoch": 1.35, "learning_rate": 2.468939393939394e-05, "loss": 6.5804, "step": 1207 }, { "epoch": 1.35, "learning_rate": 2.4681818181818184e-05, "loss": 6.566, "step": 1208 }, { "epoch": 1.36, "learning_rate": 2.4674242424242425e-05, "loss": 6.558, "step": 1209 }, { "epoch": 1.36, "learning_rate": 2.4666666666666665e-05, "loss": 6.6619, "step": 1210 }, { "epoch": 1.36, "learning_rate": 2.4659090909090913e-05, "loss": 6.6602, "step": 1211 }, { "epoch": 1.36, "learning_rate": 2.4651515151515153e-05, "loss": 6.5829, "step": 1212 }, { "epoch": 1.36, "learning_rate": 2.4643939393939394e-05, "loss": 6.7029, "step": 1213 }, { "epoch": 1.36, "learning_rate": 2.4636363636363635e-05, "loss": 6.4626, "step": 1214 }, { "epoch": 1.36, "learning_rate": 2.4628787878787882e-05, "loss": 6.6958, "step": 1215 }, { "epoch": 1.36, "learning_rate": 2.4621212121212123e-05, "loss": 6.5414, "step": 1216 }, { "epoch": 1.36, "learning_rate": 2.4613636363636363e-05, "loss": 6.5042, "step": 1217 }, { "epoch": 1.37, "learning_rate": 2.4606060606060604e-05, "loss": 6.5712, "step": 1218 }, { "epoch": 1.37, "learning_rate": 2.459848484848485e-05, "loss": 6.4932, "step": 1219 }, { "epoch": 1.37, "learning_rate": 2.4590909090909092e-05, "loss": 6.5781, "step": 1220 }, { "epoch": 1.37, "learning_rate": 2.4583333333333332e-05, "loss": 6.5244, "step": 1221 }, { "epoch": 1.37, "learning_rate": 2.4575757575757576e-05, "loss": 6.4316, "step": 1222 }, { "epoch": 1.37, "learning_rate": 2.456818181818182e-05, "loss": 6.4159, "step": 1223 }, { "epoch": 1.37, "learning_rate": 2.456060606060606e-05, "loss": 6.542, "step": 1224 }, { "epoch": 1.37, "learning_rate": 2.45530303030303e-05, "loss": 6.6044, "step": 1225 }, { "epoch": 1.37, "learning_rate": 2.454545454545455e-05, "loss": 6.5669, "step": 1226 }, { "epoch": 1.38, "learning_rate": 2.453787878787879e-05, "loss": 6.5374, "step": 1227 }, { "epoch": 1.38, "learning_rate": 2.453030303030303e-05, "loss": 6.5442, "step": 1228 }, { "epoch": 1.38, "learning_rate": 2.452272727272727e-05, "loss": 6.5453, "step": 1229 }, { "epoch": 1.38, "learning_rate": 2.4515151515151518e-05, "loss": 6.3758, "step": 1230 }, { "epoch": 1.38, "learning_rate": 2.450757575757576e-05, "loss": 6.4012, "step": 1231 }, { "epoch": 1.38, "learning_rate": 2.45e-05, "loss": 6.4423, "step": 1232 }, { "epoch": 1.38, "learning_rate": 2.4492424242424243e-05, "loss": 6.4526, "step": 1233 }, { "epoch": 1.38, "learning_rate": 2.4484848484848487e-05, "loss": 6.499, "step": 1234 }, { "epoch": 1.38, "learning_rate": 2.4477272727272728e-05, "loss": 6.4305, "step": 1235 }, { "epoch": 1.39, "learning_rate": 2.446969696969697e-05, "loss": 6.2684, "step": 1236 }, { "epoch": 1.39, "learning_rate": 2.4462121212121212e-05, "loss": 6.2699, "step": 1237 }, { "epoch": 1.39, "learning_rate": 2.4454545454545456e-05, "loss": 6.2412, "step": 1238 }, { "epoch": 1.39, "learning_rate": 2.4446969696969697e-05, "loss": 5.9701, "step": 1239 }, { "epoch": 1.39, "learning_rate": 2.4439393939393938e-05, "loss": 5.94, "step": 1240 }, { "epoch": 1.39, "learning_rate": 2.4431818181818185e-05, "loss": 5.6099, "step": 1241 }, { "epoch": 1.39, "learning_rate": 2.4424242424242426e-05, "loss": 5.6817, "step": 1242 }, { "epoch": 1.39, "learning_rate": 2.4416666666666666e-05, "loss": 7.0334, "step": 1243 }, { "epoch": 1.39, "learning_rate": 2.440909090909091e-05, "loss": 6.9049, "step": 1244 }, { "epoch": 1.4, "learning_rate": 2.4401515151515154e-05, "loss": 7.0357, "step": 1245 }, { "epoch": 1.4, "learning_rate": 2.4393939393939395e-05, "loss": 6.8491, "step": 1246 }, { "epoch": 1.4, "learning_rate": 2.4386363636363635e-05, "loss": 6.8852, "step": 1247 }, { "epoch": 1.4, "learning_rate": 2.437878787878788e-05, "loss": 6.8646, "step": 1248 }, { "epoch": 1.4, "learning_rate": 2.4371212121212123e-05, "loss": 6.802, "step": 1249 }, { "epoch": 1.4, "learning_rate": 2.4363636363636364e-05, "loss": 6.7266, "step": 1250 }, { "epoch": 1.4, "learning_rate": 2.4356060606060608e-05, "loss": 6.6146, "step": 1251 }, { "epoch": 1.4, "learning_rate": 2.434848484848485e-05, "loss": 6.4579, "step": 1252 }, { "epoch": 1.4, "learning_rate": 2.4340909090909092e-05, "loss": 6.4201, "step": 1253 }, { "epoch": 1.41, "learning_rate": 2.4333333333333333e-05, "loss": 6.6204, "step": 1254 }, { "epoch": 1.41, "learning_rate": 2.4325757575757577e-05, "loss": 6.5803, "step": 1255 }, { "epoch": 1.41, "learning_rate": 2.4318181818181818e-05, "loss": 6.3465, "step": 1256 }, { "epoch": 1.41, "learning_rate": 2.431060606060606e-05, "loss": 6.4382, "step": 1257 }, { "epoch": 1.41, "learning_rate": 2.4303030303030302e-05, "loss": 6.5965, "step": 1258 }, { "epoch": 1.41, "learning_rate": 2.4295454545454546e-05, "loss": 6.5847, "step": 1259 }, { "epoch": 1.41, "learning_rate": 2.428787878787879e-05, "loss": 6.5497, "step": 1260 }, { "epoch": 1.41, "learning_rate": 2.428030303030303e-05, "loss": 6.4795, "step": 1261 }, { "epoch": 1.41, "learning_rate": 2.4272727272727275e-05, "loss": 6.6216, "step": 1262 }, { "epoch": 1.42, "learning_rate": 2.4265151515151515e-05, "loss": 6.3692, "step": 1263 }, { "epoch": 1.42, "learning_rate": 2.425757575757576e-05, "loss": 6.5129, "step": 1264 }, { "epoch": 1.42, "learning_rate": 2.425e-05, "loss": 6.5166, "step": 1265 }, { "epoch": 1.42, "learning_rate": 2.4242424242424244e-05, "loss": 6.4256, "step": 1266 }, { "epoch": 1.42, "learning_rate": 2.4234848484848484e-05, "loss": 6.6089, "step": 1267 }, { "epoch": 1.42, "learning_rate": 2.422727272727273e-05, "loss": 6.5586, "step": 1268 }, { "epoch": 1.42, "learning_rate": 2.421969696969697e-05, "loss": 6.4557, "step": 1269 }, { "epoch": 1.42, "learning_rate": 2.4212121212121213e-05, "loss": 6.5365, "step": 1270 }, { "epoch": 1.42, "learning_rate": 2.4204545454545454e-05, "loss": 6.3723, "step": 1271 }, { "epoch": 1.43, "learning_rate": 2.4196969696969698e-05, "loss": 6.4285, "step": 1272 }, { "epoch": 1.43, "learning_rate": 2.418939393939394e-05, "loss": 6.4647, "step": 1273 }, { "epoch": 1.43, "learning_rate": 2.4181818181818182e-05, "loss": 6.471, "step": 1274 }, { "epoch": 1.43, "learning_rate": 2.4174242424242426e-05, "loss": 6.5092, "step": 1275 }, { "epoch": 1.43, "learning_rate": 2.4166666666666667e-05, "loss": 6.4166, "step": 1276 }, { "epoch": 1.43, "learning_rate": 2.415909090909091e-05, "loss": 6.6253, "step": 1277 }, { "epoch": 1.43, "learning_rate": 2.415151515151515e-05, "loss": 6.3834, "step": 1278 }, { "epoch": 1.43, "learning_rate": 2.4143939393939395e-05, "loss": 6.6312, "step": 1279 }, { "epoch": 1.43, "learning_rate": 2.4136363636363636e-05, "loss": 6.3541, "step": 1280 }, { "epoch": 1.44, "learning_rate": 2.412878787878788e-05, "loss": 6.4217, "step": 1281 }, { "epoch": 1.44, "learning_rate": 2.412121212121212e-05, "loss": 6.3666, "step": 1282 }, { "epoch": 1.44, "learning_rate": 2.4113636363636365e-05, "loss": 6.4527, "step": 1283 }, { "epoch": 1.44, "learning_rate": 2.410606060606061e-05, "loss": 6.3396, "step": 1284 }, { "epoch": 1.44, "learning_rate": 2.409848484848485e-05, "loss": 6.308, "step": 1285 }, { "epoch": 1.44, "learning_rate": 2.409090909090909e-05, "loss": 6.4148, "step": 1286 }, { "epoch": 1.44, "learning_rate": 2.4083333333333334e-05, "loss": 6.0176, "step": 1287 }, { "epoch": 1.44, "learning_rate": 2.4075757575757578e-05, "loss": 6.3755, "step": 1288 }, { "epoch": 1.45, "learning_rate": 2.4068181818181818e-05, "loss": 6.2188, "step": 1289 }, { "epoch": 1.45, "learning_rate": 2.406060606060606e-05, "loss": 5.9312, "step": 1290 }, { "epoch": 1.45, "learning_rate": 2.4053030303030303e-05, "loss": 5.9444, "step": 1291 }, { "epoch": 1.45, "learning_rate": 2.4045454545454547e-05, "loss": 5.5793, "step": 1292 }, { "epoch": 1.45, "learning_rate": 2.4037878787878787e-05, "loss": 6.8359, "step": 1293 }, { "epoch": 1.45, "learning_rate": 2.403030303030303e-05, "loss": 6.9282, "step": 1294 }, { "epoch": 1.45, "learning_rate": 2.4022727272727275e-05, "loss": 6.9443, "step": 1295 }, { "epoch": 1.45, "learning_rate": 2.4015151515151516e-05, "loss": 6.7984, "step": 1296 }, { "epoch": 1.45, "learning_rate": 2.4007575757575757e-05, "loss": 6.6523, "step": 1297 }, { "epoch": 1.46, "learning_rate": 2.4e-05, "loss": 6.7626, "step": 1298 }, { "epoch": 1.46, "learning_rate": 2.3992424242424245e-05, "loss": 6.6343, "step": 1299 }, { "epoch": 1.46, "learning_rate": 2.3984848484848485e-05, "loss": 6.7363, "step": 1300 }, { "epoch": 1.46, "learning_rate": 2.3977272727272726e-05, "loss": 6.6655, "step": 1301 }, { "epoch": 1.46, "learning_rate": 2.396969696969697e-05, "loss": 6.4777, "step": 1302 }, { "epoch": 1.46, "learning_rate": 2.3962121212121214e-05, "loss": 6.586, "step": 1303 }, { "epoch": 1.46, "learning_rate": 2.3954545454545454e-05, "loss": 6.4941, "step": 1304 }, { "epoch": 1.46, "learning_rate": 2.3946969696969695e-05, "loss": 6.5761, "step": 1305 }, { "epoch": 1.46, "learning_rate": 2.3939393939393942e-05, "loss": 6.5136, "step": 1306 }, { "epoch": 1.47, "learning_rate": 2.3931818181818183e-05, "loss": 6.4284, "step": 1307 }, { "epoch": 1.47, "learning_rate": 2.3924242424242424e-05, "loss": 6.6168, "step": 1308 }, { "epoch": 1.47, "learning_rate": 2.3916666666666668e-05, "loss": 6.4271, "step": 1309 }, { "epoch": 1.47, "learning_rate": 2.390909090909091e-05, "loss": 6.4586, "step": 1310 }, { "epoch": 1.47, "learning_rate": 2.3901515151515152e-05, "loss": 6.5155, "step": 1311 }, { "epoch": 1.47, "learning_rate": 2.3893939393939393e-05, "loss": 6.5657, "step": 1312 }, { "epoch": 1.47, "learning_rate": 2.388636363636364e-05, "loss": 6.412, "step": 1313 }, { "epoch": 1.47, "learning_rate": 2.387878787878788e-05, "loss": 6.4405, "step": 1314 }, { "epoch": 1.47, "learning_rate": 2.387121212121212e-05, "loss": 6.5153, "step": 1315 }, { "epoch": 1.48, "learning_rate": 2.3863636363636362e-05, "loss": 6.3734, "step": 1316 }, { "epoch": 1.48, "learning_rate": 2.385606060606061e-05, "loss": 6.5545, "step": 1317 }, { "epoch": 1.48, "learning_rate": 2.384848484848485e-05, "loss": 6.5565, "step": 1318 }, { "epoch": 1.48, "learning_rate": 2.384090909090909e-05, "loss": 6.6586, "step": 1319 }, { "epoch": 1.48, "learning_rate": 2.383333333333333e-05, "loss": 6.553, "step": 1320 }, { "epoch": 1.48, "learning_rate": 2.382575757575758e-05, "loss": 6.4093, "step": 1321 }, { "epoch": 1.48, "learning_rate": 2.381818181818182e-05, "loss": 6.5558, "step": 1322 }, { "epoch": 1.48, "learning_rate": 2.381060606060606e-05, "loss": 6.4449, "step": 1323 }, { "epoch": 1.48, "learning_rate": 2.3803030303030304e-05, "loss": 6.4556, "step": 1324 }, { "epoch": 1.49, "learning_rate": 2.3795454545454548e-05, "loss": 6.5031, "step": 1325 }, { "epoch": 1.49, "learning_rate": 2.3787878787878788e-05, "loss": 6.4918, "step": 1326 }, { "epoch": 1.49, "learning_rate": 2.378030303030303e-05, "loss": 6.5743, "step": 1327 }, { "epoch": 1.49, "learning_rate": 2.3772727272727276e-05, "loss": 6.5458, "step": 1328 }, { "epoch": 1.49, "learning_rate": 2.3765151515151517e-05, "loss": 6.6799, "step": 1329 }, { "epoch": 1.49, "learning_rate": 2.3757575757575757e-05, "loss": 6.4932, "step": 1330 }, { "epoch": 1.49, "learning_rate": 2.3749999999999998e-05, "loss": 6.4989, "step": 1331 }, { "epoch": 1.49, "learning_rate": 2.3742424242424245e-05, "loss": 6.4815, "step": 1332 }, { "epoch": 1.49, "learning_rate": 2.3734848484848486e-05, "loss": 6.4549, "step": 1333 }, { "epoch": 1.5, "learning_rate": 2.3727272727272726e-05, "loss": 6.4291, "step": 1334 }, { "epoch": 1.5, "learning_rate": 2.371969696969697e-05, "loss": 6.3377, "step": 1335 }, { "epoch": 1.5, "learning_rate": 2.3712121212121214e-05, "loss": 6.3472, "step": 1336 }, { "epoch": 1.5, "learning_rate": 2.3704545454545455e-05, "loss": 6.2028, "step": 1337 }, { "epoch": 1.5, "learning_rate": 2.3696969696969696e-05, "loss": 6.1291, "step": 1338 }, { "epoch": 1.5, "learning_rate": 2.368939393939394e-05, "loss": 6.0838, "step": 1339 }, { "epoch": 1.5, "learning_rate": 2.3681818181818184e-05, "loss": 5.9916, "step": 1340 }, { "epoch": 1.5, "learning_rate": 2.3674242424242424e-05, "loss": 5.7745, "step": 1341 }, { "epoch": 1.5, "learning_rate": 2.3666666666666665e-05, "loss": 5.2384, "step": 1342 }, { "epoch": 1.51, "learning_rate": 2.3659090909090912e-05, "loss": 7.172, "step": 1343 }, { "epoch": 1.51, "learning_rate": 2.3651515151515153e-05, "loss": 6.8772, "step": 1344 }, { "epoch": 1.51, "learning_rate": 2.3643939393939393e-05, "loss": 7.1029, "step": 1345 }, { "epoch": 1.51, "learning_rate": 2.3636363636363637e-05, "loss": 6.9375, "step": 1346 }, { "epoch": 1.51, "learning_rate": 2.362878787878788e-05, "loss": 6.935, "step": 1347 }, { "epoch": 1.51, "learning_rate": 2.3621212121212122e-05, "loss": 6.9154, "step": 1348 }, { "epoch": 1.51, "learning_rate": 2.3613636363636363e-05, "loss": 6.7782, "step": 1349 }, { "epoch": 1.51, "learning_rate": 2.3606060606060607e-05, "loss": 6.6765, "step": 1350 }, { "epoch": 1.51, "learning_rate": 2.359848484848485e-05, "loss": 6.7354, "step": 1351 }, { "epoch": 1.52, "learning_rate": 2.359090909090909e-05, "loss": 6.59, "step": 1352 }, { "epoch": 1.52, "learning_rate": 2.3583333333333332e-05, "loss": 6.5437, "step": 1353 }, { "epoch": 1.52, "learning_rate": 2.3575757575757576e-05, "loss": 6.6543, "step": 1354 }, { "epoch": 1.52, "learning_rate": 2.356818181818182e-05, "loss": 6.6105, "step": 1355 }, { "epoch": 1.52, "learning_rate": 2.356060606060606e-05, "loss": 6.5557, "step": 1356 }, { "epoch": 1.52, "learning_rate": 2.3553030303030304e-05, "loss": 6.5347, "step": 1357 }, { "epoch": 1.52, "learning_rate": 2.3545454545454545e-05, "loss": 6.5744, "step": 1358 }, { "epoch": 1.52, "learning_rate": 2.353787878787879e-05, "loss": 6.4213, "step": 1359 }, { "epoch": 1.52, "learning_rate": 2.353030303030303e-05, "loss": 6.3226, "step": 1360 }, { "epoch": 1.53, "learning_rate": 2.3522727272727273e-05, "loss": 6.2341, "step": 1361 }, { "epoch": 1.53, "learning_rate": 2.3515151515151517e-05, "loss": 6.5632, "step": 1362 }, { "epoch": 1.53, "learning_rate": 2.3507575757575758e-05, "loss": 6.4446, "step": 1363 }, { "epoch": 1.53, "learning_rate": 2.3500000000000002e-05, "loss": 6.5157, "step": 1364 }, { "epoch": 1.53, "learning_rate": 2.3492424242424243e-05, "loss": 6.5188, "step": 1365 }, { "epoch": 1.53, "learning_rate": 2.3484848484848487e-05, "loss": 6.5755, "step": 1366 }, { "epoch": 1.53, "learning_rate": 2.3477272727272727e-05, "loss": 6.4563, "step": 1367 }, { "epoch": 1.53, "learning_rate": 2.346969696969697e-05, "loss": 6.4703, "step": 1368 }, { "epoch": 1.53, "learning_rate": 2.3462121212121212e-05, "loss": 6.455, "step": 1369 }, { "epoch": 1.54, "learning_rate": 2.3454545454545456e-05, "loss": 6.4311, "step": 1370 }, { "epoch": 1.54, "learning_rate": 2.3446969696969696e-05, "loss": 6.4963, "step": 1371 }, { "epoch": 1.54, "learning_rate": 2.343939393939394e-05, "loss": 6.5132, "step": 1372 }, { "epoch": 1.54, "learning_rate": 2.343181818181818e-05, "loss": 6.4462, "step": 1373 }, { "epoch": 1.54, "learning_rate": 2.3424242424242425e-05, "loss": 6.3968, "step": 1374 }, { "epoch": 1.54, "learning_rate": 2.341666666666667e-05, "loss": 6.4885, "step": 1375 }, { "epoch": 1.54, "learning_rate": 2.340909090909091e-05, "loss": 6.4251, "step": 1376 }, { "epoch": 1.54, "learning_rate": 2.3401515151515153e-05, "loss": 6.2291, "step": 1377 }, { "epoch": 1.54, "learning_rate": 2.3393939393939394e-05, "loss": 6.4335, "step": 1378 }, { "epoch": 1.55, "learning_rate": 2.3386363636363638e-05, "loss": 6.5144, "step": 1379 }, { "epoch": 1.55, "learning_rate": 2.337878787878788e-05, "loss": 6.5288, "step": 1380 }, { "epoch": 1.55, "learning_rate": 2.3371212121212123e-05, "loss": 6.443, "step": 1381 }, { "epoch": 1.55, "learning_rate": 2.3363636363636363e-05, "loss": 6.4472, "step": 1382 }, { "epoch": 1.55, "learning_rate": 2.3356060606060607e-05, "loss": 6.3195, "step": 1383 }, { "epoch": 1.55, "learning_rate": 2.3348484848484848e-05, "loss": 6.3156, "step": 1384 }, { "epoch": 1.55, "learning_rate": 2.3340909090909092e-05, "loss": 6.3633, "step": 1385 }, { "epoch": 1.55, "learning_rate": 2.3333333333333336e-05, "loss": 6.1678, "step": 1386 }, { "epoch": 1.55, "learning_rate": 2.3325757575757576e-05, "loss": 6.1974, "step": 1387 }, { "epoch": 1.56, "learning_rate": 2.3318181818181817e-05, "loss": 6.1744, "step": 1388 }, { "epoch": 1.56, "learning_rate": 2.331060606060606e-05, "loss": 5.7477, "step": 1389 }, { "epoch": 1.56, "learning_rate": 2.3303030303030305e-05, "loss": 5.7442, "step": 1390 }, { "epoch": 1.56, "learning_rate": 2.3295454545454546e-05, "loss": 5.8305, "step": 1391 }, { "epoch": 1.56, "learning_rate": 2.3287878787878786e-05, "loss": 5.1165, "step": 1392 }, { "epoch": 1.56, "learning_rate": 2.328030303030303e-05, "loss": 6.6851, "step": 1393 }, { "epoch": 1.56, "learning_rate": 2.3272727272727274e-05, "loss": 6.7305, "step": 1394 }, { "epoch": 1.56, "learning_rate": 2.3265151515151515e-05, "loss": 6.7338, "step": 1395 }, { "epoch": 1.57, "learning_rate": 2.325757575757576e-05, "loss": 6.655, "step": 1396 }, { "epoch": 1.57, "learning_rate": 2.3250000000000003e-05, "loss": 6.6455, "step": 1397 }, { "epoch": 1.57, "learning_rate": 2.3242424242424243e-05, "loss": 6.6598, "step": 1398 }, { "epoch": 1.57, "learning_rate": 2.3234848484848484e-05, "loss": 6.612, "step": 1399 }, { "epoch": 1.57, "learning_rate": 2.3227272727272728e-05, "loss": 6.5589, "step": 1400 }, { "epoch": 1.57, "learning_rate": 2.3219696969696972e-05, "loss": 6.5469, "step": 1401 }, { "epoch": 1.57, "learning_rate": 2.3212121212121212e-05, "loss": 6.5711, "step": 1402 }, { "epoch": 1.57, "learning_rate": 2.3204545454545453e-05, "loss": 6.4593, "step": 1403 }, { "epoch": 1.57, "learning_rate": 2.3196969696969697e-05, "loss": 6.5287, "step": 1404 }, { "epoch": 1.58, "learning_rate": 2.318939393939394e-05, "loss": 6.428, "step": 1405 }, { "epoch": 1.58, "learning_rate": 2.318181818181818e-05, "loss": 6.5925, "step": 1406 }, { "epoch": 1.58, "learning_rate": 2.3174242424242422e-05, "loss": 6.5179, "step": 1407 }, { "epoch": 1.58, "learning_rate": 2.316666666666667e-05, "loss": 6.4426, "step": 1408 }, { "epoch": 1.58, "learning_rate": 2.315909090909091e-05, "loss": 6.5959, "step": 1409 }, { "epoch": 1.58, "learning_rate": 2.315151515151515e-05, "loss": 6.5378, "step": 1410 }, { "epoch": 1.58, "learning_rate": 2.3143939393939395e-05, "loss": 6.4764, "step": 1411 }, { "epoch": 1.58, "learning_rate": 2.313636363636364e-05, "loss": 6.5659, "step": 1412 }, { "epoch": 1.58, "learning_rate": 2.312878787878788e-05, "loss": 6.489, "step": 1413 }, { "epoch": 1.59, "learning_rate": 2.312121212121212e-05, "loss": 6.4739, "step": 1414 }, { "epoch": 1.59, "learning_rate": 2.3113636363636367e-05, "loss": 6.483, "step": 1415 }, { "epoch": 1.59, "learning_rate": 2.3106060606060608e-05, "loss": 6.3993, "step": 1416 }, { "epoch": 1.59, "learning_rate": 2.309848484848485e-05, "loss": 6.3644, "step": 1417 }, { "epoch": 1.59, "learning_rate": 2.309090909090909e-05, "loss": 6.371, "step": 1418 }, { "epoch": 1.59, "learning_rate": 2.3083333333333337e-05, "loss": 6.4391, "step": 1419 }, { "epoch": 1.59, "learning_rate": 2.3075757575757577e-05, "loss": 6.6904, "step": 1420 }, { "epoch": 1.59, "learning_rate": 2.3068181818181818e-05, "loss": 6.5358, "step": 1421 }, { "epoch": 1.59, "learning_rate": 2.3060606060606058e-05, "loss": 6.5813, "step": 1422 }, { "epoch": 1.6, "learning_rate": 2.3053030303030306e-05, "loss": 6.4515, "step": 1423 }, { "epoch": 1.6, "learning_rate": 2.3045454545454546e-05, "loss": 6.3444, "step": 1424 }, { "epoch": 1.6, "learning_rate": 2.3037878787878787e-05, "loss": 6.2243, "step": 1425 }, { "epoch": 1.6, "learning_rate": 2.303030303030303e-05, "loss": 6.326, "step": 1426 }, { "epoch": 1.6, "learning_rate": 2.3022727272727275e-05, "loss": 6.286, "step": 1427 }, { "epoch": 1.6, "learning_rate": 2.3015151515151515e-05, "loss": 6.4087, "step": 1428 }, { "epoch": 1.6, "learning_rate": 2.3007575757575756e-05, "loss": 6.3095, "step": 1429 }, { "epoch": 1.6, "learning_rate": 2.3000000000000003e-05, "loss": 6.4269, "step": 1430 }, { "epoch": 1.6, "learning_rate": 2.2992424242424244e-05, "loss": 6.4667, "step": 1431 }, { "epoch": 1.61, "learning_rate": 2.2984848484848485e-05, "loss": 6.501, "step": 1432 }, { "epoch": 1.61, "learning_rate": 2.2977272727272725e-05, "loss": 6.5379, "step": 1433 }, { "epoch": 1.61, "learning_rate": 2.2969696969696973e-05, "loss": 6.2691, "step": 1434 }, { "epoch": 1.61, "learning_rate": 2.2962121212121213e-05, "loss": 6.2627, "step": 1435 }, { "epoch": 1.61, "learning_rate": 2.2954545454545454e-05, "loss": 6.3168, "step": 1436 }, { "epoch": 1.61, "learning_rate": 2.2946969696969698e-05, "loss": 6.2913, "step": 1437 }, { "epoch": 1.61, "learning_rate": 2.2939393939393942e-05, "loss": 6.0184, "step": 1438 }, { "epoch": 1.61, "learning_rate": 2.2931818181818182e-05, "loss": 5.8459, "step": 1439 }, { "epoch": 1.61, "learning_rate": 2.2924242424242423e-05, "loss": 5.7868, "step": 1440 }, { "epoch": 1.62, "learning_rate": 2.2916666666666667e-05, "loss": 5.321, "step": 1441 }, { "epoch": 1.62, "learning_rate": 2.290909090909091e-05, "loss": 5.4116, "step": 1442 }, { "epoch": 1.62, "learning_rate": 2.290151515151515e-05, "loss": 6.6569, "step": 1443 }, { "epoch": 1.62, "learning_rate": 2.2893939393939392e-05, "loss": 6.7196, "step": 1444 }, { "epoch": 1.62, "learning_rate": 2.288636363636364e-05, "loss": 6.8785, "step": 1445 }, { "epoch": 1.62, "learning_rate": 2.287878787878788e-05, "loss": 6.7521, "step": 1446 }, { "epoch": 1.62, "learning_rate": 2.287121212121212e-05, "loss": 6.7108, "step": 1447 }, { "epoch": 1.62, "learning_rate": 2.2863636363636365e-05, "loss": 6.5794, "step": 1448 }, { "epoch": 1.62, "learning_rate": 2.285606060606061e-05, "loss": 6.6341, "step": 1449 }, { "epoch": 1.63, "learning_rate": 2.284848484848485e-05, "loss": 6.661, "step": 1450 }, { "epoch": 1.63, "learning_rate": 2.284090909090909e-05, "loss": 6.5021, "step": 1451 }, { "epoch": 1.63, "learning_rate": 2.2833333333333334e-05, "loss": 6.6725, "step": 1452 }, { "epoch": 1.63, "learning_rate": 2.2825757575757578e-05, "loss": 6.4352, "step": 1453 }, { "epoch": 1.63, "learning_rate": 2.281818181818182e-05, "loss": 6.3868, "step": 1454 }, { "epoch": 1.63, "learning_rate": 2.281060606060606e-05, "loss": 6.493, "step": 1455 }, { "epoch": 1.63, "learning_rate": 2.2803030303030303e-05, "loss": 6.4746, "step": 1456 }, { "epoch": 1.63, "learning_rate": 2.2795454545454547e-05, "loss": 6.4941, "step": 1457 }, { "epoch": 1.63, "learning_rate": 2.2787878787878788e-05, "loss": 6.4784, "step": 1458 }, { "epoch": 1.64, "learning_rate": 2.278030303030303e-05, "loss": 6.535, "step": 1459 }, { "epoch": 1.64, "learning_rate": 2.2772727272727272e-05, "loss": 6.4079, "step": 1460 }, { "epoch": 1.64, "learning_rate": 2.2765151515151516e-05, "loss": 6.4199, "step": 1461 }, { "epoch": 1.64, "learning_rate": 2.2757575757575757e-05, "loss": 6.4518, "step": 1462 }, { "epoch": 1.64, "learning_rate": 2.275e-05, "loss": 6.382, "step": 1463 }, { "epoch": 1.64, "learning_rate": 2.2742424242424245e-05, "loss": 6.6539, "step": 1464 }, { "epoch": 1.64, "learning_rate": 2.2734848484848485e-05, "loss": 6.4987, "step": 1465 }, { "epoch": 1.64, "learning_rate": 2.272727272727273e-05, "loss": 6.5062, "step": 1466 }, { "epoch": 1.64, "learning_rate": 2.271969696969697e-05, "loss": 6.3739, "step": 1467 }, { "epoch": 1.65, "learning_rate": 2.2712121212121214e-05, "loss": 6.415, "step": 1468 }, { "epoch": 1.65, "learning_rate": 2.2704545454545454e-05, "loss": 6.4559, "step": 1469 }, { "epoch": 1.65, "learning_rate": 2.26969696969697e-05, "loss": 6.4167, "step": 1470 }, { "epoch": 1.65, "learning_rate": 2.268939393939394e-05, "loss": 6.4078, "step": 1471 }, { "epoch": 1.65, "learning_rate": 2.2681818181818183e-05, "loss": 6.5476, "step": 1472 }, { "epoch": 1.65, "learning_rate": 2.2674242424242424e-05, "loss": 6.3604, "step": 1473 }, { "epoch": 1.65, "learning_rate": 2.2666666666666668e-05, "loss": 6.3507, "step": 1474 }, { "epoch": 1.65, "learning_rate": 2.2659090909090908e-05, "loss": 6.1504, "step": 1475 }, { "epoch": 1.65, "learning_rate": 2.2651515151515152e-05, "loss": 6.2906, "step": 1476 }, { "epoch": 1.66, "learning_rate": 2.2643939393939396e-05, "loss": 6.4572, "step": 1477 }, { "epoch": 1.66, "learning_rate": 2.2636363636363637e-05, "loss": 6.2954, "step": 1478 }, { "epoch": 1.66, "learning_rate": 2.262878787878788e-05, "loss": 6.4175, "step": 1479 }, { "epoch": 1.66, "learning_rate": 2.262121212121212e-05, "loss": 6.422, "step": 1480 }, { "epoch": 1.66, "learning_rate": 2.2613636363636365e-05, "loss": 6.2551, "step": 1481 }, { "epoch": 1.66, "learning_rate": 2.2606060606060606e-05, "loss": 6.3436, "step": 1482 }, { "epoch": 1.66, "learning_rate": 2.259848484848485e-05, "loss": 6.2634, "step": 1483 }, { "epoch": 1.66, "learning_rate": 2.259090909090909e-05, "loss": 6.4067, "step": 1484 }, { "epoch": 1.66, "learning_rate": 2.2583333333333335e-05, "loss": 6.3499, "step": 1485 }, { "epoch": 1.67, "learning_rate": 2.2575757575757575e-05, "loss": 6.3591, "step": 1486 }, { "epoch": 1.67, "learning_rate": 2.256818181818182e-05, "loss": 6.225, "step": 1487 }, { "epoch": 1.67, "learning_rate": 2.2560606060606063e-05, "loss": 6.0623, "step": 1488 }, { "epoch": 1.67, "learning_rate": 2.2553030303030304e-05, "loss": 5.869, "step": 1489 }, { "epoch": 1.67, "learning_rate": 2.2545454545454544e-05, "loss": 6.137, "step": 1490 }, { "epoch": 1.67, "learning_rate": 2.2537878787878788e-05, "loss": 5.6173, "step": 1491 }, { "epoch": 1.67, "learning_rate": 2.2530303030303032e-05, "loss": 5.4014, "step": 1492 }, { "epoch": 1.67, "learning_rate": 2.2522727272727273e-05, "loss": 6.7463, "step": 1493 }, { "epoch": 1.67, "learning_rate": 2.2515151515151513e-05, "loss": 6.6288, "step": 1494 }, { "epoch": 1.68, "learning_rate": 2.2507575757575757e-05, "loss": 6.6699, "step": 1495 }, { "epoch": 1.68, "learning_rate": 2.25e-05, "loss": 6.6496, "step": 1496 }, { "epoch": 1.68, "learning_rate": 2.2492424242424242e-05, "loss": 6.7079, "step": 1497 }, { "epoch": 1.68, "learning_rate": 2.2484848484848486e-05, "loss": 6.5644, "step": 1498 }, { "epoch": 1.68, "learning_rate": 2.247727272727273e-05, "loss": 6.6777, "step": 1499 }, { "epoch": 1.68, "learning_rate": 2.246969696969697e-05, "loss": 6.4601, "step": 1500 }, { "epoch": 1.68, "eval_loss": 6.180135726928711, "eval_runtime": 819.0225, "eval_samples_per_second": 3.226, "eval_steps_per_second": 0.404, "eval_wer": 1.1578341927806426, "step": 1500 }, { "epoch": 1.68, "learning_rate": 2.246212121212121e-05, "loss": 6.5695, "step": 1501 }, { "epoch": 1.68, "learning_rate": 2.2454545454545455e-05, "loss": 6.4147, "step": 1502 }, { "epoch": 1.68, "learning_rate": 2.24469696969697e-05, "loss": 6.3348, "step": 1503 }, { "epoch": 1.69, "learning_rate": 2.243939393939394e-05, "loss": 6.4682, "step": 1504 }, { "epoch": 1.69, "learning_rate": 2.243181818181818e-05, "loss": 6.4853, "step": 1505 }, { "epoch": 1.69, "learning_rate": 2.2424242424242424e-05, "loss": 6.4366, "step": 1506 }, { "epoch": 1.69, "learning_rate": 2.241666666666667e-05, "loss": 6.3061, "step": 1507 }, { "epoch": 1.69, "learning_rate": 2.240909090909091e-05, "loss": 6.4, "step": 1508 }, { "epoch": 1.69, "learning_rate": 2.240151515151515e-05, "loss": 6.3301, "step": 1509 }, { "epoch": 1.69, "learning_rate": 2.2393939393939397e-05, "loss": 6.3441, "step": 1510 }, { "epoch": 1.69, "learning_rate": 2.2386363636363637e-05, "loss": 6.4526, "step": 1511 }, { "epoch": 1.7, "learning_rate": 2.2378787878787878e-05, "loss": 6.4603, "step": 1512 }, { "epoch": 1.7, "learning_rate": 2.237121212121212e-05, "loss": 6.4997, "step": 1513 }, { "epoch": 1.7, "learning_rate": 2.2363636363636366e-05, "loss": 6.4575, "step": 1514 }, { "epoch": 1.7, "learning_rate": 2.2356060606060607e-05, "loss": 6.4645, "step": 1515 }, { "epoch": 1.7, "learning_rate": 2.2348484848484847e-05, "loss": 6.4451, "step": 1516 }, { "epoch": 1.7, "learning_rate": 2.2340909090909095e-05, "loss": 6.4197, "step": 1517 }, { "epoch": 1.7, "learning_rate": 2.2333333333333335e-05, "loss": 6.4488, "step": 1518 }, { "epoch": 1.7, "learning_rate": 2.2325757575757576e-05, "loss": 6.4524, "step": 1519 }, { "epoch": 1.7, "learning_rate": 2.2318181818181816e-05, "loss": 6.3413, "step": 1520 }, { "epoch": 1.71, "learning_rate": 2.2310606060606064e-05, "loss": 6.4992, "step": 1521 }, { "epoch": 1.71, "learning_rate": 2.2303030303030304e-05, "loss": 6.4583, "step": 1522 }, { "epoch": 1.71, "learning_rate": 2.2295454545454545e-05, "loss": 6.2452, "step": 1523 }, { "epoch": 1.71, "learning_rate": 2.2287878787878786e-05, "loss": 6.3542, "step": 1524 }, { "epoch": 1.71, "learning_rate": 2.2280303030303033e-05, "loss": 6.4856, "step": 1525 }, { "epoch": 1.71, "learning_rate": 2.2272727272727274e-05, "loss": 6.3722, "step": 1526 }, { "epoch": 1.71, "learning_rate": 2.2265151515151514e-05, "loss": 6.4022, "step": 1527 }, { "epoch": 1.71, "learning_rate": 2.2257575757575758e-05, "loss": 6.4798, "step": 1528 }, { "epoch": 1.71, "learning_rate": 2.2250000000000002e-05, "loss": 6.4091, "step": 1529 }, { "epoch": 1.72, "learning_rate": 2.2242424242424243e-05, "loss": 6.3534, "step": 1530 }, { "epoch": 1.72, "learning_rate": 2.2234848484848483e-05, "loss": 6.273, "step": 1531 }, { "epoch": 1.72, "learning_rate": 2.222727272727273e-05, "loss": 6.3591, "step": 1532 }, { "epoch": 1.72, "learning_rate": 2.221969696969697e-05, "loss": 6.2809, "step": 1533 }, { "epoch": 1.72, "learning_rate": 2.2212121212121212e-05, "loss": 6.3007, "step": 1534 }, { "epoch": 1.72, "learning_rate": 2.2204545454545452e-05, "loss": 6.1707, "step": 1535 }, { "epoch": 1.72, "learning_rate": 2.21969696969697e-05, "loss": 6.298, "step": 1536 }, { "epoch": 1.72, "learning_rate": 2.218939393939394e-05, "loss": 6.0171, "step": 1537 }, { "epoch": 1.72, "learning_rate": 2.218181818181818e-05, "loss": 5.9227, "step": 1538 }, { "epoch": 1.73, "learning_rate": 2.2174242424242425e-05, "loss": 5.8192, "step": 1539 }, { "epoch": 1.73, "learning_rate": 2.216666666666667e-05, "loss": 5.9125, "step": 1540 }, { "epoch": 1.73, "learning_rate": 2.215909090909091e-05, "loss": 5.6955, "step": 1541 }, { "epoch": 1.73, "learning_rate": 2.215151515151515e-05, "loss": 5.2539, "step": 1542 }, { "epoch": 1.73, "learning_rate": 2.2143939393939394e-05, "loss": 6.7421, "step": 1543 }, { "epoch": 1.73, "learning_rate": 2.2136363636363638e-05, "loss": 6.9187, "step": 1544 }, { "epoch": 1.73, "learning_rate": 2.212878787878788e-05, "loss": 6.8146, "step": 1545 }, { "epoch": 1.73, "learning_rate": 2.212121212121212e-05, "loss": 6.8287, "step": 1546 }, { "epoch": 1.73, "learning_rate": 2.2113636363636363e-05, "loss": 6.7992, "step": 1547 }, { "epoch": 1.74, "learning_rate": 2.2106060606060607e-05, "loss": 6.659, "step": 1548 }, { "epoch": 1.74, "learning_rate": 2.2098484848484848e-05, "loss": 6.8161, "step": 1549 }, { "epoch": 1.74, "learning_rate": 2.2090909090909092e-05, "loss": 6.5152, "step": 1550 }, { "epoch": 1.74, "learning_rate": 2.2083333333333336e-05, "loss": 6.4987, "step": 1551 }, { "epoch": 1.74, "learning_rate": 2.2075757575757577e-05, "loss": 6.6043, "step": 1552 }, { "epoch": 1.74, "learning_rate": 2.2068181818181817e-05, "loss": 6.5701, "step": 1553 }, { "epoch": 1.74, "learning_rate": 2.206060606060606e-05, "loss": 6.3272, "step": 1554 }, { "epoch": 1.74, "learning_rate": 2.2053030303030305e-05, "loss": 6.4373, "step": 1555 }, { "epoch": 1.74, "learning_rate": 2.2045454545454546e-05, "loss": 6.4491, "step": 1556 }, { "epoch": 1.75, "learning_rate": 2.2037878787878786e-05, "loss": 6.407, "step": 1557 }, { "epoch": 1.75, "learning_rate": 2.203030303030303e-05, "loss": 6.3736, "step": 1558 }, { "epoch": 1.75, "learning_rate": 2.2022727272727274e-05, "loss": 6.4506, "step": 1559 }, { "epoch": 1.75, "learning_rate": 2.2015151515151515e-05, "loss": 6.5047, "step": 1560 }, { "epoch": 1.75, "learning_rate": 2.200757575757576e-05, "loss": 6.4072, "step": 1561 }, { "epoch": 1.75, "learning_rate": 2.2e-05, "loss": 6.5289, "step": 1562 }, { "epoch": 1.75, "learning_rate": 2.1992424242424243e-05, "loss": 6.5297, "step": 1563 }, { "epoch": 1.75, "learning_rate": 2.1984848484848484e-05, "loss": 6.4059, "step": 1564 }, { "epoch": 1.75, "learning_rate": 2.1977272727272728e-05, "loss": 6.2858, "step": 1565 }, { "epoch": 1.76, "learning_rate": 2.1969696969696972e-05, "loss": 6.5416, "step": 1566 }, { "epoch": 1.76, "learning_rate": 2.1962121212121213e-05, "loss": 6.4089, "step": 1567 }, { "epoch": 1.76, "learning_rate": 2.1954545454545457e-05, "loss": 6.2685, "step": 1568 }, { "epoch": 1.76, "learning_rate": 2.1946969696969697e-05, "loss": 6.3149, "step": 1569 }, { "epoch": 1.76, "learning_rate": 2.193939393939394e-05, "loss": 6.4332, "step": 1570 }, { "epoch": 1.76, "learning_rate": 2.1931818181818182e-05, "loss": 6.3807, "step": 1571 }, { "epoch": 1.76, "learning_rate": 2.1924242424242426e-05, "loss": 6.5293, "step": 1572 }, { "epoch": 1.76, "learning_rate": 2.1916666666666666e-05, "loss": 6.4036, "step": 1573 }, { "epoch": 1.76, "learning_rate": 2.190909090909091e-05, "loss": 6.357, "step": 1574 }, { "epoch": 1.77, "learning_rate": 2.190151515151515e-05, "loss": 6.3445, "step": 1575 }, { "epoch": 1.77, "learning_rate": 2.1893939393939395e-05, "loss": 6.3598, "step": 1576 }, { "epoch": 1.77, "learning_rate": 2.1886363636363635e-05, "loss": 6.3635, "step": 1577 }, { "epoch": 1.77, "learning_rate": 2.187878787878788e-05, "loss": 6.3017, "step": 1578 }, { "epoch": 1.77, "learning_rate": 2.1871212121212123e-05, "loss": 6.4066, "step": 1579 }, { "epoch": 1.77, "learning_rate": 2.1863636363636364e-05, "loss": 6.4413, "step": 1580 }, { "epoch": 1.77, "learning_rate": 2.1856060606060605e-05, "loss": 6.3328, "step": 1581 }, { "epoch": 1.77, "learning_rate": 2.184848484848485e-05, "loss": 6.3641, "step": 1582 }, { "epoch": 1.77, "learning_rate": 2.1840909090909093e-05, "loss": 6.2866, "step": 1583 }, { "epoch": 1.78, "learning_rate": 2.1833333333333333e-05, "loss": 6.3301, "step": 1584 }, { "epoch": 1.78, "learning_rate": 2.1825757575757577e-05, "loss": 6.0793, "step": 1585 }, { "epoch": 1.78, "learning_rate": 2.1818181818181818e-05, "loss": 6.0141, "step": 1586 }, { "epoch": 1.78, "learning_rate": 2.1810606060606062e-05, "loss": 6.0441, "step": 1587 }, { "epoch": 1.78, "learning_rate": 2.1803030303030302e-05, "loss": 6.0094, "step": 1588 }, { "epoch": 1.78, "learning_rate": 2.1795454545454546e-05, "loss": 5.9993, "step": 1589 }, { "epoch": 1.78, "learning_rate": 2.178787878787879e-05, "loss": 5.8183, "step": 1590 }, { "epoch": 1.78, "learning_rate": 2.178030303030303e-05, "loss": 5.8089, "step": 1591 }, { "epoch": 1.78, "learning_rate": 2.177272727272727e-05, "loss": 4.9384, "step": 1592 }, { "epoch": 1.79, "learning_rate": 2.1765151515151516e-05, "loss": 6.7642, "step": 1593 }, { "epoch": 1.79, "learning_rate": 2.175757575757576e-05, "loss": 6.7048, "step": 1594 }, { "epoch": 1.79, "learning_rate": 2.175e-05, "loss": 6.6872, "step": 1595 }, { "epoch": 1.79, "learning_rate": 2.174242424242424e-05, "loss": 6.6541, "step": 1596 }, { "epoch": 1.79, "learning_rate": 2.1734848484848485e-05, "loss": 6.4739, "step": 1597 }, { "epoch": 1.79, "learning_rate": 2.172727272727273e-05, "loss": 6.5929, "step": 1598 }, { "epoch": 1.79, "learning_rate": 2.171969696969697e-05, "loss": 6.5725, "step": 1599 }, { "epoch": 1.79, "learning_rate": 2.1712121212121213e-05, "loss": 6.6485, "step": 1600 }, { "epoch": 1.79, "learning_rate": 2.1704545454545457e-05, "loss": 6.5038, "step": 1601 }, { "epoch": 1.8, "learning_rate": 2.1696969696969698e-05, "loss": 6.564, "step": 1602 }, { "epoch": 1.8, "learning_rate": 2.168939393939394e-05, "loss": 6.4568, "step": 1603 }, { "epoch": 1.8, "learning_rate": 2.1681818181818182e-05, "loss": 6.4886, "step": 1604 }, { "epoch": 1.8, "learning_rate": 2.1674242424242426e-05, "loss": 6.4815, "step": 1605 }, { "epoch": 1.8, "learning_rate": 2.1666666666666667e-05, "loss": 6.4702, "step": 1606 }, { "epoch": 1.8, "learning_rate": 2.1659090909090908e-05, "loss": 6.3449, "step": 1607 }, { "epoch": 1.8, "learning_rate": 2.165151515151515e-05, "loss": 6.3849, "step": 1608 }, { "epoch": 1.8, "learning_rate": 2.1643939393939396e-05, "loss": 6.4765, "step": 1609 }, { "epoch": 1.8, "learning_rate": 2.1636363636363636e-05, "loss": 6.3572, "step": 1610 }, { "epoch": 1.81, "learning_rate": 2.1628787878787877e-05, "loss": 6.4871, "step": 1611 }, { "epoch": 1.81, "learning_rate": 2.1621212121212124e-05, "loss": 