diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,28639 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0, + "global_step": 4769, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 2e-08, + "loss": 2.904, + "step": 1 + }, + { + "epoch": 0.0, + "learning_rate": 4e-08, + "loss": 2.8954, + "step": 2 + }, + { + "epoch": 0.0, + "learning_rate": 6.000000000000001e-08, + "loss": 2.9255, + "step": 3 + }, + { + "epoch": 0.0, + "learning_rate": 8e-08, + "loss": 2.9218, + "step": 4 + }, + { + "epoch": 0.0, + "learning_rate": 1.0000000000000001e-07, + "loss": 2.9576, + "step": 5 + }, + { + "epoch": 0.0, + "learning_rate": 1.2000000000000002e-07, + "loss": 2.9166, + "step": 6 + }, + { + "epoch": 0.0, + "learning_rate": 1.4e-07, + "loss": 2.8932, + "step": 7 + }, + { + "epoch": 0.0, + "learning_rate": 1.6e-07, + "loss": 2.9163, + "step": 8 + }, + { + "epoch": 0.0, + "learning_rate": 1.8e-07, + "loss": 2.8732, + "step": 9 + }, + { + "epoch": 0.0, + "learning_rate": 2.0000000000000002e-07, + "loss": 2.88, + "step": 10 + }, + { + "epoch": 0.0, + "learning_rate": 2.2e-07, + "loss": 2.8826, + "step": 11 + }, + { + "epoch": 0.0, + "learning_rate": 2.4000000000000003e-07, + "loss": 2.9134, + "step": 12 + }, + { + "epoch": 0.0, + "learning_rate": 2.6e-07, + "loss": 2.8019, + "step": 13 + }, + { + "epoch": 0.0, + "learning_rate": 2.8e-07, + "loss": 2.9251, + "step": 14 + }, + { + "epoch": 0.0, + "learning_rate": 3.0000000000000004e-07, + "loss": 2.9728, + "step": 15 + }, + { + "epoch": 0.0, + "learning_rate": 3.2e-07, + "loss": 2.9064, + "step": 16 + }, + { + "epoch": 0.0, + "learning_rate": 3.4000000000000003e-07, + "loss": 2.8646, + "step": 17 + }, + { + "epoch": 0.0, + "learning_rate": 3.6e-07, + "loss": 2.912, + "step": 18 + }, + { + "epoch": 0.0, + "learning_rate": 3.8e-07, + "loss": 2.8366, + "step": 19 + }, + { + "epoch": 0.0, + "learning_rate": 4.0000000000000003e-07, + "loss": 2.8926, + "step": 20 + }, + { + "epoch": 0.0, + "learning_rate": 4.2000000000000006e-07, + "loss": 2.8562, + "step": 21 + }, + { + "epoch": 0.0, + "learning_rate": 4.4e-07, + "loss": 2.8903, + "step": 22 + }, + { + "epoch": 0.0, + "learning_rate": 4.6000000000000004e-07, + "loss": 2.9373, + "step": 23 + }, + { + "epoch": 0.01, + "learning_rate": 4.800000000000001e-07, + "loss": 2.8644, + "step": 24 + }, + { + "epoch": 0.01, + "learning_rate": 5.000000000000001e-07, + "loss": 2.8257, + "step": 25 + }, + { + "epoch": 0.01, + "learning_rate": 5.2e-07, + "loss": 2.8438, + "step": 26 + }, + { + "epoch": 0.01, + "learning_rate": 5.4e-07, + "loss": 2.8876, + "step": 27 + }, + { + "epoch": 0.01, + "learning_rate": 5.6e-07, + "loss": 2.8682, + "step": 28 + }, + { + "epoch": 0.01, + "learning_rate": 5.800000000000001e-07, + "loss": 2.8079, + "step": 29 + }, + { + "epoch": 0.01, + "learning_rate": 6.000000000000001e-07, + "loss": 2.874, + "step": 30 + }, + { + "epoch": 0.01, + "learning_rate": 6.200000000000001e-07, + "loss": 2.861, + "step": 31 + }, + { + "epoch": 0.01, + "learning_rate": 6.4e-07, + "loss": 2.8549, + "step": 32 + }, + { + "epoch": 0.01, + "learning_rate": 6.6e-07, + "loss": 2.8561, + "step": 33 + }, + { + "epoch": 0.01, + "learning_rate": 6.800000000000001e-07, + "loss": 2.8469, + "step": 34 + }, + { + "epoch": 0.01, + "learning_rate": 7.000000000000001e-07, + "loss": 2.9127, + "step": 35 + }, + { + "epoch": 0.01, + "learning_rate": 7.2e-07, + "loss": 2.9337, + "step": 36 + }, + { + "epoch": 0.01, + "learning_rate": 7.4e-07, + "loss": 2.849, + "step": 37 + }, + { + "epoch": 0.01, + "learning_rate": 7.6e-07, + "loss": 2.9341, + "step": 38 + }, + { + "epoch": 0.01, + "learning_rate": 7.8e-07, + "loss": 2.8711, + "step": 39 + }, + { + "epoch": 0.01, + "learning_rate": 8.000000000000001e-07, + "loss": 2.8596, + "step": 40 + }, + { + "epoch": 0.01, + "learning_rate": 8.200000000000001e-07, + "loss": 2.877, + "step": 41 + }, + { + "epoch": 0.01, + "learning_rate": 8.400000000000001e-07, + "loss": 2.8907, + "step": 42 + }, + { + "epoch": 0.01, + "learning_rate": 8.6e-07, + "loss": 2.935, + "step": 43 + }, + { + "epoch": 0.01, + "learning_rate": 8.8e-07, + "loss": 2.8482, + "step": 44 + }, + { + "epoch": 0.01, + "learning_rate": 9.000000000000001e-07, + "loss": 2.8807, + "step": 45 + }, + { + "epoch": 0.01, + "learning_rate": 9.200000000000001e-07, + "loss": 2.8353, + "step": 46 + }, + { + "epoch": 0.01, + "learning_rate": 9.400000000000001e-07, + "loss": 2.8687, + "step": 47 + }, + { + "epoch": 0.01, + "learning_rate": 9.600000000000001e-07, + "loss": 2.9506, + "step": 48 + }, + { + "epoch": 0.01, + "learning_rate": 9.800000000000001e-07, + "loss": 2.8171, + "step": 49 + }, + { + "epoch": 0.01, + "learning_rate": 1.0000000000000002e-06, + "loss": 2.8962, + "step": 50 + }, + { + "epoch": 0.01, + "learning_rate": 1.02e-06, + "loss": 2.7766, + "step": 51 + }, + { + "epoch": 0.01, + "learning_rate": 1.04e-06, + "loss": 2.8165, + "step": 52 + }, + { + "epoch": 0.01, + "learning_rate": 1.06e-06, + "loss": 2.8153, + "step": 53 + }, + { + "epoch": 0.01, + "learning_rate": 1.08e-06, + "loss": 2.8423, + "step": 54 + }, + { + "epoch": 0.01, + "learning_rate": 1.1e-06, + "loss": 2.8425, + "step": 55 + }, + { + "epoch": 0.01, + "learning_rate": 1.12e-06, + "loss": 2.795, + "step": 56 + }, + { + "epoch": 0.01, + "learning_rate": 1.14e-06, + "loss": 2.8433, + "step": 57 + }, + { + "epoch": 0.01, + "learning_rate": 1.1600000000000001e-06, + "loss": 2.8655, + "step": 58 + }, + { + "epoch": 0.01, + "learning_rate": 1.1800000000000001e-06, + "loss": 2.7974, + "step": 59 + }, + { + "epoch": 0.01, + "learning_rate": 1.2000000000000002e-06, + "loss": 2.8568, + "step": 60 + }, + { + "epoch": 0.01, + "learning_rate": 1.2200000000000002e-06, + "loss": 2.8179, + "step": 61 + }, + { + "epoch": 0.01, + "learning_rate": 1.2400000000000002e-06, + "loss": 2.8589, + "step": 62 + }, + { + "epoch": 0.01, + "learning_rate": 1.26e-06, + "loss": 2.8296, + "step": 63 + }, + { + "epoch": 0.01, + "learning_rate": 1.28e-06, + "loss": 2.787, + "step": 64 + }, + { + "epoch": 0.01, + "learning_rate": 1.3e-06, + "loss": 2.8148, + "step": 65 + }, + { + "epoch": 0.01, + "learning_rate": 1.32e-06, + "loss": 2.9169, + "step": 66 + }, + { + "epoch": 0.01, + "learning_rate": 1.34e-06, + "loss": 2.7716, + "step": 67 + }, + { + "epoch": 0.01, + "learning_rate": 1.3600000000000001e-06, + "loss": 2.8049, + "step": 68 + }, + { + "epoch": 0.01, + "learning_rate": 1.3800000000000001e-06, + "loss": 2.8142, + "step": 69 + }, + { + "epoch": 0.01, + "learning_rate": 1.4000000000000001e-06, + "loss": 2.8352, + "step": 70 + }, + { + "epoch": 0.01, + "learning_rate": 1.42e-06, + "loss": 2.8208, + "step": 71 + }, + { + "epoch": 0.02, + "learning_rate": 1.44e-06, + "loss": 2.8889, + "step": 72 + }, + { + "epoch": 0.02, + "learning_rate": 1.46e-06, + "loss": 2.8063, + "step": 73 + }, + { + "epoch": 0.02, + "learning_rate": 1.48e-06, + "loss": 2.7622, + "step": 74 + }, + { + "epoch": 0.02, + "learning_rate": 1.5e-06, + "loss": 2.8007, + "step": 75 + }, + { + "epoch": 0.02, + "learning_rate": 1.52e-06, + "loss": 2.6981, + "step": 76 + }, + { + "epoch": 0.02, + "learning_rate": 1.54e-06, + "loss": 2.7913, + "step": 77 + }, + { + "epoch": 0.02, + "learning_rate": 1.56e-06, + "loss": 2.7534, + "step": 78 + }, + { + "epoch": 0.02, + "learning_rate": 1.5800000000000001e-06, + "loss": 2.7709, + "step": 79 + }, + { + "epoch": 0.02, + "learning_rate": 1.6000000000000001e-06, + "loss": 2.7763, + "step": 80 + }, + { + "epoch": 0.02, + "learning_rate": 1.6200000000000002e-06, + "loss": 2.7868, + "step": 81 + }, + { + "epoch": 0.02, + "learning_rate": 1.6400000000000002e-06, + "loss": 2.8369, + "step": 82 + }, + { + "epoch": 0.02, + "learning_rate": 1.6600000000000002e-06, + "loss": 2.79, + "step": 83 + }, + { + "epoch": 0.02, + "learning_rate": 1.6800000000000002e-06, + "loss": 2.8272, + "step": 84 + }, + { + "epoch": 0.02, + "learning_rate": 1.7000000000000002e-06, + "loss": 2.7317, + "step": 85 + }, + { + "epoch": 0.02, + "learning_rate": 1.72e-06, + "loss": 2.7402, + "step": 86 + }, + { + "epoch": 0.02, + "learning_rate": 1.74e-06, + "loss": 2.7692, + "step": 87 + }, + { + "epoch": 0.02, + "learning_rate": 1.76e-06, + "loss": 2.7757, + "step": 88 + }, + { + "epoch": 0.02, + "learning_rate": 1.7800000000000001e-06, + "loss": 2.7822, + "step": 89 + }, + { + "epoch": 0.02, + "learning_rate": 1.8000000000000001e-06, + "loss": 2.719, + "step": 90 + }, + { + "epoch": 0.02, + "learning_rate": 1.8200000000000002e-06, + "loss": 2.7086, + "step": 91 + }, + { + "epoch": 0.02, + "learning_rate": 1.8400000000000002e-06, + "loss": 2.8341, + "step": 92 + }, + { + "epoch": 0.02, + "learning_rate": 1.8600000000000002e-06, + "loss": 2.8146, + "step": 93 + }, + { + "epoch": 0.02, + "learning_rate": 1.8800000000000002e-06, + "loss": 2.7612, + "step": 94 + }, + { + "epoch": 0.02, + "learning_rate": 1.9000000000000002e-06, + "loss": 2.7453, + "step": 95 + }, + { + "epoch": 0.02, + "learning_rate": 1.9200000000000003e-06, + "loss": 2.7366, + "step": 96 + }, + { + "epoch": 0.02, + "learning_rate": 1.94e-06, + "loss": 2.707, + "step": 97 + }, + { + "epoch": 0.02, + "learning_rate": 1.9600000000000003e-06, + "loss": 2.7443, + "step": 98 + }, + { + "epoch": 0.02, + "learning_rate": 1.98e-06, + "loss": 2.7638, + "step": 99 + }, + { + "epoch": 0.02, + "learning_rate": 2.0000000000000003e-06, + "loss": 2.8238, + "step": 100 + }, + { + "epoch": 0.02, + "learning_rate": 2.02e-06, + "loss": 2.7883, + "step": 101 + }, + { + "epoch": 0.02, + "learning_rate": 2.04e-06, + "loss": 2.7674, + "step": 102 + }, + { + "epoch": 0.02, + "learning_rate": 2.06e-06, + "loss": 2.7425, + "step": 103 + }, + { + "epoch": 0.02, + "learning_rate": 2.08e-06, + "loss": 2.7795, + "step": 104 + }, + { + "epoch": 0.02, + "learning_rate": 2.1000000000000002e-06, + "loss": 2.7217, + "step": 105 + }, + { + "epoch": 0.02, + "learning_rate": 2.12e-06, + "loss": 2.7937, + "step": 106 + }, + { + "epoch": 0.02, + "learning_rate": 2.1400000000000003e-06, + "loss": 2.745, + "step": 107 + }, + { + "epoch": 0.02, + "learning_rate": 2.16e-06, + "loss": 2.7264, + "step": 108 + }, + { + "epoch": 0.02, + "learning_rate": 2.1800000000000003e-06, + "loss": 2.7458, + "step": 109 + }, + { + "epoch": 0.02, + "learning_rate": 2.2e-06, + "loss": 2.6821, + "step": 110 + }, + { + "epoch": 0.02, + "learning_rate": 2.2200000000000003e-06, + "loss": 2.6619, + "step": 111 + }, + { + "epoch": 0.02, + "learning_rate": 2.24e-06, + "loss": 2.8581, + "step": 112 + }, + { + "epoch": 0.02, + "learning_rate": 2.2600000000000004e-06, + "loss": 2.7088, + "step": 113 + }, + { + "epoch": 0.02, + "learning_rate": 2.28e-06, + "loss": 2.7115, + "step": 114 + }, + { + "epoch": 0.02, + "learning_rate": 2.3000000000000004e-06, + "loss": 2.697, + "step": 115 + }, + { + "epoch": 0.02, + "learning_rate": 2.3200000000000002e-06, + "loss": 2.7265, + "step": 116 + }, + { + "epoch": 0.02, + "learning_rate": 2.3400000000000005e-06, + "loss": 2.8132, + "step": 117 + }, + { + "epoch": 0.02, + "learning_rate": 2.3600000000000003e-06, + "loss": 2.6781, + "step": 118 + }, + { + "epoch": 0.02, + "learning_rate": 2.38e-06, + "loss": 2.6933, + "step": 119 + }, + { + "epoch": 0.03, + "learning_rate": 2.4000000000000003e-06, + "loss": 2.7092, + "step": 120 + }, + { + "epoch": 0.03, + "learning_rate": 2.42e-06, + "loss": 2.7093, + "step": 121 + }, + { + "epoch": 0.03, + "learning_rate": 