{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.26666666666666666, "eval_steps": 500, "global_step": 8000, "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": 1.9186, "step": 1 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 2.0395, "step": 2 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 1.9127, "step": 3 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 1.9807, "step": 4 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 1.9436, "step": 5 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 1.9582, "step": 6 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 1.9097, "step": 7 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 2.0254, "step": 8 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 1.9789, "step": 9 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 1.9641, "step": 10 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 1.951, "step": 11 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 1.9251, "step": 12 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 1.977, "step": 13 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 1.9826, "step": 14 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 1.9144, "step": 15 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 1.9076, "step": 16 }, { "epoch": 0.0, "learning_rate": 6.666666666666668e-08, "loss": 1.9383, "step": 17 }, { "epoch": 0.0, "learning_rate": 1.3333333333333336e-07, "loss": 2.0597, "step": 18 }, { "epoch": 0.0, "learning_rate": 2.0000000000000002e-07, "loss": 2.0222, "step": 19 }, { "epoch": 0.0, "learning_rate": 2.666666666666667e-07, "loss": 1.9147, "step": 20 }, { "epoch": 0.0, "learning_rate": 3.3333333333333335e-07, "loss": 1.9245, "step": 21 }, { "epoch": 0.0, "learning_rate": 3.3333333333333335e-07, "loss": 1.8994, "step": 22 }, { "epoch": 0.0, "learning_rate": 4.0000000000000003e-07, "loss": 1.9065, "step": 23 }, { "epoch": 0.0, "learning_rate": 4.666666666666667e-07, "loss": 1.9573, "step": 24 }, { "epoch": 0.0, "learning_rate": 5.333333333333335e-07, "loss": 1.9613, "step": 25 }, { "epoch": 0.0, "learning_rate": 6.000000000000001e-07, "loss": 1.9522, "step": 26 }, { "epoch": 0.0, "learning_rate": 6.000000000000001e-07, "loss": 1.9184, "step": 27 }, { "epoch": 0.0, "learning_rate": 6.666666666666667e-07, "loss": 1.9304, "step": 28 }, { "epoch": 0.0, "learning_rate": 7.333333333333334e-07, "loss": 1.9333, "step": 29 }, { "epoch": 0.0, "learning_rate": 8.000000000000001e-07, "loss": 1.9469, "step": 30 }, { "epoch": 0.0, "learning_rate": 8.666666666666668e-07, "loss": 1.978, "step": 31 }, { "epoch": 0.0, "learning_rate": 9.333333333333334e-07, "loss": 1.8763, "step": 32 }, { "epoch": 0.0, "learning_rate": 1.0000000000000002e-06, "loss": 1.9103, "step": 33 }, { "epoch": 0.0, "learning_rate": 1.066666666666667e-06, "loss": 1.8767, "step": 34 }, { "epoch": 0.0, "learning_rate": 1.1333333333333334e-06, "loss": 1.8744, "step": 35 }, { "epoch": 0.0, "learning_rate": 1.2000000000000002e-06, "loss": 1.9412, "step": 36 }, { "epoch": 0.0, "learning_rate": 1.2666666666666669e-06, "loss": 1.9311, "step": 37 }, { "epoch": 0.0, "learning_rate": 1.3333333333333334e-06, "loss": 1.9707, "step": 38 }, { "epoch": 0.0, "learning_rate": 1.4000000000000001e-06, "loss": 1.8933, "step": 39 }, { "epoch": 0.0, "learning_rate": 1.4666666666666669e-06, "loss": 1.9578, "step": 40 }, { "epoch": 0.0, "learning_rate": 1.5333333333333334e-06, "loss": 1.8903, "step": 41 }, { "epoch": 0.0, "learning_rate": 1.6000000000000001e-06, "loss": 1.9263, "step": 42 }, { "epoch": 0.0, "learning_rate": 1.6666666666666667e-06, "loss": 1.949, "step": 43 }, { "epoch": 0.0, "learning_rate": 1.7333333333333336e-06, "loss": 1.8755, "step": 44 }, { "epoch": 0.0, "learning_rate": 1.8000000000000001e-06, "loss": 1.8278, "step": 45 }, { "epoch": 0.0, "learning_rate": 1.8666666666666669e-06, "loss": 1.8547, "step": 46 }, { "epoch": 0.0, "learning_rate": 1.9333333333333336e-06, "loss": 1.8621, "step": 47 }, { "epoch": 0.0, "learning_rate": 2.0000000000000003e-06, "loss": 1.9388, "step": 48 }, { "epoch": 0.0, "learning_rate": 2.0666666666666666e-06, "loss": 1.8703, "step": 49 }, { "epoch": 0.0, "learning_rate": 2.133333333333334e-06, "loss": 1.8184, "step": 50 }, { "epoch": 0.0, "learning_rate": 2.2e-06, "loss": 1.7728, "step": 51 }, { "epoch": 0.0, "learning_rate": 2.266666666666667e-06, "loss": 1.7851, "step": 52 }, { "epoch": 0.0, "learning_rate": 2.3333333333333336e-06, "loss": 1.7908, "step": 53 }, { "epoch": 0.0, "learning_rate": 2.4000000000000003e-06, "loss": 1.8103, "step": 54 }, { "epoch": 0.0, "learning_rate": 2.466666666666667e-06, "loss": 1.832, "step": 55 }, { "epoch": 0.0, "learning_rate": 2.5333333333333338e-06, "loss": 1.7993, "step": 56 }, { "epoch": 0.0, "learning_rate": 2.6e-06, "loss": 1.8165, "step": 57 }, { "epoch": 0.0, "learning_rate": 2.666666666666667e-06, "loss": 1.7836, "step": 58 }, { "epoch": 0.0, "learning_rate": 2.7333333333333336e-06, "loss": 1.8548, "step": 59 }, { "epoch": 0.0, "learning_rate": 2.8000000000000003e-06, "loss": 1.7843, "step": 60 }, { "epoch": 0.0, "learning_rate": 2.866666666666667e-06, "loss": 1.8091, "step": 61 }, { "epoch": 0.0, "learning_rate": 2.9333333333333338e-06, "loss": 1.7808, "step": 62 }, { "epoch": 0.0, "learning_rate": 3e-06, "loss": 1.7489, "step": 63 }, { "epoch": 0.0, "learning_rate": 3.066666666666667e-06, "loss": 1.7997, "step": 64 }, { "epoch": 0.0, "learning_rate": 3.133333333333334e-06, "loss": 1.816, "step": 65 }, { "epoch": 0.0, "learning_rate": 3.2000000000000003e-06, "loss": 1.8102, "step": 66 }, { "epoch": 0.0, "learning_rate": 3.266666666666667e-06, "loss": 1.8187, "step": 67 }, { "epoch": 0.0, "learning_rate": 3.3333333333333333e-06, "loss": 1.8227, "step": 68 }, { "epoch": 0.0, "learning_rate": 3.4000000000000005e-06, "loss": 1.8159, "step": 69 }, { "epoch": 0.0, "learning_rate": 3.4666666666666672e-06, "loss": 1.7707, "step": 70 }, { "epoch": 0.0, "learning_rate": 3.5333333333333335e-06, "loss": 1.7788, "step": 71 }, { "epoch": 0.0, "learning_rate": 3.6000000000000003e-06, "loss": 1.7707, "step": 72 }, { "epoch": 0.0, "learning_rate": 3.6666666666666666e-06, "loss": 1.8053, "step": 73 }, { "epoch": 0.0, "learning_rate": 3.7333333333333337e-06, "loss": 1.8407, "step": 74 }, { "epoch": 0.0, "learning_rate": 3.8000000000000005e-06, "loss": 1.7704, "step": 75 }, { "epoch": 0.0, "learning_rate": 3.866666666666667e-06, "loss": 1.751, "step": 76 }, { "epoch": 0.0, "learning_rate": 3.9333333333333335e-06, "loss": 1.7764, "step": 77 }, { "epoch": 0.0, "learning_rate": 4.000000000000001e-06, "loss": 1.7225, "step": 78 }, { "epoch": 0.0, "learning_rate": 4.066666666666667e-06, "loss": 1.7711, "step": 79 }, { "epoch": 0.0, "learning_rate": 4.133333333333333e-06, "loss": 1.7279, "step": 80 }, { "epoch": 0.0, "learning_rate": 4.2000000000000004e-06, "loss": 1.7831, "step": 81 }, { "epoch": 0.0, "learning_rate": 4.266666666666668e-06, "loss": 1.7181, "step": 82 }, { "epoch": 0.0, "learning_rate": 4.333333333333334e-06, "loss": 1.7033, "step": 83 }, { "epoch": 0.0, "learning_rate": 4.4e-06, "loss": 1.769, "step": 84 }, { "epoch": 0.0, "learning_rate": 4.4666666666666665e-06, "loss": 1.7694, "step": 85 }, { "epoch": 0.0, "learning_rate": 4.533333333333334e-06, "loss": 1.7404, "step": 86 }, { "epoch": 0.0, "learning_rate": 4.600000000000001e-06, "loss": 1.7694, "step": 87 }, { "epoch": 0.0, "learning_rate": 4.666666666666667e-06, "loss": 1.7976, "step": 88 }, { "epoch": 0.0, "learning_rate": 4.7333333333333335e-06, "loss": 1.7287, "step": 89 }, { "epoch": 0.0, "learning_rate": 4.800000000000001e-06, "loss": 1.7924, "step": 90 }, { "epoch": 0.0, "learning_rate": 4.866666666666667e-06, "loss": 1.7566, "step": 91 }, { "epoch": 0.0, "learning_rate": 4.933333333333334e-06, "loss": 1.7973, "step": 92 }, { "epoch": 0.0, "learning_rate": 5e-06, "loss": 1.7102, "step": 93 }, { "epoch": 0.0, "learning_rate": 5.0666666666666676e-06, "loss": 1.7334, "step": 94 }, { "epoch": 0.0, "learning_rate": 5.133333333333334e-06, "loss": 1.7707, "step": 95 }, { "epoch": 0.0, "learning_rate": 5.2e-06, "loss": 1.7868, "step": 96 }, { "epoch": 0.0, "learning_rate": 5.2666666666666665e-06, "loss": 1.7444, "step": 97 }, { "epoch": 0.0, "learning_rate": 5.333333333333334e-06, "loss": 1.7692, "step": 98 }, { "epoch": 0.0, "learning_rate": 5.400000000000001e-06, "loss": 1.7674, "step": 99 }, { "epoch": 0.0, "learning_rate": 5.466666666666667e-06, "loss": 1.6845, "step": 100 }, { "epoch": 0.0, "learning_rate": 5.533333333333334e-06, "loss": 1.7423, "step": 101 }, { "epoch": 0.0, "learning_rate": 5.600000000000001e-06, "loss": 1.7124, "step": 102 }, { "epoch": 0.0, "learning_rate": 5.666666666666667e-06, "loss": 1.7597, "step": 103 }, { "epoch": 0.0, "learning_rate": 5.733333333333334e-06, "loss": 1.7016, "step": 104 }, { "epoch": 0.0, "learning_rate": 5.8e-06, "loss": 1.8067, "step": 105 }, { "epoch": 0.0, "learning_rate": 5.8666666666666675e-06, "loss": 1.746, "step": 106 }, { "epoch": 0.0, "learning_rate": 5.933333333333335e-06, "loss": 1.786, "step": 107 }, { "epoch": 0.0, "learning_rate": 6e-06, "loss": 1.6588, "step": 108 }, { "epoch": 0.0, "learning_rate": 6.066666666666667e-06, "loss": 1.7702, "step": 109 }, { "epoch": 0.0, "learning_rate": 6.133333333333334e-06, "loss": 1.7794, "step": 110 }, { "epoch": 0.0, "learning_rate": 6.200000000000001e-06, "loss": 1.7858, "step": 111 }, { "epoch": 0.0, "learning_rate": 6.266666666666668e-06, "loss": 1.7528, "step": 112 }, { "epoch": 0.0, "learning_rate": 6.333333333333333e-06, "loss": 1.7781, "step": 113 }, { "epoch": 0.0, "learning_rate": 6.4000000000000006e-06, "loss": 1.7645, "step": 114 }, { "epoch": 0.0, "learning_rate": 6.466666666666667e-06, "loss": 1.8199, "step": 115 }, { "epoch": 0.0, "learning_rate": 6.533333333333334e-06, "loss": 1.8067, "step": 116 }, { "epoch": 0.0, "learning_rate": 6.600000000000001e-06, "loss": 1.788, "step": 117 }, { "epoch": 0.0, "learning_rate": 6.666666666666667e-06, "loss": 1.8027, "step": 118 }, { "epoch": 0.0, "learning_rate": 6.733333333333334e-06, "loss": 1.8361, "step": 119 }, { "epoch": 0.0, "learning_rate": 6.800000000000001e-06, "loss": 1.8567, "step": 120 }, { "epoch": 0.0, "learning_rate": 6.866666666666667e-06, "loss": 1.8986, "step": 121 }, { "epoch": 0.0, "learning_rate": 6.9333333333333344e-06, "loss": 1.8875, "step": 122 }, { "epoch": 0.0, "learning_rate": 7e-06, "loss": 1.8463, "step": 123 }, { "epoch": 0.0, "learning_rate": 7.066666666666667e-06, "loss": 1.9191, "step": 124 }, { "epoch": 0.0, "learning_rate": 7.133333333333334e-06, "loss": 1.874, "step": 125 }, { "epoch": 0.0, "learning_rate": 7.2000000000000005e-06, "loss": 1.8379, "step": 126 }, { "epoch": 0.0, "learning_rate": 7.266666666666668e-06, "loss": 1.8223, "step": 127 }, { "epoch": 0.0, "learning_rate": 7.333333333333333e-06, "loss": 1.8387, "step": 128 }, { "epoch": 0.0, "learning_rate": 7.4e-06, "loss": 1.8025, "step": 129 }, { "epoch": 0.0, "learning_rate": 7.4666666666666675e-06, "loss": 1.8121, "step": 130 }, { "epoch": 0.0, "learning_rate": 7.533333333333334e-06, "loss": 1.6558, "step": 131 }, { "epoch": 0.0, "learning_rate": 7.600000000000001e-06, "loss": 1.7905, "step": 132 }, { "epoch": 0.0, "learning_rate": 7.666666666666667e-06, "loss": 1.7645, "step": 133 }, { "epoch": 0.0, "learning_rate": 7.733333333333334e-06, "loss": 1.8975, "step": 134 }, { "epoch": 0.0, "learning_rate": 7.800000000000002e-06, "loss": 1.8205, "step": 135 }, { "epoch": 0.0, "learning_rate": 7.866666666666667e-06, "loss": 1.8415, "step": 136 }, { "epoch": 0.0, "learning_rate": 7.933333333333334e-06, "loss": 1.7723, "step": 137 }, { "epoch": 0.0, "learning_rate": 8.000000000000001e-06, "loss": 1.8184, "step": 138 }, { "epoch": 0.0, "learning_rate": 8.066666666666667e-06, "loss": 1.8172, "step": 139 }, { "epoch": 0.0, "learning_rate": 8.133333333333334e-06, "loss": 1.8363, "step": 140 }, { "epoch": 0.0, "learning_rate": 8.2e-06, "loss": 1.7997, "step": 141 }, { "epoch": 0.0, "learning_rate": 8.2e-06, "loss": 1.7754, "step": 142 }, { "epoch": 0.0, "learning_rate": 8.266666666666667e-06, "loss": 1.7272, "step": 143 }, { "epoch": 0.0, "learning_rate": 8.333333333333334e-06, "loss": 1.7325, "step": 144 }, { "epoch": 0.0, "learning_rate": 8.400000000000001e-06, "loss": 1.7611, "step": 145 }, { "epoch": 0.0, "learning_rate": 8.466666666666668e-06, "loss": 1.7212, "step": 146 }, { "epoch": 0.0, "learning_rate": 8.533333333333335e-06, "loss": 1.749, "step": 147 }, { "epoch": 0.0, "learning_rate": 8.6e-06, "loss": 1.7357, "step": 148 }, { "epoch": 0.0, "learning_rate": 8.666666666666668e-06, "loss": 1.7819, "step": 149 }, { "epoch": 0.01, "learning_rate": 8.733333333333333e-06, "loss": 1.7826, "step": 150 }, { "epoch": 0.01, "learning_rate": 8.8e-06, "loss": 1.7715, "step": 151 }, { "epoch": 0.01, "learning_rate": 8.866666666666668e-06, "loss": 1.7851, "step": 152 }, { "epoch": 0.01, "learning_rate": 8.933333333333333e-06, "loss": 1.7623, "step": 153 }, { "epoch": 0.01, "learning_rate": 9e-06, "loss": 1.7616, "step": 154 }, { "epoch": 0.01, "learning_rate": 9.066666666666667e-06, "loss": 1.7483, "step": 155 }, { "epoch": 0.01, "learning_rate": 9.133333333333335e-06, "loss": 1.7837, "step": 156 }, { "epoch": 0.01, "learning_rate": 9.200000000000002e-06, "loss": 1.758, "step": 157 }, { "epoch": 0.01, "learning_rate": 9.266666666666667e-06, "loss": 1.8137, "step": 158 }, { "epoch": 0.01, "learning_rate": 9.333333333333334e-06, "loss": 1.7768, "step": 159 }, { "epoch": 0.01, "learning_rate": 9.4e-06, "loss": 1.716, "step": 160 }, { "epoch": 0.01, "learning_rate": 9.466666666666667e-06, "loss": 1.7638, "step": 161 }, { "epoch": 0.01, "learning_rate": 9.533333333333334e-06, "loss": 1.6633, "step": 162 }, { "epoch": 0.01, "learning_rate": 9.600000000000001e-06, "loss": 1.6534, "step": 163 }, { "epoch": 0.01, "learning_rate": 9.666666666666667e-06, "loss": 1.752, "step": 164 }, { "epoch": 0.01, "learning_rate": 9.733333333333334e-06, "loss": 1.6755, "step": 165 }, { "epoch": 0.01, "learning_rate": 9.800000000000001e-06, "loss": 1.6686, "step": 166 }, { "epoch": 0.01, "learning_rate": 9.866666666666668e-06, "loss": 1.7501, "step": 167 }, { "epoch": 0.01, "learning_rate": 9.933333333333334e-06, "loss": 1.7825, "step": 168 }, { "epoch": 0.01, "learning_rate": 1e-05, "loss": 1.7524, "step": 169 }, { "epoch": 0.01, "learning_rate": 1.0066666666666666e-05, "loss": 1.6896, "step": 170 }, { "epoch": 0.01, "learning_rate": 1.0133333333333335e-05, "loss": 1.6907, "step": 171 }, { "epoch": 0.01, "learning_rate": 1.02e-05, "loss": 1.7315, "step": 172 }, { "epoch": 0.01, "learning_rate": 1.0266666666666668e-05, "loss": 1.7165, "step": 173 }, { "epoch": 0.01, "learning_rate": 1.0333333333333335e-05, "loss": 1.6427, "step": 174 }, { "epoch": 0.01, "learning_rate": 1.04e-05, "loss": 1.7318, "step": 175 }, { "epoch": 0.01, "learning_rate": 1.0466666666666668e-05, "loss": 1.6821, "step": 176 }, { "epoch": 0.01, "learning_rate": 1.0533333333333333e-05, "loss": 1.7712, "step": 177 }, { "epoch": 0.01, "learning_rate": 1.0600000000000002e-05, "loss": 1.7491, "step": 178 }, { "epoch": 0.01, "learning_rate": 1.0666666666666667e-05, "loss": 1.7613, "step": 179 }, { "epoch": 0.01, "learning_rate": 1.0733333333333333e-05, "loss": 1.7526, "step": 180 }, { "epoch": 0.01, "learning_rate": 1.0800000000000002e-05, "loss": 1.6875, "step": 181 }, { "epoch": 0.01, "learning_rate": 1.0866666666666667e-05, "loss": 1.7517, "step": 182 }, { "epoch": 0.01, "learning_rate": 1.0933333333333334e-05, "loss": 1.6671, "step": 183 }, { "epoch": 0.01, "learning_rate": 1.1000000000000001e-05, "loss": 1.7398, "step": 184 }, { "epoch": 0.01, "learning_rate": 1.1066666666666669e-05, "loss": 1.7771, "step": 185 }, { "epoch": 0.01, "learning_rate": 1.1133333333333334e-05, "loss": 1.7386, "step": 186 }, { "epoch": 0.01, "learning_rate": 1.1200000000000001e-05, "loss": 1.6961, "step": 187 }, { "epoch": 0.01, "learning_rate": 1.1266666666666668e-05, "loss": 1.7061, "step": 188 }, { "epoch": 0.01, "learning_rate": 1.1333333333333334e-05, "loss": 1.7501, "step": 189 }, { "epoch": 0.01, "learning_rate": 1.14e-05, "loss": 1.6912, "step": 190 }, { "epoch": 0.01, "learning_rate": 1.1466666666666668e-05, "loss": 1.6572, "step": 191 }, { "epoch": 0.01, "learning_rate": 1.1533333333333334e-05, "loss": 1.7712, "step": 192 }, { "epoch": 0.01, "learning_rate": 1.16e-05, "loss": 1.7127, "step": 193 }, { "epoch": 0.01, "learning_rate": 1.1666666666666668e-05, "loss": 1.7661, "step": 194 }, { "epoch": 0.01, "learning_rate": 1.1733333333333335e-05, "loss": 1.7188, "step": 195 }, { "epoch": 0.01, "learning_rate": 1.18e-05, "loss": 1.7132, "step": 196 }, { "epoch": 0.01, "learning_rate": 1.186666666666667e-05, "loss": 1.7269, "step": 197 }, { "epoch": 0.01, "learning_rate": 1.1933333333333335e-05, "loss": 1.7393, "step": 198 }, { "epoch": 0.01, "learning_rate": 1.2e-05, "loss": 1.7, "step": 199 }, { "epoch": 0.01, "learning_rate": 1.206666666666667e-05, "loss": 1.7136, "step": 200 }, { "epoch": 0.01, "learning_rate": 1.2133333333333335e-05, "loss": 1.7354, "step": 201 }, { "epoch": 0.01, "learning_rate": 1.22e-05, "loss": 1.7776, "step": 202 }, { "epoch": 0.01, "learning_rate": 1.2266666666666667e-05, "loss": 1.6932, "step": 203 }, { "epoch": 0.01, "learning_rate": 1.2333333333333334e-05, "loss": 1.7788, "step": 204 }, { "epoch": 0.01, "learning_rate": 1.2400000000000002e-05, "loss": 1.7481, "step": 205 }, { "epoch": 0.01, "learning_rate": 1.2466666666666667e-05, "loss": 1.7406, "step": 206 }, { "epoch": 0.01, "learning_rate": 1.2533333333333336e-05, "loss": 1.7211, "step": 207 }, { "epoch": 0.01, "learning_rate": 1.2600000000000001e-05, "loss": 1.6875, "step": 208 }, { "epoch": 0.01, "learning_rate": 1.2666666666666667e-05, "loss": 1.7817, "step": 209 }, { "epoch": 0.01, "learning_rate": 1.2733333333333336e-05, "loss": 1.7444, "step": 210 }, { "epoch": 0.01, "learning_rate": 1.2800000000000001e-05, "loss": 1.7493, "step": 211 }, { "epoch": 0.01, "learning_rate": 1.2866666666666667e-05, "loss": 1.7014, "step": 212 }, { "epoch": 0.01, "learning_rate": 1.2933333333333334e-05, "loss": 1.784, "step": 213 }, { "epoch": 0.01, "learning_rate": 1.3000000000000001e-05, "loss": 1.7533, "step": 214 }, { "epoch": 0.01, "learning_rate": 1.3066666666666668e-05, "loss": 1.7106, "step": 215 }, { "epoch": 0.01, "learning_rate": 1.3133333333333334e-05, "loss": 1.7243, "step": 216 }, { "epoch": 0.01, "learning_rate": 1.3200000000000002e-05, "loss": 1.7969, "step": 217 }, { "epoch": 0.01, "learning_rate": 1.3266666666666668e-05, "loss": 1.7651, "step": 218 }, { "epoch": 0.01, "learning_rate": 1.3333333333333333e-05, "loss": 1.7168, "step": 219 }, { "epoch": 0.01, "learning_rate": 1.3400000000000002e-05, "loss": 1.7288, "step": 220 }, { "epoch": 0.01, "learning_rate": 1.3466666666666668e-05, "loss": 1.693, "step": 221 }, { "epoch": 0.01, "learning_rate": 1.3533333333333333e-05, "loss": 1.7519, "step": 222 }, { "epoch": 0.01, "learning_rate": 1.3600000000000002e-05, "loss": 1.7716, "step": 223 }, { "epoch": 0.01, "learning_rate": 1.3666666666666667e-05, "loss": 1.6803, "step": 224 }, { "epoch": 0.01, "learning_rate": 1.3733333333333335e-05, "loss": 1.6727, "step": 225 }, { "epoch": 0.01, "learning_rate": 1.38e-05, "loss": 1.6774, "step": 226 }, { "epoch": 0.01, "learning_rate": 1.3866666666666669e-05, "loss": 1.7483, "step": 227 }, { "epoch": 0.01, "learning_rate": 1.3933333333333334e-05, "loss": 1.7394, "step": 228 }, { "epoch": 0.01, "learning_rate": 1.4e-05, "loss": 1.6832, "step": 229 }, { "epoch": 0.01, "learning_rate": 1.4066666666666669e-05, "loss": 1.6965, "step": 230 }, { "epoch": 0.01, "learning_rate": 1.4133333333333334e-05, "loss": 1.7745, "step": 231 }, { "epoch": 0.01, "learning_rate": 1.4200000000000001e-05, "loss": 1.6316, "step": 232 }, { "epoch": 0.01, "learning_rate": 1.4266666666666668e-05, "loss": 1.7595, "step": 233 }, { "epoch": 0.01, "learning_rate": 1.4333333333333334e-05, "loss": 1.7048, "step": 234 }, { "epoch": 0.01, "learning_rate": 1.4400000000000001e-05, "loss": 1.7395, "step": 235 }, { "epoch": 0.01, "learning_rate": 1.4466666666666668e-05, "loss": 1.7643, "step": 236 }, { "epoch": 0.01, "learning_rate": 1.4533333333333335e-05, "loss": 1.7041, "step": 237 }, { "epoch": 0.01, "learning_rate": 1.46e-05, "loss": 1.7349, "step": 238 }, { "epoch": 0.01, "learning_rate": 1.4666666666666666e-05, "loss": 1.7688, "step": 239 }, { "epoch": 0.01, "learning_rate": 1.4733333333333335e-05, "loss": 1.7528, "step": 240 }, { "epoch": 0.01, "learning_rate": 1.48e-05, "loss": 1.6991, "step": 241 }, { "epoch": 0.01, "learning_rate": 1.4866666666666668e-05, "loss": 1.8094, "step": 242 }, { "epoch": 0.01, "learning_rate": 1.4933333333333335e-05, "loss": 1.6918, "step": 243 }, { "epoch": 0.01, "learning_rate": 1.5000000000000002e-05, "loss": 1.6401, "step": 244 }, { "epoch": 0.01, "learning_rate": 1.5066666666666668e-05, "loss": 1.6958, "step": 245 }, { "epoch": 0.01, "learning_rate": 1.5133333333333335e-05, "loss": 1.7534, "step": 246 }, { "epoch": 0.01, "learning_rate": 1.5200000000000002e-05, "loss": 1.7402, "step": 247 }, { "epoch": 0.01, "learning_rate": 1.5266666666666667e-05, "loss": 1.7145, "step": 248 }, { "epoch": 0.01, "learning_rate": 1.5333333333333334e-05, "loss": 1.6794, "step": 249 }, { "epoch": 0.01, "learning_rate": 1.54e-05, "loss": 1.6902, "step": 250 }, { "epoch": 0.01, "learning_rate": 1.546666666666667e-05, "loss": 1.6276, "step": 251 }, { "epoch": 0.01, "learning_rate": 1.5533333333333333e-05, "loss": 1.7621, "step": 252 }, { "epoch": 0.01, "learning_rate": 1.5600000000000003e-05, "loss": 1.728, "step": 253 }, { "epoch": 0.01, "learning_rate": 1.5666666666666667e-05, "loss": 1.6622, "step": 254 }, { "epoch": 0.01, "learning_rate": 1.5733333333333334e-05, "loss": 1.7623, "step": 255 }, { "epoch": 0.01, "learning_rate": 1.58e-05, "loss": 1.7491, "step": 256 }, { "epoch": 0.01, "learning_rate": 1.586666666666667e-05, "loss": 1.7743, "step": 257 }, { "epoch": 0.01, "learning_rate": 1.5933333333333336e-05, "loss": 1.7354, "step": 258 }, { "epoch": 0.01, "learning_rate": 1.6000000000000003e-05, "loss": 1.7296, "step": 259 }, { "epoch": 0.01, "learning_rate": 1.606666666666667e-05, "loss": 1.7819, "step": 260 }, { "epoch": 0.01, "learning_rate": 1.6133333333333334e-05, "loss": 1.6497, "step": 261 }, { "epoch": 0.01, "learning_rate": 1.62e-05, "loss": 1.6945, "step": 262 }, { "epoch": 0.01, "learning_rate": 1.6266666666666668e-05, "loss": 1.6349, "step": 263 }, { "epoch": 0.01, "learning_rate": 1.6333333333333335e-05, "loss": 1.7807, "step": 264 }, { "epoch": 0.01, "learning_rate": 1.64e-05, "loss": 1.7178, "step": 265 }, { "epoch": 0.01, "learning_rate": 1.646666666666667e-05, "loss": 1.5156, "step": 266 }, { "epoch": 0.01, "learning_rate": 1.6533333333333333e-05, "loss": 1.7474, "step": 267 }, { "epoch": 0.01, "learning_rate": 1.66e-05, "loss": 1.7243, "step": 268 }, { "epoch": 0.01, "learning_rate": 1.6666666666666667e-05, "loss": 1.6807, "step": 269 }, { "epoch": 0.01, "learning_rate": 1.6733333333333335e-05, "loss": 1.7116, "step": 270 }, { "epoch": 0.01, "learning_rate": 1.6800000000000002e-05, "loss": 1.7154, "step": 271 }, { "epoch": 0.01, "learning_rate": 1.686666666666667e-05, "loss": 1.6799, "step": 272 }, { "epoch": 0.01, "learning_rate": 1.6933333333333336e-05, "loss": 1.67, "step": 273 }, { "epoch": 0.01, "learning_rate": 1.7e-05, "loss": 1.7645, "step": 274 }, { "epoch": 0.01, "learning_rate": 1.706666666666667e-05, "loss": 1.5879, "step": 275 }, { "epoch": 0.01, "learning_rate": 1.7133333333333334e-05, "loss": 1.7205, "step": 276 }, { "epoch": 0.01, "learning_rate": 1.72e-05, "loss": 1.7202, "step": 277 }, { "epoch": 0.01, "learning_rate": 1.726666666666667e-05, "loss": 1.7341, "step": 278 }, { "epoch": 0.01, "learning_rate": 1.7333333333333336e-05, "loss": 1.7516, "step": 279 }, { "epoch": 0.01, "learning_rate": 1.7400000000000003e-05, "loss": 1.7406, "step": 280 }, { "epoch": 0.01, "learning_rate": 1.7466666666666667e-05, "loss": 1.7215, "step": 281 }, { "epoch": 0.01, "learning_rate": 1.7533333333333337e-05, "loss": 1.6123, "step": 282 }, { "epoch": 0.01, "learning_rate": 1.76e-05, "loss": 1.7594, "step": 283 }, { "epoch": 0.01, "learning_rate": 1.7666666666666668e-05, "loss": 1.6966, "step": 284 }, { "epoch": 0.01, "learning_rate": 1.7733333333333335e-05, "loss": 1.7049, "step": 285 }, { "epoch": 0.01, "learning_rate": 1.7800000000000002e-05, "loss": 1.7746, "step": 286 }, { "epoch": 0.01, "learning_rate": 1.7866666666666666e-05, "loss": 1.7757, "step": 287 }, { "epoch": 0.01, "learning_rate": 1.7933333333333333e-05, "loss": 1.7936, "step": 288 }, { "epoch": 0.01, "learning_rate": 1.8e-05, "loss": 1.7521, "step": 289 }, { "epoch": 0.01, "learning_rate": 1.8066666666666668e-05, "loss": 1.7808, "step": 290 }, { "epoch": 0.01, "learning_rate": 1.8133333333333335e-05, "loss": 1.7534, "step": 291 }, { "epoch": 0.01, "learning_rate": 1.8200000000000002e-05, "loss": 1.8655, "step": 292 }, { "epoch": 0.01, "learning_rate": 1.826666666666667e-05, "loss": 1.8176, "step": 293 }, { "epoch": 0.01, "learning_rate": 1.8333333333333333e-05, "loss": 1.802, "step": 294 }, { "epoch": 0.01, "learning_rate": 1.8400000000000003e-05, "loss": 1.7847, "step": 295 }, { "epoch": 0.01, "learning_rate": 1.8466666666666667e-05, "loss": 1.7228, "step": 296 }, { "epoch": 0.01, "learning_rate": 1.8533333333333334e-05, "loss": 1.8046, "step": 297 }, { "epoch": 0.01, "learning_rate": 1.86e-05, "loss": 1.778, "step": 298 }, { "epoch": 0.01, "learning_rate": 1.866666666666667e-05, "loss": 1.7989, "step": 299 }, { "epoch": 0.01, "learning_rate": 1.8733333333333336e-05, "loss": 1.8579, "step": 300 }, { "epoch": 0.01, "learning_rate": 1.88e-05, "loss": 1.8665, "step": 301 }, { "epoch": 0.01, "learning_rate": 1.886666666666667e-05, "loss": 1.8443, "step": 302 }, { "epoch": 0.01, "learning_rate": 1.8933333333333334e-05, "loss": 1.9039, "step": 303 }, { "epoch": 0.01, "learning_rate": 1.9e-05, "loss": 1.8245, "step": 304 }, { "epoch": 0.01, "learning_rate": 1.9066666666666668e-05, "loss": 1.8343, "step": 305 }, { "epoch": 0.01, "learning_rate": 1.9133333333333335e-05, "loss": 1.7519, "step": 306 }, { "epoch": 0.01, "learning_rate": 1.9200000000000003e-05, "loss": 1.8503, "step": 307 }, { "epoch": 0.01, "learning_rate": 1.926666666666667e-05, "loss": 1.7059, "step": 308 }, { "epoch": 0.01, "learning_rate": 1.9333333333333333e-05, "loss": 1.7276, "step": 309 }, { "epoch": 0.01, "learning_rate": 1.94e-05, "loss": 1.7797, "step": 310 }, { "epoch": 0.01, "learning_rate": 1.9466666666666668e-05, "loss": 1.6543, "step": 311 }, { "epoch": 0.01, "learning_rate": 1.9533333333333335e-05, "loss": 1.7924, "step": 312 }, { "epoch": 0.01, "learning_rate": 1.9600000000000002e-05, "loss": 1.8288, "step": 313 }, { "epoch": 0.01, "learning_rate": 1.9666666666666666e-05, "loss": 1.6645, "step": 314 }, { "epoch": 0.01, "learning_rate": 1.9733333333333336e-05, "loss": 1.8124, "step": 315 }, { "epoch": 0.01, "learning_rate": 1.9733333333333336e-05, "loss": 1.7018, "step": 316 }, { "epoch": 0.01, "learning_rate": 1.9733333333333336e-05, "loss": 1.7796, "step": 317 }, { "epoch": 0.01, "learning_rate": 1.98e-05, "loss": 1.7636, "step": 318 }, { "epoch": 0.01, "learning_rate": 1.9866666666666667e-05, "loss": 1.8111, "step": 319 }, { "epoch": 0.01, "learning_rate": 1.9933333333333334e-05, "loss": 1.7526, "step": 320 }, { "epoch": 0.01, "learning_rate": 2e-05, "loss": 1.8051, "step": 321 }, { "epoch": 0.01, "learning_rate": 1.999999994405557e-05, "loss": 1.7906, "step": 322 }, { "epoch": 0.01, "learning_rate": 1.9999999776222286e-05, "loss": 1.7678, "step": 323 }, { "epoch": 0.01, "learning_rate": 1.9999999496500138e-05, "loss": 1.8098, "step": 324 }, { "epoch": 0.01, "learning_rate": 1.999999910488914e-05, "loss": 1.7153, "step": 325 }, { "epoch": 0.01, "learning_rate": 1.999999860138929e-05, "loss": 1.7425, "step": 326 }, { "epoch": 0.01, "learning_rate": 1.9999997986000598e-05, "loss": 1.703, "step": 327 }, { "epoch": 0.01, "learning_rate": 1.9999997258723073e-05, "loss": 1.757, "step": 328 }, { "epoch": 0.01, "learning_rate": 1.9999996419556714e-05, "loss": 1.7876, "step": 329 }, { "epoch": 0.01, "learning_rate": 1.9999995468501538e-05, "loss": 1.7634, "step": 330 }, { "epoch": 0.01, "learning_rate": 1.999999440555755e-05, "loss": 1.7276, "step": 331 }, { "epoch": 0.01, "learning_rate": 1.999999323072477e-05, "loss": 1.7069, "step": 332 }, { "epoch": 0.01, "learning_rate": 1.9999991944003204e-05, "loss": 1.6352, "step": 333 }, { "epoch": 0.01, "learning_rate": 1.9999990545392867e-05, "loss": 1.7758, "step": 334 }, { "epoch": 0.01, "learning_rate": 1.9999989034893785e-05, "loss": 1.7891, "step": 335 }, { "epoch": 0.01, "learning_rate": 1.9999987412505956e-05, "loss": 1.7673, "step": 336 }, { "epoch": 0.01, "learning_rate": 1.9999985678229415e-05, "loss": 1.6696, "step": 337 }, { "epoch": 0.01, "learning_rate": 1.9999983832064173e-05, "loss": 1.6686, "step": 338 }, { "epoch": 0.01, "learning_rate": 1.999998187401025e-05, "loss": 1.776, "step": 339 }, { "epoch": 0.01, "learning_rate": 1.9999979804067673e-05, "loss": 1.7375, "step": 340 }, { "epoch": 0.01, "learning_rate": 1.9999977622236463e-05, "loss": 1.7744, "step": 341 }, { "epoch": 0.01, "learning_rate": 1.9999975328516643e-05, "loss": 1.7763, "step": 342 }, { "epoch": 0.01, "learning_rate": 1.999997292290824e-05, "loss": 1.7491, "step": 343 }, { "epoch": 0.01, "learning_rate": 1.999997040541128e-05, "loss": 1.6211, "step": 344 }, { "epoch": 0.01, "learning_rate": 1.9999967776025794e-05, "loss": 1.8335, "step": 345 }, { "epoch": 0.01, "learning_rate": 1.9999965034751806e-05, "loss": 1.7349, "step": 346 }, { "epoch": 0.01, "learning_rate": 1.9999962181589354e-05, "loss": 1.6424, "step": 347 }, { "epoch": 0.01, "learning_rate": 1.9999959216538463e-05, "loss": 1.7245, "step": 348 }, { "epoch": 0.01, "learning_rate": 1.999995613959917e-05, "loss": 1.7176, "step": 349 }, { "epoch": 0.01, "learning_rate": 1.9999952950771505e-05, "loss": 1.7128, "step": 350 }, { "epoch": 0.01, "learning_rate": 1.9999949650055512e-05, "loss": 1.7367, "step": 351 }, { "epoch": 0.01, "learning_rate": 1.999994623745122e-05, "loss": 1.7253, "step": 352 }, { "epoch": 0.01, "learning_rate": 1.9999942712958673e-05, "loss": 1.7528, "step": 353 }, { "epoch": 0.01, "learning_rate": 1.9999939076577906e-05, "loss": 1.8133, "step": 354 }, { "epoch": 0.01, "learning_rate": 1.9999935328308958e-05, "loss": 1.7158, "step": 355 }, { "epoch": 0.01, "learning_rate": 1.999993146815188e-05, "loss": 1.7413, "step": 356 }, { "epoch": 0.01, "learning_rate": 1.9999927496106708e-05, "loss": 1.6431, "step": 357 }, { "epoch": 0.01, "learning_rate": 1.9999923412173484e-05, "loss": 1.7228, "step": 358 }, { "epoch": 0.01, "learning_rate": 1.9999919216352262e-05, "loss": 1.7649, "step": 359 }, { "epoch": 0.01, "learning_rate": 1.9999914908643084e-05, "loss": 1.7568, "step": 360 }, { "epoch": 0.01, "learning_rate": 1.9999910489045997e-05, "loss": 1.8371, "step": 361 }, { "epoch": 0.01, "learning_rate": 1.9999905957561055e-05, "loss": 1.798, "step": 362 }, { "epoch": 0.01, "learning_rate": 1.9999901314188302e-05, "loss": 1.7126, "step": 363 }, { "epoch": 0.01, "learning_rate": 1.9999896558927796e-05, "loss": 1.8058, "step": 364 }, { "epoch": 0.01, "learning_rate": 1.999989169177959e-05, "loss": 1.7858, "step": 365 }, { "epoch": 0.01, "learning_rate": 1.9999886712743734e-05, "loss": 1.7532, "step": 366 }, { "epoch": 0.01, "learning_rate": 1.9999881621820286e-05, "loss": 1.6714, "step": 367 }, { "epoch": 0.01, "learning_rate": 1.9999876419009302e-05, "loss": 1.6787, "step": 368 }, { "epoch": 0.01, "learning_rate": 1.9999871104310846e-05, "loss": 1.774, "step": 369 }, { "epoch": 0.01, "learning_rate": 1.999986567772497e-05, "loss": 1.8103, "step": 370 }, { "epoch": 0.01, "learning_rate": 1.9999860139251737e-05, "loss": 1.7536, "step": 371 }, { "epoch": 0.01, "learning_rate": 1.9999854488891214e-05, "loss": 1.752, "step": 372 }, { "epoch": 0.01, "learning_rate": 1.9999848726643452e-05, "loss": 1.726, "step": 373 }, { "epoch": 0.01, "learning_rate": 1.999984285250853e-05, "loss": 1.7605, "step": 374 }, { "epoch": 0.01, "learning_rate": 1.9999836866486505e-05, "loss": 1.7503, "step": 375 }, { "epoch": 0.01, "learning_rate": 1.9999830768577445e-05, "loss": 1.7561, "step": 376 }, { "epoch": 0.01, "learning_rate": 1.9999824558781418e-05, "loss": 1.7603, "step": 377 }, { "epoch": 0.01, "learning_rate": 1.9999818237098495e-05, "loss": 1.8866, "step": 378 }, { "epoch": 0.01, "learning_rate": 1.999981180352875e-05, "loss": 1.7289, "step": 379 }, { "epoch": 0.01, "learning_rate": 1.9999805258072248e-05, "loss": 1.699, "step": 380 }, { "epoch": 0.01, "learning_rate": 1.9999798600729067e-05, "loss": 1.7002, "step": 381 }, { "epoch": 0.01, "learning_rate": 1.999979183149928e-05, "loss": 1.7647, "step": 382 }, { "epoch": 0.01, "learning_rate": 1.9999784950382965e-05, "loss": 1.7209, "step": 383 }, { "epoch": 0.01, "learning_rate": 1.999977795738019e-05, "loss": 1.746, "step": 384 }, { "epoch": 0.01, "learning_rate": 1.9999770852491047e-05, "loss": 1.7788, "step": 385 }, { "epoch": 0.01, "learning_rate": 1.9999763635715605e-05, "loss": 1.7735, "step": 386 }, { "epoch": 0.01, "learning_rate": 1.9999756307053947e-05, "loss": 1.7598, "step": 387 }, { "epoch": 0.01, "learning_rate": 1.9999748866506162e-05, "loss": 1.6748, "step": 388 }, { "epoch": 0.01, "learning_rate": 1.9999741314072322e-05, "loss": 1.741, "step": 389 }, { "epoch": 0.01, "learning_rate": 1.999973364975252e-05, "loss": 1.764, "step": 390 }, { "epoch": 0.01, "learning_rate": 1.999972587354684e-05, "loss": 1.6881, "step": 391 }, { "epoch": 0.01, "learning_rate": 1.9999717985455364e-05, "loss": 1.7021, "step": 392 }, { "epoch": 0.01, "learning_rate": 1.9999709985478188e-05, "loss": 1.7552, "step": 393 }, { "epoch": 0.01, "learning_rate": 1.9999701873615396e-05, "loss": 1.7014, "step": 394 }, { "epoch": 0.01, "learning_rate": 1.999969364986708e-05, "loss": 1.8017, "step": 395 }, { "epoch": 0.01, "learning_rate": 1.9999685314233333e-05, "loss": 1.7801, "step": 396 }, { "epoch": 0.01, "learning_rate": 1.9999676866714248e-05, "loss": 1.7607, "step": 397 }, { "epoch": 0.01, "learning_rate": 1.999966830730992e-05, "loss": 1.7759, "step": 398 }, { "epoch": 0.01, "learning_rate": 1.999965963602044e-05, "loss": 1.7352, "step": 399 }, { "epoch": 0.01, "learning_rate": 1.999965085284591e-05, "loss": 1.7172, "step": 400 }, { "epoch": 0.01, "learning_rate": 1.999964195778643e-05, "loss": 1.6856, "step": 401 }, { "epoch": 0.01, "learning_rate": 1.9999632950842095e-05, "loss": 1.7223, "step": 402 }, { "epoch": 0.01, "learning_rate": 1.999962383201301e-05, "loss": 1.7551, "step": 403 }, { "epoch": 0.01, "learning_rate": 1.999961460129927e-05, "loss": 1.6872, "step": 404 }, { "epoch": 0.01, "learning_rate": 1.9999605258700983e-05, "loss": 1.6631, "step": 405 }, { "epoch": 0.01, "learning_rate": 1.9999595804218258e-05, "loss": 1.7182, "step": 406 }, { "epoch": 0.01, "learning_rate": 1.9999586237851198e-05, "loss": 1.6832, "step": 407 }, { "epoch": 0.01, "learning_rate": 1.99995765595999e-05, "loss": 1.6863, "step": 408 }, { "epoch": 0.01, "learning_rate": 1.9999566769464483e-05, "loss": 1.6579, "step": 409 }, { "epoch": 0.01, "learning_rate": 1.999955686744506e-05, "loss": 1.6552, "step": 410 }, { "epoch": 0.01, "learning_rate": 1.9999546853541728e-05, "loss": 1.632, "step": 411 }, { "epoch": 0.01, "learning_rate": 1.9999536727754614e-05, "loss": 1.6566, "step": 412 }, { "epoch": 0.01, "learning_rate": 1.999952649008382e-05, "loss": 1.7382, "step": 413 }, { "epoch": 0.01, "learning_rate": 1.9999516140529465e-05, "loss": 1.6433, "step": 414 }, { "epoch": 0.01, "learning_rate": 1.9999505679091668e-05, "loss": 1.5754, "step": 415 }, { "epoch": 0.01, "learning_rate": 1.9999495105770536e-05, "loss": 1.6778, "step": 416 }, { "epoch": 0.01, "learning_rate": 1.9999484420566197e-05, "loss": 1.5741, "step": 417 }, { "epoch": 0.01, "learning_rate": 1.999947362347877e-05, "loss": 1.6883, "step": 418 }, { "epoch": 0.01, "learning_rate": 1.999946271450837e-05, "loss": 1.7051, "step": 419 }, { "epoch": 0.01, "learning_rate": 1.9999451693655125e-05, "loss": 1.7252, "step": 420 }, { "epoch": 0.01, "learning_rate": 1.9999440560919153e-05, "loss": 1.6869, "step": 421 }, { "epoch": 0.01, "learning_rate": 1.999942931630058e-05, "loss": 1.5641, "step": 422 }, { "epoch": 0.01, "learning_rate": 1.9999417959799538e-05, "loss": 1.6896, "step": 423 }, { "epoch": 0.01, "learning_rate": 1.9999406491416146e-05, "loss": 1.6157, "step": 424 }, { "epoch": 0.01, "learning_rate": 1.9999394911150533e-05, "loss": 1.707, "step": 425 }, { "epoch": 0.01, "learning_rate": 1.9999383219002836e-05, "loss": 1.7529, "step": 426 }, { "epoch": 0.01, "learning_rate": 1.9999371414973175e-05, "loss": 1.6552, "step": 427 }, { "epoch": 0.01, "learning_rate": 1.9999359499061694e-05, "loss": 1.6509, "step": 428 }, { "epoch": 0.01, "learning_rate": 1.9999347471268517e-05, "loss": 1.6996, "step": 429 }, { "epoch": 0.01, "learning_rate": 1.999933533159378e-05, "loss": 1.7033, "step": 430 }, { "epoch": 0.01, "learning_rate": 1.9999323080037623e-05, "loss": 1.6961, "step": 431 }, { "epoch": 0.01, "learning_rate": 1.999931071660018e-05, "loss": 1.7075, "step": 432 }, { "epoch": 0.01, "learning_rate": 1.9999298241281592e-05, "loss": 1.644, "step": 433 }, { "epoch": 0.01, "learning_rate": 1.9999285654081995e-05, "loss": 1.7677, "step": 434 }, { "epoch": 0.01, "learning_rate": 1.999927295500153e-05, "loss": 1.6569, "step": 435 }, { "epoch": 0.01, "learning_rate": 1.9999260144040342e-05, "loss": 1.6502, "step": 436 }, { "epoch": 0.01, "learning_rate": 1.9999247221198573e-05, "loss": 1.6187, "step": 437 }, { "epoch": 0.01, "learning_rate": 1.9999234186476366e-05, "loss": 1.6564, "step": 438 }, { "epoch": 0.01, "learning_rate": 1.999922103987387e-05, "loss": 1.6074, "step": 439 }, { "epoch": 0.01, "learning_rate": 1.999920778139123e-05, "loss": 1.6651, "step": 440 }, { "epoch": 0.01, "learning_rate": 1.9999194411028596e-05, "loss": 1.6969, "step": 441 }, { "epoch": 0.01, "learning_rate": 1.9999180928786113e-05, "loss": 1.5832, "step": 442 }, { "epoch": 0.01, "learning_rate": 1.9999167334663937e-05, "loss": 1.627, "step": 443 }, { "epoch": 0.01, "learning_rate": 1.9999153628662217e-05, "loss": 1.6425, "step": 444 }, { "epoch": 0.01, "learning_rate": 1.9999139810781112e-05, "loss": 1.6726, "step": 445 }, { "epoch": 0.01, "learning_rate": 1.9999125881020766e-05, "loss": 1.7015, "step": 446 }, { "epoch": 0.01, "learning_rate": 1.9999111839381346e-05, "loss": 1.6711, "step": 447 }, { "epoch": 0.01, "learning_rate": 1.9999097685863003e-05, "loss": 1.7538, "step": 448 }, { "epoch": 0.01, "learning_rate": 1.9999083420465894e-05, "loss": 1.7467, "step": 449 }, { "epoch": 0.01, "learning_rate": 1.9999069043190184e-05, "loss": 1.6285, "step": 450 }, { "epoch": 0.02, "learning_rate": 1.9999054554036032e-05, "loss": 1.7152, "step": 451 }, { "epoch": 0.02, "learning_rate": 1.9999039953003596e-05, "loss": 1.6787, "step": 452 }, { "epoch": 0.02, "learning_rate": 1.9999025240093045e-05, "loss": 1.7155, "step": 453 }, { "epoch": 0.02, "learning_rate": 1.9999010415304537e-05, "loss": 1.7105, "step": 454 }, { "epoch": 0.02, "learning_rate": 1.9998995478638246e-05, "loss": 1.7399, "step": 455 }, { "epoch": 0.02, "learning_rate": 1.9998980430094333e-05, "loss": 1.7178, "step": 456 }, { "epoch": 0.02, "learning_rate": 1.9998965269672973e-05, "loss": 1.6686, "step": 457 }, { "epoch": 0.02, "learning_rate": 1.999894999737433e-05, "loss": 1.6313, "step": 458 }, { "epoch": 0.02, "learning_rate": 1.999893461319857e-05, "loss": 1.7042, "step": 459 }, { "epoch": 0.02, "learning_rate": 1.9998919117145878e-05, "loss": 1.6602, "step": 460 }, { "epoch": 0.02, "learning_rate": 1.9998903509216415e-05, "loss": 1.6736, "step": 461 }, { "epoch": 0.02, "learning_rate": 1.9998887789410363e-05, "loss": 1.7531, "step": 462 }, { "epoch": 0.02, "learning_rate": 1.99988719577279e-05, "loss": 1.67, "step": 463 }, { "epoch": 0.02, "learning_rate": 1.9998856014169193e-05, "loss": 1.6886, "step": 464 }, { "epoch": 0.02, "learning_rate": 1.999883995873443e-05, "loss": 1.7679, "step": 465 }, { "epoch": 0.02, "learning_rate": 1.999882379142379e-05, "loss": 1.841, "step": 466 }, { "epoch": 0.02, "learning_rate": 1.9998807512237447e-05, "loss": 1.6794, "step": 467 }, { "epoch": 0.02, "learning_rate": 1.999879112117559e-05, "loss": 1.7614, "step": 468 }, { "epoch": 0.02, "learning_rate": 1.9998774618238396e-05, "loss": 1.871, "step": 469 }, { "epoch": 0.02, "learning_rate": 1.9998758003426056e-05, "loss": 1.7536, "step": 470 }, { "epoch": 0.02, "learning_rate": 1.9998741276738753e-05, "loss": 1.8715, "step": 471 }, { "epoch": 0.02, "learning_rate": 1.9998724438176674e-05, "loss": 1.738, "step": 472 }, { "epoch": 0.02, "learning_rate": 1.9998707487740007e-05, "loss": 1.7334, "step": 473 }, { "epoch": 0.02, "learning_rate": 1.9998690425428943e-05, "loss": 1.7865, "step": 474 }, { "epoch": 0.02, "learning_rate": 1.9998673251243672e-05, "loss": 1.7432, "step": 475 }, { "epoch": 0.02, "learning_rate": 1.999865596518439e-05, "loss": 1.7051, "step": 476 }, { "epoch": 0.02, "learning_rate": 1.9998638567251283e-05, "loss": 1.6956, "step": 477 }, { "epoch": 0.02, "learning_rate": 1.999862105744455e-05, "loss": 1.7605, "step": 478 }, { "epoch": 0.02, "learning_rate": 1.9998603435764387e-05, "loss": 1.7647, "step": 479 }, { "epoch": 0.02, "learning_rate": 1.9998585702210992e-05, "loss": 1.6791, "step": 480 }, { "epoch": 0.02, "learning_rate": 1.999856785678456e-05, "loss": 1.6446, "step": 481 }, { "epoch": 0.02, "learning_rate": 1.9998549899485296e-05, "loss": 1.6853, "step": 482 }, { "epoch": 0.02, "learning_rate": 1.9998531830313394e-05, "loss": 1.7483, "step": 483 }, { "epoch": 0.02, "learning_rate": 1.999851364926906e-05, "loss": 1.6887, "step": 484 }, { "epoch": 0.02, "learning_rate": 1.99984953563525e-05, "loss": 1.7486, "step": 485 }, { "epoch": 0.02, "learning_rate": 1.9998476951563914e-05, "loss": 1.7516, "step": 486 }, { "epoch": 0.02, "learning_rate": 1.999845843490351e-05, "loss": 1.6979, "step": 487 }, { "epoch": 0.02, "learning_rate": 1.9998439806371498e-05, "loss": 1.7176, "step": 488 }, { "epoch": 0.02, "learning_rate": 1.999842106596808e-05, "loss": 1.7489, "step": 489 }, { "epoch": 0.02, "learning_rate": 1.999840221369347e-05, "loss": 1.664, "step": 490 }, { "epoch": 0.02, "learning_rate": 1.9998383249547878e-05, "loss": 1.7302, "step": 491 }, { "epoch": 0.02, "learning_rate": 1.9998364173531514e-05, "loss": 1.6628, "step": 492 }, { "epoch": 0.02, "learning_rate": 1.99983449856446e-05, "loss": 1.6819, "step": 493 }, { "epoch": 0.02, "learning_rate": 1.999832568588734e-05, "loss": 1.8165, "step": 494 }, { "epoch": 0.02, "learning_rate": 1.9998306274259955e-05, "loss": 1.7561, "step": 495 }, { "epoch": 0.02, "learning_rate": 1.9998286750762662e-05, "loss": 1.865, "step": 496 }, { "epoch": 0.02, "learning_rate": 1.999826711539568e-05, "loss": 1.8683, "step": 497 }, { "epoch": 0.02, "learning_rate": 1.9998247368159225e-05, "loss": 1.7391, "step": 498 }, { "epoch": 0.02, "learning_rate": 1.999822750905352e-05, "loss": 1.6512, "step": 499 }, { "epoch": 0.02, "learning_rate": 1.9998207538078795e-05, "loss": 1.8159, "step": 500 }, { "epoch": 0.02, "learning_rate": 1.999818745523526e-05, "loss": 1.678, "step": 501 }, { "epoch": 0.02, "learning_rate": 1.9998167260523148e-05, "loss": 1.6709, "step": 502 }, { "epoch": 0.02, "learning_rate": 1.999814695394268e-05, "loss": 1.7495, "step": 503 }, { "epoch": 0.02, "learning_rate": 1.999812653549409e-05, "loss": 1.7448, "step": 504 }, { "epoch": 0.02, "learning_rate": 1.99981060051776e-05, "loss": 1.8192, "step": 505 }, { "epoch": 0.02, "learning_rate": 1.9998085362993445e-05, "loss": 1.8035, "step": 506 }, { "epoch": 0.02, "learning_rate": 1.999806460894185e-05, "loss": 1.7603, "step": 507 }, { "epoch": 0.02, "learning_rate": 1.9998043743023056e-05, "loss": 1.7631, "step": 508 }, { "epoch": 0.02, "learning_rate": 1.9998022765237287e-05, "loss": 1.6811, "step": 509 }, { "epoch": 0.02, "learning_rate": 1.999800167558478e-05, "loss": 1.5507, "step": 510 }, { "epoch": 0.02, "learning_rate": 1.9997980474065776e-05, "loss": 1.652, "step": 511 }, { "epoch": 0.02, "learning_rate": 1.999795916068051e-05, "loss": 1.6933, "step": 512 }, { "epoch": 0.02, "learning_rate": 1.999793773542922e-05, "loss": 1.7629, "step": 513 }, { "epoch": 0.02, "learning_rate": 1.999791619831214e-05, "loss": 1.6448, "step": 514 }, { "epoch": 0.02, "learning_rate": 1.9997894549329517e-05, "loss": 1.7123, "step": 515 }, { "epoch": 0.02, "learning_rate": 1.9997872788481595e-05, "loss": 1.7426, "step": 516 }, { "epoch": 0.02, "learning_rate": 1.9997850915768614e-05, "loss": 1.737, "step": 517 }, { "epoch": 0.02, "learning_rate": 1.999782893119082e-05, "loss": 1.8042, "step": 518 }, { "epoch": 0.02, "learning_rate": 1.9997806834748455e-05, "loss": 1.8087, "step": 519 }, { "epoch": 0.02, "learning_rate": 1.9997784626441773e-05, "loss": 1.7507, "step": 520 }, { "epoch": 0.02, "learning_rate": 1.999776230627102e-05, "loss": 1.7176, "step": 521 }, { "epoch": 0.02, "learning_rate": 1.9997739874236444e-05, "loss": 1.7124, "step": 522 }, { "epoch": 0.02, "learning_rate": 1.99977173303383e-05, "loss": 1.7714, "step": 523 }, { "epoch": 0.02, "learning_rate": 1.999769467457683e-05, "loss": 1.7196, "step": 524 }, { "epoch": 0.02, "learning_rate": 1.99976719069523e-05, "loss": 1.8612, "step": 525 }, { "epoch": 0.02, "learning_rate": 1.999764902746496e-05, "loss": 1.7309, "step": 526 }, { "epoch": 0.02, "learning_rate": 1.999762603611506e-05, "loss": 1.7991, "step": 527 }, { "epoch": 0.02, "learning_rate": 1.9997602932902866e-05, "loss": 1.7691, "step": 528 }, { "epoch": 0.02, "learning_rate": 1.9997579717828634e-05, "loss": 1.7297, "step": 529 }, { "epoch": 0.02, "learning_rate": 1.9997556390892623e-05, "loss": 1.6471, "step": 530 }, { "epoch": 0.02, "learning_rate": 1.9997532952095093e-05, "loss": 1.6821, "step": 531 }, { "epoch": 0.02, "learning_rate": 1.9997509401436308e-05, "loss": 1.6749, "step": 532 }, { "epoch": 0.02, "learning_rate": 1.999748573891653e-05, "loss": 1.7456, "step": 533 }, { "epoch": 0.02, "learning_rate": 1.9997461964536024e-05, "loss": 1.6575, "step": 534 }, { "epoch": 0.02, "learning_rate": 1.9997438078295057e-05, "loss": 1.7717, "step": 535 }, { "epoch": 0.02, "learning_rate": 1.99974140801939e-05, "loss": 1.8204, "step": 536 }, { "epoch": 0.02, "learning_rate": 1.999738997023281e-05, "loss": 1.6701, "step": 537 }, { "epoch": 0.02, "learning_rate": 1.9997365748412068e-05, "loss": 1.7214, "step": 538 }, { "epoch": 0.02, "learning_rate": 1.999734141473194e-05, "loss": 1.7448, "step": 539 }, { "epoch": 0.02, "learning_rate": 1.9997316969192696e-05, "loss": 1.7889, "step": 540 }, { "epoch": 0.02, "learning_rate": 1.999729241179462e-05, "loss": 1.742, "step": 541 }, { "epoch": 0.02, "learning_rate": 1.9997267742537974e-05, "loss": 1.7195, "step": 542 }, { "epoch": 0.02, "learning_rate": 1.9997242961423043e-05, "loss": 1.8383, "step": 543 }, { "epoch": 0.02, "learning_rate": 1.9997218068450096e-05, "loss": 1.6979, "step": 544 }, { "epoch": 0.02, "learning_rate": 1.999719306361942e-05, "loss": 1.6974, "step": 545 }, { "epoch": 0.02, "learning_rate": 1.9997167946931293e-05, "loss": 1.8253, "step": 546 }, { "epoch": 0.02, "learning_rate": 1.999714271838599e-05, "loss": 1.7511, "step": 547 }, { "epoch": 0.02, "learning_rate": 1.9997117377983803e-05, "loss": 1.6503, "step": 548 }, { "epoch": 0.02, "learning_rate": 1.9997091925725006e-05, "loss": 1.7162, "step": 549 }, { "epoch": 0.02, "learning_rate": 1.9997066361609887e-05, "loss": 1.6878, "step": 550 }, { "epoch": 0.02, "learning_rate": 1.999704068563873e-05, "loss": 1.6422, "step": 551 }, { "epoch": 0.02, "learning_rate": 1.9997014897811834e-05, "loss": 1.6752, "step": 552 }, { "epoch": 0.02, "learning_rate": 1.9996988998129475e-05, "loss": 1.6447, "step": 553 }, { "epoch": 0.02, "learning_rate": 1.9996962986591943e-05, "loss": 1.6488, "step": 554 }, { "epoch": 0.02, "learning_rate": 1.9996936863199537e-05, "loss": 1.8367, "step": 555 }, { "epoch": 0.02, "learning_rate": 1.999691062795254e-05, "loss": 1.7192, "step": 556 }, { "epoch": 0.02, "learning_rate": 1.999688428085125e-05, "loss": 1.7126, "step": 557 }, { "epoch": 0.02, "learning_rate": 1.9996857821895968e-05, "loss": 1.6974, "step": 558 }, { "epoch": 0.02, "learning_rate": 1.9996831251086982e-05, "loss": 1.7627, "step": 559 }, { "epoch": 0.02, "learning_rate": 1.999680456842459e-05, "loss": 1.8165, "step": 560 }, { "epoch": 0.02, "learning_rate": 1.9996777773909093e-05, "loss": 1.7664, "step": 561 }, { "epoch": 0.02, "learning_rate": 1.999675086754079e-05, "loss": 1.6643, "step": 562 }, { "epoch": 0.02, "learning_rate": 1.9996723849319978e-05, "loss": 1.771, "step": 563 }, { "epoch": 0.02, "learning_rate": 1.999669671924697e-05, "loss": 1.7321, "step": 564 }, { "epoch": 0.02, "learning_rate": 1.9996669477322056e-05, "loss": 1.7404, "step": 565 }, { "epoch": 0.02, "learning_rate": 1.999664212354555e-05, "loss": 1.7827, "step": 566 }, { "epoch": 0.02, "learning_rate": 1.999661465791776e-05, "loss": 1.7573, "step": 567 }, { "epoch": 0.02, "learning_rate": 1.9996587080438984e-05, "loss": 1.7111, "step": 568 }, { "epoch": 0.02, "learning_rate": 1.9996559391109537e-05, "loss": 1.6717, "step": 569 }, { "epoch": 0.02, "learning_rate": 1.9996531589929725e-05, "loss": 1.6994, "step": 570 }, { "epoch": 0.02, "learning_rate": 1.9996503676899863e-05, "loss": 1.7574, "step": 571 }, { "epoch": 0.02, "learning_rate": 1.9996475652020263e-05, "loss": 1.7823, "step": 572 }, { "epoch": 0.02, "learning_rate": 1.9996447515291236e-05, "loss": 1.6845, "step": 573 }, { "epoch": 0.02, "learning_rate": 1.9996419266713097e-05, "loss": 1.6744, "step": 574 }, { "epoch": 0.02, "learning_rate": 1.9996390906286165e-05, "loss": 1.744, "step": 575 }, { "epoch": 0.02, "learning_rate": 1.9996362434010754e-05, "loss": 1.7123, "step": 576 }, { "epoch": 0.02, "learning_rate": 1.9996333849887185e-05, "loss": 1.6771, "step": 577 }, { "epoch": 0.02, "learning_rate": 1.9996305153915777e-05, "loss": 1.697, "step": 578 }, { "epoch": 0.02, "learning_rate": 1.9996276346096847e-05, "loss": 1.608, "step": 579 }, { "epoch": 0.02, "learning_rate": 1.9996247426430726e-05, "loss": 1.6045, "step": 580 }, { "epoch": 0.02, "learning_rate": 1.999621839491773e-05, "loss": 1.5849, "step": 581 }, { "epoch": 0.02, "learning_rate": 1.999618925155819e-05, "loss": 1.6867, "step": 582 }, { "epoch": 0.02, "learning_rate": 1.9996159996352426e-05, "loss": 1.6069, "step": 583 }, { "epoch": 0.02, "learning_rate": 1.999613062930077e-05, "loss": 1.6674, "step": 584 }, { "epoch": 0.02, "learning_rate": 1.9996101150403543e-05, "loss": 1.6342, "step": 585 }, { "epoch": 0.02, "learning_rate": 1.9996071559661086e-05, "loss": 1.7615, "step": 586 }, { "epoch": 0.02, "learning_rate": 1.9996041857073725e-05, "loss": 1.6228, "step": 587 }, { "epoch": 0.02, "learning_rate": 1.999601204264179e-05, "loss": 1.7229, "step": 588 }, { "epoch": 0.02, "learning_rate": 1.9995982116365614e-05, "loss": 1.6625, "step": 589 }, { "epoch": 0.02, "learning_rate": 1.999595207824554e-05, "loss": 1.698, "step": 590 }, { "epoch": 0.02, "learning_rate": 1.9995921928281893e-05, "loss": 1.5784, "step": 591 }, { "epoch": 0.02, "learning_rate": 1.999589166647502e-05, "loss": 1.6698, "step": 592 }, { "epoch": 0.02, "learning_rate": 1.9995861292825252e-05, "loss": 1.6341, "step": 593 }, { "epoch": 0.02, "learning_rate": 1.9995830807332934e-05, "loss": 1.6526, "step": 594 }, { "epoch": 0.02, "learning_rate": 1.9995800209998406e-05, "loss": 1.6589, "step": 595 }, { "epoch": 0.02, "learning_rate": 1.9995769500822007e-05, "loss": 1.6467, "step": 596 }, { "epoch": 0.02, "learning_rate": 1.9995738679804086e-05, "loss": 1.6977, "step": 597 }, { "epoch": 0.02, "learning_rate": 1.9995707746944984e-05, "loss": 1.5877, "step": 598 }, { "epoch": 0.02, "learning_rate": 1.999567670224505e-05, "loss": 1.6845, "step": 599 }, { "epoch": 0.02, "learning_rate": 1.9995645545704624e-05, "loss": 1.5962, "step": 600 }, { "epoch": 0.02, "learning_rate": 1.9995614277324066e-05, "loss": 1.5771, "step": 601 }, { "epoch": 0.02, "learning_rate": 1.9995582897103717e-05, "loss": 1.64, "step": 602 }, { "epoch": 0.02, "learning_rate": 1.999555140504393e-05, "loss": 1.6044, "step": 603 }, { "epoch": 0.02, "learning_rate": 1.9995519801145063e-05, "loss": 1.6436, "step": 604 }, { "epoch": 0.02, "learning_rate": 1.9995488085407463e-05, "loss": 1.5831, "step": 605 }, { "epoch": 0.02, "learning_rate": 1.9995456257831484e-05, "loss": 1.7381, "step": 606 }, { "epoch": 0.02, "learning_rate": 1.999542431841749e-05, "loss": 1.6341, "step": 607 }, { "epoch": 0.02, "learning_rate": 1.999539226716583e-05, "loss": 1.6353, "step": 608 }, { "epoch": 0.02, "learning_rate": 1.999536010407687e-05, "loss": 1.6396, "step": 609 }, { "epoch": 0.02, "learning_rate": 1.999532782915096e-05, "loss": 1.5173, "step": 610 }, { "epoch": 0.02, "learning_rate": 1.999529544238847e-05, "loss": 1.6112, "step": 611 }, { "epoch": 0.02, "learning_rate": 1.999526294378976e-05, "loss": 1.6268, "step": 612 }, { "epoch": 0.02, "learning_rate": 1.9995230333355193e-05, "loss": 1.7229, "step": 613 }, { "epoch": 0.02, "learning_rate": 1.9995197611085136e-05, "loss": 1.6339, "step": 614 }, { "epoch": 0.02, "learning_rate": 1.999516477697995e-05, "loss": 1.7528, "step": 615 }, { "epoch": 0.02, "learning_rate": 1.9995131831040006e-05, "loss": 1.7089, "step": 616 }, { "epoch": 0.02, "learning_rate": 1.999509877326567e-05, "loss": 1.6476, "step": 617 }, { "epoch": 0.02, "learning_rate": 1.9995065603657317e-05, "loss": 1.7376, "step": 618 }, { "epoch": 0.02, "learning_rate": 1.9995032322215313e-05, "loss": 1.648, "step": 619 }, { "epoch": 0.02, "learning_rate": 1.9994998928940036e-05, "loss": 1.7282, "step": 620 }, { "epoch": 0.02, "learning_rate": 1.9994965423831853e-05, "loss": 1.7507, "step": 621 }, { "epoch": 0.02, "learning_rate": 1.9994931806891143e-05, "loss": 1.6843, "step": 622 }, { "epoch": 0.02, "learning_rate": 1.999489807811828e-05, "loss": 1.6186, "step": 623 }, { "epoch": 0.02, "learning_rate": 1.9994864237513645e-05, "loss": 1.6243, "step": 624 }, { "epoch": 0.02, "learning_rate": 1.999483028507761e-05, "loss": 1.6403, "step": 625 }, { "epoch": 0.02, "learning_rate": 1.9994796220810562e-05, "loss": 1.6816, "step": 626 }, { "epoch": 0.02, "learning_rate": 1.999476204471288e-05, "loss": 1.6629, "step": 627 }, { "epoch": 0.02, "learning_rate": 1.9994727756784945e-05, "loss": 1.7397, "step": 628 }, { "epoch": 0.02, "learning_rate": 1.999469335702714e-05, "loss": 1.6844, "step": 629 }, { "epoch": 0.02, "learning_rate": 1.9994658845439853e-05, "loss": 1.7334, "step": 630 }, { "epoch": 0.02, "learning_rate": 1.9994624222023465e-05, "loss": 1.7702, "step": 631 }, { "epoch": 0.02, "learning_rate": 1.999458948677837e-05, "loss": 1.7093, "step": 632 }, { "epoch": 0.02, "learning_rate": 1.999455463970495e-05, "loss": 1.8079, "step": 633 }, { "epoch": 0.02, "learning_rate": 1.99945196808036e-05, "loss": 1.7913, "step": 634 }, { "epoch": 0.02, "learning_rate": 1.999448461007471e-05, "loss": 1.7716, "step": 635 }, { "epoch": 0.02, "learning_rate": 1.999444942751867e-05, "loss": 1.663, "step": 636 }, { "epoch": 0.02, "learning_rate": 1.9994414133135877e-05, "loss": 1.6973, "step": 637 }, { "epoch": 0.02, "learning_rate": 1.9994378726926724e-05, "loss": 1.8234, "step": 638 }, { "epoch": 0.02, "learning_rate": 1.9994343208891607e-05, "loss": 1.8012, "step": 639 }, { "epoch": 0.02, "learning_rate": 1.9994307579030925e-05, "loss": 1.9122, "step": 640 }, { "epoch": 0.02, "learning_rate": 1.9994271837345072e-05, "loss": 1.7306, "step": 641 }, { "epoch": 0.02, "learning_rate": 1.9994235983834455e-05, "loss": 1.7048, "step": 642 }, { "epoch": 0.02, "learning_rate": 1.9994200018499473e-05, "loss": 1.7995, "step": 643 }, { "epoch": 0.02, "learning_rate": 1.9994163941340523e-05, "loss": 1.7367, "step": 644 }, { "epoch": 0.02, "learning_rate": 1.9994127752358014e-05, "loss": 1.762, "step": 645 }, { "epoch": 0.02, "learning_rate": 1.999409145155235e-05, "loss": 1.6759, "step": 646 }, { "epoch": 0.02, "learning_rate": 1.9994055038923938e-05, "loss": 1.745, "step": 647 }, { "epoch": 0.02, "learning_rate": 1.9994018514473183e-05, "loss": 1.634, "step": 648 }, { "epoch": 0.02, "learning_rate": 1.9993981878200494e-05, "loss": 1.6431, "step": 649 }, { "epoch": 0.02, "learning_rate": 1.999394513010628e-05, "loss": 1.5813, "step": 650 }, { "epoch": 0.02, "learning_rate": 1.999390827019096e-05, "loss": 1.6412, "step": 651 }, { "epoch": 0.02, "learning_rate": 1.9993871298454938e-05, "loss": 1.5756, "step": 652 }, { "epoch": 0.02, "learning_rate": 1.9993834214898626e-05, "loss": 1.736, "step": 653 }, { "epoch": 0.02, "learning_rate": 1.9993797019522446e-05, "loss": 1.6885, "step": 654 }, { "epoch": 0.02, "learning_rate": 1.9993759712326814e-05, "loss": 1.6707, "step": 655 }, { "epoch": 0.02, "learning_rate": 1.999372229331214e-05, "loss": 1.6469, "step": 656 }, { "epoch": 0.02, "learning_rate": 1.9993684762478846e-05, "loss": 1.6038, "step": 657 }, { "epoch": 0.02, "learning_rate": 1.9993647119827355e-05, "loss": 1.7185, "step": 658 }, { "epoch": 0.02, "learning_rate": 1.999360936535809e-05, "loss": 1.5861, "step": 659 }, { "epoch": 0.02, "learning_rate": 1.9993571499071465e-05, "loss": 1.6543, "step": 660 }, { "epoch": 0.02, "learning_rate": 1.999353352096791e-05, "loss": 1.7426, "step": 661 }, { "epoch": 0.02, "learning_rate": 1.999349543104785e-05, "loss": 1.608, "step": 662 }, { "epoch": 0.02, "learning_rate": 1.9993457229311707e-05, "loss": 1.6389, "step": 663 }, { "epoch": 0.02, "learning_rate": 1.9993418915759913e-05, "loss": 1.5952, "step": 664 }, { "epoch": 0.02, "learning_rate": 1.9993380490392895e-05, "loss": 1.5503, "step": 665 }, { "epoch": 0.02, "learning_rate": 1.999334195321108e-05, "loss": 1.6662, "step": 666 }, { "epoch": 0.02, "learning_rate": 1.9993303304214904e-05, "loss": 1.6334, "step": 667 }, { "epoch": 0.02, "learning_rate": 1.9993264543404794e-05, "loss": 1.6958, "step": 668 }, { "epoch": 0.02, "learning_rate": 1.9993225670781192e-05, "loss": 1.6585, "step": 669 }, { "epoch": 0.02, "learning_rate": 1.9993186686344524e-05, "loss": 1.627, "step": 670 }, { "epoch": 0.02, "learning_rate": 1.9993147590095232e-05, "loss": 1.6388, "step": 671 }, { "epoch": 0.02, "learning_rate": 1.9993108382033752e-05, "loss": 1.6022, "step": 672 }, { "epoch": 0.02, "learning_rate": 1.999306906216052e-05, "loss": 1.7699, "step": 673 }, { "epoch": 0.02, "learning_rate": 1.999302963047598e-05, "loss": 1.6571, "step": 674 }, { "epoch": 0.02, "learning_rate": 1.999299008698057e-05, "loss": 1.7136, "step": 675 }, { "epoch": 0.02, "learning_rate": 1.9992950431674732e-05, "loss": 1.6184, "step": 676 }, { "epoch": 0.02, "learning_rate": 1.9992910664558917e-05, "loss": 1.5993, "step": 677 }, { "epoch": 0.02, "learning_rate": 1.9992870785633563e-05, "loss": 1.6118, "step": 678 }, { "epoch": 0.02, "learning_rate": 1.9992830794899114e-05, "loss": 1.6144, "step": 679 }, { "epoch": 0.02, "learning_rate": 1.9992790692356022e-05, "loss": 1.5471, "step": 680 }, { "epoch": 0.02, "learning_rate": 1.999275047800474e-05, "loss": 1.6966, "step": 681 }, { "epoch": 0.02, "learning_rate": 1.9992710151845706e-05, "loss": 1.66, "step": 682 }, { "epoch": 0.02, "learning_rate": 1.999266971387938e-05, "loss": 1.687, "step": 683 }, { "epoch": 0.02, "learning_rate": 1.999262916410621e-05, "loss": 1.7545, "step": 684 }, { "epoch": 0.02, "learning_rate": 1.999258850252666e-05, "loss": 1.6397, "step": 685 }, { "epoch": 0.02, "learning_rate": 1.999254772914117e-05, "loss": 1.5977, "step": 686 }, { "epoch": 0.02, "learning_rate": 1.9992506843950207e-05, "loss": 1.6869, "step": 687 }, { "epoch": 0.02, "learning_rate": 1.9992465846954224e-05, "loss": 1.6969, "step": 688 }, { "epoch": 0.02, "learning_rate": 1.9992424738153676e-05, "loss": 1.5805, "step": 689 }, { "epoch": 0.02, "learning_rate": 1.9992383517549032e-05, "loss": 1.5886, "step": 690 }, { "epoch": 0.02, "learning_rate": 1.9992342185140748e-05, "loss": 1.6357, "step": 691 }, { "epoch": 0.02, "learning_rate": 1.9992300740929285e-05, "loss": 1.6699, "step": 692 }, { "epoch": 0.02, "learning_rate": 1.9992259184915115e-05, "loss": 1.6284, "step": 693 }, { "epoch": 0.02, "learning_rate": 1.999221751709869e-05, "loss": 1.7226, "step": 694 }, { "epoch": 0.02, "learning_rate": 1.9992175737480487e-05, "loss": 1.7684, "step": 695 }, { "epoch": 0.02, "learning_rate": 1.999213384606097e-05, "loss": 1.6376, "step": 696 }, { "epoch": 0.02, "learning_rate": 1.99920918428406e-05, "loss": 1.5482, "step": 697 }, { "epoch": 0.02, "learning_rate": 1.999204972781986e-05, "loss": 1.5492, "step": 698 }, { "epoch": 0.02, "learning_rate": 1.9992007500999216e-05, "loss": 1.4487, "step": 699 }, { "epoch": 0.02, "learning_rate": 1.999196516237914e-05, "loss": 1.5485, "step": 700 }, { "epoch": 0.02, "learning_rate": 1.9991922711960104e-05, "loss": 1.659, "step": 701 }, { "epoch": 0.02, "learning_rate": 1.9991880149742583e-05, "loss": 1.7121, "step": 702 }, { "epoch": 0.02, "learning_rate": 1.9991837475727058e-05, "loss": 1.6788, "step": 703 }, { "epoch": 0.02, "learning_rate": 1.9991794689913998e-05, "loss": 1.6016, "step": 704 }, { "epoch": 0.02, "learning_rate": 1.9991751792303892e-05, "loss": 1.7258, "step": 705 }, { "epoch": 0.02, "learning_rate": 1.9991708782897214e-05, "loss": 1.784, "step": 706 }, { "epoch": 0.02, "learning_rate": 1.9991665661694445e-05, "loss": 1.6918, "step": 707 }, { "epoch": 0.02, "learning_rate": 1.999162242869607e-05, "loss": 1.6389, "step": 708 }, { "epoch": 0.02, "learning_rate": 1.9991579083902574e-05, "loss": 1.6335, "step": 709 }, { "epoch": 0.02, "learning_rate": 1.9991535627314433e-05, "loss": 1.8193, "step": 710 }, { "epoch": 0.02, "learning_rate": 1.9991492058932143e-05, "loss": 1.6749, "step": 711 }, { "epoch": 0.02, "learning_rate": 1.9991448378756187e-05, "loss": 1.7149, "step": 712 }, { "epoch": 0.02, "learning_rate": 1.9991404586787055e-05, "loss": 1.6868, "step": 713 }, { "epoch": 0.02, "learning_rate": 1.9991360683025238e-05, "loss": 1.7268, "step": 714 }, { "epoch": 0.02, "learning_rate": 1.9991316667471222e-05, "loss": 1.6403, "step": 715 }, { "epoch": 0.02, "learning_rate": 1.9991272540125507e-05, "loss": 1.7542, "step": 716 }, { "epoch": 0.02, "learning_rate": 1.9991228300988586e-05, "loss": 1.7039, "step": 717 }, { "epoch": 0.02, "learning_rate": 1.999118395006095e-05, "loss": 1.6498, "step": 718 }, { "epoch": 0.02, "learning_rate": 1.9991139487343092e-05, "loss": 1.7826, "step": 719 }, { "epoch": 0.02, "learning_rate": 1.9991094912835514e-05, "loss": 1.7366, "step": 720 }, { "epoch": 0.02, "learning_rate": 1.999105022653872e-05, "loss": 1.6881, "step": 721 }, { "epoch": 0.02, "learning_rate": 1.9991005428453204e-05, "loss": 1.6691, "step": 722 }, { "epoch": 0.02, "learning_rate": 1.9990960518579462e-05, "loss": 1.7753, "step": 723 }, { "epoch": 0.02, "learning_rate": 1.9990915496918007e-05, "loss": 1.7289, "step": 724 }, { "epoch": 0.02, "learning_rate": 1.999087036346934e-05, "loss": 1.7188, "step": 725 }, { "epoch": 0.02, "learning_rate": 1.9990825118233958e-05, "loss": 1.7902, "step": 726 }, { "epoch": 0.02, "learning_rate": 1.9990779761212376e-05, "loss": 1.7228, "step": 727 }, { "epoch": 0.02, "learning_rate": 1.9990734292405102e-05, "loss": 1.6856, "step": 728 }, { "epoch": 0.02, "learning_rate": 1.999068871181264e-05, "loss": 1.7205, "step": 729 }, { "epoch": 0.02, "learning_rate": 1.9990643019435502e-05, "loss": 1.6737, "step": 730 }, { "epoch": 0.02, "learning_rate": 1.9990597215274196e-05, "loss": 1.6916, "step": 731 }, { "epoch": 0.02, "learning_rate": 1.999055129932924e-05, "loss": 1.7119, "step": 732 }, { "epoch": 0.02, "learning_rate": 1.9990505271601143e-05, "loss": 1.6747, "step": 733 }, { "epoch": 0.02, "learning_rate": 1.9990459132090427e-05, "loss": 1.6765, "step": 734 }, { "epoch": 0.02, "learning_rate": 1.9990412880797595e-05, "loss": 1.6359, "step": 735 }, { "epoch": 0.02, "learning_rate": 1.9990366517723178e-05, "loss": 1.6564, "step": 736 }, { "epoch": 0.02, "learning_rate": 1.999032004286769e-05, "loss": 1.6012, "step": 737 }, { "epoch": 0.02, "learning_rate": 1.999027345623165e-05, "loss": 1.6027, "step": 738 }, { "epoch": 0.02, "learning_rate": 1.9990226757815582e-05, "loss": 1.6295, "step": 739 }, { "epoch": 0.02, "learning_rate": 1.9990179947620003e-05, "loss": 1.6586, "step": 740 }, { "epoch": 0.02, "learning_rate": 1.999013302564544e-05, "loss": 1.6657, "step": 741 }, { "epoch": 0.02, "learning_rate": 1.999008599189242e-05, "loss": 1.5813, "step": 742 }, { "epoch": 0.02, "learning_rate": 1.9990038846361468e-05, "loss": 1.5426, "step": 743 }, { "epoch": 0.02, "learning_rate": 1.9989991589053107e-05, "loss": 1.6494, "step": 744 }, { "epoch": 0.02, "learning_rate": 1.9989944219967873e-05, "loss": 1.5823, "step": 745 }, { "epoch": 0.02, "learning_rate": 1.998989673910629e-05, "loss": 1.6205, "step": 746 }, { "epoch": 0.02, "learning_rate": 1.9989849146468897e-05, "loss": 1.6836, "step": 747 }, { "epoch": 0.02, "learning_rate": 1.9989801442056218e-05, "loss": 1.7232, "step": 748 }, { "epoch": 0.02, "learning_rate": 1.998975362586879e-05, "loss": 1.647, "step": 749 }, { "epoch": 0.03, "learning_rate": 1.998970569790715e-05, "loss": 1.6115, "step": 750 }, { "epoch": 0.03, "learning_rate": 1.998965765817183e-05, "loss": 1.6144, "step": 751 }, { "epoch": 0.03, "learning_rate": 1.9989609506663373e-05, "loss": 1.6567, "step": 752 }, { "epoch": 0.03, "learning_rate": 1.9989561243382313e-05, "loss": 1.7209, "step": 753 }, { "epoch": 0.03, "learning_rate": 1.9989512868329187e-05, "loss": 1.6456, "step": 754 }, { "epoch": 0.03, "learning_rate": 1.9989464381504548e-05, "loss": 1.5344, "step": 755 }, { "epoch": 0.03, "learning_rate": 1.998941578290893e-05, "loss": 1.6915, "step": 756 }, { "epoch": 0.03, "learning_rate": 1.9989367072542875e-05, "loss": 1.6844, "step": 757 }, { "epoch": 0.03, "learning_rate": 1.9989318250406935e-05, "loss": 1.68, "step": 758 }, { "epoch": 0.03, "learning_rate": 1.998926931650165e-05, "loss": 1.5659, "step": 759 }, { "epoch": 0.03, "learning_rate": 1.9989220270827573e-05, "loss": 1.6397, "step": 760 }, { "epoch": 0.03, "learning_rate": 1.998917111338525e-05, "loss": 1.5875, "step": 761 }, { "epoch": 0.03, "learning_rate": 1.9989121844175226e-05, "loss": 1.709, "step": 762 }, { "epoch": 0.03, "learning_rate": 1.998907246319806e-05, "loss": 1.696, "step": 763 }, { "epoch": 0.03, "learning_rate": 1.9989022970454302e-05, "loss": 1.6572, "step": 764 }, { "epoch": 0.03, "learning_rate": 1.9988973365944505e-05, "loss": 1.6794, "step": 765 }, { "epoch": 0.03, "learning_rate": 1.9988923649669223e-05, "loss": 1.6808, "step": 766 }, { "epoch": 0.03, "learning_rate": 1.9988873821629017e-05, "loss": 1.6831, "step": 767 }, { "epoch": 0.03, "learning_rate": 1.9988823881824436e-05, "loss": 1.6426, "step": 768 }, { "epoch": 0.03, "learning_rate": 1.998877383025605e-05, "loss": 1.7028, "step": 769 }, { "epoch": 0.03, "learning_rate": 1.998872366692441e-05, "loss": 1.7103, "step": 770 }, { "epoch": 0.03, "learning_rate": 1.9988673391830082e-05, "loss": 1.5658, "step": 771 }, { "epoch": 0.03, "learning_rate": 1.9988623004973625e-05, "loss": 1.587, "step": 772 }, { "epoch": 0.03, "learning_rate": 1.9988572506355605e-05, "loss": 1.5978, "step": 773 }, { "epoch": 0.03, "learning_rate": 1.9988521895976585e-05, "loss": 1.6715, "step": 774 }, { "epoch": 0.03, "learning_rate": 1.9988471173837135e-05, "loss": 1.6315, "step": 775 }, { "epoch": 0.03, "learning_rate": 1.9988420339937823e-05, "loss": 1.59, "step": 776 }, { "epoch": 0.03, "learning_rate": 1.998836939427921e-05, "loss": 1.5688, "step": 777 }, { "epoch": 0.03, "learning_rate": 1.9988318336861875e-05, "loss": 1.5758, "step": 778 }, { "epoch": 0.03, "learning_rate": 1.998826716768638e-05, "loss": 1.6761, "step": 779 }, { "epoch": 0.03, "learning_rate": 1.9988215886753308e-05, "loss": 1.6455, "step": 780 }, { "epoch": 0.03, "learning_rate": 1.9988164494063226e-05, "loss": 1.6121, "step": 781 }, { "epoch": 0.03, "learning_rate": 1.9988112989616716e-05, "loss": 1.6186, "step": 782 }, { "epoch": 0.03, "learning_rate": 1.9988061373414342e-05, "loss": 1.5525, "step": 783 }, { "epoch": 0.03, "learning_rate": 1.998800964545669e-05, "loss": 1.5202, "step": 784 }, { "epoch": 0.03, "learning_rate": 1.998795780574434e-05, "loss": 1.635, "step": 785 }, { "epoch": 0.03, "learning_rate": 1.9987905854277867e-05, "loss": 1.4984, "step": 786 }, { "epoch": 0.03, "learning_rate": 1.9987853791057856e-05, "loss": 1.4887, "step": 787 }, { "epoch": 0.03, "learning_rate": 1.998780161608489e-05, "loss": 1.5479, "step": 788 }, { "epoch": 0.03, "learning_rate": 1.998774932935955e-05, "loss": 1.6016, "step": 789 }, { "epoch": 0.03, "learning_rate": 1.998769693088242e-05, "loss": 1.584, "step": 790 }, { "epoch": 0.03, "learning_rate": 1.9987644420654087e-05, "loss": 1.6503, "step": 791 }, { "epoch": 0.03, "learning_rate": 1.9987591798675142e-05, "loss": 1.673, "step": 792 }, { "epoch": 0.03, "learning_rate": 1.998753906494617e-05, "loss": 1.5949, "step": 793 }, { "epoch": 0.03, "learning_rate": 1.9987486219467764e-05, "loss": 1.6583, "step": 794 }, { "epoch": 0.03, "learning_rate": 1.9987433262240518e-05, "loss": 1.6225, "step": 795 }, { "epoch": 0.03, "learning_rate": 1.9987380193265014e-05, "loss": 1.6535, "step": 796 }, { "epoch": 0.03, "learning_rate": 1.9987327012541856e-05, "loss": 1.6931, "step": 797 }, { "epoch": 0.03, "learning_rate": 1.9987273720071633e-05, "loss": 1.7307, "step": 798 }, { "epoch": 0.03, "learning_rate": 1.9987220315854946e-05, "loss": 1.6649, "step": 799 }, { "epoch": 0.03, "learning_rate": 1.9987166799892387e-05, "loss": 1.6471, "step": 800 }, { "epoch": 0.03, "learning_rate": 1.9987113172184562e-05, "loss": 1.6447, "step": 801 }, { "epoch": 0.03, "learning_rate": 1.9987059432732065e-05, "loss": 1.7242, "step": 802 }, { "epoch": 0.03, "learning_rate": 1.99870055815355e-05, "loss": 1.6949, "step": 803 }, { "epoch": 0.03, "learning_rate": 1.9986951618595466e-05, "loss": 1.7578, "step": 804 }, { "epoch": 0.03, "learning_rate": 1.998689754391257e-05, "loss": 1.6464, "step": 805 }, { "epoch": 0.03, "learning_rate": 1.9986843357487417e-05, "loss": 1.6692, "step": 806 }, { "epoch": 0.03, "learning_rate": 1.9986789059320614e-05, "loss": 1.7155, "step": 807 }, { "epoch": 0.03, "learning_rate": 1.998673464941277e-05, "loss": 1.7845, "step": 808 }, { "epoch": 0.03, "learning_rate": 1.998668012776449e-05, "loss": 1.7596, "step": 809 }, { "epoch": 0.03, "learning_rate": 1.998662549437638e-05, "loss": 1.7601, "step": 810 }, { "epoch": 0.03, "learning_rate": 1.9986570749249062e-05, "loss": 1.7745, "step": 811 }, { "epoch": 0.03, "learning_rate": 1.998651589238314e-05, "loss": 1.7939, "step": 812 }, { "epoch": 0.03, "learning_rate": 1.998646092377923e-05, "loss": 1.9128, "step": 813 }, { "epoch": 0.03, "learning_rate": 1.9986405843437952e-05, "loss": 1.7594, "step": 814 }, { "epoch": 0.03, "learning_rate": 1.9986350651359916e-05, "loss": 1.6653, "step": 815 }, { "epoch": 0.03, "learning_rate": 1.9986295347545738e-05, "loss": 1.7741, "step": 816 }, { "epoch": 0.03, "learning_rate": 1.998623993199605e-05, "loss": 1.6526, "step": 817 }, { "epoch": 0.03, "learning_rate": 1.9986184404711453e-05, "loss": 1.6472, "step": 818 }, { "epoch": 0.03, "learning_rate": 1.998612876569258e-05, "loss": 1.7879, "step": 819 }, { "epoch": 0.03, "learning_rate": 1.9986073014940053e-05, "loss": 1.688, "step": 820 }, { "epoch": 0.03, "learning_rate": 1.9986017152454497e-05, "loss": 1.7066, "step": 821 }, { "epoch": 0.03, "learning_rate": 1.9985961178236526e-05, "loss": 1.7369, "step": 822 }, { "epoch": 0.03, "learning_rate": 1.9985905092286782e-05, "loss": 1.6942, "step": 823 }, { "epoch": 0.03, "learning_rate": 1.998584889460588e-05, "loss": 1.7528, "step": 824 }, { "epoch": 0.03, "learning_rate": 1.998579258519446e-05, "loss": 1.6951, "step": 825 }, { "epoch": 0.03, "learning_rate": 1.998573616405314e-05, "loss": 1.8, "step": 826 }, { "epoch": 0.03, "learning_rate": 1.998567963118256e-05, "loss": 1.7178, "step": 827 }, { "epoch": 0.03, "learning_rate": 1.9985622986583347e-05, "loss": 1.747, "step": 828 }, { "epoch": 0.03, "learning_rate": 1.9985566230256143e-05, "loss": 1.7252, "step": 829 }, { "epoch": 0.03, "learning_rate": 1.998550936220157e-05, "loss": 1.6324, "step": 830 }, { "epoch": 0.03, "learning_rate": 1.9985452382420277e-05, "loss": 1.7256, "step": 831 }, { "epoch": 0.03, "learning_rate": 1.9985395290912893e-05, "loss": 1.6906, "step": 832 }, { "epoch": 0.03, "learning_rate": 1.998533808768006e-05, "loss": 1.6628, "step": 833 }, { "epoch": 0.03, "learning_rate": 1.9985280772722423e-05, "loss": 1.6862, "step": 834 }, { "epoch": 0.03, "learning_rate": 1.9985223346040616e-05, "loss": 1.6303, "step": 835 }, { "epoch": 0.03, "learning_rate": 1.998516580763528e-05, "loss": 1.6522, "step": 836 }, { "epoch": 0.03, "learning_rate": 1.998510815750707e-05, "loss": 1.6619, "step": 837 }, { "epoch": 0.03, "learning_rate": 1.9985050395656617e-05, "loss": 1.7161, "step": 838 }, { "epoch": 0.03, "learning_rate": 1.9984992522084577e-05, "loss": 1.5775, "step": 839 }, { "epoch": 0.03, "learning_rate": 1.9984934536791594e-05, "loss": 1.6291, "step": 840 }, { "epoch": 0.03, "learning_rate": 1.998487643977832e-05, "loss": 1.7352, "step": 841 }, { "epoch": 0.03, "learning_rate": 1.998487643977832e-05, "loss": 1.6732, "step": 842 }, { "epoch": 0.03, "learning_rate": 1.9984818231045402e-05, "loss": 1.5828, "step": 843 }, { "epoch": 0.03, "learning_rate": 1.998475991059349e-05, "loss": 1.7131, "step": 844 }, { "epoch": 0.03, "learning_rate": 1.9984701478423238e-05, "loss": 1.7096, "step": 845 }, { "epoch": 0.03, "learning_rate": 1.99846429345353e-05, "loss": 1.7751, "step": 846 }, { "epoch": 0.03, "learning_rate": 1.9984584278930333e-05, "loss": 1.5829, "step": 847 }, { "epoch": 0.03, "learning_rate": 1.998452551160899e-05, "loss": 1.6731, "step": 848 }, { "epoch": 0.03, "learning_rate": 1.998446663257193e-05, "loss": 1.5584, "step": 849 }, { "epoch": 0.03, "learning_rate": 1.9984407641819812e-05, "loss": 1.7587, "step": 850 }, { "epoch": 0.03, "learning_rate": 1.9984348539353297e-05, "loss": 1.6743, "step": 851 }, { "epoch": 0.03, "learning_rate": 1.9984289325173042e-05, "loss": 1.6542, "step": 852 }, { "epoch": 0.03, "learning_rate": 1.9984229999279713e-05, "loss": 1.7733, "step": 853 }, { "epoch": 0.03, "learning_rate": 1.9984170561673977e-05, "loss": 1.6148, "step": 854 }, { "epoch": 0.03, "learning_rate": 1.9984111012356494e-05, "loss": 1.6006, "step": 855 }, { "epoch": 0.03, "learning_rate": 1.998405135132793e-05, "loss": 1.7126, "step": 856 }, { "epoch": 0.03, "learning_rate": 1.9983991578588954e-05, "loss": 1.6158, "step": 857 }, { "epoch": 0.03, "learning_rate": 1.998393169414024e-05, "loss": 1.7279, "step": 858 }, { "epoch": 0.03, "learning_rate": 1.9983871697982448e-05, "loss": 1.7035, "step": 859 }, { "epoch": 0.03, "learning_rate": 1.9983811590116257e-05, "loss": 1.6646, "step": 860 }, { "epoch": 0.03, "learning_rate": 1.9983751370542334e-05, "loss": 1.5708, "step": 861 }, { "epoch": 0.03, "learning_rate": 1.9983691039261358e-05, "loss": 1.608, "step": 862 }, { "epoch": 0.03, "learning_rate": 1.9983630596274e-05, "loss": 1.6763, "step": 863 }, { "epoch": 0.03, "learning_rate": 1.9983570041580943e-05, "loss": 1.6307, "step": 864 }, { "epoch": 0.03, "learning_rate": 1.9983509375182853e-05, "loss": 1.6947, "step": 865 }, { "epoch": 0.03, "learning_rate": 1.9983448597080423e-05, "loss": 1.6787, "step": 866 }, { "epoch": 0.03, "learning_rate": 1.9983387707274317e-05, "loss": 1.6979, "step": 867 }, { "epoch": 0.03, "learning_rate": 1.998332670576523e-05, "loss": 1.8057, "step": 868 }, { "epoch": 0.03, "learning_rate": 1.9983265592553838e-05, "loss": 1.6643, "step": 869 }, { "epoch": 0.03, "learning_rate": 1.9983204367640823e-05, "loss": 1.7037, "step": 870 }, { "epoch": 0.03, "learning_rate": 1.998314303102688e-05, "loss": 1.6919, "step": 871 }, { "epoch": 0.03, "learning_rate": 1.9983081582712684e-05, "loss": 1.7903, "step": 872 }, { "epoch": 0.03, "learning_rate": 1.9983020022698927e-05, "loss": 1.7078, "step": 873 }, { "epoch": 0.03, "learning_rate": 1.9982958350986296e-05, "loss": 1.6435, "step": 874 }, { "epoch": 0.03, "learning_rate": 1.9982896567575487e-05, "loss": 1.6562, "step": 875 }, { "epoch": 0.03, "learning_rate": 1.9982834672467186e-05, "loss": 1.6707, "step": 876 }, { "epoch": 0.03, "learning_rate": 1.9982772665662083e-05, "loss": 1.6473, "step": 877 }, { "epoch": 0.03, "learning_rate": 1.998271054716088e-05, "loss": 1.7648, "step": 878 }, { "epoch": 0.03, "learning_rate": 1.9982648316964264e-05, "loss": 1.7189, "step": 879 }, { "epoch": 0.03, "learning_rate": 1.9982585975072938e-05, "loss": 1.5981, "step": 880 }, { "epoch": 0.03, "learning_rate": 1.9982523521487594e-05, "loss": 1.6148, "step": 881 }, { "epoch": 0.03, "learning_rate": 1.9982460956208932e-05, "loss": 1.631, "step": 882 }, { "epoch": 0.03, "learning_rate": 1.9982398279237657e-05, "loss": 1.6805, "step": 883 }, { "epoch": 0.03, "learning_rate": 1.9982335490574462e-05, "loss": 1.624, "step": 884 }, { "epoch": 0.03, "learning_rate": 1.9982272590220052e-05, "loss": 1.7178, "step": 885 }, { "epoch": 0.03, "learning_rate": 1.9982209578175136e-05, "loss": 1.6421, "step": 886 }, { "epoch": 0.03, "learning_rate": 1.998214645444042e-05, "loss": 1.7072, "step": 887 }, { "epoch": 0.03, "learning_rate": 1.99820832190166e-05, "loss": 1.7068, "step": 888 }, { "epoch": 0.03, "learning_rate": 1.998201987190439e-05, "loss": 1.5889, "step": 889 }, { "epoch": 0.03, "learning_rate": 1.99819564131045e-05, "loss": 1.6803, "step": 890 }, { "epoch": 0.03, "learning_rate": 1.998189284261764e-05, "loss": 1.7994, "step": 891 }, { "epoch": 0.03, "learning_rate": 1.9981829160444515e-05, "loss": 1.6284, "step": 892 }, { "epoch": 0.03, "learning_rate": 1.9981765366585844e-05, "loss": 1.5951, "step": 893 }, { "epoch": 0.03, "learning_rate": 1.998170146104234e-05, "loss": 1.6486, "step": 894 }, { "epoch": 0.03, "learning_rate": 1.9981637443814717e-05, "loss": 1.636, "step": 895 }, { "epoch": 0.03, "learning_rate": 1.998157331490369e-05, "loss": 1.6497, "step": 896 }, { "epoch": 0.03, "learning_rate": 1.998150907430998e-05, "loss": 1.7851, "step": 897 }, { "epoch": 0.03, "learning_rate": 1.99814447220343e-05, "loss": 1.6899, "step": 898 }, { "epoch": 0.03, "learning_rate": 1.9981380258077373e-05, "loss": 1.6245, "step": 899 }, { "epoch": 0.03, "learning_rate": 1.9981315682439926e-05, "loss": 1.6304, "step": 900 }, { "epoch": 0.03, "learning_rate": 1.9981250995122673e-05, "loss": 1.6831, "step": 901 }, { "epoch": 0.03, "learning_rate": 1.998118619612634e-05, "loss": 1.6148, "step": 902 }, { "epoch": 0.03, "learning_rate": 1.9981121285451657e-05, "loss": 1.5908, "step": 903 }, { "epoch": 0.03, "learning_rate": 1.9981056263099347e-05, "loss": 1.6673, "step": 904 }, { "epoch": 0.03, "learning_rate": 1.998099112907013e-05, "loss": 1.7509, "step": 905 }, { "epoch": 0.03, "learning_rate": 1.998092588336475e-05, "loss": 1.7374, "step": 906 }, { "epoch": 0.03, "learning_rate": 1.9980860525983924e-05, "loss": 1.5593, "step": 907 }, { "epoch": 0.03, "learning_rate": 1.998079505692839e-05, "loss": 1.7093, "step": 908 }, { "epoch": 0.03, "learning_rate": 1.9980729476198876e-05, "loss": 1.6856, "step": 909 }, { "epoch": 0.03, "learning_rate": 1.998066378379612e-05, "loss": 1.7234, "step": 910 }, { "epoch": 0.03, "learning_rate": 1.9980597979720857e-05, "loss": 1.6131, "step": 911 }, { "epoch": 0.03, "learning_rate": 1.9980532063973817e-05, "loss": 1.6931, "step": 912 }, { "epoch": 0.03, "learning_rate": 1.9980466036555746e-05, "loss": 1.6481, "step": 913 }, { "epoch": 0.03, "learning_rate": 1.998039989746738e-05, "loss": 1.5679, "step": 914 }, { "epoch": 0.03, "learning_rate": 1.9980333646709455e-05, "loss": 1.5974, "step": 915 }, { "epoch": 0.03, "learning_rate": 1.9980267284282718e-05, "loss": 1.6557, "step": 916 }, { "epoch": 0.03, "learning_rate": 1.9980200810187907e-05, "loss": 1.5759, "step": 917 }, { "epoch": 0.03, "learning_rate": 1.998013422442577e-05, "loss": 1.5331, "step": 918 }, { "epoch": 0.03, "learning_rate": 1.9980067526997044e-05, "loss": 1.6656, "step": 919 }, { "epoch": 0.03, "learning_rate": 1.9980000717902485e-05, "loss": 1.5536, "step": 920 }, { "epoch": 0.03, "learning_rate": 1.9979933797142834e-05, "loss": 1.6418, "step": 921 }, { "epoch": 0.03, "learning_rate": 1.9979866764718846e-05, "loss": 1.5756, "step": 922 }, { "epoch": 0.03, "learning_rate": 1.9979799620631264e-05, "loss": 1.6119, "step": 923 }, { "epoch": 0.03, "learning_rate": 1.9979732364880844e-05, "loss": 1.6024, "step": 924 }, { "epoch": 0.03, "learning_rate": 1.997966499746834e-05, "loss": 1.6278, "step": 925 }, { "epoch": 0.03, "learning_rate": 1.9979597518394494e-05, "loss": 1.5607, "step": 926 }, { "epoch": 0.03, "learning_rate": 1.9979529927660076e-05, "loss": 1.5649, "step": 927 }, { "epoch": 0.03, "learning_rate": 1.9979462225265834e-05, "loss": 1.5895, "step": 928 }, { "epoch": 0.03, "learning_rate": 1.9979394411212528e-05, "loss": 1.556, "step": 929 }, { "epoch": 0.03, "learning_rate": 1.9979326485500916e-05, "loss": 1.6561, "step": 930 }, { "epoch": 0.03, "learning_rate": 1.997925844813176e-05, "loss": 1.5379, "step": 931 }, { "epoch": 0.03, "learning_rate": 1.997919029910582e-05, "loss": 1.6397, "step": 932 }, { "epoch": 0.03, "learning_rate": 1.9979122038423854e-05, "loss": 1.5542, "step": 933 }, { "epoch": 0.03, "learning_rate": 1.9979053666086633e-05, "loss": 1.6292, "step": 934 }, { "epoch": 0.03, "learning_rate": 1.9978985182094918e-05, "loss": 1.5976, "step": 935 }, { "epoch": 0.03, "learning_rate": 1.9978916586449475e-05, "loss": 1.628, "step": 936 }, { "epoch": 0.03, "learning_rate": 1.9978847879151076e-05, "loss": 1.6863, "step": 937 }, { "epoch": 0.03, "learning_rate": 1.9978779060200483e-05, "loss": 1.6606, "step": 938 }, { "epoch": 0.03, "learning_rate": 1.997871012959847e-05, "loss": 1.602, "step": 939 }, { "epoch": 0.03, "learning_rate": 1.9978641087345808e-05, "loss": 1.6317, "step": 940 }, { "epoch": 0.03, "learning_rate": 1.9978571933443267e-05, "loss": 1.5922, "step": 941 }, { "epoch": 0.03, "learning_rate": 1.9978502667891626e-05, "loss": 1.6755, "step": 942 }, { "epoch": 0.03, "learning_rate": 1.9978433290691654e-05, "loss": 1.5795, "step": 943 }, { "epoch": 0.03, "learning_rate": 1.9978363801844133e-05, "loss": 1.6511, "step": 944 }, { "epoch": 0.03, "learning_rate": 1.9978294201349837e-05, "loss": 1.5967, "step": 945 }, { "epoch": 0.03, "learning_rate": 1.9978224489209547e-05, "loss": 1.6726, "step": 946 }, { "epoch": 0.03, "learning_rate": 1.9978154665424037e-05, "loss": 1.6408, "step": 947 }, { "epoch": 0.03, "learning_rate": 1.9978084729994096e-05, "loss": 1.6338, "step": 948 }, { "epoch": 0.03, "learning_rate": 1.9978014682920503e-05, "loss": 1.7053, "step": 949 }, { "epoch": 0.03, "learning_rate": 1.9977944524204037e-05, "loss": 1.6286, "step": 950 }, { "epoch": 0.03, "learning_rate": 1.9977874253845494e-05, "loss": 1.6932, "step": 951 }, { "epoch": 0.03, "learning_rate": 1.997780387184565e-05, "loss": 1.6003, "step": 952 }, { "epoch": 0.03, "learning_rate": 1.99777333782053e-05, "loss": 1.6628, "step": 953 }, { "epoch": 0.03, "learning_rate": 1.9977662772925228e-05, "loss": 1.6765, "step": 954 }, { "epoch": 0.03, "learning_rate": 1.9977592056006226e-05, "loss": 1.694, "step": 955 }, { "epoch": 0.03, "learning_rate": 1.9977521227449082e-05, "loss": 1.6779, "step": 956 }, { "epoch": 0.03, "learning_rate": 1.9977450287254595e-05, "loss": 1.7611, "step": 957 }, { "epoch": 0.03, "learning_rate": 1.9977379235423553e-05, "loss": 1.636, "step": 958 }, { "epoch": 0.03, "learning_rate": 1.997730807195675e-05, "loss": 1.6428, "step": 959 }, { "epoch": 0.03, "learning_rate": 1.997723679685499e-05, "loss": 1.6824, "step": 960 }, { "epoch": 0.03, "learning_rate": 1.9977165410119065e-05, "loss": 1.5529, "step": 961 }, { "epoch": 0.03, "learning_rate": 1.997709391174977e-05, "loss": 1.6503, "step": 962 }, { "epoch": 0.03, "learning_rate": 1.9977022301747914e-05, "loss": 1.6293, "step": 963 }, { "epoch": 0.03, "learning_rate": 1.997695058011429e-05, "loss": 1.7172, "step": 964 }, { "epoch": 0.03, "learning_rate": 1.9976878746849707e-05, "loss": 1.6457, "step": 965 }, { "epoch": 0.03, "learning_rate": 1.9976806801954963e-05, "loss": 1.6563, "step": 966 }, { "epoch": 0.03, "learning_rate": 1.997673474543087e-05, "loss": 1.749, "step": 967 }, { "epoch": 0.03, "learning_rate": 1.9976662577278226e-05, "loss": 1.7926, "step": 968 }, { "epoch": 0.03, "learning_rate": 1.9976590297497843e-05, "loss": 1.755, "step": 969 }, { "epoch": 0.03, "learning_rate": 1.9976517906090528e-05, "loss": 1.824, "step": 970 }, { "epoch": 0.03, "learning_rate": 1.9976445403057095e-05, "loss": 1.6586, "step": 971 }, { "epoch": 0.03, "learning_rate": 1.997637278839835e-05, "loss": 1.7545, "step": 972 }, { "epoch": 0.03, "learning_rate": 1.9976300062115112e-05, "loss": 1.7444, "step": 973 }, { "epoch": 0.03, "learning_rate": 1.9976227224208185e-05, "loss": 1.843, "step": 974 }, { "epoch": 0.03, "learning_rate": 1.9976154274678393e-05, "loss": 1.7045, "step": 975 }, { "epoch": 0.03, "learning_rate": 1.9976081213526545e-05, "loss": 1.9429, "step": 976 }, { "epoch": 0.03, "learning_rate": 1.9976008040753467e-05, "loss": 1.6964, "step": 977 }, { "epoch": 0.03, "learning_rate": 1.997593475635997e-05, "loss": 1.7082, "step": 978 }, { "epoch": 0.03, "learning_rate": 1.9975861360346877e-05, "loss": 1.713, "step": 979 }, { "epoch": 0.03, "learning_rate": 1.997578785271501e-05, "loss": 1.6971, "step": 980 }, { "epoch": 0.03, "learning_rate": 1.997571423346519e-05, "loss": 1.7321, "step": 981 }, { "epoch": 0.03, "learning_rate": 1.9975640502598243e-05, "loss": 1.7821, "step": 982 }, { "epoch": 0.03, "learning_rate": 1.9975566660114994e-05, "loss": 1.6841, "step": 983 }, { "epoch": 0.03, "learning_rate": 1.9975492706016263e-05, "loss": 1.7261, "step": 984 }, { "epoch": 0.03, "learning_rate": 1.9975418640302885e-05, "loss": 1.7375, "step": 985 }, { "epoch": 0.03, "learning_rate": 1.9975344462975682e-05, "loss": 1.6332, "step": 986 }, { "epoch": 0.03, "learning_rate": 1.9975270174035492e-05, "loss": 1.7357, "step": 987 }, { "epoch": 0.03, "learning_rate": 1.997519577348314e-05, "loss": 1.7845, "step": 988 }, { "epoch": 0.03, "learning_rate": 1.997512126131946e-05, "loss": 1.7529, "step": 989 }, { "epoch": 0.03, "learning_rate": 1.9975046637545287e-05, "loss": 1.8041, "step": 990 }, { "epoch": 0.03, "learning_rate": 1.9974971902161455e-05, "loss": 1.7109, "step": 991 }, { "epoch": 0.03, "learning_rate": 1.99748970551688e-05, "loss": 1.7757, "step": 992 }, { "epoch": 0.03, "learning_rate": 1.9974822096568157e-05, "loss": 1.6379, "step": 993 }, { "epoch": 0.03, "learning_rate": 1.9974747026360372e-05, "loss": 1.6741, "step": 994 }, { "epoch": 0.03, "learning_rate": 1.997467184454628e-05, "loss": 1.6765, "step": 995 }, { "epoch": 0.03, "learning_rate": 1.997459655112672e-05, "loss": 1.6029, "step": 996 }, { "epoch": 0.03, "learning_rate": 1.9974521146102535e-05, "loss": 1.7118, "step": 997 }, { "epoch": 0.03, "learning_rate": 1.9974445629474574e-05, "loss": 1.6915, "step": 998 }, { "epoch": 0.03, "learning_rate": 1.997437000124368e-05, "loss": 1.7384, "step": 999 }, { "epoch": 0.03, "learning_rate": 1.9974294261410695e-05, "loss": 1.7579, "step": 1000 }, { "epoch": 0.03, "learning_rate": 1.9974218409976468e-05, "loss": 1.6609, "step": 1001 }, { "epoch": 0.03, "learning_rate": 1.9974142446941855e-05, "loss": 1.5951, "step": 1002 }, { "epoch": 0.03, "learning_rate": 1.9974066372307694e-05, "loss": 1.6973, "step": 1003 }, { "epoch": 0.03, "learning_rate": 1.9973990186074844e-05, "loss": 1.6702, "step": 1004 }, { "epoch": 0.03, "learning_rate": 1.9973913888244154e-05, "loss": 1.7319, "step": 1005 }, { "epoch": 0.03, "learning_rate": 1.9973837478816483e-05, "loss": 1.6064, "step": 1006 }, { "epoch": 0.03, "learning_rate": 1.997376095779268e-05, "loss": 1.642, "step": 1007 }, { "epoch": 0.03, "learning_rate": 1.99736843251736e-05, "loss": 1.6762, "step": 1008 }, { "epoch": 0.03, "learning_rate": 1.997360758096011e-05, "loss": 1.5806, "step": 1009 }, { "epoch": 0.03, "learning_rate": 1.9973530725153058e-05, "loss": 1.7004, "step": 1010 }, { "epoch": 0.03, "learning_rate": 1.997345375775331e-05, "loss": 1.6861, "step": 1011 }, { "epoch": 0.03, "learning_rate": 1.9973376678761726e-05, "loss": 1.6847, "step": 1012 }, { "epoch": 0.03, "learning_rate": 1.9973299488179167e-05, "loss": 1.6302, "step": 1013 }, { "epoch": 0.03, "learning_rate": 1.9973222186006498e-05, "loss": 1.6566, "step": 1014 }, { "epoch": 0.03, "learning_rate": 1.997314477224458e-05, "loss": 1.6262, "step": 1015 }, { "epoch": 0.03, "learning_rate": 1.9973067246894288e-05, "loss": 1.6341, "step": 1016 }, { "epoch": 0.03, "learning_rate": 1.9972989609956482e-05, "loss": 1.595, "step": 1017 }, { "epoch": 0.03, "learning_rate": 1.9972911861432033e-05, "loss": 1.5541, "step": 1018 }, { "epoch": 0.03, "learning_rate": 1.997283400132181e-05, "loss": 