6.4312, "step": 1612 }, { "epoch": 1.81, "learning_rate": 2.1613636363636365e-05, "loss": 6.4756, "step": 1613 }, { "epoch": 1.81, "learning_rate": 2.1606060606060605e-05, "loss": 6.509, "step": 1614 }, { "epoch": 1.81, "learning_rate": 2.1598484848484846e-05, "loss": 6.4061, "step": 1615 }, { "epoch": 1.81, "learning_rate": 2.1590909090909093e-05, "loss": 6.3429, "step": 1616 }, { "epoch": 1.81, "learning_rate": 2.1583333333333334e-05, "loss": 6.4851, "step": 1617 }, { "epoch": 1.81, "learning_rate": 2.1575757575757575e-05, "loss": 6.6241, "step": 1618 }, { "epoch": 1.82, "learning_rate": 2.1568181818181822e-05, "loss": 6.3801, "step": 1619 }, { "epoch": 1.82, "learning_rate": 2.1560606060606062e-05, "loss": 6.2539, "step": 1620 }, { "epoch": 1.82, "learning_rate": 2.1553030303030303e-05, "loss": 6.514, "step": 1621 }, { "epoch": 1.82, "learning_rate": 2.1545454545454544e-05, "loss": 6.2726, "step": 1622 }, { "epoch": 1.82, "learning_rate": 2.153787878787879e-05, "loss": 6.3492, "step": 1623 }, { "epoch": 1.82, "learning_rate": 2.153030303030303e-05, "loss": 6.1892, "step": 1624 }, { "epoch": 1.82, "learning_rate": 2.1522727272727272e-05, "loss": 6.2476, "step": 1625 }, { "epoch": 1.82, "learning_rate": 2.1515151515151513e-05, "loss": 6.2638, "step": 1626 }, { "epoch": 1.82, "learning_rate": 2.150757575757576e-05, "loss": 6.3361, "step": 1627 }, { "epoch": 1.83, "learning_rate": 2.15e-05, "loss": 6.4685, "step": 1628 }, { "epoch": 1.83, "learning_rate": 2.149242424242424e-05, "loss": 6.3302, "step": 1629 }, { "epoch": 1.83, "learning_rate": 2.1484848484848485e-05, "loss": 6.3666, "step": 1630 }, { "epoch": 1.83, "learning_rate": 2.147727272727273e-05, "loss": 6.1151, "step": 1631 }, { "epoch": 1.83, "learning_rate": 2.146969696969697e-05, "loss": 6.2836, "step": 1632 }, { "epoch": 1.83, "learning_rate": 2.146212121212121e-05, "loss": 6.3802, "step": 1633 }, { "epoch": 1.83, "learning_rate": 2.1454545454545458e-05, "loss": 6.4205, "step": 1634 }, { "epoch": 1.83, "learning_rate": 2.14469696969697e-05, "loss": 6.2464, "step": 1635 }, { "epoch": 1.83, "learning_rate": 2.143939393939394e-05, "loss": 6.0634, "step": 1636 }, { "epoch": 1.84, "learning_rate": 2.143181818181818e-05, "loss": 6.0076, "step": 1637 }, { "epoch": 1.84, "learning_rate": 2.1424242424242427e-05, "loss": 5.93, "step": 1638 }, { "epoch": 1.84, "learning_rate": 2.1416666666666668e-05, "loss": 5.9117, "step": 1639 }, { "epoch": 1.84, "learning_rate": 2.140909090909091e-05, "loss": 5.592, "step": 1640 }, { "epoch": 1.84, "learning_rate": 2.1401515151515152e-05, "loss": 5.7251, "step": 1641 }, { "epoch": 1.84, "learning_rate": 2.1393939393939396e-05, "loss": 4.9765, "step": 1642 }, { "epoch": 1.84, "learning_rate": 2.1386363636363637e-05, "loss": 6.6352, "step": 1643 }, { "epoch": 1.84, "learning_rate": 2.1378787878787877e-05, "loss": 6.7613, "step": 1644 }, { "epoch": 1.84, "learning_rate": 2.137121212121212e-05, "loss": 6.6395, "step": 1645 }, { "epoch": 1.85, "learning_rate": 2.1363636363636365e-05, "loss": 6.6755, "step": 1646 }, { "epoch": 1.85, "learning_rate": 2.1356060606060606e-05, "loss": 6.5883, "step": 1647 }, { "epoch": 1.85, "learning_rate": 2.1348484848484847e-05, "loss": 6.5226, "step": 1648 }, { "epoch": 1.85, "learning_rate": 2.134090909090909e-05, "loss": 6.6624, "step": 1649 }, { "epoch": 1.85, "learning_rate": 2.1333333333333335e-05, "loss": 6.6252, "step": 1650 }, { "epoch": 1.85, "learning_rate": 2.1325757575757575e-05, "loss": 6.4474, "step": 1651 }, { "epoch": 1.85, "learning_rate": 2.131818181818182e-05, "loss": 6.5844, "step": 1652 }, { "epoch": 1.85, "learning_rate": 2.1310606060606063e-05, "loss": 6.3628, "step": 1653 }, { "epoch": 1.85, "learning_rate": 2.1303030303030304e-05, "loss": 6.4009, "step": 1654 }, { "epoch": 1.86, "learning_rate": 2.1295454545454544e-05, "loss": 6.3951, "step": 1655 }, { "epoch": 1.86, "learning_rate": 2.128787878787879e-05, "loss": 6.3856, "step": 1656 }, { "epoch": 1.86, "learning_rate": 2.1280303030303032e-05, "loss": 6.4223, "step": 1657 }, { "epoch": 1.86, "learning_rate": 2.1272727272727273e-05, "loss": 6.5266, "step": 1658 }, { "epoch": 1.86, "learning_rate": 2.1265151515151514e-05, "loss": 6.2856, "step": 1659 }, { "epoch": 1.86, "learning_rate": 2.1257575757575758e-05, "loss": 6.4869, "step": 1660 }, { "epoch": 1.86, "learning_rate": 2.125e-05, "loss": 6.5201, "step": 1661 }, { "epoch": 1.86, "learning_rate": 2.1242424242424242e-05, "loss": 6.3592, "step": 1662 }, { "epoch": 1.86, "learning_rate": 2.1234848484848486e-05, "loss": 6.4207, "step": 1663 }, { "epoch": 1.87, "learning_rate": 2.1227272727272727e-05, "loss": 6.4222, "step": 1664 }, { "epoch": 1.87, "learning_rate": 2.121969696969697e-05, "loss": 6.2976, "step": 1665 }, { "epoch": 1.87, "learning_rate": 2.121212121212121e-05, "loss": 6.491, "step": 1666 }, { "epoch": 1.87, "learning_rate": 2.1204545454545455e-05, "loss": 6.2637, "step": 1667 }, { "epoch": 1.87, "learning_rate": 2.11969696969697e-05, "loss": 6.4758, "step": 1668 }, { "epoch": 1.87, "learning_rate": 2.118939393939394e-05, "loss": 6.3824, "step": 1669 }, { "epoch": 1.87, "learning_rate": 2.1181818181818184e-05, "loss": 6.3344, "step": 1670 }, { "epoch": 1.87, "learning_rate": 2.1174242424242424e-05, "loss": 6.2803, "step": 1671 }, { "epoch": 1.87, "learning_rate": 2.116666666666667e-05, "loss": 6.3357, "step": 1672 }, { "epoch": 1.88, "learning_rate": 2.115909090909091e-05, "loss": 6.3862, "step": 1673 }, { "epoch": 1.88, "learning_rate": 2.1151515151515153e-05, "loss": 6.4021, "step": 1674 }, { "epoch": 1.88, "learning_rate": 2.1143939393939394e-05, "loss": 6.3403, "step": 1675 }, { "epoch": 1.88, "learning_rate": 2.1136363636363638e-05, "loss": 6.3906, "step": 1676 }, { "epoch": 1.88, "learning_rate": 2.1128787878787878e-05, "loss": 6.3249, "step": 1677 }, { "epoch": 1.88, "learning_rate": 2.1121212121212122e-05, "loss": 6.1182, "step": 1678 }, { "epoch": 1.88, "learning_rate": 2.1113636363636363e-05, "loss": 6.4157, "step": 1679 }, { "epoch": 1.88, "learning_rate": 2.1106060606060607e-05, "loss": 6.2878, "step": 1680 }, { "epoch": 1.88, "learning_rate": 2.109848484848485e-05, "loss": 6.3206, "step": 1681 }, { "epoch": 1.89, "learning_rate": 2.109090909090909e-05, "loss": 6.1811, "step": 1682 }, { "epoch": 1.89, "learning_rate": 2.1083333333333332e-05, "loss": 6.3412, "step": 1683 }, { "epoch": 1.89, "learning_rate": 2.1075757575757576e-05, "loss": 6.0665, "step": 1684 }, { "epoch": 1.89, "learning_rate": 2.106818181818182e-05, "loss": 6.0521, "step": 1685 }, { "epoch": 1.89, "learning_rate": 2.106060606060606e-05, "loss": 6.0076, "step": 1686 }, { "epoch": 1.89, "learning_rate": 2.1053030303030304e-05, "loss": 5.8959, "step": 1687 }, { "epoch": 1.89, "learning_rate": 2.1045454545454545e-05, "loss": 5.8565, "step": 1688 }, { "epoch": 1.89, "learning_rate": 2.103787878787879e-05, "loss": 5.8995, "step": 1689 }, { "epoch": 1.89, "learning_rate": 2.103030303030303e-05, "loss": 5.6771, "step": 1690 }, { "epoch": 1.9, "learning_rate": 2.1022727272727274e-05, "loss": 5.8429, "step": 1691 }, { "epoch": 1.9, "learning_rate": 2.1015151515151518e-05, "loss": 5.1249, "step": 1692 }, { "epoch": 1.9, "learning_rate": 2.1007575757575758e-05, "loss": 6.7043, "step": 1693 }, { "epoch": 1.9, "learning_rate": 2.1e-05, "loss": 6.6423, "step": 1694 }, { "epoch": 1.9, "learning_rate": 2.0992424242424243e-05, "loss": 6.6675, "step": 1695 }, { "epoch": 1.9, "learning_rate": 2.0984848484848487e-05, "loss": 6.7421, "step": 1696 }, { "epoch": 1.9, "learning_rate": 2.0977272727272727e-05, "loss": 6.6367, "step": 1697 }, { "epoch": 1.9, "learning_rate": 2.0969696969696968e-05, "loss": 6.5884, "step": 1698 }, { "epoch": 1.9, "learning_rate": 2.0962121212121212e-05, "loss": 6.5371, "step": 1699 }, { "epoch": 1.91, "learning_rate": 2.0954545454545456e-05, "loss": 6.4682, "step": 1700 }, { "epoch": 1.91, "learning_rate": 2.0946969696969697e-05, "loss": 6.4675, "step": 1701 }, { "epoch": 1.91, "learning_rate": 2.093939393939394e-05, "loss": 6.4164, "step": 1702 }, { "epoch": 1.91, "learning_rate": 2.0931818181818185e-05, "loss": 6.4212, "step": 1703 }, { "epoch": 1.91, "learning_rate": 2.0924242424242425e-05, "loss": 6.5539, "step": 1704 }, { "epoch": 1.91, "learning_rate": 2.0916666666666666e-05, "loss": 6.3942, "step": 1705 }, { "epoch": 1.91, "learning_rate": 2.090909090909091e-05, "loss": 6.354, "step": 1706 }, { "epoch": 1.91, "learning_rate": 2.0901515151515154e-05, "loss": 6.4442, "step": 1707 }, { "epoch": 1.91, "learning_rate": 2.0893939393939394e-05, "loss": 6.4392, "step": 1708 }, { "epoch": 1.92, "learning_rate": 2.0886363636363635e-05, "loss": 6.5015, "step": 1709 }, { "epoch": 1.92, "learning_rate": 2.087878787878788e-05, "loss": 6.325, "step": 1710 }, { "epoch": 1.92, "learning_rate": 2.0871212121212123e-05, "loss": 6.2169, "step": 1711 }, { "epoch": 1.92, "learning_rate": 2.0863636363636363e-05, "loss": 6.3618, "step": 1712 }, { "epoch": 1.92, "learning_rate": 2.0856060606060604e-05, "loss": 6.3376, "step": 1713 }, { "epoch": 1.92, "learning_rate": 2.084848484848485e-05, "loss": 6.2321, "step": 1714 }, { "epoch": 1.92, "learning_rate": 2.0840909090909092e-05, "loss": 6.3952, "step": 1715 }, { "epoch": 1.92, "learning_rate": 2.0833333333333333e-05, "loss": 6.4142, "step": 1716 }, { "epoch": 1.92, "learning_rate": 2.0825757575757573e-05, "loss": 6.4436, "step": 1717 }, { "epoch": 1.93, "learning_rate": 2.081818181818182e-05, "loss": 6.3833, "step": 1718 }, { "epoch": 1.93, "learning_rate": 2.081060606060606e-05, "loss": 6.2243, "step": 1719 }, { "epoch": 1.93, "learning_rate": 2.0803030303030302e-05, "loss": 6.2904, "step": 1720 }, { "epoch": 1.93, "learning_rate": 2.079545454545455e-05, "loss": 6.3758, "step": 1721 }, { "epoch": 1.93, "learning_rate": 2.078787878787879e-05, "loss": 6.3289, "step": 1722 }, { "epoch": 1.93, "learning_rate": 2.078030303030303e-05, "loss": 6.3691, "step": 1723 }, { "epoch": 1.93, "learning_rate": 2.077272727272727e-05, "loss": 6.3211, "step": 1724 }, { "epoch": 1.93, "learning_rate": 2.076515151515152e-05, "loss": 6.3905, "step": 1725 }, { "epoch": 1.93, "learning_rate": 2.075757575757576e-05, "loss": 6.1707, "step": 1726 }, { "epoch": 1.94, "learning_rate": 2.075e-05, "loss": 6.3371, "step": 1727 }, { "epoch": 1.94, "learning_rate": 2.074242424242424e-05, "loss": 6.3274, "step": 1728 }, { "epoch": 1.94, "learning_rate": 2.0734848484848488e-05, "loss": 6.2234, "step": 1729 }, { "epoch": 1.94, "learning_rate": 2.0727272727272728e-05, "loss": 6.3297, "step": 1730 }, { "epoch": 1.94, "learning_rate": 2.071969696969697e-05, "loss": 6.2254, "step": 1731 }, { "epoch": 1.94, "learning_rate": 2.0712121212121213e-05, "loss": 6.4226, "step": 1732 }, { "epoch": 1.94, "learning_rate": 2.0704545454545457e-05, "loss": 6.1227, "step": 1733 }, { "epoch": 1.94, "learning_rate": 2.0696969696969697e-05, "loss": 6.1281, "step": 1734 }, { "epoch": 1.95, "learning_rate": 2.0689393939393938e-05, "loss": 6.0881, "step": 1735 }, { "epoch": 1.95, "learning_rate": 2.0681818181818185e-05, "loss": 6.1722, "step": 1736 }, { "epoch": 1.95, "learning_rate": 2.0674242424242426e-05, "loss": 6.1378, "step": 1737 }, { "epoch": 1.95, "learning_rate": 2.0666666666666666e-05, "loss": 5.9801, "step": 1738 }, { "epoch": 1.95, "learning_rate": 2.0659090909090907e-05, "loss": 5.8506, "step": 1739 }, { "epoch": 1.95, "learning_rate": 2.0651515151515154e-05, "loss": 5.6842, "step": 1740 }, { "epoch": 1.95, "learning_rate": 2.0643939393939395e-05, "loss": 5.3519, "step": 1741 }, { "epoch": 1.95, "learning_rate": 2.0636363636363636e-05, "loss": 5.1736, "step": 1742 }, { "epoch": 1.95, "learning_rate": 2.062878787878788e-05, "loss": 6.6388, "step": 1743 }, { "epoch": 1.96, "learning_rate": 2.0621212121212124e-05, "loss": 6.5449, "step": 1744 }, { "epoch": 1.96, "learning_rate": 2.0613636363636364e-05, "loss": 6.6342, "step": 1745 }, { "epoch": 1.96, "learning_rate": 2.0606060606060605e-05, "loss": 6.4854, "step": 1746 }, { "epoch": 1.96, "learning_rate": 2.059848484848485e-05, "loss": 6.5031, "step": 1747 }, { "epoch": 1.96, "learning_rate": 2.0590909090909093e-05, "loss": 6.4256, "step": 1748 }, { "epoch": 1.96, "learning_rate": 2.0583333333333333e-05, "loss": 6.3964, "step": 1749 }, { "epoch": 1.96, "learning_rate": 2.0575757575757574e-05, "loss": 6.6619, "step": 1750 }, { "epoch": 1.96, "learning_rate": 2.0568181818181818e-05, "loss": 6.4085, "step": 1751 }, { "epoch": 1.96, "learning_rate": 2.0560606060606062e-05, "loss": 6.4842, "step": 1752 }, { "epoch": 1.97, "learning_rate": 2.0553030303030303e-05, "loss": 6.4196, "step": 1753 }, { "epoch": 1.97, "learning_rate": 2.0545454545454546e-05, "loss": 6.4503, "step": 1754 }, { "epoch": 1.97, "learning_rate": 2.053787878787879e-05, "loss": 6.2639, "step": 1755 }, { "epoch": 1.97, "learning_rate": 2.053030303030303e-05, "loss": 6.3553, "step": 1756 }, { "epoch": 1.97, "learning_rate": 2.052272727272727e-05, "loss": 6.4366, "step": 1757 }, { "epoch": 1.97, "learning_rate": 2.0515151515151516e-05, "loss": 6.3467, "step": 1758 }, { "epoch": 1.97, "learning_rate": 2.050757575757576e-05, "loss": 6.4022, "step": 1759 }, { "epoch": 1.97, "learning_rate": 2.05e-05, "loss": 6.3037, "step": 1760 }, { "epoch": 1.97, "learning_rate": 2.049242424242424e-05, "loss": 6.4166, "step": 1761 }, { "epoch": 1.98, "learning_rate": 2.0484848484848485e-05, "loss": 6.3094, "step": 1762 }, { "epoch": 1.98, "learning_rate": 2.047727272727273e-05, "loss": 6.3945, "step": 1763 }, { "epoch": 1.98, "learning_rate": 2.046969696969697e-05, "loss": 6.3168, "step": 1764 }, { "epoch": 1.98, "learning_rate": 2.0462121212121213e-05, "loss": 6.2493, "step": 1765 }, { "epoch": 1.98, "learning_rate": 2.0454545454545454e-05, "loss": 6.3813, "step": 1766 }, { "epoch": 1.98, "learning_rate": 2.0446969696969698e-05, "loss": 6.2644, "step": 1767 }, { "epoch": 1.98, "learning_rate": 2.043939393939394e-05, "loss": 6.2957, "step": 1768 }, { "epoch": 1.98, "learning_rate": 2.0431818181818183e-05, "loss": 6.2464, "step": 1769 }, { "epoch": 1.98, "learning_rate": 2.0424242424242427e-05, "loss": 6.3285, "step": 1770 }, { "epoch": 1.99, "learning_rate": 2.0416666666666667e-05, "loss": 6.2731, "step": 1771 }, { "epoch": 1.99, "learning_rate": 2.040909090909091e-05, "loss": 6.2415, "step": 1772 }, { "epoch": 1.99, "learning_rate": 2.0401515151515152e-05, "loss": 6.287, "step": 1773 }, { "epoch": 1.99, "learning_rate": 2.0393939393939396e-05, "loss": 6.3043, "step": 1774 }, { "epoch": 1.99, "learning_rate": 2.0386363636363636e-05, "loss": 6.2881, "step": 1775 }, { "epoch": 1.99, "learning_rate": 2.037878787878788e-05, "loss": 6.2979, "step": 1776 }, { "epoch": 1.99, "learning_rate": 2.037121212121212e-05, "loss": 6.2961, "step": 1777 }, { "epoch": 1.99, "learning_rate": 2.0363636363636365e-05, "loss": 6.2903, "step": 1778 }, { "epoch": 1.99, "learning_rate": 2.0356060606060605e-05, "loss": 6.1238, "step": 1779 }, { "epoch": 2.0, "learning_rate": 2.034848484848485e-05, "loss": 5.8396, "step": 1780 }, { "epoch": 2.0, "learning_rate": 2.034090909090909e-05, "loss": 5.8829, "step": 1781 }, { "epoch": 2.0, "learning_rate": 2.0333333333333334e-05, "loss": 5.8751, "step": 1782 }, { "epoch": 2.0, "learning_rate": 2.0325757575757578e-05, "loss": 5.7011, "step": 1783 }, { "epoch": 2.0, "learning_rate": 2.031818181818182e-05, "loss": 5.022, "step": 1784 }, { "epoch": 2.0, "learning_rate": 2.031060606060606e-05, "loss": 6.5951, "step": 1785 }, { "epoch": 2.0, "learning_rate": 2.0303030303030303e-05, "loss": 6.2456, "step": 1786 }, { "epoch": 2.0, "learning_rate": 2.0295454545454547e-05, "loss": 6.46, "step": 1787 }, { "epoch": 2.0, "learning_rate": 2.0287878787878788e-05, "loss": 6.537, "step": 1788 }, { "epoch": 2.01, "learning_rate": 2.0280303030303032e-05, "loss": 6.4531, "step": 1789 }, { "epoch": 2.01, "learning_rate": 2.0272727272727272e-05, "loss": 6.4222, "step": 1790 }, { "epoch": 2.01, "learning_rate": 2.0265151515151516e-05, "loss": 6.4716, "step": 1791 }, { "epoch": 2.01, "learning_rate": 2.0257575757575757e-05, "loss": 6.4666, "step": 1792 }, { "epoch": 2.01, "learning_rate": 2.025e-05, "loss": 6.4074, "step": 1793 }, { "epoch": 2.01, "learning_rate": 2.0242424242424245e-05, "loss": 6.3856, "step": 1794 }, { "epoch": 2.01, "learning_rate": 2.0234848484848486e-05, "loss": 6.4339, "step": 1795 }, { "epoch": 2.01, "learning_rate": 2.0227272727272726e-05, "loss": 6.3673, "step": 1796 }, { "epoch": 2.01, "learning_rate": 2.021969696969697e-05, "loss": 6.3789, "step": 1797 }, { "epoch": 2.02, "learning_rate": 2.0212121212121214e-05, "loss": 6.3177, "step": 1798 }, { "epoch": 2.02, "learning_rate": 2.0204545454545455e-05, "loss": 6.3696, "step": 1799 }, { "epoch": 2.02, "learning_rate": 2.0196969696969695e-05, "loss": 6.3115, "step": 1800 }, { "epoch": 2.02, "learning_rate": 2.018939393939394e-05, "loss": 6.352, "step": 1801 }, { "epoch": 2.02, "learning_rate": 2.0181818181818183e-05, "loss": 6.4706, "step": 1802 }, { "epoch": 2.02, "learning_rate": 2.0174242424242424e-05, "loss": 6.3158, "step": 1803 }, { "epoch": 2.02, "learning_rate": 2.0166666666666668e-05, "loss": 6.2602, "step": 1804 }, { "epoch": 2.02, "learning_rate": 2.0159090909090912e-05, "loss": 6.2188, "step": 1805 }, { "epoch": 2.02, "learning_rate": 2.0151515151515152e-05, "loss": 6.3196, "step": 1806 }, { "epoch": 2.03, "learning_rate": 2.0143939393939393e-05, "loss": 6.238, "step": 1807 }, { "epoch": 2.03, "learning_rate": 2.0136363636363637e-05, "loss": 6.3275, "step": 1808 }, { "epoch": 2.03, "learning_rate": 2.012878787878788e-05, "loss": 6.221, "step": 1809 }, { "epoch": 2.03, "learning_rate": 2.012121212121212e-05, "loss": 6.2595, "step": 1810 }, { "epoch": 2.03, "learning_rate": 2.0113636363636362e-05, "loss": 6.2134, "step": 1811 }, { "epoch": 2.03, "learning_rate": 2.0106060606060606e-05, "loss": 6.1232, "step": 1812 }, { "epoch": 2.03, "learning_rate": 2.009848484848485e-05, "loss": 6.2893, "step": 1813 }, { "epoch": 2.03, "learning_rate": 2.009090909090909e-05, "loss": 6.3188, "step": 1814 }, { "epoch": 2.03, "learning_rate": 2.008333333333333e-05, "loss": 6.3582, "step": 1815 }, { "epoch": 2.04, "learning_rate": 2.007575757575758e-05, "loss": 6.3518, "step": 1816 }, { "epoch": 2.04, "learning_rate": 2.006818181818182e-05, "loss": 6.1894, "step": 1817 }, { "epoch": 2.04, "learning_rate": 2.006060606060606e-05, "loss": 6.1833, "step": 1818 }, { "epoch": 2.04, "learning_rate": 2.00530303030303e-05, "loss": 6.1516, "step": 1819 }, { "epoch": 2.04, "learning_rate": 2.0045454545454548e-05, "loss": 6.1047, "step": 1820 }, { "epoch": 2.04, "learning_rate": 2.003787878787879e-05, "loss": 6.2368, "step": 1821 }, { "epoch": 2.04, "learning_rate": 2.003030303030303e-05, "loss": 6.2141, "step": 1822 }, { "epoch": 2.04, "learning_rate": 2.0022727272727276e-05, "loss": 6.159, "step": 1823 }, { "epoch": 2.04, "learning_rate": 2.0015151515151517e-05, "loss": 6.1912, "step": 1824 }, { "epoch": 2.05, "learning_rate": 2.0007575757575758e-05, "loss": 6.153, "step": 1825 }, { "epoch": 2.05, "learning_rate": 1.9999999999999998e-05, "loss": 6.2443, "step": 1826 }, { "epoch": 2.05, "learning_rate": 1.9992424242424246e-05, "loss": 5.9072, "step": 1827 }, { "epoch": 2.05, "learning_rate": 1.9984848484848486e-05, "loss": 5.9892, "step": 1828 }, { "epoch": 2.05, "learning_rate": 1.9977272727272727e-05, "loss": 5.8181, "step": 1829 }, { "epoch": 2.05, "learning_rate": 1.9969696969696967e-05, "loss": 5.7828, "step": 1830 }, { "epoch": 2.05, "learning_rate": 1.9962121212121215e-05, "loss": 5.6836, "step": 1831 }, { "epoch": 2.05, "learning_rate": 1.9954545454545455e-05, "loss": 5.4481, "step": 1832 }, { "epoch": 2.05, "learning_rate": 1.9946969696969696e-05, "loss": 5.39, "step": 1833 }, { "epoch": 2.06, "learning_rate": 1.993939393939394e-05, "loss": 4.698, "step": 1834 }, { "epoch": 2.06, "learning_rate": 1.9931818181818184e-05, "loss": 6.6882, "step": 1835 }, { "epoch": 2.06, "learning_rate": 1.9924242424242425e-05, "loss": 6.6444, "step": 1836 }, { "epoch": 2.06, "learning_rate": 1.9916666666666665e-05, "loss": 6.5563, "step": 1837 }, { "epoch": 2.06, "learning_rate": 1.9909090909090913e-05, "loss": 6.6712, "step": 1838 }, { "epoch": 2.06, "learning_rate": 1.9901515151515153e-05, "loss": 6.6636, "step": 1839 }, { "epoch": 2.06, "learning_rate": 1.9893939393939394e-05, "loss": 6.3882, "step": 1840 }, { "epoch": 2.06, "learning_rate": 1.9886363636363634e-05, "loss": 6.5123, "step": 1841 }, { "epoch": 2.07, "learning_rate": 1.987878787878788e-05, "loss": 6.3709, "step": 1842 }, { "epoch": 2.07, "learning_rate": 1.9871212121212122e-05, "loss": 6.3075, "step": 1843 }, { "epoch": 2.07, "learning_rate": 1.9863636363636363e-05, "loss": 6.3954, "step": 1844 }, { "epoch": 2.07, "learning_rate": 1.9856060606060607e-05, "loss": 6.4603, "step": 1845 }, { "epoch": 2.07, "learning_rate": 1.984848484848485e-05, "loss": 6.3419, "step": 1846 }, { "epoch": 2.07, "learning_rate": 1.984090909090909e-05, "loss": 6.1705, "step": 1847 }, { "epoch": 2.07, "learning_rate": 1.9833333333333332e-05, "loss": 6.3114, "step": 1848 }, { "epoch": 2.07, "learning_rate": 1.9825757575757576e-05, "loss": 6.4536, "step": 1849 }, { "epoch": 2.07, "learning_rate": 1.981818181818182e-05, "loss": 6.2099, "step": 1850 }, { "epoch": 2.08, "learning_rate": 1.981060606060606e-05, "loss": 6.4178, "step": 1851 }, { "epoch": 2.08, "learning_rate": 1.98030303030303e-05, "loss": 6.2553, "step": 1852 }, { "epoch": 2.08, "learning_rate": 1.9795454545454545e-05, "loss": 6.4224, "step": 1853 }, { "epoch": 2.08, "learning_rate": 1.978787878787879e-05, "loss": 6.3443, "step": 1854 }, { "epoch": 2.08, "learning_rate": 1.978030303030303e-05, "loss": 6.4468, "step": 1855 }, { "epoch": 2.08, "learning_rate": 1.9772727272727274e-05, "loss": 6.2374, "step": 1856 }, { "epoch": 2.08, "learning_rate": 1.9765151515151518e-05, "loss": 6.256, "step": 1857 }, { "epoch": 2.08, "learning_rate": 1.975757575757576e-05, "loss": 6.2146, "step": 1858 }, { "epoch": 2.08, "learning_rate": 1.975e-05, "loss": 6.3952, "step": 1859 }, { "epoch": 2.09, "learning_rate": 1.9742424242424243e-05, "loss": 6.1678, "step": 1860 }, { "epoch": 2.09, "learning_rate": 1.9734848484848487e-05, "loss": 6.3953, "step": 1861 }, { "epoch": 2.09, "learning_rate": 1.9727272727272728e-05, "loss": 6.2769, "step": 1862 }, { "epoch": 2.09, "learning_rate": 1.9719696969696968e-05, "loss": 6.1312, "step": 1863 }, { "epoch": 2.09, "learning_rate": 1.9712121212121212e-05, "loss": 6.2582, "step": 1864 }, { "epoch": 2.09, "learning_rate": 1.9704545454545456e-05, "loss": 6.2203, "step": 1865 }, { "epoch": 2.09, "learning_rate": 1.9696969696969697e-05, "loss": 6.1172, "step": 1866 }, { "epoch": 2.09, "learning_rate": 1.968939393939394e-05, "loss": 6.3582, "step": 1867 }, { "epoch": 2.09, "learning_rate": 1.968181818181818e-05, "loss": 6.2263, "step": 1868 }, { "epoch": 2.1, "learning_rate": 1.9674242424242425e-05, "loss": 6.2411, "step": 1869 }, { "epoch": 2.1, "learning_rate": 1.9666666666666666e-05, "loss": 6.1634, "step": 1870 }, { "epoch": 2.1, "learning_rate": 1.965909090909091e-05, "loss": 6.2962, "step": 1871 }, { "epoch": 2.1, "learning_rate": 1.9651515151515154e-05, "loss": 6.1034, "step": 1872 }, { "epoch": 2.1, "learning_rate": 1.9643939393939394e-05, "loss": 6.2545, "step": 1873 }, { "epoch": 2.1, "learning_rate": 1.963636363636364e-05, "loss": 6.2912, "step": 1874 }, { "epoch": 2.1, "learning_rate": 1.962878787878788e-05, "loss": 6.1602, "step": 1875 }, { "epoch": 2.1, "learning_rate": 1.9621212121212123e-05, "loss": 6.2664, "step": 1876 }, { "epoch": 2.1, "learning_rate": 1.9613636363636364e-05, "loss": 6.0515, "step": 1877 }, { "epoch": 2.11, "learning_rate": 1.9606060606060608e-05, "loss": 5.9155, "step": 1878 }, { "epoch": 2.11, "learning_rate": 1.9598484848484848e-05, "loss": 6.065, "step": 1879 }, { "epoch": 2.11, "learning_rate": 1.9590909090909092e-05, "loss": 5.7793, "step": 1880 }, { "epoch": 2.11, "learning_rate": 1.9583333333333333e-05, "loss": 5.7037, "step": 1881 }, { "epoch": 2.11, "learning_rate": 1.9575757575757577e-05, "loss": 5.7191, "step": 1882 }, { "epoch": 2.11, "learning_rate": 1.9568181818181817e-05, "loss": 5.2426, "step": 1883 }, { "epoch": 2.11, "learning_rate": 1.956060606060606e-05, "loss": 4.6078, "step": 1884 }, { "epoch": 2.11, "learning_rate": 1.9553030303030305e-05, "loss": 6.6309, "step": 1885 }, { "epoch": 2.11, "learning_rate": 1.9545454545454546e-05, "loss": 6.5253, "step": 1886 }, { "epoch": 2.12, "learning_rate": 1.9537878787878786e-05, "loss": 6.617, "step": 1887 }, { "epoch": 2.12, "learning_rate": 1.953030303030303e-05, "loss": 6.6174, "step": 1888 }, { "epoch": 2.12, "learning_rate": 1.9522727272727274e-05, "loss": 6.5535, "step": 1889 }, { "epoch": 2.12, "learning_rate": 1.9515151515151515e-05, "loss": 6.5311, "step": 1890 }, { "epoch": 2.12, "learning_rate": 1.950757575757576e-05, "loss": 6.4232, "step": 1891 }, { "epoch": 2.12, "learning_rate": 1.95e-05, "loss": 6.3947, "step": 1892 }, { "epoch": 2.12, "learning_rate": 1.9492424242424244e-05, "loss": 6.6676, "step": 1893 }, { "epoch": 2.12, "learning_rate": 1.9484848484848484e-05, "loss": 6.4253, "step": 1894 }, { "epoch": 2.12, "learning_rate": 1.9477272727272728e-05, "loss": 6.188, "step": 1895 }, { "epoch": 2.13, "learning_rate": 1.9469696969696972e-05, "loss": 6.3959, "step": 1896 }, { "epoch": 2.13, "learning_rate": 1.9462121212121213e-05, "loss": 6.2915, "step": 1897 }, { "epoch": 2.13, "learning_rate": 1.9454545454545453e-05, "loss": 6.2143, "step": 1898 }, { "epoch": 2.13, "learning_rate": 1.9446969696969697e-05, "loss": 6.1169, "step": 1899 }, { "epoch": 2.13, "learning_rate": 1.943939393939394e-05, "loss": 6.2043, "step": 1900 }, { "epoch": 2.13, "learning_rate": 1.9431818181818182e-05, "loss": 6.3782, "step": 1901 }, { "epoch": 2.13, "learning_rate": 1.9424242424242423e-05, "loss": 6.4363, "step": 1902 }, { "epoch": 2.13, "learning_rate": 1.9416666666666667e-05, "loss": 6.3553, "step": 1903 }, { "epoch": 2.13, "learning_rate": 1.940909090909091e-05, "loss": 6.264, "step": 1904 }, { "epoch": 2.14, "learning_rate": 1.940151515151515e-05, "loss": 6.207, "step": 1905 }, { "epoch": 2.14, "learning_rate": 1.9393939393939395e-05, "loss": 6.3578, "step": 1906 }, { "epoch": 2.14, "learning_rate": 1.938636363636364e-05, "loss": 6.2649, "step": 1907 }, { "epoch": 2.14, "learning_rate": 1.937878787878788e-05, "loss": 6.3111, "step": 1908 }, { "epoch": 2.14, "learning_rate": 1.937121212121212e-05, "loss": 6.2559, "step": 1909 }, { "epoch": 2.14, "learning_rate": 1.9363636363636364e-05, "loss": 6.2781, "step": 1910 }, { "epoch": 2.14, "learning_rate": 1.9356060606060608e-05, "loss": 6.2167, "step": 1911 }, { "epoch": 2.14, "learning_rate": 1.934848484848485e-05, "loss": 6.3518, "step": 1912 }, { "epoch": 2.14, "learning_rate": 1.934090909090909e-05, "loss": 6.4273, "step": 1913 }, { "epoch": 2.15, "learning_rate": 1.9333333333333333e-05, "loss": 6.1721, "step": 1914 }, { "epoch": 2.15, "learning_rate": 1.9325757575757577e-05, "loss": 6.1881, "step": 1915 }, { "epoch": 2.15, "learning_rate": 1.9318181818181818e-05, "loss": 6.2131, "step": 1916 }, { "epoch": 2.15, "learning_rate": 1.931060606060606e-05, "loss": 6.1432, "step": 1917 }, { "epoch": 2.15, "learning_rate": 1.9303030303030306e-05, "loss": 6.1769, "step": 1918 }, { "epoch": 2.15, "learning_rate": 1.9295454545454547e-05, "loss": 6.22, "step": 1919 }, { "epoch": 2.15, "learning_rate": 1.9287878787878787e-05, "loss": 6.3577, "step": 1920 }, { "epoch": 2.15, "learning_rate": 1.9280303030303028e-05, "loss": 6.0493, "step": 1921 }, { "epoch": 2.15, "learning_rate": 1.9272727272727275e-05, "loss": 6.2241, "step": 1922 }, { "epoch": 2.16, "learning_rate": 1.9265151515151516e-05, "loss": 6.1869, "step": 1923 }, { "epoch": 2.16, "learning_rate": 1.9257575757575756e-05, "loss": 6.0749, "step": 1924 }, { "epoch": 2.16, "learning_rate": 1.9250000000000004e-05, "loss": 6.2994, "step": 1925 }, { "epoch": 2.16, "learning_rate": 1.9242424242424244e-05, "loss": 5.9078, "step": 1926 }, { "epoch": 2.16, "learning_rate": 1.9234848484848485e-05, "loss": 5.96, "step": 1927 }, { "epoch": 2.16, "learning_rate": 1.9227272727272726e-05, "loss": 5.7539, "step": 1928 }, { "epoch": 2.16, "learning_rate": 1.9219696969696973e-05, "loss": 5.7196, "step": 1929 }, { "epoch": 2.16, "learning_rate": 1.9212121212121213e-05, "loss": 5.6605, "step": 1930 }, { "epoch": 2.16, "learning_rate": 1.9204545454545454e-05, "loss": 5.4221, "step": 1931 }, { "epoch": 2.17, "learning_rate": 1.9196969696969695e-05, "loss": 5.1411, "step": 1932 }, { "epoch": 2.17, "learning_rate": 1.9189393939393942e-05, "loss": 5.2827, "step": 1933 }, { "epoch": 2.17, "learning_rate": 1.9181818181818183e-05, "loss": 4.9537, "step": 1934 }, { "epoch": 2.17, "learning_rate": 1.9174242424242423e-05, "loss": 6.7323, "step": 1935 }, { "epoch": 2.17, "learning_rate": 1.9166666666666667e-05, "loss": 6.6617, "step": 1936 }, { "epoch": 2.17, "learning_rate": 1.915909090909091e-05, "loss": 6.5646, "step": 1937 }, { "epoch": 2.17, "learning_rate": 1.9151515151515152e-05, "loss": 6.4522, "step": 1938 }, { "epoch": 2.17, "learning_rate": 1.9143939393939392e-05, "loss": 6.5141, "step": 1939 }, { "epoch": 2.17, "learning_rate": 1.913636363636364e-05, "loss": 6.6695, "step": 1940 }, { "epoch": 2.18, "learning_rate": 1.912878787878788e-05, "loss": 6.4378, "step": 1941 }, { "epoch": 2.18, "learning_rate": 1.912121212121212e-05, "loss": 6.3773, "step": 1942 }, { "epoch": 2.18, "learning_rate": 1.911363636363636e-05, "loss": 6.2732, "step": 1943 }, { "epoch": 2.18, "learning_rate": 1.910606060606061e-05, "loss": 6.1961, "step": 1944 }, { "epoch": 2.18, "learning_rate": 1.909848484848485e-05, "loss": 6.4668, "step": 1945 }, { "epoch": 2.18, "learning_rate": 1.909090909090909e-05, "loss": 6.259, "step": 1946 }, { "epoch": 2.18, "learning_rate": 1.9083333333333334e-05, "loss": 6.2295, "step": 1947 }, { "epoch": 2.18, "learning_rate": 1.9075757575757578e-05, "loss": 6.165, "step": 1948 }, { "epoch": 2.18, "learning_rate": 1.906818181818182e-05, "loss": 6.5306, "step": 1949 }, { "epoch": 2.19, "learning_rate": 1.906060606060606e-05, "loss": 6.2738, "step": 1950 }, { "epoch": 2.19, "learning_rate": 1.9053030303030303e-05, "loss": 6.1947, "step": 1951 }, { "epoch": 2.19, "learning_rate": 1.9045454545454547e-05, "loss": 6.1463, "step": 1952 }, { "epoch": 2.19, "learning_rate": 1.9037878787878788e-05, "loss": 6.198, "step": 1953 }, { "epoch": 2.19, "learning_rate": 1.903030303030303e-05, "loss": 6.4132, "step": 1954 }, { "epoch": 2.19, "learning_rate": 1.9022727272727272e-05, "loss": 6.2948, "step": 1955 }, { "epoch": 2.19, "learning_rate": 1.9015151515151516e-05, "loss": 6.328, "step": 1956 }, { "epoch": 2.19, "learning_rate": 1.9007575757575757e-05, "loss": 6.3721, "step": 1957 }, { "epoch": 2.2, "learning_rate": 1.9e-05, "loss": 6.21, "step": 1958 }, { "epoch": 2.2, "learning_rate": 1.8992424242424245e-05, "loss": 6.1255, "step": 1959 }, { "epoch": 2.2, "learning_rate": 1.8984848484848486e-05, "loss": 6.4173, "step": 1960 }, { "epoch": 2.2, "learning_rate": 1.8977272727272726e-05, "loss": 6.149, "step": 1961 }, { "epoch": 2.2, "learning_rate": 1.896969696969697e-05, "loss": 6.1252, "step": 1962 }, { "epoch": 2.2, "learning_rate": 1.8962121212121214e-05, "loss": 6.2347, "step": 1963 }, { "epoch": 2.2, "learning_rate": 1.8954545454545455e-05, "loss": 6.1875, "step": 1964 }, { "epoch": 2.2, "learning_rate": 1.8946969696969695e-05, "loss": 6.1545, "step": 1965 }, { "epoch": 2.2, "learning_rate": 1.893939393939394e-05, "loss": 6.2138, "step": 1966 }, { "epoch": 2.21, "learning_rate": 1.8931818181818183e-05, "loss": 6.0947, "step": 1967 }, { "epoch": 2.21, "learning_rate": 1.8924242424242424e-05, "loss": 6.2127, "step": 1968 }, { "epoch": 2.21, "learning_rate": 1.8916666666666668e-05, "loss": 6.2961, "step": 1969 }, { "epoch": 2.21, "learning_rate": 1.890909090909091e-05, "loss": 6.1862, "step": 1970 }, { "epoch": 2.21, "learning_rate": 1.8901515151515153e-05, "loss": 6.2729, "step": 1971 }, { "epoch": 2.21, "learning_rate": 1.8893939393939393e-05, "loss": 6.0459, "step": 1972 }, { "epoch": 2.21, "learning_rate": 1.8886363636363637e-05, "loss": 5.9739, "step": 1973 }, { "epoch": 2.21, "learning_rate": 1.887878787878788e-05, "loss": 6.1887, "step": 1974 }, { "epoch": 2.21, "learning_rate": 1.8871212121212122e-05, "loss": 6.0905, "step": 1975 }, { "epoch": 2.22, "learning_rate": 1.8863636363636366e-05, "loss": 6.0177, "step": 1976 }, { "epoch": 2.22, "learning_rate": 1.8856060606060606e-05, "loss": 5.9503, "step": 1977 }, { "epoch": 2.22, "learning_rate": 1.884848484848485e-05, "loss": 5.9615, "step": 1978 }, { "epoch": 2.22, "learning_rate": 1.884090909090909e-05, "loss": 5.5775, "step": 1979 }, { "epoch": 2.22, "learning_rate": 1.8833333333333335e-05, "loss": 5.6746, "step": 1980 }, { "epoch": 2.22, "learning_rate": 1.8825757575757575e-05, "loss": 5.6284, "step": 1981 }, { "epoch": 2.22, "learning_rate": 1.881818181818182e-05, "loss": 5.4239, "step": 1982 }, { "epoch": 2.22, "learning_rate": 1.881060606060606e-05, "loss": 5.2598, "step": 1983 }, { "epoch": 2.22, "learning_rate": 1.8803030303030304e-05, "loss": 4.6313, "step": 1984 }, { "epoch": 2.23, "learning_rate": 1.8795454545454545e-05, "loss": 6.7312, "step": 1985 }, { "epoch": 2.23, "learning_rate": 1.878787878787879e-05, "loss": 6.6694, "step": 1986 }, { "epoch": 2.23, "learning_rate": 1.8780303030303033e-05, "loss": 6.727, "step": 1987 }, { "epoch": 2.23, "learning_rate": 1.8772727272727273e-05, "loss": 6.5924, "step": 1988 }, { "epoch": 2.23, "learning_rate": 1.8765151515151514e-05, "loss": 6.6222, "step": 1989 }, { "epoch": 2.23, "learning_rate": 1.8757575757575758e-05, "loss": 6.6636, "step": 1990 }, { "epoch": 2.23, "learning_rate": 1.8750000000000002e-05, "loss": 6.4683, "step": 1991 }, { "epoch": 2.23, "learning_rate": 1.8742424242424242e-05, "loss": 6.559, "step": 1992 }, { "epoch": 2.23, "learning_rate": 1.8734848484848486e-05, "loss": 6.3753, "step": 1993 }, { "epoch": 2.24, "learning_rate": 1.8727272727272727e-05, "loss": 6.3768, "step": 1994 }, { "epoch": 2.24, "learning_rate": 1.871969696969697e-05, "loss": 6.4424, "step": 1995 }, { "epoch": 2.24, "learning_rate": 1.871212121212121e-05, "loss": 6.2943, "step": 1996 }, { "epoch": 2.24, "learning_rate": 1.8704545454545455e-05, "loss": 6.2785, "step": 1997 }, { "epoch": 2.24, "learning_rate": 1.86969696969697e-05, "loss": 6.163, "step": 1998 }, { "epoch": 2.24, "learning_rate": 1.868939393939394e-05, "loss": 6.3983, "step": 1999 }, { "epoch": 2.24, "learning_rate": 1.868181818181818e-05, "loss": 6.1802, "step": 2000 }, { "epoch": 2.24, "eval_loss": 6.000200271606445, "eval_runtime": 879.6418, "eval_samples_per_second": 3.003, "eval_steps_per_second": 0.376, "eval_wer": 1.7750099166997224, "step": 2000 }, { "epoch": 2.24, "learning_rate": 1.8674242424242425e-05, "loss": 6.3019, "step": 2001 }, { "epoch": 2.24, "learning_rate": 1.866666666666667e-05, "loss": 6.1587, "step": 2002 }, { "epoch": 2.25, "learning_rate": 1.865909090909091e-05, "loss": 6.2655, "step": 2003 }, { "epoch": 2.25, "learning_rate": 1.865151515151515e-05, "loss": 6.3068, "step": 2004 }, { "epoch": 2.25, "learning_rate": 1.8643939393939394e-05, "loss": 6.2308, "step": 2005 }, { "epoch": 2.25, "learning_rate": 1.8636363636363638e-05, "loss": 6.3892, "step": 2006 }, { "epoch": 2.25, "learning_rate": 1.862878787878788e-05, "loss": 6.1984, "step": 2007 }, { "epoch": 2.25, "learning_rate": 1.862121212121212e-05, "loss": 6.2746, "step": 2008 }, { "epoch": 2.25, "learning_rate": 1.8613636363636366e-05, "loss": 6.2937, "step": 2009 }, { "epoch": 2.25, "learning_rate": 1.8606060606060607e-05, "loss": 6.271, "step": 2010 }, { "epoch": 2.25, "learning_rate": 1.8598484848484848e-05, "loss": 6.3281, "step": 2011 }, { "epoch": 2.26, "learning_rate": 1.859090909090909e-05, "loss": 6.2617, "step": 2012 }, { "epoch": 2.26, "learning_rate": 1.8583333333333336e-05, "loss": 6.2119, "step": 2013 }, { "epoch": 2.26, "learning_rate": 1.8575757575757576e-05, "loss": 6.2516, "step": 2014 }, { "epoch": 2.26, "learning_rate": 1.8568181818181817e-05, "loss": 6.1601, "step": 2015 }, { "epoch": 2.26, "learning_rate": 1.856060606060606e-05, "loss": 6.2349, "step": 2016 }, { "epoch": 2.26, "learning_rate": 1.8553030303030305e-05, "loss": 6.186, "step": 2017 }, { "epoch": 2.26, "learning_rate": 