2.4400000000000004e-06, + "loss": 2.7519, + "step": 122 + }, + { + "epoch": 0.03, + "learning_rate": 2.46e-06, + "loss": 2.6258, + "step": 123 + }, + { + "epoch": 0.03, + "learning_rate": 2.4800000000000004e-06, + "loss": 2.7133, + "step": 124 + }, + { + "epoch": 0.03, + "learning_rate": 2.5e-06, + "loss": 2.6973, + "step": 125 + }, + { + "epoch": 0.03, + "learning_rate": 2.52e-06, + "loss": 2.6987, + "step": 126 + }, + { + "epoch": 0.03, + "learning_rate": 2.5400000000000002e-06, + "loss": 2.7247, + "step": 127 + }, + { + "epoch": 0.03, + "learning_rate": 2.56e-06, + "loss": 2.7198, + "step": 128 + }, + { + "epoch": 0.03, + "learning_rate": 2.5800000000000003e-06, + "loss": 2.7382, + "step": 129 + }, + { + "epoch": 0.03, + "learning_rate": 2.6e-06, + "loss": 2.7053, + "step": 130 + }, + { + "epoch": 0.03, + "learning_rate": 2.6200000000000003e-06, + "loss": 2.7377, + "step": 131 + }, + { + "epoch": 0.03, + "learning_rate": 2.64e-06, + "loss": 2.7267, + "step": 132 + }, + { + "epoch": 0.03, + "learning_rate": 2.6600000000000004e-06, + "loss": 2.7738, + "step": 133 + }, + { + "epoch": 0.03, + "learning_rate": 2.68e-06, + "loss": 2.6547, + "step": 134 + }, + { + "epoch": 0.03, + "learning_rate": 2.7000000000000004e-06, + "loss": 2.6269, + "step": 135 + }, + { + "epoch": 0.03, + "learning_rate": 2.7200000000000002e-06, + "loss": 2.7186, + "step": 136 + }, + { + "epoch": 0.03, + "learning_rate": 2.7400000000000004e-06, + "loss": 2.7082, + "step": 137 + }, + { + "epoch": 0.03, + "learning_rate": 2.7600000000000003e-06, + "loss": 2.7233, + "step": 138 + }, + { + "epoch": 0.03, + "learning_rate": 2.7800000000000005e-06, + "loss": 2.7314, + "step": 139 + }, + { + "epoch": 0.03, + "learning_rate": 2.8000000000000003e-06, + "loss": 2.6777, + "step": 140 + }, + { + "epoch": 0.03, + "learning_rate": 2.82e-06, + "loss": 2.6482, + "step": 141 + }, + { + "epoch": 0.03, + "learning_rate": 2.84e-06, + "loss": 2.7668, + "step": 142 + }, + { + "epoch": 0.03, + "learning_rate": 2.86e-06, + "loss": 2.7234, + "step": 143 + }, + { + "epoch": 0.03, + "learning_rate": 2.88e-06, + "loss": 2.6829, + "step": 144 + }, + { + "epoch": 0.03, + "learning_rate": 2.9e-06, + "loss": 2.6545, + "step": 145 + }, + { + "epoch": 0.03, + "learning_rate": 2.92e-06, + "loss": 2.6764, + "step": 146 + }, + { + "epoch": 0.03, + "learning_rate": 2.9400000000000002e-06, + "loss": 2.654, + "step": 147 + }, + { + "epoch": 0.03, + "learning_rate": 2.96e-06, + "loss": 2.7481, + "step": 148 + }, + { + "epoch": 0.03, + "learning_rate": 2.9800000000000003e-06, + "loss": 2.7011, + "step": 149 + }, + { + "epoch": 0.03, + "learning_rate": 3e-06, + "loss": 2.6632, + "step": 150 + }, + { + "epoch": 0.03, + "learning_rate": 3.0200000000000003e-06, + "loss": 2.5988, + "step": 151 + }, + { + "epoch": 0.03, + "learning_rate": 3.04e-06, + "loss": 2.6566, + "step": 152 + }, + { + "epoch": 0.03, + "learning_rate": 3.0600000000000003e-06, + "loss": 2.7242, + "step": 153 + }, + { + "epoch": 0.03, + "learning_rate": 3.08e-06, + "loss": 2.7302, + "step": 154 + }, + { + "epoch": 0.03, + "learning_rate": 3.1000000000000004e-06, + "loss": 2.7085, + "step": 155 + }, + { + "epoch": 0.03, + "learning_rate": 3.12e-06, + "loss": 2.5801, + "step": 156 + }, + { + "epoch": 0.03, + "learning_rate": 3.1400000000000004e-06, + "loss": 2.7157, + "step": 157 + }, + { + "epoch": 0.03, + "learning_rate": 3.1600000000000002e-06, + "loss": 2.7131, + "step": 158 + }, + { + "epoch": 0.03, + "learning_rate": 3.1800000000000005e-06, + "loss": 2.6921, + "step": 159 + }, + { + "epoch": 0.03, + "learning_rate": 3.2000000000000003e-06, + "loss": 2.7074, + "step": 160 + }, + { + "epoch": 0.03, + "learning_rate": 3.2200000000000005e-06, + "loss": 2.6334, + "step": 161 + }, + { + "epoch": 0.03, + "learning_rate": 3.2400000000000003e-06, + "loss": 2.6491, + "step": 162 + }, + { + "epoch": 0.03, + "learning_rate": 3.2600000000000006e-06, + "loss": 2.7324, + "step": 163 + }, + { + "epoch": 0.03, + "learning_rate": 3.2800000000000004e-06, + "loss": 2.6472, + "step": 164 + }, + { + "epoch": 0.03, + "learning_rate": 3.3000000000000006e-06, + "loss": 2.6348, + "step": 165 + }, + { + "epoch": 0.03, + "learning_rate": 3.3200000000000004e-06, + "loss": 2.6945, + "step": 166 + }, + { + "epoch": 0.04, + "learning_rate": 3.3400000000000006e-06, + "loss": 2.6437, + "step": 167 + }, + { + "epoch": 0.04, + "learning_rate": 3.3600000000000004e-06, + "loss": 2.6649, + "step": 168 + }, + { + "epoch": 0.04, + "learning_rate": 3.3800000000000007e-06, + "loss": 2.7028, + "step": 169 + }, + { + "epoch": 0.04, + "learning_rate": 3.4000000000000005e-06, + "loss": 2.6205, + "step": 170 + }, + { + "epoch": 0.04, + "learning_rate": 3.4200000000000007e-06, + "loss": 2.6459, + "step": 171 + }, + { + "epoch": 0.04, + "learning_rate": 3.44e-06, + "loss": 2.7033, + "step": 172 + }, + { + "epoch": 0.04, + "learning_rate": 3.46e-06, + "loss": 2.6832, + "step": 173 + }, + { + "epoch": 0.04, + "learning_rate": 3.48e-06, + "loss": 2.7068, + "step": 174 + }, + { + "epoch": 0.04, + "learning_rate": 3.5e-06, + "loss": 2.6766, + "step": 175 + }, + { + "epoch": 0.04, + "learning_rate": 3.52e-06, + "loss": 2.6037, + "step": 176 + }, + { + "epoch": 0.04, + "learning_rate": 3.54e-06, + "loss": 2.6552, + "step": 177 + }, + { + "epoch": 0.04, + "learning_rate": 3.5600000000000002e-06, + "loss": 2.7111, + "step": 178 + }, + { + "epoch": 0.04, + "learning_rate": 3.58e-06, + "loss": 2.6492, + "step": 179 + }, + { + "epoch": 0.04, + "learning_rate": 3.6000000000000003e-06, + "loss": 2.6517, + "step": 180 + }, + { + "epoch": 0.04, + "learning_rate": 3.62e-06, + "loss": 2.6803, + "step": 181 + }, + { + "epoch": 0.04, + "learning_rate": 3.6400000000000003e-06, + "loss": 2.6258, + "step": 182 + }, + { + "epoch": 0.04, + "learning_rate": 3.66e-06, + "loss": 2.6104, + "step": 183 + }, + { + "epoch": 0.04, + "learning_rate": 3.6800000000000003e-06, + "loss": 2.6404, + "step": 184 + }, + { + "epoch": 0.04, + "learning_rate": 3.7e-06, + "loss": 2.6852, + "step": 185 + }, + { + "epoch": 0.04, + "learning_rate": 3.7200000000000004e-06, + "loss": 2.6703, + "step": 186 + }, + { + "epoch": 0.04, + "learning_rate": 3.74e-06, + "loss": 2.6617, + "step": 187 + }, + { + "epoch": 0.04, + "learning_rate": 3.7600000000000004e-06, + "loss": 2.618, + "step": 188 + }, + { + "epoch": 0.04, + "learning_rate": 3.7800000000000002e-06, + "loss": 2.634, + "step": 189 + }, + { + "epoch": 0.04, + "learning_rate": 3.8000000000000005e-06, + "loss": 2.6006, + "step": 190 + }, + { + "epoch": 0.04, + "learning_rate": 3.820000000000001e-06, + "loss": 2.6486, + "step": 191 + }, + { + "epoch": 0.04, + "learning_rate": 3.8400000000000005e-06, + "loss": 2.6908, + "step": 192 + }, + { + "epoch": 0.04, + "learning_rate": 3.86e-06, + "loss": 2.6407, + "step": 193 + }, + { + "epoch": 0.04, + "learning_rate": 3.88e-06, + "loss": 2.6512, + "step": 194 + }, + { + "epoch": 0.04, + "learning_rate": 3.900000000000001e-06, + "loss": 2.679, + "step": 195 + }, + { + "epoch": 0.04, + "learning_rate": 3.920000000000001e-06, + "loss": 2.6957, + "step": 196 + }, + { + "epoch": 0.04, + "learning_rate": 3.94e-06, + "loss": 2.674, + "step": 197 + }, + { + "epoch": 0.04, + "learning_rate": 3.96e-06, + "loss": 2.6031, + "step": 198 + }, + { + "epoch": 0.04, + "learning_rate": 3.980000000000001e-06, + "loss": 2.6306, + "step": 199 + }, + { + "epoch": 0.04, + "learning_rate": 4.000000000000001e-06, + "loss": 2.6644, + "step": 200 + }, + { + "epoch": 0.04, + "learning_rate": 4.0200000000000005e-06, + "loss": 2.6793, + "step": 201 + }, + { + "epoch": 0.04, + "learning_rate": 4.04e-06, + "loss": 2.6439, + "step": 202 + }, + { + "epoch": 0.04, + "learning_rate": 4.060000000000001e-06, + "loss": 2.6478, + "step": 203 + }, + { + "epoch": 0.04, + "learning_rate": 4.08e-06, + "loss": 2.59, + "step": 204 + }, + { + "epoch": 0.04, + "learning_rate": 4.1e-06, + "loss": 2.6602, + "step": 205 + }, + { + "epoch": 0.04, + "learning_rate": 4.12e-06, + "loss": 2.6733, + "step": 206 + }, + { + "epoch": 0.04, + "learning_rate": 4.14e-06, + "loss": 2.6462, + "step": 207 + }, + { + "epoch": 0.04, + "learning_rate": 4.16e-06, + "loss": 2.7422, + "step": 208 + }, + { + "epoch": 0.04, + "learning_rate": 4.18e-06, + "loss": 2.6505, + "step": 209 + }, + { + "epoch": 0.04, + "learning_rate": 4.2000000000000004e-06, + "loss": 2.704, + "step": 210 + }, + { + "epoch": 0.04, + "learning_rate": 4.22e-06, + "loss": 2.6437, + "step": 211 + }, + { + "epoch": 0.04, + "learning_rate": 4.24e-06, + "loss": 2.6348, + "step": 212 + }, + { + "epoch": 0.04, + "learning_rate": 4.26e-06, + "loss": 2.7166, + "step": 213 + }, + { + "epoch": 0.04, + "learning_rate": 4.2800000000000005e-06, + "loss": 2.7007, + "step": 214 + }, + { + "epoch": 0.05, + "learning_rate": 4.3e-06, + "loss": 2.6831, + "step": 215 + }, + { + "epoch": 0.05, + "learning_rate": 4.32e-06, + "loss": 2.596, + "step": 216 + }, + { + "epoch": 0.05, + "learning_rate": 4.34e-06, + "loss": 2.6554, + "step": 217 + }, + { + "epoch": 0.05, + "learning_rate": 4.360000000000001e-06, + "loss": 2.5943, + "step": 218 + }, + { + "epoch": 0.05, + "learning_rate": 4.38e-06, + "loss": 2.5814, + "step": 219 + }, + { + "epoch": 0.05, + "learning_rate": 4.4e-06, + "loss": 2.6784, + "step": 220 + }, + { + "epoch": 0.05, + "learning_rate": 4.42e-06, + "loss": 2.6349, + "step": 221 + }, + { + "epoch": 0.05, + "learning_rate": 4.440000000000001e-06, + "loss": 2.6699, + "step": 222 + }, + { + "epoch": 0.05, + "learning_rate": 4.4600000000000005e-06, + "loss": 2.6806, + "step": 223 + }, + { + "epoch": 0.05, + "learning_rate": 4.48e-06, + "loss": 2.6068, + "step": 224 + }, + { + "epoch": 0.05, + "learning_rate": 4.5e-06, + "loss": 2.7131, + "step": 225 + }, + { + "epoch": 0.05, + "learning_rate": 4.520000000000001e-06, + "loss": 2.657, + "step": 226 + }, + { + "epoch": 0.05, + "learning_rate": 4.540000000000001e-06, + "loss": 2.6654, + "step": 227 + }, + { + "epoch": 0.05, + "learning_rate": 4.56e-06, + "loss": 2.6943, + "step": 228 + }, + { + "epoch": 0.05, + "learning_rate": 4.58e-06, + "loss": 2.6617, + "step": 229 + }, + { + "epoch": 0.05, + "learning_rate": 4.600000000000001e-06, + "loss": 2.614, + "step": 230 + }, + { + "epoch": 0.05, + "learning_rate": 4.620000000000001e-06, + "loss": 2.6885, + "step": 231 + }, + { + "epoch": 0.05, + "learning_rate": 4.6400000000000005e-06, + "loss": 2.7078, + "step": 232 + }, + { + "epoch": 0.05, + "learning_rate": 4.66e-06, + "loss": 2.6407, + "step": 233 + }, + { + "epoch": 0.05, + "learning_rate": 4.680000000000001e-06, + "loss": 2.6499, + "step": 234 + }, + { + "epoch": 0.05, + "learning_rate": 4.7e-06, + "loss": 2.7438, + "step": 235 + }, + { + "epoch": 0.05, + "learning_rate": 4.7200000000000005e-06, + "loss": 2.689, + "step": 236 + }, + { + "epoch": 0.05, + "learning_rate": 4.74e-06, + "loss": 2.6315, + "step": 237 + }, + { + "epoch": 0.05, + "learning_rate": 4.76e-06, + "loss": 2.576, + "step": 238 + }, + { + "epoch": 0.05, + "learning_rate": 4.78e-06, + "loss": 2.6051, + "step": 239 + }, + { + "epoch": 0.05, + "learning_rate": 4.800000000000001e-06, + "loss": 2.65, + "step": 240 + }, + { + "epoch": 0.05, + "learning_rate": 4.8200000000000004e-06, + "loss": 2.6307, + "step": 241 + }, + { + "epoch": 0.05, + "learning_rate": 4.84e-06, + "loss": 2.6417, + "step": 242 + }, + { + "epoch": 0.05, + "learning_rate": 4.86e-06, + "loss": 2.7045, + "step": 243 + }, + { + "epoch": 0.05, + "learning_rate": 4.880000000000001e-06, + "loss": 2.6331, + "step": 244 + }, + { + "epoch": 0.05, + "learning_rate": 4.9000000000000005e-06, + "loss": 2.6103, + "step": 245 + }, + { + "epoch": 0.05, + "learning_rate": 4.92e-06, + "loss": 2.6674, + "step": 246 + }, + { + "epoch": 0.05, + "learning_rate": 4.94e-06, + "loss": 2.7267, + "step": 247 + }, + { + "epoch": 0.05, + "learning_rate": 4.960000000000001e-06, + "loss": 2.5947, + "step": 248 + }, + { + "epoch": 0.05, + "learning_rate": 4.980000000000001e-06, + "loss": 2.6976, + "step": 249 + }, + { + "epoch": 0.05, + "learning_rate": 5e-06, + "loss": 2.7197, + "step": 250 + }, + { + "epoch": 0.05, + "learning_rate": 5.02e-06, + "loss": 2.5839, + "step": 251 + }, + { + "epoch": 0.05, + "learning_rate": 5.04e-06, + "loss": 2.6537, + "step": 252 + }, + { + "epoch": 0.05, + "learning_rate": 5.060000000000001e-06, + "loss": 2.6167, + "step": 253 + }, + { + "epoch": 0.05, + "learning_rate": 5.0800000000000005e-06, + "loss": 2.6318, + "step": 254 + }, + { + "epoch": 0.05, + "learning_rate": 5.1e-06, + "loss": 2.6608, + "step": 255 + }, + { + "epoch": 0.05, + "learning_rate": 5.12e-06, + "loss": 2.6331, + "step": 256 + }, + { + "epoch": 0.05, + "learning_rate": 5.140000000000001e-06, + "loss": 2.629, + "step": 257 + }, + { + "epoch": 0.05, + "learning_rate": 5.1600000000000006e-06, + "loss": 2.6286, + "step": 258 + }, + { + "epoch": 0.05, + "learning_rate": 5.18e-06, + "loss": 2.7031, + "step": 259 + }, + { + "epoch": 0.05, + "learning_rate": 5.2e-06, + "loss": 2.7017, + "step": 260 + }, + { + "epoch": 0.05, + "learning_rate": 5.220000000000001e-06, + "loss": 2.6756, + "step": 261 + }, + { + "epoch": 0.05, + "learning_rate": 5.240000000000001e-06, + "loss": 2.6382, + "step": 262 + }, + { + "epoch": 0.06, + "learning_rate": 5.2600000000000005e-06, + "loss": 2.5726, + "step": 263 + }, + { + "epoch": 0.06, + "learning_rate": 5.28e-06, + "loss": 2.6359, + "step": 264 + }, + { + "epoch": 0.06, + "learning_rate": 5.300000000000001e-06, + "loss": 2.607, + "step": 265 + }, + { + "epoch": 0.06, + "learning_rate": 5.320000000000001e-06, + "loss": 2.7145, + "step": 266 + }, + { + "epoch": 0.06, + "learning_rate": 5.3400000000000005e-06, + "loss": 2.6708, + "step": 267 + }, + { + "epoch": 0.06, + "learning_rate": 5.36e-06, + "loss": 2.6088, + "step": 268 + }, + { + "epoch": 0.06, + "learning_rate": 5.380000000000001e-06, + "loss": 2.7755, + "step": 269 + }, + { + "epoch": 0.06, + "learning_rate": 5.400000000000001e-06, + "loss": 2.639, + "step": 270 + }, + { + "epoch": 0.06, + "learning_rate": 5.420000000000001e-06, + "loss": 2.7144, + "step": 271 + }, + { + "epoch": 0.06, + "learning_rate": 5.4400000000000004e-06, + "loss": 2.6536, + "step": 272 + }, + { + "epoch": 0.06, + "learning_rate": 5.460000000000001e-06, + "loss": 2.6772, + "step": 273 + }, + { + "epoch": 0.06, + "learning_rate": 5.480000000000001e-06, + "loss": 2.6416, + "step": 274 + }, + { + "epoch": 0.06, + "learning_rate": 5.500000000000001e-06, + "loss": 2.6388, + "step": 275 + }, + { + "epoch": 0.06, + "learning_rate": 5.5200000000000005e-06, + "loss": 2.6374, + "step": 276 + }, + { + "epoch": 0.06, + "learning_rate": 5.540000000000001e-06, + "loss": 2.7369, + "step": 277 + }, + { + "epoch": 0.06, + "learning_rate": 5.560000000000001e-06, + "loss": 2.5838, + "step": 278 + }, + { + "epoch": 0.06, + "learning_rate": 5.580000000000001e-06, + "loss": 2.6328, + "step": 279 + }, + { + "epoch": 0.06, + "learning_rate": 5.600000000000001e-06, + "loss": 2.6634, + "step": 280 + }, + { + "epoch": 0.06, + "learning_rate": 5.620000000000001e-06, + "loss": 2.66, + "step": 281 + }, + { + "epoch": 0.06, + "learning_rate": 5.64e-06, + "loss": 2.6362, + "step": 282 + }, + { + "epoch": 0.06, + "learning_rate": 5.66e-06, + "loss": 2.711, + "step": 283 + }, + { + "epoch": 0.06, + "learning_rate": 5.68e-06, + "loss": 2.6019, + "step": 284 + }, + { + "epoch": 0.06, + "learning_rate": 5.7e-06, + "loss": 2.6881, + "step": 285 + }, + { + "epoch": 0.06, + "learning_rate": 5.72e-06, + "loss": 2.7462, + "step": 286 + }, + { + "epoch": 0.06, + "learning_rate": 5.74e-06, + "loss": 2.6608, + "step": 287 + }, + { + "epoch": 0.06, + "learning_rate": 5.76e-06, + "loss": 2.6751, + "step": 288 + }, + { + "epoch": 0.06, + "learning_rate": 5.78e-06, + "loss": 2.7215, + "step": 289 + }, + { + "epoch": 0.06, + "learning_rate": 5.8e-06, + "loss": 2.6746, + "step": 290 + }, + { + "epoch": 0.06, + "learning_rate": 5.82e-06, + "loss": 2.6069, + "step": 291 + }, + { + "epoch": 0.06, + "learning_rate": 5.84e-06, + "loss": 2.6866, + "step": 292 + }, + { + "epoch": 0.06, + "learning_rate": 5.86e-06, + "loss": 2.6567, + "step": 293 + }, + { + "epoch": 0.06, + "learning_rate": 5.8800000000000005e-06, + "loss": 2.659, + "step": 294 + }, + { + "epoch": 0.06, + "learning_rate": 5.9e-06, + "loss": 2.6576, + "step": 295 + }, + { + "epoch": 0.06, + "learning_rate": 5.92e-06, + "loss": 2.7237, + "step": 296 + }, + { + "epoch": 0.06, + "learning_rate": 5.94e-06, + "loss": 2.6565, + "step": 297 + }, + { + "epoch": 0.06, + "learning_rate": 5.9600000000000005e-06, + "loss": 2.6672, + "step": 298 + }, + { + "epoch": 0.06, + "learning_rate": 5.98e-06, + "loss": 2.663, + "step": 299 + }, + { + "epoch": 0.06, + "learning_rate": 6e-06, + "loss": 2.6964, + "step": 300 + }, + { + "epoch": 0.06, + "learning_rate": 6.02e-06, + "loss": 2.6791, + "step": 301 + }, + { + "epoch": 0.06, + "learning_rate": 6.040000000000001e-06, + "loss": 2.6182, + "step": 302 + }, + { + "epoch": 0.06, + "learning_rate": 6.0600000000000004e-06, + "loss": 2.672, + "step": 303 + }, + { + "epoch": 0.06, + "learning_rate": 6.08e-06, + "loss": 2.6063, + "step": 304 + }, + { + "epoch": 0.06, + "learning_rate": 6.1e-06, + "loss": 2.5451, + "step": 305 + }, + { + "epoch": 0.06, + "learning_rate": 6.120000000000001e-06, + "loss": 2.6746, + "step": 306 + }, + { + "epoch": 0.06, + "learning_rate": 6.1400000000000005e-06, + "loss": 2.6669, + "step": 307 + }, + { + "epoch": 0.06, + "learning_rate": 6.16e-06, + "loss": 2.5899, + "step": 308 + }, + { + "epoch": 0.06, + "learning_rate": 6.18e-06, + "loss": 2.6187, + "step": 309 + }, + { + "epoch": 0.07, + "learning_rate": 6.200000000000001e-06, + "loss": 2.6542, + "step": 310 + }, + { + "epoch": 0.07, + "learning_rate": 6.220000000000001e-06, + "loss": 2.6532, + "step": 311 + }, + { + "epoch": 0.07, + "learning_rate": 6.24e-06, + "loss": 2.6557, + "step": 312 + }, + { + "epoch": 0.07, + "learning_rate": 6.26e-06, + "loss": 2.6446, + "step": 313 + }, + { + "epoch": 0.07, + "learning_rate": 6.280000000000001e-06, + "loss": 2.6531, + "step": 314 + }, + { + "epoch": 0.07, + "learning_rate": 6.300000000000001e-06, + "loss": 2.6009, + "step": 315 + }, + { + "epoch": 0.07, + "learning_rate": 6.3200000000000005e-06, + "loss": 2.7386, + "step": 316 + }, + { + "epoch": 0.07, + "learning_rate": 6.34e-06, + "loss": 2.6878, + "step": 317 + }, + { + "epoch": 0.07, + "learning_rate": 6.360000000000001e-06, + "loss": 2.7002, + "step": 318 + }, + { + "epoch": 0.07, + "learning_rate": 6.380000000000001e-06, + "loss": 2.642, + "step": 319 + }, + { + "epoch": 0.07, + "learning_rate": 6.4000000000000006e-06, + "loss": 2.6597, + "step": 320 + }, + { + "epoch": 0.07, + "learning_rate": 6.42e-06, + "loss": 2.7102, + "step": 321 + }, + { + "epoch": 0.07, + "learning_rate": 6.440000000000001e-06, + "loss": 2.5744, + "step": 322 + }, + { + "epoch": 0.07, + "learning_rate": 6.460000000000001e-06, + "loss": 2.6658, + "step": 323 + }, + { + "epoch": 0.07, + "learning_rate": 6.480000000000001e-06, + "loss": 2.7137, + "step": 324 + }, + { + "epoch": 0.07, + "learning_rate": 6.5000000000000004e-06, + "loss": 2.6633, + "step": 325 + }, + { + "epoch": 0.07, + "learning_rate": 6.520000000000001e-06, + "loss": 2.5941, + "step": 326 + }, + { + "epoch": 0.07, + "learning_rate": 6.540000000000001e-06, + "loss": 2.7184, + "step": 327 + }, + { + "epoch": 0.07, + "learning_rate": 6.560000000000001e-06, + "loss": 2.5913, + "step": 328 + }, + { + "epoch": 0.07, + "learning_rate": 6.5800000000000005e-06, + "loss": 2.7405, + "step": 329 + }, + { + "epoch": 0.07, + "learning_rate": 6.600000000000001e-06, + "loss": 2.6598, + "step": 330 + }, + { + "epoch": 0.07, + "learning_rate": 6.620000000000001e-06, + "loss": 2.6853, + "step": 331 + }, + { + "epoch": 0.07, + "learning_rate": 6.640000000000001e-06, + "loss": 2.6237, + "step": 332 + }, + { + "epoch": 0.07, + "learning_rate": 6.660000000000001e-06, + "loss": 2.6798, + "step": 333 + }, + { + "epoch": 0.07, + "learning_rate": 6.680000000000001e-06, + "loss": 2.616, + "step": 334 + }, + { + "epoch": 0.07, + "learning_rate": 6.700000000000001e-06, + "loss": 2.6438, + "step": 335 + }, + { + "epoch": 0.07, + "learning_rate": 6.720000000000001e-06, + "loss": 2.6679, + "step": 336 + }, + { + "epoch": 0.07, + "learning_rate": 6.740000000000001e-06, + "loss": 2.5871, + "step": 337 + }, + { + "epoch": 0.07, + "learning_rate": 6.760000000000001e-06, + "loss": 2.6121, + "step": 338 + }, + { + "epoch": 0.07, + "learning_rate": 6.780000000000001e-06, + "loss": 2.6697, + "step": 339 + }, + { + "epoch": 0.07, + "learning_rate": 6.800000000000001e-06, + "loss": 2.5997, + "step": 340 + }, + { + "epoch": 0.07, + "learning_rate": 6.820000000000001e-06, + "loss": 2.598, + "step": 341 + }, + { + "epoch": 0.07, + "learning_rate": 6.8400000000000014e-06, + "loss": 2.6637, + "step": 342 + }, + { + "epoch": 0.07, + "learning_rate": 6.860000000000001e-06, + "loss": 2.6786, + "step": 343 + }, + { + "epoch": 0.07, + "learning_rate": 6.88e-06, + "loss": 2.6308, + "step": 344 + }, + { + "epoch": 0.07, + "learning_rate": 6.9e-06, + "loss": 2.6611, + "step": 345 + }, + { + "epoch": 0.07, + "learning_rate": 6.92e-06, + "loss": 2.5419, + "step": 346 + }, + { + "epoch": 0.07, + "learning_rate": 6.9400000000000005e-06, + "loss": 2.6201, + "step": 347 + }, + { + "epoch": 0.07, + "learning_rate": 6.96e-06, + "loss": 2.7119, + "step": 348 + }, + { + "epoch": 0.07, + "learning_rate": 6.98e-06, + "loss": 2.6926, + "step": 349 + }, + { + "epoch": 0.07, + "learning_rate": 7e-06, + "loss": 2.6435, + "step": 350 + }, + { + "epoch": 0.07, + "learning_rate": 7.0200000000000006e-06, + "loss": 2.6953, + "step": 351 + }, + { + "epoch": 0.07, + "learning_rate": 7.04e-06, + "loss": 2.6546, + "step": 352 + }, + { + "epoch": 0.07, + "learning_rate": 7.06e-06, + "loss": 2.654, + "step": 353 + }, + { + "epoch": 0.07, + "learning_rate": 7.08e-06, + "loss": 2.6902, + "step": 354 + }, + { + "epoch": 0.07, + "learning_rate": 7.100000000000001e-06, + "loss": 2.6178, + "step": 355 + }, + { + "epoch": 0.07, + "learning_rate": 7.1200000000000004e-06, + "loss": 2.6457, + "step": 356 + }, + { + "epoch": 0.07, + "learning_rate": 7.14e-06, + "loss": 2.6567, + "step": 357 + }, + { + "epoch": 0.08, + "learning_rate": 7.16e-06, + "loss": 2.6741, + "step": 358 + }, + { + "epoch": 0.08, + "learning_rate": 7.180000000000001e-06, + "loss": 2.626, + "step": 359 + }, + { + "epoch": 0.08, + "learning_rate": 7.2000000000000005e-06, + "loss": 2.6413, + "step": 360 + }, + { + "epoch": 0.08, + "learning_rate": 7.22e-06, + "loss": 2.6331, + "step": 361 + }, + { + "epoch": 0.08, + "learning_rate": 7.24e-06, + "loss": 2.6287, + "step": 362 + }, + { + "epoch": 0.08, + "learning_rate": 7.260000000000001e-06, + "loss": 2.6245, + "step": 363 + }, + { + "epoch": 0.08, + "learning_rate": 