1.6367, "step": 1019 }, { "epoch": 0.03, "learning_rate": 1.997275602962669e-05, "loss": 1.6328, "step": 1020 }, { "epoch": 0.03, "learning_rate": 1.9972677946347536e-05, "loss": 1.6157, "step": 1021 }, { "epoch": 0.03, "learning_rate": 1.9972599751485225e-05, "loss": 1.6392, "step": 1022 }, { "epoch": 0.03, "learning_rate": 1.9972521445040634e-05, "loss": 1.5884, "step": 1023 }, { "epoch": 0.03, "learning_rate": 1.997244302701464e-05, "loss": 1.7174, "step": 1024 }, { "epoch": 0.03, "learning_rate": 1.9972364497408118e-05, "loss": 1.6873, "step": 1025 }, { "epoch": 0.03, "learning_rate": 1.9972285856221944e-05, "loss": 1.5334, "step": 1026 }, { "epoch": 0.03, "learning_rate": 1.9972207103457e-05, "loss": 1.6364, "step": 1027 }, { "epoch": 0.03, "learning_rate": 1.9972128239114173e-05, "loss": 1.491, "step": 1028 }, { "epoch": 0.03, "learning_rate": 1.997204926319434e-05, "loss": 1.6921, "step": 1029 }, { "epoch": 0.03, "learning_rate": 1.9971970175698387e-05, "loss": 1.6878, "step": 1030 }, { "epoch": 0.03, "learning_rate": 1.997189097662719e-05, "loss": 1.6131, "step": 1031 }, { "epoch": 0.03, "learning_rate": 1.997181166598165e-05, "loss": 1.5943, "step": 1032 }, { "epoch": 0.03, "learning_rate": 1.9971732243762643e-05, "loss": 1.5973, "step": 1033 }, { "epoch": 0.03, "learning_rate": 1.9971652709971062e-05, "loss": 1.6568, "step": 1034 }, { "epoch": 0.03, "learning_rate": 1.9971573064607797e-05, "loss": 1.6165, "step": 1035 }, { "epoch": 0.03, "learning_rate": 1.9971493307673735e-05, "loss": 1.6101, "step": 1036 }, { "epoch": 0.03, "learning_rate": 1.9971413439169777e-05, "loss": 1.562, "step": 1037 }, { "epoch": 0.03, "learning_rate": 1.9971333459096807e-05, "loss": 1.6557, "step": 1038 }, { "epoch": 0.03, "learning_rate": 1.9971253367455728e-05, "loss": 1.7349, "step": 1039 }, { "epoch": 0.03, "learning_rate": 1.997117316424743e-05, "loss": 1.6816, "step": 1040 }, { "epoch": 0.03, "learning_rate": 1.9971092849472813e-05, "loss": 1.6056, "step": 1041 }, { "epoch": 0.03, "learning_rate": 1.9971012423132776e-05, "loss": 1.7044, "step": 1042 }, { "epoch": 0.03, "learning_rate": 1.997093188522822e-05, "loss": 1.6103, "step": 1043 }, { "epoch": 0.03, "learning_rate": 1.9970851235760042e-05, "loss": 1.6705, "step": 1044 }, { "epoch": 0.03, "learning_rate": 1.9970770474729146e-05, "loss": 1.723, "step": 1045 }, { "epoch": 0.03, "learning_rate": 1.997068960213644e-05, "loss": 1.6675, "step": 1046 }, { "epoch": 0.03, "learning_rate": 1.9970608617982824e-05, "loss": 1.6938, "step": 1047 }, { "epoch": 0.03, "learning_rate": 1.9970527522269204e-05, "loss": 1.6141, "step": 1048 }, { "epoch": 0.03, "learning_rate": 1.997044631499649e-05, "loss": 1.6162, "step": 1049 }, { "epoch": 0.04, "learning_rate": 1.997036499616559e-05, "loss": 1.6147, "step": 1050 }, { "epoch": 0.04, "learning_rate": 1.997028356577741e-05, "loss": 1.7823, "step": 1051 }, { "epoch": 0.04, "learning_rate": 1.997020202383287e-05, "loss": 1.6498, "step": 1052 }, { "epoch": 0.04, "learning_rate": 1.9970120370332876e-05, "loss": 1.645, "step": 1053 }, { "epoch": 0.04, "learning_rate": 1.997003860527834e-05, "loss": 1.6857, "step": 1054 }, { "epoch": 0.04, "learning_rate": 1.996995672867018e-05, "loss": 1.6193, "step": 1055 }, { "epoch": 0.04, "learning_rate": 1.996987474050931e-05, "loss": 1.6017, "step": 1056 }, { "epoch": 0.04, "learning_rate": 1.996979264079665e-05, "loss": 1.6858, "step": 1057 }, { "epoch": 0.04, "learning_rate": 1.9969710429533115e-05, "loss": 1.7157, "step": 1058 }, { "epoch": 0.04, "learning_rate": 1.9969628106719632e-05, "loss": 1.6304, "step": 1059 }, { "epoch": 0.04, "learning_rate": 1.9969545672357117e-05, "loss": 1.7122, "step": 1060 }, { "epoch": 0.04, "learning_rate": 1.996946312644649e-05, "loss": 1.7069, "step": 1061 }, { "epoch": 0.04, "learning_rate": 1.9969380468988676e-05, "loss": 1.6491, "step": 1062 }, { "epoch": 0.04, "learning_rate": 1.9969297699984606e-05, "loss": 1.6682, "step": 1063 }, { "epoch": 0.04, "learning_rate": 1.9969214819435198e-05, "loss": 1.6906, "step": 1064 }, { "epoch": 0.04, "learning_rate": 1.9969131827341383e-05, "loss": 1.7027, "step": 1065 }, { "epoch": 0.04, "learning_rate": 1.996904872370409e-05, "loss": 1.7267, "step": 1066 }, { "epoch": 0.04, "learning_rate": 1.9968965508524242e-05, "loss": 1.7498, "step": 1067 }, { "epoch": 0.04, "learning_rate": 1.9968882181802783e-05, "loss": 1.7185, "step": 1068 }, { "epoch": 0.04, "learning_rate": 1.9968798743540638e-05, "loss": 1.6427, "step": 1069 }, { "epoch": 0.04, "learning_rate": 1.9968715193738738e-05, "loss": 1.7569, "step": 1070 }, { "epoch": 0.04, "learning_rate": 1.996863153239802e-05, "loss": 1.7847, "step": 1071 }, { "epoch": 0.04, "learning_rate": 1.9968547759519426e-05, "loss": 1.7303, "step": 1072 }, { "epoch": 0.04, "learning_rate": 1.9968463875103886e-05, "loss": 1.6199, "step": 1073 }, { "epoch": 0.04, "learning_rate": 1.996837987915234e-05, "loss": 1.7297, "step": 1074 }, { "epoch": 0.04, "learning_rate": 1.9968295771665727e-05, "loss": 1.6584, "step": 1075 }, { "epoch": 0.04, "learning_rate": 1.9968211552644993e-05, "loss": 1.7117, "step": 1076 }, { "epoch": 0.04, "learning_rate": 1.9968127222091073e-05, "loss": 1.6944, "step": 1077 }, { "epoch": 0.04, "learning_rate": 1.9968042780004917e-05, "loss": 1.6571, "step": 1078 }, { "epoch": 0.04, "learning_rate": 1.9967958226387466e-05, "loss": 1.5196, "step": 1079 }, { "epoch": 0.04, "learning_rate": 1.9967873561239672e-05, "loss": 1.6713, "step": 1080 }, { "epoch": 0.04, "learning_rate": 1.9967788784562474e-05, "loss": 1.6329, "step": 1081 }, { "epoch": 0.04, "learning_rate": 1.996770389635682e-05, "loss": 1.6804, "step": 1082 }, { "epoch": 0.04, "learning_rate": 1.996761889662367e-05, "loss": 1.5993, "step": 1083 }, { "epoch": 0.04, "learning_rate": 1.996753378536397e-05, "loss": 1.6395, "step": 1084 }, { "epoch": 0.04, "learning_rate": 1.996744856257867e-05, "loss": 1.5706, "step": 1085 }, { "epoch": 0.04, "learning_rate": 1.9967363228268725e-05, "loss": 1.7494, "step": 1086 }, { "epoch": 0.04, "learning_rate": 1.996727778243509e-05, "loss": 1.6058, "step": 1087 }, { "epoch": 0.04, "learning_rate": 1.996719222507872e-05, "loss": 1.6054, "step": 1088 }, { "epoch": 0.04, "learning_rate": 1.9967106556200573e-05, "loss": 1.6273, "step": 1089 }, { "epoch": 0.04, "learning_rate": 1.9967020775801605e-05, "loss": 1.608, "step": 1090 }, { "epoch": 0.04, "learning_rate": 1.9966934883882783e-05, "loss": 1.6215, "step": 1091 }, { "epoch": 0.04, "learning_rate": 1.996684888044506e-05, "loss": 1.6245, "step": 1092 }, { "epoch": 0.04, "learning_rate": 1.9966762765489407e-05, "loss": 1.5929, "step": 1093 }, { "epoch": 0.04, "learning_rate": 1.996667653901678e-05, "loss": 1.7475, "step": 1094 }, { "epoch": 0.04, "learning_rate": 1.9966590201028144e-05, "loss": 1.6368, "step": 1095 }, { "epoch": 0.04, "learning_rate": 1.9966503751524467e-05, "loss": 1.6528, "step": 1096 }, { "epoch": 0.04, "learning_rate": 1.9966417190506718e-05, "loss": 1.5756, "step": 1097 }, { "epoch": 0.04, "learning_rate": 1.996633051797586e-05, "loss": 1.6495, "step": 1098 }, { "epoch": 0.04, "learning_rate": 1.9966243733932873e-05, "loss": 1.5452, "step": 1099 }, { "epoch": 0.04, "learning_rate": 1.996615683837872e-05, "loss": 1.6365, "step": 1100 }, { "epoch": 0.04, "learning_rate": 1.996606983131437e-05, "loss": 1.6203, "step": 1101 }, { "epoch": 0.04, "learning_rate": 1.996598271274081e-05, "loss": 1.6138, "step": 1102 }, { "epoch": 0.04, "learning_rate": 1.9965895482659e-05, "loss": 1.6175, "step": 1103 }, { "epoch": 0.04, "learning_rate": 1.9965808141069922e-05, "loss": 1.581, "step": 1104 }, { "epoch": 0.04, "learning_rate": 1.9965720687974555e-05, "loss": 1.7168, "step": 1105 }, { "epoch": 0.04, "learning_rate": 1.9965633123373877e-05, "loss": 1.683, "step": 1106 }, { "epoch": 0.04, "learning_rate": 1.9965545447268867e-05, "loss": 1.6853, "step": 1107 }, { "epoch": 0.04, "learning_rate": 1.9965457659660504e-05, "loss": 1.6572, "step": 1108 }, { "epoch": 0.04, "learning_rate": 1.9965369760549774e-05, "loss": 1.5633, "step": 1109 }, { "epoch": 0.04, "learning_rate": 1.9965281749937653e-05, "loss": 1.6127, "step": 1110 }, { "epoch": 0.04, "learning_rate": 1.9965193627825138e-05, "loss": 1.6609, "step": 1111 }, { "epoch": 0.04, "learning_rate": 1.9965105394213203e-05, "loss": 1.5762, "step": 1112 }, { "epoch": 0.04, "learning_rate": 1.9965017049102845e-05, "loss": 1.621, "step": 1113 }, { "epoch": 0.04, "learning_rate": 1.9964928592495046e-05, "loss": 1.5878, "step": 1114 }, { "epoch": 0.04, "learning_rate": 1.9964840024390797e-05, "loss": 1.6146, "step": 1115 }, { "epoch": 0.04, "learning_rate": 1.9964751344791092e-05, "loss": 1.6824, "step": 1116 }, { "epoch": 0.04, "learning_rate": 1.9964662553696915e-05, "loss": 1.5614, "step": 1117 }, { "epoch": 0.04, "learning_rate": 1.996457365110927e-05, "loss": 1.6623, "step": 1118 }, { "epoch": 0.04, "learning_rate": 1.9964484637029147e-05, "loss": 1.6372, "step": 1119 }, { "epoch": 0.04, "learning_rate": 1.9964395511457543e-05, "loss": 1.5624, "step": 1120 }, { "epoch": 0.04, "learning_rate": 1.996430627439545e-05, "loss": 1.5364, "step": 1121 }, { "epoch": 0.04, "learning_rate": 1.9964216925843876e-05, "loss": 1.6404, "step": 1122 }, { "epoch": 0.04, "learning_rate": 1.9964127465803812e-05, "loss": 1.5201, "step": 1123 }, { "epoch": 0.04, "learning_rate": 1.9964037894276263e-05, "loss": 1.6285, "step": 1124 }, { "epoch": 0.04, "learning_rate": 1.9963948211262233e-05, "loss": 1.6399, "step": 1125 }, { "epoch": 0.04, "learning_rate": 1.996385841676272e-05, "loss": 1.6279, "step": 1126 }, { "epoch": 0.04, "learning_rate": 1.996376851077873e-05, "loss": 1.5886, "step": 1127 }, { "epoch": 0.04, "learning_rate": 1.9963678493311273e-05, "loss": 1.6545, "step": 1128 }, { "epoch": 0.04, "learning_rate": 1.9963588364361356e-05, "loss": 1.5104, "step": 1129 }, { "epoch": 0.04, "learning_rate": 1.9963498123929984e-05, "loss": 1.6277, "step": 1130 }, { "epoch": 0.04, "learning_rate": 1.9963407772018163e-05, "loss": 1.6694, "step": 1131 }, { "epoch": 0.04, "learning_rate": 1.9963317308626916e-05, "loss": 1.6967, "step": 1132 }, { "epoch": 0.04, "learning_rate": 1.9963226733757242e-05, "loss": 1.6006, "step": 1133 }, { "epoch": 0.04, "learning_rate": 1.996313604741017e-05, "loss": 1.6555, "step": 1134 }, { "epoch": 0.04, "learning_rate": 1.9963045249586696e-05, "loss": 1.5769, "step": 1135 }, { "epoch": 0.04, "learning_rate": 1.996295434028785e-05, "loss": 1.6999, "step": 1136 }, { "epoch": 0.04, "learning_rate": 1.996286331951464e-05, "loss": 1.6305, "step": 1137 }, { "epoch": 0.04, "learning_rate": 1.9962772187268093e-05, "loss": 1.635, "step": 1138 }, { "epoch": 0.04, "learning_rate": 1.9962680943549224e-05, "loss": 1.5488, "step": 1139 }, { "epoch": 0.04, "learning_rate": 1.996258958835905e-05, "loss": 1.7434, "step": 1140 }, { "epoch": 0.04, "learning_rate": 1.9962498121698602e-05, "loss": 1.7038, "step": 1141 }, { "epoch": 0.04, "learning_rate": 1.99624065435689e-05, "loss": 1.7913, "step": 1142 }, { "epoch": 0.04, "learning_rate": 1.9962314853970962e-05, "loss": 1.8211, "step": 1143 }, { "epoch": 0.04, "learning_rate": 1.9962223052905823e-05, "loss": 1.6677, "step": 1144 }, { "epoch": 0.04, "learning_rate": 1.9962131140374506e-05, "loss": 1.8, "step": 1145 }, { "epoch": 0.04, "learning_rate": 1.996203911637804e-05, "loss": 1.7083, "step": 1146 }, { "epoch": 0.04, "learning_rate": 1.9961946980917457e-05, "loss": 1.6554, "step": 1147 }, { "epoch": 0.04, "learning_rate": 1.9961854733993785e-05, "loss": 1.7406, "step": 1148 }, { "epoch": 0.04, "learning_rate": 1.9961762375608055e-05, "loss": 1.6063, "step": 1149 }, { "epoch": 0.04, "learning_rate": 1.9961669905761303e-05, "loss": 1.6888, "step": 1150 }, { "epoch": 0.04, "learning_rate": 1.9961577324454564e-05, "loss": 1.8049, "step": 1151 }, { "epoch": 0.04, "learning_rate": 1.996148463168887e-05, "loss": 1.7518, "step": 1152 }, { "epoch": 0.04, "learning_rate": 1.996139182746526e-05, "loss": 1.8164, "step": 1153 }, { "epoch": 0.04, "learning_rate": 1.9961298911784773e-05, "loss": 1.7415, "step": 1154 }, { "epoch": 0.04, "learning_rate": 1.9961205884648452e-05, "loss": 1.7367, "step": 1155 }, { "epoch": 0.04, "learning_rate": 1.996111274605733e-05, "loss": 1.7261, "step": 1156 }, { "epoch": 0.04, "learning_rate": 1.9961019496012458e-05, "loss": 1.6759, "step": 1157 }, { "epoch": 0.04, "learning_rate": 1.9960926134514875e-05, "loss": 1.7504, "step": 1158 }, { "epoch": 0.04, "learning_rate": 1.9960832661565625e-05, "loss": 1.6158, "step": 1159 }, { "epoch": 0.04, "learning_rate": 1.9960739077165755e-05, "loss": 1.6092, "step": 1160 }, { "epoch": 0.04, "learning_rate": 1.996064538131631e-05, "loss": 1.6698, "step": 1161 }, { "epoch": 0.04, "learning_rate": 1.996055157401834e-05, "loss": 1.7019, "step": 1162 }, { "epoch": 0.04, "learning_rate": 1.9960457655272893e-05, "loss": 1.6293, "step": 1163 }, { "epoch": 0.04, "learning_rate": 1.9960363625081026e-05, "loss": 1.6684, "step": 1164 }, { "epoch": 0.04, "learning_rate": 1.9960269483443785e-05, "loss": 1.6083, "step": 1165 }, { "epoch": 0.04, "learning_rate": 1.9960175230362222e-05, "loss": 1.707, "step": 1166 }, { "epoch": 0.04, "learning_rate": 1.9960080865837398e-05, "loss": 1.7343, "step": 1167 }, { "epoch": 0.04, "learning_rate": 1.9959986389870364e-05, "loss": 1.7035, "step": 1168 }, { "epoch": 0.04, "learning_rate": 1.995989180246218e-05, "loss": 1.6434, "step": 1169 }, { "epoch": 0.04, "learning_rate": 1.99597971036139e-05, "loss": 1.7325, "step": 1170 }, { "epoch": 0.04, "learning_rate": 1.9959702293326585e-05, "loss": 1.6456, "step": 1171 }, { "epoch": 0.04, "learning_rate": 1.9959607371601303e-05, "loss": 1.6235, "step": 1172 }, { "epoch": 0.04, "learning_rate": 1.9959512338439106e-05, "loss": 1.6831, "step": 1173 }, { "epoch": 0.04, "learning_rate": 1.995941719384106e-05, "loss": 1.71, "step": 1174 }, { "epoch": 0.04, "learning_rate": 1.9959321937808236e-05, "loss": 1.7065, "step": 1175 }, { "epoch": 0.04, "learning_rate": 1.995922657034169e-05, "loss": 1.612, "step": 1176 }, { "epoch": 0.04, "learning_rate": 1.9959131091442497e-05, "loss": 1.5624, "step": 1177 }, { "epoch": 0.04, "learning_rate": 1.995903550111172e-05, "loss": 1.5983, "step": 1178 }, { "epoch": 0.04, "learning_rate": 1.9958939799350433e-05, "loss": 1.6131, "step": 1179 }, { "epoch": 0.04, "learning_rate": 1.9958843986159705e-05, "loss": 1.6805, "step": 1180 }, { "epoch": 0.04, "learning_rate": 1.9958748061540607e-05, "loss": 1.7235, "step": 1181 }, { "epoch": 0.04, "learning_rate": 1.9958652025494212e-05, "loss": 1.6126, "step": 1182 }, { "epoch": 0.04, "learning_rate": 1.9958555878021596e-05, "loss": 1.5745, "step": 1183 }, { "epoch": 0.04, "learning_rate": 1.9958459619123832e-05, "loss": 1.5886, "step": 1184 }, { "epoch": 0.04, "learning_rate": 1.9958363248802004e-05, "loss": 1.611, "step": 1185 }, { "epoch": 0.04, "learning_rate": 1.9958266767057183e-05, "loss": 1.6842, "step": 1186 }, { "epoch": 0.04, "learning_rate": 1.995817017389045e-05, "loss": 1.6989, "step": 1187 }, { "epoch": 0.04, "learning_rate": 1.9958073469302888e-05, "loss": 1.7419, "step": 1188 }, { "epoch": 0.04, "learning_rate": 1.9957976653295576e-05, "loss": 1.5281, "step": 1189 }, { "epoch": 0.04, "learning_rate": 1.9957879725869604e-05, "loss": 1.5022, "step": 1190 }, { "epoch": 0.04, "learning_rate": 1.9957782687026046e-05, "loss": 1.5748, "step": 1191 }, { "epoch": 0.04, "learning_rate": 1.9957685536765998e-05, "loss": 1.5864, "step": 1192 }, { "epoch": 0.04, "learning_rate": 1.995758827509054e-05, "loss": 1.5447, "step": 1193 }, { "epoch": 0.04, "learning_rate": 1.995749090200076e-05, "loss": 1.5961, "step": 1194 }, { "epoch": 0.04, "learning_rate": 1.9957393417497753e-05, "loss": 1.558, "step": 1195 }, { "epoch": 0.04, "learning_rate": 1.9957295821582608e-05, "loss": 1.5548, "step": 1196 }, { "epoch": 0.04, "learning_rate": 1.9957198114256414e-05, "loss": 1.6064, "step": 1197 }, { "epoch": 0.04, "learning_rate": 1.9957100295520267e-05, "loss": 1.624, "step": 1198 }, { "epoch": 0.04, "learning_rate": 1.995700236537526e-05, "loss": 1.6727, "step": 1199 }, { "epoch": 0.04, "learning_rate": 1.995690432382249e-05, "loss": 1.6714, "step": 1200 }, { "epoch": 0.04, "learning_rate": 1.995680617086305e-05, "loss": 1.6606, "step": 1201 }, { "epoch": 0.04, "learning_rate": 1.9956707906498046e-05, "loss": 1.6599, "step": 1202 }, { "epoch": 0.04, "learning_rate": 1.9956609530728574e-05, "loss": 1.6398, "step": 1203 }, { "epoch": 0.04, "learning_rate": 1.995651104355573e-05, "loss": 1.5876, "step": 1204 }, { "epoch": 0.04, "learning_rate": 1.995641244498062e-05, "loss": 1.7105, "step": 1205 }, { "epoch": 0.04, "learning_rate": 1.9956313735004345e-05, "loss": 1.6303, "step": 1206 }, { "epoch": 0.04, "learning_rate": 1.9956214913628015e-05, "loss": 1.5312, "step": 1207 }, { "epoch": 0.04, "learning_rate": 1.995611598085273e-05, "loss": 1.6545, "step": 1208 }, { "epoch": 0.04, "learning_rate": 1.99560169366796e-05, "loss": 1.5524, "step": 1209 }, { "epoch": 0.04, "learning_rate": 1.995591778110973e-05, "loss": 1.5889, "step": 1210 }, { "epoch": 0.04, "learning_rate": 1.9955818514144237e-05, "loss": 1.5672, "step": 1211 }, { "epoch": 0.04, "learning_rate": 1.995571913578422e-05, "loss": 1.5557, "step": 1212 }, { "epoch": 0.04, "learning_rate": 1.99556196460308e-05, "loss": 1.6762, "step": 1213 }, { "epoch": 0.04, "learning_rate": 1.995552004488509e-05, "loss": 1.59, "step": 1214 }, { "epoch": 0.04, "learning_rate": 1.99554203323482e-05, "loss": 1.6603, "step": 1215 }, { "epoch": 0.04, "learning_rate": 1.9955320508421247e-05, "loss": 1.6528, "step": 1216 }, { "epoch": 0.04, "learning_rate": 1.9955220573105345e-05, "loss": 1.603, "step": 1217 }, { "epoch": 0.04, "learning_rate": 1.9955120526401618e-05, "loss": 1.6216, "step": 1218 }, { "epoch": 0.04, "learning_rate": 1.9955020368311185e-05, "loss": 1.5285, "step": 1219 }, { "epoch": 0.04, "learning_rate": 1.9954920098835164e-05, "loss": 1.5793, "step": 1220 }, { "epoch": 0.04, "learning_rate": 1.9954819717974677e-05, "loss": 1.6158, "step": 1221 }, { "epoch": 0.04, "learning_rate": 1.9954719225730847e-05, "loss": 1.5877, "step": 1222 }, { "epoch": 0.04, "learning_rate": 1.99546186221048e-05, "loss": 1.4566, "step": 1223 }, { "epoch": 0.04, "learning_rate": 1.9954517907097663e-05, "loss": 1.653, "step": 1224 }, { "epoch": 0.04, "learning_rate": 1.9954417080710557e-05, "loss": 1.557, "step": 1225 }, { "epoch": 0.04, "learning_rate": 1.9954316142944613e-05, "loss": 1.6278, "step": 1226 }, { "epoch": 0.04, "learning_rate": 1.9954215093800964e-05, "loss": 1.6598, "step": 1227 }, { "epoch": 0.04, "learning_rate": 1.9954113933280737e-05, "loss": 1.5343, "step": 1228 }, { "epoch": 0.04, "learning_rate": 1.9954012661385062e-05, "loss": 1.5572, "step": 1229 }, { "epoch": 0.04, "learning_rate": 1.9953911278115078e-05, "loss": 1.6109, "step": 1230 }, { "epoch": 0.04, "learning_rate": 1.9953809783471917e-05, "loss": 1.6002, "step": 1231 }, { "epoch": 0.04, "learning_rate": 1.9953708177456713e-05, "loss": 1.6471, "step": 1232 }, { "epoch": 0.04, "learning_rate": 1.9953606460070602e-05, "loss": 1.6386, "step": 1233 }, { "epoch": 0.04, "learning_rate": 1.9953504631314722e-05, "loss": 1.6224, "step": 1234 }, { "epoch": 0.04, "learning_rate": 1.9953402691190218e-05, "loss": 1.5986, "step": 1235 }, { "epoch": 0.04, "learning_rate": 1.9953300639698226e-05, "loss": 1.6137, "step": 1236 }, { "epoch": 0.04, "learning_rate": 1.9953198476839886e-05, "loss": 1.6884, "step": 1237 }, { "epoch": 0.04, "learning_rate": 1.9953096202616345e-05, "loss": 1.671, "step": 1238 }, { "epoch": 0.04, "learning_rate": 1.9952993817028744e-05, "loss": 1.6154, "step": 1239 }, { "epoch": 0.04, "learning_rate": 1.9952891320078235e-05, "loss": 1.6024, "step": 1240 }, { "epoch": 0.04, "learning_rate": 1.9952788711765957e-05, "loss": 1.6306, "step": 1241 }, { "epoch": 0.04, "learning_rate": 1.9952685992093064e-05, "loss": 1.6228, "step": 1242 }, { "epoch": 0.04, "learning_rate": 1.9952583161060702e-05, "loss": 1.5974, "step": 1243 }, { "epoch": 0.04, "learning_rate": 1.995248021867002e-05, "loss": 1.5174, "step": 1244 }, { "epoch": 0.04, "learning_rate": 1.9952377164922173e-05, "loss": 1.6512, "step": 1245 }, { "epoch": 0.04, "learning_rate": 1.9952273999818312e-05, "loss": 1.5648, "step": 1246 }, { "epoch": 0.04, "learning_rate": 1.9952170723359594e-05, "loss": 1.6121, "step": 1247 }, { "epoch": 0.04, "learning_rate": 1.9952067335547174e-05, "loss": 1.6075, "step": 1248 }, { "epoch": 0.04, "learning_rate": 1.9951963836382206e-05, "loss": 1.5739, "step": 1249 }, { "epoch": 0.04, "learning_rate": 1.995186022586585e-05, "loss": 1.6385, "step": 1250 }, { "epoch": 0.04, "learning_rate": 1.9951756503999268e-05, "loss": 1.5274, "step": 1251 }, { "epoch": 0.04, "learning_rate": 1.9951652670783615e-05, "loss": 1.6857, "step": 1252 }, { "epoch": 0.04, "learning_rate": 1.9951548726220057e-05, "loss": 1.6547, "step": 1253 }, { "epoch": 0.04, "learning_rate": 1.9951444670309752e-05, "loss": 1.5757, "step": 1254 }, { "epoch": 0.04, "learning_rate": 1.995134050305387e-05, "loss": 1.6347, "step": 1255 }, { "epoch": 0.04, "learning_rate": 1.9951236224453575e-05, "loss": 1.5946, "step": 1256 }, { "epoch": 0.04, "learning_rate": 1.9951131834510034e-05, "loss": 1.5859, "step": 1257 }, { "epoch": 0.04, "learning_rate": 1.995102733322441e-05, "loss": 1.5915, "step": 1258 }, { "epoch": 0.04, "learning_rate": 1.995092272059788e-05, "loss": 1.6645, "step": 1259 }, { "epoch": 0.04, "learning_rate": 1.995081799663161e-05, "loss": 1.5909, "step": 1260 }, { "epoch": 0.04, "learning_rate": 1.995071316132677e-05, "loss": 1.6137, "step": 1261 }, { "epoch": 0.04, "learning_rate": 1.9950608214684536e-05, "loss": 1.5934, "step": 1262 }, { "epoch": 0.04, "learning_rate": 1.9950503156706086e-05, "loss": 1.6403, "step": 1263 }, { "epoch": 0.04, "learning_rate": 1.9950397987392588e-05, "loss": 1.6143, "step": 1264 }, { "epoch": 0.04, "learning_rate": 1.995029270674522e-05, "loss": 1.5759, "step": 1265 }, { "epoch": 0.04, "learning_rate": 1.9950187314765163e-05, "loss": 1.6703, "step": 1266 }, { "epoch": 0.04, "learning_rate": 1.9950081811453598e-05, "loss": 1.5353, "step": 1267 }, { "epoch": 0.04, "learning_rate": 1.99499761968117e-05, "loss": 1.5204, "step": 1268 }, { "epoch": 0.04, "learning_rate": 1.9949870470840653e-05, "loss": 1.5727, "step": 1269 }, { "epoch": 0.04, "learning_rate": 1.9949764633541645e-05, "loss": 1.543, "step": 1270 }, { "epoch": 0.04, "learning_rate": 1.994965868491585e-05, "loss": 1.4907, "step": 1271 }, { "epoch": 0.04, "learning_rate": 1.994955262496446e-05, "loss": 1.6271, "step": 1272 }, { "epoch": 0.04, "learning_rate": 1.994944645368866e-05, "loss": 1.7154, "step": 1273 }, { "epoch": 0.04, "learning_rate": 1.9949340171089642e-05, "loss": 1.6924, "step": 1274 }, { "epoch": 0.04, "learning_rate": 1.9949233777168588e-05, "loss": 1.7103, "step": 1275 }, { "epoch": 0.04, "learning_rate": 1.9949127271926696e-05, "loss": 1.6344, "step": 1276 }, { "epoch": 0.04, "learning_rate": 1.994902065536515e-05, "loss": 1.5972, "step": 1277 }, { "epoch": 0.04, "learning_rate": 1.9948913927485147e-05, "loss": 1.6229, "step": 1278 }, { "epoch": 0.04, "learning_rate": 1.9948807088287884e-05, "loss": 1.6236, "step": 1279 }, { "epoch": 0.04, "learning_rate": 1.994870013777455e-05, "loss": 1.65, "step": 1280 }, { "epoch": 0.04, "learning_rate": 1.9948593075946348e-05, "loss": 1.6507, "step": 1281 }, { "epoch": 0.04, "learning_rate": 1.9948485902804472e-05, "loss": 1.5627, "step": 1282 }, { "epoch": 0.04, "learning_rate": 1.9948378618350118e-05, "loss": 1.5484, "step": 1283 }, { "epoch": 0.04, "learning_rate": 1.9948271222584498e-05, "loss": 1.592, "step": 1284 }, { "epoch": 0.04, "learning_rate": 1.99481637155088e-05, "loss": 1.6963, "step": 1285 }, { "epoch": 0.04, "learning_rate": 1.9948056097124236e-05, "loss": 1.7066, "step": 1286 }, { "epoch": 0.04, "learning_rate": 1.9947948367432004e-05, "loss": 1.5904, "step": 1287 }, { "epoch": 0.04, "learning_rate": 1.9947840526433316e-05, "loss": 1.656, "step": 1288 }, { "epoch": 0.04, "learning_rate": 1.9947732574129374e-05, "loss": 1.5153, "step": 1289 }, { "epoch": 0.04, "learning_rate": 1.9947624510521385e-05, "loss": 1.6165, "step": 1290 }, { "epoch": 0.04, "learning_rate": 1.9947516335610563e-05, "loss": 1.615, "step": 1291 }, { "epoch": 0.04, "learning_rate": 1.994740804939811e-05, "loss": 1.617, "step": 1292 }, { "epoch": 0.04, "learning_rate": 1.994729965188525e-05, "loss": 1.5532, "step": 1293 }, { "epoch": 0.04, "learning_rate": 1.9947191143073185e-05, "loss": 1.6387, "step": 1294 }, { "epoch": 0.04, "learning_rate": 1.9947082522963134e-05, "loss": 1.6722, "step": 1295 }, { "epoch": 0.04, "learning_rate": 1.994697379155631e-05, "loss": 1.496, "step": 1296 }, { "epoch": 0.04, "learning_rate": 1.9946864948853936e-05, "loss": 1.5604, "step": 1297 }, { "epoch": 0.04, "learning_rate": 1.994675599485722e-05, "loss": 1.6629, "step": 1298 }, { "epoch": 0.04, "learning_rate": 1.9946646929567386e-05, "loss": 1.6254, "step": 1299 }, { "epoch": 0.04, "learning_rate": 1.9946537752985653e-05, "loss": 1.6184, "step": 1300 }, { "epoch": 0.04, "learning_rate": 1.9946428465113244e-05, "loss": 1.672, "step": 1301 }, { "epoch": 0.04, "learning_rate": 1.9946319065951382e-05, "loss": 1.689, "step": 1302 }, { "epoch": 0.04, "learning_rate": 1.9946209555501293e-05, "loss": 1.5271, "step": 1303 }, { "epoch": 0.04, "learning_rate": 1.9946099933764197e-05, "loss": 1.599, "step": 1304 }, { "epoch": 0.04, "learning_rate": 1.9945990200741324e-05, "loss": 1.5929, "step": 1305 }, { "epoch": 0.04, "learning_rate": 1.9945880356433904e-05, "loss": 1.7457, "step": 1306 }, { "epoch": 0.04, "learning_rate": 1.994577040084316e-05, "loss": 1.684, "step": 1307 }, { "epoch": 0.04, "learning_rate": 1.9945660333970327e-05, "loss": 1.6994, "step": 1308 }, { "epoch": 0.04, "learning_rate": 1.9945550155816634e-05, "loss": 1.7208, "step": 1309 }, { "epoch": 0.04, "learning_rate": 1.9945439866383314e-05, "loss": 1.6097, "step": 1310 }, { "epoch": 0.04, "learning_rate": 1.9945329465671602e-05, "loss": 1.6127, "step": 1311 }, { "epoch": 0.04, "learning_rate": 1.9945218953682736e-05, "loss": 1.6854, "step": 1312 }, { "epoch": 0.04, "learning_rate": 1.9945108330417946e-05, "loss": 1.6602, "step": 1313 }, { "epoch": 0.04, "learning_rate": 1.9944997595878474e-05, "loss": 1.6267, "step": 1314 }, { "epoch": 0.04, "learning_rate": 1.994488675006556e-05, "loss": 1.616, "step": 1315 }, { "epoch": 0.04, "learning_rate": 1.9944775792980442e-05, "loss": 1.5941, "step": 1316 }, { "epoch": 0.04, "learning_rate": 1.994466472462436e-05, "loss": 1.6722, "step": 1317 }, { "epoch": 0.04, "learning_rate": 1.9944553544998563e-05, "loss": 1.6764, "step": 1318 }, { "epoch": 0.04, "learning_rate": 1.9944442254104287e-05, "loss": 1.7032, "step": 1319 }, { "epoch": 0.04, "learning_rate": 1.994433085194278e-05, "loss": 1.7122, "step": 1320 }, { "epoch": 0.04, "learning_rate": 1.994421933851529e-05, "loss": 1.7186, "step": 1321 }, { "epoch": 0.04, "learning_rate": 1.9944107713823068e-05, "loss": 1.6648, "step": 1322 }, { "epoch": 0.04, "learning_rate": 1.9943995977867358e-05, "loss": 1.7077, "step": 1323 }, { "epoch": 0.04, "learning_rate": 1.9943884130649408e-05, "loss": 1.5965, "step": 1324 }, { "epoch": 0.04, "learning_rate": 1.9943772172170475e-05, "loss": 1.6794, "step": 1325 }, { "epoch": 0.04, "learning_rate": 1.994366010243181e-05, "loss": 1.7242, "step": 1326 }, { "epoch": 0.04, "learning_rate": 1.9943547921434666e-05, "loss": 1.8306, "step": 1327 }, { "epoch": 0.04, "learning_rate": 1.99434356291803e-05, "loss": 1.6413, "step": 1328 }, { "epoch": 0.04, "learning_rate": 1.9943323225669962e-05, "loss": 1.8242, "step": 1329 }, { "epoch": 0.04, "learning_rate": 1.994321071090492e-05, "loss": 1.7054, "step": 1330 }, { "epoch": 0.04, "learning_rate": 1.9943098084886423e-05, "loss": 1.7294, "step": 1331 }, { "epoch": 0.04, "learning_rate": 1.994298534761574e-05, "loss": 1.7649, "step": 1332 }, { "epoch": 0.04, "learning_rate": 1.9942872499094125e-05, "loss": 1.7519, "step": 1333 }, { "epoch": 0.04, "learning_rate": 1.9942759539322845e-05, "loss": 1.7147, "step": 1334 }, { "epoch": 0.04, "learning_rate": 1.9942646468303165e-05, "loss": 1.6697, "step": 1335 }, { "epoch": 0.04, "learning_rate": 1.9942533286036343e-05, "loss": 1.8079, "step": 1336 }, { "epoch": 0.04, "learning_rate": 1.9942419992523652e-05, "loss": 1.7039, "step": 1337 }, { "epoch": 0.04, "learning_rate": 1.9942306587766357e-05, "loss": 1.7065, "step": 1338 }, { "epoch": 0.04, "learning_rate": 1.9942193071765728e-05, "loss": 1.7696, "step": 1339 }, { "epoch": 0.04, "learning_rate": 1.994207944452304e-05, "loss": 1.7799, "step": 1340 }, { "epoch": 0.04, "learning_rate": 1.9941965706039554e-05, "loss": 1.7041, "step": 1341 }, { "epoch": 0.04, "learning_rate": 1.994185185631655e-05, "loss": 1.6312, "step": 1342 }, { "epoch": 0.04, "learning_rate": 1.9941737895355296e-05, "loss": 1.6625, "step": 1343 }, { "epoch": 0.04, "learning_rate": 1.9941623823157076e-05, "loss": 1.6658, "step": 1344 }, { "epoch": 0.04, "learning_rate": 1.9941509639723155e-05, "loss": 1.6824, "step": 1345 }, { "epoch": 0.04, "learning_rate": 1.994139534505482e-05, "loss": 1.6928, "step": 1346 }, { "epoch": 0.04, "learning_rate": 1.9941280939153347e-05, "loss": 1.5455, "step": 1347 }, { "epoch": 0.04, "learning_rate": 1.9941166422020016e-05, "loss": 1.7014, "step": 1348 }, { "epoch": 0.04, "learning_rate": 1.9941051793656106e-05, "loss": 1.771, "step": 1349 }, { "epoch": 0.04, "learning_rate": 1.99409370540629e-05, "loss": 1.7288, "step": 1350 }, { "epoch": 0.05, "learning_rate": 1.9940822203241684e-05, "loss": 1.65, "step": 1351 }, { "epoch": 0.05, "learning_rate": 1.9940707241193746e-05, "loss": 1.5895, "step": 1352 }, { "epoch": 0.05, "learning_rate": 1.9940592167920367e-05, "loss": 1.7042, "step": 1353 }, { "epoch": 0.05, "learning_rate": 1.9940476983422834e-05, "loss": 1.6684, "step": 1354 }, { "epoch": 0.05, "learning_rate": 1.994036168770244e-05, "loss": 1.6267, "step": 1355 }, { "epoch": 0.05, "learning_rate": 1.9940246280760473e-05, "loss": 1.7436, "step": 1356 }, { "epoch": 0.05, "learning_rate": 1.9940130762598224e-05, "loss": 1.6783, "step": 1357 }, { "epoch": 0.05, "learning_rate": 1.9940015133216985e-05, "loss": 1.5508, "step": 1358 }, { "epoch": 0.05, "learning_rate": 1.993989939261805e-05, "loss": 1.5019, "step": 1359 }, { "epoch": 0.05, "learning_rate": 1.9939783540802715e-05, "loss": 1.6227, "step": 1360 }, { "epoch": 0.05, "learning_rate": 1.9939667577772278e-05, "loss": 1.5657, "step": 1361 }, { "epoch": 0.05, "learning_rate": 1.993955150352803e-05, "loss": 1.6167, "step": 1362 }, { "epoch": 0.05, "learning_rate": 1.9939435318071277e-05, "loss": 1.5727, "step": 1363 }, { "epoch": 0.05, "learning_rate": 1.9939319021403316e-05, "loss": 1.6162, "step": 1364 }, { "epoch": 0.05, "learning_rate": 1.9939202613525448e-05, "loss": 1.584, "step": 1365 }, { "epoch": 0.05, "learning_rate": 1.9939086094438975e-05, "loss": 1.6398, "step": 1366 }, { "epoch": 0.05, "learning_rate": 1.99389694641452e-05, "loss": 1.6015, "step": 1367 }, { "epoch": 0.05, "learning_rate": 1.9938852722645432e-05, "loss": 1.5844, "step": 1368 }, { "epoch": 0.05, "learning_rate": 1.9938735869940972e-05, "loss": 1.6408, "step": 1369 }, { "epoch": 0.05, "learning_rate": 1.9938618906033134e-05, "loss": 1.626, "step": 1370 }, { "epoch": 0.05, "learning_rate": 1.993850183092322e-05, "loss": 1.5115, "step": 1371 }, { "epoch": 0.05, "learning_rate": 1.9938384644612542e-05, "loss": 1.6032, "step": 1372 }, { "epoch": 0.05, "learning_rate": 1.9938267347102414e-05, "loss": 1.5745, "step": 1373 }, { "epoch": 0.05, "learning_rate": 1.9938149938394143e-05, "loss": 1.6143, "step": 1374 }, { "epoch": 0.05, "learning_rate": 1.993803241848905e-05, "loss": 1.7406, "step": 1375 }, { "epoch": 0.05, "learning_rate": 1.9937914787388447e-05, "loss": 1.6007, "step": 1376 }, { "epoch": 0.05, "learning_rate": 1.9937797045093645e-05, "loss": 1.6194, "step": 1377 }, { "epoch": 0.05, "learning_rate": 1.9937679191605964e-05, "loss": 1.5828, "step": 1378 }, { "epoch": 0.05, "learning_rate": 1.993756122692673e-05, "loss": 1.609, "step": 1379 }, { "epoch": 0.05, "learning_rate": 1.993744315105725e-05, "loss": 1.6348, "step": 1380 }, { "epoch": 0.05, "learning_rate": 1.9937324963998858e-05, "loss": 1.6401, "step": 1381 }, { "epoch": 0.05, "learning_rate": 1.993720666575287e-05, "loss": 1.6189, "step": 1382 }, { "epoch": 0.05, "learning_rate": 1.9937088256320606e-05, "loss": 1.68, "step": 1383 }, { "epoch": 0.05, "learning_rate": 1.9936969735703396e-05, "loss": 1.5399, "step": 1384 }, { "epoch": 0.05, "learning_rate": 1.993685110390257e-05, "loss": 1.6824, "step": 1385 }, { "epoch": 0.05, "learning_rate": 1.9936732360919447e-05, "loss": 1.5792, "step": 1386 }, { "epoch": 0.05, "learning_rate": 1.9936613506755357e-05, "loss": 1.5224, "step": 1387 }, { "epoch": 0.05, "learning_rate": 1.9936494541411633e-05, "loss": 1.6781, "step": 1388 }, { "epoch": 0.05, "learning_rate": 1.9936375464889608e-05, "loss": 1.6119, "step": 1389 }, { "epoch": 0.05, "learning_rate": 1.9936256277190608e-05, "loss": 1.6698, "step": 1390 }, { "epoch": 0.05, "learning_rate": 1.9936136978315972e-05, "loss": 1.6278, "step": 1391 }, { "epoch": 0.05, "learning_rate": 1.9936017568267032e-05, "loss": 1.5831, "step": 1392 }, { "epoch": 0.05, "learning_rate": 1.9935898047045126e-05, "loss": 1.5519, "step": 1393 }, { "epoch": 0.05, "learning_rate": 1.993577841465159e-05, "loss": 1.6717, "step": 1394 }, { "epoch": 0.05, "learning_rate": 1.993565867108776e-05, "loss": 1.6077, "step": 1395 }, { "epoch": 0.05, "learning_rate": 1.993553881635498e-05, "loss": 1.5602, "step": 1396 }, { "epoch": 0.05, "learning_rate": 1.993541885045459e-05, "loss": 1.7193, "step": 1397 }, { "epoch": 0.05, "learning_rate": 1.993529877338793e-05, "loss": 1.6941, "step": 1398 }, { "epoch": 0.05, "learning_rate": 1.993517858515635e-05, "loss": 1.7771, "step": 1399 }, { "epoch": 0.05, "learning_rate": 1.9935058285761185e-05, "loss": 1.7347, "step": 1400 }, { "epoch": 0.05, "learning_rate": 1.9934937875203788e-05, "loss": 1.7422, "step": 1401 }, { "epoch": 0.05, "learning_rate": 1.99348173534855e-05, "loss": 1.6905, "step": 1402 }, { "epoch": 0.05, "learning_rate": 1.9934696720607682e-05, "loss": 1.6943, "step": 1403 }, { "epoch": 0.05, "learning_rate": 1.9934575976571673e-05, "loss": 1.6391, "step": 1404 }, { "epoch": 0.05, "learning_rate": 1.9934455121378822e-05, "loss": 1.584, "step": 1405 }, { "epoch": 0.05, "learning_rate": 1.9934334155030492e-05, "loss": 1.7205, "step": 1406 }, { "epoch": 0.05, "learning_rate": 1.9934213077528026e-05, "loss": 1.6994, "step": 1407 }, { "epoch": 0.05, "learning_rate": 1.9934091888872785e-05, "loss": 1.6834, "step": 1408 }, { "epoch": 0.05, "learning_rate": 1.9933970589066126e-05, "loss": 1.6448, "step": 1409 }, { "epoch": 0.05, "learning_rate": 1.99338491781094e-05, "loss": 1.637, "step": 1410 }, { "epoch": 0.05, "learning_rate": 1.9933727656003964e-05, "loss": 1.6507, "step": 1411 }, { "epoch": 0.05, "learning_rate": 1.9933606022751188e-05, "loss": 1.6369, "step": 1412 }, { "epoch": 0.05, "learning_rate": 1.9933484278352426e-05, "loss": 1.6344, "step": 1413 }, { "epoch": 0.05, "learning_rate": 1.9933362422809043e-05, "loss": 1.7271, "step": 1414 }, { "epoch": 0.05, "learning_rate": 1.9933240456122397e-05, "loss": 1.6245, "step": 1415 }, { "epoch": 0.05, "learning_rate": 1.9933118378293862e-05, "loss": 1.6642, "step": 1416 }, { "epoch": 0.05, "learning_rate": 1.9932996189324796e-05, "loss": 1.6563, "step": 1417 }, { "epoch": 0.05, "learning_rate": 1.993287388921657e-05, "loss": 1.6208, "step": 1418 }, { "epoch": 0.05, "learning_rate": 1.9932751477970547e-05, "loss": 1.5515, "step": 1419 }, { "epoch": 0.05, "learning_rate": 1.9932628955588103e-05, "loss": 1.5433, "step": 1420 }, { "epoch": 0.05, "learning_rate": 1.993250632207061e-05, "loss": 1.6546, "step": 1421 }, { "epoch": 0.05, "learning_rate": 1.9932383577419432e-05, "loss": 1.6549, "step": 1422 }, { "epoch": 0.05, "learning_rate": 1.9932260721635946e-05, "loss": 1.5856, "step": 1423 }, { "epoch": 0.05, "learning_rate": 1.993213775472153e-05, "loss": 1.5211, "step": 1424 }, { "epoch": 0.05, "learning_rate": 1.9932014676677562e-05, "loss": 1.5634, "step": 1425 }, { "epoch": 0.05, "learning_rate": 1.993189148750541e-05, "loss": 1.7107, "step": 1426 }, { "epoch": 0.05, "learning_rate": 1.993176818720646e-05, "loss": 1.6024, "step": 1427 }, { "epoch": 0.05, "learning_rate": 1.9931644775782085e-05, "loss": 1.7109, "step": 1428 }, { "epoch": 0.05, "learning_rate": 1.993152125323367e-05, "loss": 1.6633, "step": 1429 }, { "epoch": 0.05, "learning_rate": 1.9931397619562597e-05, "loss": 1.5953, "step": 1430 }, { "epoch": 0.05, "learning_rate": 1.993127387477025e-05, "loss": 1.6051, "step": 1431 }, { "epoch": 0.05, "learning_rate": 1.9931150018858013e-05, "loss": 1.5875, "step": 1432 }, { "epoch": 0.05, "learning_rate": 1.993102605182727e-05, "loss": 1.7736, "step": 1433 }, { "epoch": 0.05, "learning_rate": 1.9930901973679408e-05, "loss": 1.6392, "step": 1434 }, { "epoch": 0.05, "learning_rate": 1.993077778441582e-05, "loss": 1.7044, "step": 1435 }, { "epoch": 0.05, "learning_rate": 1.9930653484037886e-05, "loss": 1.75, "step": 1436 }, { "epoch": 0.05, "learning_rate": 1.9930529072547008e-05, "loss": 1.5086, "step": 1437 }, { "epoch": 0.05, "learning_rate": 1.9930404549944577e-05, "loss": 1.5598, "step": 1438 }, { "epoch": 0.05, "learning_rate": 1.9930279916231975e-05, "loss": 1.6463, "step": 1439 }, { "epoch": 0.05, "learning_rate": 1.9930155171410606e-05, "loss": 1.479, "step": 1440 }, { "epoch": 0.05, "learning_rate": 1.9930030315481862e-05, "loss": 1.566, "step": 1441 }, { "epoch": 0.05, "learning_rate": 1.9929905348447143e-05, "loss": 1.5889, "step": 1442 }, { "epoch": 0.05, "learning_rate": 1.9929780270307846e-05, "loss": 1.6152, "step": 1443 }, { "epoch": 0.05, "learning_rate": 1.992965508106537e-05, "loss": 1.6965, "step": 1444 }, { "epoch": 0.05, "learning_rate": 1.992952978072112e-05, "loss": 1.6589, "step": 1445 }, { "epoch": 0.05, "learning_rate": 1.992940436927649e-05, "loss": 1.5212, "step": 1446 }, { "epoch": 0.05, "learning_rate": 1.9929278846732883e-05, "loss": 1.6293, "step": 1447 }, { "epoch": 0.05, "learning_rate": 1.9929153213091714e-05, "loss": 1.6139, "step": 1448 }, { "epoch": 0.05, "learning_rate": 1.9929027468354382e-05, "loss": 1.5531, "step": 1449 }, { "epoch": 0.05, "learning_rate": 1.992890161252229e-05, "loss": 1.5551, "step": 1450 }, { "epoch": 0.05, "learning_rate": 1.992877564559685e-05, "loss": 1.5414, "step": 1451 }, { "epoch": 0.05, "learning_rate": 1.9928649567579476e-05, "loss": 1.502, "step": 1452 }, { "epoch": 0.05, "learning_rate": 1.9928523378471573e-05, "loss": 1.521, "step": 1453 }, { "epoch": 0.05, "learning_rate": 1.9928397078274554e-05, "loss": 1.5405, "step": 1454 }, { "epoch": 0.05, "learning_rate": 1.9928270666989835e-05, "loss": 1.5948, "step": 1455 }, { "epoch": 0.05, "learning_rate": 1.9928144144618824e-05, "loss": 1.5975, "step": 1456 }, { "epoch": 0.05, "learning_rate": 1.9928017511162943e-05, "loss": 1.6154, "step": 1457 }, { "epoch": 0.05, "learning_rate": 1.9927890766623607e-05, "loss": 1.4858, "step": 1458 }, { "epoch": 0.05, "learning_rate": 1.9927763911002232e-05, "loss": 1.632, "step": 1459 }, { "epoch": 0.05, "learning_rate": 1.9927636944300238e-05, "loss": 1.6599, "step": 1460 }, { "epoch": 0.05, "learning_rate": 1.9927509866519052e-05, "loss": 1.641, "step": 1461 }, { "epoch": 0.05, "learning_rate": 1.992738267766009e-05, "loss": 1.6098, "step": 1462 }, { "epoch": 0.05, "learning_rate": 1.992725537772477e-05, "loss": 1.655, "step": 1463 }, { "epoch": 0.05, "learning_rate": 1.9927127966714527e-05, "loss": 1.5982, "step": 1464 }, { "epoch": 0.05, "learning_rate": 1.9927000444630776e-05, "loss": 1.5973, "step": 1465 }, { "epoch": 0.05, "learning_rate": 1.9926872811474952e-05, "loss": 1.5782, "step": 1466 }, { "epoch": 0.05, "learning_rate": 1.992674506724848e-05, "loss": 1.5643, "step": 1467 }, { "epoch": 0.05, "learning_rate": 1.9926617211952793e-05, "loss": 1.5919, "step": 1468 }, { "epoch": 0.05, "learning_rate": 1.9926489245589315e-05, "loss": 1.5572, "step": 1469 }, { "epoch": 0.05, "learning_rate": 1.9926361168159482e-05, "loss": 1.5989, "step": 1470 }, { "epoch": 0.05, "learning_rate": 1.9926232979664727e-05, "loss": 1.6261, "step": 1471 }, { "epoch": 0.05, "learning_rate": 1.9926104680106484e-05, "loss": 1.5254, "step": 1472 }, { "epoch": 0.05, "learning_rate": 1.9925976269486188e-05, "loss": 1.6487, "step": 1473 }, { "epoch": 0.05, "learning_rate": 1.9925847747805274e-05, "loss": 1.5753, "step": 1474 }, { "epoch": 0.05, "learning_rate": 1.9925719115065184e-05, "loss": 1.6428, "step": 1475 }, { "epoch": 0.05, "learning_rate": 1.9925590371267352e-05, "loss": 1.6491, "step": 1476 }, { "epoch": 0.05, "learning_rate": 1.9925461516413224e-05, "loss": 1.6211, "step": 1477 }, { "epoch": 0.05, "learning_rate": 1.9925332550504234e-05, "loss": 1.6518, "step": 1478 }, { "epoch": 0.05, "learning_rate": 1.9925203473541838e-05, "loss": 1.5852, "step": 1479 }, { "epoch": 0.05, "learning_rate": 1.9925074285527467e-05, "loss": 1.5501, "step": 1480 }, { "epoch": 0.05, "learning_rate": 1.9924944986462576e-05, "loss": 1.6937, "step": 1481 }, { "epoch": 0.05, "learning_rate": 1.9924815576348604e-05, "loss": 1.6511, "step": 1482 }, { "epoch": 0.05, "learning_rate": 1.9924686055187003e-05, "loss": 1.4932, "step": 1483 }, { "epoch": 0.05, "learning_rate": 1.9924556422979223e-05, "loss": 1.5726, "step": 1484 }, { "epoch": 0.05, "learning_rate": 1.9924426679726715e-05, "loss": 1.574, "step": 1485 }, { "epoch": 0.05, "learning_rate": 1.9924296825430928e-05, "loss": 1.5853, "step": 1486 }, { "epoch": 0.05, "learning_rate": 1.9924166860093313e-05, "loss": 1.5669, "step": 1487 }, { "epoch": 0.05, "learning_rate": 1.992403678371533e-05, "loss": 1.595, "step": 1488 }, { "epoch": 0.05, "learning_rate": 1.992390659629843e-05, "loss": 1.6222, "step": 1489 }, { "epoch": 0.05, "learning_rate": 1.9923776297844074e-05, "loss": 1.5283, "step": 1490 }, { "epoch": 0.05, "learning_rate": 1.9923645888353715e-05, "loss": 1.582, "step": 1491 }, { "epoch": 0.05, "learning_rate": 1.9923515367828812e-05, "loss": 1.7117, "step": 1492 }, { "epoch": 0.05, "learning_rate": 1.992338473627083e-05, "loss": 1.7249, "step": 1493 }, { "epoch": 0.05, "learning_rate": 1.9923253993681226e-05, "loss": 1.613, "step": 1494 }, { "epoch": 0.05, "learning_rate": 1.9923123140061467e-05, "loss": 1.6258, "step": 1495 }, { "epoch": 0.05, "learning_rate": 1.9922992175413016e-05, "loss": 1.609, "step": 1496 }, { "epoch": 0.05, "learning_rate": 1.9922861099737335e-05, "loss": 1.7442, "step": 1497 }, { "epoch": 0.05, "learning_rate": 1.9922729913035893e-05, "loss": 1.766, "step": 1498 }, { "epoch": 0.05, "learning_rate": 1.9922598615310157e-05, "loss": 1.7545, "step": 1499 }, { "epoch": 0.05, "learning_rate": 1.99224672065616e-05, "loss": 1.5772, "step": 1500 }, { "epoch": 0.05, "learning_rate": 1.992233568679169e-05, "loss": 1.619, "step": 1501 }, { "epoch": 0.05, "learning_rate": 1.9922204056001896e-05, "loss": 1.5486, "step": 1502 }, { "epoch": 0.05, "learning_rate": 1.9922072314193692e-05, "loss": 1.5604, "step": 1503 }, { "epoch": 0.05, "learning_rate": 1.9921940461368552e-05, "loss": 1.5375, "step": 1504 }, { "epoch": 0.05, "learning_rate": 1.992180849752795e-05, "loss": 1.5536, "step": 1505 }, { "epoch": 0.05, "learning_rate": 1.992167642267337e-05, "loss": 1.5595, "step": 1506 }, { "epoch": 0.05, "learning_rate": 1.9921544236806284e-05, "loss": 1.6689, "step": 1507 }, { "epoch": 0.05, "learning_rate": 1.9921411939928167e-05, "loss": 1.6274, "step": 1508 }, { "epoch": 0.05, "learning_rate": 1.9921279532040504e-05, "loss": 1.5988, "step": 1509 }, { "epoch": 0.05, "learning_rate": 1.9921147013144782e-05, "loss": 1.6223, "step": 1510 }, { "epoch": 0.05, "learning_rate": 1.9921014383242472e-05, "loss": 1.6844, "step": 1511 }, { "epoch": 0.05, "learning_rate": 1.9920881642335064e-05, "loss": 1.7422, "step": 1512 }, { "epoch": 0.05, "learning_rate": 1.9920748790424043e-05, "loss": 1.6293, "step": 1513 }, { "epoch": 0.05, "learning_rate": 1.99206158275109e-05, "loss": 1.6952, "step": 1514 }, { "epoch": 0.05, "learning_rate": 1.9920482753597113e-05, "loss": 1.7939, "step": 1515 }, { "epoch": 0.05, "learning_rate": 1.992034956868418e-05, "loss": 1.6995, "step": 1516 }, { "epoch": 0.05, "learning_rate": 1.9920216272773584e-05, "loss": 1.7335, "step": 1517 }, { "epoch": 0.05, "learning_rate": 1.992008286586682e-05, "loss": 1.6663, "step": 1518 }, { "epoch": 0.05, "learning_rate": 1.991994934796538e-05, "loss": 1.7134, "step": 1519 }, { "epoch": 0.05, "learning_rate": 1.991981571907076e-05, "loss": 1.6596, "step": 1520 }, { "epoch": 0.05, "learning_rate": 1.9919681979184452e-05, "loss": 1.7435, "step": 1521 }, { "epoch": 0.05, "learning_rate": 1.9919548128307954e-05, "loss": 1.772, "step": 1522 }, { "epoch": 0.05, "learning_rate": 1.9919414166442763e-05, "loss": 1.6876, "step": 1523 }, { "epoch": 0.05, "learning_rate": 1.991928009359038e-05, "loss": 1.7233, "step": 1524 }, { "epoch": 0.05, "learning_rate": 1.9919145909752305e-05, "loss": 1.6906, "step": 1525 }, { "epoch": 0.05, "learning_rate": 1.9919011614930036e-05, "loss": 1.722, "step": 1526 }, { "epoch": 0.05, "learning_rate": 1.991887720912508e-05, "loss": 1.7439, "step": 1527 }, { "epoch": 0.05, "learning_rate": 1.991874269233893e-05, "loss": 1.6555, "step": 1528 }, { "epoch": 0.05, "learning_rate": 1.9918608064573107e-05, "loss": 1.5964, "step": 1529 }, { "epoch": 0.05, "learning_rate": 1.9918473325829108e-05, "loss": 1.6349, "step": 1530 }, { "epoch": 0.05, "learning_rate": 1.991833847610844e-05, "loss": 1.5417, "step": 1531 }, { "epoch": 0.05, "learning_rate": 1.9918203515412616e-05, "loss": 1.7047, "step": 1532 }, { "epoch": 0.05, "learning_rate": 1.9918068443743147e-05, "loss": 1.6198, "step": 1533 }, { "epoch": 0.05, "learning_rate": 1.991793326110154e-05, "loss": 1.6172, "step": 1534 }, { "epoch": 0.05, "learning_rate": 1.991779796748931e-05, "loss": 1.7319, "step": 1535 }, { "epoch": 0.05, "learning_rate": 1.9917662562907963e-05, "loss": 1.6024, "step": 1536 }, { "epoch": 0.05, "learning_rate": 1.991752704735903e-05, "loss": 1.6104, "step": 1537 }, { "epoch": 0.05, "learning_rate": 1.9917391420844012e-05, "loss": 1.6089, "step": 1538 }, { "epoch": 0.05, "learning_rate": 1.9917255683364437e-05, "loss": 1.8177, "step": 1539 }, { "epoch": 0.05, "learning_rate": 1.9917119834921818e-05, "loss": 1.554, "step": 1540 }, { "epoch": 0.05, "learning_rate": 1.9916983875517672e-05, "loss": 1.6214, "step": 1541 }, { "epoch": 0.05, "learning_rate": 1.9916847805153528e-05, "loss": 1.6718, "step": 1542 }, { "epoch": 0.05, "learning_rate": 1.9916711623830904e-05, "loss": 1.6154, "step": 1543 }, { "epoch": 0.05, "learning_rate": 1.9916575331551328e-05, "loss": 1.6384, "step": 1544 }, { "epoch": 0.05, "learning_rate": 1.9916438928316318e-05, "loss": 1.6323, "step": 1545 }, { "epoch": 0.05, "learning_rate": 1.9916302414127408e-05, "loss": 1.5907, "step": 1546 }, { "epoch": 0.05, "learning_rate": 1.991616578898612e-05, "loss": 1.6018, "step": 1547 }, { "epoch": 0.05, "learning_rate": 1.9916029052893982e-05, "loss": 1.7244, "step": 1548 }, { "epoch": 0.05, "learning_rate": 1.9915892205852528e-05, "loss": 1.6609, "step": 1549 }, { "epoch": 0.05, "learning_rate": 1.9915755247863284e-05, "loss": 1.6487, "step": 1550 }, { "epoch": 0.05, "learning_rate": 1.991561817892779e-05, "loss": 1.4927, "step": 1551 }, { "epoch": 0.05, "learning_rate": 1.9915480999047573e-05, "loss": 1.5983, "step": 1552 }, { "epoch": 0.05, "learning_rate": 1.991534370822417e-05, "loss": 1.5455, "step": 1553 }, { "epoch": 0.05, "learning_rate": 1.9915206306459117e-05, "loss": 1.7186, "step": 1554 }, { "epoch": 0.05, "learning_rate": 1.9915068793753952e-05, "loss": 1.5706, "step": 1555 }, { "epoch": 0.05, "learning_rate": 1.9914931170110216e-05, "loss": 1.6138, "step": 1556 }, { "epoch": 0.05, "learning_rate": 1.9914793435529446e-05, "loss": 1.5778, "step": 1557 }, { "epoch": 0.05, "learning_rate": 1.9914655590013177e-05, "loss": 1.6456, "step": 1558 }, { "epoch": 0.05, "learning_rate": 1.9914517633562962e-05, "loss": 1.6103, "step": 1559 }, { "epoch": 0.05, "learning_rate": 1.9914379566180342e-05, "loss": 1.6497, "step": 1560 }, { "epoch": 0.05, "learning_rate": 1.9914241387866858e-05, "loss": 1.5231, "step": 1561 }, { "epoch": 0.05, "learning_rate": 1.9914103098624055e-05, "loss": 1.5421, "step": 1562 }, { "epoch": 0.05, "learning_rate": 1.9913964698453486e-05, "loss": 1.5404, "step": 1563 }, { "epoch": 0.05, "learning_rate": 1.99138261873567e-05, "loss": 1.6193, "step": 1564 }, { "epoch": 0.05, "learning_rate": 1.9913687565335237e-05, "loss": 1.5608, "step": 1565 }, { "epoch": 0.05, "learning_rate": 1.9913548832390658e-05, "loss": 1.6118, "step": 1566 }, { "epoch": 0.05, "learning_rate": 1.991340998852451e-05, "loss": 1.6904, "step": 1567 }, { "epoch": 0.05, "learning_rate": 1.9913271033738354e-05, "loss": 1.6504, "step": 1568 }, { "epoch": 0.05, "learning_rate": 1.9913131968033734e-05, "loss": 1.6743, "step": 1569 }, { "epoch": 0.05, "learning_rate": 1.9912992791412212e-05, "loss": 1.6521, "step": 1570 }, { "epoch": 0.05, "learning_rate": 1.991285350387535e-05, "loss": 1.6093, "step": 1571 }, { "epoch": 0.05, "learning_rate": 1.9912714105424694e-05, "loss": 1.6298, "step": 1572 }, { "epoch": 0.05, "learning_rate": 1.991257459606181e-05, "loss": 1.6585, "step": 1573 }, { "epoch": 0.05, "learning_rate": 1.9912434975788263e-05, "loss": 1.6816, "step": 1574 }, { "epoch": 0.05, "learning_rate": 1.991229524460561e-05, "loss": 1.7307, "step": 1575 }, { "epoch": 0.05, "learning_rate": 1.991215540251542e-05, "loss": 1.7305, "step": 1576 }, { "epoch": 0.05, "learning_rate": 1.991201544951925e-05, "loss": 1.6997, "step": 1577 }, { "epoch": 0.05, "learning_rate": 1.991187538561867e-05, "loss": 1.7011, "step": 1578 }, { "epoch": 0.05, "learning_rate": 1.991173521081525e-05, "loss": 1.5956, "step": 1579 }, { "epoch": 0.05, "learning_rate": 1.991159492511055e-05, "loss": 1.6537, "step": 1580 }, { "epoch": 0.05, "learning_rate": 1.9911454528506148e-05, "loss": 1.5996, "step": 1581 }, { "epoch": 0.05, "learning_rate": 1.9911314021003614e-05, "loss": 1.5909, "step": 1582 }, { "epoch": 0.05, "learning_rate": 1.9911173402604517e-05, "loss": 1.6676, "step": 1583 }, { "epoch": 0.05, "learning_rate": 1.9911032673310427e-05, "loss": 1.5641, "step": 1584 }, { "epoch": 0.05, "learning_rate": 1.9910891833122926e-05, "loss": 1.6612, "step": 1585 }, { "epoch": 0.05, "learning_rate": 1.991075088204359e-05, "loss": 1.5844, "step": 1586 }, { "epoch": 0.05, "learning_rate": 1.9910609820073986e-05, "loss": 1.6367, "step": 1587 }, { "epoch": 0.05, "learning_rate": 1.9910468647215708e-05, "loss": 1.66, "step": 1588 }, { "epoch": 0.05, "learning_rate": 1.991032736347032e-05, "loss": 1.7084, "step": 1589 }, { "epoch": 0.05, "learning_rate": 1.991018596883941e-05, "loss": 1.6064, "step": 1590 }, { "epoch": 0.05, "learning_rate": 1.9910044463324564e-05, "loss": 1.6112, "step": 1591 }, { "epoch": 0.05, "learning_rate": 1.9909902846927355e-05, "loss": 1.7017, "step": 1592 }, { "epoch": 0.05, "learning_rate": 1.9909761119649377e-05, "loss": 1.6031, "step": 1593 }, { "epoch": 0.05, "learning_rate": 1.990961928149221e-05, "loss": 1.5863, "step": 1594 }, { "epoch": 0.05, "learning_rate": 1.9909477332457447e-05, "loss": 1.752, "step": 1595 }, { "epoch": 0.05, "learning_rate": 1.990933527254667e-05, "loss": 1.6845, "step": 1596 }, { "epoch": 0.05, "learning_rate": 1.990919310176147e-05, "loss": 1.5922, "step": 1597 }, { "epoch": 0.05, "learning_rate": 1.990905082010344e-05, "loss": 1.6785, "step": 1598 }, { "epoch": 0.05, "learning_rate": 1.9908908427574172e-05, "loss": 1.6501, "step": 1599 }, { "epoch": 0.05, "learning_rate": 1.9908765924175256e-05, "loss": 1.5012, "step": 1600 }, { "epoch": 0.05, "learning_rate": 1.9908623309908292e-05, "loss": 1.5435, "step": 1601 }, { "epoch": 0.05, "learning_rate": 1.9908480584774867e-05, "loss": 1.6428, "step": 1602 }, { "epoch": 0.05, "learning_rate": 1.9908337748776585e-05, "loss": 1.5628, "step": 1603 }, { "epoch": 0.05, "learning_rate": 1.9908194801915043e-05, "loss": 1.5244, "step": 1604 }, { "epoch": 0.05, "learning_rate": 1.9908051744191837e-05, "loss": 1.6173, "step": 1605 }, { "epoch": 0.05, "learning_rate": 1.9907908575608573e-05, "loss": 1.6007, "step": 1606 }, { "epoch": 0.05, "learning_rate": 1.990776529616685e-05, "loss": 1.5923, "step": 1607 }, { "epoch": 0.05, "learning_rate": 1.990762190586827e-05, "loss": 1.4939, "step": 1608 }, { "epoch": 0.05, "learning_rate": 1.9907478404714438e-05, "loss": 1.6715, "step": 1609 }, { "epoch": 0.05, "learning_rate": 1.990733479270696e-05, "loss": 1.4889, "step": 1610 }, { "epoch": 0.05, "learning_rate": 1.9907191069847445e-05, "loss": 1.5912, "step": 1611 }, { "epoch": 0.05, "learning_rate": 1.99070472361375e-05, "loss": 1.6904, "step": 1612 }, { "epoch": 0.05, "learning_rate": 1.990690329157873e-05, "loss": 1.5936, "step": 1613 }, { "epoch": 0.05, "learning_rate": 1.9906759236172754e-05, "loss": 1.5951, "step": 1614 }, { "epoch": 0.05, "learning_rate": 1.9906615069921175e-05, "loss": 1.6328, "step": 1615 }, { "epoch": 0.05, "learning_rate": 1.990647079282561e-05, "loss": 1.5812, "step": 1616 }, { "epoch": 0.05, "learning_rate": 1.9906326404887675e-05, "loss": 1.6324, "step": 1617 }, { "epoch": 0.05, "learning_rate": 1.9906181906108983e-05, "loss": 1.6393, "step": 1618 }, { "epoch": 0.05, "learning_rate": 1.990603729649115e-05, "loss": 1.5389, "step": 1619 }, { "epoch": 0.05, "learning_rate": 1.9905892576035798e-05, "loss": 1.4796, "step": 1620 }, { "epoch": 0.05, "learning_rate": 1.9905747744744547e-05, "loss": 1.6249, "step": 1621 }, { "epoch": 0.05, "learning_rate": 1.990560280261901e-05, "loss": 1.5247, "step": 1622 }, { "epoch": 0.05, "learning_rate": 1.990545774966081e-05, "loss": 1.547, "step": 1623 }, { "epoch": 0.05, "learning_rate": 1.990531258587158e-05, "loss": 1.5541, "step": 1624 }, { "epoch": 0.05, "learning_rate": 1.9905167311252933e-05, "loss": 1.5774, "step": 1625 }, { "epoch": 0.05, "learning_rate": 1.99050219258065e-05, "loss": 1.5681, "step": 1626 }, { "epoch": 0.05, "learning_rate": 1.9904876429533912e-05, "loss": 1.5917, "step": 1627 }, { "epoch": 0.05, "learning_rate": 1.9904730822436786e-05, "loss": 1.5848, "step": 1628 }, { "epoch": 0.05, "learning_rate": 1.9904585104516752e-05, "loss": 1.6195, "step": 1629 }, { "epoch": 0.05, "learning_rate": 1.990443927577545e-05, "loss": 1.6123, "step": 1630 }, { "epoch": 0.05, "learning_rate": 1.9904293336214508e-05, "loss": 1.5615, "step": 1631 }, { "epoch": 0.05, "learning_rate": 1.9904147285835557e-05, "loss": 1.7189, "step": 1632 }, { "epoch": 0.05, "learning_rate": 1.990400112464023e-05, "loss": 1.5202, "step": 1633 }, { "epoch": 0.05, "learning_rate": 1.9903854852630167e-05, "loss": 1.5384, "step": 1634 }, { "epoch": 0.05, "learning_rate": 1.9903708469807e-05, "loss": 1.6806, "step": 1635 }, { "epoch": 0.05, "learning_rate": 1.9903561976172367e-05, "loss": 1.5484, "step": 1636 }, { "epoch": 0.05, "learning_rate": 1.990341537172791e-05, "loss": 1.5839, "step": 1637 }, { "epoch": 0.05, "learning_rate": 1.990326865647527e-05, "loss": 1.5739, "step": 1638 }, { "epoch": 0.05, "learning_rate": 1.9903121830416085e-05, "loss": 1.5857, "step": 1639 }, { "epoch": 0.05, "learning_rate": 1.9902974893552e-05, "loss": 1.5014, "step": 1640 }, { "epoch": 0.05, "learning_rate": 1.9902827845884658e-05, "loss": 1.4184, "step": 1641 }, { "epoch": 0.05, "learning_rate": 1.9902680687415704e-05, "loss": 1.4572, "step": 1642 }, { "epoch": 0.05, "learning_rate": 1.9902533418146784e-05, "loss": 1.555, "step": 1643 }, { "epoch": 0.05, "learning_rate": 1.990238603807955e-05, "loss": 1.4731, "step": 1644 }, { "epoch": 0.05, "learning_rate": 1.9902238547215652e-05, "loss": 1.4838, "step": 1645 }, { "epoch": 0.05, "learning_rate": 1.990209094555673e-05, "loss": 1.5053, "step": 1646 }, { "epoch": 0.05, "learning_rate": 1.9901943233104444e-05, "loss": 1.53, "step": 1647 }, { "epoch": 0.05, "learning_rate": 1.9901795409860445e-05, "loss": 1.5205, "step": 1648 }, { "epoch": 0.05, "learning_rate": 1.9901647475826387e-05, "loss": 1.3975, "step": 1649 }, { "epoch": 0.06, "learning_rate": 1.9901499431003923e-05, "loss": 1.5608, "step": 1650 }, { "epoch": 0.06, "learning_rate": 1.9901351275394712e-05, "loss": 1.5403, "step": 1651 }, { "epoch": 0.06, "learning_rate": 1.9901203009000413e-05, "loss": 1.498, "step": 1652 }, { "epoch": 0.06, "learning_rate": 1.990105463182268e-05, "loss": 1.5227, "step": 1653 }, { "epoch": 0.06, "learning_rate": 1.990090614386318e-05, "loss": 1.5104, "step": 1654 }, { "epoch": 0.06, "learning_rate": 1.9900757545123564e-05, "loss": 1.4581, "step": 1655 }, { "epoch": 0.06, "learning_rate": 1.990060883560551e-05, "loss": 1.5647, "step": 1656 }, { "epoch": 0.06, "learning_rate": 1.9900460015310667e-05, "loss": 1.5567, "step": 1657 }, { "epoch": 0.06, "learning_rate": 1.9900311084240703e-05, "loss": 1.6768, "step": 1658 }, { "epoch": 0.06, "learning_rate": 1.9900162042397295e-05, "loss": 1.6046, "step": 1659 }, { "epoch": 0.06, "learning_rate": 1.99000128897821e-05, "loss": 1.5633, "step": 1660 }, { "epoch": 0.06, "learning_rate": 1.989986362639679e-05, "loss": 1.7139, "step": 1661 }, { "epoch": 0.06, "learning_rate": 1.9899714252243037e-05, "loss": 1.5789, "step": 1662 }, { "epoch": 0.06, "learning_rate": 1.989956476732251e-05, "loss": 1.7587, "step": 1663 }, { "epoch": 0.06, "learning_rate": 1.989941517163688e-05, "loss": 1.7593, "step": 1664 }, { "epoch": 0.06, "learning_rate": 1.9899265465187823e-05, "loss": 1.6886, "step": 1665 }, { "epoch": 0.06, "learning_rate": 1.9899115647977015e-05, "loss": 1.6715, "step": 1666 }, { "epoch": 0.06, "learning_rate": 1.989896572000613e-05, "loss": 1.5999, "step": 1667 }, { "epoch": 0.06, "learning_rate": 1.9898815681276847e-05, "loss": 1.6688, "step": 1668 }, { "epoch": 0.06, "learning_rate": 1.9898665531790846e-05, "loss": 1.6367, "step": 1669 }, { "epoch": 0.06, "learning_rate": 1.9898515271549805e-05, "loss": 1.6261, "step": 1670 }, { "epoch": 0.06, "learning_rate": 1.9898364900555405e-05, "loss": 1.7993, "step": 1671 }, { "epoch": 0.06, "learning_rate": 1.989821441880933e-05, "loss": 1.79, "step": 1672 }, { "epoch": 0.06, "learning_rate": 1.989806382631326e-05, "loss": 1.7234, "step": 1673 }, { "epoch": 0.06, "learning_rate": 1.9897913123068887e-05, "loss": 1.5697, "step": 1674 }, { "epoch": 0.06, "learning_rate": 1.989776230907789e-05, "loss": 1.712, "step": 1675 }, { "epoch": 0.06, "learning_rate": 1.9897611384341963e-05, "loss": 1.72, "step": 1676 }, { "epoch": 0.06, "learning_rate": 1.9897460348862785e-05, "loss": 1.7317, "step": 1677 }, { "epoch": 0.06, "learning_rate": 1.989730920264206e-05, "loss": 1.7265, "step": 1678 }, { "epoch": 0.06, "learning_rate": 1.989715794568147e-05, "loss": 1.6748, "step": 1679 }, { "epoch": 0.06, "learning_rate": 1.9897006577982702e-05, "loss": 1.6891, "step": 1680 }, { "epoch": 0.06, "learning_rate": 1.9896855099547462e-05, "loss": 1.6817, "step": 1681 }, { "epoch": 0.06, "learning_rate": 1.9896703510377436e-05, "loss": 1.7089, "step": 1682 }, { "epoch": 0.06, "learning_rate": 1.989655181047433e-05, "loss": 1.6785, "step": 1683 }, { "epoch": 0.06, "learning_rate": 1.9896399999839828e-05, "loss": 1.6672, "step": 1684 }, { "epoch": 0.06, "learning_rate": 1.9896248078475636e-05, "loss": 1.6927, "step": 1685 }, { "epoch": 0.06, "learning_rate": 1.9896096046383456e-05, "loss": 1.6124, "step": 1686 }, { "epoch": 0.06, "learning_rate": 1.989594390356498e-05, "loss": 1.6399, "step": 1687 }, { "epoch": 0.06, "learning_rate": 1.9895791650021926e-05, "loss": 1.6557, "step": 1688 }, { "epoch": 0.06, "learning_rate": 1.989563928575598e-05, "loss": 1.5971, "step": 1689 }, { "epoch": 0.06, "learning_rate": 1.989548681076886e-05, "loss": 1.7183, "step": 1690 }, { "epoch": 0.06, "learning_rate": 1.9895334225062264e-05, "loss": 1.6397, "step": 1691 }, { "epoch": 0.06, "learning_rate": 1.98951815286379e-05, "loss": 1.6344, "step": 1692 }, { "epoch": 0.06, "learning_rate": 1.9895028721497482e-05, "loss": 1.6484, "step": 1693 }, { "epoch": 0.06, "learning_rate": 1.9894875803642715e-05, "loss": 1.7213, "step": 1694 }, { "epoch": 0.06, "learning_rate": 1.989472277507531e-05, "loss": 1.6691, "step": 1695 }, { "epoch": 0.06, "learning_rate": 1.9894569635796987e-05, "loss": 1.6297, "step": 1696 }, { "epoch": 0.06, "learning_rate": 1.9894416385809444e-05, "loss": 1.7651, "step": 1697 }, { "epoch": 0.06, "learning_rate": 1.9894263025114414e-05, "loss": 1.5895, "step": 1698 }, { "epoch": 0.06, "learning_rate": 1.9894109553713597e-05, "loss": 1.6218, "step": 1699 }, { "epoch": 0.06, "learning_rate": 1.989395597160872e-05, "loss": 1.6958, "step": 1700 }, { "epoch": 0.06, "learning_rate": 1.9893802278801496e-05, "loss": 1.7562, "step": 1701 }, { "epoch": 0.06, "learning_rate": 1.9893648475293646e-05, "loss": 1.6363, "step": 1702 }, { "epoch": 0.06, "learning_rate": 1.9893494561086895e-05, "loss": 1.6075, "step": 1703 }, { "epoch": 0.06, "learning_rate": 1.989334053618296e-05, "loss": 1.6797, "step": 1704 }, { "epoch": 0.06, "learning_rate": 1.9893186400583568e-05, "loss": 1.5979, "step": 1705 }, { "epoch": 0.06, "learning_rate": 1.9893032154290438e-05, "loss": 1.6787, "step": 1706 }, { "epoch": 0.06, "learning_rate": 1.98928777973053e-05, "loss": 1.6579, "step": 1707 }, { "epoch": 0.06, "learning_rate": 1.9892723329629885e-05, "loss": 1.6069, "step": 1708 }, { "epoch": 0.06, "learning_rate": 1.9892568751265914e-05, "loss": 1.6876, "step": 1709 }, { "epoch": 0.06, "learning_rate": 1.9892414062215123e-05, "loss": 1.7183, "step": 1710 }, { "epoch": 0.06, "learning_rate": 1.9892259262479238e-05, "loss": 1.5502, "step": 1711 }, { "epoch": 0.06, "learning_rate": 1.9892104352059988e-05, "loss": 1.6571, "step": 1712 }, { "epoch": 0.06, "learning_rate": 1.9891949330959116e-05, "loss": 1.6601, "step": 1713 }, { "epoch": 0.06, "learning_rate": 1.989179419917835e-05, "loss": 1.6564, "step": 1714 }, { "epoch": 0.06, "learning_rate": 1.9891638956719426e-05, "loss": 1.5891, "step": 1715 }, { "epoch": 0.06, "learning_rate": 1.9891483603584082e-05, "loss": 1.5044, "step": 1716 }, { "epoch": 0.06, "learning_rate": 1.9891328139774057e-05, "loss": 1.5097, "step": 1717 }, { "epoch": 0.06, "learning_rate": 1.989117256529109e-05, "loss": 1.6384, "step": 1718 }, { "epoch": 0.06, "learning_rate": 1.9891016880136923e-05, "loss": 1.626, "step": 1719 }, { "epoch": 0.06, "learning_rate": 1.9890861084313293e-05, "loss": 1.5575, "step": 1720 }, { "epoch": 0.06, "learning_rate": 1.989070517782195e-05, "loss": 1.599, "step": 1721 }, { "epoch": 0.06, "learning_rate": 1.9890549160664633e-05, "loss": 1.622, "step": 1722 }, { "epoch": 0.06, "learning_rate": 1.989039303284309e-05, "loss": 1.594, "step": 1723 }, { "epoch": 0.06, "learning_rate": 1.9890236794359065e-05, "loss": 1.7382, "step": 1724 }, { "epoch": 0.06, "learning_rate": 1.989008044521431e-05, "loss": 1.584, "step": 1725 }, { "epoch": 0.06, "learning_rate": 1.9889923985410576e-05, "loss": 1.6502, "step": 1726 }, { "epoch": 0.06, "learning_rate": 1.9889767414949606e-05, "loss": 1.658, "step": 1727 }, { "epoch": 0.06, "learning_rate": 1.988961073383316e-05, "loss": 1.6597, "step": 1728 }, { "epoch": 0.06, "learning_rate": 1.9889453942062988e-05, "loss": 1.5839, "step": 1729 }, { "epoch": 0.06, "learning_rate": 1.988929703964084e-05, "loss": 1.6568, "step": 1730 }, { "epoch": 0.06, "learning_rate": 1.988914002656848e-05, "loss": 1.5891, "step": 1731 }, { "epoch": 0.06, "learning_rate": 1.9888982902847658e-05, "loss": 1.6004, "step": 1732 }, { "epoch": 0.06, "learning_rate": 1.9888825668480133e-05, "loss": 1.6287, "step": 1733 }, { "epoch": 0.06, "learning_rate": 1.988866832346767e-05, "loss": 1.5897, "step": 1734 }, { "epoch": 0.06, "learning_rate": 1.9888510867812022e-05, "loss": 1.5955, "step": 1735 }, { "epoch": 0.06, "learning_rate": 1.9888353301514953e-05, "loss": 1.6448, "step": 1736 }, { "epoch": 0.06, "learning_rate": 1.988819562457823e-05, "loss": 1.564, "step": 1737 }, { "epoch": 0.06, "learning_rate": 1.988803783700361e-05, "loss": 1.6283, "step": 1738 }, { "epoch": 0.06, "learning_rate": 1.9887879938792866e-05, "loss": 1.721, "step": 1739 }, { "epoch": 0.06, "learning_rate": 1.988772192994776e-05, "loss": 1.6441, "step": 1740 }, { "epoch": 0.06, "learning_rate": 1.988756381047006e-05, "loss": 1.6102, "step": 1741 }, { "epoch": 0.06, "learning_rate": 1.988740558036154e-05, "loss": 1.7015, "step": 1742 }, { "epoch": 0.06, "learning_rate": 1.9887247239623964e-05, "loss": 1.6103, "step": 1743 }, { "epoch": 0.06, "learning_rate": 1.9887088788259105e-05, "loss": 1.6474, "step": 1744 }, { "epoch": 0.06, "learning_rate": 1.9886930226268738e-05, "loss": 1.5906, "step": 1745 }, { "epoch": 0.06, "learning_rate": 1.9886771553654637e-05, "loss": 1.642, "step": 1746 }, { "epoch": 0.06, "learning_rate": 1.988661277041858e-05, "loss": 1.6043, "step": 1747 }, { "epoch": 0.06, "learning_rate": 1.9886453876562337e-05, "loss": 1.5959, "step": 1748 }, { "epoch": 0.06, "learning_rate": 1.988629487208769e-05, "loss": 1.6846, "step": 1749 }, { "epoch": 0.06, "learning_rate": 1.988613575699642e-05, "loss": 1.5737, "step": 1750 }, { "epoch": 0.06, "learning_rate": 1.98859765312903e-05, "loss": 1.5726, "step": 1751 }, { "epoch": 0.06, "learning_rate": 1.9885817194971116e-05, "loss": 1.6333, "step": 1752 }, { "epoch": 0.06, "learning_rate": 1.9885657748040655e-05, "loss": 1.619, "step": 1753 }, { "epoch": 0.06, "learning_rate": 1.9885498190500697e-05, "loss": 1.6317, "step": 1754 }, { "epoch": 0.06, "learning_rate": 1.9885338522353026e-05, "loss": 1.6235, "step": 1755 }, { "epoch": 0.06, "learning_rate": 1.988517874359943e-05, "loss": 1.5203, "step": 1756 }, { "epoch": 0.06, "learning_rate": 1.9885018854241695e-05, "loss": 1.6985, "step": 1757 }, { "epoch": 0.06, "learning_rate": 1.9884858854281613e-05, "loss": 1.5791, "step": 1758 }, { "epoch": 0.06, "learning_rate": 1.9884698743720973e-05, "loss": 1.678, "step": 1759 }, { "epoch": 0.06, "learning_rate": 1.9884538522561567e-05, "loss": 1.6029, "step": 1760 }, { "epoch": 0.06, "learning_rate": 1.9884378190805188e-05, "loss": 1.498, "step": 1761 }, { "epoch": 0.06, "learning_rate": 1.9884217748453625e-05, "loss": 1.5159, "step": 1762 }, { "epoch": 0.06, "learning_rate": 1.9884057195508683e-05, "loss": 1.5791, "step": 1763 }, { "epoch": 0.06, "learning_rate": 1.9883896531972145e-05, "loss": 1.5841, "step": 1764 }, { "epoch": 0.06, "learning_rate": 1.9883735757845822e-05, "loss": 1.6456, "step": 1765 }, { "epoch": 0.06, "learning_rate": 1.9883574873131502e-05, "loss": 1.6742, "step": 1766 }, { "epoch": 0.06, "learning_rate": 1.9883413877830996e-05, "loss": 1.6463, "step": 1767 }, { "epoch": 0.06, "learning_rate": 1.9883252771946094e-05, "loss": 1.5969, "step": 1768 }, { "epoch": 0.06, "learning_rate": 1.9883091555478608e-05, "loss": 1.5891, "step": 1769 }, { "epoch": 0.06, "learning_rate": 1.9882930228430337e-05, "loss": 1.6622, "step": 1770 }, { "epoch": 0.06, "learning_rate": 1.9882768790803086e-05, "loss": 1.6605, "step": 1771 }, { "epoch": 0.06, "learning_rate": 1.9882607242598663e-05, "loss": 1.5896, "step": 1772 }, { "epoch": 0.06, "learning_rate": 1.9882445583818876e-05, "loss": 1.7623, "step": 1773 }, { "epoch": 0.06, "learning_rate": 1.988228381446553e-05, "loss": 1.6148, "step": 1774 }, { "epoch": 0.06, "learning_rate": 1.988212193454044e-05, "loss": 1.6211, "step": 1775 }, { "epoch": 0.06, "learning_rate": 1.988195994404541e-05, "loss": 1.6248, "step": 1776 }, { "epoch": 0.06, "learning_rate": 1.9881797842982265e-05, "loss": 1.5066, "step": 1777 }, { "epoch": 0.06, "learning_rate": 1.9881635631352808e-05, "loss": 1.6312, "step": 1778 }, { "epoch": 0.06, "learning_rate": 1.9881473309158856e-05, "loss": 1.6256, "step": 1779 }, { "epoch": 0.06, "learning_rate": 1.9881310876402225e-05, "loss": 1.6138, "step": 1780 }, { "epoch": 0.06, "learning_rate": 1.9881148333084735e-05, "loss": 1.5626, "step": 1781 }, { "epoch": 0.06, "learning_rate": 1.9880985679208206e-05, "loss": 1.6555, "step": 1782 }, { "epoch": 0.06, "learning_rate": 1.9880822914774453e-05, "loss": 1.5315, "step": 1783 }, { "epoch": 0.06, "learning_rate": 1.98806600397853e-05, "loss": 1.6259, "step": 1784 }, { "epoch": 0.06, "learning_rate": 1.9880497054242566e-05, "loss": 1.5083, "step": 1785 }, { "epoch": 0.06, "learning_rate": 1.9880333958148085e-05, "loss": 1.527, "step": 1786 }, { "epoch": 0.06, "learning_rate": 1.9880170751503668e-05, "loss": 1.6651, "step": 1787 }, { "epoch": 0.06, "learning_rate": 1.9880007434311148e-05, "loss": 1.6086, "step": 1788 }, { "epoch": 0.06, "learning_rate": 1.987984400657235e-05, "loss": 1.5883, "step": 1789 }, { "epoch": 0.06, "learning_rate": 1.987968046828911e-05, "loss": 1.5732, "step": 1790 }, { "epoch": 0.06, "learning_rate": 1.987951681946325e-05, "loss": 1.5826, "step": 1791 }, { "epoch": 0.06, "learning_rate": 1.98793530600966e-05, "loss": 1.5085, "step": 1792 }, { "epoch": 0.06, "learning_rate": 1.9879189190191e-05, "loss": 1.5552, "step": 1793 }, { "epoch": 0.06, "learning_rate": 1.987902520974828e-05, "loss": 1.5526, "step": 1794 }, { "epoch": 0.06, "learning_rate": 1.987886111877027e-05, "loss": 1.544, "step": 1795 }, { "epoch": 0.06, "learning_rate": 1.987869691725881e-05, "loss": 1.5076, "step": 1796 }, { "epoch": 0.06, "learning_rate": 1.9878532605215738e-05, "loss": 1.5303, "step": 1797 }, { "epoch": 0.06, "learning_rate": 1.987836818264289e-05, "loss": 1.5458, "step": 1798 }, { "epoch": 0.06, "learning_rate": 1.987820364954211e-05, "loss": 1.5872, "step": 1799 }, { "epoch": 0.06, "learning_rate": 1.9878039005915233e-05, "loss": 1.4696, "step": 1800 }, { "epoch": 0.06, "learning_rate": 1.9877874251764108e-05, "loss": 1.5505, "step": 1801 }, { "epoch": 0.06, "learning_rate": 1.9877709387090573e-05, "loss": 1.5759, "step": 1802 }, { "epoch": 0.06, "learning_rate": 1.9877544411896474e-05, "loss": 1.678, "step": 1803 }, { "epoch": 0.06, "learning_rate": 1.9877379326183656e-05, "loss": 1.6529, "step": 1804 }, { "epoch": 0.06, "learning_rate": 1.987721412995397e-05, "loss": 1.5875, "step": 1805 }, { "epoch": 0.06, "learning_rate": 1.9877048823209262e-05, "loss": 1.5749, "step": 1806 }, { "epoch": 0.06, "learning_rate": 1.9876883405951378e-05, "loss": 1.569, "step": 1807 }, { "epoch": 0.06, "learning_rate": 1.9876717878182176e-05, "loss": 1.4581, "step": 1808 }, { "epoch": 0.06, "learning_rate": 1.9876552239903503e-05, "loss": 1.639, "step": 1809 }, { "epoch": 0.06, "learning_rate": 1.987638649111721e-05, "loss": 1.5783, "step": 1810 }, { "epoch": 0.06, "learning_rate": 1.987622063182516e-05, "loss": 1.5727, "step": 1811 }, { "epoch": 0.06, "learning_rate": 1.9876054662029203e-05, "loss": 1.6095, "step": 1812 }, { "epoch": 0.06, "learning_rate": 1.9875888581731194e-05, "loss": 1.5292, "step": 1813 }, { "epoch": 0.06, "learning_rate": 1.9875722390932997e-05, "loss": 1.5872, "step": 1814 }, { "epoch": 0.06, "learning_rate": 1.987555608963647e-05, "loss": 1.5987, "step": 1815 }, { "epoch": 0.06, "learning_rate": 1.987538967784347e-05, "loss": 1.5103, "step": 1816 }, { "epoch": 0.06, "learning_rate": 1.987522315555586e-05, "loss": 1.7136, "step": 1817 }, { "epoch": 0.06, "learning_rate": 1.9875056522775506e-05, "loss": 1.6294, "step": 1818 }, { "epoch": 0.06, "learning_rate": 1.9874889779504274e-05, "loss": 1.5777, "step": 1819 }, { "epoch": 0.06, "learning_rate": 1.987472292574402e-05, "loss": 1.6704, "step": 1820 }, { "epoch": 0.06, "learning_rate": 1.9874555961496624e-05, "loss": 1.6197, "step": 1821 }, { "epoch": 0.06, "learning_rate": 1.9874388886763944e-05, "loss": 1.5907, "step": 1822 }, { "epoch": 0.06, "learning_rate": 1.9874221701547857e-05, "loss": 1.4978, "step": 1823 }, { "epoch": 0.06, "learning_rate": 1.9874054405850223e-05, "loss": 1.556, "step": 1824 }, { "epoch": 0.06, "learning_rate": 1.9873886999672927e-05, "loss": 1.6204, "step": 1825 }, { "epoch": 0.06, "learning_rate": 1.9873719483017836e-05, "loss": 1.6142, "step": 1826 }, { "epoch": 0.06, "learning_rate": 1.9873551855886818e-05, "loss": 1.5899, "step": 1827 }, { "epoch": 0.06, "learning_rate": 1.987338411828176e-05, "loss": 1.5694, "step": 1828 }, { "epoch": 0.06, "learning_rate": 1.987321627020453e-05, "loss": 1.6616, "step": 1829 }, { "epoch": 0.06, "learning_rate": 1.9873048311657007e-05, "loss": 1.5922, "step": 1830 }, { "epoch": 0.06, "learning_rate": 1.9872880242641078e-05, "loss": 1.5649, "step": 1831 }, { "epoch": 0.06, "learning_rate": 1.9872712063158615e-05, "loss": 1.5453, "step": 1832 }, { "epoch": 0.06, "learning_rate": 1.9872543773211506e-05, "loss": 1.5903, "step": 1833 }, { "epoch": 0.06, "learning_rate": 1.9872375372801627e-05, "loss": 1.5157, "step": 1834 }, { "epoch": 0.06, "learning_rate": 1.987220686193087e-05, "loss": 1.6111, "step": 1835 }, { "epoch": 0.06, "learning_rate": 1.9872038240601114e-05, "loss": 1.5733, "step": 1836 }, { "epoch": 0.06, "learning_rate": 1.987186950881425e-05, "loss": 1.4643, "step": 1837 }, { "epoch": 0.06, "learning_rate": 1.9871700666572163e-05, "loss": 1.589, "step": 1838 }, { "epoch": 0.06, "learning_rate": 1.987153171387674e-05, "loss": 1.5513, "step": 1839 }, { "epoch": 0.06, "learning_rate": 1.987136265072988e-05, "loss": 1.617, "step": 1840 }, { "epoch": 0.06, "learning_rate": 1.9871193477133467e-05, "loss": 1.6073, "step": 1841 }, { "epoch": 0.06, "learning_rate": 1.98710241930894e-05, "loss": 1.6137, "step": 1842 }, { "epoch": 0.06, "learning_rate": 1.9870854798599563e-05, "loss": 1.7213, "step": 1843 }, { "epoch": 0.06, "learning_rate": 1.987068529366586e-05, "loss": 1.5787, "step": 1844 }, { "epoch": 0.06, "learning_rate": 1.9870515678290184e-05, "loss": 1.764, "step": 1845 }, { "epoch": 0.06, "learning_rate": 1.9870345952474436e-05, "loss": 1.6755, "step": 1846 }, { "epoch": 0.06, "learning_rate": 1.9870176116220512e-05, "loss": 1.6687, "step": 1847 }, { "epoch": 0.06, "learning_rate": 1.9870006169530315e-05, "loss": 1.6083, "step": 1848 }, { "epoch": 0.06, "learning_rate": 1.9869836112405744e-05, "loss": 1.6412, "step": 1849 }, { "epoch": 0.06, "learning_rate": 1.9869665944848703e-05, "loss": 1.552, "step": 1850 }, { "epoch": 0.06, "learning_rate": 1.9869495666861094e-05, "loss": 1.7306, "step": 1851 }, { "epoch": 0.06, "learning_rate": 1.9869325278444824e-05, "loss": 1.6682, "step": 1852 }, { "epoch": 0.06, "learning_rate": 1.98691547796018e-05, "loss": 1.6559, "step": 1853 }, { "epoch": 0.06, "learning_rate": 1.986898417033393e-05, "loss": 1.7122, "step": 1854 }, { "epoch": 0.06, "learning_rate": 1.9868813450643118e-05, "loss": 1.7514, "step": 1855 }, { "epoch": 0.06, "learning_rate": 1.986864262053128e-05, "loss": 1.7008, "step": 1856 }, { "epoch": 0.06, "learning_rate": 1.986847168000033e-05, "loss": 1.6705, "step": 1857 }, { "epoch": 0.06, "learning_rate": 1.986830062905217e-05, "loss": 1.7135, "step": 1858 }, { "epoch": 0.06, "learning_rate": 1.9868129467688723e-05, "loss": 1.6231, "step": 1859 }, { "epoch": 0.06, "learning_rate": 1.98679581959119e-05, "loss": 1.7028, "step": 1860 }, { "epoch": 0.06, "learning_rate": 1.9867786813723615e-05, "loss": 1.6731, "step": 1861 }, { "epoch": 0.06, "learning_rate": 1.9867615321125796e-05, "loss": 1.6925, "step": 1862 }, { "epoch": 0.06, "learning_rate": 1.986744371812035e-05, "loss": 1.5961, "step": 1863 }, { "epoch": 0.06, "learning_rate": 1.98672720047092e-05, "loss": 1.6556, "step": 1864 }, { "epoch": 0.06, "learning_rate": 1.9867100180894276e-05, "loss": 1.588, "step": 1865 }, { "epoch": 0.06, "learning_rate": 1.9866928246677486e-05, "loss": 1.5849, "step": 1866 }, { "epoch": 0.06, "learning_rate": 1.9866756202060764e-05, "loss": 1.4719, "step": 1867 }, { "epoch": 0.06, "learning_rate": 1.9866584047046036e-05, "loss": 1.5045, "step": 1868 }, { "epoch": 0.06, "learning_rate": 1.986641178163522e-05, "loss": 1.4766, "step": 1869 }, { "epoch": 0.06, "learning_rate": 1.986623940583025e-05, "loss": 1.5873, "step": 1870 }, { "epoch": 0.06, "learning_rate": 1.9866066919633052e-05, "loss": 1.5969, "step": 1871 }, { "epoch": 0.06, "learning_rate": 1.9865894323045558e-05, "loss": 1.6104, "step": 1872 }, { "epoch": 0.06, "learning_rate": 1.9865721616069695e-05, "loss": 1.5449, "step": 1873 }, { "epoch": 0.06, "learning_rate": 1.98655487987074e-05, "loss": 1.5799, "step": 1874 }, { "epoch": 0.06, "learning_rate": 1.9865375870960604e-05, "loss": 1.5393, "step": 1875 }, { "epoch": 0.06, "learning_rate": 1.9865202832831243e-05, "loss": 1.5184, "step": 1876 }, { "epoch": 0.06, "learning_rate": 1.9865029684321255e-05, "loss": 1.5377, "step": 1877 }, { "epoch": 0.06, "learning_rate": 1.9864856425432574e-05, "loss": 1.5818, "step": 1878 }, { "epoch": 0.06, "learning_rate": 1.9864683056167137e-05, "loss": 1.5274, "step": 1879 }, { "epoch": 0.06, "learning_rate": 1.986450957652689e-05, "loss": 1.6211, "step": 1880 }, { "epoch": 0.06, "learning_rate": 1.986433598651377e-05, "loss": 1.5563, "step": 1881 }, { "epoch": 0.06, "learning_rate": 1.986416228612972e-05, "loss": 1.4773, "step": 1882 }, { "epoch": 0.06, "learning_rate": 1.986398847537668e-05, "loss": 1.6542, "step": 1883 }, { "epoch": 0.06, "learning_rate": 1.98638145542566e-05, "loss": 1.5041, "step": 1884 }, { "epoch": 0.06, "learning_rate": 1.986364052277143e-05, "loss": 1.5133, "step": 1885 }, { "epoch": 0.06, "learning_rate": 1.9863466380923104e-05, "loss": 1.4752, "step": 1886 }, { "epoch": 0.06, "learning_rate": 1.986329212871358e-05, "loss": 1.564, "step": 1887 }, { "epoch": 0.06, "learning_rate": 1.9863117766144807e-05, "loss": 1.4981, "step": 1888 }, { "epoch": 0.06, "learning_rate": 1.986294329321873e-05, "loss": 1.6623, "step": 1889 }, { "epoch": 0.06, "learning_rate": 1.986276870993731e-05, "loss": 1.4757, "step": 1890 }, { "epoch": 0.06, "learning_rate": 1.9862594016302493e-05, "loss": 1.5604, "step": 1891 }, { "epoch": 0.06, "learning_rate": 1.986241921231624e-05, "loss": 1.4498, "step": 1892 }, { "epoch": 0.06, "learning_rate": 1.9862244297980497e-05, "loss": 1.6365, "step": 1893 }, { "epoch": 0.06, "learning_rate": 1.9862069273297233e-05, "loss": 1.5084, "step": 1894 }, { "epoch": 0.06, "learning_rate": 1.9861894138268402e-05, "loss": 1.5484, "step": 1895 }, { "epoch": 0.06, "learning_rate": 1.9861718892895955e-05, "loss": 1.5209, "step": 1896 }, { "epoch": 0.06, "learning_rate": 1.986154353718187e-05, "loss": 1.5446, "step": 1897 }, { "epoch": 0.06, "learning_rate": 1.9861368071128093e-05, "loss": 1.5362, "step": 1898 }, { "epoch": 0.06, "learning_rate": 1.9861192494736593e-05, "loss": 1.5162, "step": 1899 }, { "epoch": 0.06, "learning_rate": 1.9861016808009335e-05, "loss": 1.5404, "step": 1900 }, { "epoch": 0.06, "learning_rate": 1.9860841010948287e-05, "loss": 1.6197, "step": 1901 }, { "epoch": 0.06, "learning_rate": 1.9860665103555418e-05, "loss": 1.5312, "step": 1902 }, { "epoch": 0.06, "learning_rate": 1.9860489085832685e-05, "loss": 1.5775, "step": 1903 }, { "epoch": 0.06, "learning_rate": 1.9860312957782068e-05, "loss": 1.5274, "step": 1904 }, { "epoch": 0.06, "learning_rate": 1.986013671940553e-05, "loss": 1.5853, "step": 1905 }, { "epoch": 0.06, "learning_rate": 1.985996037070505e-05, "loss": 1.4914, "step": 1906 }, { "epoch": 0.06, "learning_rate": 1.98597839116826e-05, "loss": 1.5502, "step": 1907 }, { "epoch": 0.06, "learning_rate": 1.985960734234015e-05, "loss": 1.5248, "step": 1908 }, { "epoch": 0.06, "learning_rate": 1.9859430662679676e-05, "loss": 1.6398, "step": 1909 }, { "epoch": 0.06, "learning_rate": 1.985925387270316e-05, "loss": 1.6477, "step": 1910 }, { "epoch": 0.06, "learning_rate": 1.985907697241258e-05, "loss": 1.5275, "step": 1911 }, { "epoch": 0.06, "learning_rate": 1.9858899961809904e-05, "loss": 1.6823, "step": 1912 }, { "epoch": 0.06, "learning_rate": 1.985872284089713e-05, "loss": 1.5868, "step": 1913 }, { "epoch": 0.06, "learning_rate": 1.9858545609676226e-05, "loss": 1.6564, "step": 1914 }, { "epoch": 0.06, "learning_rate": 1.985836826814918e-05, "loss": 1.6595, "step": 1915 }, { "epoch": 0.06, "learning_rate": 1.9858190816317976e-05, "loss": 1.5932, "step": 1916 }, { "epoch": 0.06, "learning_rate": 1.9858013254184597e-05, "loss": 1.7086, "step": 1917 }, { "epoch": 0.06, "learning_rate": 1.9857835581751038e-05, "loss": 1.586, "step": 1918 }, { "epoch": 0.06, "learning_rate": 1.9857657799019275e-05, "loss": 1.5873, "step": 1919 }, { "epoch": 0.06, "learning_rate": 1.985747990599131e-05, "loss": 1.5927, "step": 1920 }, { "epoch": 0.06, "learning_rate": 1.985730190266912e-05, "loss": 1.5749, "step": 1921 }, { "epoch": 0.06, "learning_rate": 1.9857123789054707e-05, "loss": 1.6266, "step": 1922 }, { "epoch": 0.06, "learning_rate": 1.9856945565150058e-05, "loss": 1.6795, "step": 1923 }, { "epoch": 0.06, "learning_rate": 1.9856767230957173e-05, "loss": 1.5503, "step": 1924 }, { "epoch": 0.06, "learning_rate": 1.985658878647804e-05, "loss": 1.5911, "step": 1925 }, { "epoch": 0.06, "learning_rate": 1.985641023171466e-05, "loss": 1.5453, "step": 1926 }, { "epoch": 0.06, "learning_rate": 1.9856231566669036e-05, "loss": 1.5743, "step": 1927 }, { "epoch": 0.06, "learning_rate": 1.9856052791343153e-05, "loss": 1.5286, "step": 1928 }, { "epoch": 0.06, "learning_rate": 1.9855873905739026e-05, "loss": 1.5548, "step": 1929 }, { "epoch": 0.06, "learning_rate": 1.9855694909858645e-05, "loss": 1.5896, "step": 1930 }, { "epoch": 0.06, "learning_rate": 1.985551580370402e-05, "loss": 1.6637, "step": 1931 }, { "epoch": 0.06, "learning_rate": 1.9855336587277153e-05, "loss": 1.563, "step": 1932 }, { "epoch": 0.06, "learning_rate": 1.9855157260580052e-05, "loss": 1.5519, "step": 1933 }, { "epoch": 0.06, "learning_rate": 1.9854977823614718e-05, "loss": 1.5673, "step": 1934 }, { "epoch": 0.06, "learning_rate": 1.985479827638316e-05, "loss": 1.5261, "step": 1935 }, { "epoch": 0.06, "learning_rate": 1.985461861888739e-05, "loss": 1.5604, "step": 1936 }, { "epoch": 0.06, "learning_rate": 1.9854438851129414e-05, "loss": 1.5725, "step": 1937 }, { "epoch": 0.06, "learning_rate": 1.985425897311125e-05, "loss": 1.5204, "step": 1938 }, { "epoch": 0.06, "learning_rate": 1.9854078984834904e-05, "loss": 1.5047, "step": 1939 }, { "epoch": 0.06, "learning_rate": 1.985389888630239e-05, "loss": 1.5137, "step": 1940 }, { "epoch": 0.06, "learning_rate": 1.9853718677515727e-05, "loss": 1.4759, "step": 1941 }, { "epoch": 0.06, "learning_rate": 1.9853538358476933e-05, "loss": 1.5971, "step": 1942 }, { "epoch": 0.06, "learning_rate": 1.985335792918802e-05, "loss": 1.5476, "step": 1943 }, { "epoch": 0.06, "learning_rate": 1.9853177389651007e-05, "loss": 1.7202, "step": 1944 }, { "epoch": 0.06, "learning_rate": 1.9852996739867918e-05, "loss": 1.5089, "step": 1945 }, { "epoch": 0.06, "learning_rate": 1.9852815979840773e-05, "loss": 1.5753, "step": 1946 }, { "epoch": 0.06, "learning_rate": 1.9852635109571592e-05, "loss": 1.5813, "step": 1947 }, { "epoch": 0.06, "learning_rate": 1.98524541290624e-05, "loss": 1.6358, "step": 1948 }, { "epoch": 0.06, "learning_rate": 1.9852273038315223e-05, "loss": 1.586, "step": 1949 }, { "epoch": 0.07, "learning_rate": 1.985209183733209e-05, "loss": 1.6528, "step": 1950 }, { "epoch": 0.07, "learning_rate": 1.9851910526115023e-05, "loss": 1.574, "step": 1951 }, { "epoch": 0.07, "learning_rate": 1.9851729104666053e-05, "loss": 1.5359, "step": 1952 }, { "epoch": 0.07, "learning_rate": 1.985154757298721e-05, "loss": 1.6175, "step": 1953 }, { "epoch": 0.07, "learning_rate": 1.9851365931080525e-05, "loss": 1.6325, "step": 1954 }, { "epoch": 0.07, "learning_rate": 1.985118417894803e-05, "loss": 1.5266, "step": 1955 }, { "epoch": 0.07, "learning_rate": 1.985100231659176e-05, "loss": 1.6152, "step": 1956 }, { "epoch": 0.07, "learning_rate": 1.985082034401375e-05, "loss": 1.4999, "step": 1957 }, { "epoch": 0.07, "learning_rate": 1.985063826121603e-05, "loss": 1.564, "step": 1958 }, { "epoch": 0.07, "learning_rate": 1.9850456068200645e-05, "loss": 1.452, "step": 1959 }, { "epoch": 0.07, "learning_rate": 1.9850273764969632e-05, "loss": 1.5572, "step": 1960 }, { "epoch": 0.07, "learning_rate": 1.985009135152503e-05, "loss": 1.6576, "step": 1961 }, { "epoch": 0.07, "learning_rate": 1.9849908827868874e-05, "loss": 1.5003, "step": 1962 }, { "epoch": 0.07, "learning_rate": 1.9849726194003215e-05, "loss": 1.569, "step": 1963 }, { "epoch": 0.07, "learning_rate": 1.9849543449930096e-05, "loss": 1.6, "step": 1964 }, { "epoch": 0.07, "learning_rate": 1.9849360595651553e-05, "loss": 1.5967, "step": 1965 }, { "epoch": 0.07, "learning_rate": 1.9849177631169643e-05, "loss": 1.617, "step": 1966 }, { "epoch": 0.07, "learning_rate": 1.984899455648641e-05, "loss": 1.558, "step": 1967 }, { "epoch": 0.07, "learning_rate": 1.984881137160389e-05, "loss": 1.5279, "step": 1968 }, { "epoch": 0.07, "learning_rate": 1.984862807652415e-05, "loss": 1.6691, "step": 1969 }, { "epoch": 0.07, "learning_rate": 1.9848444671249235e-05, "loss": 1.6026, "step": 1970 }, { "epoch": 0.07, "learning_rate": 1.984826115578119e-05, "loss": 1.6636, "step": 1971 }, { "epoch": 0.07, "learning_rate": 1.9848077530122083e-05, "loss": 1.5439, "step": 1972 }, { "epoch": 0.07, "learning_rate": 1.9847893794273956e-05, "loss": 1.5713, "step": 1973 }, { "epoch": 0.07, "learning_rate": 1.9847709948238867e-05, "loss": 1.6689, "step": 1974 }, { "epoch": 0.07, "learning_rate": 1.9847525992018877e-05, "loss": 1.6827, "step": 1975 }, { "epoch": 0.07, "learning_rate": 1.9847341925616037e-05, "loss": 1.6784, "step": 1976 }, { "epoch": 0.07, "learning_rate": 1.984715774903242e-05, "loss": 1.6274, "step": 1977 }, { "epoch": 0.07, "learning_rate": 1.984697346227007e-05, "loss": 1.5273, "step": 1978 }, { "epoch": 0.07, "learning_rate": 1.984678906533106e-05, "loss": 1.538, "step": 1979 }, { "epoch": 0.07, "learning_rate": 1.984660455821745e-05, "loss": 1.4975, "step": 1980 }, { "epoch": 0.07, "learning_rate": 1.9846419940931305e-05, "loss": 1.424, "step": 1981 }, { "epoch": 0.07, "learning_rate": 1.9846235213474694e-05, "loss": 1.5245, "step": 1982 }, { "epoch": 0.07, "learning_rate": 1.9846050375849674e-05, "loss": 1.6123, "step": 1983 }, { "epoch": 0.07, "learning_rate": 1.9845865428058325e-05, "loss": 1.5876, "step": 1984 }, { "epoch": 0.07, "learning_rate": 1.984568037010271e-05, "loss": 1.5443, "step": 1985 }, { "epoch": 0.07, "learning_rate": 1.9845495201984898e-05, "loss": 1.5423, "step": 1986 }, { "epoch": 0.07, "learning_rate": 1.9845309923706965e-05, "loss": 1.4398, "step": 1987 }, { "epoch": 0.07, "learning_rate": 1.9845124535270978e-05, "loss": 1.4729, "step": 1988 }, { "epoch": 0.07, "learning_rate": 1.984493903667902e-05, "loss": 1.4489, "step": 1989 }, { "epoch": 0.07, "learning_rate": 1.9844753427933163e-05, "loss": 1.4621, "step": 1990 }, { "epoch": 0.07, "learning_rate": 1.984456770903548e-05, "loss": 1.4972, "step": 1991 }, { "epoch": 0.07, "learning_rate": 1.9844381879988056e-05, "loss": 1.5433, "step": 1992 }, { "epoch": 0.07, "learning_rate": 1.984419594079296e-05, "loss": 1.5548, "step": 1993 }, { "epoch": 0.07, "learning_rate": 1.984400989145228e-05, "loss": 1.5598, "step": 1994 }, { "epoch": 0.07, "learning_rate": 1.9843823731968103e-05, "loss": 1.5489, "step": 1995 }, { "epoch": 0.07, "learning_rate": 1.9843637462342498e-05, "loss": 1.6096, "step": 1996 }, { "epoch": 0.07, "learning_rate": 1.9843451082577557e-05, "loss": 1.4981, "step": 1997 }, { "epoch": 0.07, "learning_rate": 1.9843264592675365e-05, "loss": 1.5299, "step": 1998 }, { "epoch": 0.07, "learning_rate": 1.984307799263801e-05, "loss": 1.6421, "step": 1999 }, { "epoch": 0.07, "learning_rate": 1.9842891282467576e-05, "loss": 1.5746, "step": 2000 }, { "epoch": 0.07, "learning_rate": 1.9842704462166157e-05, "loss": 1.5324, "step": 2001 }, { "epoch": 0.07, "learning_rate": 1.9842517531735837e-05, "loss": 1.5964, "step": 2002 }, { "epoch": 0.07, "learning_rate": 1.9842330491178712e-05, "loss": 1.5929, "step": 2003 }, { "epoch": 0.07, "learning_rate": 1.984214334049688e-05, "loss": 1.5707, "step": 2004 }, { "epoch": 0.07, "learning_rate": 1.984195607969242e-05, "loss": 1.5788, "step": 2005 }, { "epoch": 0.07, "learning_rate": 1.984176870876744e-05, "loss": 1.7217, "step": 2006 }, { "epoch": 0.07, "learning_rate": 1.9841581227724033e-05, "loss": 1.622, "step": 2007 }, { "epoch": 0.07, "learning_rate": 1.9841393636564295e-05, "loss": 1.5957, "step": 2008 }, { "epoch": 0.07, "learning_rate": 1.9841205935290327e-05, "loss": 1.4797, "step": 2009 }, { "epoch": 0.07, "learning_rate": 1.9841018123904228e-05, "loss": 1.6357, "step": 2010 }, { "epoch": 0.07, "learning_rate": 1.98408302024081e-05, "loss": 1.5963, "step": 2011 }, { "epoch": 0.07, "learning_rate": 1.9840642170804044e-05, "loss": 1.4215, "step": 2012 }, { "epoch": 0.07, "learning_rate": 1.9840454029094167e-05, "loss": 1.4095, "step": 2013 }, { "epoch": 0.07, "learning_rate": 1.984026577728057e-05, "loss": 1.4331, "step": 2014 }, { "epoch": 0.07, "learning_rate": 1.9840077415365368e-05, "loss": 1.5752, "step": 2015 }, { "epoch": 0.07, "learning_rate": 1.9839888943350656e-05, "loss": 