1.8545454545454545e-05, "loss": 6.1217, "step": 2018 }, { "epoch": 2.26, "learning_rate": 1.8537878787878786e-05, "loss": 6.2804, "step": 2019 }, { "epoch": 2.26, "learning_rate": 1.8530303030303033e-05, "loss": 6.3131, "step": 2020 }, { "epoch": 2.27, "learning_rate": 1.8522727272727274e-05, "loss": 6.0962, "step": 2021 }, { "epoch": 2.27, "learning_rate": 1.8515151515151514e-05, "loss": 6.223, "step": 2022 }, { "epoch": 2.27, "learning_rate": 1.8507575757575755e-05, "loss": 6.2346, "step": 2023 }, { "epoch": 2.27, "learning_rate": 1.8500000000000002e-05, "loss": 6.0723, "step": 2024 }, { "epoch": 2.27, "learning_rate": 1.8492424242424243e-05, "loss": 6.0273, "step": 2025 }, { "epoch": 2.27, "learning_rate": 1.8484848484848484e-05, "loss": 6.0558, "step": 2026 }, { "epoch": 2.27, "learning_rate": 1.8477272727272728e-05, "loss": 6.0263, "step": 2027 }, { "epoch": 2.27, "learning_rate": 1.846969696969697e-05, "loss": 5.7521, "step": 2028 }, { "epoch": 2.27, "learning_rate": 1.8462121212121212e-05, "loss": 5.6549, "step": 2029 }, { "epoch": 2.28, "learning_rate": 1.8454545454545453e-05, "loss": 5.8606, "step": 2030 }, { "epoch": 2.28, "learning_rate": 1.84469696969697e-05, "loss": 5.3422, "step": 2031 }, { "epoch": 2.28, "learning_rate": 1.843939393939394e-05, "loss": 5.4787, "step": 2032 }, { "epoch": 2.28, "learning_rate": 1.843181818181818e-05, "loss": 5.0512, "step": 2033 }, { "epoch": 2.28, "learning_rate": 1.8424242424242422e-05, "loss": 4.8726, "step": 2034 }, { "epoch": 2.28, "learning_rate": 1.841666666666667e-05, "loss": 6.5656, "step": 2035 }, { "epoch": 2.28, "learning_rate": 1.840909090909091e-05, "loss": 6.5941, "step": 2036 }, { "epoch": 2.28, "learning_rate": 1.840151515151515e-05, "loss": 6.5075, "step": 2037 }, { "epoch": 2.28, "learning_rate": 1.8393939393939395e-05, "loss": 6.519, "step": 2038 }, { "epoch": 2.29, "learning_rate": 1.838636363636364e-05, "loss": 6.4105, "step": 2039 }, { "epoch": 2.29, "learning_rate": 1.837878787878788e-05, "loss": 6.5528, "step": 2040 }, { "epoch": 2.29, "learning_rate": 1.837121212121212e-05, "loss": 6.3433, "step": 2041 }, { "epoch": 2.29, "learning_rate": 1.8363636363636364e-05, "loss": 6.4499, "step": 2042 }, { "epoch": 2.29, "learning_rate": 1.8356060606060608e-05, "loss": 6.3573, "step": 2043 }, { "epoch": 2.29, "learning_rate": 1.8348484848484848e-05, "loss": 6.1922, "step": 2044 }, { "epoch": 2.29, "learning_rate": 1.834090909090909e-05, "loss": 6.2771, "step": 2045 }, { "epoch": 2.29, "learning_rate": 1.8333333333333336e-05, "loss": 6.2291, "step": 2046 }, { "epoch": 2.29, "learning_rate": 1.8325757575757577e-05, "loss": 6.3164, "step": 2047 }, { "epoch": 2.3, "learning_rate": 1.8318181818181817e-05, "loss": 6.1285, "step": 2048 }, { "epoch": 2.3, "learning_rate": 1.831060606060606e-05, "loss": 6.4506, "step": 2049 }, { "epoch": 2.3, "learning_rate": 1.8303030303030305e-05, "loss": 6.242, "step": 2050 }, { "epoch": 2.3, "learning_rate": 1.8295454545454546e-05, "loss": 6.3939, "step": 2051 }, { "epoch": 2.3, "learning_rate": 1.8287878787878787e-05, "loss": 6.1249, "step": 2052 }, { "epoch": 2.3, "learning_rate": 1.828030303030303e-05, "loss": 6.246, "step": 2053 }, { "epoch": 2.3, "learning_rate": 1.8272727272727275e-05, "loss": 6.3196, "step": 2054 }, { "epoch": 2.3, "learning_rate": 1.8265151515151515e-05, "loss": 6.3665, "step": 2055 }, { "epoch": 2.3, "learning_rate": 1.8257575757575756e-05, "loss": 6.1542, "step": 2056 }, { "epoch": 2.31, "learning_rate": 1.825e-05, "loss": 6.2504, "step": 2057 }, { "epoch": 2.31, "learning_rate": 1.8242424242424244e-05, "loss": 6.0431, "step": 2058 }, { "epoch": 2.31, "learning_rate": 1.8234848484848484e-05, "loss": 6.2982, "step": 2059 }, { "epoch": 2.31, "learning_rate": 1.822727272727273e-05, "loss": 6.275, "step": 2060 }, { "epoch": 2.31, "learning_rate": 1.8219696969696972e-05, "loss": 6.4536, "step": 2061 }, { "epoch": 2.31, "learning_rate": 1.8212121212121213e-05, "loss": 6.3655, "step": 2062 }, { "epoch": 2.31, "learning_rate": 1.8204545454545454e-05, "loss": 6.175, "step": 2063 }, { "epoch": 2.31, "learning_rate": 1.8196969696969697e-05, "loss": 6.3511, "step": 2064 }, { "epoch": 2.32, "learning_rate": 1.818939393939394e-05, "loss": 6.1601, "step": 2065 }, { "epoch": 2.32, "learning_rate": 1.8181818181818182e-05, "loss": 6.0346, "step": 2066 }, { "epoch": 2.32, "learning_rate": 1.8174242424242423e-05, "loss": 6.1596, "step": 2067 }, { "epoch": 2.32, "learning_rate": 1.8166666666666667e-05, "loss": 6.1505, "step": 2068 }, { "epoch": 2.32, "learning_rate": 1.815909090909091e-05, "loss": 6.1257, "step": 2069 }, { "epoch": 2.32, "learning_rate": 1.815151515151515e-05, "loss": 6.1547, "step": 2070 }, { "epoch": 2.32, "learning_rate": 1.8143939393939395e-05, "loss": 6.0609, "step": 2071 }, { "epoch": 2.32, "learning_rate": 1.8136363636363636e-05, "loss": 6.1685, "step": 2072 }, { "epoch": 2.32, "learning_rate": 1.812878787878788e-05, "loss": 5.9542, "step": 2073 }, { "epoch": 2.33, "learning_rate": 1.812121212121212e-05, "loss": 6.1899, "step": 2074 }, { "epoch": 2.33, "learning_rate": 1.8113636363636364e-05, "loss": 6.0647, "step": 2075 }, { "epoch": 2.33, "learning_rate": 1.8106060606060605e-05, "loss": 6.0034, "step": 2076 }, { "epoch": 2.33, "learning_rate": 1.809848484848485e-05, "loss": 5.9611, "step": 2077 }, { "epoch": 2.33, "learning_rate": 1.8090909090909093e-05, "loss": 5.8959, "step": 2078 }, { "epoch": 2.33, "learning_rate": 1.8083333333333334e-05, "loss": 5.6503, "step": 2079 }, { "epoch": 2.33, "learning_rate": 1.8075757575757578e-05, "loss": 5.6743, "step": 2080 }, { "epoch": 2.33, "learning_rate": 1.8068181818181818e-05, "loss": 5.4787, "step": 2081 }, { "epoch": 2.33, "learning_rate": 1.8060606060606062e-05, "loss": 5.1239, "step": 2082 }, { "epoch": 2.34, "learning_rate": 1.8053030303030303e-05, "loss": 4.8965, "step": 2083 }, { "epoch": 2.34, "learning_rate": 1.8045454545454547e-05, "loss": 4.5085, "step": 2084 }, { "epoch": 2.34, "learning_rate": 1.8037878787878787e-05, "loss": 6.5608, "step": 2085 }, { "epoch": 2.34, "learning_rate": 1.803030303030303e-05, "loss": 6.613, "step": 2086 }, { "epoch": 2.34, "learning_rate": 1.8022727272727272e-05, "loss": 6.6967, "step": 2087 }, { "epoch": 2.34, "learning_rate": 1.8015151515151516e-05, "loss": 6.5498, "step": 2088 }, { "epoch": 2.34, "learning_rate": 1.800757575757576e-05, "loss": 6.4536, "step": 2089 }, { "epoch": 2.34, "learning_rate": 1.8e-05, "loss": 6.5189, "step": 2090 }, { "epoch": 2.34, "learning_rate": 1.799242424242424e-05, "loss": 6.4021, "step": 2091 }, { "epoch": 2.35, "learning_rate": 1.7984848484848485e-05, "loss": 6.2398, "step": 2092 }, { "epoch": 2.35, "learning_rate": 1.797727272727273e-05, "loss": 6.4002, "step": 2093 }, { "epoch": 2.35, "learning_rate": 1.796969696969697e-05, "loss": 6.528, "step": 2094 }, { "epoch": 2.35, "learning_rate": 1.7962121212121214e-05, "loss": 6.2887, "step": 2095 }, { "epoch": 2.35, "learning_rate": 1.7954545454545454e-05, "loss": 6.4979, "step": 2096 }, { "epoch": 2.35, "learning_rate": 1.7946969696969698e-05, "loss": 6.3785, "step": 2097 }, { "epoch": 2.35, "learning_rate": 1.793939393939394e-05, "loss": 6.4825, "step": 2098 }, { "epoch": 2.35, "learning_rate": 1.7931818181818183e-05, "loss": 6.1861, "step": 2099 }, { "epoch": 2.35, "learning_rate": 1.7924242424242427e-05, "loss": 6.2524, "step": 2100 }, { "epoch": 2.36, "learning_rate": 1.7916666666666667e-05, "loss": 6.1037, "step": 2101 }, { "epoch": 2.36, "learning_rate": 1.7909090909090908e-05, "loss": 6.3354, "step": 2102 }, { "epoch": 2.36, "learning_rate": 1.7901515151515152e-05, "loss": 6.3081, "step": 2103 }, { "epoch": 2.36, "learning_rate": 1.7893939393939396e-05, "loss": 6.2077, "step": 2104 }, { "epoch": 2.36, "learning_rate": 1.7886363636363637e-05, "loss": 6.2373, "step": 2105 }, { "epoch": 2.36, "learning_rate": 1.7878787878787877e-05, "loss": 6.249, "step": 2106 }, { "epoch": 2.36, "learning_rate": 1.787121212121212e-05, "loss": 6.1671, "step": 2107 }, { "epoch": 2.36, "learning_rate": 1.7863636363636365e-05, "loss": 6.2671, "step": 2108 }, { "epoch": 2.36, "learning_rate": 1.7856060606060606e-05, "loss": 6.1272, "step": 2109 }, { "epoch": 2.37, "learning_rate": 1.7848484848484846e-05, "loss": 6.2354, "step": 2110 }, { "epoch": 2.37, "learning_rate": 1.7840909090909094e-05, "loss": 6.2194, "step": 2111 }, { "epoch": 2.37, "learning_rate": 1.7833333333333334e-05, "loss": 6.1477, "step": 2112 }, { "epoch": 2.37, "learning_rate": 1.7825757575757575e-05, "loss": 6.1971, "step": 2113 }, { "epoch": 2.37, "learning_rate": 1.781818181818182e-05, "loss": 6.1474, "step": 2114 }, { "epoch": 2.37, "learning_rate": 1.7810606060606063e-05, "loss": 6.1129, "step": 2115 }, { "epoch": 2.37, "learning_rate": 1.7803030303030303e-05, "loss": 6.1468, "step": 2116 }, { "epoch": 2.37, "learning_rate": 1.7795454545454544e-05, "loss": 6.0674, "step": 2117 }, { "epoch": 2.37, "learning_rate": 1.7787878787878788e-05, "loss": 6.1174, "step": 2118 }, { "epoch": 2.38, "learning_rate": 1.7780303030303032e-05, "loss": 6.158, "step": 2119 }, { "epoch": 2.38, "learning_rate": 1.7772727272727273e-05, "loss": 6.2855, "step": 2120 }, { "epoch": 2.38, "learning_rate": 1.7765151515151513e-05, "loss": 6.1578, "step": 2121 }, { "epoch": 2.38, "learning_rate": 1.775757575757576e-05, "loss": 6.1146, "step": 2122 }, { "epoch": 2.38, "learning_rate": 1.775e-05, "loss": 6.0309, "step": 2123 }, { "epoch": 2.38, "learning_rate": 1.7742424242424242e-05, "loss": 6.1105, "step": 2124 }, { "epoch": 2.38, "learning_rate": 1.7734848484848482e-05, "loss": 6.1803, "step": 2125 }, { "epoch": 2.38, "learning_rate": 1.772727272727273e-05, "loss": 6.0015, "step": 2126 }, { "epoch": 2.38, "learning_rate": 1.771969696969697e-05, "loss": 5.9725, "step": 2127 }, { "epoch": 2.39, "learning_rate": 1.771212121212121e-05, "loss": 5.7605, "step": 2128 }, { "epoch": 2.39, "learning_rate": 1.7704545454545455e-05, "loss": 5.8779, "step": 2129 }, { "epoch": 2.39, "learning_rate": 1.76969696969697e-05, "loss": 5.6369, "step": 2130 }, { "epoch": 2.39, "learning_rate": 1.768939393939394e-05, "loss": 5.3335, "step": 2131 }, { "epoch": 2.39, "learning_rate": 1.768181818181818e-05, "loss": 5.345, "step": 2132 }, { "epoch": 2.39, "learning_rate": 1.7674242424242427e-05, "loss": 5.209, "step": 2133 }, { "epoch": 2.39, "learning_rate": 1.7666666666666668e-05, "loss": 4.5496, "step": 2134 }, { "epoch": 2.39, "learning_rate": 1.765909090909091e-05, "loss": 6.5626, "step": 2135 }, { "epoch": 2.39, "learning_rate": 1.765151515151515e-05, "loss": 6.6633, "step": 2136 }, { "epoch": 2.4, "learning_rate": 1.7643939393939397e-05, "loss": 6.478, "step": 2137 }, { "epoch": 2.4, "learning_rate": 1.7636363636363637e-05, "loss": 6.4729, "step": 2138 }, { "epoch": 2.4, "learning_rate": 1.7628787878787878e-05, "loss": 6.567, "step": 2139 }, { "epoch": 2.4, "learning_rate": 1.7621212121212122e-05, "loss": 6.4651, "step": 2140 }, { "epoch": 2.4, "learning_rate": 1.7613636363636366e-05, "loss": 6.2488, "step": 2141 }, { "epoch": 2.4, "learning_rate": 1.7606060606060606e-05, "loss": 6.6085, "step": 2142 }, { "epoch": 2.4, "learning_rate": 1.7598484848484847e-05, "loss": 6.3854, "step": 2143 }, { "epoch": 2.4, "learning_rate": 1.759090909090909e-05, "loss": 6.1812, "step": 2144 }, { "epoch": 2.4, "learning_rate": 1.7583333333333335e-05, "loss": 6.1669, "step": 2145 }, { "epoch": 2.41, "learning_rate": 1.7575757575757576e-05, "loss": 6.2759, "step": 2146 }, { "epoch": 2.41, "learning_rate": 1.7568181818181816e-05, "loss": 6.4001, "step": 2147 }, { "epoch": 2.41, "learning_rate": 1.7560606060606064e-05, "loss": 6.2398, "step": 2148 }, { "epoch": 2.41, "learning_rate": 1.7553030303030304e-05, "loss": 6.2594, "step": 2149 }, { "epoch": 2.41, "learning_rate": 1.7545454545454545e-05, "loss": 6.3477, "step": 2150 }, { "epoch": 2.41, "learning_rate": 1.753787878787879e-05, "loss": 6.3323, "step": 2151 }, { "epoch": 2.41, "learning_rate": 1.7530303030303033e-05, "loss": 6.3467, "step": 2152 }, { "epoch": 2.41, "learning_rate": 1.7522727272727273e-05, "loss": 6.1617, "step": 2153 }, { "epoch": 2.41, "learning_rate": 1.7515151515151514e-05, "loss": 6.3141, "step": 2154 }, { "epoch": 2.42, "learning_rate": 1.7507575757575758e-05, "loss": 6.3743, "step": 2155 }, { "epoch": 2.42, "learning_rate": 1.7500000000000002e-05, "loss": 6.198, "step": 2156 }, { "epoch": 2.42, "learning_rate": 1.7492424242424242e-05, "loss": 6.1177, "step": 2157 }, { "epoch": 2.42, "learning_rate": 1.7484848484848483e-05, "loss": 6.1189, "step": 2158 }, { "epoch": 2.42, "learning_rate": 1.7477272727272727e-05, "loss": 6.1955, "step": 2159 }, { "epoch": 2.42, "learning_rate": 1.746969696969697e-05, "loss": 6.2785, "step": 2160 }, { "epoch": 2.42, "learning_rate": 1.746212121212121e-05, "loss": 6.2275, "step": 2161 }, { "epoch": 2.42, "learning_rate": 1.7454545454545456e-05, "loss": 6.0509, "step": 2162 }, { "epoch": 2.42, "learning_rate": 1.74469696969697e-05, "loss": 6.1924, "step": 2163 }, { "epoch": 2.43, "learning_rate": 1.743939393939394e-05, "loss": 6.2093, "step": 2164 }, { "epoch": 2.43, "learning_rate": 1.743181818181818e-05, "loss": 6.2134, "step": 2165 }, { "epoch": 2.43, "learning_rate": 1.7424242424242425e-05, "loss": 6.1109, "step": 2166 }, { "epoch": 2.43, "learning_rate": 1.741666666666667e-05, "loss": 6.2305, "step": 2167 }, { "epoch": 2.43, "learning_rate": 1.740909090909091e-05, "loss": 6.1764, "step": 2168 }, { "epoch": 2.43, "learning_rate": 1.740151515151515e-05, "loss": 6.1159, "step": 2169 }, { "epoch": 2.43, "learning_rate": 1.7393939393939394e-05, "loss": 6.1775, "step": 2170 }, { "epoch": 2.43, "learning_rate": 1.7386363636363638e-05, "loss": 6.2543, "step": 2171 }, { "epoch": 2.43, "learning_rate": 1.737878787878788e-05, "loss": 6.2607, "step": 2172 }, { "epoch": 2.44, "learning_rate": 1.7371212121212123e-05, "loss": 6.0468, "step": 2173 }, { "epoch": 2.44, "learning_rate": 1.7363636363636363e-05, "loss": 5.9744, "step": 2174 }, { "epoch": 2.44, "learning_rate": 1.7356060606060607e-05, "loss": 6.1642, "step": 2175 }, { "epoch": 2.44, "learning_rate": 1.7348484848484848e-05, "loss": 5.9042, "step": 2176 }, { "epoch": 2.44, "learning_rate": 1.734090909090909e-05, "loss": 6.0546, "step": 2177 }, { "epoch": 2.44, "learning_rate": 1.7333333333333332e-05, "loss": 5.9854, "step": 2178 }, { "epoch": 2.44, "learning_rate": 1.7325757575757576e-05, "loss": 5.7982, "step": 2179 }, { "epoch": 2.44, "learning_rate": 1.731818181818182e-05, "loss": 5.5125, "step": 2180 }, { "epoch": 2.45, "learning_rate": 1.731060606060606e-05, "loss": 5.4571, "step": 2181 }, { "epoch": 2.45, "learning_rate": 1.7303030303030305e-05, "loss": 5.464, "step": 2182 }, { "epoch": 2.45, "learning_rate": 1.7295454545454545e-05, "loss": 4.7765, "step": 2183 }, { "epoch": 2.45, "learning_rate": 1.728787878787879e-05, "loss": 4.7393, "step": 2184 }, { "epoch": 2.45, "learning_rate": 1.728030303030303e-05, "loss": 6.5339, "step": 2185 }, { "epoch": 2.45, "learning_rate": 1.7272727272727274e-05, "loss": 6.5982, "step": 2186 }, { "epoch": 2.45, "learning_rate": 1.7265151515151515e-05, "loss": 6.6287, "step": 2187 }, { "epoch": 2.45, "learning_rate": 1.725757575757576e-05, "loss": 6.551, "step": 2188 }, { "epoch": 2.45, "learning_rate": 1.725e-05, "loss": 6.4877, "step": 2189 }, { "epoch": 2.46, "learning_rate": 1.7242424242424243e-05, "loss": 6.3503, "step": 2190 }, { "epoch": 2.46, "learning_rate": 1.7234848484848487e-05, "loss": 6.44, "step": 2191 }, { "epoch": 2.46, "learning_rate": 1.7227272727272728e-05, "loss": 6.4668, "step": 2192 }, { "epoch": 2.46, "learning_rate": 1.721969696969697e-05, "loss": 6.3961, "step": 2193 }, { "epoch": 2.46, "learning_rate": 1.7212121212121212e-05, "loss": 6.3723, "step": 2194 }, { "epoch": 2.46, "learning_rate": 1.7204545454545456e-05, "loss": 6.3272, "step": 2195 }, { "epoch": 2.46, "learning_rate": 1.7196969696969697e-05, "loss": 6.2044, "step": 2196 }, { "epoch": 2.46, "learning_rate": 1.718939393939394e-05, "loss": 6.1727, "step": 2197 }, { "epoch": 2.46, "learning_rate": 1.718181818181818e-05, "loss": 6.1624, "step": 2198 }, { "epoch": 2.47, "learning_rate": 1.7174242424242425e-05, "loss": 6.3122, "step": 2199 }, { "epoch": 2.47, "learning_rate": 1.7166666666666666e-05, "loss": 6.2299, "step": 2200 }, { "epoch": 2.47, "learning_rate": 1.715909090909091e-05, "loss": 6.2892, "step": 2201 }, { "epoch": 2.47, "learning_rate": 1.7151515151515154e-05, "loss": 6.2339, "step": 2202 }, { "epoch": 2.47, "learning_rate": 1.7143939393939395e-05, "loss": 6.1226, "step": 2203 }, { "epoch": 2.47, "learning_rate": 1.7136363636363635e-05, "loss": 6.3123, "step": 2204 }, { "epoch": 2.47, "learning_rate": 1.712878787878788e-05, "loss": 6.2653, "step": 2205 }, { "epoch": 2.47, "learning_rate": 1.7121212121212123e-05, "loss": 6.2088, "step": 2206 }, { "epoch": 2.47, "learning_rate": 1.7113636363636364e-05, "loss": 6.2609, "step": 2207 }, { "epoch": 2.48, "learning_rate": 1.7106060606060604e-05, "loss": 6.288, "step": 2208 }, { "epoch": 2.48, "learning_rate": 1.709848484848485e-05, "loss": 6.277, "step": 2209 }, { "epoch": 2.48, "learning_rate": 1.7090909090909092e-05, "loss": 6.0645, "step": 2210 }, { "epoch": 2.48, "learning_rate": 1.7083333333333333e-05, "loss": 6.3715, "step": 2211 }, { "epoch": 2.48, "learning_rate": 1.7075757575757574e-05, "loss": 6.1733, "step": 2212 }, { "epoch": 2.48, "learning_rate": 1.706818181818182e-05, "loss": 6.36, "step": 2213 }, { "epoch": 2.48, "learning_rate": 1.706060606060606e-05, "loss": 6.2126, "step": 2214 }, { "epoch": 2.48, "learning_rate": 1.7053030303030302e-05, "loss": 6.3901, "step": 2215 }, { "epoch": 2.48, "learning_rate": 1.7045454545454546e-05, "loss": 6.1325, "step": 2216 }, { "epoch": 2.49, "learning_rate": 1.703787878787879e-05, "loss": 6.1698, "step": 2217 }, { "epoch": 2.49, "learning_rate": 1.703030303030303e-05, "loss": 6.0617, "step": 2218 }, { "epoch": 2.49, "learning_rate": 1.702272727272727e-05, "loss": 6.1063, "step": 2219 }, { "epoch": 2.49, "learning_rate": 1.7015151515151515e-05, "loss": 6.2368, "step": 2220 }, { "epoch": 2.49, "learning_rate": 1.700757575757576e-05, "loss": 6.1298, "step": 2221 }, { "epoch": 2.49, "learning_rate": 1.7e-05, "loss": 6.091, "step": 2222 }, { "epoch": 2.49, "learning_rate": 1.699242424242424e-05, "loss": 6.234, "step": 2223 }, { "epoch": 2.49, "learning_rate": 1.6984848484848488e-05, "loss": 6.0621, "step": 2224 }, { "epoch": 2.49, "learning_rate": 1.697727272727273e-05, "loss": 5.9083, "step": 2225 }, { "epoch": 2.5, "learning_rate": 1.696969696969697e-05, "loss": 5.8474, "step": 2226 }, { "epoch": 2.5, "learning_rate": 1.696212121212121e-05, "loss": 5.859, "step": 2227 }, { "epoch": 2.5, "learning_rate": 1.6954545454545457e-05, "loss": 5.515, "step": 2228 }, { "epoch": 2.5, "learning_rate": 1.6946969696969698e-05, "loss": 5.6851, "step": 2229 }, { "epoch": 2.5, "learning_rate": 1.6939393939393938e-05, "loss": 5.4398, "step": 2230 }, { "epoch": 2.5, "learning_rate": 1.6931818181818182e-05, "loss": 5.3542, "step": 2231 }, { "epoch": 2.5, "learning_rate": 1.6924242424242426e-05, "loss": 5.4139, "step": 2232 }, { "epoch": 2.5, "learning_rate": 1.6916666666666667e-05, "loss": 5.2254, "step": 2233 }, { "epoch": 2.5, "learning_rate": 1.6909090909090907e-05, "loss": 4.7844, "step": 2234 }, { "epoch": 2.51, "learning_rate": 1.6901515151515155e-05, "loss": 6.6234, "step": 2235 }, { "epoch": 2.51, "learning_rate": 1.6893939393939395e-05, "loss": 6.54, "step": 2236 }, { "epoch": 2.51, "learning_rate": 1.6886363636363636e-05, "loss": 6.6488, "step": 2237 }, { "epoch": 2.51, "learning_rate": 1.6878787878787877e-05, "loss": 6.6699, "step": 2238 }, { "epoch": 2.51, "learning_rate": 1.6871212121212124e-05, "loss": 6.389, "step": 2239 }, { "epoch": 2.51, "learning_rate": 1.6863636363636364e-05, "loss": 6.4478, "step": 2240 }, { "epoch": 2.51, "learning_rate": 1.6856060606060605e-05, "loss": 6.3789, "step": 2241 }, { "epoch": 2.51, "learning_rate": 1.684848484848485e-05, "loss": 6.407, "step": 2242 }, { "epoch": 2.51, "learning_rate": 1.6840909090909093e-05, "loss": 6.3832, "step": 2243 }, { "epoch": 2.52, "learning_rate": 1.6833333333333334e-05, "loss": 6.3429, "step": 2244 }, { "epoch": 2.52, "learning_rate": 1.6825757575757574e-05, "loss": 6.1699, "step": 2245 }, { "epoch": 2.52, "learning_rate": 1.6818181818181818e-05, "loss": 6.1983, "step": 2246 }, { "epoch": 2.52, "learning_rate": 1.6810606060606062e-05, "loss": 6.2892, "step": 2247 }, { "epoch": 2.52, "learning_rate": 1.6803030303030303e-05, "loss": 6.3612, "step": 2248 }, { "epoch": 2.52, "learning_rate": 1.6795454545454543e-05, "loss": 6.0923, "step": 2249 }, { "epoch": 2.52, "learning_rate": 1.678787878787879e-05, "loss": 6.1724, "step": 2250 }, { "epoch": 2.52, "learning_rate": 1.678030303030303e-05, "loss": 6.1309, "step": 2251 }, { "epoch": 2.52, "learning_rate": 1.6772727272727272e-05, "loss": 6.3413, "step": 2252 }, { "epoch": 2.53, "learning_rate": 1.6765151515151516e-05, "loss": 6.1572, "step": 2253 }, { "epoch": 2.53, "learning_rate": 1.675757575757576e-05, "loss": 6.2519, "step": 2254 }, { "epoch": 2.53, "learning_rate": 1.675e-05, "loss": 6.2772, "step": 2255 }, { "epoch": 2.53, "learning_rate": 1.674242424242424e-05, "loss": 6.1342, "step": 2256 }, { "epoch": 2.53, "learning_rate": 1.6734848484848485e-05, "loss": 6.1476, "step": 2257 }, { "epoch": 2.53, "learning_rate": 1.672727272727273e-05, "loss": 6.076, "step": 2258 }, { "epoch": 2.53, "learning_rate": 1.671969696969697e-05, "loss": 6.2645, "step": 2259 }, { "epoch": 2.53, "learning_rate": 1.671212121212121e-05, "loss": 6.2382, "step": 2260 }, { "epoch": 2.53, "learning_rate": 1.6704545454545454e-05, "loss": 6.067, "step": 2261 }, { "epoch": 2.54, "learning_rate": 1.6696969696969698e-05, "loss": 6.2117, "step": 2262 }, { "epoch": 2.54, "learning_rate": 1.668939393939394e-05, "loss": 6.2231, "step": 2263 }, { "epoch": 2.54, "learning_rate": 1.6681818181818183e-05, "loss": 6.3249, "step": 2264 }, { "epoch": 2.54, "learning_rate": 1.6674242424242427e-05, "loss": 6.217, "step": 2265 }, { "epoch": 2.54, "learning_rate": 1.6666666666666667e-05, "loss": 6.1476, "step": 2266 }, { "epoch": 2.54, "learning_rate": 1.6659090909090908e-05, "loss": 6.2188, "step": 2267 }, { "epoch": 2.54, "learning_rate": 1.6651515151515152e-05, "loss": 6.2504, "step": 2268 }, { "epoch": 2.54, "learning_rate": 1.6643939393939396e-05, "loss": 6.2463, "step": 2269 }, { "epoch": 2.54, "learning_rate": 1.6636363636363637e-05, "loss": 6.2853, "step": 2270 }, { "epoch": 2.55, "learning_rate": 1.6628787878787877e-05, "loss": 6.1603, "step": 2271 }, { "epoch": 2.55, "learning_rate": 1.662121212121212e-05, "loss": 6.0146, "step": 2272 }, { "epoch": 2.55, "learning_rate": 1.6613636363636365e-05, "loss": 6.1538, "step": 2273 }, { "epoch": 2.55, "learning_rate": 1.6606060606060606e-05, "loss": 6.0606, "step": 2274 }, { "epoch": 2.55, "learning_rate": 1.659848484848485e-05, "loss": 6.0985, "step": 2275 }, { "epoch": 2.55, "learning_rate": 1.659090909090909e-05, "loss": 6.0173, "step": 2276 }, { "epoch": 2.55, "learning_rate": 1.6583333333333334e-05, "loss": 5.9564, "step": 2277 }, { "epoch": 2.55, "learning_rate": 1.6575757575757575e-05, "loss": 5.991, "step": 2278 }, { "epoch": 2.55, "learning_rate": 1.656818181818182e-05, "loss": 6.0018, "step": 2279 }, { "epoch": 2.56, "learning_rate": 1.656060606060606e-05, "loss": 5.7132, "step": 2280 }, { "epoch": 2.56, "learning_rate": 1.6553030303030304e-05, "loss": 5.5543, "step": 2281 }, { "epoch": 2.56, "learning_rate": 1.6545454545454548e-05, "loss": 5.3687, "step": 2282 }, { "epoch": 2.56, "learning_rate": 1.6537878787878788e-05, "loss": 5.052, "step": 2283 }, { "epoch": 2.56, "learning_rate": 1.6530303030303032e-05, "loss": 4.6106, "step": 2284 }, { "epoch": 2.56, "learning_rate": 1.6522727272727273e-05, "loss": 6.6253, "step": 2285 }, { "epoch": 2.56, "learning_rate": 1.6515151515151517e-05, "loss": 6.3646, "step": 2286 }, { "epoch": 2.56, "learning_rate": 1.6507575757575757e-05, "loss": 6.3842, "step": 2287 }, { "epoch": 2.57, "learning_rate": 1.65e-05, "loss": 6.4279, "step": 2288 }, { "epoch": 2.57, "learning_rate": 1.6492424242424242e-05, "loss": 6.3439, "step": 2289 }, { "epoch": 2.57, "learning_rate": 1.6484848484848486e-05, "loss": 6.3667, "step": 2290 }, { "epoch": 2.57, "learning_rate": 1.6477272727272726e-05, "loss": 6.3497, "step": 2291 }, { "epoch": 2.57, "learning_rate": 1.646969696969697e-05, "loss": 6.2358, "step": 2292 }, { "epoch": 2.57, "learning_rate": 1.6462121212121214e-05, "loss": 6.3865, "step": 2293 }, { "epoch": 2.57, "learning_rate": 1.6454545454545455e-05, "loss": 6.329, "step": 2294 }, { "epoch": 2.57, "learning_rate": 1.6446969696969696e-05, "loss": 6.2545, "step": 2295 }, { "epoch": 2.57, "learning_rate": 1.643939393939394e-05, "loss": 6.3754, "step": 2296 }, { "epoch": 2.58, "learning_rate": 1.6431818181818184e-05, "loss": 6.3176, "step": 2297 }, { "epoch": 2.58, "learning_rate": 1.6424242424242424e-05, "loss": 6.3208, "step": 2298 }, { "epoch": 2.58, "learning_rate": 1.6416666666666668e-05, "loss": 6.3146, "step": 2299 }, { "epoch": 2.58, "learning_rate": 1.640909090909091e-05, "loss": 6.1334, "step": 2300 }, { "epoch": 2.58, "learning_rate": 1.6401515151515153e-05, "loss": 6.2501, "step": 2301 }, { "epoch": 2.58, "learning_rate": 1.6393939393939393e-05, "loss": 6.1707, "step": 2302 }, { "epoch": 2.58, "learning_rate": 1.6386363636363637e-05, "loss": 6.1745, "step": 2303 }, { "epoch": 2.58, "learning_rate": 1.637878787878788e-05, "loss": 6.2051, "step": 2304 }, { "epoch": 2.58, "learning_rate": 1.6371212121212122e-05, "loss": 6.1013, "step": 2305 }, { "epoch": 2.59, "learning_rate": 1.6363636363636363e-05, "loss": 6.1721, "step": 2306 }, { "epoch": 2.59, "learning_rate": 1.6356060606060606e-05, "loss": 6.2579, "step": 2307 }, { "epoch": 2.59, "learning_rate": 1.634848484848485e-05, "loss": 6.2452, "step": 2308 }, { "epoch": 2.59, "learning_rate": 1.634090909090909e-05, "loss": 6.3156, "step": 2309 }, { "epoch": 2.59, "learning_rate": 1.633333333333333e-05, "loss": 6.2178, "step": 2310 }, { "epoch": 2.59, "learning_rate": 1.6325757575757576e-05, "loss": 6.097, "step": 2311 }, { "epoch": 2.59, "learning_rate": 1.631818181818182e-05, "loss": 6.1595, "step": 2312 }, { "epoch": 2.59, "learning_rate": 1.631060606060606e-05, "loss": 6.2152, "step": 2313 }, { "epoch": 2.59, "learning_rate": 1.63030303030303e-05, "loss": 6.2021, "step": 2314 }, { "epoch": 2.6, "learning_rate": 1.6295454545454548e-05, "loss": 6.0207, "step": 2315 }, { "epoch": 2.6, "learning_rate": 1.628787878787879e-05, "loss": 6.0533, "step": 2316 }, { "epoch": 2.6, "learning_rate": 1.628030303030303e-05, "loss": 6.1329, "step": 2317 }, { "epoch": 2.6, "learning_rate": 1.6272727272727273e-05, "loss": 6.0721, "step": 2318 }, { "epoch": 2.6, "learning_rate": 1.6265151515151517e-05, "loss": 6.2194, "step": 2319 }, { "epoch": 2.6, "learning_rate": 1.6257575757575758e-05, "loss": 6.2839, "step": 2320 }, { "epoch": 2.6, "learning_rate": 1.625e-05, "loss": 6.1655, "step": 2321 }, { "epoch": 2.6, "learning_rate": 1.6242424242424243e-05, "loss": 6.1059, "step": 2322 }, { "epoch": 2.6, "learning_rate": 1.6234848484848487e-05, "loss": 6.1311, "step": 2323 }, { "epoch": 2.61, "learning_rate": 1.6227272727272727e-05, "loss": 6.008, "step": 2324 }, { "epoch": 2.61, "learning_rate": 1.6219696969696968e-05, "loss": 6.1076, "step": 2325 }, { "epoch": 2.61, "learning_rate": 1.6212121212121215e-05, "loss": 5.956, "step": 2326 }, { "epoch": 2.61, "learning_rate": 1.6204545454545456e-05, "loss": 5.9741, "step": 2327 }, { "epoch": 2.61, "learning_rate": 1.6196969696969696e-05, "loss": 5.745, "step": 2328 }, { "epoch": 2.61, "learning_rate": 1.6189393939393937e-05, "loss": 5.6903, "step": 2329 }, { "epoch": 2.61, "learning_rate": 1.6181818181818184e-05, "loss": 5.5106, "step": 2330 }, { "epoch": 2.61, "learning_rate": 1.6174242424242425e-05, "loss": 5.381, "step": 2331 }, { "epoch": 2.61, "learning_rate": 1.6166666666666665e-05, "loss": 5.0386, "step": 2332 }, { "epoch": 2.62, "learning_rate": 1.615909090909091e-05, "loss": 5.138, "step": 2333 }, { "epoch": 2.62, "learning_rate": 1.6151515151515153e-05, "loss": 4.2886, "step": 2334 }, { "epoch": 2.62, "learning_rate": 1.6143939393939394e-05, "loss": 6.5621, "step": 2335 }, { "epoch": 2.62, "learning_rate": 1.6136363636363635e-05, "loss": 6.5151, "step": 2336 }, { "epoch": 2.62, "learning_rate": 1.6128787878787882e-05, "loss": 6.4804, "step": 2337 }, { "epoch": 2.62, "learning_rate": 1.6121212121212123e-05, "loss": 6.4992, "step": 2338 }, { "epoch": 2.62, "learning_rate": 1.6113636363636363e-05, "loss": 6.3768, "step": 2339 }, { "epoch": 2.62, "learning_rate": 1.6106060606060604e-05, "loss": 6.51, "step": 2340 }, { "epoch": 2.62, "learning_rate": 1.609848484848485e-05, "loss": 6.404, "step": 2341 }, { "epoch": 2.63, "learning_rate": 1.6090909090909092e-05, "loss": 6.3597, "step": 2342 }, { "epoch": 2.63, "learning_rate": 1.6083333333333332e-05, "loss": 6.3633, "step": 2343 }, { "epoch": 2.63, "learning_rate": 1.6075757575757576e-05, "loss": 6.2775, "step": 2344 }, { "epoch": 2.63, "learning_rate": 1.606818181818182e-05, "loss": 6.263, "step": 2345 }, { "epoch": 2.63, "learning_rate": 1.606060606060606e-05, "loss": 6.2348, "step": 2346 }, { "epoch": 2.63, "learning_rate": 1.60530303030303e-05, "loss": 6.2594, "step": 2347 }, { "epoch": 2.63, "learning_rate": 1.6045454545454546e-05, "loss": 6.31, "step": 2348 }, { "epoch": 2.63, "learning_rate": 1.603787878787879e-05, "loss": 6.175, "step": 2349 }, { "epoch": 2.63, "learning_rate": 1.603030303030303e-05, "loss": 6.381, "step": 2350 }, { "epoch": 2.64, "learning_rate": 1.602272727272727e-05, "loss": 6.2303, "step": 2351 }, { "epoch": 2.64, "learning_rate": 1.6015151515151518e-05, "loss": 6.172, "step": 2352 }, { "epoch": 2.64, "learning_rate": 1.600757575757576e-05, "loss": 6.0969, "step": 2353 }, { "epoch": 2.64, "learning_rate": 1.6e-05, "loss": 6.1309, "step": 2354 }, { "epoch": 2.64, "learning_rate": 1.5992424242424243e-05, "loss": 6.2066, "step": 2355 }, { "epoch": 2.64, "learning_rate": 1.5984848484848487e-05, "loss": 6.2233, "step": 2356 }, { "epoch": 2.64, "learning_rate": 1.5977272727272728e-05, "loss": 6.1099, "step": 2357 }, { "epoch": 2.64, "learning_rate": 1.596969696969697e-05, "loss": 6.097, "step": 2358 }, { "epoch": 2.64, "learning_rate": 1.5962121212121212e-05, "loss": 6.3359, "step": 2359 }, { "epoch": 2.65, "learning_rate": 1.5954545454545456e-05, "loss": 6.0746, "step": 2360 }, { "epoch": 2.65, "learning_rate": 1.5946969696969697e-05, "loss": 6.2887, "step": 2361 }, { "epoch": 2.65, "learning_rate": 1.5939393939393938e-05, "loss": 6.1928, "step": 2362 }, { "epoch": 2.65, "learning_rate": 1.593181818181818e-05, "loss": 6.159, "step": 2363 }, { "epoch": 2.65, "learning_rate": 1.5924242424242426e-05, "loss": 6.1558, "step": 2364 }, { "epoch": 2.65, "learning_rate": 1.5916666666666666e-05, "loss": 6.088, "step": 2365 }, { "epoch": 2.65, "learning_rate": 1.590909090909091e-05, "loss": 6.2103, "step": 2366 }, { "epoch": 2.65, "learning_rate": 1.5901515151515154e-05, "loss": 6.2256, "step": 2367 }, { "epoch": 2.65, "learning_rate": 1.5893939393939395e-05, "loss": 6.0581, "step": 2368 }, { "epoch": 2.66, "learning_rate": 1.5886363636363635e-05, "loss": 6.2129, "step": 2369 }, { "epoch": 2.66, "learning_rate": 1.587878787878788e-05, "loss": 6.1059, "step": 2370 }, { "epoch": 2.66, "learning_rate": 1.5871212121212123e-05, "loss": 6.1742, "step": 2371 }, { "epoch": 2.66, "learning_rate": 1.5863636363636364e-05, "loss": 6.2758, "step": 2372 }, { "epoch": 2.66, "learning_rate": 1.5856060606060605e-05, "loss": 5.9828, "step": 2373 }, { "epoch": 2.66, "learning_rate": 1.584848484848485e-05, "loss": 5.9689, "step": 2374 }, { "epoch": 2.66, "learning_rate": 1.5840909090909092e-05, "loss": 6.096, "step": 2375 }, { "epoch": 2.66, "learning_rate": 1.5833333333333333e-05, "loss": 5.9787, "step": 2376 }, { "epoch": 2.66, "learning_rate": 1.5825757575757577e-05, "loss": 5.9504, "step": 2377 }, { "epoch": 2.67, "learning_rate": 1.5818181818181818e-05, "loss": 5.8789, "step": 2378 }, { "epoch": 2.67, "learning_rate": 1.581060606060606e-05, "loss": 5.6282, "step": 2379 }, { "epoch": 2.67, "learning_rate": 1.5803030303030302e-05, "loss": 5.662, "step": 2380 }, { "epoch": 2.67, "learning_rate": 1.5795454545454546e-05, "loss": 5.6678, "step": 2381 }, { "epoch": 2.67, "learning_rate": 1.5787878787878787e-05, "loss": 5.2082, "step": 2382 }, { "epoch": 2.67, "learning_rate": 1.578030303030303e-05, "loss": 5.1731, "step": 2383 }, { "epoch": 2.67, "learning_rate": 1.5772727272727275e-05, "loss": 4.5313, "step": 2384 }, { "epoch": 2.67, "learning_rate": 1.5765151515151515e-05, "loss": 6.5904, "step": 2385 }, { "epoch": 2.67, "learning_rate": 1.575757575757576e-05, "loss": 6.5336, "step": 2386 }, { "epoch": 2.68, "learning_rate": 1.575e-05, "loss": 6.4042, "step": 2387 }, { "epoch": 2.68, "learning_rate": 1.5742424242424244e-05, "loss": 6.4659, "step": 2388 }, { "epoch": 2.68, "learning_rate": 1.5734848484848485e-05, "loss": 6.3525, "step": 2389 }, { "epoch": 2.68, "learning_rate": 1.572727272727273e-05, "loss": 6.3983, "step": 2390 }, { "epoch": 2.68, "learning_rate": 1.571969696969697e-05, "loss": 6.2894, "step": 2391 }, { "epoch": 2.68, "learning_rate": 1.5712121212121213e-05, "loss": 6.3466, "step": 2392 }, { "epoch": 2.68, "learning_rate": 1.5704545454545454e-05, "loss": 6.3099, "step": 2393 }, { "epoch": 2.68, "learning_rate": 1.5696969696969698e-05, "loss": 6.0441, "step": 2394 }, { "epoch": 2.68, "learning_rate": 1.5689393939393942e-05, "loss": 6.0321, "step": 2395 }, { "epoch": 2.69, "learning_rate": 1.5681818181818182e-05, "loss": 6.314, "step": 2396 }, { "epoch": 2.69, "learning_rate": 1.5674242424242423e-05, "loss": 6.1446, "step": 2397 }, { "epoch": 2.69, "learning_rate": 1.5666666666666667e-05, "loss": 6.0762, "step": 2398 }, { "epoch": 2.69, "learning_rate": 1.565909090909091e-05, "loss": 6.1385, "step": 2399 }, { "epoch": 2.69, "learning_rate": 1.565151515151515e-05, "loss": 6.1033, "step": 2400 }, { "epoch": 2.69, "learning_rate": 1.5643939393939395e-05, "loss": 6.2918, "step": 2401 }, { "epoch": 2.69, "learning_rate": 1.5636363636363636e-05, "loss": 6.1287, "step": 2402 }, { "epoch": 2.69, "learning_rate": 1.562878787878788e-05, "loss": 6.2078, "step": 2403 }, { "epoch": 2.7, "learning_rate": 1.562121212121212e-05, "loss": 6.2242, "step": 2404 }, { "epoch": 2.7, "learning_rate": 1.5613636363636365e-05, "loss": 6.0876, "step": 2405 }, { "epoch": 2.7, "learning_rate": 1.560606060606061e-05, "loss": 6.322, "step": 2406 }, { "epoch": 2.7, "learning_rate": 1.559848484848485e-05, "loss": 6.3674, "step": 2407 }, { "epoch": 2.7, "learning_rate": 1.559090909090909e-05, "loss": 6.1519, "step": 2408 }, { "epoch": 2.7, "learning_rate": 1.5583333333333334e-05, "loss": 6.1022, "step": 2409 }, { "epoch": 2.7, "learning_rate": 1.5575757575757578e-05, "loss": 6.0965, "step": 2410 }, { "epoch": 2.7, "learning_rate": 1.556818181818182e-05, "loss": 6.1361, "step": 2411 }, { "epoch": 2.7, "learning_rate": 1.556060606060606e-05, "loss": 6.0671, "step": 2412 }, { "epoch": 2.71, "learning_rate": 1.5553030303030303e-05, "loss": 6.1983, "step": 2413 }, { "epoch": 2.71, "learning_rate": 1.5545454545454547e-05, "loss": 6.1084, "step": 2414 }, { "epoch": 2.71, "learning_rate": 1.5537878787878788e-05, "loss": 6.1555, "step": 2415 }, { "epoch": 2.71, "learning_rate": 1.5530303030303028e-05, "loss": 6.1856, "step": 2416 }, { "epoch": 2.71, "learning_rate": 1.5522727272727275e-05, "loss": 6.0217, "step": 2417 }, { "epoch": 2.71, "learning_rate": 1.5515151515151516e-05, "loss": 6.1515, "step": 2418 }, { "epoch": 2.71, "learning_rate": 1.5507575757575757e-05, "loss": 5.9787, "step": 2419 }, { "epoch": 2.71, "learning_rate": 1.55e-05, "loss": 6.1138, "step": 2420 }, { "epoch": 2.71, "learning_rate": 1.5492424242424245e-05, "loss": 6.1708, "step": 2421 }, { "epoch": 2.72, "learning_rate": 1.5484848484848485e-05, "loss": 6.1242, "step": 2422 }, { "epoch": 2.72, "learning_rate": 1.5477272727272726e-05, "loss": 5.974, "step": 2423 }, { "epoch": 2.72, "learning_rate": 1.546969696969697e-05, "loss": 5.9227, "step": 2424 }, { "epoch": 2.72, "learning_rate": 1.5462121212121214e-05, "loss": 6.1227, "step": 2425 }, { "epoch": 2.72, "learning_rate": 1.5454545454545454e-05, "loss": 5.8331, "step": 