7.280000000000001e-06, + "loss": 2.6825, + "step": 364 + }, + { + "epoch": 0.08, + "learning_rate": 7.3e-06, + "loss": 2.6394, + "step": 365 + }, + { + "epoch": 0.08, + "learning_rate": 7.32e-06, + "loss": 2.6457, + "step": 366 + }, + { + "epoch": 0.08, + "learning_rate": 7.340000000000001e-06, + "loss": 2.6285, + "step": 367 + }, + { + "epoch": 0.08, + "learning_rate": 7.360000000000001e-06, + "loss": 2.6595, + "step": 368 + }, + { + "epoch": 0.08, + "learning_rate": 7.3800000000000005e-06, + "loss": 2.6503, + "step": 369 + }, + { + "epoch": 0.08, + "learning_rate": 7.4e-06, + "loss": 2.753, + "step": 370 + }, + { + "epoch": 0.08, + "learning_rate": 7.420000000000001e-06, + "loss": 2.6086, + "step": 371 + }, + { + "epoch": 0.08, + "learning_rate": 7.440000000000001e-06, + "loss": 2.6348, + "step": 372 + }, + { + "epoch": 0.08, + "learning_rate": 7.4600000000000006e-06, + "loss": 2.6628, + "step": 373 + }, + { + "epoch": 0.08, + "learning_rate": 7.48e-06, + "loss": 2.6399, + "step": 374 + }, + { + "epoch": 0.08, + "learning_rate": 7.500000000000001e-06, + "loss": 2.6811, + "step": 375 + }, + { + "epoch": 0.08, + "learning_rate": 7.520000000000001e-06, + "loss": 2.6206, + "step": 376 + }, + { + "epoch": 0.08, + "learning_rate": 7.540000000000001e-06, + "loss": 2.6644, + "step": 377 + }, + { + "epoch": 0.08, + "learning_rate": 7.5600000000000005e-06, + "loss": 2.5965, + "step": 378 + }, + { + "epoch": 0.08, + "learning_rate": 7.58e-06, + "loss": 2.6529, + "step": 379 + }, + { + "epoch": 0.08, + "learning_rate": 7.600000000000001e-06, + "loss": 2.5284, + "step": 380 + }, + { + "epoch": 0.08, + "learning_rate": 7.620000000000001e-06, + "loss": 2.635, + "step": 381 + }, + { + "epoch": 0.08, + "learning_rate": 7.640000000000001e-06, + "loss": 2.6574, + "step": 382 + }, + { + "epoch": 0.08, + "learning_rate": 7.660000000000001e-06, + "loss": 2.6826, + "step": 383 + }, + { + "epoch": 0.08, + "learning_rate": 7.680000000000001e-06, + "loss": 2.6407, + "step": 384 + }, + { + "epoch": 0.08, + "learning_rate": 7.7e-06, + "loss": 2.642, + "step": 385 + }, + { + "epoch": 0.08, + "learning_rate": 7.72e-06, + "loss": 2.6291, + "step": 386 + }, + { + "epoch": 0.08, + "learning_rate": 7.74e-06, + "loss": 2.6691, + "step": 387 + }, + { + "epoch": 0.08, + "learning_rate": 7.76e-06, + "loss": 2.6138, + "step": 388 + }, + { + "epoch": 0.08, + "learning_rate": 7.78e-06, + "loss": 2.6328, + "step": 389 + }, + { + "epoch": 0.08, + "learning_rate": 7.800000000000002e-06, + "loss": 2.6893, + "step": 390 + }, + { + "epoch": 0.08, + "learning_rate": 7.820000000000001e-06, + "loss": 2.527, + "step": 391 + }, + { + "epoch": 0.08, + "learning_rate": 7.840000000000001e-06, + "loss": 2.7547, + "step": 392 + }, + { + "epoch": 0.08, + "learning_rate": 7.860000000000001e-06, + "loss": 2.7153, + "step": 393 + }, + { + "epoch": 0.08, + "learning_rate": 7.88e-06, + "loss": 2.6948, + "step": 394 + }, + { + "epoch": 0.08, + "learning_rate": 7.9e-06, + "loss": 2.6645, + "step": 395 + }, + { + "epoch": 0.08, + "learning_rate": 7.92e-06, + "loss": 2.5865, + "step": 396 + }, + { + "epoch": 0.08, + "learning_rate": 7.94e-06, + "loss": 2.5877, + "step": 397 + }, + { + "epoch": 0.08, + "learning_rate": 7.960000000000002e-06, + "loss": 2.6676, + "step": 398 + }, + { + "epoch": 0.08, + "learning_rate": 7.980000000000002e-06, + "loss": 2.6788, + "step": 399 + }, + { + "epoch": 0.08, + "learning_rate": 8.000000000000001e-06, + "loss": 2.6323, + "step": 400 + }, + { + "epoch": 0.08, + "learning_rate": 8.020000000000001e-06, + "loss": 2.6804, + "step": 401 + }, + { + "epoch": 0.08, + "learning_rate": 8.040000000000001e-06, + "loss": 2.7035, + "step": 402 + }, + { + "epoch": 0.08, + "learning_rate": 8.06e-06, + "loss": 2.6179, + "step": 403 + }, + { + "epoch": 0.08, + "learning_rate": 8.08e-06, + "loss": 2.6633, + "step": 404 + }, + { + "epoch": 0.08, + "learning_rate": 8.1e-06, + "loss": 2.6037, + "step": 405 + }, + { + "epoch": 0.09, + "learning_rate": 8.120000000000002e-06, + "loss": 2.7244, + "step": 406 + }, + { + "epoch": 0.09, + "learning_rate": 8.14e-06, + "loss": 2.5889, + "step": 407 + }, + { + "epoch": 0.09, + "learning_rate": 8.16e-06, + "loss": 2.713, + "step": 408 + }, + { + "epoch": 0.09, + "learning_rate": 8.18e-06, + "loss": 2.589, + "step": 409 + }, + { + "epoch": 0.09, + "learning_rate": 8.2e-06, + "loss": 2.6463, + "step": 410 + }, + { + "epoch": 0.09, + "learning_rate": 8.220000000000001e-06, + "loss": 2.7026, + "step": 411 + }, + { + "epoch": 0.09, + "learning_rate": 8.24e-06, + "loss": 2.6963, + "step": 412 + }, + { + "epoch": 0.09, + "learning_rate": 8.26e-06, + "loss": 2.6659, + "step": 413 + }, + { + "epoch": 0.09, + "learning_rate": 8.28e-06, + "loss": 2.6471, + "step": 414 + }, + { + "epoch": 0.09, + "learning_rate": 8.3e-06, + "loss": 2.6469, + "step": 415 + }, + { + "epoch": 0.09, + "learning_rate": 8.32e-06, + "loss": 2.6439, + "step": 416 + }, + { + "epoch": 0.09, + "learning_rate": 8.34e-06, + "loss": 2.623, + "step": 417 + }, + { + "epoch": 0.09, + "learning_rate": 8.36e-06, + "loss": 2.6562, + "step": 418 + }, + { + "epoch": 0.09, + "learning_rate": 8.380000000000001e-06, + "loss": 2.669, + "step": 419 + }, + { + "epoch": 0.09, + "learning_rate": 8.400000000000001e-06, + "loss": 2.6713, + "step": 420 + }, + { + "epoch": 0.09, + "learning_rate": 8.42e-06, + "loss": 2.6205, + "step": 421 + }, + { + "epoch": 0.09, + "learning_rate": 8.44e-06, + "loss": 2.5721, + "step": 422 + }, + { + "epoch": 0.09, + "learning_rate": 8.46e-06, + "loss": 2.626, + "step": 423 + }, + { + "epoch": 0.09, + "learning_rate": 8.48e-06, + "loss": 2.6611, + "step": 424 + }, + { + "epoch": 0.09, + "learning_rate": 8.5e-06, + "loss": 2.6192, + "step": 425 + }, + { + "epoch": 0.09, + "learning_rate": 8.52e-06, + "loss": 2.6562, + "step": 426 + }, + { + "epoch": 0.09, + "learning_rate": 8.540000000000001e-06, + "loss": 2.6378, + "step": 427 + }, + { + "epoch": 0.09, + "learning_rate": 8.560000000000001e-06, + "loss": 2.6421, + "step": 428 + }, + { + "epoch": 0.09, + "learning_rate": 8.580000000000001e-06, + "loss": 2.6713, + "step": 429 + }, + { + "epoch": 0.09, + "learning_rate": 8.6e-06, + "loss": 2.7092, + "step": 430 + }, + { + "epoch": 0.09, + "learning_rate": 8.62e-06, + "loss": 2.6449, + "step": 431 + }, + { + "epoch": 0.09, + "learning_rate": 8.64e-06, + "loss": 2.6893, + "step": 432 + }, + { + "epoch": 0.09, + "learning_rate": 8.66e-06, + "loss": 2.6157, + "step": 433 + }, + { + "epoch": 0.09, + "learning_rate": 8.68e-06, + "loss": 2.597, + "step": 434 + }, + { + "epoch": 0.09, + "learning_rate": 8.700000000000001e-06, + "loss": 2.6545, + "step": 435 + }, + { + "epoch": 0.09, + "learning_rate": 8.720000000000001e-06, + "loss": 2.6653, + "step": 436 + }, + { + "epoch": 0.09, + "learning_rate": 8.740000000000001e-06, + "loss": 2.6919, + "step": 437 + }, + { + "epoch": 0.09, + "learning_rate": 8.76e-06, + "loss": 2.6525, + "step": 438 + }, + { + "epoch": 0.09, + "learning_rate": 8.78e-06, + "loss": 2.582, + "step": 439 + }, + { + "epoch": 0.09, + "learning_rate": 8.8e-06, + "loss": 2.6406, + "step": 440 + }, + { + "epoch": 0.09, + "learning_rate": 8.82e-06, + "loss": 2.6308, + "step": 441 + }, + { + "epoch": 0.09, + "learning_rate": 8.84e-06, + "loss": 2.6115, + "step": 442 + }, + { + "epoch": 0.09, + "learning_rate": 8.860000000000002e-06, + "loss": 2.6383, + "step": 443 + }, + { + "epoch": 0.09, + "learning_rate": 8.880000000000001e-06, + "loss": 2.6466, + "step": 444 + }, + { + "epoch": 0.09, + "learning_rate": 8.900000000000001e-06, + "loss": 2.6419, + "step": 445 + }, + { + "epoch": 0.09, + "learning_rate": 8.920000000000001e-06, + "loss": 2.6747, + "step": 446 + }, + { + "epoch": 0.09, + "learning_rate": 8.94e-06, + "loss": 2.5933, + "step": 447 + }, + { + "epoch": 0.09, + "learning_rate": 8.96e-06, + "loss": 2.6729, + "step": 448 + }, + { + "epoch": 0.09, + "learning_rate": 8.98e-06, + "loss": 2.6456, + "step": 449 + }, + { + "epoch": 0.09, + "learning_rate": 9e-06, + "loss": 2.6645, + "step": 450 + }, + { + "epoch": 0.09, + "learning_rate": 9.020000000000002e-06, + "loss": 2.5955, + "step": 451 + }, + { + "epoch": 0.09, + "learning_rate": 9.040000000000002e-06, + "loss": 2.6567, + "step": 452 + }, + { + "epoch": 0.09, + "learning_rate": 9.060000000000001e-06, + "loss": 2.6677, + "step": 453 + }, + { + "epoch": 0.1, + "learning_rate": 9.080000000000001e-06, + "loss": 2.6252, + "step": 454 + }, + { + "epoch": 0.1, + "learning_rate": 9.100000000000001e-06, + "loss": 2.6543, + "step": 455 + }, + { + "epoch": 0.1, + "learning_rate": 9.12e-06, + "loss": 2.616, + "step": 456 + }, + { + "epoch": 0.1, + "learning_rate": 9.14e-06, + "loss": 2.67, + "step": 457 + }, + { + "epoch": 0.1, + "learning_rate": 9.16e-06, + "loss": 2.5737, + "step": 458 + }, + { + "epoch": 0.1, + "learning_rate": 9.180000000000002e-06, + "loss": 2.6604, + "step": 459 + }, + { + "epoch": 0.1, + "learning_rate": 9.200000000000002e-06, + "loss": 2.6908, + "step": 460 + }, + { + "epoch": 0.1, + "learning_rate": 9.220000000000002e-06, + "loss": 2.6625, + "step": 461 + }, + { + "epoch": 0.1, + "learning_rate": 9.240000000000001e-06, + "loss": 2.5731, + "step": 462 + }, + { + "epoch": 0.1, + "learning_rate": 9.260000000000001e-06, + "loss": 2.6091, + "step": 463 + }, + { + "epoch": 0.1, + "learning_rate": 9.280000000000001e-06, + "loss": 2.6811, + "step": 464 + }, + { + "epoch": 0.1, + "learning_rate": 9.3e-06, + "loss": 2.6964, + "step": 465 + }, + { + "epoch": 0.1, + "learning_rate": 9.32e-06, + "loss": 2.6075, + "step": 466 + }, + { + "epoch": 0.1, + "learning_rate": 9.340000000000002e-06, + "loss": 2.6927, + "step": 467 + }, + { + "epoch": 0.1, + "learning_rate": 9.360000000000002e-06, + "loss": 2.6686, + "step": 468 + }, + { + "epoch": 0.1, + "learning_rate": 9.38e-06, + "loss": 2.6425, + "step": 469 + }, + { + "epoch": 0.1, + "learning_rate": 9.4e-06, + "loss": 2.7061, + "step": 470 + }, + { + "epoch": 0.1, + "learning_rate": 9.42e-06, + "loss": 2.6467, + "step": 471 + }, + { + "epoch": 0.1, + "learning_rate": 9.440000000000001e-06, + "loss": 2.5739, + "step": 472 + }, + { + "epoch": 0.1, + "learning_rate": 9.460000000000001e-06, + "loss": 2.6558, + "step": 473 + }, + { + "epoch": 0.1, + "learning_rate": 9.48e-06, + "loss": 2.6405, + "step": 474 + }, + { + "epoch": 0.1, + "learning_rate": 9.5e-06, + "loss": 2.667, + "step": 475 + }, + { + "epoch": 0.1, + "learning_rate": 9.52e-06, + "loss": 2.6547, + "step": 476 + }, + { + "epoch": 0.1, + "learning_rate": 9.54e-06, + "loss": 2.6691, + "step": 477 + }, + { + "epoch": 0.1, + "learning_rate": 9.56e-06, + "loss": 2.6919, + "step": 478 + }, + { + "epoch": 0.1, + "learning_rate": 9.58e-06, + "loss": 2.6178, + "step": 479 + }, + { + "epoch": 0.1, + "learning_rate": 9.600000000000001e-06, + "loss": 2.6346, + "step": 480 + }, + { + "epoch": 0.1, + "learning_rate": 9.620000000000001e-06, + "loss": 2.6164, + "step": 481 + }, + { + "epoch": 0.1, + "learning_rate": 9.640000000000001e-06, + "loss": 2.6418, + "step": 482 + }, + { + "epoch": 0.1, + "learning_rate": 9.66e-06, + "loss": 2.639, + "step": 483 + }, + { + "epoch": 0.1, + "learning_rate": 9.68e-06, + "loss": 2.581, + "step": 484 + }, + { + "epoch": 0.1, + "learning_rate": 9.7e-06, + "loss": 2.598, + "step": 485 + }, + { + "epoch": 0.1, + "learning_rate": 9.72e-06, + "loss": 2.6466, + "step": 486 + }, + { + "epoch": 0.1, + "learning_rate": 9.74e-06, + "loss": 2.6558, + "step": 