1.5439, "step": 2016 }, { "epoch": 0.07, "learning_rate": 1.9839700361238548e-05, "loss": 1.612, "step": 2017 }, { "epoch": 0.07, "learning_rate": 1.983951166903116e-05, "loss": 1.5571, "step": 2018 }, { "epoch": 0.07, "learning_rate": 1.9839322866730595e-05, "loss": 1.6583, "step": 2019 }, { "epoch": 0.07, "learning_rate": 1.9839133954338972e-05, "loss": 1.763, "step": 2020 }, { "epoch": 0.07, "learning_rate": 1.9838944931858395e-05, "loss": 1.6992, "step": 2021 }, { "epoch": 0.07, "learning_rate": 1.9838755799290993e-05, "loss": 1.6358, "step": 2022 }, { "epoch": 0.07, "learning_rate": 1.9838566556638872e-05, "loss": 1.7275, "step": 2023 }, { "epoch": 0.07, "learning_rate": 1.983837720390415e-05, "loss": 1.7458, "step": 2024 }, { "epoch": 0.07, "learning_rate": 1.9838187741088953e-05, "loss": 1.6311, "step": 2025 }, { "epoch": 0.07, "learning_rate": 1.983799816819539e-05, "loss": 1.583, "step": 2026 }, { "epoch": 0.07, "learning_rate": 1.983780848522559e-05, "loss": 1.709, "step": 2027 }, { "epoch": 0.07, "learning_rate": 1.9837618692181674e-05, "loss": 1.7471, "step": 2028 }, { "epoch": 0.07, "learning_rate": 1.983742878906576e-05, "loss": 1.6771, "step": 2029 }, { "epoch": 0.07, "learning_rate": 1.9837238775879982e-05, "loss": 1.7142, "step": 2030 }, { "epoch": 0.07, "learning_rate": 1.983704865262646e-05, "loss": 1.5811, "step": 2031 }, { "epoch": 0.07, "learning_rate": 1.9836858419307325e-05, "loss": 1.6678, "step": 2032 }, { "epoch": 0.07, "learning_rate": 1.9836668075924698e-05, "loss": 1.6022, "step": 2033 }, { "epoch": 0.07, "learning_rate": 1.9836477622480718e-05, "loss": 1.5886, "step": 2034 }, { "epoch": 0.07, "learning_rate": 1.983628705897751e-05, "loss": 1.5998, "step": 2035 }, { "epoch": 0.07, "learning_rate": 1.983609638541721e-05, "loss": 1.5868, "step": 2036 }, { "epoch": 0.07, "learning_rate": 1.983590560180195e-05, "loss": 1.5852, "step": 2037 }, { "epoch": 0.07, "learning_rate": 1.983571470813386e-05, "loss": 1.6456, "step": 2038 }, { "epoch": 0.07, "learning_rate": 1.9835523704415082e-05, "loss": 1.7522, "step": 2039 }, { "epoch": 0.07, "learning_rate": 1.9835332590647754e-05, "loss": 1.617, "step": 2040 }, { "epoch": 0.07, "learning_rate": 1.9835141366834006e-05, "loss": 1.4446, "step": 2041 }, { "epoch": 0.07, "learning_rate": 1.9834950032975988e-05, "loss": 1.5612, "step": 2042 }, { "epoch": 0.07, "learning_rate": 1.9834758589075836e-05, "loss": 1.599, "step": 2043 }, { "epoch": 0.07, "learning_rate": 1.983456703513569e-05, "loss": 1.5513, "step": 2044 }, { "epoch": 0.07, "learning_rate": 1.9834375371157697e-05, "loss": 1.6981, "step": 2045 }, { "epoch": 0.07, "learning_rate": 1.9834183597143995e-05, "loss": 1.6783, "step": 2046 }, { "epoch": 0.07, "learning_rate": 1.9833991713096742e-05, "loss": 1.5896, "step": 2047 }, { "epoch": 0.07, "learning_rate": 1.9833799719018073e-05, "loss": 1.5168, "step": 2048 }, { "epoch": 0.07, "learning_rate": 1.983360761491014e-05, "loss": 1.4952, "step": 2049 }, { "epoch": 0.07, "learning_rate": 1.9833415400775092e-05, "loss": 1.5412, "step": 2050 }, { "epoch": 0.07, "learning_rate": 1.9833223076615087e-05, "loss": 1.5939, "step": 2051 }, { "epoch": 0.07, "learning_rate": 1.9833030642432264e-05, "loss": 1.5774, "step": 2052 }, { "epoch": 0.07, "learning_rate": 1.9832838098228786e-05, "loss": 1.531, "step": 2053 }, { "epoch": 0.07, "learning_rate": 1.9832645444006804e-05, "loss": 1.457, "step": 2054 }, { "epoch": 0.07, "learning_rate": 1.9832452679768473e-05, "loss": 1.4704, "step": 2055 }, { "epoch": 0.07, "learning_rate": 1.9832259805515954e-05, "loss": 1.498, "step": 2056 }, { "epoch": 0.07, "learning_rate": 1.9832066821251395e-05, "loss": 1.5781, "step": 2057 }, { "epoch": 0.07, "learning_rate": 1.9831873726976966e-05, "loss": 1.577, "step": 2058 }, { "epoch": 0.07, "learning_rate": 1.9831680522694823e-05, "loss": 1.6252, "step": 2059 }, { "epoch": 0.07, "learning_rate": 1.9831487208407126e-05, "loss": 1.6133, "step": 2060 }, { "epoch": 0.07, "learning_rate": 1.9831293784116043e-05, "loss": 1.6726, "step": 2061 }, { "epoch": 0.07, "learning_rate": 1.9831100249823732e-05, "loss": 1.5414, "step": 2062 }, { "epoch": 0.07, "learning_rate": 1.9830906605532365e-05, "loss": 1.5227, "step": 2063 }, { "epoch": 0.07, "learning_rate": 1.9830712851244106e-05, "loss": 1.7105, "step": 2064 }, { "epoch": 0.07, "learning_rate": 1.983051898696112e-05, "loss": 1.6578, "step": 2065 }, { "epoch": 0.07, "learning_rate": 1.983032501268558e-05, "loss": 1.6983, "step": 2066 }, { "epoch": 0.07, "learning_rate": 1.983013092841965e-05, "loss": 1.5635, "step": 2067 }, { "epoch": 0.07, "learning_rate": 1.9829936734165512e-05, "loss": 1.5172, "step": 2068 }, { "epoch": 0.07, "learning_rate": 1.982974242992533e-05, "loss": 1.5859, "step": 2069 }, { "epoch": 0.07, "learning_rate": 1.9829548015701282e-05, "loss": 1.6181, "step": 2070 }, { "epoch": 0.07, "learning_rate": 1.9829353491495545e-05, "loss": 1.6068, "step": 2071 }, { "epoch": 0.07, "learning_rate": 1.9829158857310288e-05, "loss": 1.6453, "step": 2072 }, { "epoch": 0.07, "learning_rate": 1.98289641131477e-05, "loss": 1.7126, "step": 2073 }, { "epoch": 0.07, "learning_rate": 1.9828769259009947e-05, "loss": 1.5008, "step": 2074 }, { "epoch": 0.07, "learning_rate": 1.982857429489922e-05, "loss": 1.6506, "step": 2075 }, { "epoch": 0.07, "learning_rate": 1.98283792208177e-05, "loss": 1.649, "step": 2076 }, { "epoch": 0.07, "learning_rate": 1.9828184036767556e-05, "loss": 1.7264, "step": 2077 }, { "epoch": 0.07, "learning_rate": 1.982798874275099e-05, "loss": 1.5685, "step": 2078 }, { "epoch": 0.07, "learning_rate": 1.9827793338770174e-05, "loss": 1.5431, "step": 2079 }, { "epoch": 0.07, "learning_rate": 1.9827597824827306e-05, "loss": 1.6422, "step": 2080 }, { "epoch": 0.07, "learning_rate": 1.982740220092456e-05, "loss": 1.733, "step": 2081 }, { "epoch": 0.07, "learning_rate": 1.9827206467064133e-05, "loss": 1.6395, "step": 2082 }, { "epoch": 0.07, "learning_rate": 1.9827010623248217e-05, "loss": 1.5781, "step": 2083 }, { "epoch": 0.07, "learning_rate": 1.9826814669478998e-05, "loss": 1.5674, "step": 2084 }, { "epoch": 0.07, "learning_rate": 1.982661860575867e-05, "loss": 1.5707, "step": 2085 }, { "epoch": 0.07, "learning_rate": 1.982642243208943e-05, "loss": 1.5738, "step": 2086 }, { "epoch": 0.07, "learning_rate": 1.982622614847347e-05, "loss": 1.6421, "step": 2087 }, { "epoch": 0.07, "learning_rate": 1.9826029754912984e-05, "loss": 1.6648, "step": 2088 }, { "epoch": 0.07, "learning_rate": 1.9825833251410173e-05, "loss": 1.6814, "step": 2089 }, { "epoch": 0.07, "learning_rate": 1.9825636637967232e-05, "loss": 1.6804, "step": 2090 }, { "epoch": 0.07, "learning_rate": 1.9825439914586368e-05, "loss": 1.645, "step": 2091 }, { "epoch": 0.07, "learning_rate": 1.9825243081269778e-05, "loss": 1.5887, "step": 2092 }, { "epoch": 0.07, "learning_rate": 1.9825046138019658e-05, "loss": 1.5176, "step": 2093 }, { "epoch": 0.07, "learning_rate": 1.982484908483822e-05, "loss": 1.5511, "step": 2094 }, { "epoch": 0.07, "learning_rate": 1.982465192172767e-05, "loss": 1.5112, "step": 2095 }, { "epoch": 0.07, "learning_rate": 1.9824454648690205e-05, "loss": 1.5869, "step": 2096 }, { "epoch": 0.07, "learning_rate": 1.982425726572804e-05, "loss": 1.505, "step": 2097 }, { "epoch": 0.07, "learning_rate": 1.982405977284338e-05, "loss": 1.567, "step": 2098 }, { "epoch": 0.07, "learning_rate": 1.982386217003844e-05, "loss": 1.5993, "step": 2099 }, { "epoch": 0.07, "learning_rate": 1.9823664457315422e-05, "loss": 1.6253, "step": 2100 }, { "epoch": 0.07, "learning_rate": 1.9823466634676544e-05, "loss": 1.5684, "step": 2101 }, { "epoch": 0.07, "learning_rate": 1.982326870212402e-05, "loss": 1.5383, "step": 2102 }, { "epoch": 0.07, "learning_rate": 1.9823070659660062e-05, "loss": 1.6979, "step": 2103 }, { "epoch": 0.07, "learning_rate": 1.982287250728689e-05, "loss": 1.5595, "step": 2104 }, { "epoch": 0.07, "learning_rate": 1.9822674245006714e-05, "loss": 1.5399, "step": 2105 }, { "epoch": 0.07, "learning_rate": 1.9822475872821757e-05, "loss": 1.5675, "step": 2106 }, { "epoch": 0.07, "learning_rate": 1.982227739073424e-05, "loss": 1.6915, "step": 2107 }, { "epoch": 0.07, "learning_rate": 1.9822078798746378e-05, "loss": 1.6844, "step": 2108 }, { "epoch": 0.07, "learning_rate": 1.98218800968604e-05, "loss": 1.6282, "step": 2109 }, { "epoch": 0.07, "learning_rate": 1.9821681285078522e-05, "loss": 1.5642, "step": 2110 }, { "epoch": 0.07, "learning_rate": 1.9821482363402976e-05, "loss": 1.6383, "step": 2111 }, { "epoch": 0.07, "learning_rate": 1.9821283331835982e-05, "loss": 1.5588, "step": 2112 }, { "epoch": 0.07, "learning_rate": 1.982108419037977e-05, "loss": 1.7664, "step": 2113 }, { "epoch": 0.07, "learning_rate": 1.982088493903657e-05, "loss": 1.7595, "step": 2114 }, { "epoch": 0.07, "learning_rate": 1.9820685577808604e-05, "loss": 1.674, "step": 2115 }, { "epoch": 0.07, "learning_rate": 1.982048610669811e-05, "loss": 1.6518, "step": 2116 }, { "epoch": 0.07, "learning_rate": 1.9820286525707318e-05, "loss": 1.7174, "step": 2117 }, { "epoch": 0.07, "learning_rate": 1.9820086834838456e-05, "loss": 1.6041, "step": 2118 }, { "epoch": 0.07, "learning_rate": 1.9819887034093768e-05, "loss": 1.581, "step": 2119 }, { "epoch": 0.07, "learning_rate": 1.981968712347548e-05, "loss": 1.6573, "step": 2120 }, { "epoch": 0.07, "learning_rate": 1.9819487102985833e-05, "loss": 1.6421, "step": 2121 }, { "epoch": 0.07, "learning_rate": 1.9819286972627066e-05, "loss": 1.5734, "step": 2122 }, { "epoch": 0.07, "learning_rate": 1.981908673240142e-05, "loss": 1.6532, "step": 2123 }, { "epoch": 0.07, "learning_rate": 1.9818886382311132e-05, "loss": 1.5475, "step": 2124 }, { "epoch": 0.07, "learning_rate": 1.981868592235844e-05, "loss": 1.6368, "step": 2125 }, { "epoch": 0.07, "learning_rate": 1.9818485352545595e-05, "loss": 1.6759, "step": 2126 }, { "epoch": 0.07, "learning_rate": 1.9818284672874834e-05, "loss": 1.6224, "step": 2127 }, { "epoch": 0.07, "learning_rate": 1.981808388334841e-05, "loss": 1.6835, "step": 2128 }, { "epoch": 0.07, "learning_rate": 1.9817882983968562e-05, "loss": 1.6184, "step": 2129 }, { "epoch": 0.07, "learning_rate": 1.9817681974737543e-05, "loss": 1.5894, "step": 2130 }, { "epoch": 0.07, "learning_rate": 1.98174808556576e-05, "loss": 1.6003, "step": 2131 }, { "epoch": 0.07, "learning_rate": 1.981727962673098e-05, "loss": 1.5621, "step": 2132 }, { "epoch": 0.07, "learning_rate": 1.9817078287959943e-05, "loss": 1.5028, "step": 2133 }, { "epoch": 0.07, "learning_rate": 1.9816876839346734e-05, "loss": 1.5442, "step": 2134 }, { "epoch": 0.07, "learning_rate": 1.9816675280893613e-05, "loss": 1.5454, "step": 2135 }, { "epoch": 0.07, "learning_rate": 1.981647361260283e-05, "loss": 1.5754, "step": 2136 }, { "epoch": 0.07, "learning_rate": 1.9816271834476642e-05, "loss": 1.6094, "step": 2137 }, { "epoch": 0.07, "learning_rate": 1.981606994651731e-05, "loss": 1.5383, "step": 2138 }, { "epoch": 0.07, "learning_rate": 1.9815867948727087e-05, "loss": 1.5307, "step": 2139 }, { "epoch": 0.07, "learning_rate": 1.981566584110824e-05, "loss": 1.5252, "step": 2140 }, { "epoch": 0.07, "learning_rate": 1.9815463623663033e-05, "loss": 1.5154, "step": 2141 }, { "epoch": 0.07, "learning_rate": 1.9815261296393715e-05, "loss": 1.6688, "step": 2142 }, { "epoch": 0.07, "learning_rate": 1.9815058859302563e-05, "loss": 1.6707, "step": 2143 }, { "epoch": 0.07, "learning_rate": 1.9814856312391835e-05, "loss": 1.6247, "step": 2144 }, { "epoch": 0.07, "learning_rate": 1.98146536556638e-05, "loss": 1.5474, "step": 2145 }, { "epoch": 0.07, "learning_rate": 1.9814450889120725e-05, "loss": 1.588, "step": 2146 }, { "epoch": 0.07, "learning_rate": 1.9814248012764877e-05, "loss": 1.4247, "step": 2147 }, { "epoch": 0.07, "learning_rate": 1.981404502659853e-05, "loss": 1.5755, "step": 2148 }, { "epoch": 0.07, "learning_rate": 1.981384193062395e-05, "loss": 1.6317, "step": 2149 }, { "epoch": 0.07, "learning_rate": 1.9813638724843413e-05, "loss": 1.6276, "step": 2150 }, { "epoch": 0.07, "learning_rate": 1.9813435409259193e-05, "loss": 1.628, "step": 2151 }, { "epoch": 0.07, "learning_rate": 1.9813231983873563e-05, "loss": 1.561, "step": 2152 }, { "epoch": 0.07, "learning_rate": 1.9813028448688797e-05, "loss": 1.5105, "step": 2153 }, { "epoch": 0.07, "learning_rate": 1.9812824803707178e-05, "loss": 1.5298, "step": 2154 }, { "epoch": 0.07, "learning_rate": 1.981262104893098e-05, "loss": 1.569, "step": 2155 }, { "epoch": 0.07, "learning_rate": 1.9812417184362488e-05, "loss": 1.6191, "step": 2156 }, { "epoch": 0.07, "learning_rate": 1.9812213210003976e-05, "loss": 1.5122, "step": 2157 }, { "epoch": 0.07, "learning_rate": 1.981200912585773e-05, "loss": 1.5537, "step": 2158 }, { "epoch": 0.07, "learning_rate": 1.981180493192603e-05, "loss": 1.4382, "step": 2159 }, { "epoch": 0.07, "learning_rate": 1.981160062821117e-05, "loss": 1.4781, "step": 2160 }, { "epoch": 0.07, "learning_rate": 1.981139621471543e-05, "loss": 1.5349, "step": 2161 }, { "epoch": 0.07, "learning_rate": 1.9811191691441092e-05, "loss": 1.5671, "step": 2162 }, { "epoch": 0.07, "learning_rate": 1.981098705839045e-05, "loss": 1.5135, "step": 2163 }, { "epoch": 0.07, "learning_rate": 1.9810782315565795e-05, "loss": 1.5141, "step": 2164 }, { "epoch": 0.07, "learning_rate": 1.9810577462969415e-05, "loss": 1.4713, "step": 2165 }, { "epoch": 0.07, "learning_rate": 1.9810372500603603e-05, "loss": 1.53, "step": 2166 }, { "epoch": 0.07, "learning_rate": 1.9810167428470653e-05, "loss": 1.5409, "step": 2167 }, { "epoch": 0.07, "learning_rate": 1.9809962246572857e-05, "loss": 1.6292, "step": 2168 }, { "epoch": 0.07, "learning_rate": 1.9809756954912514e-05, "loss": 1.6233, "step": 2169 }, { "epoch": 0.07, "learning_rate": 1.9809551553491918e-05, "loss": 1.5096, "step": 2170 }, { "epoch": 0.07, "learning_rate": 1.9809346042313368e-05, "loss": 1.4446, "step": 2171 }, { "epoch": 0.07, "learning_rate": 1.9809140421379168e-05, "loss": 1.6195, "step": 2172 }, { "epoch": 0.07, "learning_rate": 1.980893469069161e-05, "loss": 1.5614, "step": 2173 }, { "epoch": 0.07, "learning_rate": 1.9808728850253e-05, "loss": 1.6611, "step": 2174 }, { "epoch": 0.07, "learning_rate": 1.9808522900065648e-05, "loss": 1.5425, "step": 2175 }, { "epoch": 0.07, "learning_rate": 1.9808316840131846e-05, "loss": 1.5273, "step": 2176 }, { "epoch": 0.07, "learning_rate": 1.9808110670453908e-05, "loss": 1.5695, "step": 2177 }, { "epoch": 0.07, "learning_rate": 1.980790439103414e-05, "loss": 1.5708, "step": 2178 }, { "epoch": 0.07, "learning_rate": 1.9807698001874848e-05, "loss": 1.5305, "step": 2179 }, { "epoch": 0.07, "learning_rate": 1.9807491502978337e-05, "loss": 1.5267, "step": 2180 }, { "epoch": 0.07, "learning_rate": 1.980728489434693e-05, "loss": 1.6482, "step": 2181 }, { "epoch": 0.07, "learning_rate": 1.9807078175982925e-05, "loss": 1.7131, "step": 2182 }, { "epoch": 0.07, "learning_rate": 1.9806871347888645e-05, "loss": 1.553, "step": 2183 }, { "epoch": 0.07, "learning_rate": 1.9806664410066396e-05, "loss": 1.5952, "step": 2184 }, { "epoch": 0.07, "learning_rate": 1.98064573625185e-05, "loss": 1.5982, "step": 2185 }, { "epoch": 0.07, "learning_rate": 1.980625020524727e-05, "loss": 1.5945, "step": 2186 }, { "epoch": 0.07, "learning_rate": 1.980604293825503e-05, "loss": 1.6909, "step": 2187 }, { "epoch": 0.07, "learning_rate": 1.9805835561544086e-05, "loss": 1.6335, "step": 2188 }, { "epoch": 0.07, "learning_rate": 1.980562807511677e-05, "loss": 1.7555, "step": 2189 }, { "epoch": 0.07, "learning_rate": 1.9805420478975403e-05, "loss": 1.7077, "step": 2190 }, { "epoch": 0.07, "learning_rate": 1.9805212773122303e-05, "loss": 1.6637, "step": 2191 }, { "epoch": 0.07, "learning_rate": 1.9805004957559795e-05, "loss": 1.7854, "step": 2192 }, { "epoch": 0.07, "learning_rate": 1.9804797032290203e-05, "loss": 1.616, "step": 2193 }, { "epoch": 0.07, "learning_rate": 1.9804588997315858e-05, "loss": 1.6914, "step": 2194 }, { "epoch": 0.07, "learning_rate": 1.9804380852639084e-05, "loss": 1.5988, "step": 2195 }, { "epoch": 0.07, "learning_rate": 1.9804172598262213e-05, "loss": 1.7199, "step": 2196 }, { "epoch": 0.07, "learning_rate": 1.980396423418757e-05, "loss": 1.6723, "step": 2197 }, { "epoch": 0.07, "learning_rate": 1.9803755760417495e-05, "loss": 1.6651, "step": 2198 }, { "epoch": 0.07, "learning_rate": 1.980354717695431e-05, "loss": 1.5651, "step": 2199 }, { "epoch": 0.07, "learning_rate": 1.9803338483800355e-05, "loss": 1.7179, "step": 2200 }, { "epoch": 0.07, "learning_rate": 1.980312968095796e-05, "loss": 1.6333, "step": 2201 }, { "epoch": 0.07, "learning_rate": 1.9802920768429473e-05, "loss": 1.7317, "step": 2202 }, { "epoch": 0.07, "learning_rate": 1.9802711746217222e-05, "loss": 1.7432, "step": 2203 }, { "epoch": 0.07, "learning_rate": 1.9802502614323545e-05, "loss": 1.7317, "step": 2204 }, { "epoch": 0.07, "learning_rate": 1.9802293372750783e-05, "loss": 1.7219, "step": 2205 }, { "epoch": 0.07, "learning_rate": 1.9802084021501282e-05, "loss": 1.628, "step": 2206 }, { "epoch": 0.07, "learning_rate": 1.980187456057738e-05, "loss": 1.6401, "step": 2207 }, { "epoch": 0.07, "learning_rate": 1.980166498998142e-05, "loss": 1.6075, "step": 2208 }, { "epoch": 0.07, "learning_rate": 1.9801455309715748e-05, "loss": 1.6609, "step": 2209 }, { "epoch": 0.07, "learning_rate": 1.9801245519782713e-05, "loss": 1.6574, "step": 2210 }, { "epoch": 0.07, "learning_rate": 1.9801035620184657e-05, "loss": 1.7194, "step": 2211 }, { "epoch": 0.07, "learning_rate": 1.9800825610923937e-05, "loss": 1.6918, "step": 2212 }, { "epoch": 0.07, "learning_rate": 1.980061549200289e-05, "loss": 1.6444, "step": 2213 }, { "epoch": 0.07, "learning_rate": 1.980040526342388e-05, "loss": 1.5197, "step": 2214 }, { "epoch": 0.07, "learning_rate": 1.980019492518925e-05, "loss": 1.684, "step": 2215 }, { "epoch": 0.07, "learning_rate": 1.9799984477301358e-05, "loss": 1.677, "step": 2216 }, { "epoch": 0.07, "learning_rate": 1.979977391976256e-05, "loss": 1.6071, "step": 2217 }, { "epoch": 0.07, "learning_rate": 1.9799563252575208e-05, "loss": 1.6023, "step": 2218 }, { "epoch": 0.07, "learning_rate": 1.9799352475741656e-05, "loss": 1.603, "step": 2219 }, { "epoch": 0.07, "learning_rate": 1.979914158926427e-05, "loss": 1.5467, "step": 2220 }, { "epoch": 0.07, "learning_rate": 1.9798930593145412e-05, "loss": 1.5874, "step": 2221 }, { "epoch": 0.07, "learning_rate": 1.979871948738743e-05, "loss": 1.5164, "step": 2222 }, { "epoch": 0.07, "learning_rate": 1.9798508271992697e-05, "loss": 1.609, "step": 2223 }, { "epoch": 0.07, "learning_rate": 1.9798296946963572e-05, "loss": 1.6028, "step": 2224 }, { "epoch": 0.07, "learning_rate": 1.9798085512302418e-05, "loss": 1.4893, "step": 2225 }, { "epoch": 0.07, "learning_rate": 1.9797873968011605e-05, "loss": 1.6292, "step": 2226 }, { "epoch": 0.07, "learning_rate": 1.9797662314093496e-05, "loss": 1.6416, "step": 2227 }, { "epoch": 0.07, "learning_rate": 1.9797450550550465e-05, "loss": 1.5398, "step": 2228 }, { "epoch": 0.07, "learning_rate": 1.9797238677384874e-05, "loss": 1.6501, "step": 2229 }, { "epoch": 0.07, "learning_rate": 1.9797026694599097e-05, "loss": 1.6482, "step": 2230 }, { "epoch": 0.07, "learning_rate": 1.9796814602195507e-05, "loss": 1.594, "step": 2231 }, { "epoch": 0.07, "learning_rate": 1.9796602400176476e-05, "loss": 1.5221, "step": 2232 }, { "epoch": 0.07, "learning_rate": 1.9796390088544377e-05, "loss": 1.5614, "step": 2233 }, { "epoch": 0.07, "learning_rate": 1.9796177667301588e-05, "loss": 1.5541, "step": 2234 }, { "epoch": 0.07, "learning_rate": 1.9795965136450485e-05, "loss": 1.5142, "step": 2235 }, { "epoch": 0.07, "learning_rate": 1.979575249599344e-05, "loss": 1.5858, "step": 2236 }, { "epoch": 0.07, "learning_rate": 1.9795539745932847e-05, "loss": 1.6925, "step": 2237 }, { "epoch": 0.07, "learning_rate": 1.979532688627107e-05, "loss": 1.6158, "step": 2238 }, { "epoch": 0.07, "learning_rate": 1.97951139170105e-05, "loss": 1.6511, "step": 2239 }, { "epoch": 0.07, "learning_rate": 1.9794900838153517e-05, "loss": 1.6544, "step": 2240 }, { "epoch": 0.07, "learning_rate": 1.9794687649702507e-05, "loss": 1.6406, "step": 2241 }, { "epoch": 0.07, "learning_rate": 1.9794474351659854e-05, "loss": 1.6409, "step": 2242 }, { "epoch": 0.07, "learning_rate": 1.9794260944027942e-05, "loss": 1.555, "step": 2243 }, { "epoch": 0.07, "learning_rate": 1.9794047426809168e-05, "loss": 1.5794, "step": 2244 }, { "epoch": 0.07, "learning_rate": 1.979383380000591e-05, "loss": 1.5482, "step": 2245 }, { "epoch": 0.07, "learning_rate": 1.979362006362056e-05, "loss": 1.5436, "step": 2246 }, { "epoch": 0.07, "learning_rate": 1.9793406217655516e-05, "loss": 1.5186, "step": 2247 }, { "epoch": 0.07, "learning_rate": 1.9793192262113167e-05, "loss": 1.593, "step": 2248 }, { "epoch": 0.07, "learning_rate": 1.9792978196995908e-05, "loss": 1.65, "step": 2249 }, { "epoch": 0.07, "learning_rate": 1.979276402230613e-05, "loss": 1.5669, "step": 2250 }, { "epoch": 0.08, "learning_rate": 1.9792549738046232e-05, "loss": 1.6316, "step": 2251 }, { "epoch": 0.08, "learning_rate": 1.9792335344218614e-05, "loss": 1.6028, "step": 2252 }, { "epoch": 0.08, "learning_rate": 1.9792120840825674e-05, "loss": 1.7098, "step": 2253 }, { "epoch": 0.08, "learning_rate": 1.979190622786981e-05, "loss": 1.5955, "step": 2254 }, { "epoch": 0.08, "learning_rate": 1.9791691505353423e-05, "loss": 1.5669, "step": 2255 }, { "epoch": 0.08, "learning_rate": 1.9791476673278917e-05, "loss": 1.5598, "step": 2256 }, { "epoch": 0.08, "learning_rate": 1.9791261731648694e-05, "loss": 1.5666, "step": 2257 }, { "epoch": 0.08, "learning_rate": 1.979104668046516e-05, "loss": 1.5324, "step": 2258 }, { "epoch": 0.08, "learning_rate": 1.9790831519730728e-05, "loss": 1.5335, "step": 2259 }, { "epoch": 0.08, "learning_rate": 1.9790616249447795e-05, "loss": 1.528, "step": 2260 }, { "epoch": 0.08, "learning_rate": 1.9790400869618772e-05, "loss": 1.6957, "step": 2261 }, { "epoch": 0.08, "learning_rate": 1.979018538024607e-05, "loss": 1.5334, "step": 2262 }, { "epoch": 0.08, "learning_rate": 1.9789969781332102e-05, "loss": 1.5139, "step": 2263 }, { "epoch": 0.08, "learning_rate": 1.978975407287928e-05, "loss": 1.6272, "step": 2264 }, { "epoch": 0.08, "learning_rate": 1.9789538254890017e-05, "loss": 1.6155, "step": 2265 }, { "epoch": 0.08, "learning_rate": 1.9789322327366722e-05, "loss": 1.5228, "step": 2266 }, { "epoch": 0.08, "learning_rate": 1.9789106290311818e-05, "loss": 1.6546, "step": 2267 }, { "epoch": 0.08, "learning_rate": 1.9788890143727724e-05, "loss": 1.6999, "step": 2268 }, { "epoch": 0.08, "learning_rate": 1.9788673887616852e-05, "loss": 1.5995, "step": 2269 }, { "epoch": 0.08, "learning_rate": 1.9788457521981622e-05, "loss": 1.5561, "step": 2270 }, { "epoch": 0.08, "learning_rate": 1.9788241046824463e-05, "loss": 1.6411, "step": 2271 }, { "epoch": 0.08, "learning_rate": 1.978802446214779e-05, "loss": 1.5858, "step": 2272 }, { "epoch": 0.08, "learning_rate": 1.9787807767954025e-05, "loss": 1.5436, "step": 2273 }, { "epoch": 0.08, "learning_rate": 1.9787590964245596e-05, "loss": 1.5254, "step": 2274 }, { "epoch": 0.08, "learning_rate": 1.978737405102493e-05, "loss": 1.5808, "step": 2275 }, { "epoch": 0.08, "learning_rate": 1.9787157028294453e-05, "loss": 1.6005, "step": 2276 }, { "epoch": 0.08, "learning_rate": 1.978693989605659e-05, "loss": 1.6726, "step": 2277 }, { "epoch": 0.08, "learning_rate": 1.9786722654313773e-05, "loss": 1.6388, "step": 2278 }, { "epoch": 0.08, "learning_rate": 1.9786505303068434e-05, "loss": 1.7068, "step": 2279 }, { "epoch": 0.08, "learning_rate": 1.9786287842323002e-05, "loss": 1.7314, "step": 2280 }, { "epoch": 0.08, "learning_rate": 1.978607027207991e-05, "loss": 1.6604, "step": 2281 }, { "epoch": 0.08, "learning_rate": 1.97858525923416e-05, "loss": 1.4522, "step": 2282 }, { "epoch": 0.08, "learning_rate": 1.9785634803110496e-05, "loss": 1.6399, "step": 2283 }, { "epoch": 0.08, "learning_rate": 1.9785416904389044e-05, "loss": 1.6326, "step": 2284 }, { "epoch": 0.08, "learning_rate": 1.9785198896179675e-05, "loss": 1.6564, "step": 2285 }, { "epoch": 0.08, "learning_rate": 1.9784980778484834e-05, "loss": 1.57, "step": 2286 }, { "epoch": 0.08, "learning_rate": 1.978476255130696e-05, "loss": 1.5415, "step": 2287 }, { "epoch": 0.08, "learning_rate": 1.978454421464849e-05, "loss": 1.6425, "step": 2288 }, { "epoch": 0.08, "learning_rate": 1.9784325768511873e-05, "loss": 1.715, "step": 2289 }, { "epoch": 0.08, "learning_rate": 1.978410721289955e-05, "loss": 1.6474, "step": 2290 }, { "epoch": 0.08, "learning_rate": 1.978388854781397e-05, "loss": 1.5282, "step": 2291 }, { "epoch": 0.08, "learning_rate": 1.9783669773257577e-05, "loss": 1.5074, "step": 2292 }, { "epoch": 0.08, "learning_rate": 1.9783450889232818e-05, "loss": 1.5371, "step": 2293 }, { "epoch": 0.08, "learning_rate": 1.9783231895742145e-05, "loss": 1.6437, "step": 2294 }, { "epoch": 0.08, "learning_rate": 1.9783012792788005e-05, "loss": 1.5923, "step": 2295 }, { "epoch": 0.08, "learning_rate": 1.9782793580372848e-05, "loss": 1.6799, "step": 2296 }, { "epoch": 0.08, "learning_rate": 1.978257425849913e-05, "loss": 1.621, "step": 2297 }, { "epoch": 0.08, "learning_rate": 1.978235482716931e-05, "loss": 1.6641, "step": 2298 }, { "epoch": 0.08, "learning_rate": 1.978213528638583e-05, "loss": 1.6943, "step": 2299 }, { "epoch": 0.08, "learning_rate": 1.9781915636151158e-05, "loss": 1.5333, "step": 2300 }, { "epoch": 0.08, "learning_rate": 1.978169587646775e-05, "loss": 1.5273, "step": 2301 }, { "epoch": 0.08, "learning_rate": 1.9781476007338058e-05, "loss": 1.5461, "step": 2302 }, { "epoch": 0.08, "learning_rate": 1.9781256028764547e-05, "loss": 1.5288, "step": 2303 }, { "epoch": 0.08, "learning_rate": 1.9781035940749684e-05, "loss": 1.6302, "step": 2304 }, { "epoch": 0.08, "learning_rate": 1.978081574329592e-05, "loss": 1.5768, "step": 2305 }, { "epoch": 0.08, "learning_rate": 1.9780595436405724e-05, "loss": 1.6802, "step": 2306 }, { "epoch": 0.08, "learning_rate": 1.9780375020081565e-05, "loss": 1.581, "step": 2307 }, { "epoch": 0.08, "learning_rate": 1.9780154494325902e-05, "loss": 1.5859, "step": 2308 }, { "epoch": 0.08, "learning_rate": 1.977993385914121e-05, "loss": 1.5308, "step": 2309 }, { "epoch": 0.08, "learning_rate": 1.977971311452995e-05, "loss": 1.5296, "step": 2310 }, { "epoch": 0.08, "learning_rate": 1.9779492260494596e-05, "loss": 1.5772, "step": 2311 }, { "epoch": 0.08, "learning_rate": 1.977927129703762e-05, "loss": 1.4934, "step": 2312 }, { "epoch": 0.08, "learning_rate": 1.9779050224161494e-05, "loss": 1.523, "step": 2313 }, { "epoch": 0.08, "learning_rate": 1.9778829041868687e-05, "loss": 1.5192, "step": 2314 }, { "epoch": 0.08, "learning_rate": 1.9778607750161682e-05, "loss": 1.5864, "step": 2315 }, { "epoch": 0.08, "learning_rate": 1.9778386349042943e-05, "loss": 1.6169, "step": 2316 }, { "epoch": 0.08, "learning_rate": 1.977816483851496e-05, "loss": 1.534, "step": 2317 }, { "epoch": 0.08, "learning_rate": 1.9777943218580206e-05, "loss": 1.5885, "step": 2318 }, { "epoch": 0.08, "learning_rate": 1.9777721489241158e-05, "loss": 1.604, "step": 2319 }, { "epoch": 0.08, "learning_rate": 1.9777499650500303e-05, "loss": 1.555, "step": 2320 }, { "epoch": 0.08, "learning_rate": 1.9777277702360118e-05, "loss": 1.6179, "step": 2321 }, { "epoch": 0.08, "learning_rate": 1.9777055644823087e-05, "loss": 1.5049, "step": 2322 }, { "epoch": 0.08, "learning_rate": 1.9776833477891696e-05, "loss": 1.5799, "step": 2323 }, { "epoch": 0.08, "learning_rate": 1.9776611201568434e-05, "loss": 1.57, "step": 2324 }, { "epoch": 0.08, "learning_rate": 1.977638881585578e-05, "loss": 1.4597, "step": 2325 }, { "epoch": 0.08, "learning_rate": 1.9776166320756228e-05, "loss": 1.5483, "step": 2326 }, { "epoch": 0.08, "learning_rate": 1.9775943716272266e-05, "loss": 1.5012, "step": 2327 }, { "epoch": 0.08, "learning_rate": 1.9775721002406385e-05, "loss": 1.5816, "step": 2328 }, { "epoch": 0.08, "learning_rate": 1.9775498179161077e-05, "loss": 1.5024, "step": 2329 }, { "epoch": 0.08, "learning_rate": 1.9775275246538834e-05, "loss": 1.5092, "step": 2330 }, { "epoch": 0.08, "learning_rate": 1.9775052204542155e-05, "loss": 1.5016, "step": 2331 }, { "epoch": 0.08, "learning_rate": 1.977482905317353e-05, "loss": 1.5363, "step": 2332 }, { "epoch": 0.08, "learning_rate": 1.9774605792435456e-05, "loss": 1.5752, "step": 2333 }, { "epoch": 0.08, "learning_rate": 1.9774382422330433e-05, "loss": 1.5563, "step": 2334 }, { "epoch": 0.08, "learning_rate": 1.9774158942860962e-05, "loss": 1.5886, "step": 2335 }, { "epoch": 0.08, "learning_rate": 1.977393535402954e-05, "loss": 1.5133, "step": 2336 }, { "epoch": 0.08, "learning_rate": 1.977371165583867e-05, "loss": 1.548, "step": 2337 }, { "epoch": 0.08, "learning_rate": 1.9773487848290856e-05, "loss": 1.6316, "step": 2338 }, { "epoch": 0.08, "learning_rate": 1.97732639313886e-05, "loss": 1.6268, "step": 2339 }, { "epoch": 0.08, "learning_rate": 1.977303990513441e-05, "loss": 1.5623, "step": 2340 }, { "epoch": 0.08, "learning_rate": 1.977281576953079e-05, "loss": 1.5226, "step": 2341 }, { "epoch": 0.08, "learning_rate": 1.977259152458025e-05, "loss": 1.5546, "step": 2342 }, { "epoch": 0.08, "learning_rate": 1.97723671702853e-05, "loss": 1.5743, "step": 2343 }, { "epoch": 0.08, "learning_rate": 1.9772142706648446e-05, "loss": 1.5375, "step": 2344 }, { "epoch": 0.08, "learning_rate": 1.9771918133672203e-05, "loss": 1.5042, "step": 2345 }, { "epoch": 0.08, "learning_rate": 1.977169345135908e-05, "loss": 1.6308, "step": 2346 }, { "epoch": 0.08, "learning_rate": 1.9771468659711595e-05, "loss": 1.5564, "step": 2347 }, { "epoch": 0.08, "learning_rate": 1.9771243758732265e-05, "loss": 1.4983, "step": 2348 }, { "epoch": 0.08, "learning_rate": 1.97710187484236e-05, "loss": 1.5784, "step": 2349 }, { "epoch": 0.08, "learning_rate": 1.9770793628788123e-05, "loss": 1.5526, "step": 2350 }, { "epoch": 0.08, "learning_rate": 1.9770568399828346e-05, "loss": 1.6642, "step": 2351 }, { "epoch": 0.08, "learning_rate": 1.9770343061546797e-05, "loss": 1.6817, "step": 2352 }, { "epoch": 0.08, "learning_rate": 1.9770117613945996e-05, "loss": 1.6838, "step": 2353 }, { "epoch": 0.08, "learning_rate": 1.976989205702846e-05, "loss": 1.6669, "step": 2354 }, { "epoch": 0.08, "learning_rate": 1.976966639079672e-05, "loss": 1.6972, "step": 2355 }, { "epoch": 0.08, "learning_rate": 1.9769440615253295e-05, "loss": 1.7048, "step": 2356 }, { "epoch": 0.08, "learning_rate": 1.976921473040071e-05, "loss": 1.6914, "step": 2357 }, { "epoch": 0.08, "learning_rate": 1.97689887362415e-05, "loss": 1.6367, "step": 2358 }, { "epoch": 0.08, "learning_rate": 1.9768762632778188e-05, "loss": 1.628, "step": 2359 }, { "epoch": 0.08, "learning_rate": 1.9768536420013305e-05, "loss": 1.7888, "step": 2360 }, { "epoch": 0.08, "learning_rate": 1.9768310097949385e-05, "loss": 1.7657, "step": 2361 }, { "epoch": 0.08, "learning_rate": 1.9768083666588954e-05, "loss": 1.6889, "step": 2362 }, { "epoch": 0.08, "learning_rate": 1.976785712593455e-05, "loss": 1.6826, "step": 2363 }, { "epoch": 0.08, "learning_rate": 1.9767630475988707e-05, "loss": 1.7045, "step": 2364 }, { "epoch": 0.08, "learning_rate": 1.976740371675396e-05, "loss": 1.8096, "step": 2365 }, { "epoch": 0.08, "learning_rate": 1.9767176848232848e-05, "loss": 1.7141, "step": 2366 }, { "epoch": 0.08, "learning_rate": 1.976694987042791e-05, "loss": 1.738, "step": 2367 }, { "epoch": 0.08, "learning_rate": 1.9766722783341682e-05, "loss": 1.6805, "step": 2368 }, { "epoch": 0.08, "learning_rate": 1.9766495586976706e-05, "loss": 1.684, "step": 2369 }, { "epoch": 0.08, "learning_rate": 1.9766268281335525e-05, "loss": 1.6803, "step": 2370 }, { "epoch": 0.08, "learning_rate": 1.9766040866420684e-05, "loss": 1.6445, "step": 2371 }, { "epoch": 0.08, "learning_rate": 1.9765813342234726e-05, "loss": 1.6562, "step": 2372 }, { "epoch": 0.08, "learning_rate": 1.9765585708780194e-05, "loss": 1.6892, "step": 2373 }, { "epoch": 0.08, "learning_rate": 1.9765357966059638e-05, "loss": 1.6594, "step": 2374 }, { "epoch": 0.08, "learning_rate": 1.9765130114075606e-05, "loss": 1.6594, "step": 2375 }, { "epoch": 0.08, "learning_rate": 1.976490215283065e-05, "loss": 1.6426, "step": 2376 }, { "epoch": 0.08, "learning_rate": 1.9764674082327313e-05, "loss": 1.5749, "step": 2377 }, { "epoch": 0.08, "learning_rate": 1.9764445902568154e-05, "loss": 1.6318, "step": 2378 }, { "epoch": 0.08, "learning_rate": 1.976421761355572e-05, "loss": 1.6258, "step": 2379 }, { "epoch": 0.08, "learning_rate": 1.9763989215292576e-05, "loss": 1.6167, "step": 2380 }, { "epoch": 0.08, "learning_rate": 1.976376070778127e-05, "loss": 1.5936, "step": 2381 }, { "epoch": 0.08, "learning_rate": 1.976353209102435e-05, "loss": 1.5341, "step": 2382 }, { "epoch": 0.08, "learning_rate": 1.9763303365024392e-05, "loss": 1.574, "step": 2383 }, { "epoch": 0.08, "learning_rate": 1.9763074529783943e-05, "loss": 1.5288, "step": 2384 }, { "epoch": 0.08, "learning_rate": 1.9762845585305567e-05, "loss": 1.5722, "step": 2385 }, { "epoch": 0.08, "learning_rate": 1.9762616531591826e-05, "loss": 1.5782, "step": 2386 }, { "epoch": 0.08, "learning_rate": 1.9762387368645283e-05, "loss": 1.5573, "step": 2387 }, { "epoch": 0.08, "learning_rate": 1.9762158096468498e-05, "loss": 1.6159, "step": 2388 }, { "epoch": 0.08, "learning_rate": 1.976192871506404e-05, "loss": 1.5745, "step": 2389 }, { "epoch": 0.08, "learning_rate": 1.9761699224434476e-05, "loss": 1.5384, "step": 2390 }, { "epoch": 0.08, "learning_rate": 1.9761469624582375e-05, "loss": 1.5793, "step": 2391 }, { "epoch": 0.08, "learning_rate": 1.9761239915510302e-05, "loss": 1.6042, "step": 2392 }, { "epoch": 0.08, "learning_rate": 1.976101009722083e-05, "loss": 1.5891, "step": 2393 }, { "epoch": 0.08, "learning_rate": 1.976078016971653e-05, "loss": 1.597, "step": 2394 }, { "epoch": 0.08, "learning_rate": 1.976055013299997e-05, "loss": 1.6589, "step": 2395 }, { "epoch": 0.08, "learning_rate": 1.9760319987073734e-05, "loss": 1.709, "step": 2396 }, { "epoch": 0.08, "learning_rate": 1.9760089731940384e-05, "loss": 1.6709, "step": 2397 }, { "epoch": 0.08, "learning_rate": 1.975985936760251e-05, "loss": 1.5864, "step": 2398 }, { "epoch": 0.08, "learning_rate": 1.975962889406268e-05, "loss": 1.5077, "step": 2399 }, { "epoch": 0.08, "learning_rate": 1.9759398311323474e-05, "loss": 1.5306, "step": 2400 }, { "epoch": 0.08, "learning_rate": 1.9759167619387474e-05, "loss": 1.5785, "step": 2401 }, { "epoch": 0.08, "learning_rate": 1.9758936818257263e-05, "loss": 1.5257, "step": 2402 }, { "epoch": 0.08, "learning_rate": 1.975870590793542e-05, "loss": 1.491, "step": 2403 }, { "epoch": 0.08, "learning_rate": 1.975847488842453e-05, "loss": 1.4754, "step": 2404 }, { "epoch": 0.08, "learning_rate": 1.975824375972718e-05, "loss": 1.4554, "step": 2405 }, { "epoch": 0.08, "learning_rate": 1.975801252184595e-05, "loss": 1.5886, "step": 2406 }, { "epoch": 0.08, "learning_rate": 1.975778117478343e-05, "loss": 1.5965, "step": 2407 }, { "epoch": 0.08, "learning_rate": 1.975754971854221e-05, "loss": 1.5256, "step": 2408 }, { "epoch": 0.08, "learning_rate": 1.975731815312488e-05, "loss": 1.5327, "step": 2409 }, { "epoch": 0.08, "learning_rate": 1.975708647853403e-05, "loss": 1.6185, "step": 2410 }, { "epoch": 0.08, "learning_rate": 1.9756854694772258e-05, "loss": 1.7395, "step": 2411 }, { "epoch": 0.08, "learning_rate": 1.9756622801842144e-05, "loss": 1.6559, "step": 2412 }, { "epoch": 0.08, "learning_rate": 1.9756390799746295e-05, "loss": 1.5122, "step": 2413 }, { "epoch": 0.08, "learning_rate": 1.97561586884873e-05, "loss": 1.5461, "step": 2414 }, { "epoch": 0.08, "learning_rate": 1.9755926468067757e-05, "loss": 1.5049, "step": 2415 }, { "epoch": 0.08, "learning_rate": 1.9755694138490268e-05, "loss": 1.5803, "step": 2416 }, { "epoch": 0.08, "learning_rate": 1.975546169975743e-05, "loss": 1.7122, "step": 2417 }, { "epoch": 0.08, "learning_rate": 1.975522915187184e-05, "loss": 1.6426, "step": 2418 }, { "epoch": 0.08, "learning_rate": 1.975499649483611e-05, "loss": 1.709, "step": 2419 }, { "epoch": 0.08, "learning_rate": 1.9754763728652834e-05, "loss": 1.6282, "step": 2420 }, { "epoch": 0.08, "learning_rate": 1.975453085332462e-05, "loss": 1.6344, "step": 2421 }, { "epoch": 0.08, "learning_rate": 1.9754297868854075e-05, "loss": 1.7003, "step": 2422 }, { "epoch": 0.08, "learning_rate": 1.9754064775243797e-05, "loss": 1.7333, "step": 2423 }, { "epoch": 0.08, "learning_rate": 1.9753831572496405e-05, "loss": 1.6937, "step": 2424 }, { "epoch": 0.08, "learning_rate": 1.9753598260614506e-05, "loss": 1.6301, "step": 2425 }, { "epoch": 0.08, "learning_rate": 1.9753364839600705e-05, "loss": 1.5346, "step": 2426 }, { "epoch": 0.08, "learning_rate": 1.975313130945762e-05, "loss": 1.6508, "step": 2427 }, { "epoch": 0.08, "learning_rate": 1.975289767018786e-05, "loss": 1.5637, "step": 2428 }, { "epoch": 0.08, "learning_rate": 1.975266392179404e-05, "loss": 1.6423, "step": 2429 }, { "epoch": 0.08, "learning_rate": 1.975243006427878e-05, "loss": 1.6021, "step": 2430 }, { "epoch": 0.08, "learning_rate": 1.9752196097644687e-05, "loss": 1.6352, "step": 2431 }, { "epoch": 0.08, "learning_rate": 1.9751962021894385e-05, "loss": 1.5953, "step": 2432 }, { "epoch": 0.08, "learning_rate": 1.9751727837030494e-05, "loss": 1.5539, "step": 2433 }, { "epoch": 0.08, "learning_rate": 1.9751493543055634e-05, "loss": 1.6249, "step": 2434 }, { "epoch": 0.08, "learning_rate": 1.9751259139972425e-05, "loss": 1.4641, "step": 2435 }, { "epoch": 0.08, "learning_rate": 1.9751024627783487e-05, "loss": 1.5637, "step": 2436 }, { "epoch": 0.08, "learning_rate": 1.9750790006491447e-05, "loss": 1.5541, "step": 2437 }, { "epoch": 0.08, "learning_rate": 1.975055527609893e-05, "loss": 1.6387, "step": 2438 }, { "epoch": 0.08, "learning_rate": 1.9750320436608565e-05, "loss": 1.7406, "step": 2439 }, { "epoch": 0.08, "learning_rate": 1.9750085488022974e-05, "loss": 1.6306, "step": 2440 }, { "epoch": 0.08, "learning_rate": 1.974985043034479e-05, "loss": 1.5758, "step": 2441 }, { "epoch": 0.08, "learning_rate": 1.9749615263576642e-05, "loss": 1.5595, "step": 2442 }, { "epoch": 0.08, "learning_rate": 1.974937998772116e-05, "loss": 1.4882, "step": 2443 }, { "epoch": 0.08, "learning_rate": 1.9749144602780976e-05, "loss": 1.6651, "step": 2444 }, { "epoch": 0.08, "learning_rate": 1.9748909108758727e-05, "loss": 1.677, "step": 2445 }, { "epoch": 0.08, "learning_rate": 1.9748673505657045e-05, "loss": 1.6117, "step": 2446 }, { "epoch": 0.08, "learning_rate": 1.974843779347857e-05, "loss": 1.7334, "step": 2447 }, { "epoch": 0.08, "learning_rate": 1.9748201972225933e-05, "loss": 1.6659, "step": 2448 }, { "epoch": 0.08, "learning_rate": 1.9747966041901776e-05, "loss": 1.632, "step": 2449 }, { "epoch": 0.08, "learning_rate": 1.974773000250874e-05, "loss": 1.6088, "step": 2450 }, { "epoch": 0.08, "learning_rate": 1.9747493854049468e-05, "loss": 1.5915, "step": 2451 }, { "epoch": 0.08, "learning_rate": 1.9747257596526594e-05, "loss": 1.6289, "step": 2452 }, { "epoch": 0.08, "learning_rate": 1.974702122994277e-05, "loss": 1.6116, "step": 2453 }, { "epoch": 0.08, "learning_rate": 1.9746784754300636e-05, "loss": 1.5854, "step": 2454 }, { "epoch": 0.08, "learning_rate": 1.9746548169602843e-05, "loss": 1.6115, "step": 2455 }, { "epoch": 0.08, "learning_rate": 1.9746311475852028e-05, "loss": 1.5532, "step": 2456 }, { "epoch": 0.08, "learning_rate": 1.9746074673050848e-05, "loss": 1.4704, "step": 2457 }, { "epoch": 0.08, "learning_rate": 1.9745837761201948e-05, "loss": 1.6337, "step": 2458 }, { "epoch": 0.08, "learning_rate": 1.9745600740307988e-05, "loss": 1.496, "step": 2459 }, { "epoch": 0.08, "learning_rate": 1.9745363610371608e-05, "loss": 1.4797, "step": 2460 }, { "epoch": 0.08, "learning_rate": 1.9745126371395463e-05, "loss": 1.5359, "step": 2461 }, { "epoch": 0.08, "learning_rate": 1.9744889023382217e-05, "loss": 1.4257, "step": 2462 }, { "epoch": 0.08, "learning_rate": 1.9744651566334516e-05, "loss": 1.5815, "step": 2463 }, { "epoch": 0.08, "learning_rate": 1.974441400025502e-05, "loss": 1.5657, "step": 2464 }, { "epoch": 0.08, "learning_rate": 1.9744176325146388e-05, "loss": 1.5477, "step": 2465 }, { "epoch": 0.08, "learning_rate": 1.974393854101128e-05, "loss": 1.5463, "step": 2466 }, { "epoch": 0.08, "learning_rate": 1.9743700647852356e-05, "loss": 1.6416, "step": 2467 }, { "epoch": 0.08, "learning_rate": 1.974346264567227e-05, "loss": 1.6568, "step": 2468 }, { "epoch": 0.08, "learning_rate": 1.97432245344737e-05, "loss": 1.5494, "step": 2469 }, { "epoch": 0.08, "learning_rate": 1.97429863142593e-05, "loss": 1.5438, "step": 2470 }, { "epoch": 0.08, "learning_rate": 1.9742747985031733e-05, "loss": 1.4898, "step": 2471 }, { "epoch": 0.08, "learning_rate": 1.9742509546793673e-05, "loss": 1.4836, "step": 2472 }, { "epoch": 0.08, "learning_rate": 1.974227099954779e-05, "loss": 1.5315, "step": 2473 }, { "epoch": 0.08, "learning_rate": 1.974203234329674e-05, "loss": 1.5448, "step": 2474 }, { "epoch": 0.08, "learning_rate": 1.9741793578043204e-05, "loss": 1.6018, "step": 2475 }, { "epoch": 0.08, "learning_rate": 1.9741554703789852e-05, "loss": 1.6803, "step": 2476 }, { "epoch": 0.08, "learning_rate": 1.9741315720539355e-05, "loss": 1.512, "step": 2477 }, { "epoch": 0.08, "learning_rate": 1.9741076628294387e-05, "loss": 1.5579, "step": 2478 }, { "epoch": 0.08, "learning_rate": 1.9740837427057627e-05, "loss": 1.6123, "step": 2479 }, { "epoch": 0.08, "learning_rate": 1.9740598116831746e-05, "loss": 1.6422, "step": 2480 }, { "epoch": 0.08, "learning_rate": 1.9740358697619422e-05, "loss": 1.5295, "step": 2481 }, { "epoch": 0.08, "learning_rate": 1.9740119169423337e-05, "loss": 1.4615, "step": 2482 }, { "epoch": 0.08, "learning_rate": 1.973987953224617e-05, "loss": 1.4433, "step": 2483 }, { "epoch": 0.08, "learning_rate": 1.97396397860906e-05, "loss": 1.6342, "step": 2484 }, { "epoch": 0.08, "learning_rate": 1.9739399930959316e-05, "loss": 1.468, "step": 2485 }, { "epoch": 0.08, "learning_rate": 1.9739159966854994e-05, "loss": 1.5256, "step": 2486 }, { "epoch": 0.08, "learning_rate": 1.9738919893780323e-05, "loss": 1.6149, "step": 2487 }, { "epoch": 0.08, "learning_rate": 1.9738679711737984e-05, "loss": 1.5058, "step": 2488 }, { "epoch": 0.08, "learning_rate": 1.9738439420730674e-05, "loss": 1.3951, "step": 2489 }, { "epoch": 0.08, "learning_rate": 1.9738199020761073e-05, "loss": 1.5075, "step": 2490 }, { "epoch": 0.08, "learning_rate": 1.9737958511831878e-05, "loss": 1.5183, "step": 2491 }, { "epoch": 0.08, "learning_rate": 1.973771789394577e-05, "loss": 1.5151, "step": 2492 }, { "epoch": 0.08, "learning_rate": 1.973747716710545e-05, "loss": 1.5815, "step": 2493 }, { "epoch": 0.08, "learning_rate": 1.973723633131361e-05, "loss": 1.6267, "step": 2494 }, { "epoch": 0.08, "learning_rate": 1.9736995386572942e-05, "loss": 1.5236, "step": 2495 }, { "epoch": 0.08, "learning_rate": 1.9736754332886142e-05, "loss": 1.4869, "step": 2496 }, { "epoch": 0.08, "learning_rate": 1.973651317025591e-05, "loss": 1.5299, "step": 2497 }, { "epoch": 0.08, "learning_rate": 1.9736271898684945e-05, "loss": 1.383, "step": 2498 }, { "epoch": 0.08, "learning_rate": 1.973603051817594e-05, "loss": 1.5724, "step": 2499 }, { "epoch": 0.08, "learning_rate": 1.9735789028731603e-05, "loss": 1.5609, "step": 2500 }, { "epoch": 0.08, "learning_rate": 1.9735547430354635e-05, "loss": 1.6482, "step": 2501 }, { "epoch": 0.08, "learning_rate": 1.973530572304773e-05, "loss": 1.5945, "step": 2502 }, { "epoch": 0.08, "learning_rate": 1.9735063906813608e-05, "loss": 1.5895, "step": 2503 }, { "epoch": 0.08, "learning_rate": 1.9734821981654963e-05, "loss": 1.506, "step": 2504 }, { "epoch": 0.08, "learning_rate": 1.9734579947574506e-05, "loss": 1.5201, "step": 2505 }, { "epoch": 0.08, "learning_rate": 1.9734337804574943e-05, "loss": 1.5077, "step": 2506 }, { "epoch": 0.08, "learning_rate": 1.9734095552658984e-05, "loss": 1.5517, "step": 2507 }, { "epoch": 0.08, "learning_rate": 1.9733853191829343e-05, "loss": 1.4852, "step": 2508 }, { "epoch": 0.08, "learning_rate": 1.9733610722088724e-05, "loss": 1.5431, "step": 2509 }, { "epoch": 0.08, "learning_rate": 1.973336814343985e-05, "loss": 1.6509, "step": 2510 }, { "epoch": 0.08, "learning_rate": 1.973312545588543e-05, "loss": 1.4484, "step": 2511 }, { "epoch": 0.08, "learning_rate": 1.973288265942818e-05, "loss": 1.5461, "step": 2512 }, { "epoch": 0.08, "learning_rate": 1.9732639754070812e-05, "loss": 1.6609, "step": 2513 }, { "epoch": 0.08, "learning_rate": 1.973239673981605e-05, "loss": 1.5161, "step": 2514 }, { "epoch": 0.08, "learning_rate": 1.9732153616666608e-05, "loss": 1.5084, "step": 2515 }, { "epoch": 0.08, "learning_rate": 1.9731910384625214e-05, "loss": 1.5284, "step": 2516 }, { "epoch": 0.08, "learning_rate": 1.9731667043694583e-05, "loss": 1.529, "step": 2517 }, { "epoch": 0.08, "learning_rate": 1.973142359387744e-05, "loss": 1.5338, "step": 2518 }, { "epoch": 0.08, "learning_rate": 1.973118003517651e-05, "loss": 1.5476, "step": 2519 }, { "epoch": 0.08, "learning_rate": 1.9730936367594514e-05, "loss": 1.4885, "step": 2520 }, { "epoch": 0.08, "learning_rate": 1.9730692591134182e-05, "loss": 1.5644, "step": 2521 }, { "epoch": 0.08, "learning_rate": 1.973044870579824e-05, "loss": 1.5343, "step": 2522 }, { "epoch": 0.08, "learning_rate": 1.9730204711589418e-05, "loss": 1.4119, "step": 2523 }, { "epoch": 0.08, "learning_rate": 1.9729960608510444e-05, "loss": 1.4257, "step": 2524 }, { "epoch": 0.08, "learning_rate": 1.972971639656405e-05, "loss": 1.5317, "step": 2525 }, { "epoch": 0.08, "learning_rate": 1.9729472075752974e-05, "loss": 1.663, "step": 2526 }, { "epoch": 0.08, "learning_rate": 1.9729227646079942e-05, "loss": 1.4778, "step": 2527 }, { "epoch": 0.08, "learning_rate": 1.972898310754769e-05, "loss": 1.689, "step": 2528 }, { "epoch": 0.08, "learning_rate": 1.9728738460158957e-05, "loss": 1.5871, "step": 2529 }, { "epoch": 0.08, "learning_rate": 1.972849370391648e-05, "loss": 1.5193, "step": 2530 }, { "epoch": 0.08, "learning_rate": 1.9728248838822996e-05, "loss": 1.5137, "step": 2531 }, { "epoch": 0.08, "learning_rate": 1.9728003864881242e-05, "loss": 1.5625, "step": 2532 }, { "epoch": 0.08, "learning_rate": 1.972775878209397e-05, "loss": 1.5459, "step": 2533 }, { "epoch": 0.08, "learning_rate": 1.9727513590463906e-05, "loss": 1.5906, "step": 2534 }, { "epoch": 0.08, "learning_rate": 1.9727268289993807e-05, "loss": 1.6463, "step": 2535 }, { "epoch": 0.08, "learning_rate": 1.9727022880686413e-05, "loss": 1.709, "step": 2536 }, { "epoch": 0.08, "learning_rate": 1.972677736254447e-05, "loss": 1.7294, "step": 2537 }, { "epoch": 0.08, "learning_rate": 1.9726531735570723e-05, "loss": 1.6308, "step": 2538 }, { "epoch": 0.08, "learning_rate": 1.972628599976792e-05, "loss": 1.6702, "step": 2539 }, { "epoch": 0.08, "learning_rate": 1.9726040155138814e-05, "loss": 1.7269, "step": 2540 }, { "epoch": 0.08, "learning_rate": 1.9725794201686155e-05, "loss": 1.6725, "step": 2541 }, { "epoch": 0.08, "learning_rate": 1.9725548139412693e-05, "loss": 1.8005, "step": 2542 }, { "epoch": 0.08, "learning_rate": 1.9725301968321182e-05, "loss": 1.7078, "step": 2543 }, { "epoch": 0.08, "learning_rate": 1.9725055688414378e-05, "loss": 1.7744, "step": 2544 }, { "epoch": 0.08, "learning_rate": 1.9724809299695035e-05, "loss": 1.7491, "step": 2545 }, { "epoch": 0.08, "learning_rate": 1.972456280216591e-05, "loss": 1.6443, "step": 2546 }, { "epoch": 0.08, "learning_rate": 1.972431619582976e-05, "loss": 1.6532, "step": 2547 }, { "epoch": 0.08, "learning_rate": 1.9724069480689346e-05, "loss": 1.7238, "step": 2548 }, { "epoch": 0.08, "learning_rate": 1.972382265674743e-05, "loss": 1.6181, "step": 2549 }, { "epoch": 0.09, "learning_rate": 1.972357572400677e-05, "loss": 1.662, "step": 2550 }, { "epoch": 0.09, "learning_rate": 1.9723328682470127e-05, "loss": 1.5708, "step": 2551 }, { "epoch": 0.09, "learning_rate": 1.9723081532140272e-05, "loss": 1.6606, "step": 2552 }, { "epoch": 0.09, "learning_rate": 1.9722834273019966e-05, "loss": 1.6621, "step": 2553 }, { "epoch": 0.09, "learning_rate": 1.9722586905111975e-05, "loss": 1.6525, "step": 2554 }, { "epoch": 0.09, "learning_rate": 1.972233942841907e-05, "loss": 1.6631, "step": 2555 }, { "epoch": 0.09, "learning_rate": 1.972209184294402e-05, "loss": 1.6694, "step": 2556 }, { "epoch": 0.09, "learning_rate": 1.972184414868959e-05, "loss": 1.5748, "step": 2557 }, { "epoch": 0.09, "learning_rate": 1.9721596345658554e-05, "loss": 1.5731, "step": 2558 }, { "epoch": 0.09, "learning_rate": 1.972134843385369e-05, "loss": 1.5519, "step": 2559 }, { "epoch": 0.09, "learning_rate": 1.9721100413277763e-05, "loss": 1.518, "step": 2560 }, { "epoch": 0.09, "learning_rate": 1.9720852283933553e-05, "loss": 1.6326, "step": 2561 }, { "epoch": 0.09, "learning_rate": 1.9720604045823835e-05, "loss": 1.6491, "step": 2562 }, { "epoch": 0.09, "learning_rate": 1.9720355698951387e-05, "loss": 1.657, "step": 2563 }, { "epoch": 0.09, "learning_rate": 1.972010724331899e-05, "loss": 1.6488, "step": 2564 }, { "epoch": 0.09, "learning_rate": 1.9719858678929422e-05, "loss": 1.6047, "step": 2565 }, { "epoch": 0.09, "learning_rate": 1.9719610005785466e-05, "loss": 1.5896, "step": 2566 }, { "epoch": 0.09, "learning_rate": 1.9719361223889896e-05, "loss": 1.5762, "step": 2567 }, { "epoch": 0.09, "learning_rate": 1.9719112333245508e-05, "loss": 1.6207, "step": 2568 }, { "epoch": 0.09, "learning_rate": 1.971886333385508e-05, "loss": 1.6023, "step": 2569 }, { "epoch": 0.09, "learning_rate": 1.971861422572139e-05, "loss": 1.5276, "step": 2570 }, { "epoch": 0.09, "learning_rate": 1.9718365008847247e-05, "loss": 1.6458, "step": 2571 }, { "epoch": 0.09, "learning_rate": 1.9718115683235418e-05, "loss": 1.6201, "step": 2572 }, { "epoch": 0.09, "learning_rate": 1.9717866248888705e-05, "loss": 1.5268, "step": 2573 }, { "epoch": 0.09, "learning_rate": 1.971761670580989e-05, "loss": 1.4889, "step": 2574 }, { "epoch": 0.09, "learning_rate": 1.9717367054001774e-05, "loss": 1.4311, "step": 2575 }, { "epoch": 0.09, "learning_rate": 1.9717117293467146e-05, "loss": 1.5327, "step": 2576 }, { "epoch": 0.09, "learning_rate": 1.9716867424208805e-05, "loss": 1.5347, "step": 2577 }, { "epoch": 0.09, "learning_rate": 1.9716617446229537e-05, "loss": 1.4106, "step": 2578 }, { "epoch": 0.09, "learning_rate": 1.9716367359532146e-05, "loss": 1.4516, "step": 2579 }, { "epoch": 0.09, "learning_rate": 1.971611716411943e-05, "loss": 1.6337, "step": 2580 }, { "epoch": 0.09, "learning_rate": 1.9715866859994187e-05, "loss": 1.6064, "step": 2581 }, { "epoch": 0.09, "learning_rate": 1.971561644715922e-05, "loss": 1.5406, "step": 2582 }, { "epoch": 0.09, "learning_rate": 1.9715365925617325e-05, "loss": 1.5054, "step": 2583 }, { "epoch": 0.09, "learning_rate": 1.9715115295371313e-05, "loss": 1.6712, "step": 2584 }, { "epoch": 0.09, "learning_rate": 1.971486455642398e-05, "loss": 1.5994, "step": 2585 }, { "epoch": 0.09, "learning_rate": 1.9714613708778137e-05, "loss": 1.5487, "step": 2586 }, { "epoch": 0.09, "learning_rate": 1.971436275243659e-05, "loss": 1.6329, "step": 2587 }, { "epoch": 0.09, "learning_rate": 1.9714111687402146e-05, "loss": 1.6534, "step": 2588 }, { "epoch": 0.09, "learning_rate": 1.9713860513677617e-05, "loss": 1.5932, "step": 2589 }, { "epoch": 0.09, "learning_rate": 1.9713609231265807e-05, "loss": 1.5031, "step": 2590 }, { "epoch": 0.09, "learning_rate": 1.971335784016953e-05, "loss": 1.6615, "step": 2591 }, { "epoch": 0.09, "learning_rate": 1.9713106340391607e-05, "loss": 1.5447, "step": 2592 }, { "epoch": 0.09, "learning_rate": 1.9712854731934837e-05, "loss": 1.5434, "step": 2593 }, { "epoch": 0.09, "learning_rate": 1.971260301480205e-05, "loss": 1.5616, "step": 2594 }, { "epoch": 0.09, "learning_rate": 1.971235118899605e-05, "loss": 1.5682, "step": 2595 }, { "epoch": 0.09, "learning_rate": 1.9712099254519666e-05, "loss": 1.5483, "step": 2596 }, { "epoch": 0.09, "learning_rate": 1.971184721137571e-05, "loss": 1.6004, "step": 2597 }, { "epoch": 0.09, "learning_rate": 1.9711595059566998e-05, "loss": 1.7153, "step": 2598 }, { "epoch": 0.09, "learning_rate": 1.971134279909636e-05, "loss": 1.5961, "step": 2599 }, { "epoch": 0.09, "learning_rate": 1.9711090429966617e-05, "loss": 1.5993, "step": 2600 }, { "epoch": 0.09, "learning_rate": 1.971083795218059e-05, "loss": 1.6434, "step": 2601 }, { "epoch": 0.09, "learning_rate": 1.9710585365741105e-05, "loss": 1.611, "step": 2602 }, { "epoch": 0.09, "learning_rate": 1.9710332670650988e-05, "loss": 1.5569, "step": 2603 }, { "epoch": 0.09, "learning_rate": 1.9710079866913067e-05, "loss": 1.5588, "step": 2604 }, { "epoch": 0.09, "learning_rate": 1.970982695453017e-05, "loss": 1.4581, "step": 2605 }, { "epoch": 0.09, "learning_rate": 1.9709573933505123e-05, "loss": 1.736, "step": 2606 }, { "epoch": 0.09, "learning_rate": 1.9709320803840765e-05, "loss": 1.6408, "step": 2607 }, { "epoch": 0.09, "learning_rate": 1.970906756553992e-05, "loss": 1.5429, "step": 2608 }, { "epoch": 0.09, "learning_rate": 1.970881421860543e-05, "loss": 1.5345, "step": 2609 }, { "epoch": 0.09, "learning_rate": 1.970856076304012e-05, "loss": 1.5488, "step": 2610 }, { "epoch": 0.09, "learning_rate": 1.9708307198846837e-05, "loss": 1.6491, "step": 2611 }, { "epoch": 0.09, "learning_rate": 1.970805352602841e-05, "loss": 1.5747, "step": 2612 }, { "epoch": 0.09, "learning_rate": 1.970779974458768e-05, "loss": 1.6097, "step": 2613 }, { "epoch": 0.09, "learning_rate": 1.970754585452748e-05, "loss": 1.6466, "step": 2614 }, { "epoch": 0.09, "learning_rate": 1.9707291855850664e-05, "loss": 1.5216, "step": 2615 }, { "epoch": 0.09, "learning_rate": 1.9707037748560065e-05, "loss": 1.5776, "step": 2616 }, { "epoch": 0.09, "learning_rate": 1.9706783532658528e-05, "loss": 1.5026, "step": 2617 }, { "epoch": 0.09, "learning_rate": 1.97065292081489e-05, "loss": 1.508, "step": 2618 }, { "epoch": 0.09, "learning_rate": 1.9706274775034015e-05, "loss": 1.4504, "step": 2619 }, { "epoch": 0.09, "learning_rate": 1.9706020233316735e-05, "loss": 1.6012, "step": 2620 }, { "epoch": 0.09, "learning_rate": 1.97057655829999e-05, "loss": 1.4946, "step": 2621 }, { "epoch": 0.09, "learning_rate": 1.970551082408636e-05, "loss": 1.5749, "step": 2622 }, { "epoch": 0.09, "learning_rate": 1.9705255956578972e-05, "loss": 1.5672, "step": 2623 }, { "epoch": 0.09, "learning_rate": 1.9705000980480578e-05, "loss": 1.5608, "step": 2624 }, { "epoch": 0.09, "learning_rate": 1.9704745895794036e-05, "loss": 1.5458, "step": 2625 }, { "epoch": 0.09, "learning_rate": 1.9704490702522198e-05, "loss": 1.6673, "step": 2626 }, { "epoch": 0.09, "learning_rate": 1.9704235400667922e-05, "loss": 1.7599, "step": 2627 }, { "epoch": 0.09, "learning_rate": 1.970397999023406e-05, "loss": 1.5765, "step": 2628 }, { "epoch": 0.09, "learning_rate": 1.9703724471223475e-05, "loss": 1.6099, "step": 2629 }, { "epoch": 0.09, "learning_rate": 1.9703468843639023e-05, "loss": 1.7077, "step": 2630 }, { "epoch": 0.09, "learning_rate": 1.970321310748357e-05, "loss": 1.5466, "step": 2631 }, { "epoch": 0.09, "learning_rate": 1.9702957262759964e-05, "loss": 1.4896, "step": 2632 }, { "epoch": 0.09, "learning_rate": 1.9702701309471083e-05, "loss": 1.6112, "step": 2633 }, { "epoch": 0.09, "learning_rate": 1.970244524761978e-05, "loss": 1.4499, "step": 2634 }, { "epoch": 0.09, "learning_rate": 1.9702189077208922e-05, "loss": 1.629, "step": 2635 }, { "epoch": 0.09, "learning_rate": 1.9701932798241382e-05, "loss": 1.6576, "step": 2636 }, { "epoch": 0.09, "learning_rate": 1.9701676410720024e-05, "loss": 1.4623, "step": 2637 }, { "epoch": 0.09, "learning_rate": 1.970141991464771e-05, "loss": 1.607, "step": 2638 }, { "epoch": 0.09, "learning_rate": 1.9701163310027316e-05, "loss": 1.4993, "step": 2639 }, { "epoch": 0.09, "learning_rate": 1.9700906596861715e-05, "loss": 1.744, "step": 2640 }, { "epoch": 0.09, "learning_rate": 1.9700649775153775e-05, "loss": 1.5576, "step": 2641 }, { "epoch": 0.09, "learning_rate": 1.970039284490637e-05, "loss": 1.6203, "step": 2642 }, { "epoch": 0.09, "learning_rate": 1.9700135806122378e-05, "loss": 1.6793, "step": 2643 }, { "epoch": 0.09, "learning_rate": 1.9699878658804673e-05, "loss": 1.5225, "step": 2644 }, { "epoch": 0.09, "learning_rate": 1.9699621402956132e-05, "loss": 1.5162, "step": 2645 }, { "epoch": 0.09, "learning_rate": 1.969936403857963e-05, "loss": 1.5801, "step": 2646 }, { "epoch": 0.09, "learning_rate": 1.969910656567805e-05, "loss": 1.5964, "step": 2647 }, { "epoch": 0.09, "learning_rate": 1.969884898425428e-05, "loss": 1.5982, "step": 2648 }, { "epoch": 0.09, "learning_rate": 1.9698591294311193e-05, "loss": 1.4792, "step": 2649 }, { "epoch": 0.09, "learning_rate": 1.9698333495851672e-05, "loss": 1.5479, "step": 2650 }, { "epoch": 0.09, "learning_rate": 1.96980755888786e-05, "loss": 1.5333, "step": 2651 }, { "epoch": 0.09, "learning_rate": 1.9697817573394874e-05, "loss": 1.5143, "step": 2652 }, { "epoch": 0.09, "learning_rate": 1.969755944940337e-05, "loss": 1.5211, "step": 2653 }, { "epoch": 0.09, "learning_rate": 1.969730121690698e-05, "loss": 1.5719, "step": 2654 }, { "epoch": 0.09, "learning_rate": 1.969704287590859e-05, "loss": 1.5685, "step": 2655 }, { "epoch": 0.09, "learning_rate": 1.9696784426411097e-05, "loss": 1.5194, "step": 2656 }, { "epoch": 0.09, "learning_rate": 1.9696525868417387e-05, "loss": 1.4923, "step": 2657 }, { "epoch": 0.09, "learning_rate": 1.9696267201930354e-05, "loss": 1.5929, "step": 2658 }, { "epoch": 0.09, "learning_rate": 1.9696008426952897e-05, "loss": 1.6055, "step": 2659 }, { "epoch": 0.09, "learning_rate": 1.9695749543487905e-05, "loss": 1.4696, "step": 2660 }, { "epoch": 0.09, "learning_rate": 1.9695490551538277e-05, "loss": 1.5767, "step": 2661 }, { "epoch": 0.09, "learning_rate": 1.9695231451106914e-05, "loss": 1.4765, "step": 2662 }, { "epoch": 0.09, "learning_rate": 1.9694972242196712e-05, "loss": 1.5965, "step": 2663 }, { "epoch": 0.09, "learning_rate": 1.9694712924810568e-05, "loss": 1.4989, "step": 2664 }, { "epoch": 0.09, "learning_rate": 1.9694453498951392e-05, "loss": 1.5389, "step": 2665 }, { "epoch": 0.09, "learning_rate": 1.9694193964622075e-05, "loss": 1.5814, "step": 2666 }, { "epoch": 0.09, "learning_rate": 1.969393432182553e-05, "loss": 1.6155, "step": 2667 }, { "epoch": 0.09, "learning_rate": 1.9693674570564663e-05, "loss": 1.5597, "step": 2668 }, { "epoch": 0.09, "learning_rate": 1.9693414710842377e-05, "loss": 1.6596, "step": 2669 }, { "epoch": 0.09, "learning_rate": 1.9693154742661575e-05, "loss": 1.6428, "step": 2670 }, { "epoch": 0.09, "learning_rate": 1.9692894666025174e-05, "loss": 1.483, "step": 2671 }, { "epoch": 0.09, "learning_rate": 1.969263448093608e-05, "loss": 1.636, "step": 2672 }, { "epoch": 0.09, "learning_rate": 1.96923741873972e-05, "loss": 1.6168, "step": 2673 }, { "epoch": 0.09, "learning_rate": 1.9692113785411456e-05, "loss": 1.5267, "step": 2674 }, { "epoch": 0.09, "learning_rate": 1.9691853274981754e-05, "loss": 1.5343, "step": 2675 }, { "epoch": 0.09, "learning_rate": 1.969159265611101e-05, "loss": 1.55, "step": 2676 }, { "epoch": 0.09, "learning_rate": 1.9691331928802144e-05, "loss": 1.5472, "step": 2677 }, { "epoch": 0.09, "learning_rate": 1.969107109305807e-05, "loss": 1.6011, "step": 2678 }, { "epoch": 0.09, "learning_rate": 1.9690810148881706e-05, "loss": 1.6189, "step": 2679 }, { "epoch": 0.09, "learning_rate": 1.9690549096275972e-05, "loss": 1.4427, "step": 2680 }, { "epoch": 0.09, "learning_rate": 1.9690287935243793e-05, "loss": 1.5086, "step": 2681 }, { "epoch": 0.09, "learning_rate": 1.9690026665788084e-05, "loss": 1.5623, "step": 2682 }, { "epoch": 0.09, "learning_rate": 1.9689765287911774e-05, "loss": 1.5889, "step": 2683 }, { "epoch": 0.09, "learning_rate": 1.9689503801617785e-05, "loss": 1.4979, "step": 2684 }, { "epoch": 0.09, "learning_rate": 1.9689242206909044e-05, "loss": 1.5646, "step": 2685 }, { "epoch": 0.09, "learning_rate": 1.9688980503788474e-05, "loss": 1.6258, "step": 2686 }, { "epoch": 0.09, "learning_rate": 1.9688718692259007e-05, "loss": 1.4153, "step": 2687 }, { "epoch": 0.09, "learning_rate": 1.9688456772323573e-05, "loss": 1.4745, "step": 2688 }, { "epoch": 0.09, "learning_rate": 1.9688194743985103e-05, "loss": 1.5028, "step": 2689 }, { "epoch": 0.09, "learning_rate": 1.968793260724653e-05, "loss": 1.5174, "step": 2690 }, { "epoch": 0.09, "learning_rate": 1.9687670362110775e-05, "loss": 1.5119, "step": 2691 }, { "epoch": 0.09, "learning_rate": 1.9687408008580785e-05, "loss": 1.5264, "step": 2692 }, { "epoch": 0.09, "learning_rate": 1.9687145546659495e-05, "loss": 1.5567, "step": 2693 }, { "epoch": 0.09, "learning_rate": 1.9686882976349837e-05, "loss": 1.5383, "step": 2694 }, { "epoch": 0.09, "learning_rate": 1.9686620297654747e-05, "loss": 1.4142, "step": 2695 }, { "epoch": 0.09, "learning_rate": 1.9686357510577174e-05, "loss": 1.5045, "step": 2696 }, { "epoch": 0.09, "learning_rate": 1.9686094615120047e-05, "loss": 1.6123, "step": 2697 }, { "epoch": 0.09, "learning_rate": 1.9685831611286312e-05, "loss": 1.5643, "step": 2698 }, { "epoch": 0.09, "learning_rate": 1.9685568499078913e-05, "loss": 1.48, "step": 2699 }, { "epoch": 0.09, "learning_rate": 1.9685305278500794e-05, "loss": 1.6695, "step": 2700 }, { "epoch": 0.09, "learning_rate": 1.9685041949554896e-05, "loss": 1.6954, "step": 2701 }, { "epoch": 0.09, "learning_rate": 1.9684778512244172e-05, "loss": 1.4651, "step": 2702 }, { "epoch": 0.09, "learning_rate": 1.9684514966571566e-05, "loss": 1.607, "step": 2703 }, { "epoch": 0.09, "learning_rate": 1.9684251312540023e-05, "loss": 1.7125, "step": 2704 }, { "epoch": 0.09, "learning_rate": 1.9683987550152498e-05, "loss": 1.655, "step": 2705 }, { "epoch": 0.09, "learning_rate": 1.9683723679411943e-05, "loss": 1.726, "step": 2706 }, { "epoch": 0.09, "learning_rate": 1.9683459700321305e-05, "loss": 1.5906, "step": 2707 }, { "epoch": 0.09, "learning_rate": 1.9683195612883545e-05, "loss": 1.5562, "step": 2708 }, { "epoch": 0.09, "learning_rate": 1.968293141710161e-05, "loss": 1.621, "step": 2709 }, { "epoch": 0.09, "learning_rate": 1.9682667112978464e-05, "loss": 1.712, "step": 2710 }, { "epoch": 0.09, "learning_rate": 1.968240270051706e-05, "loss": 1.7522, "step": 2711 }, { "epoch": 0.09, "learning_rate": 1.968213817972035e-05, "loss": 1.6612, "step": 2712 }, { "epoch": 0.09, "learning_rate": 1.9681873550591306e-05, "loss": 1.6384, "step": 2713 }, { "epoch": 0.09, "learning_rate": 1.9681608813132883e-05, "loss": 1.6679, "step": 2714 }, { "epoch": 0.09, "learning_rate": 1.9681343967348043e-05, "loss": 1.6525, "step": 2715 }, { "epoch": 0.09, "learning_rate": 1.9681079013239748e-05, "loss": 1.6563, "step": 2716 }, { "epoch": 0.09, "learning_rate": 1.9680813950810965e-05, "loss": 1.7266, "step": 2717 }, { "epoch": 0.09, "learning_rate": 1.968054878006466e-05, "loss": 1.7623, "step": 2718 }, { "epoch": 0.09, "learning_rate": 1.96802835010038e-05, "loss": 1.5916, "step": 2719 }, { "epoch": 0.09, "learning_rate": 1.9680018113631347e-05, "loss": 1.7194, "step": 2720 }, { "epoch": 0.09, "learning_rate": 1.967975261795028e-05, "loss": 1.6386, "step": 2721 }, { "epoch": 0.09, "learning_rate": 1.9679487013963566e-05, "loss": 1.7301, "step": 2722 }, { "epoch": 0.09, "learning_rate": 1.9679221301674173e-05, "loss": 1.5197, "step": 2723 }, { "epoch": 0.09, "learning_rate": 1.9678955481085077e-05, "loss": 1.7004, "step": 2724 }, { "epoch": 0.09, "learning_rate": 1.9678689552199252e-05, "loss": 1.6483, "step": 2725 }, { "epoch": 0.09, "learning_rate": 1.9678423515019675e-05, "loss": 1.6671, "step": 2726 }, { "epoch": 0.09, "learning_rate": 1.967815736954932e-05, "loss": 1.5986, "step": 2727 }, { "epoch": 0.09, "learning_rate": 1.967789111579117e-05, "loss": 1.6431, "step": 2728 }, { "epoch": 0.09, "learning_rate": 1.9677624753748198e-05, "loss": 1.552, "step": 2729 }, { "epoch": 0.09, "learning_rate": 1.967735828342338e-05, "loss": 1.6929, "step": 2730 }, { "epoch": 0.09, "learning_rate": 1.9677091704819714e-05, "loss": 1.5052, "step": 2731 }, { "epoch": 0.09, "learning_rate": 1.9676825017940167e-05, "loss": 1.624, "step": 2732 }, { "epoch": 0.09, "learning_rate": 1.9676558222787733e-05, "loss": 1.4395, "step": 2733 }, { "epoch": 0.09, "learning_rate": 1.967629131936539e-05, "loss": 1.5158, "step": 2734 }, { "epoch": 0.09, "learning_rate": 1.9676024307676127e-05, "loss": 1.5504, "step": 2735 }, { "epoch": 0.09, "learning_rate": 1.967575718772293e-05, "loss": 1.496, "step": 2736 }, { "epoch": 0.09, "learning_rate": 1.9675489959508794e-05, "loss": 1.5466, "step": 2737 }, { "epoch": 0.09, "learning_rate": 1.96752226230367e-05, "loss": 1.5954, "step": 2738 }, { "epoch": 0.09, "learning_rate": 1.967495517830965e-05, "loss": 1.5065, "step": 2739 }, { "epoch": 0.09, "learning_rate": 1.9674687625330623e-05, "loss": 1.4647, "step": 2740 }, { "epoch": 0.09, "learning_rate": 1.9674419964102622e-05, "loss": 1.4508, "step": 2741 }, { "epoch": 0.09, "learning_rate": 1.967415219462864e-05, "loss": 1.6167, "step": 2742 }, { "epoch": 0.09, "learning_rate": 1.9673884316911673e-05, "loss": 1.6134, "step": 2743 }, { "epoch": 0.09, "learning_rate": 1.967361633095472e-05, "loss": 1.5145, "step": 2744 }, { "epoch": 0.09, "learning_rate": 1.9673348236760774e-05, "loss": 1.5148, "step": 2745 }, { "epoch": 0.09, "learning_rate": 1.967308003433284e-05, "loss": 1.5537, "step": 2746 }, { "epoch": 0.09, "learning_rate": 1.9672811723673914e-05, "loss": 1.53, "step": 2747 }, { "epoch": 0.09, "learning_rate": 1.9672543304787e-05, "loss": 1.5959, "step": 2748 }, { "epoch": 0.09, "learning_rate": 1.9672274777675108e-05, "loss": 1.3898, "step": 2749 }, { "epoch": 0.09, "learning_rate": 1.9672006142341234e-05, "loss": 1.5033, "step": 2750 }, { "epoch": 0.09, "learning_rate": 1.9671737398788385e-05, "loss": 1.5778, "step": 2751 }, { "epoch": 0.09, "learning_rate": 1.9671468547019575e-05, "loss": 1.5122, "step": 2752 }, { "epoch": 0.09, "learning_rate": 1.96711995870378e-05, "loss": 1.5782, "step": 2753 }, { "epoch": 0.09, "learning_rate": 1.967093051884608e-05, "loss": 1.614, "step": 2754 }, { "epoch": 0.09, "learning_rate": 1.967066134244742e-05, "loss": 1.4297, "step": 2755 }, { "epoch": 0.09, "learning_rate": 1.967039205784483e-05, "loss": 1.5784, "step": 2756 }, { "epoch": 0.09, "learning_rate": 1.9670122665041332e-05, "loss": 1.5847, "step": 2757 }, { "epoch": 0.09, "learning_rate": 1.9669853164039935e-05, "loss": 1.5701, "step": 2758 }, { "epoch": 0.09, "learning_rate": 1.966958355484365e-05, "loss": 1.55, "step": 2759 }, { "epoch": 0.09, "learning_rate": 1.96693138374555e-05, "loss": 1.4803, "step": 2760 }, { "epoch": 0.09, "learning_rate": 1.9669044011878497e-05, "loss": 1.542, "step": 2761 }, { "epoch": 0.09, "learning_rate": 1.9668774078115668e-05, "loss": 1.5863, "step": 2762 }, { "epoch": 0.09, "learning_rate": 1.9668504036170025e-05, "loss": 1.4599, "step": 2763 }, { "epoch": 0.09, "learning_rate": 1.9668233886044597e-05, "loss": 1.6144, "step": 2764 }, { "epoch": 0.09, "learning_rate": 1.96679636277424e-05, "loss": 1.4511, "step": 2765 }, { "epoch": 0.09, "learning_rate": 1.966769326126646e-05, "loss": 1.6391, "step": 2766 }, { "epoch": 0.09, "learning_rate": 1.9667422786619804e-05, "loss": 1.6741, "step": 2767 }, { "epoch": 0.09, "learning_rate": 1.9667152203805457e-05, "loss": 1.5679, "step": 2768 }, { "epoch": 0.09, "learning_rate": 1.9666881512826446e-05, "loss": 1.6826, "step": 2769 }, { "epoch": 0.09, "learning_rate": 1.9666610713685804e-05, "loss": 1.6148, "step": 2770 }, { "epoch": 0.09, "learning_rate": 1.9666339806386555e-05, "loss": 1.696, "step": 2771 }, { "epoch": 0.09, "learning_rate": 1.9666068790931733e-05, "loss": 1.6057, "step": 2772 }, { "epoch": 0.09, "learning_rate": 1.966579766732437e-05, "loss": 1.5823, "step": 2773 }, { "epoch": 0.09, "learning_rate": 1.96655264355675e-05, "loss": 1.6262, "step": 2774 }, { "epoch": 0.09, "learning_rate": 1.9665255095664155e-05, "loss": 1.6252, "step": 2775 }, { "epoch": 0.09, "learning_rate": 1.9664983647617375e-05, "loss": 1.6119, "step": 2776 }, { "epoch": 0.09, "learning_rate": 1.9664712091430194e-05, "loss": 1.5309, "step": 2777 }, { "epoch": 0.09, "learning_rate": 1.9664440427105654e-05, "loss": 1.6073, "step": 2778 }, { "epoch": 0.09, "learning_rate": 1.966416865464679e-05, "loss": 1.5267, "step": 2779 }, { "epoch": 0.09, "learning_rate": 1.9663896774056646e-05, "loss": 1.5073, "step": 2780 }, { "epoch": 0.09, "learning_rate": 1.9663624785338264e-05, "loss": 1.436, "step": 2781 }, { "epoch": 0.09, "learning_rate": 1.9663352688494686e-05, "loss": 1.6529, "step": 2782 }, { "epoch": 0.09, "learning_rate": 1.9663080483528958e-05, "loss": 1.6164, "step": 2783 }, { "epoch": 0.09, "learning_rate": 1.9662808170444126e-05, "loss": 1.5972, "step": 2784 }, { "epoch": 0.09, "learning_rate": 1.9662535749243233e-05, "loss": 1.7045, "step": 2785 }, { "epoch": 0.09, "learning_rate": 1.966226321992933e-05, "loss": 1.6253, "step": 2786 }, { "epoch": 0.09, "learning_rate": 1.9661990582505464e-05, "loss": 1.5842, "step": 2787 }, { "epoch": 0.09, "learning_rate": 1.9661717836974694e-05, "loss": 1.6749, "step": 2788 }, { "epoch": 0.09, "learning_rate": 1.966144498334006e-05, "loss": 1.5757, "step": 2789 }, { "epoch": 0.09, "learning_rate": 1.966117202160462e-05, "loss": 1.6646, "step": 2790 }, { "epoch": 0.09, "learning_rate": 1.966089895177143e-05, "loss": 1.4754, "step": 2791 }, { "epoch": 0.09, "learning_rate": 1.9660625773843542e-05, "loss": 1.6141, "step": 2792 }, { "epoch": 0.09, "learning_rate": 1.9660352487824016e-05, "loss": 1.5741, "step": 2793 }, { "epoch": 0.09, "learning_rate": 1.9660079093715906e-05, "loss": 1.5186, "step": 2794 }, { "epoch": 0.09, "learning_rate": 1.9659805591522275e-05, "loss": 1.583, "step": 2795 }, { "epoch": 0.09, "learning_rate": 1.965953198124618e-05, "loss": 1.4174, "step": 2796 }, { "epoch": 0.09, "learning_rate": 1.9659258262890683e-05, "loss": 1.6093, "step": 2797 }, { "epoch": 0.09, "learning_rate": 1.965898443645885e-05, "loss": 1.481, "step": 2798 }, { "epoch": 0.09, "learning_rate": 1.965871050195374e-05, "loss": 1.4389, "step": 2799 }, { "epoch": 0.09, "learning_rate": 1.9658436459378422e-05, "loss": 1.394, "step": 2800 }, { "epoch": 0.09, "learning_rate": 1.9658162308735954e-05, "loss": 1.5207, "step": 2801 }, { "epoch": 0.09, "learning_rate": 1.9657888050029416e-05, "loss": 1.4702, "step": 2802 }, { "epoch": 0.09, "learning_rate": 1.9657613683261866e-05, "loss": 1.5427, "step": 2803 }, { "epoch": 0.09, "learning_rate": 1.9657339208436383e-05, "loss": 1.4125, "step": 2804 }, { "epoch": 0.09, "learning_rate": 1.9657064625556027e-05, "loss": 1.5935, "step": 2805 }, { "epoch": 0.09, "learning_rate": 1.965678993462388e-05, "loss": 1.5016, "step": 2806 }, { "epoch": 0.09, "learning_rate": 1.9656515135643014e-05, "loss": 1.5572, "step": 2807 }, { "epoch": 0.09, "learning_rate": 1.9656240228616496e-05, "loss": 1.5546, "step": 2808 }, { "epoch": 0.09, "learning_rate": 1.965596521354741e-05, "loss": 1.6011, "step": 2809 }, { "epoch": 0.09, "learning_rate": 1.9655690090438834e-05, "loss": 1.6313, "step": 2810 }, { "epoch": 0.09, "learning_rate": 1.965541485929384e-05, "loss": 1.5569, "step": 2811 }, { "epoch": 0.09, "learning_rate": 1.965513952011551e-05, "loss": 1.5828, "step": 2812 }, { "epoch": 0.09, "learning_rate": 1.9654864072906926e-05, "loss": 1.7647, "step": 2813 }, { "epoch": 0.09, "learning_rate": 1.965458851767117e-05, "loss": 1.6088, "step": 2814 }, { "epoch": 0.09, "learning_rate": 1.9654312854411325e-05, "loss": 1.6654, "step": 2815 }, { "epoch": 0.09, "learning_rate": 1.9654037083130473e-05, "loss": 1.6091, "step": 2816 }, { "epoch": 0.09, "learning_rate": 1.96537612038317e-05, "loss": 1.5904, "step": 2817 }, { "epoch": 0.09, "learning_rate": 1.9653485216518094e-05, "loss": 1.5946, "step": 2818 }, { "epoch": 0.09, "learning_rate": 1.9653209121192747e-05, "loss": 1.6517, "step": 2819 }, { "epoch": 0.09, "learning_rate": 1.9652932917858743e-05, "loss": 1.5444, "step": 2820 }, { "epoch": 0.09, "learning_rate": 1.9652656606519174e-05, "loss": 1.5858, "step": 2821 }, { "epoch": 0.09, "learning_rate": 1.9652380187177128e-05, "loss": 1.6718, "step": 2822 }, { "epoch": 0.09, "learning_rate": 1.9652103659835704e-05, "loss": 1.6282, "step": 2823 }, { "epoch": 0.09, "learning_rate": 1.9651827024497993e-05, "loss": 1.6948, "step": 2824 }, { "epoch": 0.09, "learning_rate": 1.9651550281167088e-05, "loss": 1.5843, "step": 2825 }, { "epoch": 0.09, "learning_rate": 1.9651273429846092e-05, "loss": 1.5631, "step": 2826 }, { "epoch": 0.09, "learning_rate": 1.9650996470538093e-05, "loss": 1.4059, "step": 2827 }, { "epoch": 0.09, "learning_rate": 1.9650719403246203e-05, "loss": 1.5318, "step": 2828 }, { "epoch": 0.09, "learning_rate": 1.9650442227973506e-05, "loss": 1.4596, "step": 2829 }, { "epoch": 0.09, "learning_rate": 1.9650164944723116e-05, "loss": 1.5702, "step": 2830 }, { "epoch": 0.09, "learning_rate": 1.964988755349813e-05, "loss": 1.5737, "step": 2831 }, { "epoch": 0.09, "learning_rate": 1.9649610054301656e-05, "loss": 1.4969, "step": 2832 }, { "epoch": 0.09, "learning_rate": 1.964933244713679e-05, "loss": 1.5052, "step": 2833 }, { "epoch": 0.09, "learning_rate": 1.9649054732006648e-05, "loss": 1.5895, "step": 2834 }, { "epoch": 0.09, "learning_rate": 1.9648776908914337e-05, "loss": 1.5271, "step": 2835 }, { "epoch": 0.09, "learning_rate": 1.9648498977862954e-05, "loss": 1.5834, "step": 2836 }, { "epoch": 0.09, "learning_rate": 1.964822093885562e-05, "loss": 1.5464, "step": 2837 }, { "epoch": 0.09, "learning_rate": 1.9647942791895446e-05, "loss": 1.6752, "step": 2838 }, { "epoch": 0.09, "learning_rate": 1.9647664536985536e-05, "loss": 1.5673, "step": 2839 }, { "epoch": 0.09, "learning_rate": 1.9647386174129008e-05, "loss": 1.5575, "step": 2840 }, { "epoch": 0.09, "learning_rate": 1.964710770332898e-05, "loss": 1.4727, "step": 2841 }, { "epoch": 0.09, "learning_rate": 1.964682912458856e-05, "loss": 1.5855, "step": 2842 }, { "epoch": 0.09, "learning_rate": 1.964655043791087e-05, "loss": 1.6423, "step": 2843 }, { "epoch": 0.09, "learning_rate": 1.964627164329903e-05, "loss": 1.6193, "step": 2844 }, { "epoch": 0.09, "learning_rate": 1.9645992740756153e-05, "loss": 1.5996, "step": 2845 }, { "epoch": 0.09, "learning_rate": 1.9645713730285366e-05, "loss": 1.599, "step": 2846 }, { "epoch": 0.09, "learning_rate": 1.964543461188979e-05, "loss": 1.5109, "step": 2847 }, { "epoch": 0.09, "learning_rate": 1.9645155385572545e-05, "loss": 1.538, "step": 2848 }, { "epoch": 0.09, "learning_rate": 1.9644876051336756e-05, "loss": 1.4604, "step": 2849 }, { "epoch": 0.1, "learning_rate": 1.9644596609185547e-05, "loss": 1.6387, "step": 2850 }, { "epoch": 0.1, "learning_rate": 1.964431705912205e-05, "loss": 1.6075, "step": 2851 }, { "epoch": 0.1, "learning_rate": 1.964403740114939e-05, "loss": 1.4802, "step": 2852 }, { "epoch": 0.1, "learning_rate": 1.9643757635270694e-05, "loss": 1.5858, "step": 2853 }, { "epoch": 0.1, "learning_rate": 1.9643477761489097e-05, "loss": 1.5851, "step": 2854 }, { "epoch": 0.1, "learning_rate": 1.9643197779807724e-05, "loss": 1.6054, "step": 2855 }, { "epoch": 0.1, "learning_rate": 1.964291769022971e-05, "loss": 1.6113, "step": 2856 }, { "epoch": 0.1, "learning_rate": 1.9642637492758193e-05, "loss": 1.5833, "step": 2857 }, { "epoch": 0.1, "learning_rate": 1.96423571873963e-05, "loss": 1.544, "step": 2858 }, { "epoch": 0.1, "learning_rate": 1.964207677414718e-05, "loss": 1.4863, "step": 2859 }, { "epoch": 0.1, "learning_rate": 1.9641796253013957e-05, "loss": 1.579, "step": 2860 }, { "epoch": 0.1, "learning_rate": 1.9641515623999776e-05, "loss": 1.5751, "step": 2861 }, { "epoch": 0.1, "learning_rate": 1.9641234887107777e-05, "loss": 1.5932, "step": 2862 }, { "epoch": 0.1, "learning_rate": 1.96409540423411e-05, "loss": 1.5645, "step": 2863 }, { "epoch": 0.1, "learning_rate": 1.9640673089702892e-05, "loss": 1.6016, "step": 2864 }, { "epoch": 0.1, "learning_rate": 1.9640392029196288e-05, "loss": 1.6076, "step": 2865 }, { "epoch": 0.1, "learning_rate": 1.9640110860824442e-05, "loss": 1.7017, "step": 2866 }, { "epoch": 0.1, "learning_rate": 1.9639829584590492e-05, "loss": 1.6155, "step": 2867 }, { "epoch": 0.1, "learning_rate": 1.9639548200497586e-05, "loss": 1.5513, "step": 2868 }, { "epoch": 0.1, "learning_rate": 1.9639266708548878e-05, "loss": 1.552, "step": 2869 }, { "epoch": 0.1, "learning_rate": 1.9638985108747516e-05, "loss": 1.5173, "step": 2870 }, { "epoch": 0.1, "learning_rate": 1.9638703401096647e-05, "loss": 1.5075, "step": 2871 }, { "epoch": 0.1, "learning_rate": 1.9638421585599422e-05, "loss": 1.5505, "step": 2872 }, { "epoch": 0.1, "learning_rate": 1.9638139662259e-05, "loss": 1.5369, "step": 2873 }, { "epoch": 0.1, "learning_rate": 1.9637857631078532e-05, "loss": 1.642, "step": 2874 }, { "epoch": 0.1, "learning_rate": 1.9637575492061176e-05, "loss": 1.5716, "step": 2875 }, { "epoch": 0.1, "learning_rate": 1.963729324521009e-05, "loss": 1.5547, "step": 2876 }, { "epoch": 0.1, "learning_rate": 1.963701089052842e-05, "loss": 1.5903, "step": 2877 }, { "epoch": 0.1, "learning_rate": 1.963672842801934e-05, "loss": 1.4205, "step": 2878 }, { "epoch": 0.1, "learning_rate": 1.963644585768601e-05, "loss": 1.5946, "step": 2879 }, { "epoch": 0.1, "learning_rate": 1.9636163179531576e-05, "loss": 1.4754, "step": 2880 }, { "epoch": 0.1, "learning_rate": 1.963588039355922e-05, "loss": 1.5725, "step": 2881 }, { "epoch": 0.1, "learning_rate": 1.9635597499772096e-05, "loss": 1.6823, "step": 2882 }, { "epoch": 0.1, "learning_rate": 1.9635314498173364e-05, "loss": 1.7189, "step": 2883 }, { "epoch": 0.1, "learning_rate": 1.9635031388766204e-05, "loss": 1.5617, "step": 2884 }, { "epoch": 0.1, "learning_rate": 1.9634748171553775e-05, "loss": 1.4847, "step": 2885 }, { "epoch": 0.1, "learning_rate": 1.9634464846539244e-05, "loss": 1.64, "step": 2886 }, { "epoch": 0.1, "learning_rate": 1.963418141372579e-05, "loss": 1.5387, "step": 2887 }, { "epoch": 0.1, "learning_rate": 1.963389787311658e-05, "loss": 1.5362, "step": 2888 }, { "epoch": 0.1, "learning_rate": 1.963361422471478e-05, "loss": 1.5403, "step": 2889 }, { "epoch": 0.1, "learning_rate": 1.9633330468523572e-05, "loss": 1.5021, "step": 2890 }, { "epoch": 0.1, "learning_rate": 1.963304660454613e-05, "loss": 1.4219, "step": 2891 }, { "epoch": 0.1, "learning_rate": 1.963276263278563e-05, "loss": 1.4953, "step": 2892 }, { "epoch": 0.1, "learning_rate": 1.9632478553245243e-05, "loss": 1.5143, "step": 2893 }, { "epoch": 0.1, "learning_rate": 1.9632194365928152e-05, "loss": 1.5271, "step": 2894 }, { "epoch": 0.1, "learning_rate": 1.963191007083754e-05, "loss": 1.6852, "step": 2895 }, { "epoch": 0.1, "learning_rate": 1.9631625667976584e-05, "loss": 1.6001, "step": 2896 }, { "epoch": 0.1, "learning_rate": 1.9631341157348464e-05, "loss": 1.5633, "step": 2897 }, { "epoch": 0.1, "learning_rate": 1.963105653895637e-05, "loss": 1.564, "step": 2898 }, { "epoch": 0.1, "learning_rate": 1.9630771812803484e-05, "loss": 1.4516, "step": 2899 }, { "epoch": 0.1, "learning_rate": 1.9630486978892988e-05, "loss": 1.5928, "step": 2900 }, { "epoch": 0.1, "learning_rate": 1.9630202037228073e-05, "loss": 1.5174, "step": 2901 }, { "epoch": 0.1, "learning_rate": 1.9629916987811924e-05, "loss": 1.6556, "step": 2902 }, { "epoch": 0.1, "learning_rate": 1.9629631830647736e-05, "loss": 1.6053, "step": 2903 }, { "epoch": 0.1, "learning_rate": 1.9629346565738694e-05, "loss": 1.5278, "step": 2904 }, { "epoch": 0.1, "learning_rate": 1.962906119308799e-05, "loss": 1.5711, "step": 2905 }, { "epoch": 0.1, "learning_rate": 1.962877571269882e-05, "loss": 1.5568, "step": 2906 }, { "epoch": 0.1, "learning_rate": 1.962849012457438e-05, "loss": 1.5651, "step": 2907 }, { "epoch": 0.1, "learning_rate": 1.9628204428717856e-05, "loss": 1.4668, "step": 2908 }, { "epoch": 0.1, "learning_rate": 1.9627918625132455e-05, "loss": 1.5086, "step": 2909 }, { "epoch": 0.1, "learning_rate": 1.962763271382137e-05, "loss": 1.4814, "step": 2910 }, { "epoch": 0.1, "learning_rate": 1.9627346694787798e-05, "loss": 1.5876, "step": 2911 }, { "epoch": 0.1, "learning_rate": 1.962706056803495e-05, "loss": 1.5931, "step": 2912 }, { "epoch": 0.1, "learning_rate": 1.9626774333566012e-05, "loss": 1.5783, "step": 2913 }, { "epoch": 0.1, "learning_rate": 1.9626487991384194e-05, "loss": 1.6448, "step": 2914 }, { "epoch": 0.1, "learning_rate": 1.96262015414927e-05, "loss": 1.6516, "step": 2915 }, { "epoch": 0.1, "learning_rate": 1.962591498389474e-05, "loss": 1.6606, "step": 2916 }, { "epoch": 0.1, "learning_rate": 1.9625628318593514e-05, "loss": 1.582, "step": 2917 }, { "epoch": 0.1, "learning_rate": 1.9625341545592226e-05, "loss": 1.6109, "step": 2918 }, { "epoch": 0.1, "learning_rate": 1.9625054664894094e-05, "loss": 1.6285, "step": 2919 }, { "epoch": 0.1, "learning_rate": 1.962476767650232e-05, "loss": 1.5893, "step": 2920 }, { "epoch": 0.1, "learning_rate": 1.9624480580420122e-05, "loss": 1.5142, "step": 2921 }, { "epoch": 0.1, "learning_rate": 1.9624193376650708e-05, "loss": 1.6288, "step": 2922 }, { "epoch": 0.1, "learning_rate": 1.9623906065197288e-05, "loss": 1.5349, "step": 2923 }, { "epoch": 0.1, "learning_rate": 1.9623618646063087e-05, "loss": 1.5019, "step": 2924 }, { "epoch": 0.1, "learning_rate": 1.962333111925131e-05, "loss": 1.603, "step": 2925 }, { "epoch": 0.1, "learning_rate": 1.962304348476518e-05, "loss": 1.5174, "step": 2926 }, { "epoch": 0.1, "learning_rate": 1.9622755742607915e-05, "loss": 1.6111, "step": 2927 }, { "epoch": 0.1, "learning_rate": 1.9622467892782732e-05, "loss": 1.6266, "step": 2928 }, { "epoch": 0.1, "learning_rate": 1.9622179935292855e-05, "loss": 1.5473, "step": 2929 }, { "epoch": 0.1, "learning_rate": 1.9621891870141505e-05, "loss": 1.602, "step": 2930 }, { "epoch": 0.1, "learning_rate": 1.9621603697331902e-05, "loss": 1.7134, "step": 2931 }, { "epoch": 0.1, "learning_rate": 1.9621315416867274e-05, "loss": 1.5524, "step": 2932 }, { "epoch": 0.1, "learning_rate": 1.9621027028750844e-05, "loss": 1.615, "step": 2933 }, { "epoch": 0.1, "learning_rate": 1.962073853298584e-05, "loss": 1.6207, "step": 2934 }, { "epoch": 0.1, "learning_rate": 1.9620449929575495e-05, "loss": 1.6667, "step": 2935 }, { "epoch": 0.1, "learning_rate": 1.9620161218523027e-05, "loss": 1.5751, "step": 2936 }, { "epoch": 0.1, "learning_rate": 1.9619872399831676e-05, "loss": 1.6412, "step": 2937 }, { "epoch": 0.1, "learning_rate": 1.961958347350467e-05, "loss": 1.6573, "step": 2938 }, { "epoch": 0.1, "learning_rate": 1.9619294439545238e-05, "loss": 1.6406, "step": 2939 }, { "epoch": 0.1, "learning_rate": 1.9619005297956623e-05, "loss": 1.6411, "step": 2940 }, { "epoch": 0.1, "learning_rate": 1.961871604874205e-05, "loss": 1.6232, "step": 2941 }, { "epoch": 0.1, "learning_rate": 1.961842669190476e-05, "loss": 1.5402, "step": 2942 }, { "epoch": 0.1, "learning_rate": 1.9618137227447996e-05, "loss": 1.6424, "step": 2943 }, { "epoch": 0.1, "learning_rate": 1.9617847655374988e-05, "loss": 1.4757, "step": 2944 }, { "epoch": 0.1, "learning_rate": 1.961755797568898e-05, "loss": 1.5931, "step": 2945 }, { "epoch": 0.1, "learning_rate": 1.9617268188393214e-05, "loss": 1.4921, "step": 2946 }, { "epoch": 0.1, "learning_rate": 1.961697829349093e-05, "loss": 1.578, "step": 2947 }, { "epoch": 0.1, "learning_rate": 1.9616688290985374e-05, "loss": 1.5981, "step": 2948 }, { "epoch": 0.1, "learning_rate": 1.961639818087979e-05, "loss": 1.5611, "step": 2949 }, { "epoch": 0.1, "learning_rate": 1.9616107963177423e-05, "loss": 1.5667, "step": 2950 }, { "epoch": 0.1, "learning_rate": 1.961581763788152e-05, "loss": 1.5131, "step": 2951 }, { "epoch": 0.1, "learning_rate": 1.9615527204995332e-05, "loss": 1.5821, "step": 2952 }, { "epoch": 0.1, "learning_rate": 1.9615236664522108e-05, "loss": 1.5772, "step": 2953 }, { "epoch": 0.1, "learning_rate": 1.9614946016465094e-05, "loss": 1.6394, "step": 2954 }, { "epoch": 0.1, "learning_rate": 1.9614655260827547e-05, "loss": 1.5512, "step": 2955 }, { "epoch": 0.1, "learning_rate": 1.9614364397612723e-05, "loss": 1.3635, "step": 2956 }, { "epoch": 0.1, "learning_rate": 1.9614073426823868e-05, "loss": 1.5017, "step": 2957 }, { "epoch": 0.1, "learning_rate": 1.9613782348464244e-05, "loss": 1.52, "step": 2958 }, { "epoch": 0.1, "learning_rate": 1.9613491162537105e-05, "loss": 1.5487, "step": 2959 }, { "epoch": 0.1, "learning_rate": 1.961319986904571e-05, "loss": 1.6192, "step": 2960 }, { "epoch": 0.1, "learning_rate": 1.9612908467993318e-05, "loss": 1.5811, "step": 2961 }, { "epoch": 0.1, "learning_rate": 1.961261695938319e-05, "loss": 1.5818, "step": 2962 }, { "epoch": 0.1, "learning_rate": 1.9612325343218587e-05, "loss": 1.5924, "step": 2963 }, { "epoch": 0.1, "learning_rate": 1.9612033619502772e-05, "loss": 1.6479, "step": 2964 }, { "epoch": 0.1, "learning_rate": 1.961174178823901e-05, "loss": 1.5538, "step": 2965 }, { "epoch": 0.1, "learning_rate": 1.9611449849430566e-05, "loss": 1.5288, "step": 2966 }, { "epoch": 0.1, "learning_rate": 1.9611157803080708e-05, "loss": 1.5707, "step": 2967 }, { "epoch": 0.1, "learning_rate": 1.9610865649192695e-05, "loss": 1.482, "step": 2968 }, { "epoch": 0.1, "learning_rate": 1.961057338776981e-05, "loss": 1.5297, "step": 2969 }, { "epoch": 0.1, "learning_rate": 1.961028101881531e-05, "loss": 1.6547, "step": 2970 }, { "epoch": 0.1, "learning_rate": 1.9609988542332473e-05, "loss": 1.5572, "step": 2971 }, { "epoch": 0.1, "learning_rate": 1.960969595832457e-05, "loss": 1.6309, "step": 2972 }, { "epoch": 0.1, "learning_rate": 1.960940326679488e-05, "loss": 1.6454, "step": 2973 }, { "epoch": 0.1, "learning_rate": 1.960911046774667e-05, "loss": 1.6263, "step": 2974 }, { "epoch": 0.1, "learning_rate": 1.960881756118322e-05, "loss": 1.6321, "step": 2975 }, { "epoch": 0.1, "learning_rate": 1.9608524547107802e-05, "loss": 1.6175, "step": 2976 }, { "epoch": 0.1, "learning_rate": 1.9608231425523702e-05, "loss": 1.5155, "step": 2977 }, { "epoch": 0.1, "learning_rate": 1.9607938196434198e-05, "loss": 1.5279, "step": 2978 }, { "epoch": 0.1, "learning_rate": 1.9607644859842568e-05, "loss": 1.6217, "step": 2979 }, { "epoch": 0.1, "learning_rate": 1.9607351415752096e-05, "loss": 1.6428, "step": 2980 }, { "epoch": 0.1, "learning_rate": 1.9607057864166062e-05, "loss": 1.6084, "step": 2981 }, { "epoch": 0.1, "learning_rate": 1.960676420508776e-05, "loss": 1.5923, "step": 2982 }, { "epoch": 0.1, "learning_rate": 1.960647043852046e-05, "loss": 1.568, "step": 2983 }, { "epoch": 0.1, "learning_rate": 1.9606176564467465e-05, "loss": 1.5943, "step": 2984 }, { "epoch": 0.1, "learning_rate": 1.9605882582932054e-05, "loss": 1.6521, "step": 2985 }, { "epoch": 0.1, "learning_rate": 1.960558849391752e-05, "loss": 1.5016, "step": 2986 }, { "epoch": 0.1, "learning_rate": 1.960529429742715e-05, "loss": 1.5143, "step": 2987 }, { "epoch": 0.1, "learning_rate": 1.960499999346424e-05, "loss": 1.5473, "step": 2988 }, { "epoch": 0.1, "learning_rate": 1.960470558203208e-05, "loss": 1.5381, "step": 2989 }, { "epoch": 0.1, "learning_rate": 1.960441106313396e-05, "loss": 1.5969, "step": 2990 }, { "epoch": 0.1, "learning_rate": 1.9604116436773184e-05, "loss": 1.5801, "step": 2991 }, { "epoch": 0.1, "learning_rate": 1.9603821702953047e-05, "loss": 1.5477, "step": 2992 }, { "epoch": 0.1, "learning_rate": 1.960352686167684e-05, "loss": 1.6451, "step": 2993 }, { "epoch": 0.1, "learning_rate": 1.9603231912947873e-05, "loss": 1.5444, "step": 2994 }, { "epoch": 0.1, "learning_rate": 1.9602936856769432e-05, "loss": 1.4948, "step": 2995 }, { "epoch": 0.1, "learning_rate": 1.960264169314483e-05, "loss": 1.5939, "step": 2996 }, { "epoch": 0.1, "learning_rate": 1.9602346422077364e-05, "loss": 1.4766, "step": 2997 }, { "epoch": 0.1, "learning_rate": 1.960205104357034e-05, "loss": 1.6251, "step": 2998 }, { "epoch": 0.1, "learning_rate": 1.960175555762706e-05, "loss": 1.5403, "step": 2999 }, { "epoch": 0.1, "learning_rate": 1.9601459964250838e-05, "loss": 1.6249, "step": 3000 }, { "epoch": 0.1, "learning_rate": 1.960116426344497e-05, "loss": 1.6097, "step": 3001 }, { "epoch": 0.1, "learning_rate": 1.960086845521277e-05, "loss": 1.4657, "step": 3002 }, { "epoch": 0.1, "learning_rate": 1.960057253955755e-05, "loss": 1.615, "step": 3003 }, { "epoch": 0.1, "learning_rate": 1.9600276516482623e-05, "loss": 1.5616, "step": 3004 }, { "epoch": 0.1, "learning_rate": 1.9599980385991292e-05, "loss": 1.5606, "step": 3005 }, { "epoch": 0.1, "learning_rate": 1.9599684148086876e-05, "loss": 1.5637, "step": 3006 }, { "epoch": 0.1, "learning_rate": 1.9599387802772693e-05, "loss": 1.6604, "step": 3007 }, { "epoch": 0.1, "learning_rate": 1.959909135005205e-05, "loss": 1.4932, "step": 3008 }, { "epoch": 0.1, "learning_rate": 1.959879478992827e-05, "loss": 1.5855, "step": 3009 }, { "epoch": 0.1, "learning_rate": 1.9598498122404674e-05, "loss": 1.6236, "step": 3010 }, { "epoch": 0.1, "learning_rate": 1.959820134748457e-05, "loss": 1.5949, "step": 3011 }, { "epoch": 0.1, "learning_rate": 1.9597904465171293e-05, "loss": 1.4775, "step": 3012 }, { "epoch": 0.1, "learning_rate": 1.959760747546815e-05, "loss": 1.5379, "step": 3013 }, { "epoch": 0.1, "learning_rate": 1.9597310378378478e-05, "loss": 1.544, "step": 3014 }, { "epoch": 0.1, "learning_rate": 