2426 }, { "epoch": 2.72, "learning_rate": 1.5446969696969695e-05, "loss": 5.9052, "step": 2427 }, { "epoch": 2.72, "learning_rate": 1.5439393939393942e-05, "loss": 5.9818, "step": 2428 }, { "epoch": 2.72, "learning_rate": 1.5431818181818183e-05, "loss": 5.7801, "step": 2429 }, { "epoch": 2.72, "learning_rate": 1.5424242424242424e-05, "loss": 5.569, "step": 2430 }, { "epoch": 2.73, "learning_rate": 1.5416666666666664e-05, "loss": 5.5634, "step": 2431 }, { "epoch": 2.73, "learning_rate": 1.540909090909091e-05, "loss": 5.4026, "step": 2432 }, { "epoch": 2.73, "learning_rate": 1.5401515151515152e-05, "loss": 4.9031, "step": 2433 }, { "epoch": 2.73, "learning_rate": 1.5393939393939393e-05, "loss": 4.1569, "step": 2434 }, { "epoch": 2.73, "learning_rate": 1.5386363636363637e-05, "loss": 6.7125, "step": 2435 }, { "epoch": 2.73, "learning_rate": 1.537878787878788e-05, "loss": 6.4626, "step": 2436 }, { "epoch": 2.73, "learning_rate": 1.537121212121212e-05, "loss": 6.6231, "step": 2437 }, { "epoch": 2.73, "learning_rate": 1.5363636363636362e-05, "loss": 6.556, "step": 2438 }, { "epoch": 2.73, "learning_rate": 1.535606060606061e-05, "loss": 6.4251, "step": 2439 }, { "epoch": 2.74, "learning_rate": 1.534848484848485e-05, "loss": 6.4817, "step": 2440 }, { "epoch": 2.74, "learning_rate": 1.534090909090909e-05, "loss": 6.5593, "step": 2441 }, { "epoch": 2.74, "learning_rate": 1.533333333333333e-05, "loss": 6.374, "step": 2442 }, { "epoch": 2.74, "learning_rate": 1.532575757575758e-05, "loss": 6.2992, "step": 2443 }, { "epoch": 2.74, "learning_rate": 1.531818181818182e-05, "loss": 6.2547, "step": 2444 }, { "epoch": 2.74, "learning_rate": 1.531060606060606e-05, "loss": 6.2012, "step": 2445 }, { "epoch": 2.74, "learning_rate": 1.5303030303030304e-05, "loss": 6.2569, "step": 2446 }, { "epoch": 2.74, "learning_rate": 1.5295454545454548e-05, "loss": 6.0617, "step": 2447 }, { "epoch": 2.74, "learning_rate": 1.5287878787878788e-05, "loss": 6.3644, "step": 2448 }, { "epoch": 2.75, "learning_rate": 1.528030303030303e-05, "loss": 6.267, "step": 2449 }, { "epoch": 2.75, "learning_rate": 1.5272727272727273e-05, "loss": 6.2829, "step": 2450 }, { "epoch": 2.75, "learning_rate": 1.5265151515151517e-05, "loss": 6.3796, "step": 2451 }, { "epoch": 2.75, "learning_rate": 1.5257575757575757e-05, "loss": 6.2653, "step": 2452 }, { "epoch": 2.75, "learning_rate": 1.525e-05, "loss": 6.0968, "step": 2453 }, { "epoch": 2.75, "learning_rate": 1.5242424242424244e-05, "loss": 6.357, "step": 2454 }, { "epoch": 2.75, "learning_rate": 1.5234848484848486e-05, "loss": 6.1704, "step": 2455 }, { "epoch": 2.75, "learning_rate": 1.5227272727272728e-05, "loss": 6.2998, "step": 2456 }, { "epoch": 2.75, "learning_rate": 1.5219696969696969e-05, "loss": 6.2565, "step": 2457 }, { "epoch": 2.76, "learning_rate": 1.5212121212121215e-05, "loss": 6.1454, "step": 2458 }, { "epoch": 2.76, "learning_rate": 1.5204545454545455e-05, "loss": 6.1968, "step": 2459 }, { "epoch": 2.76, "learning_rate": 1.5196969696969697e-05, "loss": 6.1256, "step": 2460 }, { "epoch": 2.76, "learning_rate": 1.5189393939393938e-05, "loss": 6.2607, "step": 2461 }, { "epoch": 2.76, "learning_rate": 1.5181818181818184e-05, "loss": 6.1487, "step": 2462 }, { "epoch": 2.76, "learning_rate": 1.5174242424242424e-05, "loss": 6.2005, "step": 2463 }, { "epoch": 2.76, "learning_rate": 1.5166666666666667e-05, "loss": 6.2415, "step": 2464 }, { "epoch": 2.76, "learning_rate": 1.5159090909090907e-05, "loss": 6.2879, "step": 2465 }, { "epoch": 2.76, "learning_rate": 1.5151515151515153e-05, "loss": 6.1119, "step": 2466 }, { "epoch": 2.77, "learning_rate": 1.5143939393939395e-05, "loss": 6.1503, "step": 2467 }, { "epoch": 2.77, "learning_rate": 1.5136363636363636e-05, "loss": 6.2391, "step": 2468 }, { "epoch": 2.77, "learning_rate": 1.5128787878787881e-05, "loss": 6.1167, "step": 2469 }, { "epoch": 2.77, "learning_rate": 1.5121212121212122e-05, "loss": 6.0606, "step": 2470 }, { "epoch": 2.77, "learning_rate": 1.5113636363636364e-05, "loss": 6.1354, "step": 2471 }, { "epoch": 2.77, "learning_rate": 1.5106060606060605e-05, "loss": 6.0569, "step": 2472 }, { "epoch": 2.77, "learning_rate": 1.509848484848485e-05, "loss": 5.9763, "step": 2473 }, { "epoch": 2.77, "learning_rate": 1.5090909090909091e-05, "loss": 5.9956, "step": 2474 }, { "epoch": 2.77, "learning_rate": 1.5083333333333333e-05, "loss": 6.0861, "step": 2475 }, { "epoch": 2.78, "learning_rate": 1.5075757575757576e-05, "loss": 5.9716, "step": 2476 }, { "epoch": 2.78, "learning_rate": 1.506818181818182e-05, "loss": 5.8972, "step": 2477 }, { "epoch": 2.78, "learning_rate": 1.5060606060606062e-05, "loss": 5.8129, "step": 2478 }, { "epoch": 2.78, "learning_rate": 1.5053030303030303e-05, "loss": 5.6568, "step": 2479 }, { "epoch": 2.78, "learning_rate": 1.5045454545454545e-05, "loss": 5.5304, "step": 2480 }, { "epoch": 2.78, "learning_rate": 1.5037878787878789e-05, "loss": 5.4146, "step": 2481 }, { "epoch": 2.78, "learning_rate": 1.5030303030303031e-05, "loss": 5.2958, "step": 2482 }, { "epoch": 2.78, "learning_rate": 1.5022727272727272e-05, "loss": 4.9631, "step": 2483 }, { "epoch": 2.78, "learning_rate": 1.5015151515151514e-05, "loss": 4.497, "step": 2484 }, { "epoch": 2.79, "learning_rate": 1.5007575757575758e-05, "loss": 6.4746, "step": 2485 }, { "epoch": 2.79, "learning_rate": 1.5e-05, "loss": 6.539, "step": 2486 }, { "epoch": 2.79, "learning_rate": 1.4992424242424244e-05, "loss": 6.5087, "step": 2487 }, { "epoch": 2.79, "learning_rate": 1.4984848484848485e-05, "loss": 6.5744, "step": 2488 }, { "epoch": 2.79, "learning_rate": 1.4977272727272729e-05, "loss": 6.5638, "step": 2489 }, { "epoch": 2.79, "learning_rate": 1.496969696969697e-05, "loss": 6.4194, "step": 2490 }, { "epoch": 2.79, "learning_rate": 1.4962121212121214e-05, "loss": 6.3728, "step": 2491 }, { "epoch": 2.79, "learning_rate": 1.4954545454545454e-05, "loss": 6.4915, "step": 2492 }, { "epoch": 2.79, "learning_rate": 1.4946969696969698e-05, "loss": 6.2857, "step": 2493 }, { "epoch": 2.8, "learning_rate": 1.4939393939393939e-05, "loss": 6.285, "step": 2494 }, { "epoch": 2.8, "learning_rate": 1.4931818181818183e-05, "loss": 6.2391, "step": 2495 }, { "epoch": 2.8, "learning_rate": 1.4924242424242423e-05, "loss": 6.3514, "step": 2496 }, { "epoch": 2.8, "learning_rate": 1.4916666666666667e-05, "loss": 6.2142, "step": 2497 }, { "epoch": 2.8, "learning_rate": 1.490909090909091e-05, "loss": 6.0427, "step": 2498 }, { "epoch": 2.8, "learning_rate": 1.4901515151515152e-05, "loss": 6.1209, "step": 2499 }, { "epoch": 2.8, "learning_rate": 1.4893939393939394e-05, "loss": 6.0888, "step": 2500 }, { "epoch": 2.8, "eval_loss": 5.845263481140137, "eval_runtime": 881.8914, "eval_samples_per_second": 2.996, "eval_steps_per_second": 0.375, "eval_wer": 1.7581118603728678, "step": 2500 }, { "epoch": 2.8, "learning_rate": 1.4886363636363636e-05, "loss": 6.2666, "step": 2501 }, { "epoch": 2.8, "learning_rate": 1.4878787878787879e-05, "loss": 6.1491, "step": 2502 }, { "epoch": 2.81, "learning_rate": 1.4871212121212121e-05, "loss": 6.0389, "step": 2503 }, { "epoch": 2.81, "learning_rate": 1.4863636363636365e-05, "loss": 6.1544, "step": 2504 }, { "epoch": 2.81, "learning_rate": 1.4856060606060606e-05, "loss": 6.2012, "step": 2505 }, { "epoch": 2.81, "learning_rate": 1.484848484848485e-05, "loss": 6.2036, "step": 2506 }, { "epoch": 2.81, "learning_rate": 1.4840909090909092e-05, "loss": 6.2521, "step": 2507 }, { "epoch": 2.81, "learning_rate": 1.4833333333333334e-05, "loss": 6.1957, "step": 2508 }, { "epoch": 2.81, "learning_rate": 1.4825757575757576e-05, "loss": 6.1968, "step": 2509 }, { "epoch": 2.81, "learning_rate": 1.4818181818181819e-05, "loss": 6.2595, "step": 2510 }, { "epoch": 2.82, "learning_rate": 1.4810606060606061e-05, "loss": 6.3676, "step": 2511 }, { "epoch": 2.82, "learning_rate": 1.4803030303030303e-05, "loss": 6.0907, "step": 2512 }, { "epoch": 2.82, "learning_rate": 1.4795454545454546e-05, "loss": 6.1561, "step": 2513 }, { "epoch": 2.82, "learning_rate": 1.4787878787878788e-05, "loss": 6.0659, "step": 2514 }, { "epoch": 2.82, "learning_rate": 1.478030303030303e-05, "loss": 6.2186, "step": 2515 }, { "epoch": 2.82, "learning_rate": 1.4772727272727273e-05, "loss": 6.0755, "step": 2516 }, { "epoch": 2.82, "learning_rate": 1.4765151515151515e-05, "loss": 6.1092, "step": 2517 }, { "epoch": 2.82, "learning_rate": 1.4757575757575759e-05, "loss": 6.1964, "step": 2518 }, { "epoch": 2.82, "learning_rate": 1.475e-05, "loss": 6.1073, "step": 2519 }, { "epoch": 2.83, "learning_rate": 1.4742424242424243e-05, "loss": 6.0145, "step": 2520 }, { "epoch": 2.83, "learning_rate": 1.4734848484848486e-05, "loss": 6.2631, "step": 2521 }, { "epoch": 2.83, "learning_rate": 1.4727272727272728e-05, "loss": 6.1163, "step": 2522 }, { "epoch": 2.83, "learning_rate": 1.471969696969697e-05, "loss": 6.1631, "step": 2523 }, { "epoch": 2.83, "learning_rate": 1.4712121212121213e-05, "loss": 5.9733, "step": 2524 }, { "epoch": 2.83, "learning_rate": 1.4704545454545455e-05, "loss": 5.9928, "step": 2525 }, { "epoch": 2.83, "learning_rate": 1.4696969696969697e-05, "loss": 6.0008, "step": 2526 }, { "epoch": 2.83, "learning_rate": 1.4689393939393941e-05, "loss": 6.0646, "step": 2527 }, { "epoch": 2.83, "learning_rate": 1.4681818181818182e-05, "loss": 5.8081, "step": 2528 }, { "epoch": 2.84, "learning_rate": 1.4674242424242426e-05, "loss": 5.7113, "step": 2529 }, { "epoch": 2.84, "learning_rate": 1.4666666666666666e-05, "loss": 5.7223, "step": 2530 }, { "epoch": 2.84, "learning_rate": 1.465909090909091e-05, "loss": 5.2575, "step": 2531 }, { "epoch": 2.84, "learning_rate": 1.4651515151515151e-05, "loss": 5.3393, "step": 2532 }, { "epoch": 2.84, "learning_rate": 1.4643939393939395e-05, "loss": 5.1757, "step": 2533 }, { "epoch": 2.84, "learning_rate": 1.4636363636363635e-05, "loss": 4.5787, "step": 2534 }, { "epoch": 2.84, "learning_rate": 1.462878787878788e-05, "loss": 6.4461, "step": 2535 }, { "epoch": 2.84, "learning_rate": 1.462121212121212e-05, "loss": 6.467, "step": 2536 }, { "epoch": 2.84, "learning_rate": 1.4613636363636364e-05, "loss": 6.3725, "step": 2537 }, { "epoch": 2.85, "learning_rate": 1.4606060606060608e-05, "loss": 6.5094, "step": 2538 }, { "epoch": 2.85, "learning_rate": 1.4598484848484849e-05, "loss": 6.3196, "step": 2539 }, { "epoch": 2.85, "learning_rate": 1.4590909090909093e-05, "loss": 6.3793, "step": 2540 }, { "epoch": 2.85, "learning_rate": 1.4583333333333333e-05, "loss": 6.4023, "step": 2541 }, { "epoch": 2.85, "learning_rate": 1.4575757575757577e-05, "loss": 6.4038, "step": 2542 }, { "epoch": 2.85, "learning_rate": 1.4568181818181818e-05, "loss": 6.339, "step": 2543 }, { "epoch": 2.85, "learning_rate": 1.4560606060606062e-05, "loss": 6.3062, "step": 2544 }, { "epoch": 2.85, "learning_rate": 1.4553030303030302e-05, "loss": 6.2636, "step": 2545 }, { "epoch": 2.85, "learning_rate": 1.4545454545454546e-05, "loss": 6.2378, "step": 2546 }, { "epoch": 2.86, "learning_rate": 1.4537878787878787e-05, "loss": 6.2669, "step": 2547 }, { "epoch": 2.86, "learning_rate": 1.4530303030303031e-05, "loss": 6.1103, "step": 2548 }, { "epoch": 2.86, "learning_rate": 1.4522727272727273e-05, "loss": 6.1708, "step": 2549 }, { "epoch": 2.86, "learning_rate": 1.4515151515151516e-05, "loss": 6.1639, "step": 2550 }, { "epoch": 2.86, "learning_rate": 1.4507575757575758e-05, "loss": 6.205, "step": 2551 }, { "epoch": 2.86, "learning_rate": 1.45e-05, "loss": 6.2056, "step": 2552 }, { "epoch": 2.86, "learning_rate": 1.4492424242424242e-05, "loss": 6.0338, "step": 2553 }, { "epoch": 2.86, "learning_rate": 1.4484848484848485e-05, "loss": 6.1466, "step": 2554 }, { "epoch": 2.86, "learning_rate": 1.4477272727272729e-05, "loss": 6.0014, "step": 2555 }, { "epoch": 2.87, "learning_rate": 1.446969696969697e-05, "loss": 6.0874, "step": 2556 }, { "epoch": 2.87, "learning_rate": 1.4462121212121213e-05, "loss": 6.2096, "step": 2557 }, { "epoch": 2.87, "learning_rate": 1.4454545454545456e-05, "loss": 6.1184, "step": 2558 }, { "epoch": 2.87, "learning_rate": 1.4446969696969698e-05, "loss": 6.2408, "step": 2559 }, { "epoch": 2.87, "learning_rate": 1.443939393939394e-05, "loss": 6.0439, "step": 2560 }, { "epoch": 2.87, "learning_rate": 1.4431818181818182e-05, "loss": 6.068, "step": 2561 }, { "epoch": 2.87, "learning_rate": 1.4424242424242425e-05, "loss": 6.0175, "step": 2562 }, { "epoch": 2.87, "learning_rate": 1.4416666666666667e-05, "loss": 6.023, "step": 2563 }, { "epoch": 2.87, "learning_rate": 1.440909090909091e-05, "loss": 6.1184, "step": 2564 }, { "epoch": 2.88, "learning_rate": 1.4401515151515152e-05, "loss": 5.9487, "step": 2565 }, { "epoch": 2.88, "learning_rate": 1.4393939393939394e-05, "loss": 5.9513, "step": 2566 }, { "epoch": 2.88, "learning_rate": 1.4386363636363636e-05, "loss": 5.9061, "step": 2567 }, { "epoch": 2.88, "learning_rate": 1.4378787878787878e-05, "loss": 6.0247, "step": 2568 }, { "epoch": 2.88, "learning_rate": 1.4371212121212122e-05, "loss": 6.2227, "step": 2569 }, { "epoch": 2.88, "learning_rate": 1.4363636363636363e-05, "loss": 5.9537, "step": 2570 }, { "epoch": 2.88, "learning_rate": 1.4356060606060607e-05, "loss": 6.1902, "step": 2571 }, { "epoch": 2.88, "learning_rate": 1.434848484848485e-05, "loss": 6.1231, "step": 2572 }, { "epoch": 2.88, "learning_rate": 1.4340909090909092e-05, "loss": 5.9834, "step": 2573 }, { "epoch": 2.89, "learning_rate": 1.4333333333333334e-05, "loss": 5.9828, "step": 2574 }, { "epoch": 2.89, "learning_rate": 1.4325757575757576e-05, "loss": 5.921, "step": 2575 }, { "epoch": 2.89, "learning_rate": 1.4318181818181818e-05, "loss": 5.8787, "step": 2576 }, { "epoch": 2.89, "learning_rate": 1.431060606060606e-05, "loss": 5.838, "step": 2577 }, { "epoch": 2.89, "learning_rate": 1.4303030303030305e-05, "loss": 5.7256, "step": 2578 }, { "epoch": 2.89, "learning_rate": 1.4295454545454545e-05, "loss": 5.6374, "step": 2579 }, { "epoch": 2.89, "learning_rate": 1.428787878787879e-05, "loss": 5.5289, "step": 2580 }, { "epoch": 2.89, "learning_rate": 1.428030303030303e-05, "loss": 5.4498, "step": 2581 }, { "epoch": 2.89, "learning_rate": 1.4272727272727274e-05, "loss": 5.1088, "step": 2582 }, { "epoch": 2.9, "learning_rate": 1.4265151515151515e-05, "loss": 4.8553, "step": 2583 }, { "epoch": 2.9, "learning_rate": 1.4257575757575758e-05, "loss": 4.3758, "step": 2584 }, { "epoch": 2.9, "learning_rate": 1.4249999999999999e-05, "loss": 6.5679, "step": 2585 }, { "epoch": 2.9, "learning_rate": 1.4242424242424243e-05, "loss": 6.3611, "step": 2586 }, { "epoch": 2.9, "learning_rate": 1.4234848484848484e-05, "loss": 6.4009, "step": 2587 }, { "epoch": 2.9, "learning_rate": 1.4227272727272728e-05, "loss": 6.433, "step": 2588 }, { "epoch": 2.9, "learning_rate": 1.4219696969696972e-05, "loss": 6.309, "step": 2589 }, { "epoch": 2.9, "learning_rate": 1.4212121212121212e-05, "loss": 6.4418, "step": 2590 }, { "epoch": 2.9, "learning_rate": 1.4204545454545456e-05, "loss": 6.2833, "step": 2591 }, { "epoch": 2.91, "learning_rate": 1.4196969696969697e-05, "loss": 6.3231, "step": 2592 }, { "epoch": 2.91, "learning_rate": 1.418939393939394e-05, "loss": 6.2211, "step": 2593 }, { "epoch": 2.91, "learning_rate": 1.4181818181818181e-05, "loss": 6.192, "step": 2594 }, { "epoch": 2.91, "learning_rate": 1.4174242424242425e-05, "loss": 6.2101, "step": 2595 }, { "epoch": 2.91, "learning_rate": 1.4166666666666666e-05, "loss": 6.1113, "step": 2596 }, { "epoch": 2.91, "learning_rate": 1.415909090909091e-05, "loss": 6.2036, "step": 2597 }, { "epoch": 2.91, "learning_rate": 1.415151515151515e-05, "loss": 6.1407, "step": 2598 }, { "epoch": 2.91, "learning_rate": 1.4143939393939395e-05, "loss": 6.0975, "step": 2599 }, { "epoch": 2.91, "learning_rate": 1.4136363636363637e-05, "loss": 6.195, "step": 2600 }, { "epoch": 2.92, "learning_rate": 1.4128787878787879e-05, "loss": 6.2409, "step": 2601 }, { "epoch": 2.92, "learning_rate": 1.4121212121212121e-05, "loss": 6.146, "step": 2602 }, { "epoch": 2.92, "learning_rate": 1.4113636363636364e-05, "loss": 6.0791, "step": 2603 }, { "epoch": 2.92, "learning_rate": 1.4106060606060606e-05, "loss": 6.2553, "step": 2604 }, { "epoch": 2.92, "learning_rate": 1.4098484848484848e-05, "loss": 6.0885, "step": 2605 }, { "epoch": 2.92, "learning_rate": 1.4090909090909092e-05, "loss": 6.0652, "step": 2606 }, { "epoch": 2.92, "learning_rate": 1.4083333333333333e-05, "loss": 6.003, "step": 2607 }, { "epoch": 2.92, "learning_rate": 1.4075757575757577e-05, "loss": 6.0275, "step": 2608 }, { "epoch": 2.92, "learning_rate": 1.406818181818182e-05, "loss": 6.0687, "step": 2609 }, { "epoch": 2.93, "learning_rate": 1.4060606060606061e-05, "loss": 6.1724, "step": 2610 }, { "epoch": 2.93, "learning_rate": 1.4053030303030304e-05, "loss": 6.1112, "step": 2611 }, { "epoch": 2.93, "learning_rate": 1.4045454545454546e-05, "loss": 5.9664, "step": 2612 }, { "epoch": 2.93, "learning_rate": 1.4037878787878788e-05, "loss": 6.2046, "step": 2613 }, { "epoch": 2.93, "learning_rate": 1.403030303030303e-05, "loss": 6.1502, "step": 2614 }, { "epoch": 2.93, "learning_rate": 1.4022727272727273e-05, "loss": 6.1934, "step": 2615 }, { "epoch": 2.93, "learning_rate": 1.4015151515151515e-05, "loss": 6.1615, "step": 2616 }, { "epoch": 2.93, "learning_rate": 1.4007575757575757e-05, "loss": 6.0933, "step": 2617 }, { "epoch": 2.93, "learning_rate": 1.4e-05, "loss": 5.9867, "step": 2618 }, { "epoch": 2.94, "learning_rate": 1.3992424242424242e-05, "loss": 6.0406, "step": 2619 }, { "epoch": 2.94, "learning_rate": 1.3984848484848486e-05, "loss": 5.9274, "step": 2620 }, { "epoch": 2.94, "learning_rate": 1.3977272727272727e-05, "loss": 5.8517, "step": 2621 }, { "epoch": 2.94, "learning_rate": 1.396969696969697e-05, "loss": 6.1097, "step": 2622 }, { "epoch": 2.94, "learning_rate": 1.3962121212121213e-05, "loss": 6.0706, "step": 2623 }, { "epoch": 2.94, "learning_rate": 1.3954545454545455e-05, "loss": 6.0108, "step": 2624 }, { "epoch": 2.94, "learning_rate": 1.3946969696969698e-05, "loss": 5.8261, "step": 2625 }, { "epoch": 2.94, "learning_rate": 1.393939393939394e-05, "loss": 5.8692, "step": 2626 }, { "epoch": 2.95, "learning_rate": 1.3931818181818182e-05, "loss": 5.732, "step": 2627 }, { "epoch": 2.95, "learning_rate": 1.3924242424242424e-05, "loss": 5.6379, "step": 2628 }, { "epoch": 2.95, "learning_rate": 1.3916666666666668e-05, "loss": 5.7437, "step": 2629 }, { "epoch": 2.95, "learning_rate": 1.3909090909090909e-05, "loss": 5.8382, "step": 2630 }, { "epoch": 2.95, "learning_rate": 1.3901515151515153e-05, "loss": 5.1888, "step": 2631 }, { "epoch": 2.95, "learning_rate": 1.3893939393939394e-05, "loss": 4.9007, "step": 2632 }, { "epoch": 2.95, "learning_rate": 1.3886363636363638e-05, "loss": 4.8791, "step": 2633 }, { "epoch": 2.95, "learning_rate": 1.3878787878787878e-05, "loss": 4.3732, "step": 2634 }, { "epoch": 2.95, "learning_rate": 1.3871212121212122e-05, "loss": 6.57, "step": 2635 }, { "epoch": 2.96, "learning_rate": 1.3863636363636363e-05, "loss": 6.4589, "step": 2636 }, { "epoch": 2.96, "learning_rate": 1.3856060606060607e-05, "loss": 6.5854, "step": 2637 }, { "epoch": 2.96, "learning_rate": 1.3848484848484847e-05, "loss": 6.3936, "step": 2638 }, { "epoch": 2.96, "learning_rate": 1.3840909090909091e-05, "loss": 6.3889, "step": 2639 }, { "epoch": 2.96, "learning_rate": 1.3833333333333335e-05, "loss": 6.2643, "step": 2640 }, { "epoch": 2.96, "learning_rate": 1.3825757575757576e-05, "loss": 6.2905, "step": 2641 }, { "epoch": 2.96, "learning_rate": 1.381818181818182e-05, "loss": 6.328, "step": 2642 }, { "epoch": 2.96, "learning_rate": 1.381060606060606e-05, "loss": 6.3227, "step": 2643 }, { "epoch": 2.96, "learning_rate": 1.3803030303030304e-05, "loss": 6.0856, "step": 2644 }, { "epoch": 2.97, "learning_rate": 1.3795454545454545e-05, "loss": 6.1553, "step": 2645 }, { "epoch": 2.97, "learning_rate": 1.3787878787878789e-05, "loss": 6.2453, "step": 2646 }, { "epoch": 2.97, "learning_rate": 1.378030303030303e-05, "loss": 6.2721, "step": 2647 }, { "epoch": 2.97, "learning_rate": 1.3772727272727274e-05, "loss": 6.2114, "step": 2648 }, { "epoch": 2.97, "learning_rate": 1.3765151515151514e-05, "loss": 6.2997, "step": 2649 }, { "epoch": 2.97, "learning_rate": 1.3757575757575758e-05, "loss": 6.0543, "step": 2650 }, { "epoch": 2.97, "learning_rate": 1.375e-05, "loss": 6.1979, "step": 2651 }, { "epoch": 2.97, "learning_rate": 1.3742424242424243e-05, "loss": 6.1353, "step": 2652 }, { "epoch": 2.97, "learning_rate": 1.3734848484848485e-05, "loss": 6.1883, "step": 2653 }, { "epoch": 2.98, "learning_rate": 1.3727272727272727e-05, "loss": 6.0755, "step": 2654 }, { "epoch": 2.98, "learning_rate": 1.371969696969697e-05, "loss": 6.0917, "step": 2655 }, { "epoch": 2.98, "learning_rate": 1.3712121212121212e-05, "loss": 6.1798, "step": 2656 }, { "epoch": 2.98, "learning_rate": 1.3704545454545456e-05, "loss": 6.1191, "step": 2657 }, { "epoch": 2.98, "learning_rate": 1.3696969696969697e-05, "loss": 6.1833, "step": 2658 }, { "epoch": 2.98, "learning_rate": 1.368939393939394e-05, "loss": 6.0788, "step": 2659 }, { "epoch": 2.98, "learning_rate": 1.3681818181818183e-05, "loss": 5.9864, "step": 2660 }, { "epoch": 2.98, "learning_rate": 1.3674242424242425e-05, "loss": 6.1196, "step": 2661 }, { "epoch": 2.98, "learning_rate": 1.3666666666666667e-05, "loss": 6.2538, "step": 2662 }, { "epoch": 2.99, "learning_rate": 1.365909090909091e-05, "loss": 6.0898, "step": 2663 }, { "epoch": 2.99, "learning_rate": 1.3651515151515152e-05, "loss": 6.185, "step": 2664 }, { "epoch": 2.99, "learning_rate": 1.3643939393939394e-05, "loss": 5.7188, "step": 2665 }, { "epoch": 2.99, "learning_rate": 1.3636363636363637e-05, "loss": 5.9985, "step": 2666 }, { "epoch": 2.99, "learning_rate": 1.3628787878787879e-05, "loss": 6.1204, "step": 2667 }, { "epoch": 2.99, "learning_rate": 1.3621212121212121e-05, "loss": 6.0411, "step": 2668 }, { "epoch": 2.99, "learning_rate": 1.3613636363636363e-05, "loss": 5.8924, "step": 2669 }, { "epoch": 2.99, "learning_rate": 1.3606060606060606e-05, "loss": 5.9221, "step": 2670 }, { "epoch": 2.99, "learning_rate": 1.359848484848485e-05, "loss": 5.7824, "step": 2671 }, { "epoch": 3.0, "learning_rate": 1.359090909090909e-05, "loss": 5.597, "step": 2672 }, { "epoch": 3.0, "learning_rate": 1.3583333333333334e-05, "loss": 5.6367, "step": 2673 }, { "epoch": 3.0, "learning_rate": 1.3575757575757577e-05, "loss": 5.1606, "step": 2674 }, { "epoch": 3.0, "learning_rate": 1.3568181818181819e-05, "loss": 5.0194, "step": 2675 }, { "epoch": 3.0, "learning_rate": 1.3560606060606061e-05, "loss": 4.1676, "step": 2676 }, { "epoch": 3.0, "learning_rate": 1.3553030303030303e-05, "loss": 6.3327, "step": 2677 }, { "epoch": 3.0, "learning_rate": 1.3545454545454546e-05, "loss": 6.2865, "step": 2678 }, { "epoch": 3.0, "learning_rate": 1.3537878787878788e-05, "loss": 6.3395, "step": 2679 }, { "epoch": 3.0, "learning_rate": 1.353030303030303e-05, "loss": 6.3106, "step": 2680 }, { "epoch": 3.01, "learning_rate": 1.3522727272727273e-05, "loss": 6.3632, "step": 2681 }, { "epoch": 3.01, "learning_rate": 1.3515151515151517e-05, "loss": 6.3705, "step": 2682 }, { "epoch": 3.01, "learning_rate": 1.3507575757575757e-05, "loss": 6.1918, "step": 2683 }, { "epoch": 3.01, "learning_rate": 1.3500000000000001e-05, "loss": 6.2926, "step": 2684 }, { "epoch": 3.01, "learning_rate": 1.3492424242424242e-05, "loss": 6.2737, "step": 2685 }, { "epoch": 3.01, "learning_rate": 1.3484848484848486e-05, "loss": 6.1884, "step": 2686 }, { "epoch": 3.01, "learning_rate": 1.3477272727272726e-05, "loss": 6.1732, "step": 2687 }, { "epoch": 3.01, "learning_rate": 1.346969696969697e-05, "loss": 6.0071, "step": 2688 }, { "epoch": 3.01, "learning_rate": 1.3462121212121211e-05, "loss": 6.2152, "step": 2689 }, { "epoch": 3.02, "learning_rate": 1.3454545454545455e-05, "loss": 6.0995, "step": 2690 }, { "epoch": 3.02, "learning_rate": 1.3446969696969699e-05, "loss": 6.1984, "step": 2691 }, { "epoch": 3.02, "learning_rate": 1.343939393939394e-05, "loss": 6.1829, "step": 2692 }, { "epoch": 3.02, "learning_rate": 1.3431818181818184e-05, "loss": 6.0885, "step": 2693 }, { "epoch": 3.02, "learning_rate": 1.3424242424242424e-05, "loss": 6.1808, "step": 2694 }, { "epoch": 3.02, "learning_rate": 1.3416666666666668e-05, "loss": 6.1814, "step": 2695 }, { "epoch": 3.02, "learning_rate": 1.3409090909090909e-05, "loss": 6.0656, "step": 2696 }, { "epoch": 3.02, "learning_rate": 1.3401515151515153e-05, "loss": 6.0741, "step": 2697 }, { "epoch": 3.02, "learning_rate": 1.3393939393939393e-05, "loss": 5.9853, "step": 2698 }, { "epoch": 3.03, "learning_rate": 1.3386363636363637e-05, "loss": 6.0419, "step": 2699 }, { "epoch": 3.03, "learning_rate": 1.3378787878787878e-05, "loss": 6.117, "step": 2700 }, { "epoch": 3.03, "learning_rate": 1.3371212121212122e-05, "loss": 6.0867, "step": 2701 }, { "epoch": 3.03, "learning_rate": 1.3363636363636364e-05, "loss": 6.0902, "step": 2702 }, { "epoch": 3.03, "learning_rate": 1.3356060606060606e-05, "loss": 6.2407, "step": 2703 }, { "epoch": 3.03, "learning_rate": 1.3348484848484849e-05, "loss": 5.9879, "step": 2704 }, { "epoch": 3.03, "learning_rate": 1.3340909090909091e-05, "loss": 6.1059, "step": 2705 }, { "epoch": 3.03, "learning_rate": 1.3333333333333333e-05, "loss": 6.087, "step": 2706 }, { "epoch": 3.03, "learning_rate": 1.3325757575757576e-05, "loss": 5.9207, "step": 2707 }, { "epoch": 3.04, "learning_rate": 1.331818181818182e-05, "loss": 5.8997, "step": 2708 }, { "epoch": 3.04, "learning_rate": 1.331060606060606e-05, "loss": 5.9126, "step": 2709 }, { "epoch": 3.04, "learning_rate": 1.3303030303030304e-05, "loss": 5.9661, "step": 2710 }, { "epoch": 3.04, "learning_rate": 1.3295454545454546e-05, "loss": 5.9483, "step": 2711 }, { "epoch": 3.04, "learning_rate": 1.3287878787878789e-05, "loss": 6.0193, "step": 2712 }, { "epoch": 3.04, "learning_rate": 1.3280303030303031e-05, "loss": 5.9905, "step": 2713 }, { "epoch": 3.04, "learning_rate": 1.3272727272727273e-05, "loss": 5.946, "step": 2714 }, { "epoch": 3.04, "learning_rate": 1.3265151515151516e-05, "loss": 5.8088, "step": 2715 }, { "epoch": 3.04, "learning_rate": 1.3257575757575758e-05, "loss": 5.9265, "step": 2716 }, { "epoch": 3.05, "learning_rate": 1.325e-05, "loss": 5.8953, "step": 2717 }, { "epoch": 3.05, "learning_rate": 1.3242424242424242e-05, "loss": 5.7927, "step": 2718 }, { "epoch": 3.05, "learning_rate": 1.3234848484848485e-05, "loss": 5.835, "step": 2719 }, { "epoch": 3.05, "learning_rate": 1.3227272727272727e-05, "loss": 5.8176, "step": 2720 }, { "epoch": 3.05, "learning_rate": 1.321969696969697e-05, "loss": 5.4691, "step": 2721 }, { "epoch": 3.05, "learning_rate": 1.3212121212121213e-05, "loss": 5.479, "step": 2722 }, { "epoch": 3.05, "learning_rate": 1.3204545454545454e-05, "loss": 5.3383, "step": 2723 }, { "epoch": 3.05, "learning_rate": 1.3196969696969698e-05, "loss": 5.0216, "step": 2724 }, { "epoch": 3.05, "learning_rate": 1.318939393939394e-05, "loss": 4.7733, "step": 2725 }, { "epoch": 3.06, "learning_rate": 1.3181818181818183e-05, "loss": 3.6825, "step": 2726 }, { "epoch": 3.06, "learning_rate": 1.3174242424242425e-05, "loss": 6.3076, "step": 2727 }, { "epoch": 3.06, "learning_rate": 1.3166666666666667e-05, "loss": 6.3033, "step": 2728 }, { "epoch": 3.06, "learning_rate": 1.315909090909091e-05, "loss": 6.2731, "step": 2729 }, { "epoch": 3.06, "learning_rate": 1.3151515151515152e-05, "loss": 6.1465, "step": 2730 }, { "epoch": 3.06, "learning_rate": 1.3143939393939394e-05, "loss": 6.3055, "step": 2731 }, { "epoch": 3.06, "learning_rate": 1.3136363636363636e-05, "loss": 6.2977, "step": 2732 }, { "epoch": 3.06, "learning_rate": 1.312878787878788e-05, "loss": 6.402, "step": 2733 }, { "epoch": 3.07, "learning_rate": 1.312121212121212e-05, "loss": 6.1693, "step": 2734 }, { "epoch": 3.07, "learning_rate": 1.3113636363636365e-05, "loss": 6.0351, "step": 2735 }, { "epoch": 3.07, "learning_rate": 1.3106060606060605e-05, "loss": 6.1118, "step": 2736 }, { "epoch": 3.07, "learning_rate": 1.309848484848485e-05, "loss": 6.0256, "step": 2737 }, { "epoch": 3.07, "learning_rate": 1.309090909090909e-05, "loss": 6.0979, "step": 2738 }, { "epoch": 3.07, "learning_rate": 1.3083333333333334e-05, "loss": 6.0932, "step": 2739 }, { "epoch": 3.07, "learning_rate": 1.3075757575757575e-05, "loss": 6.0504, "step": 2740 }, { "epoch": 3.07, "learning_rate": 1.3068181818181819e-05, "loss": 6.1479, "step": 2741 }, { "epoch": 3.07, "learning_rate": 1.3060606060606061e-05, "loss": 6.1473, "step": 2742 }, { "epoch": 3.08, "learning_rate": 1.3053030303030303e-05, "loss": 6.0779, "step": 2743 }, { "epoch": 3.08, "learning_rate": 1.3045454545454547e-05, "loss": 6.1422, "step": 2744 }, { "epoch": 3.08, "learning_rate": 1.3037878787878788e-05, "loss": 6.0463, "step": 2745 }, { "epoch": 3.08, "learning_rate": 1.3030303030303032e-05, "loss": 6.0867, "step": 2746 }, { "epoch": 3.08, "learning_rate": 1.3022727272727272e-05, "loss": 5.9629, "step": 2747 }, { "epoch": 3.08, "learning_rate": 1.3015151515151516e-05, "loss": 5.9759, "step": 2748 }, { "epoch": 3.08, "learning_rate": 1.3007575757575757e-05, "loss": 6.148, "step": 2749 }, { "epoch": 3.08, "learning_rate": 1.3000000000000001e-05, "loss": 6.1167, "step": 2750 }, { "epoch": 3.08, "learning_rate": 1.2992424242424241e-05, "loss": 6.1132, "step": 2751 }, { "epoch": 3.09, "learning_rate": 1.2984848484848485e-05, "loss": 6.0355, "step": 2752 }, { "epoch": 3.09, "learning_rate": 1.2977272727272728e-05, "loss": 6.0977, "step": 2753 }, { "epoch": 3.09, "learning_rate": 1.296969696969697e-05, "loss": 5.9838, "step": 2754 }, { "epoch": 3.09, "learning_rate": 1.2962121212121212e-05, "loss": 6.2097, "step": 2755 }, { "epoch": 3.09, "learning_rate": 1.2954545454545455e-05, "loss": 5.9733, "step": 2756 }, { "epoch": 3.09, "learning_rate": 1.2946969696969697e-05, "loss": 5.9807, "step": 2757 }, { "epoch": 3.09, "learning_rate": 1.293939393939394e-05, "loss": 5.9994, "step": 2758 }, { "epoch": 3.09, "learning_rate": 1.2931818181818182e-05, "loss": 6.0416, "step": 2759 }, { "epoch": 3.09, "learning_rate": 1.2924242424242424e-05, "loss": 5.9098, "step": 2760 }, { "epoch": 3.1, "learning_rate": 1.2916666666666668e-05, "loss": 5.9817, "step": 2761 }, { "epoch": 3.1, "learning_rate": 1.290909090909091e-05, "loss": 5.951, "step": 2762 }, { "epoch": 3.1, "learning_rate": 1.2901515151515152e-05, "loss": 6.1318, "step": 2763 }, { "epoch": 3.1, "learning_rate": 1.2893939393939395e-05, "loss": 6.0098, "step": 2764 }, { "epoch": 3.1, "learning_rate": 1.2886363636363637e-05, "loss": 5.9125, "step": 2765 }, { "epoch": 3.1, "learning_rate": 1.287878787878788e-05, "loss": 5.9097, "step": 2766 }, { "epoch": 3.1, "learning_rate": 1.2871212121212122e-05, "loss": 5.9798, "step": 2767 }, { "epoch": 3.1, "learning_rate": 1.2863636363636364e-05, "loss": 5.8757, "step": 2768 }, { "epoch": 3.1, "learning_rate": 1.2856060606060606e-05, "loss": 5.6691, "step": 2769 }, { "epoch": 3.11, "learning_rate": 1.2848484848484848e-05, "loss": 5.8003, "step": 2770 }, { "epoch": 3.11, "learning_rate": 1.284090909090909e-05, "loss": 5.4288, "step": 2771 }, { "epoch": 3.11, "learning_rate": 1.2833333333333333e-05, "loss": 5.3083, "step": 2772 }, { "epoch": 3.11, "learning_rate": 1.2825757575757577e-05, "loss": 5.2724, "step": 2773 }, { "epoch": 3.11, "learning_rate": 1.2818181818181818e-05, "loss": 4.8527, "step": 2774 }, { "epoch": 3.11, "learning_rate": 1.2810606060606062e-05, "loss": 4.8897, "step": 2775 }, { "epoch": 3.11, "learning_rate": 1.2803030303030302e-05, "loss": 4.2398, "step": 2776 }, { "epoch": 3.11, "learning_rate": 1.2795454545454546e-05, "loss": 6.3437, "step": 2777 }, { "epoch": 3.11, "learning_rate": 1.2787878787878788e-05, "loss": 6.3307, "step": 2778 }, { "epoch": 3.12, "learning_rate": 1.278030303030303e-05, "loss": 6.1326, "step": 2779 }, { "epoch": 3.12, "learning_rate": 1.2772727272727273e-05, "loss": 6.2245, "step": 2780 }, { "epoch": 3.12, "learning_rate": 1.2765151515151515e-05, "loss": 6.2183, "step": 2781 }, { "epoch": 3.12, "learning_rate": 1.2757575757575758e-05, "loss": 6.2559, "step": 2782 }, { "epoch": 3.12, "learning_rate": 1.275e-05, "loss": 6.2463, "step": 2783 }, { "epoch": 3.12, "learning_rate": 1.2742424242424244e-05, "loss": 6.0644, "step": 2784 }, { "epoch": 3.12, "learning_rate": 1.2734848484848484e-05, "loss": 6.1446, "step": 2785 }, { "epoch": 3.12, "learning_rate": 1.2727272727272728e-05, "loss": 6.1169, "step": 2786 }, { "epoch": 3.12, "learning_rate": 1.2719696969696969e-05, "loss": 5.9059, "step": 2787 }, { "epoch": 3.13, "learning_rate": 1.2712121212121213e-05, "loss": 6.2276, "step": 2788 }, { "epoch": 3.13, "learning_rate": 1.2704545454545454e-05, "loss": 6.0476, "step": 2789 }, { "epoch": 3.13, "learning_rate": 1.2696969696969698e-05, "loss": 6.1381, "step": 2790 }, { "epoch": 3.13, "learning_rate": 1.2689393939393938e-05, "loss": 6.1015, "step": 2791 }, { "epoch": 3.13, "learning_rate": 1.2681818181818182e-05, "loss": 6.1161, "step": 2792 }, { "epoch": 3.13, "learning_rate": 1.2674242424242425e-05, "loss": 6.1949, "step": 2793 }, { "epoch": 3.13, "learning_rate": 1.2666666666666667e-05, "loss": 6.1161, "step": 2794 }, { "epoch": 3.13, "learning_rate": 1.265909090909091e-05, "loss": 5.9767, "step": 2795 }, { "epoch": 3.13, "learning_rate": 1.2651515151515151e-05, "loss": 5.9796, "step": 2796 }, { "epoch": 3.14, "learning_rate": 1.2643939393939395e-05, "loss": 6.1194, "step": 2797 }, { "epoch": 3.14, "learning_rate": 1.2636363636363636e-05, "loss": 6.1285, "step": 2798 }, { "epoch": 3.14, "learning_rate": 1.262878787878788e-05, "loss": 6.0551, "step": 2799 }, { "epoch": 3.14, "learning_rate": 1.262121212121212e-05, "loss": 6.1235, "step": 2800 }, { "epoch": 3.14, "learning_rate": 1.2613636363636365e-05, "loss": 5.9844, "step": 2801 }, { "epoch": 3.14, "learning_rate": 1.2606060606060605e-05, "loss": 6.0144, "step": 2802 }, { "epoch": 3.14, "learning_rate": 1.2598484848484849e-05, "loss": 6.0452, "step": 2803 }, { "epoch": 3.14, "learning_rate": 1.2590909090909091e-05, "loss": 5.9843, "step": 2804 }, { "epoch": 3.14, "learning_rate": 1.2583333333333334e-05, "loss": 5.9164, "step": 2805 }, { "epoch": 3.15, "learning_rate": 1.2575757575757576e-05, "loss": 6.0514, "step": 2806 }, { "epoch": 3.15, "learning_rate": 1.2568181818181818e-05, "loss": 5.9055, "step": 2807 }, { "epoch": 3.15, "learning_rate": 1.256060606060606e-05, "loss": 5.9087, "step": 2808 }, { "epoch": 3.15, "learning_rate": 1.2553030303030303e-05, "loss": 6.0116, "step": 2809 }, { "epoch": 3.15, "learning_rate": 1.2545454545454545e-05, "loss": 6.0002, "step": 2810 }, { "epoch": 3.15, "learning_rate": 1.2537878787878787e-05, "loss": 5.9371, "step": 2811 }, { "epoch": 3.15, "learning_rate": 1.2530303030303031e-05, "loss": 5.9367, "step": 2812 }, { "epoch": 3.15, "learning_rate": 1.2522727272727274e-05, "loss": 5.9301, "step": 2813 }, { "epoch": 3.15, "learning_rate": 1.2515151515151516e-05, "loss": 5.9678, "step": 2814 }, { "epoch": 3.16, "learning_rate": 1.2507575757575758e-05, "loss": 5.9743, "step": 2815 }, { "epoch": 3.16, "learning_rate": 1.25e-05, "loss": 5.9469, "step": 2816 }, { "epoch": 3.16, "learning_rate": 1.2492424242424243e-05, "loss": 5.9136, "step": 2817 }, { "epoch": 3.16, "learning_rate": 1.2484848484848485e-05, "loss": 5.8197, "step": 2818 }, { "epoch": 3.16, "learning_rate": 1.2477272727272727e-05, "loss": 5.6004, "step": 2819 }, { "epoch": 3.16, "learning_rate": 1.246969696969697e-05, "loss": 5.5001, "step": 2820 }, { "epoch": 3.16, "learning_rate": 1.2462121212121212e-05, "loss": 5.5068, "step": 2821 }, { "epoch": 3.16, "learning_rate": 1.2454545454545454e-05, "loss": 5.2867, "step": 2822 }, { "epoch": 3.16, "learning_rate": 1.2446969696969697e-05, "loss": 5.3019, "step": 2823 }, { "epoch": 3.17, "learning_rate": 1.243939393939394e-05, "loss": 5.132, "step": 2824 }, { "epoch": 3.17, "learning_rate": 1.2431818181818181e-05, "loss": 4.8367, "step": 2825 }, { "epoch": 3.17, "learning_rate": 1.2424242424242425e-05, "loss": 3.998, "step": 2826 }, { "epoch": 3.17, "learning_rate": 1.2416666666666666e-05, "loss": 6.4595, "step": 2827 }, { "epoch": 3.17, "learning_rate": 1.240909090909091e-05, "loss": 6.3013, "step": 2828 }, { "epoch": 3.17, "learning_rate": 1.2401515151515152e-05, "loss": 6.3502, "step": 2829 }, { "epoch": 3.17, "learning_rate": 1.2393939393939394e-05, "loss": 6.3557, "step": 2830 }, { "epoch": 3.17, "learning_rate": 1.2386363636363637e-05, "loss": 6.3595, "step": 