487 + }, + { + "epoch": 0.1, + "learning_rate": 9.760000000000001e-06, + "loss": 2.6934, + "step": 488 + }, + { + "epoch": 0.1, + "learning_rate": 9.780000000000001e-06, + "loss": 2.6804, + "step": 489 + }, + { + "epoch": 0.1, + "learning_rate": 9.800000000000001e-06, + "loss": 2.6494, + "step": 490 + }, + { + "epoch": 0.1, + "learning_rate": 9.820000000000001e-06, + "loss": 2.6624, + "step": 491 + }, + { + "epoch": 0.1, + "learning_rate": 9.84e-06, + "loss": 2.5917, + "step": 492 + }, + { + "epoch": 0.1, + "learning_rate": 9.86e-06, + "loss": 2.6236, + "step": 493 + }, + { + "epoch": 0.1, + "learning_rate": 9.88e-06, + "loss": 2.6067, + "step": 494 + }, + { + "epoch": 0.1, + "learning_rate": 9.9e-06, + "loss": 2.6552, + "step": 495 + }, + { + "epoch": 0.1, + "learning_rate": 9.920000000000002e-06, + "loss": 2.5844, + "step": 496 + }, + { + "epoch": 0.1, + "learning_rate": 9.940000000000001e-06, + "loss": 2.6031, + "step": 497 + }, + { + "epoch": 0.1, + "learning_rate": 9.960000000000001e-06, + "loss": 2.6164, + "step": 498 + }, + { + "epoch": 0.1, + "learning_rate": 9.980000000000001e-06, + "loss": 2.5958, + "step": 499 + }, + { + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 2.5818, + "step": 500 + }, + { + "epoch": 0.11, + "learning_rate": 9.999998646097393e-06, + "loss": 2.6299, + "step": 501 + }, + { + "epoch": 0.11, + "learning_rate": 9.999994584390302e-06, + "loss": 2.7155, + "step": 502 + }, + { + "epoch": 0.11, + "learning_rate": 9.999987814880927e-06, + "loss": 2.6293, + "step": 503 + }, + { + "epoch": 0.11, + "learning_rate": 9.999978337572935e-06, + "loss": 2.6665, + "step": 504 + }, + { + "epoch": 0.11, + "learning_rate": 9.99996615247146e-06, + "loss": 2.6861, + "step": 505 + }, + { + "epoch": 0.11, + "learning_rate": 9.999951259583096e-06, + "loss": 2.6985, + "step": 506 + }, + { + "epoch": 0.11, + "learning_rate": 9.999933658915914e-06, + "loss": 2.6425, + "step": 507 + }, + { + "epoch": 0.11, + "learning_rate": 9.999913350479443e-06, + "loss": 2.6117, + "step": 508 + }, + { + "epoch": 0.11, + "learning_rate": 9.999890334284681e-06, + "loss": 2.6097, + "step": 509 + }, + { + "epoch": 0.11, + "learning_rate": 9.999864610344095e-06, + "loss": 2.6846, + "step": 510 + }, + { + "epoch": 0.11, + "learning_rate": 9.999836178671614e-06, + "loss": 2.7102, + "step": 511 + }, + { + "epoch": 0.11, + "learning_rate": 9.999805039282636e-06, + "loss": 2.5869, + "step": 512 + }, + { + "epoch": 0.11, + "learning_rate": 9.999771192194025e-06, + "loss": 2.7877, + "step": 513 + }, + { + "epoch": 0.11, + "learning_rate": 9.99973463742411e-06, + "loss": 2.6353, + "step": 514 + }, + { + "epoch": 0.11, + "learning_rate": 9.99969537499269e-06, + "loss": 2.6514, + "step": 515 + }, + { + "epoch": 0.11, + "learning_rate": 9.999653404921026e-06, + "loss": 2.6014, + "step": 516 + }, + { + "epoch": 0.11, + "learning_rate": 9.999608727231848e-06, + "loss": 2.6663, + "step": 517 + }, + { + "epoch": 0.11, + "learning_rate": 9.999561341949354e-06, + "loss": 2.6809, + "step": 518 + }, + { + "epoch": 0.11, + "learning_rate": 9.999511249099201e-06, + "loss": 2.6582, + "step": 519 + }, + { + "epoch": 0.11, + "learning_rate": 9.999458448708522e-06, + "loss": 2.707, + "step": 520 + }, + { + "epoch": 0.11, + "learning_rate": 9.999402940805908e-06, + "loss": 2.6701, + "step": 521 + }, + { + "epoch": 0.11, + "learning_rate": 9.999344725421425e-06, + "loss": 2.5734, + "step": 522 + }, + { + "epoch": 0.11, + "learning_rate": 9.999283802586595e-06, + "loss": 2.5933, + "step": 523 + }, + { + "epoch": 0.11, + "learning_rate": 9.999220172334414e-06, + "loss": 2.5568, + "step": 524 + }, + { + "epoch": 0.11, + "learning_rate": 9.99915383469934e-06, + "loss": 2.614, + "step": 525 + }, + { + "epoch": 0.11, + "learning_rate": 9.999084789717299e-06, + "loss": 2.581, + "step": 526 + }, + { + "epoch": 0.11, + "learning_rate": 9.999013037425686e-06, + "loss": 2.6889, + "step": 527 + }, + { + "epoch": 0.11, + "learning_rate": 9.998938577863357e-06, + "loss": 2.641, + "step": 528 + }, + { + "epoch": 0.11, + "learning_rate": 9.998861411070636e-06, + "loss": 2.6743, + "step": 529 + }, + { + "epoch": 0.11, + "learning_rate": 9.998781537089316e-06, + "loss": 2.6075, + "step": 530 + }, + { + "epoch": 0.11, + "learning_rate": 9.998698955962649e-06, + "loss": 2.6413, + "step": 531 + }, + { + "epoch": 0.11, + "learning_rate": 9.998613667735361e-06, + "loss": 2.6764, + "step": 532 + }, + { + "epoch": 0.11, + "learning_rate": 9.998525672453642e-06, + "loss": 2.6931, + "step": 533 + }, + { + "epoch": 0.11, + "learning_rate": 9.998434970165145e-06, + "loss": 2.6283, + "step": 534 + }, + { + "epoch": 0.11, + "learning_rate": 9.99834156091899e-06, + "loss": 2.6365, + "step": 535 + }, + { + "epoch": 0.11, + "learning_rate": 9.998245444765764e-06, + "loss": 2.6418, + "step": 536 + }, + { + "epoch": 0.11, + "learning_rate": 9.998146621757522e-06, + "loss": 2.5555, + "step": 537 + }, + { + "epoch": 0.11, + "learning_rate": 9.99804509194778e-06, + "loss": 2.612, + "step": 538 + }, + { + "epoch": 0.11, + "learning_rate": 9.997940855391525e-06, + "loss": 2.6419, + "step": 539 + }, + { + "epoch": 0.11, + "learning_rate": 9.997833912145206e-06, + "loss": 2.6567, + "step": 540 + }, + { + "epoch": 0.11, + "learning_rate": 9.997724262266738e-06, + "loss": 2.5447, + "step": 541 + }, + { + "epoch": 0.11, + "learning_rate": 9.997611905815508e-06, + "loss": 2.71, + "step": 542 + }, + { + "epoch": 0.11, + "learning_rate": 9.99749684285236e-06, + "loss": 2.6079, + "step": 543 + }, + { + "epoch": 0.11, + "learning_rate": 9.997379073439604e-06, + "loss": 2.6233, + "step": 544 + }, + { + "epoch": 0.11, + "learning_rate": 9.997258597641027e-06, + "loss": 2.6078, + "step": 545 + }, + { + "epoch": 0.11, + "learning_rate": 9.99713541552187e-06, + "loss": 2.644, + "step": 546 + }, + { + "epoch": 0.11, + "learning_rate": 9.997009527148845e-06, + "loss": 2.6422, + "step": 547 + }, + { + "epoch": 0.11, + "learning_rate": 9.996880932590125e-06, + "loss": 2.6723, + "step": 548 + }, + { + "epoch": 0.12, + "learning_rate": 9.996749631915355e-06, + "loss": 2.6505, + "step": 549 + }, + { + "epoch": 0.12, + "learning_rate": 9.996615625195643e-06, + "loss": 2.6895, + "step": 550 + }, + { + "epoch": 0.12, + "learning_rate": 9.996478912503557e-06, + "loss": 2.6951, + "step": 551 + }, + { + "epoch": 0.12, + "learning_rate": 9.996339493913143e-06, + "loss": 2.7075, + "step": 552 + }, + { + "epoch": 0.12, + "learning_rate": 9.996197369499898e-06, + "loss": 2.6155, + "step": 553 + }, + { + "epoch": 0.12, + "learning_rate": 9.996052539340793e-06, + "loss": 2.6419, + "step": 554 + }, + { + "epoch": 0.12, + "learning_rate": 9.995905003514262e-06, + "loss": 2.7584, + "step": 555 + }, + { + "epoch": 0.12, + "learning_rate": 9.995754762100207e-06, + "loss": 2.5918, + "step": 556 + }, + { + "epoch": 0.12, + "learning_rate": 9.99560181517999e-06, + "loss": 2.6495, + "step": 557 + }, + { + "epoch": 0.12, + "learning_rate": 9.995446162836443e-06, + "loss": 2.6874, + "step": 558 + }, + { + "epoch": 0.12, + "learning_rate": 9.99528780515386e-06, + "loss": 2.7173, + "step": 559 + }, + { + "epoch": 0.12, + "learning_rate": 9.995126742218002e-06, + "loss": 2.6463, + "step": 560 + }, + { + "epoch": 0.12, + "learning_rate": 9.994962974116097e-06, + "loss": 2.7188, + "step": 561 + }, + { + "epoch": 0.12, + "learning_rate": 9.994796500936829e-06, + "loss": 2.7249, + "step": 562 + }, + { + "epoch": 0.12, + "learning_rate": 9.994627322770358e-06, + "loss": 2.656, + "step": 563 + }, + { + "epoch": 0.12, + "learning_rate": 9.994455439708304e-06, + "loss": 2.6656, + "step": 564 + }, + { + "epoch": 0.12, + "learning_rate": 9.994280851843751e-06, + "loss": 2.6373, + "step": 565 + }, + { + "epoch": 0.12, + "learning_rate": 9.994103559271252e-06, + "loss": 2.6844, + "step": 566 + }, + { + "epoch": 0.12, + "learning_rate": 9.993923562086817e-06, + "loss": 2.6477, + "step": 567 + }, + { + "epoch": 0.12, + "learning_rate": 9.993740860387928e-06, + "loss": 2.63, + "step": 568 + }, + { + "epoch": 0.12, + "learning_rate": 9.99355545427353e-06, + "loss": 2.6357, + "step": 569 + }, + { + "epoch": 0.12, + "learning_rate": 9.993367343844029e-06, + "loss": 2.6499, + "step": 570 + }, + { + "epoch": 0.12, + "learning_rate": 9.993176529201303e-06, + "loss": 2.6594, + "step": 571 + }, + { + "epoch": 0.12, + "learning_rate": 9.992983010448684e-06, + "loss": 2.5985, + "step": 572 + }, + { + "epoch": 0.12, + "learning_rate": 9.992786787690978e-06, + "loss": 2.662, + "step": 573 + }, + { + "epoch": 0.12, + "learning_rate": 9.99258786103445e-06, + "loss": 2.7439, + "step": 574 + }, + { + "epoch": 0.12, + "learning_rate": 9.992386230586832e-06, + "loss": 2.6223, + "step": 575 + }, + { + "epoch": 0.12, + "learning_rate": 9.992181896457318e-06, + "loss": 2.6464, + "step": 576 + }, + { + "epoch": 0.12, + "learning_rate": 9.991974858756568e-06, + "loss": 2.6552, + "step": 577 + }, + { + "epoch": 0.12, + "learning_rate": 9.991765117596705e-06, + "loss": 2.6463, + "step": 578 + }, + { + "epoch": 0.12, + "learning_rate": 9.991552673091321e-06, + "loss": 2.5778, + "step": 579 + }, + { + "epoch": 0.12, + "learning_rate": 9.99133752535546e-06, + "loss": 2.6233, + "step": 580 + }, + { + "epoch": 0.12, + "learning_rate": 9.991119674505643e-06, + "loss": 2.6629, + "step": 581 + }, + { + "epoch": 0.12, + "learning_rate": 9.990899120659847e-06, + "loss": 2.6217, + "step": 582 + }, + { + "epoch": 0.12, + "learning_rate": 9.990675863937518e-06, + "loss": 2.6548, + "step": 583 + }, + { + "epoch": 0.12, + "learning_rate": 9.990449904459561e-06, + "loss": 2.6158, + "step": 584 + }, + { + "epoch": 0.12, + "learning_rate": 9.990221242348348e-06, + "loss": 2.6694, + "step": 585 + }, + { + "epoch": 0.12, + "learning_rate": 9.989989877727713e-06, + "loss": 2.598, + "step": 586 + }, + { + "epoch": 0.12, + "learning_rate": 9.989755810722951e-06, + "loss": 2.6575, + "step": 587 + }, + { + "epoch": 0.12, + "learning_rate": 9.989519041460831e-06, + "loss": 2.6573, + "step": 588 + }, + { + "epoch": 0.12, + "learning_rate": 9.989279570069572e-06, + "loss": 2.7157, + "step": 589 + }, + { + "epoch": 0.12, + "learning_rate": 9.989037396678863e-06, + "loss": 2.7018, + "step": 590 + }, + { + "epoch": 0.12, + "learning_rate": 9.988792521419858e-06, + "loss": 2.6644, + "step": 591 + }, + { + "epoch": 0.12, + "learning_rate": 9.98854494442517e-06, + "loss": 2.6572, + "step": 592 + }, + { + "epoch": 0.12, + "learning_rate": 9.988294665828877e-06, + "loss": 2.6205, + "step": 593 + }, + { + "epoch": 0.12, + "learning_rate": 9.988041685766521e-06, + "loss": 2.637, + "step": 594 + }, + { + "epoch": 0.12, + "learning_rate": 9.987786004375106e-06, + "loss": 2.6677, + "step": 595 + }, + { + "epoch": 0.12, + "learning_rate": 9.9875276217931e-06, + "loss": 2.6747, + "step": 596 + }, + { + "epoch": 0.13, + "learning_rate": 9.98726653816043e-06, + "loss": 2.6453, + "step": 597 + }, + { + "epoch": 0.13, + "learning_rate": 9.987002753618493e-06, + "loss": 2.6028, + "step": 598 + }, + { + "epoch": 0.13, + "learning_rate": 9.98673626831014e-06, + "loss": 2.5864, + "step": 599 + }, + { + "epoch": 0.13, + "learning_rate": 9.986467082379692e-06, + "loss": 2.648, + "step": 600 + }, + { + "epoch": 0.13, + "learning_rate": 9.986195195972928e-06, + "loss": 2.6394, + "step": 601 + }, + { + "epoch": 0.13, + "learning_rate": 9.985920609237092e-06, + "loss": 2.5973, + "step": 