1.9597013173905596e-05, "loss": 1.667, "step": 3015 }, { "epoch": 0.1, "learning_rate": 1.9596715862052823e-05, "loss": 1.5571, "step": 3016 }, { "epoch": 0.1, "learning_rate": 1.9596418442823495e-05, "loss": 1.5353, "step": 3017 }, { "epoch": 0.1, "learning_rate": 1.9596120916220935e-05, "loss": 1.7241, "step": 3018 }, { "epoch": 0.1, "learning_rate": 1.959582328224847e-05, "loss": 1.5311, "step": 3019 }, { "epoch": 0.1, "learning_rate": 1.9595525540909434e-05, "loss": 1.5339, "step": 3020 }, { "epoch": 0.1, "learning_rate": 1.9595227692207158e-05, "loss": 1.507, "step": 3021 }, { "epoch": 0.1, "learning_rate": 1.9594929736144978e-05, "loss": 1.4961, "step": 3022 }, { "epoch": 0.1, "learning_rate": 1.9594631672726216e-05, "loss": 1.5116, "step": 3023 }, { "epoch": 0.1, "learning_rate": 1.9594333501954218e-05, "loss": 1.4368, "step": 3024 }, { "epoch": 0.1, "learning_rate": 1.959403522383232e-05, "loss": 1.557, "step": 3025 }, { "epoch": 0.1, "learning_rate": 1.9593736838363853e-05, "loss": 1.5223, "step": 3026 }, { "epoch": 0.1, "learning_rate": 1.9593438345552158e-05, "loss": 1.4661, "step": 3027 }, { "epoch": 0.1, "learning_rate": 1.9593139745400575e-05, "loss": 1.6464, "step": 3028 }, { "epoch": 0.1, "learning_rate": 1.9592841037912448e-05, "loss": 1.4649, "step": 3029 }, { "epoch": 0.1, "learning_rate": 1.9592542223091118e-05, "loss": 1.5872, "step": 3030 }, { "epoch": 0.1, "learning_rate": 1.9592243300939926e-05, "loss": 1.537, "step": 3031 }, { "epoch": 0.1, "learning_rate": 1.9591944271462216e-05, "loss": 1.5147, "step": 3032 }, { "epoch": 0.1, "learning_rate": 1.9591645134661336e-05, "loss": 1.4921, "step": 3033 }, { "epoch": 0.1, "learning_rate": 1.9591345890540635e-05, "loss": 1.5554, "step": 3034 }, { "epoch": 0.1, "learning_rate": 1.959104653910346e-05, "loss": 1.5813, "step": 3035 }, { "epoch": 0.1, "learning_rate": 1.9590747080353158e-05, "loss": 1.4686, "step": 3036 }, { "epoch": 0.1, "learning_rate": 1.959044751429308e-05, "loss": 1.5778, "step": 3037 }, { "epoch": 0.1, "learning_rate": 1.9590147840926578e-05, "loss": 1.4714, "step": 3038 }, { "epoch": 0.1, "learning_rate": 1.9589848060257007e-05, "loss": 1.5994, "step": 3039 }, { "epoch": 0.1, "learning_rate": 1.958954817228772e-05, "loss": 1.5907, "step": 3040 }, { "epoch": 0.1, "learning_rate": 1.958924817702207e-05, "loss": 1.6164, "step": 3041 }, { "epoch": 0.1, "learning_rate": 1.9588948074463418e-05, "loss": 1.5827, "step": 3042 }, { "epoch": 0.1, "learning_rate": 1.9588647864615118e-05, "loss": 1.605, "step": 3043 }, { "epoch": 0.1, "learning_rate": 1.9588347547480534e-05, "loss": 1.4873, "step": 3044 }, { "epoch": 0.1, "learning_rate": 1.958804712306302e-05, "loss": 1.5308, "step": 3045 }, { "epoch": 0.1, "learning_rate": 1.958774659136594e-05, "loss": 1.5354, "step": 3046 }, { "epoch": 0.1, "learning_rate": 1.958744595239266e-05, "loss": 1.5086, "step": 3047 }, { "epoch": 0.1, "learning_rate": 1.958714520614654e-05, "loss": 1.6526, "step": 3048 }, { "epoch": 0.1, "learning_rate": 1.9586844352630943e-05, "loss": 1.5603, "step": 3049 }, { "epoch": 0.1, "learning_rate": 1.9586543391849243e-05, "loss": 1.6478, "step": 3050 }, { "epoch": 0.1, "learning_rate": 1.95862423238048e-05, "loss": 1.6425, "step": 3051 }, { "epoch": 0.1, "learning_rate": 1.9585941148500987e-05, "loss": 1.7196, "step": 3052 }, { "epoch": 0.1, "learning_rate": 1.958563986594117e-05, "loss": 1.7978, "step": 3053 }, { "epoch": 0.1, "learning_rate": 1.958533847612872e-05, "loss": 1.7125, "step": 3054 }, { "epoch": 0.1, "learning_rate": 1.9585036979067015e-05, "loss": 1.757, "step": 3055 }, { "epoch": 0.1, "learning_rate": 1.9584735374759422e-05, "loss": 1.6268, "step": 3056 }, { "epoch": 0.1, "learning_rate": 1.958443366320932e-05, "loss": 1.5704, "step": 3057 }, { "epoch": 0.1, "learning_rate": 1.9584131844420084e-05, "loss": 1.619, "step": 3058 }, { "epoch": 0.1, "learning_rate": 1.9583829918395087e-05, "loss": 1.8038, "step": 3059 }, { "epoch": 0.1, "learning_rate": 1.958352788513771e-05, "loss": 1.585, "step": 3060 }, { "epoch": 0.1, "learning_rate": 1.9583225744651334e-05, "loss": 1.7085, "step": 3061 }, { "epoch": 0.1, "learning_rate": 1.9582923496939337e-05, "loss": 1.7252, "step": 3062 }, { "epoch": 0.1, "learning_rate": 1.9582621142005106e-05, "loss": 1.6485, "step": 3063 }, { "epoch": 0.1, "learning_rate": 1.9582318679852018e-05, "loss": 1.7464, "step": 3064 }, { "epoch": 0.1, "learning_rate": 1.9582016110483458e-05, "loss": 1.7622, "step": 3065 }, { "epoch": 0.1, "learning_rate": 1.9581713433902813e-05, "loss": 1.6171, "step": 3066 }, { "epoch": 0.1, "learning_rate": 1.958141065011347e-05, "loss": 1.7394, "step": 3067 }, { "epoch": 0.1, "learning_rate": 1.9581107759118815e-05, "loss": 1.5763, "step": 3068 }, { "epoch": 0.1, "learning_rate": 1.958080476092224e-05, "loss": 1.5998, "step": 3069 }, { "epoch": 0.1, "learning_rate": 1.9580501655527132e-05, "loss": 1.5229, "step": 3070 }, { "epoch": 0.1, "learning_rate": 1.9580198442936886e-05, "loss": 1.6364, "step": 3071 }, { "epoch": 0.1, "learning_rate": 1.957989512315489e-05, "loss": 1.5885, "step": 3072 }, { "epoch": 0.1, "learning_rate": 1.957959169618454e-05, "loss": 1.6106, "step": 3073 }, { "epoch": 0.1, "learning_rate": 1.9579288162029235e-05, "loss": 1.6097, "step": 3074 }, { "epoch": 0.1, "learning_rate": 1.9578984520692362e-05, "loss": 1.6324, "step": 3075 }, { "epoch": 0.1, "learning_rate": 1.9578680772177327e-05, "loss": 1.5621, "step": 3076 }, { "epoch": 0.1, "learning_rate": 1.9578376916487528e-05, "loss": 1.5343, "step": 3077 }, { "epoch": 0.1, "learning_rate": 1.9578072953626357e-05, "loss": 1.4755, "step": 3078 }, { "epoch": 0.1, "learning_rate": 1.9577768883597225e-05, "loss": 1.5991, "step": 3079 }, { "epoch": 0.1, "learning_rate": 1.9577464706403525e-05, "loss": 1.5761, "step": 3080 }, { "epoch": 0.1, "learning_rate": 1.9577160422048667e-05, "loss": 1.5332, "step": 3081 }, { "epoch": 0.1, "learning_rate": 1.9576856030536055e-05, "loss": 1.521, "step": 3082 }, { "epoch": 0.1, "learning_rate": 1.9576551531869092e-05, "loss": 1.616, "step": 3083 }, { "epoch": 0.1, "learning_rate": 1.9576246926051187e-05, "loss": 1.5109, "step": 3084 }, { "epoch": 0.1, "learning_rate": 1.9575942213085746e-05, "loss": 1.4295, "step": 3085 }, { "epoch": 0.1, "learning_rate": 1.957563739297618e-05, "loss": 1.5753, "step": 3086 }, { "epoch": 0.1, "learning_rate": 1.9575332465725897e-05, "loss": 1.5995, "step": 3087 }, { "epoch": 0.1, "learning_rate": 1.9575027431338317e-05, "loss": 1.5522, "step": 3088 }, { "epoch": 0.1, "learning_rate": 1.9574722289816847e-05, "loss": 1.6233, "step": 3089 }, { "epoch": 0.1, "learning_rate": 1.9574417041164893e-05, "loss": 1.5508, "step": 3090 }, { "epoch": 0.1, "learning_rate": 1.9574111685385887e-05, "loss": 1.4625, "step": 3091 }, { "epoch": 0.1, "learning_rate": 1.9573806222483234e-05, "loss": 1.5512, "step": 3092 }, { "epoch": 0.1, "learning_rate": 1.9573500652460357e-05, "loss": 1.5341, "step": 3093 }, { "epoch": 0.1, "learning_rate": 1.9573194975320672e-05, "loss": 1.5352, "step": 3094 }, { "epoch": 0.1, "learning_rate": 1.9572889191067604e-05, "loss": 1.5888, "step": 3095 }, { "epoch": 0.1, "learning_rate": 1.957258329970457e-05, "loss": 1.5299, "step": 3096 }, { "epoch": 0.1, "learning_rate": 1.957227730123499e-05, "loss": 1.5613, "step": 3097 }, { "epoch": 0.1, "learning_rate": 1.9571971195662292e-05, "loss": 1.5834, "step": 3098 }, { "epoch": 0.1, "learning_rate": 1.95716649829899e-05, "loss": 1.5983, "step": 3099 }, { "epoch": 0.1, "learning_rate": 1.957135866322124e-05, "loss": 1.5647, "step": 3100 }, { "epoch": 0.1, "learning_rate": 1.957105223635974e-05, "loss": 1.5066, "step": 3101 }, { "epoch": 0.1, "learning_rate": 1.957074570240883e-05, "loss": 1.4618, "step": 3102 }, { "epoch": 0.1, "learning_rate": 1.9570439061371936e-05, "loss": 1.5934, "step": 3103 }, { "epoch": 0.1, "learning_rate": 1.957013231325249e-05, "loss": 1.5405, "step": 3104 }, { "epoch": 0.1, "learning_rate": 1.956982545805393e-05, "loss": 1.571, "step": 3105 }, { "epoch": 0.1, "learning_rate": 1.9569518495779682e-05, "loss": 1.6262, "step": 3106 }, { "epoch": 0.1, "learning_rate": 1.9569211426433183e-05, "loss": 1.5408, "step": 3107 }, { "epoch": 0.1, "learning_rate": 1.9568904250017867e-05, "loss": 1.5926, "step": 3108 }, { "epoch": 0.1, "learning_rate": 1.9568596966537177e-05, "loss": 1.561, "step": 3109 }, { "epoch": 0.1, "learning_rate": 1.9568289575994544e-05, "loss": 1.6652, "step": 3110 }, { "epoch": 0.1, "learning_rate": 1.956798207839341e-05, "loss": 1.5973, "step": 3111 }, { "epoch": 0.1, "learning_rate": 1.956767447373722e-05, "loss": 1.5764, "step": 3112 }, { "epoch": 0.1, "learning_rate": 1.956736676202941e-05, "loss": 1.6419, "step": 3113 }, { "epoch": 0.1, "learning_rate": 1.956705894327342e-05, "loss": 1.5253, "step": 3114 }, { "epoch": 0.1, "learning_rate": 1.9566751017472704e-05, "loss": 1.5697, "step": 3115 }, { "epoch": 0.1, "learning_rate": 1.95664429846307e-05, "loss": 1.645, "step": 3116 }, { "epoch": 0.1, "learning_rate": 1.9566134844750856e-05, "loss": 1.7054, "step": 3117 }, { "epoch": 0.1, "learning_rate": 1.9565826597836623e-05, "loss": 1.6125, "step": 3118 }, { "epoch": 0.1, "learning_rate": 1.9565518243891445e-05, "loss": 1.6816, "step": 3119 }, { "epoch": 0.1, "learning_rate": 1.9565209782918772e-05, "loss": 1.6006, "step": 3120 }, { "epoch": 0.1, "learning_rate": 1.9564901214922063e-05, "loss": 1.5453, "step": 3121 }, { "epoch": 0.1, "learning_rate": 1.956459253990476e-05, "loss": 1.5768, "step": 3122 }, { "epoch": 0.1, "learning_rate": 1.9564283757870327e-05, "loss": 1.6599, "step": 3123 }, { "epoch": 0.1, "learning_rate": 1.9563974868822212e-05, "loss": 1.49, "step": 3124 }, { "epoch": 0.1, "learning_rate": 1.9563665872763875e-05, "loss": 1.5717, "step": 3125 }, { "epoch": 0.1, "learning_rate": 1.956335676969877e-05, "loss": 1.6084, "step": 3126 }, { "epoch": 0.1, "learning_rate": 1.9563047559630356e-05, "loss": 1.5965, "step": 3127 }, { "epoch": 0.1, "learning_rate": 1.9562738242562097e-05, "loss": 1.7026, "step": 3128 }, { "epoch": 0.1, "learning_rate": 1.9562428818497445e-05, "loss": 1.6555, "step": 3129 }, { "epoch": 0.1, "learning_rate": 1.9562119287439874e-05, "loss": 1.4805, "step": 3130 }, { "epoch": 0.1, "learning_rate": 1.9561809649392836e-05, "loss": 1.6776, "step": 3131 }, { "epoch": 0.1, "learning_rate": 1.9561499904359805e-05, "loss": 1.6051, "step": 3132 }, { "epoch": 0.1, "learning_rate": 1.956119005234424e-05, "loss": 1.7033, "step": 3133 }, { "epoch": 0.1, "learning_rate": 1.9560880093349607e-05, "loss": 1.7107, "step": 3134 }, { "epoch": 0.1, "learning_rate": 1.956057002737938e-05, "loss": 1.595, "step": 3135 }, { "epoch": 0.1, "learning_rate": 1.9560259854437026e-05, "loss": 1.7072, "step": 3136 }, { "epoch": 0.1, "learning_rate": 1.9559949574526015e-05, "loss": 1.6213, "step": 3137 }, { "epoch": 0.1, "learning_rate": 1.9559639187649817e-05, "loss": 1.5528, "step": 3138 }, { "epoch": 0.1, "learning_rate": 1.955932869381191e-05, "loss": 1.7263, "step": 3139 }, { "epoch": 0.1, "learning_rate": 1.955901809301576e-05, "loss": 1.5441, "step": 3140 }, { "epoch": 0.1, "learning_rate": 1.955870738526485e-05, "loss": 1.598, "step": 3141 }, { "epoch": 0.1, "learning_rate": 1.955839657056265e-05, "loss": 1.6032, "step": 3142 }, { "epoch": 0.1, "learning_rate": 1.955808564891265e-05, "loss": 1.4409, "step": 3143 }, { "epoch": 0.1, "learning_rate": 1.955777462031831e-05, "loss": 1.5379, "step": 3144 }, { "epoch": 0.1, "learning_rate": 1.9557463484783125e-05, "loss": 1.4475, "step": 3145 }, { "epoch": 0.1, "learning_rate": 1.955715224231057e-05, "loss": 1.6884, "step": 3146 }, { "epoch": 0.1, "learning_rate": 1.9556840892904127e-05, "loss": 1.57, "step": 3147 }, { "epoch": 0.1, "learning_rate": 1.9556529436567287e-05, "loss": 1.5417, "step": 3148 }, { "epoch": 0.1, "learning_rate": 1.9556217873303526e-05, "loss": 1.5857, "step": 3149 }, { "epoch": 0.1, "learning_rate": 1.955590620311633e-05, "loss": 1.5129, "step": 3150 }, { "epoch": 0.11, "learning_rate": 1.9555594426009193e-05, "loss": 1.6084, "step": 3151 }, { "epoch": 0.11, "learning_rate": 1.9555282541985598e-05, "loss": 1.5124, "step": 3152 }, { "epoch": 0.11, "learning_rate": 1.9554970551049037e-05, "loss": 1.5601, "step": 3153 }, { "epoch": 0.11, "learning_rate": 1.9554658453203003e-05, "loss": 1.4732, "step": 3154 }, { "epoch": 0.11, "learning_rate": 1.9554346248450985e-05, "loss": 1.4837, "step": 3155 }, { "epoch": 0.11, "learning_rate": 1.9554033936796474e-05, "loss": 1.6057, "step": 3156 }, { "epoch": 0.11, "learning_rate": 1.955372151824297e-05, "loss": 1.7003, "step": 3157 }, { "epoch": 0.11, "learning_rate": 1.955340899279396e-05, "loss": 1.5404, "step": 3158 }, { "epoch": 0.11, "learning_rate": 1.9553096360452953e-05, "loss": 1.5143, "step": 3159 }, { "epoch": 0.11, "learning_rate": 1.9552783621223437e-05, "loss": 1.5641, "step": 3160 }, { "epoch": 0.11, "learning_rate": 1.9552470775108917e-05, "loss": 1.5243, "step": 3161 }, { "epoch": 0.11, "learning_rate": 1.955215782211289e-05, "loss": 1.5265, "step": 3162 }, { "epoch": 0.11, "learning_rate": 1.955184476223886e-05, "loss": 1.6533, "step": 3163 }, { "epoch": 0.11, "learning_rate": 1.9551531595490324e-05, "loss": 1.5132, "step": 3164 }, { "epoch": 0.11, "learning_rate": 1.9551218321870795e-05, "loss": 1.697, "step": 3165 }, { "epoch": 0.11, "learning_rate": 1.955090494138377e-05, "loss": 1.562, "step": 3166 }, { "epoch": 0.11, "learning_rate": 1.9550591454032764e-05, "loss": 1.6377, "step": 3167 }, { "epoch": 0.11, "learning_rate": 1.955027785982128e-05, "loss": 1.6095, "step": 3168 }, { "epoch": 0.11, "learning_rate": 1.9549964158752825e-05, "loss": 1.6102, "step": 3169 }, { "epoch": 0.11, "learning_rate": 1.9549650350830903e-05, "loss": 1.6917, "step": 3170 }, { "epoch": 0.11, "learning_rate": 1.954933643605904e-05, "loss": 1.5139, "step": 3171 }, { "epoch": 0.11, "learning_rate": 1.9549022414440738e-05, "loss": 1.5998, "step": 3172 }, { "epoch": 0.11, "learning_rate": 1.954870828597952e-05, "loss": 1.6092, "step": 3173 }, { "epoch": 0.11, "learning_rate": 1.9548394050678883e-05, "loss": 1.5454, "step": 3174 }, { "epoch": 0.11, "learning_rate": 1.9548079708542365e-05, "loss": 1.5096, "step": 3175 }, { "epoch": 0.11, "learning_rate": 1.9547765259573467e-05, "loss": 1.4835, "step": 3176 }, { "epoch": 0.11, "learning_rate": 1.9547450703775714e-05, "loss": 1.5207, "step": 3177 }, { "epoch": 0.11, "learning_rate": 1.954713604115262e-05, "loss": 1.4381, "step": 3178 }, { "epoch": 0.11, "learning_rate": 1.9546821271707713e-05, "loss": 1.5986, "step": 3179 }, { "epoch": 0.11, "learning_rate": 1.9546506395444514e-05, "loss": 1.5669, "step": 3180 }, { "epoch": 0.11, "learning_rate": 1.9546191412366543e-05, "loss": 1.5831, "step": 3181 }, { "epoch": 0.11, "learning_rate": 1.954587632247732e-05, "loss": 1.5736, "step": 3182 }, { "epoch": 0.11, "learning_rate": 1.954556112578038e-05, "loss": 1.636, "step": 3183 }, { "epoch": 0.11, "learning_rate": 1.9545245822279243e-05, "loss": 1.5302, "step": 3184 }, { "epoch": 0.11, "learning_rate": 1.9544930411977444e-05, "loss": 1.5406, "step": 3185 }, { "epoch": 0.11, "learning_rate": 1.9544614894878505e-05, "loss": 1.552, "step": 3186 }, { "epoch": 0.11, "learning_rate": 1.9544299270985958e-05, "loss": 1.5266, "step": 3187 }, { "epoch": 0.11, "learning_rate": 1.9543983540303335e-05, "loss": 1.6037, "step": 3188 }, { "epoch": 0.11, "learning_rate": 1.954366770283417e-05, "loss": 1.6596, "step": 3189 }, { "epoch": 0.11, "learning_rate": 1.9543351758581995e-05, "loss": 1.6668, "step": 3190 }, { "epoch": 0.11, "learning_rate": 1.9543035707550343e-05, "loss": 1.6448, "step": 3191 }, { "epoch": 0.11, "learning_rate": 1.9542719549742757e-05, "loss": 1.6064, "step": 3192 }, { "epoch": 0.11, "learning_rate": 1.954240328516277e-05, "loss": 1.572, "step": 3193 }, { "epoch": 0.11, "learning_rate": 1.954208691381392e-05, "loss": 1.5664, "step": 3194 }, { "epoch": 0.11, "learning_rate": 1.954177043569975e-05, "loss": 1.5717, "step": 3195 }, { "epoch": 0.11, "learning_rate": 1.9541453850823796e-05, "loss": 1.5052, "step": 3196 }, { "epoch": 0.11, "learning_rate": 1.9541137159189606e-05, "loss": 1.4548, "step": 3197 }, { "epoch": 0.11, "learning_rate": 1.954082036080072e-05, "loss": 1.5792, "step": 3198 }, { "epoch": 0.11, "learning_rate": 1.954050345566068e-05, "loss": 1.5265, "step": 3199 }, { "epoch": 0.11, "learning_rate": 1.9540186443773038e-05, "loss": 1.4333, "step": 3200 }, { "epoch": 0.11, "learning_rate": 1.953986932514134e-05, "loss": 1.5623, "step": 3201 }, { "epoch": 0.11, "learning_rate": 1.9539552099769128e-05, "loss": 1.5401, "step": 3202 }, { "epoch": 0.11, "learning_rate": 1.9539234767659957e-05, "loss": 1.5237, "step": 3203 }, { "epoch": 0.11, "learning_rate": 1.9538917328817377e-05, "loss": 1.5214, "step": 3204 }, { "epoch": 0.11, "learning_rate": 1.953859978324494e-05, "loss": 1.4559, "step": 3205 }, { "epoch": 0.11, "learning_rate": 1.9538282130946198e-05, "loss": 1.5173, "step": 3206 }, { "epoch": 0.11, "learning_rate": 1.9537964371924706e-05, "loss": 1.4466, "step": 3207 }, { "epoch": 0.11, "learning_rate": 1.9537646506184016e-05, "loss": 1.4874, "step": 3208 }, { "epoch": 0.11, "learning_rate": 1.9537328533727688e-05, "loss": 1.5317, "step": 3209 }, { "epoch": 0.11, "learning_rate": 1.953701045455928e-05, "loss": 1.4526, "step": 3210 }, { "epoch": 0.11, "learning_rate": 1.9536692268682348e-05, "loss": 1.5219, "step": 3211 }, { "epoch": 0.11, "learning_rate": 1.9536373976100454e-05, "loss": 1.4866, "step": 3212 }, { "epoch": 0.11, "learning_rate": 1.953605557681716e-05, "loss": 1.5579, "step": 3213 }, { "epoch": 0.11, "learning_rate": 1.9535737070836028e-05, "loss": 1.5924, "step": 3214 }, { "epoch": 0.11, "learning_rate": 1.9535418458160625e-05, "loss": 1.6403, "step": 3215 }, { "epoch": 0.11, "learning_rate": 1.953509973879451e-05, "loss": 1.5999, "step": 3216 }, { "epoch": 0.11, "learning_rate": 1.953478091274125e-05, "loss": 1.4633, "step": 3217 }, { "epoch": 0.11, "learning_rate": 1.9534461980004415e-05, "loss": 1.5709, "step": 3218 }, { "epoch": 0.11, "learning_rate": 1.9534142940587572e-05, "loss": 1.6709, "step": 3219 }, { "epoch": 0.11, "learning_rate": 1.9533823794494294e-05, "loss": 1.7166, "step": 3220 }, { "epoch": 0.11, "learning_rate": 1.9533504541728144e-05, "loss": 1.6622, "step": 3221 }, { "epoch": 0.11, "learning_rate": 1.9533185182292705e-05, "loss": 1.5889, "step": 3222 }, { "epoch": 0.11, "learning_rate": 1.953286571619154e-05, "loss": 1.6888, "step": 3223 }, { "epoch": 0.11, "learning_rate": 1.953254614342823e-05, "loss": 1.7558, "step": 3224 }, { "epoch": 0.11, "learning_rate": 1.9532226464006346e-05, "loss": 1.7023, "step": 3225 }, { "epoch": 0.11, "learning_rate": 1.9531906677929472e-05, "loss": 1.6774, "step": 3226 }, { "epoch": 0.11, "learning_rate": 1.9531586785201178e-05, "loss": 1.6651, "step": 3227 }, { "epoch": 0.11, "learning_rate": 1.9531266785825046e-05, "loss": 1.7124, "step": 3228 }, { "epoch": 0.11, "learning_rate": 1.953094667980466e-05, "loss": 1.7232, "step": 3229 }, { "epoch": 0.11, "learning_rate": 1.9530626467143597e-05, "loss": 1.4859, "step": 3230 }, { "epoch": 0.11, "learning_rate": 1.9530306147845445e-05, "loss": 1.5915, "step": 3231 }, { "epoch": 0.11, "learning_rate": 1.952998572191378e-05, "loss": 1.7152, "step": 3232 }, { "epoch": 0.11, "learning_rate": 1.9529665189352194e-05, "loss": 1.7121, "step": 3233 }, { "epoch": 0.11, "learning_rate": 1.952934455016427e-05, "loss": 1.6762, "step": 3234 }, { "epoch": 0.11, "learning_rate": 1.9529023804353598e-05, "loss": 1.6889, "step": 3235 }, { "epoch": 0.11, "learning_rate": 1.9528702951923765e-05, "loss": 1.7325, "step": 3236 }, { "epoch": 0.11, "learning_rate": 1.9528381992878362e-05, "loss": 1.6616, "step": 3237 }, { "epoch": 0.11, "learning_rate": 1.952806092722098e-05, "loss": 1.6746, "step": 3238 }, { "epoch": 0.11, "learning_rate": 1.952773975495521e-05, "loss": 1.6382, "step": 3239 }, { "epoch": 0.11, "learning_rate": 1.952741847608465e-05, "loss": 1.5027, "step": 3240 }, { "epoch": 0.11, "learning_rate": 1.9527097090612888e-05, "loss": 1.654, "step": 3241 }, { "epoch": 0.11, "learning_rate": 1.9526775598543522e-05, "loss": 1.5507, "step": 3242 }, { "epoch": 0.11, "learning_rate": 1.9526453999880153e-05, "loss": 1.6318, "step": 3243 }, { "epoch": 0.11, "learning_rate": 1.9526132294626377e-05, "loss": 1.6523, "step": 3244 }, { "epoch": 0.11, "learning_rate": 1.9525810482785794e-05, "loss": 1.5497, "step": 3245 }, { "epoch": 0.11, "learning_rate": 1.9525488564362e-05, "loss": 1.6382, "step": 3246 }, { "epoch": 0.11, "learning_rate": 1.9525166539358608e-05, "loss": 1.5928, "step": 3247 }, { "epoch": 0.11, "learning_rate": 1.9524844407779208e-05, "loss": 1.5861, "step": 3248 }, { "epoch": 0.11, "learning_rate": 1.9524522169627414e-05, "loss": 1.6012, "step": 3249 }, { "epoch": 0.11, "learning_rate": 1.9524199824906826e-05, "loss": 1.6625, "step": 3250 }, { "epoch": 0.11, "learning_rate": 1.952387737362105e-05, "loss": 1.5588, "step": 3251 }, { "epoch": 0.11, "learning_rate": 1.9523554815773703e-05, "loss": 1.6876, "step": 3252 }, { "epoch": 0.11, "learning_rate": 1.9523232151368383e-05, "loss": 1.6401, "step": 3253 }, { "epoch": 0.11, "learning_rate": 1.9522909380408704e-05, "loss": 1.5358, "step": 3254 }, { "epoch": 0.11, "learning_rate": 1.952258650289828e-05, "loss": 1.6471, "step": 3255 }, { "epoch": 0.11, "learning_rate": 1.952226351884072e-05, "loss": 1.607, "step": 3256 }, { "epoch": 0.11, "learning_rate": 1.9521940428239644e-05, "loss": 1.5697, "step": 3257 }, { "epoch": 0.11, "learning_rate": 1.952161723109866e-05, "loss": 1.5607, "step": 3258 }, { "epoch": 0.11, "learning_rate": 1.9521293927421388e-05, "loss": 1.4533, "step": 3259 }, { "epoch": 0.11, "learning_rate": 1.9520970517211443e-05, "loss": 1.5366, "step": 3260 }, { "epoch": 0.11, "learning_rate": 1.9520647000472447e-05, "loss": 1.5478, "step": 3261 }, { "epoch": 0.11, "learning_rate": 1.9520323377208017e-05, "loss": 1.5884, "step": 3262 }, { "epoch": 0.11, "learning_rate": 1.9519999647421773e-05, "loss": 1.5253, "step": 3263 }, { "epoch": 0.11, "learning_rate": 1.9519675811117338e-05, "loss": 1.4831, "step": 3264 }, { "epoch": 0.11, "learning_rate": 1.9519351868298337e-05, "loss": 1.548, "step": 3265 }, { "epoch": 0.11, "learning_rate": 1.9519027818968396e-05, "loss": 1.4872, "step": 3266 }, { "epoch": 0.11, "learning_rate": 1.951870366313114e-05, "loss": 1.5776, "step": 3267 }, { "epoch": 0.11, "learning_rate": 1.9518379400790192e-05, "loss": 1.6053, "step": 3268 }, { "epoch": 0.11, "learning_rate": 1.951805503194918e-05, "loss": 1.6046, "step": 3269 }, { "epoch": 0.11, "learning_rate": 1.951773055661174e-05, "loss": 1.5375, "step": 3270 }, { "epoch": 0.11, "learning_rate": 1.9517405974781495e-05, "loss": 1.5401, "step": 3271 }, { "epoch": 0.11, "learning_rate": 1.9517081286462082e-05, "loss": 1.6118, "step": 3272 }, { "epoch": 0.11, "learning_rate": 1.9516756491657132e-05, "loss": 1.561, "step": 3273 }, { "epoch": 0.11, "learning_rate": 1.951643159037028e-05, "loss": 1.5959, "step": 3274 }, { "epoch": 0.11, "learning_rate": 1.9516106582605156e-05, "loss": 1.5077, "step": 3275 }, { "epoch": 0.11, "learning_rate": 1.9515781468365406e-05, "loss": 1.6632, "step": 3276 }, { "epoch": 0.11, "learning_rate": 1.951545624765466e-05, "loss": 1.5043, "step": 3277 }, { "epoch": 0.11, "learning_rate": 1.9515130920476563e-05, "loss": 1.6169, "step": 3278 }, { "epoch": 0.11, "learning_rate": 1.951480548683475e-05, "loss": 1.5101, "step": 3279 }, { "epoch": 0.11, "learning_rate": 1.951447994673286e-05, "loss": 1.5937, "step": 3280 }, { "epoch": 0.11, "learning_rate": 1.9514154300174542e-05, "loss": 1.604, "step": 3281 }, { "epoch": 0.11, "learning_rate": 1.951382854716344e-05, "loss": 1.8234, "step": 3282 }, { "epoch": 0.11, "learning_rate": 1.951350268770319e-05, "loss": 1.6392, "step": 3283 }, { "epoch": 0.11, "learning_rate": 1.9513176721797445e-05, "loss": 1.602, "step": 3284 }, { "epoch": 0.11, "learning_rate": 1.951285064944985e-05, "loss": 1.573, "step": 3285 }, { "epoch": 0.11, "learning_rate": 1.9512524470664058e-05, "loss": 1.6012, "step": 3286 }, { "epoch": 0.11, "learning_rate": 1.9512198185443712e-05, "loss": 1.5161, "step": 3287 }, { "epoch": 0.11, "learning_rate": 1.9511871793792467e-05, "loss": 1.6632, "step": 3288 }, { "epoch": 0.11, "learning_rate": 1.9511545295713975e-05, "loss": 1.6081, "step": 3289 }, { "epoch": 0.11, "learning_rate": 1.9511218691211885e-05, "loss": 1.5738, "step": 3290 }, { "epoch": 0.11, "learning_rate": 1.9510891980289853e-05, "loss": 1.5499, "step": 3291 }, { "epoch": 0.11, "learning_rate": 1.9510565162951538e-05, "loss": 1.5264, "step": 3292 }, { "epoch": 0.11, "learning_rate": 1.9510238239200592e-05, "loss": 1.4763, "step": 3293 }, { "epoch": 0.11, "learning_rate": 1.950991120904068e-05, "loss": 1.5632, "step": 3294 }, { "epoch": 0.11, "learning_rate": 1.950958407247545e-05, "loss": 1.5881, "step": 3295 }, { "epoch": 0.11, "learning_rate": 1.950925682950857e-05, "loss": 1.6959, "step": 3296 }, { "epoch": 0.11, "learning_rate": 1.9508929480143704e-05, "loss": 1.6873, "step": 3297 }, { "epoch": 0.11, "learning_rate": 1.9508602024384507e-05, "loss": 1.6488, "step": 3298 }, { "epoch": 0.11, "learning_rate": 1.9508274462234648e-05, "loss": 1.5285, "step": 3299 }, { "epoch": 0.11, "learning_rate": 1.950794679369779e-05, "loss": 1.6912, "step": 3300 }, { "epoch": 0.11, "learning_rate": 1.9507619018777597e-05, "loss": 1.612, "step": 3301 }, { "epoch": 0.11, "learning_rate": 1.9507291137477744e-05, "loss": 1.6516, "step": 3302 }, { "epoch": 0.11, "learning_rate": 1.9506963149801894e-05, "loss": 1.6637, "step": 3303 }, { "epoch": 0.11, "learning_rate": 1.9506635055753714e-05, "loss": 1.7245, "step": 3304 }, { "epoch": 0.11, "learning_rate": 1.9506306855336885e-05, "loss": 1.6262, "step": 3305 }, { "epoch": 0.11, "learning_rate": 1.9505978548555066e-05, "loss": 1.5875, "step": 3306 }, { "epoch": 0.11, "learning_rate": 1.950565013541194e-05, "loss": 1.6491, "step": 3307 }, { "epoch": 0.11, "learning_rate": 1.9505321615911183e-05, "loss": 1.6241, "step": 3308 }, { "epoch": 0.11, "learning_rate": 1.9504992990056465e-05, "loss": 1.6496, "step": 3309 }, { "epoch": 0.11, "learning_rate": 1.9504664257851464e-05, "loss": 1.6543, "step": 3310 }, { "epoch": 0.11, "learning_rate": 1.9504335419299857e-05, "loss": 1.6467, "step": 3311 }, { "epoch": 0.11, "learning_rate": 1.9504006474405327e-05, "loss": 1.6845, "step": 3312 }, { "epoch": 0.11, "learning_rate": 1.950367742317155e-05, "loss": 1.7151, "step": 3313 }, { "epoch": 0.11, "learning_rate": 1.9503348265602212e-05, "loss": 1.5955, "step": 3314 }, { "epoch": 0.11, "learning_rate": 1.9503019001700996e-05, "loss": 1.5036, "step": 3315 }, { "epoch": 0.11, "learning_rate": 1.9502689631471582e-05, "loss": 1.5013, "step": 3316 }, { "epoch": 0.11, "learning_rate": 1.950236015491766e-05, "loss": 1.5832, "step": 3317 }, { "epoch": 0.11, "learning_rate": 1.950203057204291e-05, "loss": 1.5309, "step": 3318 }, { "epoch": 0.11, "learning_rate": 1.950170088285103e-05, "loss": 1.5974, "step": 3319 }, { "epoch": 0.11, "learning_rate": 1.95013710873457e-05, "loss": 1.5502, "step": 3320 }, { "epoch": 0.11, "learning_rate": 1.9501041185530608e-05, "loss": 1.5316, "step": 3321 }, { "epoch": 0.11, "learning_rate": 1.9500711177409456e-05, "loss": 1.5597, "step": 3322 }, { "epoch": 0.11, "learning_rate": 1.9500381062985928e-05, "loss": 1.6937, "step": 3323 }, { "epoch": 0.11, "learning_rate": 1.950005084226372e-05, "loss": 1.5502, "step": 3324 }, { "epoch": 0.11, "learning_rate": 1.9499720515246524e-05, "loss": 1.5115, "step": 3325 }, { "epoch": 0.11, "learning_rate": 1.9499390081938045e-05, "loss": 1.5322, "step": 3326 }, { "epoch": 0.11, "learning_rate": 1.9499059542341968e-05, "loss": 1.5604, "step": 3327 }, { "epoch": 0.11, "learning_rate": 1.9498728896462002e-05, "loss": 1.5931, "step": 3328 }, { "epoch": 0.11, "learning_rate": 1.949839814430184e-05, "loss": 1.5936, "step": 3329 }, { "epoch": 0.11, "learning_rate": 1.9498067285865185e-05, "loss": 1.527, "step": 3330 }, { "epoch": 0.11, "learning_rate": 1.9497736321155737e-05, "loss": 1.6508, "step": 3331 }, { "epoch": 0.11, "learning_rate": 1.9497405250177203e-05, "loss": 1.5217, "step": 3332 }, { "epoch": 0.11, "learning_rate": 1.9497074072933283e-05, "loss": 1.6319, "step": 3333 }, { "epoch": 0.11, "learning_rate": 1.9496742789427684e-05, "loss": 1.5521, "step": 3334 }, { "epoch": 0.11, "learning_rate": 1.9496411399664116e-05, "loss": 1.5791, "step": 3335 }, { "epoch": 0.11, "learning_rate": 1.9496079903646282e-05, "loss": 1.5031, "step": 3336 }, { "epoch": 0.11, "learning_rate": 1.9495748301377895e-05, "loss": 1.5176, "step": 3337 }, { "epoch": 0.11, "learning_rate": 1.9495416592862663e-05, "loss": 1.491, "step": 3338 }, { "epoch": 0.11, "learning_rate": 1.9495084778104295e-05, "loss": 1.4764, "step": 3339 }, { "epoch": 0.11, "learning_rate": 1.949475285710651e-05, "loss": 1.5853, "step": 3340 }, { "epoch": 0.11, "learning_rate": 1.9494420829873014e-05, "loss": 1.5568, "step": 3341 }, { "epoch": 0.11, "learning_rate": 1.9494088696407532e-05, "loss": 1.5318, "step": 3342 }, { "epoch": 0.11, "learning_rate": 1.949375645671377e-05, "loss": 1.5468, "step": 3343 }, { "epoch": 0.11, "learning_rate": 1.9493424110795455e-05, "loss": 1.5641, "step": 3344 }, { "epoch": 0.11, "learning_rate": 1.9493091658656293e-05, "loss": 1.5344, "step": 3345 }, { "epoch": 0.11, "learning_rate": 1.949275910030002e-05, "loss": 1.5254, "step": 3346 }, { "epoch": 0.11, "learning_rate": 1.949242643573034e-05, "loss": 1.4669, "step": 3347 }, { "epoch": 0.11, "learning_rate": 1.9492093664950988e-05, "loss": 1.5877, "step": 3348 }, { "epoch": 0.11, "learning_rate": 1.949176078796568e-05, "loss": 1.55, "step": 3349 }, { "epoch": 0.11, "learning_rate": 1.9491427804778148e-05, "loss": 1.5329, "step": 3350 }, { "epoch": 0.11, "learning_rate": 1.949109471539211e-05, "loss": 1.6008, "step": 3351 }, { "epoch": 0.11, "learning_rate": 1.9490761519811295e-05, "loss": 1.4448, "step": 3352 }, { "epoch": 0.11, "learning_rate": 1.9490428218039432e-05, "loss": 1.6062, "step": 3353 }, { "epoch": 0.11, "learning_rate": 1.9490094810080253e-05, "loss": 1.5968, "step": 3354 }, { "epoch": 0.11, "learning_rate": 1.9489761295937483e-05, "loss": 1.4938, "step": 3355 }, { "epoch": 0.11, "learning_rate": 1.9489427675614855e-05, "loss": 1.4892, "step": 3356 }, { "epoch": 0.11, "learning_rate": 1.9489093949116108e-05, "loss": 1.597, "step": 3357 }, { "epoch": 0.11, "learning_rate": 1.9488760116444966e-05, "loss": 1.5242, "step": 3358 }, { "epoch": 0.11, "learning_rate": 1.9488426177605173e-05, "loss": 1.5206, "step": 3359 }, { "epoch": 0.11, "learning_rate": 1.948809213260046e-05, "loss": 1.5055, "step": 3360 }, { "epoch": 0.11, "learning_rate": 1.9487757981434568e-05, "loss": 1.5148, "step": 3361 }, { "epoch": 0.11, "learning_rate": 1.9487423724111233e-05, "loss": 1.6495, "step": 3362 }, { "epoch": 0.11, "learning_rate": 1.9487089360634196e-05, "loss": 1.4111, "step": 3363 }, { "epoch": 0.11, "learning_rate": 1.9486754891007197e-05, "loss": 1.5317, "step": 3364 }, { "epoch": 0.11, "learning_rate": 1.9486420315233982e-05, "loss": 1.4712, "step": 3365 }, { "epoch": 0.11, "learning_rate": 1.9486085633318293e-05, "loss": 1.6252, "step": 3366 }, { "epoch": 0.11, "learning_rate": 1.9485750845263874e-05, "loss": 1.5077, "step": 3367 }, { "epoch": 0.11, "learning_rate": 1.9485415951074468e-05, "loss": 1.5251, "step": 3368 }, { "epoch": 0.11, "learning_rate": 1.948508095075383e-05, "loss": 1.4423, "step": 3369 }, { "epoch": 0.11, "learning_rate": 1.9484745844305694e-05, "loss": 1.5247, "step": 3370 }, { "epoch": 0.11, "learning_rate": 1.9484410631733825e-05, "loss": 1.5683, "step": 3371 }, { "epoch": 0.11, "learning_rate": 1.9484075313041968e-05, "loss": 1.506, "step": 3372 }, { "epoch": 0.11, "learning_rate": 1.948373988823387e-05, "loss": 1.5463, "step": 3373 }, { "epoch": 0.11, "learning_rate": 1.9483404357313295e-05, "loss": 1.501, "step": 3374 }, { "epoch": 0.11, "learning_rate": 1.9483068720283983e-05, "loss": 1.497, "step": 3375 }, { "epoch": 0.11, "learning_rate": 1.94827329771497e-05, "loss": 1.4851, "step": 3376 }, { "epoch": 0.11, "learning_rate": 1.94823971279142e-05, "loss": 1.4443, "step": 3377 }, { "epoch": 0.11, "learning_rate": 1.948206117258124e-05, "loss": 1.568, "step": 3378 }, { "epoch": 0.11, "learning_rate": 1.9481725111154577e-05, "loss": 1.5067, "step": 3379 }, { "epoch": 0.11, "learning_rate": 1.9481388943637976e-05, "loss": 1.4665, "step": 3380 }, { "epoch": 0.11, "learning_rate": 1.9481052670035195e-05, "loss": 1.5558, "step": 3381 }, { "epoch": 0.11, "learning_rate": 1.9480716290349998e-05, "loss": 1.575, "step": 3382 }, { "epoch": 0.11, "learning_rate": 1.9480379804586144e-05, "loss": 1.4633, "step": 3383 }, { "epoch": 0.11, "learning_rate": 1.9480043212747406e-05, "loss": 1.5741, "step": 3384 }, { "epoch": 0.11, "learning_rate": 1.9479706514837544e-05, "loss": 1.4978, "step": 3385 }, { "epoch": 0.11, "learning_rate": 1.9479369710860326e-05, "loss": 1.4628, "step": 3386 }, { "epoch": 0.11, "learning_rate": 1.9479032800819523e-05, "loss": 1.5027, "step": 3387 }, { "epoch": 0.11, "learning_rate": 1.9478695784718905e-05, "loss": 1.4731, "step": 3388 }, { "epoch": 0.11, "learning_rate": 1.9478358662562238e-05, "loss": 1.4595, "step": 3389 }, { "epoch": 0.11, "learning_rate": 1.9478021434353298e-05, "loss": 1.4811, "step": 3390 }, { "epoch": 0.11, "learning_rate": 1.947768410009586e-05, "loss": 1.5303, "step": 3391 }, { "epoch": 0.11, "learning_rate": 1.9477346659793692e-05, "loss": 1.4226, "step": 3392 }, { "epoch": 0.11, "learning_rate": 1.9477009113450574e-05, "loss": 1.4626, "step": 3393 }, { "epoch": 0.11, "learning_rate": 1.9476671461070286e-05, "loss": 1.4769, "step": 3394 }, { "epoch": 0.11, "learning_rate": 1.9476333702656597e-05, "loss": 1.472, "step": 3395 }, { "epoch": 0.11, "learning_rate": 1.9475995838213292e-05, "loss": 1.5504, "step": 3396 }, { "epoch": 0.11, "learning_rate": 1.9475657867744153e-05, "loss": 1.518, "step": 3397 }, { "epoch": 0.11, "learning_rate": 1.9475319791252956e-05, "loss": 1.7063, "step": 3398 }, { "epoch": 0.11, "learning_rate": 1.9474981608743487e-05, "loss": 1.59, "step": 3399 }, { "epoch": 0.11, "learning_rate": 1.9474643320219534e-05, "loss": 1.8203, "step": 3400 }, { "epoch": 0.11, "learning_rate": 1.947430492568487e-05, "loss": 1.5918, "step": 3401 }, { "epoch": 0.11, "learning_rate": 1.9473966425143292e-05, "loss": 1.6628, "step": 3402 }, { "epoch": 0.11, "learning_rate": 1.947362781859859e-05, "loss": 1.6606, "step": 3403 }, { "epoch": 0.11, "learning_rate": 1.947328910605454e-05, "loss": 1.7131, "step": 3404 }, { "epoch": 0.11, "learning_rate": 1.9472950287514944e-05, "loss": 1.6519, "step": 3405 }, { "epoch": 0.11, "learning_rate": 1.9472611362983583e-05, "loss": 1.5776, "step": 3406 }, { "epoch": 0.11, "learning_rate": 1.9472272332464257e-05, "loss": 1.617, "step": 3407 }, { "epoch": 0.11, "learning_rate": 1.9471933195960757e-05, "loss": 1.7534, "step": 3408 }, { "epoch": 0.11, "learning_rate": 1.9471593953476873e-05, "loss": 1.712, "step": 3409 }, { "epoch": 0.11, "learning_rate": 1.947125460501641e-05, "loss": 1.7298, "step": 3410 }, { "epoch": 0.11, "learning_rate": 1.947091515058315e-05, "loss": 1.7987, "step": 3411 }, { "epoch": 0.11, "learning_rate": 1.947057559018091e-05, "loss": 1.6746, "step": 3412 }, { "epoch": 0.11, "learning_rate": 1.947023592381348e-05, "loss": 1.8254, "step": 3413 }, { "epoch": 0.11, "learning_rate": 1.9469896151484657e-05, "loss": 1.6631, "step": 3414 }, { "epoch": 0.11, "learning_rate": 1.9469556273198246e-05, "loss": 1.6713, "step": 3415 }, { "epoch": 0.11, "learning_rate": 1.946921628895805e-05, "loss": 1.5507, "step": 3416 }, { "epoch": 0.11, "learning_rate": 1.9468876198767876e-05, "loss": 1.605, "step": 3417 }, { "epoch": 0.11, "learning_rate": 1.946853600263152e-05, "loss": 1.6237, "step": 3418 }, { "epoch": 0.11, "learning_rate": 1.94681957005528e-05, "loss": 1.6576, "step": 3419 }, { "epoch": 0.11, "learning_rate": 1.946785529253552e-05, "loss": 1.6496, "step": 3420 }, { "epoch": 0.11, "learning_rate": 1.9467514778583484e-05, "loss": 1.5894, "step": 3421 }, { "epoch": 0.11, "learning_rate": 1.9467174158700507e-05, "loss": 1.6426, "step": 3422 }, { "epoch": 0.11, "learning_rate": 1.9466833432890396e-05, "loss": 1.7746, "step": 3423 }, { "epoch": 0.11, "learning_rate": 1.9466492601156964e-05, "loss": 1.6985, "step": 3424 }, { "epoch": 0.11, "learning_rate": 1.946615166350403e-05, "loss": 1.5856, "step": 3425 }, { "epoch": 0.11, "learning_rate": 1.9465810619935404e-05, "loss": 1.7704, "step": 3426 }, { "epoch": 0.11, "learning_rate": 1.94654694704549e-05, "loss": 1.7134, "step": 3427 }, { "epoch": 0.11, "learning_rate": 1.9465128215066342e-05, "loss": 1.5798, "step": 3428 }, { "epoch": 0.11, "learning_rate": 1.946478685377354e-05, "loss": 1.6121, "step": 3429 }, { "epoch": 0.11, "learning_rate": 1.946444538658032e-05, "loss": 1.5466, "step": 3430 }, { "epoch": 0.11, "learning_rate": 1.94641038134905e-05, "loss": 1.6135, "step": 3431 }, { "epoch": 0.11, "learning_rate": 1.94637621345079e-05, "loss": 1.675, "step": 3432 }, { "epoch": 0.11, "learning_rate": 1.9463420349636348e-05, "loss": 1.6215, "step": 3433 }, { "epoch": 0.11, "learning_rate": 1.9463078458879665e-05, "loss": 1.5443, "step": 3434 }, { "epoch": 0.11, "learning_rate": 1.9462736462241672e-05, "loss": 1.6227, "step": 3435 }, { "epoch": 0.11, "learning_rate": 1.9462394359726208e-05, "loss": 1.6429, "step": 3436 }, { "epoch": 0.11, "learning_rate": 1.9462052151337086e-05, "loss": 1.6353, "step": 3437 }, { "epoch": 0.11, "learning_rate": 1.9461709837078143e-05, "loss": 1.6085, "step": 3438 }, { "epoch": 0.11, "learning_rate": 1.946136741695321e-05, "loss": 1.6707, "step": 3439 }, { "epoch": 0.11, "learning_rate": 1.9461024890966115e-05, "loss": 1.6272, "step": 3440 }, { "epoch": 0.11, "learning_rate": 1.9460682259120693e-05, "loss": 1.6416, "step": 3441 }, { "epoch": 0.11, "learning_rate": 1.946033952142077e-05, "loss": 1.6683, "step": 3442 }, { "epoch": 0.11, "learning_rate": 1.9459996677870197e-05, "loss": 1.6152, "step": 3443 }, { "epoch": 0.11, "learning_rate": 1.9459653728472794e-05, "loss": 1.6218, "step": 3444 }, { "epoch": 0.11, "learning_rate": 1.9459310673232407e-05, "loss": 1.632, "step": 3445 }, { "epoch": 0.11, "learning_rate": 1.9458967512152872e-05, "loss": 1.5516, "step": 3446 }, { "epoch": 0.11, "learning_rate": 1.9458624245238027e-05, "loss": 1.6055, "step": 3447 }, { "epoch": 0.11, "learning_rate": 1.9458280872491713e-05, "loss": 1.5676, "step": 3448 }, { "epoch": 0.11, "learning_rate": 1.9457937393917775e-05, "loss": 1.4231, "step": 3449 }, { "epoch": 0.12, "learning_rate": 1.9457593809520056e-05, "loss": 1.5387, "step": 3450 }, { "epoch": 0.12, "learning_rate": 1.9457250119302396e-05, "loss": 1.6081, "step": 3451 }, { "epoch": 0.12, "learning_rate": 1.9456906323268643e-05, "loss": 1.4825, "step": 3452 }, { "epoch": 0.12, "learning_rate": 1.9456562421422648e-05, "loss": 1.5308, "step": 3453 }, { "epoch": 0.12, "learning_rate": 1.945621841376825e-05, "loss": 1.6558, "step": 3454 }, { "epoch": 0.12, "learning_rate": 1.9455874300309304e-05, "loss": 1.606, "step": 3455 }, { "epoch": 0.12, "learning_rate": 1.945553008104966e-05, "loss": 1.4854, "step": 3456 }, { "epoch": 0.12, "learning_rate": 1.945518575599317e-05, "loss": 1.6152, "step": 3457 }, { "epoch": 0.12, "learning_rate": 1.9454841325143684e-05, "loss": 1.5579, "step": 3458 }, { "epoch": 0.12, "learning_rate": 1.9454496788505055e-05, "loss": 1.5471, "step": 3459 }, { "epoch": 0.12, "learning_rate": 1.945415214608114e-05, "loss": 1.6732, "step": 3460 }, { "epoch": 0.12, "learning_rate": 1.94538073978758e-05, "loss": 1.6204, "step": 3461 }, { "epoch": 0.12, "learning_rate": 1.945346254389288e-05, "loss": 1.5973, "step": 3462 }, { "epoch": 0.12, "learning_rate": 1.9453117584136253e-05, "loss": 1.573, "step": 3463 }, { "epoch": 0.12, "learning_rate": 1.9452772518609767e-05, "loss": 1.6568, "step": 3464 }, { "epoch": 0.12, "learning_rate": 1.945242734731729e-05, "loss": 1.5045, "step": 3465 }, { "epoch": 0.12, "learning_rate": 1.9452082070262678e-05, "loss": 1.5275, "step": 3466 }, { "epoch": 0.12, "learning_rate": 1.94517366874498e-05, "loss": 1.6343, "step": 3467 }, { "epoch": 0.12, "learning_rate": 1.945139119888252e-05, "loss": 1.5179, "step": 3468 }, { "epoch": 0.12, "learning_rate": 1.94510456045647e-05, "loss": 1.5736, "step": 3469 }, { "epoch": 0.12, "learning_rate": 1.945069990450021e-05, "loss": 1.5105, "step": 3470 }, { "epoch": 0.12, "learning_rate": 1.9450354098692915e-05, "loss": 1.5158, "step": 3471 }, { "epoch": 0.12, "learning_rate": 1.9450008187146685e-05, "loss": 1.6744, "step": 3472 }, { "epoch": 0.12, "learning_rate": 1.9449662169865394e-05, "loss": 1.5927, "step": 3473 }, { "epoch": 0.12, "learning_rate": 1.944931604685291e-05, "loss": 1.5214, "step": 3474 }, { "epoch": 0.12, "learning_rate": 1.944896981811311e-05, "loss": 1.6145, "step": 3475 }, { "epoch": 0.12, "learning_rate": 1.944862348364986e-05, "loss": 1.5449, "step": 3476 }, { "epoch": 0.12, "learning_rate": 1.9448277043467043e-05, "loss": 1.6015, "step": 3477 }, { "epoch": 0.12, "learning_rate": 1.9447930497568528e-05, "loss": 1.5707, "step": 3478 }, { "epoch": 0.12, "learning_rate": 1.9447583845958198e-05, "loss": 1.551, "step": 3479 }, { "epoch": 0.12, "learning_rate": 1.9447237088639928e-05, "loss": 1.6543, "step": 3480 }, { "epoch": 0.12, "learning_rate": 1.9446890225617604e-05, "loss": 1.6291, "step": 3481 }, { "epoch": 0.12, "learning_rate": 1.94465432568951e-05, "loss": 1.5163, "step": 3482 }, { "epoch": 0.12, "learning_rate": 1.9446196182476304e-05, "loss": 1.656, "step": 3483 }, { "epoch": 0.12, "learning_rate": 1.9445849002365092e-05, "loss": 1.6432, "step": 3484 }, { "epoch": 0.12, "learning_rate": 1.944550171656536e-05, "loss": 1.6072, "step": 3485 }, { "epoch": 0.12, "learning_rate": 1.9445154325080978e-05, "loss": 1.6457, "step": 3486 }, { "epoch": 0.12, "learning_rate": 1.9444806827915848e-05, "loss": 1.5663, "step": 3487 }, { "epoch": 0.12, "learning_rate": 1.944445922507385e-05, "loss": 1.4972, "step": 3488 }, { "epoch": 0.12, "learning_rate": 1.944411151655887e-05, "loss": 1.6828, "step": 3489 }, { "epoch": 0.12, "learning_rate": 1.944376370237481e-05, "loss": 1.5496, "step": 3490 }, { "epoch": 0.12, "learning_rate": 1.9443415782525556e-05, "loss": 1.5709, "step": 3491 }, { "epoch": 0.12, "learning_rate": 1.9443067757015e-05, "loss": 1.5308, "step": 3492 }, { "epoch": 0.12, "learning_rate": 1.9442719625847032e-05, "loss": 1.5643, "step": 3493 }, { "epoch": 0.12, "learning_rate": 1.9442371389025552e-05, "loss": 1.6546, "step": 3494 }, { "epoch": 0.12, "learning_rate": 1.9442023046554456e-05, "loss": 1.5069, "step": 3495 }, { "epoch": 0.12, "learning_rate": 1.944167459843764e-05, "loss": 1.6169, "step": 3496 }, { "epoch": 0.12, "learning_rate": 1.9441326044679007e-05, "loss": 1.5396, "step": 3497 }, { "epoch": 0.12, "learning_rate": 1.9440977385282453e-05, "loss": 1.5251, "step": 3498 }, { "epoch": 0.12, "learning_rate": 1.9440628620251874e-05, "loss": 1.6243, "step": 3499 }, { "epoch": 0.12, "learning_rate": 1.9440279749591185e-05, "loss": 1.4507, "step": 3500 }, { "epoch": 0.12, "learning_rate": 1.9439930773304284e-05, "loss": 1.5334, "step": 3501 }, { "epoch": 0.12, "learning_rate": 1.943958169139507e-05, "loss": 1.5751, "step": 3502 }, { "epoch": 0.12, "learning_rate": 1.9439232503867455e-05, "loss": 1.5357, "step": 3503 }, { "epoch": 0.12, "learning_rate": 1.9438883210725342e-05, "loss": 1.547, "step": 3504 }, { "epoch": 0.12, "learning_rate": 1.9438533811972645e-05, "loss": 1.452, "step": 3505 }, { "epoch": 0.12, "learning_rate": 1.9438184307613265e-05, "loss": 1.5994, "step": 3506 }, { "epoch": 0.12, "learning_rate": 1.943783469765112e-05, "loss": 1.4833, "step": 3507 }, { "epoch": 0.12, "learning_rate": 1.9437484982090122e-05, "loss": 1.6039, "step": 3508 }, { "epoch": 0.12, "learning_rate": 1.9437135160934176e-05, "loss": 1.552, "step": 3509 }, { "epoch": 0.12, "learning_rate": 1.9436785234187204e-05, "loss": 1.5485, "step": 3510 }, { "epoch": 0.12, "learning_rate": 1.943643520185312e-05, "loss": 1.6076, "step": 3511 }, { "epoch": 0.12, "learning_rate": 1.9436085063935837e-05, "loss": 1.5878, "step": 3512 }, { "epoch": 0.12, "learning_rate": 1.9435734820439274e-05, "loss": 1.5854, "step": 3513 }, { "epoch": 0.12, "learning_rate": 1.943538447136735e-05, "loss": 1.6004, "step": 3514 }, { "epoch": 0.12, "learning_rate": 1.9435034016723987e-05, "loss": 1.6014, "step": 3515 }, { "epoch": 0.12, "learning_rate": 1.9434683456513105e-05, "loss": 1.5968, "step": 3516 }, { "epoch": 0.12, "learning_rate": 1.9434332790738625e-05, "loss": 1.5818, "step": 3517 }, { "epoch": 0.12, "learning_rate": 1.9433982019404473e-05, "loss": 1.5118, "step": 3518 }, { "epoch": 0.12, "learning_rate": 1.9433631142514572e-05, "loss": 1.5199, "step": 3519 }, { "epoch": 0.12, "learning_rate": 1.9433280160072846e-05, "loss": 1.5824, "step": 3520 }, { "epoch": 0.12, "learning_rate": 1.9432929072083223e-05, "loss": 1.7072, "step": 3521 }, { "epoch": 0.12, "learning_rate": 1.9432577878549635e-05, "loss": 1.5898, "step": 3522 }, { "epoch": 0.12, "learning_rate": 1.943222657947601e-05, "loss": 1.5331, "step": 3523 }, { "epoch": 0.12, "learning_rate": 1.943187517486628e-05, "loss": 1.4847, "step": 3524 }, { "epoch": 0.12, "learning_rate": 1.943152366472437e-05, "loss": 1.5334, "step": 3525 }, { "epoch": 0.12, "learning_rate": 1.943117204905422e-05, "loss": 1.6807, "step": 3526 }, { "epoch": 0.12, "learning_rate": 1.9430820327859757e-05, "loss": 1.5649, "step": 3527 }, { "epoch": 0.12, "learning_rate": 1.9430468501144928e-05, "loss": 1.624, "step": 3528 }, { "epoch": 0.12, "learning_rate": 1.9430116568913656e-05, "loss": 1.4994, "step": 3529 }, { "epoch": 0.12, "learning_rate": 1.9429764531169892e-05, "loss": 1.4954, "step": 3530 }, { "epoch": 0.12, "learning_rate": 1.9429412387917566e-05, "loss": 1.6289, "step": 3531 }, { "epoch": 0.12, "learning_rate": 1.942906013916062e-05, "loss": 1.4375, "step": 3532 }, { "epoch": 0.12, "learning_rate": 1.9428707784902994e-05, "loss": 1.5125, "step": 3533 }, { "epoch": 0.12, "learning_rate": 1.9428355325148632e-05, "loss": 1.5645, "step": 3534 }, { "epoch": 0.12, "learning_rate": 1.942800275990148e-05, "loss": 1.5315, "step": 3535 }, { "epoch": 0.12, "learning_rate": 1.9427650089165478e-05, "loss": 1.5566, "step": 3536 }, { "epoch": 0.12, "learning_rate": 1.9427297312944577e-05, "loss": 1.5875, "step": 3537 }, { "epoch": 0.12, "learning_rate": 1.942694443124272e-05, "loss": 1.6305, "step": 3538 }, { "epoch": 0.12, "learning_rate": 1.942659144406386e-05, "loss": 1.6121, "step": 3539 }, { "epoch": 0.12, "learning_rate": 1.942623835141194e-05, "loss": 1.6514, "step": 3540 }, { "epoch": 0.12, "learning_rate": 1.942588515329092e-05, "loss": 1.6134, "step": 3541 }, { "epoch": 0.12, "learning_rate": 1.942553184970474e-05, "loss": 1.6059, "step": 3542 }, { "epoch": 0.12, "learning_rate": 1.9425178440657362e-05, "loss": 1.512, "step": 3543 }, { "epoch": 0.12, "learning_rate": 1.9424824926152736e-05, "loss": 1.639, "step": 3544 }, { "epoch": 0.12, "learning_rate": 1.9424471306194822e-05, "loss": 1.6335, "step": 3545 }, { "epoch": 0.12, "learning_rate": 1.9424117580787567e-05, "loss": 1.5601, "step": 3546 }, { "epoch": 0.12, "learning_rate": 1.9423763749934942e-05, "loss": 1.3522, "step": 3547 }, { "epoch": 0.12, "learning_rate": 1.9423409813640895e-05, "loss": 1.6047, "step": 3548 }, { "epoch": 0.12, "learning_rate": 1.9423055771909393e-05, "loss": 1.5119, "step": 3549 }, { "epoch": 0.12, "learning_rate": 1.9422701624744396e-05, "loss": 1.4196, "step": 3550 }, { "epoch": 0.12, "learning_rate": 1.9422347372149862e-05, "loss": 1.5786, "step": 3551 }, { "epoch": 0.12, "learning_rate": 1.9421993014129762e-05, "loss": 1.4887, "step": 3552 }, { "epoch": 0.12, "learning_rate": 1.9421638550688057e-05, "loss": 1.4939, "step": 3553 }, { "epoch": 0.12, "learning_rate": 1.9421283981828707e-05, "loss": 1.5196, "step": 3554 }, { "epoch": 0.12, "learning_rate": 1.942092930755569e-05, "loss": 1.5692, "step": 3555 }, { "epoch": 0.12, "learning_rate": 1.942057452787297e-05, "loss": 1.6758, "step": 3556 }, { "epoch": 0.12, "learning_rate": 1.9420219642784514e-05, "loss": 1.5227, "step": 3557 }, { "epoch": 0.12, "learning_rate": 1.9419864652294294e-05, "loss": 1.6269, "step": 3558 }, { "epoch": 0.12, "learning_rate": 1.9419509556406287e-05, "loss": 1.509, "step": 3559 }, { "epoch": 0.12, "learning_rate": 1.9419154355124458e-05, "loss": 1.6015, "step": 3560 }, { "epoch": 0.12, "learning_rate": 1.9418799048452787e-05, "loss": 1.4921, "step": 3561 }, { "epoch": 0.12, "learning_rate": 1.941844363639525e-05, "loss": 1.4394, "step": 3562 }, { "epoch": 0.12, "learning_rate": 1.9418088118955816e-05, "loss": 1.4789, "step": 3563 }, { "epoch": 0.12, "learning_rate": 1.9417732496138476e-05, "loss": 1.5233, "step": 3564 }, { "epoch": 0.12, "learning_rate": 1.9417376767947194e-05, "loss": 1.583, "step": 3565 }, { "epoch": 0.12, "learning_rate": 1.9417020934385963e-05, "loss": 1.592, "step": 3566 }, { "epoch": 0.12, "learning_rate": 1.9416664995458756e-05, "loss": 1.6237, "step": 3567 }, { "epoch": 0.12, "learning_rate": 1.9416308951169557e-05, "loss": 1.5966, "step": 3568 }, { "epoch": 0.12, "learning_rate": 1.9415952801522352e-05, "loss": 1.5031, "step": 3569 }, { "epoch": 0.12, "learning_rate": 1.9415596546521128e-05, "loss": 1.6034, "step": 3570 }, { "epoch": 0.12, "learning_rate": 1.9415240186169866e-05, "loss": 1.4293, "step": 3571 }, { "epoch": 0.12, "learning_rate": 1.9414883720472557e-05, "loss": 1.4764, "step": 3572 }, { "epoch": 0.12, "learning_rate": 1.941452714943319e-05, "loss": 1.5182, "step": 3573 }, { "epoch": 0.12, "learning_rate": 1.9414170473055747e-05, "loss": 1.5293, "step": 3574 }, { "epoch": 0.12, "learning_rate": 1.941381369134423e-05, "loss": 1.5986, "step": 3575 }, { "epoch": 0.12, "learning_rate": 1.941345680430262e-05, "loss": 1.7069, "step": 3576 }, { "epoch": 0.12, "learning_rate": 1.941309981193492e-05, "loss": 1.7064, "step": 3577 }, { "epoch": 0.12, "learning_rate": 1.941274271424512e-05, "loss": 1.621, "step": 3578 }, { "epoch": 0.12, "learning_rate": 1.941238551123721e-05, "loss": 1.4783, "step": 3579 }, { "epoch": 0.12, "learning_rate": 1.94120282029152e-05, "loss": 1.6368, "step": 3580 }, { "epoch": 0.12, "learning_rate": 1.941167078928307e-05, "loss": 1.5489, "step": 3581 }, { "epoch": 0.12, "learning_rate": 1.9411313270344836e-05, "loss": 1.6877, "step": 3582 }, { "epoch": 0.12, "learning_rate": 1.9410955646104492e-05, "loss": 1.6042, "step": 3583 }, { "epoch": 0.12, "learning_rate": 1.9410597916566036e-05, "loss": 1.6686, "step": 3584 }, { "epoch": 0.12, "learning_rate": 1.9410240081733477e-05, "loss": 1.6349, "step": 3585 }, { "epoch": 0.12, "learning_rate": 1.9409882141610813e-05, "loss": 1.6263, "step": 3586 }, { "epoch": 0.12, "learning_rate": 1.9409524096202048e-05, "loss": 1.6413, "step": 3587 }, { "epoch": 0.12, "learning_rate": 1.9409165945511195e-05, "loss": 1.5749, "step": 3588 }, { "epoch": 0.12, "learning_rate": 1.9408807689542257e-05, "loss": 1.6174, "step": 3589 }, { "epoch": 0.12, "learning_rate": 1.9408449328299243e-05, "loss": 1.7363, "step": 3590 }, { "epoch": 0.12, "learning_rate": 1.940809086178616e-05, "loss": 1.692, "step": 3591 }, { "epoch": 0.12, "learning_rate": 1.9407732290007023e-05, "loss": 1.7127, "step": 3592 }, { "epoch": 0.12, "learning_rate": 1.9407373612965844e-05, "loss": 1.6863, "step": 3593 }, { "epoch": 0.12, "learning_rate": 1.9407014830666635e-05, "loss": 1.5424, "step": 3594 }, { "epoch": 0.12, "learning_rate": 1.9406655943113408e-05, "loss": 1.5898, "step": 3595 }, { "epoch": 0.12, "learning_rate": 1.940629695031018e-05, "loss": 1.6092, "step": 3596 }, { "epoch": 0.12, "learning_rate": 1.940593785226097e-05, "loss": 1.5685, "step": 3597 }, { "epoch": 0.12, "learning_rate": 1.9405578648969796e-05, "loss": 1.7436, "step": 3598 }, { "epoch": 0.12, "learning_rate": 1.9405219340440675e-05, "loss": 1.7118, "step": 3599 }, { "epoch": 0.12, "learning_rate": 1.9404859926677625e-05, "loss": 1.5635, "step": 3600 }, { "epoch": 0.12, "learning_rate": 1.9404500407684673e-05, "loss": 1.5604, "step": 3601 }, { "epoch": 0.12, "learning_rate": 1.9404140783465837e-05, "loss": 1.6759, "step": 3602 }, { "epoch": 0.12, "learning_rate": 1.940378105402514e-05, "loss": 1.6627, "step": 3603 }, { "epoch": 0.12, "learning_rate": 1.940342121936661e-05, "loss": 1.6895, "step": 3604 }, { "epoch": 0.12, "learning_rate": 1.9403061279494276e-05, "loss": 1.6076, "step": 3605 }, { "epoch": 0.12, "learning_rate": 1.9402701234412163e-05, "loss": 1.589, "step": 3606 }, { "epoch": 0.12, "learning_rate": 1.9402341084124298e-05, "loss": 1.6362, "step": 3607 }, { "epoch": 0.12, "learning_rate": 1.9401980828634704e-05, "loss": 1.6565, "step": 3608 }, { "epoch": 0.12, "learning_rate": 1.9401620467947426e-05, "loss": 1.5711, "step": 3609 }, { "epoch": 0.12, "learning_rate": 1.940126000206649e-05, "loss": 1.6334, "step": 3610 }, { "epoch": 0.12, "learning_rate": 1.9400899430995923e-05, "loss": 1.6286, "step": 3611 }, { "epoch": 0.12, "learning_rate": 1.9400538754739765e-05, "loss": 1.4641, "step": 3612 }, { "epoch": 0.12, "learning_rate": 1.9400177973302054e-05, "loss": 1.6005, "step": 3613 }, { "epoch": 0.12, "learning_rate": 1.9399817086686826e-05, "loss": 1.5018, "step": 3614 }, { "epoch": 0.12, "learning_rate": 1.939945609489811e-05, "loss": 1.4441, "step": 3615 }, { "epoch": 0.12, "learning_rate": 1.9399094997939957e-05, "loss": 1.6059, "step": 3616 }, { "epoch": 0.12, "learning_rate": 1.93987337958164e-05, "loss": 1.5775, "step": 3617 }, { "epoch": 0.12, "learning_rate": 1.9398372488531484e-05, "loss": 1.6202, "step": 3618 }, { "epoch": 0.12, "learning_rate": 1.939801107608925e-05, "loss": 1.6194, "step": 3619 }, { "epoch": 0.12, "learning_rate": 1.939764955849374e-05, "loss": 1.5854, "step": 3620 }, { "epoch": 0.12, "learning_rate": 1.9397287935749006e-05, "loss": 1.4882, "step": 3621 }, { "epoch": 0.12, "learning_rate": 1.9396926207859085e-05, "loss": 1.4753, "step": 3622 }, { "epoch": 0.12, "learning_rate": 1.9396564374828032e-05, "loss": 1.6019, "step": 3623 }, { "epoch": 0.12, "learning_rate": 1.939620243665989e-05, "loss": 1.6804, "step": 3624 }, { "epoch": 0.12, "learning_rate": 1.9395840393358707e-05, "loss": 1.6499, "step": 3625 }, { "epoch": 0.12, "learning_rate": 1.9395478244928542e-05, "loss": 1.6603, "step": 3626 }, { "epoch": 0.12, "learning_rate": 1.9395115991373444e-05, "loss": 1.7266, "step": 3627 }, { "epoch": 0.12, "learning_rate": 1.9394753632697464e-05, "loss": 1.766, "step": 3628 }, { "epoch": 0.12, "learning_rate": 1.9394391168904652e-05, "loss": 1.6172, "step": 3629 }, { "epoch": 0.12, "learning_rate": 1.9394028599999072e-05, "loss": 1.6796, "step": 3630 }, { "epoch": 0.12, "learning_rate": 1.939366592598478e-05, "loss": 1.6218, "step": 3631 }, { "epoch": 0.12, "learning_rate": 1.939330314686583e-05, "loss": 1.6367, "step": 3632 }, { "epoch": 0.12, "learning_rate": 1.9392940262646284e-05, "loss": 1.6423, "step": 3633 }, { "epoch": 0.12, "learning_rate": 1.93925772733302e-05, "loss": 1.6704, "step": 3634 }, { "epoch": 0.12, "learning_rate": 1.9392214178921635e-05, "loss": 1.6206, "step": 3635 }, { "epoch": 0.12, "learning_rate": 1.939185097942466e-05, "loss": 1.5471, "step": 3636 }, { "epoch": 0.12, "learning_rate": 1.939148767484334e-05, "loss": 1.5521, "step": 3637 }, { "epoch": 0.12, "learning_rate": 1.939112426518173e-05, "loss": 1.6532, "step": 3638 }, { "epoch": 0.12, "learning_rate": 1.9390760750443908e-05, "loss": 1.5808, "step": 3639 }, { "epoch": 0.12, "learning_rate": 1.9390397130633932e-05, "loss": 1.6554, "step": 3640 }, { "epoch": 0.12, "learning_rate": 1.939003340575587e-05, "loss": 1.5352, "step": 3641 }, { "epoch": 0.12, "learning_rate": 1.93896695758138e-05, "loss": 1.5588, "step": 3642 }, { "epoch": 0.12, "learning_rate": 1.938930564081179e-05, "loss": 1.5956, "step": 3643 }, { "epoch": 0.12, "learning_rate": 1.9388941600753902e-05, "loss": 1.5841, "step": 3644 }, { "epoch": 0.12, "learning_rate": 1.9388577455644223e-05, "loss": 1.6055, "step": 3645 }, { "epoch": 0.12, "learning_rate": 1.9388213205486823e-05, "loss": 1.7103, "step": 3646 }, { "epoch": 0.12, "learning_rate": 1.9387848850285773e-05, "loss": 1.6471, "step": 3647 }, { "epoch": 0.12, "learning_rate": 1.9387484390045153e-05, "loss": 1.5639, "step": 3648 }, { "epoch": 0.12, "learning_rate": 1.9387119824769043e-05, "loss": 1.5631, "step": 3649 }, { "epoch": 0.12, "learning_rate": 1.9386755154461522e-05, "loss": 1.6504, "step": 3650 }, { "epoch": 0.12, "learning_rate": 1.9386390379126666e-05, "loss": 1.6095, "step": 3651 }, { "epoch": 0.12, "learning_rate": 1.938602549876856e-05, "loss": 1.7513, "step": 3652 }, { "epoch": 0.12, "learning_rate": 1.938566051339128e-05, "loss": 1.5318, "step": 3653 }, { "epoch": 0.12, "learning_rate": 1.938529542299892e-05, "loss": 1.514, "step": 3654 }, { "epoch": 0.12, "learning_rate": 1.938493022759556e-05, "loss": 1.6, "step": 3655 }, { "epoch": 0.12, "learning_rate": 1.938456492718529e-05, "loss": 1.6485, "step": 3656 }, { "epoch": 0.12, "learning_rate": 1.9384199521772188e-05, "loss": 1.6078, "step": 3657 }, { "epoch": 0.12, "learning_rate": 1.9383834011360347e-05, "loss": 1.6292, "step": 3658 }, { "epoch": 0.12, "learning_rate": 1.938346839595386e-05, "loss": 1.562, "step": 3659 }, { "epoch": 0.12, "learning_rate": 1.9383102675556812e-05, "loss": 1.5743, "step": 3660 }, { "epoch": 0.12, "learning_rate": 1.9382736850173303e-05, "loss": 1.5943, "step": 3661 }, { "epoch": 0.12, "learning_rate": 1.938237091980742e-05, "loss": 1.4773, "step": 3662 }, { "epoch": 0.12, "learning_rate": 1.9382004884463258e-05, "loss": 1.5793, "step": 3663 }, { "epoch": 0.12, "learning_rate": 1.9381638744144914e-05, "loss": 1.5477, "step": 3664 }, { "epoch": 0.12, "learning_rate": 1.9381272498856487e-05, "loss": 1.506, "step": 3665 }, { "epoch": 0.12, "learning_rate": 1.938090614860207e-05, "loss": 1.4293, "step": 3666 }, { "epoch": 0.12, "learning_rate": 1.9380539693385763e-05, "loss": 1.4849, "step": 3667 }, { "epoch": 0.12, "learning_rate": 1.938017313321167e-05, "loss": 1.5362, "step": 3668 }, { "epoch": 0.12, "learning_rate": 1.9379806468083888e-05, "loss": 1.5476, "step": 3669 }, { "epoch": 0.12, "learning_rate": 1.9379439698006522e-05, "loss": 1.5757, "step": 3670 }, { "epoch": 0.12, "learning_rate": 1.9379072822983675e-05, "loss": 1.4258, "step": 3671 }, { "epoch": 0.12, "learning_rate": 1.937870584301945e-05, "loss": 1.5479, "step": 3672 }, { "epoch": 0.12, "learning_rate": 1.937833875811796e-05, "loss": 1.4776, "step": 3673 }, { "epoch": 0.12, "learning_rate": 1.9377971568283302e-05, "loss": 1.5835, "step": 3674 }, { "epoch": 0.12, "learning_rate": 1.9377604273519593e-05, "loss": 1.5465, "step": 3675 }, { "epoch": 0.12, "learning_rate": 1.937723687383094e-05, "loss": 1.6021, "step": 3676 }, { "epoch": 0.12, "learning_rate": 1.937686936922145e-05, "loss": 1.5985, "step": 3677 }, { "epoch": 0.12, "learning_rate": 1.937650175969524e-05, "loss": 1.575, "step": 3678 }, { "epoch": 0.12, "learning_rate": 1.9376134045256424e-05, "loss": 1.5811, "step": 3679 }, { "epoch": 0.12, "learning_rate": 1.937576622590911e-05, "loss": 1.6804, "step": 3680 }, { "epoch": 0.12, "learning_rate": 1.9375398301657417e-05, "loss": 1.509, "step": 3681 }, { "epoch": 0.12, "learning_rate": 1.9375030272505463e-05, "loss": 1.5477, "step": 3682 }, { "epoch": 0.12, "learning_rate": 1.9374662138457366e-05, "loss": 1.4755, "step": 3683 }, { "epoch": 0.12, "learning_rate": 1.9374293899517243e-05, "loss": 1.5391, "step": 3684 }, { "epoch": 0.12, "learning_rate": 1.9373925555689212e-05, "loss": 1.4384, "step": 3685 }, { "epoch": 0.12, "learning_rate": 1.93735571069774e-05, "loss": 1.5634, "step": 3686 }, { "epoch": 0.12, "learning_rate": 1.937318855338593e-05, "loss": 1.5081, "step": 3687 }, { "epoch": 0.12, "learning_rate": 1.937281989491892e-05, "loss": 1.5837, "step": 3688 }, { "epoch": 0.12, "learning_rate": 1.937245113158049e-05, "loss": 1.5097, "step": 3689 }, { "epoch": 0.12, "learning_rate": 1.9372082263374783e-05, "loss": 1.5779, "step": 3690 }, { "epoch": 0.12, "learning_rate": 1.937171329030591e-05, "loss": 1.4862, "step": 3691 }, { "epoch": 0.12, "learning_rate": 1.9371344212378013e-05, "loss": 1.5456, "step": 3692 }, { "epoch": 0.12, "learning_rate": 1.937097502959521e-05, "loss": 1.6174, "step": 3693 }, { "epoch": 0.12, "learning_rate": 1.9370605741961638e-05, "loss": 1.5352, "step": 3694 }, { "epoch": 0.12, "learning_rate": 1.9370236349481425e-05, "loss": 1.6147, "step": 3695 }, { "epoch": 0.12, "learning_rate": 1.936986685215871e-05, "loss": 1.6738, "step": 3696 }, { "epoch": 0.12, "learning_rate": 1.936949724999762e-05, "loss": 1.619, "step": 3697 }, { "epoch": 0.12, "learning_rate": 1.9369127543002297e-05, "loss": 1.5811, "step": 3698 }, { "epoch": 0.12, "learning_rate": 1.936875773117687e-05, "loss": 1.5334, "step": 3699 }, { "epoch": 0.12, "learning_rate": 1.9368387814525483e-05, "loss": 1.581, "step": 3700 }, { "epoch": 0.12, "learning_rate": 1.9368017793052278e-05, "loss": 1.5831, "step": 3701 }, { "epoch": 0.12, "learning_rate": 1.9367647666761384e-05, "loss": 1.6286, "step": 3702 }, { "epoch": 0.12, "learning_rate": 1.9367277435656953e-05, "loss": 1.7165, "step": 3703 }, { "epoch": 0.12, "learning_rate": 1.9366907099743125e-05, "loss": 1.6259, "step": 3704 }, { "epoch": 0.12, "learning_rate": 1.9366536659024037e-05, "loss": 1.6194, "step": 3705 }, { "epoch": 0.12, "learning_rate": 1.9366166113503843e-05, "loss": 1.6907, "step": 3706 }, { "epoch": 0.12, "learning_rate": 1.9365795463186682e-05, "loss": 1.5393, "step": 3707 }, { "epoch": 0.12, "learning_rate": 1.9365424708076703e-05, "loss": 1.5057, "step": 3708 }, { "epoch": 0.12, "learning_rate": 1.9365053848178058e-05, "loss": 1.6385, "step": 3709 }, { "epoch": 0.12, "learning_rate": 1.9364682883494892e-05, "loss": 1.5841, "step": 3710 }, { "epoch": 0.12, "learning_rate": 1.9364311814031358e-05, "loss": 1.5087, "step": 3711 }, { "epoch": 0.12, "learning_rate": 1.9363940639791607e-05, "loss": 1.5923, "step": 3712 }, { "epoch": 0.12, "learning_rate": 1.9363569360779793e-05, "loss": 1.6313, "step": 3713 }, { "epoch": 0.12, "learning_rate": 1.9363197977000066e-05, "loss": 1.6597, "step": 3714 }, { "epoch": 0.12, "learning_rate": 1.936282648845659e-05, "loss": 1.5967, "step": 3715 }, { "epoch": 0.12, "learning_rate": 1.9362454895153516e-05, "loss": 1.6286, "step": 3716 }, { "epoch": 0.12, "learning_rate": 1.9362083197095e-05, "loss": 1.6249, "step": 3717 }, { "epoch": 0.12, "learning_rate": 1.9361711394285202e-05, "loss": 1.5572, "step": 3718 }, { "epoch": 0.12, "learning_rate": 1.9361339486728286e-05, "loss": 1.6868, "step": 3719 }, { "epoch": 0.12, "learning_rate": 1.9360967474428408e-05, "loss": 1.6852, "step": 3720 }, { "epoch": 0.12, "learning_rate": 1.9360595357389735e-05, "loss": 1.6277, "step": 3721 }, { "epoch": 0.12, "learning_rate": 1.9360223135616423e-05, "loss": 1.5668, "step": 3722 }, { "epoch": 0.12, "learning_rate": 1.935985080911265e-05, "loss": 1.6533, "step": 3723 }, { "epoch": 0.12, "learning_rate": 1.9359478377882567e-05, "loss": 1.6357, "step": 3724 }, { "epoch": 0.12, "learning_rate": 1.9359105841930352e-05, "loss": 1.6192, "step": 3725 }, { "epoch": 0.12, "learning_rate": 1.935873320126017e-05, "loss": 1.5869, "step": 3726 }, { "epoch": 0.12, "learning_rate": 1.935836045587619e-05, "loss": 1.6464, "step": 3727 }, { "epoch": 0.12, "learning_rate": 1.9357987605782576e-05, "loss": 1.5993, "step": 3728 }, { "epoch": 0.12, "learning_rate": 1.9357614650983513e-05, "loss": 1.514, "step": 3729 }, { "epoch": 0.12, "learning_rate": 1.9357241591483165e-05, "loss": 1.6055, "step": 3730 }, { "epoch": 0.12, "learning_rate": 1.9356868427285706e-05, "loss": 1.5748, "step": 3731 }, { "epoch": 0.12, "learning_rate": 1.9356495158395317e-05, "loss": 1.6191, "step": 3732 }, { "epoch": 0.12, "learning_rate": 1.9356121784816165e-05, "loss": 1.5673, "step": 3733 }, { "epoch": 0.12, "learning_rate": 1.935574830655244e-05, "loss": 1.6208, "step": 3734 }, { "epoch": 0.12, "learning_rate": 1.9355374723608314e-05, "loss": 1.595, "step": 3735 }, { "epoch": 0.12, "learning_rate": 1.9355001035987966e-05, "loss": 1.584, "step": 3736 }, { "epoch": 0.12, "learning_rate": 1.9354627243695577e-05, "loss": 1.591, "step": 3737 }, { "epoch": 0.12, "learning_rate": 1.9354253346735332e-05, "loss": 1.5067, "step": 3738 }, { "epoch": 0.12, "learning_rate": 1.935387934511141e-05, "loss": 1.6132, "step": 3739 }, { "epoch": 0.12, "learning_rate": 1.9353505238828004e-05, "loss": 1.5829, "step": 3740 }, { "epoch": 0.12, "learning_rate": 1.9353131027889293e-05, "loss": 1.5822, "step": 3741 }, { "epoch": 0.12, "learning_rate": 1.9352756712299467e-05, "loss": 1.4932, "step": 3742 }, { "epoch": 0.12, "learning_rate": 1.9352382292062712e-05, "loss": 1.5578, "step": 3743 }, { "epoch": 0.12, "learning_rate": 1.935200776718322e-05, "loss": 1.5983, "step": 3744 }, { "epoch": 0.12, "learning_rate": 1.9351633137665175e-05, "loss": 1.5149, "step": 3745 }, { "epoch": 0.12, "learning_rate": 1.9351258403512778e-05, "loss": 1.5188, "step": 3746 }, { "epoch": 0.12, "learning_rate": 1.9350883564730213e-05, "loss": 1.4373, "step": 3747 }, { "epoch": 0.12, "learning_rate": 1.9350508621321685e-05, "loss": 1.5008, "step": 3748 }, { "epoch": 0.12, "learning_rate": 1.9350133573291377e-05, "loss": 1.5146, "step": 3749 }, { "epoch": 0.12, "learning_rate": 1.9349758420643494e-05, "loss": 1.5877, "step": 3750 }, { "epoch": 0.13, "learning_rate": 1.934938316338223e-05, "loss": 1.479, "step": 3751 }, { "epoch": 0.13, "learning_rate": 1.9349007801511784e-05, "loss": 1.4775, "step": 3752 }, { "epoch": 0.13, "learning_rate": 1.9348632335036354e-05, "loss": 1.7168, "step": 3753 }, { "epoch": 0.13, "learning_rate": 1.9348256763960146e-05, "loss": 1.5675, "step": 3754 }, { "epoch": 0.13, "learning_rate": 1.934788108828736e-05, "loss": 1.6318, "step": 3755 }, { "epoch": 0.13, "learning_rate": 1.9347505308022197e-05, "loss": 1.5753, "step": 3756 }, { "epoch": 0.13, "learning_rate": 1.934712942316886e-05, "loss": 1.529, "step": 3757 }, { "epoch": 0.13, "learning_rate": 1.9346753433731563e-05, "loss": 1.5337, "step": 3758 }, { "epoch": 0.13, "learning_rate": 1.934637733971451e-05, "loss": 1.5149, "step": 3759 }, { "epoch": 0.13, "learning_rate": 1.9346001141121903e-05, "loss": 1.6095, "step": 3760 }, { "epoch": 0.13, "learning_rate": 1.9345624837957954e-05, "loss": 1.6274, "step": 3761 }, { "epoch": 0.13, "learning_rate": 1.934524843022688e-05, "loss": 1.548, "step": 3762 }, { "epoch": 0.13, "learning_rate": 1.9344871917932884e-05, "loss": 1.6188, "step": 3763 }, { "epoch": 0.13, "learning_rate": 1.934449530108018e-05, "loss": 1.6303, "step": 3764 }, { "epoch": 0.13, "learning_rate": 1.9344118579672987e-05, "loss": 1.674, "step": 3765 }, { "epoch": 0.13, "learning_rate": 1.934374175371552e-05, "loss": 1.5454, "step": 3766 }, { "epoch": 0.13, "learning_rate": 1.9343364823211982e-05, "loss": 1.6461, "step": 3767 }, { "epoch": 0.13, "learning_rate": 1.934298778816661e-05, "loss": 1.5045, "step": 3768 }, { "epoch": 0.13, "learning_rate": 1.934261064858361e-05, "loss": 1.5582, "step": 3769 }, { "epoch": 0.13, "learning_rate": 1.9342233404467207e-05, "loss": 1.6152, "step": 3770 }, { "epoch": 0.13, "learning_rate": 1.9341856055821618e-05, "loss": 1.625, "step": 3771 }, { "epoch": 0.13, "learning_rate": 1.9341478602651068e-05, "loss": 1.5569, "step": 3772 }, { "epoch": 0.13, "learning_rate": 1.934110104495978e-05, "loss": 1.514, "step": 3773 }, { "epoch": 0.13, "learning_rate": 1.934072338275198e-05, "loss": 1.4283, "step": 3774 }, { "epoch": 0.13, "learning_rate": 1.9340345616031885e-05, "loss": 1.5517, "step": 3775 }, { "epoch": 0.13, "learning_rate": 1.9339967744803735e-05, "loss": 1.4371, "step": 3776 }, { "epoch": 0.13, "learning_rate": 1.933958976907175e-05, "loss": 1.5244, "step": 3777 }, { "epoch": 0.13, "learning_rate": 1.933921168884016e-05, "loss": 1.4995, "step": 3778 }, { "epoch": 0.13, "learning_rate": 1.9338833504113192e-05, "loss": 1.6329, "step": 3779 }, { "epoch": 0.13, "learning_rate": 1.9338455214895086e-05, "loss": 1.4346, "step": 3780 }, { "epoch": 0.13, "learning_rate": 1.9338076821190065e-05, "loss": 1.546, "step": 3781 }, { "epoch": 0.13, "learning_rate": 1.933769832300237e-05, "loss": 1.5109, "step": 3782 }, { "epoch": 0.13, "learning_rate": 1.9337319720336236e-05, "loss": 1.5355, "step": 3783 }, { "epoch": 0.13, "learning_rate": 1.9336941013195892e-05, "loss": 1.5153, "step": 3784 }, { "epoch": 0.13, "learning_rate": 1.9336562201585588e-05, "loss": 1.5846, "step": 3785 }, { "epoch": 0.13, "learning_rate": 1.9336183285509548e-05, "loss": 1.5671, "step": 3786 }, { "epoch": 0.13, "learning_rate": 1.9335804264972018e-05, "loss": 1.6427, "step": 3787 }, { "epoch": 0.13, "learning_rate": 1.9335425139977242e-05, "loss": 1.519, "step": 3788 }, { "epoch": 0.13, "learning_rate": 1.9335045910529458e-05, "loss": 1.6149, "step": 3789 }, { "epoch": 0.13, "learning_rate": 1.933466657663291e-05, "loss": 1.6301, "step": 3790 }, { "epoch": 0.13, "learning_rate": 1.933428713829184e-05, "loss": 1.5498, "step": 3791 }, { "epoch": 0.13, "learning_rate": 1.93339075955105e-05, "loss": 1.5773, "step": 3792 }, { "epoch": 0.13, "learning_rate": 1.9333527948293128e-05, "loss": 1.6923, "step": 3793 }, { "epoch": 0.13, "learning_rate": 1.9333148196643982e-05, "loss": 1.5397, "step": 3794 }, { "epoch": 0.13, "learning_rate": 1.9332768340567304e-05, "loss": 1.6117, "step": 3795 }, { "epoch": 0.13, "learning_rate": 1.933238838006734e-05, "loss": 1.5624, "step": 3796 }, { "epoch": 0.13, "learning_rate": 1.9332008315148355e-05, "loss": 1.6498, "step": 3797 }, { "epoch": 0.13, "learning_rate": 1.9331628145814587e-05, "loss": 1.6306, "step": 3798 }, { "epoch": 0.13, "learning_rate": 1.93312478720703e-05, "loss": 1.5368, "step": 3799 }, { "epoch": 0.13, "learning_rate": 1.9330867493919743e-05, "loss": 1.5581, "step": 3800 }, { "epoch": 0.13, "learning_rate": 1.9330487011367177e-05, "loss": 1.7191, "step": 3801 }, { "epoch": 0.13, "learning_rate": 1.9330106424416852e-05, "loss": 1.6101, "step": 3802 }, { "epoch": 0.13, "learning_rate": 1.9329725733073034e-05, "loss": 1.6538, "step": 3803 }, { "epoch": 0.13, "learning_rate": 1.9329344937339977e-05, "loss": 1.5173, "step": 3804 }, { "epoch": 0.13, "learning_rate": 1.9328964037221944e-05, "loss": 1.5833, "step": 3805 }, { "epoch": 0.13, "learning_rate": 1.9328583032723194e-05, "loss": 1.6177, "step": 3806 }, { "epoch": 0.13, "learning_rate": 1.9328201923847995e-05, "loss": 1.5754, "step": 3807 }, { "epoch": 0.13, "learning_rate": 1.932782071060061e-05, "loss": 1.5682, "step": 3808 }, { "epoch": 0.13, "learning_rate": 1.93274393929853e-05, "loss": 1.5906, "step": 3809 }, { "epoch": 0.13, "learning_rate": 1.932705797100634e-05, "loss": 1.6877, "step": 3810 }, { "epoch": 0.13, "learning_rate": 1.9326676444667988e-05, "loss": 1.6013, "step": 3811 }, { "epoch": 0.13, "learning_rate": 1.9326294813974517e-05, "loss": 1.5536, "step": 3812 }, { "epoch": 0.13, "learning_rate": 1.9325913078930196e-05, "loss": 1.7136, "step": 3813 }, { "epoch": 0.13, "learning_rate": 1.9325531239539303e-05, "loss": 1.5251, "step": 3814 }, { "epoch": 0.13, "learning_rate": 1.9325149295806097e-05, "loss": 1.5538, "step": 3815 }, { "epoch": 0.13, "learning_rate": 1.9324767247734865e-05, "loss": 1.7002, "step": 3816 }, { "epoch": 0.13, "learning_rate": 1.9324385095329875e-05, "loss": 1.6255, "step": 3817 }, { "epoch": 0.13, "learning_rate": 1.93240028385954e-05, "loss": 1.5563, "step": 3818 }, { "epoch": 0.13, "learning_rate": 1.9323620477535726e-05, "loss": 1.4537, "step": 3819 }, { "epoch": 0.13, "learning_rate": 1.9323238012155125e-05, "loss": 1.6621, "step": 3820 }, { "epoch": 0.13, "learning_rate": 1.9322855442457876e-05, "loss": 1.5471, "step": 3821 }, { "epoch": 0.13, "learning_rate": 1.932247276844826e-05, "loss": 1.5869, "step": 3822 }, { "epoch": 0.13, "learning_rate": 1.932208999013056e-05, "loss": 1.604, "step": 3823 }, { "epoch": 0.13, "learning_rate": 1.932170710750906e-05, "loss": 1.6437, "step": 3824 }, { "epoch": 0.13, "learning_rate": 1.9321324120588042e-05, "loss": 1.6665, "step": 3825 }, { "epoch": 0.13, "learning_rate": 1.932094102937179e-05, "loss": 1.6133, "step": 3826 }, { "epoch": 0.13, "learning_rate": 1.9320557833864595e-05, "loss": 1.5561, "step": 3827 }, { "epoch": 0.13, "learning_rate": 1.932017453407074e-05, "loss": 1.5068, "step": 3828 }, { "epoch": 0.13, "learning_rate": 1.9319791129994515e-05, "loss": 1.6569, "step": 3829 }, { "epoch": 0.13, "learning_rate": 1.931940762164021e-05, "loss": 1.5566, "step": 3830 }, { "epoch": 0.13, "learning_rate": 1.9319024009012114e-05, "loss": 1.5438, "step": 3831 }, { "epoch": 0.13, "learning_rate": 1.9318640292114526e-05, "loss": 1.4558, "step": 3832 }, { "epoch": 0.13, "learning_rate": 1.9318256470951732e-05, "loss": 1.4928, "step": 3833 }, { "epoch": 0.13, "learning_rate": 1.931787254552803e-05, "loss": 1.475, "step": 3834 }, { "epoch": 0.13, "learning_rate": 1.9317488515847717e-05, "loss": 1.5021, "step": 3835 }, { "epoch": 0.13, "learning_rate": 1.9317104381915083e-05, "loss": 1.5055, "step": 3836 }, { "epoch": 0.13, "learning_rate": 1.9316720143734433e-05, "loss": 1.3587, "step": 3837 }, { "epoch": 0.13, "learning_rate": 1.9316335801310064e-05, "loss": 1.4466, "step": 3838 }, { "epoch": 0.13, "learning_rate": 1.931595135464628e-05, "loss": 1.4426, "step": 3839 }, { "epoch": 0.13, "learning_rate": 1.931556680374737e-05, "loss": 1.4985, "step": 3840 }, { "epoch": 0.13, "learning_rate": 1.9315182148617655e-05, "loss": 1.6058, "step": 3841 }, { "epoch": 0.13, "learning_rate": 1.9314797389261426e-05, "loss": 1.3301, "step": 3842 }, { "epoch": 0.13, "learning_rate": 1.931441252568299e-05, "loss": 1.5112, "step": 3843 }, { "epoch": 0.13, "learning_rate": 1.9314027557886655e-05, "loss": 1.4683, "step": 3844 }, { "epoch": 0.13, "learning_rate": 1.9313642485876732e-05, "loss": 1.4319, "step": 3845 }, { "epoch": 0.13, "learning_rate": 1.9313642485876732e-05, "loss": 1.4614, "step": 3846 }, { "epoch": 0.13, "learning_rate": 1.931325730965752e-05, "loss": 1.4154, "step": 3847 }, { "epoch": 0.13, "learning_rate": 1.931287202923334e-05, "loss": 1.4033, "step": 3848 }, { "epoch": 0.13, "learning_rate": 1.9312486644608492e-05, "loss": 1.4862, "step": 3849 }, { "epoch": 0.13, "learning_rate": 1.9312101155787296e-05, "loss": 1.4507, "step": 3850 }, { "epoch": 0.13, "learning_rate": 1.931171556277406e-05, "loss": 1.6362, "step": 3851 }, { "epoch": 0.13, "learning_rate": 1.9311329865573105e-05, "loss": 1.5459, "step": 3852 }, { "epoch": 0.13, "learning_rate": 1.931094406418874e-05, "loss": 1.5506, "step": 3853 }, { "epoch": 0.13, "learning_rate": 1.9310558158625286e-05, "loss": 1.5328, "step": 3854 }, { "epoch": 0.13, "learning_rate": 1.9310172148887054e-05, "loss": 1.5552, "step": 3855 }, { "epoch": 0.13, "learning_rate": 1.9309786034978372e-05, "loss": 1.563, "step": 3856 }, { "epoch": 0.13, "learning_rate": 1.9309399816903554e-05, "loss": 1.6713, "step": 3857 }, { "epoch": 0.13, "learning_rate": 1.9309013494666923e-05, "loss": 1.6071, "step": 3858 }, { "epoch": 0.13, "learning_rate": 1.9308627068272804e-05, "loss": 1.5351, "step": 3859 }, { "epoch": 0.13, "learning_rate": 1.930824053772552e-05, "loss": 1.4839, "step": 3860 }, { "epoch": 0.13, "learning_rate": 1.9307853903029387e-05, "loss": 1.3962, "step": 3861 }, { "epoch": 0.13, "learning_rate": 1.9307467164188745e-05, "loss": 1.4438, "step": 3862 }, { "epoch": 0.13, "learning_rate": 1.9307080321207913e-05, "loss": 1.5779, "step": 3863 }, { "epoch": 0.13, "learning_rate": 1.930669337409122e-05, "loss": 1.5253, "step": 3864 }, { "epoch": 0.13, "learning_rate": 1.9306306322842994e-05, "loss": 1.6378, "step": 3865 }, { "epoch": 0.13, "learning_rate": 1.9305919167467575e-05, "loss": 1.5945, "step": 3866 }, { "epoch": 0.13, "learning_rate": 1.9305531907969285e-05, "loss": 1.4876, "step": 3867 }, { "epoch": 0.13, "learning_rate": 1.9305144544352455e-05, "loss": 1.5102, "step": 3868 }, { "epoch": 0.13, "learning_rate": 1.930475707662143e-05, "loss": 1.5401, "step": 3869 }, { "epoch": 0.13, "learning_rate": 1.9304369504780536e-05, "loss": 1.5732, "step": 3870 }, { "epoch": 0.13, "learning_rate": 1.9303981828834114e-05, "loss": 1.5638, "step": 3871 }, { "epoch": 0.13, "learning_rate": 1.9303594048786505e-05, "loss": 1.4554, "step": 3872 }, { "epoch": 0.13, "learning_rate": 1.9303206164642037e-05, "loss": 1.5075, "step": 3873 }, { "epoch": 0.13, "learning_rate": 1.930281817640506e-05, "loss": 1.5362, "step": 3874 }, { "epoch": 0.13, "learning_rate": 1.930243008407991e-05, "loss": 1.5416, "step": 3875 }, { "epoch": 0.13, "learning_rate": 1.930204188767093e-05, "loss": 1.6344, "step": 3876 }, { "epoch": 0.13, "learning_rate": 1.9301653587182467e-05, "loss": 1.4982, "step": 3877 }, { "epoch": 0.13, "learning_rate": 1.9301265182618862e-05, "loss": 1.4651, "step": 3878 }, { "epoch": 0.13, "learning_rate": 1.9300876673984463e-05, "loss": 1.4993, "step": 3879 }, { "epoch": 0.13, "learning_rate": 1.9300488061283613e-05, "loss": 1.533, "step": 3880 }, { "epoch": 0.13, "learning_rate": 1.9300099344520665e-05, "loss": 1.6373, "step": 3881 }, { "epoch": 0.13, "learning_rate": 1.9299710523699965e-05, "loss": 1.6178, "step": 3882 }, { "epoch": 0.13, "learning_rate": 1.9299321598825867e-05, "loss": 1.6454, "step": 3883 }, { "epoch": 0.13, "learning_rate": 1.9298932569902717e-05, "loss": 1.624, "step": 3884 }, { "epoch": 0.13, "learning_rate": 1.9298543436934874e-05, "loss": 1.6686, "step": 3885 }, { "epoch": 0.13, "learning_rate": 1.929815419992669e-05, "loss": 1.6236, "step": 3886 }, { "epoch": 0.13, "learning_rate": 1.9297764858882516e-05, "loss": 1.5662, "step": 3887 }, { "epoch": 0.13, "learning_rate": 1.9297375413806712e-05, "loss": 1.6821, "step": 3888 }, { "epoch": 0.13, "learning_rate": 1.9296985864703638e-05, "loss": 1.6367, "step": 3889 }, { "epoch": 0.13, "learning_rate": 1.9296596211577646e-05, "loss": 1.6223, "step": 3890 }, { "epoch": 0.13, "learning_rate": 1.92962064544331e-05, "loss": 1.6338, "step": 3891 }, { "epoch": 0.13, "learning_rate": 1.9295816593274364e-05, "loss": 1.536, "step": 3892 }, { "epoch": 0.13, "learning_rate": 1.9295426628105792e-05, "loss": 1.6567, "step": 3893 }, { "epoch": 0.13, "learning_rate": 1.9295036558931755e-05, "loss": 1.5204, "step": 3894 }, { "epoch": 0.13, "learning_rate": 1.929464638575661e-05, "loss": 1.5252, "step": 3895 }, { "epoch": 0.13, "learning_rate": 1.9294256108584734e-05, "loss": 1.5951, "step": 3896 }, { "epoch": 0.13, "learning_rate": 1.9293865727420483e-05, "loss": 1.4975, "step": 3897 }, { "epoch": 0.13, "learning_rate": 1.9293475242268224e-05, "loss": 1.4736, "step": 3898 }, { "epoch": 0.13, "learning_rate": 1.9293084653132336e-05, "loss": 1.5002, "step": 3899 }, { "epoch": 0.13, "learning_rate": 1.9292693960017185e-05, "loss": 1.4472, "step": 3900 }, { "epoch": 0.13, "learning_rate": 1.9292303162927136e-05, "loss": 1.5411, "step": 3901 }, { "epoch": 0.13, "learning_rate": 1.9291912261866568e-05, "loss": 1.5623, "step": 3902 }, { "epoch": 0.13, "learning_rate": 1.929152125683986e-05, "loss": 1.462, "step": 3903 }, { "epoch": 0.13, "learning_rate": 1.9291130147851374e-05, "loss": 1.4521, "step": 3904 }, { "epoch": 0.13, "learning_rate": 1.9290738934905492e-05, "loss": 1.6322, "step": 3905 }, { "epoch": 0.13, "learning_rate": 1.9290347618006595e-05, "loss": 1.4944, "step": 3906 }, { "epoch": 0.13, "learning_rate": 1.9289956197159058e-05, "loss": 1.5046, "step": 3907 }, { "epoch": 0.13, "learning_rate": 1.928956467236726e-05, "loss": 1.5326, "step": 3908 }, { "epoch": 0.13, "learning_rate": 1.9289173043635584e-05, "loss": 1.6357, "step": 3909 }, { "epoch": 0.13, "learning_rate": 1.9288781310968406e-05, "loss": 1.5697, "step": 3910 }, { "epoch": 0.13, "learning_rate": 1.9288389474370116e-05, "loss": 1.5524, "step": 3911 }, { "epoch": 0.13, "learning_rate": 1.9287997533845098e-05, "loss": 1.5774, "step": 3912 }, { "epoch": 0.13, "learning_rate": 1.9287605489397732e-05, "loss": 1.4837, "step": 3913 }, { "epoch": 0.13, "learning_rate": 1.928721334103241e-05, "loss": 1.5469, "step": 3914 }, { "epoch": 0.13, "learning_rate": 1.9286821088753515e-05, "loss": 1.5759, "step": 3915 }, { "epoch": 0.13, "learning_rate": 1.9286428732565442e-05, "loss": 1.5776, "step": 3916 }, { "epoch": 0.13, "learning_rate": 1.9286036272472572e-05, "loss": 1.5463, "step": 3917 }, { "epoch": 0.13, "learning_rate": 1.9285643708479304e-05, "loss": 1.545, "step": 3918 }, { "epoch": 0.13, "learning_rate": 1.9285251040590028e-05, "loss": 1.588, "step": 3919 }, { "epoch": 0.13, "learning_rate": 1.9284858268809135e-05, "loss": 1.5703, "step": 3920 }, { "epoch": 0.13, "learning_rate": 1.9284465393141024e-05, "loss": 1.5, "step": 3921 }, { "epoch": 0.13, "learning_rate": 1.928407241359009e-05, "loss": 1.4887, "step": 3922 }, { "epoch": 0.13, "learning_rate": 1.9283679330160726e-05, "loss": 1.639, "step": 3923 }, { "epoch": 0.13, "learning_rate": 1.9283286142857335e-05, "loss": 1.5853, "step": 3924 }, { "epoch": 0.13, "learning_rate": 1.9282892851684316e-05, "loss": 1.5303, "step": 3925 }, { "epoch": 0.13, "learning_rate": 1.9282499456646064e-05, "loss": 1.5223, "step": 3926 }, { "epoch": 0.13, "learning_rate": 1.9282105957746987e-05, "loss": 1.6146, "step": 3927 }, { "epoch": 0.13, "learning_rate": 1.9281712354991483e-05, "loss": 1.517, "step": 3928 }, { "epoch": 0.13, "learning_rate": 1.928131864838396e-05, "loss": 1.6229, "step": 3929 }, { "epoch": 0.13, "learning_rate": 1.928092483792882e-05, "loss": 1.5514, "step": 3930 }, { "epoch": 0.13, "learning_rate": 1.928053092363047e-05, "loss": 1.6291, "step": 3931 }, { "epoch": 0.13, "learning_rate": 1.9280136905493322e-05, "loss": 1.4815, "step": 3932 }, { "epoch": 0.13, "learning_rate": 1.9279742783521776e-05, "loss": 1.5644, "step": 3933 }, { "epoch": 0.13, "learning_rate": 1.9279348557720247e-05, "loss": 1.4593, "step": 3934 }, { "epoch": 0.13, "learning_rate": 1.927895422809315e-05, "loss": 1.5824, "step": 3935 }, { "epoch": 0.13, "learning_rate": 1.9278559794644888e-05, "loss": 1.5281, "step": 3936 }, { "epoch": 0.13, "learning_rate": 1.927816525737988e-05, "loss": 1.5798, "step": 3937 }, { "epoch": 0.13, "learning_rate": 1.927777061630254e-05, "loss": 1.4702, "step": 3938 }, { "epoch": 0.13, "learning_rate": 1.9277375871417282e-05, "loss": 1.5427, "step": 3939 }, { "epoch": 0.13, "learning_rate": 1.9276981022728523e-05, "loss": 1.5519, "step": 3940 }, { "epoch": 0.13, "learning_rate": 1.9276586070240684e-05, "loss": 1.4778, "step": 3941 }, { "epoch": 0.13, "learning_rate": 1.927619101395818e-05, "loss": 1.5278, "step": 3942 }, { "epoch": 0.13, "learning_rate": 1.9275795853885433e-05, "loss": 1.4876, "step": 3943 }, { "epoch": 0.13, "learning_rate": 1.9275400590026866e-05, "loss": 1.4705, "step": 3944 }, { "epoch": 0.13, "learning_rate": 1.92750052223869e-05, "loss": 1.4439, "step": 3945 }, { "epoch": 0.13, "learning_rate": 1.9274609750969953e-05, "loss": 1.3987, "step": 3946 }, { "epoch": 0.13, "learning_rate": 1.927421417578046e-05, "loss": 1.4644, "step": 3947 }, { "epoch": 0.13, "learning_rate": 1.9273818496822843e-05, "loss": 1.5758, "step": 3948 }, { "epoch": 0.13, "learning_rate": 1.9273422714101525e-05, "loss": 1.4105, "step": 3949 }, { "epoch": 0.13, "learning_rate": 1.9273026827620942e-05, "loss": 1.475, "step": 3950 }, { "epoch": 0.13, "learning_rate": 1.9272630837385518e-05, "loss": 1.4349, "step": 3951 }, { "epoch": 0.13, "learning_rate": 1.9272234743399684e-05, "loss": 1.4572, "step": 3952 }, { "epoch": 0.13, "learning_rate": 1.9271838545667876e-05, "loss": 1.4263, "step": 3953 }, { "epoch": 0.13, "learning_rate": 1.927144224419452e-05, "loss": 1.4822, "step": 3954 }, { "epoch": 0.13, "learning_rate": 1.927104583898406e-05, "loss": 1.4516, "step": 3955 }, { "epoch": 0.13, "learning_rate": 1.927064933004092e-05, "loss": 1.4046, "step": 3956 }, { "epoch": 0.13, "learning_rate": 1.9270252717369546e-05, "loss": 1.3845, "step": 3957 }, { "epoch": 0.13, "learning_rate": 1.926985600097437e-05, "loss": 1.3288, "step": 3958 }, { "epoch": 0.13, "learning_rate": 1.9269459180859834e-05, "loss": 1.5526, "step": 3959 }, { "epoch": 0.13, "learning_rate": 1.9269062257030377e-05, "loss": 1.5542, "step": 3960 }, { "epoch": 0.13, "learning_rate": 1.9268665229490436e-05, "loss": 1.6019, "step": 3961 }, { "epoch": 0.13, "learning_rate": 1.926826809824446e-05, "loss": 1.5088, "step": 3962 }, { "epoch": 0.13, "learning_rate": 1.9267870863296888e-05, "loss": 1.5703, "step": 3963 }, { "epoch": 0.13, "learning_rate": 1.9267473524652168e-05, "loss": 1.4702, "step": 3964 }, { "epoch": 0.13, "learning_rate": 1.926707608231474e-05, "loss": 1.4639, "step": 3965 }, { "epoch": 0.13, "learning_rate": 1.926667853628906e-05, "loss": 1.4278, "step": 3966 }, { "epoch": 0.13, "learning_rate": 1.9266280886579568e-05, "loss": 1.5505, "step": 3967 }, { "epoch": 0.13, "learning_rate": 1.9265883133190715e-05, "loss": 1.4581, "step": 3968 }, { "epoch": 0.13, "learning_rate": 1.9265485276126953e-05, "loss": 1.5718, "step": 3969 }, { "epoch": 0.13, "learning_rate": 1.9265087315392736e-05, "loss": 1.4395, "step": 3970 }, { "epoch": 0.13, "learning_rate": 1.9264689250992514e-05, "loss": 1.5978, "step": 3971 }, { "epoch": 0.13, "learning_rate": 1.9264291082930736e-05, "loss": 1.4999, "step": 3972 }, { "epoch": 0.13, "learning_rate": 1.9263892811211865e-05, "loss": 1.6118, "step": 3973 }, { "epoch": 0.13, "learning_rate": 1.9263494435840355e-05, "loss": 1.6451, "step": 3974 }, { "epoch": 0.13, "learning_rate": 1.926309595682066e-05, "loss": 1.5852, "step": 3975 }, { "epoch": 0.13, "learning_rate": 1.926269737415724e-05, "loss": 1.7579, "step": 3976 }, { "epoch": 0.13, "learning_rate": 1.926229868785456e-05, "loss": 1.5916, "step": 3977 }, { "epoch": 0.13, "learning_rate": 1.9261899897917073e-05, "loss": 1.5844, "step": 3978 }, { "epoch": 0.13, "learning_rate": 1.9261501004349248e-05, "loss": 1.8153, "step": 3979 }, { "epoch": 0.13, "learning_rate": 1.926110200715554e-05, "loss": 1.6153, "step": 3980 }, { "epoch": 0.13, "learning_rate": 1.926070290634042e-05, "loss": 1.6153, "step": 3981 }, { "epoch": 0.13, "learning_rate": 1.9260303701908354e-05, "loss": 1.6466, "step": 3982 }, { "epoch": 0.13, "learning_rate": 1.9259904393863804e-05, "loss": 1.6197, "step": 3983 }, { "epoch": 0.13, "learning_rate": 1.9259504982211242e-05, "loss": 1.5452, "step": 3984 }, { "epoch": 0.13, "learning_rate": 1.9259105466955132e-05, "loss": 1.6451, "step": 3985 }, { "epoch": 0.13, "learning_rate": 1.925870584809995e-05, "loss": 1.5622, "step": 3986 }, { "epoch": 0.13, "learning_rate": 1.9258306125650163e-05, "loss": 1.4626, "step": 3987 }, { "epoch": 0.13, "learning_rate": 1.9257906299610245e-05, "loss": 1.6599, "step": 3988 }, { "epoch": 0.13, "learning_rate": 1.925750636998467e-05, "loss": 1.5519, "step": 3989 }, { "epoch": 0.13, "learning_rate": 1.925710633677791e-05, "loss": 1.632, "step": 3990 }, { "epoch": 0.13, "learning_rate": 1.925670619999444e-05, "loss": 1.5986, "step": 3991 }, { "epoch": 0.13, "learning_rate": 1.9256305959638748e-05, "loss": 1.6159, "step": 3992 }, { "epoch": 0.13, "learning_rate": 1.92559056157153e-05, "loss": 1.6121, "step": 3993 }, { "epoch": 0.13, "learning_rate": 1.925550516822858e-05, "loss": 1.6256, "step": 3994 }, { "epoch": 0.13, "learning_rate": 1.9255104617183068e-05, "loss": 1.6942, "step": 3995 }, { "epoch": 0.13, "learning_rate": 1.925470396258325e-05, "loss": 1.5815, "step": 3996 }, { "epoch": 0.13, "learning_rate": 1.9254303204433602e-05, "loss": 1.6448, "step": 3997 }, { "epoch": 0.13, "learning_rate": 1.9253902342738612e-05, "loss": 1.5015, "step": 3998 }, { "epoch": 0.13, "learning_rate": 1.9253501377502765e-05, "loss": 1.6044, "step": 3999 }, { "epoch": 0.13, "learning_rate": 1.9253100308730547e-05, "loss": 1.603, "step": 4000 }, { "epoch": 0.13, "learning_rate": 1.925269913642644e-05, "loss": 1.5827, "step": 4001 }, { "epoch": 0.13, "learning_rate": 1.9252297860594945e-05, "loss": 1.6072, "step": 4002 }, { "epoch": 0.13, "learning_rate": 1.9251896481240536e-05, "loss": 1.5591, "step": 4003 }, { "epoch": 0.13, "learning_rate": 1.925149499836772e-05, "loss": 1.6758, "step": 4004 }, { "epoch": 0.13, "learning_rate": 1.925109341198098e-05, "loss": 1.6258, "step": 4005 }, { "epoch": 0.13, "learning_rate": 1.925069172208481e-05, "loss": 1.6788, "step": 4006 }, { "epoch": 0.13, "learning_rate": 1.9250289928683706e-05, "loss": 1.5792, "step": 4007 }, { "epoch": 0.13, "learning_rate": 1.924988803178216e-05, "loss": 1.438, "step": 4008 }, { "epoch": 0.13, "learning_rate": 1.924948603138468e-05, "loss": 1.5849, "step": 4009 }, { "epoch": 0.13, "learning_rate": 1.924908392749575e-05, "loss": 1.5747, "step": 4010 }, { "epoch": 0.13, "learning_rate": 1.9248681720119875e-05, "loss": 1.5412, "step": 4011 }, { "epoch": 0.13, "learning_rate": 1.9248279409261555e-05, "loss": 1.5782, "step": 4012 }, { "epoch": 0.13, "learning_rate": 