2831 }, { "epoch": 3.17, "learning_rate": 1.2378787878787879e-05, "loss": 6.4059, "step": 2832 }, { "epoch": 3.18, "learning_rate": 1.2371212121212121e-05, "loss": 6.2026, "step": 2833 }, { "epoch": 3.18, "learning_rate": 1.2363636363636364e-05, "loss": 6.3708, "step": 2834 }, { "epoch": 3.18, "learning_rate": 1.2356060606060608e-05, "loss": 6.1253, "step": 2835 }, { "epoch": 3.18, "learning_rate": 1.2348484848484848e-05, "loss": 5.9994, "step": 2836 }, { "epoch": 3.18, "learning_rate": 1.2340909090909092e-05, "loss": 6.237, "step": 2837 }, { "epoch": 3.18, "learning_rate": 1.2333333333333333e-05, "loss": 6.1435, "step": 2838 }, { "epoch": 3.18, "learning_rate": 1.2325757575757577e-05, "loss": 6.3098, "step": 2839 }, { "epoch": 3.18, "learning_rate": 1.2318181818181817e-05, "loss": 6.2421, "step": 2840 }, { "epoch": 3.18, "learning_rate": 1.2310606060606061e-05, "loss": 6.1335, "step": 2841 }, { "epoch": 3.19, "learning_rate": 1.2303030303030302e-05, "loss": 5.9469, "step": 2842 }, { "epoch": 3.19, "learning_rate": 1.2295454545454546e-05, "loss": 6.0816, "step": 2843 }, { "epoch": 3.19, "learning_rate": 1.2287878787878788e-05, "loss": 6.1629, "step": 2844 }, { "epoch": 3.19, "learning_rate": 1.228030303030303e-05, "loss": 6.0302, "step": 2845 }, { "epoch": 3.19, "learning_rate": 1.2272727272727274e-05, "loss": 6.1572, "step": 2846 }, { "epoch": 3.19, "learning_rate": 1.2265151515151515e-05, "loss": 6.0628, "step": 2847 }, { "epoch": 3.19, "learning_rate": 1.2257575757575759e-05, "loss": 6.1832, "step": 2848 }, { "epoch": 3.19, "learning_rate": 1.225e-05, "loss": 6.117, "step": 2849 }, { "epoch": 3.2, "learning_rate": 1.2242424242424244e-05, "loss": 6.0308, "step": 2850 }, { "epoch": 3.2, "learning_rate": 1.2234848484848484e-05, "loss": 6.1066, "step": 2851 }, { "epoch": 3.2, "learning_rate": 1.2227272727272728e-05, "loss": 6.0451, "step": 2852 }, { "epoch": 3.2, "learning_rate": 1.2219696969696969e-05, "loss": 5.947, "step": 2853 }, { "epoch": 3.2, "learning_rate": 1.2212121212121213e-05, "loss": 6.1145, "step": 2854 }, { "epoch": 3.2, "learning_rate": 1.2204545454545455e-05, "loss": 5.9557, "step": 2855 }, { "epoch": 3.2, "learning_rate": 1.2196969696969697e-05, "loss": 6.1921, "step": 2856 }, { "epoch": 3.2, "learning_rate": 1.218939393939394e-05, "loss": 6.1758, "step": 2857 }, { "epoch": 3.2, "learning_rate": 1.2181818181818182e-05, "loss": 5.8689, "step": 2858 }, { "epoch": 3.21, "learning_rate": 1.2174242424242424e-05, "loss": 6.0666, "step": 2859 }, { "epoch": 3.21, "learning_rate": 1.2166666666666667e-05, "loss": 5.9561, "step": 2860 }, { "epoch": 3.21, "learning_rate": 1.2159090909090909e-05, "loss": 5.8791, "step": 2861 }, { "epoch": 3.21, "learning_rate": 1.2151515151515151e-05, "loss": 6.0524, "step": 2862 }, { "epoch": 3.21, "learning_rate": 1.2143939393939395e-05, "loss": 5.9111, "step": 2863 }, { "epoch": 3.21, "learning_rate": 1.2136363636363637e-05, "loss": 5.9558, "step": 2864 }, { "epoch": 3.21, "learning_rate": 1.212878787878788e-05, "loss": 5.865, "step": 2865 }, { "epoch": 3.21, "learning_rate": 1.2121212121212122e-05, "loss": 5.9116, "step": 2866 }, { "epoch": 3.21, "learning_rate": 1.2113636363636364e-05, "loss": 5.8095, "step": 2867 }, { "epoch": 3.22, "learning_rate": 1.2106060606060607e-05, "loss": 5.9733, "step": 2868 }, { "epoch": 3.22, "learning_rate": 1.2098484848484849e-05, "loss": 5.5788, "step": 2869 }, { "epoch": 3.22, "learning_rate": 1.2090909090909091e-05, "loss": 5.4515, "step": 2870 }, { "epoch": 3.22, "learning_rate": 1.2083333333333333e-05, "loss": 5.3323, "step": 2871 }, { "epoch": 3.22, "learning_rate": 1.2075757575757576e-05, "loss": 5.2368, "step": 2872 }, { "epoch": 3.22, "learning_rate": 1.2068181818181818e-05, "loss": 4.9858, "step": 2873 }, { "epoch": 3.22, "learning_rate": 1.206060606060606e-05, "loss": 4.9115, "step": 2874 }, { "epoch": 3.22, "learning_rate": 1.2053030303030304e-05, "loss": 4.6216, "step": 2875 }, { "epoch": 3.22, "learning_rate": 1.2045454545454545e-05, "loss": 3.9144, "step": 2876 }, { "epoch": 3.23, "learning_rate": 1.2037878787878789e-05, "loss": 6.3547, "step": 2877 }, { "epoch": 3.23, "learning_rate": 1.203030303030303e-05, "loss": 6.3025, "step": 2878 }, { "epoch": 3.23, "learning_rate": 1.2022727272727273e-05, "loss": 6.4685, "step": 2879 }, { "epoch": 3.23, "learning_rate": 1.2015151515151516e-05, "loss": 6.4242, "step": 2880 }, { "epoch": 3.23, "learning_rate": 1.2007575757575758e-05, "loss": 6.2113, "step": 2881 }, { "epoch": 3.23, "learning_rate": 1.2e-05, "loss": 6.1956, "step": 2882 }, { "epoch": 3.23, "learning_rate": 1.1992424242424243e-05, "loss": 6.2195, "step": 2883 }, { "epoch": 3.23, "learning_rate": 1.1984848484848485e-05, "loss": 6.2668, "step": 2884 }, { "epoch": 3.23, "learning_rate": 1.1977272727272727e-05, "loss": 6.1192, "step": 2885 }, { "epoch": 3.24, "learning_rate": 1.1969696969696971e-05, "loss": 6.0004, "step": 2886 }, { "epoch": 3.24, "learning_rate": 1.1962121212121212e-05, "loss": 6.0849, "step": 2887 }, { "epoch": 3.24, "learning_rate": 1.1954545454545456e-05, "loss": 6.2679, "step": 2888 }, { "epoch": 3.24, "learning_rate": 1.1946969696969696e-05, "loss": 6.2139, "step": 2889 }, { "epoch": 3.24, "learning_rate": 1.193939393939394e-05, "loss": 5.9697, "step": 2890 }, { "epoch": 3.24, "learning_rate": 1.1931818181818181e-05, "loss": 6.1767, "step": 2891 }, { "epoch": 3.24, "learning_rate": 1.1924242424242425e-05, "loss": 6.1799, "step": 2892 }, { "epoch": 3.24, "learning_rate": 1.1916666666666666e-05, "loss": 6.0493, "step": 2893 }, { "epoch": 3.24, "learning_rate": 1.190909090909091e-05, "loss": 6.1082, "step": 2894 }, { "epoch": 3.25, "learning_rate": 1.1901515151515152e-05, "loss": 6.0175, "step": 2895 }, { "epoch": 3.25, "learning_rate": 1.1893939393939394e-05, "loss": 6.039, "step": 2896 }, { "epoch": 3.25, "learning_rate": 1.1886363636363638e-05, "loss": 6.078, "step": 2897 }, { "epoch": 3.25, "learning_rate": 1.1878787878787879e-05, "loss": 6.0527, "step": 2898 }, { "epoch": 3.25, "learning_rate": 1.1871212121212123e-05, "loss": 5.8994, "step": 2899 }, { "epoch": 3.25, "learning_rate": 1.1863636363636363e-05, "loss": 5.9851, "step": 2900 }, { "epoch": 3.25, "learning_rate": 1.1856060606060607e-05, "loss": 5.8876, "step": 2901 }, { "epoch": 3.25, "learning_rate": 1.1848484848484848e-05, "loss": 5.9174, "step": 2902 }, { "epoch": 3.25, "learning_rate": 1.1840909090909092e-05, "loss": 6.0145, "step": 2903 }, { "epoch": 3.26, "learning_rate": 1.1833333333333332e-05, "loss": 6.0673, "step": 2904 }, { "epoch": 3.26, "learning_rate": 1.1825757575757576e-05, "loss": 6.1447, "step": 2905 }, { "epoch": 3.26, "learning_rate": 1.1818181818181819e-05, "loss": 6.1951, "step": 2906 }, { "epoch": 3.26, "learning_rate": 1.1810606060606061e-05, "loss": 5.9909, "step": 2907 }, { "epoch": 3.26, "learning_rate": 1.1803030303030303e-05, "loss": 5.944, "step": 2908 }, { "epoch": 3.26, "learning_rate": 1.1795454545454546e-05, "loss": 6.0911, "step": 2909 }, { "epoch": 3.26, "learning_rate": 1.1787878787878788e-05, "loss": 6.051, "step": 2910 }, { "epoch": 3.26, "learning_rate": 1.178030303030303e-05, "loss": 5.9802, "step": 2911 }, { "epoch": 3.26, "learning_rate": 1.1772727272727272e-05, "loss": 5.8403, "step": 2912 }, { "epoch": 3.27, "learning_rate": 1.1765151515151515e-05, "loss": 5.9803, "step": 2913 }, { "epoch": 3.27, "learning_rate": 1.1757575757575759e-05, "loss": 5.9136, "step": 2914 }, { "epoch": 3.27, "learning_rate": 1.1750000000000001e-05, "loss": 5.9589, "step": 2915 }, { "epoch": 3.27, "learning_rate": 1.1742424242424243e-05, "loss": 5.9216, "step": 2916 }, { "epoch": 3.27, "learning_rate": 1.1734848484848486e-05, "loss": 5.9201, "step": 2917 }, { "epoch": 3.27, "learning_rate": 1.1727272727272728e-05, "loss": 5.7474, "step": 2918 }, { "epoch": 3.27, "learning_rate": 1.171969696969697e-05, "loss": 5.9139, "step": 2919 }, { "epoch": 3.27, "learning_rate": 1.1712121212121212e-05, "loss": 5.6268, "step": 2920 }, { "epoch": 3.27, "learning_rate": 1.1704545454545455e-05, "loss": 5.388, "step": 2921 }, { "epoch": 3.28, "learning_rate": 1.1696969696969697e-05, "loss": 5.3447, "step": 2922 }, { "epoch": 3.28, "learning_rate": 1.168939393939394e-05, "loss": 5.2647, "step": 2923 }, { "epoch": 3.28, "learning_rate": 1.1681818181818182e-05, "loss": 4.918, "step": 2924 }, { "epoch": 3.28, "learning_rate": 1.1674242424242424e-05, "loss": 4.7745, "step": 2925 }, { "epoch": 3.28, "learning_rate": 1.1666666666666668e-05, "loss": 4.0022, "step": 2926 }, { "epoch": 3.28, "learning_rate": 1.1659090909090909e-05, "loss": 6.5563, "step": 2927 }, { "epoch": 3.28, "learning_rate": 1.1651515151515152e-05, "loss": 6.384, "step": 2928 }, { "epoch": 3.28, "learning_rate": 1.1643939393939393e-05, "loss": 6.3865, "step": 2929 }, { "epoch": 3.28, "learning_rate": 1.1636363636363637e-05, "loss": 6.3067, "step": 2930 }, { "epoch": 3.29, "learning_rate": 1.162878787878788e-05, "loss": 6.2381, "step": 2931 }, { "epoch": 3.29, "learning_rate": 1.1621212121212122e-05, "loss": 6.2478, "step": 2932 }, { "epoch": 3.29, "learning_rate": 1.1613636363636364e-05, "loss": 6.3514, "step": 2933 }, { "epoch": 3.29, "learning_rate": 1.1606060606060606e-05, "loss": 6.2807, "step": 2934 }, { "epoch": 3.29, "learning_rate": 1.1598484848484849e-05, "loss": 6.2674, "step": 2935 }, { "epoch": 3.29, "learning_rate": 1.159090909090909e-05, "loss": 5.9925, "step": 2936 }, { "epoch": 3.29, "learning_rate": 1.1583333333333335e-05, "loss": 6.1002, "step": 2937 }, { "epoch": 3.29, "learning_rate": 1.1575757575757575e-05, "loss": 6.0631, "step": 2938 }, { "epoch": 3.29, "learning_rate": 1.156818181818182e-05, "loss": 6.0201, "step": 2939 }, { "epoch": 3.3, "learning_rate": 1.156060606060606e-05, "loss": 6.0653, "step": 2940 }, { "epoch": 3.3, "learning_rate": 1.1553030303030304e-05, "loss": 6.2587, "step": 2941 }, { "epoch": 3.3, "learning_rate": 1.1545454545454545e-05, "loss": 6.1217, "step": 2942 }, { "epoch": 3.3, "learning_rate": 1.1537878787878789e-05, "loss": 5.9783, "step": 2943 }, { "epoch": 3.3, "learning_rate": 1.1530303030303029e-05, "loss": 6.0408, "step": 2944 }, { "epoch": 3.3, "learning_rate": 1.1522727272727273e-05, "loss": 6.0999, "step": 2945 }, { "epoch": 3.3, "learning_rate": 1.1515151515151515e-05, "loss": 6.0859, "step": 2946 }, { "epoch": 3.3, "learning_rate": 1.1507575757575758e-05, "loss": 6.0194, "step": 2947 }, { "epoch": 3.3, "learning_rate": 1.1500000000000002e-05, "loss": 6.1582, "step": 2948 }, { "epoch": 3.31, "learning_rate": 1.1492424242424242e-05, "loss": 6.0165, "step": 2949 }, { "epoch": 3.31, "learning_rate": 1.1484848484848486e-05, "loss": 5.9581, "step": 2950 }, { "epoch": 3.31, "learning_rate": 1.1477272727272727e-05, "loss": 6.0901, "step": 2951 }, { "epoch": 3.31, "learning_rate": 1.1469696969696971e-05, "loss": 6.0761, "step": 2952 }, { "epoch": 3.31, "learning_rate": 1.1462121212121211e-05, "loss": 5.9194, "step": 2953 }, { "epoch": 3.31, "learning_rate": 1.1454545454545455e-05, "loss": 6.0594, "step": 2954 }, { "epoch": 3.31, "learning_rate": 1.1446969696969696e-05, "loss": 6.0704, "step": 2955 }, { "epoch": 3.31, "learning_rate": 1.143939393939394e-05, "loss": 6.141, "step": 2956 }, { "epoch": 3.32, "learning_rate": 1.1431818181818182e-05, "loss": 6.0867, "step": 2957 }, { "epoch": 3.32, "learning_rate": 1.1424242424242425e-05, "loss": 6.1224, "step": 2958 }, { "epoch": 3.32, "learning_rate": 1.1416666666666667e-05, "loss": 6.0147, "step": 2959 }, { "epoch": 3.32, "learning_rate": 1.140909090909091e-05, "loss": 5.9816, "step": 2960 }, { "epoch": 3.32, "learning_rate": 1.1401515151515151e-05, "loss": 5.9855, "step": 2961 }, { "epoch": 3.32, "learning_rate": 1.1393939393939394e-05, "loss": 6.2395, "step": 2962 }, { "epoch": 3.32, "learning_rate": 1.1386363636363636e-05, "loss": 5.9338, "step": 2963 }, { "epoch": 3.32, "learning_rate": 1.1378787878787878e-05, "loss": 6.064, "step": 2964 }, { "epoch": 3.32, "learning_rate": 1.1371212121212122e-05, "loss": 6.0427, "step": 2965 }, { "epoch": 3.33, "learning_rate": 1.1363636363636365e-05, "loss": 5.9604, "step": 2966 }, { "epoch": 3.33, "learning_rate": 1.1356060606060607e-05, "loss": 5.8552, "step": 2967 }, { "epoch": 3.33, "learning_rate": 1.134848484848485e-05, "loss": 5.7128, "step": 2968 }, { "epoch": 3.33, "learning_rate": 1.1340909090909092e-05, "loss": 5.7473, "step": 2969 }, { "epoch": 3.33, "learning_rate": 1.1333333333333334e-05, "loss": 5.6238, "step": 2970 }, { "epoch": 3.33, "learning_rate": 1.1325757575757576e-05, "loss": 5.5046, "step": 2971 }, { "epoch": 3.33, "learning_rate": 1.1318181818181818e-05, "loss": 5.5068, "step": 2972 }, { "epoch": 3.33, "learning_rate": 1.131060606060606e-05, "loss": 5.0002, "step": 2973 }, { "epoch": 3.33, "learning_rate": 1.1303030303030303e-05, "loss": 5.0581, "step": 2974 }, { "epoch": 3.34, "learning_rate": 1.1295454545454545e-05, "loss": 4.7729, "step": 2975 }, { "epoch": 3.34, "learning_rate": 1.1287878787878788e-05, "loss": 4.1735, "step": 2976 }, { "epoch": 3.34, "learning_rate": 1.1280303030303032e-05, "loss": 6.4848, "step": 2977 }, { "epoch": 3.34, "learning_rate": 1.1272727272727272e-05, "loss": 6.2888, "step": 2978 }, { "epoch": 3.34, "learning_rate": 1.1265151515151516e-05, "loss": 6.2162, "step": 2979 }, { "epoch": 3.34, "learning_rate": 1.1257575757575757e-05, "loss": 6.275, "step": 2980 }, { "epoch": 3.34, "learning_rate": 1.125e-05, "loss": 6.3532, "step": 2981 }, { "epoch": 3.34, "learning_rate": 1.1242424242424243e-05, "loss": 6.2881, "step": 2982 }, { "epoch": 3.34, "learning_rate": 1.1234848484848485e-05, "loss": 6.2206, "step": 2983 }, { "epoch": 3.35, "learning_rate": 1.1227272727272728e-05, "loss": 6.1535, "step": 2984 }, { "epoch": 3.35, "learning_rate": 1.121969696969697e-05, "loss": 6.2174, "step": 2985 }, { "epoch": 3.35, "learning_rate": 1.1212121212121212e-05, "loss": 6.1428, "step": 2986 }, { "epoch": 3.35, "learning_rate": 1.1204545454545454e-05, "loss": 6.1961, "step": 2987 }, { "epoch": 3.35, "learning_rate": 1.1196969696969698e-05, "loss": 5.9338, "step": 2988 }, { "epoch": 3.35, "learning_rate": 1.1189393939393939e-05, "loss": 6.1103, "step": 2989 }, { "epoch": 3.35, "learning_rate": 1.1181818181818183e-05, "loss": 6.0411, "step": 2990 }, { "epoch": 3.35, "learning_rate": 1.1174242424242424e-05, "loss": 6.1064, "step": 2991 }, { "epoch": 3.35, "learning_rate": 1.1166666666666668e-05, "loss": 6.1179, "step": 2992 }, { "epoch": 3.36, "learning_rate": 1.1159090909090908e-05, "loss": 6.1252, "step": 2993 }, { "epoch": 3.36, "learning_rate": 1.1151515151515152e-05, "loss": 6.0475, "step": 2994 }, { "epoch": 3.36, "learning_rate": 1.1143939393939393e-05, "loss": 6.0511, "step": 2995 }, { "epoch": 3.36, "learning_rate": 1.1136363636363637e-05, "loss": 6.1529, "step": 2996 }, { "epoch": 3.36, "learning_rate": 1.1128787878787879e-05, "loss": 6.0826, "step": 2997 }, { "epoch": 3.36, "learning_rate": 1.1121212121212121e-05, "loss": 6.0413, "step": 2998 }, { "epoch": 3.36, "learning_rate": 1.1113636363636365e-05, "loss": 5.9451, "step": 2999 }, { "epoch": 3.36, "learning_rate": 1.1106060606060606e-05, "loss": 6.0993, "step": 3000 }, { "epoch": 3.36, "eval_loss": 5.770209789276123, "eval_runtime": 881.4753, "eval_samples_per_second": 2.997, "eval_steps_per_second": 0.376, "eval_wer": 1.4096191987306623, "step": 3000 }, { "epoch": 3.36, "learning_rate": 1.109848484848485e-05, "loss": 6.0002, "step": 3001 }, { "epoch": 3.37, "learning_rate": 1.109090909090909e-05, "loss": 6.0435, "step": 3002 }, { "epoch": 3.37, "learning_rate": 1.1083333333333335e-05, "loss": 6.0997, "step": 3003 }, { "epoch": 3.37, "learning_rate": 1.1075757575757575e-05, "loss": 6.1266, "step": 3004 }, { "epoch": 3.37, "learning_rate": 1.1068181818181819e-05, "loss": 6.0179, "step": 3005 }, { "epoch": 3.37, "learning_rate": 1.106060606060606e-05, "loss": 6.0858, "step": 3006 }, { "epoch": 3.37, "learning_rate": 1.1053030303030304e-05, "loss": 5.8385, "step": 3007 }, { "epoch": 3.37, "learning_rate": 1.1045454545454546e-05, "loss": 5.9072, "step": 3008 }, { "epoch": 3.37, "learning_rate": 1.1037878787878788e-05, "loss": 5.8373, "step": 3009 }, { "epoch": 3.37, "learning_rate": 1.103030303030303e-05, "loss": 6.0065, "step": 3010 }, { "epoch": 3.38, "learning_rate": 1.1022727272727273e-05, "loss": 6.1609, "step": 3011 }, { "epoch": 3.38, "learning_rate": 1.1015151515151515e-05, "loss": 6.0093, "step": 3012 }, { "epoch": 3.38, "learning_rate": 1.1007575757575757e-05, "loss": 6.0269, "step": 3013 }, { "epoch": 3.38, "learning_rate": 1.1e-05, "loss": 5.926, "step": 3014 }, { "epoch": 3.38, "learning_rate": 1.0992424242424242e-05, "loss": 5.8557, "step": 3015 }, { "epoch": 3.38, "learning_rate": 1.0984848484848486e-05, "loss": 5.8708, "step": 3016 }, { "epoch": 3.38, "learning_rate": 1.0977272727272728e-05, "loss": 5.9527, "step": 3017 }, { "epoch": 3.38, "learning_rate": 1.096969696969697e-05, "loss": 5.5124, "step": 3018 }, { "epoch": 3.38, "learning_rate": 1.0962121212121213e-05, "loss": 5.8577, "step": 3019 }, { "epoch": 3.39, "learning_rate": 1.0954545454545455e-05, "loss": 5.707, "step": 3020 }, { "epoch": 3.39, "learning_rate": 1.0946969696969697e-05, "loss": 5.6377, "step": 3021 }, { "epoch": 3.39, "learning_rate": 1.093939393939394e-05, "loss": 5.4422, "step": 3022 }, { "epoch": 3.39, "learning_rate": 1.0931818181818182e-05, "loss": 4.9724, "step": 3023 }, { "epoch": 3.39, "learning_rate": 1.0924242424242424e-05, "loss": 4.8427, "step": 3024 }, { "epoch": 3.39, "learning_rate": 1.0916666666666667e-05, "loss": 4.6496, "step": 3025 }, { "epoch": 3.39, "learning_rate": 1.0909090909090909e-05, "loss": 3.879, "step": 3026 }, { "epoch": 3.39, "learning_rate": 1.0901515151515151e-05, "loss": 6.5474, "step": 3027 }, { "epoch": 3.39, "learning_rate": 1.0893939393939395e-05, "loss": 6.2804, "step": 3028 }, { "epoch": 3.4, "learning_rate": 1.0886363636363636e-05, "loss": 6.347, "step": 3029 }, { "epoch": 3.4, "learning_rate": 1.087878787878788e-05, "loss": 6.3937, "step": 3030 }, { "epoch": 3.4, "learning_rate": 1.087121212121212e-05, "loss": 6.1148, "step": 3031 }, { "epoch": 3.4, "learning_rate": 1.0863636363636364e-05, "loss": 6.2733, "step": 3032 }, { "epoch": 3.4, "learning_rate": 1.0856060606060607e-05, "loss": 6.2659, "step": 3033 }, { "epoch": 3.4, "learning_rate": 1.0848484848484849e-05, "loss": 6.2108, "step": 3034 }, { "epoch": 3.4, "learning_rate": 1.0840909090909091e-05, "loss": 6.1291, "step": 3035 }, { "epoch": 3.4, "learning_rate": 1.0833333333333334e-05, "loss": 6.2991, "step": 3036 }, { "epoch": 3.4, "learning_rate": 1.0825757575757576e-05, "loss": 6.1573, "step": 3037 }, { "epoch": 3.41, "learning_rate": 1.0818181818181818e-05, "loss": 6.1326, "step": 3038 }, { "epoch": 3.41, "learning_rate": 1.0810606060606062e-05, "loss": 6.1355, "step": 3039 }, { "epoch": 3.41, "learning_rate": 1.0803030303030303e-05, "loss": 6.111, "step": 3040 }, { "epoch": 3.41, "learning_rate": 1.0795454545454547e-05, "loss": 6.0561, "step": 3041 }, { "epoch": 3.41, "learning_rate": 1.0787878787878787e-05, "loss": 5.9972, "step": 3042 }, { "epoch": 3.41, "learning_rate": 1.0780303030303031e-05, "loss": 6.1046, "step": 3043 }, { "epoch": 3.41, "learning_rate": 1.0772727272727272e-05, "loss": 6.0881, "step": 3044 }, { "epoch": 3.41, "learning_rate": 1.0765151515151516e-05, "loss": 5.947, "step": 3045 }, { "epoch": 3.41, "learning_rate": 1.0757575757575756e-05, "loss": 6.0035, "step": 3046 }, { "epoch": 3.42, "learning_rate": 1.075e-05, "loss": 5.878, "step": 3047 }, { "epoch": 3.42, "learning_rate": 1.0742424242424243e-05, "loss": 5.98, "step": 3048 }, { "epoch": 3.42, "learning_rate": 1.0734848484848485e-05, "loss": 6.0004, "step": 3049 }, { "epoch": 3.42, "learning_rate": 1.0727272727272729e-05, "loss": 6.0753, "step": 3050 }, { "epoch": 3.42, "learning_rate": 1.071969696969697e-05, "loss": 6.0061, "step": 3051 }, { "epoch": 3.42, "learning_rate": 1.0712121212121214e-05, "loss": 6.0032, "step": 3052 }, { "epoch": 3.42, "learning_rate": 1.0704545454545454e-05, "loss": 6.0593, "step": 3053 }, { "epoch": 3.42, "learning_rate": 1.0696969696969698e-05, "loss": 6.2167, "step": 3054 }, { "epoch": 3.42, "learning_rate": 1.0689393939393939e-05, "loss": 6.0656, "step": 3055 }, { "epoch": 3.43, "learning_rate": 1.0681818181818183e-05, "loss": 5.7744, "step": 3056 }, { "epoch": 3.43, "learning_rate": 1.0674242424242423e-05, "loss": 6.0111, "step": 3057 }, { "epoch": 3.43, "learning_rate": 1.0666666666666667e-05, "loss": 6.0364, "step": 3058 }, { "epoch": 3.43, "learning_rate": 1.065909090909091e-05, "loss": 5.9357, "step": 3059 }, { "epoch": 3.43, "learning_rate": 1.0651515151515152e-05, "loss": 5.9229, "step": 3060 }, { "epoch": 3.43, "learning_rate": 1.0643939393939394e-05, "loss": 6.2102, "step": 3061 }, { "epoch": 3.43, "learning_rate": 1.0636363636363636e-05, "loss": 6.0377, "step": 3062 }, { "epoch": 3.43, "learning_rate": 1.0628787878787879e-05, "loss": 5.912, "step": 3063 }, { "epoch": 3.43, "learning_rate": 1.0621212121212121e-05, "loss": 6.0793, "step": 3064 }, { "epoch": 3.44, "learning_rate": 1.0613636363636363e-05, "loss": 5.9411, "step": 3065 }, { "epoch": 3.44, "learning_rate": 1.0606060606060606e-05, "loss": 5.7931, "step": 3066 }, { "epoch": 3.44, "learning_rate": 1.059848484848485e-05, "loss": 5.8981, "step": 3067 }, { "epoch": 3.44, "learning_rate": 1.0590909090909092e-05, "loss": 5.8399, "step": 3068 }, { "epoch": 3.44, "learning_rate": 1.0583333333333334e-05, "loss": 5.9336, "step": 3069 }, { "epoch": 3.44, "learning_rate": 1.0575757575757577e-05, "loss": 5.6477, "step": 3070 }, { "epoch": 3.44, "learning_rate": 1.0568181818181819e-05, "loss": 5.4915, "step": 3071 }, { "epoch": 3.44, "learning_rate": 1.0560606060606061e-05, "loss": 5.4761, "step": 3072 }, { "epoch": 3.45, "learning_rate": 1.0553030303030303e-05, "loss": 5.0007, "step": 3073 }, { "epoch": 3.45, "learning_rate": 1.0545454545454546e-05, "loss": 5.07, "step": 3074 }, { "epoch": 3.45, "learning_rate": 1.0537878787878788e-05, "loss": 4.6934, "step": 3075 }, { "epoch": 3.45, "learning_rate": 1.053030303030303e-05, "loss": 4.0287, "step": 3076 }, { "epoch": 3.45, "learning_rate": 1.0522727272727273e-05, "loss": 6.31, "step": 3077 }, { "epoch": 3.45, "learning_rate": 1.0515151515151515e-05, "loss": 6.3855, "step": 3078 }, { "epoch": 3.45, "learning_rate": 1.0507575757575759e-05, "loss": 6.4226, "step": 3079 }, { "epoch": 3.45, "learning_rate": 1.05e-05, "loss": 6.3444, "step": 3080 }, { "epoch": 3.45, "learning_rate": 1.0492424242424243e-05, "loss": 6.2597, "step": 3081 }, { "epoch": 3.46, "learning_rate": 1.0484848484848484e-05, "loss": 6.295, "step": 3082 }, { "epoch": 3.46, "learning_rate": 1.0477272727272728e-05, "loss": 6.2445, "step": 3083 }, { "epoch": 3.46, "learning_rate": 1.046969696969697e-05, "loss": 6.2017, "step": 3084 }, { "epoch": 3.46, "learning_rate": 1.0462121212121213e-05, "loss": 6.1275, "step": 3085 }, { "epoch": 3.46, "learning_rate": 1.0454545454545455e-05, "loss": 6.1797, "step": 3086 }, { "epoch": 3.46, "learning_rate": 1.0446969696969697e-05, "loss": 6.1682, "step": 3087 }, { "epoch": 3.46, "learning_rate": 1.043939393939394e-05, "loss": 6.1193, "step": 3088 }, { "epoch": 3.46, "learning_rate": 1.0431818181818182e-05, "loss": 6.0864, "step": 3089 }, { "epoch": 3.46, "learning_rate": 1.0424242424242426e-05, "loss": 6.0325, "step": 3090 }, { "epoch": 3.47, "learning_rate": 1.0416666666666666e-05, "loss": 6.0845, "step": 3091 }, { "epoch": 3.47, "learning_rate": 1.040909090909091e-05, "loss": 6.0404, "step": 3092 }, { "epoch": 3.47, "learning_rate": 1.0401515151515151e-05, "loss": 6.075, "step": 3093 }, { "epoch": 3.47, "learning_rate": 1.0393939393939395e-05, "loss": 6.091, "step": 3094 }, { "epoch": 3.47, "learning_rate": 1.0386363636363635e-05, "loss": 5.9912, "step": 3095 }, { "epoch": 3.47, "learning_rate": 1.037878787878788e-05, "loss": 6.1101, "step": 3096 }, { "epoch": 3.47, "learning_rate": 1.037121212121212e-05, "loss": 6.1094, "step": 3097 }, { "epoch": 3.47, "learning_rate": 1.0363636363636364e-05, "loss": 6.0213, "step": 3098 }, { "epoch": 3.47, "learning_rate": 1.0356060606060606e-05, "loss": 6.0304, "step": 3099 }, { "epoch": 3.48, "learning_rate": 1.0348484848484849e-05, "loss": 5.8756, "step": 3100 }, { "epoch": 3.48, "learning_rate": 1.0340909090909093e-05, "loss": 6.0791, "step": 3101 }, { "epoch": 3.48, "learning_rate": 1.0333333333333333e-05, "loss": 6.0658, "step": 3102 }, { "epoch": 3.48, "learning_rate": 1.0325757575757577e-05, "loss": 6.0121, "step": 3103 }, { "epoch": 3.48, "learning_rate": 1.0318181818181818e-05, "loss": 5.9451, "step": 3104 }, { "epoch": 3.48, "learning_rate": 1.0310606060606062e-05, "loss": 6.0398, "step": 3105 }, { "epoch": 3.48, "learning_rate": 1.0303030303030302e-05, "loss": 5.9369, "step": 3106 }, { "epoch": 3.48, "learning_rate": 1.0295454545454546e-05, "loss": 6.0814, "step": 3107 }, { "epoch": 3.48, "learning_rate": 1.0287878787878787e-05, "loss": 5.9369, "step": 3108 }, { "epoch": 3.49, "learning_rate": 1.0280303030303031e-05, "loss": 6.0422, "step": 3109 }, { "epoch": 3.49, "learning_rate": 1.0272727272727273e-05, "loss": 6.0667, "step": 3110 }, { "epoch": 3.49, "learning_rate": 1.0265151515151516e-05, "loss": 5.867, "step": 3111 }, { "epoch": 3.49, "learning_rate": 1.0257575757575758e-05, "loss": 6.0244, "step": 3112 }, { "epoch": 3.49, "learning_rate": 1.025e-05, "loss": 5.9211, "step": 3113 }, { "epoch": 3.49, "learning_rate": 1.0242424242424242e-05, "loss": 5.9538, "step": 3114 }, { "epoch": 3.49, "learning_rate": 1.0234848484848485e-05, "loss": 5.8176, "step": 3115 }, { "epoch": 3.49, "learning_rate": 1.0227272727272727e-05, "loss": 5.8052, "step": 3116 }, { "epoch": 3.49, "learning_rate": 1.021969696969697e-05, "loss": 5.8502, "step": 3117 }, { "epoch": 3.5, "learning_rate": 1.0212121212121213e-05, "loss": 5.5941, "step": 3118 }, { "epoch": 3.5, "learning_rate": 1.0204545454545456e-05, "loss": 5.5501, "step": 3119 }, { "epoch": 3.5, "learning_rate": 1.0196969696969698e-05, "loss": 5.644, "step": 3120 }, { "epoch": 3.5, "learning_rate": 1.018939393939394e-05, "loss": 5.3506, "step": 3121 }, { "epoch": 3.5, "learning_rate": 1.0181818181818182e-05, "loss": 5.3255, "step": 3122 }, { "epoch": 3.5, "learning_rate": 1.0174242424242425e-05, "loss": 5.496, "step": 3123 }, { "epoch": 3.5, "learning_rate": 1.0166666666666667e-05, "loss": 4.9926, "step": 3124 }, { "epoch": 3.5, "learning_rate": 1.015909090909091e-05, "loss": 4.5274, "step": 3125 }, { "epoch": 3.5, "learning_rate": 1.0151515151515152e-05, "loss": 4.1335, "step": 3126 }, { "epoch": 3.51, "learning_rate": 1.0143939393939394e-05, "loss": 6.2812, "step": 3127 }, { "epoch": 3.51, "learning_rate": 1.0136363636363636e-05, "loss": 6.3156, "step": 3128 }, { "epoch": 3.51, "learning_rate": 1.0128787878787878e-05, "loss": 6.3577, "step": 3129 }, { "epoch": 3.51, "learning_rate": 1.0121212121212122e-05, "loss": 6.2707, "step": 3130 }, { "epoch": 3.51, "learning_rate": 1.0113636363636363e-05, "loss": 6.1539, "step": 3131 }, { "epoch": 3.51, "learning_rate": 1.0106060606060607e-05, "loss": 6.1595, "step": 3132 }, { "epoch": 3.51, "learning_rate": 1.0098484848484848e-05, "loss": 6.3983, "step": 3133 }, { "epoch": 3.51, "learning_rate": 1.0090909090909092e-05, "loss": 6.2183, "step": 3134 }, { "epoch": 3.51, "learning_rate": 1.0083333333333334e-05, "loss": 6.1978, "step": 3135 }, { "epoch": 3.52, "learning_rate": 1.0075757575757576e-05, "loss": 6.0939, "step": 3136 }, { "epoch": 3.52, "learning_rate": 1.0068181818181819e-05, "loss": 6.2363, "step": 3137 }, { "epoch": 3.52, "learning_rate": 1.006060606060606e-05, "loss": 6.0419, "step": 3138 }, { "epoch": 3.52, "learning_rate": 1.0053030303030303e-05, "loss": 5.895, "step": 3139 }, { "epoch": 3.52, "learning_rate": 1.0045454545454545e-05, "loss": 6.0496, "step": 3140 }, { "epoch": 3.52, "learning_rate": 1.003787878787879e-05, "loss": 6.1931, "step": 3141 }, { "epoch": 3.52, "learning_rate": 1.003030303030303e-05, "loss": 6.1425, "step": 3142 }, { "epoch": 3.52, "learning_rate": 1.0022727272727274e-05, "loss": 6.0857, "step": 3143 }, { "epoch": 3.52, "learning_rate": 1.0015151515151515e-05, "loss": 6.1065, "step": 3144 }, { "epoch": 3.53, "learning_rate": 1.0007575757575759e-05, "loss": 5.9491, "step": 3145 }, { "epoch": 3.53, "learning_rate": 9.999999999999999e-06, "loss": 6.1445, "step": 3146 }, { "epoch": 3.53, "learning_rate": 9.992424242424243e-06, "loss": 6.0558, "step": 3147 }, { "epoch": 3.53, "learning_rate": 9.984848484848484e-06, "loss": 6.1288, "step": 3148 }, { "epoch": 3.53, "learning_rate": 9.977272727272728e-06, "loss": 6.022, "step": 3149 }, { "epoch": 3.53, "learning_rate": 9.96969696969697e-06, "loss": 5.9757, "step": 3150 }, { "epoch": 3.53, "learning_rate": 9.962121212121212e-06, "loss": 6.0333, "step": 3151 }, { "epoch": 3.53, "learning_rate": 9.954545454545456e-06, "loss": 5.9891, "step": 3152 }, { "epoch": 3.53, "learning_rate": 9.946969696969697e-06, "loss": 5.9725, "step": 3153 }, { "epoch": 3.54, "learning_rate": 9.93939393939394e-06, "loss": 6.1012, "step": 3154 }, { "epoch": 3.54, "learning_rate": 9.931818181818181e-06, "loss": 6.0649, "step": 3155 }, { "epoch": 3.54, "learning_rate": 9.924242424242425e-06, "loss": 6.0404, "step": 3156 }, { "epoch": 3.54, "learning_rate": 9.916666666666666e-06, "loss": 6.016, "step": 3157 }, { "epoch": 3.54, "learning_rate": 9.90909090909091e-06, "loss": 5.9059, "step": 3158 }, { "epoch": 3.54, "learning_rate": 9.90151515151515e-06, "loss": 5.9029, "step": 3159 }, { "epoch": 3.54, "learning_rate": 9.893939393939395e-06, "loss": 6.0406, "step": 3160 }, { "epoch": 3.54, "learning_rate": 9.886363636363637e-06, "loss": 5.9844, "step": 3161 }, { "epoch": 3.54, "learning_rate": 9.87878787878788e-06, "loss": 5.8963, "step": 3162 }, { "epoch": 3.55, "learning_rate": 9.871212121212121e-06, "loss": 5.8716, "step": 3163 }, { "epoch": 3.55, "learning_rate": 9.863636363636364e-06, "loss": 5.7271, "step": 3164 }, { "epoch": 3.55, "learning_rate": 9.856060606060606e-06, "loss": 5.9779, "step": 3165 }, { "epoch": 3.55, "learning_rate": 9.848484848484848e-06, "loss": 5.8718, "step": 3166 }, { "epoch": 3.55, "learning_rate": 9.84090909090909e-06, "loss": 5.8872, "step": 3167 }, { "epoch": 3.55, "learning_rate": 9.833333333333333e-06, "loss": 5.6183, "step": 3168 }, { "epoch": 3.55, "learning_rate": 9.825757575757577e-06, "loss": 5.4232, "step": 3169 }, { "epoch": 3.55, "learning_rate": 9.81818181818182e-06, "loss": 5.4623, "step": 3170 }, { "epoch": 3.55, "learning_rate": 9.810606060606061e-06, "loss": 5.5047, "step": 3171 }, { "epoch": 3.56, "learning_rate": 9.803030303030304e-06, "loss": 5.2013, "step": 3172 }, { "epoch": 3.56, "learning_rate": 9.795454545454546e-06, "loss": 4.9982, "step": 3173 }, { "epoch": 3.56, "learning_rate": 9.787878787878788e-06, "loss": 5.0807, "step": 3174 }, { "epoch": 3.56, "learning_rate": 9.78030303030303e-06, "loss": 4.5435, "step": 3175 }, { "epoch": 3.56, "learning_rate": 9.772727272727273e-06, "loss": 4.0904, "step": 3176 }, { "epoch": 3.56, "learning_rate": 9.765151515151515e-06, "loss": 6.1924, "step": 3177 }, { "epoch": 3.56, "learning_rate": 9.757575757575758e-06, "loss": 6.1912, "step": 3178 }, { "epoch": 3.56, "learning_rate": 9.75e-06, "loss": 6.1883, "step": 3179 }, { "epoch": 3.57, "learning_rate": 9.742424242424242e-06, "loss": 6.409, "step": 3180 }, { "epoch": 3.57, "learning_rate": 9.734848484848486e-06, "loss": 6.385, "step": 3181 }, { "epoch": 3.57, "learning_rate": 9.727272727272727e-06, "loss": 6.1951, "step": 3182 }, { "epoch": 3.57, "learning_rate": 9.71969696969697e-06, "loss": 6.1971, "step": 3183 }, { "epoch": 3.57, "learning_rate": 9.712121212121211e-06, "loss": 6.0317, "step": 3184 }, { "epoch": 3.57, "learning_rate": 9.704545454545455e-06, "loss": 6.1087, "step": 3185 }, { "epoch": 3.57, "learning_rate": 9.696969696969698e-06, "loss": 6.183, "step": 3186 }, { "epoch": 3.57, "learning_rate": 9.68939393939394e-06, "loss": 6.1276, "step": 3187 }, { "epoch": 3.57, "learning_rate": 9.681818181818182e-06, "loss": 6.1228, "step": 3188 }, { "epoch": 3.58, "learning_rate": 9.674242424242424e-06, "loss": 6.0602, "step": 3189 }, { "epoch": 3.58, "learning_rate": 9.666666666666667e-06, "loss": 6.0649, "step": 3190 }, { "epoch": 3.58, "learning_rate": 9.659090909090909e-06, "loss": 6.0853, "step": 3191 }, { "epoch": 3.58, "learning_rate": 9.651515151515153e-06, "loss": 6.2326, "step": 3192 }, { "epoch": 3.58, "learning_rate": 9.643939393939394e-06, "loss": 6.1293, "step": 3193 }, { "epoch": 3.58, "learning_rate": 9.636363636363638e-06, "loss": 6.0395, "step": 3194 }, { "epoch": 3.58, "learning_rate": 9.628787878787878e-06, "loss": 5.9826, "step": 3195 }, { "epoch": 3.58, "learning_rate": 9.621212121212122e-06, "loss": 5.9549, "step": 3196 }, { "epoch": 3.58, "learning_rate": 9.613636363636363e-06, "loss": 6.0249, "step": 3197 }, { "epoch": 3.59, "learning_rate": 9.606060606060607e-06, "loss": 6.0483, "step": 3198 }, { "epoch": 3.59, "learning_rate": 9.598484848484847e-06, "loss": 6.0135, "step": 3199 }, { "epoch": 3.59, "learning_rate": 9.590909090909091e-06, "loss": 6.0822, "step": 3200 }, { "epoch": 3.59, "learning_rate": 9.583333333333334e-06, "loss": 6.0149, "step": 3201 }, { "epoch": 3.59, "learning_rate": 9.575757575757576e-06, "loss": 5.93, "step": 3202 }, { "epoch": 3.59, "learning_rate": 9.56818181818182e-06, "loss": 5.9912, "step": 3203 }, { "epoch": 3.59, "learning_rate": 9.56060606060606e-06, "loss": 5.894, "step": 3204 }, { "epoch": 3.59, "learning_rate": 9.553030303030304e-06, "loss": 5.8374, "step": 3205 }, { "epoch": 3.59, "learning_rate": 9.545454545454545e-06, "loss": 6.1222, "step": 3206 }, { "epoch": 3.6, "learning_rate": 9.537878787878789e-06, "loss": 6.0707, "step": 3207 }, { "epoch": 3.6, "learning_rate": 9.53030303030303e-06, "loss": 6.0342, "step": 3208 }, { "epoch": 3.6, "learning_rate": 9.522727272727274e-06, "loss": 5.9023, "step": 3209 }, { "epoch": 3.6, "learning_rate": 9.515151515151514e-06, "loss": 5.9425, "step": 3210 }, { "epoch": 3.6, "learning_rate": 9.507575757575758e-06, "loss": 5.9372, "step": 3211 }, { "epoch": 3.6, "learning_rate": 9.5e-06, "loss": 5.9863, "step": 3212 }, { "epoch": 3.6, "learning_rate": 9.492424242424243e-06, "loss": 5.8839, "step": 3213 }, { "epoch": 3.6, "learning_rate": 9.484848484848485e-06, "loss": 5.7893, "step": 3214 }, { "epoch": 3.6, "learning_rate": 9.477272727272727e-06, "loss": 5.8477, "step": 3215 }, { "epoch": 3.61, "learning_rate": 9.46969696969697e-06, "loss": 6.0492, "step": 3216 }, { "epoch": 3.61, "learning_rate": 9.462121212121212e-06, "loss": 5.8273, "step": 3217 }, { "epoch": 3.61, "learning_rate": 9.454545454545454e-06, "loss": 5.6117, "step": 3218 }, { "epoch": 3.61, "learning_rate": 9.446969696969697e-06, "loss": 5.7431, "step": 3219 }, { "epoch": 3.61, "learning_rate": 9.43939393939394e-06, "loss": 5.6493, "step": 3220 }, { "epoch": 3.61, "learning_rate": 9.431818181818183e-06, "loss": 5.5633, "step": 3221 }, { "epoch": 3.61, "learning_rate": 9.424242424242425e-06, "loss": 5.3781, "step": 3222 }, { "epoch": 3.61, "learning_rate": 9.416666666666667e-06, "loss": 5.0526, "step": 3223 }, { "epoch": 3.61, "learning_rate": 9.40909090909091e-06, "loss": 4.776, "step": 3224 }, { "epoch": 3.62, "learning_rate": 9.401515151515152e-06, "loss": 4.8805, "step": 3225 }, { "epoch": 3.62, "learning_rate": 9.393939393939394e-06, "loss": 4.0805, "step": 3226 }, { "epoch": 3.62, "learning_rate": 9.386363636363637e-06, "loss": 6.4089, "step": 3227 }, { "epoch": 3.62, "learning_rate": 9.378787878787879e-06, "loss": 6.5216, "step": 3228 }, { "epoch": 3.62, "learning_rate": 9.371212121212121e-06, "loss": 6.4216, "step": 3229 }, { "epoch": 3.62, "learning_rate": 9.363636363636363e-06, "loss": 6.1218, "step": 3230 }, { "epoch": 3.62, "learning_rate": 9.356060606060606e-06, "loss": 6.2012, "step": 3231 }, { "epoch": 3.62, "learning_rate": 9.34848484848485e-06, "loss": 6.2533, "step": 3232 }, { "epoch": 3.62, "learning_rate": 9.34090909090909e-06, "loss": 6.1856, "step": 3233 }, { "epoch": 3.63, "learning_rate": 9.333333333333334e-06, "loss": 6.0602, "step": 3234 }, { "epoch": 3.63, "learning_rate": 9.325757575757575e-06, "loss": 6.1835, "step": 3235 }, { "epoch": 3.63, "learning_rate": 9.318181818181819e-06, "loss": 6.2629, "step": 3236 }, { "epoch": 3.63, "learning_rate": 9.31060606060606e-06, "loss": 6.1854, "step": 3237 }, { "epoch": 3.63, "learning_rate": 9.303030303030303e-06, "loss": 6.1444, "step": 3238 }, { "epoch": 3.63, "learning_rate": 9.295454545454546e-06, "loss": 6.0965, "step": 3239 }, { "epoch": 3.63, "learning_rate": 9.287878787878788e-06, "loss": 5.9767, "step": 3240 }, { "epoch": 3.63, "learning_rate": 9.28030303030303e-06, "loss": 6.0341, "step": 3241 }, { "epoch": 3.63, "learning_rate": 9.272727272727273e-06, "loss": 6.0102, "step": 3242 }, { "epoch": 3.64, "learning_rate": 9.265151515151517e-06, "loss": 5.9513, "step": 3243 }, { "epoch": 3.64, "learning_rate": 9.257575757575757e-06, "loss": 6.0204, "step": 3244 }, { "epoch": 3.64, "learning_rate": 9.250000000000001e-06, "loss": 5.9961, "step": 3245 }, { "epoch": 3.64, "learning_rate": 9.242424242424242e-06, "loss": 6.1252, "step": 3246 }, { "epoch": 3.64, "learning_rate": 9.234848484848486e-06, "loss": 6.0768, "step": 3247 }, { "epoch": 3.64, "learning_rate": 9.227272727272726e-06, "loss": 5.9993, "step": 3248 }, { "epoch": 3.64, "learning_rate": 9.21969696969697e-06, "loss": 6.0925, "step": 3249 }, { "epoch": 3.64, "learning_rate": 9.212121212121211e-06, "loss": 6.1221, "step": 3250 }, { "epoch": 3.64, "learning_rate": 9.204545454545455e-06, "loss": 6.0426, "step": 3251 }, { "epoch": 3.65, "learning_rate": 9.196969696969697e-06, "loss": 5.9109, "step": 3252 }, { "epoch": 3.65, "learning_rate": 9.18939393939394e-06, "loss": 5.9599, "step": 3253 }, { "epoch": 3.65, "learning_rate": 9.181818181818182e-06, "loss": 5.9607, "step": 3254 }, { "epoch": 3.65, "learning_rate": 9.174242424242424e-06, "loss": 6.0742, "step": 3255 }, { "epoch": 3.65, "learning_rate": 9.166666666666668e-06, "loss": 5.9653, "step": 3256 }, { "epoch": 3.65, "learning_rate": 9.159090909090909e-06, "loss": 6.0707, "step": 3257 }, { "epoch": 3.65, "learning_rate": 9.151515151515153e-06, "loss": 6.0373, "step": 3258 }, { "epoch": 3.65, "learning_rate": 9.143939393939393e-06, "loss": 6.0122, "step": 3259 }, { "epoch": 3.65, "learning_rate": 9.136363636363637e-06, "loss": 5.9696, "step": 3260 }, { "epoch": 3.66, "learning_rate": 9.128787878787878e-06, "loss": 5.9998, "step": 3261 }, { "epoch": 3.66, "learning_rate": 9.121212121212122e-06, "loss": 5.8565, "step": 3262 }, { "epoch": 3.66, "learning_rate": 9.113636363636364e-06, "loss": 5.9253, "step": 3263 }, { "epoch": 3.66, "learning_rate": 9.106060606060606e-06, "loss": 5.9662, "step": 3264 }, { "epoch": 3.66, "learning_rate": 9.098484848484849e-06, "loss": 6.0112, "step": 3265 }, { "epoch": 3.66, "learning_rate": 9.090909090909091e-06, "loss": 5.7195, "step": 3266 }, { "epoch": 3.66, "learning_rate": 9.083333333333333e-06, "loss": 5.9412, "step": 3267 }, { "epoch": 3.66, "learning_rate": 9.075757575757576e-06, "loss": 5.5053, "step": 3268 }, { "epoch": 3.66, "learning_rate": 9.068181818181818e-06, "loss": 5.7433, "step": 3269 }, { "epoch": 3.67, "learning_rate": 9.06060606060606e-06, "loss": 5.6069, "step": 3270 }, { "epoch": 3.67, "learning_rate": 9.053030303030302e-06, "loss": 5.5765, "step": 3271 }, { "epoch": 3.67, "learning_rate": 9.045454545454546e-06, "loss": 5.1281, "step": 3272 }, { "epoch": 3.67, "learning_rate": 9.037878787878789e-06, "loss": 5.1235, "step": 3273 }, { "epoch": 3.67, "learning_rate": 9.030303030303031e-06, "loss": 4.8744, "step": 3274 }, { "epoch": 3.67, "learning_rate": 9.022727272727273e-06, "loss": 4.7366, "step": 3275 }, { "epoch": 3.67, "learning_rate": 9.015151515151516e-06, "loss": 4.0879, "step": 3276 }, { "epoch": 3.67, "learning_rate": 9.007575757575758e-06, "loss": 6.4885, "step": 3277 }, { "epoch": 3.67, "learning_rate": 9e-06, "loss": 6.0749, "step": 3278 }, { "epoch": 3.68, "learning_rate": 8.992424242424243e-06, "loss": 6.119, "step": 3279 }, { "epoch": 3.68, "learning_rate": 8.984848484848485e-06, "loss": 6.2292, "step": 3280 }, { "epoch": 3.68, "learning_rate": 8.977272727272727e-06, "loss": 6.2212, "step": 3281 }, { "epoch": 3.68, "learning_rate": 8.96969696969697e-06, "loss": 6.0858, "step": 3282 }, { "epoch": 3.68, "learning_rate": 8.962121212121213e-06, "loss": 6.2413, "step": 3283 }, { "epoch": 3.68, "learning_rate": 8.954545454545454e-06, "loss": 6.2011, "step": 3284 }, { "epoch": 3.68, "learning_rate": 8.946969696969698e-06, "loss": 6.0624, "step": 3285 }, { "epoch": 3.68, "learning_rate": 8.939393939393939e-06, "loss": 6.1338, "step": 3286 }, { "epoch": 3.68, "learning_rate": 8.931818181818183e-06, "loss": 6.1297, "step": 3287 }, { "epoch": 3.69, "learning_rate": 8.924242424242423e-06, "loss": 5.93, "step": 3288 }, { "epoch": 3.69, "learning_rate": 8.916666666666667e-06, "loss": 6.0967, "step": 3289 }, { "epoch": 3.69, "learning_rate": 8.90909090909091e-06, "loss": 5.9118, "step": 3290 }, { "epoch": 3.69, "learning_rate": 8.901515151515152e-06, "loss": 5.8433, "step": 3291 }, { "epoch": 3.69, "learning_rate": 8.893939393939394e-06, "loss": 6.1135, "step": 3292 }, { "epoch": 3.69, "learning_rate": 8.886363636363636e-06, "loss": 6.0063, "step": 3293 }, { "epoch": 3.69, "learning_rate": 8.87878787878788e-06, "loss": 6.0394, "step": 3294 }, { "epoch": 3.69, "learning_rate": 8.871212121212121e-06, "loss": 6.0667, "step": 3295 }, { "epoch": 3.7, "learning_rate": 8.863636363636365e-06, "loss": 5.8964, "step": 3296 }, { "epoch": 3.7, "learning_rate": 8.856060606060605e-06, "loss": 5.9865, "step": 3297 }, { "epoch": 3.7, "learning_rate": 8.84848484848485e-06, "loss": 6.0968, "step": 3298 }, { "epoch": 3.7, "learning_rate": 8.84090909090909e-06, "loss": 6.0512, "step": 3299 }, { "epoch": 3.7, "learning_rate": 8.833333333333334e-06, "loss": 6.0153, "step": 3300 }, { "epoch": 3.7, "learning_rate": 8.825757575757575e-06, "loss": 5.8984, "step": 3301 }, { "epoch": 3.7, "learning_rate": 8.818181818181819e-06, "loss": 5.9947, "step": 3302 }, { "epoch": 3.7, "learning_rate": 8.810606060606061e-06, "loss": 6.1704, "step": 3303 }, { "epoch": 3.7, "learning_rate": 8.803030303030303e-06, "loss": 6.0721, "step": 3304 }, { "epoch": 3.71, "learning_rate": 8.795454545454545e-06, "loss": 5.9846, "step": 3305 }, { "epoch": 3.71, "learning_rate": 8.787878787878788e-06, "loss": 5.8902, "step": 3306 }, { "epoch": 3.71, "learning_rate": 8.780303030303032e-06, "loss": 5.9474, "step": 3307 }, { "epoch": 3.71, "learning_rate": 8.772727272727272e-06, "loss": 5.9768, "step": 3308 }, { "epoch": 3.71, "learning_rate": 8.765151515151516e-06, "loss": 5.8644, "step": 3309 }, { "epoch": 3.71, "learning_rate": 8.757575757575757e-06, "loss": 5.8595, "step": 3310 }, { "epoch": 3.71, "learning_rate": 8.750000000000001e-06, "loss": 5.9499, "step": 3311 }, { "epoch": 3.71, "learning_rate": 8.742424242424242e-06, "loss": 5.7839, "step": 3312 }, { "epoch": 3.71, "learning_rate": 8.734848484848486e-06, "loss": 5.958, "step": 3313 }, { "epoch": 3.72, "learning_rate": 8.727272727272728e-06, "loss": 5.9132, "step": 3314 }, { "epoch": 3.72, "learning_rate": 8.71969696969697e-06, "loss": 5.8704, "step": 3315 }, { "epoch": 3.72, "learning_rate": 8.712121212121212e-06, "loss": 5.8749, "step": 3316 }, { "epoch": 3.72, "learning_rate": 8.704545454545455e-06, "loss": 5.6569, "step": 3317 }, { "epoch": 3.72, "learning_rate": 8.696969696969697e-06, "loss": 5.7849, "step": 3318 }, { "epoch": 3.72, "learning_rate": 8.68939393939394e-06, "loss": 5.8084, "step": 3319 }, { "epoch": 3.72, "learning_rate": 8.681818181818182e-06, "loss": 5.6062, "step": 3320 }, { "epoch": 3.72, "learning_rate": 8.674242424242424e-06, "loss": 5.4582, "step": 3321 }, { "epoch": 3.72, "learning_rate": 8.666666666666666e-06, "loss": 5.4227, "step": 3322 }, { "epoch": 3.73, "learning_rate": 8.65909090909091e-06, "loss": 4.9563, "step": 3323 }, { "epoch": 3.73, "learning_rate": 8.651515151515152e-06, "loss": 5.038, "step": 3324 }, { "epoch": 3.73, "learning_rate": 8.643939393939395e-06, "loss": 4.5636, "step": 3325 }, { "epoch": 3.73, "learning_rate": 8.636363636363637e-06, "loss": 3.9352, "step": 3326 }, { "epoch": 3.73, "learning_rate": 8.62878787878788e-06, "loss": 6.2813, "step": 3327 }, { "epoch": 3.73, "learning_rate": 8.621212121212122e-06, "loss": 6.12, "step": 3328 }, { "epoch": 3.73, "learning_rate": 8.613636363636364e-06, "loss": 6.4493, "step": 3329 }, { "epoch": 3.73, "learning_rate": 8.606060606060606e-06, "loss": 6.4303, "step": 3330 }, { "epoch": 3.73, "learning_rate": 8.598484848484848e-06, "loss": 6.2316, "step": 3331 }, { "epoch": 3.74, "learning_rate": 8.59090909090909e-06, "loss": 6.4408, "step": 3332 }, { "epoch": 3.74, "learning_rate": 8.583333333333333e-06, "loss": 6.2567, "step": 3333 }, { "epoch": 3.74, "learning_rate": 8.575757575757577e-06, "loss": 6.2573, "step": 3334 }, { "epoch": 3.74, "learning_rate": 8.568181818181818e-06, "loss": 6.2579, "step": 3335 }, { "epoch": 3.74, "learning_rate": 8.560606060606062e-06, "loss": 6.1379, "step": 3336 }, { "epoch": 3.74, "learning_rate": 8.553030303030302e-06, "loss": 6.1123, "step": 3337 }, { "epoch": 3.74, "learning_rate": 8.545454545454546e-06, "loss": 5.9657, "step": 3338 }, { "epoch": 3.74, "learning_rate": 8.537878787878787e-06, "loss": 6.1307, "step": 3339 }, { "epoch": 3.74, "learning_rate": 8.53030303030303e-06, "loss": 6.0211, "step": 3340 }, { "epoch": 3.75, "learning_rate": 8.522727272727273e-06, "loss": 6.1154, "step": 3341 }, { "epoch": 3.75, "learning_rate": 8.515151515151515e-06, "loss": 5.9056, "step": 3342 }, { "epoch": 3.75, "learning_rate": 8.507575757575758e-06, "loss": 6.072, "step": 3343 }, { "epoch": 3.75, "learning_rate": 8.5e-06, "loss": 5.9293, "step": 3344 }, { "epoch": 3.75, "learning_rate": 8.492424242424244e-06, "loss": 6.1472, "step": 3345 }, { "epoch": 3.75, "learning_rate": 8.484848484848485e-06, "loss": 5.9286, "step": 3346 }, { "epoch": 3.75, "learning_rate": 8.477272727272729e-06, "loss": 5.9796, "step": 3347 }, { "epoch": 3.75, "learning_rate": 8.469696969696969e-06, "loss": 5.9553, "step": 3348 }, { "epoch": 3.75, "learning_rate": 8.462121212121213e-06, "loss": 5.9068, "step": 3349 }, { "epoch": 3.76, "learning_rate": 8.454545454545454e-06, "loss": 6.0022, "step": 3350 }, { "epoch": 3.76, "learning_rate": 8.446969696969698e-06, "loss": 6.002, "step": 3351 }, { "epoch": 3.76, "learning_rate": 8.439393939393938e-06, "loss": 6.0681, "step": 3352 }, { "epoch": 3.76, "learning_rate": 8.431818181818182e-06, "loss": 5.9636, "step": 3353 }, { "epoch": 3.76, "learning_rate": 8.424242424242425e-06, "loss": 6.1638, "step": 3354 }, { "epoch": 3.76, "learning_rate": 8.416666666666667e-06, "loss": 6.0777, "step": 3355 }, { "epoch": 3.76, "learning_rate": 8.409090909090909e-06, "loss": 6.0215, "step": 3356 }, { "epoch": 3.76, "learning_rate": 8.401515151515151e-06, "loss": 6.032, "step": 3357 }, { "epoch": 3.76, "learning_rate": 8.393939393939395e-06, "loss": 5.9103, "step": 3358 }, { "epoch": 3.77, "learning_rate": 8.386363636363636e-06, "loss": 5.9178, "step": 3359 }, { "epoch": 3.77, "learning_rate": 8.37878787878788e-06, "loss": 5.8993, "step": 3360 }, { "epoch": 3.77, "learning_rate": 8.37121212121212e-06, "loss": 5.929, "step": 3361 }, { "epoch": 3.77, "learning_rate": 8.363636363636365e-06, "loss": 5.927, "step": 3362 }, { "epoch": 3.77, "learning_rate": 8.356060606060605e-06, "loss": 5.8923, "step": 3363 }, { "epoch": 3.77, "learning_rate": 8.348484848484849e-06, "loss": 5.8625, "step": 3364 }, { "epoch": 3.77, "learning_rate": 8.340909090909091e-06, "loss": 5.8981, "step": 3365 }, { "epoch": 3.77, "learning_rate": 8.333333333333334e-06, "loss": 5.9284, "step": 3366 }, { "epoch": 3.77, "learning_rate": 8.325757575757576e-06, "loss": 5.8924, "step": 3367 }, { "epoch": 3.78, "learning_rate": 8.318181818181818e-06, "loss": 5.8364, "step": 3368 }, { "epoch": 3.78, "learning_rate": 8.31060606060606e-06, "loss": 5.758, "step": 3369 }, { "epoch": 3.78, "learning_rate": 8.303030303030303e-06, "loss": 5.6088, "step": 3370 }, { "epoch": 3.78, "learning_rate": 8.295454545454545e-06, "loss": 5.3329, "step": 3371 }, { "epoch": 3.78, "learning_rate": 8.287878787878787e-06, "loss": 4.9342, "step": 3372 }, { "epoch": 3.78, "learning_rate": 8.28030303030303e-06, "loss": 4.9826, "step": 3373 }, { "epoch": 3.78, "learning_rate": 8.272727272727274e-06, "loss": 4.6738, "step": 3374 }, { "epoch": 3.78, "learning_rate": 8.265151515151516e-06, "loss": 4.5348, "step": 3375 }, { "epoch": 3.78, "learning_rate": 8.257575757575758e-06, "loss": 3.8387, "step": 3376 }, { "epoch": 3.79, "learning_rate": 8.25e-06, "loss": 6.2261, "step": 3377 }, { "epoch": 3.79, "learning_rate": 8.242424242424243e-06, "loss": 6.3137, "step": 3378 }, { "epoch": 3.79, "learning_rate": 8.234848484848485e-06, "loss": 6.3197, "step": 3379 }, { "epoch": 3.79, "learning_rate": 8.227272727272728e-06, "loss": 6.4337, "step": 3380 }, { "epoch": 3.79, "learning_rate": 8.21969696969697e-06, "loss": 6.1621, "step": 3381 }, { "epoch": 3.79, "learning_rate": 8.212121212121212e-06, "loss": 6.1759, "step": 3382 }, { "epoch": 3.79, "learning_rate": 8.204545454545454e-06, "loss": 6.285, "step": 3383 }, { "epoch": 3.79, "learning_rate": 8.196969696969697e-06, "loss": 6.0709, "step": 3384 }, { "epoch": 3.79, "learning_rate": 8.18939393939394e-06, "loss": 6.2609, "step": 3385 }, { "epoch": 3.8, "learning_rate": 8.181818181818181e-06, "loss": 6.0927, "step": 3386 }, { "epoch": 3.8, "learning_rate": 8.174242424242425e-06, "loss": 6.0236, "step": 3387 }, { "epoch": 3.8, "learning_rate": 8.166666666666666e-06, "loss": 6.1494, "step": 3388 }, { "epoch": 3.8, "learning_rate": 8.15909090909091e-06, "loss": 5.9113, "step": 3389 }, { "epoch": 3.8, "learning_rate": 8.15151515151515e-06, "loss": 5.9526, "step": 3390 }, { "epoch": 3.8, "learning_rate": 8.143939393939394e-06, "loss": 6.0984, "step": 3391 }, { "epoch": 3.8, "learning_rate": 8.136363636363637e-06, "loss": 6.0633, "step": 3392 }, { "epoch": 3.8, "learning_rate": 8.128787878787879e-06, "loss": 5.9816, "step": 3393 }, { "epoch": 3.8, "learning_rate": 8.121212121212121e-06, "loss": 5.8659, "step": 3394 }, { "epoch": 3.81, "learning_rate": 8.113636363636364e-06, "loss": 5.9684, "step": 3395 }, { "epoch": 3.81, "learning_rate": 8.106060606060608e-06, "loss": 6.1124, "step": 3396 }, { "epoch": 3.81, "learning_rate": 8.098484848484848e-06, "loss": 6.1247, "step": 3397 }, { "epoch": 3.81, "learning_rate": 8.090909090909092e-06, "loss": 5.9349, "step": 3398 }, { "epoch": 3.81, "learning_rate": 8.083333333333333e-06, "loss": 5.959, "step": 3399 }, { "epoch": 3.81, "learning_rate": 8.075757575757577e-06, "loss": 5.9369, "step": 3400 }, { "epoch": 3.81, "learning_rate": 8.068181818181817e-06, "loss": 6.1117, "step": 3401 }, { "epoch": 3.81, "learning_rate": 8.060606060606061e-06, "loss": 5.9074, "step": 3402 }, { "epoch": 3.82, "learning_rate": 8.053030303030302e-06, "loss": 6.0317, "step": 3403 }, { "epoch": 3.82, "learning_rate": 8.045454545454546e-06, "loss": 5.837, "step": 3404 }, { "epoch": 3.82, "learning_rate": 8.037878787878788e-06, "loss": 5.8569, "step": 3405 }, { "epoch": 3.82, "learning_rate": 8.03030303030303e-06, "loss": 5.9651, "step": 3406 }, { "epoch": 3.82, "learning_rate": 8.022727272727273e-06, "loss": 5.989, "step": 3407 }, { "epoch": 3.82, "learning_rate": 8.015151515151515e-06, "loss": 5.9735, "step": 3408 }, { "epoch": 3.82, "learning_rate": 8.007575757575759e-06, "loss": 6.0281, "step": 3409 }, { "epoch": 3.82, "learning_rate": 8e-06, "loss": 5.9034, "step": 3410 }, { "epoch": 3.82, "learning_rate": 7.992424242424244e-06, "loss": 5.663, "step": 3411 }, { "epoch": 3.83, "learning_rate": 7.984848484848484e-06, "loss": 5.8484, "step": 3412 }, { "epoch": 3.83, "learning_rate": 7.977272727272728e-06, "loss": 6.0519, "step": 3413 }, { "epoch": 3.83, "learning_rate": 7.969696969696969e-06, "loss": 5.9932, "step": 3414 }, { "epoch": 3.83, "learning_rate": 7.962121212121213e-06, "loss": 5.6767, "step": 3415 }, { "epoch": 3.83, "learning_rate": 7.954545454545455e-06, "loss": 5.8676, "step": 3416 }, { "epoch": 3.83, "learning_rate": 7.946969696969697e-06, "loss": 5.8001, "step": 3417 }, { "epoch": 3.83, "learning_rate": 7.93939393939394e-06, "loss": 5.7278, "step": 3418 }, { "epoch": 3.83, "learning_rate": 7.931818181818182e-06, "loss": 5.549, "step": 3419 }, { "epoch": 3.83, "learning_rate": 7.924242424242424e-06, "loss": 5.717, "step": 3420 }, { "epoch": 3.84, "learning_rate": 7.916666666666667e-06, "loss": 5.4291, "step": 3421 }, { "epoch": 3.84, "learning_rate": 7.909090909090909e-06, "loss": 5.362, "step": 3422 }, { "epoch": 3.84, "learning_rate": 7.901515151515151e-06, "loss": 5.0988, "step": 3423 }, { "epoch": 3.84, "learning_rate": 7.893939393939393e-06, "loss": 5.0448, "step": 3424 }, { "epoch": 3.84, "learning_rate": 7.886363636363637e-06, "loss": 4.5766, "step": 3425 }, { "epoch": 3.84, "learning_rate": 7.87878787878788e-06, "loss": 3.7608, "step": 3426 }, { "epoch": 3.84, "learning_rate": 7.871212121212122e-06, "loss": 6.2875, "step": 3427 }, { "epoch": 3.84, "learning_rate": 7.863636363636364e-06, "loss": 6.2032, "step": 3428 }, { "epoch": 3.84, "learning_rate": 7.856060606060607e-06, "loss": 6.1482, "step": 3429 }, { "epoch": 3.85, "learning_rate": 7.848484848484849e-06, "loss": 6.3295, "step": 3430 }, { "epoch": 3.85, "learning_rate": 7.840909090909091e-06, "loss": 6.0971, "step": 3431 }, { "epoch": 3.85, "learning_rate": 7.833333333333333e-06, "loss": 6.2574, "step": 3432 }, { "epoch": 3.85, "learning_rate": 7.825757575757576e-06, "loss": 6.1002, "step": 3433 }, { "epoch": 3.85, "learning_rate": 7.818181818181818e-06, "loss": 6.2944, "step": 3434 }, { "epoch": 3.85, "learning_rate": 7.81060606060606e-06, "loss": 6.0351, "step": 3435 }, { "epoch": 3.85, "learning_rate": 7.803030303030304e-06, "loss": 6.0179, "step": 3436 }, { "epoch": 3.85, "learning_rate": 7.795454545454545e-06, "loss": 6.0207, "step": 3437 }, { "epoch": 3.85, "learning_rate": 7.787878787878789e-06, "loss": 6.0006, "step": 3438 }, { "epoch": 3.86, "learning_rate": 7.78030303030303e-06, "loss": 6.0821, "step": 3439 }, { "epoch": 3.86, "learning_rate": 7.772727272727273e-06, "loss": 6.1248, "step": 3440 }, { "epoch": 3.86, "learning_rate": 7.765151515151514e-06, "loss": 5.9488, "step": 3441 }, { "epoch": 3.86, "learning_rate": 7.757575757575758e-06, "loss": 6.023, "step": 3442 }, { "epoch": 3.86, "learning_rate": 7.75e-06, "loss": 6.0645, "step": 3443 }, { "epoch": 3.86, "learning_rate": 7.742424242424243e-06, "loss": 5.9618, "step": 3444 }, { "epoch": 3.86, "learning_rate": 7.734848484848485e-06, "loss": 5.9519, "step": 3445 }, { "epoch": 3.86, "learning_rate": 7.727272727272727e-06, "loss": 6.028, "step": 3446 }, { "epoch": 3.86, "learning_rate": 7.719696969696971e-06, "loss": 6.0897, "step": 3447 }, { "epoch": 3.87, "learning_rate": 7.712121212121212e-06, "loss": 5.9684, "step": 3448 }, { "epoch": 3.87, "learning_rate": 7.704545454545456e-06, "loss": 6.0151, "step": 3449 }, { "epoch": 3.87, "learning_rate": 7.696969696969696e-06, "loss": 5.928, "step": 3450 }, { "epoch": 3.87, "learning_rate": 7.68939393939394e-06, "loss": 5.9767, "step": 3451 }, { "epoch": 3.87, "learning_rate": 7.681818181818181e-06, "loss": 6.0172, "step": 3452 }, { "epoch": 3.87, "learning_rate": 7.674242424242425e-06, "loss": 5.8787, "step": 3453 }, { "epoch": 3.87, "learning_rate": 7.666666666666666e-06, "loss": 5.8308, "step": 3454 }, { "epoch": 3.87, "learning_rate": 7.65909090909091e-06, "loss": 5.8781, "step": 3455 }, { "epoch": 3.87, "learning_rate": 7.651515151515152e-06, "loss": 5.9719, "step": 3456 }, { "epoch": 3.88, "learning_rate": 7.643939393939394e-06, "loss": 5.8568, "step": 3457 }, { "epoch": 3.88, "learning_rate": 7.636363636363636e-06, "loss": 5.9317, "step": 3458 }, { "epoch": 3.88, "learning_rate": 7.628787878787879e-06, "loss": 5.8254, "step": 3459 }, { "epoch": 3.88, "learning_rate": 7.621212121212122e-06, "loss": 5.9783, "step": 3460 }, { "epoch": 3.88, "learning_rate": 7.613636363636364e-06, "loss": 5.891, "step": 3461 }, { "epoch": 3.88, "learning_rate": 7.606060606060607e-06, "loss": 5.723, "step": 3462 }, { "epoch": 3.88, "learning_rate": 7.598484848484849e-06, "loss": 5.9023, "step": 3463 }, { "epoch": 3.88, "learning_rate": 7.590909090909092e-06, "loss": 5.8852, "step": 3464 }, { "epoch": 3.88, "learning_rate": 7.583333333333333e-06, "loss": 5.7222, "step": 3465 }, { "epoch": 3.89, "learning_rate": 7.5757575757575764e-06, "loss": 5.7953, "step": 3466 }, { "epoch": 3.89, "learning_rate": 7.568181818181818e-06, "loss": 5.7369, "step": 3467 }, { "epoch": 3.89, "learning_rate": 7.560606060606061e-06, "loss": 5.7211, "step": 3468 }, { "epoch": 3.89, "learning_rate": 7.5530303030303025e-06, "loss": 5.3314, "step": 3469 }, { "epoch": 3.89, "learning_rate": 7.545454545454546e-06, "loss": 5.3624, "step": 3470 }, { "epoch": 3.89, "learning_rate": 7.537878787878788e-06, "loss": 5.4281, "step": 3471 }, { "epoch": 3.89, "learning_rate": 7.530303030303031e-06, "loss": 5.0483, "step": 3472 }, { "epoch": 3.89, "learning_rate": 7.5227272727272725e-06, "loss": 4.987, "step": 3473 }, { "epoch": 3.89, "learning_rate": 7.515151515151516e-06, "loss": 4.8409, "step": 3474 }, { "epoch": 3.9, "learning_rate": 7.507575757575757e-06, "loss": 3.9948, "step": 3475 }, { "epoch": 3.9, "learning_rate": 7.5e-06, "loss": 3.6253, "step": 3476 }, { "epoch": 3.9, "learning_rate": 7.4924242424242425e-06, "loss": 6.3815, "step": 3477 }, { "epoch": 3.9, "learning_rate": 7.484848484848485e-06, "loss": 6.192, "step": 3478 }, { "epoch": 3.9, "learning_rate": 7.477272727272727e-06, "loss": 6.1795, "step": 3479 }, { "epoch": 3.9, "learning_rate": 7.469696969696969e-06, "loss": 6.2256, "step": 3480 }, { "epoch": 3.9, "learning_rate": 7.462121212121212e-06, "loss": 6.2418, "step": 3481 }, { "epoch": 3.9, "learning_rate": 7.454545454545455e-06, "loss": 5.9793, "step": 3482 }, { "epoch": 3.9, "learning_rate": 7.446969696969697e-06, "loss": 6.0414, "step": 3483 }, { "epoch": 3.91, "learning_rate": 7.439393939393939e-06, "loss": 6.1743, "step": 3484 }, { "epoch": 3.91, "learning_rate": 7.4318181818181825e-06, "loss": 6.132, "step": 3485 }, { "epoch": 3.91, "learning_rate": 7.424242424242425e-06, "loss": 6.0686, "step": 3486 }, { "epoch": 3.91, "learning_rate": 7.416666666666667e-06, "loss": 6.0082, "step": 3487 }, { "epoch": 3.91, "learning_rate": 7.409090909090909e-06, "loss": 6.0215, "step": 3488 }, { "epoch": 3.91, "learning_rate": 7.401515151515152e-06, "loss": 6.0033, "step": 3489 }, { "epoch": 3.91, "learning_rate": 7.393939393939394e-06, "loss": 6.1981, "step": 3490 }, { "epoch": 3.91, "learning_rate": 7.386363636363636e-06, "loss": 6.0291, "step": 3491 }, { "epoch": 3.91, "learning_rate": 7.378787878787879e-06, "loss": 6.0074, "step": 3492 }, { "epoch": 3.92, "learning_rate": 7.371212121212122e-06, "loss": 6.144, "step": 3493 }, { "epoch": 3.92, "learning_rate": 7.363636363636364e-06, "loss": 5.9657, "step": 3494 }, { "epoch": 3.92, "learning_rate": 7.356060606060606e-06, "loss": 6.198, "step": 3495 }, { "epoch": 3.92, "learning_rate": 7.3484848484848486e-06, "loss": 5.918, "step": 3496 }, { "epoch": 3.92, "learning_rate": 7.340909090909091e-06, "loss": 5.9337, "step": 3497 }, { "epoch": 3.92, "learning_rate": 7.333333333333333e-06, "loss": 5.9963, "step": 3498 }, { "epoch": 3.92, "learning_rate": 7.3257575757575754e-06, "loss": 5.9742, "step": 3499 }, { "epoch": 3.92, "learning_rate": 7.318181818181818e-06, "loss": 6.0851, "step": 3500 }, { "epoch": 3.92, "eval_loss": 5.663444995880127, "eval_runtime": 837.6848, "eval_samples_per_second": 3.154, "eval_steps_per_second": 0.395, "eval_wer": 1.0943673145577153, "step": 3500 }, { "epoch": 3.92, "learning_rate": 7.31060606060606e-06, "loss": 6.1085, "step": 3501 }, { "epoch": 3.93, "learning_rate": 7.303030303030304e-06, "loss": 6.1166, "step": 3502 }, { "epoch": 3.93, "learning_rate": 7.295454545454546e-06, "loss": 5.9004, "step": 3503 }, { "epoch": 3.93, "learning_rate": 7.287878787878789e-06, "loss": 5.9489, "step": 3504 }, { "epoch": 3.93, "learning_rate": 7.280303030303031e-06, "loss": 6.0072, "step": 3505 }, { "epoch": 3.93, "learning_rate": 7.272727272727273e-06, "loss": 5.9029, "step": 3506 }, { "epoch": 3.93, "learning_rate": 7.2651515151515155e-06, "loss": 5.8868, "step": 3507 }, { "epoch": 3.93, "learning_rate": 7.257575757575758e-06, "loss": 5.879, "step": 3508 }, { "epoch": 3.93, "learning_rate": 7.25e-06, "loss": 5.81, "step": 3509 }, { "epoch": 3.93, "learning_rate": 7.242424242424242e-06, "loss": 5.8775, "step": 3510 }, { "epoch": 3.94, "learning_rate": 7.234848484848485e-06, "loss": 5.9748, "step": 3511 }, { "epoch": 3.94, "learning_rate": 7.227272727272728e-06, "loss": 5.8427, "step": 3512 }, { "epoch": 3.94, "learning_rate": 7.21969696969697e-06, "loss": 5.887, "step": 3513 }, { "epoch": 3.94, "learning_rate": 7.212121212121212e-06, "loss": 5.7038, "step": 3514 }, { "epoch": 3.94, "learning_rate": 7.204545454545455e-06, "loss": 5.9018, "step": 3515 }, { "epoch": 3.94, "learning_rate": 7.196969696969697e-06, "loss": 5.8585, "step": 3516 }, { "epoch": 3.94, "learning_rate": 7.189393939393939e-06, "loss": 5.6106, "step": 3517 }, { "epoch": 3.94, "learning_rate": 7.1818181818181815e-06, "loss": 5.8303, "step": 3518 }, { "epoch": 3.95, "learning_rate": 7.174242424242425e-06, "loss": 5.5915, "step": 3519 }, { "epoch": 3.95, "learning_rate": 7.166666666666667e-06, "loss": 5.4184, "step": 3520 }, { "epoch": 3.95, "learning_rate": 7.159090909090909e-06, "loss": 5.3998, "step": 3521 }, { "epoch": 3.95, "learning_rate": 7.151515151515152e-06, "loss": 5.4068, "step": 3522 }, { "epoch": 3.95, "learning_rate": 7.143939393939395e-06, "loss": 5.0318, "step": 3523 }, { "epoch": 3.95, "learning_rate": 7.136363636363637e-06, "loss": 4.8644, "step": 3524 }, { "epoch": 3.95, "learning_rate": 7.128787878787879e-06, "loss": 4.2077, "step": 3525 }, { "epoch": 3.95, "learning_rate": 7.1212121212121215e-06, "loss": 4.0022, "step": 3526 }, { "epoch": 3.95, "learning_rate": 7.113636363636364e-06, "loss": 6.3453, "step": 3527 }, { "epoch": 3.96, "learning_rate": 7.106060606060606e-06, "loss": 6.142, "step": 3528 }, { "epoch": 3.96, "learning_rate": 7.098484848484848e-06, "loss": 6.1543, "step": 3529 }, { "epoch": 3.96, "learning_rate": 7.090909090909091e-06, "loss": 6.289, "step": 3530 }, { "epoch": 3.96, "learning_rate": 7.083333333333333e-06, "loss": 6.15, "step": 3531 }, { "epoch": 3.96, "learning_rate": 7.075757575757575e-06, "loss": 6.17, "step": 3532 }, { "epoch": 3.96, "learning_rate": 7.0681818181818184e-06, "loss": 6.1746, "step": 3533 }, { "epoch": 3.96, "learning_rate": 7.060606060606061e-06, "loss": 6.12, "step": 3534 }, { "epoch": 3.96, "learning_rate": 7.053030303030303e-06, "loss": 6.0361, "step": 3535 }, { "epoch": 3.96, "learning_rate": 7.045454545454546e-06, "loss": 6.0805, "step": 3536 }, { "epoch": 3.97, "learning_rate": 7.0378787878787884e-06, "loss": 6.0783, "step": 3537 }, { "epoch": 3.97, "learning_rate": 7.030303030303031e-06, "loss": 6.0286, "step": 3538 }, { "epoch": 3.97, "learning_rate": 7.022727272727273e-06, "loss": 5.9548, "step": 3539 }, { "epoch": 3.97, "learning_rate": 7.015151515151515e-06, "loss": 6.027, "step": 3540 }, { "epoch": 3.97, "learning_rate": 7.007575757575758e-06, "loss": 5.9973, "step": 3541 }, { "epoch": 3.97, "learning_rate": 7e-06, "loss": 6.0594, "step": 3542 }, { "epoch": 3.97, "learning_rate": 6.992424242424243e-06, "loss": 6.0276, "step": 3543 }, { "epoch": 3.97, "learning_rate": 6.984848484848485e-06, "loss": 6.1397, "step": 3544 }, { "epoch": 3.97, "learning_rate": 6.977272727272728e-06, "loss": 6.02, "step": 3545 }, { "epoch": 3.98, "learning_rate": 6.96969696969697e-06, "loss": 6.0413, "step": 3546 }, { "epoch": 3.98, "learning_rate": 6.962121212121212e-06, "loss": 5.9935, "step": 3547 }, { "epoch": 3.98, "learning_rate": 6.9545454545454545e-06, "loss": 5.9335, "step": 3548 }, { "epoch": 3.98, "learning_rate": 6.946969696969697e-06, "loss": 6.0299, "step": 3549 }, { "epoch": 3.98, "learning_rate": 6.939393939393939e-06, "loss": 5.8552, "step": 3550 }, { "epoch": 3.98, "learning_rate": 6.931818181818181e-06, "loss": 6.1905, "step": 3551 }, { "epoch": 3.98, "learning_rate": 6.924242424242424e-06, "loss": 6.0016, "step": 3552 }, { "epoch": 3.98, "learning_rate": 6.916666666666668e-06, "loss": 5.89, "step": 3553 }, { "epoch": 3.98, "learning_rate": 6.90909090909091e-06, "loss": 5.8131, "step": 3554 }, { "epoch": 3.99, "learning_rate": 6.901515151515152e-06, "loss": 6.087, "step": 3555 }, { "epoch": 3.99, "learning_rate": 6.8939393939393945e-06, "loss": 5.9167, "step": 3556 }, { "epoch": 3.99, "learning_rate": 6.886363636363637e-06, "loss": 5.9302, "step": 3557 }, { "epoch": 3.99, "learning_rate": 6.878787878787879e-06, "loss": 5.9125, "step": 3558 }, { "epoch": 3.99, "learning_rate": 6.871212121212121e-06, "loss": 5.7922, "step": 3559 }, { "epoch": 3.99, "learning_rate": 6.863636363636364e-06, "loss": 5.8991, "step": 3560 }, { "epoch": 3.99, "learning_rate": 6.856060606060606e-06, "loss": 5.6873, "step": 3561 }, { "epoch": 3.99, "learning_rate": 6.848484848484848e-06, "loss": 5.7785, "step": 3562 }, { "epoch": 3.99, "learning_rate": 6.840909090909091e-06, "loss": 5.432, "step": 3563 }, { "epoch": 4.0, "learning_rate": 6.833333333333334e-06, "loss": 5.3162, "step": 3564 }, { "epoch": 4.0, "learning_rate": 6.825757575757576e-06, "loss": 4.9657, "step": 3565 }, { "epoch": 4.0, "learning_rate": 6.818181818181818e-06, "loss": 4.5967, "step": 3566 }, { "epoch": 4.0, "learning_rate": 6.8106060606060606e-06, "loss": 4.5212, "step": 3567 }, { "epoch": 4.0, "learning_rate": 6.803030303030303e-06, "loss": 4.0882, "step": 3568 }, { "epoch": 4.0, "learning_rate": 6.795454545454545e-06, "loss": 6.3473, "step": 3569 }, { "epoch": 4.0, "learning_rate": 6.787878787878788e-06, "loss": 6.2654, "step": 3570 }, { "epoch": 4.0, "learning_rate": 6.780303030303031e-06, "loss": 6.1246, "step": 3571 }, { "epoch": 4.0, "learning_rate": 6.772727272727273e-06, "loss": 6.258, "step": 3572 }, { "epoch": 4.01, "learning_rate": 6.765151515151515e-06, "loss": 6.1418, "step": 3573 }, { "epoch": 4.01, "learning_rate": 6.757575757575758e-06, "loss": 6.1822, "step": 3574 }, { "epoch": 4.01, "learning_rate": 6.750000000000001e-06, "loss": 6.1613, "step": 3575 }, { "epoch": 4.01, "learning_rate": 6.742424242424243e-06, "loss": 6.1422, "step": 3576 }, { "epoch": 4.01, "learning_rate": 6.734848484848485e-06, "loss": 6.1248, "step": 3577 }, { "epoch": 4.01, "learning_rate": 6.7272727272727275e-06, "loss": 6.0361, "step": 3578 }, { "epoch": 4.01, "learning_rate": 6.71969696969697e-06, "loss": 5.9663, "step": 3579 }, { "epoch": 4.01, "learning_rate": 6.712121212121212e-06, "loss": 5.9254, "step": 3580 }, { "epoch": 4.01, "learning_rate": 6.704545454545454e-06, "loss": 5.9715, "step": 3581 }, { "epoch": 4.02, "learning_rate": 6.696969696969697e-06, "loss": 5.9497, "step": 3582 }, { "epoch": 4.02, "learning_rate": 6.689393939393939e-06, "loss": 6.0343, "step": 3583 }, { "epoch": 4.02, "learning_rate": 6.681818181818182e-06, "loss": 6.1459, "step": 3584 }, { "epoch": 4.02, "learning_rate": 6.674242424242424e-06, "loss": 6.0018, "step": 3585 }, { "epoch": 4.02, "learning_rate": 6.666666666666667e-06, "loss": 5.9507, "step": 3586 }, { "epoch": 4.02, "learning_rate": 6.65909090909091e-06, "loss": 6.0171, "step": 3587 }, { "epoch": 4.02, "learning_rate": 6.651515151515152e-06, "loss": 5.9358, "step": 3588 }, { "epoch": 4.02, "learning_rate": 6.643939393939394e-06, "loss": 5.9271, "step": 3589 }, { "epoch": 4.02, "learning_rate": 6.636363636363637e-06, "loss": 6.0003, "step": 3590 }, { "epoch": 4.03, "learning_rate": 6.628787878787879e-06, "loss": 6.1401, "step": 3591 }, { "epoch": 4.03, "learning_rate": 6.621212121212121e-06, "loss": 5.9403, "step": 3592 }, { "epoch": 4.03, "learning_rate": 6.6136363636363635e-06, "loss": 5.8425, "step": 3593 }, { "epoch": 4.03, "learning_rate": 6.606060606060607e-06, "loss": 5.8615, "step": 3594 }, { "epoch": 4.03, "learning_rate": 6.598484848484849e-06, "loss": 5.9871, "step": 3595 }, { "epoch": 4.03, "learning_rate": 6.590909090909091e-06, "loss": 5.7851, "step": 3596 }, { "epoch": 4.03, "learning_rate": 6.5833333333333335e-06, "loss": 6.119, "step": 3597 }, { "epoch": 4.03, "learning_rate": 6.575757575757576e-06, "loss": 5.9322, "step": 3598 }, { "epoch": 4.03, "learning_rate": 6.568181818181818e-06, "loss": 5.978, "step": 3599 }, { "epoch": 4.04, "learning_rate": 6.56060606060606e-06, "loss": 5.9748, "step": 3600 }, { "epoch": 4.04, "learning_rate": 6.553030303030303e-06, "loss": 5.7632, "step": 3601 }, { "epoch": 4.04, "learning_rate": 6.545454545454545e-06, "loss": 5.8264, "step": 3602 }, { "epoch": 4.04, "learning_rate": 6.537878787878787e-06, "loss": 5.8639, "step": 3603 }, { "epoch": 4.04, "learning_rate": 6.5303030303030304e-06, "loss": 5.8214, "step": 3604 }, { "epoch": 4.04, "learning_rate": 6.5227272727272736e-06, "loss": 5.7623, "step": 3605 }, { "epoch": 4.04, "learning_rate": 6.515151515151516e-06, "loss": 5.9149, "step": 3606 }, { "epoch": 4.04, "learning_rate": 6.507575757575758e-06, "loss": 5.7844, "step": 3607 }, { "epoch": 4.04, "learning_rate": 6.5000000000000004e-06, "loss": 5.642, "step": 3608 }, { "epoch": 4.05, "learning_rate": 6.492424242424243e-06, "loss": 5.6442, "step": 3609 }, { "epoch": 4.05, "learning_rate": 6.484848484848485e-06, "loss": 5.7908, "step": 3610 }, { "epoch": 4.05, "learning_rate": 6.477272727272727e-06, "loss": 5.5181, "step": 3611 }, { "epoch": 4.05, "learning_rate": 6.46969696969697e-06, "loss": 5.4017, "step": 3612 }, { "epoch": 4.05, "learning_rate": 6.462121212121212e-06, "loss": 5.2099, "step": 3613 }, { "epoch": 4.05, "learning_rate": 6.454545454545455e-06, "loss": 5.2354, "step": 3614 }, { "epoch": 4.05, "learning_rate": 6.446969696969697e-06, "loss": 4.8586, "step": 3615 }, { "epoch": 4.05, "learning_rate": 6.43939393939394e-06, "loss": 4.66, "step": 3616 }, { "epoch": 4.05, "learning_rate": 6.431818181818182e-06, "loss": 4.7236, "step": 3617 }, { "epoch": 4.06, "learning_rate": 6.424242424242424e-06, "loss": 3.4448, "step": 3618 }, { "epoch": 4.06, "learning_rate": 6.4166666666666665e-06, "loss": 6.222, "step": 3619 }, { "epoch": 4.06, "learning_rate": 6.409090909090909e-06, "loss": 6.1911, "step": 3620 }, { "epoch": 4.06, "learning_rate": 6.401515151515151e-06, "loss": 6.1733, "step": 3621 }, { "epoch": 4.06, "learning_rate": 6.393939393939394e-06, "loss": 6.1485, "step": 3622 }, { "epoch": 4.06, "learning_rate": 6.3863636363636365e-06, "loss": 6.1566, "step": 3623 }, { "epoch": 4.06, "learning_rate": 6.378787878787879e-06, "loss": 6.2254, "step": 3624 }, { "epoch": 4.06, "learning_rate": 6.371212121212122e-06, "loss": 6.1799, "step": 3625 }, { "epoch": 4.07, "learning_rate": 6.363636363636364e-06, "loss": 6.1105, "step": 3626 }, { "epoch": 4.07, "learning_rate": 6.3560606060606065e-06, "loss": 6.0448, "step": 3627 }, { "epoch": 4.07, "learning_rate": 6.348484848484849e-06, "loss": 6.0516, "step": 3628 }, { "epoch": 4.07, "learning_rate": 6.340909090909091e-06, "loss": 6.0081, "step": 3629 }, { "epoch": 4.07, "learning_rate": 6.333333333333333e-06, "loss": 5.9957, "step": 3630 }, { "epoch": 4.07, "learning_rate": 6.325757575757576e-06, "loss": 5.9397, "step": 3631 }, { "epoch": 4.07, "learning_rate": 6.318181818181818e-06, "loss": 6.0154, "step": 3632 }, { "epoch": 4.07, "learning_rate": 6.31060606060606e-06, "loss": 6.0524, "step": 3633 }, { "epoch": 4.07, "learning_rate": 6.3030303030303026e-06, "loss": 6.0831, "step": 3634 }, { "epoch": 4.08, "learning_rate": 6.295454545454546e-06, "loss": 5.9741, "step": 3635 }, { "epoch": 4.08, "learning_rate": 6.287878787878788e-06, "loss": 5.8778, "step": 3636 }, { "epoch": 4.08, "learning_rate": 6.28030303030303e-06, "loss": 5.752, "step": 3637 }, { "epoch": 4.08, "learning_rate": 6.272727272727273e-06, "loss": 5.9756, "step": 3638 }, { "epoch": 4.08, "learning_rate": 6.265151515151516e-06, "loss": 5.9938, "step": 3639 }, { "epoch": 4.08, "learning_rate": 6.257575757575758e-06, "loss": 5.9178, "step": 3640 }, { "epoch": 4.08, "learning_rate": 6.25e-06, "loss": 5.8519, "step": 3641 }, { "epoch": 4.08, "learning_rate": 6.242424242424243e-06, "loss": 5.9986, "step": 3642 }, { "epoch": 4.08, "learning_rate": 6.234848484848485e-06, "loss": 5.8762, "step": 3643 }, { "epoch": 4.09, "learning_rate": 6.227272727272727e-06, "loss": 5.9681, "step": 3644 }, { "epoch": 4.09, "learning_rate": 6.21969696969697e-06, "loss": 5.8132, "step": 3645 }, { "epoch": 4.09, "learning_rate": 6.212121212121213e-06, "loss": 5.8719, "step": 3646 }, { "epoch": 4.09, "learning_rate": 