602 + }, + { + "epoch": 0.13, + "learning_rate": 9.98564332232089e-06, + "loss": 2.6454, + "step": 603 + }, + { + "epoch": 0.13, + "learning_rate": 9.98536333537449e-06, + "loss": 2.6416, + "step": 604 + }, + { + "epoch": 0.13, + "learning_rate": 9.98508064854952e-06, + "loss": 2.7182, + "step": 605 + }, + { + "epoch": 0.13, + "learning_rate": 9.984795261999072e-06, + "loss": 2.6655, + "step": 606 + }, + { + "epoch": 0.13, + "learning_rate": 9.984507175877704e-06, + "loss": 2.6902, + "step": 607 + }, + { + "epoch": 0.13, + "learning_rate": 9.984216390341428e-06, + "loss": 2.6469, + "step": 608 + }, + { + "epoch": 0.13, + "learning_rate": 9.983922905547724e-06, + "loss": 2.6633, + "step": 609 + }, + { + "epoch": 0.13, + "learning_rate": 9.983626721655533e-06, + "loss": 2.6387, + "step": 610 + }, + { + "epoch": 0.13, + "learning_rate": 9.983327838825256e-06, + "loss": 2.6952, + "step": 611 + }, + { + "epoch": 0.13, + "learning_rate": 9.983026257218753e-06, + "loss": 2.6782, + "step": 612 + }, + { + "epoch": 0.13, + "learning_rate": 9.982721976999354e-06, + "loss": 2.6319, + "step": 613 + }, + { + "epoch": 0.13, + "learning_rate": 9.982414998331842e-06, + "loss": 2.6287, + "step": 614 + }, + { + "epoch": 0.13, + "learning_rate": 9.982105321382467e-06, + "loss": 2.6911, + "step": 615 + }, + { + "epoch": 0.13, + "learning_rate": 9.981792946318934e-06, + "loss": 2.6469, + "step": 616 + }, + { + "epoch": 0.13, + "learning_rate": 9.981477873310416e-06, + "loss": 2.6138, + "step": 617 + }, + { + "epoch": 0.13, + "learning_rate": 9.981160102527545e-06, + "loss": 2.5851, + "step": 618 + }, + { + "epoch": 0.13, + "learning_rate": 9.980839634142415e-06, + "loss": 2.6497, + "step": 619 + }, + { + "epoch": 0.13, + "learning_rate": 9.980516468328571e-06, + "loss": 2.7079, + "step": 620 + }, + { + "epoch": 0.13, + "learning_rate": 9.980190605261037e-06, + "loss": 2.65, + "step": 621 + }, + { + "epoch": 0.13, + "learning_rate": 9.979862045116282e-06, + "loss": 2.6805, + "step": 622 + }, + { + "epoch": 0.13, + "learning_rate": 9.979530788072241e-06, + "loss": 2.6551, + "step": 623 + }, + { + "epoch": 0.13, + "learning_rate": 9.979196834308313e-06, + "loss": 2.7097, + "step": 624 + }, + { + "epoch": 0.13, + "learning_rate": 9.978860184005352e-06, + "loss": 2.6897, + "step": 625 + }, + { + "epoch": 0.13, + "learning_rate": 9.978520837345678e-06, + "loss": 2.6049, + "step": 626 + }, + { + "epoch": 0.13, + "learning_rate": 9.978178794513063e-06, + "loss": 2.6248, + "step": 627 + }, + { + "epoch": 0.13, + "learning_rate": 9.977834055692746e-06, + "loss": 2.6877, + "step": 628 + }, + { + "epoch": 0.13, + "learning_rate": 9.977486621071425e-06, + "loss": 2.6936, + "step": 629 + }, + { + "epoch": 0.13, + "learning_rate": 9.97713649083726e-06, + "loss": 2.6283, + "step": 630 + }, + { + "epoch": 0.13, + "learning_rate": 9.976783665179862e-06, + "loss": 2.6689, + "step": 631 + }, + { + "epoch": 0.13, + "learning_rate": 9.97642814429031e-06, + "loss": 2.6762, + "step": 632 + }, + { + "epoch": 0.13, + "learning_rate": 9.97606992836114e-06, + "loss": 2.6635, + "step": 633 + }, + { + "epoch": 0.13, + "learning_rate": 9.97570901758635e-06, + "loss": 2.6838, + "step": 634 + }, + { + "epoch": 0.13, + "learning_rate": 9.975345412161395e-06, + "loss": 2.7145, + "step": 635 + }, + { + "epoch": 0.13, + "learning_rate": 9.974979112283185e-06, + "loss": 2.5656, + "step": 636 + }, + { + "epoch": 0.13, + "learning_rate": 9.974610118150098e-06, + "loss": 2.6535, + "step": 637 + }, + { + "epoch": 0.13, + "learning_rate": 9.974238429961965e-06, + "loss": 2.631, + "step": 638 + }, + { + "epoch": 0.13, + "learning_rate": 9.973864047920079e-06, + "loss": 2.6506, + "step": 639 + }, + { + "epoch": 0.13, + "learning_rate": 9.97348697222719e-06, + "loss": 2.6557, + "step": 640 + }, + { + "epoch": 0.13, + "learning_rate": 9.973107203087508e-06, + "loss": 2.6416, + "step": 641 + }, + { + "epoch": 0.13, + "learning_rate": 9.972724740706701e-06, + "loss": 2.6633, + "step": 642 + }, + { + "epoch": 0.13, + "learning_rate": 9.972339585291897e-06, + "loss": 2.6538, + "step": 643 + }, + { + "epoch": 0.14, + "learning_rate": 9.971951737051677e-06, + "loss": 2.6492, + "step": 644 + }, + { + "epoch": 0.14, + "learning_rate": 9.97156119619609e-06, + "loss": 2.6859, + "step": 645 + }, + { + "epoch": 0.14, + "learning_rate": 9.971167962936633e-06, + "loss": 2.7502, + "step": 646 + }, + { + "epoch": 0.14, + "learning_rate": 9.97077203748627e-06, + "loss": 2.6367, + "step": 647 + }, + { + "epoch": 0.14, + "learning_rate": 9.970373420059413e-06, + "loss": 2.6612, + "step": 648 + }, + { + "epoch": 0.14, + "learning_rate": 9.969972110871943e-06, + "loss": 2.6692, + "step": 649 + }, + { + "epoch": 0.14, + "learning_rate": 9.969568110141194e-06, + "loss": 2.6318, + "step": 650 + }, + { + "epoch": 0.14, + "learning_rate": 9.969161418085953e-06, + "loss": 2.6273, + "step": 651 + }, + { + "epoch": 0.14, + "learning_rate": 9.96875203492647e-06, + "loss": 2.6371, + "step": 652 + }, + { + "epoch": 0.14, + "learning_rate": 9.968339960884452e-06, + "loss": 2.7055, + "step": 653 + }, + { + "epoch": 0.14, + "learning_rate": 9.967925196183061e-06, + "loss": 2.6082, + "step": 654 + }, + { + "epoch": 0.14, + "learning_rate": 9.967507741046918e-06, + "loss": 2.6439, + "step": 655 + }, + { + "epoch": 0.14, + "learning_rate": 9.967087595702101e-06, + "loss": 2.7128, + "step": 656 + }, + { + "epoch": 0.14, + "learning_rate": 9.966664760376144e-06, + "loss": 2.6161, + "step": 657 + }, + { + "epoch": 0.14, + "learning_rate": 9.966239235298036e-06, + "loss": 2.6884, + "step": 658 + }, + { + "epoch": 0.14, + "learning_rate": 9.965811020698228e-06, + "loss": 2.6828, + "step": 659 + }, + { + "epoch": 0.14, + "learning_rate": 9.965380116808623e-06, + "loss": 2.6599, + "step": 660 + }, + { + "epoch": 0.14, + "learning_rate": 9.964946523862582e-06, + "loss": 2.6287, + "step": 661 + }, + { + "epoch": 0.14, + "learning_rate": 9.964510242094922e-06, + "loss": 2.7071, + "step": 662 + }, + { + "epoch": 0.14, + "learning_rate": 9.964071271741918e-06, + "loss": 2.6772, + "step": 663 + }, + { + "epoch": 0.14, + "learning_rate": 9.963629613041295e-06, + "loss": 2.7028, + "step": 664 + }, + { + "epoch": 0.14, + "learning_rate": 9.96318526623224e-06, + "loss": 2.6452, + "step": 665 + }, + { + "epoch": 0.14, + "learning_rate": 9.962738231555395e-06, + "loss": 2.6471, + "step": 666 + }, + { + "epoch": 0.14, + "learning_rate": 9.962288509252855e-06, + "loss": 2.6696, + "step": 667 + }, + { + "epoch": 0.14, + "learning_rate": 9.961836099568174e-06, + "loss": 2.624, + "step": 668 + }, + { + "epoch": 0.14, + "learning_rate": 9.96138100274636e-06, + "loss": 2.6633, + "step": 669 + }, + { + "epoch": 0.14, + "learning_rate": 9.960923219033873e-06, + "loss": 2.7073, + "step": 670 + }, + { + "epoch": 0.14, + "learning_rate": 9.960462748678632e-06, + "loss": 2.6751, + "step": 671 + }, + { + "epoch": 0.14, + "learning_rate": 9.95999959193001e-06, + "loss": 2.6528, + "step": 672 + }, + { + "epoch": 0.14, + "learning_rate": 9.959533749038834e-06, + "loss": 2.7003, + "step": 673 + }, + { + "epoch": 0.14, + "learning_rate": 9.959065220257388e-06, + "loss": 2.6681, + "step": 674 + }, + { + "epoch": 0.14, + "learning_rate": 9.958594005839405e-06, + "loss": 2.5618, + "step": 675 + }, + { + "epoch": 0.14, + "learning_rate": 9.958120106040082e-06, + "loss": 2.6813, + "step": 676 + }, + { + "epoch": 0.14, + "learning_rate": 9.957643521116059e-06, + "loss": 2.6567, + "step": 677 + }, + { + "epoch": 0.14, + "learning_rate": 9.957164251325441e-06, + "loss": 2.5926, + "step": 678 + }, + { + "epoch": 0.14, + "learning_rate": 9.956682296927779e-06, + "loss": 2.6359, + "step": 679 + }, + { + "epoch": 0.14, + "learning_rate": 9.956197658184082e-06, + "loss": 2.6647, + "step": 680 + }, + { + "epoch": 0.14, + "learning_rate": 9.95571033535681e-06, + "loss": 2.6848, + "step": 681 + }, + { + "epoch": 0.14, + "learning_rate": 9.955220328709879e-06, + "loss": 2.6174, + "step": 682 + }, + { + "epoch": 0.14, + "learning_rate": 9.954727638508655e-06, + "loss": 2.6435, + "step": 683 + }, + { + "epoch": 0.14, + "learning_rate": 9.954232265019965e-06, + "loss": 2.6143, + "step": 684 + }, + { + "epoch": 0.14, + "learning_rate": 9.95373420851208e-06, + "loss": 2.671, + "step": 685 + }, + { + "epoch": 0.14, + "learning_rate": 9.953233469254728e-06, + "loss": 2.6134, + "step": 686 + }, + { + "epoch": 0.14, + "learning_rate": 9.952730047519093e-06, + "loss": 2.6983, + "step": 687 + }, + { + "epoch": 0.14, + "learning_rate": 9.952223943577806e-06, + "loss": 2.6356, + "step": 688 + }, + { + "epoch": 0.14, + "learning_rate": 9.951715157704954e-06, + "loss": 2.6943, + "step": 689 + }, + { + "epoch": 0.14, + "learning_rate": 9.951203690176073e-06, + "loss": 2.6993, + "step": 690 + }, + { + "epoch": 0.14, + "learning_rate": 9.950689541268157e-06, + "loss": 2.6705, + "step": 691 + }, + { + "epoch": 0.15, + "learning_rate": 9.950172711259651e-06, + "loss": 2.6084, + "step": 692 + }, + { + "epoch": 0.15, + "learning_rate": 9.949653200430445e-06, + "loss": 2.6108, + "step": 693 + }, + { + "epoch": 0.15, + "learning_rate": 9.949131009061888e-06, + "loss": 2.6643, + "step": 694 + }, + { + "epoch": 0.15, + "learning_rate": 9.948606137436779e-06, + "loss": 2.6333, + "step": 695 + }, + { + "epoch": 0.15, + "learning_rate": 9.948078585839368e-06, + "loss": 2.608, + "step": 696 + }, + { + "epoch": 0.15, + "learning_rate": 9.947548354555355e-06, + "loss": 2.5904, + "step": 697 + }, + { + "epoch": 0.15, + "learning_rate": 9.947015443871894e-06, + "loss": 2.6329, + "step": 698 + }, + { + "epoch": 0.15, + "learning_rate": 9.94647985407759e-06, + "loss": 2.6722, + "step": 699 + }, + { + "epoch": 0.15, + "learning_rate": 9.945941585462493e-06, + "loss": 2.6716, + "step": 700 + }, + { + "epoch": 0.15, + "learning_rate": 9.945400638318113e-06, + "loss": 2.6327, + "step": 701 + }, + { + "epoch": 0.15, + "learning_rate": 9.944857012937403e-06, + "loss": 2.599, + "step": 702 + }, + { + "epoch": 0.15, + "learning_rate": 9.94431070961477e-06, + "loss": 2.5956, + "step": 703 + }, + { + "epoch": 0.15, + "learning_rate": 9.943761728646072e-06, + "loss": 2.6851, + "step": 704 + }, + { + "epoch": 0.15, + "learning_rate": 9.943210070328614e-06, + "loss": 2.6805, + "step": 705 + }, + { + "epoch": 0.15, + "learning_rate": 9.942655734961155e-06, + "loss": 2.604, + "step": 706 + }, + { + "epoch": 0.15, + "learning_rate": 9.942098722843898e-06, + "loss": 2.6351, + "step": 707 + }, + { + "epoch": 0.15, + "learning_rate": 9.9415390342785e-06, + "loss": 2.6892, + "step": 708 + }, + { + "epoch": 0.15, + "learning_rate": 9.94097666956807e-06, + "loss": 2.6315, + "step": 709 + }, + { + "epoch": 0.15, + "learning_rate": 9.940411629017159e-06, + "loss": 2.6615, + "step": 710 + }, + { + "epoch": 0.15, + "learning_rate": 9.93984391293177e-06, + "loss": 2.7145, + "step": 711 + }, + { + "epoch": 0.15, + "learning_rate": 9.939273521619361e-06, + "loss": 2.6576, + "step": 712 + }, + { + "epoch": 0.15, + "learning_rate": 9.93870045538883e-06, + "loss": 2.5922, + "step": 713 + }, + { + "epoch": 0.15, + "learning_rate": 9.938124714550528e-06, + "loss": 2.6672, + "step": 714 + }, + { + "epoch": 0.15, + "learning_rate": 9.937546299416254e-06, + "loss": 2.6957, + "step": 715 + }, + { + "epoch": 0.15, + "learning_rate": 9.936965210299254e-06, + "loss": 2.6691, + "step": 716 + }, + { + "epoch": 