1.9247876994925293e-05, "loss": 1.477, "step": 4013 }, { "epoch": 0.13, "learning_rate": 1.924747447711559e-05, "loss": 1.4601, "step": 4014 }, { "epoch": 0.13, "learning_rate": 1.9247071855836948e-05, "loss": 1.6103, "step": 4015 }, { "epoch": 0.13, "learning_rate": 1.9246669131093875e-05, "loss": 1.6084, "step": 4016 }, { "epoch": 0.13, "learning_rate": 1.924626630289088e-05, "loss": 1.4254, "step": 4017 }, { "epoch": 0.13, "learning_rate": 1.924586337123246e-05, "loss": 1.5238, "step": 4018 }, { "epoch": 0.13, "learning_rate": 1.9245460336123136e-05, "loss": 1.5077, "step": 4019 }, { "epoch": 0.13, "learning_rate": 1.9245057197567405e-05, "loss": 1.4954, "step": 4020 }, { "epoch": 0.13, "learning_rate": 1.9244653955569784e-05, "loss": 1.5759, "step": 4021 }, { "epoch": 0.13, "learning_rate": 1.9244250610134787e-05, "loss": 1.5659, "step": 4022 }, { "epoch": 0.13, "learning_rate": 1.9243847161266924e-05, "loss": 1.5977, "step": 4023 }, { "epoch": 0.13, "learning_rate": 1.9243443608970706e-05, "loss": 1.5623, "step": 4024 }, { "epoch": 0.13, "learning_rate": 1.9243039953250654e-05, "loss": 1.5176, "step": 4025 }, { "epoch": 0.13, "learning_rate": 1.9242636194111284e-05, "loss": 1.5213, "step": 4026 }, { "epoch": 0.13, "learning_rate": 1.9242232331557113e-05, "loss": 1.691, "step": 4027 }, { "epoch": 0.13, "learning_rate": 1.9241828365592653e-05, "loss": 1.6265, "step": 4028 }, { "epoch": 0.13, "learning_rate": 1.9241424296222435e-05, "loss": 1.5808, "step": 4029 }, { "epoch": 0.13, "learning_rate": 1.9241020123450972e-05, "loss": 1.7255, "step": 4030 }, { "epoch": 0.13, "learning_rate": 1.924061584728279e-05, "loss": 1.5582, "step": 4031 }, { "epoch": 0.13, "learning_rate": 1.924021146772241e-05, "loss": 1.5578, "step": 4032 }, { "epoch": 0.13, "learning_rate": 1.923980698477436e-05, "loss": 1.6382, "step": 4033 }, { "epoch": 0.13, "learning_rate": 1.923940239844316e-05, "loss": 1.556, "step": 4034 }, { "epoch": 0.13, "learning_rate": 1.9238997708733343e-05, "loss": 1.6361, "step": 4035 }, { "epoch": 0.13, "learning_rate": 1.9238592915649437e-05, "loss": 1.5135, "step": 4036 }, { "epoch": 0.13, "learning_rate": 1.9238188019195964e-05, "loss": 1.5623, "step": 4037 }, { "epoch": 0.13, "learning_rate": 1.9237783019377462e-05, "loss": 1.7187, "step": 4038 }, { "epoch": 0.13, "learning_rate": 1.923737791619846e-05, "loss": 1.5614, "step": 4039 }, { "epoch": 0.13, "learning_rate": 1.9236972709663487e-05, "loss": 1.6605, "step": 4040 }, { "epoch": 0.13, "learning_rate": 1.9236567399777086e-05, "loss": 1.6082, "step": 4041 }, { "epoch": 0.13, "learning_rate": 1.923616198654378e-05, "loss": 1.5921, "step": 4042 }, { "epoch": 0.13, "learning_rate": 1.9235756469968112e-05, "loss": 1.5972, "step": 4043 }, { "epoch": 0.13, "learning_rate": 1.923535085005462e-05, "loss": 1.53, "step": 4044 }, { "epoch": 0.13, "learning_rate": 1.923494512680784e-05, "loss": 1.5859, "step": 4045 }, { "epoch": 0.13, "learning_rate": 1.9234539300232312e-05, "loss": 1.5242, "step": 4046 }, { "epoch": 0.13, "learning_rate": 1.923413337033258e-05, "loss": 1.5904, "step": 4047 }, { "epoch": 0.13, "learning_rate": 1.9233727337113176e-05, "loss": 1.5916, "step": 4048 }, { "epoch": 0.13, "learning_rate": 1.9233321200578657e-05, "loss": 1.6604, "step": 4049 }, { "epoch": 0.14, "learning_rate": 1.9232914960733556e-05, "loss": 1.4637, "step": 4050 }, { "epoch": 0.14, "learning_rate": 1.9232508617582426e-05, "loss": 1.5837, "step": 4051 }, { "epoch": 0.14, "learning_rate": 1.923210217112981e-05, "loss": 1.5861, "step": 4052 }, { "epoch": 0.14, "learning_rate": 1.9231695621380257e-05, "loss": 1.5561, "step": 4053 }, { "epoch": 0.14, "learning_rate": 1.923128896833831e-05, "loss": 1.6433, "step": 4054 }, { "epoch": 0.14, "learning_rate": 1.9230882212008528e-05, "loss": 1.615, "step": 4055 }, { "epoch": 0.14, "learning_rate": 1.9230475352395457e-05, "loss": 1.5618, "step": 4056 }, { "epoch": 0.14, "learning_rate": 1.923006838950365e-05, "loss": 1.5505, "step": 4057 }, { "epoch": 0.14, "learning_rate": 1.922966132333766e-05, "loss": 1.52, "step": 4058 }, { "epoch": 0.14, "learning_rate": 1.922925415390204e-05, "loss": 1.5864, "step": 4059 }, { "epoch": 0.14, "learning_rate": 1.9228846881201352e-05, "loss": 1.6001, "step": 4060 }, { "epoch": 0.14, "learning_rate": 1.9228439505240147e-05, "loss": 1.6198, "step": 4061 }, { "epoch": 0.14, "learning_rate": 1.9228032026022984e-05, "loss": 1.6118, "step": 4062 }, { "epoch": 0.14, "learning_rate": 1.9227624443554425e-05, "loss": 1.595, "step": 4063 }, { "epoch": 0.14, "learning_rate": 1.9227216757839028e-05, "loss": 1.5744, "step": 4064 }, { "epoch": 0.14, "learning_rate": 1.9226808968881356e-05, "loss": 1.6729, "step": 4065 }, { "epoch": 0.14, "learning_rate": 1.922640107668597e-05, "loss": 1.6491, "step": 4066 }, { "epoch": 0.14, "learning_rate": 1.9225993081257435e-05, "loss": 1.602, "step": 4067 }, { "epoch": 0.14, "learning_rate": 1.9225584982600317e-05, "loss": 1.7026, "step": 4068 }, { "epoch": 0.14, "learning_rate": 1.922517678071918e-05, "loss": 1.7563, "step": 4069 }, { "epoch": 0.14, "learning_rate": 1.922476847561859e-05, "loss": 1.7741, "step": 4070 }, { "epoch": 0.14, "learning_rate": 1.9224360067303123e-05, "loss": 1.6066, "step": 4071 }, { "epoch": 0.14, "learning_rate": 1.922395155577734e-05, "loss": 1.7484, "step": 4072 }, { "epoch": 0.14, "learning_rate": 1.9223542941045817e-05, "loss": 1.6455, "step": 4073 }, { "epoch": 0.14, "learning_rate": 1.9223134223113122e-05, "loss": 1.579, "step": 4074 }, { "epoch": 0.14, "learning_rate": 1.9222725401983832e-05, "loss": 1.5512, "step": 4075 }, { "epoch": 0.14, "learning_rate": 1.9222316477662517e-05, "loss": 1.5734, "step": 4076 }, { "epoch": 0.14, "learning_rate": 1.9221907450153755e-05, "loss": 1.5645, "step": 4077 }, { "epoch": 0.14, "learning_rate": 1.9221498319462125e-05, "loss": 1.5921, "step": 4078 }, { "epoch": 0.14, "learning_rate": 1.9221089085592203e-05, "loss": 1.536, "step": 4079 }, { "epoch": 0.14, "learning_rate": 1.922067974854856e-05, "loss": 1.5461, "step": 4080 }, { "epoch": 0.14, "learning_rate": 1.9220270308335793e-05, "loss": 1.5734, "step": 4081 }, { "epoch": 0.14, "learning_rate": 1.9219860764958466e-05, "loss": 1.6429, "step": 4082 }, { "epoch": 0.14, "learning_rate": 1.9219451118421173e-05, "loss": 1.5448, "step": 4083 }, { "epoch": 0.14, "learning_rate": 1.9219041368728494e-05, "loss": 1.558, "step": 4084 }, { "epoch": 0.14, "learning_rate": 1.9218631515885007e-05, "loss": 1.4791, "step": 4085 }, { "epoch": 0.14, "learning_rate": 1.9218221559895306e-05, "loss": 1.5693, "step": 4086 }, { "epoch": 0.14, "learning_rate": 1.9217811500763975e-05, "loss": 1.4747, "step": 4087 }, { "epoch": 0.14, "learning_rate": 1.9217401338495605e-05, "loss": 1.5306, "step": 4088 }, { "epoch": 0.14, "learning_rate": 1.921699107309478e-05, "loss": 1.4945, "step": 4089 }, { "epoch": 0.14, "learning_rate": 1.9216580704566095e-05, "loss": 1.5337, "step": 4090 }, { "epoch": 0.14, "learning_rate": 1.921617023291414e-05, "loss": 1.5522, "step": 4091 }, { "epoch": 0.14, "learning_rate": 1.9215759658143507e-05, "loss": 1.54, "step": 4092 }, { "epoch": 0.14, "learning_rate": 1.921534898025879e-05, "loss": 1.5324, "step": 4093 }, { "epoch": 0.14, "learning_rate": 1.9214938199264584e-05, "loss": 1.5912, "step": 4094 }, { "epoch": 0.14, "learning_rate": 1.9214527315165487e-05, "loss": 1.5019, "step": 4095 }, { "epoch": 0.14, "learning_rate": 1.9214116327966095e-05, "loss": 1.4412, "step": 4096 }, { "epoch": 0.14, "learning_rate": 1.9213705237671007e-05, "loss": 1.4222, "step": 4097 }, { "epoch": 0.14, "learning_rate": 1.9213294044284822e-05, "loss": 1.4673, "step": 4098 }, { "epoch": 0.14, "learning_rate": 1.9212882747812136e-05, "loss": 1.5089, "step": 4099 }, { "epoch": 0.14, "learning_rate": 1.9212471348257562e-05, "loss": 1.5084, "step": 4100 }, { "epoch": 0.14, "learning_rate": 1.9212059845625693e-05, "loss": 1.5562, "step": 4101 }, { "epoch": 0.14, "learning_rate": 1.921164823992114e-05, "loss": 1.5843, "step": 4102 }, { "epoch": 0.14, "learning_rate": 1.92112365311485e-05, "loss": 1.4834, "step": 4103 }, { "epoch": 0.14, "learning_rate": 1.9210824719312395e-05, "loss": 1.5577, "step": 4104 }, { "epoch": 0.14, "learning_rate": 1.9210412804417413e-05, "loss": 1.5036, "step": 4105 }, { "epoch": 0.14, "learning_rate": 1.9210000786468178e-05, "loss": 1.4746, "step": 4106 }, { "epoch": 0.14, "learning_rate": 1.9209588665469294e-05, "loss": 1.507, "step": 4107 }, { "epoch": 0.14, "learning_rate": 1.9209176441425376e-05, "loss": 1.4801, "step": 4108 }, { "epoch": 0.14, "learning_rate": 1.9208764114341028e-05, "loss": 1.5631, "step": 4109 }, { "epoch": 0.14, "learning_rate": 1.920835168422087e-05, "loss": 1.5417, "step": 4110 }, { "epoch": 0.14, "learning_rate": 1.9207939151069515e-05, "loss": 1.3907, "step": 4111 }, { "epoch": 0.14, "learning_rate": 1.9207526514891582e-05, "loss": 1.5236, "step": 4112 }, { "epoch": 0.14, "learning_rate": 1.9207113775691682e-05, "loss": 1.6069, "step": 4113 }, { "epoch": 0.14, "learning_rate": 1.920670093347444e-05, "loss": 1.6447, "step": 4114 }, { "epoch": 0.14, "learning_rate": 1.9206287988244467e-05, "loss": 1.5278, "step": 4115 }, { "epoch": 0.14, "learning_rate": 1.920587494000639e-05, "loss": 1.5415, "step": 4116 }, { "epoch": 0.14, "learning_rate": 1.920546178876483e-05, "loss": 1.5148, "step": 4117 }, { "epoch": 0.14, "learning_rate": 1.9205048534524405e-05, "loss": 1.5463, "step": 4118 }, { "epoch": 0.14, "learning_rate": 1.9204635177289744e-05, "loss": 1.4978, "step": 4119 }, { "epoch": 0.14, "learning_rate": 1.9204221717065468e-05, "loss": 1.4463, "step": 4120 }, { "epoch": 0.14, "learning_rate": 1.9203808153856206e-05, "loss": 1.4927, "step": 4121 }, { "epoch": 0.14, "learning_rate": 1.9203394487666585e-05, "loss": 1.5817, "step": 4122 }, { "epoch": 0.14, "learning_rate": 1.920298071850123e-05, "loss": 1.445, "step": 4123 }, { "epoch": 0.14, "learning_rate": 1.9202566846364782e-05, "loss": 1.3726, "step": 4124 }, { "epoch": 0.14, "learning_rate": 1.9202152871261853e-05, "loss": 1.4554, "step": 4125 }, { "epoch": 0.14, "learning_rate": 1.9201738793197092e-05, "loss": 1.498, "step": 4126 }, { "epoch": 0.14, "learning_rate": 1.9201324612175123e-05, "loss": 1.4802, "step": 4127 }, { "epoch": 0.14, "learning_rate": 1.9200910328200583e-05, "loss": 1.4991, "step": 4128 }, { "epoch": 0.14, "learning_rate": 1.9200495941278105e-05, "loss": 1.599, "step": 4129 }, { "epoch": 0.14, "learning_rate": 1.920008145141233e-05, "loss": 1.4623, "step": 4130 }, { "epoch": 0.14, "learning_rate": 1.9199666858607892e-05, "loss": 1.629, "step": 4131 }, { "epoch": 0.14, "learning_rate": 1.919925216286943e-05, "loss": 1.5226, "step": 4132 }, { "epoch": 0.14, "learning_rate": 1.9198837364201587e-05, "loss": 1.5077, "step": 4133 }, { "epoch": 0.14, "learning_rate": 1.9198422462608998e-05, "loss": 1.5801, "step": 4134 }, { "epoch": 0.14, "learning_rate": 1.9198007458096313e-05, "loss": 1.5609, "step": 4135 }, { "epoch": 0.14, "learning_rate": 1.919759235066817e-05, "loss": 1.5142, "step": 4136 }, { "epoch": 0.14, "learning_rate": 1.9197177140329216e-05, "loss": 1.4281, "step": 4137 }, { "epoch": 0.14, "learning_rate": 1.9196761827084097e-05, "loss": 1.5137, "step": 4138 }, { "epoch": 0.14, "learning_rate": 1.9196346410937455e-05, "loss": 1.5282, "step": 4139 }, { "epoch": 0.14, "learning_rate": 1.9195930891893946e-05, "loss": 1.6069, "step": 4140 }, { "epoch": 0.14, "learning_rate": 1.9195515269958217e-05, "loss": 1.5679, "step": 4141 }, { "epoch": 0.14, "learning_rate": 1.9195099545134913e-05, "loss": 1.5397, "step": 4142 }, { "epoch": 0.14, "learning_rate": 1.9194683717428688e-05, "loss": 1.4899, "step": 4143 }, { "epoch": 0.14, "learning_rate": 1.91942677868442e-05, "loss": 1.5908, "step": 4144 }, { "epoch": 0.14, "learning_rate": 1.9193851753386095e-05, "loss": 1.6611, "step": 4145 }, { "epoch": 0.14, "learning_rate": 1.9193435617059034e-05, "loss": 1.6552, "step": 4146 }, { "epoch": 0.14, "learning_rate": 1.919301937786767e-05, "loss": 1.5771, "step": 4147 }, { "epoch": 0.14, "learning_rate": 1.9192603035816657e-05, "loss": 1.5119, "step": 4148 }, { "epoch": 0.14, "learning_rate": 1.9192186590910663e-05, "loss": 1.6116, "step": 4149 }, { "epoch": 0.14, "learning_rate": 1.9191770043154336e-05, "loss": 1.6618, "step": 4150 }, { "epoch": 0.14, "learning_rate": 1.9191353392552346e-05, "loss": 1.5868, "step": 4151 }, { "epoch": 0.14, "learning_rate": 1.919093663910935e-05, "loss": 1.5814, "step": 4152 }, { "epoch": 0.14, "learning_rate": 1.9190519782830016e-05, "loss": 1.6931, "step": 4153 }, { "epoch": 0.14, "learning_rate": 1.9190102823719e-05, "loss": 1.6609, "step": 4154 }, { "epoch": 0.14, "learning_rate": 1.9189685761780974e-05, "loss": 1.6991, "step": 4155 }, { "epoch": 0.14, "learning_rate": 1.9189268597020597e-05, "loss": 1.5761, "step": 4156 }, { "epoch": 0.14, "learning_rate": 1.918885132944255e-05, "loss": 1.5794, "step": 4157 }, { "epoch": 0.14, "learning_rate": 1.918843395905149e-05, "loss": 1.5904, "step": 4158 }, { "epoch": 0.14, "learning_rate": 1.9188016485852092e-05, "loss": 1.6979, "step": 4159 }, { "epoch": 0.14, "learning_rate": 1.9187598909849023e-05, "loss": 1.6072, "step": 4160 }, { "epoch": 0.14, "learning_rate": 1.918718123104696e-05, "loss": 1.6742, "step": 4161 }, { "epoch": 0.14, "learning_rate": 1.9186763449450572e-05, "loss": 1.5501, "step": 4162 }, { "epoch": 0.14, "learning_rate": 1.918634556506454e-05, "loss": 1.4705, "step": 4163 }, { "epoch": 0.14, "learning_rate": 1.918592757789353e-05, "loss": 1.518, "step": 4164 }, { "epoch": 0.14, "learning_rate": 1.9185509487942226e-05, "loss": 1.641, "step": 4165 }, { "epoch": 0.14, "learning_rate": 1.9185091295215305e-05, "loss": 1.5767, "step": 4166 }, { "epoch": 0.14, "learning_rate": 1.9184672999717442e-05, "loss": 1.6301, "step": 4167 }, { "epoch": 0.14, "learning_rate": 1.918425460145332e-05, "loss": 1.54, "step": 4168 }, { "epoch": 0.14, "learning_rate": 1.9183836100427627e-05, "loss": 1.5809, "step": 4169 }, { "epoch": 0.14, "learning_rate": 1.918341749664504e-05, "loss": 1.5068, "step": 4170 }, { "epoch": 0.14, "learning_rate": 1.9182998790110236e-05, "loss": 1.5445, "step": 4171 }, { "epoch": 0.14, "learning_rate": 1.9182579980827908e-05, "loss": 1.5626, "step": 4172 }, { "epoch": 0.14, "learning_rate": 1.9182161068802742e-05, "loss": 1.6062, "step": 4173 }, { "epoch": 0.14, "learning_rate": 1.918174205403942e-05, "loss": 1.6148, "step": 4174 }, { "epoch": 0.14, "learning_rate": 1.9181322936542638e-05, "loss": 1.6489, "step": 4175 }, { "epoch": 0.14, "learning_rate": 1.9180903716317077e-05, "loss": 1.5118, "step": 4176 }, { "epoch": 0.14, "learning_rate": 1.9180484393367434e-05, "loss": 1.6541, "step": 4177 }, { "epoch": 0.14, "learning_rate": 1.91800649676984e-05, "loss": 1.5069, "step": 4178 }, { "epoch": 0.14, "learning_rate": 1.917964543931466e-05, "loss": 1.5794, "step": 4179 }, { "epoch": 0.14, "learning_rate": 1.9179225808220918e-05, "loss": 1.5323, "step": 4180 }, { "epoch": 0.14, "learning_rate": 1.9178806074421866e-05, "loss": 1.543, "step": 4181 }, { "epoch": 0.14, "learning_rate": 1.91783862379222e-05, "loss": 1.6269, "step": 4182 }, { "epoch": 0.14, "learning_rate": 1.9177966298726615e-05, "loss": 1.4472, "step": 4183 }, { "epoch": 0.14, "learning_rate": 1.9177546256839814e-05, "loss": 1.5286, "step": 4184 }, { "epoch": 0.14, "learning_rate": 1.9177126112266493e-05, "loss": 1.4736, "step": 4185 }, { "epoch": 0.14, "learning_rate": 1.9176705865011356e-05, "loss": 1.4928, "step": 4186 }, { "epoch": 0.14, "learning_rate": 1.9176285515079102e-05, "loss": 1.6045, "step": 4187 }, { "epoch": 0.14, "learning_rate": 1.917586506247444e-05, "loss": 1.4821, "step": 4188 }, { "epoch": 0.14, "learning_rate": 1.9175444507202068e-05, "loss": 1.4855, "step": 4189 }, { "epoch": 0.14, "learning_rate": 1.9175023849266697e-05, "loss": 1.648, "step": 4190 }, { "epoch": 0.14, "learning_rate": 1.9174603088673026e-05, "loss": 1.541, "step": 4191 }, { "epoch": 0.14, "learning_rate": 1.917418222542577e-05, "loss": 1.4894, "step": 4192 }, { "epoch": 0.14, "learning_rate": 1.9173761259529634e-05, "loss": 1.5939, "step": 4193 }, { "epoch": 0.14, "learning_rate": 1.9173340190989332e-05, "loss": 1.4882, "step": 4194 }, { "epoch": 0.14, "learning_rate": 1.9172919019809572e-05, "loss": 1.4299, "step": 4195 }, { "epoch": 0.14, "learning_rate": 1.9172497745995068e-05, "loss": 1.4362, "step": 4196 }, { "epoch": 0.14, "learning_rate": 1.917207636955053e-05, "loss": 1.3769, "step": 4197 }, { "epoch": 0.14, "learning_rate": 1.917165489048068e-05, "loss": 1.4122, "step": 4198 }, { "epoch": 0.14, "learning_rate": 1.9171233308790225e-05, "loss": 1.4157, "step": 4199 }, { "epoch": 0.14, "learning_rate": 1.9170811624483893e-05, "loss": 1.4229, "step": 4200 }, { "epoch": 0.14, "learning_rate": 1.9170389837566394e-05, "loss": 1.4522, "step": 4201 }, { "epoch": 0.14, "learning_rate": 1.9169967948042444e-05, "loss": 1.4699, "step": 4202 }, { "epoch": 0.14, "learning_rate": 1.9169545955916773e-05, "loss": 1.4235, "step": 4203 }, { "epoch": 0.14, "learning_rate": 1.9169123861194097e-05, "loss": 1.4747, "step": 4204 }, { "epoch": 0.14, "learning_rate": 1.9168701663879143e-05, "loss": 1.5462, "step": 4205 }, { "epoch": 0.14, "learning_rate": 1.9168279363976627e-05, "loss": 1.5312, "step": 4206 }, { "epoch": 0.14, "learning_rate": 1.916785696149128e-05, "loss": 1.5089, "step": 4207 }, { "epoch": 0.14, "learning_rate": 1.916743445642783e-05, "loss": 1.5346, "step": 4208 }, { "epoch": 0.14, "learning_rate": 1.9167011848791e-05, "loss": 1.6103, "step": 4209 }, { "epoch": 0.14, "learning_rate": 1.9166589138585517e-05, "loss": 1.4733, "step": 4210 }, { "epoch": 0.14, "learning_rate": 1.916616632581612e-05, "loss": 1.4816, "step": 4211 }, { "epoch": 0.14, "learning_rate": 1.916574341048753e-05, "loss": 1.6065, "step": 4212 }, { "epoch": 0.14, "learning_rate": 1.916532039260448e-05, "loss": 1.5756, "step": 4213 }, { "epoch": 0.14, "learning_rate": 1.916489727217171e-05, "loss": 1.5339, "step": 4214 }, { "epoch": 0.14, "learning_rate": 1.9164474049193947e-05, "loss": 1.5933, "step": 4215 }, { "epoch": 0.14, "learning_rate": 1.9164050723675932e-05, "loss": 1.5062, "step": 4216 }, { "epoch": 0.14, "learning_rate": 1.9163627295622397e-05, "loss": 1.5504, "step": 4217 }, { "epoch": 0.14, "learning_rate": 1.9163203765038082e-05, "loss": 1.4622, "step": 4218 }, { "epoch": 0.14, "learning_rate": 1.9162780131927724e-05, "loss": 1.5273, "step": 4219 }, { "epoch": 0.14, "learning_rate": 1.9162356396296068e-05, "loss": 1.5294, "step": 4220 }, { "epoch": 0.14, "learning_rate": 1.916193255814785e-05, "loss": 1.5916, "step": 4221 }, { "epoch": 0.14, "learning_rate": 1.916150861748781e-05, "loss": 1.6022, "step": 4222 }, { "epoch": 0.14, "learning_rate": 1.9161084574320696e-05, "loss": 1.5395, "step": 4223 }, { "epoch": 0.14, "learning_rate": 1.9160660428651257e-05, "loss": 1.4667, "step": 4224 }, { "epoch": 0.14, "learning_rate": 1.9160236180484227e-05, "loss": 1.4539, "step": 4225 }, { "epoch": 0.14, "learning_rate": 1.9159811829824364e-05, "loss": 1.5225, "step": 4226 }, { "epoch": 0.14, "learning_rate": 1.9159387376676408e-05, "loss": 1.6224, "step": 4227 }, { "epoch": 0.14, "learning_rate": 1.9158962821045113e-05, "loss": 1.5764, "step": 4228 }, { "epoch": 0.14, "learning_rate": 1.9158538162935227e-05, "loss": 1.4057, "step": 4229 }, { "epoch": 0.14, "learning_rate": 1.91581134023515e-05, "loss": 1.5369, "step": 4230 }, { "epoch": 0.14, "learning_rate": 1.915768853929869e-05, "loss": 1.5238, "step": 4231 }, { "epoch": 0.14, "learning_rate": 1.915726357378155e-05, "loss": 1.5494, "step": 4232 }, { "epoch": 0.14, "learning_rate": 1.9156838505804826e-05, "loss": 1.5851, "step": 4233 }, { "epoch": 0.14, "learning_rate": 1.9156413335373284e-05, "loss": 1.6678, "step": 4234 }, { "epoch": 0.14, "learning_rate": 1.915598806249168e-05, "loss": 1.6518, "step": 4235 }, { "epoch": 0.14, "learning_rate": 1.9155562687164765e-05, "loss": 1.4865, "step": 4236 }, { "epoch": 0.14, "learning_rate": 1.915513720939731e-05, "loss": 1.5316, "step": 4237 }, { "epoch": 0.14, "learning_rate": 1.9154711629194062e-05, "loss": 1.511, "step": 4238 }, { "epoch": 0.14, "learning_rate": 1.9154285946559792e-05, "loss": 1.3839, "step": 4239 }, { "epoch": 0.14, "learning_rate": 1.9153860161499263e-05, "loss": 1.4527, "step": 4240 }, { "epoch": 0.14, "learning_rate": 1.9153434274017234e-05, "loss": 1.5312, "step": 4241 }, { "epoch": 0.14, "learning_rate": 1.9153008284118476e-05, "loss": 1.498, "step": 4242 }, { "epoch": 0.14, "learning_rate": 1.915258219180775e-05, "loss": 1.5475, "step": 4243 }, { "epoch": 0.14, "learning_rate": 1.9152155997089824e-05, "loss": 1.549, "step": 4244 }, { "epoch": 0.14, "learning_rate": 1.9151729699969475e-05, "loss": 1.6117, "step": 4245 }, { "epoch": 0.14, "learning_rate": 1.915130330045146e-05, "loss": 1.6242, "step": 4246 }, { "epoch": 0.14, "learning_rate": 1.915087679854056e-05, "loss": 1.557, "step": 4247 }, { "epoch": 0.14, "learning_rate": 1.915045019424154e-05, "loss": 1.6036, "step": 4248 }, { "epoch": 0.14, "learning_rate": 1.915002348755918e-05, "loss": 1.5435, "step": 4249 }, { "epoch": 0.14, "learning_rate": 1.914959667849825e-05, "loss": 1.5114, "step": 4250 }, { "epoch": 0.14, "learning_rate": 1.9149169767063528e-05, "loss": 1.4597, "step": 4251 }, { "epoch": 0.14, "learning_rate": 1.9148742753259783e-05, "loss": 1.4925, "step": 4252 }, { "epoch": 0.14, "learning_rate": 1.9148315637091805e-05, "loss": 1.4944, "step": 4253 }, { "epoch": 0.14, "learning_rate": 1.9147888418564364e-05, "loss": 1.5045, "step": 4254 }, { "epoch": 0.14, "learning_rate": 1.9147461097682244e-05, "loss": 1.4603, "step": 4255 }, { "epoch": 0.14, "learning_rate": 1.9147033674450223e-05, "loss": 1.5109, "step": 4256 }, { "epoch": 0.14, "learning_rate": 1.9146606148873088e-05, "loss": 1.4998, "step": 4257 }, { "epoch": 0.14, "learning_rate": 1.914617852095562e-05, "loss": 1.6172, "step": 4258 }, { "epoch": 0.14, "learning_rate": 1.9145750790702606e-05, "loss": 1.4158, "step": 4259 }, { "epoch": 0.14, "learning_rate": 1.9145322958118826e-05, "loss": 1.4384, "step": 4260 }, { "epoch": 0.14, "learning_rate": 1.9144895023209072e-05, "loss": 1.487, "step": 4261 }, { "epoch": 0.14, "learning_rate": 1.914446698597813e-05, "loss": 1.5653, "step": 4262 }, { "epoch": 0.14, "learning_rate": 1.914403884643079e-05, "loss": 1.5476, "step": 4263 }, { "epoch": 0.14, "learning_rate": 1.914361060457184e-05, "loss": 1.469, "step": 4264 }, { "epoch": 0.14, "learning_rate": 1.914318226040608e-05, "loss": 1.597, "step": 4265 }, { "epoch": 0.14, "learning_rate": 1.9142753813938293e-05, "loss": 1.4687, "step": 4266 }, { "epoch": 0.14, "learning_rate": 1.914232526517328e-05, "loss": 1.5311, "step": 4267 }, { "epoch": 0.14, "learning_rate": 1.9141896614115824e-05, "loss": 1.5287, "step": 4268 }, { "epoch": 0.14, "learning_rate": 1.9141467860770737e-05, "loss": 1.4717, "step": 4269 }, { "epoch": 0.14, "learning_rate": 1.9141039005142807e-05, "loss": 1.5552, "step": 4270 }, { "epoch": 0.14, "learning_rate": 1.9140610047236834e-05, "loss": 1.5954, "step": 4271 }, { "epoch": 0.14, "learning_rate": 1.914018098705762e-05, "loss": 1.4574, "step": 4272 }, { "epoch": 0.14, "learning_rate": 1.913975182460996e-05, "loss": 1.5749, "step": 4273 }, { "epoch": 0.14, "learning_rate": 1.913932255989866e-05, "loss": 1.5497, "step": 4274 }, { "epoch": 0.14, "learning_rate": 1.9138893192928523e-05, "loss": 1.5294, "step": 4275 }, { "epoch": 0.14, "learning_rate": 1.9138463723704352e-05, "loss": 1.4841, "step": 4276 }, { "epoch": 0.14, "learning_rate": 1.9138034152230952e-05, "loss": 1.5842, "step": 4277 }, { "epoch": 0.14, "learning_rate": 1.9137604478513136e-05, "loss": 1.4595, "step": 4278 }, { "epoch": 0.14, "learning_rate": 1.91371747025557e-05, "loss": 1.5552, "step": 4279 }, { "epoch": 0.14, "learning_rate": 1.913674482436346e-05, "loss": 1.501, "step": 4280 }, { "epoch": 0.14, "learning_rate": 1.9136314843941223e-05, "loss": 1.5016, "step": 4281 }, { "epoch": 0.14, "learning_rate": 1.9135884761293803e-05, "loss": 1.5171, "step": 4282 }, { "epoch": 0.14, "learning_rate": 1.913545457642601e-05, "loss": 1.5006, "step": 4283 }, { "epoch": 0.14, "learning_rate": 1.9135024289342658e-05, "loss": 1.5077, "step": 4284 }, { "epoch": 0.14, "learning_rate": 1.9134593900048562e-05, "loss": 1.5047, "step": 4285 }, { "epoch": 0.14, "learning_rate": 1.9134163408548537e-05, "loss": 1.5492, "step": 4286 }, { "epoch": 0.14, "learning_rate": 1.9133732814847396e-05, "loss": 1.451, "step": 4287 }, { "epoch": 0.14, "learning_rate": 1.9133302118949966e-05, "loss": 1.4811, "step": 4288 }, { "epoch": 0.14, "learning_rate": 1.9132871320861056e-05, "loss": 1.4738, "step": 4289 }, { "epoch": 0.14, "learning_rate": 1.913244042058549e-05, "loss": 1.4813, "step": 4290 }, { "epoch": 0.14, "learning_rate": 1.9132009418128094e-05, "loss": 1.577, "step": 4291 }, { "epoch": 0.14, "learning_rate": 1.9131578313493685e-05, "loss": 1.6388, "step": 4292 }, { "epoch": 0.14, "learning_rate": 1.9131147106687085e-05, "loss": 1.4448, "step": 4293 }, { "epoch": 0.14, "learning_rate": 1.9130715797713123e-05, "loss": 1.5284, "step": 4294 }, { "epoch": 0.14, "learning_rate": 1.9130284386576627e-05, "loss": 1.3667, "step": 4295 }, { "epoch": 0.14, "learning_rate": 1.9129852873282417e-05, "loss": 1.5812, "step": 4296 }, { "epoch": 0.14, "learning_rate": 1.9129421257835322e-05, "loss": 1.4029, "step": 4297 }, { "epoch": 0.14, "learning_rate": 1.9128989540240178e-05, "loss": 1.4368, "step": 4298 }, { "epoch": 0.14, "learning_rate": 1.912855772050181e-05, "loss": 1.4298, "step": 4299 }, { "epoch": 0.14, "learning_rate": 1.9128125798625054e-05, "loss": 1.4604, "step": 4300 }, { "epoch": 0.14, "learning_rate": 1.9127693774614738e-05, "loss": 1.5218, "step": 4301 }, { "epoch": 0.14, "learning_rate": 1.9127261648475698e-05, "loss": 1.3902, "step": 4302 }, { "epoch": 0.14, "learning_rate": 1.9126829420212764e-05, "loss": 1.5701, "step": 4303 }, { "epoch": 0.14, "learning_rate": 1.912639708983078e-05, "loss": 1.5785, "step": 4304 }, { "epoch": 0.14, "learning_rate": 1.912596465733458e-05, "loss": 1.4604, "step": 4305 }, { "epoch": 0.14, "learning_rate": 1.9125532122729002e-05, "loss": 1.561, "step": 4306 }, { "epoch": 0.14, "learning_rate": 1.9125099486018887e-05, "loss": 1.4516, "step": 4307 }, { "epoch": 0.14, "learning_rate": 1.9124666747209075e-05, "loss": 1.5733, "step": 4308 }, { "epoch": 0.14, "learning_rate": 1.9124233906304406e-05, "loss": 1.4492, "step": 4309 }, { "epoch": 0.14, "learning_rate": 1.9123800963309725e-05, "loss": 1.4605, "step": 4310 }, { "epoch": 0.14, "learning_rate": 1.9123367918229875e-05, "loss": 1.6116, "step": 4311 }, { "epoch": 0.14, "learning_rate": 1.9122934771069704e-05, "loss": 1.7266, "step": 4312 }, { "epoch": 0.14, "learning_rate": 1.9122501521834052e-05, "loss": 1.5776, "step": 4313 }, { "epoch": 0.14, "learning_rate": 1.912206817052778e-05, "loss": 1.433, "step": 4314 }, { "epoch": 0.14, "learning_rate": 1.9121634717155718e-05, "loss": 1.5132, "step": 4315 }, { "epoch": 0.14, "learning_rate": 1.9121201161722732e-05, "loss": 1.5394, "step": 4316 }, { "epoch": 0.14, "learning_rate": 1.9120767504233663e-05, "loss": 1.5385, "step": 4317 }, { "epoch": 0.14, "learning_rate": 1.912033374469337e-05, "loss": 1.6755, "step": 4318 }, { "epoch": 0.14, "learning_rate": 1.9119899883106702e-05, "loss": 1.5687, "step": 4319 }, { "epoch": 0.14, "learning_rate": 1.9119465919478513e-05, "loss": 1.6751, "step": 4320 }, { "epoch": 0.14, "learning_rate": 1.9119031853813662e-05, "loss": 1.7236, "step": 4321 }, { "epoch": 0.14, "learning_rate": 1.9118597686117e-05, "loss": 1.6294, "step": 4322 }, { "epoch": 0.14, "learning_rate": 1.9118163416393392e-05, "loss": 1.6246, "step": 4323 }, { "epoch": 0.14, "learning_rate": 1.9117729044647694e-05, "loss": 1.6443, "step": 4324 }, { "epoch": 0.14, "learning_rate": 1.9117294570884764e-05, "loss": 1.7282, "step": 4325 }, { "epoch": 0.14, "learning_rate": 1.9116859995109466e-05, "loss": 1.5927, "step": 4326 }, { "epoch": 0.14, "learning_rate": 1.911642531732666e-05, "loss": 1.5232, "step": 4327 }, { "epoch": 0.14, "learning_rate": 1.9115990537541217e-05, "loss": 1.6314, "step": 4328 }, { "epoch": 0.14, "learning_rate": 1.9115555655757987e-05, "loss": 1.7172, "step": 4329 }, { "epoch": 0.14, "learning_rate": 1.911512067198185e-05, "loss": 1.6639, "step": 4330 }, { "epoch": 0.14, "learning_rate": 1.9114685586217666e-05, "loss": 1.6616, "step": 4331 }, { "epoch": 0.14, "learning_rate": 1.9114250398470303e-05, "loss": 1.6569, "step": 4332 }, { "epoch": 0.14, "learning_rate": 1.9113815108744635e-05, "loss": 1.5483, "step": 4333 }, { "epoch": 0.14, "learning_rate": 1.9113379717045528e-05, "loss": 1.6029, "step": 4334 }, { "epoch": 0.14, "learning_rate": 1.9112944223377857e-05, "loss": 1.5875, "step": 4335 }, { "epoch": 0.14, "learning_rate": 1.9112508627746487e-05, "loss": 1.5935, "step": 4336 }, { "epoch": 0.14, "learning_rate": 1.91120729301563e-05, "loss": 1.6522, "step": 4337 }, { "epoch": 0.14, "learning_rate": 1.9111637130612172e-05, "loss": 1.5604, "step": 4338 }, { "epoch": 0.14, "learning_rate": 1.9111201229118974e-05, "loss": 1.5993, "step": 4339 }, { "epoch": 0.14, "learning_rate": 1.9110765225681582e-05, "loss": 1.6535, "step": 4340 }, { "epoch": 0.14, "learning_rate": 1.9110329120304877e-05, "loss": 1.6597, "step": 4341 }, { "epoch": 0.14, "learning_rate": 1.9109892912993744e-05, "loss": 1.6339, "step": 4342 }, { "epoch": 0.14, "learning_rate": 1.910945660375305e-05, "loss": 1.577, "step": 4343 }, { "epoch": 0.14, "learning_rate": 1.9109020192587693e-05, "loss": 1.597, "step": 4344 }, { "epoch": 0.14, "learning_rate": 1.9108583679502545e-05, "loss": 1.6482, "step": 4345 }, { "epoch": 0.14, "learning_rate": 1.9108147064502494e-05, "loss": 1.5598, "step": 4346 }, { "epoch": 0.14, "learning_rate": 1.9107710347592422e-05, "loss": 1.6173, "step": 4347 }, { "epoch": 0.14, "learning_rate": 1.910727352877722e-05, "loss": 1.6675, "step": 4348 }, { "epoch": 0.14, "learning_rate": 1.910683660806177e-05, "loss": 1.7176, "step": 4349 }, { "epoch": 0.14, "learning_rate": 1.910639958545097e-05, "loss": 1.6935, "step": 4350 }, { "epoch": 0.15, "learning_rate": 1.9105962460949698e-05, "loss": 1.4884, "step": 4351 }, { "epoch": 0.15, "learning_rate": 1.9105525234562858e-05, "loss": 1.6611, "step": 4352 }, { "epoch": 0.15, "learning_rate": 1.910508790629533e-05, "loss": 1.5691, "step": 4353 }, { "epoch": 0.15, "learning_rate": 1.910465047615201e-05, "loss": 1.6369, "step": 4354 }, { "epoch": 0.15, "learning_rate": 1.9104212944137796e-05, "loss": 1.5805, "step": 4355 }, { "epoch": 0.15, "learning_rate": 1.910377531025758e-05, "loss": 1.4921, "step": 4356 }, { "epoch": 0.15, "learning_rate": 1.9103337574516266e-05, "loss": 1.7077, "step": 4357 }, { "epoch": 0.15, "learning_rate": 1.9102899736918742e-05, "loss": 1.6258, "step": 4358 }, { "epoch": 0.15, "learning_rate": 1.910246179746991e-05, "loss": 1.4937, "step": 4359 }, { "epoch": 0.15, "learning_rate": 1.9102023756174675e-05, "loss": 1.4569, "step": 4360 }, { "epoch": 0.15, "learning_rate": 1.9101585613037933e-05, "loss": 1.4388, "step": 4361 }, { "epoch": 0.15, "learning_rate": 1.910114736806459e-05, "loss": 1.5092, "step": 4362 }, { "epoch": 0.15, "learning_rate": 1.9100709021259544e-05, "loss": 1.5187, "step": 4363 }, { "epoch": 0.15, "learning_rate": 1.9100270572627705e-05, "loss": 1.4984, "step": 4364 }, { "epoch": 0.15, "learning_rate": 1.909983202217397e-05, "loss": 1.5301, "step": 4365 }, { "epoch": 0.15, "learning_rate": 1.9099393369903262e-05, "loss": 1.4903, "step": 4366 }, { "epoch": 0.15, "learning_rate": 1.9098954615820477e-05, "loss": 1.5065, "step": 4367 }, { "epoch": 0.15, "learning_rate": 1.909851575993052e-05, "loss": 1.5025, "step": 4368 }, { "epoch": 0.15, "learning_rate": 1.9098076802238318e-05, "loss": 1.5207, "step": 4369 }, { "epoch": 0.15, "learning_rate": 1.9097637742748768e-05, "loss": 1.489, "step": 4370 }, { "epoch": 0.15, "learning_rate": 1.9097198581466785e-05, "loss": 1.5521, "step": 4371 }, { "epoch": 0.15, "learning_rate": 1.9096759318397287e-05, "loss": 1.4474, "step": 4372 }, { "epoch": 0.15, "learning_rate": 1.9096319953545186e-05, "loss": 1.4603, "step": 4373 }, { "epoch": 0.15, "learning_rate": 1.9095880486915395e-05, "loss": 1.4852, "step": 4374 }, { "epoch": 0.15, "learning_rate": 1.909544091851284e-05, "loss": 1.5462, "step": 4375 }, { "epoch": 0.15, "learning_rate": 1.9095001248342436e-05, "loss": 1.5237, "step": 4376 }, { "epoch": 0.15, "learning_rate": 1.9094561476409095e-05, "loss": 1.6084, "step": 4377 }, { "epoch": 0.15, "learning_rate": 1.9094121602717747e-05, "loss": 1.5894, "step": 4378 }, { "epoch": 0.15, "learning_rate": 1.9093681627273306e-05, "loss": 1.5308, "step": 4379 }, { "epoch": 0.15, "learning_rate": 1.90932415500807e-05, "loss": 1.5633, "step": 4380 }, { "epoch": 0.15, "learning_rate": 1.909280137114485e-05, "loss": 1.6356, "step": 4381 }, { "epoch": 0.15, "learning_rate": 1.9092361090470688e-05, "loss": 1.607, "step": 4382 }, { "epoch": 0.15, "learning_rate": 1.909192070806313e-05, "loss": 1.5578, "step": 4383 }, { "epoch": 0.15, "learning_rate": 1.909148022392711e-05, "loss": 1.6372, "step": 4384 }, { "epoch": 0.15, "learning_rate": 1.9091039638067555e-05, "loss": 1.5773, "step": 4385 }, { "epoch": 0.15, "learning_rate": 1.9090598950489393e-05, "loss": 1.5205, "step": 4386 }, { "epoch": 0.15, "learning_rate": 1.9090158161197557e-05, "loss": 1.4287, "step": 4387 }, { "epoch": 0.15, "learning_rate": 1.9089717270196982e-05, "loss": 1.4356, "step": 4388 }, { "epoch": 0.15, "learning_rate": 1.9089276277492593e-05, "loss": 1.5137, "step": 4389 }, { "epoch": 0.15, "learning_rate": 1.9088835183089328e-05, "loss": 1.4718, "step": 4390 }, { "epoch": 0.15, "learning_rate": 1.9088393986992124e-05, "loss": 1.591, "step": 4391 }, { "epoch": 0.15, "learning_rate": 1.9087952689205918e-05, "loss": 1.5609, "step": 4392 }, { "epoch": 0.15, "learning_rate": 1.9087511289735646e-05, "loss": 1.6612, "step": 4393 }, { "epoch": 0.15, "learning_rate": 1.9087069788586245e-05, "loss": 1.5633, "step": 4394 }, { "epoch": 0.15, "learning_rate": 1.9086628185762658e-05, "loss": 1.5409, "step": 4395 }, { "epoch": 0.15, "learning_rate": 1.908618648126982e-05, "loss": 1.5497, "step": 4396 }, { "epoch": 0.15, "learning_rate": 1.9085744675112682e-05, "loss": 1.538, "step": 4397 }, { "epoch": 0.15, "learning_rate": 1.9085302767296183e-05, "loss": 1.5985, "step": 4398 }, { "epoch": 0.15, "learning_rate": 1.9084860757825268e-05, "loss": 1.4962, "step": 4399 }, { "epoch": 0.15, "learning_rate": 1.9084418646704884e-05, "loss": 1.5368, "step": 4400 }, { "epoch": 0.15, "learning_rate": 1.9083976433939973e-05, "loss": 1.5359, "step": 4401 }, { "epoch": 0.15, "learning_rate": 1.9083534119535486e-05, "loss": 1.5418, "step": 4402 }, { "epoch": 0.15, "learning_rate": 1.9083091703496373e-05, "loss": 1.4896, "step": 4403 }, { "epoch": 0.15, "learning_rate": 1.9082649185827583e-05, "loss": 1.5353, "step": 4404 }, { "epoch": 0.15, "learning_rate": 1.9082206566534068e-05, "loss": 1.482, "step": 4405 }, { "epoch": 0.15, "learning_rate": 1.9081763845620777e-05, "loss": 1.5102, "step": 4406 }, { "epoch": 0.15, "learning_rate": 1.908132102309267e-05, "loss": 1.513, "step": 4407 }, { "epoch": 0.15, "learning_rate": 1.9080878098954697e-05, "loss": 1.4219, "step": 4408 }, { "epoch": 0.15, "learning_rate": 1.9080435073211812e-05, "loss": 1.5078, "step": 4409 }, { "epoch": 0.15, "learning_rate": 1.9079991945868976e-05, "loss": 1.5644, "step": 4410 }, { "epoch": 0.15, "learning_rate": 1.9079548716931147e-05, "loss": 1.6665, "step": 4411 }, { "epoch": 0.15, "learning_rate": 1.9079105386403283e-05, "loss": 1.5507, "step": 4412 }, { "epoch": 0.15, "learning_rate": 1.907866195429034e-05, "loss": 1.5057, "step": 4413 }, { "epoch": 0.15, "learning_rate": 1.907821842059729e-05, "loss": 1.5888, "step": 4414 }, { "epoch": 0.15, "learning_rate": 1.907777478532909e-05, "loss": 1.5402, "step": 4415 }, { "epoch": 0.15, "learning_rate": 1.90773310484907e-05, "loss": 1.6146, "step": 4416 }, { "epoch": 0.15, "learning_rate": 1.9076887210087086e-05, "loss": 1.5545, "step": 4417 }, { "epoch": 0.15, "learning_rate": 1.9076443270123222e-05, "loss": 1.6027, "step": 4418 }, { "epoch": 0.15, "learning_rate": 1.9075999228604068e-05, "loss": 1.4779, "step": 4419 }, { "epoch": 0.15, "learning_rate": 1.9075555085534594e-05, "loss": 1.5723, "step": 4420 }, { "epoch": 0.15, "learning_rate": 1.9075110840919765e-05, "loss": 1.6638, "step": 4421 }, { "epoch": 0.15, "learning_rate": 1.907466649476456e-05, "loss": 1.528, "step": 4422 }, { "epoch": 0.15, "learning_rate": 1.9074222047073945e-05, "loss": 1.6888, "step": 4423 }, { "epoch": 0.15, "learning_rate": 1.90737774978529e-05, "loss": 1.6196, "step": 4424 }, { "epoch": 0.15, "learning_rate": 1.907333284710639e-05, "loss": 1.4725, "step": 4425 }, { "epoch": 0.15, "learning_rate": 1.907288809483939e-05, "loss": 1.4048, "step": 4426 }, { "epoch": 0.15, "learning_rate": 1.9072443241056884e-05, "loss": 1.537, "step": 4427 }, { "epoch": 0.15, "learning_rate": 1.9071998285763845e-05, "loss": 1.5447, "step": 4428 }, { "epoch": 0.15, "learning_rate": 1.907155322896525e-05, "loss": 1.4902, "step": 4429 }, { "epoch": 0.15, "learning_rate": 1.907110807066608e-05, "loss": 1.5169, "step": 4430 }, { "epoch": 0.15, "learning_rate": 1.9070662810871325e-05, "loss": 1.4432, "step": 4431 }, { "epoch": 0.15, "learning_rate": 1.9070217449585948e-05, "loss": 1.5218, "step": 4432 }, { "epoch": 0.15, "learning_rate": 1.9069771986814948e-05, "loss": 1.4234, "step": 4433 }, { "epoch": 0.15, "learning_rate": 1.9069326422563302e-05, "loss": 1.5463, "step": 4434 }, { "epoch": 0.15, "learning_rate": 1.9068880756836e-05, "loss": 1.5318, "step": 4435 }, { "epoch": 0.15, "learning_rate": 1.9068434989638023e-05, "loss": 1.5932, "step": 4436 }, { "epoch": 0.15, "learning_rate": 1.906798912097436e-05, "loss": 1.5588, "step": 4437 }, { "epoch": 0.15, "learning_rate": 1.9067543150850003e-05, "loss": 1.41, "step": 4438 }, { "epoch": 0.15, "learning_rate": 1.9067097079269942e-05, "loss": 1.4542, "step": 4439 }, { "epoch": 0.15, "learning_rate": 1.9066650906239163e-05, "loss": 1.516, "step": 4440 }, { "epoch": 0.15, "learning_rate": 1.9066204631762663e-05, "loss": 1.5186, "step": 4441 }, { "epoch": 0.15, "learning_rate": 1.9065758255845432e-05, "loss": 1.4671, "step": 4442 }, { "epoch": 0.15, "learning_rate": 1.906531177849247e-05, "loss": 1.5222, "step": 4443 }, { "epoch": 0.15, "learning_rate": 1.9064865199708765e-05, "loss": 1.4427, "step": 4444 }, { "epoch": 0.15, "learning_rate": 1.9064418519499316e-05, "loss": 1.4737, "step": 4445 }, { "epoch": 0.15, "learning_rate": 1.9063971737869123e-05, "loss": 1.6121, "step": 4446 }, { "epoch": 0.15, "learning_rate": 1.9063524854823184e-05, "loss": 1.5244, "step": 4447 }, { "epoch": 0.15, "learning_rate": 1.9063077870366504e-05, "loss": 1.5493, "step": 4448 }, { "epoch": 0.15, "learning_rate": 1.9062630784504072e-05, "loss": 1.6221, "step": 4449 }, { "epoch": 0.15, "learning_rate": 1.90621835972409e-05, "loss": 1.5144, "step": 4450 }, { "epoch": 0.15, "learning_rate": 1.9061736308581996e-05, "loss": 1.5643, "step": 4451 }, { "epoch": 0.15, "learning_rate": 1.906128891853235e-05, "loss": 1.4844, "step": 4452 }, { "epoch": 0.15, "learning_rate": 1.9060841427096978e-05, "loss": 1.5804, "step": 4453 }, { "epoch": 0.15, "learning_rate": 1.906039383428089e-05, "loss": 1.5815, "step": 4454 }, { "epoch": 0.15, "learning_rate": 1.905994614008908e-05, "loss": 1.4371, "step": 4455 }, { "epoch": 0.15, "learning_rate": 1.9059498344526576e-05, "loss": 1.5112, "step": 4456 }, { "epoch": 0.15, "learning_rate": 1.905905044759837e-05, "loss": 1.4878, "step": 4457 }, { "epoch": 0.15, "learning_rate": 1.9058602449309487e-05, "loss": 1.4084, "step": 4458 }, { "epoch": 0.15, "learning_rate": 1.9058154349664932e-05, "loss": 1.5972, "step": 4459 }, { "epoch": 0.15, "learning_rate": 1.905770614866972e-05, "loss": 1.4319, "step": 4460 }, { "epoch": 0.15, "learning_rate": 1.905725784632887e-05, "loss": 1.4213, "step": 4461 }, { "epoch": 0.15, "learning_rate": 1.9056809442647392e-05, "loss": 1.5196, "step": 4462 }, { "epoch": 0.15, "learning_rate": 1.905636093763031e-05, "loss": 1.4504, "step": 4463 }, { "epoch": 0.15, "learning_rate": 1.9055912331282637e-05, "loss": 1.4685, "step": 4464 }, { "epoch": 0.15, "learning_rate": 1.9055463623609393e-05, "loss": 1.4973, "step": 4465 }, { "epoch": 0.15, "learning_rate": 1.90550148146156e-05, "loss": 1.508, "step": 4466 }, { "epoch": 0.15, "learning_rate": 1.9054565904306283e-05, "loss": 1.4718, "step": 4467 }, { "epoch": 0.15, "learning_rate": 1.9054116892686453e-05, "loss": 1.5362, "step": 4468 }, { "epoch": 0.15, "learning_rate": 1.9053667779761147e-05, "loss": 1.4707, "step": 4469 }, { "epoch": 0.15, "learning_rate": 1.9053218565535383e-05, "loss": 1.4645, "step": 4470 }, { "epoch": 0.15, "learning_rate": 1.9052769250014192e-05, "loss": 1.5257, "step": 4471 }, { "epoch": 0.15, "learning_rate": 1.9052319833202596e-05, "loss": 1.4574, "step": 4472 }, { "epoch": 0.15, "learning_rate": 1.9051870315105626e-05, "loss": 1.4505, "step": 4473 }, { "epoch": 0.15, "learning_rate": 1.9051420695728312e-05, "loss": 1.522, "step": 4474 }, { "epoch": 0.15, "learning_rate": 1.9050970975075685e-05, "loss": 1.5371, "step": 4475 }, { "epoch": 0.15, "learning_rate": 1.9050521153152773e-05, "loss": 1.455, "step": 4476 }, { "epoch": 0.15, "learning_rate": 1.9050071229964617e-05, "loss": 1.4459, "step": 4477 }, { "epoch": 0.15, "learning_rate": 1.9049621205516243e-05, "loss": 1.5765, "step": 4478 }, { "epoch": 0.15, "learning_rate": 1.904917107981269e-05, "loss": 1.4931, "step": 4479 }, { "epoch": 0.15, "learning_rate": 1.9048720852858997e-05, "loss": 1.406, "step": 4480 }, { "epoch": 0.15, "learning_rate": 1.9048270524660197e-05, "loss": 1.4789, "step": 4481 }, { "epoch": 0.15, "learning_rate": 1.904782009522133e-05, "loss": 1.5672, "step": 4482 }, { "epoch": 0.15, "learning_rate": 1.9047369564547434e-05, "loss": 1.5425, "step": 4483 }, { "epoch": 0.15, "learning_rate": 1.9046918932643555e-05, "loss": 1.5804, "step": 4484 }, { "epoch": 0.15, "learning_rate": 1.9046468199514732e-05, "loss": 1.5188, "step": 4485 }, { "epoch": 0.15, "learning_rate": 1.9046017365166004e-05, "loss": 1.4279, "step": 4486 }, { "epoch": 0.15, "learning_rate": 1.9045566429602425e-05, "loss": 1.3948, "step": 4487 }, { "epoch": 0.15, "learning_rate": 1.9045115392829032e-05, "loss": 1.3983, "step": 4488 }, { "epoch": 0.15, "learning_rate": 1.9044664254850874e-05, "loss": 1.3993, "step": 4489 }, { "epoch": 0.15, "learning_rate": 1.9044213015672998e-05, "loss": 1.4837, "step": 4490 }, { "epoch": 0.15, "learning_rate": 1.904376167530046e-05, "loss": 1.5473, "step": 4491 }, { "epoch": 0.15, "learning_rate": 1.90433102337383e-05, "loss": 1.5978, "step": 4492 }, { "epoch": 0.15, "learning_rate": 1.9042858690991574e-05, "loss": 1.5075, "step": 4493 }, { "epoch": 0.15, "learning_rate": 1.9042407047065333e-05, "loss": 1.5951, "step": 4494 }, { "epoch": 0.15, "learning_rate": 1.9041955301964632e-05, "loss": 1.6801, "step": 4495 }, { "epoch": 0.15, "learning_rate": 1.9041503455694522e-05, "loss": 1.6491, "step": 4496 }, { "epoch": 0.15, "learning_rate": 1.9041051508260066e-05, "loss": 1.6013, "step": 4497 }, { "epoch": 0.15, "learning_rate": 1.9040599459666314e-05, "loss": 1.6992, "step": 4498 }, { "epoch": 0.15, "learning_rate": 1.9040147309918326e-05, "loss": 1.6733, "step": 4499 }, { "epoch": 0.15, "learning_rate": 1.903969505902116e-05, "loss": 1.7617, "step": 4500 }, { "epoch": 0.15, "learning_rate": 1.903924270697988e-05, "loss": 1.7275, "step": 4501 }, { "epoch": 0.15, "learning_rate": 1.903879025379954e-05, "loss": 1.7432, "step": 4502 }, { "epoch": 0.15, "learning_rate": 1.9038337699485207e-05, "loss": 1.7456, "step": 4503 }, { "epoch": 0.15, "learning_rate": 1.9037885044041947e-05, "loss": 1.7459, "step": 4504 }, { "epoch": 0.15, "learning_rate": 1.903743228747482e-05, "loss": 1.8348, "step": 4505 }, { "epoch": 0.15, "learning_rate": 1.9036979429788895e-05, "loss": 1.7908, "step": 4506 }, { "epoch": 0.15, "learning_rate": 1.903652647098924e-05, "loss": 1.6974, "step": 4507 }, { "epoch": 0.15, "learning_rate": 1.9036073411080917e-05, "loss": 1.7168, "step": 4508 }, { "epoch": 0.15, "learning_rate": 1.9035620250069e-05, "loss": 1.6183, "step": 4509 }, { "epoch": 0.15, "learning_rate": 1.9035166987958563e-05, "loss": 1.6162, "step": 4510 }, { "epoch": 0.15, "learning_rate": 1.903471362475467e-05, "loss": 1.704, "step": 4511 }, { "epoch": 0.15, "learning_rate": 1.9034260160462398e-05, "loss": 1.6206, "step": 4512 }, { "epoch": 0.15, "learning_rate": 1.903380659508682e-05, "loss": 1.4306, "step": 4513 }, { "epoch": 0.15, "learning_rate": 1.903335292863301e-05, "loss": 1.6001, "step": 4514 }, { "epoch": 0.15, "learning_rate": 1.9032899161106045e-05, "loss": 1.6544, "step": 4515 }, { "epoch": 0.15, "learning_rate": 1.9032445292510998e-05, "loss": 1.5303, "step": 4516 }, { "epoch": 0.15, "learning_rate": 1.9031991322852956e-05, "loss": 1.5361, "step": 4517 }, { "epoch": 0.15, "learning_rate": 1.9031537252136992e-05, "loss": 1.549, "step": 4518 }, { "epoch": 0.15, "learning_rate": 1.9031083080368185e-05, "loss": 1.579, "step": 4519 }, { "epoch": 0.15, "learning_rate": 1.903062880755162e-05, "loss": 1.5746, "step": 4520 }, { "epoch": 0.15, "learning_rate": 1.903017443369238e-05, "loss": 1.5572, "step": 4521 }, { "epoch": 0.15, "learning_rate": 1.902971995879555e-05, "loss": 1.5292, "step": 4522 }, { "epoch": 0.15, "learning_rate": 1.9029265382866216e-05, "loss": 1.5572, "step": 4523 }, { "epoch": 0.15, "learning_rate": 1.9028810705909456e-05, "loss": 1.605, "step": 4524 }, { "epoch": 0.15, "learning_rate": 1.9028355927930363e-05, "loss": 1.5828, "step": 4525 }, { "epoch": 0.15, "learning_rate": 1.902790104893403e-05, "loss": 1.4733, "step": 4526 }, { "epoch": 0.15, "learning_rate": 1.902744606892554e-05, "loss": 1.4452, "step": 4527 }, { "epoch": 0.15, "learning_rate": 1.9026990987909985e-05, "loss": 1.4826, "step": 4528 }, { "epoch": 0.15, "learning_rate": 1.9026535805892456e-05, "loss": 1.5692, "step": 4529 }, { "epoch": 0.15, "learning_rate": 1.902608052287805e-05, "loss": 1.4404, "step": 4530 }, { "epoch": 0.15, "learning_rate": 1.9025625138871856e-05, "loss": 1.4718, "step": 4531 }, { "epoch": 0.15, "learning_rate": 1.9025169653878973e-05, "loss": 1.4161, "step": 4532 }, { "epoch": 0.15, "learning_rate": 1.9024714067904496e-05, "loss": 1.4908, "step": 4533 }, { "epoch": 0.15, "learning_rate": 1.9024258380953524e-05, "loss": 1.5291, "step": 4534 }, { "epoch": 0.15, "learning_rate": 1.9023802593031156e-05, "loss": 1.528, "step": 4535 }, { "epoch": 0.15, "learning_rate": 1.9023346704142488e-05, "loss": 1.5465, "step": 4536 }, { "epoch": 0.15, "learning_rate": 1.9022890714292625e-05, "loss": 1.5191, "step": 4537 }, { "epoch": 0.15, "learning_rate": 1.902243462348666e-05, "loss": 1.4748, "step": 4538 }, { "epoch": 0.15, "learning_rate": 1.902197843172971e-05, "loss": 1.6293, "step": 4539 }, { "epoch": 0.15, "learning_rate": 1.9021522139026873e-05, "loss": 1.5488, "step": 4540 }, { "epoch": 0.15, "learning_rate": 1.9021065745383248e-05, "loss": 1.4722, "step": 4541 }, { "epoch": 0.15, "learning_rate": 1.9020609250803952e-05, "loss": 1.4728, "step": 4542 }, { "epoch": 0.15, "learning_rate": 1.9020152655294087e-05, "loss": 1.5047, "step": 4543 }, { "epoch": 0.15, "learning_rate": 1.9019695958858762e-05, "loss": 1.4092, "step": 4544 }, { "epoch": 0.15, "learning_rate": 1.901923916150309e-05, "loss": 1.4167, "step": 4545 }, { "epoch": 0.15, "learning_rate": 1.9018782263232177e-05, "loss": 1.5924, "step": 4546 }, { "epoch": 0.15, "learning_rate": 1.901832526405114e-05, "loss": 1.5561, "step": 4547 }, { "epoch": 0.15, "learning_rate": 1.901786816396509e-05, "loss": 1.4662, "step": 4548 }, { "epoch": 0.15, "learning_rate": 1.9017410962979143e-05, "loss": 1.619, "step": 4549 }, { "epoch": 0.15, "learning_rate": 1.901695366109841e-05, "loss": 1.5584, "step": 4550 }, { "epoch": 0.15, "learning_rate": 1.9016496258328012e-05, "loss": 1.7069, "step": 4551 }, { "epoch": 0.15, "learning_rate": 1.9016038754673064e-05, "loss": 1.5675, "step": 4552 }, { "epoch": 0.15, "learning_rate": 1.9015581150138693e-05, "loss": 1.5687, "step": 4553 }, { "epoch": 0.15, "learning_rate": 1.9015123444730007e-05, "loss": 1.5801, "step": 4554 }, { "epoch": 0.15, "learning_rate": 1.9014665638452138e-05, "loss": 1.4992, "step": 4555 }, { "epoch": 0.15, "learning_rate": 1.9014207731310202e-05, "loss": 1.5914, "step": 4556 }, { "epoch": 0.15, "learning_rate": 1.901374972330932e-05, "loss": 1.5604, "step": 4557 }, { "epoch": 0.15, "learning_rate": 1.9013291614454622e-05, "loss": 1.5266, "step": 4558 }, { "epoch": 0.15, "learning_rate": 1.9012833404751237e-05, "loss": 1.4887, "step": 4559 }, { "epoch": 0.15, "learning_rate": 1.9012375094204283e-05, "loss": 1.5801, "step": 4560 }, { "epoch": 0.15, "learning_rate": 1.901191668281889e-05, "loss": 1.4922, "step": 4561 }, { "epoch": 0.15, "learning_rate": 1.9011458170600195e-05, "loss": 1.5289, "step": 4562 }, { "epoch": 0.15, "learning_rate": 1.9010999557553317e-05, "loss": 1.6079, "step": 4563 }, { "epoch": 0.15, "learning_rate": 1.9010540843683396e-05, "loss": 1.5879, "step": 4564 }, { "epoch": 0.15, "learning_rate": 1.901008202899556e-05, "loss": 1.52, "step": 4565 }, { "epoch": 0.15, "learning_rate": 1.9009623113494945e-05, "loss": 1.6146, "step": 4566 }, { "epoch": 0.15, "learning_rate": 1.9009164097186686e-05, "loss": 1.606, "step": 4567 }, { "epoch": 0.15, "learning_rate": 1.9008704980075915e-05, "loss": 1.5022, "step": 4568 }, { "epoch": 0.15, "learning_rate": 1.9008245762167773e-05, "loss": 1.5405, "step": 4569 }, { "epoch": 0.15, "learning_rate": 1.9007786443467396e-05, "loss": 1.6011, "step": 4570 }, { "epoch": 0.15, "learning_rate": 1.9007327023979924e-05, "loss": 1.5723, "step": 4571 }, { "epoch": 0.15, "learning_rate": 1.90068675037105e-05, "loss": 1.4267, "step": 4572 }, { "epoch": 0.15, "learning_rate": 1.9006407882664256e-05, "loss": 1.5293, "step": 4573 }, { "epoch": 0.15, "learning_rate": 1.9005948160846347e-05, "loss": 1.4981, "step": 4574 }, { "epoch": 0.15, "learning_rate": 1.900548833826191e-05, "loss": 1.4689, "step": 4575 }, { "epoch": 0.15, "learning_rate": 1.9005028414916092e-05, "loss": 1.5742, "step": 4576 }, { "epoch": 0.15, "learning_rate": 1.900456839081404e-05, "loss": 1.4968, "step": 4577 }, { "epoch": 0.15, "learning_rate": 1.9004108265960893e-05, "loss": 1.3279, "step": 4578 }, { "epoch": 0.15, "learning_rate": 1.900364804036181e-05, "loss": 1.4374, "step": 4579 }, { "epoch": 0.15, "learning_rate": 1.9003187714021936e-05, "loss": 1.389, "step": 4580 }, { "epoch": 0.15, "learning_rate": 1.900272728694642e-05, "loss": 1.4879, "step": 4581 }, { "epoch": 0.15, "learning_rate": 1.900226675914042e-05, "loss": 1.4489, "step": 4582 }, { "epoch": 0.15, "learning_rate": 1.900180613060908e-05, "loss": 1.5185, "step": 4583 }, { "epoch": 0.15, "learning_rate": 1.9001345401357557e-05, "loss": 1.5312, "step": 4584 }, { "epoch": 0.15, "learning_rate": 1.900088457139101e-05, "loss": 1.5403, "step": 4585 }, { "epoch": 0.15, "learning_rate": 1.9000423640714595e-05, "loss": 1.6327, "step": 4586 }, { "epoch": 0.15, "learning_rate": 1.899996260933346e-05, "loss": 1.4884, "step": 4587 }, { "epoch": 0.15, "learning_rate": 1.899950147725277e-05, "loss": 1.4761, "step": 4588 }, { "epoch": 0.15, "learning_rate": 1.899904024447769e-05, "loss": 1.4995, "step": 4589 }, { "epoch": 0.15, "learning_rate": 1.8998578911013372e-05, "loss": 1.6233, "step": 4590 }, { "epoch": 0.15, "learning_rate": 1.8998117476864984e-05, "loss": 1.4559, "step": 4591 }, { "epoch": 0.15, "learning_rate": 1.899765594203768e-05, "loss": 1.6164, "step": 4592 }, { "epoch": 0.15, "learning_rate": 1.8997194306536635e-05, "loss": 1.5892, "step": 4593 }, { "epoch": 0.15, "learning_rate": 1.8996732570367007e-05, "loss": 1.6473, "step": 4594 }, { "epoch": 0.15, "learning_rate": 1.8996270733533965e-05, "loss": 1.4766, "step": 4595 }, { "epoch": 0.15, "learning_rate": 1.8995808796042676e-05, "loss": 1.4834, "step": 4596 }, { "epoch": 0.15, "learning_rate": 1.8995346757898308e-05, "loss": 1.5929, "step": 4597 }, { "epoch": 0.15, "learning_rate": 1.8994884619106034e-05, "loss": 1.5317, "step": 4598 }, { "epoch": 0.15, "learning_rate": 1.899442237967102e-05, "loss": 1.5155, "step": 4599 }, { "epoch": 0.15, "learning_rate": 1.899396003959844e-05, "loss": 1.4364, "step": 4600 }, { "epoch": 0.15, "learning_rate": 1.8993497598893465e-05, "loss": 1.4556, "step": 4601 }, { "epoch": 0.15, "learning_rate": 1.8993035057561274e-05, "loss": 1.4243, "step": 4602 }, { "epoch": 0.15, "learning_rate": 1.899257241560704e-05, "loss": 1.4248, "step": 4603 }, { "epoch": 0.15, "learning_rate": 1.8992109673035936e-05, "loss": 1.4361, "step": 4604 }, { "epoch": 0.15, "learning_rate": 1.8991646829853146e-05, "loss": 1.5403, "step": 4605 }, { "epoch": 0.15, "learning_rate": 1.899118388606384e-05, "loss": 1.4976, "step": 4606 }, { "epoch": 0.15, "learning_rate": 1.8990720841673206e-05, "loss": 1.5745, "step": 4607 }, { "epoch": 0.15, "learning_rate": 1.8990257696686424e-05, "loss": 1.5134, "step": 4608 }, { "epoch": 0.15, "learning_rate": 1.8989794451108673e-05, "loss": 1.5456, "step": 4609 }, { "epoch": 0.15, "learning_rate": 1.8989331104945135e-05, "loss": 1.5127, "step": 4610 }, { "epoch": 0.15, "learning_rate": 1.8988867658201002e-05, "loss": 1.4788, "step": 4611 }, { "epoch": 0.15, "learning_rate": 1.898840411088145e-05, "loss": 1.548, "step": 4612 }, { "epoch": 0.15, "learning_rate": 1.8987940462991673e-05, "loss": 1.5836, "step": 4613 }, { "epoch": 0.15, "learning_rate": 1.898747671453685e-05, "loss": 1.5066, "step": 4614 }, { "epoch": 0.15, "learning_rate": 1.898701286552218e-05, "loss": 1.5678, "step": 4615 }, { "epoch": 0.15, "learning_rate": 1.8986548915952843e-05, "loss": 1.6283, "step": 4616 }, { "epoch": 0.15, "learning_rate": 1.8986084865834044e-05, "loss": 1.5443, "step": 4617 }, { "epoch": 0.15, "learning_rate": 1.898562071517096e-05, "loss": 1.5149, "step": 4618 }, { "epoch": 0.15, "learning_rate": 1.898515646396879e-05, "loss": 1.594, "step": 4619 }, { "epoch": 0.15, "learning_rate": 1.8984692112232732e-05, "loss": 1.6684, "step": 4620 }, { "epoch": 0.15, "learning_rate": 1.8984227659967978e-05, "loss": 1.5812, "step": 4621 }, { "epoch": 0.15, "learning_rate": 1.8983763107179726e-05, "loss": 1.5807, "step": 4622 }, { "epoch": 0.15, "learning_rate": 1.8983298453873172e-05, "loss": 1.5131, "step": 4623 }, { "epoch": 0.15, "learning_rate": 1.8982833700053518e-05, "loss": 1.5484, "step": 4624 }, { "epoch": 0.15, "learning_rate": 1.898236884572596e-05, "loss": 1.5871, "step": 4625 }, { "epoch": 0.15, "learning_rate": 1.8981903890895705e-05, "loss": 1.6645, "step": 4626 }, { "epoch": 0.15, "learning_rate": 1.8981438835567948e-05, "loss": 1.6682, "step": 4627 }, { "epoch": 0.15, "learning_rate": 1.8980973679747897e-05, "loss": 1.5748, "step": 4628 }, { "epoch": 0.15, "learning_rate": 1.898050842344076e-05, "loss": 1.5415, "step": 4629 }, { "epoch": 0.15, "learning_rate": 1.8980043066651737e-05, "loss": 1.4868, "step": 4630 }, { "epoch": 0.15, "learning_rate": 1.8979577609386033e-05, "loss": 1.5787, "step": 4631 }, { "epoch": 0.15, "learning_rate": 1.8979112051648863e-05, "loss": 1.5486, "step": 4632 }, { "epoch": 0.15, "learning_rate": 1.897864639344543e-05, "loss": 1.6234, "step": 4633 }, { "epoch": 0.15, "learning_rate": 1.897818063478095e-05, "loss": 1.5715, "step": 4634 }, { "epoch": 0.15, "learning_rate": 1.897771477566063e-05, "loss": 1.5943, "step": 4635 }, { "epoch": 0.15, "learning_rate": 1.8977248816089678e-05, "loss": 1.5368, "step": 4636 }, { "epoch": 0.15, "learning_rate": 1.8976782756073323e-05, "loss": 1.4192, "step": 4637 }, { "epoch": 0.15, "learning_rate": 1.897631659561676e-05, "loss": 1.4066, "step": 4638 }, { "epoch": 0.15, "learning_rate": 1.897585033472522e-05, "loss": 1.475, "step": 4639 }, { "epoch": 0.15, "learning_rate": 1.8975383973403915e-05, "loss": 1.3361, "step": 4640 }, { "epoch": 0.15, "learning_rate": 1.8974917511658058e-05, "loss": 1.5032, "step": 4641 }, { "epoch": 0.15, "learning_rate": 1.8974450949492874e-05, "loss": 1.5085, "step": 4642 }, { "epoch": 0.15, "learning_rate": 1.8973984286913584e-05, "loss": 1.4628, "step": 4643 }, { "epoch": 0.15, "learning_rate": 1.897351752392541e-05, "loss": 1.4738, "step": 4644 }, { "epoch": 0.15, "learning_rate": 1.8973050660533568e-05, "loss": 1.4319, "step": 4645 }, { "epoch": 0.15, "learning_rate": 1.8972583696743284e-05, "loss": 1.4114, "step": 4646 }, { "epoch": 0.15, "learning_rate": 1.8972116632559788e-05, "loss": 1.4063, "step": 4647 }, { "epoch": 0.15, "learning_rate": 1.8971649467988303e-05, "loss": 1.3611, "step": 4648 }, { "epoch": 0.15, "learning_rate": 1.8971182203034055e-05, "loss": 1.4336, "step": 4649 }, { "epoch": 0.15, "learning_rate": 1.8970714837702272e-05, "loss": 1.4465, "step": 4650 }, { "epoch": 0.16, "learning_rate": 1.8970247371998184e-05, "loss": 1.4564, "step": 4651 }, { "epoch": 0.16, "learning_rate": 1.896977980592702e-05, "loss": 1.4378, "step": 4652 }, { "epoch": 0.16, "learning_rate": 1.8969312139494015e-05, "loss": 1.4728, "step": 4653 }, { "epoch": 0.16, "learning_rate": 1.8968844372704394e-05, "loss": 1.5517, "step": 4654 }, { "epoch": 0.16, "learning_rate": 1.8968376505563402e-05, "loss": 1.5005, "step": 4655 }, { "epoch": 0.16, "learning_rate": 1.8967908538076267e-05, "loss": 1.537, "step": 4656 }, { "epoch": 0.16, "learning_rate": 1.8967440470248227e-05, "loss": 1.5485, "step": 4657 }, { "epoch": 0.16, "learning_rate": 1.8966972302084516e-05, "loss": 1.5307, "step": 4658 }, { "epoch": 0.16, "learning_rate": 1.8966504033590382e-05, "loss": 1.4275, "step": 4659 }, { "epoch": 0.16, "learning_rate": 1.8966035664771047e-05, "loss": 1.534, "step": 4660 }, { "epoch": 0.16, "learning_rate": 1.8965567195631766e-05, "loss": 1.6143, "step": 4661 }, { "epoch": 0.16, "learning_rate": 1.8965098626177777e-05, "loss": 1.5766, "step": 4662 }, { "epoch": 0.16, "learning_rate": 1.896462995641432e-05, "loss": 1.4275, "step": 4663 }, { "epoch": 0.16, "learning_rate": 1.896416118634665e-05, "loss": 1.6074, "step": 4664 }, { "epoch": 0.16, "learning_rate": 1.896369231597999e-05, "loss": 1.4731, "step": 4665 }, { "epoch": 0.16, "learning_rate": 1.896322334531961e-05, "loss": 1.518, "step": 4666 }, { "epoch": 0.16, "learning_rate": 1.896275427437074e-05, "loss": 1.488, "step": 4667 }, { "epoch": 0.16, "learning_rate": 1.8962285103138637e-05, "loss": 1.4711, "step": 4668 }, { "epoch": 0.16, "learning_rate": 1.8961815831628545e-05, "loss": 1.6113, "step": 4669 }, { "epoch": 0.16, "learning_rate": 1.8961346459845724e-05, "loss": 1.6288, "step": 4670 }, { "epoch": 0.16, "learning_rate": 1.8960876987795414e-05, "loss": 1.4819, "step": 4671 }, { "epoch": 0.16, "learning_rate": 1.8960407415482876e-05, "loss": 1.4939, "step": 4672 }, { "epoch": 0.16, "learning_rate": 1.895993774291336e-05, "loss": 1.5457, "step": 4673 }, { "epoch": 0.16, "learning_rate": 1.8959467970092123e-05, "loss": 1.5735, "step": 4674 }, { "epoch": 0.16, "learning_rate": 1.8959467970092123e-05, "loss": 1.7228, "step": 4675 }, { "epoch": 0.16, "learning_rate": 1.8958998097024423e-05, "loss": 1.5776, "step": 4676 }, { "epoch": 0.16, "learning_rate": 1.8958528123715513e-05, "loss": 1.7048, "step": 4677 }, { "epoch": 0.16, "learning_rate": 1.8958058050170655e-05, "loss": 1.6684, "step": 4678 }, { "epoch": 0.16, "learning_rate": 1.8957587876395108e-05, "loss": 1.5535, "step": 4679 }, { "epoch": 0.16, "learning_rate": 1.895711760239413e-05, "loss": 1.628, "step": 4680 }, { "epoch": 0.16, "learning_rate": 1.8956647228172986e-05, "loss": 1.4386, "step": 4681 }, { "epoch": 0.16, "learning_rate": 1.8956176753736938e-05, "loss": 1.6161, "step": 4682 }, { "epoch": 0.16, "learning_rate": 1.8955706179091248e-05, "loss": 1.4656, "step": 4683 }, { "epoch": 0.16, "learning_rate": 1.8955235504241187e-05, "loss": 1.6211, "step": 4684 }, { "epoch": 0.16, "learning_rate": 1.895476472919201e-05, "loss": 1.6371, "step": 4685 }, { "epoch": 0.16, "learning_rate": 1.8954293853949002e-05, "loss": 1.6277, "step": 4686 }, { "epoch": 0.16, "learning_rate": 1.8953822878517414e-05, "loss": 1.5298, "step": 4687 }, { "epoch": 0.16, "learning_rate": 1.8953351802902527e-05, "loss": 1.581, "step": 4688 }, { "epoch": 0.16, "learning_rate": 1.8952880627109606e-05, "loss": 1.5903, "step": 4689 }, { "epoch": 0.16, "learning_rate": 1.8952409351143925e-05, "loss": 1.5533, "step": 4690 }, { "epoch": 0.16, "learning_rate": 1.895193797501076e-05, "loss": 1.6079, "step": 4691 }, { "epoch": 0.16, "learning_rate": 1.8951466498715378e-05, "loss": 1.5325, "step": 4692 }, { "epoch": 0.16, "learning_rate": 1.895099492226306e-05, "loss": 1.5602, "step": 4693 }, { "epoch": 0.16, "learning_rate": 1.895052324565908e-05, "loss": 1.5212, "step": 4694 }, { "epoch": 0.16, "learning_rate": 1.8950051468908718e-05, "loss": 1.5742, "step": 4695 }, { "epoch": 0.16, "learning_rate": 1.894957959201725e-05, "loss": 1.5412, "step": 4696 }, { "epoch": 0.16, "learning_rate": 1.894910761498996e-05, "loss": 1.514, "step": 4697 }, { "epoch": 0.16, "learning_rate": 1.894863553783212e-05, "loss": 1.5811, "step": 4698 }, { "epoch": 0.16, "learning_rate": 1.894816336054902e-05, "loss": 1.642, "step": 4699 }, { "epoch": 0.16, "learning_rate": 1.8947691083145944e-05, "loss": 1.5444, "step": 4700 }, { "epoch": 0.16, "learning_rate": 1.894721870562817e-05, "loss": 1.5591, "step": 4701 }, { "epoch": 0.16, "learning_rate": 1.8946746228000987e-05, "loss": 1.7128, "step": 4702 }, { "epoch": 0.16, "learning_rate": 1.894627365026968e-05, "loss": 1.627, "step": 4703 }, { "epoch": 0.16, "learning_rate": 1.894580097243954e-05, "loss": 1.6222, "step": 4704 }, { "epoch": 0.16, "learning_rate": 1.8945328194515847e-05, "loss": 1.5269, "step": 4705 }, { "epoch": 0.16, "learning_rate": 1.8944855316503906e-05, "loss": 1.5948, "step": 4706 }, { "epoch": 0.16, "learning_rate": 1.8944382338408994e-05, "loss": 1.5109, "step": 4707 }, { "epoch": 0.16, "learning_rate": 1.894390926023641e-05, "loss": 1.5408, "step": 4708 }, { "epoch": 0.16, "learning_rate": 1.8943436081991445e-05, "loss": 1.5896, "step": 4709 }, { "epoch": 0.16, "learning_rate": 1.8942962803679393e-05, "loss": 1.5656, "step": 4710 }, { "epoch": 0.16, "learning_rate": 1.8942489425305553e-05, "loss": 1.6367, "step": 4711 }, { "epoch": 0.16, "learning_rate": 1.8942015946875215e-05, "loss": 1.6639, "step": 4712 }, { "epoch": 0.16, "learning_rate": 1.8941542368393683e-05, "loss": 1.5232, "step": 4713 }, { "epoch": 0.16, "learning_rate": 1.8941068689866255e-05, "loss": 1.596, "step": 4714 }, { "epoch": 0.16, "learning_rate": 1.8940594911298228e-05, "loss": 1.5079, "step": 4715 }, { "epoch": 0.16, "learning_rate": 1.8940121032694902e-05, "loss": 1.4263, "step": 4716 }, { "epoch": 0.16, "learning_rate": 1.8939647054061582e-05, "loss": 1.4947, "step": 4717 }, { "epoch": 0.16, "learning_rate": 1.8939172975403572e-05, "loss": 1.4275, "step": 4718 }, { "epoch": 0.16, "learning_rate": 1.8938698796726177e-05, "loss": 1.4983, "step": 4719 }, { "epoch": 0.16, "learning_rate": 1.8938224518034698e-05, "loss": 1.4978, "step": 4720 }, { "epoch": 0.16, "learning_rate": 1.8937750139334447e-05, "loss": 1.5104, "step": 4721 }, { "epoch": 0.16, "learning_rate": 1.8937275660630727e-05, "loss": 1.5231, "step": 4722 }, { "epoch": 0.16, "learning_rate": 1.893680108192885e-05, "loss": 1.4149, "step": 4723 }, { "epoch": 0.16, "learning_rate": 1.8936326403234125e-05, "loss": 1.5125, "step": 4724 }, { "epoch": 0.16, "learning_rate": 1.893585162455186e-05, "loss": 1.4562, "step": 4725 }, { "epoch": 0.16, "learning_rate": 1.8935376745887377e-05, "loss": 1.4682, "step": 4726 }, { "epoch": 0.16, "learning_rate": 1.8934901767245977e-05, "loss": 1.4992, "step": 4727 }, { "epoch": 0.16, "learning_rate": 1.8934426688632986e-05, "loss": 1.4641, "step": 4728 }, { "epoch": 0.16, "learning_rate": 1.893395151005371e-05, "loss": 1.5136, "step": 4729 }, { "epoch": 0.16, "learning_rate": 1.893347623151347e-05, "loss": 1.6094, "step": 4730 }, { "epoch": 0.16, "learning_rate": 1.8933000853017584e-05, "loss": 1.5296, "step": 4731 }, { "epoch": 0.16, "learning_rate": 1.893252537457137e-05, "loss": 1.4123, "step": 4732 }, { "epoch": 0.16, "learning_rate": 1.8932049796180147e-05, "loss": 1.4903, "step": 4733 }, { "epoch": 0.16, "learning_rate": 1.893157411784924e-05, "loss": 1.4543, "step": 4734 }, { "epoch": 0.16, "learning_rate": 1.893109833958397e-05, "loss": 1.4143, "step": 4735 }, { "epoch": 0.16, "learning_rate": 1.8930622461389656e-05, "loss": 1.5249, "step": 4736 }, { "epoch": 0.16, "learning_rate": 1.893014648327163e-05, "loss": 1.5842, "step": 4737 }, { "epoch": 0.16, "learning_rate": 1.8929670405235206e-05, "loss": 1.4784, "step": 4738 }, { "epoch": 0.16, "learning_rate": 1.8929194227285728e-05, "loss": 1.5366, "step": 4739 }, { "epoch": 0.16, "learning_rate": 1.8928717949428508e-05, "loss": 1.6452, "step": 4740 }, { "epoch": 0.16, "learning_rate": 1.8928241571668887e-05, "loss": 1.5297, "step": 4741 }, { "epoch": 0.16, "learning_rate": 1.8927765094012183e-05, "loss": 1.5399, "step": 4742 }, { "epoch": 0.16, "learning_rate": 1.8927288516463738e-05, "loss": 1.5055, "step": 4743 }, { "epoch": 0.16, "learning_rate": 1.892681183902888e-05, "loss": 1.5693, "step": 4744 }, { "epoch": 0.16, "learning_rate": 1.8926335061712938e-05, "loss": 1.5339, "step": 4745 }, { "epoch": 0.16, "learning_rate": 1.892585818452126e-05, "loss": 1.654, "step": 4746 }, { "epoch": 0.16, "learning_rate": 1.8925381207459163e-05, "loss": 1.5639, "step": 4747 }, { "epoch": 0.16, "learning_rate": 1.8924904130532e-05, "loss": 1.5015, "step": 4748 }, { "epoch": 0.16, "learning_rate": 1.89244269537451e-05, "loss": 1.5829, "step": 4749 }, { "epoch": 0.16, "learning_rate": 1.8923949677103807e-05, "loss": 1.4848, "step": 4750 }, { "epoch": 0.16, "learning_rate": 1.8923472300613456e-05, "loss": 1.6106, "step": 4751 }, { "epoch": 0.16, "learning_rate": 1.8922994824279394e-05, "loss": 1.6993, "step": 4752 }, { "epoch": 0.16, "learning_rate": 1.892251724810696e-05, "loss": 1.5781, "step": 4753 }, { "epoch": 0.16, "learning_rate": 1.89220395721015e-05, "loss": 1.4895, "step": 4754 }, { "epoch": 0.16, "learning_rate": 1.8921561796268354e-05, "loss": 1.5081, "step": 4755 }, { "epoch": 0.16, "learning_rate": 1.892108392061287e-05, "loss": 1.4611, "step": 4756 }, { "epoch": 0.16, "learning_rate": 1.8920605945140396e-05, "loss": 1.514, "step": 4757 }, { "epoch": 0.16, "learning_rate": 1.8920127869856283e-05, "loss": 1.4748, "step": 4758 }, { "epoch": 0.16, "learning_rate": 1.8919649694765875e-05, "loss": 1.439, "step": 4759 }, { "epoch": 0.16, "learning_rate": 1.8919171419874524e-05, "loss": 1.5611, "step": 4760 }, { "epoch": 0.16, "learning_rate": 1.891869304518758e-05, "loss": 1.4342, "step": 4761 }, { "epoch": 0.16, "learning_rate": 1.89182145707104e-05, "loss": 1.4207, "step": 4762 }, { "epoch": 0.16, "learning_rate": 1.8917735996448332e-05, "loss": 1.5159, "step": 4763 }, { "epoch": 0.16, "learning_rate": 1.8917257322406735e-05, "loss": 1.5568, "step": 4764 }, { "epoch": 0.16, "learning_rate": 1.8916778548590963e-05, "loss": 1.6568, "step": 4765 }, { "epoch": 0.16, "learning_rate": 1.891629967500637e-05, "loss": 1.6399, "step": 4766 }, { "epoch": 0.16, "learning_rate": 1.891582070165832e-05, "loss": 1.5409, "step": 4767 }, { "epoch": 0.16, "learning_rate": 1.8915341628552166e-05, "loss": 1.5501, "step": 4768 }, { "epoch": 0.16, "learning_rate": 1.8914862455693273e-05, "loss": 1.5718, "step": 4769 }, { "epoch": 0.16, "learning_rate": 1.8914383183087e-05, "loss": 1.5418, "step": 4770 }, { "epoch": 0.16, "learning_rate": 1.891390381073871e-05, "loss": 1.5935, "step": 4771 }, { "epoch": 0.16, "learning_rate": 1.891342433865377e-05, "loss": 1.5185, "step": 4772 }, { "epoch": 0.16, "learning_rate": 1.8912944766837542e-05, "loss": 1.5133, "step": 4773 }, { "epoch": 0.16, "learning_rate": 1.891246509529539e-05, "loss": 1.6059, "step": 4774 }, { "epoch": 0.16, "learning_rate": 1.8911985324032685e-05, "loss": 1.6695, "step": 4775 }, { "epoch": 0.16, "learning_rate": 1.891150545305479e-05, "loss": 1.5463, "step": 4776 }, { "epoch": 0.16, "learning_rate": 1.8911025482367077e-05, "loss": 1.5209, "step": 4777 }, { "epoch": 0.16, "learning_rate": 1.8910545411974916e-05, "loss": 1.6086, "step": 4778 }, { "epoch": 0.16, "learning_rate": 1.891006524188368e-05, "loss": 1.501, "step": 4779 }, { "epoch": 0.16, "learning_rate": 1.890958497209874e-05, "loss": 1.5664, "step": 4780 }, { "epoch": 0.16, "learning_rate": 1.8909104602625472e-05, "loss": 1.5161, "step": 4781 }, { "epoch": 0.16, "learning_rate": 1.8908624133469248e-05, "loss": 1.5725, "step": 4782 }, { "epoch": 0.16, "learning_rate": 1.890814356463544e-05, "loss": 1.5361, "step": 4783 }, { "epoch": 0.16, "learning_rate": 1.8907662896129435e-05, "loss": 1.5982, "step": 4784 }, { "epoch": 0.16, "learning_rate": 1.8907182127956604e-05, "loss": 1.5161, "step": 4785 }, { "epoch": 0.16, "learning_rate": 1.890670126012233e-05, "loss": 1.484, "step": 4786 }, { "epoch": 0.16, "learning_rate": 1.890622029263199e-05, "loss": 1.4688, "step": 4787 }, { "epoch": 0.16, "learning_rate": 1.8905739225490966e-05, "loss": 1.5305, "step": 4788 }, { "epoch": 0.16, "learning_rate": 1.8905258058704644e-05, "loss": 1.4387, "step": 4789 }, { "epoch": 0.16, "learning_rate": 1.8904776792278403e-05, "loss": 1.4505, "step": 4790 }, { "epoch": 0.16, "learning_rate": 1.8904295426217632e-05, "loss": 1.5112, "step": 4791 }, { "epoch": 0.16, "learning_rate": 1.8903813960527714e-05, "loss": 1.4829, "step": 4792 }, { "epoch": 0.16, "learning_rate": 1.8903332395214037e-05, "loss": 1.5054, "step": 4793 }, { "epoch": 0.16, "learning_rate": 1.8902850730281993e-05, "loss": 1.5886, "step": 4794 }, { "epoch": 0.16, "learning_rate": 1.8902368965736967e-05, "loss": 1.5107, "step": 4795 }, { "epoch": 0.16, "learning_rate": 1.8901887101584347e-05, "loss": 1.5433, "step": 4796 }, { "epoch": 0.16, "learning_rate": 1.890140513782953e-05, "loss": 1.5358, "step": 4797 }, { "epoch": 0.16, "learning_rate": 1.8900923074477903e-05, "loss": 1.5196, "step": 4798 }, { "epoch": 0.16, "learning_rate": 1.8900440911534864e-05, "loss": 1.5361, "step": 4799 }, { "epoch": 0.16, "learning_rate": 1.889995864900581e-05, "loss": 1.5296, "step": 4800 }, { "epoch": 0.16, "learning_rate": 1.889947628689613e-05, "loss": 1.5046, "step": 4801 }, { "epoch": 0.16, "learning_rate": 1.8898993825211228e-05, "loss": 1.5875, "step": 4802 }, { "epoch": 0.16, "learning_rate": 1.88985112639565e-05, "loss": 1.5409, "step": 4803 }, { "epoch": 0.16, "learning_rate": 1.8898028603137343e-05, "loss": 1.4278, "step": 4804 }, { "epoch": 0.16, "learning_rate": 1.8897545842759154e-05, "loss": 1.5422, "step": 4805 }, { "epoch": 0.16, "learning_rate": 1.8897062982827347e-05, "loss": 1.4895, "step": 4806 }, { "epoch": 0.16, "learning_rate": 1.8896580023347315e-05, "loss": 1.4922, "step": 4807 }, { "epoch": 0.16, "learning_rate": 1.8896096964324462e-05, "loss": 1.5672, "step": 4808 }, { "epoch": 0.16, "learning_rate": 1.8895613805764196e-05, "loss": 1.4023, "step": 4809 }, { "epoch": 0.16, "learning_rate": 1.889513054767192e-05, "loss": 1.4686, "step": 4810 }, { "epoch": 0.16, "learning_rate": 1.8894647190053047e-05, "loss": 1.5453, "step": 4811 }, { "epoch": 0.16, "learning_rate": 1.889416373291298e-05, "loss": 1.5719, "step": 4812 }, { "epoch": 0.16, "learning_rate": 1.8893680176257125e-05, "loss": 1.6043, "step": 4813 }, { "epoch": 0.16, "learning_rate": 1.88931965200909e-05, "loss": 1.5017, "step": 4814 }, { "epoch": 0.16, "learning_rate": 1.8892712764419716e-05, "loss": 1.5119, "step": 4815 }, { "epoch": 0.16, "learning_rate": 1.889222890924898e-05, "loss": 1.5732, "step": 4816 }, { "epoch": 0.16, "learning_rate": 1.889174495458411e-05, "loss": 1.5222, "step": 4817 }, { "epoch": 0.16, "learning_rate": 1.8891260900430518e-05, "loss": 1.551, "step": 4818 }, { "epoch": 0.16, "learning_rate": 1.8890776746793628e-05, "loss": 1.5291, "step": 4819 }, { "epoch": 0.16, "learning_rate": 1.8890292493678844e-05, "loss": 1.4554, "step": 4820 }, { "epoch": 0.16, "learning_rate": 1.8889808141091598e-05, "loss": 1.5497, "step": 4821 }, { "epoch": 0.16, "learning_rate": 1.8889323689037296e-05, "loss": 1.5073, "step": 4822 }, { "epoch": 0.16, "learning_rate": 1.888883913752137e-05, "loss": 1.5926, "step": 4823 }, { "epoch": 0.16, "learning_rate": 1.8888354486549238e-05, "loss": 1.568, "step": 4824 }, { "epoch": 0.16, "learning_rate": 1.8887869736126317e-05, "loss": 1.6031, "step": 4825 }, { "epoch": 0.16, "learning_rate": 1.8887384886258038e-05, "loss": 1.5922, "step": 4826 }, { "epoch": 0.16, "learning_rate": 1.8886899936949822e-05, "loss": 1.4717, "step": 4827 }, { "epoch": 0.16, "learning_rate": 1.8886414888207098e-05, "loss": 1.5939, "step": 4828 }, { "epoch": 0.16, "learning_rate": 1.8885929740035292e-05, "loss": 1.4696, "step": 4829 }, { "epoch": 0.16, "learning_rate": 1.888544449243983e-05, "loss": 1.5801, "step": 4830 }, { "epoch": 0.16, "learning_rate": 1.888495914542614e-05, "loss": 1.5245, "step": 4831 }, { "epoch": 0.16, "learning_rate": 1.888447369899966e-05, "loss": 1.5678, "step": 4832 }, { "epoch": 0.16, "learning_rate": 1.8883988153165815e-05, "loss": 1.6724, "step": 4833 }, { "epoch": 0.16, "learning_rate": 1.8883502507930044e-05, "loss": 1.5226, "step": 4834 }, { "epoch": 0.16, "learning_rate": 1.8883016763297773e-05, "loss": 1.49, "step": 4835 }, { "epoch": 0.16, "learning_rate": 1.8882530919274442e-05, "loss": 1.503, "step": 4836 }, { "epoch": 0.16, "learning_rate": 1.8882044975865486e-05, "loss": 1.4969, "step": 4837 }, { "epoch": 0.16, "learning_rate": 1.888155893307634e-05, "loss": 1.456, "step": 4838 }, { "epoch": 0.16, "learning_rate": 1.8881072790912445e-05, "loss": 1.5435, "step": 4839 }, { "epoch": 0.16, "learning_rate": 1.888058654937924e-05, "loss": 1.4828, "step": 4840 }, { "epoch": 0.16, "learning_rate": 1.8880100208482167e-05, "loss": 1.4063, "step": 4841 }, { "epoch": 0.16, "learning_rate": 1.8879613768226662e-05, "loss": 1.4853, "step": 4842 }, { "epoch": 0.16, "learning_rate": 1.8879127228618174e-05, "loss": 1.5062, "step": 4843 }, { "epoch": 0.16, "learning_rate": 1.8878640589662144e-05, "loss": 1.3951, "step": 4844 }, { "epoch": 0.16, "learning_rate": 1.8878153851364013e-05, "loss": 1.5752, "step": 4845 }, { "epoch": 0.16, "learning_rate": 1.887766701372924e-05, "loss": 1.5731, "step": 4846 }, { "epoch": 0.16, "learning_rate": 1.8877180076763257e-05, "loss": 1.5388, "step": 4847 }, { "epoch": 0.16, "learning_rate": 1.8876693040471518e-05, "loss": 1.5769, "step": 4848 }, { "epoch": 0.16, "learning_rate": 1.8876205904859474e-05, "loss": 1.5612, "step": 4849 }, { "epoch": 0.16, "learning_rate": 1.8875718669932577e-05, "loss": 1.7673, "step": 4850 }, { "epoch": 0.16, "learning_rate": 1.8875231335696277e-05, "loss": 1.656, "step": 4851 }, { "epoch": 0.16, "learning_rate": 1.8874743902156024e-05, "loss": 1.6555, "step": 4852 }, { "epoch": 0.16, "learning_rate": 1.8874256369317274e-05, "loss": 1.8185, "step": 4853 }, { "epoch": 0.16, "learning_rate": 1.887376873718548e-05, "loss": 1.5587, "step": 4854 }, { "epoch": 0.16, "learning_rate": 1.88732810057661e-05, "loss": 1.66, "step": 4855 }, { "epoch": 0.16, "learning_rate": 1.8872793175064594e-05, "loss": 1.7152, "step": 4856 }, { "epoch": 0.16, "learning_rate": 1.8872305245086414e-05, "loss": 1.6262, "step": 4857 }, { "epoch": 0.16, "learning_rate": 1.8871817215837027e-05, "loss": 1.7725, "step": 4858 }, { "epoch": 0.16, "learning_rate": 1.8871329087321886e-05, "loss": 1.5728, "step": 4859 }, { "epoch": 0.16, "learning_rate": 1.8870840859546455e-05, "loss": 1.5509, "step": 4860 }, { "epoch": 0.16, "learning_rate": 1.8870352532516197e-05, "loss": 1.6391, "step": 4861 }, { "epoch": 0.16, "learning_rate": 1.8869864106236582e-05, "loss": 1.5379, "step": 4862 }, { "epoch": 0.16, "learning_rate": 1.8869375580713064e-05, "loss": 1.6048, "step": 4863 }, { "epoch": 0.16, "learning_rate": 1.8868886955951115e-05, "loss": 1.5112, "step": 4864 }, { "epoch": 0.16, "learning_rate": 1.8868398231956203e-05, "loss": 1.4784, "step": 4865 }, { "epoch": 0.16, "learning_rate": 1.8867909408733794e-05, "loss": 1.4971, "step": 4866 }, { "epoch": 0.16, "learning_rate": 1.886742048628936e-05, "loss": 1.407, "step": 4867 }, { "epoch": 0.16, "learning_rate": 1.8866931464628365e-05, "loss": 1.4678, "step": 4868 }, { "epoch": 0.16, "learning_rate": 1.8866442343756288e-05, "loss": 1.5905, "step": 4869 }, { "epoch": 0.16, "learning_rate": 1.88659531236786e-05, "loss": 1.5057, "step": 4870 }, { "epoch": 0.16, "learning_rate": 1.8865463804400776e-05, "loss": 1.5591, "step": 4871 }, { "epoch": 0.16, "learning_rate": 1.8864974385928284e-05, "loss": 1.4615, "step": 4872 }, { "epoch": 0.16, "learning_rate": 1.8864484868266605e-05, "loss": 1.4534, "step": 4873 }, { "epoch": 0.16, "learning_rate": 1.886399525142122e-05, "loss": 1.4762, "step": 4874 }, { "epoch": 0.16, "learning_rate": 1.8863505535397597e-05, "loss": 1.4552, "step": 4875 }, { "epoch": 0.16, "learning_rate": 1.8863015720201225e-05, "loss": 1.4742, "step": 4876 }, { "epoch": 0.16, "learning_rate": 1.8862525805837584e-05, "loss": 1.61, "step": 4877 }, { "epoch": 0.16, "learning_rate": 1.8862035792312148e-05, "loss": 1.5733, "step": 4878 }, { "epoch": 0.16, "learning_rate": 1.8861545679630406e-05, "loss": 1.626, "step": 4879 }, { "epoch": 0.16, "learning_rate": 1.8861055467797843e-05, "loss": 1.6017, "step": 4880 }, { "epoch": 0.16, "learning_rate": 1.8860565156819935e-05, "loss": 1.5562, "step": 4881 }, { "epoch": 0.16, "learning_rate": 1.886007474670218e-05, "loss": 1.5632, "step": 4882 }, { "epoch": 0.16, "learning_rate": 1.885958423745006e-05, "loss": 1.5752, "step": 4883 }, { "epoch": 0.16, "learning_rate": 1.8859093629069057e-05, "loss": 1.5788, "step": 4884 }, { "epoch": 0.16, "learning_rate": 1.885860292156467e-05, "loss": 1.6617, "step": 4885 }, { "epoch": 0.16, "learning_rate": 1.8858112114942386e-05, "loss": 1.6228, "step": 4886 }, { "epoch": 0.16, "learning_rate": 1.8857621209207693e-05, "loss": 1.593, "step": 4887 }, { "epoch": 0.16, "learning_rate": 1.8857130204366087e-05, "loss": 1.6095, "step": 4888 }, { "epoch": 0.16, "learning_rate": 1.8856639100423062e-05, "loss": 1.5061, "step": 4889 }, { "epoch": 0.16, "learning_rate": 1.8856147897384112e-05, "loss": 1.5893, "step": 4890 }, { "epoch": 0.16, "learning_rate": 1.8855656595254734e-05, "loss": 1.5678, "step": 4891 }, { "epoch": 0.16, "learning_rate": 1.8855165194040425e-05, "loss": 1.4321, "step": 4892 }, { "epoch": 0.16, "learning_rate": 1.8854673693746683e-05, "loss": 1.4995, "step": 4893 }, { "epoch": 0.16, "learning_rate": 1.8854182094379007e-05, "loss": 1.6516, "step": 4894 }, { "epoch": 0.16, "learning_rate": 1.8853690395942896e-05, "loss": 1.5679, "step": 4895 }, { "epoch": 0.16, "learning_rate": 1.885319859844385e-05, "loss": 1.4724, "step": 4896 }, { "epoch": 0.16, "learning_rate": 1.8852706701887383e-05, "loss": 1.6228, "step": 4897 }, { "epoch": 0.16, "learning_rate": 1.8852214706278986e-05, "loss": 1.5478, "step": 4898 }, { "epoch": 0.16, "learning_rate": 1.8851722611624166e-05, "loss": 1.6108, "step": 4899 }, { "epoch": 0.16, "learning_rate": 1.8851230417928433e-05, "loss": 1.582, "step": 4900 }, { "epoch": 0.16, "learning_rate": 1.8850738125197297e-05, "loss": 1.5653, "step": 4901 }, { "epoch": 0.16, "learning_rate": 1.8850245733436255e-05, "loss": 1.6709, "step": 4902 }, { "epoch": 0.16, "learning_rate": 1.8849753242650828e-05, "loss": 1.4776, "step": 4903 }, { "epoch": 0.16, "learning_rate": 1.884926065284652e-05, "loss": 1.6273, "step": 4904 }, { "epoch": 0.16, "learning_rate": 1.8848767964028846e-05, "loss": 1.611, "step": 4905 }, { "epoch": 0.16, "learning_rate": 1.8848275176203313e-05, "loss": 1.4501, "step": 4906 }, { "epoch": 0.16, "learning_rate": 1.8847782289375437e-05, "loss": 1.4971, "step": 4907 }, { "epoch": 0.16, "learning_rate": 1.8847289303550738e-05, "loss": 1.5707, "step": 4908 }, { "epoch": 0.16, "learning_rate": 1.8846796218734725e-05, "loss": 1.6126, "step": 4909 }, { "epoch": 0.16, "learning_rate": 1.884630303493292e-05, "loss": 1.6264, "step": 4910 }, { "epoch": 0.16, "learning_rate": 1.884580975215084e-05, "loss": 1.5337, "step": 4911 }, { "epoch": 0.16, "learning_rate": 1.8845316370394004e-05, "loss": 1.5484, "step": 4912 }, { "epoch": 0.16, "learning_rate": 1.884482288966793e-05, "loss": 1.5526, "step": 4913 }, { "epoch": 0.16, "learning_rate": 1.8844329309978146e-05, "loss": 1.4924, "step": 4914 }, { "epoch": 0.16, "learning_rate": 1.8843835631330166e-05, "loss": 1.467, "step": 4915 }, { "epoch": 0.16, "learning_rate": 1.8843341853729517e-05, "loss": 1.4505, "step": 4916 }, { "epoch": 0.16, "learning_rate": 1.8842847977181732e-05, "loss": 1.4762, "step": 4917 }, { "epoch": 0.16, "learning_rate": 1.884235400169232e-05, "loss": 1.4495, "step": 4918 }, { "epoch": 0.16, "learning_rate": 1.8841859927266823e-05, "loss": 1.5719, "step": 4919 }, { "epoch": 0.16, "learning_rate": 1.8841365753910765e-05, "loss": 1.5858, "step": 4920 }, { "epoch": 0.16, "learning_rate": 1.884087148162967e-05, "loss": 1.3941, "step": 4921 }, { "epoch": 0.16, "learning_rate": 1.8840377110429075e-05, "loss": 1.4626, "step": 4922 }, { "epoch": 0.16, "learning_rate": 1.8839882640314512e-05, "loss": 1.5135, "step": 4923 }, { "epoch": 0.16, "learning_rate": 1.8839388071291506e-05, "loss": 1.5625, "step": 4924 }, { "epoch": 0.16, "learning_rate": 1.8838893403365595e-05, "loss": 1.474, "step": 4925 }, { "epoch": 0.16, "learning_rate": 1.8838398636542316e-05, "loss": 1.4083, "step": 4926 }, { "epoch": 0.16, "learning_rate": 1.8837903770827202e-05, "loss": 1.3985, "step": 4927 }, { "epoch": 0.16, "learning_rate": 1.883740880622579e-05, "loss": 1.3859, "step": 4928 }, { "epoch": 0.16, "learning_rate": 1.883691374274362e-05, "loss": 1.6596, "step": 4929 }, { "epoch": 0.16, "learning_rate": 1.8836418580386227e-05, "loss": 1.5628, "step": 4930 }, { "epoch": 0.16, "learning_rate": 1.8835923319159158e-05, "loss": 1.6066, "step": 4931 }, { "epoch": 0.16, "learning_rate": 1.8835427959067952e-05, "loss": 1.6647, "step": 4932 }, { "epoch": 0.16, "learning_rate": 1.8834932500118148e-05, "loss": 1.4465, "step": 4933 }, { "epoch": 0.16, "learning_rate": 1.8834436942315293e-05, "loss": 1.563, "step": 4934 }, { "epoch": 0.16, "learning_rate": 1.883394128566493e-05, "loss": 1.555, "step": 4935 }, { "epoch": 0.16, "learning_rate": 1.8833445530172607e-05, "loss": 1.535, "step": 4936 }, { "epoch": 0.16, "learning_rate": 1.8832949675843866e-05, "loss": 1.5289, "step": 4937 }, { "epoch": 0.16, "learning_rate": 1.8832453722684263e-05, "loss": 1.5262, "step": 4938 }, { "epoch": 0.16, "learning_rate": 1.883195767069934e-05, "loss": 1.644, "step": 4939 }, { "epoch": 0.16, "learning_rate": 1.883146151989465e-05, "loss": 1.5508, "step": 4940 }, { "epoch": 0.16, "learning_rate": 1.8830965270275746e-05, "loss": 1.6099, "step": 4941 }, { "epoch": 0.16, "learning_rate": 1.883046892184818e-05, "loss": 1.5723, "step": 4942 }, { "epoch": 0.16, "learning_rate": 1.8829972474617505e-05, "loss": 1.5977, "step": 4943 }, { "epoch": 0.16, "learning_rate": 1.8829475928589272e-05, "loss": 1.7459, "step": 4944 }, { "epoch": 0.16, "learning_rate": 1.882897928376904e-05, "loss": 1.6048, "step": 4945 }, { "epoch": 0.16, "learning_rate": 1.882848254016237e-05, "loss": 1.6174, "step": 4946 }, { "epoch": 0.16, "learning_rate": 1.882798569777481e-05, "loss": 1.5925, "step": 4947 }, { "epoch": 0.16, "learning_rate": 1.882748875661193e-05, "loss": 1.6306, "step": 4948 }, { "epoch": 0.16, "learning_rate": 1.8826991716679283e-05, "loss": 1.4176, "step": 4949 }, { "epoch": 0.17, "learning_rate": 1.8826494577982432e-05, "loss": 1.4959, "step": 4950 }, { "epoch": 0.17, "learning_rate": 1.882599734052694e-05, "loss": 1.5306, "step": 4951 }, { "epoch": 0.17, "learning_rate": 1.8825500004318372e-05, "loss": 1.7164, "step": 4952 }, { "epoch": 0.17, "learning_rate": 1.882500256936229e-05, "loss": 1.6552, "step": 4953 }, { "epoch": 0.17, "learning_rate": 1.882450503566426e-05, "loss": 1.4997, "step": 4954 }, { "epoch": 0.17, "learning_rate": 1.8824007403229852e-05, "loss": 1.5625, "step": 4955 }, { "epoch": 0.17, "learning_rate": 1.882350967206463e-05, "loss": 1.5656, "step": 4956 }, { "epoch": 0.17, "learning_rate": 1.8823011842174166e-05, "loss": 1.5532, "step": 4957 }, { "epoch": 0.17, "learning_rate": 1.8822513913564028e-05, "loss": 1.5537, "step": 4958 }, { "epoch": 0.17, "learning_rate": 1.882201588623979e-05, "loss": 1.5174, "step": 4959 }, { "epoch": 0.17, "learning_rate": 1.882151776020702e-05, "loss": 1.5354, "step": 4960 }, { "epoch": 0.17, "learning_rate": 1.8821019535471295e-05, "loss": 1.4963, "step": 4961 }, { "epoch": 0.17, "learning_rate": 1.882052121203819e-05, "loss": 1.4871, "step": 4962 }, { "epoch": 0.17, "learning_rate": 1.882002278991328e-05, "loss": 1.5664, "step": 4963 }, { "epoch": 0.17, "learning_rate": 1.8819524269102138e-05, "loss": 1.496, "step": 4964 }, { "epoch": 0.17, "learning_rate": 1.8819025649610346e-05, "loss": 1.482, "step": 4965 }, { "epoch": 0.17, "learning_rate": 1.881852693144348e-05, "loss": 1.3728, "step": 4966 }, { "epoch": 0.17, "learning_rate": 1.8818028114607124e-05, "loss": 1.4346, "step": 4967 }, { "epoch": 0.17, "learning_rate": 1.8817529199106858e-05, "loss": 1.3092, "step": 4968 }, { "epoch": 0.17, "learning_rate": 1.8817030184948263e-05, "loss": 1.58, "step": 4969 }, { "epoch": 0.17, "learning_rate": 1.8816531072136924e-05, "loss": 1.4434, "step": 4970 }, { "epoch": 0.17, "learning_rate": 1.8816031860678423e-05, "loss": 1.3619, "step": 4971 }, { "epoch": 0.17, "learning_rate": 1.881553255057835e-05, "loss": 1.4988, "step": 4972 }, { "epoch": 0.17, "learning_rate": 1.8815033141842286e-05, "loss": 1.506, "step": 4973 }, { "epoch": 0.17, "learning_rate": 1.881453363447582e-05, "loss": 1.5241, "step": 4974 }, { "epoch": 0.17, "learning_rate": 1.8814034028484545e-05, "loss": 1.5806, "step": 4975 }, { "epoch": 0.17, "learning_rate": 1.881353432387405e-05, "loss": 1.6061, "step": 4976 }, { "epoch": 0.17, "learning_rate": 1.8813034520649923e-05, "loss": 1.4853, "step": 4977 }, { "epoch": 0.17, "learning_rate": 1.8812534618817758e-05, "loss": 1.4658, "step": 4978 }, { "epoch": 0.17, "learning_rate": 1.881203461838315e-05, "loss": 1.548, "step": 4979 }, { "epoch": 0.17, "learning_rate": 1.881153451935169e-05, "loss": 1.521, "step": 4980 }, { "epoch": 0.17, "learning_rate": 1.8811034321728975e-05, "loss": 1.5542, "step": 4981 }, { "epoch": 0.17, "learning_rate": 1.8810534025520603e-05, "loss": 1.5645, "step": 4982 }, { "epoch": 0.17, "learning_rate": 1.8810033630732172e-05, "loss": 1.5066, "step": 4983 }, { "epoch": 0.17, "learning_rate": 1.880953313736928e-05, "loss": 1.5335, "step": 4984 }, { "epoch": 0.17, "learning_rate": 1.8809032545437525e-05, "loss": 1.4555, "step": 4985 }, { "epoch": 0.17, "learning_rate": 1.880853185494251e-05, "loss": 1.4695, "step": 4986 }, { "epoch": 0.17, "learning_rate": 1.8808031065889834e-05, "loss": 1.5265, "step": 4987 }, { "epoch": 0.17, "learning_rate": 1.880753017828511e-05, "loss": 1.5139, "step": 4988 }, { "epoch": 0.17, "learning_rate": 1.8807029192133927e-05, "loss": 1.4429, "step": 4989 }, { "epoch": 0.17, "learning_rate": 1.8806528107441905e-05, "loss": 1.5886, "step": 4990 }, { "epoch": 0.17, "learning_rate": 1.880602692421464e-05, "loss": 1.4529, "step": 4991 }, { "epoch": 0.17, "learning_rate": 1.880552564245775e-05, "loss": 1.3756, "step": 4992 }, { "epoch": 0.17, "learning_rate": 1.8805024262176836e-05, "loss": 1.4892, "step": 4993 }, { "epoch": 0.17, "learning_rate": 1.880452278337751e-05, "loss": 1.5849, "step": 4994 }, { "epoch": 0.17, "learning_rate": 1.8804021206065378e-05, "loss": 1.5509, "step": 4995 }, { "epoch": 0.17, "learning_rate": 1.8803519530246063e-05, "loss": 1.4472, "step": 4996 }, { "epoch": 0.17, "learning_rate": 1.8803017755925168e-05, "loss": 1.4662, "step": 4997 }, { "epoch": 0.17, "learning_rate": 1.8802515883108314e-05, "loss": 1.5297, "step": 4998 }, { "epoch": 0.17, "learning_rate": 1.880201391180111e-05, "loss": 1.4418, "step": 4999 }, { "epoch": 0.17, "learning_rate": 1.8801511842009184e-05, "loss": 1.4678, "step": 5000 }, { "epoch": 0.17, "learning_rate": 1.880100967373814e-05, "loss": 1.4796, "step": 5001 }, { "epoch": 0.17, "learning_rate": 1.8800507406993605e-05, "loss": 1.4803, "step": 5002 }, { "epoch": 0.17, "learning_rate": 1.8800005041781194e-05, "loss": 1.4997, "step": 5003 }, { "epoch": 0.17, "learning_rate": 1.8799502578106533e-05, "loss": 1.4362, "step": 5004 }, { "epoch": 0.17, "learning_rate": 1.8799000015975242e-05, "loss": 1.5393, "step": 5005 }, { "epoch": 0.17, "learning_rate": 1.8798497355392942e-05, "loss": 1.52, "step": 5006 }, { "epoch": 0.17, "learning_rate": 1.8797994596365258e-05, "loss": 1.4828, "step": 5007 }, { "epoch": 0.17, "learning_rate": 1.8797491738897817e-05, "loss": 1.5847, "step": 5008 }, { "epoch": 0.17, "learning_rate": 1.8796988782996244e-05, "loss": 1.5695, "step": 5009 }, { "epoch": 