6.204545454545455e-06, "loss": 5.9498, "step": 3647 }, { "epoch": 4.09, "learning_rate": 6.196969696969697e-06, "loss": 5.8171, "step": 3648 }, { "epoch": 4.09, "learning_rate": 6.1893939393939395e-06, "loss": 5.7877, "step": 3649 }, { "epoch": 4.09, "learning_rate": 6.181818181818182e-06, "loss": 5.9197, "step": 3650 }, { "epoch": 4.09, "learning_rate": 6.174242424242424e-06, "loss": 5.7186, "step": 3651 }, { "epoch": 4.09, "learning_rate": 6.166666666666666e-06, "loss": 5.9149, "step": 3652 }, { "epoch": 4.1, "learning_rate": 6.159090909090909e-06, "loss": 5.7665, "step": 3653 }, { "epoch": 4.1, "learning_rate": 6.151515151515151e-06, "loss": 5.8835, "step": 3654 }, { "epoch": 4.1, "learning_rate": 6.143939393939394e-06, "loss": 5.9414, "step": 3655 }, { "epoch": 4.1, "learning_rate": 6.136363636363637e-06, "loss": 5.835, "step": 3656 }, { "epoch": 4.1, "learning_rate": 6.1287878787878795e-06, "loss": 5.7599, "step": 3657 }, { "epoch": 4.1, "learning_rate": 6.121212121212122e-06, "loss": 5.7924, "step": 3658 }, { "epoch": 4.1, "learning_rate": 6.113636363636364e-06, "loss": 5.6162, "step": 3659 }, { "epoch": 4.1, "learning_rate": 6.106060606060606e-06, "loss": 5.6177, "step": 3660 }, { "epoch": 4.1, "learning_rate": 6.098484848484849e-06, "loss": 5.3788, "step": 3661 }, { "epoch": 4.11, "learning_rate": 6.090909090909091e-06, "loss": 5.3297, "step": 3662 }, { "epoch": 4.11, "learning_rate": 6.083333333333333e-06, "loss": 5.2662, "step": 3663 }, { "epoch": 4.11, "learning_rate": 6.0757575757575755e-06, "loss": 5.0729, "step": 3664 }, { "epoch": 4.11, "learning_rate": 6.068181818181819e-06, "loss": 4.7153, "step": 3665 }, { "epoch": 4.11, "learning_rate": 6.060606060606061e-06, "loss": 4.3963, "step": 3666 }, { "epoch": 4.11, "learning_rate": 6.053030303030303e-06, "loss": 4.5125, "step": 3667 }, { "epoch": 4.11, "learning_rate": 6.0454545454545456e-06, "loss": 3.4096, "step": 3668 }, { "epoch": 4.11, "learning_rate": 6.037878787878788e-06, "loss": 6.219, "step": 3669 }, { "epoch": 4.11, "learning_rate": 6.03030303030303e-06, "loss": 6.2165, "step": 3670 }, { "epoch": 4.12, "learning_rate": 6.0227272727272724e-06, "loss": 6.2158, "step": 3671 }, { "epoch": 4.12, "learning_rate": 6.015151515151515e-06, "loss": 6.2219, "step": 3672 }, { "epoch": 4.12, "learning_rate": 6.007575757575758e-06, "loss": 5.9513, "step": 3673 }, { "epoch": 4.12, "learning_rate": 6e-06, "loss": 6.0967, "step": 3674 }, { "epoch": 4.12, "learning_rate": 5.9924242424242424e-06, "loss": 6.2341, "step": 3675 }, { "epoch": 4.12, "learning_rate": 5.984848484848486e-06, "loss": 6.0827, "step": 3676 }, { "epoch": 4.12, "learning_rate": 5.977272727272728e-06, "loss": 6.0541, "step": 3677 }, { "epoch": 4.12, "learning_rate": 5.96969696969697e-06, "loss": 5.9176, "step": 3678 }, { "epoch": 4.12, "learning_rate": 5.9621212121212125e-06, "loss": 6.064, "step": 3679 }, { "epoch": 4.13, "learning_rate": 5.954545454545455e-06, "loss": 5.8664, "step": 3680 }, { "epoch": 4.13, "learning_rate": 5.946969696969697e-06, "loss": 5.9766, "step": 3681 }, { "epoch": 4.13, "learning_rate": 5.939393939393939e-06, "loss": 5.8698, "step": 3682 }, { "epoch": 4.13, "learning_rate": 5.931818181818182e-06, "loss": 6.0524, "step": 3683 }, { "epoch": 4.13, "learning_rate": 5.924242424242424e-06, "loss": 5.8664, "step": 3684 }, { "epoch": 4.13, "learning_rate": 5.916666666666666e-06, "loss": 5.7876, "step": 3685 }, { "epoch": 4.13, "learning_rate": 5.909090909090909e-06, "loss": 6.0159, "step": 3686 }, { "epoch": 4.13, "learning_rate": 5.901515151515152e-06, "loss": 5.9626, "step": 3687 }, { "epoch": 4.13, "learning_rate": 5.893939393939394e-06, "loss": 5.8817, "step": 3688 }, { "epoch": 4.14, "learning_rate": 5.886363636363636e-06, "loss": 5.9509, "step": 3689 }, { "epoch": 4.14, "learning_rate": 5.878787878787879e-06, "loss": 5.883, "step": 3690 }, { "epoch": 4.14, "learning_rate": 5.871212121212122e-06, "loss": 5.9617, "step": 3691 }, { "epoch": 4.14, "learning_rate": 5.863636363636364e-06, "loss": 5.7139, "step": 3692 }, { "epoch": 4.14, "learning_rate": 5.856060606060606e-06, "loss": 5.8619, "step": 3693 }, { "epoch": 4.14, "learning_rate": 5.8484848484848485e-06, "loss": 5.7971, "step": 3694 }, { "epoch": 4.14, "learning_rate": 5.840909090909091e-06, "loss": 5.8234, "step": 3695 }, { "epoch": 4.14, "learning_rate": 5.833333333333334e-06, "loss": 5.8033, "step": 3696 }, { "epoch": 4.14, "learning_rate": 5.825757575757576e-06, "loss": 5.8825, "step": 3697 }, { "epoch": 4.15, "learning_rate": 5.8181818181818185e-06, "loss": 6.1077, "step": 3698 }, { "epoch": 4.15, "learning_rate": 5.810606060606061e-06, "loss": 5.8805, "step": 3699 }, { "epoch": 4.15, "learning_rate": 5.803030303030303e-06, "loss": 5.9134, "step": 3700 }, { "epoch": 4.15, "learning_rate": 5.795454545454545e-06, "loss": 5.8988, "step": 3701 }, { "epoch": 4.15, "learning_rate": 5.787878787878788e-06, "loss": 5.7953, "step": 3702 }, { "epoch": 4.15, "learning_rate": 5.78030303030303e-06, "loss": 5.7571, "step": 3703 }, { "epoch": 4.15, "learning_rate": 5.772727272727272e-06, "loss": 5.7925, "step": 3704 }, { "epoch": 4.15, "learning_rate": 5.7651515151515146e-06, "loss": 5.9651, "step": 3705 }, { "epoch": 4.15, "learning_rate": 5.757575757575758e-06, "loss": 5.849, "step": 3706 }, { "epoch": 4.16, "learning_rate": 5.750000000000001e-06, "loss": 5.9224, "step": 3707 }, { "epoch": 4.16, "learning_rate": 5.742424242424243e-06, "loss": 5.7967, "step": 3708 }, { "epoch": 4.16, "learning_rate": 5.7348484848484854e-06, "loss": 5.649, "step": 3709 }, { "epoch": 4.16, "learning_rate": 5.727272727272728e-06, "loss": 5.7653, "step": 3710 }, { "epoch": 4.16, "learning_rate": 5.71969696969697e-06, "loss": 5.6402, "step": 3711 }, { "epoch": 4.16, "learning_rate": 5.712121212121212e-06, "loss": 5.5705, "step": 3712 }, { "epoch": 4.16, "learning_rate": 5.704545454545455e-06, "loss": 5.4154, "step": 3713 }, { "epoch": 4.16, "learning_rate": 5.696969696969697e-06, "loss": 5.0866, "step": 3714 }, { "epoch": 4.16, "learning_rate": 5.689393939393939e-06, "loss": 4.9838, "step": 3715 }, { "epoch": 4.17, "learning_rate": 5.681818181818182e-06, "loss": 4.3339, "step": 3716 }, { "epoch": 4.17, "learning_rate": 5.674242424242425e-06, "loss": 4.165, "step": 3717 }, { "epoch": 4.17, "learning_rate": 5.666666666666667e-06, "loss": 3.6336, "step": 3718 }, { "epoch": 4.17, "learning_rate": 5.659090909090909e-06, "loss": 6.3108, "step": 3719 }, { "epoch": 4.17, "learning_rate": 5.6515151515151515e-06, "loss": 6.2601, "step": 3720 }, { "epoch": 4.17, "learning_rate": 5.643939393939394e-06, "loss": 6.2083, "step": 3721 }, { "epoch": 4.17, "learning_rate": 5.636363636363636e-06, "loss": 6.3002, "step": 3722 }, { "epoch": 4.17, "learning_rate": 5.628787878787878e-06, "loss": 6.2639, "step": 3723 }, { "epoch": 4.17, "learning_rate": 5.6212121212121215e-06, "loss": 6.165, "step": 3724 }, { "epoch": 4.18, "learning_rate": 5.613636363636364e-06, "loss": 6.1525, "step": 3725 }, { "epoch": 4.18, "learning_rate": 5.606060606060606e-06, "loss": 6.139, "step": 3726 }, { "epoch": 4.18, "learning_rate": 5.598484848484849e-06, "loss": 6.0458, "step": 3727 }, { "epoch": 4.18, "learning_rate": 5.5909090909090915e-06, "loss": 6.1306, "step": 3728 }, { "epoch": 4.18, "learning_rate": 5.583333333333334e-06, "loss": 5.9946, "step": 3729 }, { "epoch": 4.18, "learning_rate": 5.575757575757576e-06, "loss": 5.9834, "step": 3730 }, { "epoch": 4.18, "learning_rate": 5.568181818181818e-06, "loss": 5.9739, "step": 3731 }, { "epoch": 4.18, "learning_rate": 5.560606060606061e-06, "loss": 5.9836, "step": 3732 }, { "epoch": 4.18, "learning_rate": 5.553030303030303e-06, "loss": 5.9514, "step": 3733 }, { "epoch": 4.19, "learning_rate": 5.545454545454545e-06, "loss": 5.9728, "step": 3734 }, { "epoch": 4.19, "learning_rate": 5.5378787878787876e-06, "loss": 5.9318, "step": 3735 }, { "epoch": 4.19, "learning_rate": 5.53030303030303e-06, "loss": 5.758, "step": 3736 }, { "epoch": 4.19, "learning_rate": 5.522727272727273e-06, "loss": 5.9901, "step": 3737 }, { "epoch": 4.19, "learning_rate": 5.515151515151515e-06, "loss": 5.9045, "step": 3738 }, { "epoch": 4.19, "learning_rate": 5.5075757575757576e-06, "loss": 6.0238, "step": 3739 }, { "epoch": 4.19, "learning_rate": 5.5e-06, "loss": 5.8389, "step": 3740 }, { "epoch": 4.19, "learning_rate": 5.492424242424243e-06, "loss": 5.987, "step": 3741 }, { "epoch": 4.2, "learning_rate": 5.484848484848485e-06, "loss": 5.9235, "step": 3742 }, { "epoch": 4.2, "learning_rate": 5.477272727272728e-06, "loss": 6.1513, "step": 3743 }, { "epoch": 4.2, "learning_rate": 5.46969696969697e-06, "loss": 6.1079, "step": 3744 }, { "epoch": 4.2, "learning_rate": 5.462121212121212e-06, "loss": 5.8862, "step": 3745 }, { "epoch": 4.2, "learning_rate": 5.4545454545454545e-06, "loss": 5.7632, "step": 3746 }, { "epoch": 4.2, "learning_rate": 5.446969696969698e-06, "loss": 5.9325, "step": 3747 }, { "epoch": 4.2, "learning_rate": 5.43939393939394e-06, "loss": 5.8638, "step": 3748 }, { "epoch": 4.2, "learning_rate": 5.431818181818182e-06, "loss": 5.7334, "step": 3749 }, { "epoch": 4.2, "learning_rate": 5.4242424242424245e-06, "loss": 5.9388, "step": 3750 }, { "epoch": 4.21, "learning_rate": 5.416666666666667e-06, "loss": 5.9432, "step": 3751 }, { "epoch": 4.21, "learning_rate": 5.409090909090909e-06, "loss": 5.9289, "step": 3752 }, { "epoch": 4.21, "learning_rate": 5.401515151515151e-06, "loss": 5.8485, "step": 3753 }, { "epoch": 4.21, "learning_rate": 5.393939393939394e-06, "loss": 5.8985, "step": 3754 }, { "epoch": 4.21, "learning_rate": 5.386363636363636e-06, "loss": 5.8366, "step": 3755 }, { "epoch": 4.21, "learning_rate": 5.378787878787878e-06, "loss": 5.6838, "step": 3756 }, { "epoch": 4.21, "learning_rate": 5.371212121212121e-06, "loss": 5.8562, "step": 3757 }, { "epoch": 4.21, "learning_rate": 5.3636363636363645e-06, "loss": 5.7613, "step": 3758 }, { "epoch": 4.21, "learning_rate": 5.356060606060607e-06, "loss": 5.5334, "step": 3759 }, { "epoch": 4.22, "learning_rate": 5.348484848484849e-06, "loss": 5.6425, "step": 3760 }, { "epoch": 4.22, "learning_rate": 5.340909090909091e-06, "loss": 5.428, "step": 3761 }, { "epoch": 4.22, "learning_rate": 5.333333333333334e-06, "loss": 5.5259, "step": 3762 }, { "epoch": 4.22, "learning_rate": 5.325757575757576e-06, "loss": 5.1761, "step": 3763 }, { "epoch": 4.22, "learning_rate": 5.318181818181818e-06, "loss": 5.0898, "step": 3764 }, { "epoch": 4.22, "learning_rate": 5.3106060606060605e-06, "loss": 4.9686, "step": 3765 }, { "epoch": 4.22, "learning_rate": 5.303030303030303e-06, "loss": 4.7025, "step": 3766 }, { "epoch": 4.22, "learning_rate": 5.295454545454546e-06, "loss": 4.5768, "step": 3767 }, { "epoch": 4.22, "learning_rate": 5.287878787878788e-06, "loss": 3.4667, "step": 3768 }, { "epoch": 4.23, "learning_rate": 5.2803030303030305e-06, "loss": 6.4196, "step": 3769 }, { "epoch": 4.23, "learning_rate": 5.272727272727273e-06, "loss": 6.1487, "step": 3770 }, { "epoch": 4.23, "learning_rate": 5.265151515151515e-06, "loss": 6.1623, "step": 3771 }, { "epoch": 4.23, "learning_rate": 5.257575757575757e-06, "loss": 6.2331, "step": 3772 }, { "epoch": 4.23, "learning_rate": 5.25e-06, "loss": 6.2376, "step": 3773 }, { "epoch": 4.23, "learning_rate": 5.242424242424242e-06, "loss": 6.0933, "step": 3774 }, { "epoch": 4.23, "learning_rate": 5.234848484848485e-06, "loss": 6.1468, "step": 3775 }, { "epoch": 4.23, "learning_rate": 5.2272727272727274e-06, "loss": 6.2323, "step": 3776 }, { "epoch": 4.23, "learning_rate": 5.21969696969697e-06, "loss": 6.0953, "step": 3777 }, { "epoch": 4.24, "learning_rate": 5.212121212121213e-06, "loss": 6.1373, "step": 3778 }, { "epoch": 4.24, "learning_rate": 5.204545454545455e-06, "loss": 5.9309, "step": 3779 }, { "epoch": 4.24, "learning_rate": 5.1969696969696974e-06, "loss": 6.0563, "step": 3780 }, { "epoch": 4.24, "learning_rate": 5.18939393939394e-06, "loss": 5.9478, "step": 3781 }, { "epoch": 4.24, "learning_rate": 5.181818181818182e-06, "loss": 5.8973, "step": 3782 }, { "epoch": 4.24, "learning_rate": 5.174242424242424e-06, "loss": 5.8524, "step": 3783 }, { "epoch": 4.24, "learning_rate": 5.166666666666667e-06, "loss": 5.9851, "step": 3784 }, { "epoch": 4.24, "learning_rate": 5.159090909090909e-06, "loss": 5.9809, "step": 3785 }, { "epoch": 4.24, "learning_rate": 5.151515151515151e-06, "loss": 6.032, "step": 3786 }, { "epoch": 4.25, "learning_rate": 5.1439393939393935e-06, "loss": 6.0004, "step": 3787 }, { "epoch": 4.25, "learning_rate": 5.136363636363637e-06, "loss": 5.9845, "step": 3788 }, { "epoch": 4.25, "learning_rate": 5.128787878787879e-06, "loss": 6.1306, "step": 3789 }, { "epoch": 4.25, "learning_rate": 5.121212121212121e-06, "loss": 5.8952, "step": 3790 }, { "epoch": 4.25, "learning_rate": 5.1136363636363635e-06, "loss": 6.1098, "step": 3791 }, { "epoch": 4.25, "learning_rate": 5.106060606060607e-06, "loss": 5.9496, "step": 3792 }, { "epoch": 4.25, "learning_rate": 5.098484848484849e-06, "loss": 5.7889, "step": 3793 }, { "epoch": 4.25, "learning_rate": 5.090909090909091e-06, "loss": 5.9141, "step": 3794 }, { "epoch": 4.25, "learning_rate": 5.0833333333333335e-06, "loss": 5.8948, "step": 3795 }, { "epoch": 4.26, "learning_rate": 5.075757575757576e-06, "loss": 5.9788, "step": 3796 }, { "epoch": 4.26, "learning_rate": 5.068181818181818e-06, "loss": 5.8652, "step": 3797 }, { "epoch": 4.26, "learning_rate": 5.060606060606061e-06, "loss": 5.8521, "step": 3798 }, { "epoch": 4.26, "learning_rate": 5.0530303030303035e-06, "loss": 5.9079, "step": 3799 }, { "epoch": 4.26, "learning_rate": 5.045454545454546e-06, "loss": 5.7668, "step": 3800 }, { "epoch": 4.26, "learning_rate": 5.037878787878788e-06, "loss": 5.8593, "step": 3801 }, { "epoch": 4.26, "learning_rate": 5.03030303030303e-06, "loss": 5.9812, "step": 3802 }, { "epoch": 4.26, "learning_rate": 5.022727272727273e-06, "loss": 5.8682, "step": 3803 }, { "epoch": 4.26, "learning_rate": 5.015151515151515e-06, "loss": 5.8605, "step": 3804 }, { "epoch": 4.27, "learning_rate": 5.007575757575757e-06, "loss": 5.9533, "step": 3805 }, { "epoch": 4.27, "learning_rate": 4.9999999999999996e-06, "loss": 5.7291, "step": 3806 }, { "epoch": 4.27, "learning_rate": 4.992424242424242e-06, "loss": 5.7806, "step": 3807 }, { "epoch": 4.27, "learning_rate": 4.984848484848485e-06, "loss": 5.8513, "step": 3808 }, { "epoch": 4.27, "learning_rate": 4.977272727272728e-06, "loss": 5.6283, "step": 3809 }, { "epoch": 4.27, "learning_rate": 4.96969696969697e-06, "loss": 5.6557, "step": 3810 }, { "epoch": 4.27, "learning_rate": 4.962121212121213e-06, "loss": 5.4192, "step": 3811 }, { "epoch": 4.27, "learning_rate": 4.954545454545455e-06, "loss": 5.4663, "step": 3812 }, { "epoch": 4.27, "learning_rate": 4.946969696969697e-06, "loss": 5.3652, "step": 3813 }, { "epoch": 4.28, "learning_rate": 4.93939393939394e-06, "loss": 5.0649, "step": 3814 }, { "epoch": 4.28, "learning_rate": 4.931818181818182e-06, "loss": 4.8314, "step": 3815 }, { "epoch": 4.28, "learning_rate": 4.924242424242424e-06, "loss": 4.3923, "step": 3816 }, { "epoch": 4.28, "learning_rate": 4.9166666666666665e-06, "loss": 4.0663, "step": 3817 }, { "epoch": 4.28, "learning_rate": 4.90909090909091e-06, "loss": 3.609, "step": 3818 }, { "epoch": 4.28, "learning_rate": 4.901515151515152e-06, "loss": 6.2083, "step": 3819 }, { "epoch": 4.28, "learning_rate": 4.893939393939394e-06, "loss": 6.1557, "step": 3820 }, { "epoch": 4.28, "learning_rate": 4.8863636363636365e-06, "loss": 6.1204, "step": 3821 }, { "epoch": 4.28, "learning_rate": 4.878787878787879e-06, "loss": 6.1795, "step": 3822 }, { "epoch": 4.29, "learning_rate": 4.871212121212121e-06, "loss": 6.1765, "step": 3823 }, { "epoch": 4.29, "learning_rate": 4.863636363636363e-06, "loss": 6.018, "step": 3824 }, { "epoch": 4.29, "learning_rate": 4.856060606060606e-06, "loss": 5.9167, "step": 3825 }, { "epoch": 4.29, "learning_rate": 4.848484848484849e-06, "loss": 6.1957, "step": 3826 }, { "epoch": 4.29, "learning_rate": 4.840909090909091e-06, "loss": 5.9125, "step": 3827 }, { "epoch": 4.29, "learning_rate": 4.833333333333333e-06, "loss": 5.9597, "step": 3828 }, { "epoch": 4.29, "learning_rate": 4.8257575757575765e-06, "loss": 5.9627, "step": 3829 }, { "epoch": 4.29, "learning_rate": 4.818181818181819e-06, "loss": 6.0977, "step": 3830 }, { "epoch": 4.29, "learning_rate": 4.810606060606061e-06, "loss": 6.0889, "step": 3831 }, { "epoch": 4.3, "learning_rate": 4.803030303030303e-06, "loss": 6.081, "step": 3832 }, { "epoch": 4.3, "learning_rate": 4.795454545454546e-06, "loss": 6.0213, "step": 3833 }, { "epoch": 4.3, "learning_rate": 4.787878787878788e-06, "loss": 5.9242, "step": 3834 }, { "epoch": 4.3, "learning_rate": 4.78030303030303e-06, "loss": 5.9288, "step": 3835 }, { "epoch": 4.3, "learning_rate": 4.7727272727272725e-06, "loss": 6.0178, "step": 3836 }, { "epoch": 4.3, "learning_rate": 4.765151515151515e-06, "loss": 5.9715, "step": 3837 }, { "epoch": 4.3, "learning_rate": 4.757575757575757e-06, "loss": 5.9594, "step": 3838 }, { "epoch": 4.3, "learning_rate": 4.75e-06, "loss": 5.928, "step": 3839 }, { "epoch": 4.3, "learning_rate": 4.7424242424242426e-06, "loss": 5.8587, "step": 3840 }, { "epoch": 4.31, "learning_rate": 4.734848484848485e-06, "loss": 5.9341, "step": 3841 }, { "epoch": 4.31, "learning_rate": 4.727272727272727e-06, "loss": 5.9721, "step": 3842 }, { "epoch": 4.31, "learning_rate": 4.71969696969697e-06, "loss": 5.8805, "step": 3843 }, { "epoch": 4.31, "learning_rate": 4.7121212121212126e-06, "loss": 6.0605, "step": 3844 }, { "epoch": 4.31, "learning_rate": 4.704545454545455e-06, "loss": 5.8341, "step": 3845 }, { "epoch": 4.31, "learning_rate": 4.696969696969697e-06, "loss": 5.9429, "step": 3846 }, { "epoch": 4.31, "learning_rate": 4.6893939393939394e-06, "loss": 5.8651, "step": 3847 }, { "epoch": 4.31, "learning_rate": 4.681818181818182e-06, "loss": 5.8785, "step": 3848 }, { "epoch": 4.32, "learning_rate": 4.674242424242425e-06, "loss": 5.9414, "step": 3849 }, { "epoch": 4.32, "learning_rate": 4.666666666666667e-06, "loss": 6.0253, "step": 3850 }, { "epoch": 4.32, "learning_rate": 4.6590909090909095e-06, "loss": 5.8767, "step": 3851 }, { "epoch": 4.32, "learning_rate": 4.651515151515152e-06, "loss": 5.8151, "step": 3852 }, { "epoch": 4.32, "learning_rate": 4.643939393939394e-06, "loss": 5.882, "step": 3853 }, { "epoch": 4.32, "learning_rate": 4.636363636363636e-06, "loss": 5.8548, "step": 3854 }, { "epoch": 4.32, "learning_rate": 4.628787878787879e-06, "loss": 5.8692, "step": 3855 }, { "epoch": 4.32, "learning_rate": 4.621212121212121e-06, "loss": 5.7408, "step": 3856 }, { "epoch": 4.32, "learning_rate": 4.613636363636363e-06, "loss": 5.8268, "step": 3857 }, { "epoch": 4.33, "learning_rate": 4.6060606060606055e-06, "loss": 5.6376, "step": 3858 }, { "epoch": 4.33, "learning_rate": 4.598484848484849e-06, "loss": 5.8963, "step": 3859 }, { "epoch": 4.33, "learning_rate": 4.590909090909091e-06, "loss": 5.5266, "step": 3860 }, { "epoch": 4.33, "learning_rate": 4.583333333333334e-06, "loss": 5.4584, "step": 3861 }, { "epoch": 4.33, "learning_rate": 4.575757575757576e-06, "loss": 5.4794, "step": 3862 }, { "epoch": 4.33, "learning_rate": 4.568181818181819e-06, "loss": 5.181, "step": 3863 }, { "epoch": 4.33, "learning_rate": 4.560606060606061e-06, "loss": 5.0722, "step": 3864 }, { "epoch": 4.33, "learning_rate": 4.553030303030303e-06, "loss": 4.8418, "step": 3865 }, { "epoch": 4.33, "learning_rate": 4.5454545454545455e-06, "loss": 5.0351, "step": 3866 }, { "epoch": 4.34, "learning_rate": 4.537878787878788e-06, "loss": 4.3706, "step": 3867 }, { "epoch": 4.34, "learning_rate": 4.53030303030303e-06, "loss": 3.7141, "step": 3868 }, { "epoch": 4.34, "learning_rate": 4.522727272727273e-06, "loss": 6.3218, "step": 3869 }, { "epoch": 4.34, "learning_rate": 4.5151515151515155e-06, "loss": 6.1795, "step": 3870 }, { "epoch": 4.34, "learning_rate": 4.507575757575758e-06, "loss": 6.1349, "step": 3871 }, { "epoch": 4.34, "learning_rate": 4.5e-06, "loss": 6.0989, "step": 3872 }, { "epoch": 4.34, "learning_rate": 4.492424242424242e-06, "loss": 6.1737, "step": 3873 }, { "epoch": 4.34, "learning_rate": 4.484848484848485e-06, "loss": 6.214, "step": 3874 }, { "epoch": 4.34, "learning_rate": 4.477272727272727e-06, "loss": 6.1944, "step": 3875 }, { "epoch": 4.35, "learning_rate": 4.469696969696969e-06, "loss": 6.1607, "step": 3876 }, { "epoch": 4.35, "learning_rate": 4.4621212121212116e-06, "loss": 5.9978, "step": 3877 }, { "epoch": 4.35, "learning_rate": 4.454545454545455e-06, "loss": 6.0706, "step": 3878 }, { "epoch": 4.35, "learning_rate": 4.446969696969697e-06, "loss": 5.9615, "step": 3879 }, { "epoch": 4.35, "learning_rate": 4.43939393939394e-06, "loss": 6.1976, "step": 3880 }, { "epoch": 4.35, "learning_rate": 4.4318181818181824e-06, "loss": 5.9765, "step": 3881 }, { "epoch": 4.35, "learning_rate": 4.424242424242425e-06, "loss": 5.8143, "step": 3882 }, { "epoch": 4.35, "learning_rate": 4.416666666666667e-06, "loss": 6.0442, "step": 3883 }, { "epoch": 4.35, "learning_rate": 4.409090909090909e-06, "loss": 5.9172, "step": 3884 }, { "epoch": 4.36, "learning_rate": 4.401515151515152e-06, "loss": 5.8243, "step": 3885 }, { "epoch": 4.36, "learning_rate": 4.393939393939394e-06, "loss": 5.8842, "step": 3886 }, { "epoch": 4.36, "learning_rate": 4.386363636363636e-06, "loss": 5.9967, "step": 3887 }, { "epoch": 4.36, "learning_rate": 4.3787878787878785e-06, "loss": 6.007, "step": 3888 }, { "epoch": 4.36, "learning_rate": 4.371212121212121e-06, "loss": 5.8805, "step": 3889 }, { "epoch": 4.36, "learning_rate": 4.363636363636364e-06, "loss": 5.8779, "step": 3890 }, { "epoch": 4.36, "learning_rate": 4.356060606060606e-06, "loss": 6.0065, "step": 3891 }, { "epoch": 4.36, "learning_rate": 4.3484848484848485e-06, "loss": 5.9082, "step": 3892 }, { "epoch": 4.36, "learning_rate": 4.340909090909091e-06, "loss": 5.9072, "step": 3893 }, { "epoch": 4.37, "learning_rate": 4.333333333333333e-06, "loss": 6.0222, "step": 3894 }, { "epoch": 4.37, "learning_rate": 4.325757575757576e-06, "loss": 5.912, "step": 3895 }, { "epoch": 4.37, "learning_rate": 4.3181818181818185e-06, "loss": 5.7917, "step": 3896 }, { "epoch": 4.37, "learning_rate": 4.310606060606061e-06, "loss": 5.8869, "step": 3897 }, { "epoch": 4.37, "learning_rate": 4.303030303030303e-06, "loss": 5.9186, "step": 3898 }, { "epoch": 4.37, "learning_rate": 4.295454545454545e-06, "loss": 5.8161, "step": 3899 }, { "epoch": 4.37, "learning_rate": 4.2878787878787885e-06, "loss": 5.7857, "step": 3900 }, { "epoch": 4.37, "learning_rate": 4.280303030303031e-06, "loss": 5.9493, "step": 3901 }, { "epoch": 4.37, "learning_rate": 4.272727272727273e-06, "loss": 5.8237, "step": 3902 }, { "epoch": 4.38, "learning_rate": 4.265151515151515e-06, "loss": 5.9383, "step": 3903 }, { "epoch": 4.38, "learning_rate": 4.257575757575758e-06, "loss": 5.8752, "step": 3904 }, { "epoch": 4.38, "learning_rate": 4.25e-06, "loss": 5.8824, "step": 3905 }, { "epoch": 4.38, "learning_rate": 4.242424242424242e-06, "loss": 5.9408, "step": 3906 }, { "epoch": 4.38, "learning_rate": 4.2348484848484845e-06, "loss": 5.7186, "step": 3907 }, { "epoch": 4.38, "learning_rate": 4.227272727272727e-06, "loss": 5.7788, "step": 3908 }, { "epoch": 4.38, "learning_rate": 4.219696969696969e-06, "loss": 5.7062, "step": 3909 }, { "epoch": 4.38, "learning_rate": 4.212121212121212e-06, "loss": 5.7976, "step": 3910 }, { "epoch": 4.38, "learning_rate": 4.2045454545454546e-06, "loss": 5.5834, "step": 3911 }, { "epoch": 4.39, "learning_rate": 4.196969696969698e-06, "loss": 5.479, "step": 3912 }, { "epoch": 4.39, "learning_rate": 4.18939393939394e-06, "loss": 5.0965, "step": 3913 }, { "epoch": 4.39, "learning_rate": 4.181818181818182e-06, "loss": 5.1172, "step": 3914 }, { "epoch": 4.39, "learning_rate": 4.1742424242424246e-06, "loss": 4.9772, "step": 3915 }, { "epoch": 4.39, "learning_rate": 4.166666666666667e-06, "loss": 4.6952, "step": 3916 }, { "epoch": 4.39, "learning_rate": 4.159090909090909e-06, "loss": 4.4114, "step": 3917 }, { "epoch": 4.39, "learning_rate": 4.1515151515151514e-06, "loss": 3.648, "step": 3918 }, { "epoch": 4.39, "learning_rate": 4.143939393939394e-06, "loss": 6.2379, "step": 3919 }, { "epoch": 4.39, "learning_rate": 4.136363636363637e-06, "loss": 6.071, "step": 3920 }, { "epoch": 4.4, "learning_rate": 4.128787878787879e-06, "loss": 5.9092, "step": 3921 }, { "epoch": 4.4, "learning_rate": 4.1212121212121215e-06, "loss": 6.1844, "step": 3922 }, { "epoch": 4.4, "learning_rate": 4.113636363636364e-06, "loss": 6.0741, "step": 3923 }, { "epoch": 4.4, "learning_rate": 4.106060606060606e-06, "loss": 6.0012, "step": 3924 }, { "epoch": 4.4, "learning_rate": 4.098484848484848e-06, "loss": 6.0003, "step": 3925 }, { "epoch": 4.4, "learning_rate": 4.090909090909091e-06, "loss": 6.1987, "step": 3926 }, { "epoch": 4.4, "learning_rate": 4.083333333333333e-06, "loss": 6.0125, "step": 3927 }, { "epoch": 4.4, "learning_rate": 4.075757575757575e-06, "loss": 6.1388, "step": 3928 }, { "epoch": 4.4, "learning_rate": 4.068181818181818e-06, "loss": 5.9194, "step": 3929 }, { "epoch": 4.41, "learning_rate": 4.060606060606061e-06, "loss": 5.8873, "step": 3930 }, { "epoch": 4.41, "learning_rate": 4.053030303030304e-06, "loss": 6.0066, "step": 3931 }, { "epoch": 4.41, "learning_rate": 4.045454545454546e-06, "loss": 5.8983, "step": 3932 }, { "epoch": 4.41, "learning_rate": 4.037878787878788e-06, "loss": 5.9107, "step": 3933 }, { "epoch": 4.41, "learning_rate": 4.030303030303031e-06, "loss": 5.9148, "step": 3934 }, { "epoch": 4.41, "learning_rate": 4.022727272727273e-06, "loss": 5.8426, "step": 3935 }, { "epoch": 4.41, "learning_rate": 4.015151515151515e-06, "loss": 5.9893, "step": 3936 }, { "epoch": 4.41, "learning_rate": 4.0075757575757575e-06, "loss": 5.7977, "step": 3937 }, { "epoch": 4.41, "learning_rate": 4e-06, "loss": 6.0869, "step": 3938 }, { "epoch": 4.42, "learning_rate": 3.992424242424242e-06, "loss": 6.0277, "step": 3939 }, { "epoch": 4.42, "learning_rate": 3.984848484848484e-06, "loss": 5.867, "step": 3940 }, { "epoch": 4.42, "learning_rate": 3.9772727272727275e-06, "loss": 6.0596, "step": 3941 }, { "epoch": 4.42, "learning_rate": 3.96969696969697e-06, "loss": 5.8799, "step": 3942 }, { "epoch": 4.42, "learning_rate": 3.962121212121212e-06, "loss": 6.0214, "step": 3943 }, { "epoch": 4.42, "learning_rate": 3.954545454545454e-06, "loss": 5.8961, "step": 3944 }, { "epoch": 4.42, "learning_rate": 3.946969696969697e-06, "loss": 5.9133, "step": 3945 }, { "epoch": 4.42, "learning_rate": 3.93939393939394e-06, "loss": 6.049, "step": 3946 }, { "epoch": 4.42, "learning_rate": 3.931818181818182e-06, "loss": 5.8181, "step": 3947 }, { "epoch": 4.43, "learning_rate": 3.9242424242424244e-06, "loss": 5.7887, "step": 3948 }, { "epoch": 4.43, "learning_rate": 3.916666666666667e-06, "loss": 5.9129, "step": 3949 }, { "epoch": 4.43, "learning_rate": 3.909090909090909e-06, "loss": 5.7827, "step": 3950 }, { "epoch": 4.43, "learning_rate": 3.901515151515152e-06, "loss": 5.7391, "step": 3951 }, { "epoch": 4.43, "learning_rate": 3.8939393939393944e-06, "loss": 5.7624, "step": 3952 }, { "epoch": 4.43, "learning_rate": 3.886363636363637e-06, "loss": 5.9552, "step": 3953 }, { "epoch": 4.43, "learning_rate": 3.878787878787879e-06, "loss": 5.9008, "step": 3954 }, { "epoch": 4.43, "learning_rate": 3.871212121212121e-06, "loss": 5.944, "step": 3955 }, { "epoch": 4.43, "learning_rate": 3.863636363636364e-06, "loss": 5.9644, "step": 3956 }, { "epoch": 4.44, "learning_rate": 3.856060606060606e-06, "loss": 5.726, "step": 3957 }, { "epoch": 4.44, "learning_rate": 3.848484848484848e-06, "loss": 5.802, "step": 3958 }, { "epoch": 4.44, "learning_rate": 3.8409090909090905e-06, "loss": 5.6843, "step": 3959 }, { "epoch": 4.44, "learning_rate": 3.833333333333333e-06, "loss": 5.7764, "step": 3960 }, { "epoch": 4.44, "learning_rate": 3.825757575757576e-06, "loss": 5.5066, "step": 3961 }, { "epoch": 4.44, "learning_rate": 3.818181818181818e-06, "loss": 5.4873, "step": 3962 }, { "epoch": 4.44, "learning_rate": 3.810606060606061e-06, "loss": 5.2033, "step": 3963 }, { "epoch": 4.44, "learning_rate": 3.8030303030303036e-06, "loss": 5.1215, "step": 3964 }, { "epoch": 4.45, "learning_rate": 3.795454545454546e-06, "loss": 4.8641, "step": 3965 }, { "epoch": 4.45, "learning_rate": 3.7878787878787882e-06, "loss": 4.7577, "step": 3966 }, { "epoch": 4.45, "learning_rate": 3.7803030303030305e-06, "loss": 4.2188, "step": 3967 }, { "epoch": 4.45, "learning_rate": 3.772727272727273e-06, "loss": 3.5566, "step": 3968 }, { "epoch": 4.45, "learning_rate": 3.7651515151515155e-06, "loss": 6.2738, "step": 3969 }, { "epoch": 4.45, "learning_rate": 3.757575757575758e-06, "loss": 6.1988, "step": 3970 }, { "epoch": 4.45, "learning_rate": 3.75e-06, "loss": 6.2322, "step": 3971 }, { "epoch": 4.45, "learning_rate": 3.7424242424242424e-06, "loss": 5.9918, "step": 3972 }, { "epoch": 4.45, "learning_rate": 3.7348484848484847e-06, "loss": 6.2242, "step": 3973 }, { "epoch": 4.46, "learning_rate": 3.7272727272727274e-06, "loss": 6.0757, "step": 3974 }, { "epoch": 4.46, "learning_rate": 3.7196969696969697e-06, "loss": 6.0371, "step": 3975 }, { "epoch": 4.46, "learning_rate": 3.7121212121212124e-06, "loss": 6.0265, "step": 3976 }, { "epoch": 4.46, "learning_rate": 3.7045454545454547e-06, "loss": 6.079, "step": 3977 }, { "epoch": 4.46, "learning_rate": 3.696969696969697e-06, "loss": 5.9745, "step": 3978 }, { "epoch": 4.46, "learning_rate": 3.6893939393939397e-06, "loss": 5.7729, "step": 3979 }, { "epoch": 4.46, "learning_rate": 3.681818181818182e-06, "loss": 5.963, "step": 3980 }, { "epoch": 4.46, "learning_rate": 3.6742424242424243e-06, "loss": 5.8666, "step": 3981 }, { "epoch": 4.46, "learning_rate": 3.6666666666666666e-06, "loss": 5.8753, "step": 3982 }, { "epoch": 4.47, "learning_rate": 3.659090909090909e-06, "loss": 6.0223, "step": 3983 }, { "epoch": 4.47, "learning_rate": 3.651515151515152e-06, "loss": 5.8683, "step": 3984 }, { "epoch": 4.47, "learning_rate": 3.6439393939393943e-06, "loss": 5.8273, "step": 3985 }, { "epoch": 4.47, "learning_rate": 3.6363636363636366e-06, "loss": 5.844, "step": 3986 }, { "epoch": 4.47, "learning_rate": 3.628787878787879e-06, "loss": 6.0019, "step": 3987 }, { "epoch": 4.47, "learning_rate": 3.621212121212121e-06, "loss": 6.1198, "step": 3988 }, { "epoch": 4.47, "learning_rate": 3.613636363636364e-06, "loss": 5.9026, "step": 3989 }, { "epoch": 4.47, "learning_rate": 3.606060606060606e-06, "loss": 5.9284, "step": 3990 }, { "epoch": 4.47, "learning_rate": 3.5984848484848485e-06, "loss": 5.8772, "step": 3991 }, { "epoch": 4.48, "learning_rate": 3.5909090909090908e-06, "loss": 5.8742, "step": 3992 }, { "epoch": 4.48, "learning_rate": 3.5833333333333335e-06, "loss": 5.8556, "step": 3993 }, { "epoch": 4.48, "learning_rate": 3.575757575757576e-06, "loss": 5.9246, "step": 3994 }, { "epoch": 4.48, "learning_rate": 3.5681818181818185e-06, "loss": 5.8857, "step": 3995 }, { "epoch": 4.48, "learning_rate": 3.5606060606060608e-06, "loss": 5.9292, "step": 3996 }, { "epoch": 4.48, "learning_rate": 3.553030303030303e-06, "loss": 5.8697, "step": 3997 }, { "epoch": 4.48, "learning_rate": 3.5454545454545454e-06, "loss": 5.8805, "step": 3998 }, { "epoch": 4.48, "learning_rate": 3.5378787878787876e-06, "loss": 5.6532, "step": 3999 }, { "epoch": 4.48, "learning_rate": 3.5303030303030304e-06, "loss": 5.9357, "step": 4000 }, { "epoch": 4.48, "eval_loss": 5.612024307250977, "eval_runtime": 823.4353, "eval_samples_per_second": 3.209, "eval_steps_per_second": 0.402, "eval_wer": 1.0267155890519635, "step": 4000 }, { "epoch": 4.49, "learning_rate": 3.522727272727273e-06, "loss": 5.8247, "step": 4001 }, { "epoch": 4.49, "learning_rate": 3.5151515151515154e-06, "loss": 5.8036, "step": 4002 }, { "epoch": 4.49, "learning_rate": 3.5075757575757577e-06, "loss": 5.9761, "step": 4003 }, { "epoch": 4.49, "learning_rate": 3.5e-06, "loss": 5.7666, "step": 4004 }, { "epoch": 4.49, "learning_rate": 3.4924242424242427e-06, "loss": 5.743, "step": 4005 }, { "epoch": 4.49, "learning_rate": 3.484848484848485e-06, "loss": 5.6684, "step": 4006 }, { "epoch": 4.49, "learning_rate": 3.4772727272727272e-06, "loss": 5.8398, "step": 4007 }, { "epoch": 4.49, "learning_rate": 3.4696969696969695e-06, "loss": 5.8143, "step": 4008 }, { "epoch": 4.49, "learning_rate": 3.462121212121212e-06, "loss": 5.7507, "step": 4009 }, { "epoch": 4.5, "learning_rate": 3.454545454545455e-06, "loss": 5.4821, "step": 4010 }, { "epoch": 4.5, "learning_rate": 3.4469696969696973e-06, "loss": 5.4335, "step": 4011 }, { "epoch": 4.5, "learning_rate": 3.4393939393939395e-06, "loss": 5.6151, "step": 4012 }, { "epoch": 4.5, "learning_rate": 3.431818181818182e-06, "loss": 5.2547, "step": 4013 }, { "epoch": 4.5, "learning_rate": 3.424242424242424e-06, "loss": 4.8481, "step": 4014 }, { "epoch": 4.5, "learning_rate": 3.416666666666667e-06, "loss": 4.7193, "step": 4015 }, { "epoch": 4.5, "learning_rate": 3.409090909090909e-06, "loss": 4.5573, "step": 4016 }, { "epoch": 4.5, "learning_rate": 3.4015151515151514e-06, "loss": 3.9217, "step": 4017 }, { "epoch": 4.5, "learning_rate": 3.393939393939394e-06, "loss": 3.5834, "step": 4018 }, { "epoch": 4.51, "learning_rate": 3.3863636363636364e-06, "loss": 6.1817, "step": 4019 }, { "epoch": 4.51, "learning_rate": 3.378787878787879e-06, "loss": 6.2577, "step": 4020 }, { "epoch": 4.51, "learning_rate": 3.3712121212121214e-06, "loss": 6.2622, "step": 4021 }, { "epoch": 4.51, "learning_rate": 3.3636363636363637e-06, "loss": 6.1538, "step": 4022 }, { "epoch": 4.51, "learning_rate": 3.356060606060606e-06, "loss": 6.1377, "step": 4023 }, { "epoch": 4.51, "learning_rate": 3.3484848484848483e-06, "loss": 5.9713, "step": 4024 }, { "epoch": 4.51, "learning_rate": 3.340909090909091e-06, "loss": 6.1848, "step": 4025 }, { "epoch": 4.51, "learning_rate": 3.3333333333333333e-06, "loss": 6.0562, "step": 4026 }, { "epoch": 4.51, "learning_rate": 3.325757575757576e-06, "loss": 5.9116, "step": 4027 }, { "epoch": 4.52, "learning_rate": 3.3181818181818183e-06, "loss": 6.0162, "step": 4028 }, { "epoch": 4.52, "learning_rate": 3.3106060606060606e-06, "loss": 5.9158, "step": 4029 }, { "epoch": 4.52, "learning_rate": 3.3030303030303033e-06, "loss": 5.82, "step": 4030 }, { "epoch": 4.52, "learning_rate": 3.2954545454545456e-06, "loss": 6.1441, "step": 4031 }, { "epoch": 4.52, "learning_rate": 3.287878787878788e-06, "loss": 6.0003, "step": 4032 }, { "epoch": 4.52, "learning_rate": 3.28030303030303e-06, "loss": 5.9714, "step": 4033 }, { "epoch": 4.52, "learning_rate": 3.2727272727272725e-06, "loss": 6.0075, "step": 4034 }, { "epoch": 4.52, "learning_rate": 3.2651515151515152e-06, "loss": 5.8875, "step": 4035 }, { "epoch": 4.52, "learning_rate": 3.257575757575758e-06, "loss": 6.0526, "step": 4036 }, { "epoch": 4.53, "learning_rate": 3.2500000000000002e-06, "loss": 5.9145, "step": 4037 }, { "epoch": 4.53, "learning_rate": 3.2424242424242425e-06, "loss": 5.9201, "step": 4038 }, { "epoch": 4.53, "learning_rate": 3.234848484848485e-06, "loss": 6.0167, "step": 4039 }, { "epoch": 4.53, "learning_rate": 3.2272727272727275e-06, "loss": 5.8785, "step": 4040 }, { "epoch": 4.53, "learning_rate": 3.21969696969697e-06, "loss": 5.9095, "step": 4041 }, { "epoch": 4.53, "learning_rate": 3.212121212121212e-06, "loss": 6.0626, "step": 4042 }, { "epoch": 4.53, "learning_rate": 3.2045454545454544e-06, "loss": 5.9586, "step": 4043 }, { "epoch": 4.53, "learning_rate": 3.196969696969697e-06, "loss": 5.8299, "step": 4044 }, { "epoch": 4.53, "learning_rate": 3.1893939393939394e-06, "loss": 5.8729, "step": 4045 }, { "epoch": 4.54, "learning_rate": 3.181818181818182e-06, "loss": 5.8147, "step": 4046 }, { "epoch": 4.54, "learning_rate": 3.1742424242424244e-06, "loss": 5.9045, "step": 4047 }, { "epoch": 4.54, "learning_rate": 3.1666666666666667e-06, "loss": 5.9084, "step": 4048 }, { "epoch": 4.54, "learning_rate": 3.159090909090909e-06, "loss": 5.9675, "step": 4049 }, { "epoch": 4.54, "learning_rate": 3.1515151515151513e-06, "loss": 5.8205, "step": 4050 }, { "epoch": 4.54, "learning_rate": 3.143939393939394e-06, "loss": 5.8813, "step": 4051 }, { "epoch": 4.54, "learning_rate": 3.1363636363636363e-06, "loss": 5.9438, "step": 4052 }, { "epoch": 4.54, "learning_rate": 3.128787878787879e-06, "loss": 5.9785, "step": 4053 }, { "epoch": 4.54, "learning_rate": 3.1212121212121213e-06, "loss": 5.7725, "step": 4054 }, { "epoch": 