0.15, + "learning_rate": 9.936381447514225e-06, + "loss": 2.6017, + "step": 717 + }, + { + "epoch": 0.15, + "learning_rate": 9.93579501137731e-06, + "loss": 2.6705, + "step": 718 + }, + { + "epoch": 0.15, + "learning_rate": 9.935205902206098e-06, + "loss": 2.7009, + "step": 719 + }, + { + "epoch": 0.15, + "learning_rate": 9.93461412031963e-06, + "loss": 2.7041, + "step": 720 + }, + { + "epoch": 0.15, + "learning_rate": 9.93401966603839e-06, + "loss": 2.6109, + "step": 721 + }, + { + "epoch": 0.15, + "learning_rate": 9.933422539684314e-06, + "loss": 2.6471, + "step": 722 + }, + { + "epoch": 0.15, + "learning_rate": 9.932822741580779e-06, + "loss": 2.6379, + "step": 723 + }, + { + "epoch": 0.15, + "learning_rate": 9.932220272052615e-06, + "loss": 2.6675, + "step": 724 + }, + { + "epoch": 0.15, + "learning_rate": 9.931615131426094e-06, + "loss": 2.649, + "step": 725 + }, + { + "epoch": 0.15, + "learning_rate": 9.931007320028939e-06, + "loss": 2.664, + "step": 726 + }, + { + "epoch": 0.15, + "learning_rate": 9.930396838190313e-06, + "loss": 2.6689, + "step": 727 + }, + { + "epoch": 0.15, + "learning_rate": 9.929783686240833e-06, + "loss": 2.5982, + "step": 728 + }, + { + "epoch": 0.15, + "learning_rate": 9.929167864512557e-06, + "loss": 2.6668, + "step": 729 + }, + { + "epoch": 0.15, + "learning_rate": 9.928549373338991e-06, + "loss": 2.5918, + "step": 730 + }, + { + "epoch": 0.15, + "learning_rate": 9.927928213055082e-06, + "loss": 2.6889, + "step": 731 + }, + { + "epoch": 0.15, + "learning_rate": 9.927304383997232e-06, + "loss": 2.6367, + "step": 732 + }, + { + "epoch": 0.15, + "learning_rate": 9.926677886503278e-06, + "loss": 2.6211, + "step": 733 + }, + { + "epoch": 0.15, + "learning_rate": 9.926048720912509e-06, + "loss": 2.5755, + "step": 734 + }, + { + "epoch": 0.15, + "learning_rate": 9.925416887565653e-06, + "loss": 2.7025, + "step": 735 + }, + { + "epoch": 0.15, + "learning_rate": 9.924782386804891e-06, + "loss": 2.5744, + "step": 736 + }, + { + "epoch": 0.15, + "learning_rate": 9.924145218973841e-06, + "loss": 2.6629, + "step": 737 + }, + { + "epoch": 0.15, + "learning_rate": 9.92350538441757e-06, + "loss": 2.6481, + "step": 738 + }, + { + "epoch": 0.15, + "learning_rate": 9.922862883482586e-06, + "loss": 2.6443, + "step": 739 + }, + { + "epoch": 0.16, + "learning_rate": 9.922217716516843e-06, + "loss": 2.6371, + "step": 740 + }, + { + "epoch": 0.16, + "learning_rate": 9.92156988386974e-06, + "loss": 2.5897, + "step": 741 + }, + { + "epoch": 0.16, + "learning_rate": 9.920919385892114e-06, + "loss": 2.5825, + "step": 742 + }, + { + "epoch": 0.16, + "learning_rate": 9.920266222936252e-06, + "loss": 2.6657, + "step": 743 + }, + { + "epoch": 0.16, + "learning_rate": 9.919610395355882e-06, + "loss": 2.6236, + "step": 744 + }, + { + "epoch": 0.16, + "learning_rate": 9.918951903506172e-06, + "loss": 2.6644, + "step": 745 + }, + { + "epoch": 0.16, + "learning_rate": 9.91829074774374e-06, + "loss": 2.6701, + "step": 746 + }, + { + "epoch": 0.16, + "learning_rate": 9.91762692842664e-06, + "loss": 2.6067, + "step": 747 + }, + { + "epoch": 0.16, + "learning_rate": 9.916960445914368e-06, + "loss": 2.655, + "step": 748 + }, + { + "epoch": 0.16, + "learning_rate": 9.916291300567868e-06, + "loss": 2.6269, + "step": 749 + }, + { + "epoch": 0.16, + "learning_rate": 9.915619492749524e-06, + "loss": 2.6673, + "step": 750 + }, + { + "epoch": 0.16, + "learning_rate": 9.914945022823157e-06, + "loss": 2.627, + "step": 751 + }, + { + "epoch": 0.16, + "learning_rate": 9.914267891154037e-06, + "loss": 2.675, + "step": 752 + }, + { + "epoch": 0.16, + "learning_rate": 9.913588098108872e-06, + "loss": 2.6673, + "step": 753 + }, + { + "epoch": 0.16, + "learning_rate": 9.91290564405581e-06, + "loss": 2.6193, + "step": 754 + }, + { + "epoch": 0.16, + "learning_rate": 9.912220529364441e-06, + "loss": 2.596, + "step": 755 + }, + { + "epoch": 0.16, + "learning_rate": 9.911532754405797e-06, + "loss": 2.6261, + "step": 756 + }, + { + "epoch": 0.16, + "learning_rate": 9.910842319552351e-06, + "loss": 2.6207, + "step": 757 + }, + { + "epoch": 0.16, + "learning_rate": 9.910149225178018e-06, + "loss": 2.6266, + "step": 758 + }, + { + "epoch": 0.16, + "learning_rate": 9.909453471658145e-06, + "loss": 2.6406, + "step": 759 + }, + { + "epoch": 0.16, + "learning_rate": 9.90875505936953e-06, + "loss": 2.6512, + "step": 760 + }, + { + "epoch": 0.16, + "learning_rate": 9.908053988690403e-06, + "loss": 2.6429, + "step": 761 + }, + { + "epoch": 0.16, + "learning_rate": 9.907350260000437e-06, + "loss": 2.671, + "step": 762 + }, + { + "epoch": 0.16, + "learning_rate": 9.906643873680746e-06, + "loss": 2.6646, + "step": 763 + }, + { + "epoch": 0.16, + "learning_rate": 9.905934830113878e-06, + "loss": 2.6559, + "step": 764 + }, + { + "epoch": 0.16, + "learning_rate": 9.905223129683828e-06, + "loss": 2.6183, + "step": 765 + }, + { + "epoch": 0.16, + "learning_rate": 9.90450877277602e-06, + "loss": 2.6842, + "step": 766 + }, + { + "epoch": 0.16, + "learning_rate": 9.903791759777326e-06, + "loss": 2.6405, + "step": 767 + }, + { + "epoch": 0.16, + "learning_rate": 9.903072091076049e-06, + "loss": 2.7203, + "step": 768 + }, + { + "epoch": 0.16, + "learning_rate": 9.902349767061937e-06, + "loss": 2.6545, + "step": 769 + }, + { + "epoch": 0.16, + "learning_rate": 9.901624788126169e-06, + "loss": 2.6514, + "step": 770 + }, + { + "epoch": 0.16, + "learning_rate": 9.90089715466137e-06, + "loss": 2.6403, + "step": 771 + }, + { + "epoch": 0.16, + "learning_rate": 9.900166867061593e-06, + "loss": 2.6125, + "step": 772 + }, + { + "epoch": 0.16, + "learning_rate": 9.899433925722334e-06, + "loss": 2.6529, + "step": 773 + }, + { + "epoch": 0.16, + "learning_rate": 9.89869833104053e-06, + "loss": 2.5998, + "step": 774 + }, + { + "epoch": 0.16, + "learning_rate": 9.897960083414546e-06, + "loss": 2.5513, + "step": 775 + }, + { + "epoch": 0.16, + "learning_rate": 9.897219183244188e-06, + "loss": 2.6331, + "step": 776 + }, + { + "epoch": 0.16, + "learning_rate": 9.896475630930703e-06, + "loss": 2.5989, + "step": 777 + }, + { + "epoch": 0.16, + "learning_rate": 9.895729426876765e-06, + "loss": 2.6236, + "step": 778 + }, + { + "epoch": 0.16, + "learning_rate": 9.894980571486492e-06, + "loss": 2.6751, + "step": 779 + }, + { + "epoch": 0.16, + "learning_rate": 9.894229065165433e-06, + "loss": 2.6023, + "step": 780 + }, + { + "epoch": 0.16, + "learning_rate": 9.893474908320577e-06, + "loss": 2.6789, + "step": 781 + }, + { + "epoch": 0.16, + "learning_rate": 9.892718101360344e-06, + "loss": 2.5993, + "step": 782 + }, + { + "epoch": 0.16, + "learning_rate": 9.89195864469459e-06, + "loss": 2.6585, + "step": 783 + }, + { + "epoch": 0.16, + "learning_rate": 9.891196538734612e-06, + "loss": 2.613, + "step": 784 + }, + { + "epoch": 0.16, + "learning_rate": 9.89043178389313e-06, + "loss": 2.6641, + "step": 785 + }, + { + "epoch": 0.16, + "learning_rate": 9.889664380584312e-06, + "loss": 2.6673, + "step": 786 + }, + { + "epoch": 0.17, + "learning_rate": 9.888894329223751e-06, + "loss": 2.6562, + "step": 787 + }, + { + "epoch": 0.17, + "learning_rate": 9.888121630228476e-06, + "loss": 2.6245, + "step": 788 + }, + { + "epoch": 0.17, + "learning_rate": 9.88734628401695e-06, + "loss": 2.6599, + "step": 789 + }, + { + "epoch": 0.17, + "learning_rate": 9.886568291009074e-06, + "loss": 2.601, + "step": 790 + }, + { + "epoch": 0.17, + "learning_rate": 9.885787651626176e-06, + "loss": 2.6579, + "step": 791 + }, + { + "epoch": 0.17, + "learning_rate": 9.88500436629102e-06, + "loss": 2.6087, + "step": 792 + }, + { + "epoch": 0.17, + "learning_rate": 9.884218435427802e-06, + "loss": 2.647, + "step": 793 + }, + { + "epoch": 0.17, + "learning_rate": 9.883429859462155e-06, + "loss": 2.5845, + "step": 794 + }, + { + "epoch": 0.17, + "learning_rate": 9.882638638821136e-06, + "loss": 2.6389, + "step": 795 + }, + { + "epoch": 0.17, + "learning_rate": 9.881844773933243e-06, + "loss": 2.6165, + "step": 796 + }, + { + "epoch": 0.17, + "learning_rate": 9.881048265228402e-06, + "loss": 2.6051, + "step": 797 + }, + { + "epoch": 0.17, + "learning_rate": 9.88024911313797e-06, + "loss": 2.6493, + "step": 798 + }, + { + "epoch": 0.17, + "learning_rate": 9.879447318094737e-06, + "loss": 2.6211, + "step": 799 + }, + { + "epoch": 0.17, + "learning_rate": 9.878642880532923e-06, + "loss": 2.668, + "step": 800 + }, + { + "epoch": 0.17, + "learning_rate": 9.877835800888183e-06, + "loss": 2.549, + "step": 801 + }, + { + "epoch": 0.17, + "learning_rate": 9.877026079597595e-06, + "loss": 2.6899, + "step": 802 + }, + { + "epoch": 0.17, + "learning_rate": 9.876213717099678e-06, + "loss": 2.6659, + "step": 803 + }, + { + "epoch": 0.17, + "learning_rate": 9.875398713834372e-06, + "loss": 2.7021, + "step": 804 + }, + { + "epoch": 0.17, + "learning_rate": 9.874581070243053e-06, + "loss": 2.6572, + "step": 805 + }, + { + "epoch": 0.17, + "learning_rate": 9.873760786768524e-06, + "loss": 2.6526, + "step": 806 + }, + { + "epoch": 0.17, + "learning_rate": 9.87293786385502e-06, + "loss": 2.6161, + "step": 807 + }, + { + "epoch": 0.17, + "learning_rate": 9.8721123019482e-06, + "loss": 2.6693, + "step": 808 + }, + { + "epoch": 0.17, + "learning_rate": 9.87128410149516e-06, + "loss": 2.634, + "step": 809 + }, + { + "epoch": 0.17, + "learning_rate": 9.87045326294442e-06, + "loss": 2.6079, + "step": 810 + }, + { + "epoch": 0.17, + "learning_rate": 9.869619786745931e-06, + "loss": 2.7352, + "step": 811 + }, + { + "epoch": 0.17, + "learning_rate": 9.868783673351069e-06, + "loss": 2.6228, + "step": 812 + }, + { + "epoch": 0.17, + "learning_rate": 9.867944923212642e-06, + "loss": 2.6478, + "step": 813 + }, + { + "epoch": 0.17, + "learning_rate": 9.867103536784883e-06, + "loss": 2.6299, + "step": 814 + }, + { + "epoch": 0.17, + "learning_rate": 9.866259514523456e-06, + "loss": 2.6474, + "step": 815 + }, + { + "epoch": 0.17, + "learning_rate": 9.86541285688545e-06, + "loss": 2.6767, + "step": 816 + }, + { + "epoch": 0.17, + "learning_rate": 9.86456356432938e-06, + "loss": 2.6866, + "step": 817 + }, + { + "epoch": 0.17, + "learning_rate": 9.863711637315193e-06, + "loss": 2.6242, + "step": 818 + }, + { + "epoch": 0.17, + "learning_rate": 9.862857076304256e-06, + "loss": 2.7202, + "step": 819 + }, + { + "epoch": 0.17, + "learning_rate": 9.861999881759367e-06, + "loss": 2.6433, + "step": 820 + }, + { + "epoch": 0.17, + "learning_rate": 9.86114005414475e-06, + "loss": 2.6141, + "step": 821 + }, + { + "epoch": 0.17, + "learning_rate": 9.860277593926057e-06, + "loss": 2.6949, + "step": 822 + }, + { + "epoch": 0.17, + "learning_rate": 9.859412501570357e-06, + "loss": 2.5736, + "step": 823 + }, + { + "epoch": 0.17, + "learning_rate": 9.858544777546153e-06, + "loss": 2.6286, + "step": 824 + }, + { + "epoch": 0.17, + "learning_rate": 9.857674422323371e-06, + "loss": 2.6303, + "step": 825 + }, + { + "epoch": 0.17, + "learning_rate": 9.856801436373363e-06, + "loss": 2.6915, + "step": 826 + }, + { + "epoch": 0.17, + "learning_rate": 9.8559258201689e-06, + "loss": 2.6343, + "step": 827 + }, + { + "epoch": 0.17, + "learning_rate": 9.855047574184185e-06, + "loss": 2.6532, + "step": 828 + }, + { + "epoch": 0.17, + "learning_rate": 9.85416669889484e-06, + "loss": 2.6296, + "step": 829 + }, + { + "epoch": 0.17, + "learning_rate": 9.853283194777913e-06, + "loss": 2.669, + "step": 830 + }, + { + "epoch": 0.17, + "learning_rate": 