0.17, "learning_rate": 1.879648572866617e-05, "loss": 1.4972, "step": 5010 }, { "epoch": 0.17, "learning_rate": 1.8795982575913213e-05, "loss": 1.5242, "step": 5011 }, { "epoch": 0.17, "learning_rate": 1.8795479324743015e-05, "loss": 1.5645, "step": 5012 }, { "epoch": 0.17, "learning_rate": 1.8794975975161206e-05, "loss": 1.5392, "step": 5013 }, { "epoch": 0.17, "learning_rate": 1.879447252717341e-05, "loss": 1.5233, "step": 5014 }, { "epoch": 0.17, "learning_rate": 1.8793968980785263e-05, "loss": 1.5578, "step": 5015 }, { "epoch": 0.17, "learning_rate": 1.87934653360024e-05, "loss": 1.5227, "step": 5016 }, { "epoch": 0.17, "learning_rate": 1.8792961592830456e-05, "loss": 1.4612, "step": 5017 }, { "epoch": 0.17, "learning_rate": 1.8792457751275072e-05, "loss": 1.5341, "step": 5018 }, { "epoch": 0.17, "learning_rate": 1.879195381134188e-05, "loss": 1.5338, "step": 5019 }, { "epoch": 0.17, "learning_rate": 1.8791449773036517e-05, "loss": 1.5817, "step": 5020 }, { "epoch": 0.17, "learning_rate": 1.8790945636364628e-05, "loss": 1.5391, "step": 5021 }, { "epoch": 0.17, "learning_rate": 1.8790441401331848e-05, "loss": 1.6319, "step": 5022 }, { "epoch": 0.17, "learning_rate": 1.8789937067943825e-05, "loss": 1.6051, "step": 5023 }, { "epoch": 0.17, "learning_rate": 1.8789432636206197e-05, "loss": 1.7677, "step": 5024 }, { "epoch": 0.17, "learning_rate": 1.8788928106124608e-05, "loss": 1.6606, "step": 5025 }, { "epoch": 0.17, "learning_rate": 1.878842347770471e-05, "loss": 1.4968, "step": 5026 }, { "epoch": 0.17, "learning_rate": 1.878791875095214e-05, "loss": 1.6743, "step": 5027 }, { "epoch": 0.17, "learning_rate": 1.878741392587255e-05, "loss": 1.6599, "step": 5028 }, { "epoch": 0.17, "learning_rate": 1.878690900247159e-05, "loss": 1.7003, "step": 5029 }, { "epoch": 0.17, "learning_rate": 1.8786403980754908e-05, "loss": 1.7028, "step": 5030 }, { "epoch": 0.17, "learning_rate": 1.8785898860728152e-05, "loss": 1.5458, "step": 5031 }, { "epoch": 0.17, "learning_rate": 1.8785393642396976e-05, "loss": 1.5451, "step": 5032 }, { "epoch": 0.17, "learning_rate": 1.8784888325767033e-05, "loss": 1.6995, "step": 5033 }, { "epoch": 0.17, "learning_rate": 1.8784382910843978e-05, "loss": 1.5659, "step": 5034 }, { "epoch": 0.17, "learning_rate": 1.878387739763346e-05, "loss": 1.5328, "step": 5035 }, { "epoch": 0.17, "learning_rate": 1.8783371786141146e-05, "loss": 1.6812, "step": 5036 }, { "epoch": 0.17, "learning_rate": 1.8782866076372685e-05, "loss": 1.5789, "step": 5037 }, { "epoch": 0.17, "learning_rate": 1.878236026833373e-05, "loss": 1.6204, "step": 5038 }, { "epoch": 0.17, "learning_rate": 1.8781854362029957e-05, "loss": 1.5715, "step": 5039 }, { "epoch": 0.17, "learning_rate": 1.8781348357467017e-05, "loss": 1.5101, "step": 5040 }, { "epoch": 0.17, "learning_rate": 1.8780842254650565e-05, "loss": 1.6532, "step": 5041 }, { "epoch": 0.17, "learning_rate": 1.8780336053586278e-05, "loss": 1.5482, "step": 5042 }, { "epoch": 0.17, "learning_rate": 1.8779829754279806e-05, "loss": 1.4319, "step": 5043 }, { "epoch": 0.17, "learning_rate": 1.8779323356736824e-05, "loss": 1.6094, "step": 5044 }, { "epoch": 0.17, "learning_rate": 1.8778816860962995e-05, "loss": 1.4378, "step": 5045 }, { "epoch": 0.17, "learning_rate": 1.8778310266963985e-05, "loss": 1.63, "step": 5046 }, { "epoch": 0.17, "learning_rate": 1.877780357474546e-05, "loss": 1.4957, "step": 5047 }, { "epoch": 0.17, "learning_rate": 1.8777296784313095e-05, "loss": 1.5449, "step": 5048 }, { "epoch": 0.17, "learning_rate": 1.8776789895672557e-05, "loss": 1.5527, "step": 5049 }, { "epoch": 0.17, "learning_rate": 1.8776282908829516e-05, "loss": 1.5503, "step": 5050 }, { "epoch": 0.17, "learning_rate": 1.877577582378965e-05, "loss": 1.6082, "step": 5051 }, { "epoch": 0.17, "learning_rate": 1.877526864055863e-05, "loss": 1.4905, "step": 5052 }, { "epoch": 0.17, "learning_rate": 1.8774761359142127e-05, "loss": 1.5303, "step": 5053 }, { "epoch": 0.17, "learning_rate": 1.877425397954582e-05, "loss": 1.5387, "step": 5054 }, { "epoch": 0.17, "learning_rate": 1.8773746501775388e-05, "loss": 1.6552, "step": 5055 }, { "epoch": 0.17, "learning_rate": 1.8773238925836507e-05, "loss": 1.4664, "step": 5056 }, { "epoch": 0.17, "learning_rate": 1.8772731251734853e-05, "loss": 1.6205, "step": 5057 }, { "epoch": 0.17, "learning_rate": 1.8772223479476114e-05, "loss": 1.601, "step": 5058 }, { "epoch": 0.17, "learning_rate": 1.8771715609065966e-05, "loss": 1.4467, "step": 5059 }, { "epoch": 0.17, "learning_rate": 1.8771207640510092e-05, "loss": 1.5327, "step": 5060 }, { "epoch": 0.17, "learning_rate": 1.8770699573814176e-05, "loss": 1.6966, "step": 5061 }, { "epoch": 0.17, "learning_rate": 1.8770191408983903e-05, "loss": 1.5654, "step": 5062 }, { "epoch": 0.17, "learning_rate": 1.876968314602496e-05, "loss": 1.4928, "step": 5063 }, { "epoch": 0.17, "learning_rate": 1.8769174784943032e-05, "loss": 1.6451, "step": 5064 }, { "epoch": 0.17, "learning_rate": 1.876866632574381e-05, "loss": 1.5191, "step": 5065 }, { "epoch": 0.17, "learning_rate": 1.8768157768432976e-05, "loss": 1.5371, "step": 5066 }, { "epoch": 0.17, "learning_rate": 1.8767649113016225e-05, "loss": 1.6152, "step": 5067 }, { "epoch": 0.17, "learning_rate": 1.876714035949925e-05, "loss": 1.4931, "step": 5068 }, { "epoch": 0.17, "learning_rate": 1.8766631507887744e-05, "loss": 1.54, "step": 5069 }, { "epoch": 0.17, "learning_rate": 1.8766122558187394e-05, "loss": 1.4319, "step": 5070 }, { "epoch": 0.17, "learning_rate": 1.8765613510403898e-05, "loss": 1.3908, "step": 5071 }, { "epoch": 0.17, "learning_rate": 1.8765104364542955e-05, "loss": 1.4582, "step": 5072 }, { "epoch": 0.17, "learning_rate": 1.8764595120610258e-05, "loss": 1.5073, "step": 5073 }, { "epoch": 0.17, "learning_rate": 1.8764085778611507e-05, "loss": 1.4587, "step": 5074 }, { "epoch": 0.17, "learning_rate": 1.87635763385524e-05, "loss": 1.4045, "step": 5075 }, { "epoch": 0.17, "learning_rate": 1.8763066800438638e-05, "loss": 1.5433, "step": 5076 }, { "epoch": 0.17, "learning_rate": 1.8762557164275922e-05, "loss": 1.5249, "step": 5077 }, { "epoch": 0.17, "learning_rate": 1.8762047430069947e-05, "loss": 1.3957, "step": 5078 }, { "epoch": 0.17, "learning_rate": 1.8761537597826426e-05, "loss": 1.4408, "step": 5079 }, { "epoch": 0.17, "learning_rate": 1.8761027667551063e-05, "loss": 1.4495, "step": 5080 }, { "epoch": 0.17, "learning_rate": 1.876051763924956e-05, "loss": 1.5011, "step": 5081 }, { "epoch": 0.17, "learning_rate": 1.8760007512927624e-05, "loss": 1.6069, "step": 5082 }, { "epoch": 0.17, "learning_rate": 1.8759497288590963e-05, "loss": 1.5211, "step": 5083 }, { "epoch": 0.17, "learning_rate": 1.8758986966245282e-05, "loss": 1.4656, "step": 5084 }, { "epoch": 0.17, "learning_rate": 1.87584765458963e-05, "loss": 1.5676, "step": 5085 }, { "epoch": 0.17, "learning_rate": 1.8757966027549722e-05, "loss": 1.6107, "step": 5086 }, { "epoch": 0.17, "learning_rate": 1.875745541121126e-05, "loss": 1.5538, "step": 5087 }, { "epoch": 0.17, "learning_rate": 1.875694469688663e-05, "loss": 1.5145, "step": 5088 }, { "epoch": 0.17, "learning_rate": 1.875643388458154e-05, "loss": 1.5756, "step": 5089 }, { "epoch": 0.17, "learning_rate": 1.8755922974301716e-05, "loss": 1.6408, "step": 5090 }, { "epoch": 0.17, "learning_rate": 1.8755411966052867e-05, "loss": 1.585, "step": 5091 }, { "epoch": 0.17, "learning_rate": 1.875490085984071e-05, "loss": 1.5859, "step": 5092 }, { "epoch": 0.17, "learning_rate": 1.8754389655670967e-05, "loss": 1.5976, "step": 5093 }, { "epoch": 0.17, "learning_rate": 1.8753878353549357e-05, "loss": 1.473, "step": 5094 }, { "epoch": 0.17, "learning_rate": 1.8753366953481603e-05, "loss": 1.5665, "step": 5095 }, { "epoch": 0.17, "learning_rate": 1.875285545547342e-05, "loss": 1.5983, "step": 5096 }, { "epoch": 0.17, "learning_rate": 1.875234385953054e-05, "loss": 1.5845, "step": 5097 }, { "epoch": 0.17, "learning_rate": 1.8751832165658682e-05, "loss": 1.6469, "step": 5098 }, { "epoch": 0.17, "learning_rate": 1.875132037386357e-05, "loss": 1.6461, "step": 5099 }, { "epoch": 0.17, "learning_rate": 1.8750808484150936e-05, "loss": 1.566, "step": 5100 }, { "epoch": 0.17, "learning_rate": 1.87502964965265e-05, "loss": 1.5916, "step": 5101 }, { "epoch": 0.17, "learning_rate": 1.8749784410995996e-05, "loss": 1.536, "step": 5102 }, { "epoch": 0.17, "learning_rate": 1.8749272227565153e-05, "loss": 1.5662, "step": 5103 }, { "epoch": 0.17, "learning_rate": 1.87487599462397e-05, "loss": 1.4989, "step": 5104 }, { "epoch": 0.17, "learning_rate": 1.874824756702537e-05, "loss": 1.3919, "step": 5105 }, { "epoch": 0.17, "learning_rate": 1.8747735089927898e-05, "loss": 1.516, "step": 5106 }, { "epoch": 0.17, "learning_rate": 1.8747222514953015e-05, "loss": 1.4908, "step": 5107 }, { "epoch": 0.17, "learning_rate": 1.8746709842106457e-05, "loss": 1.4595, "step": 5108 }, { "epoch": 0.17, "learning_rate": 1.874619707139396e-05, "loss": 1.4148, "step": 5109 }, { "epoch": 0.17, "learning_rate": 1.8745684202821262e-05, "loss": 1.492, "step": 5110 }, { "epoch": 0.17, "learning_rate": 1.87451712363941e-05, "loss": 1.4924, "step": 5111 }, { "epoch": 0.17, "learning_rate": 1.8744658172118215e-05, "loss": 1.3918, "step": 5112 }, { "epoch": 0.17, "learning_rate": 1.8744145009999348e-05, "loss": 1.4626, "step": 5113 }, { "epoch": 0.17, "learning_rate": 1.874363175004324e-05, "loss": 1.4853, "step": 5114 }, { "epoch": 0.17, "learning_rate": 1.874311839225563e-05, "loss": 1.4607, "step": 5115 }, { "epoch": 0.17, "learning_rate": 1.874260493664227e-05, "loss": 1.5084, "step": 5116 }, { "epoch": 0.17, "learning_rate": 1.87420913832089e-05, "loss": 1.5282, "step": 5117 }, { "epoch": 0.17, "learning_rate": 1.8741577731961266e-05, "loss": 1.5604, "step": 5118 }, { "epoch": 0.17, "learning_rate": 1.8741063982905117e-05, "loss": 1.4953, "step": 5119 }, { "epoch": 0.17, "learning_rate": 1.8740550136046195e-05, "loss": 1.587, "step": 5120 }, { "epoch": 0.17, "learning_rate": 1.874003619139026e-05, "loss": 1.51, "step": 5121 }, { "epoch": 0.17, "learning_rate": 1.8739522148943055e-05, "loss": 1.5745, "step": 5122 }, { "epoch": 0.17, "learning_rate": 1.8739008008710334e-05, "loss": 1.6193, "step": 5123 }, { "epoch": 0.17, "learning_rate": 1.873849377069785e-05, "loss": 1.6549, "step": 5124 }, { "epoch": 0.17, "learning_rate": 1.873797943491136e-05, "loss": 1.6258, "step": 5125 }, { "epoch": 0.17, "learning_rate": 1.8737465001356607e-05, "loss": 1.6096, "step": 5126 }, { "epoch": 0.17, "learning_rate": 1.8736950470039355e-05, "loss": 1.634, "step": 5127 }, { "epoch": 0.17, "learning_rate": 1.8736435840965367e-05, "loss": 1.6184, "step": 5128 }, { "epoch": 0.17, "learning_rate": 1.873592111414039e-05, "loss": 1.5416, "step": 5129 }, { "epoch": 0.17, "learning_rate": 1.8735406289570193e-05, "loss": 1.5451, "step": 5130 }, { "epoch": 0.17, "learning_rate": 1.8734891367260528e-05, "loss": 1.5218, "step": 5131 }, { "epoch": 0.17, "learning_rate": 1.8734376347217158e-05, "loss": 1.572, "step": 5132 }, { "epoch": 0.17, "learning_rate": 1.873386122944585e-05, "loss": 1.6004, "step": 5133 }, { "epoch": 0.17, "learning_rate": 1.8733346013952362e-05, "loss": 1.4652, "step": 5134 }, { "epoch": 0.17, "learning_rate": 1.8732830700742466e-05, "loss": 1.5888, "step": 5135 }, { "epoch": 0.17, "learning_rate": 1.873231528982192e-05, "loss": 1.5257, "step": 5136 }, { "epoch": 0.17, "learning_rate": 1.8731799781196498e-05, "loss": 1.523, "step": 5137 }, { "epoch": 0.17, "learning_rate": 1.8731284174871962e-05, "loss": 1.5402, "step": 5138 }, { "epoch": 0.17, "learning_rate": 1.8730768470854085e-05, "loss": 1.4778, "step": 5139 }, { "epoch": 0.17, "learning_rate": 1.8730252669148633e-05, "loss": 1.5696, "step": 5140 }, { "epoch": 0.17, "learning_rate": 1.872973676976138e-05, "loss": 1.4976, "step": 5141 }, { "epoch": 0.17, "learning_rate": 1.8729220772698096e-05, "loss": 1.4585, "step": 5142 }, { "epoch": 0.17, "learning_rate": 1.872870467796456e-05, "loss": 1.4758, "step": 5143 }, { "epoch": 0.17, "learning_rate": 1.8728188485566545e-05, "loss": 1.4969, "step": 5144 }, { "epoch": 0.17, "learning_rate": 1.872767219550982e-05, "loss": 1.4826, "step": 5145 }, { "epoch": 0.17, "learning_rate": 1.872715580780017e-05, "loss": 1.6604, "step": 5146 }, { "epoch": 0.17, "learning_rate": 1.8726639322443367e-05, "loss": 1.5054, "step": 5147 }, { "epoch": 0.17, "learning_rate": 1.872612273944519e-05, "loss": 1.6894, "step": 5148 }, { "epoch": 0.17, "learning_rate": 1.8725606058811424e-05, "loss": 1.6013, "step": 5149 }, { "epoch": 0.17, "learning_rate": 1.8725089280547847e-05, "loss": 1.6418, "step": 5150 }, { "epoch": 0.17, "learning_rate": 1.8724572404660243e-05, "loss": 1.5343, "step": 5151 }, { "epoch": 0.17, "learning_rate": 1.872405543115439e-05, "loss": 1.5775, "step": 5152 }, { "epoch": 0.17, "learning_rate": 1.872353836003608e-05, "loss": 1.5799, "step": 5153 }, { "epoch": 0.17, "learning_rate": 1.872302119131109e-05, "loss": 1.534, "step": 5154 }, { "epoch": 0.17, "learning_rate": 1.872250392498521e-05, "loss": 1.4942, "step": 5155 }, { "epoch": 0.17, "learning_rate": 1.8721986561064233e-05, "loss": 1.4319, "step": 5156 }, { "epoch": 0.17, "learning_rate": 1.8721469099553943e-05, "loss": 1.4798, "step": 5157 }, { "epoch": 0.17, "learning_rate": 1.8720951540460128e-05, "loss": 1.4813, "step": 5158 }, { "epoch": 0.17, "learning_rate": 1.872043388378858e-05, "loss": 1.4905, "step": 5159 }, { "epoch": 0.17, "learning_rate": 1.8719916129545094e-05, "loss": 1.5248, "step": 5160 }, { "epoch": 0.17, "learning_rate": 1.871939827773546e-05, "loss": 1.5275, "step": 5161 }, { "epoch": 0.17, "learning_rate": 1.8718880328365475e-05, "loss": 1.447, "step": 5162 }, { "epoch": 0.17, "learning_rate": 1.871836228144093e-05, "loss": 1.4326, "step": 5163 }, { "epoch": 0.17, "learning_rate": 1.8717844136967626e-05, "loss": 1.4378, "step": 5164 }, { "epoch": 0.17, "learning_rate": 1.8717325894951356e-05, "loss": 1.4936, "step": 5165 }, { "epoch": 0.17, "learning_rate": 1.8716807555397924e-05, "loss": 1.4237, "step": 5166 }, { "epoch": 0.17, "learning_rate": 1.8716289118313124e-05, "loss": 1.428, "step": 5167 }, { "epoch": 0.17, "learning_rate": 1.8715770583702763e-05, "loss": 1.5025, "step": 5168 }, { "epoch": 0.17, "learning_rate": 1.8715251951572635e-05, "loss": 1.4668, "step": 5169 }, { "epoch": 0.17, "learning_rate": 1.8714733221928547e-05, "loss": 1.4238, "step": 5170 }, { "epoch": 0.17, "learning_rate": 1.8714214394776306e-05, "loss": 1.5349, "step": 5171 }, { "epoch": 0.17, "learning_rate": 1.8713695470121714e-05, "loss": 1.5259, "step": 5172 }, { "epoch": 0.17, "learning_rate": 1.8713176447970576e-05, "loss": 1.5444, "step": 5173 }, { "epoch": 0.17, "learning_rate": 1.87126573283287e-05, "loss": 1.4841, "step": 5174 }, { "epoch": 0.17, "learning_rate": 1.8712138111201898e-05, "loss": 1.5827, "step": 5175 }, { "epoch": 0.17, "learning_rate": 1.8711618796595972e-05, "loss": 1.4715, "step": 5176 }, { "epoch": 0.17, "learning_rate": 1.871109938451674e-05, "loss": 1.5072, "step": 5177 }, { "epoch": 0.17, "learning_rate": 1.871057987497001e-05, "loss": 1.5274, "step": 5178 }, { "epoch": 0.17, "learning_rate": 1.8710060267961596e-05, "loss": 1.6669, "step": 5179 }, { "epoch": 0.17, "learning_rate": 1.8709540563497312e-05, "loss": 1.4065, "step": 5180 }, { "epoch": 0.17, "learning_rate": 1.8709020761582967e-05, "loss": 1.482, "step": 5181 }, { "epoch": 0.17, "learning_rate": 1.8708500862224387e-05, "loss": 1.5492, "step": 5182 }, { "epoch": 0.17, "learning_rate": 1.8707980865427382e-05, "loss": 1.5556, "step": 5183 }, { "epoch": 0.17, "learning_rate": 1.8707460771197773e-05, "loss": 1.5554, "step": 5184 }, { "epoch": 0.17, "learning_rate": 1.870694057954138e-05, "loss": 1.4911, "step": 5185 }, { "epoch": 0.17, "learning_rate": 1.870642029046402e-05, "loss": 1.6081, "step": 5186 }, { "epoch": 0.17, "learning_rate": 1.870589990397152e-05, "loss": 1.4921, "step": 5187 }, { "epoch": 0.17, "learning_rate": 1.8705379420069695e-05, "loss": 1.3952, "step": 5188 }, { "epoch": 0.17, "learning_rate": 1.8704858838764374e-05, "loss": 1.5274, "step": 5189 }, { "epoch": 0.17, "learning_rate": 1.870433816006138e-05, "loss": 1.4398, "step": 5190 }, { "epoch": 0.17, "learning_rate": 1.8703817383966542e-05, "loss": 1.4764, "step": 5191 }, { "epoch": 0.17, "learning_rate": 1.870329651048568e-05, "loss": 1.5299, "step": 5192 }, { "epoch": 0.17, "learning_rate": 1.8702775539624628e-05, "loss": 1.3726, "step": 5193 }, { "epoch": 0.17, "learning_rate": 1.8702254471389214e-05, "loss": 1.5775, "step": 5194 }, { "epoch": 0.17, "learning_rate": 1.8701733305785265e-05, "loss": 1.5678, "step": 5195 }, { "epoch": 0.17, "learning_rate": 1.8701212042818616e-05, "loss": 1.5257, "step": 5196 }, { "epoch": 0.17, "learning_rate": 1.87006906824951e-05, "loss": 1.7061, "step": 5197 }, { "epoch": 0.17, "learning_rate": 1.8700169224820545e-05, "loss": 1.5256, "step": 5198 }, { "epoch": 0.17, "learning_rate": 1.869964766980079e-05, "loss": 1.5006, "step": 5199 }, { "epoch": 0.17, "learning_rate": 1.8699126017441672e-05, "loss": 1.4816, "step": 5200 }, { "epoch": 0.17, "learning_rate": 1.8698604267749024e-05, "loss": 1.6194, "step": 5201 }, { "epoch": 0.17, "learning_rate": 1.8698082420728685e-05, "loss": 1.6004, "step": 5202 }, { "epoch": 0.17, "learning_rate": 1.8697560476386494e-05, "loss": 1.5466, "step": 5203 }, { "epoch": 0.17, "learning_rate": 1.869703843472829e-05, "loss": 1.5312, "step": 5204 }, { "epoch": 0.17, "learning_rate": 1.8696516295759914e-05, "loss": 1.5499, "step": 5205 }, { "epoch": 0.17, "learning_rate": 1.869599405948721e-05, "loss": 1.5822, "step": 5206 }, { "epoch": 0.17, "learning_rate": 1.8695471725916023e-05, "loss": 1.6214, "step": 5207 }, { "epoch": 0.17, "learning_rate": 1.869494929505219e-05, "loss": 1.5462, "step": 5208 }, { "epoch": 0.17, "learning_rate": 1.8694426766901567e-05, "loss": 1.6284, "step": 5209 }, { "epoch": 0.17, "learning_rate": 1.869390414146999e-05, "loss": 1.6037, "step": 5210 }, { "epoch": 0.17, "learning_rate": 1.8693381418763312e-05, "loss": 1.633, "step": 5211 }, { "epoch": 0.17, "learning_rate": 1.869285859878738e-05, "loss": 1.5403, "step": 5212 }, { "epoch": 0.17, "learning_rate": 1.8692335681548052e-05, "loss": 1.4952, "step": 5213 }, { "epoch": 0.17, "learning_rate": 1.8691812667051164e-05, "loss": 1.5597, "step": 5214 }, { "epoch": 0.17, "learning_rate": 1.8691289555302578e-05, "loss": 1.5208, "step": 5215 }, { "epoch": 0.17, "learning_rate": 1.8690766346308144e-05, "loss": 1.5738, "step": 5216 }, { "epoch": 0.17, "learning_rate": 1.8690243040073715e-05, "loss": 1.5745, "step": 5217 }, { "epoch": 0.17, "learning_rate": 1.868971963660515e-05, "loss": 1.4905, "step": 5218 }, { "epoch": 0.17, "learning_rate": 1.8689196135908303e-05, "loss": 1.6819, "step": 5219 }, { "epoch": 0.17, "learning_rate": 1.8688672537989032e-05, "loss": 1.5604, "step": 5220 }, { "epoch": 0.17, "learning_rate": 1.868814884285319e-05, "loss": 1.6279, "step": 5221 }, { "epoch": 0.17, "learning_rate": 1.868762505050665e-05, "loss": 1.5893, "step": 5222 }, { "epoch": 0.17, "learning_rate": 1.868710116095526e-05, "loss": 1.4608, "step": 5223 }, { "epoch": 0.17, "learning_rate": 1.8686577174204887e-05, "loss": 1.5144, "step": 5224 }, { "epoch": 0.17, "learning_rate": 1.8686053090261395e-05, "loss": 1.5615, "step": 5225 }, { "epoch": 0.17, "learning_rate": 1.8685528909130643e-05, "loss": 1.4838, "step": 5226 }, { "epoch": 0.17, "learning_rate": 1.8685004630818496e-05, "loss": 1.6659, "step": 5227 }, { "epoch": 0.17, "learning_rate": 1.8684480255330827e-05, "loss": 1.6049, "step": 5228 }, { "epoch": 0.17, "learning_rate": 1.8683955782673496e-05, "loss": 1.4958, "step": 5229 }, { "epoch": 0.17, "learning_rate": 1.868343121285238e-05, "loss": 1.5609, "step": 5230 }, { "epoch": 0.17, "learning_rate": 1.8682906545873337e-05, "loss": 1.4814, "step": 5231 }, { "epoch": 0.17, "learning_rate": 1.8682381781742246e-05, "loss": 1.6414, "step": 5232 }, { "epoch": 0.17, "learning_rate": 1.8681856920464972e-05, "loss": 1.5857, "step": 5233 }, { "epoch": 0.17, "learning_rate": 1.8681331962047396e-05, "loss": 1.4705, "step": 5234 }, { "epoch": 0.17, "learning_rate": 1.8680806906495384e-05, "loss": 1.5276, "step": 5235 }, { "epoch": 0.17, "learning_rate": 1.8680281753814813e-05, "loss": 1.5888, "step": 5236 }, { "epoch": 0.17, "learning_rate": 1.8679756504011563e-05, "loss": 1.5651, "step": 5237 }, { "epoch": 0.17, "learning_rate": 1.8679231157091507e-05, "loss": 1.569, "step": 5238 }, { "epoch": 0.17, "learning_rate": 1.867870571306052e-05, "loss": 1.5483, "step": 5239 }, { "epoch": 0.17, "learning_rate": 1.867818017192449e-05, "loss": 1.4366, "step": 5240 }, { "epoch": 0.17, "learning_rate": 1.8677654533689287e-05, "loss": 1.5129, "step": 5241 }, { "epoch": 0.17, "learning_rate": 1.86771287983608e-05, "loss": 1.7271, "step": 5242 }, { "epoch": 0.17, "learning_rate": 1.8676602965944906e-05, "loss": 1.5753, "step": 5243 }, { "epoch": 0.17, "learning_rate": 1.867607703644749e-05, "loss": 1.5836, "step": 5244 }, { "epoch": 0.17, "learning_rate": 1.8675551009874443e-05, "loss": 1.4339, "step": 5245 }, { "epoch": 0.17, "learning_rate": 1.8675024886231642e-05, "loss": 1.4728, "step": 5246 }, { "epoch": 0.17, "learning_rate": 1.867449866552498e-05, "loss": 1.5039, "step": 5247 }, { "epoch": 0.17, "learning_rate": 1.8673972347760338e-05, "loss": 1.3076, "step": 5248 }, { "epoch": 0.17, "learning_rate": 1.8673445932943612e-05, "loss": 1.5006, "step": 5249 }, { "epoch": 0.17, "learning_rate": 1.867291942108069e-05, "loss": 1.4811, "step": 5250 }, { "epoch": 0.18, "learning_rate": 1.867239281217746e-05, "loss": 1.4218, "step": 5251 }, { "epoch": 0.18, "learning_rate": 1.8671866106239812e-05, "loss": 1.53, "step": 5252 }, { "epoch": 0.18, "learning_rate": 1.8671339303273647e-05, "loss": 1.4987, "step": 5253 }, { "epoch": 0.18, "learning_rate": 1.8670812403284858e-05, "loss": 1.5118, "step": 5254 }, { "epoch": 0.18, "learning_rate": 1.8670285406279335e-05, "loss": 1.4575, "step": 5255 }, { "epoch": 0.18, "learning_rate": 1.866975831226298e-05, "loss": 1.4139, "step": 5256 }, { "epoch": 0.18, "learning_rate": 1.8669231121241683e-05, "loss": 1.4474, "step": 5257 }, { "epoch": 0.18, "learning_rate": 1.8668703833221355e-05, "loss": 1.526, "step": 5258 }, { "epoch": 0.18, "learning_rate": 1.8668176448207883e-05, "loss": 1.5142, "step": 5259 }, { "epoch": 0.18, "learning_rate": 1.8667648966207175e-05, "loss": 1.5632, "step": 5260 }, { "epoch": 0.18, "learning_rate": 1.8667121387225134e-05, "loss": 1.6416, "step": 5261 }, { "epoch": 0.18, "learning_rate": 1.8666593711267656e-05, "loss": 1.5535, "step": 5262 }, { "epoch": 0.18, "learning_rate": 1.866606593834065e-05, "loss": 1.5899, "step": 5263 }, { "epoch": 0.18, "learning_rate": 1.8665538068450023e-05, "loss": 1.559, "step": 5264 }, { "epoch": 0.18, "learning_rate": 1.866501010160168e-05, "loss": 1.5279, "step": 5265 }, { "epoch": 0.18, "learning_rate": 1.8664482037801527e-05, "loss": 1.5077, "step": 5266 }, { "epoch": 0.18, "learning_rate": 1.8663953877055472e-05, "loss": 1.5034, "step": 5267 }, { "epoch": 0.18, "learning_rate": 1.8663425619369424e-05, "loss": 1.5561, "step": 5268 }, { "epoch": 0.18, "learning_rate": 1.8662897264749295e-05, "loss": 1.5279, "step": 5269 }, { "epoch": 0.18, "learning_rate": 1.8662368813201e-05, "loss": 1.5679, "step": 5270 }, { "epoch": 0.18, "learning_rate": 1.8661840264730445e-05, "loss": 1.5708, "step": 5271 }, { "epoch": 0.18, "learning_rate": 1.8661311619343546e-05, "loss": 1.5581, "step": 5272 }, { "epoch": 0.18, "learning_rate": 1.8660782877046225e-05, "loss": 1.5089, "step": 5273 }, { "epoch": 0.18, "learning_rate": 1.866025403784439e-05, "loss": 1.5367, "step": 5274 }, { "epoch": 0.18, "learning_rate": 1.8659725101743956e-05, "loss": 1.5799, "step": 5275 }, { "epoch": 0.18, "learning_rate": 1.865919606875085e-05, "loss": 1.6016, "step": 5276 }, { "epoch": 0.18, "learning_rate": 1.8658666938870985e-05, "loss": 1.5587, "step": 5277 }, { "epoch": 0.18, "learning_rate": 1.8658137712110287e-05, "loss": 1.5961, "step": 5278 }, { "epoch": 0.18, "learning_rate": 1.865760838847467e-05, "loss": 1.5419, "step": 5279 }, { "epoch": 0.18, "learning_rate": 1.8657078967970063e-05, "loss": 1.5538, "step": 5280 }, { "epoch": 0.18, "learning_rate": 1.8656549450602382e-05, "loss": 1.623, "step": 5281 }, { "epoch": 0.18, "learning_rate": 1.8656019836377562e-05, "loss": 1.522, "step": 5282 }, { "epoch": 0.18, "learning_rate": 1.8655490125301523e-05, "loss": 1.4939, "step": 5283 }, { "epoch": 0.18, "learning_rate": 1.865496031738019e-05, "loss": 1.5139, "step": 5284 }, { "epoch": 0.18, "learning_rate": 1.8654430412619494e-05, "loss": 1.5645, "step": 5285 }, { "epoch": 0.18, "learning_rate": 1.8653900411025365e-05, "loss": 1.5646, "step": 5286 }, { "epoch": 0.18, "learning_rate": 1.8653370312603728e-05, "loss": 1.5357, "step": 5287 }, { "epoch": 0.18, "learning_rate": 1.865284011736052e-05, "loss": 1.5803, "step": 5288 }, { "epoch": 0.18, "learning_rate": 1.865230982530167e-05, "loss": 1.6079, "step": 5289 }, { "epoch": 0.18, "learning_rate": 1.8651779436433112e-05, "loss": 1.4371, "step": 5290 }, { "epoch": 0.18, "learning_rate": 1.865124895076078e-05, "loss": 1.4796, "step": 5291 }, { "epoch": 0.18, "learning_rate": 1.8650718368290613e-05, "loss": 1.5101, "step": 5292 }, { "epoch": 0.18, "learning_rate": 1.8650187689028544e-05, "loss": 1.5759, "step": 5293 }, { "epoch": 0.18, "learning_rate": 1.8649656912980508e-05, "loss": 1.6038, "step": 5294 }, { "epoch": 0.18, "learning_rate": 1.8649126040152454e-05, "loss": 1.489, "step": 5295 }, { "epoch": 0.18, "learning_rate": 1.8648595070550312e-05, "loss": 1.4674, "step": 5296 }, { "epoch": 0.18, "learning_rate": 1.8648064004180023e-05, "loss": 1.4921, "step": 5297 }, { "epoch": 0.18, "learning_rate": 1.8647532841047536e-05, "loss": 1.5579, "step": 5298 }, { "epoch": 0.18, "learning_rate": 1.864700158115879e-05, "loss": 1.5751, "step": 5299 }, { "epoch": 0.18, "learning_rate": 1.864647022451973e-05, "loss": 1.636, "step": 5300 }, { "epoch": 0.18, "learning_rate": 1.8645938771136303e-05, "loss": 1.5316, "step": 5301 }, { "epoch": 0.18, "learning_rate": 1.864540722101445e-05, "loss": 1.4083, "step": 5302 }, { "epoch": 0.18, "learning_rate": 1.864487557416012e-05, "loss": 1.5465, "step": 5303 }, { "epoch": 0.18, "learning_rate": 1.864434383057927e-05, "loss": 1.5822, "step": 5304 }, { "epoch": 0.18, "learning_rate": 1.864381199027784e-05, "loss": 1.5385, "step": 5305 }, { "epoch": 0.18, "learning_rate": 1.864328005326178e-05, "loss": 1.4763, "step": 5306 }, { "epoch": 0.18, "learning_rate": 1.864274801953705e-05, "loss": 1.4317, "step": 5307 }, { "epoch": 0.18, "learning_rate": 1.8642215889109594e-05, "loss": 1.4728, "step": 5308 }, { "epoch": 0.18, "learning_rate": 1.864168366198537e-05, "loss": 1.4148, "step": 5309 }, { "epoch": 0.18, "learning_rate": 1.864115133817034e-05, "loss": 1.6244, "step": 5310 }, { "epoch": 0.18, "learning_rate": 1.864061891767045e-05, "loss": 1.4453, "step": 5311 }, { "epoch": 0.18, "learning_rate": 1.8640086400491658e-05, "loss": 1.4545, "step": 5312 }, { "epoch": 0.18, "learning_rate": 1.8639553786639927e-05, "loss": 1.4345, "step": 5313 }, { "epoch": 0.18, "learning_rate": 1.8639021076121214e-05, "loss": 1.3711, "step": 5314 }, { "epoch": 0.18, "learning_rate": 1.863848826894148e-05, "loss": 1.493, "step": 5315 }, { "epoch": 0.18, "learning_rate": 1.8637955365106685e-05, "loss": 1.4215, "step": 5316 }, { "epoch": 0.18, "learning_rate": 1.863742236462279e-05, "loss": 1.4386, "step": 5317 }, { "epoch": 0.18, "learning_rate": 1.8636889267495767e-05, "loss": 1.4518, "step": 5318 }, { "epoch": 0.18, "learning_rate": 1.863635607373157e-05, "loss": 1.4712, "step": 5319 }, { "epoch": 0.18, "learning_rate": 1.8635822783336174e-05, "loss": 1.4377, "step": 5320 }, { "epoch": 0.18, "learning_rate": 1.8635289396315543e-05, "loss": 1.4587, "step": 5321 }, { "epoch": 0.18, "learning_rate": 1.863475591267564e-05, "loss": 1.5101, "step": 5322 }, { "epoch": 0.18, "learning_rate": 1.863422233242244e-05, "loss": 1.3815, "step": 5323 }, { "epoch": 0.18, "learning_rate": 1.863368865556191e-05, "loss": 1.4913, "step": 5324 }, { "epoch": 0.18, "learning_rate": 1.8633154882100024e-05, "loss": 1.5495, "step": 5325 }, { "epoch": 0.18, "learning_rate": 1.8632621012042755e-05, "loss": 1.5403, "step": 5326 }, { "epoch": 0.18, "learning_rate": 1.863208704539607e-05, "loss": 1.6209, "step": 5327 }, { "epoch": 0.18, "learning_rate": 1.8631552982165946e-05, "loss": 1.5551, "step": 5328 }, { "epoch": 0.18, "learning_rate": 1.8631018822358363e-05, "loss": 1.5277, "step": 5329 }, { "epoch": 0.18, "learning_rate": 1.8630484565979298e-05, "loss": 1.5284, "step": 5330 }, { "epoch": 0.18, "learning_rate": 1.862995021303472e-05, "loss": 1.675, "step": 5331 }, { "epoch": 0.18, "learning_rate": 1.862941576353062e-05, "loss": 1.5292, "step": 5332 }, { "epoch": 0.18, "learning_rate": 1.8628881217472967e-05, "loss": 1.6005, "step": 5333 }, { "epoch": 0.18, "learning_rate": 1.8628346574867748e-05, "loss": 1.603, "step": 5334 }, { "epoch": 0.18, "learning_rate": 1.8627811835720942e-05, "loss": 1.541, "step": 5335 }, { "epoch": 0.18, "learning_rate": 1.8627277000038534e-05, "loss": 1.4808, "step": 5336 }, { "epoch": 0.18, "learning_rate": 1.8626742067826506e-05, "loss": 1.4564, "step": 5337 }, { "epoch": 0.18, "learning_rate": 1.8626207039090845e-05, "loss": 1.5168, "step": 5338 }, { "epoch": 0.18, "learning_rate": 1.8625671913837542e-05, "loss": 1.4468, "step": 5339 }, { "epoch": 0.18, "learning_rate": 1.8625136692072577e-05, "loss": 1.4977, "step": 5340 }, { "epoch": 0.18, "learning_rate": 1.862460137380194e-05, "loss": 1.6004, "step": 5341 }, { "epoch": 0.18, "learning_rate": 1.8624065959031626e-05, "loss": 1.5265, "step": 5342 }, { "epoch": 0.18, "learning_rate": 1.8623530447767617e-05, "loss": 1.5453, "step": 5343 }, { "epoch": 0.18, "learning_rate": 1.862299484001591e-05, "loss": 1.4977, "step": 5344 }, { "epoch": 0.18, "learning_rate": 1.8622459135782503e-05, "loss": 1.5991, "step": 5345 }, { "epoch": 0.18, "learning_rate": 1.8621923335073378e-05, "loss": 1.547, "step": 5346 }, { "epoch": 0.18, "learning_rate": 1.8621387437894538e-05, "loss": 1.5396, "step": 5347 }, { "epoch": 0.18, "learning_rate": 1.8620851444251978e-05, "loss": 1.5858, "step": 5348 }, { "epoch": 0.18, "learning_rate": 1.8620315354151695e-05, "loss": 1.4897, "step": 5349 }, { "epoch": 0.18, "learning_rate": 1.8619779167599685e-05, "loss": 1.4351, "step": 5350 }, { "epoch": 0.18, "learning_rate": 1.8619242884601953e-05, "loss": 1.6434, "step": 5351 }, { "epoch": 0.18, "learning_rate": 1.861870650516449e-05, "loss": 1.6027, "step": 5352 }, { "epoch": 0.18, "learning_rate": 1.861817002929331e-05, "loss": 1.5803, "step": 5353 }, { "epoch": 0.18, "learning_rate": 1.86176334569944e-05, "loss": 1.5933, "step": 5354 }, { "epoch": 0.18, "learning_rate": 1.8617096788273778e-05, "loss": 1.4871, "step": 5355 }, { "epoch": 0.18, "learning_rate": 1.8616560023137445e-05, "loss": 1.5875, "step": 5356 }, { "epoch": 0.18, "learning_rate": 1.8616023161591398e-05, "loss": 1.5765, "step": 5357 }, { "epoch": 0.18, "learning_rate": 1.8615486203641655e-05, "loss": 1.6146, "step": 5358 }, { "epoch": 0.18, "learning_rate": 1.861494914929422e-05, "loss": 1.6622, "step": 5359 }, { "epoch": 0.18, "learning_rate": 1.86144119985551e-05, "loss": 1.4676, "step": 5360 }, { "epoch": 0.18, "learning_rate": 1.8613874751430304e-05, "loss": 1.5035, "step": 5361 }, { "epoch": 0.18, "learning_rate": 1.8613337407925854e-05, "loss": 1.5104, "step": 5362 }, { "epoch": 0.18, "learning_rate": 1.8612799968047748e-05, "loss": 1.3623, "step": 5363 }, { "epoch": 0.18, "learning_rate": 1.861226243180201e-05, "loss": 1.5646, "step": 5364 }, { "epoch": 0.18, "learning_rate": 1.8611724799194646e-05, "loss": 1.4773, "step": 5365 }, { "epoch": 0.18, "learning_rate": 1.8611187070231676e-05, "loss": 1.5037, "step": 5366 }, { "epoch": 0.18, "learning_rate": 1.8610649244919114e-05, "loss": 1.5642, "step": 5367 }, { "epoch": 0.18, "learning_rate": 1.8610111323262986e-05, "loss": 1.5524, "step": 5368 }, { "epoch": 0.18, "learning_rate": 1.86095733052693e-05, "loss": 1.5546, "step": 5369 }, { "epoch": 0.18, "learning_rate": 1.860903519094408e-05, "loss": 1.5923, "step": 5370 }, { "epoch": 0.18, "learning_rate": 1.860849698029335e-05, "loss": 1.6019, "step": 5371 }, { "epoch": 0.18, "learning_rate": 1.8607958673323126e-05, "loss": 1.5924, "step": 5372 }, { "epoch": 0.18, "learning_rate": 1.860742027003944e-05, "loss": 1.5127, "step": 5373 }, { "epoch": 0.18, "learning_rate": 1.8606881770448305e-05, "loss": 1.6293, "step": 5374 }, { "epoch": 0.18, "learning_rate": 1.8606343174555753e-05, "loss": 1.6081, "step": 5375 }, { "epoch": 0.18, "learning_rate": 1.860580448236781e-05, "loss": 1.6897, "step": 5376 }, { "epoch": 0.18, "learning_rate": 1.86052656938905e-05, "loss": 1.6264, "step": 5377 }, { "epoch": 0.18, "learning_rate": 1.860472680912986e-05, "loss": 1.5744, "step": 5378 }, { "epoch": 0.18, "learning_rate": 1.8604187828091906e-05, "loss": 1.5528, "step": 5379 }, { "epoch": 0.18, "learning_rate": 1.8603648750782676e-05, "loss": 1.7647, "step": 5380 }, { "epoch": 0.18, "learning_rate": 1.8603109577208206e-05, "loss": 1.5954, "step": 5381 }, { "epoch": 0.18, "learning_rate": 1.8602570307374523e-05, "loss": 1.6398, "step": 5382 }, { "epoch": 0.18, "learning_rate": 1.860203094128766e-05, "loss": 1.7104, "step": 5383 }, { "epoch": 0.18, "learning_rate": 1.860149147895366e-05, "loss": 1.621, "step": 5384 }, { "epoch": 0.18, "learning_rate": 1.8600951920378546e-05, "loss": 1.626, "step": 5385 }, { "epoch": 0.18, "learning_rate": 1.860041226556836e-05, "loss": 1.5966, "step": 5386 }, { "epoch": 0.18, "learning_rate": 1.859987251452915e-05, "loss": 1.6299, "step": 5387 }, { "epoch": 0.18, "learning_rate": 1.8599332667266942e-05, "loss": 1.3978, "step": 5388 }, { "epoch": 0.18, "learning_rate": 1.8598792723787784e-05, "loss": 1.4971, "step": 5389 }, { "epoch": 0.18, "learning_rate": 1.8598252684097716e-05, "loss": 1.5798, "step": 5390 }, { "epoch": 0.18, "learning_rate": 1.859771254820278e-05, "loss": 1.5426, "step": 5391 }, { "epoch": 0.18, "learning_rate": 1.8597172316109015e-05, "loss": 1.5535, "step": 5392 }, { "epoch": 0.18, "learning_rate": 1.859663198782247e-05, "loss": 1.5216, "step": 5393 }, { "epoch": 0.18, "learning_rate": 1.859609156334919e-05, "loss": 1.5711, "step": 5394 }, { "epoch": 0.18, "learning_rate": 1.859555104269523e-05, "loss": 1.5775, "step": 5395 }, { "epoch": 0.18, "learning_rate": 1.8595010425866622e-05, "loss": 1.5991, "step": 5396 }, { "epoch": 0.18, "learning_rate": 1.8594469712869425e-05, "loss": 1.6539, "step": 5397 }, { "epoch": 0.18, "learning_rate": 1.8593928903709685e-05, "loss": 1.6277, "step": 5398 }, { "epoch": 0.18, "learning_rate": 1.8593387998393457e-05, "loss": 1.6023, "step": 5399 }, { "epoch": 0.18, "learning_rate": 1.8592846996926793e-05, "loss": 1.63, "step": 5400 }, { "epoch": 0.18, "learning_rate": 1.859230589931574e-05, "loss": 1.6616, "step": 5401 }, { "epoch": 0.18, "learning_rate": 1.8591764705566358e-05, "loss": 1.5487, "step": 5402 }, { "epoch": 0.18, "learning_rate": 1.85912234156847e-05, "loss": 1.5563, "step": 5403 }, { "epoch": 0.18, "learning_rate": 1.8590682029676825e-05, "loss": 1.6344, "step": 5404 }, { "epoch": 0.18, "learning_rate": 1.859014054754879e-05, "loss": 1.5235, "step": 5405 }, { "epoch": 0.18, "learning_rate": 1.8589598969306646e-05, "loss": 1.4296, "step": 5406 }, { "epoch": 0.18, "learning_rate": 1.8589057294956463e-05, "loss": 1.5858, "step": 5407 }, { "epoch": 0.18, "learning_rate": 1.8588515524504296e-05, "loss": 1.5571, "step": 5408 }, { "epoch": 0.18, "learning_rate": 1.858797365795621e-05, "loss": 1.5427, "step": 5409 }, { "epoch": 0.18, "learning_rate": 1.8587431695318266e-05, "loss": 1.4864, "step": 5410 }, { "epoch": 0.18, "learning_rate": 1.858688963659653e-05, "loss": 1.473, "step": 5411 }, { "epoch": 0.18, "learning_rate": 1.8586347481797064e-05, "loss": 1.5631, "step": 5412 }, { "epoch": 0.18, "learning_rate": 1.8585805230925933e-05, "loss": 1.5546, "step": 5413 }, { "epoch": 0.18, "learning_rate": 1.858526288398921e-05, "loss": 1.494, "step": 5414 }, { "epoch": 0.18, "learning_rate": 1.858472044099296e-05, "loss": 1.4692, "step": 5415 }, { "epoch": 0.18, "learning_rate": 1.858417790194325e-05, "loss": 1.3674, "step": 5416 }, { "epoch": 0.18, "learning_rate": 1.8583635266846155e-05, "loss": 1.5196, "step": 5417 }, { "epoch": 0.18, "learning_rate": 1.8583092535707742e-05, "loss": 1.5325, "step": 5418 }, { "epoch": 0.18, "learning_rate": 1.8582549708534086e-05, "loss": 1.5162, "step": 5419 }, { "epoch": 0.18, "learning_rate": 1.8582006785331258e-05, "loss": 1.3979, "step": 5420 }, { "epoch": 0.18, "learning_rate": 1.858146376610534e-05, "loss": 1.4857, "step": 5421 }, { "epoch": 0.18, "learning_rate": 1.85809206508624e-05, "loss": 1.5003, "step": 5422 }, { "epoch": 0.18, "learning_rate": 1.8580377439608517e-05, "loss": 1.6019, "step": 5423 }, { "epoch": 0.18, "learning_rate": 1.8579834132349773e-05, "loss": 1.5179, "step": 5424 }, { "epoch": 0.18, "learning_rate": 1.8579290729092242e-05, "loss": 1.5452, "step": 5425 }, { "epoch": 0.18, "learning_rate": 1.8578747229842005e-05, "loss": 1.5253, "step": 5426 }, { "epoch": 0.18, "learning_rate": 1.857820363460514e-05, "loss": 1.5377, "step": 5427 }, { "epoch": 0.18, "learning_rate": 1.8577659943387737e-05, "loss": 1.4068, "step": 5428 }, { "epoch": 0.18, "learning_rate": 1.8577116156195877e-05, "loss": 1.529, "step": 5429 }, { "epoch": 0.18, "learning_rate": 1.857657227303564e-05, "loss": 1.4608, "step": 5430 }, { "epoch": 0.18, "learning_rate": 1.8576028293913114e-05, "loss": 1.4031, "step": 5431 }, { "epoch": 0.18, "learning_rate": 1.857548421883439e-05, "loss": 1.4668, "step": 5432 }, { "epoch": 0.18, "learning_rate": 1.8574940047805547e-05, "loss": 1.4368, "step": 5433 }, { "epoch": 0.18, "learning_rate": 1.857439578083268e-05, "loss": 1.469, "step": 5434 }, { "epoch": 0.18, "learning_rate": 1.8573851417921873e-05, "loss": 1.597, "step": 5435 }, { "epoch": 0.18, "learning_rate": 1.857330695907922e-05, "loss": 1.4324, "step": 5436 }, { "epoch": 0.18, "learning_rate": 1.857276240431082e-05, "loss": 1.512, "step": 5437 }, { "epoch": 0.18, "learning_rate": 1.8572217753622755e-05, "loss": 1.5656, "step": 5438 }, { "epoch": 0.18, "learning_rate": 1.8571673007021124e-05, "loss": 1.5833, "step": 5439 }, { "epoch": 0.18, "learning_rate": 1.8571128164512022e-05, "loss": 1.5353, "step": 5440 }, { "epoch": 0.18, "learning_rate": 1.8570583226101545e-05, "loss": 1.5422, "step": 5441 }, { "epoch": 0.18, "learning_rate": 1.857003819179579e-05, "loss": 1.5007, "step": 5442 }, { "epoch": 0.18, "learning_rate": 1.856949306160085e-05, "loss": 1.6081, "step": 5443 }, { "epoch": 0.18, "learning_rate": 1.8568947835522836e-05, "loss": 1.4504, "step": 5444 }, { "epoch": 0.18, "learning_rate": 1.856840251356784e-05, "loss": 1.6253, "step": 5445 }, { "epoch": 0.18, "learning_rate": 1.8567857095741968e-05, "loss": 1.6296, "step": 5446 }, { "epoch": 0.18, "learning_rate": 1.8567311582051318e-05, "loss": 1.6212, "step": 5447 }, { "epoch": 0.18, "learning_rate": 1.8566765972501994e-05, "loss": 1.479, "step": 5448 }, { "epoch": 0.18, "learning_rate": 1.8566220267100107e-05, "loss": 1.5429, "step": 5449 }, { "epoch": 0.18, "learning_rate": 1.8565674465851753e-05, "loss": 1.5855, "step": 5450 }, { "epoch": 0.18, "learning_rate": 1.856512856876305e-05, "loss": 1.5455, "step": 5451 }, { "epoch": 0.18, "learning_rate": 1.85645825758401e-05, "loss": 1.4519, "step": 5452 }, { "epoch": 0.18, "learning_rate": 1.856403648708901e-05, "loss": 1.5935, "step": 5453 }, { "epoch": 0.18, "learning_rate": 1.856349030251589e-05, "loss": 1.6011, "step": 5454 }, { "epoch": 0.18, "learning_rate": 1.8562944022126857e-05, "loss": 1.539, "step": 5455 }, { "epoch": 0.18, "learning_rate": 1.8562397645928022e-05, "loss": 1.5865, "step": 5456 }, { "epoch": 0.18, "learning_rate": 1.8561851173925495e-05, "loss": 1.4947, "step": 5457 }, { "epoch": 0.18, "learning_rate": 1.8561304606125392e-05, "loss": 1.5776, "step": 5458 }, { "epoch": 0.18, "learning_rate": 1.8560757942533828e-05, "loss": 1.616, "step": 5459 }, { "epoch": 0.18, "learning_rate": 1.8560211183156918e-05, "loss": 1.6606, "step": 5460 }, { "epoch": 0.18, "learning_rate": 1.8559664328000782e-05, "loss": 1.7053, "step": 5461 }, { "epoch": 0.18, "learning_rate": 1.855911737707154e-05, "loss": 1.5875, "step": 5462 }, { "epoch": 0.18, "learning_rate": 1.855857033037531e-05, "loss": 1.561, "step": 5463 }, { "epoch": 0.18, "learning_rate": 1.8558023187918214e-05, "loss": 1.6003, "step": 5464 }, { "epoch": 0.18, "learning_rate": 1.855747594970637e-05, "loss": 1.6646, "step": 5465 }, { "epoch": 0.18, "learning_rate": 1.8556928615745903e-05, "loss": 1.7833, "step": 5466 }, { "epoch": 0.18, "learning_rate": 1.855638118604294e-05, "loss": 1.5476, "step": 5467 }, { "epoch": 0.18, "learning_rate": 1.8555833660603608e-05, "loss": 1.6605, "step": 5468 }, { "epoch": 0.18, "learning_rate": 1.8555286039434022e-05, "loss": 1.5419, "step": 5469 }, { "epoch": 0.18, "learning_rate": 1.855473832254032e-05, "loss": 1.5841, "step": 5470 }, { "epoch": 0.18, "learning_rate": 1.8554190509928625e-05, "loss": 1.5246, "step": 5471 }, { "epoch": 0.18, "learning_rate": 1.855364260160507e-05, "loss": 1.6836, "step": 5472 }, { "epoch": 0.18, "learning_rate": 1.8553094597575782e-05, "loss": 1.5874, "step": 5473 }, { "epoch": 0.18, "learning_rate": 1.8552546497846893e-05, "loss": 1.5436, "step": 5474 }, { "epoch": 0.18, "learning_rate": 1.8551998302424538e-05, "loss": 1.6033, "step": 5475 }, { "epoch": 0.18, "learning_rate": 1.8551450011314853e-05, "loss": 1.4496, "step": 5476 }, { "epoch": 0.18, "learning_rate": 1.8550901624523963e-05, "loss": 1.5024, "step": 5477 }, { "epoch": 0.18, "learning_rate": 1.8550353142058016e-05, "loss": 1.5193, "step": 5478 }, { "epoch": 0.18, "learning_rate": 1.8549804563923136e-05, "loss": 1.4449, "step": 5479 }, { "epoch": 0.18, "learning_rate": 1.8549255890125476e-05, "loss": 1.3976, "step": 5480 }, { "epoch": 0.18, "learning_rate": 1.854870712067116e-05, "loss": 1.4488, "step": 5481 }, { "epoch": 0.18, "learning_rate": 1.8548158255566342e-05, "loss": 1.5263, "step": 5482 }, { "epoch": 0.18, "learning_rate": 1.854760929481715e-05, "loss": 1.4855, "step": 5483 }, { "epoch": 0.18, "learning_rate": 1.8547060238429737e-05, "loss": 1.4334, "step": 5484 }, { "epoch": 0.18, "learning_rate": 1.854651108641024e-05, "loss": 1.5558, "step": 5485 }, { "epoch": 0.18, "learning_rate": 1.8545961838764806e-05, "loss": 1.3705, "step": 5486 }, { "epoch": 0.18, "learning_rate": 1.854541249549958e-05, "loss": 1.5187, "step": 5487 }, { "epoch": 0.18, "learning_rate": 1.8544863056620708e-05, "loss": 1.438, "step": 5488 }, { "epoch": 0.18, "learning_rate": 1.8544313522134335e-05, "loss": 1.6124, "step": 5489 }, { "epoch": 0.18, "learning_rate": 1.8543763892046618e-05, "loss": 1.4119, "step": 5490 }, { "epoch": 0.18, "learning_rate": 1.85432141663637e-05, "loss": 1.5006, "step": 5491 }, { "epoch": 0.18, "learning_rate": 1.8542664345091732e-05, "loss": 1.4273, "step": 5492 }, { "epoch": 0.18, "learning_rate": 1.8542114428236864e-05, "loss": 1.4992, "step": 5493 }, { "epoch": 0.18, "learning_rate": 1.854156441580526e-05, "loss": 1.4968, "step": 5494 }, { "epoch": 0.18, "learning_rate": 1.8541014307803058e-05, "loss": 1.4871, "step": 5495 }, { "epoch": 0.18, "learning_rate": 1.8540464104236428e-05, "loss": 1.5493, "step": 5496 }, { "epoch": 0.18, "learning_rate": 1.8539913805111512e-05, "loss": 1.4994, "step": 5497 }, { "epoch": 0.18, "learning_rate": 1.853936341043448e-05, "loss": 1.4898, "step": 5498 }, { "epoch": 0.18, "learning_rate": 1.8538812920211484e-05, "loss": 1.4816, "step": 5499 }, { "epoch": 0.18, "learning_rate": 1.8538262334448683e-05, "loss": 1.4957, "step": 5500 }, { "epoch": 0.18, "learning_rate": 1.853771165315224e-05, "loss": 1.4143, "step": 5501 }, { "epoch": 0.18, "learning_rate": 1.8537160876328313e-05, "loss": 1.4981, "step": 5502 }, { "epoch": 0.18, "learning_rate": 1.8536610003983066e-05, "loss": 1.4165, "step": 5503 }, { "epoch": 0.18, "learning_rate": 1.853605903612267e-05, "loss": 1.4888, "step": 5504 }, { "epoch": 0.18, "learning_rate": 1.8535507972753275e-05, "loss": 1.3621, "step": 5505 }, { "epoch": 0.18, "learning_rate": 1.8534956813881058e-05, "loss": 1.5669, "step": 5506 }, { "epoch": 0.18, "learning_rate": 1.8534405559512184e-05, "loss": 1.42, "step": 5507 }, { "epoch": 0.18, "learning_rate": 1.853385420965282e-05, "loss": 1.472, "step": 5508 }, { "epoch": 0.18, "learning_rate": 1.8533302764309134e-05, "loss": 1.4647, "step": 5509 }, { "epoch": 0.18, "learning_rate": 1.8532751223487294e-05, "loss": 1.4793, "step": 5510 }, { "epoch": 0.18, "learning_rate": 1.8532199587193477e-05, "loss": 1.4654, "step": 5511 }, { "epoch": 0.18, "learning_rate": 1.8531647855433853e-05, "loss": 1.5085, "step": 5512 }, { "epoch": 0.18, "learning_rate": 1.853109602821459e-05, "loss": 1.4903, "step": 5513 }, { "epoch": 0.18, "learning_rate": 1.8530544105541872e-05, "loss": 1.5163, "step": 5514 }, { "epoch": 0.18, "learning_rate": 1.8529992087421866e-05, "loss": 1.4928, "step": 5515 }, { "epoch": 0.18, "learning_rate": 1.852943997386075e-05, "loss": 1.4007, "step": 5516 }, { "epoch": 0.18, "learning_rate": 1.852888776486471e-05, "loss": 1.4965, "step": 5517 }, { "epoch": 0.18, "learning_rate": 1.852833546043991e-05, "loss": 1.5093, "step": 5518 }, { "epoch": 0.18, "learning_rate": 1.852778306059254e-05, "loss": 1.4794, "step": 5519 }, { "epoch": 0.18, "learning_rate": 1.852723056532878e-05, "loss": 1.6406, "step": 5520 }, { "epoch": 0.18, "learning_rate": 1.852667797465481e-05, "loss": 1.6267, "step": 5521 }, { "epoch": 0.18, "learning_rate": 1.8526125288576807e-05, "loss": 1.4101, "step": 5522 }, { "epoch": 0.18, "learning_rate": 1.8525572507100964e-05, "loss": 1.4955, "step": 5523 }, { "epoch": 0.18, "learning_rate": 1.8525019630233463e-05, "loss": 1.492, "step": 5524 }, { "epoch": 0.18, "learning_rate": 1.852446665798049e-05, "loss": 1.5377, "step": 5525 }, { "epoch": 0.18, "learning_rate": 1.852391359034823e-05, "loss": 1.4485, "step": 5526 }, { "epoch": 0.18, "learning_rate": 1.8523360427342877e-05, "loss": 1.4685, "step": 5527 }, { "epoch": 0.18, "learning_rate": 1.8522807168970616e-05, "loss": 1.5369, "step": 5528 }, { "epoch": 0.18, "learning_rate": 1.8522253815237636e-05, "loss": 1.4552, "step": 5529 }, { "epoch": 0.18, "learning_rate": 1.852170036615013e-05, "loss": 1.569, "step": 5530 }, { "epoch": 0.18, "learning_rate": 1.852114682171429e-05, "loss": 1.5785, "step": 5531 }, { "epoch": 0.18, "learning_rate": 1.8520593181936312e-05, "loss": 1.5786, "step": 5532 }, { "epoch": 0.18, "learning_rate": 1.8520039446822388e-05, "loss": 1.5798, "step": 5533 }, { "epoch": 0.18, "learning_rate": 1.8519485616378717e-05, "loss": 1.4686, "step": 5534 }, { "epoch": 0.18, "learning_rate": 1.851893169061149e-05, "loss": 1.5875, "step": 5535 }, { "epoch": 0.18, "learning_rate": 1.851837766952691e-05, "loss": 1.5932, "step": 5536 }, { "epoch": 0.18, "learning_rate": 1.851782355313117e-05, "loss": 1.6229, "step": 5537 }, { "epoch": 0.18, "learning_rate": 1.851726934143048e-05, "loss": 1.6172, "step": 5538 }, { "epoch": 0.18, "learning_rate": 1.8516715034431028e-05, "loss": 1.4591, "step": 5539 }, { "epoch": 0.18, "learning_rate": 1.8516160632139026e-05, "loss": 1.6183, "step": 5540 }, { "epoch": 0.18, "learning_rate": 1.8515606134560676e-05, "loss": 1.5, "step": 5541 }, { "epoch": 0.18, "learning_rate": 1.851505154170218e-05, "loss": 1.5748, "step": 5542 }, { "epoch": 0.18, "learning_rate": 1.851449685356974e-05, "loss": 1.6971, "step": 5543 }, { "epoch": 0.18, "learning_rate": 1.8513942070169572e-05, "loss": 1.6576, "step": 5544 }, { "epoch": 0.18, "learning_rate": 1.8513387191507874e-05, "loss": 1.6393, "step": 5545 }, { "epoch": 0.18, "learning_rate": 1.8512832217590856e-05, "loss": 1.7095, "step": 5546 }, { "epoch": 0.18, "learning_rate": 1.8512277148424732e-05, "loss": 1.7756, "step": 5547 }, { "epoch": 0.18, "learning_rate": 1.851172198401571e-05, "loss": 1.882, "step": 5548 }, { "epoch": 0.18, "learning_rate": 1.8511166724369997e-05, "loss": 1.6712, "step": 5549 }, { "epoch": 0.18, "learning_rate": 1.8510611369493815e-05, "loss": 1.6864, "step": 5550 }, { "epoch": 0.19, "learning_rate": 1.8510055919393374e-05, "loss": 1.6285, "step": 5551 }, { "epoch": 0.19, "learning_rate": 1.8509500374074886e-05, "loss": 1.5038, "step": 5552 }, { "epoch": 0.19, "learning_rate": 1.8508944733544568e-05, "loss": 1.596, "step": 5553 }, { "epoch": 0.19, "learning_rate": 1.8508388997808638e-05, "loss": 1.7277, "step": 5554 }, { "epoch": 0.19, "learning_rate": 1.8507833166873317e-05, "loss": 1.5471, "step": 5555 }, { "epoch": 0.19, "learning_rate": 1.8507277240744818e-05, "loss": 1.5691, "step": 5556 }, { "epoch": 0.19, "learning_rate": 1.8506721219429368e-05, "loss": 1.6004, "step": 5557 }, { "epoch": 0.19, "learning_rate": 1.850616510293318e-05, "loss": 1.48, "step": 5558 }, { "epoch": 0.19, "learning_rate": 1.8505608891262487e-05, "loss": 1.5139, "step": 5559 }, { "epoch": 0.19, "learning_rate": 1.85050525844235e-05, "loss": 1.5401, "step": 5560 }, { "epoch": 0.19, "learning_rate": 1.8504496182422455e-05, "loss": 1.4531, "step": 5561 }, { "epoch": 0.19, "learning_rate": 1.850393968526557e-05, "loss": 1.5746, "step": 5562 }, { "epoch": 0.19, "learning_rate": 1.850338309295907e-05, "loss": 1.461, "step": 5563 }, { "epoch": 0.19, "learning_rate": 1.850282640550919e-05, "loss": 1.5101, "step": 5564 }, { "epoch": 0.19, "learning_rate": 1.8502269622922157e-05, "loss": 1.6135, "step": 5565 }, { "epoch": 0.19, "learning_rate": 1.8501712745204197e-05, "loss": 1.606, "step": 5566 }, { "epoch": 0.19, "learning_rate": 1.8501155772361543e-05, "loss": 1.5607, "step": 5567 }, { "epoch": 0.19, "learning_rate": 1.8500598704400427e-05, "loss": 1.5965, "step": 5568 }, { "epoch": 0.19, "learning_rate": 1.8500041541327082e-05, "loss": 1.5965, "step": 5569 }, { "epoch": 0.19, "learning_rate": 1.849948428314774e-05, "loss": 1.4409, "step": 5570 }, { "epoch": 0.19, "learning_rate": 1.849892692986864e-05, "loss": 1.6034, "step": 5571 }, { "epoch": 0.19, "learning_rate": 1.8498369481496015e-05, "loss": 1.487, "step": 5572 }, { "epoch": 0.19, "learning_rate": 1.8497811938036105e-05, "loss": 1.5504, "step": 5573 }, { "epoch": 0.19, "learning_rate": 1.8497254299495147e-05, "loss": 1.6481, "step": 5574 }, { "epoch": 0.19, "learning_rate": 1.8496696565879376e-05, "loss": 1.496, "step": 5575 }, { "epoch": 0.19, "learning_rate": 1.849613873719504e-05, "loss": 1.5632, "step": 5576 }, { "epoch": 0.19, "learning_rate": 1.8495580813448374e-05, "loss": 1.6049, "step": 5577 }, { "epoch": 0.19, "learning_rate": 1.8495022794645625e-05, "loss": 1.602, "step": 5578 }, { "epoch": 0.19, "learning_rate": 1.8494464680793033e-05, "loss": 1.452, "step": 5579 }, { "epoch": 0.19, "learning_rate": 1.8493906471896846e-05, "loss": 1.5036, "step": 5580 }, { "epoch": 0.19, "learning_rate": 1.849334816796331e-05, "loss": 1.4374, "step": 5581 }, { "epoch": 0.19, "learning_rate": 1.8492789768998668e-05, "loss": 1.4025, "step": 5582 }, { "epoch": 0.19, "learning_rate": 1.8492231275009172e-05, "loss": 1.4442, "step": 5583 }, { "epoch": 0.19, "learning_rate": 1.849167268600107e-05, "loss": 1.5462, "step": 5584 }, { "epoch": 0.19, "learning_rate": 1.849111400198061e-05, "loss": 1.4351, "step": 5585 }, { "epoch": 0.19, "learning_rate": 1.849055522295404e-05, "loss": 1.5207, "step": 5586 }, { "epoch": 0.19, "learning_rate": 1.848999634892762e-05, "loss": 1.5109, "step": 5587 }, { "epoch": 0.19, "learning_rate": 1.8489437379907596e-05, "loss": 1.6635, "step": 5588 }, { "epoch": 0.19, "learning_rate": 1.8488878315900228e-05, "loss": 1.6082, "step": 5589 }, { "epoch": 0.19, "learning_rate": 1.848831915691177e-05, "loss": 1.5526, "step": 5590 }, { "epoch": 0.19, "learning_rate": 1.8487759902948476e-05, "loss": 1.5941, "step": 5591 }, { "epoch": 0.19, "learning_rate": 1.8487200554016602e-05, "loss": 1.4588, "step": 5592 }, { "epoch": 0.19, "learning_rate": 1.848664111012241e-05, "loss": 1.6319, "step": 5593 }, { "epoch": 0.19, "learning_rate": 1.8486081571272163e-05, "loss": 1.5726, "step": 5594 }, { "epoch": 0.19, "learning_rate": 1.8485521937472114e-05, "loss": 1.4956, "step": 5595 }, { "epoch": 0.19, "learning_rate": 1.8484962208728528e-05, "loss": 1.4741, "step": 5596 }, { "epoch": 0.19, "learning_rate": 1.8484402385047667e-05, "loss": 1.5257, "step": 5597 }, { "epoch": 0.19, "learning_rate": 1.8483842466435798e-05, "loss": 1.6281, "step": 5598 }, { "epoch": 0.19, "learning_rate": 1.8483282452899183e-05, "loss": 1.5452, "step": 5599 }, { "epoch": 0.19, "learning_rate": 1.8482722344444086e-05, "loss": 1.4947, "step": 5600 }, { "epoch": 0.19, "learning_rate": 1.848216214107678e-05, "loss": 1.6128, "step": 5601 }, { "epoch": 0.19, "learning_rate": 1.8481601842803524e-05, "loss": 1.4309, "step": 5602 }, { "epoch": 0.19, "learning_rate": 1.8481041449630598e-05, "loss": 1.555, "step": 5603 }, { "epoch": 0.19, "learning_rate": 1.848048096156426e-05, "loss": 1.5984, "step": 5604 }, { "epoch": 0.19, "learning_rate": 1.8479920378610795e-05, "loss": 1.5459, "step": 5605 }, { "epoch": 0.19, "learning_rate": 1.8479359700776465e-05, "loss": 1.4265, "step": 5606 }, { "epoch": 0.19, "learning_rate": 1.8478798928067544e-05, "loss": 1.3496, "step": 5607 }, { "epoch": 0.19, "learning_rate": 1.8478238060490312e-05, "loss": 1.4712, "step": 5608 }, { "epoch": 0.19, "learning_rate": 1.847767709805104e-05, "loss": 1.5577, "step": 5609 }, { "epoch": 0.19, "learning_rate": 1.8477116040756007e-05, "loss": 1.4908, "step": 5610 }, { "epoch": 0.19, "learning_rate": 1.847655488861149e-05, "loss": 1.5037, "step": 5611 }, { "epoch": 0.19, "learning_rate": 1.8475993641623766e-05, "loss": 1.5458, "step": 5612 }, { "epoch": 0.19, "learning_rate": 1.847543229979912e-05, "loss": 1.5865, "step": 5613 }, { "epoch": 0.19, "learning_rate": 1.8474870863143822e-05, "loss": 1.4806, "step": 5614 }, { "epoch": 0.19, "learning_rate": 1.8474309331664165e-05, "loss": 1.5131, "step": 5615 }, { "epoch": 0.19, "learning_rate": 1.8473747705366427e-05, "loss": 1.5514, "step": 5616 }, { "epoch": 0.19, "learning_rate": 1.8473185984256893e-05, "loss": 1.6432, "step": 5617 }, { "epoch": 0.19, "learning_rate": 1.8472624168341848e-05, "loss": 1.5426, "step": 5618 }, { "epoch": 0.19, "learning_rate": 1.8472062257627573e-05, "loss": 1.6609, "step": 5619 }, { "epoch": 0.19, "learning_rate": 1.8471500252120364e-05, "loss": 1.4939, "step": 5620 }, { "epoch": 0.19, "learning_rate": 1.8470938151826506e-05, "loss": 1.4846, "step": 5621 }, { "epoch": 0.19, "learning_rate": 1.8470375956752283e-05, "loss": 1.5574, "step": 5622 }, { "epoch": 0.19, "learning_rate": 1.8469813666903994e-05, "loss": 1.4974, "step": 5623 }, { "epoch": 0.19, "learning_rate": 1.8469251282287925e-05, "loss": 1.533, "step": 5624 }, { "epoch": 0.19, "learning_rate": 1.846868880291037e-05, "loss": 1.5639, "step": 5625 }, { "epoch": 0.19, "learning_rate": 1.8468126228777617e-05, "loss": 1.425, "step": 5626 }, { "epoch": 0.19, "learning_rate": 1.8467563559895972e-05, "loss": 1.4744, "step": 5627 }, { "epoch": 0.19, "learning_rate": 1.8467000796271717e-05, "loss": 1.6261, "step": 5628 }, { "epoch": 0.19, "learning_rate": 1.8466437937911164e-05, "loss": 1.4076, "step": 5629 }, { "epoch": 0.19, "learning_rate": 1.8465874984820597e-05, "loss": 1.4806, "step": 5630 }, { "epoch": 0.19, "learning_rate": 1.8465311937006323e-05, "loss": 1.497, "step": 5631 }, { "epoch": 0.19, "learning_rate": 1.8464748794474635e-05, "loss": 1.4902, "step": 5632 }, { "epoch": 0.19, "learning_rate": 1.8464185557231843e-05, "loss": 1.5901, "step": 5633 }, { "epoch": 0.19, "learning_rate": 1.8463622225284242e-05, "loss": 1.6561, "step": 5634 }, { "epoch": 0.19, "learning_rate": 1.846305879863814e-05, "loss": 1.5818, "step": 5635 }, { "epoch": 0.19, "learning_rate": 1.8462495277299838e-05, "loss": 1.5616, "step": 5636 }, { "epoch": 0.19, "learning_rate": 1.8461931661275642e-05, "loss": 1.6103, "step": 5637 }, { "epoch": 0.19, "learning_rate": 1.8461367950571855e-05, "loss": 1.6098, "step": 5638 }, { "epoch": 0.19, "learning_rate": 1.8460804145194792e-05, "loss": 1.4636, "step": 5639 }, { "epoch": 0.19, "learning_rate": 1.846024024515075e-05, "loss": 1.4672, "step": 5640 }, { "epoch": 0.19, "learning_rate": 1.845967625044605e-05, "loss": 1.5436, "step": 5641 }, { "epoch": 0.19, "learning_rate": 1.8459112161086995e-05, "loss": 1.5266, "step": 5642 }, { "epoch": 0.19, "learning_rate": 1.8458547977079903e-05, "loss": 1.5538, "step": 5643 }, { "epoch": 0.19, "learning_rate": 1.8457983698431078e-05, "loss": 1.5646, "step": 5644 }, { "epoch": 0.19, "learning_rate": 1.8457419325146837e-05, "loss": 1.4742, "step": 5645 }, { "epoch": 0.19, "learning_rate": 1.8456854857233498e-05, "loss": 1.5067, "step": 5646 }, { "epoch": 0.19, "learning_rate": 1.8456290294697376e-05, "loss": 1.552, "step": 5647 }, { "epoch": 0.19, "learning_rate": 1.8455725637544784e-05, "loss": 1.5145, "step": 5648 }, { "epoch": 0.19, "learning_rate": 1.8455160885782045e-05, "loss": 1.4842, "step": 5649 }, { "epoch": 0.19, "learning_rate": 1.8454596039415472e-05, "loss": 1.4849, "step": 5650 }, { "epoch": 0.19, "learning_rate": 1.845403109845139e-05, "loss": 1.4851, "step": 5651 }, { "epoch": 0.19, "learning_rate": 1.8453466062896122e-05, "loss": 1.4763, "step": 5652 }, { "epoch": 0.19, "learning_rate": 1.8452900932755982e-05, "loss": 1.4619, "step": 5653 }, { "epoch": 0.19, "learning_rate": 1.84523357080373e-05, "loss": 1.4677, "step": 5654 }, { "epoch": 0.19, "learning_rate": 1.8451770388746398e-05, "loss": 1.4882, "step": 5655 }, { "epoch": 0.19, "learning_rate": 1.8451204974889597e-05, "loss": 1.4264, "step": 5656 }, { "epoch": 0.19, "learning_rate": 1.845063946647323e-05, "loss": 1.4625, "step": 5657 }, { "epoch": 0.19, "learning_rate": 1.8450073863503622e-05, "loss": 1.481, "step": 5658 }, { "epoch": 0.19, "learning_rate": 1.8449508165987106e-05, "loss": 1.4237, "step": 5659 }, { "epoch": 0.19, "learning_rate": 1.844894237393e-05, "loss": 1.472, "step": 5660 }, { "epoch": 0.19, "learning_rate": 1.8448376487338647e-05, "loss": 1.538, "step": 5661 }, { "epoch": 0.19, "learning_rate": 1.844781050621937e-05, "loss": 1.5277, "step": 5662 }, { "epoch": 0.19, "learning_rate": 1.8447244430578505e-05, "loss": 1.5463, "step": 5663 }, { "epoch": 0.19, "learning_rate": 1.8446678260422388e-05, "loss": 1.4729, "step": 5664 }, { "epoch": 0.19, "learning_rate": 1.8446111995757346e-05, "loss": 1.5067, "step": 5665 }, { "epoch": 0.19, "learning_rate": 1.8445545636589724e-05, "loss": 1.5149, "step": 5666 }, { "epoch": 0.19, "learning_rate": 1.8444979182925855e-05, "loss": 1.5304, "step": 5667 }, { "epoch": 0.19, "learning_rate": 1.8444412634772076e-05, "loss": 1.603, "step": 5668 }, { "epoch": 0.19, "learning_rate": 1.844384599213473e-05, "loss": 1.636, "step": 5669 }, { "epoch": 0.19, "learning_rate": 1.8443279255020153e-05, "loss": 1.5708, "step": 5670 }, { "epoch": 0.19, "learning_rate": 1.8442712423434686e-05, "loss": 1.5642, "step": 5671 }, { "epoch": 0.19, "learning_rate": 1.8442145497384673e-05, "loss": 1.5007, "step": 5672 }, { "epoch": 0.19, "learning_rate": 1.844157847687646e-05, "loss": 1.6591, "step": 5673 }, { "epoch": 0.19, "learning_rate": 1.8441011361916387e-05, "loss": 1.5419, "step": 5674 }, { "epoch": 0.19, "learning_rate": 1.84404441525108e-05, "loss": 1.4171, "step": 5675 }, { "epoch": 0.19, "learning_rate": 1.8439876848666048e-05, "loss": 1.4729, "step": 5676 }, { "epoch": 0.19, "learning_rate": 1.843930945038847e-05, "loss": 1.5268, "step": 5677 }, { "epoch": 0.19, "learning_rate": 1.843874195768443e-05, "loss": 1.4726, "step": 5678 }, { "epoch": 0.19, "learning_rate": 1.8438174370560263e-05, "loss": 1.4426, "step": 5679 }, { "epoch": 0.19, "learning_rate": 1.843760668902233e-05, "loss": 1.4536, "step": 5680 }, { "epoch": 0.19, "learning_rate": 1.8437038913076974e-05, "loss": 1.4759, "step": 5681 }, { "epoch": 0.19, "learning_rate": 1.8436471042730555e-05, "loss": 1.4721, "step": 5682 }, { "epoch": 0.19, "learning_rate": 1.8435903077989423e-05, "loss": 1.4415, "step": 5683 }, { "epoch": 0.19, "learning_rate": 1.8435335018859933e-05, "loss": 1.5111, "step": 5684 }, { "epoch": 0.19, "learning_rate": 1.843476686534844e-05, "loss": 1.5576, "step": 5685 }, { "epoch": 0.19, "learning_rate": 1.8434198617461304e-05, "loss": 1.4728, "step": 5686 }, { "epoch": 0.19, "learning_rate": 1.8433630275204882e-05, "loss": 1.3745, "step": 5687 }, { "epoch": 0.19, "learning_rate": 1.8433061838585537e-05, "loss": 1.4295, "step": 5688 }, { "epoch": 0.19, "learning_rate": 1.843249330760962e-05, "loss": 1.5889, "step": 5689 }, { "epoch": 0.19, "learning_rate": 1.84319246822835e-05, "loss": 1.3457, "step": 5690 }, { "epoch": 0.19, "learning_rate": 1.8431355962613535e-05, "loss": 1.5302, "step": 5691 }, { "epoch": 0.19, "learning_rate": 1.8430787148606087e-05, "loss": 1.4902, "step": 5692 }, { "epoch": 0.19, "learning_rate": 1.8430218240267527e-05, "loss": 1.4921, "step": 5693 }, { "epoch": 0.19, "learning_rate": 1.8429649237604215e-05, "loss": 1.4422, "step": 5694 }, { "epoch": 0.19, "learning_rate": 1.842908014062252e-05, "loss": 1.455, "step": 5695 }, { "epoch": 0.19, "learning_rate": 1.842851094932881e-05, "loss": 1.3592, "step": 5696 }, { "epoch": 0.19, "learning_rate": 1.8427941663729453e-05, "loss": 1.4214, "step": 5697 }, { "epoch": 0.19, "learning_rate": 1.8427372283830816e-05, "loss": 1.4918, "step": 5698 }, { "epoch": 0.19, "learning_rate": 1.842680280963927e-05, "loss": 1.3967, "step": 5699 }, { "epoch": 0.19, "learning_rate": 1.8426233241161193e-05, "loss": 1.5379, "step": 5700 }, { "epoch": 0.19, "learning_rate": 1.842566357840295e-05, "loss": 1.383, "step": 5701 }, { "epoch": 0.19, "learning_rate": 1.842509382137092e-05, "loss": 1.4708, "step": 5702 }, { "epoch": 0.19, "learning_rate": 1.842452397007148e-05, "loss": 1.399, "step": 5703 }, { "epoch": 0.19, "learning_rate": 1.8423954024510995e-05, "loss": 1.4318, "step": 5704 }, { "epoch": 0.19, "learning_rate": 1.8423383984695853e-05, "loss": 1.5642, "step": 5705 }, { "epoch": 0.19, "learning_rate": 1.842281385063243e-05, "loss": 1.5023, "step": 5706 }, { "epoch": 0.19, "learning_rate": 1.84222436223271e-05, "loss": 1.5143, "step": 5707 }, { "epoch": 0.19, "learning_rate": 1.842167329978625e-05, "loss": 1.6215, "step": 5708 }, { "epoch": 0.19, "learning_rate": 1.8421102883016253e-05, "loss": 1.5894, "step": 5709 }, { "epoch": 0.19, "learning_rate": 1.84205323720235e-05, "loss": 1.5503, "step": 5710 }, { "epoch": 0.19, "learning_rate": 1.841996176681437e-05, "loss": 1.6373, "step": 5711 }, { "epoch": 0.19, "learning_rate": 1.8419391067395248e-05, "loss": 1.6496, "step": 5712 }, { "epoch": 0.19, "learning_rate": 1.8418820273772523e-05, "loss": 1.7024, "step": 5713 }, { "epoch": 0.19, "learning_rate": 1.8418249385952575e-05, "loss": 1.6682, "step": 5714 }, { "epoch": 0.19, "learning_rate": 1.8417678403941798e-05, "loss": 1.6298, "step": 5715 }, { "epoch": 0.19, "learning_rate": 1.8417107327746574e-05, "loss": 1.6113, "step": 5716 }, { "epoch": 0.19, "learning_rate": 1.8416536157373298e-05, "loss": 1.6266, "step": 5717 }, { "epoch": 0.19, "learning_rate": 1.841596489282836e-05, "loss": 1.5657, "step": 5718 }, { "epoch": 0.19, "learning_rate": 1.8415393534118147e-05, "loss": 1.62, "step": 5719 }, { "epoch": 0.19, "learning_rate": 1.841482208124906e-05, "loss": 1.747, "step": 5720 }, { "epoch": 0.19, "learning_rate": 1.8414250534227485e-05, "loss": 1.6617, "step": 5721 }, { "epoch": 0.19, "learning_rate": 1.8413678893059823e-05, "loss": 1.5595, "step": 5722 }, { "epoch": 0.19, "learning_rate": 1.8413107157752467e-05, "loss": 1.6738, "step": 5723 }, { "epoch": 0.19, "learning_rate": 1.8412535328311813e-05, "loss": 1.4597, "step": 5724 }, { "epoch": 0.19, "learning_rate": 1.8411963404744263e-05, "loss": 1.5963, "step": 5725 }, { "epoch": 0.19, "learning_rate": 1.841139138705621e-05, "loss": 1.4048, "step": 5726 }, { "epoch": 0.19, "learning_rate": 1.8410819275254065e-05, "loss": 1.6092, "step": 5727 }, { "epoch": 0.19, "learning_rate": 1.841024706934422e-05, "loss": 1.5242, "step": 5728 }, { "epoch": 0.19, "learning_rate": 1.8409674769333077e-05, "loss": 1.5589, "step": 5729 }, { "epoch": 0.19, "learning_rate": 1.8409102375227042e-05, "loss": 1.5953, "step": 5730 }, { "epoch": 0.19, "learning_rate": 1.8408529887032524e-05, "loss": 1.567, "step": 5731 }, { "epoch": 0.19, "learning_rate": 1.8407957304755922e-05, "loss": 1.5673, "step": 5732 }, { "epoch": 0.19, "learning_rate": 1.8407384628403642e-05, "loss": 1.6308, "step": 5733 }, { "epoch": 0.19, "learning_rate": 1.84068118579821e-05, "loss": 1.5961, "step": 5734 }, { "epoch": 0.19, "learning_rate": 1.8406238993497697e-05, "loss": 1.4893, "step": 5735 }, { "epoch": 0.19, "learning_rate": 1.8405666034956842e-05, "loss": 1.4924, "step": 5736 }, { "epoch": 0.19, "learning_rate": 1.840509298236595e-05, "loss": 1.5196, "step": 5737 }, { "epoch": 0.19, "learning_rate": 1.8404519835731436e-05, "loss": 1.4397, "step": 5738 }, { "epoch": 0.19, "learning_rate": 1.8403946595059705e-05, "loss": 1.4905, "step": 5739 }, { "epoch": 0.19, "learning_rate": 1.8403373260357172e-05, "loss": 1.408, "step": 5740 }, { "epoch": 0.19, "learning_rate": 1.8402799831630256e-05, "loss": 1.4126, "step": 5741 }, { "epoch": 0.19, "learning_rate": 1.840222630888537e-05, "loss": 1.5377, "step": 5742 }, { "epoch": 0.19, "learning_rate": 1.8401652692128933e-05, "loss": 1.4699, "step": 5743 }, { "epoch": 0.19, "learning_rate": 1.8401078981367364e-05, "loss": 1.4321, "step": 5744 }, { "epoch": 0.19, "learning_rate": 1.840050517660708e-05, "loss": 1.619, "step": 5745 }, { "epoch": 0.19, "learning_rate": 1.83999312778545e-05, "loss": 1.4989, "step": 5746 }, { "epoch": 0.19, "learning_rate": 1.8399357285116045e-05, "loss": 1.4538, "step": 5747 }, { "epoch": 0.19, "learning_rate": 1.8398783198398146e-05, "loss": 1.5104, "step": 5748 }, { "epoch": 0.19, "learning_rate": 1.8398209017707214e-05, "loss": 1.4228, "step": 5749 }, { "epoch": 0.19, "learning_rate": 1.8397634743049683e-05, "loss": 1.4188, "step": 5750 }, { "epoch": 0.19, "learning_rate": 1.8397060374431972e-05, "loss": 1.3191, "step": 5751 }, { "epoch": 0.19, "learning_rate": 1.8396485911860514e-05, "loss": 1.5197, "step": 5752 }, { "epoch": 0.19, "learning_rate": 1.839591135534173e-05, "loss": 1.5144, "step": 5753 }, { "epoch": 0.19, "learning_rate": 1.839533670488205e-05, "loss": 1.4587, "step": 5754 }, { "epoch": 0.19, "learning_rate": 1.8394761960487907e-05, "loss": 1.462, "step": 5755 }, { "epoch": 0.19, "learning_rate": 1.839418712216573e-05, "loss": 1.5164, "step": 5756 }, { "epoch": 0.19, "learning_rate": 1.8393612189921953e-05, "loss": 1.5055, "step": 5757 }, { "epoch": 0.19, "learning_rate": 1.8393037163763005e-05, "loss": 1.5183, "step": 5758 }, { "epoch": 0.19, "learning_rate": 1.839246204369532e-05, "loss": 1.5442, "step": 5759 }, { "epoch": 0.19, "learning_rate": 1.8391886829725334e-05, "loss": 1.4757, "step": 5760 }, { "epoch": 0.19, "learning_rate": 1.8391311521859484e-05, "loss": 1.5424, "step": 5761 }, { "epoch": 0.19, "learning_rate": 1.8390736120104212e-05, "loss": 1.5704, "step": 5762 }, { "epoch": 0.19, "learning_rate": 1.8390160624465946e-05, "loss": 1.5394, "step": 5763 }, { "epoch": 0.19, "learning_rate": 1.838958503495113e-05, "loss": 1.4374, "step": 5764 }, { "epoch": 0.19, "learning_rate": 1.8389009351566206e-05, "loss": 1.425, "step": 5765 }, { "epoch": 0.19, "learning_rate": 1.838843357431761e-05, "loss": 1.5333, "step": 5766 }, { "epoch": 0.19, "learning_rate": 1.8387857703211787e-05, "loss": 1.4053, "step": 5767 }, { "epoch": 0.19, "learning_rate": 1.8387281738255184e-05, "loss": 1.4335, "step": 5768 }, { "epoch": 0.19, "learning_rate": 1.8386705679454243e-05, "loss": 1.5282, "step": 5769 }, { "epoch": 0.19, "learning_rate": 1.8386129526815407e-05, "loss": 1.4814, "step": 5770 }, { "epoch": 0.19, "learning_rate": 1.8385553280345124e-05, "loss": 1.6297, "step": 5771 }, { "epoch": 0.19, "learning_rate": 1.8384976940049842e-05, "loss": 1.5423, "step": 5772 }, { "epoch": 0.19, "learning_rate": 1.8384400505936008e-05, "loss": 1.6224, "step": 5773 }, { "epoch": 0.19, "learning_rate": 1.8383823978010077e-05, "loss": 1.4791, "step": 5774 }, { "epoch": 0.19, "learning_rate": 1.8383247356278496e-05, "loss": 1.5584, "step": 5775 }, { "epoch": 0.19, "learning_rate": 1.8382670640747713e-05, "loss": 1.5025, "step": 5776 }, { "epoch": 0.19, "learning_rate": 1.8382093831424186e-05, "loss": 1.5487, "step": 5777 }, { "epoch": 0.19, "learning_rate": 1.838151692831437e-05, "loss": 1.6152, "step": 5778 }, { "epoch": 0.19, "learning_rate": 1.8380939931424712e-05, "loss": 1.4514, "step": 5779 }, { "epoch": 0.19, "learning_rate": 1.8380362840761675e-05, "loss": 1.5076, "step": 5780 }, { "epoch": 0.19, "learning_rate": 1.8379785656331714e-05, "loss": 1.6092, "step": 5781 }, { "epoch": 0.19, "learning_rate": 1.837920837814129e-05, "loss": 1.5126, "step": 5782 }, { "epoch": 0.19, "learning_rate": 1.8378631006196854e-05, "loss": 1.5136, "step": 5783 }, { "epoch": 0.19, "learning_rate": 1.8378053540504874e-05, "loss": 1.5204, "step": 5784 }, { "epoch": 0.19, "learning_rate": 1.837747598107181e-05, "loss": 1.4664, "step": 5785 }, { "epoch": 0.19, "learning_rate": 1.837689832790412e-05, "loss": 1.5405, "step": 5786 }, { "epoch": 0.19, "learning_rate": 1.8376320581008274e-05, "loss": 1.5599, "step": 5787 }, { "epoch": 0.19, "learning_rate": 1.837574274039073e-05, "loss": 1.602, "step": 5788 }, { "epoch": 0.19, "learning_rate": 1.8375164806057957e-05, "loss": 1.4835, "step": 5789 }, { "epoch": 0.19, "learning_rate": 1.837458677801642e-05, "loss": 1.4226, "step": 5790 }, { "epoch": 0.19, "learning_rate": 1.8374008656272585e-05, "loss": 1.6002, "step": 5791 }, { "epoch": 0.19, "learning_rate": 1.8373430440832925e-05, "loss": 1.3979, "step": 5792 }, { "epoch": 0.19, "learning_rate": 1.8372852131703903e-05, "loss": 1.4832, "step": 5793 }, { "epoch": 0.19, "learning_rate": 1.8372273728891997e-05, "loss": 1.3578, "step": 5794 }, { "epoch": 0.19, "learning_rate": 1.8371695232403674e-05, "loss": 1.5548, "step": 5795 }, { "epoch": 0.19, "learning_rate": 1.837111664224541e-05, "loss": 1.4921, "step": 5796 }, { "epoch": 0.19, "learning_rate": 1.8370537958423673e-05, "loss": 1.4164, "step": 5797 }, { "epoch": 0.19, "learning_rate": 1.8369959180944947e-05, "loss": 1.2672, "step": 5798 }, { "epoch": 0.19, "learning_rate": 1.83693803098157e-05, "loss": 1.4878, "step": 5799 }, { "epoch": 0.19, "learning_rate": 1.8368801345042407e-05, "loss": 1.3965, "step": 5800 }, { "epoch": 0.19, "learning_rate": 1.8368222286631557e-05, "loss": 1.5596, "step": 5801 }, { "epoch": 0.19, "learning_rate": 1.836764313458962e-05, "loss": 1.5614, "step": 5802 }, { "epoch": 0.19, "learning_rate": 1.8367063888923078e-05, "loss": 1.4581, "step": 5803 }, { "epoch": 0.19, "learning_rate": 1.836648454963841e-05, "loss": 1.5995, "step": 5804 }, { "epoch": 0.19, "learning_rate": 1.8365905116742103e-05, "loss": 1.4923, "step": 5805 }, { "epoch": 0.19, "learning_rate": 1.836532559024064e-05, "loss": 1.5166, "step": 5806 }, { "epoch": 0.19, "learning_rate": 1.83647459701405e-05, "loss": 1.5777, "step": 5807 }, { "epoch": 0.19, "learning_rate": 1.8364166256448173e-05, "loss": 1.5457, "step": 5808 }, { "epoch": 0.19, "learning_rate": 1.8363586449170144e-05, "loss": 1.6073, "step": 5809 }, { "epoch": 0.19, "learning_rate": 1.8363006548312897e-05, "loss": 1.6246, "step": 5810 }, { "epoch": 0.19, "learning_rate": 1.8362426553882932e-05, "loss": 1.6644, "step": 5811 }, { "epoch": 0.19, "learning_rate": 1.8361846465886722e-05, "loss": 1.6263, "step": 5812 }, { "epoch": 0.19, "learning_rate": 1.836126628433077e-05, "loss": 1.7202, "step": 5813 }, { "epoch": 0.19, "learning_rate": 1.836068600922156e-05, "loss": 1.6219, "step": 5814 }, { "epoch": 0.19, "learning_rate": 1.8360105640565588e-05, "loss": 1.5599, "step": 5815 }, { "epoch": 0.19, "learning_rate": 1.835952517836935e-05, "loss": 1.597, "step": 5816 }, { "epoch": 0.19, "learning_rate": 1.835894462263934e-05, "loss": 1.5725, "step": 5817 }, { "epoch": 0.19, "learning_rate": 1.8358363973382046e-05, "loss": 1.5579, "step": 5818 }, { "epoch": 0.19, "learning_rate": 1.8357783230603977e-05, "loss": 1.5636, "step": 5819 }, { "epoch": 0.19, "learning_rate": 1.8357202394311624e-05, "loss": 1.4562, "step": 5820 }, { "epoch": 0.19, "learning_rate": 1.8356621464511483e-05, "loss": 1.5605, "step": 5821 }, { "epoch": 0.19, "learning_rate": 1.835604044121006e-05, "loss": 1.5239, "step": 5822 }, { "epoch": 0.19, "learning_rate": 1.8355459324413854e-05, "loss": 1.5143, "step": 5823 }, { "epoch": 0.19, "learning_rate": 1.8354878114129368e-05, "loss": 1.6306, "step": 5824 }, { "epoch": 0.19, "learning_rate": 1.83542968103631e-05, "loss": 1.5494, "step": 5825 }, { "epoch": 0.19, "learning_rate": 1.835371541312156e-05, "loss": 1.5898, "step": 5826 }, { "epoch": 0.19, "learning_rate": 1.8353133922411252e-05, "loss": 1.4912, "step": 5827 }, { "epoch": 0.19, "learning_rate": 1.835255233823868e-05, "loss": 1.4527, "step": 5828 }, { "epoch": 0.19, "learning_rate": 1.835197066061035e-05, "loss": 1.5164, "step": 5829 }, { "epoch": 0.19, "learning_rate": 1.835138888953278e-05, "loss": 1.527, "step": 5830 }, { "epoch": 0.19, "learning_rate": 1.8350807025012466e-05, "loss": 1.5564, "step": 5831 }, { "epoch": 0.19, "learning_rate": 1.8350225067055927e-05, "loss": 1.5824, "step": 5832 }, { "epoch": 0.19, "learning_rate": 1.8349643015669675e-05, "loss": 1.466, "step": 5833 }, { "epoch": 0.19, "learning_rate": 1.8349060870860215e-05, "loss": 1.5213, "step": 5834 }, { "epoch": 0.19, "learning_rate": 1.8348478632634067e-05, "loss": 1.5664, "step": 5835 }, { "epoch": 0.19, "learning_rate": 1.8347896300997745e-05, "loss": 1.5037, "step": 5836 }, { "epoch": 0.19, "learning_rate": 1.8347313875957762e-05, "loss": 1.5382, "step": 5837 }, { "epoch": 0.19, "learning_rate": 1.8346731357520637e-05, "loss": 1.431, "step": 5838 }, { "epoch": 0.19, "learning_rate": 1.8346148745692887e-05, "loss": 1.3964, "step": 5839 }, { "epoch": 0.19, "learning_rate": 1.834556604048103e-05, "loss": 1.361, "step": 5840 }, { "epoch": 0.19, "learning_rate": 1.8344983241891588e-05, "loss": 1.4242, "step": 5841 }, { "epoch": 0.19, "learning_rate": 1.834440034993108e-05, "loss": 1.5309, "step": 5842 }, { "epoch": 0.19, "learning_rate": 1.8343817364606025e-05, "loss": 1.3624, "step": 5843 }, { "epoch": 0.19, "learning_rate": 1.8343234285922955e-05, "loss": 1.4621, "step": 5844 }, { "epoch": 0.19, "learning_rate": 1.8342651113888387e-05, "loss": 1.3396, "step": 5845 }, { "epoch": 0.19, "learning_rate": 1.8342067848508843e-05, "loss": 1.3901, "step": 5846 }, { "epoch": 0.19, "learning_rate": 1.8341484489790856e-05, "loss": 1.2983, "step": 5847 }, { "epoch": 0.19, "learning_rate": 1.834090103774095e-05, "loss": 1.4183, "step": 5848 }, { "epoch": 0.19, "learning_rate": 1.8340317492365654e-05, "loss": 1.5059, "step": 5849 }, { "epoch": 0.2, "learning_rate": 1.8339733853671497e-05, "loss": 1.5095, "step": 5850 }, { "epoch": 0.2, "learning_rate": 1.833915012166501e-05, "loss": 1.5379, "step": 5851 }, { "epoch": 0.2, "learning_rate": 1.8338566296352723e-05, "loss": 1.4248, "step": 5852 }, { "epoch": 0.2, "learning_rate": 1.8337982377741167e-05, "loss": 1.3874, "step": 5853 }, { "epoch": 0.2, "learning_rate": 1.8337398365836884e-05, "loss": 1.5428, "step": 5854 }, { "epoch": 0.2, "learning_rate": 1.8336814260646395e-05, "loss": 1.4549, "step": 5855 }, { "epoch": 0.2, "learning_rate": 1.8336230062176245e-05, "loss": 1.4399, "step": 5856 }, { "epoch": 0.2, "learning_rate": 1.8335645770432963e-05, "loss": 1.5083, "step": 5857 }, { "epoch": 0.2, "learning_rate": 1.8335061385423095e-05, "loss": 1.5088, "step": 5858 }, { "epoch": 0.2, "learning_rate": 1.8334476907153177e-05, "loss": 1.6185, "step": 5859 }, { "epoch": 0.2, "learning_rate": 1.8333892335629747e-05, "loss": 1.5867, "step": 5860 }, { "epoch": 0.2, "learning_rate": 1.8333307670859346e-05, "loss": 1.5414, "step": 5861 }, { "epoch": 0.2, "learning_rate": 1.8332722912848513e-05, "loss": 1.5483, "step": 5862 }, { "epoch": 0.2, "learning_rate": 1.8332138061603802e-05, "loss": 1.5653, "step": 5863 }, { "epoch": 0.2, "learning_rate": 1.833155311713174e-05, "loss": 1.5634, "step": 5864 }, { "epoch": 0.2, "learning_rate": 1.8330968079438887e-05, "loss": 1.4638, "step": 5865 }, { "epoch": 0.2, "learning_rate": 1.8330382948531778e-05, "loss": 1.4197, "step": 5866 }, { "epoch": 0.2, "learning_rate": 1.8329797724416963e-05, "loss": 1.5845, "step": 5867 }, { "epoch": 0.2, "learning_rate": 1.8329212407100996e-05, "loss": 1.5535, "step": 5868 }, { "epoch": 0.2, "learning_rate": 1.832862699659042e-05, "loss": 1.5416, "step": 5869 }, { "epoch": 0.2, "learning_rate": 1.832804149289179e-05, "loss": 1.464, "step": 5870 }, { "epoch": 0.2, "learning_rate": 1.8327455896011645e-05, "loss": 1.6066, "step": 5871 }, { "epoch": 0.2, "learning_rate": 1.8326870205956553e-05, "loss": 1.5967, "step": 5872 }, { "epoch": 0.2, "learning_rate": 1.8326284422733057e-05, "loss": 1.5812, "step": 5873 }, { "epoch": 0.2, "learning_rate": 1.8325698546347714e-05, "loss": 1.5344, "step": 5874 }, { "epoch": 0.2, "learning_rate": 1.8325112576807083e-05, "loss": 1.4323, "step": 5875 }, { "epoch": 0.2, "learning_rate": 1.8324526514117715e-05, "loss": 1.3768, "step": 5876 }, { "epoch": 0.2, "learning_rate": 1.832394035828617e-05, "loss": 1.5082, "step": 5877 }, { "epoch": 0.2, "learning_rate": 1.8323354109319003e-05, "loss": 1.5721, "step": 5878 }, { "epoch": 0.2, "learning_rate": 1.832276776722278e-05, "loss": 1.6055, "step": 5879 }, { "epoch": 0.2, "learning_rate": 1.8322181332004057e-05, "loss": 1.5653, "step": 5880 }, { "epoch": 0.2, "learning_rate": 1.8321594803669395e-05, "loss": 1.4367, "step": 5881 }, { "epoch": 0.2, "learning_rate": 1.8321008182225357e-05, "loss": 1.6045, "step": 5882 }, { "epoch": 0.2, "learning_rate": 1.8320421467678507e-05, "loss": 1.5986, "step": 5883 }, { "epoch": 0.2, "learning_rate": 1.8319834660035415e-05, "loss": 1.5328, "step": 5884 }, { "epoch": 0.2, "learning_rate": 1.8319247759302642e-05, "loss": 1.5311, "step": 5885 }, { "epoch": 0.2, "learning_rate": 1.831866076548675e-05, "loss": 1.5333, "step": 5886 }, { "epoch": 0.2, "learning_rate": 1.8318073678594316e-05, "loss": 1.6283, "step": 5887 }, { "epoch": 0.2, "learning_rate": 1.83174864986319e-05, "loss": 1.5594, "step": 5888 }, { "epoch": 0.2, "learning_rate": 1.8316899225606078e-05, "loss": 1.4741, "step": 5889 }, { "epoch": 0.2, "learning_rate": 1.831631185952342e-05, "loss": 1.6704, "step": 5890 }, { "epoch": 0.2, "learning_rate": 1.83157244003905e-05, "loss": 1.5379, "step": 5891 }, { "epoch": 0.2, "learning_rate": 1.8315136848213883e-05, "loss": 1.6268, "step": 5892 }, { "epoch": 0.2, "learning_rate": 1.831454920300015e-05, "loss": 1.6215, "step": 5893 }, { "epoch": 0.2, "learning_rate": 1.8313961464755876e-05, "loss": 1.6078, "step": 5894 }, { "epoch": 0.2, "learning_rate": 1.8313373633487633e-05, "loss": 1.746, "step": 5895 }, { "epoch": 0.2, "learning_rate": 1.8312785709202002e-05, "loss": 1.7354, "step": 5896 }, { "epoch": 0.2, "learning_rate": 1.8312197691905564e-05, "loss": 1.683, "step": 5897 }, { "epoch": 0.2, "learning_rate": 1.8311609581604887e-05, "loss": 1.6031, "step": 5898 }, { "epoch": 0.2, "learning_rate": 1.8311021378306563e-05, "loss": 1.6391, "step": 5899 }, { "epoch": 0.2, "learning_rate": 1.8310433082017167e-05, "loss": 1.5961, "step": 5900 }, { "epoch": 0.2, "learning_rate": 1.8309844692743283e-05, "loss": 1.5445, "step": 5901 }, { "epoch": 0.2, "learning_rate": 1.8309256210491496e-05, "loss": 1.5943, "step": 5902 }, { "epoch": 0.2, "learning_rate": 1.830866763526839e-05, "loss": 1.6509, "step": 5903 }, { "epoch": 0.2, "learning_rate": 1.8308078967080547e-05, "loss": 1.5768, "step": 5904 }, { "epoch": 0.2, "learning_rate": 1.8307490205934556e-05, "loss": 1.5672, "step": 5905 }, { "epoch": 0.2, "learning_rate": 1.8306901351837007e-05, "loss": 1.5473, "step": 5906 }, { "epoch": 0.2, "learning_rate": 1.8306312404794485e-05, "loss": 1.4668, "step": 5907 }, { "epoch": 0.2, "learning_rate": 1.830572336481358e-05, "loss": 1.4959, "step": 5908 }, { "epoch": 0.2, "learning_rate": 1.830513423190089e-05, "loss": 1.4726, "step": 5909 }, { "epoch": 0.2, "learning_rate": 1.830454500606299e-05, "loss": 1.5189, "step": 5910 }, { "epoch": 0.2, "learning_rate": 1.830395568730649e-05, "loss": 1.5417, "step": 5911 }, { "epoch": 0.2, "learning_rate": 1.8303366275637977e-05, "loss": 1.5423, "step": 5912 }, { "epoch": 0.2, "learning_rate": 1.8302776771064044e-05, "loss": 1.5805, "step": 5913 }, { "epoch": 0.2, "learning_rate": 1.830218717359129e-05, "loss": 1.57, "step": 5914 }, { "epoch": 0.2, "learning_rate": 1.830159748322631e-05, "loss": 1.5511, "step": 5915 }, { "epoch": 0.2, "learning_rate": 1.8301007699975704e-05, "loss": 1.5605, "step": 5916 }, { "epoch": 0.2, "learning_rate": 1.8300417823846066e-05, "loss": 1.58, "step": 5917 }, { "epoch": 0.2, "learning_rate": 1.8299827854844005e-05, "loss": 1.4332, "step": 5918 }, { "epoch": 0.2, "learning_rate": 1.829923779297611e-05, "loss": 1.5424, "step": 5919 }, { "epoch": 0.2, "learning_rate": 1.8298647638248995e-05, "loss": 1.4662, "step": 5920 }, { "epoch": 0.2, "learning_rate": 1.829805739066926e-05, "loss": 1.5791, "step": 5921 }, { "epoch": 0.2, "learning_rate": 1.8297467050243503e-05, "loss": 1.507, "step": 5922 }, { "epoch": 0.2, "learning_rate": 1.8296876616978337e-05, "loss": 1.4311, "step": 5923 }, { "epoch": 0.2, "learning_rate": 1.8296286090880362e-05, "loss": 1.5514, "step": 5924 }, { "epoch": 0.2, "learning_rate": 1.829569547195619e-05, "loss": 1.527, "step": 5925 }, { "epoch": 0.2, "learning_rate": 1.8295104760212424e-05, "loss": 1.5553, "step": 5926 }, { "epoch": 0.2, "learning_rate": 1.8294513955655685e-05, "loss": 1.4147, "step": 5927 }, { "epoch": 0.2, "learning_rate": 1.829392305829257e-05, "loss": 1.502, "step": 5928 }, { "epoch": 0.2, "learning_rate": 1.8293332068129697e-05, "loss": 1.4882, "step": 5929 }, { "epoch": 0.2, "learning_rate": 1.8292740985173675e-05, "loss": 1.5582, "step": 5930 }, { "epoch": 0.2, "learning_rate": 1.8292149809431123e-05, "loss": 1.5714, "step": 5931 }, { "epoch": 0.2, "learning_rate": 1.829155854090865e-05, "loss": 1.5067, "step": 5932 }, { "epoch": 0.2, "learning_rate": 1.8290967179612877e-05, "loss": 1.4733, "step": 5933 }, { "epoch": 0.2, "learning_rate": 1.8290375725550417e-05, "loss": 1.3343, "step": 5934 }, { "epoch": 0.2, "learning_rate": 1.828978417872789e-05, "loss": 1.431, "step": 5935 }, { "epoch": 0.2, "learning_rate": 1.8289192539151913e-05, "loss": 1.5029, "step": 5936 }, { "epoch": 0.2, "learning_rate": 1.8288600806829104e-05, "loss": 1.5074, "step": 5937 }, { "epoch": 0.2, "learning_rate": 1.828800898176609e-05, "loss": 1.5307, "step": 5938 }, { "epoch": 0.2, "learning_rate": 1.8287417063969487e-05, "loss": 1.3963, "step": 5939 }, { "epoch": 0.2, "learning_rate": 1.8286825053445916e-05, "loss": 1.4531, "step": 5940 }, { "epoch": 0.2, "learning_rate": 1.828623295020201e-05, "loss": 1.424, "step": 5941 }, { "epoch": 0.2, "learning_rate": 1.8285640754244384e-05, "loss": 1.4435, "step": 5942 }, { "epoch": 0.2, "learning_rate": 1.8285048465579672e-05, "loss": 1.4843, "step": 5943 }, { "epoch": 0.2, "learning_rate": 1.8284456084214496e-05, "loss": 1.5725, "step": 5944 }, { "epoch": 0.2, "learning_rate": 1.828386361015549e-05, "loss": 1.5169, "step": 5945 }, { "epoch": 0.2, "learning_rate": 1.8283271043409275e-05, "loss": 1.519, "step": 5946 }, { "epoch": 0.2, "learning_rate": 1.8282678383982485e-05, "loss": 1.4898, "step": 5947 }, { "epoch": 0.2, "learning_rate": 1.8282085631881755e-05, "loss": 1.4577, "step": 5948 }, { "epoch": 0.2, "learning_rate": 1.8281492787113707e-05, "loss": 1.6505, "step": 5949 }, { "epoch": 0.2, "learning_rate": 1.8280899849684987e-05, "loss": 1.4664, "step": 5950 }, { "epoch": 0.2, "learning_rate": 1.8280306819602224e-05, "loss": 1.3863, "step": 5951 }, { "epoch": 0.2, "learning_rate": 1.8279713696872047e-05, "loss": 1.5763, "step": 5952 }, { "epoch": 0.2, "learning_rate": 1.8279120481501104e-05, "loss": 1.5918, "step": 5953 }, { "epoch": 0.2, "learning_rate": 1.8278527173496022e-05, "loss": 1.5208, "step": 5954 }, { "epoch": 0.2, "learning_rate": 1.8277933772863443e-05, "loss": 1.3418, "step": 5955 }, { "epoch": 0.2, "learning_rate": 1.827734027961001e-05, "loss": 1.5123, "step": 5956 }, { "epoch": 0.2, "learning_rate": 1.827674669374236e-05, "loss": 1.6052, "step": 5957 }, { "epoch": 0.2, "learning_rate": 1.8276153015267133e-05, "loss": 1.4713, "step": 5958 }, { "epoch": 0.2, "learning_rate": 1.8275559244190975e-05, "loss": 1.5712, "step": 5959 }, { "epoch": 0.2, "learning_rate": 1.827496538052053e-05, "loss": 1.4691, "step": 5960 }, { "epoch": 0.2, "learning_rate": 1.8274371424262442e-05, "loss": 1.6768, "step": 5961 }, { "epoch": 0.2, "learning_rate": 1.8273777375423354e-05, "loss": 1.5254, "step": 5962 }, { "epoch": 0.2, "learning_rate": 1.8273183234009914e-05, "loss": 1.549, "step": 5963 }, { "epoch": 0.2, "learning_rate": 1.8272589000028774e-05, "loss": 1.4436, "step": 5964 }, { "epoch": 0.2, "learning_rate": 1.8271994673486577e-05, "loss": 1.5119, "step": 5965 }, { "epoch": 0.2, "learning_rate": 1.8271400254389974e-05, "loss": 1.6053, "step": 5966 }, { "epoch": 0.2, "learning_rate": 1.827080574274562e-05, "loss": 1.4119, "step": 5967 }, { "epoch": 0.2, "learning_rate": 1.827021113856016e-05, "loss": 1.5015, "step": 5968 }, { "epoch": 0.2, "learning_rate": 1.8269616441840255e-05, "loss": 1.5867, "step": 5969 }, { "epoch": 0.2, "learning_rate": 1.8269021652592555e-05, "loss": 1.4278, "step": 5970 }, { "epoch": 0.2, "learning_rate": 1.8268426770823713e-05, "loss": 1.5525, "step": 5971 }, { "epoch": 0.2, "learning_rate": 1.8267831796540388e-05, "loss": 1.5386, "step": 5972 }, { "epoch": 0.2, "learning_rate": 1.8267236729749234e-05, "loss": 1.6217, "step": 5973 }, { "epoch": 0.2, "learning_rate": 1.8266641570456915e-05, "loss": 1.5382, "step": 5974 }, { "epoch": 0.2, "learning_rate": 1.826604631867008e-05, "loss": 1.4921, "step": 5975 }, { "epoch": 0.2, "learning_rate": 1.8265450974395403e-05, "loss": 1.5373, "step": 5976 }, { "epoch": 0.2, "learning_rate": 1.8264855537639535e-05, "loss": 1.4163, "step": 5977 }, { "epoch": 0.2, "learning_rate": 1.8264260008409138e-05, "loss": 1.5653, "step": 5978 }, { "epoch": 0.2, "learning_rate": 1.826366438671088e-05, "loss": 1.6073, "step": 5979 }, { "epoch": 0.2, "learning_rate": 1.8263068672551428e-05, "loss": 1.4256, "step": 5980 }, { "epoch": 0.2, "learning_rate": 1.826247286593744e-05, "loss": 1.5184, "step": 5981 }, { "epoch": 0.2, "learning_rate": 1.8261876966875583e-05, "loss": 1.6035, "step": 5982 }, { "epoch": 0.2, "learning_rate": 1.8261280975372533e-05, "loss": 1.38, "step": 5983 }, { "epoch": 0.2, "learning_rate": 1.826068489143495e-05, "loss": 1.4821, "step": 5984 }, { "epoch": 0.2, "learning_rate": 1.8260088715069506e-05, "loss": 1.4848, "step": 5985 }, { "epoch": 0.2, "learning_rate": 1.8259492446282868e-05, "loss": 1.5624, "step": 5986 }, { "epoch": 0.2, "learning_rate": 1.8258896085081717e-05, "loss": 1.5428, "step": 5987 }, { "epoch": 0.2, "learning_rate": 1.8258299631472717e-05, "loss": 1.5285, "step": 5988 }, { "epoch": 0.2, "learning_rate": 1.8257703085462542e-05, "loss": 1.4715, "step": 5989 }, { "epoch": 0.2, "learning_rate": 1.8257106447057872e-05, "loss": 1.4886, "step": 5990 }, { "epoch": 0.2, "learning_rate": 1.825650971626538e-05, "loss": 1.5144, "step": 5991 }, { "epoch": 0.2, "learning_rate": 1.825591289309174e-05, "loss": 1.5934, "step": 5992 }, { "epoch": 0.2, "learning_rate": 1.8255315977543637e-05, "loss": 1.5483, "step": 5993 }, { "epoch": 0.2, "learning_rate": 1.825471896962774e-05, "loss": 1.6475, "step": 5994 }, { "epoch": 0.2, "learning_rate": 1.8254121869350737e-05, "loss": 1.5455, "step": 5995 }, { "epoch": 0.2, "learning_rate": 1.8253524676719307e-05, "loss": 1.6081, "step": 5996 }, { "epoch": 0.2, "learning_rate": 1.825292739174013e-05, "loss": 1.5229, "step": 5997 }, { "epoch": 0.2, "learning_rate": 1.825233001441989e-05, "loss": 1.5502, "step": 5998 }, { "epoch": 0.2, "learning_rate": 1.8251732544765267e-05, "loss": 1.5119, "step": 5999 }, { "epoch": 0.2, "learning_rate": 1.8251134982782952e-05, "loss": 1.4215, "step": 6000 }, { "epoch": 0.2, "learning_rate": 1.825053732847963e-05, "loss": 1.6029, "step": 6001 }, { "epoch": 0.2, "learning_rate": 1.8249939581861985e-05, "loss": 1.5258, "step": 6002 }, { "epoch": 0.2, "learning_rate": 1.824934174293671e-05, "loss": 1.4277, "step": 6003 }, { "epoch": 0.2, "learning_rate": 1.824874381171049e-05, "loss": 1.4301, "step": 6004 }, { "epoch": 0.2, "learning_rate": 1.8248145788190018e-05, "loss": 1.4543, "step": 6005 }, { "epoch": 0.2, "learning_rate": 1.824754767238198e-05, "loss": 1.4098, "step": 6006 }, { "epoch": 0.2, "learning_rate": 1.8246949464293077e-05, "loss": 1.4412, "step": 6007 }, { "epoch": 0.2, "learning_rate": 