4.55, "learning_rate": 3.1136363636363636e-06, "loss": 6.0377, "step": 4055 }, { "epoch": 4.55, "learning_rate": 3.1060606060606063e-06, "loss": 5.8234, "step": 4056 }, { "epoch": 4.55, "learning_rate": 3.0984848484848486e-06, "loss": 5.6636, "step": 4057 }, { "epoch": 4.55, "learning_rate": 3.090909090909091e-06, "loss": 5.7357, "step": 4058 }, { "epoch": 4.55, "learning_rate": 3.083333333333333e-06, "loss": 5.7947, "step": 4059 }, { "epoch": 4.55, "learning_rate": 3.0757575757575755e-06, "loss": 5.5759, "step": 4060 }, { "epoch": 4.55, "learning_rate": 3.0681818181818186e-06, "loss": 5.7065, "step": 4061 }, { "epoch": 4.55, "learning_rate": 3.060606060606061e-06, "loss": 5.6179, "step": 4062 }, { "epoch": 4.55, "learning_rate": 3.053030303030303e-06, "loss": 5.4578, "step": 4063 }, { "epoch": 4.56, "learning_rate": 3.0454545454545455e-06, "loss": 5.2009, "step": 4064 }, { "epoch": 4.56, "learning_rate": 3.0378787878787878e-06, "loss": 4.9667, "step": 4065 }, { "epoch": 4.56, "learning_rate": 3.0303030303030305e-06, "loss": 4.9777, "step": 4066 }, { "epoch": 4.56, "learning_rate": 3.0227272727272728e-06, "loss": 4.4069, "step": 4067 }, { "epoch": 4.56, "learning_rate": 3.015151515151515e-06, "loss": 3.7542, "step": 4068 }, { "epoch": 4.56, "learning_rate": 3.0075757575757574e-06, "loss": 6.2947, "step": 4069 }, { "epoch": 4.56, "learning_rate": 3e-06, "loss": 6.1143, "step": 4070 }, { "epoch": 4.56, "learning_rate": 2.992424242424243e-06, "loss": 6.1571, "step": 4071 }, { "epoch": 4.57, "learning_rate": 2.984848484848485e-06, "loss": 6.118, "step": 4072 }, { "epoch": 4.57, "learning_rate": 2.9772727272727274e-06, "loss": 6.1772, "step": 4073 }, { "epoch": 4.57, "learning_rate": 2.9696969696969697e-06, "loss": 6.0503, "step": 4074 }, { "epoch": 4.57, "learning_rate": 2.962121212121212e-06, "loss": 6.0828, "step": 4075 }, { "epoch": 4.57, "learning_rate": 2.9545454545454547e-06, "loss": 6.0393, "step": 4076 }, { "epoch": 4.57, "learning_rate": 2.946969696969697e-06, "loss": 6.1639, "step": 4077 }, { "epoch": 4.57, "learning_rate": 2.9393939393939397e-06, "loss": 6.009, "step": 4078 }, { "epoch": 4.57, "learning_rate": 2.931818181818182e-06, "loss": 5.9363, "step": 4079 }, { "epoch": 4.57, "learning_rate": 2.9242424242424243e-06, "loss": 5.8817, "step": 4080 }, { "epoch": 4.58, "learning_rate": 2.916666666666667e-06, "loss": 6.0066, "step": 4081 }, { "epoch": 4.58, "learning_rate": 2.9090909090909093e-06, "loss": 5.9484, "step": 4082 }, { "epoch": 4.58, "learning_rate": 2.9015151515151516e-06, "loss": 5.8783, "step": 4083 }, { "epoch": 4.58, "learning_rate": 2.893939393939394e-06, "loss": 5.822, "step": 4084 }, { "epoch": 4.58, "learning_rate": 2.886363636363636e-06, "loss": 6.0421, "step": 4085 }, { "epoch": 4.58, "learning_rate": 2.878787878787879e-06, "loss": 5.9203, "step": 4086 }, { "epoch": 4.58, "learning_rate": 2.8712121212121216e-06, "loss": 5.7648, "step": 4087 }, { "epoch": 4.58, "learning_rate": 2.863636363636364e-06, "loss": 5.7705, "step": 4088 }, { "epoch": 4.58, "learning_rate": 2.856060606060606e-06, "loss": 5.9404, "step": 4089 }, { "epoch": 4.59, "learning_rate": 2.8484848484848484e-06, "loss": 5.9146, "step": 4090 }, { "epoch": 4.59, "learning_rate": 2.840909090909091e-06, "loss": 5.9234, "step": 4091 }, { "epoch": 4.59, "learning_rate": 2.8333333333333335e-06, "loss": 5.9421, "step": 4092 }, { "epoch": 4.59, "learning_rate": 2.8257575757575757e-06, "loss": 5.7846, "step": 4093 }, { "epoch": 4.59, "learning_rate": 2.818181818181818e-06, "loss": 5.8647, "step": 4094 }, { "epoch": 4.59, "learning_rate": 2.8106060606060607e-06, "loss": 5.971, "step": 4095 }, { "epoch": 4.59, "learning_rate": 2.803030303030303e-06, "loss": 5.8213, "step": 4096 }, { "epoch": 4.59, "learning_rate": 2.7954545454545458e-06, "loss": 5.9, "step": 4097 }, { "epoch": 4.59, "learning_rate": 2.787878787878788e-06, "loss": 5.8497, "step": 4098 }, { "epoch": 4.6, "learning_rate": 2.7803030303030303e-06, "loss": 5.8861, "step": 4099 }, { "epoch": 4.6, "learning_rate": 2.7727272727272726e-06, "loss": 5.7035, "step": 4100 }, { "epoch": 4.6, "learning_rate": 2.765151515151515e-06, "loss": 5.7704, "step": 4101 }, { "epoch": 4.6, "learning_rate": 2.7575757575757576e-06, "loss": 5.7689, "step": 4102 }, { "epoch": 4.6, "learning_rate": 2.75e-06, "loss": 5.84, "step": 4103 }, { "epoch": 4.6, "learning_rate": 2.7424242424242426e-06, "loss": 5.7764, "step": 4104 }, { "epoch": 4.6, "learning_rate": 2.734848484848485e-06, "loss": 5.7576, "step": 4105 }, { "epoch": 4.6, "learning_rate": 2.7272727272727272e-06, "loss": 5.6927, "step": 4106 }, { "epoch": 4.6, "learning_rate": 2.71969696969697e-06, "loss": 5.7037, "step": 4107 }, { "epoch": 4.61, "learning_rate": 2.7121212121212122e-06, "loss": 5.8329, "step": 4108 }, { "epoch": 4.61, "learning_rate": 2.7045454545454545e-06, "loss": 5.7694, "step": 4109 }, { "epoch": 4.61, "learning_rate": 2.696969696969697e-06, "loss": 5.7154, "step": 4110 }, { "epoch": 4.61, "learning_rate": 2.689393939393939e-06, "loss": 5.4245, "step": 4111 }, { "epoch": 4.61, "learning_rate": 2.6818181818181822e-06, "loss": 5.4756, "step": 4112 }, { "epoch": 4.61, "learning_rate": 2.6742424242424245e-06, "loss": 5.2513, "step": 4113 }, { "epoch": 4.61, "learning_rate": 2.6742424242424245e-06, "loss": 5.0783, "step": 4114 }, { "epoch": 4.61, "learning_rate": 2.666666666666667e-06, "loss": 4.948, "step": 4115 }, { "epoch": 4.61, "learning_rate": 2.659090909090909e-06, "loss": 4.744, "step": 4116 }, { "epoch": 4.62, "learning_rate": 2.6515151515151514e-06, "loss": 4.5605, "step": 4117 }, { "epoch": 4.62, "learning_rate": 2.643939393939394e-06, "loss": 3.4786, "step": 4118 }, { "epoch": 4.62, "learning_rate": 2.6363636363636364e-06, "loss": 6.0706, "step": 4119 }, { "epoch": 4.62, "learning_rate": 2.6287878787878787e-06, "loss": 6.1418, "step": 4120 }, { "epoch": 4.62, "learning_rate": 2.621212121212121e-06, "loss": 6.167, "step": 4121 }, { "epoch": 4.62, "learning_rate": 2.6136363636363637e-06, "loss": 6.138, "step": 4122 }, { "epoch": 4.62, "learning_rate": 2.6060606060606064e-06, "loss": 6.0118, "step": 4123 }, { "epoch": 4.62, "learning_rate": 2.5984848484848487e-06, "loss": 6.0067, "step": 4124 }, { "epoch": 4.62, "learning_rate": 2.590909090909091e-06, "loss": 6.045, "step": 4125 }, { "epoch": 4.63, "learning_rate": 2.5833333333333333e-06, "loss": 5.9513, "step": 4126 }, { "epoch": 4.63, "learning_rate": 2.5757575757575756e-06, "loss": 6.1646, "step": 4127 }, { "epoch": 4.63, "learning_rate": 2.5681818181818183e-06, "loss": 5.985, "step": 4128 }, { "epoch": 4.63, "learning_rate": 2.5606060606060606e-06, "loss": 6.0708, "step": 4129 }, { "epoch": 4.63, "learning_rate": 2.5530303030303033e-06, "loss": 5.9871, "step": 4130 }, { "epoch": 4.63, "learning_rate": 2.5454545454545456e-06, "loss": 6.0665, "step": 4131 }, { "epoch": 4.63, "learning_rate": 2.537878787878788e-06, "loss": 5.9259, "step": 4132 }, { "epoch": 4.63, "learning_rate": 2.5303030303030306e-06, "loss": 5.8849, "step": 4133 }, { "epoch": 4.63, "learning_rate": 2.522727272727273e-06, "loss": 6.211, "step": 4134 }, { "epoch": 4.64, "learning_rate": 2.515151515151515e-06, "loss": 5.9949, "step": 4135 }, { "epoch": 4.64, "learning_rate": 2.5075757575757575e-06, "loss": 6.0339, "step": 4136 }, { "epoch": 4.64, "learning_rate": 2.4999999999999998e-06, "loss": 5.8767, "step": 4137 }, { "epoch": 4.64, "learning_rate": 2.4924242424242425e-06, "loss": 5.9768, "step": 4138 }, { "epoch": 4.64, "learning_rate": 2.484848484848485e-06, "loss": 5.8958, "step": 4139 }, { "epoch": 4.64, "learning_rate": 2.4772727272727275e-06, "loss": 5.904, "step": 4140 }, { "epoch": 4.64, "learning_rate": 2.46969696969697e-06, "loss": 5.8161, "step": 4141 }, { "epoch": 4.64, "learning_rate": 2.462121212121212e-06, "loss": 5.8527, "step": 4142 }, { "epoch": 4.64, "learning_rate": 2.454545454545455e-06, "loss": 5.938, "step": 4143 }, { "epoch": 4.65, "learning_rate": 2.446969696969697e-06, "loss": 5.8682, "step": 4144 }, { "epoch": 4.65, "learning_rate": 2.4393939393939394e-06, "loss": 5.9935, "step": 4145 }, { "epoch": 4.65, "learning_rate": 2.4318181818181817e-06, "loss": 5.9548, "step": 4146 }, { "epoch": 4.65, "learning_rate": 2.4242424242424244e-06, "loss": 5.9264, "step": 4147 }, { "epoch": 4.65, "learning_rate": 2.4166666666666667e-06, "loss": 5.818, "step": 4148 }, { "epoch": 4.65, "learning_rate": 2.4090909090909094e-06, "loss": 5.8824, "step": 4149 }, { "epoch": 4.65, "learning_rate": 2.4015151515151517e-06, "loss": 5.8198, "step": 4150 }, { "epoch": 4.65, "learning_rate": 2.393939393939394e-06, "loss": 5.8051, "step": 4151 }, { "epoch": 4.65, "learning_rate": 2.3863636363636363e-06, "loss": 5.7026, "step": 4152 }, { "epoch": 4.66, "learning_rate": 2.3787878787878786e-06, "loss": 5.7879, "step": 4153 }, { "epoch": 4.66, "learning_rate": 2.3712121212121213e-06, "loss": 5.7849, "step": 4154 }, { "epoch": 4.66, "learning_rate": 2.3636363636363636e-06, "loss": 5.6779, "step": 4155 }, { "epoch": 4.66, "learning_rate": 2.3560606060606063e-06, "loss": 5.7784, "step": 4156 }, { "epoch": 4.66, "learning_rate": 2.3484848484848486e-06, "loss": 5.6773, "step": 4157 }, { "epoch": 4.66, "learning_rate": 2.340909090909091e-06, "loss": 5.7083, "step": 4158 }, { "epoch": 4.66, "learning_rate": 2.3333333333333336e-06, "loss": 5.6353, "step": 4159 }, { "epoch": 4.66, "learning_rate": 2.325757575757576e-06, "loss": 5.5358, "step": 4160 }, { "epoch": 4.66, "learning_rate": 2.318181818181818e-06, "loss": 5.6865, "step": 4161 }, { "epoch": 4.67, "learning_rate": 2.3106060606060605e-06, "loss": 5.6409, "step": 4162 }, { "epoch": 4.67, "learning_rate": 2.3030303030303027e-06, "loss": 5.5184, "step": 4163 }, { "epoch": 4.67, "learning_rate": 2.2954545454545455e-06, "loss": 5.286, "step": 4164 }, { "epoch": 4.67, "learning_rate": 2.287878787878788e-06, "loss": 5.2988, "step": 4165 }, { "epoch": 4.67, "learning_rate": 2.2803030303030305e-06, "loss": 4.8275, "step": 4166 }, { "epoch": 4.67, "learning_rate": 2.2727272727272728e-06, "loss": 4.5244, "step": 4167 }, { "epoch": 4.67, "learning_rate": 2.265151515151515e-06, "loss": 3.9256, "step": 4168 }, { "epoch": 4.67, "learning_rate": 2.2575757575757578e-06, "loss": 6.1269, "step": 4169 }, { "epoch": 4.67, "learning_rate": 2.25e-06, "loss": 6.1317, "step": 4170 }, { "epoch": 4.68, "learning_rate": 2.2424242424242423e-06, "loss": 6.1496, "step": 4171 }, { "epoch": 4.68, "learning_rate": 2.2348484848484846e-06, "loss": 6.1576, "step": 4172 }, { "epoch": 4.68, "learning_rate": 2.2272727272727274e-06, "loss": 6.1363, "step": 4173 }, { "epoch": 4.68, "learning_rate": 2.21969696969697e-06, "loss": 6.151, "step": 4174 }, { "epoch": 4.68, "learning_rate": 2.2121212121212124e-06, "loss": 6.1636, "step": 4175 }, { "epoch": 4.68, "learning_rate": 2.2045454545454547e-06, "loss": 6.0563, "step": 4176 }, { "epoch": 4.68, "learning_rate": 2.196969696969697e-06, "loss": 6.0521, "step": 4177 }, { "epoch": 4.68, "learning_rate": 2.1893939393939392e-06, "loss": 6.0082, "step": 4178 }, { "epoch": 4.68, "learning_rate": 2.181818181818182e-06, "loss": 5.9894, "step": 4179 }, { "epoch": 4.69, "learning_rate": 2.1742424242424242e-06, "loss": 5.9513, "step": 4180 }, { "epoch": 4.69, "learning_rate": 2.1666666666666665e-06, "loss": 5.8799, "step": 4181 }, { "epoch": 4.69, "learning_rate": 2.1590909090909092e-06, "loss": 6.0155, "step": 4182 }, { "epoch": 4.69, "learning_rate": 2.1515151515151515e-06, "loss": 5.9136, "step": 4183 }, { "epoch": 4.69, "learning_rate": 2.1439393939393943e-06, "loss": 6.0188, "step": 4184 }, { "epoch": 4.69, "learning_rate": 2.1363636363636365e-06, "loss": 5.9726, "step": 4185 }, { "epoch": 4.69, "learning_rate": 2.128787878787879e-06, "loss": 5.9218, "step": 4186 }, { "epoch": 4.69, "learning_rate": 2.121212121212121e-06, "loss": 5.9752, "step": 4187 }, { "epoch": 4.7, "learning_rate": 2.1136363636363634e-06, "loss": 6.0989, "step": 4188 }, { "epoch": 4.7, "learning_rate": 2.106060606060606e-06, "loss": 5.9007, "step": 4189 }, { "epoch": 4.7, "learning_rate": 2.098484848484849e-06, "loss": 5.9881, "step": 4190 }, { "epoch": 4.7, "learning_rate": 2.090909090909091e-06, "loss": 5.8722, "step": 4191 }, { "epoch": 4.7, "learning_rate": 2.0833333333333334e-06, "loss": 5.8469, "step": 4192 }, { "epoch": 4.7, "learning_rate": 2.0757575757575757e-06, "loss": 5.9824, "step": 4193 }, { "epoch": 4.7, "learning_rate": 2.0681818181818184e-06, "loss": 5.7765, "step": 4194 }, { "epoch": 4.7, "learning_rate": 2.0606060606060607e-06, "loss": 6.0604, "step": 4195 }, { "epoch": 4.7, "learning_rate": 2.053030303030303e-06, "loss": 5.882, "step": 4196 }, { "epoch": 4.71, "learning_rate": 2.0454545454545453e-06, "loss": 5.9139, "step": 4197 }, { "epoch": 4.71, "learning_rate": 2.0378787878787876e-06, "loss": 5.978, "step": 4198 }, { "epoch": 4.71, "learning_rate": 2.0303030303030303e-06, "loss": 5.9107, "step": 4199 }, { "epoch": 4.71, "learning_rate": 2.022727272727273e-06, "loss": 6.0502, "step": 4200 }, { "epoch": 4.71, "learning_rate": 2.0151515151515153e-06, "loss": 5.7611, "step": 4201 }, { "epoch": 4.71, "learning_rate": 2.0075757575757576e-06, "loss": 5.7486, "step": 4202 }, { "epoch": 4.71, "learning_rate": 2e-06, "loss": 5.8021, "step": 4203 }, { "epoch": 4.71, "learning_rate": 1.992424242424242e-06, "loss": 5.8075, "step": 4204 }, { "epoch": 4.71, "learning_rate": 1.984848484848485e-06, "loss": 5.736, "step": 4205 }, { "epoch": 4.72, "learning_rate": 1.977272727272727e-06, "loss": 5.8593, "step": 4206 }, { "epoch": 4.72, "learning_rate": 1.96969696969697e-06, "loss": 5.7307, "step": 4207 }, { "epoch": 4.72, "learning_rate": 1.9621212121212122e-06, "loss": 5.75, "step": 4208 }, { "epoch": 4.72, "learning_rate": 1.9545454545454545e-06, "loss": 5.8129, "step": 4209 }, { "epoch": 4.72, "learning_rate": 1.9469696969696972e-06, "loss": 5.7835, "step": 4210 }, { "epoch": 4.72, "learning_rate": 1.9393939393939395e-06, "loss": 5.4234, "step": 4211 }, { "epoch": 4.72, "learning_rate": 1.931818181818182e-06, "loss": 5.5873, "step": 4212 }, { "epoch": 4.72, "learning_rate": 1.924242424242424e-06, "loss": 5.2161, "step": 4213 }, { "epoch": 4.72, "learning_rate": 1.9166666666666664e-06, "loss": 5.1992, "step": 4214 }, { "epoch": 4.73, "learning_rate": 1.909090909090909e-06, "loss": 4.9386, "step": 4215 }, { "epoch": 4.73, "learning_rate": 1.9015151515151518e-06, "loss": 4.6453, "step": 4216 }, { "epoch": 4.73, "learning_rate": 1.8939393939393941e-06, "loss": 4.37, "step": 4217 }, { "epoch": 4.73, "learning_rate": 1.8863636363636364e-06, "loss": 3.7118, "step": 4218 }, { "epoch": 4.73, "learning_rate": 1.878787878787879e-06, "loss": 6.2455, "step": 4219 }, { "epoch": 4.73, "learning_rate": 1.8712121212121212e-06, "loss": 6.1829, "step": 4220 }, { "epoch": 4.73, "learning_rate": 1.8636363636363637e-06, "loss": 6.1839, "step": 4221 }, { "epoch": 4.73, "learning_rate": 1.8560606060606062e-06, "loss": 6.1917, "step": 4222 }, { "epoch": 4.73, "learning_rate": 1.8484848484848485e-06, "loss": 6.0508, "step": 4223 }, { "epoch": 4.74, "learning_rate": 1.840909090909091e-06, "loss": 5.9759, "step": 4224 }, { "epoch": 4.74, "learning_rate": 1.8333333333333333e-06, "loss": 6.2126, "step": 4225 }, { "epoch": 4.74, "learning_rate": 1.825757575757576e-06, "loss": 6.0086, "step": 4226 }, { "epoch": 4.74, "learning_rate": 1.8181818181818183e-06, "loss": 6.0312, "step": 4227 }, { "epoch": 4.74, "learning_rate": 1.8106060606060606e-06, "loss": 6.0597, "step": 4228 }, { "epoch": 4.74, "learning_rate": 1.803030303030303e-06, "loss": 6.0616, "step": 4229 }, { "epoch": 4.74, "learning_rate": 1.7954545454545454e-06, "loss": 5.9607, "step": 4230 }, { "epoch": 4.74, "learning_rate": 1.787878787878788e-06, "loss": 5.876, "step": 4231 }, { "epoch": 4.74, "learning_rate": 1.7803030303030304e-06, "loss": 5.9563, "step": 4232 }, { "epoch": 4.75, "learning_rate": 1.7727272727272727e-06, "loss": 5.8637, "step": 4233 }, { "epoch": 4.75, "learning_rate": 1.7651515151515152e-06, "loss": 5.9113, "step": 4234 }, { "epoch": 4.75, "learning_rate": 1.7575757575757577e-06, "loss": 5.9517, "step": 4235 }, { "epoch": 4.75, "learning_rate": 1.75e-06, "loss": 5.9473, "step": 4236 }, { "epoch": 4.75, "learning_rate": 1.7424242424242425e-06, "loss": 6.0238, "step": 4237 }, { "epoch": 4.75, "learning_rate": 1.7348484848484848e-06, "loss": 5.8729, "step": 4238 }, { "epoch": 4.75, "learning_rate": 1.7272727272727275e-06, "loss": 5.8971, "step": 4239 }, { "epoch": 4.75, "learning_rate": 1.7196969696969698e-06, "loss": 5.8575, "step": 4240 }, { "epoch": 4.75, "learning_rate": 1.712121212121212e-06, "loss": 5.7706, "step": 4241 }, { "epoch": 4.76, "learning_rate": 1.7045454545454546e-06, "loss": 5.9286, "step": 4242 }, { "epoch": 4.76, "learning_rate": 1.696969696969697e-06, "loss": 5.9477, "step": 4243 }, { "epoch": 4.76, "learning_rate": 1.6893939393939396e-06, "loss": 5.9668, "step": 4244 }, { "epoch": 4.76, "learning_rate": 1.6818181818181819e-06, "loss": 5.9684, "step": 4245 }, { "epoch": 4.76, "learning_rate": 1.6742424242424242e-06, "loss": 5.9264, "step": 4246 }, { "epoch": 4.76, "learning_rate": 1.6666666666666667e-06, "loss": 5.8608, "step": 4247 }, { "epoch": 4.76, "learning_rate": 1.6590909090909092e-06, "loss": 5.7802, "step": 4248 }, { "epoch": 4.76, "learning_rate": 1.6515151515151517e-06, "loss": 5.7884, "step": 4249 }, { "epoch": 4.76, "learning_rate": 1.643939393939394e-06, "loss": 5.8461, "step": 4250 }, { "epoch": 4.77, "learning_rate": 1.6363636363636363e-06, "loss": 5.8566, "step": 4251 }, { "epoch": 4.77, "learning_rate": 1.628787878787879e-06, "loss": 5.8185, "step": 4252 }, { "epoch": 4.77, "learning_rate": 1.6212121212121213e-06, "loss": 5.8064, "step": 4253 }, { "epoch": 4.77, "learning_rate": 1.6136363636363638e-06, "loss": 5.7483, "step": 4254 }, { "epoch": 4.77, "learning_rate": 1.606060606060606e-06, "loss": 5.7328, "step": 4255 }, { "epoch": 4.77, "learning_rate": 1.5984848484848486e-06, "loss": 5.7705, "step": 4256 }, { "epoch": 4.77, "learning_rate": 1.590909090909091e-06, "loss": 5.5803, "step": 4257 }, { "epoch": 4.77, "learning_rate": 1.5833333333333333e-06, "loss": 5.6551, "step": 4258 }, { "epoch": 4.77, "learning_rate": 1.5757575757575756e-06, "loss": 5.784, "step": 4259 }, { "epoch": 4.78, "learning_rate": 1.5681818181818181e-06, "loss": 5.5734, "step": 4260 }, { "epoch": 4.78, "learning_rate": 1.5606060606060606e-06, "loss": 5.5223, "step": 4261 }, { "epoch": 4.78, "learning_rate": 1.5530303030303032e-06, "loss": 5.2248, "step": 4262 }, { "epoch": 4.78, "learning_rate": 1.5454545454545454e-06, "loss": 5.2213, "step": 4263 }, { "epoch": 4.78, "learning_rate": 1.5378787878787877e-06, "loss": 5.1248, "step": 4264 }, { "epoch": 4.78, "learning_rate": 1.5303030303030304e-06, "loss": 5.0354, "step": 4265 }, { "epoch": 4.78, "learning_rate": 1.5227272727272727e-06, "loss": 4.7034, "step": 4266 }, { "epoch": 4.78, "learning_rate": 1.5151515151515152e-06, "loss": 4.3145, "step": 4267 }, { "epoch": 4.78, "learning_rate": 1.5075757575757575e-06, "loss": 3.3278, "step": 4268 }, { "epoch": 4.79, "learning_rate": 1.5e-06, "loss": 6.1039, "step": 4269 }, { "epoch": 4.79, "learning_rate": 1.4924242424242425e-06, "loss": 6.1825, "step": 4270 }, { "epoch": 4.79, "learning_rate": 1.4848484848484848e-06, "loss": 6.1063, "step": 4271 }, { "epoch": 4.79, "learning_rate": 1.4772727272727273e-06, "loss": 6.0015, "step": 4272 }, { "epoch": 4.79, "learning_rate": 1.4696969696969698e-06, "loss": 6.1495, "step": 4273 }, { "epoch": 4.79, "learning_rate": 1.4621212121212121e-06, "loss": 6.1767, "step": 4274 }, { "epoch": 4.79, "learning_rate": 1.4545454545454546e-06, "loss": 6.0219, "step": 4275 }, { "epoch": 4.79, "learning_rate": 1.446969696969697e-06, "loss": 6.1531, "step": 4276 }, { "epoch": 4.79, "learning_rate": 1.4393939393939394e-06, "loss": 5.986, "step": 4277 }, { "epoch": 4.8, "learning_rate": 1.431818181818182e-06, "loss": 5.9164, "step": 4278 }, { "epoch": 4.8, "learning_rate": 1.4242424242424242e-06, "loss": 5.8912, "step": 4279 }, { "epoch": 4.8, "learning_rate": 1.4166666666666667e-06, "loss": 6.067, "step": 4280 }, { "epoch": 4.8, "learning_rate": 1.409090909090909e-06, "loss": 5.9407, "step": 4281 }, { "epoch": 4.8, "learning_rate": 1.4015151515151515e-06, "loss": 5.8927, "step": 4282 }, { "epoch": 4.8, "learning_rate": 1.393939393939394e-06, "loss": 6.1013, "step": 4283 }, { "epoch": 4.8, "learning_rate": 1.3863636363636363e-06, "loss": 5.839, "step": 4284 }, { "epoch": 4.8, "learning_rate": 1.3787878787878788e-06, "loss": 5.9786, "step": 4285 }, { "epoch": 4.8, "learning_rate": 1.3712121212121213e-06, "loss": 5.8447, "step": 4286 }, { "epoch": 4.81, "learning_rate": 1.3636363636363636e-06, "loss": 5.9586, "step": 4287 }, { "epoch": 4.81, "learning_rate": 1.3560606060606061e-06, "loss": 5.9137, "step": 4288 }, { "epoch": 4.81, "learning_rate": 1.3484848484848484e-06, "loss": 5.9352, "step": 4289 }, { "epoch": 4.81, "learning_rate": 1.3409090909090911e-06, "loss": 5.9197, "step": 4290 }, { "epoch": 4.81, "learning_rate": 1.3333333333333334e-06, "loss": 5.844, "step": 4291 }, { "epoch": 4.81, "learning_rate": 1.3257575757575757e-06, "loss": 5.8715, "step": 4292 }, { "epoch": 4.81, "learning_rate": 1.3181818181818182e-06, "loss": 5.9229, "step": 4293 }, { "epoch": 4.81, "learning_rate": 1.3106060606060605e-06, "loss": 5.8181, "step": 4294 }, { "epoch": 4.82, "learning_rate": 1.3030303030303032e-06, "loss": 5.7632, "step": 4295 }, { "epoch": 4.82, "learning_rate": 1.2954545454545455e-06, "loss": 5.9798, "step": 4296 }, { "epoch": 4.82, "learning_rate": 1.2878787878787878e-06, "loss": 5.8335, "step": 4297 }, { "epoch": 4.82, "learning_rate": 1.2803030303030303e-06, "loss": 5.9716, "step": 4298 }, { "epoch": 4.82, "learning_rate": 1.2727272727272728e-06, "loss": 5.8307, "step": 4299 }, { "epoch": 4.82, "learning_rate": 1.2651515151515153e-06, "loss": 5.6712, "step": 4300 }, { "epoch": 4.82, "learning_rate": 1.2575757575757576e-06, "loss": 5.8163, "step": 4301 }, { "epoch": 4.82, "learning_rate": 1.2499999999999999e-06, "loss": 5.785, "step": 4302 }, { "epoch": 4.82, "learning_rate": 1.2424242424242426e-06, "loss": 5.9307, "step": 4303 }, { "epoch": 4.83, "learning_rate": 1.234848484848485e-06, "loss": 5.7519, "step": 4304 }, { "epoch": 4.83, "learning_rate": 1.2272727272727274e-06, "loss": 5.7651, "step": 4305 }, { "epoch": 4.83, "learning_rate": 1.2196969696969697e-06, "loss": 5.8788, "step": 4306 }, { "epoch": 4.83, "learning_rate": 1.2121212121212122e-06, "loss": 5.6922, "step": 4307 }, { "epoch": 4.83, "learning_rate": 1.2045454545454547e-06, "loss": 5.6885, "step": 4308 }, { "epoch": 4.83, "learning_rate": 1.196969696969697e-06, "loss": 5.5855, "step": 4309 }, { "epoch": 4.83, "learning_rate": 1.1893939393939393e-06, "loss": 5.4312, "step": 4310 }, { "epoch": 4.83, "learning_rate": 1.1818181818181818e-06, "loss": 5.5097, "step": 4311 }, { "epoch": 4.83, "learning_rate": 1.1742424242424243e-06, "loss": 5.5256, "step": 4312 }, { "epoch": 4.84, "learning_rate": 1.1666666666666668e-06, "loss": 5.3268, "step": 4313 }, { "epoch": 4.84, "learning_rate": 1.159090909090909e-06, "loss": 5.3588, "step": 4314 }, { "epoch": 4.84, "learning_rate": 1.1515151515151514e-06, "loss": 4.7586, "step": 4315 }, { "epoch": 4.84, "learning_rate": 1.143939393939394e-06, "loss": 4.8012, "step": 4316 }, { "epoch": 4.84, "learning_rate": 1.1363636363636364e-06, "loss": 4.458, "step": 4317 }, { "epoch": 4.84, "learning_rate": 1.1287878787878789e-06, "loss": 3.7137, "step": 4318 }, { "epoch": 4.84, "learning_rate": 1.1212121212121212e-06, "loss": 6.1743, "step": 4319 }, { "epoch": 4.84, "learning_rate": 1.1136363636363637e-06, "loss": 6.0547, "step": 4320 }, { "epoch": 4.84, "learning_rate": 1.1060606060606062e-06, "loss": 5.9966, "step": 4321 }, { "epoch": 4.85, "learning_rate": 1.0984848484848485e-06, "loss": 6.2657, "step": 4322 }, { "epoch": 4.85, "learning_rate": 1.090909090909091e-06, "loss": 6.0092, "step": 4323 }, { "epoch": 4.85, "learning_rate": 1.0833333333333333e-06, "loss": 6.1257, "step": 4324 }, { "epoch": 4.85, "learning_rate": 1.0757575757575758e-06, "loss": 6.1605, "step": 4325 }, { "epoch": 4.85, "learning_rate": 1.0681818181818183e-06, "loss": 6.011, "step": 4326 }, { "epoch": 4.85, "learning_rate": 1.0606060606060606e-06, "loss": 6.0989, "step": 4327 }, { "epoch": 4.85, "learning_rate": 1.053030303030303e-06, "loss": 5.9341, "step": 4328 }, { "epoch": 4.85, "learning_rate": 1.0454545454545456e-06, "loss": 6.0907, "step": 4329 }, { "epoch": 4.85, "learning_rate": 1.0378787878787879e-06, "loss": 6.0903, "step": 4330 }, { "epoch": 4.86, "learning_rate": 1.0303030303030304e-06, "loss": 5.8949, "step": 4331 }, { "epoch": 4.86, "learning_rate": 1.0227272727272727e-06, "loss": 5.9525, "step": 4332 }, { "epoch": 4.86, "learning_rate": 1.0151515151515152e-06, "loss": 6.0088, "step": 4333 }, { "epoch": 4.86, "learning_rate": 1.0075757575757577e-06, "loss": 5.9159, "step": 4334 }, { "epoch": 4.86, "learning_rate": 1e-06, "loss": 6.0243, "step": 4335 }, { "epoch": 4.86, "learning_rate": 9.924242424242425e-07, "loss": 5.7624, "step": 4336 }, { "epoch": 4.86, "learning_rate": 9.84848484848485e-07, "loss": 5.9382, "step": 4337 }, { "epoch": 4.86, "learning_rate": 9.772727272727273e-07, "loss": 5.9658, "step": 4338 }, { "epoch": 4.86, "learning_rate": 9.696969696969698e-07, "loss": 5.8132, "step": 4339 }, { "epoch": 4.87, "learning_rate": 9.62121212121212e-07, "loss": 5.975, "step": 4340 }, { "epoch": 4.87, "learning_rate": 9.545454545454546e-07, "loss": 5.8409, "step": 4341 }, { "epoch": 4.87, "learning_rate": 9.469696969696971e-07, "loss": 6.0587, "step": 4342 }, { "epoch": 4.87, "learning_rate": 9.393939393939395e-07, "loss": 5.8754, "step": 4343 }, { "epoch": 4.87, "learning_rate": 9.318181818181818e-07, "loss": 5.8916, "step": 4344 }, { "epoch": 4.87, "learning_rate": 9.242424242424242e-07, "loss": 6.0487, "step": 4345 }, { "epoch": 4.87, "learning_rate": 9.166666666666666e-07, "loss": 5.9416, "step": 4346 }, { "epoch": 4.87, "learning_rate": 9.090909090909091e-07, "loss": 5.8493, "step": 4347 }, { "epoch": 4.87, "learning_rate": 9.015151515151515e-07, "loss": 5.8351, "step": 4348 }, { "epoch": 4.88, "learning_rate": 8.93939393939394e-07, "loss": 6.0182, "step": 4349 }, { "epoch": 4.88, "learning_rate": 8.863636363636363e-07, "loss": 5.949, "step": 4350 }, { "epoch": 4.88, "learning_rate": 8.787878787878788e-07, "loss": 5.8944, "step": 4351 }, { "epoch": 4.88, "learning_rate": 8.712121212121212e-07, "loss": 5.8651, "step": 4352 }, { "epoch": 4.88, "learning_rate": 8.636363636363637e-07, "loss": 5.7474, "step": 4353 }, { "epoch": 4.88, "learning_rate": 8.56060606060606e-07, "loss": 5.8869, "step": 4354 }, { "epoch": 4.88, "learning_rate": 8.484848484848485e-07, "loss": 5.9369, "step": 4355 }, { "epoch": 4.88, "learning_rate": 8.409090909090909e-07, "loss": 5.782, "step": 4356 }, { "epoch": 4.88, "learning_rate": 8.333333333333333e-07, "loss": 5.8052, "step": 4357 }, { "epoch": 4.89, "learning_rate": 8.257575757575758e-07, "loss": 5.5784, "step": 4358 }, { "epoch": 4.89, "learning_rate": 8.181818181818181e-07, "loss": 5.5387, "step": 4359 }, { "epoch": 4.89, "learning_rate": 8.106060606060606e-07, "loss": 5.5537, "step": 4360 }, { "epoch": 4.89, "learning_rate": 8.03030303030303e-07, "loss": 5.4226, "step": 4361 }, { "epoch": 4.89, "learning_rate": 7.954545454545455e-07, "loss": 5.1039, "step": 4362 }, { "epoch": 4.89, "learning_rate": 7.878787878787878e-07, "loss": 5.235, "step": 4363 }, { "epoch": 4.89, "learning_rate": 7.803030303030303e-07, "loss": 5.1453, "step": 4364 }, { "epoch": 4.89, "learning_rate": 7.727272727272727e-07, "loss": 5.1506, "step": 4365 }, { "epoch": 4.89, "learning_rate": 7.651515151515152e-07, "loss": 4.6338, "step": 4366 }, { "epoch": 4.9, "learning_rate": 7.575757575757576e-07, "loss": 4.4236, "step": 4367 }, { "epoch": 4.9, "learning_rate": 7.5e-07, "loss": 3.8034, "step": 4368 }, { "epoch": 4.9, "learning_rate": 7.424242424242424e-07, "loss": 6.2571, "step": 4369 }, { "epoch": 4.9, "learning_rate": 7.348484848484849e-07, "loss": 6.0681, "step": 4370 }, { "epoch": 4.9, "learning_rate": 7.272727272727273e-07, "loss": 6.0155, "step": 4371 }, { "epoch": 4.9, "learning_rate": 7.196969696969697e-07, "loss": 6.0834, "step": 4372 }, { "epoch": 4.9, "learning_rate": 7.121212121212121e-07, "loss": 6.0411, "step": 4373 }, { "epoch": 4.9, "learning_rate": 7.045454545454545e-07, "loss": 6.0578, "step": 4374 }, { "epoch": 4.9, "learning_rate": 6.96969696969697e-07, "loss": 6.1776, "step": 4375 }, { "epoch": 4.91, "learning_rate": 6.893939393939394e-07, "loss": 5.9837, "step": 4376 }, { "epoch": 4.91, "learning_rate": 6.818181818181818e-07, "loss": 6.0588, "step": 4377 }, { "epoch": 4.91, "learning_rate": 6.742424242424242e-07, "loss": 5.9814, "step": 4378 }, { "epoch": 4.91, "learning_rate": 6.666666666666667e-07, "loss": 6.0949, "step": 4379 }, { "epoch": 4.91, "learning_rate": 6.590909090909091e-07, "loss": 5.8556, "step": 4380 }, { "epoch": 4.91, "learning_rate": 6.515151515151516e-07, "loss": 5.8973, "step": 4381 }, { "epoch": 4.91, "learning_rate": 6.439393939393939e-07, "loss": 5.9929, "step": 4382 }, { "epoch": 4.91, "learning_rate": 6.363636363636364e-07, "loss": 5.9753, "step": 4383 }, { "epoch": 4.91, "learning_rate": 6.287878787878788e-07, "loss": 6.1357, "step": 4384 }, { "epoch": 4.92, "learning_rate": 6.212121212121213e-07, "loss": 6.0967, "step": 4385 }, { "epoch": 4.92, "learning_rate": 6.136363636363637e-07, "loss": 5.8857, "step": 4386 }, { "epoch": 4.92, "learning_rate": 6.060606060606061e-07, "loss": 5.8596, "step": 4387 }, { "epoch": 4.92, "learning_rate": 5.984848484848485e-07, "loss": 5.8176, "step": 4388 }, { "epoch": 4.92, "learning_rate": 5.909090909090909e-07, "loss": 5.7731, "step": 4389 }, { "epoch": 4.92, "learning_rate": 5.833333333333334e-07, "loss": 5.8458, "step": 4390 }, { "epoch": 4.92, "learning_rate": 5.757575757575757e-07, "loss": 5.9058, "step": 4391 }, { "epoch": 4.92, "learning_rate": 5.681818181818182e-07, "loss": 6.0349, "step": 4392 }, { "epoch": 4.92, "learning_rate": 5.606060606060606e-07, "loss": 5.8406, "step": 4393 }, { "epoch": 4.93, "learning_rate": 5.530303030303031e-07, "loss": 5.8716, "step": 4394 }, { "epoch": 4.93, "learning_rate": 5.454545454545455e-07, "loss": 5.9578, "step": 4395 }, { "epoch": 4.93, "learning_rate": 5.378787878787879e-07, "loss": 5.8902, "step": 4396 }, { "epoch": 4.93, "learning_rate": 5.303030303030303e-07, "loss": 5.7837, "step": 4397 }, { "epoch": 4.93, "learning_rate": 5.227272727272728e-07, "loss": 5.7964, "step": 4398 }, { "epoch": 4.93, "learning_rate": 5.151515151515152e-07, "loss": 5.8155, "step": 4399 }, { "epoch": 4.93, "learning_rate": 5.075757575757576e-07, "loss": 5.9266, "step": 4400 }, { "epoch": 4.93, "learning_rate": 5e-07, "loss": 5.8827, "step": 4401 }, { "epoch": 4.93, "learning_rate": 4.924242424242425e-07, "loss": 5.7171, "step": 4402 }, { "epoch": 4.94, "learning_rate": 4.848484848484849e-07, "loss": 5.7663, "step": 4403 }, { "epoch": 4.94, "learning_rate": 4.772727272727273e-07, "loss": 5.7453, "step": 4404 }, { "epoch": 4.94, "learning_rate": 4.696969696969697e-07, "loss": 5.822, "step": 4405 }, { "epoch": 4.94, "learning_rate": 4.621212121212121e-07, "loss": 5.785, "step": 4406 }, { "epoch": 4.94, "learning_rate": 4.5454545454545457e-07, "loss": 5.7835, "step": 4407 }, { "epoch": 4.94, "learning_rate": 4.46969696969697e-07, "loss": 5.7226, "step": 4408 }, { "epoch": 4.94, "learning_rate": 4.393939393939394e-07, "loss": 5.6343, "step": 4409 }, { "epoch": 4.94, "learning_rate": 4.3181818181818187e-07, "loss": 5.4581, "step": 4410 }, { "epoch": 4.95, "learning_rate": 4.2424242424242427e-07, "loss": 5.5606, "step": 4411 }, { "epoch": 4.95, "learning_rate": 4.1666666666666667e-07, "loss": 5.4199, "step": 4412 }, { "epoch": 4.95, "learning_rate": 4.0909090909090906e-07, "loss": 5.2994, "step": 4413 }, { "epoch": 4.95, "learning_rate": 4.015151515151515e-07, "loss": 4.9215, "step": 4414 }, { "epoch": 4.95, "learning_rate": 3.939393939393939e-07, "loss": 4.7728, "step": 4415 }, { "epoch": 4.95, "learning_rate": 3.8636363636363636e-07, "loss": 4.6235, "step": 4416 }, { "epoch": 4.95, "learning_rate": 3.787878787878788e-07, "loss": 4.2827, "step": 4417 }, { "epoch": 4.95, "learning_rate": 3.712121212121212e-07, "loss": 3.6713, "step": 4418 }, { "epoch": 4.95, "learning_rate": 3.6363636363636366e-07, "loss": 5.9763, "step": 4419 }, { "epoch": 4.96, "learning_rate": 3.5606060606060606e-07, "loss": 6.1134, "step": 4420 }, { "epoch": 4.96, "learning_rate": 3.484848484848485e-07, "loss": 6.1949, "step": 4421 }, { "epoch": 4.96, "learning_rate": 3.409090909090909e-07, "loss": 6.0498, "step": 4422 }, { "epoch": 4.96, "learning_rate": 3.3333333333333335e-07, "loss": 5.951, "step": 4423 }, { "epoch": 4.96, "learning_rate": 3.257575757575758e-07, "loss": 6.0914, "step": 4424 }, { "epoch": 4.96, "learning_rate": 3.181818181818182e-07, "loss": 5.924, "step": 4425 }, { "epoch": 4.96, "learning_rate": 3.1060606060606065e-07, "loss": 6.0032, "step": 4426 }, { "epoch": 4.96, "learning_rate": 3.0303030303030305e-07, "loss": 5.9572, "step": 4427 }, { "epoch": 4.96, "learning_rate": 2.9545454545454545e-07, "loss": 5.8292, "step": 4428 }, { "epoch": 4.97, "learning_rate": 2.8787878787878784e-07, "loss": 5.9096, "step": 4429 }, { "epoch": 4.97, "learning_rate": 2.803030303030303e-07, "loss": 5.9862, "step": 4430 }, { "epoch": 4.97, "learning_rate": 2.7272727272727274e-07, "loss": 6.0106, "step": 4431 }, { "epoch": 4.97, "learning_rate": 2.6515151515151514e-07, "loss": 5.9465, "step": 4432 }, { "epoch": 4.97, "learning_rate": 2.575757575757576e-07, "loss": 5.959, "step": 4433 }, { "epoch": 4.97, "learning_rate": 2.5e-07, "loss": 5.9248, "step": 4434 }, { "epoch": 4.97, "learning_rate": 2.4242424242424244e-07, "loss": 5.9356, "step": 4435 }, { "epoch": 4.97, "learning_rate": 2.3484848484848486e-07, "loss": 5.9166, "step": 4436 }, { "epoch": 4.97, "learning_rate": 2.2727272727272729e-07, "loss": 5.7964, "step": 4437 }, { "epoch": 4.98, "learning_rate": 2.196969696969697e-07, "loss": 6.048, "step": 4438 }, { "epoch": 4.98, "learning_rate": 2.1212121212121213e-07, "loss": 5.8211, "step": 4439 }, { "epoch": 4.98, "learning_rate": 2.0454545454545453e-07, "loss": 5.9737, "step": 4440 }, { "epoch": 4.98, "learning_rate": 1.9696969696969696e-07, "loss": 5.8772, "step": 4441 }, { "epoch": 4.98, "learning_rate": 1.893939393939394e-07, "loss": 5.987, "step": 4442 }, { "epoch": 4.98, "learning_rate": 1.8181818181818183e-07, "loss": 5.9606, "step": 4443 }, { "epoch": 4.98, "learning_rate": 1.7424242424242425e-07, "loss": 5.8327, "step": 4444 }, { "epoch": 4.98, "learning_rate": 1.6666666666666668e-07, "loss": 5.7469, "step": 4445 }, { "epoch": 4.98, "learning_rate": 1.590909090909091e-07, "loss": 5.8319, "step": 4446 }, { "epoch": 4.99, "learning_rate": 1.5151515151515152e-07, "loss": 5.9524, "step": 4447 }, { "epoch": 4.99, "learning_rate": 1.4393939393939392e-07, "loss": 5.9809, "step": 4448 }, { "epoch": 4.99, "learning_rate": 1.3636363636363637e-07, "loss": 5.686, "step": 4449 }, { "epoch": 4.99, "learning_rate": 1.287878787878788e-07, "loss": 5.7107, "step": 4450 }, { "epoch": 4.99, "learning_rate": 1.2121212121212122e-07, "loss": 5.6012, "step": 4451 }, { "epoch": 4.99, "learning_rate": 1.1363636363636364e-07, "loss": 5.8222, "step": 4452 }, { "epoch": 4.99, "learning_rate": 1.0606060606060607e-07, "loss": 5.7081, "step": 4453 }, { "epoch": 4.99, "learning_rate": 9.848484848484848e-08, "loss": 5.5819, "step": 4454 }, { "epoch": 4.99, "learning_rate": 9.090909090909091e-08, "loss": 5.4983, "step": 4455 }, { "epoch": 5.0, "learning_rate": 8.333333333333334e-08, "loss": 5.3402, "step": 4456 }, { "epoch": 5.0, "learning_rate": 7.575757575757576e-08, "loss": 4.9594, "step": 4457 }, { "epoch": 5.0, "learning_rate": 6.818181818181819e-08, "loss": 4.7658, "step": 4458 }, { "epoch": 5.0, "learning_rate": 6.060606060606061e-08, "loss": 4.793, "step": 4459 }, { "epoch": 5.0, "learning_rate": 5.3030303030303034e-08, "loss": 3.1112, "step": 4460 }, { "epoch": 5.0, "step": 4460, "total_flos": 0.0, "train_loss": 6.37029180826093, "train_runtime": 31215.1698, "train_samples_per_second": 4.571, "train_steps_per_second": 0.143 } ], "max_steps": 4460, "num_train_epochs": 5, "total_flos": 0.0, "trial_name": null, "trial_params": null }