1.824635116392999e-05, "loss": 1.4579, "step": 6008 }, { "epoch": 0.2, "learning_rate": 1.8245752771299426e-05, "loss": 1.5312, "step": 6009 }, { "epoch": 0.2, "learning_rate": 1.8245154286408074e-05, "loss": 1.5222, "step": 6010 }, { "epoch": 0.2, "learning_rate": 1.8244555709262627e-05, "loss": 1.5644, "step": 6011 }, { "epoch": 0.2, "learning_rate": 1.824395703986979e-05, "loss": 1.5347, "step": 6012 }, { "epoch": 0.2, "learning_rate": 1.824335827823626e-05, "loss": 1.5371, "step": 6013 }, { "epoch": 0.2, "learning_rate": 1.8242759424368733e-05, "loss": 1.5039, "step": 6014 }, { "epoch": 0.2, "learning_rate": 1.8242160478273908e-05, "loss": 1.4084, "step": 6015 }, { "epoch": 0.2, "learning_rate": 1.8241561439958496e-05, "loss": 1.4434, "step": 6016 }, { "epoch": 0.2, "learning_rate": 1.8240962309429186e-05, "loss": 1.3889, "step": 6017 }, { "epoch": 0.2, "learning_rate": 1.8240363086692695e-05, "loss": 1.42, "step": 6018 }, { "epoch": 0.2, "learning_rate": 1.8239763771755716e-05, "loss": 1.3648, "step": 6019 }, { "epoch": 0.2, "learning_rate": 1.8239164364624964e-05, "loss": 1.4338, "step": 6020 }, { "epoch": 0.2, "learning_rate": 1.823856486530714e-05, "loss": 1.496, "step": 6021 }, { "epoch": 0.2, "learning_rate": 1.823796527380895e-05, "loss": 1.559, "step": 6022 }, { "epoch": 0.2, "learning_rate": 1.8237365590137114e-05, "loss": 1.5359, "step": 6023 }, { "epoch": 0.2, "learning_rate": 1.8236765814298328e-05, "loss": 1.4467, "step": 6024 }, { "epoch": 0.2, "learning_rate": 1.823616594629931e-05, "loss": 1.5, "step": 6025 }, { "epoch": 0.2, "learning_rate": 1.8235565986146775e-05, "loss": 1.5719, "step": 6026 }, { "epoch": 0.2, "learning_rate": 1.823496593384743e-05, "loss": 1.4974, "step": 6027 }, { "epoch": 0.2, "learning_rate": 1.8234365789407985e-05, "loss": 1.5419, "step": 6028 }, { "epoch": 0.2, "learning_rate": 1.8233765552835165e-05, "loss": 1.5124, "step": 6029 }, { "epoch": 0.2, "learning_rate": 1.823316522413568e-05, "loss": 1.478, "step": 6030 }, { "epoch": 0.2, "learning_rate": 1.823256480331625e-05, "loss": 1.5744, "step": 6031 }, { "epoch": 0.2, "learning_rate": 1.823196429038359e-05, "loss": 1.493, "step": 6032 }, { "epoch": 0.2, "learning_rate": 1.8231363685344422e-05, "loss": 1.5449, "step": 6033 }, { "epoch": 0.2, "learning_rate": 1.8230762988205463e-05, "loss": 1.4471, "step": 6034 }, { "epoch": 0.2, "learning_rate": 1.8230162198973436e-05, "loss": 1.4307, "step": 6035 }, { "epoch": 0.2, "learning_rate": 1.822956131765506e-05, "loss": 1.5059, "step": 6036 }, { "epoch": 0.2, "learning_rate": 1.8228960344257068e-05, "loss": 1.5105, "step": 6037 }, { "epoch": 0.2, "learning_rate": 1.822835927878617e-05, "loss": 1.4708, "step": 6038 }, { "epoch": 0.2, "learning_rate": 1.8227758121249108e-05, "loss": 1.4539, "step": 6039 }, { "epoch": 0.2, "learning_rate": 1.822715687165259e-05, "loss": 1.3537, "step": 6040 }, { "epoch": 0.2, "learning_rate": 1.8226555530003355e-05, "loss": 1.5518, "step": 6041 }, { "epoch": 0.2, "learning_rate": 1.8225954096308133e-05, "loss": 1.4151, "step": 6042 }, { "epoch": 0.2, "learning_rate": 1.8225352570573643e-05, "loss": 1.4556, "step": 6043 }, { "epoch": 0.2, "learning_rate": 1.8224750952806626e-05, "loss": 1.4875, "step": 6044 }, { "epoch": 0.2, "learning_rate": 1.8224149243013805e-05, "loss": 1.475, "step": 6045 }, { "epoch": 0.2, "learning_rate": 1.8223547441201916e-05, "loss": 1.4908, "step": 6046 }, { "epoch": 0.2, "learning_rate": 1.8222945547377696e-05, "loss": 1.525, "step": 6047 }, { "epoch": 0.2, "learning_rate": 1.8222343561547876e-05, "loss": 1.5724, "step": 6048 }, { "epoch": 0.2, "learning_rate": 1.822174148371919e-05, "loss": 1.523, "step": 6049 }, { "epoch": 0.2, "learning_rate": 1.8221139313898373e-05, "loss": 1.5794, "step": 6050 }, { "epoch": 0.2, "learning_rate": 1.822053705209217e-05, "loss": 1.5903, "step": 6051 }, { "epoch": 0.2, "learning_rate": 1.821993469830732e-05, "loss": 1.5885, "step": 6052 }, { "epoch": 0.2, "learning_rate": 1.821933225255055e-05, "loss": 1.4654, "step": 6053 }, { "epoch": 0.2, "learning_rate": 1.8218729714828612e-05, "loss": 1.4793, "step": 6054 }, { "epoch": 0.2, "learning_rate": 1.8218127085148246e-05, "loss": 1.4921, "step": 6055 }, { "epoch": 0.2, "learning_rate": 1.8217524363516192e-05, "loss": 1.5243, "step": 6056 }, { "epoch": 0.2, "learning_rate": 1.8216921549939196e-05, "loss": 1.5151, "step": 6057 }, { "epoch": 0.2, "learning_rate": 1.8216318644424004e-05, "loss": 1.5269, "step": 6058 }, { "epoch": 0.2, "learning_rate": 1.8215715646977355e-05, "loss": 1.5926, "step": 6059 }, { "epoch": 0.2, "learning_rate": 1.8215112557606005e-05, "loss": 1.5303, "step": 6060 }, { "epoch": 0.2, "learning_rate": 1.82145093763167e-05, "loss": 1.5741, "step": 6061 }, { "epoch": 0.2, "learning_rate": 1.821390610311618e-05, "loss": 1.6668, "step": 6062 }, { "epoch": 0.2, "learning_rate": 1.8213302738011204e-05, "loss": 1.6109, "step": 6063 }, { "epoch": 0.2, "learning_rate": 1.821269928100852e-05, "loss": 1.5543, "step": 6064 }, { "epoch": 0.2, "learning_rate": 1.8212095732114884e-05, "loss": 1.7143, "step": 6065 }, { "epoch": 0.2, "learning_rate": 1.821149209133704e-05, "loss": 1.6673, "step": 6066 }, { "epoch": 0.2, "learning_rate": 1.8210888358681752e-05, "loss": 1.811, "step": 6067 }, { "epoch": 0.2, "learning_rate": 1.8210284534155772e-05, "loss": 1.6709, "step": 6068 }, { "epoch": 0.2, "learning_rate": 1.820968061776585e-05, "loss": 1.6124, "step": 6069 }, { "epoch": 0.2, "learning_rate": 1.8209076609518752e-05, "loss": 1.6777, "step": 6070 }, { "epoch": 0.2, "learning_rate": 1.8208472509421227e-05, "loss": 1.6892, "step": 6071 }, { "epoch": 0.2, "learning_rate": 1.8207868317480046e-05, "loss": 1.3974, "step": 6072 }, { "epoch": 0.2, "learning_rate": 1.8207264033701963e-05, "loss": 1.5145, "step": 6073 }, { "epoch": 0.2, "learning_rate": 1.820665965809373e-05, "loss": 1.6479, "step": 6074 }, { "epoch": 0.2, "learning_rate": 1.8206055190662128e-05, "loss": 1.5722, "step": 6075 }, { "epoch": 0.2, "learning_rate": 1.8205450631413904e-05, "loss": 1.6045, "step": 6076 }, { "epoch": 0.2, "learning_rate": 1.8204845980355834e-05, "loss": 1.4768, "step": 6077 }, { "epoch": 0.2, "learning_rate": 1.8204241237494672e-05, "loss": 1.6244, "step": 6078 }, { "epoch": 0.2, "learning_rate": 1.8203636402837196e-05, "loss": 1.4858, "step": 6079 }, { "epoch": 0.2, "learning_rate": 1.8203031476390167e-05, "loss": 1.6691, "step": 6080 }, { "epoch": 0.2, "learning_rate": 1.8202426458160353e-05, "loss": 1.527, "step": 6081 }, { "epoch": 0.2, "learning_rate": 1.8201821348154526e-05, "loss": 1.5369, "step": 6082 }, { "epoch": 0.2, "learning_rate": 1.8201216146379455e-05, "loss": 1.6425, "step": 6083 }, { "epoch": 0.2, "learning_rate": 1.8200610852841913e-05, "loss": 1.5853, "step": 6084 }, { "epoch": 0.2, "learning_rate": 1.820000546754867e-05, "loss": 1.6618, "step": 6085 }, { "epoch": 0.2, "learning_rate": 1.81993999905065e-05, "loss": 1.5363, "step": 6086 }, { "epoch": 0.2, "learning_rate": 1.8198794421722183e-05, "loss": 1.579, "step": 6087 }, { "epoch": 0.2, "learning_rate": 1.8198188761202487e-05, "loss": 1.666, "step": 6088 }, { "epoch": 0.2, "learning_rate": 1.8197583008954196e-05, "loss": 1.5904, "step": 6089 }, { "epoch": 0.2, "learning_rate": 1.819697716498408e-05, "loss": 1.5802, "step": 6090 }, { "epoch": 0.2, "learning_rate": 1.8196371229298925e-05, "loss": 1.8104, "step": 6091 }, { "epoch": 0.2, "learning_rate": 1.8195765201905503e-05, "loss": 1.5911, "step": 6092 }, { "epoch": 0.2, "learning_rate": 1.81951590828106e-05, "loss": 1.7316, "step": 6093 }, { "epoch": 0.2, "learning_rate": 1.8194552872020997e-05, "loss": 1.6075, "step": 6094 }, { "epoch": 0.2, "learning_rate": 1.8193946569543478e-05, "loss": 1.6548, "step": 6095 }, { "epoch": 0.2, "learning_rate": 1.8193340175384826e-05, "loss": 1.507, "step": 6096 }, { "epoch": 0.2, "learning_rate": 1.8192733689551826e-05, "loss": 1.5508, "step": 6097 }, { "epoch": 0.2, "learning_rate": 1.819212711205126e-05, "loss": 1.5395, "step": 6098 }, { "epoch": 0.2, "learning_rate": 1.819152044288992e-05, "loss": 1.4862, "step": 6099 }, { "epoch": 0.2, "learning_rate": 1.819091368207459e-05, "loss": 1.4825, "step": 6100 }, { "epoch": 0.2, "learning_rate": 1.8190306829612065e-05, "loss": 1.5878, "step": 6101 }, { "epoch": 0.2, "learning_rate": 1.8189699885509128e-05, "loss": 1.4996, "step": 6102 }, { "epoch": 0.2, "learning_rate": 1.8189092849772577e-05, "loss": 1.6462, "step": 6103 }, { "epoch": 0.2, "learning_rate": 1.8188485722409196e-05, "loss": 1.5595, "step": 6104 }, { "epoch": 0.2, "learning_rate": 1.818787850342578e-05, "loss": 1.5012, "step": 6105 }, { "epoch": 0.2, "learning_rate": 1.8187271192829136e-05, "loss": 1.6043, "step": 6106 }, { "epoch": 0.2, "learning_rate": 1.818666379062604e-05, "loss": 1.5525, "step": 6107 }, { "epoch": 0.2, "learning_rate": 1.8186056296823297e-05, "loss": 1.44, "step": 6108 }, { "epoch": 0.2, "learning_rate": 1.8185448711427707e-05, "loss": 1.556, "step": 6109 }, { "epoch": 0.2, "learning_rate": 1.818484103444606e-05, "loss": 1.4823, "step": 6110 }, { "epoch": 0.2, "learning_rate": 1.8184233265885165e-05, "loss": 1.5228, "step": 6111 }, { "epoch": 0.2, "learning_rate": 1.8183625405751816e-05, "loss": 1.6345, "step": 6112 }, { "epoch": 0.2, "learning_rate": 1.8183017454052816e-05, "loss": 1.5417, "step": 6113 }, { "epoch": 0.2, "learning_rate": 1.818240941079497e-05, "loss": 1.5616, "step": 6114 }, { "epoch": 0.2, "learning_rate": 1.8181801275985075e-05, "loss": 1.5331, "step": 6115 }, { "epoch": 0.2, "learning_rate": 1.818119304962994e-05, "loss": 1.5643, "step": 6116 }, { "epoch": 0.2, "learning_rate": 1.8180584731736367e-05, "loss": 1.6211, "step": 6117 }, { "epoch": 0.2, "learning_rate": 1.8179976322311164e-05, "loss": 1.5445, "step": 6118 }, { "epoch": 0.2, "learning_rate": 1.817936782136114e-05, "loss": 1.5686, "step": 6119 }, { "epoch": 0.2, "learning_rate": 1.8178759228893108e-05, "loss": 1.4697, "step": 6120 }, { "epoch": 0.2, "learning_rate": 1.8178150544913867e-05, "loss": 1.6091, "step": 6121 }, { "epoch": 0.2, "learning_rate": 1.8177541769430234e-05, "loss": 1.5336, "step": 6122 }, { "epoch": 0.2, "learning_rate": 1.8176932902449017e-05, "loss": 1.5009, "step": 6123 }, { "epoch": 0.2, "learning_rate": 1.8176323943977034e-05, "loss": 1.5698, "step": 6124 }, { "epoch": 0.2, "learning_rate": 1.8175714894021088e-05, "loss": 1.6133, "step": 6125 }, { "epoch": 0.2, "learning_rate": 1.8175105752588006e-05, "loss": 1.431, "step": 6126 }, { "epoch": 0.2, "learning_rate": 1.8174496519684596e-05, "loss": 1.6324, "step": 6127 }, { "epoch": 0.2, "learning_rate": 1.817388719531768e-05, "loss": 1.6934, "step": 6128 }, { "epoch": 0.2, "learning_rate": 1.817327777949407e-05, "loss": 1.5257, "step": 6129 }, { "epoch": 0.2, "learning_rate": 1.817266827222059e-05, "loss": 1.6474, "step": 6130 }, { "epoch": 0.2, "learning_rate": 1.8172058673504055e-05, "loss": 1.5248, "step": 6131 }, { "epoch": 0.2, "learning_rate": 1.8171448983351284e-05, "loss": 1.4862, "step": 6132 }, { "epoch": 0.2, "learning_rate": 1.817083920176911e-05, "loss": 1.5624, "step": 6133 }, { "epoch": 0.2, "learning_rate": 1.8170229328764343e-05, "loss": 1.6251, "step": 6134 }, { "epoch": 0.2, "learning_rate": 1.8169619364343814e-05, "loss": 1.7139, "step": 6135 }, { "epoch": 0.2, "learning_rate": 1.8169009308514346e-05, "loss": 1.655, "step": 6136 }, { "epoch": 0.2, "learning_rate": 1.8168399161282765e-05, "loss": 1.6164, "step": 6137 }, { "epoch": 0.2, "learning_rate": 1.8167788922655893e-05, "loss": 1.58, "step": 6138 }, { "epoch": 0.2, "learning_rate": 1.8167178592640565e-05, "loss": 1.6601, "step": 6139 }, { "epoch": 0.2, "learning_rate": 1.816656817124361e-05, "loss": 1.5854, "step": 6140 }, { "epoch": 0.2, "learning_rate": 1.8165957658471854e-05, "loss": 1.5414, "step": 6141 }, { "epoch": 0.2, "learning_rate": 1.8165347054332128e-05, "loss": 1.5405, "step": 6142 }, { "epoch": 0.2, "learning_rate": 1.8164736358831265e-05, "loss": 1.664, "step": 6143 }, { "epoch": 0.2, "learning_rate": 1.81641255719761e-05, "loss": 1.5628, "step": 6144 }, { "epoch": 0.2, "learning_rate": 1.8163514693773464e-05, "loss": 1.5383, "step": 6145 }, { "epoch": 0.2, "learning_rate": 1.8162903724230192e-05, "loss": 1.5947, "step": 6146 }, { "epoch": 0.2, "learning_rate": 1.8162292663353126e-05, "loss": 1.651, "step": 6147 }, { "epoch": 0.2, "learning_rate": 1.8161681511149094e-05, "loss": 1.6559, "step": 6148 }, { "epoch": 0.2, "learning_rate": 1.8161070267624938e-05, "loss": 1.6171, "step": 6149 }, { "epoch": 0.2, "learning_rate": 1.81604589327875e-05, "loss": 1.6229, "step": 6150 }, { "epoch": 0.21, "learning_rate": 1.8159847506643617e-05, "loss": 1.4442, "step": 6151 }, { "epoch": 0.21, "learning_rate": 1.8159235989200132e-05, "loss": 1.6528, "step": 6152 }, { "epoch": 0.21, "learning_rate": 1.8158624380463888e-05, "loss": 1.7479, "step": 6153 }, { "epoch": 0.21, "learning_rate": 1.8158012680441723e-05, "loss": 1.6152, "step": 6154 }, { "epoch": 0.21, "learning_rate": 1.8157400889140486e-05, "loss": 1.5396, "step": 6155 }, { "epoch": 0.21, "learning_rate": 1.8156789006567018e-05, "loss": 1.5208, "step": 6156 }, { "epoch": 0.21, "learning_rate": 1.8156177032728172e-05, "loss": 1.6098, "step": 6157 }, { "epoch": 0.21, "learning_rate": 1.8155564967630792e-05, "loss": 1.6147, "step": 6158 }, { "epoch": 0.21, "learning_rate": 1.8154952811281723e-05, "loss": 1.6394, "step": 6159 }, { "epoch": 0.21, "learning_rate": 1.815434056368782e-05, "loss": 1.7069, "step": 6160 }, { "epoch": 0.21, "learning_rate": 1.815372822485593e-05, "loss": 1.6938, "step": 6161 }, { "epoch": 0.21, "learning_rate": 1.8153115794792904e-05, "loss": 1.7345, "step": 6162 }, { "epoch": 0.21, "learning_rate": 1.8152503273505595e-05, "loss": 1.5543, "step": 6163 }, { "epoch": 0.21, "learning_rate": 1.8151890661000857e-05, "loss": 1.5127, "step": 6164 }, { "epoch": 0.21, "learning_rate": 1.815127795728554e-05, "loss": 1.5919, "step": 6165 }, { "epoch": 0.21, "learning_rate": 1.8150665162366512e-05, "loss": 1.5543, "step": 6166 }, { "epoch": 0.21, "learning_rate": 1.8150052276250617e-05, "loss": 1.381, "step": 6167 }, { "epoch": 0.21, "learning_rate": 1.8149439298944716e-05, "loss": 1.6432, "step": 6168 }, { "epoch": 0.21, "learning_rate": 1.8148826230455672e-05, "loss": 1.5667, "step": 6169 }, { "epoch": 0.21, "learning_rate": 1.8148213070790337e-05, "loss": 1.5506, "step": 6170 }, { "epoch": 0.21, "learning_rate": 1.814759981995558e-05, "loss": 1.6721, "step": 6171 }, { "epoch": 0.21, "learning_rate": 1.8146986477958252e-05, "loss": 1.6049, "step": 6172 }, { "epoch": 0.21, "learning_rate": 1.8146373044805226e-05, "loss": 1.677, "step": 6173 }, { "epoch": 0.21, "learning_rate": 1.814575952050336e-05, "loss": 1.5185, "step": 6174 }, { "epoch": 0.21, "learning_rate": 1.814514590505952e-05, "loss": 1.6109, "step": 6175 }, { "epoch": 0.21, "learning_rate": 1.814453219848057e-05, "loss": 1.5301, "step": 6176 }, { "epoch": 0.21, "learning_rate": 1.8143918400773378e-05, "loss": 1.5003, "step": 6177 }, { "epoch": 0.21, "learning_rate": 1.8143304511944813e-05, "loss": 1.519, "step": 6178 }, { "epoch": 0.21, "learning_rate": 1.8142690532001744e-05, "loss": 1.4265, "step": 6179 }, { "epoch": 0.21, "learning_rate": 1.814207646095104e-05, "loss": 1.4157, "step": 6180 }, { "epoch": 0.21, "learning_rate": 1.8141462298799568e-05, "loss": 1.4115, "step": 6181 }, { "epoch": 0.21, "learning_rate": 1.8140848045554203e-05, "loss": 1.5674, "step": 6182 }, { "epoch": 0.21, "learning_rate": 1.8140233701221822e-05, "loss": 1.4733, "step": 6183 }, { "epoch": 0.21, "learning_rate": 1.8139619265809293e-05, "loss": 1.554, "step": 6184 }, { "epoch": 0.21, "learning_rate": 1.813900473932349e-05, "loss": 1.5551, "step": 6185 }, { "epoch": 0.21, "learning_rate": 1.8138390121771295e-05, "loss": 1.4839, "step": 6186 }, { "epoch": 0.21, "learning_rate": 1.813777541315958e-05, "loss": 1.4925, "step": 6187 }, { "epoch": 0.21, "learning_rate": 1.813716061349522e-05, "loss": 1.4858, "step": 6188 }, { "epoch": 0.21, "learning_rate": 1.81365457227851e-05, "loss": 1.4393, "step": 6189 }, { "epoch": 0.21, "learning_rate": 1.81359307410361e-05, "loss": 1.3833, "step": 6190 }, { "epoch": 0.21, "learning_rate": 1.81353156682551e-05, "loss": 1.5195, "step": 6191 }, { "epoch": 0.21, "learning_rate": 1.8134700504448982e-05, "loss": 1.5061, "step": 6192 }, { "epoch": 0.21, "learning_rate": 1.8134085249624623e-05, "loss": 1.535, "step": 6193 }, { "epoch": 0.21, "learning_rate": 1.8133469903788915e-05, "loss": 1.5131, "step": 6194 }, { "epoch": 0.21, "learning_rate": 1.8132854466948743e-05, "loss": 1.4541, "step": 6195 }, { "epoch": 0.21, "learning_rate": 1.8132238939110985e-05, "loss": 1.5555, "step": 6196 }, { "epoch": 0.21, "learning_rate": 1.8131623320282535e-05, "loss": 1.4866, "step": 6197 }, { "epoch": 0.21, "learning_rate": 1.8131007610470278e-05, "loss": 1.4174, "step": 6198 }, { "epoch": 0.21, "learning_rate": 1.8130391809681105e-05, "loss": 1.4905, "step": 6199 }, { "epoch": 0.21, "learning_rate": 1.8129775917921906e-05, "loss": 1.5251, "step": 6200 }, { "epoch": 0.21, "learning_rate": 1.812915993519957e-05, "loss": 1.4918, "step": 6201 }, { "epoch": 0.21, "learning_rate": 1.8128543861520994e-05, "loss": 1.4129, "step": 6202 }, { "epoch": 0.21, "learning_rate": 1.8127927696893068e-05, "loss": 1.5178, "step": 6203 }, { "epoch": 0.21, "learning_rate": 1.812731144132268e-05, "loss": 1.5618, "step": 6204 }, { "epoch": 0.21, "learning_rate": 1.8126695094816736e-05, "loss": 1.4336, "step": 6205 }, { "epoch": 0.21, "learning_rate": 1.812607865738213e-05, "loss": 1.4993, "step": 6206 }, { "epoch": 0.21, "learning_rate": 1.812546212902575e-05, "loss": 1.4283, "step": 6207 }, { "epoch": 0.21, "learning_rate": 1.8124845509754507e-05, "loss": 1.414, "step": 6208 }, { "epoch": 0.21, "learning_rate": 1.8124228799575295e-05, "loss": 1.4722, "step": 6209 }, { "epoch": 0.21, "learning_rate": 1.812361199849501e-05, "loss": 1.4517, "step": 6210 }, { "epoch": 0.21, "learning_rate": 1.8122995106520558e-05, "loss": 1.5007, "step": 6211 }, { "epoch": 0.21, "learning_rate": 1.812237812365884e-05, "loss": 1.5368, "step": 6212 }, { "epoch": 0.21, "learning_rate": 1.812176104991676e-05, "loss": 1.5364, "step": 6213 }, { "epoch": 0.21, "learning_rate": 1.812114388530122e-05, "loss": 1.4167, "step": 6214 }, { "epoch": 0.21, "learning_rate": 1.8120526629819132e-05, "loss": 1.4988, "step": 6215 }, { "epoch": 0.21, "learning_rate": 1.8119909283477394e-05, "loss": 1.5263, "step": 6216 }, { "epoch": 0.21, "learning_rate": 1.811929184628292e-05, "loss": 1.4521, "step": 6217 }, { "epoch": 0.21, "learning_rate": 1.8118674318242613e-05, "loss": 1.4201, "step": 6218 }, { "epoch": 0.21, "learning_rate": 1.8118056699363386e-05, "loss": 1.512, "step": 6219 }, { "epoch": 0.21, "learning_rate": 1.811743898965215e-05, "loss": 1.5864, "step": 6220 }, { "epoch": 0.21, "learning_rate": 1.8116821189115813e-05, "loss": 1.517, "step": 6221 }, { "epoch": 0.21, "learning_rate": 1.8116203297761293e-05, "loss": 1.4571, "step": 6222 }, { "epoch": 0.21, "learning_rate": 1.8115585315595496e-05, "loss": 1.5759, "step": 6223 }, { "epoch": 0.21, "learning_rate": 1.8114967242625342e-05, "loss": 1.411, "step": 6224 }, { "epoch": 0.21, "learning_rate": 1.811434907885775e-05, "loss": 1.5037, "step": 6225 }, { "epoch": 0.21, "learning_rate": 1.8113730824299624e-05, "loss": 1.4894, "step": 6226 }, { "epoch": 0.21, "learning_rate": 1.8113112478957894e-05, "loss": 1.4599, "step": 6227 }, { "epoch": 0.21, "learning_rate": 1.811249404283947e-05, "loss": 1.4507, "step": 6228 }, { "epoch": 0.21, "learning_rate": 1.811187551595128e-05, "loss": 1.489, "step": 6229 }, { "epoch": 0.21, "learning_rate": 1.8111256898300237e-05, "loss": 1.4392, "step": 6230 }, { "epoch": 0.21, "learning_rate": 1.8110638189893267e-05, "loss": 1.5446, "step": 6231 }, { "epoch": 0.21, "learning_rate": 1.811001939073729e-05, "loss": 1.6367, "step": 6232 }, { "epoch": 0.21, "learning_rate": 1.8109400500839235e-05, "loss": 1.6283, "step": 6233 }, { "epoch": 0.21, "learning_rate": 1.810878152020602e-05, "loss": 1.5671, "step": 6234 }, { "epoch": 0.21, "learning_rate": 1.8108162448844575e-05, "loss": 1.468, "step": 6235 }, { "epoch": 0.21, "learning_rate": 1.8107543286761825e-05, "loss": 1.5512, "step": 6236 }, { "epoch": 0.21, "learning_rate": 1.8106924033964696e-05, "loss": 1.5967, "step": 6237 }, { "epoch": 0.21, "learning_rate": 1.810630469046012e-05, "loss": 1.5407, "step": 6238 }, { "epoch": 0.21, "learning_rate": 1.8105685256255026e-05, "loss": 1.6093, "step": 6239 }, { "epoch": 0.21, "learning_rate": 1.8105065731356343e-05, "loss": 1.6067, "step": 6240 }, { "epoch": 0.21, "learning_rate": 1.8104446115771004e-05, "loss": 1.5439, "step": 6241 }, { "epoch": 0.21, "learning_rate": 1.8103826409505944e-05, "loss": 1.6019, "step": 6242 }, { "epoch": 0.21, "learning_rate": 1.8103206612568094e-05, "loss": 1.5867, "step": 6243 }, { "epoch": 0.21, "learning_rate": 1.8102586724964387e-05, "loss": 1.5879, "step": 6244 }, { "epoch": 0.21, "learning_rate": 1.8101966746701763e-05, "loss": 1.6494, "step": 6245 }, { "epoch": 0.21, "learning_rate": 1.8101346677787157e-05, "loss": 1.5539, "step": 6246 }, { "epoch": 0.21, "learning_rate": 1.810072651822751e-05, "loss": 1.6642, "step": 6247 }, { "epoch": 0.21, "learning_rate": 1.8100106268029755e-05, "loss": 1.5865, "step": 6248 }, { "epoch": 0.21, "learning_rate": 1.809948592720084e-05, "loss": 1.563, "step": 6249 }, { "epoch": 0.21, "learning_rate": 1.8098865495747694e-05, "loss": 1.4801, "step": 6250 }, { "epoch": 0.21, "learning_rate": 1.8098244973677267e-05, "loss": 1.499, "step": 6251 }, { "epoch": 0.21, "learning_rate": 1.80976243609965e-05, "loss": 1.366, "step": 6252 }, { "epoch": 0.21, "learning_rate": 1.8097003657712343e-05, "loss": 1.6151, "step": 6253 }, { "epoch": 0.21, "learning_rate": 1.809638286383173e-05, "loss": 1.3709, "step": 6254 }, { "epoch": 0.21, "learning_rate": 1.8095761979361615e-05, "loss": 1.4739, "step": 6255 }, { "epoch": 0.21, "learning_rate": 1.8095141004308942e-05, "loss": 1.5577, "step": 6256 }, { "epoch": 0.21, "learning_rate": 1.809451993868066e-05, "loss": 1.4644, "step": 6257 }, { "epoch": 0.21, "learning_rate": 1.8093898782483718e-05, "loss": 1.6095, "step": 6258 }, { "epoch": 0.21, "learning_rate": 1.8093277535725068e-05, "loss": 1.4818, "step": 6259 }, { "epoch": 0.21, "learning_rate": 1.8092656198411654e-05, "loss": 1.4998, "step": 6260 }, { "epoch": 0.21, "learning_rate": 1.8092034770550438e-05, "loss": 1.4933, "step": 6261 }, { "epoch": 0.21, "learning_rate": 1.8091413252148366e-05, "loss": 1.4591, "step": 6262 }, { "epoch": 0.21, "learning_rate": 1.8090791643212393e-05, "loss": 1.5234, "step": 6263 }, { "epoch": 0.21, "learning_rate": 1.8090169943749477e-05, "loss": 1.4281, "step": 6264 }, { "epoch": 0.21, "learning_rate": 1.8089548153766568e-05, "loss": 1.3802, "step": 6265 }, { "epoch": 0.21, "learning_rate": 1.8088926273270633e-05, "loss": 1.3787, "step": 6266 }, { "epoch": 0.21, "learning_rate": 1.8088304302268622e-05, "loss": 1.3508, "step": 6267 }, { "epoch": 0.21, "learning_rate": 1.8087682240767496e-05, "loss": 1.4629, "step": 6268 }, { "epoch": 0.21, "learning_rate": 1.808706008877422e-05, "loss": 1.5033, "step": 6269 }, { "epoch": 0.21, "learning_rate": 1.808643784629575e-05, "loss": 1.4921, "step": 6270 }, { "epoch": 0.21, "learning_rate": 1.8085815513339046e-05, "loss": 1.4572, "step": 6271 }, { "epoch": 0.21, "learning_rate": 1.8085193089911075e-05, "loss": 1.6094, "step": 6272 }, { "epoch": 0.21, "learning_rate": 1.80845705760188e-05, "loss": 1.5266, "step": 6273 }, { "epoch": 0.21, "learning_rate": 1.808394797166919e-05, "loss": 1.5266, "step": 6274 }, { "epoch": 0.21, "learning_rate": 1.8083325276869207e-05, "loss": 1.7071, "step": 6275 }, { "epoch": 0.21, "learning_rate": 1.808270249162582e-05, "loss": 1.4896, "step": 6276 }, { "epoch": 0.21, "learning_rate": 1.8082079615945995e-05, "loss": 1.5012, "step": 6277 }, { "epoch": 0.21, "learning_rate": 1.8081456649836707e-05, "loss": 1.4935, "step": 6278 }, { "epoch": 0.21, "learning_rate": 1.8080833593304917e-05, "loss": 1.6881, "step": 6279 }, { "epoch": 0.21, "learning_rate": 1.8080210446357606e-05, "loss": 1.591, "step": 6280 }, { "epoch": 0.21, "learning_rate": 1.807958720900174e-05, "loss": 1.5745, "step": 6281 }, { "epoch": 0.21, "learning_rate": 1.8078963881244296e-05, "loss": 1.4975, "step": 6282 }, { "epoch": 0.21, "learning_rate": 1.8078340463092247e-05, "loss": 1.5585, "step": 6283 }, { "epoch": 0.21, "learning_rate": 1.8077716954552564e-05, "loss": 1.6004, "step": 6284 }, { "epoch": 0.21, "learning_rate": 1.8077093355632232e-05, "loss": 1.6085, "step": 6285 }, { "epoch": 0.21, "learning_rate": 1.807646966633822e-05, "loss": 1.5527, "step": 6286 }, { "epoch": 0.21, "learning_rate": 1.8075845886677515e-05, "loss": 1.5588, "step": 6287 }, { "epoch": 0.21, "learning_rate": 1.807522201665709e-05, "loss": 1.5248, "step": 6288 }, { "epoch": 0.21, "learning_rate": 1.8074598056283922e-05, "loss": 1.5629, "step": 6289 }, { "epoch": 0.21, "learning_rate": 1.8073974005565002e-05, "loss": 1.6365, "step": 6290 }, { "epoch": 0.21, "learning_rate": 1.8073349864507307e-05, "loss": 1.4612, "step": 6291 }, { "epoch": 0.21, "learning_rate": 1.807272563311782e-05, "loss": 1.5016, "step": 6292 }, { "epoch": 0.21, "learning_rate": 1.807210131140353e-05, "loss": 1.5207, "step": 6293 }, { "epoch": 0.21, "learning_rate": 1.8071476899371414e-05, "loss": 1.5668, "step": 6294 }, { "epoch": 0.21, "learning_rate": 1.8070852397028468e-05, "loss": 1.5979, "step": 6295 }, { "epoch": 0.21, "learning_rate": 1.8070227804381676e-05, "loss": 1.7013, "step": 6296 }, { "epoch": 0.21, "learning_rate": 1.806960312143802e-05, "loss": 1.5317, "step": 6297 }, { "epoch": 0.21, "learning_rate": 1.80689783482045e-05, "loss": 1.4957, "step": 6298 }, { "epoch": 0.21, "learning_rate": 1.8068353484688098e-05, "loss": 1.4803, "step": 6299 }, { "epoch": 0.21, "learning_rate": 1.8067728530895812e-05, "loss": 1.4345, "step": 6300 }, { "epoch": 0.21, "learning_rate": 1.8067103486834633e-05, "loss": 1.5768, "step": 6301 }, { "epoch": 0.21, "learning_rate": 1.8066478352511552e-05, "loss": 1.3405, "step": 6302 }, { "epoch": 0.21, "learning_rate": 1.8065853127933563e-05, "loss": 1.4273, "step": 6303 }, { "epoch": 0.21, "learning_rate": 1.8065227813107667e-05, "loss": 1.4601, "step": 6304 }, { "epoch": 0.21, "learning_rate": 1.806460240804085e-05, "loss": 1.5308, "step": 6305 }, { "epoch": 0.21, "learning_rate": 1.8063976912740125e-05, "loss": 1.4747, "step": 6306 }, { "epoch": 0.21, "learning_rate": 1.8063351327212476e-05, "loss": 1.4613, "step": 6307 }, { "epoch": 0.21, "learning_rate": 1.8062725651464913e-05, "loss": 1.4541, "step": 6308 }, { "epoch": 0.21, "learning_rate": 1.806209988550443e-05, "loss": 1.4984, "step": 6309 }, { "epoch": 0.21, "learning_rate": 1.8061474029338032e-05, "loss": 1.4215, "step": 6310 }, { "epoch": 0.21, "learning_rate": 1.806084808297272e-05, "loss": 1.4836, "step": 6311 }, { "epoch": 0.21, "learning_rate": 1.8060222046415502e-05, "loss": 1.5297, "step": 6312 }, { "epoch": 0.21, "learning_rate": 1.805959591967337e-05, "loss": 1.4708, "step": 6313 }, { "epoch": 0.21, "learning_rate": 1.8058969702753346e-05, "loss": 1.4208, "step": 6314 }, { "epoch": 0.21, "learning_rate": 1.805834339566243e-05, "loss": 1.4365, "step": 6315 }, { "epoch": 0.21, "learning_rate": 1.8057716998407624e-05, "loss": 1.4696, "step": 6316 }, { "epoch": 0.21, "learning_rate": 1.8057090510995946e-05, "loss": 1.4249, "step": 6317 }, { "epoch": 0.21, "learning_rate": 1.80564639334344e-05, "loss": 1.5329, "step": 6318 }, { "epoch": 0.21, "learning_rate": 1.8055837265729996e-05, "loss": 1.5478, "step": 6319 }, { "epoch": 0.21, "learning_rate": 1.8055210507889748e-05, "loss": 1.4789, "step": 6320 }, { "epoch": 0.21, "learning_rate": 1.805458365992067e-05, "loss": 1.5267, "step": 6321 }, { "epoch": 0.21, "learning_rate": 1.805395672182977e-05, "loss": 1.5518, "step": 6322 }, { "epoch": 0.21, "learning_rate": 1.8053329693624074e-05, "loss": 1.5295, "step": 6323 }, { "epoch": 0.21, "learning_rate": 1.8052702575310588e-05, "loss": 1.4458, "step": 6324 }, { "epoch": 0.21, "learning_rate": 1.805207536689633e-05, "loss": 1.5847, "step": 6325 }, { "epoch": 0.21, "learning_rate": 1.805144806838832e-05, "loss": 1.4395, "step": 6326 }, { "epoch": 0.21, "learning_rate": 1.8050820679793578e-05, "loss": 1.4396, "step": 6327 }, { "epoch": 0.21, "learning_rate": 1.805019320111912e-05, "loss": 1.5691, "step": 6328 }, { "epoch": 0.21, "learning_rate": 1.8049565632371968e-05, "loss": 1.4344, "step": 6329 }, { "epoch": 0.21, "learning_rate": 1.804893797355914e-05, "loss": 1.413, "step": 6330 }, { "epoch": 0.21, "learning_rate": 1.804831022468767e-05, "loss": 1.4548, "step": 6331 }, { "epoch": 0.21, "learning_rate": 1.8047682385764572e-05, "loss": 1.4794, "step": 6332 }, { "epoch": 0.21, "learning_rate": 1.804705445679688e-05, "loss": 1.4957, "step": 6333 }, { "epoch": 0.21, "learning_rate": 1.8046426437791603e-05, "loss": 1.5548, "step": 6334 }, { "epoch": 0.21, "learning_rate": 1.8045798328755785e-05, "loss": 1.4261, "step": 6335 }, { "epoch": 0.21, "learning_rate": 1.8045170129696444e-05, "loss": 1.5384, "step": 6336 }, { "epoch": 0.21, "learning_rate": 1.8044541840620615e-05, "loss": 1.5011, "step": 6337 }, { "epoch": 0.21, "learning_rate": 1.8043913461535325e-05, "loss": 1.5101, "step": 6338 }, { "epoch": 0.21, "learning_rate": 1.8043284992447603e-05, "loss": 1.5121, "step": 6339 }, { "epoch": 0.21, "learning_rate": 1.8042656433364484e-05, "loss": 1.4147, "step": 6340 }, { "epoch": 0.21, "learning_rate": 1.8042027784292998e-05, "loss": 1.4738, "step": 6341 }, { "epoch": 0.21, "learning_rate": 1.8041399045240183e-05, "loss": 1.4111, "step": 6342 }, { "epoch": 0.21, "learning_rate": 1.8040770216213067e-05, "loss": 1.3793, "step": 6343 }, { "epoch": 0.21, "learning_rate": 1.8040141297218693e-05, "loss": 1.4873, "step": 6344 }, { "epoch": 0.21, "learning_rate": 1.8039512288264098e-05, "loss": 1.4384, "step": 6345 }, { "epoch": 0.21, "learning_rate": 1.803888318935631e-05, "loss": 1.3964, "step": 6346 }, { "epoch": 0.21, "learning_rate": 1.8038254000502383e-05, "loss": 1.4385, "step": 6347 }, { "epoch": 0.21, "learning_rate": 1.8037624721709344e-05, "loss": 1.5306, "step": 6348 }, { "epoch": 0.21, "learning_rate": 1.803699535298424e-05, "loss": 1.4127, "step": 6349 }, { "epoch": 0.21, "learning_rate": 1.8036365894334113e-05, "loss": 1.4613, "step": 6350 }, { "epoch": 0.21, "learning_rate": 1.8035736345766e-05, "loss": 1.5706, "step": 6351 }, { "epoch": 0.21, "learning_rate": 1.8035106707286957e-05, "loss": 1.5122, "step": 6352 }, { "epoch": 0.21, "learning_rate": 1.8034476978904013e-05, "loss": 1.4349, "step": 6353 }, { "epoch": 0.21, "learning_rate": 1.803384716062423e-05, "loss": 1.5086, "step": 6354 }, { "epoch": 0.21, "learning_rate": 1.803321725245464e-05, "loss": 1.5079, "step": 6355 }, { "epoch": 0.21, "learning_rate": 1.803258725440231e-05, "loss": 1.5068, "step": 6356 }, { "epoch": 0.21, "learning_rate": 1.8031957166474265e-05, "loss": 1.4146, "step": 6357 }, { "epoch": 0.21, "learning_rate": 1.8031326988677577e-05, "loss": 1.5144, "step": 6358 }, { "epoch": 0.21, "learning_rate": 1.8030696721019282e-05, "loss": 1.46, "step": 6359 }, { "epoch": 0.21, "learning_rate": 1.8030066363506438e-05, "loss": 1.4648, "step": 6360 }, { "epoch": 0.21, "learning_rate": 1.80294359161461e-05, "loss": 1.5079, "step": 6361 }, { "epoch": 0.21, "learning_rate": 1.802880537894532e-05, "loss": 1.5051, "step": 6362 }, { "epoch": 0.21, "learning_rate": 1.8028174751911147e-05, "loss": 1.6176, "step": 6363 }, { "epoch": 0.21, "learning_rate": 1.8027544035050647e-05, "loss": 1.5605, "step": 6364 }, { "epoch": 0.21, "learning_rate": 1.802691322837087e-05, "loss": 1.514, "step": 6365 }, { "epoch": 0.21, "learning_rate": 1.8026282331878877e-05, "loss": 1.5105, "step": 6366 }, { "epoch": 0.21, "learning_rate": 1.8025651345581726e-05, "loss": 1.4974, "step": 6367 }, { "epoch": 0.21, "learning_rate": 1.802502026948648e-05, "loss": 1.4784, "step": 6368 }, { "epoch": 0.21, "learning_rate": 1.8024389103600196e-05, "loss": 1.5241, "step": 6369 }, { "epoch": 0.21, "learning_rate": 1.802375784792994e-05, "loss": 1.567, "step": 6370 }, { "epoch": 0.21, "learning_rate": 1.802312650248277e-05, "loss": 1.4355, "step": 6371 }, { "epoch": 0.21, "learning_rate": 1.802249506726575e-05, "loss": 1.5596, "step": 6372 }, { "epoch": 0.21, "learning_rate": 1.8021863542285953e-05, "loss": 1.5095, "step": 6373 }, { "epoch": 0.21, "learning_rate": 1.802123192755044e-05, "loss": 1.5349, "step": 6374 }, { "epoch": 0.21, "learning_rate": 1.8020600223066277e-05, "loss": 1.4647, "step": 6375 }, { "epoch": 0.21, "learning_rate": 1.801996842884053e-05, "loss": 1.5041, "step": 6376 }, { "epoch": 0.21, "learning_rate": 1.8019336544880276e-05, "loss": 1.4387, "step": 6377 }, { "epoch": 0.21, "learning_rate": 1.8018704571192573e-05, "loss": 1.4978, "step": 6378 }, { "epoch": 0.21, "learning_rate": 1.8018072507784505e-05, "loss": 1.3752, "step": 6379 }, { "epoch": 0.21, "learning_rate": 1.8017440354663138e-05, "loss": 1.4445, "step": 6380 }, { "epoch": 0.21, "learning_rate": 1.8016808111835544e-05, "loss": 1.5204, "step": 6381 }, { "epoch": 0.21, "learning_rate": 1.8016175779308798e-05, "loss": 1.4576, "step": 6382 }, { "epoch": 0.21, "learning_rate": 1.8015543357089976e-05, "loss": 1.3838, "step": 6383 }, { "epoch": 0.21, "learning_rate": 1.8014910845186154e-05, "loss": 1.4054, "step": 6384 }, { "epoch": 0.21, "learning_rate": 1.8014278243604407e-05, "loss": 1.4218, "step": 6385 }, { "epoch": 0.21, "learning_rate": 1.801364555235182e-05, "loss": 1.5068, "step": 6386 }, { "epoch": 0.21, "learning_rate": 1.801301277143546e-05, "loss": 1.4247, "step": 6387 }, { "epoch": 0.21, "learning_rate": 1.801237990086242e-05, "loss": 1.443, "step": 6388 }, { "epoch": 0.21, "learning_rate": 1.8011746940639773e-05, "loss": 1.478, "step": 6389 }, { "epoch": 0.21, "learning_rate": 1.8011113890774603e-05, "loss": 1.5802, "step": 6390 }, { "epoch": 0.21, "learning_rate": 1.8010480751273994e-05, "loss": 1.5388, "step": 6391 }, { "epoch": 0.21, "learning_rate": 1.800984752214503e-05, "loss": 1.5072, "step": 6392 }, { "epoch": 0.21, "learning_rate": 1.8009214203394795e-05, "loss": 1.488, "step": 6393 }, { "epoch": 0.21, "learning_rate": 1.8008580795030377e-05, "loss": 1.5517, "step": 6394 }, { "epoch": 0.21, "learning_rate": 1.8007947297058862e-05, "loss": 1.4109, "step": 6395 }, { "epoch": 0.21, "learning_rate": 1.8007313709487334e-05, "loss": 1.4658, "step": 6396 }, { "epoch": 0.21, "learning_rate": 1.8006680032322893e-05, "loss": 1.4334, "step": 6397 }, { "epoch": 0.21, "learning_rate": 1.8006046265572618e-05, "loss": 1.4864, "step": 6398 }, { "epoch": 0.21, "learning_rate": 1.8005412409243604e-05, "loss": 1.5451, "step": 6399 }, { "epoch": 0.21, "learning_rate": 1.8004778463342947e-05, "loss": 1.5963, "step": 6400 }, { "epoch": 0.21, "learning_rate": 1.8004144427877738e-05, "loss": 1.5975, "step": 6401 }, { "epoch": 0.21, "learning_rate": 1.8003510302855064e-05, "loss": 1.5224, "step": 6402 }, { "epoch": 0.21, "learning_rate": 1.800287608828203e-05, "loss": 1.4776, "step": 6403 }, { "epoch": 0.21, "learning_rate": 1.800224178416573e-05, "loss": 1.2736, "step": 6404 }, { "epoch": 0.21, "learning_rate": 1.8001607390513258e-05, "loss": 1.5301, "step": 6405 }, { "epoch": 0.21, "learning_rate": 1.8000972907331708e-05, "loss": 1.5525, "step": 6406 }, { "epoch": 0.21, "learning_rate": 1.800033833462819e-05, "loss": 1.4904, "step": 6407 }, { "epoch": 0.21, "learning_rate": 1.7999703672409802e-05, "loss": 1.5086, "step": 6408 }, { "epoch": 0.21, "learning_rate": 1.7999068920683635e-05, "loss": 1.4516, "step": 6409 }, { "epoch": 0.21, "learning_rate": 1.7998434079456803e-05, "loss": 1.489, "step": 6410 }, { "epoch": 0.21, "learning_rate": 1.7997799148736403e-05, "loss": 1.519, "step": 6411 }, { "epoch": 0.21, "learning_rate": 1.799716412852954e-05, "loss": 1.62, "step": 6412 }, { "epoch": 0.21, "learning_rate": 1.7996529018843324e-05, "loss": 1.6004, "step": 6413 }, { "epoch": 0.21, "learning_rate": 1.799589381968485e-05, "loss": 1.5697, "step": 6414 }, { "epoch": 0.21, "learning_rate": 1.7995258531061237e-05, "loss": 1.5104, "step": 6415 }, { "epoch": 0.21, "learning_rate": 1.7994623152979587e-05, "loss": 1.553, "step": 6416 }, { "epoch": 0.21, "learning_rate": 1.7993987685447008e-05, "loss": 1.6726, "step": 6417 }, { "epoch": 0.21, "learning_rate": 1.7993352128470617e-05, "loss": 1.5842, "step": 6418 }, { "epoch": 0.21, "learning_rate": 1.799271648205752e-05, "loss": 1.4178, "step": 6419 }, { "epoch": 0.21, "learning_rate": 1.799208074621483e-05, "loss": 1.3722, "step": 6420 }, { "epoch": 0.21, "learning_rate": 1.799144492094966e-05, "loss": 1.5288, "step": 6421 }, { "epoch": 0.21, "learning_rate": 1.7990809006269123e-05, "loss": 1.5924, "step": 6422 }, { "epoch": 0.21, "learning_rate": 1.7990173002180336e-05, "loss": 1.5263, "step": 6423 }, { "epoch": 0.21, "learning_rate": 1.7989536908690413e-05, "loss": 1.5044, "step": 6424 }, { "epoch": 0.21, "learning_rate": 1.7988900725806476e-05, "loss": 1.5082, "step": 6425 }, { "epoch": 0.21, "learning_rate": 1.798826445353564e-05, "loss": 1.4605, "step": 6426 }, { "epoch": 0.21, "learning_rate": 1.798762809188502e-05, "loss": 1.4607, "step": 6427 }, { "epoch": 0.21, "learning_rate": 1.7986991640861748e-05, "loss": 1.4162, "step": 6428 }, { "epoch": 0.21, "learning_rate": 1.798635510047293e-05, "loss": 1.5398, "step": 6429 }, { "epoch": 0.21, "learning_rate": 1.7985718470725698e-05, "loss": 1.3945, "step": 6430 }, { "epoch": 0.21, "learning_rate": 1.7985081751627174e-05, "loss": 1.5045, "step": 6431 }, { "epoch": 0.21, "learning_rate": 1.798444494318448e-05, "loss": 1.4281, "step": 6432 }, { "epoch": 0.21, "learning_rate": 1.7983808045404744e-05, "loss": 1.4397, "step": 6433 }, { "epoch": 0.21, "learning_rate": 1.798317105829509e-05, "loss": 1.4027, "step": 6434 }, { "epoch": 0.21, "learning_rate": 1.7982533981862645e-05, "loss": 1.5941, "step": 6435 }, { "epoch": 0.21, "learning_rate": 1.7981896816114538e-05, "loss": 1.5468, "step": 6436 }, { "epoch": 0.21, "learning_rate": 1.79812595610579e-05, "loss": 1.5466, "step": 6437 }, { "epoch": 0.21, "learning_rate": 1.7980622216699853e-05, "loss": 1.5951, "step": 6438 }, { "epoch": 0.21, "learning_rate": 1.797998478304754e-05, "loss": 1.6026, "step": 6439 }, { "epoch": 0.21, "learning_rate": 1.7979347260108088e-05, "loss": 1.5135, "step": 6440 }, { "epoch": 0.21, "learning_rate": 1.797870964788863e-05, "loss": 1.4779, "step": 6441 }, { "epoch": 0.21, "learning_rate": 1.7978071946396296e-05, "loss": 1.5588, "step": 6442 }, { "epoch": 0.21, "learning_rate": 1.7977434155638226e-05, "loss": 1.5096, "step": 6443 }, { "epoch": 0.21, "learning_rate": 1.7976796275621556e-05, "loss": 1.5341, "step": 6444 }, { "epoch": 0.21, "learning_rate": 1.797615830635342e-05, "loss": 1.5486, "step": 6445 }, { "epoch": 0.21, "learning_rate": 1.7975520247840962e-05, "loss": 1.3981, "step": 6446 }, { "epoch": 0.21, "learning_rate": 1.797488210009132e-05, "loss": 1.4407, "step": 6447 }, { "epoch": 0.21, "learning_rate": 1.7974243863111627e-05, "loss": 1.5917, "step": 6448 }, { "epoch": 0.21, "learning_rate": 1.797360553690903e-05, "loss": 1.4635, "step": 6449 }, { "epoch": 0.21, "learning_rate": 1.797296712149067e-05, "loss": 1.4972, "step": 6450 }, { "epoch": 0.22, "learning_rate": 1.797232861686369e-05, "loss": 1.5429, "step": 6451 }, { "epoch": 0.22, "learning_rate": 1.7971690023035236e-05, "loss": 1.6447, "step": 6452 }, { "epoch": 0.22, "learning_rate": 1.797105134001245e-05, "loss": 1.494, "step": 6453 }, { "epoch": 0.22, "learning_rate": 1.797041256780248e-05, "loss": 1.5586, "step": 6454 }, { "epoch": 0.22, "learning_rate": 1.7969773706412476e-05, "loss": 1.5776, "step": 6455 }, { "epoch": 0.22, "learning_rate": 1.796913475584958e-05, "loss": 1.5318, "step": 6456 }, { "epoch": 0.22, "learning_rate": 1.7968495716120945e-05, "loss": 1.4003, "step": 6457 }, { "epoch": 0.22, "learning_rate": 1.796785658723372e-05, "loss": 1.5675, "step": 6458 }, { "epoch": 0.22, "learning_rate": 1.7967217369195058e-05, "loss": 1.5409, "step": 6459 }, { "epoch": 0.22, "learning_rate": 1.796657806201211e-05, "loss": 1.5451, "step": 6460 }, { "epoch": 0.22, "learning_rate": 1.7965938665692027e-05, "loss": 1.4414, "step": 6461 }, { "epoch": 0.22, "learning_rate": 1.7965299180241963e-05, "loss": 1.4885, "step": 6462 }, { "epoch": 0.22, "learning_rate": 1.7964659605669078e-05, "loss": 1.4317, "step": 6463 }, { "epoch": 0.22, "learning_rate": 1.7964019941980524e-05, "loss": 1.5322, "step": 6464 }, { "epoch": 0.22, "learning_rate": 1.796338018918346e-05, "loss": 1.4961, "step": 6465 }, { "epoch": 0.22, "learning_rate": 1.7962740347285046e-05, "loss": 1.4845, "step": 6466 }, { "epoch": 0.22, "learning_rate": 1.7962100416292437e-05, "loss": 1.5379, "step": 6467 }, { "epoch": 0.22, "learning_rate": 1.796146039621279e-05, "loss": 1.4409, "step": 6468 }, { "epoch": 0.22, "learning_rate": 1.7960820287053276e-05, "loss": 1.5543, "step": 6469 }, { "epoch": 0.22, "learning_rate": 1.796018008882105e-05, "loss": 1.5963, "step": 6470 }, { "epoch": 0.22, "learning_rate": 1.795953980152328e-05, "loss": 1.5588, "step": 6471 }, { "epoch": 0.22, "learning_rate": 1.7958899425167123e-05, "loss": 1.5897, "step": 6472 }, { "epoch": 0.22, "learning_rate": 1.7958258959759747e-05, "loss": 1.6123, "step": 6473 }, { "epoch": 0.22, "learning_rate": 1.7957618405308323e-05, "loss": 1.4138, "step": 6474 }, { "epoch": 0.22, "learning_rate": 1.7956977761820015e-05, "loss": 1.5342, "step": 6475 }, { "epoch": 0.22, "learning_rate": 1.7956337029301987e-05, "loss": 1.6002, "step": 6476 }, { "epoch": 0.22, "learning_rate": 1.7955696207761414e-05, "loss": 1.5519, "step": 6477 }, { "epoch": 0.22, "learning_rate": 1.795505529720546e-05, "loss": 1.5788, "step": 6478 }, { "epoch": 0.22, "learning_rate": 1.7954414297641307e-05, "loss": 1.3979, "step": 6479 }, { "epoch": 0.22, "learning_rate": 1.795377320907611e-05, "loss": 1.5019, "step": 6480 }, { "epoch": 0.22, "learning_rate": 1.795313203151706e-05, "loss": 1.5867, "step": 6481 }, { "epoch": 0.22, "learning_rate": 1.795249076497132e-05, "loss": 1.4219, "step": 6482 }, { "epoch": 0.22, "learning_rate": 1.7951849409446068e-05, "loss": 1.5149, "step": 6483 }, { "epoch": 0.22, "learning_rate": 1.795120796494848e-05, "loss": 1.4583, "step": 6484 }, { "epoch": 0.22, "learning_rate": 1.7950566431485733e-05, "loss": 1.4987, "step": 6485 }, { "epoch": 0.22, "learning_rate": 1.7949924809065005e-05, "loss": 1.5892, "step": 6486 }, { "epoch": 0.22, "learning_rate": 1.794928309769347e-05, "loss": 1.6361, "step": 6487 }, { "epoch": 0.22, "learning_rate": 1.794864129737832e-05, "loss": 1.445, "step": 6488 }, { "epoch": 0.22, "learning_rate": 1.794799940812673e-05, "loss": 1.4417, "step": 6489 }, { "epoch": 0.22, "learning_rate": 1.794735742994588e-05, "loss": 1.5178, "step": 6490 }, { "epoch": 0.22, "learning_rate": 1.7946715362842955e-05, "loss": 1.4834, "step": 6491 }, { "epoch": 0.22, "learning_rate": 1.7946073206825134e-05, "loss": 1.5002, "step": 6492 }, { "epoch": 0.22, "learning_rate": 1.794543096189961e-05, "loss": 1.4655, "step": 6493 }, { "epoch": 0.22, "learning_rate": 1.794478862807356e-05, "loss": 1.5022, "step": 6494 }, { "epoch": 0.22, "learning_rate": 1.7944146205354182e-05, "loss": 1.4867, "step": 6495 }, { "epoch": 0.22, "learning_rate": 1.7943503693748658e-05, "loss": 1.4502, "step": 6496 }, { "epoch": 0.22, "learning_rate": 1.7942861093264176e-05, "loss": 1.5187, "step": 6497 }, { "epoch": 0.22, "learning_rate": 1.7942218403907923e-05, "loss": 1.5502, "step": 6498 }, { "epoch": 0.22, "learning_rate": 1.7941575625687103e-05, "loss": 1.4781, "step": 6499 }, { "epoch": 0.22, "learning_rate": 1.7940932758608895e-05, "loss": 1.4829, "step": 6500 }, { "epoch": 0.22, "learning_rate": 1.7940289802680494e-05, "loss": 1.4138, "step": 6501 }, { "epoch": 0.22, "learning_rate": 1.79396467579091e-05, "loss": 1.4105, "step": 6502 }, { "epoch": 0.22, "learning_rate": 1.7939003624301897e-05, "loss": 1.4486, "step": 6503 }, { "epoch": 0.22, "learning_rate": 1.7938360401866096e-05, "loss": 1.4641, "step": 6504 }, { "epoch": 0.22, "learning_rate": 1.7937717090608882e-05, "loss": 1.4395, "step": 6505 }, { "epoch": 0.22, "learning_rate": 1.793707369053746e-05, "loss": 1.4715, "step": 6506 }, { "epoch": 0.22, "learning_rate": 1.793643020165902e-05, "loss": 1.5585, "step": 6507 }, { "epoch": 0.22, "learning_rate": 1.793578662398077e-05, "loss": 1.4898, "step": 6508 }, { "epoch": 0.22, "learning_rate": 1.7935142957509915e-05, "loss": 1.68, "step": 6509 }, { "epoch": 0.22, "learning_rate": 1.7934499202253643e-05, "loss": 1.5937, "step": 6510 }, { "epoch": 0.22, "learning_rate": 1.793385535821917e-05, "loss": 1.5626, "step": 6511 }, { "epoch": 0.22, "learning_rate": 1.7933211425413688e-05, "loss": 1.4521, "step": 6512 }, { "epoch": 0.22, "learning_rate": 1.7932567403844413e-05, "loss": 1.5168, "step": 6513 }, { "epoch": 0.22, "learning_rate": 1.7931923293518546e-05, "loss": 1.5811, "step": 6514 }, { "epoch": 0.22, "learning_rate": 1.793127909444329e-05, "loss": 1.3602, "step": 6515 }, { "epoch": 0.22, "learning_rate": 1.793063480662586e-05, "loss": 1.5968, "step": 6516 }, { "epoch": 0.22, "learning_rate": 1.7929990430073463e-05, "loss": 1.4787, "step": 6517 }, { "epoch": 0.22, "learning_rate": 1.7929345964793306e-05, "loss": 1.435, "step": 6518 }, { "epoch": 0.22, "learning_rate": 1.79287014107926e-05, "loss": 1.5156, "step": 6519 }, { "epoch": 0.22, "learning_rate": 1.792805676807856e-05, "loss": 1.636, "step": 6520 }, { "epoch": 0.22, "learning_rate": 1.79274120366584e-05, "loss": 1.6121, "step": 6521 }, { "epoch": 0.22, "learning_rate": 1.792676721653933e-05, "loss": 1.5915, "step": 6522 }, { "epoch": 0.22, "learning_rate": 1.7926122307728562e-05, "loss": 1.5779, "step": 6523 }, { "epoch": 0.22, "learning_rate": 1.792547731023332e-05, "loss": 1.6167, "step": 6524 }, { "epoch": 0.22, "learning_rate": 1.792483222406081e-05, "loss": 1.4446, "step": 6525 }, { "epoch": 0.22, "learning_rate": 1.7924187049218264e-05, "loss": 1.4987, "step": 6526 }, { "epoch": 0.22, "learning_rate": 1.792354178571289e-05, "loss": 1.5411, "step": 6527 }, { "epoch": 0.22, "learning_rate": 1.792289643355191e-05, "loss": 1.4623, "step": 6528 }, { "epoch": 0.22, "learning_rate": 1.7922250992742546e-05, "loss": 1.4472, "step": 6529 }, { "epoch": 0.22, "learning_rate": 1.7921605463292018e-05, "loss": 1.4577, "step": 6530 }, { "epoch": 0.22, "learning_rate": 1.7920959845207552e-05, "loss": 1.5206, "step": 6531 }, { "epoch": 0.22, "learning_rate": 1.7920314138496365e-05, "loss": 1.4961, "step": 6532 }, { "epoch": 0.22, "learning_rate": 1.791966834316569e-05, "loss": 1.4324, "step": 6533 }, { "epoch": 0.22, "learning_rate": 1.7919022459222754e-05, "loss": 1.476, "step": 6534 }, { "epoch": 0.22, "learning_rate": 1.7918376486674773e-05, "loss": 1.501, "step": 6535 }, { "epoch": 0.22, "learning_rate": 1.791773042552898e-05, "loss": 1.5834, "step": 6536 }, { "epoch": 0.22, "learning_rate": 1.7917084275792607e-05, "loss": 1.507, "step": 6537 }, { "epoch": 0.22, "learning_rate": 1.7916438037472884e-05, "loss": 1.4759, "step": 6538 }, { "epoch": 0.22, "learning_rate": 1.7915791710577035e-05, "loss": 1.481, "step": 6539 }, { "epoch": 0.22, "learning_rate": 1.7915145295112297e-05, "loss": 1.4789, "step": 6540 }, { "epoch": 0.22, "learning_rate": 1.79144987910859e-05, "loss": 1.5514, "step": 6541 }, { "epoch": 0.22, "learning_rate": 1.791385219850508e-05, "loss": 1.5635, "step": 6542 }, { "epoch": 0.22, "learning_rate": 1.791320551737707e-05, "loss": 1.3558, "step": 6543 }, { "epoch": 0.22, "learning_rate": 1.7912558747709105e-05, "loss": 1.497, "step": 6544 }, { "epoch": 0.22, "learning_rate": 1.7911911889508425e-05, "loss": 1.4515, "step": 6545 }, { "epoch": 0.22, "learning_rate": 1.7911264942782264e-05, "loss": 1.431, "step": 6546 }, { "epoch": 0.22, "learning_rate": 1.7910617907537863e-05, "loss": 1.4752, "step": 6547 }, { "epoch": 0.22, "learning_rate": 1.7909970783782457e-05, "loss": 1.3247, "step": 6548 }, { "epoch": 0.22, "learning_rate": 1.7909323571523295e-05, "loss": 1.4997, "step": 6549 }, { "epoch": 0.22, "learning_rate": 1.7908676270767608e-05, "loss": 1.4717, "step": 6550 }, { "epoch": 0.22, "learning_rate": 1.790802888152265e-05, "loss": 1.3604, "step": 6551 }, { "epoch": 0.22, "learning_rate": 1.7907381403795655e-05, "loss": 1.3626, "step": 6552 }, { "epoch": 0.22, "learning_rate": 1.7906733837593876e-05, "loss": 1.3605, "step": 6553 }, { "epoch": 0.22, "learning_rate": 1.790608618292455e-05, "loss": 1.3627, "step": 6554 }, { "epoch": 0.22, "learning_rate": 1.790543843979493e-05, "loss": 1.4157, "step": 6555 }, { "epoch": 0.22, "learning_rate": 1.790479060821226e-05, "loss": 1.4172, "step": 6556 }, { "epoch": 0.22, "learning_rate": 1.790414268818379e-05, "loss": 1.4002, "step": 6557 }, { "epoch": 0.22, "learning_rate": 1.7903494679716767e-05, "loss": 1.4863, "step": 6558 }, { "epoch": 0.22, "learning_rate": 1.7902846582818444e-05, "loss": 1.3854, "step": 6559 }, { "epoch": 0.22, "learning_rate": 1.7902198397496073e-05, "loss": 1.4834, "step": 6560 }, { "epoch": 0.22, "learning_rate": 1.7901550123756906e-05, "loss": 1.4218, "step": 6561 }, { "epoch": 0.22, "learning_rate": 1.7900901761608197e-05, "loss": 1.3106, "step": 6562 }, { "epoch": 0.22, "learning_rate": 1.7900253311057193e-05, "loss": 1.3691, "step": 6563 }, { "epoch": 0.22, "learning_rate": 1.7899604772111163e-05, "loss": 1.4215, "step": 6564 }, { "epoch": 0.22, "learning_rate": 1.7898956144777354e-05, "loss": 1.5133, "step": 6565 }, { "epoch": 0.22, "learning_rate": 1.7898307429063025e-05, "loss": 1.5543, "step": 6566 }, { "epoch": 0.22, "learning_rate": 1.789765862497544e-05, "loss": 1.3489, "step": 6567 }, { "epoch": 0.22, "learning_rate": 1.7897009732521848e-05, "loss": 1.376, "step": 6568 }, { "epoch": 0.22, "learning_rate": 1.7896360751709517e-05, "loss": 1.4141, "step": 6569 }, { "epoch": 0.22, "learning_rate": 1.7895711682545704e-05, "loss": 1.4317, "step": 6570 }, { "epoch": 0.22, "learning_rate": 1.7895062525037675e-05, "loss": 1.4469, "step": 6571 }, { "epoch": 0.22, "learning_rate": 1.7894413279192693e-05, "loss": 1.401, "step": 6572 }, { "epoch": 0.22, "learning_rate": 1.789376394501802e-05, "loss": 1.5053, "step": 6573 }, { "epoch": 0.22, "learning_rate": 1.789311452252092e-05, "loss": 1.4303, "step": 6574 }, { "epoch": 0.22, "learning_rate": 1.7892465011708668e-05, "loss": 1.6182, "step": 6575 }, { "epoch": 0.22, "learning_rate": 1.7891815412588524e-05, "loss": 1.5378, "step": 6576 }, { "epoch": 0.22, "learning_rate": 1.7891165725167752e-05, "loss": 1.4752, "step": 6577 }, { "epoch": 0.22, "learning_rate": 1.7890515949453632e-05, "loss": 1.5204, "step": 6578 }, { "epoch": 0.22, "learning_rate": 1.788986608545343e-05, "loss": 1.5473, "step": 6579 }, { "epoch": 0.22, "learning_rate": 1.788921613317441e-05, "loss": 1.6026, "step": 6580 }, { "epoch": 0.22, "learning_rate": 1.788856609262386e-05, "loss": 1.609, "step": 6581 }, { "epoch": 0.22, "learning_rate": 1.7887915963809036e-05, "loss": 1.4162, "step": 6582 }, { "epoch": 0.22, "learning_rate": 1.7887265746737224e-05, "loss": 1.6348, "step": 6583 }, { "epoch": 0.22, "learning_rate": 1.7886615441415693e-05, "loss": 1.6573, "step": 6584 }, { "epoch": 0.22, "learning_rate": 1.7885965047851722e-05, "loss": 1.5972, "step": 6585 }, { "epoch": 0.22, "learning_rate": 1.788531456605259e-05, "loss": 1.6205, "step": 6586 }, { "epoch": 0.22, "learning_rate": 1.788466399602557e-05, "loss": 1.6366, "step": 6587 }, { "epoch": 0.22, "learning_rate": 1.7884013337777944e-05, "loss": 1.5884, "step": 6588 }, { "epoch": 0.22, "learning_rate": 1.7883362591316992e-05, "loss": 1.6412, "step": 6589 }, { "epoch": 0.22, "learning_rate": 1.7882711756649996e-05, "loss": 1.6494, "step": 6590 }, { "epoch": 0.22, "learning_rate": 1.7882060833784237e-05, "loss": 1.6433, "step": 6591 }, { "epoch": 0.22, "learning_rate": 1.7881409822727e-05, "loss": 1.6182, "step": 6592 }, { "epoch": 0.22, "learning_rate": 1.7880758723485564e-05, "loss": 1.6294, "step": 6593 }, { "epoch": 0.22, "learning_rate": 1.788010753606722e-05, "loss": 1.5167, "step": 6594 }, { "epoch": 0.22, "learning_rate": 1.7879456260479254e-05, "loss": 1.6387, "step": 6595 }, { "epoch": 0.22, "learning_rate": 1.7878804896728946e-05, "loss": 1.5205, "step": 6596 }, { "epoch": 0.22, "learning_rate": 1.787815344482359e-05, "loss": 1.6849, "step": 6597 }, { "epoch": 0.22, "learning_rate": 1.7877501904770476e-05, "loss": 1.576, "step": 6598 }, { "epoch": 0.22, "learning_rate": 1.787685027657689e-05, "loss": 1.5634, "step": 6599 }, { "epoch": 0.22, "learning_rate": 1.7876198560250127e-05, "loss": 1.5742, "step": 6600 }, { "epoch": 0.22, "learning_rate": 1.7875546755797474e-05, "loss": 1.4255, "step": 6601 }, { "epoch": 0.22, "learning_rate": 1.7874894863226226e-05, "loss": 1.6347, "step": 6602 }, { "epoch": 0.22, "learning_rate": 1.787424288254368e-05, "loss": 1.5698, "step": 6603 }, { "epoch": 0.22, "learning_rate": 1.787359081375713e-05, "loss": 1.4908, "step": 6604 }, { "epoch": 0.22, "learning_rate": 1.7872938656873864e-05, "loss": 1.5947, "step": 6605 }, { "epoch": 0.22, "learning_rate": 1.7872286411901192e-05, "loss": 1.6789, "step": 6606 }, { "epoch": 0.22, "learning_rate": 1.7871634078846404e-05, "loss": 1.5849, "step": 6607 }, { "epoch": 0.22, "learning_rate": 1.7870981657716798e-05, "loss": 1.6193, "step": 6608 }, { "epoch": 0.22, "learning_rate": 1.7870329148519675e-05, "loss": 1.4584, "step": 6609 }, { "epoch": 0.22, "learning_rate": 1.786967655126234e-05, "loss": 1.6377, "step": 6610 }, { "epoch": 0.22, "learning_rate": 1.7869023865952093e-05, "loss": 1.5405, "step": 6611 }, { "epoch": 0.22, "learning_rate": 1.7868371092596233e-05, "loss": 1.6119, "step": 6612 }, { "epoch": 0.22, "learning_rate": 1.7867718231202065e-05, "loss": 1.4887, "step": 6613 }, { "epoch": 0.22, "learning_rate": 1.78670652817769e-05, "loss": 1.5758, "step": 6614 }, { "epoch": 0.22, "learning_rate": 1.7866412244328034e-05, "loss": 1.5149, "step": 6615 }, { "epoch": 0.22, "learning_rate": 1.7865759118862784e-05, "loss": 1.5404, "step": 6616 }, { "epoch": 0.22, "learning_rate": 1.786510590538845e-05, "loss": 1.6501, "step": 6617 }, { "epoch": 0.22, "learning_rate": 1.786445260391234e-05, "loss": 1.5109, "step": 6618 }, { "epoch": 0.22, "learning_rate": 1.786379921444177e-05, "loss": 1.4583, "step": 6619 }, { "epoch": 0.22, "learning_rate": 1.7863145736984043e-05, "loss": 1.4577, "step": 6620 }, { "epoch": 0.22, "learning_rate": 1.7862492171546478e-05, "loss": 1.5053, "step": 6621 }, { "epoch": 0.22, "learning_rate": 1.7861838518136386e-05, "loss": 1.5715, "step": 6622 }, { "epoch": 0.22, "learning_rate": 1.7861184776761082e-05, "loss": 1.5625, "step": 6623 }, { "epoch": 0.22, "learning_rate": 1.7860530947427878e-05, "loss": 1.3903, "step": 6624 }, { "epoch": 0.22, "learning_rate": 1.7859877030144085e-05, "loss": 1.4109, "step": 6625 }, { "epoch": 0.22, "learning_rate": 1.785922302491703e-05, "loss": 1.4148, "step": 6626 }, { "epoch": 0.22, "learning_rate": 1.785856893175402e-05, "loss": 1.4407, "step": 6627 }, { "epoch": 0.22, "learning_rate": 1.7857914750662382e-05, "loss": 1.4716, "step": 6628 }, { "epoch": 0.22, "learning_rate": 1.7857260481649432e-05, "loss": 1.5465, "step": 6629 }, { "epoch": 0.22, "learning_rate": 1.785660612472249e-05, "loss": 1.5219, "step": 6630 }, { "epoch": 0.22, "learning_rate": 1.7855951679888877e-05, "loss": 1.4801, "step": 6631 }, { "epoch": 0.22, "learning_rate": 1.785529714715592e-05, "loss": 1.52, "step": 6632 }, { "epoch": 0.22, "learning_rate": 1.7854642526530935e-05, "loss": 1.3373, "step": 6633 }, { "epoch": 0.22, "learning_rate": 1.7853987818021252e-05, "loss": 1.521, "step": 6634 }, { "epoch": 0.22, "learning_rate": 1.7853333021634195e-05, "loss": 1.4664, "step": 6635 }, { "epoch": 0.22, "learning_rate": 1.785267813737709e-05, "loss": 1.4553, "step": 6636 }, { "epoch": 0.22, "learning_rate": 1.785202316525727e-05, "loss": 1.3959, "step": 6637 }, { "epoch": 0.22, "learning_rate": 1.7851368105282054e-05, "loss": 1.3948, "step": 6638 }, { "epoch": 0.22, "learning_rate": 1.7850712957458777e-05, "loss": 1.2621, "step": 6639 }, { "epoch": 0.22, "learning_rate": 1.785005772179477e-05, "loss": 1.5654, "step": 6640 }, { "epoch": 0.22, "learning_rate": 1.784940239829736e-05, "loss": 1.5259, "step": 6641 }, { "epoch": 0.22, "learning_rate": 1.7848746986973883e-05, "loss": 1.5394, "step": 6642 }, { "epoch": 0.22, "learning_rate": 1.784809148783167e-05, "loss": 1.5364, "step": 6643 }, { "epoch": 0.22, "learning_rate": 1.784743590087806e-05, "loss": 1.606, "step": 6644 }, { "epoch": 0.22, "learning_rate": 1.784678022612038e-05, "loss": 1.6105, "step": 6645 }, { "epoch": 0.22, "learning_rate": 1.784612446356597e-05, "loss": 1.4176, "step": 6646 }, { "epoch": 0.22, "learning_rate": 1.7845468613222177e-05, "loss": 1.5118, "step": 6647 }, { "epoch": 0.22, "learning_rate": 1.7844812675096323e-05, "loss": 1.3897, "step": 6648 }, { "epoch": 0.22, "learning_rate": 1.784415664919576e-05, "loss": 1.5704, "step": 6649 }, { "epoch": 0.22, "learning_rate": 1.784350053552782e-05, "loss": 1.5834, "step": 6650 }, { "epoch": 0.22, "learning_rate": 1.7842844334099845e-05, "loss": 1.5052, "step": 6651 }, { "epoch": 0.22, "learning_rate": 1.784218804491918e-05, "loss": 1.6038, "step": 6652 }, { "epoch": 0.22, "learning_rate": 1.7841531667993174e-05, "loss": 1.5779, "step": 6653 }, { "epoch": 0.22, "learning_rate": 1.784087520332916e-05, "loss": 1.6162, "step": 6654 }, { "epoch": 0.22, "learning_rate": 1.7840218650934488e-05, "loss": 1.6361, "step": 6655 }, { "epoch": 0.22, "learning_rate": 1.7839562010816503e-05, "loss": 1.5532, "step": 6656 }, { "epoch": 0.22, "learning_rate": 1.7838905282982554e-05, "loss": 1.6632, "step": 6657 }, { "epoch": 0.22, "learning_rate": 1.7838248467439987e-05, "loss": 1.5465, "step": 6658 }, { "epoch": 0.22, "learning_rate": 1.783759156419615e-05, "loss": 1.5689, "step": 6659 }, { "epoch": 0.22, "learning_rate": 1.78369345732584e-05, "loss": 1.4494, "step": 6660 }, { "epoch": 0.22, "learning_rate": 1.783627749463408e-05, "loss": 1.5635, "step": 6661 }, { "epoch": 0.22, "learning_rate": 1.7835620328330544e-05, "loss": 1.4855, "step": 6662 }, { "epoch": 0.22, "learning_rate": 1.783496307435515e-05, "loss": 1.5382, "step": 6663 }, { "epoch": 0.22, "learning_rate": 1.7834305732715243e-05, "loss": 1.5706, "step": 6664 }, { "epoch": 0.22, "learning_rate": 1.783364830341819e-05, "loss": 1.4973, "step": 6665 }, { "epoch": 0.22, "learning_rate": 1.783299078647133e-05, "loss": 1.5495, "step": 6666 }, { "epoch": 0.22, "learning_rate": 1.7832333181882035e-05, "loss": 1.5353, "step": 6667 }, { "epoch": 0.22, "learning_rate": 1.7831675489657657e-05, "loss": 1.5012, "step": 6668 }, { "epoch": 0.22, "learning_rate": 1.7831017709805555e-05, "loss": 1.4017, "step": 6669 }, { "epoch": 0.22, "learning_rate": 1.783035984233309e-05, "loss": 1.4043, "step": 6670 }, { "epoch": 0.22, "learning_rate": 1.7829701887247618e-05, "loss": 1.6338, "step": 6671 }, { "epoch": 0.22, "learning_rate": 1.782904384455651e-05, "loss": 1.529, "step": 6672 }, { "epoch": 0.22, "learning_rate": 1.7828385714267122e-05, "loss": 1.5147, "step": 6673 }, { "epoch": 0.22, "learning_rate": 1.782772749638682e-05, "loss": 1.5149, "step": 6674 }, { "epoch": 0.22, "learning_rate": 1.7827069190922964e-05, "loss": 1.4905, "step": 6675 }, { "epoch": 0.22, "learning_rate": 1.782641079788293e-05, "loss": 1.5247, "step": 6676 }, { "epoch": 0.22, "learning_rate": 1.7825752317274073e-05, "loss": 1.5429, "step": 6677 }, { "epoch": 0.22, "learning_rate": 1.7825093749103766e-05, "loss": 1.6192, "step": 6678 }, { "epoch": 0.22, "learning_rate": 1.782443509337938e-05, "loss": 1.6332, "step": 6679 }, { "epoch": 0.22, "learning_rate": 1.782377635010828e-05, "loss": 1.5172, "step": 6680 }, { "epoch": 0.22, "learning_rate": 1.782311751929784e-05, "loss": 1.6584, "step": 6681 }, { "epoch": 0.22, "learning_rate": 1.7822458600955432e-05, "loss": 1.5831, "step": 6682 }, { "epoch": 0.22, "learning_rate": 1.7821799595088425e-05, "loss": 1.5398, "step": 6683 }, { "epoch": 0.22, "learning_rate": 1.7821140501704195e-05, "loss": 1.5049, "step": 6684 }, { "epoch": 0.22, "learning_rate": 1.7820481320810118e-05, "loss": 1.5402, "step": 6685 }, { "epoch": 0.22, "learning_rate": 1.7819822052413564e-05, "loss": 1.5832, "step": 6686 }, { "epoch": 0.22, "learning_rate": 1.7819162696521914e-05, "loss": 1.5567, "step": 6687 }, { "epoch": 0.22, "learning_rate": 1.7818503253142546e-05, "loss": 1.61, "step": 6688 }, { "epoch": 0.22, "learning_rate": 1.7817843722282834e-05, "loss": 1.4281, "step": 6689 }, { "epoch": 0.22, "learning_rate": 1.7817184103950166e-05, "loss": 1.5031, "step": 6690 }, { "epoch": 0.22, "learning_rate": 1.781652439815191e-05, "loss": 1.5572, "step": 6691 }, { "epoch": 0.22, "learning_rate": 1.781586460489546e-05, "loss": 1.5324, "step": 6692 }, { "epoch": 0.22, "learning_rate": 1.781520472418819e-05, "loss": 1.5282, "step": 6693 }, { "epoch": 0.22, "learning_rate": 1.7814544756037486e-05, "loss": 1.5218, "step": 6694 }, { "epoch": 0.22, "learning_rate": 1.781388470045073e-05, "loss": 1.5434, "step": 6695 }, { "epoch": 0.22, "learning_rate": 1.7813224557435313e-05, "loss": 1.4119, "step": 6696 }, { "epoch": 0.22, "learning_rate": 1.7812564326998615e-05, "loss": 1.4254, "step": 6697 }, { "epoch": 0.22, "learning_rate": 1.7811904009148026e-05, "loss": 1.4805, "step": 6698 }, { "epoch": 0.22, "learning_rate": 1.7811243603890934e-05, "loss": 1.4976, "step": 6699 }, { "epoch": 0.22, "learning_rate": 1.781058311123473e-05, "loss": 1.5835, "step": 6700 }, { "epoch": 0.22, "learning_rate": 1.7809922531186803e-05, "loss": 1.5674, "step": 6701 }, { "epoch": 0.22, "learning_rate": 1.7809261863754542e-05, "loss": 1.4916, "step": 6702 }, { "epoch": 0.22, "learning_rate": 1.7808601108945344e-05, "loss": 1.5923, "step": 6703 }, { "epoch": 0.22, "learning_rate": 1.7807940266766595e-05, "loss": 1.5143, "step": 6704 }, { "epoch": 0.22, "learning_rate": 1.7807279337225694e-05, "loss": 1.6083, "step": 6705 }, { "epoch": 0.22, "learning_rate": 1.7806618320330035e-05, "loss": 1.4593, "step": 6706 }, { "epoch": 0.22, "learning_rate": 1.7805957216087013e-05, "loss": 1.476, "step": 6707 }, { "epoch": 0.22, "learning_rate": 1.7805296024504026e-05, "loss": 1.4965, "step": 6708 }, { "epoch": 0.22, "learning_rate": 1.7804634745588475e-05, "loss": 1.5827, "step": 6709 }, { "epoch": 0.22, "learning_rate": 1.7803973379347758e-05, "loss": 1.6525, "step": 6710 }, { "epoch": 0.22, "learning_rate": 1.7803311925789267e-05, "loss": 1.5864, "step": 6711 }, { "epoch": 0.22, "learning_rate": 1.7802650384920413e-05, "loss": 1.5119, "step": 6712 }, { "epoch": 0.22, "learning_rate": 1.780198875674859e-05, "loss": 1.5456, "step": 6713 }, { "epoch": 0.22, "learning_rate": 1.780132704128121e-05, "loss": 1.6026, "step": 6714 }, { "epoch": 0.22, "learning_rate": 1.780066523852567e-05, "loss": 1.5935, "step": 6715 }, { "epoch": 0.22, "learning_rate": 1.7800003348489372e-05, "loss": 1.423, "step": 6716 }, { "epoch": 0.22, "learning_rate": 1.779934137117973e-05, "loss": 1.5445, "step": 6717 }, { "epoch": 0.22, "learning_rate": 1.7798679306604147e-05, "loss": 1.6522, "step": 6718 }, { "epoch": 0.22, "learning_rate": 1.779801715477003e-05, "loss": 1.418, "step": 6719 }, { "epoch": 0.22, "learning_rate": 1.779735491568479e-05, "loss": 1.3997, "step": 6720 }, { "epoch": 0.22, "learning_rate": 1.7796692589355835e-05, "loss": 1.6126, "step": 6721 }, { "epoch": 0.22, "learning_rate": 1.7796030175790574e-05, "loss": 1.4508, "step": 6722 }, { "epoch": 0.22, "learning_rate": 1.779536767499642e-05, "loss": 1.5019, "step": 6723 }, { "epoch": 0.22, "learning_rate": 1.779470508698079e-05, "loss": 1.5161, "step": 6724 }, { "epoch": 0.22, "learning_rate": 1.7794042411751092e-05, "loss": 1.4365, "step": 6725 }, { "epoch": 0.22, "learning_rate": 1.7793379649314743e-05, "loss": 1.444, "step": 6726 }, { "epoch": 0.22, "learning_rate": 1.7792716799679157e-05, "loss": 1.5566, "step": 6727 }, { "epoch": 0.22, "learning_rate": 1.7792053862851752e-05, "loss": 1.6264, "step": 6728 }, { "epoch": 0.22, "learning_rate": 1.7791390838839946e-05, "loss": 1.4115, "step": 6729 }, { "epoch": 0.22, "learning_rate": 1.7790727727651158e-05, "loss": 1.4666, "step": 6730 }, { "epoch": 0.22, "learning_rate": 1.7790064529292802e-05, "loss": 1.48, "step": 6731 }, { "epoch": 0.22, "learning_rate": 1.7789401243772307e-05, "loss": 1.5129, "step": 6732 }, { "epoch": 0.22, "learning_rate": 1.7788737871097085e-05, "loss": 1.3973, "step": 6733 }, { "epoch": 0.22, "learning_rate": 1.7788074411274566e-05, "loss": 1.3342, "step": 6734 }, { "epoch": 0.22, "learning_rate": 1.778741086431217e-05, "loss": 1.3873, "step": 6735 }, { "epoch": 0.22, "learning_rate": 1.7786747230217327e-05, "loss": 1.379, "step": 6736 }, { "epoch": 0.22, "learning_rate": 1.7786083508997452e-05, "loss": 1.4713, "step": 6737 }, { "epoch": 0.22, "learning_rate": 1.7785419700659982e-05, "loss": 1.417, "step": 6738 }, { "epoch": 0.22, "learning_rate": 1.7784755805212332e-05, "loss": 1.4303, "step": 6739 }, { "epoch": 0.22, "learning_rate": 1.7784091822661943e-05, "loss": 1.4533, "step": 6740 }, { "epoch": 0.22, "learning_rate": 1.7783427753016234e-05, "loss": 1.4239, "step": 6741 }, { "epoch": 0.22, "learning_rate": 1.778276359628264e-05, "loss": 1.4426, "step": 6742 }, { "epoch": 0.22, "learning_rate": 1.7782099352468594e-05, "loss": 1.5914, "step": 6743 }, { "epoch": 0.22, "learning_rate": 1.7781435021581527e-05, "loss": 1.7067, "step": 6744 }, { "epoch": 0.22, "learning_rate": 1.778077060362887e-05, "loss": 1.5713, "step": 6745 }, { "epoch": 0.22, "learning_rate": 1.7780106098618056e-05, "loss": 1.4081, "step": 6746 }, { "epoch": 0.22, "learning_rate": 1.7779441506556528e-05, "loss": 1.4599, "step": 6747 }, { "epoch": 0.22, "learning_rate": 1.7778776827451715e-05, "loss": 1.5526, "step": 6748 }, { "epoch": 0.22, "learning_rate": 1.777811206131105e-05, "loss": 1.4204, "step": 6749 }, { "epoch": 0.23, "learning_rate": 1.777744720814198e-05, "loss": 1.4931, "step": 6750 }, { "epoch": 0.23, "learning_rate": 1.777678226795194e-05, "loss": 1.3819, "step": 6751 }, { "epoch": 0.23, "learning_rate": 1.777611724074837e-05, "loss": 1.3857, "step": 6752 }, { "epoch": 0.23, "learning_rate": 1.777545212653871e-05, "loss": 1.4154, "step": 6753 }, { "epoch": 0.23, "learning_rate": 1.7774786925330406e-05, "loss": 1.373, "step": 6754 }, { "epoch": 0.23, "learning_rate": 1.77741216371309e-05, "loss": 1.3825, "step": 6755 }, { "epoch": 0.23, "learning_rate": 1.7773456261947627e-05, "loss": 1.4642, "step": 6756 }, { "epoch": 0.23, "learning_rate": 1.777279079978804e-05, "loss": 1.4698, "step": 6757 }, { "epoch": 0.23, "learning_rate": 1.7772125250659587e-05, "loss": 1.4541, "step": 6758 }, { "epoch": 0.23, "learning_rate": 1.777145961456971e-05, "loss": 1.4913, "step": 6759 }, { "epoch": 0.23, "learning_rate": 1.7770793891525857e-05, "loss": 1.3632, "step": 6760 }, { "epoch": 0.23, "learning_rate": 1.7770128081535477e-05, "loss": 1.4019, "step": 6761 }, { "epoch": 0.23, "learning_rate": 1.7769462184606022e-05, "loss": 1.3716, "step": 6762 }, { "epoch": 0.23, "learning_rate": 1.7768796200744942e-05, "loss": 1.4897, "step": 6763 }, { "epoch": 0.23, "learning_rate": 1.7768130129959685e-05, "loss": 1.5091, "step": 6764 }, { "epoch": 0.23, "learning_rate": 1.7767463972257706e-05, "loss": 1.486, "step": 6765 }, { "epoch": 0.23, "learning_rate": 1.7766797727646458e-05, "loss": 1.4281, "step": 6766 }, { "epoch": 0.23, "learning_rate": 1.77661313961334e-05, "loss": 1.4998, "step": 6767 }, { "epoch": 0.23, "learning_rate": 1.7765464977725977e-05, "loss": 1.5621, "step": 6768 }, { "epoch": 0.23, "learning_rate": 1.7764798472431658e-05, "loss": 1.4559, "step": 6769 }, { "epoch": 0.23, "learning_rate": 1.7764131880257892e-05, "loss": 1.5705, "step": 6770 }, { "epoch": 0.23, "learning_rate": 1.776346520121214e-05, "loss": 1.5326, "step": 6771 }, { "epoch": 0.23, "learning_rate": 1.7762798435301863e-05, "loss": 1.5768, "step": 6772 }, { "epoch": 0.23, "learning_rate": 1.7762131582534518e-05, "loss": 1.4701, "step": 6773 }, { "epoch": 0.23, "learning_rate": 1.776146464291757e-05, "loss": 1.5614, "step": 6774 }, { "epoch": 0.23, "learning_rate": 1.776079761645848e-05, "loss": 1.4494, "step": 6775 }, { "epoch": 0.23, "learning_rate": 1.7760130503164708e-05, "loss": 1.4834, "step": 6776 }, { "epoch": 0.23, "learning_rate": 1.7759463303043723e-05, "loss": 1.557, "step": 6777 }, { "epoch": 0.23, "learning_rate": 1.7758796016102987e-05, "loss": 1.4966, "step": 6778 }, { "epoch": 0.23, "learning_rate": 1.775812864234997e-05, "loss": 1.5267, "step": 6779 }, { "epoch": 0.23, "learning_rate": 1.775746118179213e-05, "loss": 1.5068, "step": 6780 }, { "epoch": 0.23, "learning_rate": 1.7756793634436947e-05, "loss": 1.5059, "step": 6781 }, { "epoch": 0.23, "learning_rate": 1.7756126000291883e-05, "loss": 1.532, "step": 6782 }, { "epoch": 0.23, "learning_rate": 1.7755458279364414e-05, "loss": 1.4512, "step": 6783 }, { "epoch": 0.23, "learning_rate": 1.7754790471662003e-05, "loss": 1.5079, "step": 6784 }, { "epoch": 0.23, "learning_rate": 1.7754122577192127e-05, "loss": 1.5572, "step": 6785 }, { "epoch": 0.23, "learning_rate": 1.7753454595962257e-05, "loss": 1.5716, "step": 6786 }, { "epoch": 0.23, "learning_rate": 1.7752786527979867e-05, "loss": 1.5178, "step": 6787 }, { "epoch": 0.23, "learning_rate": 1.7752118373252436e-05, "loss": 1.7995, "step": 6788 }, { "epoch": 0.23, "learning_rate": 1.7751450131787435e-05, "loss": 1.655, "step": 6789 }, { "epoch": 0.23, "learning_rate": 1.775078180359234e-05, "loss": 1.5706, "step": 6790 }, { "epoch": 0.23, "learning_rate": 1.7750113388674638e-05, "loss": 1.6316, "step": 6791 }, { "epoch": 0.23, "learning_rate": 1.7749444887041797e-05, "loss": 1.6641, "step": 6792 }, { "epoch": 0.23, "learning_rate": 1.7748776298701303e-05, "loss": 1.5969, "step": 6793 }, { "epoch": 0.23, "learning_rate": 1.7748107623660635e-05, "loss": 1.519, "step": 6794 }, { "epoch": 0.23, "learning_rate": 1.7747438861927272e-05, "loss": 1.546, "step": 6795 }, { "epoch": 0.23, "learning_rate": 1.77467700135087e-05, "loss": 1.4406, "step": 6796 }, { "epoch": 0.23, "learning_rate": 1.7746101078412404e-05, "loss": 1.3824, "step": 6797 }, { "epoch": 0.23, "learning_rate": 1.7745432056645862e-05, "loss": 1.5404, "step": 6798 }, { "epoch": 0.23, "learning_rate": 1.7744762948216568e-05, "loss": 1.5266, "step": 6799 }, { "epoch": 0.23, "learning_rate": 1.7744093753132006e-05, "loss": 1.5634, "step": 6800 }, { "epoch": 0.23, "learning_rate": 1.7743424471399662e-05, "loss": 1.6141, "step": 6801 }, { "epoch": 0.23, "learning_rate": 1.7742755103027022e-05, "loss": 1.6658, "step": 6802 }, { "epoch": 0.23, "learning_rate": 1.774208564802158e-05, "loss": 1.5922, "step": 6803 }, { "epoch": 0.23, "learning_rate": 1.7741416106390828e-05, "loss": 1.6263, "step": 6804 }, { "epoch": 0.23, "learning_rate": 1.774074647814225e-05, "loss": 1.4813, "step": 6805 }, { "epoch": 0.23, "learning_rate": 1.774007676328334e-05, "loss": 1.6287, "step": 6806 }, { "epoch": 0.23, "learning_rate": 1.77394069618216e-05, "loss": 1.6115, "step": 6807 }, { "epoch": 0.23, "learning_rate": 1.7738737073764516e-05, "loss": 1.3777, "step": 6808 }, { "epoch": 0.23, "learning_rate": 1.7738067099119587e-05, "loss": 1.4613, "step": 6809 }, { "epoch": 0.23, "learning_rate": 1.7737397037894305e-05, "loss": 1.533, "step": 6810 }, { "epoch": 0.23, "learning_rate": 1.7736726890096174e-05, "loss": 1.5487, "step": 6811 }, { "epoch": 0.23, "learning_rate": 1.7736056655732685e-05, "loss": 1.504, "step": 6812 }, { "epoch": 0.23, "learning_rate": 1.7735386334811343e-05, "loss": 1.4631, "step": 6813 }, { "epoch": 0.23, "learning_rate": 1.7734715927339642e-05, "loss": 1.5961, "step": 6814 }, { "epoch": 0.23, "learning_rate": 1.7734045433325088e-05, "loss": 1.4829, "step": 6815 }, { "epoch": 0.23, "learning_rate": 1.7733374852775184e-05, "loss": 1.4444, "step": 6816 }, { "epoch": 0.23, "learning_rate": 1.7732704185697426e-05, "loss": 1.5645, "step": 6817 }, { "epoch": 0.23, "learning_rate": 1.773203343209933e-05, "loss": 1.4847, "step": 6818 }, { "epoch": 0.23, "learning_rate": 1.773136259198839e-05, "loss": 1.6216, "step": 6819 }, { "epoch": 0.23, "learning_rate": 1.7730691665372117e-05, "loss": 1.6058, "step": 6820 }, { "epoch": 0.23, "learning_rate": 1.7730020652258015e-05, "loss": 1.5001, "step": 6821 }, { "epoch": 0.23, "learning_rate": 1.7729349552653596e-05, "loss": 1.4942, "step": 6822 }, { "epoch": 0.23, "learning_rate": 1.7728678366566364e-05, "loss": 1.4011, "step": 6823 }, { "epoch": 0.23, "learning_rate": 1.772800709400383e-05, "loss": 1.3772, "step": 6824 }, { "epoch": 0.23, "learning_rate": 1.7727335734973512e-05, "loss": 1.4581, "step": 6825 }, { "epoch": 0.23, "learning_rate": 1.7726664289482914e-05, "loss": 1.4541, "step": 6826 }, { "epoch": 0.23, "learning_rate": 1.772599275753955e-05, "loss": 1.4278, "step": 6827 }, { "epoch": 0.23, "learning_rate": 1.772532113915093e-05, "loss": 1.4432, "step": 6828 }, { "epoch": 0.23, "learning_rate": 1.772464943432458e-05, "loss": 1.4986, "step": 6829 }, { "epoch": 0.23, "learning_rate": 1.7723977643068005e-05, "loss": 1.5555, "step": 6830 }, { "epoch": 0.23, "learning_rate": 1.7723305765388726e-05, "loss": 1.491, "step": 6831 }, { "epoch": 0.23, "learning_rate": 1.7722633801294258e-05, "loss": 1.5727, "step": 6832 }, { "epoch": 0.23, "learning_rate": 1.7721961750792123e-05, "loss": 1.5275, "step": 6833 }, { "epoch": 0.23, "learning_rate": 1.7721289613889835e-05, "loss": 1.5291, "step": 6834 }, { "epoch": 0.23, "learning_rate": 1.7720617390594923e-05, "loss": 1.5553, "step": 6835 }, { "epoch": 0.23, "learning_rate": 1.7719945080914902e-05, "loss": 1.3221, "step": 6836 }, { "epoch": 0.23, "learning_rate": 1.7719272684857297e-05, "loss": 1.4321, "step": 6837 }, { "epoch": 0.23, "learning_rate": 1.771860020242963e-05, "loss": 1.4761, "step": 6838 }, { "epoch": 0.23, "learning_rate": 1.7717927633639424e-05, "loss": 1.6236, "step": 6839 }, { "epoch": 0.23, "learning_rate": 1.771725497849421e-05, "loss": 1.4376, "step": 6840 }, { "epoch": 0.23, "learning_rate": 1.7716582237001506e-05, "loss": 1.4483, "step": 6841 }, { "epoch": 0.23, "learning_rate": 1.771590940916885e-05, "loss": 1.5335, "step": 6842 }, { "epoch": 0.23, "learning_rate": 1.7715236495003755e-05, "loss": 1.4553, "step": 6843 }, { "epoch": 0.23, "learning_rate": 1.7714563494513766e-05, "loss": 1.4069, "step": 6844 }, { "epoch": 0.23, "learning_rate": 1.77138904077064e-05, "loss": 1.6409, "step": 6845 }, { "epoch": 0.23, "learning_rate": 1.77132172345892e-05, "loss": 1.5859, "step": 6846 }, { "epoch": 0.23, "learning_rate": 1.7712543975169687e-05, "loss": 1.4892, "step": 6847 }, { "epoch": 0.23, "learning_rate": 1.7711870629455405e-05, "loss": 1.5128, "step": 6848 }, { "epoch": 0.23, "learning_rate": 1.771119719745388e-05, "loss": 1.5235, "step": 6849 }, { "epoch": 0.23, "learning_rate": 1.7710523679172646e-05, "loss": 1.5153, "step": 6850 }, { "epoch": 0.23, "learning_rate": 1.7709850074619243e-05, "loss": 1.4749, "step": 6851 }, { "epoch": 0.23, "learning_rate": 1.770917638380121e-05, "loss": 1.5621, "step": 6852 }, { "epoch": 0.23, "learning_rate": 1.770850260672608e-05, "loss": 1.5084, "step": 6853 }, { "epoch": 0.23, "learning_rate": 1.770782874340139e-05, "loss": 1.4838, "step": 6854 }, { "epoch": 0.23, "learning_rate": 1.770715479383469e-05, "loss": 1.4918, "step": 6855 }, { "epoch": 0.23, "learning_rate": 1.7706480758033514e-05, "loss": 1.5007, "step": 6856 }, { "epoch": 0.23, "learning_rate": 1.7705806636005397e-05, "loss": 1.5375, "step": 6857 }, { "epoch": 0.23, "learning_rate": 1.7705132427757895e-05, "loss": 1.4834, "step": 6858 }, { "epoch": 0.23, "learning_rate": 1.7704458133298542e-05, "loss": 1.5543, "step": 6859 }, { "epoch": 0.23, "learning_rate": 1.7703783752634888e-05, "loss": 1.6386, "step": 6860 }, { "epoch": 0.23, "learning_rate": 1.7703109285774474e-05, "loss": 1.6337, "step": 6861 }, { "epoch": 0.23, "learning_rate": 1.770243473272485e-05, "loss": 1.6164, "step": 6862 }, { "epoch": 0.23, "learning_rate": 1.770176009349356e-05, "loss": 1.663, "step": 6863 }, { "epoch": 0.23, "learning_rate": 1.7701085368088157e-05, "loss": 1.5573, "step": 6864 }, { "epoch": 0.23, "learning_rate": 1.770041055651619e-05, "loss": 1.4953, "step": 6865 }, { "epoch": 0.23, "learning_rate": 1.7699735658785203e-05, "loss": 1.5735, "step": 6866 }, { "epoch": 0.23, "learning_rate": 1.7699060674902756e-05, "loss": 1.5997, "step": 6867 }, { "epoch": 0.23, "learning_rate": 1.7698385604876398e-05, "loss": 1.5361, "step": 6868 }, { "epoch": 0.23, "learning_rate": 1.769771044871368e-05, "loss": 1.5323, "step": 6869 }, { "epoch": 0.23, "learning_rate": 1.769703520642216e-05, "loss": 1.5832, "step": 6870 }, { "epoch": 0.23, "learning_rate": 1.7696359878009386e-05, "loss": 1.4944, "step": 6871 }, { "epoch": 0.23, "learning_rate": 1.7695684463482924e-05, "loss": 1.5653, "step": 6872 }, { "epoch": 0.23, "learning_rate": 1.7695684463482924e-05, "loss": 1.5809, "step": 6873 }, { "epoch": 0.23, "learning_rate": 1.769500896285032e-05, "loss": 1.5773, "step": 6874 }, { "epoch": 0.23, "learning_rate": 1.7694333376119144e-05, "loss": 1.5797, "step": 6875 }, { "epoch": 0.23, "learning_rate": 1.769365770329695e-05, "loss": 1.5164, "step": 6876 }, { "epoch": 0.23, "learning_rate": 1.7692981944391293e-05, "loss": 1.571, "step": 6877 }, { "epoch": 0.23, "learning_rate": 1.7692306099409745e-05, "loss": 1.5444, "step": 6878 }, { "epoch": 0.23, "learning_rate": 1.7691630168359857e-05, "loss": 1.5946, "step": 6879 }, { "epoch": 0.23, "learning_rate": 1.7690954151249196e-05, "loss": 1.488, "step": 6880 }, { "epoch": 0.23, "learning_rate": 1.7690278048085327e-05, "loss": 1.4053, "step": 6881 }, { "epoch": 0.23, "learning_rate": 1.7689601858875817e-05, "loss": 1.4724, "step": 6882 }, { "epoch": 0.23, "learning_rate": 1.7688925583628226e-05, "loss": 1.4706, "step": 6883 }, { "epoch": 0.23, "learning_rate": 1.7688249222350125e-05, "loss": 1.4757, "step": 6884 }, { "epoch": 0.23, "learning_rate": 1.768757277504908e-05, "loss": 1.5311, "step": 6885 }, { "epoch": 0.23, "learning_rate": 1.768689624173266e-05, "loss": 1.5552, "step": 6886 }, { "epoch": 0.23, "learning_rate": 1.7686219622408435e-05, "loss": 1.3609, "step": 6887 }, { "epoch": 0.23, "learning_rate": 1.7685542917083975e-05, "loss": 1.4014, "step": 6888 }, { "epoch": 0.23, "learning_rate": 1.7684866125766853e-05, "loss": 1.4199, "step": 6889 }, { "epoch": 0.23, "learning_rate": 1.768418924846464e-05, "loss": 1.5236, "step": 6890 }, { "epoch": 0.23, "learning_rate": 1.7683512285184908e-05, "loss": 1.6259, "step": 6891 }, { "epoch": 0.23, "learning_rate": 1.7682835235935236e-05, "loss": 1.5256, "step": 6892 }, { "epoch": 0.23, "learning_rate": 1.76821581007232e-05, "loss": 1.4603, "step": 6893 }, { "epoch": 0.23, "learning_rate": 1.7681480879556365e-05, "loss": 1.5131, "step": 6894 }, { "epoch": 0.23, "learning_rate": 1.768080357244232e-05, "loss": 1.5228, "step": 6895 }, { "epoch": 0.23, "learning_rate": 1.768012617938864e-05, "loss": 1.3403, "step": 6896 }, { "epoch": 0.23, "learning_rate": 1.7679448700402904e-05, "loss": 1.3703, "step": 6897 }, { "epoch": 0.23, "learning_rate": 1.7678771135492696e-05, "loss": 1.3951, "step": 6898 }, { "epoch": 0.23, "learning_rate": 1.767809348466559e-05, "loss": 1.4697, "step": 6899 }, { "epoch": 0.23, "learning_rate": 1.7677415747929173e-05, "loss": 1.301, "step": 6900 }, { "epoch": 0.23, "learning_rate": 1.7676737925291027e-05, "loss": 1.3882, "step": 6901 }, { "epoch": 0.23, "learning_rate": 1.767606001675874e-05, "loss": 1.409, "step": 6902 }, { "epoch": 0.23, "learning_rate": 1.767538202233989e-05, "loss": 1.3886, "step": 6903 }, { "epoch": 0.23, "learning_rate": 1.767470394204206e-05, "loss": 1.3556, "step": 6904 }, { "epoch": 0.23, "learning_rate": 1.767402577587285e-05, "loss": 1.3407, "step": 6905 }, { "epoch": 0.23, "learning_rate": 1.767334752383984e-05, "loss": 1.4221, "step": 6906 }, { "epoch": 0.23, "learning_rate": 1.767266918595062e-05, "loss": 1.3412, "step": 6907 }, { "epoch": 0.23, "learning_rate": 1.7671990762212778e-05, "loss": 1.4419, "step": 6908 }, { "epoch": 0.23, "learning_rate": 1.7671312252633906e-05, "loss": 1.4256, "step": 6909 }, { "epoch": 0.23, "learning_rate": 1.7670633657221602e-05, "loss": 1.5023, "step": 6910 }, { "epoch": 0.23, "learning_rate": 1.7669954975983447e-05, "loss": 1.5722, "step": 6911 }, { "epoch": 0.23, "learning_rate": 1.766927620892704e-05, "loss": 1.3958, "step": 6912 }, { "epoch": 0.23, "learning_rate": 1.7668597356059977e-05, "loss": 1.5385, "step": 6913 }, { "epoch": 0.23, "learning_rate": 1.7667918417389857e-05, "loss": 1.4538, "step": 6914 }, { "epoch": 0.23, "learning_rate": 1.766723939292427e-05, "loss": 1.429, "step": 6915 }, { "epoch": 0.23, "learning_rate": 1.7666560282670814e-05, "loss": 1.4714, "step": 6916 }, { "epoch": 0.23, "learning_rate": 1.7665881086637092e-05, "loss": 1.4206, "step": 6917 }, { "epoch": 0.23, "learning_rate": 1.7665201804830704e-05, "loss": 1.4633, "step": 6918 }, { "epoch": 0.23, "learning_rate": 1.766452243725924e-05, "loss": 1.4062, "step": 6919 }, { "epoch": 0.23, "learning_rate": 1.7663842983930314e-05, "loss": 1.4752, "step": 6920 }, { "epoch": 0.23, "learning_rate": 1.7663163444851522e-05, "loss": 1.4629, "step": 6921 }, { "epoch": 0.23, "learning_rate": 1.7662483820030468e-05, "loss": 1.4022, "step": 6922 }, { "epoch": 0.23, "learning_rate": 1.7661804109474755e-05, "loss": 1.451, "step": 6923 }, { "epoch": 0.23, "learning_rate": 1.7661124313191992e-05, "loss": 1.5613, "step": 6924 }, { "epoch": 0.23, "learning_rate": 1.766044443118978e-05, "loss": 1.4477, "step": 6925 }, { "epoch": 0.23, "learning_rate": 1.7659764463475732e-05, "loss": 1.4269, "step": 6926 }, { "epoch": 0.23, "learning_rate": 1.7659084410057453e-05, "loss": 1.3861, "step": 6927 }, { "epoch": 0.23, "learning_rate": 1.7658404270942554e-05, "loss": 1.4057, "step": 6928 }, { "epoch": 0.23, "learning_rate": 1.765772404613864e-05, "loss": 1.4903, "step": 6929 }, { "epoch": 0.23, "learning_rate": 1.7657043735653324e-05, "loss": 1.446, "step": 6930 }, { "epoch": 0.23, "learning_rate": 1.7656363339494222e-05, "loss": 1.4811, "step": 6931 }, { "epoch": 0.23, "learning_rate": 1.7655682857668942e-05, "loss": 1.4199, "step": 6932 }, { "epoch": 0.23, "learning_rate": 1.76550022901851e-05, "loss": 1.5518, "step": 6933 }, { "epoch": 0.23, "learning_rate": 1.7654321637050314e-05, "loss": 1.4703, "step": 6934 }, { "epoch": 0.23, "learning_rate": 1.7653640898272192e-05, "loss": 1.4025, "step": 6935 }, { "epoch": 0.23, "learning_rate": 1.765296007385836e-05, "loss": 1.4387, "step": 6936 }, { "epoch": 0.23, "learning_rate": 1.7652279163816422e-05, "loss": 1.5023, "step": 6937 }, { "epoch": 0.23, "learning_rate": 1.7651598168154014e-05, "loss": 1.4674, "step": 6938 }, { "epoch": 0.23, "learning_rate": 1.7650917086878744e-05, "loss": 1.5855, "step": 6939 }, { "epoch": 0.23, "learning_rate": 1.7650235919998234e-05, "loss": 1.478, "step": 6940 }, { "epoch": 0.23, "learning_rate": 1.7649554667520105e-05, "loss": 1.4372, "step": 6941 }, { "epoch": 0.23, "learning_rate": 1.7648873329451987e-05, "loss": 1.4959, "step": 6942 }, { "epoch": 0.23, "learning_rate": 1.7648191905801493e-05, "loss": 1.5494, "step": 6943 }, { "epoch": 0.23, "learning_rate": 1.7647510396576256e-05, "loss": 1.588, "step": 6944 }, { "epoch": 0.23, "learning_rate": 1.7646828801783897e-05, "loss": 1.4036, "step": 6945 }, { "epoch": 0.23, "learning_rate": 1.764614712143204e-05, "loss": 1.4463, "step": 6946 }, { "epoch": 0.23, "learning_rate": 1.7645465355528317e-05, "loss": 1.4076, "step": 6947 }, { "epoch": 0.23, "learning_rate": 1.7644783504080353e-05, "loss": 1.554, "step": 6948 }, { "epoch": 0.23, "learning_rate": 1.764410156709578e-05, "loss": 1.495, "step": 6949 }, { "epoch": 0.23, "learning_rate": 1.7643419544582225e-05, "loss": 1.4433, "step": 6950 }, { "epoch": 0.23, "learning_rate": 1.764273743654732e-05, "loss": 1.4366, "step": 6951 }, { "epoch": 0.23, "learning_rate": 1.76420552429987e-05, "loss": 1.5032, "step": 6952 }, { "epoch": 0.23, "learning_rate": 1.7641372963943996e-05, "loss": 1.4483, "step": 6953 }, { "epoch": 0.23, "learning_rate": 1.764069059939084e-05, "loss": 1.4658, "step": 6954 }, { "epoch": 0.23, "learning_rate": 1.7640008149346866e-05, "loss": 1.3526, "step": 6955 }, { "epoch": 0.23, "learning_rate": 1.763932561381972e-05, "loss": 1.5202, "step": 6956 }, { "epoch": 0.23, "learning_rate": 1.763864299281702e-05, "loss": 1.6435, "step": 6957 }, { "epoch": 0.23, "learning_rate": 1.7637960286346423e-05, "loss": 1.6714, "step": 6958 }, { "epoch": 0.23, "learning_rate": 1.763727749441556e-05, "loss": 1.6291, "step": 6959 }, { "epoch": 0.23, "learning_rate": 1.7636594617032067e-05, "loss": 1.5267, "step": 6960 }, { "epoch": 0.23, "learning_rate": 1.7635911654203588e-05, "loss": 1.6604, "step": 6961 }, { "epoch": 0.23, "learning_rate": 1.7635228605937765e-05, "loss": 1.6762, "step": 6962 }, { "epoch": 0.23, "learning_rate": 1.763454547224224e-05, "loss": 1.4742, "step": 6963 }, { "epoch": 0.23, "learning_rate": 1.7633862253124656e-05, "loss": 1.5453, "step": 6964 }, { "epoch": 0.23, "learning_rate": 1.763317894859266e-05, "loss": 1.5029, "step": 6965 }, { "epoch": 0.23, "learning_rate": 1.7632495558653893e-05, "loss": 1.4358, "step": 6966 }, { "epoch": 0.23, "learning_rate": 1.7631812083316003e-05, "loss": 1.4449, "step": 6967 }, { "epoch": 0.23, "learning_rate": 1.763112852258664e-05, "loss": 1.5857, "step": 6968 }, { "epoch": 0.23, "learning_rate": 1.763044487647345e-05, "loss": 1.4841, "step": 6969 }, { "epoch": 0.23, "learning_rate": 1.7629761144984087e-05, "loss": 1.4779, "step": 6970 }, { "epoch": 0.23, "learning_rate": 1.7629077328126195e-05, "loss": 1.481, "step": 6971 }, { "epoch": 0.23, "learning_rate": 1.762839342590742e-05, "loss": 1.3963, "step": 6972 }, { "epoch": 0.23, "learning_rate": 1.762770943833543e-05, "loss": 1.4714, "step": 6973 }, { "epoch": 0.23, "learning_rate": 1.7627025365417865e-05, "loss": 1.4487, "step": 6974 }, { "epoch": 0.23, "learning_rate": 1.762634120716238e-05, "loss": 1.4945, "step": 6975 }, { "epoch": 0.23, "learning_rate": 1.762565696357664e-05, "loss": 1.5787, "step": 6976 }, { "epoch": 0.23, "learning_rate": 1.762497263466829e-05, "loss": 1.5449, "step": 6977 }, { "epoch": 0.23, "learning_rate": 1.762428822044499e-05, "loss": 1.4445, "step": 6978 }, { "epoch": 0.23, "learning_rate": 1.76236037209144e-05, "loss": 1.4339, "step": 6979 }, { "epoch": 0.23, "learning_rate": 1.7622919136084183e-05, "loss": 1.469, "step": 6980 }, { "epoch": 0.23, "learning_rate": 1.7622234465961988e-05, "loss": 1.3554, "step": 6981 }, { "epoch": 0.23, "learning_rate": 1.762154971055548e-05, "loss": 1.4618, "step": 6982 }, { "epoch": 0.23, "learning_rate": 1.7620864869872324e-05, "loss": 1.4803, "step": 6983 }, { "epoch": 0.23, "learning_rate": 1.7620179943920178e-05, "loss": 1.4004, "step": 6984 }, { "epoch": 0.23, "learning_rate": 1.761949493270671e-05, "loss": 1.4574, "step": 6985 }, { "epoch": 0.23, "learning_rate": 1.7618809836239582e-05, "loss": 1.4113, "step": 6986 }, { "epoch": 0.23, "learning_rate": 1.7618124654526457e-05, "loss": 1.6169, "step": 6987 }, { "epoch": 0.23, "learning_rate": 1.7617439387575007e-05, "loss": 1.4529, "step": 6988 }, { "epoch": 0.23, "learning_rate": 1.7616754035392897e-05, "loss": 1.427, "step": 6989 }, { "epoch": 0.23, "learning_rate": 1.7616068597987793e-05, "loss": 1.5597, "step": 6990 }, { "epoch": 0.23, "learning_rate": 1.761538307536737e-05, "loss": 1.481, "step": 6991 }, { "epoch": 0.23, "learning_rate": 1.761469746753929e-05, "loss": 1.4318, "step": 6992 }, { "epoch": 0.23, "learning_rate": 1.761401177451123e-05, "loss": 1.327, "step": 6993 }, { "epoch": 0.23, "learning_rate": 1.761332599629086e-05, "loss": 1.498, "step": 6994 }, { "epoch": 0.23, "learning_rate": 1.7612640132885857e-05, "loss": 1.5266, "step": 6995 }, { "epoch": 0.23, "learning_rate": 1.761195418430389e-05, "loss": 1.5208, "step": 6996 }, { "epoch": 0.23, "learning_rate": 1.7611268150552637e-05, "loss": 1.508, "step": 6997 }, { "epoch": 0.23, "learning_rate": 1.7610582031639772e-05, "loss": 1.4652, "step": 6998 }, { "epoch": 0.23, "learning_rate": 1.760989582757297e-05, "loss": 1.4704, "step": 6999 }, { "epoch": 0.23, "learning_rate": 1.7609209538359917e-05, "loss": 1.5169, "step": 7000 }, { "epoch": 0.23, "learning_rate": 1.760852316400828e-05, "loss": 1.5439, "step": 7001 }, { "epoch": 0.23, "learning_rate": 1.760783670452575e-05, "loss": 1.3967, "step": 7002 }, { "epoch": 0.23, "learning_rate": 1.760715015992e-05, "loss": 1.4009, "step": 7003 }, { "epoch": 0.23, "learning_rate": 1.760646353019872e-05, "loss": 1.3884, "step": 7004 }, { "epoch": 0.23, "learning_rate": 1.7605776815369583e-05, "loss": 1.4524, "step": 7005 }, { "epoch": 0.23, "learning_rate": 1.7605090015440277e-05, "loss": 1.4034, "step": 7006 }, { "epoch": 0.23, "learning_rate": 1.7604403130418485e-05, "loss": 1.3886, "step": 7007 }, { "epoch": 0.23, "learning_rate": 1.7603716160311898e-05, "loss": 1.4835, "step": 7008 }, { "epoch": 0.23, "learning_rate": 1.7603029105128193e-05, "loss": 1.4144, "step": 7009 }, { "epoch": 0.23, "learning_rate": 1.7602341964875066e-05, "loss": 1.5045, "step": 7010 }, { "epoch": 0.23, "learning_rate": 1.76016547395602e-05, "loss": 1.457, "step": 7011 }, { "epoch": 0.23, "learning_rate": 1.760096742919129e-05, "loss": 1.5109, "step": 7012 }, { "epoch": 0.23, "learning_rate": 1.7600280033776018e-05, "loss": 1.4463, "step": 7013 }, { "epoch": 0.23, "learning_rate": 1.7599592553322082e-05, "loss": 1.505, "step": 7014 }, { "epoch": 0.23, "learning_rate": 1.759890498783717e-05, "loss": 1.4964, "step": 7015 }, { "epoch": 0.23, "learning_rate": 1.759821733732898e-05, "loss": 1.525, "step": 7016 }, { "epoch": 0.23, "learning_rate": 1.75975296018052e-05, "loss": 1.4187, "step": 7017 }, { "epoch": 0.23, "learning_rate": 1.759684178127353e-05, "loss": 1.5463, "step": 7018 }, { "epoch": 0.23, "learning_rate": 1.7596153875741663e-05, "loss": 1.5521, "step": 7019 }, { "epoch": 0.23, "learning_rate": 1.7595465885217296e-05, "loss": 1.4718, "step": 7020 }, { "epoch": 0.23, "learning_rate": 1.759477780970813e-05, "loss": 1.5707, "step": 7021 }, { "epoch": 0.23, "learning_rate": 1.7594089649221858e-05, "loss": 1.4527, "step": 7022 }, { "epoch": 0.23, "learning_rate": 1.7593401403766185e-05, "loss": 1.521, "step": 7023 }, { "epoch": 0.23, "learning_rate": 1.759271307334881e-05, "loss": 1.4673, "step": 7024 }, { "epoch": 0.23, "learning_rate": 1.7592024657977432e-05, "loss": 1.4155, "step": 7025 }, { "epoch": 0.23, "learning_rate": 1.759133615765976e-05, "loss": 1.5388, "step": 7026 }, { "epoch": 0.23, "learning_rate": 1.7590647572403493e-05, "loss": 1.524, "step": 7027 }, { "epoch": 0.23, "learning_rate": 1.7589958902216335e-05, "loss": 1.4883, "step": 7028 }, { "epoch": 0.23, "learning_rate": 1.7589270147105993e-05, "loss": 1.5719, "step": 7029 }, { "epoch": 0.23, "learning_rate": 1.758858130708017e-05, "loss": 1.4562, "step": 7030 }, { "epoch": 0.23, "learning_rate": 1.758789238214658e-05, "loss": 1.3974, "step": 7031 }, { "epoch": 0.23, "learning_rate": 1.7587203372312927e-05, "loss": 1.5203, "step": 7032 }, { "epoch": 0.23, "learning_rate": 1.758651427758692e-05, "loss": 1.4483, "step": 7033 }, { "epoch": 0.23, "learning_rate": 1.7585825097976273e-05, "loss": 1.4543, "step": 7034 }, { "epoch": 0.23, "learning_rate": 1.7585135833488692e-05, "loss": 1.4649, "step": 7035 }, { "epoch": 0.23, "learning_rate": 1.758444648413189e-05, "loss": 1.4945, "step": 7036 }, { "epoch": 0.23, "learning_rate": 1.7583757049913583e-05, "loss": 1.4231, "step": 7037 }, { "epoch": 0.23, "learning_rate": 1.758306753084148e-05, "loss": 1.5754, "step": 7038 }, { "epoch": 0.23, "learning_rate": 1.7582377926923306e-05, "loss": 1.3662, "step": 7039 }, { "epoch": 0.23, "learning_rate": 1.7581688238166768e-05, "loss": 1.5422, "step": 7040 }, { "epoch": 0.23, "learning_rate": 1.7580998464579585e-05, "loss": 1.4435, "step": 7041 }, { "epoch": 0.23, "learning_rate": 1.7580308606169473e-05, "loss": 1.5389, "step": 7042 }, { "epoch": 0.23, "learning_rate": 1.7579618662944152e-05, "loss": 1.5453, "step": 7043 }, { "epoch": 0.23, "learning_rate": 1.7578928634911345e-05, "loss": 1.5781, "step": 7044 }, { "epoch": 0.23, "learning_rate": 1.757823852207877e-05, "loss": 1.4636, "step": 7045 }, { "epoch": 0.23, "learning_rate": 1.7577548324454148e-05, "loss": 1.5994, "step": 7046 }, { "epoch": 0.23, "learning_rate": 1.75768580420452e-05, "loss": 1.4942, "step": 7047 }, { "epoch": 0.23, "learning_rate": 1.7576167674859656e-05, "loss": 1.4622, "step": 7048 }, { "epoch": 0.23, "learning_rate": 1.7575477222905233e-05, "loss": 1.6057, "step": 7049 }, { "epoch": 0.23, "learning_rate": 1.757478668618966e-05, "loss": 1.6046, "step": 7050 }, { "epoch": 0.24, "learning_rate": 1.7574096064720663e-05, "loss": 1.4114, "step": 7051 }, { "epoch": 0.24, "learning_rate": 1.7573405358505967e-05, "loss": 1.3601, "step": 7052 }, { "epoch": 0.24, "learning_rate": 1.7572714567553305e-05, "loss": 1.4182, "step": 7053 }, { "epoch": 0.24, "learning_rate": 1.7572023691870403e-05, "loss": 1.4565, "step": 7054 }, { "epoch": 0.24, "learning_rate": 1.757133273146499e-05, "loss": 1.4871, "step": 7055 }, { "epoch": 0.24, "learning_rate": 1.7570641686344803e-05, "loss": 1.5806, "step": 7056 }, { "epoch": 0.24, "learning_rate": 1.7569950556517566e-05, "loss": 1.4348, "step": 7057 }, { "epoch": 0.24, "learning_rate": 1.7569259341991015e-05, "loss": 1.5509, "step": 7058 }, { "epoch": 0.24, "learning_rate": 1.756856804277289e-05, "loss": 1.4779, "step": 7059 }, { "epoch": 0.24, "learning_rate": 1.7567876658870917e-05, "loss": 1.432, "step": 7060 }, { "epoch": 0.24, "learning_rate": 1.7567185190292832e-05, "loss": 1.5477, "step": 7061 }, { "epoch": 0.24, "learning_rate": 1.756649363704638e-05, "loss": 1.5161, "step": 7062 }, { "epoch": 0.24, "learning_rate": 1.7565801999139292e-05, "loss": 1.5166, "step": 7063 }, { "epoch": 0.24, "learning_rate": 1.7565110276579306e-05, "loss": 1.3836, "step": 7064 }, { "epoch": 0.24, "learning_rate": 1.7564418469374167e-05, "loss": 1.4855, "step": 7065 }, { "epoch": 0.24, "learning_rate": 1.756372657753161e-05, "loss": 1.4004, "step": 7066 }, { "epoch": 0.24, "learning_rate": 1.7563034601059384e-05, "loss": 1.4874, "step": 7067 }, { "epoch": 0.24, "learning_rate": 1.7562342539965223e-05, "loss": 1.4013, "step": 7068 }, { "epoch": 0.24, "learning_rate": 1.7561650394256874e-05, "loss": 1.486, "step": 7069 }, { "epoch": 0.24, "learning_rate": 1.756095816394208e-05, "loss": 1.5545, "step": 7070 }, { "epoch": 0.24, "learning_rate": 1.756026584902859e-05, "loss": 1.4391, "step": 7071 }, { "epoch": 0.24, "learning_rate": 1.755957344952415e-05, "loss": 1.4207, "step": 7072 }, { "epoch": 0.24, "learning_rate": 1.7558880965436502e-05, "loss": 1.5085, "step": 7073 }, { "epoch": 0.24, "learning_rate": 1.75581883967734e-05, "loss": 1.475, "step": 7074 }, { "epoch": 0.24, "learning_rate": 1.7557495743542586e-05, "loss": 1.377, "step": 7075 }, { "epoch": 0.24, "learning_rate": 1.7556803005751814e-05, "loss": 1.4453, "step": 7076 }, { "epoch": 0.24, "learning_rate": 1.7556110183408837e-05, "loss": 1.4844, "step": 7077 }, { "epoch": 0.24, "learning_rate": 1.7555417276521405e-05, "loss": 1.4963, "step": 7078 }, { "epoch": 0.24, "learning_rate": 1.7554724285097272e-05, "loss": 1.4383, "step": 7079 }, { "epoch": 0.24, "learning_rate": 1.7554031209144188e-05, "loss": 1.4523, "step": 7080 }, { "epoch": 0.24, "learning_rate": 1.755333804866991e-05, "loss": 1.5505, "step": 7081 }, { "epoch": 0.24, "learning_rate": 1.75526448036822e-05, "loss": 1.4065, "step": 7082 }, { "epoch": 0.24, "learning_rate": 1.7551951474188803e-05, "loss": 1.4897, "step": 7083 }, { "epoch": 0.24, "learning_rate": 1.7551258060197484e-05, "loss": 1.4551, "step": 7084 }, { "epoch": 0.24, "learning_rate": 1.7550564561716e-05, "loss": 1.5536, "step": 7085 }, { "epoch": 0.24, "learning_rate": 1.7549870978752114e-05, "loss": 1.4928, "step": 7086 }, { "epoch": 0.24, "learning_rate": 1.7549177311313575e-05, "loss": 1.5674, "step": 7087 }, { "epoch": 0.24, "learning_rate": 1.754848355940816e-05, "loss": 1.587, "step": 7088 }, { "epoch": 0.24, "learning_rate": 1.754778972304362e-05, "loss": 1.4436, "step": 7089 }, { "epoch": 0.24, "learning_rate": 1.7547095802227723e-05, "loss": 1.4758, "step": 7090 }, { "epoch": 0.24, "learning_rate": 1.754640179696823e-05, "loss": 1.6177, "step": 7091 }, { "epoch": 0.24, "learning_rate": 1.7545707707272907e-05, "loss": 1.589, "step": 7092 }, { "epoch": 0.24, "learning_rate": 1.7545013533149525e-05, "loss": 1.503, "step": 7093 }, { "epoch": 0.24, "learning_rate": 1.7544319274605845e-05, "loss": 1.6271, "step": 7094 }, { "epoch": 0.24, "learning_rate": 1.7543624931649636e-05, "loss": 1.4622, "step": 7095 }, { "epoch": 0.24, "learning_rate": 1.754293050428867e-05, "loss": 1.4926, "step": 7096 }, { "epoch": 0.24, "learning_rate": 1.7542235992530716e-05, "loss": 1.4786, "step": 7097 }, { "epoch": 0.24, "learning_rate": 1.7541541396383542e-05, "loss": 1.5091, "step": 7098 }, { "epoch": 0.24, "learning_rate": 1.7540846715854926e-05, "loss": 1.4173, "step": 7099 }, { "epoch": 0.24, "learning_rate": 1.7540151950952632e-05, "loss": 1.5326, "step": 7100 }, { "epoch": 0.24, "learning_rate": 1.7539457101684434e-05, "loss": 1.5104, "step": 7101 }, { "epoch": 0.24, "learning_rate": 1.7538762168058118e-05, "loss": 1.3728, "step": 7102 }, { "epoch": 0.24, "learning_rate": 1.753806715008145e-05, "loss": 1.4298, "step": 7103 }, { "epoch": 0.24, "learning_rate": 1.7537372047762205e-05, "loss": 1.3934, "step": 7104 }, { "epoch": 0.24, "learning_rate": 1.7536676861108167e-05, "loss": 1.4183, "step": 7105 }, { "epoch": 0.24, "learning_rate": 1.753598159012711e-05, "loss": 1.4962, "step": 7106 }, { "epoch": 0.24, "learning_rate": 1.7535286234826813e-05, "loss": 1.2898, "step": 7107 }, { "epoch": 0.24, "learning_rate": 1.7534590795215062e-05, "loss": 1.4414, "step": 7108 }, { "epoch": 0.24, "learning_rate": 1.7533895271299633e-05, "loss": 1.4251, "step": 7109 }, { "epoch": 0.24, "learning_rate": 1.753319966308831e-05, "loss": 1.4707, "step": 7110 }, { "epoch": 0.24, "learning_rate": 1.753250397058887e-05, "loss": 1.4843, "step": 7111 }, { "epoch": 0.24, "learning_rate": 1.7531808193809106e-05, "loss": 1.367, "step": 7112 }, { "epoch": 0.24, "learning_rate": 1.75311123327568e-05, "loss": 1.387, "step": 7113 }, { "epoch": 0.24, "learning_rate": 1.7530416387439736e-05, "loss": 1.3847, "step": 7114 }, { "epoch": 0.24, "learning_rate": 1.75297203578657e-05, "loss": 1.5542, "step": 7115 }, { "epoch": 0.24, "learning_rate": 1.7529024244042484e-05, "loss": 1.5007, "step": 7116 }, { "epoch": 0.24, "learning_rate": 1.7528328045977876e-05, "loss": 1.4942, "step": 7117 }, { "epoch": 0.24, "learning_rate": 1.7527631763679663e-05, "loss": 1.6103, "step": 7118 }, { "epoch": 0.24, "learning_rate": 1.752693539715564e-05, "loss": 1.6198, "step": 7119 }, { "epoch": 0.24, "learning_rate": 1.752623894641359e-05, "loss": 1.5936, "step": 7120 }, { "epoch": 0.24, "learning_rate": 1.7525542411461313e-05, "loss": 1.6201, "step": 7121 }, { "epoch": 0.24, "learning_rate": 1.75248457923066e-05, "loss": 1.5699, "step": 7122 }, { "epoch": 0.24, "learning_rate": 1.7524149088957244e-05, "loss": 1.5569, "step": 7123 }, { "epoch": 0.24, "learning_rate": 1.7523452301421046e-05, "loss": 1.6117, "step": 7124 }, { "epoch": 0.24, "learning_rate": 1.75227554297058e-05, "loss": 1.5019, "step": 7125 }, { "epoch": 0.24, "learning_rate": 1.7522058473819295e-05, "loss": 1.4904, "step": 7126 }, { "epoch": 0.24, "learning_rate": 1.7521361433769338e-05, "loss": 1.5108, "step": 7127 }, { "epoch": 0.24, "learning_rate": 1.7520664309563727e-05, "loss": 1.6095, "step": 7128 }, { "epoch": 0.24, "learning_rate": 1.7519967101210264e-05, "loss": 1.5036, "step": 7129 }, { "epoch": 0.24, "learning_rate": 1.751926980871674e-05, "loss": 1.5171, "step": 7130 }, { "epoch": 0.24, "learning_rate": 1.7518572432090966e-05, "loss": 1.5409, "step": 7131 }, { "epoch": 0.24, "learning_rate": 1.7517874971340747e-05, "loss": 1.6, "step": 7132 }, { "epoch": 0.24, "learning_rate": 1.751717742647388e-05, "loss": 1.5004, "step": 7133 }, { "epoch": 0.24, "learning_rate": 1.7516479797498172e-05, "loss": 1.4774, "step": 7134 }, { "epoch": 0.24, "learning_rate": 1.7515782084421426e-05, "loss": 1.6601, "step": 7135 }, { "epoch": 0.24, "learning_rate": 1.7515084287251455e-05, "loss": 1.4925, "step": 7136 }, { "epoch": 0.24, "learning_rate": 1.7514386405996066e-05, "loss": 1.4993, "step": 7137 }, { "epoch": 0.24, "learning_rate": 1.7513688440663062e-05, "loss": 1.4923, "step": 7138 }, { "epoch": 0.24, "learning_rate": 1.7512990391260252e-05, "loss": 1.4043, "step": 7139 }, { "epoch": 0.24, "learning_rate": 1.7512292257795453e-05, "loss": 1.3698, "step": 7140 }, { "epoch": 0.24, "learning_rate": 1.7511594040276468e-05, "loss": 1.5097, "step": 7141 }, { "epoch": 0.24, "learning_rate": 1.751089573871112e-05, "loss": 1.4632, "step": 7142 }, { "epoch": 0.24, "learning_rate": 1.7510197353107215e-05, "loss": 1.5041, "step": 7143 }, { "epoch": 0.24, "learning_rate": 1.7509498883472566e-05, "loss": 1.562, "step": 7144 }, { "epoch": 0.24, "learning_rate": 1.7508800329814993e-05, "loss": 1.5656, "step": 7145 }, { "epoch": 0.24, "learning_rate": 1.750810169214231e-05, "loss": 1.6185, "step": 7146 }, { "epoch": 0.24, "learning_rate": 1.750740297046233e-05, "loss": 1.5923, "step": 7147 }, { "epoch": 0.24, "learning_rate": 1.750670416478288e-05, "loss": 1.6129, "step": 7148 }, { "epoch": 0.24, "learning_rate": 1.750600527511177e-05, "loss": 1.4584, "step": 7149 }, { "epoch": 0.24, "learning_rate": 1.7505306301456823e-05, "loss": 1.4605, "step": 7150 }, { "epoch": 0.24, "learning_rate": 1.7504607243825858e-05, "loss": 1.4167, "step": 7151 }, { "epoch": 0.24, "learning_rate": 1.7503908102226704e-05, "loss": 1.4294, "step": 7152 }, { "epoch": 0.24, "learning_rate": 1.7503208876667175e-05, "loss": 1.4256, "step": 7153 }, { "epoch": 0.24, "learning_rate": 1.7502509567155096e-05, "loss": 1.4483, "step": 7154 }, { "epoch": 0.24, "learning_rate": 1.75018101736983e-05, "loss": 1.63, "step": 7155 }, { "epoch": 0.24, "learning_rate": 1.7501110696304598e-05, "loss": 1.5511, "step": 7156 }, { "epoch": 0.24, "learning_rate": 1.7500411134981825e-05, "loss": 1.5462, "step": 7157 }, { "epoch": 0.24, "learning_rate": 1.749971148973781e-05, "loss": 1.5403, "step": 7158 }, { "epoch": 0.24, "learning_rate": 1.7499011760580377e-05, "loss": 1.4631, "step": 7159 }, { "epoch": 0.24, "learning_rate": 1.749831194751736e-05, "loss": 1.4542, "step": 7160 }, { "epoch": 0.24, "learning_rate": 1.7497612050556584e-05, "loss": 1.351, "step": 7161 }, { "epoch": 0.24, "learning_rate": 1.749691206970588e-05, "loss": 1.4897, "step": 7162 }, { "epoch": 0.24, "learning_rate": 1.7496212004973086e-05, "loss": 1.3433, "step": 7163 }, { "epoch": 0.24, "learning_rate": 1.7495511856366026e-05, "loss": 1.4246, "step": 7164 }, { "epoch": 0.24, "learning_rate": 1.7494811623892543e-05, "loss": 1.4894, "step": 7165 }, { "epoch": 0.24, "learning_rate": 1.7494111307560464e-05, "loss": 1.594, "step": 7166 }, { "epoch": 0.24, "learning_rate": 1.749341090737763e-05, "loss": 1.4064, "step": 7167 }, { "epoch": 0.24, "learning_rate": 1.7492710423351878e-05, "loss": 1.5058, "step": 7168 }, { "epoch": 0.24, "learning_rate": 1.7492009855491045e-05, "loss": 1.5553, "step": 7169 }, { "epoch": 0.24, "learning_rate": 1.7491309203802965e-05, "loss": 1.5471, "step": 7170 }, { "epoch": 0.24, "learning_rate": 1.749060846829548e-05, "loss": 1.5994, "step": 7171 }, { "epoch": 0.24, "learning_rate": 1.7489907648976433e-05, "loss": 1.6936, "step": 7172 }, { "epoch": 0.24, "learning_rate": 1.7489206745853664e-05, "loss": 1.6728, "step": 7173 }, { "epoch": 0.24, "learning_rate": 1.7488505758935016e-05, "loss": 1.5904, "step": 7174 }, { "epoch": 0.24, "learning_rate": 1.7487804688228327e-05, "loss": 1.6558, "step": 7175 }, { "epoch": 0.24, "learning_rate": 1.7487103533741446e-05, "loss": 1.6851, "step": 7176 }, { "epoch": 0.24, "learning_rate": 1.7486402295482224e-05, "loss": 1.6621, "step": 7177 }, { "epoch": 0.24, "learning_rate": 1.7485700973458494e-05, "loss": 1.5244, "step": 7178 }, { "epoch": 0.24, "learning_rate": 1.748499956767811e-05, "loss": 1.5683, "step": 7179 }, { "epoch": 0.24, "learning_rate": 1.7484298078148926e-05, "loss": 1.5535, "step": 7180 }, { "epoch": 0.24, "learning_rate": 1.7483596504878778e-05, "loss": 1.409, "step": 7181 }, { "epoch": 0.24, "learning_rate": 1.7482894847875527e-05, "loss": 1.4763, "step": 7182 }, { "epoch": 0.24, "learning_rate": 1.7482193107147012e-05, "loss": 1.5385, "step": 7183 }, { "epoch": 0.24, "learning_rate": 1.74814912827011e-05, "loss": 1.5689, "step": 7184 }, { "epoch": 0.24, "learning_rate": 1.7480789374545633e-05, "loss": 1.5514, "step": 7185 }, { "epoch": 0.24, "learning_rate": 1.7480087382688466e-05, "loss": 1.5787, "step": 7186 }, { "epoch": 0.24, "learning_rate": 1.7479385307137457e-05, "loss": 1.4972, "step": 7187 }, { "epoch": 0.24, "learning_rate": 1.7478683147900453e-05, "loss": 1.4788, "step": 7188 }, { "epoch": 0.24, "learning_rate": 1.747798090498532e-05, "loss": 1.5769, "step": 7189 }, { "epoch": 0.24, "learning_rate": 1.7477278578399913e-05, "loss": 1.5438, "step": 7190 }, { "epoch": 0.24, "learning_rate": 1.7476576168152085e-05, "loss": 1.4999, "step": 7191 }, { "epoch": 0.24, "learning_rate": 1.7475873674249704e-05, "loss": 1.5159, "step": 7192 }, { "epoch": 0.24, "learning_rate": 1.7475171096700624e-05, "loss": 1.6306, "step": 7193 }, { "epoch": 0.24, "learning_rate": 1.7474468435512702e-05, "loss": 1.5987, "step": 7194 }, { "epoch": 0.24, "learning_rate": 1.7473765690693812e-05, "loss": 1.5369, "step": 7195 }, { "epoch": 0.24, "learning_rate": 1.7473062862251805e-05, "loss": 1.645, "step": 7196 }, { "epoch": 0.24, "learning_rate": 1.7472359950194554e-05, "loss": 1.436, "step": 7197 }, { "epoch": 0.24, "learning_rate": 1.747165695452992e-05, "loss": 1.5166, "step": 7198 }, { "epoch": 0.24, "learning_rate": 1.7470953875265766e-05, "loss": 1.5617, "step": 7199 }, { "epoch": 0.24, "learning_rate": 1.7470250712409963e-05, "loss": 1.5839, "step": 7200 }, { "epoch": 0.24, "learning_rate": 1.7469547465970373e-05, "loss": 1.5895, "step": 7201 }, { "epoch": 0.24, "learning_rate": 1.7468844135954875e-05, "loss": 1.5492, "step": 7202 }, { "epoch": 0.24, "learning_rate": 1.7468140722371326e-05, "loss": 1.5073, "step": 7203 }, { "epoch": 0.24, "learning_rate": 1.7467437225227605e-05, "loss": 1.6159, "step": 7204 }, { "epoch": 0.24, "learning_rate": 1.7466733644531578e-05, "loss": 1.5373, "step": 7205 }, { "epoch": 0.24, "learning_rate": 1.7466029980291126e-05, "loss": 1.6069, "step": 7206 }, { "epoch": 0.24, "learning_rate": 1.7465326232514113e-05, "loss": 1.7046, "step": 7207 }, { "epoch": 0.24, "learning_rate": 1.7464622401208417e-05, "loss": 1.55, "step": 7208 }, { "epoch": 0.24, "learning_rate": 1.746391848638191e-05, "loss": 1.503, "step": 7209 }, { "epoch": 0.24, "learning_rate": 1.7463214488042472e-05, "loss": 1.5163, "step": 7210 }, { "epoch": 0.24, "learning_rate": 1.746251040619798e-05, "loss": 1.5079, "step": 7211 }, { "epoch": 0.24, "learning_rate": 1.7461806240856308e-05, "loss": 1.5644, "step": 7212 }, { "epoch": 0.24, "learning_rate": 1.7461101992025334e-05, "loss": 1.5874, "step": 7213 }, { "epoch": 0.24, "learning_rate": 1.7460397659712947e-05, "loss": 1.5898, "step": 7214 }, { "epoch": 0.24, "learning_rate": 1.745969324392702e-05, "loss": 1.6012, "step": 7215 }, { "epoch": 0.24, "learning_rate": 1.7458988744675433e-05, "loss": 1.5418, "step": 7216 }, { "epoch": 0.24, "learning_rate": 1.7458284161966074e-05, "loss": 1.4123, "step": 7217 }, { "epoch": 0.24, "learning_rate": 1.7457579495806825e-05, "loss": 1.5681, "step": 7218 }, { "epoch": 0.24, "learning_rate": 1.7456874746205568e-05, "loss": 1.408, "step": 7219 }, { "epoch": 0.24, "learning_rate": 1.7456169913170194e-05, "loss": 1.4302, "step": 7220 }, { "epoch": 0.24, "learning_rate": 1.7455464996708583e-05, "loss": 1.3776, "step": 7221 }, { "epoch": 0.24, "learning_rate": 1.7454759996828622e-05, "loss": 1.3485, "step": 7222 }, { "epoch": 0.24, "learning_rate": 1.7454054913538207e-05, "loss": 1.4493, "step": 7223 }, { "epoch": 0.24, "learning_rate": 1.7453349746845218e-05, "loss": 1.3852, "step": 7224 }, { "epoch": 0.24, "learning_rate": 1.745264449675755e-05, "loss": 1.4363, "step": 7225 }, { "epoch": 0.24, "learning_rate": 1.745193916328309e-05, "loss": 1.5098, "step": 7226 }, { "epoch": 0.24, "learning_rate": 1.7451233746429736e-05, "loss": 1.4855, "step": 7227 }, { "epoch": 0.24, "learning_rate": 1.7450528246205374e-05, "loss": 1.3535, "step": 7228 }, { "epoch": 0.24, "learning_rate": 1.7449822662617907e-05, "loss": 1.3846, "step": 7229 }, { "epoch": 0.24, "learning_rate": 1.7449116995675218e-05, "loss": 1.4123, "step": 7230 }, { "epoch": 0.24, "learning_rate": 1.7448411245385215e-05, "loss": 1.5276, "step": 7231 }, { "epoch": 0.24, "learning_rate": 1.7447705411755783e-05, "loss": 1.4913, "step": 7232 }, { "epoch": 0.24, "learning_rate": 1.744699949479483e-05, "loss": 1.4302, "step": 7233 }, { "epoch": 0.24, "learning_rate": 1.7446293494510246e-05, "loss": 1.5249, "step": 7234 }, { "epoch": 0.24, "learning_rate": 1.7445587410909932e-05, "loss": 1.4139, "step": 7235 }, { "epoch": 0.24, "learning_rate": 1.7444881244001787e-05, "loss": 1.3777, "step": 7236 }, { "epoch": 0.24, "learning_rate": 1.744417499379372e-05, "loss": 1.3951, "step": 7237 }, { "epoch": 0.24, "learning_rate": 1.7443468660293627e-05, "loss": 1.4231, "step": 7238 }, { "epoch": 0.24, "learning_rate": 1.7442762243509414e-05, "loss": 1.413, "step": 7239 }, { "epoch": 0.24, "learning_rate": 1.744205574344898e-05, "loss": 1.435, "step": 7240 }, { "epoch": 0.24, "learning_rate": 1.7441349160120234e-05, "loss": 1.3777, "step": 7241 }, { "epoch": 0.24, "learning_rate": 1.7440642493531084e-05, "loss": 1.3558, "step": 7242 }, { "epoch": 0.24, "learning_rate": 1.743993574368943e-05, "loss": 1.3723, "step": 7243 }, { "epoch": 0.24, "learning_rate": 1.7439228910603184e-05, "loss": 1.5164, "step": 7244 }, { "epoch": 0.24, "learning_rate": 1.7438521994280257e-05, "loss": 1.4122, "step": 7245 }, { "epoch": 0.24, "learning_rate": 1.7437814994728553e-05, "loss": 1.4196, "step": 7246 }, { "epoch": 0.24, "learning_rate": 1.7437107911955992e-05, "loss": 1.3904, "step": 7247 }, { "epoch": 0.24, "learning_rate": 1.743640074597047e-05, "loss": 1.4438, "step": 7248 }, { "epoch": 0.24, "learning_rate": 1.743569349677991e-05, "loss": 1.5058, "step": 7249 }, { "epoch": 0.24, "learning_rate": 1.743498616439223e-05, "loss": 1.3905, "step": 7250 }, { "epoch": 0.24, "learning_rate": 1.7434278748815333e-05, "loss": 1.4532, "step": 7251 }, { "epoch": 0.24, "learning_rate": 1.743357125005714e-05, "loss": 1.4302, "step": 7252 }, { "epoch": 0.24, "learning_rate": 1.7432863668125564e-05, "loss": 1.3733, "step": 7253 }, { "epoch": 0.24, "learning_rate": 1.7432156003028528e-05, "loss": 1.4505, "step": 7254 }, { "epoch": 0.24, "learning_rate": 1.7431448254773943e-05, "loss": 1.4144, "step": 7255 }, { "epoch": 0.24, "learning_rate": 1.7430740423369735e-05, "loss": 1.3638, "step": 7256 }, { "epoch": 0.24, "learning_rate": 1.7430032508823816e-05, "loss": 1.4143, "step": 7257 }, { "epoch": 0.24, "learning_rate": 1.742932451114411e-05, "loss": 1.4416, "step": 7258 }, { "epoch": 0.24, "learning_rate": 1.7428616430338546e-05, "loss": 1.439, "step": 7259 }, { "epoch": 0.24, "learning_rate": 1.7427908266415035e-05, "loss": 1.4715, "step": 7260 }, { "epoch": 0.24, "learning_rate": 1.7427200019381504e-05, "loss": 1.4337, "step": 7261 }, { "epoch": 0.24, "learning_rate": 1.7426491689245883e-05, "loss": 1.5174, "step": 7262 }, { "epoch": 0.24, "learning_rate": 1.7425783276016096e-05, "loss": 1.5387, "step": 7263 }, { "epoch": 0.24, "learning_rate": 1.742507477970006e-05, "loss": 1.3909, "step": 7264 }, { "epoch": 0.24, "learning_rate": 1.7424366200305717e-05, "loss": 1.4432, "step": 7265 }, { "epoch": 0.24, "learning_rate": 1.7423657537840978e-05, "loss": 1.3299, "step": 7266 }, { "epoch": 0.24, "learning_rate": 1.742294879231379e-05, "loss": 1.3413, "step": 7267 }, { "epoch": 0.24, "learning_rate": 1.7422239963732072e-05, "loss": 1.4338, "step": 7268 }, { "epoch": 0.24, "learning_rate": 1.742153105210376e-05, "loss": 1.4086, "step": 7269 }, { "epoch": 0.24, "learning_rate": 1.7420822057436777e-05, "loss": 1.399, "step": 7270 }, { "epoch": 0.24, "learning_rate": 1.742011297973907e-05, "loss": 1.411, "step": 7271 }, { "epoch": 0.24, "learning_rate": 1.7419403819018555e-05, "loss": 1.3943, "step": 7272 }, { "epoch": 0.24, "learning_rate": 1.7418694575283185e-05, "loss": 1.4456, "step": 7273 }, { "epoch": 0.24, "learning_rate": 1.7417985248540887e-05, "loss": 1.4767, "step": 7274 }, { "epoch": 0.24, "learning_rate": 1.7417275838799596e-05, "loss": 1.4487, "step": 7275 }, { "epoch": 0.24, "learning_rate": 1.7416566346067253e-05, "loss": 1.5007, "step": 7276 }, { "epoch": 0.24, "learning_rate": 1.7415856770351797e-05, "loss": 1.4776, "step": 7277 }, { "epoch": 0.24, "learning_rate": 1.7415147111661162e-05, "loss": 1.5845, "step": 7278 }, { "epoch": 0.24, "learning_rate": 1.7414437370003295e-05, "loss": 1.4606, "step": 7279 }, { "epoch": 0.24, "learning_rate": 1.7413727545386134e-05, "loss": 1.5395, "step": 7280 }, { "epoch": 0.24, "learning_rate": 1.741301763781762e-05, "loss": 1.5614, "step": 7281 }, { "epoch": 0.24, "learning_rate": 1.7412307647305694e-05, "loss": 1.5119, "step": 7282 }, { "epoch": 0.24, "learning_rate": 1.7411597573858306e-05, "loss": 1.4796, "step": 7283 }, { "epoch": 0.24, "learning_rate": 1.74108874174834e-05, "loss": 1.616, "step": 7284 }, { "epoch": 0.24, "learning_rate": 1.7410177178188917e-05, "loss": 1.4725, "step": 7285 }, { "epoch": 0.24, "learning_rate": 1.740946685598281e-05, "loss": 1.6023, "step": 7286 }, { "epoch": 0.24, "learning_rate": 1.740875645087302e-05, "loss": 1.6603, "step": 7287 }, { "epoch": 0.24, "learning_rate": 1.74080459628675e-05, "loss": 1.6198, "step": 7288 }, { "epoch": 0.24, "learning_rate": 1.7407335391974204e-05, "loss": 1.7509, "step": 7289 }, { "epoch": 0.24, "learning_rate": 1.740662473820107e-05, "loss": 1.7988, "step": 7290 }, { "epoch": 0.24, "learning_rate": 1.7405914001556058e-05, "loss": 1.6363, "step": 7291 }, { "epoch": 0.24, "learning_rate": 1.7405203182047122e-05, "loss": 1.6082, "step": 7292 }, { "epoch": 0.24, "learning_rate": 1.7404492279682213e-05, "loss": 1.5692, "step": 7293 }, { "epoch": 0.24, "learning_rate": 1.7403781294469282e-05, "loss": 1.6088, "step": 7294 }, { "epoch": 0.24, "learning_rate": 1.740307022641629e-05, "loss": 1.6466, "step": 7295 }, { "epoch": 0.24, "learning_rate": 1.7402359075531184e-05, "loss": 1.6384, "step": 7296 }, { "epoch": 0.24, "learning_rate": 1.7401647841821928e-05, "loss": 1.6329, "step": 7297 }, { "epoch": 0.24, "learning_rate": 1.740093652529648e-05, "loss": 1.516, "step": 7298 }, { "epoch": 0.24, "learning_rate": 1.7400225125962796e-05, "loss": 1.5084, "step": 7299 }, { "epoch": 0.24, "learning_rate": 1.739951364382884e-05, "loss": 1.5797, "step": 7300 }, { "epoch": 0.24, "learning_rate": 1.7398802078902568e-05, "loss": 1.4714, "step": 7301 }, { "epoch": 0.24, "learning_rate": 1.7398090431191945e-05, "loss": 1.527, "step": 7302 }, { "epoch": 0.24, "learning_rate": 1.739737870070493e-05, "loss": 1.5636, "step": 7303 }, { "epoch": 0.24, "learning_rate": 1.739666688744949e-05, "loss": 1.5075, "step": 7304 }, { "epoch": 0.24, "learning_rate": 1.7395954991433588e-05, "loss": 1.5127, "step": 7305 }, { "epoch": 0.24, "learning_rate": 1.739524301266519e-05, "loss": 1.5578, "step": 7306 }, { "epoch": 0.24, "learning_rate": 1.7394530951152256e-05, "loss": 1.4833, "step": 7307 }, { "epoch": 0.24, "learning_rate": 1.7393818806902768e-05, "loss": 1.5041, "step": 7308 }, { "epoch": 0.24, "learning_rate": 1.7393106579924676e-05, "loss": 1.5426, "step": 7309 }, { "epoch": 0.24, "learning_rate": 1.739239427022596e-05, "loss": 1.4956, "step": 7310 }, { "epoch": 0.24, "learning_rate": 1.739168187781459e-05, "loss": 1.5919, "step": 7311 }, { "epoch": 0.24, "learning_rate": 1.7390969402698537e-05, "loss": 1.6292, "step": 7312 }, { "epoch": 0.24, "learning_rate": 1.7390256844885764e-05, "loss": 1.5112, "step": 7313 }, { "epoch": 0.24, "learning_rate": 1.7389544204384253e-05, "loss": 1.4472, "step": 7314 }, { "epoch": 0.24, "learning_rate": 1.738883148120198e-05, "loss": 1.6402, "step": 7315 }, { "epoch": 0.24, "learning_rate": 1.7388118675346907e-05, "loss": 1.5331, "step": 7316 }, { "epoch": 0.24, "learning_rate": 1.7387405786827017e-05, "loss": 1.5003, "step": 7317 }, { "epoch": 0.24, "learning_rate": 1.738669281565029e-05, "loss": 1.568, "step": 7318 }, { "epoch": 0.24, "learning_rate": 1.7385979761824697e-05, "loss": 1.4178, "step": 7319 }, { "epoch": 0.24, "learning_rate": 1.7385266625358223e-05, "loss": 1.5614, "step": 7320 }, { "epoch": 0.24, "learning_rate": 1.7384553406258842e-05, "loss": 1.4224, "step": 7321 }, { "epoch": 0.24, "learning_rate": 1.7383840104534532e-05, "loss": 1.5968, "step": 7322 }, { "epoch": 0.24, "learning_rate": 1.738312672019328e-05, "loss": 1.504, "step": 7323 }, { "epoch": 0.24, "learning_rate": 1.738241325324306e-05, "loss": 1.5211, "step": 7324 }, { "epoch": 0.24, "learning_rate": 1.7381699703691866e-05, "loss": 1.5124, "step": 7325 }, { "epoch": 0.24, "learning_rate": 1.7380986071547677e-05, "loss": 1.5439, "step": 7326 }, { "epoch": 0.24, "learning_rate": 1.7380272356818474e-05, "loss": 1.6589, "step": 7327 }, { "epoch": 0.24, "learning_rate": 1.7379558559512245e-05, "loss": 1.5695, "step": 7328 }, { "epoch": 0.24, "learning_rate": 1.7378844679636978e-05, "loss": 1.4872, "step": 7329 }, { "epoch": 0.24, "learning_rate": 1.737813071720066e-05, "loss": 1.4727, "step": 7330 }, { "epoch": 0.24, "learning_rate": 1.7377416672211277e-05, "loss": 1.4735, "step": 7331 }, { "epoch": 0.24, "learning_rate": 1.7376702544676823e-05, "loss": 1.481, "step": 7332 }, { "epoch": 0.24, "learning_rate": 1.7375988334605283e-05, "loss": 1.4633, "step": 7333 }, { "epoch": 0.24, "learning_rate": 1.7375274042004654e-05, "loss": 1.4224, "step": 7334 }, { "epoch": 0.24, "learning_rate": 1.7374559666882923e-05, "loss": 1.4131, "step": 7335 }, { "epoch": 0.24, "learning_rate": 1.7373845209248088e-05, "loss": 1.4079, "step": 7336 }, { "epoch": 0.24, "learning_rate": 1.7373130669108137e-05, "loss": 1.5585, "step": 7337 }, { "epoch": 0.24, "learning_rate": 1.7372416046471067e-05, "loss": 1.4279, "step": 7338 }, { "epoch": 0.24, "learning_rate": 1.7371701341344877e-05, "loss": 1.5139, "step": 7339 }, { "epoch": 0.24, "learning_rate": 1.7370986553737562e-05, "loss": 1.4823, "step": 7340 }, { "epoch": 0.24, "learning_rate": 1.7370271683657117e-05, "loss": 1.4662, "step": 7341 }, { "epoch": 0.24, "learning_rate": 1.7369556731111545e-05, "loss": 1.5241, "step": 7342 }, { "epoch": 0.24, "learning_rate": 1.736884169610884e-05, "loss": 1.5915, "step": 7343 }, { "epoch": 0.24, "learning_rate": 1.736812657865701e-05, "loss": 1.5058, "step": 7344 }, { "epoch": 0.24, "learning_rate": 1.736741137876405e-05, "loss": 1.6658, "step": 7345 }, { "epoch": 0.24, "learning_rate": 1.7366696096437964e-05, "loss": 1.4795, "step": 7346 }, { "epoch": 0.24, "learning_rate": 1.7365980731686756e-05, "loss": 1.5439, "step": 7347 }, { "epoch": 0.24, "learning_rate": 1.7365265284518434e-05, "loss": 1.5543, "step": 7348 }, { "epoch": 0.24, "learning_rate": 1.7364549754940995e-05, "loss": 1.5194, "step": 7349 }, { "epoch": 0.24, "learning_rate": 1.736383414296245e-05, "loss": 1.4759, "step": 7350 }, { "epoch": 0.25, "learning_rate": 1.7363118448590803e-05, "loss": 1.6392, "step": 7351 }, { "epoch": 0.25, "learning_rate": 1.7362402671834062e-05, "loss": 1.5503, "step": 7352 }, { "epoch": 0.25, "learning_rate": 1.7361686812700244e-05, "loss": 1.5167, "step": 7353 }, { "epoch": 0.25, "learning_rate": 1.7360970871197347e-05, "loss": 1.5578, "step": 7354 }, { "epoch": 0.25, "learning_rate": 1.7360254847333388e-05, "loss": 1.5352, "step": 7355 }, { "epoch": 0.25, "learning_rate": 1.7359538741116372e-05, "loss": 1.5715, "step": 7356 }, { "epoch": 0.25, "learning_rate": 1.735882255255432e-05, "loss": 1.561, "step": 7357 }, { "epoch": 0.25, "learning_rate": 1.7358106281655245e-05, "loss": 1.5188, "step": 7358 }, { "epoch": 0.25, "learning_rate": 1.7357389928427156e-05, "loss": 1.5128, "step": 7359 }, { "epoch": 0.25, "learning_rate": 1.7356673492878073e-05, "loss": 1.5066, "step": 7360 }, { "epoch": 0.25, "learning_rate": 1.7355956975016005e-05, "loss": 1.364, "step": 7361 }, { "epoch": 0.25, "learning_rate": 1.7355240374848975e-05, "loss": 1.4047, "step": 7362 }, { "epoch": 0.25, "learning_rate": 1.7354523692385e-05, "loss": 1.4294, "step": 7363 }, { "epoch": 0.25, "learning_rate": 1.73538069276321e-05, "loss": 1.5272, "step": 7364 }, { "epoch": 0.25, "learning_rate": 1.735309008059829e-05, "loss": 1.5951, "step": 7365 }, { "epoch": 0.25, "learning_rate": 1.7352373151291593e-05, "loss": 1.5884, "step": 7366 }, { "epoch": 0.25, "learning_rate": 1.7351656139720034e-05, "loss": 1.487, "step": 7367 }, { "epoch": 0.25, "learning_rate": 1.7350939045891632e-05, "loss": 1.4238, "step": 7368 }, { "epoch": 0.25, "learning_rate": 1.7350221869814413e-05, "loss": 1.441, "step": 7369 }, { "epoch": 0.25, "learning_rate": 1.73495046114964e-05, "loss": 1.5316, "step": 7370 }, { "epoch": 0.25, "learning_rate": 1.7348787270945613e-05, "loss": 1.3788, "step": 7371 }, { "epoch": 0.25, "learning_rate": 1.734806984817009e-05, "loss": 1.4965, "step": 7372 }, { "epoch": 0.25, "learning_rate": 1.734735234317785e-05, "loss": 1.5074, "step": 7373 }, { "epoch": 0.25, "learning_rate": 1.734663475597692e-05, "loss": 1.5523, "step": 7374 }, { "epoch": 0.25, "learning_rate": 1.734591708657533e-05, "loss": 1.5479, "step": 7375 }, { "epoch": 0.25, "learning_rate": 1.734519933498112e-05, "loss": 1.4655, "step": 7376 }, { "epoch": 0.25, "learning_rate": 1.7344481501202305e-05, "loss": 1.6595, "step": 7377 }, { "epoch": 0.25, "learning_rate": 1.734376358524693e-05, "loss": 1.5376, "step": 7378 }, { "epoch": 0.25, "learning_rate": 1.7343045587123018e-05, "loss": 1.5286, "step": 7379 }, { "epoch": 0.25, "learning_rate": 1.734232750683861e-05, "loss": 1.6168, "step": 7380 }, { "epoch": 0.25, "learning_rate": 1.734160934440173e-05, "loss": 1.3826, "step": 7381 }, { "epoch": 0.25, "learning_rate": 1.7340891099820428e-05, "loss": 1.5243, "step": 7382 }, { "epoch": 0.25, "learning_rate": 1.734017277310273e-05, "loss": 1.5375, "step": 7383 }, { "epoch": 0.25, "learning_rate": 1.7339454364256676e-05, "loss": 1.602, "step": 7384 }, { "epoch": 0.25, "learning_rate": 1.7338735873290304e-05, "loss": 1.6099, "step": 7385 }, { "epoch": 0.25, "learning_rate": 1.7338017300211654e-05, "loss": 1.5868, "step": 7386 }, { "epoch": 0.25, "learning_rate": 1.7337298645028764e-05, "loss": 1.6472, "step": 7387 }, { "epoch": 0.25, "learning_rate": 1.733657990774968e-05, "loss": 1.5696, "step": 7388 }, { "epoch": 0.25, "learning_rate": 1.7335861088382438e-05, "loss": 1.5704, "step": 7389 }, { "epoch": 0.25, "learning_rate": 1.7335142186935083e-05, "loss": 1.5882, "step": 7390 }, { "epoch": 0.25, "learning_rate": 1.7334423203415658e-05, "loss": 1.5475, "step": 7391 }, { "epoch": 0.25, "learning_rate": 1.733370413783221e-05, "loss": 1.556, "step": 7392 }, { "epoch": 0.25, "learning_rate": 1.7332984990192782e-05, "loss": 1.4371, "step": 7393 }, { "epoch": 0.25, "learning_rate": 1.733226576050542e-05, "loss": 1.5232, "step": 7394 }, { "epoch": 0.25, "learning_rate": 1.7331546448778175e-05, "loss": 1.4481, "step": 7395 }, { "epoch": 0.25, "learning_rate": 1.7330827055019094e-05, "loss": 1.4239, "step": 7396 }, { "epoch": 0.25, "learning_rate": 1.7330107579236225e-05, "loss": 1.5, "step": 7397 }, { "epoch": 0.25, "learning_rate": 1.7329388021437615e-05, "loss": 1.5113, "step": 7398 }, { "epoch": 0.25, "learning_rate": 1.732866838163132e-05, "loss": 1.5114, "step": 7399 }, { "epoch": 0.25, "learning_rate": 1.732794865982539e-05, "loss": 1.5169, "step": 7400 }, { "epoch": 0.25, "learning_rate": 1.732722885602788e-05, "loss": 1.5263, "step": 7401 }, { "epoch": 0.25, "learning_rate": 1.732650897024684e-05, "loss": 1.4715, "step": 7402 }, { "epoch": 0.25, "learning_rate": 1.732578900249033e-05, "loss": 1.6334, "step": 7403 }, { "epoch": 0.25, "learning_rate": 1.7325068952766403e-05, "loss": 1.5803, "step": 7404 }, { "epoch": 0.25, "learning_rate": 1.732434882108311e-05, "loss": 1.5926, "step": 7405 }, { "epoch": 0.25, "learning_rate": 1.7323628607448517e-05, "loss": 1.5014, "step": 7406 }, { "epoch": 0.25, "learning_rate": 1.732290831187068e-05, "loss": 1.5359, "step": 7407 }, { "epoch": 0.25, "learning_rate": 1.7322187934357652e-05, "loss": 1.4937, "step": 7408 }, { "epoch": 0.25, "learning_rate": 1.7321467474917502e-05, "loss": 1.5323, "step": 7409 }, { "epoch": 0.25, "learning_rate": 1.732074693355829e-05, "loss": 1.5018, "step": 7410 }, { "epoch": 0.25, "learning_rate": 1.7320026310288072e-05, "loss": 1.4222, "step": 7411 }, { "epoch": 0.25, "learning_rate": 1.7319305605114916e-05, "loss": 1.4336, "step": 7412 }, { "epoch": 0.25, "learning_rate": 1.7318584818046884e-05, "loss": 1.5264, "step": 7413 }, { "epoch": 0.25, "learning_rate": 1.7317863949092043e-05, "loss": 1.4808, "step": 7414 }, { "epoch": 0.25, "learning_rate": 1.7317142998258454e-05, "loss": 1.4776, "step": 7415 }, { "epoch": 0.25, "learning_rate": 1.7316421965554188e-05, "loss": 1.4252, "step": 7416 }, { "epoch": 0.25, "learning_rate": 1.731570085098731e-05, "loss": 1.3427, "step": 7417 }, { "epoch": 0.25, "learning_rate": 1.731497965456589e-05, "loss": 1.5015, "step": 7418 }, { "epoch": 0.25, "learning_rate": 1.7314258376298e-05, "loss": 1.3137, "step": 7419 }, { "epoch": 0.25, "learning_rate": 1.7313537016191706e-05, "loss": 1.5758, "step": 7420 }, { "epoch": 0.25, "learning_rate": 1.7312815574255083e-05, "loss": 1.4566, "step": 7421 }, { "epoch": 0.25, "learning_rate": 1.7312094050496194e-05, "loss": 1.5388, "step": 7422 }, { "epoch": 0.25, "learning_rate": 1.7311372444923122e-05, "loss": 1.4697, "step": 7423 }, { "epoch": 0.25, "learning_rate": 1.731065075754394e-05, "loss": 1.4522, "step": 7424 }, { "epoch": 0.25, "learning_rate": 1.730992898836672e-05, "loss": 1.4112, "step": 7425 }, { "epoch": 0.25, "learning_rate": 1.7309207137399535e-05, "loss": 1.4444, "step": 7426 }, { "epoch": 0.25, "learning_rate": 1.7308485204650466e-05, "loss": 1.4542, "step": 7427 }, { "epoch": 0.25, "learning_rate": 1.730776319012759e-05, "loss": 1.3492, "step": 7428 }, { "epoch": 0.25, "learning_rate": 1.7307041093838986e-05, "loss": 1.501, "step": 7429 }, { "epoch": 0.25, "learning_rate": 1.7306318915792734e-05, "loss": 1.4841, "step": 7430 }, { "epoch": 0.25, "learning_rate": 1.7305596655996916e-05, "loss": 1.4992, "step": 7431 }, { "epoch": 0.25, "learning_rate": 1.7304874314459603e-05, "loss": 1.4888, "step": 7432 }, { "epoch": 0.25, "learning_rate": 1.730415189118889e-05, "loss": 1.5788, "step": 7433 }, { "epoch": 0.25, "learning_rate": 1.7303429386192848e-05, "loss": 1.4775, "step": 7434 }, { "epoch": 0.25, "learning_rate": 1.7302706799479575e-05, "loss": 1.5287, "step": 7435 }, { "epoch": 0.25, "learning_rate": 1.7301984131057147e-05, "loss": 1.6026, "step": 7436 }, { "epoch": 0.25, "learning_rate": 1.7301261380933647e-05, "loss": 1.4973, "step": 7437 }, { "epoch": 0.25, "learning_rate": 1.730053854911717e-05, "loss": 1.5037, "step": 7438 }, { "epoch": 0.25, "learning_rate": 1.72998156356158e-05, "loss": 1.5679, "step": 7439 }, { "epoch": 0.25, "learning_rate": 1.7299092640437627e-05, "loss": 1.4909, "step": 7440 }, { "epoch": 0.25, "learning_rate": 1.7298369563590734e-05, "loss": 1.4995, "step": 7441 }, { "epoch": 0.25, "learning_rate": 1.729764640508322e-05, "loss": 1.4601, "step": 7442 }, { "epoch": 0.25, "learning_rate": 1.7296923164923173e-05, "loss": 1.4156, "step": 7443 }, { "epoch": 0.25, "learning_rate": 1.7296199843118685e-05, "loss": 1.503, "step": 7444 }, { "epoch": 0.25, "learning_rate": 1.7295476439677847e-05, "loss": 1.5119, "step": 7445 }, { "epoch": 0.25, "learning_rate": 1.729475295460876e-05, "loss": 1.4088, "step": 7446 }, { "epoch": 0.25, "learning_rate": 1.7294029387919508e-05, "loss": 1.4381, "step": 7447 }, { "epoch": 0.25, "learning_rate": 1.7293305739618195e-05, "loss": 1.4056, "step": 7448 }, { "epoch": 0.25, "learning_rate": 1.729258200971292e-05, "loss": 1.3882, "step": 7449 }, { "epoch": 0.25, "learning_rate": 1.7291858198211772e-05, "loss": 1.5075, "step": 7450 }, { "epoch": 0.25, "learning_rate": 1.7291134305122858e-05, "loss": 1.461, "step": 7451 }, { "epoch": 0.25, "learning_rate": 1.7290410330454272e-05, "loss": 1.5899, "step": 7452 }, { "epoch": 0.25, "learning_rate": 1.7289686274214116e-05, "loss": 1.4316, "step": 7453 }, { "epoch": 0.25, "learning_rate": 1.728896213641049e-05, "loss": 1.5111, "step": 7454 }, { "epoch": 0.25, "learning_rate": 1.7288237917051506e-05, "loss": 1.4598, "step": 7455 }, { "epoch": 0.25, "learning_rate": 1.728751361614525e-05, "loss": 1.4306, "step": 7456 }, { "epoch": 0.25, "learning_rate": 1.728678923369984e-05, "loss": 1.468, "step": 7457 }, { "epoch": 0.25, "learning_rate": 1.7286064769723378e-05, "loss": 1.5091, "step": 7458 }, { "epoch": 0.25, "learning_rate": 1.7285340224223965e-05, "loss": 1.4772, "step": 7459 }, { "epoch": 0.25, "learning_rate": 1.7284615597209712e-05, "loss": 1.6067, "step": 7460 }, { "epoch": 0.25, "learning_rate": 1.7283890888688728e-05, "loss": 1.6076, "step": 7461 }, { "epoch": 0.25, "learning_rate": 1.7283166098669118e-05, "loss": 1.7184, "step": 7462 }, { "epoch": 0.25, "learning_rate": 1.728244122715899e-05, "loss": 1.5277, "step": 7463 }, { "epoch": 0.25, "learning_rate": 1.7281716274166464e-05, "loss": 1.5936, "step": 7464 }, { "epoch": 0.25, "learning_rate": 1.7280991239699643e-05, "loss": 1.7412, "step": 7465 }, { "epoch": 0.25, "learning_rate": 1.728026612376664e-05, "loss": 1.7397, "step": 7466 }, { "epoch": 0.25, "learning_rate": 1.7279540926375573e-05, "loss": 1.6958, "step": 7467 }, { "epoch": 0.25, "learning_rate": 1.727881564753455e-05, "loss": 1.5969, "step": 7468 }, { "epoch": 0.25, "learning_rate": 1.7278090287251687e-05, "loss": 1.6476, "step": 7469 }, { "epoch": 0.25, "learning_rate": 1.7277364845535108e-05, "loss": 1.5891, "step": 7470 }, { "epoch": 0.25, "learning_rate": 1.727663932239292e-05, "loss": 1.7067, "step": 7471 }, { "epoch": 0.25, "learning_rate": 1.7275913717833247e-05, "loss": 1.57, "step": 7472 }, { "epoch": 0.25, "learning_rate": 1.72751880318642e-05, "loss": 1.4626, "step": 7473 }, { "epoch": 0.25, "learning_rate": 1.727446226449391e-05, "loss": 1.5764, "step": 7474 }, { "epoch": 0.25, "learning_rate": 1.7273736415730488e-05, "loss": 1.5823, "step": 7475 }, { "epoch": 0.25, "learning_rate": 1.727301048558206e-05, "loss": 1.6169, "step": 7476 }, { "epoch": 0.25, "learning_rate": 1.7272284474056748e-05, "loss": 1.7196, "step": 7477 }, { "epoch": 0.25, "learning_rate": 1.7271558381162676e-05, "loss": 1.658, "step": 7478 }, { "epoch": 0.25, "learning_rate": 1.7270832206907965e-05, "loss": 1.49, "step": 7479 }, { "epoch": 0.25, "learning_rate": 1.727010595130074e-05, "loss": 1.6762, "step": 7480 }, { "epoch": 0.25, "learning_rate": 1.7269379614349128e-05, "loss": 1.4826, "step": 7481 }, { "epoch": 0.25, "learning_rate": 1.726865319606126e-05, "loss": 1.5856, "step": 7482 }, { "epoch": 0.25, "learning_rate": 1.7267926696445257e-05, "loss": 1.5719, "step": 7483 }, { "epoch": 0.25, "learning_rate": 1.7267200115509256e-05, "loss": 1.5024, "step": 7484 }, { "epoch": 0.25, "learning_rate": 1.7266473453261378e-05, "loss": 1.4877, "step": 7485 }, { "epoch": 0.25, "learning_rate": 1.7265746709709762e-05, "loss": 1.4748, "step": 7486 }, { "epoch": 0.25, "learning_rate": 1.726501988486253e-05, "loss": 1.4374, "step": 7487 }, { "epoch": 0.25, "learning_rate": 1.7264292978727822e-05, "loss": 1.4202, "step": 7488 }, { "epoch": 0.25, "learning_rate": 1.7263565991313767e-05, "loss": 1.4573, "step": 7489 }, { "epoch": 0.25, "learning_rate": 1.72628389226285e-05, "loss": 1.3064, "step": 7490 }, { "epoch": 0.25, "learning_rate": 1.726211177268016e-05, "loss": 1.4342, "step": 7491 }, { "epoch": 0.25, "learning_rate": 1.7261384541476877e-05, "loss": 1.4373, "step": 7492 }, { "epoch": 0.25, "learning_rate": 1.726065722902679e-05, "loss": 1.5574, "step": 7493 }, { "epoch": 0.25, "learning_rate": 1.725992983533804e-05, "loss": 1.566, "step": 7494 }, { "epoch": 0.25, "learning_rate": 1.7259202360418765e-05, "loss": 1.7429, "step": 7495 }, { "epoch": 0.25, "learning_rate": 1.7258474804277095e-05, "loss": 1.6062, "step": 7496 }, { "epoch": 0.25, "learning_rate": 1.7257747166921186e-05, "loss": 1.5447, "step": 7497 }, { "epoch": 0.25, "learning_rate": 1.7257019448359167e-05, "loss": 1.5071, "step": 7498 }, { "epoch": 0.25, "learning_rate": 1.7256291648599187e-05, "loss": 1.4572, "step": 7499 }, { "epoch": 0.25, "learning_rate": 1.725556376764939e-05, "loss": 1.4803, "step": 7500 }, { "epoch": 0.25, "learning_rate": 1.7254835805517913e-05, "loss": 1.426, "step": 7501 }, { "epoch": 0.25, "learning_rate": 1.725410776221291e-05, "loss": 1.5846, "step": 7502 }, { "epoch": 0.25, "learning_rate": 1.7253379637742516e-05, "loss": 1.4091, "step": 7503 }, { "epoch": 0.25, "learning_rate": 1.7252651432114892e-05, "loss": 1.376, "step": 7504 }, { "epoch": 0.25, "learning_rate": 1.7251923145338175e-05, "loss": 1.4784, "step": 7505 }, { "epoch": 0.25, "learning_rate": 1.725119477742052e-05, "loss": 1.3232, "step": 7506 }, { "epoch": 0.25, "learning_rate": 1.725046632837007e-05, "loss": 1.4323, "step": 7507 }, { "epoch": 0.25, "learning_rate": 1.7249737798194982e-05, "loss": 1.4969, "step": 7508 }, { "epoch": 0.25, "learning_rate": 1.7249009186903402e-05, "loss": 1.3869, "step": 7509 }, { "epoch": 0.25, "learning_rate": 1.724828049450349e-05, "loss": 1.5574, "step": 7510 }, { "epoch": 0.25, "learning_rate": 1.724755172100339e-05, "loss": 1.4129, "step": 7511 }, { "epoch": 0.25, "learning_rate": 1.724682286641126e-05, "loss": 1.5019, "step": 7512 }, { "epoch": 0.25, "learning_rate": 1.724609393073526e-05, "loss": 1.4253, "step": 7513 }, { "epoch": 0.25, "learning_rate": 1.7245364913983537e-05, "loss": 1.502, "step": 7514 }, { "epoch": 0.25, "learning_rate": 1.7244635816164254e-05, "loss": 1.4986, "step": 7515 }, { "epoch": 0.25, "learning_rate": 1.724390663728557e-05, "loss": 1.4345, "step": 7516 }, { "epoch": 0.25, "learning_rate": 1.7243177377355636e-05, "loss": 1.6018, "step": 7517 }, { "epoch": 0.25, "learning_rate": 1.724244803638262e-05, "loss": 1.3876, "step": 7518 }, { "epoch": 0.25, "learning_rate": 1.7241718614374678e-05, "loss": 1.6853, "step": 7519 }, { "epoch": 0.25, "learning_rate": 1.724098911133997e-05, "loss": 1.3908, "step": 7520 }, { "epoch": 0.25, "learning_rate": 1.7240259527286666e-05, "loss": 1.4156, "step": 7521 }, { "epoch": 0.25, "learning_rate": 1.723952986222292e-05, "loss": 1.482, "step": 7522 }, { "epoch": 0.25, "learning_rate": 1.7238800116156903e-05, "loss": 1.4927, "step": 7523 }, { "epoch": 0.25, "learning_rate": 1.7238070289096775e-05, "loss": 1.5664, "step": 7524 }, { "epoch": 0.25, "learning_rate": 1.72373403810507e-05, "loss": 1.4761, "step": 7525 }, { "epoch": 0.25, "learning_rate": 1.7236610392026856e-05, "loss": 1.3825, "step": 7526 }, { "epoch": 0.25, "learning_rate": 1.7235880322033403e-05, "loss": 1.3678, "step": 7527 }, { "epoch": 0.25, "learning_rate": 1.7235150171078508e-05, "loss": 1.2489, "step": 7528 }, { "epoch": 0.25, "learning_rate": 1.7234419939170343e-05, "loss": 1.4943, "step": 7529 }, { "epoch": 0.25, "learning_rate": 1.723368962631708e-05, "loss": 1.5466, "step": 7530 }, { "epoch": 0.25, "learning_rate": 1.723295923252689e-05, "loss": 1.6566, "step": 7531 }, { "epoch": 0.25, "learning_rate": 1.723222875780794e-05, "loss": 1.5847, "step": 7532 }, { "epoch": 0.25, "learning_rate": 1.723149820216841e-05, "loss": 1.6916, "step": 7533 }, { "epoch": 0.25, "learning_rate": 1.7230767565616477e-05, "loss": 1.5275, "step": 7534 }, { "epoch": 0.25, "learning_rate": 1.7230036848160302e-05, "loss": 1.5431, "step": 7535 }, { "epoch": 0.25, "learning_rate": 1.7229306049808073e-05, "loss": 1.689, "step": 7536 }, { "epoch": 0.25, "learning_rate": 1.722857517056796e-05, "loss": 1.5661, "step": 7537 }, { "epoch": 0.25, "learning_rate": 1.722784421044815e-05, "loss": 1.5912, "step": 7538 }, { "epoch": 0.25, "learning_rate": 1.7227113169456814e-05, "loss": 1.4445, "step": 7539 }, { "epoch": 0.25, "learning_rate": 1.722638204760213e-05, "loss": 1.5421, "step": 7540 }, { "epoch": 0.25, "learning_rate": 1.722565084489228e-05, "loss": 1.5567, "step": 7541 }, { "epoch": 0.25, "learning_rate": 1.722491956133545e-05, "loss": 1.4711, "step": 7542 }, { "epoch": 0.25, "learning_rate": 1.722418819693982e-05, "loss": 1.6462, "step": 7543 }, { "epoch": 0.25, "learning_rate": 1.7223456751713573e-05, "loss": 1.5336, "step": 7544 }, { "epoch": 0.25, "learning_rate": 1.722272522566489e-05, "loss": 1.5404, "step": 7545 }, { "epoch": 0.25, "learning_rate": 1.722199361880196e-05, "loss": 1.4732, "step": 7546 }, { "epoch": 0.25, "learning_rate": 1.7221261931132965e-05, "loss": 1.4586, "step": 7547 }, { "epoch": 0.25, "learning_rate": 1.7220530162666095e-05, "loss": 1.4066, "step": 7548 }, { "epoch": 0.25, "learning_rate": 1.7219798313409538e-05, "loss": 1.4689, "step": 7549 }, { "epoch": 0.25, "learning_rate": 1.721906638337148e-05, "loss": 1.4794, "step": 7550 }, { "epoch": 0.25, "learning_rate": 1.7218334372560114e-05, "loss": 1.5337, "step": 7551 }, { "epoch": 0.25, "learning_rate": 1.7217602280983622e-05, "loss": 1.5114, "step": 7552 }, { "epoch": 0.25, "learning_rate": 1.7216870108650206e-05, "loss": 1.6635, "step": 7553 }, { "epoch": 0.25, "learning_rate": 1.7216137855568055e-05, "loss": 1.5191, "step": 7554 }, { "epoch": 0.25, "learning_rate": 1.7215405521745358e-05, "loss": 1.4532, "step": 7555 }, { "epoch": 0.25, "learning_rate": 1.721467310719031e-05, "loss": 1.4489, "step": 7556 }, { "epoch": 0.25, "learning_rate": 1.721394061191111e-05, "loss": 1.6027, "step": 7557 }, { "epoch": 0.25, "learning_rate": 1.721320803591595e-05, "loss": 1.4741, "step": 7558 }, { "epoch": 0.25, "learning_rate": 1.7212475379213032e-05, "loss": 1.3932, "step": 7559 }, { "epoch": 0.25, "learning_rate": 1.7211742641810543e-05, "loss": 1.4608, "step": 7560 }, { "epoch": 0.25, "learning_rate": 1.7211009823716695e-05, "loss": 1.5664, "step": 7561 }, { "epoch": 0.25, "learning_rate": 1.7210276924939678e-05, "loss": 1.4976, "step": 7562 }, { "epoch": 0.25, "learning_rate": 1.7209543945487696e-05, "loss": 1.5454, "step": 7563 }, { "epoch": 0.25, "learning_rate": 1.7208810885368948e-05, "loss": 1.5091, "step": 7564 }, { "epoch": 0.25, "learning_rate": 1.7208077744591633e-05, "loss": 1.5484, "step": 7565 }, { "epoch": 0.25, "learning_rate": 1.7207344523163965e-05, "loss": 1.5226, "step": 7566 }, { "epoch": 0.25, "learning_rate": 1.720661122109414e-05, "loss": 1.4827, "step": 7567 }, { "epoch": 0.25, "learning_rate": 1.7205877838390363e-05, "loss": 1.5782, "step": 7568 }, { "epoch": 0.25, "learning_rate": 1.7205144375060842e-05, "loss": 1.532, "step": 7569 }, { "epoch": 0.25, "learning_rate": 1.7204410831113778e-05, "loss": 1.415, "step": 7570 }, { "epoch": 0.25, "learning_rate": 1.7203677206557386e-05, "loss": 1.422, "step": 7571 }, { "epoch": 0.25, "learning_rate": 1.720294350139987e-05, "loss": 1.4631, "step": 7572 }, { "epoch": 0.25, "learning_rate": 1.7202209715649444e-05, "loss": 1.6065, "step": 7573 }, { "epoch": 0.25, "learning_rate": 1.720147584931431e-05, "loss": 1.4272, "step": 7574 }, { "epoch": 0.25, "learning_rate": 1.720074190240269e-05, "loss": 1.5434, "step": 7575 }, { "epoch": 0.25, "learning_rate": 1.7200007874922788e-05, "loss": 1.4198, "step": 7576 }, { "epoch": 0.25, "learning_rate": 1.7199273766882816e-05, "loss": 1.5474, "step": 7577 }, { "epoch": 0.25, "learning_rate": 1.7198539578290997e-05, "loss": 1.5445, "step": 7578 }, { "epoch": 0.25, "learning_rate": 1.7197805309155537e-05, "loss": 1.4619, "step": 7579 }, { "epoch": 0.25, "learning_rate": 1.7197070959484655e-05, "loss": 1.5295, "step": 7580 }, { "epoch": 0.25, "learning_rate": 1.7196336529286567e-05, "loss": 1.5969, "step": 7581 }, { "epoch": 0.25, "learning_rate": 1.7195602018569487e-05, "loss": 1.5641, "step": 7582 }, { "epoch": 0.25, "learning_rate": 1.719486742734164e-05, "loss": 1.4503, "step": 7583 }, { "epoch": 0.25, "learning_rate": 1.719413275561124e-05, "loss": 1.3771, "step": 7584 }, { "epoch": 0.25, "learning_rate": 1.7193398003386514e-05, "loss": 1.3876, "step": 7585 }, { "epoch": 0.25, "learning_rate": 1.7192663170675673e-05, "loss": 1.5289, "step": 7586 }, { "epoch": 0.25, "learning_rate": 1.7191928257486947e-05, "loss": 1.48, "step": 7587 }, { "epoch": 0.25, "learning_rate": 1.7191193263828557e-05, "loss": 1.486, "step": 7588 }, { "epoch": 0.25, "learning_rate": 1.719045818970872e-05, "loss": 1.4611, "step": 7589 }, { "epoch": 0.25, "learning_rate": 1.7189723035135674e-05, "loss": 1.4123, "step": 7590 }, { "epoch": 0.25, "learning_rate": 1.7188987800117634e-05, "loss": 1.5806, "step": 7591 }, { "epoch": 0.25, "learning_rate": 1.7188252484662828e-05, "loss": 1.4669, "step": 7592 }, { "epoch": 0.25, "learning_rate": 1.7187517088779486e-05, "loss": 1.4497, "step": 7593 }, { "epoch": 0.25, "learning_rate": 1.7186781612475837e-05, "loss": 1.4931, "step": 7594 }, { "epoch": 0.25, "learning_rate": 1.7186046055760105e-05, "loss": 1.4903, "step": 7595 }, { "epoch": 0.25, "learning_rate": 1.7185310418640525e-05, "loss": 1.5116, "step": 7596 }, { "epoch": 0.25, "learning_rate": 1.718457470112533e-05, "loss": 1.4446, "step": 7597 }, { "epoch": 0.25, "learning_rate": 1.718383890322274e-05, "loss": 1.3218, "step": 7598 }, { "epoch": 0.25, "learning_rate": 1.7183103024941003e-05, "loss": 1.3487, "step": 7599 }, { "epoch": 0.25, "learning_rate": 1.7182367066288344e-05, "loss": 1.3807, "step": 7600 }, { "epoch": 0.25, "learning_rate": 1.7181631027272997e-05, "loss": 1.5365, "step": 7601 }, { "epoch": 0.25, "learning_rate": 1.7180894907903202e-05, "loss": 1.456, "step": 7602 }, { "epoch": 0.25, "learning_rate": 1.718015870818719e-05, "loss": 1.6533, "step": 7603 }, { "epoch": 0.25, "learning_rate": 1.7179422428133203e-05, "loss": 1.4604, "step": 7604 }, { "epoch": 0.25, "learning_rate": 1.717868606774948e-05, "loss": 1.5106, "step": 7605 }, { "epoch": 0.25, "learning_rate": 1.717794962704425e-05, "loss": 1.4828, "step": 7606 }, { "epoch": 0.25, "learning_rate": 1.7177213106025768e-05, "loss": 1.3292, "step": 7607 }, { "epoch": 0.25, "learning_rate": 1.7176476504702266e-05, "loss": 1.454, "step": 7608 }, { "epoch": 0.25, "learning_rate": 1.7175739823081985e-05, "loss": 1.3897, "step": 7609 }, { "epoch": 0.25, "learning_rate": 1.717500306117317e-05, "loss": 1.4685, "step": 7610 }, { "epoch": 0.25, "learning_rate": 1.7174266218984064e-05, "loss": 1.4552, "step": 7611 }, { "epoch": 0.25, "learning_rate": 1.7173529296522914e-05, "loss": 1.4797, "step": 7612 }, { "epoch": 0.25, "learning_rate": 1.717279229379796e-05, "loss": 1.472, "step": 7613 }, { "epoch": 0.25, "learning_rate": 1.7172055210817452e-05, "loss": 1.3575, "step": 7614 }, { "epoch": 0.25, "learning_rate": 1.7171318047589637e-05, "loss": 1.3204, "step": 7615 }, { "epoch": 0.25, "learning_rate": 1.7170580804122763e-05, "loss": 1.3786, "step": 7616 }, { "epoch": 0.25, "learning_rate": 1.7169843480425074e-05, "loss": 1.5011, "step": 7617 }, { "epoch": 0.25, "learning_rate": 1.716910607650483e-05, "loss": 1.4104, "step": 7618 }, { "epoch": 0.25, "learning_rate": 1.716836859237027e-05, "loss": 1.5262, "step": 7619 }, { "epoch": 0.25, "learning_rate": 1.7167631028029658e-05, "loss": 1.3869, "step": 7620 }, { "epoch": 0.25, "learning_rate": 1.7166893383491237e-05, "loss": 1.4861, "step": 7621 }, { "epoch": 0.25, "learning_rate": 1.7166155658763265e-05, "loss": 1.4831, "step": 7622 }, { "epoch": 0.25, "learning_rate": 1.7165417853853988e-05, "loss": 1.4629, "step": 7623 }, { "epoch": 0.25, "learning_rate": 1.7164679968771674e-05, "loss": 1.6344, "step": 7624 }, { "epoch": 0.25, "learning_rate": 1.7163942003524574e-05, "loss": 1.5689, "step": 7625 }, { "epoch": 0.25, "learning_rate": 1.7163203958120942e-05, "loss": 1.463, "step": 7626 }, { "epoch": 0.25, "learning_rate": 1.7162465832569038e-05, "loss": 1.5368, "step": 7627 }, { "epoch": 0.25, "learning_rate": 1.716172762687712e-05, "loss": 1.5483, "step": 7628 }, { "epoch": 0.25, "learning_rate": 1.716098934105345e-05, "loss": 1.5017, "step": 7629 }, { "epoch": 0.25, "learning_rate": 1.7160250975106286e-05, "loss": 1.4375, "step": 7630 }, { "epoch": 0.25, "learning_rate": 1.7159512529043894e-05, "loss": 1.4263, "step": 7631 }, { "epoch": 0.25, "learning_rate": 1.715877400287453e-05, "loss": 1.6167, "step": 7632 }, { "epoch": 0.25, "learning_rate": 1.7158035396606457e-05, "loss": 1.6255, "step": 7633 }, { "epoch": 0.25, "learning_rate": 1.715729671024795e-05, "loss": 1.5588, "step": 7634 }, { "epoch": 0.25, "learning_rate": 1.715655794380726e-05, "loss": 1.5108, "step": 7635 }, { "epoch": 0.25, "learning_rate": 1.7155819097292662e-05, "loss": 1.5861, "step": 7636 }, { "epoch": 0.25, "learning_rate": 1.7155080170712423e-05, "loss": 1.5397, "step": 7637 }, { "epoch": 0.25, "learning_rate": 1.7154341164074803e-05, "loss": 1.6017, "step": 7638 }, { "epoch": 0.25, "learning_rate": 1.715360207738808e-05, "loss": 1.4643, "step": 7639 }, { "epoch": 0.25, "learning_rate": 1.7152862910660516e-05, "loss": 1.4776, "step": 7640 }, { "epoch": 0.25, "learning_rate": 1.7152123663900386e-05, "loss": 1.5421, "step": 7641 }, { "epoch": 0.25, "learning_rate": 1.715138433711596e-05, "loss": 1.6607, "step": 7642 }, { "epoch": 0.25, "learning_rate": 1.7150644930315512e-05, "loss": 1.5526, "step": 7643 }, { "epoch": 0.25, "learning_rate": 1.7149905443507315e-05, "loss": 1.6675, "step": 7644 }, { "epoch": 0.25, "learning_rate": 1.7149165876699635e-05, "loss": 1.644, "step": 7645 }, { "epoch": 0.25, "learning_rate": 1.7148426229900762e-05, "loss": 1.4803, "step": 7646 }, { "epoch": 0.25, "learning_rate": 1.7147686503118956e-05, "loss": 1.5514, "step": 7647 }, { "epoch": 0.25, "learning_rate": 1.7146946696362502e-05, "loss": 1.4635, "step": 7648 }, { "epoch": 0.25, "learning_rate": 1.714620680963968e-05, "loss": 1.4016, "step": 7649 }, { "epoch": 0.26, "learning_rate": 1.7145466842958764e-05, "loss": 1.6817, "step": 7650 }, { "epoch": 0.26, "learning_rate": 1.7144726796328034e-05, "loss": 1.3043, "step": 7651 }, { "epoch": 0.26, "learning_rate": 1.714398666975577e-05, "loss": 1.4391, "step": 7652 }, { "epoch": 0.26, "learning_rate": 1.714324646325026e-05, "loss": 1.4359, "step": 7653 }, { "epoch": 0.26, "learning_rate": 1.7142506176819775e-05, "loss": 1.3623, "step": 7654 }, { "epoch": 0.26, "learning_rate": 1.7141765810472603e-05, "loss": 1.503, "step": 7655 }, { "epoch": 0.26, "learning_rate": 1.7141025364217028e-05, "loss": 1.4508, "step": 7656 }, { "epoch": 0.26, "learning_rate": 1.7140284838061336e-05, "loss": 1.4438, "step": 7657 }, { "epoch": 0.26, "learning_rate": 1.7139544232013808e-05, "loss": 1.5307, "step": 7658 }, { "epoch": 0.26, "learning_rate": 1.7138803546082735e-05, "loss": 1.5258, "step": 7659 }, { "epoch": 0.26, "learning_rate": 1.7138062780276404e-05, "loss": 1.5334, "step": 7660 }, { "epoch": 0.26, "learning_rate": 1.713732193460311e-05, "loss": 1.4445, "step": 7661 }, { "epoch": 0.26, "learning_rate": 1.7136581009071126e-05, "loss": 1.4541, "step": 7662 }, { "epoch": 0.26, "learning_rate": 1.7135840003688753e-05, "loss": 1.5462, "step": 7663 }, { "epoch": 0.26, "learning_rate": 1.7135098918464282e-05, "loss": 1.5538, "step": 7664 }, { "epoch": 0.26, "learning_rate": 1.7134357753406003e-05, "loss": 1.5244, "step": 7665 }, { "epoch": 0.26, "learning_rate": 1.7133616508522205e-05, "loss": 1.5023, "step": 7666 }, { "epoch": 0.26, "learning_rate": 1.713287518382119e-05, "loss": 1.5072, "step": 7667 }, { "epoch": 0.26, "learning_rate": 1.713213377931125e-05, "loss": 1.5755, "step": 7668 }, { "epoch": 0.26, "learning_rate": 1.7131392295000676e-05, "loss": 1.5507, "step": 7669 }, { "epoch": 0.26, "learning_rate": 1.7130650730897766e-05, "loss": 1.5059, "step": 7670 }, { "epoch": 0.26, "learning_rate": 1.7129909087010817e-05, "loss": 1.6584, "step": 7671 }, { "epoch": 0.26, "learning_rate": 1.712916736334813e-05, "loss": 1.5138, "step": 7672 }, { "epoch": 0.26, "learning_rate": 1.7128425559918006e-05, "loss": 1.5662, "step": 7673 }, { "epoch": 0.26, "learning_rate": 1.7127683676728735e-05, "loss": 1.5213, "step": 7674 }, { "epoch": 0.26, "learning_rate": 1.7126941713788633e-05, "loss": 1.3662, "step": 7675 }, { "epoch": 0.26, "learning_rate": 1.7126199671105986e-05, "loss": 1.5713, "step": 7676 }, { "epoch": 0.26, "learning_rate": 1.7125457548689105e-05, "loss": 1.4256, "step": 7677 }, { "epoch": 0.26, "learning_rate": 1.7124715346546295e-05, "loss": 1.418, "step": 7678 }, { "epoch": 0.26, "learning_rate": 1.7123973064685857e-05, "loss": 1.5277, "step": 7679 }, { "epoch": 0.26, "learning_rate": 1.7123230703116094e-05, "loss": 1.5426, "step": 7680 }, { "epoch": 0.26, "learning_rate": 1.7122488261845316e-05, "loss": 1.4628, "step": 7681 }, { "epoch": 0.26, "learning_rate": 1.712174574088183e-05, "loss": 1.4371, "step": 7682 }, { "epoch": 0.26, "learning_rate": 1.7121003140233944e-05, "loss": 1.4671, "step": 7683 }, { "epoch": 0.26, "learning_rate": 1.712026045990997e-05, "loss": 1.5534, "step": 7684 }, { "epoch": 0.26, "learning_rate": 1.7119517699918206e-05, "loss": 1.5899, "step": 7685 }, { "epoch": 0.26, "learning_rate": 1.7118774860266975e-05, "loss": 1.5279, "step": 7686 }, { "epoch": 0.26, "learning_rate": 1.7118031940964584e-05, "loss": 1.554, "step": 7687 }, { "epoch": 0.26, "learning_rate": 1.7117288942019344e-05, "loss": 1.464, "step": 7688 }, { "epoch": 0.26, "learning_rate": 1.711654586343957e-05, "loss": 1.5713, "step": 7689 }, { "epoch": 0.26, "learning_rate": 1.7115802705233576e-05, "loss": 1.5182, "step": 7690 }, { "epoch": 0.26, "learning_rate": 1.711505946740968e-05, "loss": 1.5317, "step": 7691 }, { "epoch": 0.26, "learning_rate": 1.7114316149976193e-05, "loss": 1.5155, "step": 7692 }, { "epoch": 0.26, "learning_rate": 1.7113572752941436e-05, "loss": 1.4841, "step": 7693 }, { "epoch": 0.26, "learning_rate": 1.711282927631372e-05, "loss": 1.4483, "step": 7694 }, { "epoch": 0.26, "learning_rate": 1.711208572010137e-05, "loss": 1.4916, "step": 7695 }, { "epoch": 0.26, "learning_rate": 1.711134208431271e-05, "loss": 1.343, "step": 7696 }, { "epoch": 0.26, "learning_rate": 1.7110598368956052e-05, "loss": 1.4338, "step": 7697 }, { "epoch": 0.26, "learning_rate": 1.7109854574039718e-05, "loss": 1.3746, "step": 7698 }, { "epoch": 0.26, "learning_rate": 1.7109110699572032e-05, "loss": 1.4527, "step": 7699 }, { "epoch": 0.26, "learning_rate": 1.7108366745561318e-05, "loss": 1.5294, "step": 7700 }, { "epoch": 0.26, "learning_rate": 1.7107622712015898e-05, "loss": 1.5043, "step": 7701 }, { "epoch": 0.26, "learning_rate": 1.7106878598944102e-05, "loss": 1.4409, "step": 7702 }, { "epoch": 0.26, "learning_rate": 1.7106134406354246e-05, "loss": 1.4727, "step": 7703 }, { "epoch": 0.26, "learning_rate": 1.7105390134254668e-05, "loss": 1.5654, "step": 7704 }, { "epoch": 0.26, "learning_rate": 1.710464578265369e-05, "loss": 1.5281, "step": 7705 }, { "epoch": 0.26, "learning_rate": 1.710390135155964e-05, "loss": 1.4061, "step": 7706 }, { "epoch": 0.26, "learning_rate": 1.7103156840980845e-05, "loss": 1.5324, "step": 7707 }, { "epoch": 0.26, "learning_rate": 1.710241225092564e-05, "loss": 1.5234, "step": 7708 }, { "epoch": 0.26, "learning_rate": 1.7101667581402357e-05, "loss": 1.6268, "step": 7709 }, { "epoch": 0.26, "learning_rate": 1.7100922832419326e-05, "loss": 1.527, "step": 7710 }, { "epoch": 0.26, "learning_rate": 1.7100178003984876e-05, "loss": 1.3761, "step": 7711 }, { "epoch": 0.26, "learning_rate": 1.7099433096107344e-05, "loss": 1.4815, "step": 7712 }, { "epoch": 0.26, "learning_rate": 1.7098688108795067e-05, "loss": 1.5533, "step": 7713 }, { "epoch": 0.26, "learning_rate": 1.7097943042056377e-05, "loss": 1.5461, "step": 7714 }, { "epoch": 0.26, "learning_rate": 1.7097197895899614e-05, "loss": 1.5694, "step": 7715 }, { "epoch": 0.26, "learning_rate": 1.7096452670333113e-05, "loss": 1.5012, "step": 7716 }, { "epoch": 0.26, "learning_rate": 1.709570736536521e-05, "loss": 1.5445, "step": 7717 }, { "epoch": 0.26, "learning_rate": 1.7094961981004247e-05, "loss": 1.5459, "step": 7718 }, { "epoch": 0.26, "learning_rate": 1.709421651725857e-05, "loss": 1.4707, "step": 7719 }, { "epoch": 0.26, "learning_rate": 1.7093470974136505e-05, "loss": 1.4134, "step": 7720 }, { "epoch": 0.26, "learning_rate": 1.7092725351646407e-05, "loss": 1.6019, "step": 7721 }, { "epoch": 0.26, "learning_rate": 1.7091979649796613e-05, "loss": 1.541, "step": 7722 }, { "epoch": 0.26, "learning_rate": 1.7091233868595465e-05, "loss": 1.607, "step": 7723 }, { "epoch": 0.26, "learning_rate": 1.7090488008051315e-05, "loss": 1.4958, "step": 7724 }, { "epoch": 0.26, "learning_rate": 1.70897420681725e-05, "loss": 1.4975, "step": 7725 }, { "epoch": 0.26, "learning_rate": 1.7088996048967375e-05, "loss": 1.6302, "step": 7726 }, { "epoch": 0.26, "learning_rate": 1.7088249950444277e-05, "loss": 1.577, "step": 7727 }, { "epoch": 0.26, "learning_rate": 1.708750377261156e-05, "loss": 1.566, "step": 7728 }, { "epoch": 0.26, "learning_rate": 1.708675751547757e-05, "loss": 1.5815, "step": 7729 }, { "epoch": 0.26, "learning_rate": 1.708601117905066e-05, "loss": 1.5708, "step": 7730 }, { "epoch": 0.26, "learning_rate": 1.708526476333918e-05, "loss": 1.6613, "step": 7731 }, { "epoch": 0.26, "learning_rate": 1.708451826835148e-05, "loss": 1.6595, "step": 7732 }, { "epoch": 0.26, "learning_rate": 1.7083771694095915e-05, "loss": 1.6202, "step": 7733 }, { "epoch": 0.26, "learning_rate": 1.708302504058083e-05, "loss": 1.5079, "step": 7734 }, { "epoch": 0.26, "learning_rate": 1.7082278307814593e-05, "loss": 1.5959, "step": 7735 }, { "epoch": 0.26, "learning_rate": 1.708153149580555e-05, "loss": 1.5466, "step": 7736 }, { "epoch": 0.26, "learning_rate": 1.708078460456206e-05, "loss": 1.5064, "step": 7737 }, { "epoch": 0.26, "learning_rate": 1.7080037634092476e-05, "loss": 1.3572, "step": 7738 }, { "epoch": 0.26, "learning_rate": 1.7079290584405158e-05, "loss": 1.4897, "step": 7739 }, { "epoch": 0.26, "learning_rate": 1.7078543455508468e-05, "loss": 1.4895, "step": 7740 }, { "epoch": 0.26, "learning_rate": 1.707779624741076e-05, "loss": 1.506, "step": 7741 }, { "epoch": 0.26, "learning_rate": 1.7077048960120398e-05, "loss": 1.3587, "step": 7742 }, { "epoch": 0.26, "learning_rate": 1.7076301593645745e-05, "loss": 1.4808, "step": 7743 }, { "epoch": 0.26, "learning_rate": 1.707555414799516e-05, "loss": 1.4026, "step": 7744 }, { "epoch": 0.26, "learning_rate": 1.7074806623177004e-05, "loss": 1.5352, "step": 7745 }, { "epoch": 0.26, "learning_rate": 1.7074059019199646e-05, "loss": 1.5153, "step": 7746 }, { "epoch": 0.26, "learning_rate": 1.7073311336071447e-05, "loss": 1.4605, "step": 7747 }, { "epoch": 0.26, "learning_rate": 1.7072563573800774e-05, "loss": 1.3802, "step": 7748 }, { "epoch": 0.26, "learning_rate": 1.7071815732395995e-05, "loss": 1.4865, "step": 7749 }, { "epoch": 0.26, "learning_rate": 1.7071067811865477e-05, "loss": 1.4107, "step": 7750 }, { "epoch": 0.26, "learning_rate": 1.7070319812217586e-05, "loss": 1.4447, "step": 7751 }, { "epoch": 0.26, "learning_rate": 1.7069571733460697e-05, "loss": 1.3709, "step": 7752 }, { "epoch": 0.26, "learning_rate": 1.7068823575603172e-05, "loss": 1.4308, "step": 7753 }, { "epoch": 0.26, "learning_rate": 1.7068075338653386e-05, "loss": 1.4122, "step": 7754 }, { "epoch": 0.26, "learning_rate": 1.7067327022619716e-05, "loss": 1.5083, "step": 7755 }, { "epoch": 0.26, "learning_rate": 1.7066578627510525e-05, "loss": 1.3325, "step": 7756 }, { "epoch": 0.26, "learning_rate": 1.7065830153334194e-05, "loss": 1.5454, "step": 7757 }, { "epoch": 0.26, "learning_rate": 1.7065081600099097e-05, "loss": 1.4922, "step": 7758 }, { "epoch": 0.26, "learning_rate": 1.7064332967813607e-05, "loss": 1.5143, "step": 7759 }, { "epoch": 0.26, "learning_rate": 1.70635842564861e-05, "loss": 1.448, "step": 7760 }, { "epoch": 0.26, "learning_rate": 1.7062835466124953e-05, "loss": 1.4516, "step": 7761 }, { "epoch": 0.26, "learning_rate": 1.706208659673855e-05, "loss": 1.4508, "step": 7762 }, { "epoch": 0.26, "learning_rate": 1.7061337648335264e-05, "loss": 1.4758, "step": 7763 }, { "epoch": 0.26, "learning_rate": 1.7060588620923476e-05, "loss": 1.5754, "step": 7764 }, { "epoch": 0.26, "learning_rate": 1.7059839514511565e-05, "loss": 1.5737, "step": 7765 }, { "epoch": 0.26, "learning_rate": 1.7059090329107916e-05, "loss": 1.5006, "step": 7766 }, { "epoch": 0.26, "learning_rate": 1.7058341064720915e-05, "loss": 1.4594, "step": 7767 }, { "epoch": 0.26, "learning_rate": 1.7057591721358936e-05, "loss": 1.5943, "step": 7768 }, { "epoch": 0.26, "learning_rate": 1.705684229903037e-05, "loss": 1.3999, "step": 7769 }, { "epoch": 0.26, "learning_rate": 1.70560927977436e-05, "loss": 1.3989, "step": 7770 }, { "epoch": 0.26, "learning_rate": 1.7055343217507016e-05, "loss": 1.507, "step": 7771 }, { "epoch": 0.26, "learning_rate": 1.7054593558328996e-05, "loss": 1.4151, "step": 7772 }, { "epoch": 0.26, "learning_rate": 1.7053843820217934e-05, "loss": 1.4307, "step": 7773 }, { "epoch": 0.26, "learning_rate": 1.705309400318222e-05, "loss": 1.468, "step": 7774 }, { "epoch": 0.26, "learning_rate": 1.7052344107230244e-05, "loss": 1.4172, "step": 7775 }, { "epoch": 0.26, "learning_rate": 1.705159413237039e-05, "loss": 1.3832, "step": 7776 }, { "epoch": 0.26, "learning_rate": 1.7050844078611058e-05, "loss": 1.4539, "step": 7777 }, { "epoch": 0.26, "learning_rate": 1.705009394596063e-05, "loss": 1.5969, "step": 7778 }, { "epoch": 0.26, "learning_rate": 1.704934373442751e-05, "loss": 1.5772, "step": 7779 }, { "epoch": 0.26, "learning_rate": 1.7048593444020084e-05, "loss": 1.4278, "step": 7780 }, { "epoch": 0.26, "learning_rate": 1.7047843074746752e-05, "loss": 1.52, "step": 7781 }, { "epoch": 0.26, "learning_rate": 1.704709262661591e-05, "loss": 1.3961, "step": 7782 }, { "epoch": 0.26, "learning_rate": 1.7046342099635948e-05, "loss": 1.5793, "step": 7783 }, { "epoch": 0.26, "learning_rate": 1.704559149381527e-05, "loss": 1.5536, "step": 7784 }, { "epoch": 0.26, "learning_rate": 1.704484080916227e-05, "loss": 1.5206, "step": 7785 }, { "epoch": 0.26, "learning_rate": 1.7044090045685353e-05, "loss": 1.4796, "step": 7786 }, { "epoch": 0.26, "learning_rate": 1.7043339203392916e-05, "loss": 1.553, "step": 7787 }, { "epoch": 0.26, "learning_rate": 1.704258828229336e-05, "loss": 1.4392, "step": 7788 }, { "epoch": 0.26, "learning_rate": 1.7041837282395084e-05, "loss": 1.4453, "step": 7789 }, { "epoch": 0.26, "learning_rate": 1.7041086203706498e-05, "loss": 1.4948, "step": 7790 }, { "epoch": 0.26, "learning_rate": 1.7040335046235997e-05, "loss": 1.5235, "step": 7791 }, { "epoch": 0.26, "learning_rate": 1.7039583809991992e-05, "loss": 1.5265, "step": 7792 }, { "epoch": 0.26, "learning_rate": 1.703883249498289e-05, "loss": 1.4748, "step": 7793 }, { "epoch": 0.26, "learning_rate": 1.7038081101217093e-05, "loss": 1.5498, "step": 7794 }, { "epoch": 0.26, "learning_rate": 1.7037329628703005e-05, "loss": 1.3659, "step": 7795 }, { "epoch": 0.26, "learning_rate": 1.7036578077449046e-05, "loss": 1.2807, "step": 7796 }, { "epoch": 0.26, "learning_rate": 1.703582644746361e-05, "loss": 1.4383, "step": 7797 }, { "epoch": 0.26, "learning_rate": 1.7035074738755114e-05, "loss": 1.4747, "step": 7798 }, { "epoch": 0.26, "learning_rate": 1.7034322951331977e-05, "loss": 1.5556, "step": 7799 }, { "epoch": 0.26, "learning_rate": 1.7033571085202597e-05, "loss": 1.5408, "step": 7800 }, { "epoch": 0.26, "learning_rate": 1.7032819140375395e-05, "loss": 1.5474, "step": 7801 }, { "epoch": 0.26, "learning_rate": 1.7032067116858782e-05, "loss": 1.4532, "step": 7802 }, { "epoch": 0.26, "learning_rate": 1.7031315014661173e-05, "loss": 1.4021, "step": 7803 }, { "epoch": 0.26, "learning_rate": 1.7030562833790978e-05, "loss": 1.5064, "step": 7804 }, { "epoch": 0.26, "learning_rate": 1.702981057425662e-05, "loss": 1.5077, "step": 7805 }, { "epoch": 0.26, "learning_rate": 1.7029058236066515e-05, "loss": 1.5583, "step": 7806 }, { "epoch": 0.26, "learning_rate": 1.7028305819229075e-05, "loss": 1.5091, "step": 7807 }, { "epoch": 0.26, "learning_rate": 1.7027553323752727e-05, "loss": 1.4631, "step": 7808 }, { "epoch": 0.26, "learning_rate": 1.7026800749645884e-05, "loss": 1.4608, "step": 7809 }, { "epoch": 0.26, "learning_rate": 1.702604809691697e-05, "loss": 1.6026, "step": 7810 }, { "epoch": 0.26, "learning_rate": 1.7025295365574403e-05, "loss": 1.5452, "step": 7811 }, { "epoch": 0.26, "learning_rate": 1.7024542555626608e-05, "loss": 1.5961, "step": 7812 }, { "epoch": 0.26, "learning_rate": 1.7023789667082012e-05, "loss": 1.5431, "step": 7813 }, { "epoch": 0.26, "learning_rate": 1.702303669994903e-05, "loss": 1.7138, "step": 7814 }, { "epoch": 0.26, "learning_rate": 1.7022283654236093e-05, "loss": 1.5635, "step": 7815 }, { "epoch": 0.26, "learning_rate": 1.7021530529951627e-05, "loss": 1.6488, "step": 7816 }, { "epoch": 0.26, "learning_rate": 1.7020777327104055e-05, "loss": 1.5404, "step": 7817 }, { "epoch": 0.26, "learning_rate": 1.7020024045701804e-05, "loss": 1.5096, "step": 7818 }, { "epoch": 0.26, "learning_rate": 1.701927068575331e-05, "loss": 1.5105, "step": 7819 }, { "epoch": 0.26, "learning_rate": 1.7018517247266993e-05, "loss": 1.3689, "step": 7820 }, { "epoch": 0.26, "learning_rate": 1.701776373025129e-05, "loss": 1.421, "step": 7821 }, { "epoch": 0.26, "learning_rate": 1.7017010134714628e-05, "loss": 1.4383, "step": 7822 }, { "epoch": 0.26, "learning_rate": 1.701625646066544e-05, "loss": 1.3463, "step": 7823 }, { "epoch": 0.26, "learning_rate": 1.7015502708112157e-05, "loss": 1.4297, "step": 7824 }, { "epoch": 0.26, "learning_rate": 1.7014748877063212e-05, "loss": 1.5873, "step": 7825 }, { "epoch": 0.26, "learning_rate": 1.7013994967527047e-05, "loss": 1.4914, "step": 7826 }, { "epoch": 0.26, "learning_rate": 1.701324097951209e-05, "loss": 1.5399, "step": 7827 }, { "epoch": 0.26, "learning_rate": 1.7012486913026784e-05, "loss": 1.5765, "step": 7828 }, { "epoch": 0.26, "learning_rate": 1.7011732768079555e-05, "loss": 1.4742, "step": 7829 }, { "epoch": 0.26, "learning_rate": 1.7010978544678855e-05, "loss": 1.4923, "step": 7830 }, { "epoch": 0.26, "learning_rate": 1.701022424283311e-05, "loss": 1.5888, "step": 7831 }, { "epoch": 0.26, "learning_rate": 1.700946986255077e-05, "loss": 1.5184, "step": 7832 }, { "epoch": 0.26, "learning_rate": 1.7008715403840267e-05, "loss": 1.4631, "step": 7833 }, { "epoch": 0.26, "learning_rate": 1.700796086671005e-05, "loss": 1.5122, "step": 7834 }, { "epoch": 0.26, "learning_rate": 1.7007206251168557e-05, "loss": 1.5098, "step": 7835 }, { "epoch": 0.26, "learning_rate": 1.7006451557224233e-05, "loss": 1.4514, "step": 7836 }, { "epoch": 0.26, "learning_rate": 1.7005696784885518e-05, "loss": 1.4945, "step": 7837 }, { "epoch": 0.26, "learning_rate": 1.7004941934160866e-05, "loss": 1.5062, "step": 7838 }, { "epoch": 0.26, "learning_rate": 1.7004187005058718e-05, "loss": 1.4425, "step": 7839 }, { "epoch": 0.26, "learning_rate": 1.7003431997587516e-05, "loss": 1.3231, "step": 7840 }, { "epoch": 0.26, "learning_rate": 1.7002676911755717e-05, "loss": 1.5992, "step": 7841 }, { "epoch": 0.26, "learning_rate": 1.700192174757176e-05, "loss": 1.4704, "step": 7842 }, { "epoch": 0.26, "learning_rate": 1.7001166505044106e-05, "loss": 1.499, "step": 7843 }, { "epoch": 0.26, "learning_rate": 1.7000411184181193e-05, "loss": 1.594, "step": 7844 }, { "epoch": 0.26, "learning_rate": 1.699965578499148e-05, "loss": 1.5179, "step": 7845 }, { "epoch": 0.26, "learning_rate": 1.6998900307483413e-05, "loss": 1.4664, "step": 7846 }, { "epoch": 0.26, "learning_rate": 1.6998144751665453e-05, "loss": 1.482, "step": 7847 }, { "epoch": 0.26, "learning_rate": 1.699738911754605e-05, "loss": 1.4196, "step": 7848 }, { "epoch": 0.26, "learning_rate": 1.6996633405133656e-05, "loss": 1.4314, "step": 7849 }, { "epoch": 0.26, "learning_rate": 1.699587761443673e-05, "loss": 1.375, "step": 7850 }, { "epoch": 0.26, "learning_rate": 1.6995121745463726e-05, "loss": 1.5791, "step": 7851 }, { "epoch": 0.26, "learning_rate": 1.6994365798223103e-05, "loss": 1.4125, "step": 7852 }, { "epoch": 0.26, "learning_rate": 1.699360977272332e-05, "loss": 1.4374, "step": 7853 }, { "epoch": 0.26, "learning_rate": 1.6992853668972835e-05, "loss": 1.4676, "step": 7854 }, { "epoch": 0.26, "learning_rate": 1.6992097486980107e-05, "loss": 1.4724, "step": 7855 }, { "epoch": 0.26, "learning_rate": 1.6991341226753603e-05, "loss": 1.3572, "step": 7856 }, { "epoch": 0.26, "learning_rate": 1.6990584888301774e-05, "loss": 1.4358, "step": 7857 }, { "epoch": 0.26, "learning_rate": 1.6989828471633086e-05, "loss": 1.4892, "step": 7858 }, { "epoch": 0.26, "learning_rate": 1.698907197675601e-05, "loss": 1.4837, "step": 7859 }, { "epoch": 0.26, "learning_rate": 1.6988315403679e-05, "loss": 1.4309, "step": 7860 }, { "epoch": 0.26, "learning_rate": 1.698755875241053e-05, "loss": 1.4442, "step": 7861 }, { "epoch": 0.26, "learning_rate": 1.6986802022959062e-05, "loss": 1.5376, "step": 7862 }, { "epoch": 0.26, "learning_rate": 1.698604521533306e-05, "loss": 1.5706, "step": 7863 }, { "epoch": 0.26, "learning_rate": 1.6985288329540996e-05, "loss": 1.4642, "step": 7864 }, { "epoch": 0.26, "learning_rate": 1.698453136559134e-05, "loss": 1.4691, "step": 7865 }, { "epoch": 0.26, "learning_rate": 1.698377432349256e-05, "loss": 1.5234, "step": 7866 }, { "epoch": 0.26, "learning_rate": 1.698301720325312e-05, "loss": 1.5089, "step": 7867 }, { "epoch": 0.26, "learning_rate": 1.69822600048815e-05, "loss": 1.4751, "step": 7868 }, { "epoch": 0.26, "learning_rate": 1.698150272838617e-05, "loss": 1.6123, "step": 7869 }, { "epoch": 0.26, "learning_rate": 1.6980745373775604e-05, "loss": 1.5552, "step": 7870 }, { "epoch": 0.26, "learning_rate": 1.6979987941058274e-05, "loss": 1.5405, "step": 7871 }, { "epoch": 0.26, "learning_rate": 1.697923043024265e-05, "loss": 1.3835, "step": 7872 }, { "epoch": 0.26, "learning_rate": 1.697847284133722e-05, "loss": 1.5012, "step": 7873 }, { "epoch": 0.26, "learning_rate": 1.697771517435045e-05, "loss": 1.3386, "step": 7874 }, { "epoch": 0.26, "learning_rate": 1.697695742929082e-05, "loss": 1.4292, "step": 7875 }, { "epoch": 0.26, "learning_rate": 1.697619960616681e-05, "loss": 1.5319, "step": 7876 }, { "epoch": 0.26, "learning_rate": 1.6975441704986897e-05, "loss": 1.529, "step": 7877 }, { "epoch": 0.26, "learning_rate": 1.6974683725759563e-05, "loss": 1.4596, "step": 7878 }, { "epoch": 0.26, "learning_rate": 1.697392566849329e-05, "loss": 1.5688, "step": 7879 }, { "epoch": 0.26, "learning_rate": 1.6973167533196556e-05, "loss": 1.5441, "step": 7880 }, { "epoch": 0.26, "learning_rate": 1.6972409319877847e-05, "loss": 1.5627, "step": 7881 }, { "epoch": 0.26, "learning_rate": 1.697165102854565e-05, "loss": 1.4697, "step": 7882 }, { "epoch": 0.26, "learning_rate": 1.6970892659208438e-05, "loss": 1.5448, "step": 7883 }, { "epoch": 0.26, "learning_rate": 1.6970134211874706e-05, "loss": 1.5815, "step": 7884 }, { "epoch": 0.26, "learning_rate": 1.696937568655294e-05, "loss": 1.5306, "step": 7885 }, { "epoch": 0.26, "learning_rate": 1.6968617083251618e-05, "loss": 1.4539, "step": 7886 }, { "epoch": 0.26, "learning_rate": 1.6967858401979237e-05, "loss": 1.4493, "step": 7887 }, { "epoch": 0.26, "learning_rate": 1.6967099642744284e-05, "loss": 1.4639, "step": 7888 }, { "epoch": 0.26, "learning_rate": 1.6966340805555248e-05, "loss": 1.5058, "step": 7889 }, { "epoch": 0.26, "learning_rate": 1.6965581890420618e-05, "loss": 1.4381, "step": 7890 }, { "epoch": 0.26, "learning_rate": 1.696482289734889e-05, "loss": 1.5345, "step": 7891 }, { "epoch": 0.26, "learning_rate": 1.6964063826348553e-05, "loss": 1.5332, "step": 7892 }, { "epoch": 0.26, "learning_rate": 1.6963304677428096e-05, "loss": 1.5077, "step": 7893 }, { "epoch": 0.26, "learning_rate": 1.6962545450596017e-05, "loss": 1.4913, "step": 7894 }, { "epoch": 0.26, "learning_rate": 1.6961786145860814e-05, "loss": 1.4845, "step": 7895 }, { "epoch": 0.26, "learning_rate": 1.696102676323098e-05, "loss": 1.442, "step": 7896 }, { "epoch": 0.26, "learning_rate": 1.6960267302715014e-05, "loss": 1.5109, "step": 7897 }, { "epoch": 0.26, "learning_rate": 1.6959507764321408e-05, "loss": 1.4385, "step": 7898 }, { "epoch": 0.26, "learning_rate": 1.695874814805866e-05, "loss": 1.5438, "step": 7899 }, { "epoch": 0.26, "learning_rate": 1.6957988453935276e-05, "loss": 1.5705, "step": 7900 }, { "epoch": 0.26, "learning_rate": 1.6957228681959755e-05, "loss": 1.4489, "step": 7901 }, { "epoch": 0.26, "learning_rate": 1.6956468832140592e-05, "loss": 1.535, "step": 7902 }, { "epoch": 0.26, "learning_rate": 1.6955708904486297e-05, "loss": 1.4444, "step": 7903 }, { "epoch": 0.26, "learning_rate": 1.6954948899005365e-05, "loss": 1.4944, "step": 7904 }, { "epoch": 0.26, "learning_rate": 1.6954188815706306e-05, "loss": 1.5318, "step": 7905 }, { "epoch": 0.26, "learning_rate": 1.695342865459762e-05, "loss": 1.5192, "step": 7906 }, { "epoch": 0.26, "learning_rate": 1.6952668415687813e-05, "loss": 1.5856, "step": 7907 }, { "epoch": 0.26, "learning_rate": 1.6951908098985393e-05, "loss": 1.4379, "step": 7908 }, { "epoch": 0.26, "learning_rate": 1.695114770449887e-05, "loss": 1.3623, "step": 7909 }, { "epoch": 0.26, "learning_rate": 1.6950387232236743e-05, "loss": 1.5681, "step": 7910 }, { "epoch": 0.26, "learning_rate": 1.6949626682207528e-05, "loss": 1.573, "step": 7911 }, { "epoch": 0.26, "learning_rate": 1.6948866054419732e-05, "loss": 1.4894, "step": 7912 }, { "epoch": 0.26, "learning_rate": 1.6948105348881868e-05, "loss": 1.553, "step": 7913 }, { "epoch": 0.26, "learning_rate": 1.6947344565602442e-05, "loss": 1.4733, "step": 7914 }, { "epoch": 0.26, "learning_rate": 1.6946583704589973e-05, "loss": 1.3741, "step": 7915 }, { "epoch": 0.26, "learning_rate": 1.6945822765852974e-05, "loss": 1.5116, "step": 7916 }, { "epoch": 0.26, "learning_rate": 1.694506174939995e-05, "loss": 1.5616, "step": 7917 }, { "epoch": 0.26, "learning_rate": 1.694430065523943e-05, "loss": 1.5924, "step": 7918 }, { "epoch": 0.26, "learning_rate": 1.6943539483379915e-05, "loss": 1.4987, "step": 7919 }, { "epoch": 0.26, "learning_rate": 1.694277823382993e-05, "loss": 1.4481, "step": 7920 }, { "epoch": 0.26, "learning_rate": 1.6942016906597997e-05, "loss": 1.4693, "step": 7921 }, { "epoch": 0.26, "learning_rate": 1.6941255501692624e-05, "loss": 1.4637, "step": 7922 }, { "epoch": 0.26, "learning_rate": 1.6940494019122336e-05, "loss": 1.5229, "step": 7923 }, { "epoch": 0.26, "learning_rate": 1.693973245889565e-05, "loss": 1.3812, "step": 7924 }, { "epoch": 0.26, "learning_rate": 1.693897082102109e-05, "loss": 1.4347, "step": 7925 }, { "epoch": 0.26, "learning_rate": 1.6938209105507177e-05, "loss": 1.3998, "step": 7926 }, { "epoch": 0.26, "learning_rate": 1.6937447312362437e-05, "loss": 1.53, "step": 7927 }, { "epoch": 0.26, "learning_rate": 1.6936685441595388e-05, "loss": 1.3269, "step": 7928 }, { "epoch": 0.26, "learning_rate": 1.6935923493214557e-05, "loss": 1.3538, "step": 7929 }, { "epoch": 0.26, "learning_rate": 1.6935161467228466e-05, "loss": 1.4885, "step": 7930 }, { "epoch": 0.26, "learning_rate": 1.693439936364565e-05, "loss": 1.3821, "step": 7931 }, { "epoch": 0.26, "learning_rate": 1.693363718247463e-05, "loss": 1.5406, "step": 7932 }, { "epoch": 0.26, "learning_rate": 1.693287492372393e-05, "loss": 1.4805, "step": 7933 }, { "epoch": 0.26, "learning_rate": 1.693211258740209e-05, "loss": 1.5581, "step": 7934 }, { "epoch": 0.26, "learning_rate": 1.6931350173517625e-05, "loss": 1.5225, "step": 7935 }, { "epoch": 0.26, "learning_rate": 1.693058768207908e-05, "loss": 1.4985, "step": 7936 }, { "epoch": 0.26, "learning_rate": 1.6929825113094972e-05, "loss": 1.3834, "step": 7937 }, { "epoch": 0.26, "learning_rate": 1.6929062466573848e-05, "loss": 1.4358, "step": 7938 }, { "epoch": 0.26, "learning_rate": 1.6928299742524236e-05, "loss": 1.4431, "step": 7939 }, { "epoch": 0.26, "learning_rate": 1.6927536940954663e-05, "loss": 1.5337, "step": 7940 }, { "epoch": 0.26, "learning_rate": 1.692677406187367e-05, "loss": 1.4104, "step": 7941 }, { "epoch": 0.26, "learning_rate": 1.6926011105289797e-05, "loss": 1.3615, "step": 7942 }, { "epoch": 0.26, "learning_rate": 1.6925248071211574e-05, "loss": 1.3896, "step": 7943 }, { "epoch": 0.26, "learning_rate": 1.6924484959647536e-05, "loss": 1.4011, "step": 7944 }, { "epoch": 0.26, "learning_rate": 1.692372177060623e-05, "loss": 1.3712, "step": 7945 }, { "epoch": 0.26, "learning_rate": 1.692295850409619e-05, "loss": 1.3677, "step": 7946 }, { "epoch": 0.26, "learning_rate": 1.6922195160125956e-05, "loss": 1.4061, "step": 7947 }, { "epoch": 0.26, "learning_rate": 1.692143173870407e-05, "loss": 1.4621, "step": 7948 }, { "epoch": 0.26, "learning_rate": 1.6920668239839074e-05, "loss": 1.5302, "step": 7949 }, { "epoch": 0.27, "learning_rate": 1.691990466353951e-05, "loss": 1.5371, "step": 7950 }, { "epoch": 0.27, "learning_rate": 1.6919141009813924e-05, "loss": 1.5533, "step": 7951 }, { "epoch": 0.27, "learning_rate": 1.6918377278670856e-05, "loss": 1.4562, "step": 7952 }, { "epoch": 0.27, "learning_rate": 1.6917613470118855e-05, "loss": 1.5029, "step": 7953 }, { "epoch": 0.27, "learning_rate": 1.6916849584166465e-05, "loss": 1.5675, "step": 7954 }, { "epoch": 0.27, "learning_rate": 1.6916085620822236e-05, "loss": 1.4304, "step": 7955 }, { "epoch": 0.27, "learning_rate": 1.6915321580094713e-05, "loss": 1.472, "step": 7956 }, { "epoch": 0.27, "learning_rate": 1.6914557461992445e-05, "loss": 1.4952, "step": 7957 }, { "epoch": 0.27, "learning_rate": 1.6913793266523987e-05, "loss": 1.4167, "step": 7958 }, { "epoch": 0.27, "learning_rate": 1.6913028993697877e-05, "loss": 1.4523, "step": 7959 }, { "epoch": 0.27, "learning_rate": 1.691226464352268e-05, "loss": 1.5199, "step": 7960 }, { "epoch": 0.27, "learning_rate": 1.6911500216006936e-05, "loss": 1.4454, "step": 7961 }, { "epoch": 0.27, "learning_rate": 1.6910735711159213e-05, "loss": 1.482, "step": 7962 }, { "epoch": 0.27, "learning_rate": 1.690997112898805e-05, "loss": 1.5548, "step": 7963 }, { "epoch": 0.27, "learning_rate": 1.690920646950201e-05, "loss": 1.5006, "step": 7964 }, { "epoch": 0.27, "learning_rate": 1.690844173270965e-05, "loss": 1.4825, "step": 7965 }, { "epoch": 0.27, "learning_rate": 1.690767691861952e-05, "loss": 1.501, "step": 7966 }, { "epoch": 0.27, "learning_rate": 1.690691202724018e-05, "loss": 1.4738, "step": 7967 }, { "epoch": 0.27, "learning_rate": 1.690614705858019e-05, "loss": 1.5151, "step": 7968 }, { "epoch": 0.27, "learning_rate": 1.690538201264811e-05, "loss": 1.4922, "step": 7969 }, { "epoch": 0.27, "learning_rate": 1.6904616889452497e-05, "loss": 1.4824, "step": 7970 }, { "epoch": 0.27, "learning_rate": 1.6903851689001914e-05, "loss": 1.5179, "step": 7971 }, { "epoch": 0.27, "learning_rate": 1.6903086411304924e-05, "loss": 1.4569, "step": 7972 }, { "epoch": 0.27, "learning_rate": 1.6902321056370083e-05, "loss": 1.5445, "step": 7973 }, { "epoch": 0.27, "learning_rate": 1.6901555624205963e-05, "loss": 1.5283, "step": 7974 }, { "epoch": 0.27, "learning_rate": 1.6900790114821122e-05, "loss": 1.5991, "step": 7975 }, { "epoch": 0.27, "learning_rate": 1.690002452822413e-05, "loss": 1.6033, "step": 7976 }, { "epoch": 0.27, "learning_rate": 1.6899258864423548e-05, "loss": 1.5673, "step": 7977 }, { "epoch": 0.27, "learning_rate": 1.6898493123427945e-05, "loss": 1.6144, "step": 7978 }, { "epoch": 0.27, "learning_rate": 1.689772730524589e-05, "loss": 1.5525, "step": 7979 }, { "epoch": 0.27, "learning_rate": 1.6896961409885953e-05, "loss": 1.5164, "step": 7980 }, { "epoch": 0.27, "learning_rate": 1.68961954373567e-05, "loss": 1.6408, "step": 7981 }, { "epoch": 0.27, "learning_rate": 1.6895429387666702e-05, "loss": 1.5187, "step": 7982 }, { "epoch": 0.27, "learning_rate": 1.6894663260824533e-05, "loss": 1.4727, "step": 7983 }, { "epoch": 0.27, "learning_rate": 1.6893897056838763e-05, "loss": 1.5157, "step": 7984 }, { "epoch": 0.27, "learning_rate": 1.689313077571796e-05, "loss": 1.4941, "step": 7985 }, { "epoch": 0.27, "learning_rate": 1.689236441747071e-05, "loss": 1.6131, "step": 7986 }, { "epoch": 0.27, "learning_rate": 1.6891597982105583e-05, "loss": 1.6406, "step": 7987 }, { "epoch": 0.27, "learning_rate": 1.6890831469631145e-05, "loss": 1.5436, "step": 7988 }, { "epoch": 0.27, "learning_rate": 1.6890064880055985e-05, "loss": 1.4942, "step": 7989 }, { "epoch": 0.27, "learning_rate": 1.688929821338867e-05, "loss": 1.5358, "step": 7990 }, { "epoch": 0.27, "learning_rate": 1.6888531469637784e-05, "loss": 1.5255, "step": 7991 }, { "epoch": 0.27, "learning_rate": 1.688776464881191e-05, "loss": 1.4882, "step": 7992 }, { "epoch": 0.27, "learning_rate": 1.688699775091962e-05, "loss": 1.5386, "step": 7993 }, { "epoch": 0.27, "learning_rate": 1.6886230775969495e-05, "loss": 1.5402, "step": 7994 }, { "epoch": 0.27, "learning_rate": 1.688546372397012e-05, "loss": 1.5737, "step": 7995 }, { "epoch": 0.27, "learning_rate": 1.6884696594930077e-05, "loss": 1.4746, "step": 7996 }, { "epoch": 0.27, "learning_rate": 1.6883929388857954e-05, "loss": 1.5147, "step": 7997 }, { "epoch": 0.27, "learning_rate": 1.6883162105762327e-05, "loss": 1.4377, "step": 7998 }, { "epoch": 0.27, "learning_rate": 1.6882394745651784e-05, "loss": 1.6049, "step": 7999 }, { "epoch": 0.27, "learning_rate": 1.688162730853491e-05, "loss": 1.5158, "step": 8000 } ], "logging_steps": 1.0, "max_steps": 30000, "num_input_tokens_seen": 0, "num_train_epochs": 9223372036854775807, "save_steps": 2000, "total_flos": 3.355230384370483e+19, "train_batch_size": 6, "trial_name": null, "trial_params": null }