9.852397062311876e-06, + "loss": 2.6872, + "step": 831 + }, + { + "epoch": 0.17, + "learning_rate": 9.851508301976625e-06, + "loss": 2.5946, + "step": 832 + }, + { + "epoch": 0.17, + "learning_rate": 9.850616914253476e-06, + "loss": 2.7062, + "step": 833 + }, + { + "epoch": 0.17, + "learning_rate": 9.84972289962517e-06, + "loss": 2.6386, + "step": 834 + }, + { + "epoch": 0.18, + "learning_rate": 9.848826258575871e-06, + "loss": 2.596, + "step": 835 + }, + { + "epoch": 0.18, + "learning_rate": 9.847926991591165e-06, + "loss": 2.6662, + "step": 836 + }, + { + "epoch": 0.18, + "learning_rate": 9.84702509915806e-06, + "loss": 2.634, + "step": 837 + }, + { + "epoch": 0.18, + "learning_rate": 9.846120581764986e-06, + "loss": 2.6369, + "step": 838 + }, + { + "epoch": 0.18, + "learning_rate": 9.845213439901795e-06, + "loss": 2.7727, + "step": 839 + }, + { + "epoch": 0.18, + "learning_rate": 9.844303674059756e-06, + "loss": 2.6957, + "step": 840 + }, + { + "epoch": 0.18, + "learning_rate": 9.843391284731569e-06, + "loss": 2.6147, + "step": 841 + }, + { + "epoch": 0.18, + "learning_rate": 9.842476272411343e-06, + "loss": 2.6914, + "step": 842 + }, + { + "epoch": 0.18, + "learning_rate": 9.841558637594614e-06, + "loss": 2.6317, + "step": 843 + }, + { + "epoch": 0.18, + "learning_rate": 9.84063838077834e-06, + "loss": 2.6653, + "step": 844 + }, + { + "epoch": 0.18, + "learning_rate": 9.839715502460894e-06, + "loss": 2.6167, + "step": 845 + }, + { + "epoch": 0.18, + "learning_rate": 9.838790003142071e-06, + "loss": 2.6449, + "step": 846 + }, + { + "epoch": 0.18, + "learning_rate": 9.837861883323086e-06, + "loss": 2.6829, + "step": 847 + }, + { + "epoch": 0.18, + "learning_rate": 9.836931143506572e-06, + "loss": 2.6753, + "step": 848 + }, + { + "epoch": 0.18, + "learning_rate": 9.835997784196583e-06, + "loss": 2.6788, + "step": 849 + }, + { + "epoch": 0.18, + "learning_rate": 9.835061805898588e-06, + "loss": 2.6232, + "step": 850 + }, + { + "epoch": 0.18, + "learning_rate": 9.834123209119478e-06, + "loss": 2.6799, + "step": 851 + }, + { + "epoch": 0.18, + "learning_rate": 9.833181994367558e-06, + "loss": 2.6365, + "step": 852 + }, + { + "epoch": 0.18, + "learning_rate": 9.832238162152556e-06, + "loss": 2.656, + "step": 853 + }, + { + "epoch": 0.18, + "learning_rate": 9.831291712985613e-06, + "loss": 2.64, + "step": 854 + }, + { + "epoch": 0.18, + "learning_rate": 9.830342647379292e-06, + "loss": 2.7695, + "step": 855 + }, + { + "epoch": 0.18, + "learning_rate": 9.829390965847565e-06, + "loss": 2.5788, + "step": 856 + }, + { + "epoch": 0.18, + "learning_rate": 9.828436668905829e-06, + "loss": 2.6604, + "step": 857 + }, + { + "epoch": 0.18, + "learning_rate": 9.827479757070895e-06, + "loss": 2.6853, + "step": 858 + }, + { + "epoch": 0.18, + "learning_rate": 9.826520230860986e-06, + "loss": 2.625, + "step": 859 + }, + { + "epoch": 0.18, + "learning_rate": 9.825558090795747e-06, + "loss": 2.6634, + "step": 860 + }, + { + "epoch": 0.18, + "learning_rate": 9.824593337396232e-06, + "loss": 2.6197, + "step": 861 + }, + { + "epoch": 0.18, + "learning_rate": 9.823625971184916e-06, + "loss": 2.6796, + "step": 862 + }, + { + "epoch": 0.18, + "learning_rate": 9.822655992685687e-06, + "loss": 2.595, + "step": 863 + }, + { + "epoch": 0.18, + "learning_rate": 9.821683402423849e-06, + "loss": 2.6652, + "step": 864 + }, + { + "epoch": 0.18, + "learning_rate": 9.820708200926116e-06, + "loss": 2.5941, + "step": 865 + }, + { + "epoch": 0.18, + "learning_rate": 9.819730388720622e-06, + "loss": 2.7133, + "step": 866 + }, + { + "epoch": 0.18, + "learning_rate": 9.818749966336908e-06, + "loss": 2.6614, + "step": 867 + }, + { + "epoch": 0.18, + "learning_rate": 9.817766934305936e-06, + "loss": 2.6487, + "step": 868 + }, + { + "epoch": 0.18, + "learning_rate": 9.816781293160079e-06, + "loss": 2.6859, + "step": 869 + }, + { + "epoch": 0.18, + "learning_rate": 9.815793043433118e-06, + "loss": 2.6591, + "step": 870 + }, + { + "epoch": 0.18, + "learning_rate": 9.814802185660252e-06, + "loss": 2.6589, + "step": 871 + }, + { + "epoch": 0.18, + "learning_rate": 9.81380872037809e-06, + "loss": 2.606, + "step": 872 + }, + { + "epoch": 0.18, + "learning_rate": 9.812812648124659e-06, + "loss": 2.6339, + "step": 873 + }, + { + "epoch": 0.18, + "learning_rate": 9.811813969439388e-06, + "loss": 2.6454, + "step": 874 + }, + { + "epoch": 0.18, + "learning_rate": 9.810812684863123e-06, + "loss": 2.5752, + "step": 875 + }, + { + "epoch": 0.18, + "learning_rate": 9.809808794938121e-06, + "loss": 2.6135, + "step": 876 + }, + { + "epoch": 0.18, + "learning_rate": 9.808802300208052e-06, + "loss": 2.6524, + "step": 877 + }, + { + "epoch": 0.18, + "learning_rate": 9.80779320121799e-06, + "loss": 2.608, + "step": 878 + }, + { + "epoch": 0.18, + "learning_rate": 9.806781498514428e-06, + "loss": 2.6488, + "step": 879 + }, + { + "epoch": 0.18, + "learning_rate": 9.805767192645264e-06, + "loss": 2.7013, + "step": 880 + }, + { + "epoch": 0.18, + "learning_rate": 9.804750284159802e-06, + "loss": 2.6086, + "step": 881 + }, + { + "epoch": 0.18, + "learning_rate": 9.803730773608768e-06, + "loss": 2.6259, + "step": 882 + }, + { + "epoch": 0.19, + "learning_rate": 9.802708661544282e-06, + "loss": 2.5922, + "step": 883 + }, + { + "epoch": 0.19, + "learning_rate": 9.801683948519885e-06, + "loss": 2.6594, + "step": 884 + }, + { + "epoch": 0.19, + "learning_rate": 9.800656635090517e-06, + "loss": 2.6267, + "step": 885 + }, + { + "epoch": 0.19, + "learning_rate": 9.799626721812534e-06, + "loss": 2.6694, + "step": 886 + }, + { + "epoch": 0.19, + "learning_rate": 9.798594209243697e-06, + "loss": 2.6439, + "step": 887 + }, + { + "epoch": 0.19, + "learning_rate": 9.797559097943174e-06, + "loss": 2.6325, + "step": 888 + }, + { + "epoch": 0.19, + "learning_rate": 9.79652138847154e-06, + "loss": 2.6856, + "step": 889 + }, + { + "epoch": 0.19, + "learning_rate": 9.79548108139078e-06, + "loss": 2.6554, + "step": 890 + }, + { + "epoch": 0.19, + "learning_rate": 9.794438177264282e-06, + "loss": 2.6177, + "step": 891 + }, + { + "epoch": 0.19, + "learning_rate": 9.793392676656844e-06, + "loss": 2.6233, + "step": 892 + }, + { + "epoch": 0.19, + "learning_rate": 9.792344580134664e-06, + "loss": 2.6421, + "step": 893 + }, + { + "epoch": 0.19, + "learning_rate": 9.791293888265357e-06, + "loss": 2.6137, + "step": 894 + }, + { + "epoch": 0.19, + "learning_rate": 9.790240601617933e-06, + "loss": 2.6151, + "step": 895 + }, + { + "epoch": 0.19, + "learning_rate": 9.78918472076281e-06, + "loss": 2.7187, + "step": 896 + }, + { + "epoch": 0.19, + "learning_rate": 9.788126246271814e-06, + "loss": 2.5671, + "step": 897 + }, + { + "epoch": 0.19, + "learning_rate": 9.787065178718174e-06, + "loss": 2.6443, + "step": 898 + }, + { + "epoch": 0.19, + "learning_rate": 9.78600151867652e-06, + "loss": 2.5923, + "step": 899 + }, + { + "epoch": 0.19, + "learning_rate": 9.784935266722892e-06, + "loss": 2.6632, + "step": 900 + }, + { + "epoch": 0.19, + "learning_rate": 9.783866423434728e-06, + "loss": 2.6239, + "step": 901 + }, + { + "epoch": 0.19, + "learning_rate": 9.782794989390874e-06, + "loss": 2.6924, + "step": 902 + }, + { + "epoch": 0.19, + "learning_rate": 9.781720965171575e-06, + "loss": 2.6291, + "step": 903 + }, + { + "epoch": 0.19, + "learning_rate": 9.780644351358482e-06, + "loss": 2.631, + "step": 904 + }, + { + "epoch": 0.19, + "learning_rate": 9.779565148534645e-06, + "loss": 2.607, + "step": 905 + }, + { + "epoch": 0.19, + "learning_rate": 9.778483357284523e-06, + "loss": 2.6212, + "step": 906 + }, + { + "epoch": 0.19, + "learning_rate": 9.777398978193966e-06, + "loss": 2.6939, + "step": 907 + }, + { + "epoch": 0.19, + "learning_rate": 9.776312011850236e-06, + "loss": 2.6617, + "step": 908 + }, + { + "epoch": 0.19, + "learning_rate": 9.77522245884199e-06, + "loss": 2.6639, + "step": 909 + }, + { + "epoch": 0.19, + "learning_rate": 9.774130319759287e-06, + "loss": 2.6788, + "step": 910 + }, + { + "epoch": 0.19, + "learning_rate": 9.773035595193588e-06, + "loss": 2.6322, + "step": 911 + }, + { + "epoch": 0.19, + "learning_rate": 9.771938285737752e-06, + "loss": 2.5979, + "step": 912 + }, + { + "epoch": 0.19, + "learning_rate": 9.77083839198604e-06, + "loss": 2.6539, + "step": 913 + }, + { + "epoch": 0.19, + "learning_rate": 9.76973591453411e-06, + "loss": 2.6506, + "step": 914 + }, + { + "epoch": 0.19, + "learning_rate": 9.768630853979023e-06, + "loss": 2.5869, + "step": 915 + }, + { + "epoch": 0.19, + "learning_rate": 9.767523210919236e-06, + "loss": 2.6086, + "step": 916 + }, + { + "epoch": 0.19, + "learning_rate": 9.766412985954605e-06, + "loss": 2.6539, + "step": 917 + }, + { + "epoch": 0.19, + "learning_rate": 9.765300179686386e-06, + "loss": 2.5556, + "step": 918 + }, + { + "epoch": 0.19, + "learning_rate": 9.764184792717229e-06, + "loss": 2.6593, + "step": 919 + }, + { + "epoch": 0.19, + "learning_rate": 9.763066825651186e-06, + "loss": 2.6297, + "step": 920 + }, + { + "epoch": 0.19, + "learning_rate": 9.761946279093703e-06, + "loss": 2.645, + "step": 921 + }, + { + "epoch": 0.19, + "learning_rate": 9.760823153651627e-06, + "loss": 2.6641, + "step": 922 + }, + { + "epoch": 0.19, + "learning_rate": 9.759697449933194e-06, + "loss": 2.6646, + "step": 923 + }, + { + "epoch": 0.19, + "learning_rate": 9.758569168548047e-06, + "loss": 2.6289, + "step": 924 + }, + { + "epoch": 0.19, + "learning_rate": 9.757438310107217e-06, + "loss": 2.6394, + "step": 925 + }, + { + "epoch": 0.19, + "learning_rate": 9.75630487522313e-06, + "loss": 2.6205, + "step": 926 + }, + { + "epoch": 0.19, + "learning_rate": 9.755168864509616e-06, + "loss": 2.6301, + "step": 927 + }, + { + "epoch": 0.19, + "learning_rate": 9.75403027858189e-06, + "loss": 2.6561, + "step": 928 + }, + { + "epoch": 0.19, + "learning_rate": 9.752889118056565e-06, + "loss": 2.6445, + "step": 929 + }, + { + "epoch": 0.2, + "learning_rate": 9.751745383551652e-06, + "loss": 2.6458, + "step": 930 + }, + { + "epoch": 0.2, + "learning_rate": 9.750599075686553e-06, + "loss": 2.6603, + "step": 931 + }, + { + "epoch": 0.2, + "learning_rate": 9.749450195082059e-06, + "loss": 2.7342, + "step": 932 + }, + { + "epoch": 0.2, + "learning_rate": 9.748298742360365e-06, + "loss": 2.6077, + "step": 933 + }, + { + "epoch": 0.2, + "learning_rate": 9.747144718145052e-06, + "loss": 2.6677, + "step": 934 + }, + { + "epoch": 0.2, + "learning_rate": 9.74598812306109e-06, + "loss": 2.6343, + "step": 935 + }, + { + "epoch": 0.2, + "learning_rate": 9.744828957734851e-06, + "loss": 2.5954, + "step": 936 + }, + { + "epoch": 0.2, + "learning_rate": 9.743667222794091e-06, + "loss": 2.7133, + "step": 937 + }, + { + "epoch": 0.2, + "learning_rate": 9.742502918867959e-06, + "loss": 2.6347, + "step": 938 + }, + { + "epoch": 0.2, + "learning_rate": 9.741336046587001e-06, + "loss": 2.6492, + "step": 939 + }, + { + "epoch": 0.2, + "learning_rate": 9.740166606583147e-06, + "loss": 2.6241, + "step": 940 + }, + { + "epoch": 0.2, + "learning_rate": 9.73899459948972e-06, + "loss": 2.622, + "step": 941 + }, + { + "epoch": 0.2, + "learning_rate": 9.737820025941434e-06, + "loss": 2.6747, + "step": 942 + }, + { + "epoch": 0.2, + "learning_rate": 9.73664288657439e-06, + "loss": 2.7095, + "step": 943 + }, + { + "epoch": 0.2, + "learning_rate": 9.735463182026085e-06, + "loss": 2.6597, + "step": 944 + }, + { + "epoch": 0.2, + "learning_rate": 9.734280912935398e-06, + "loss": 2.7326, + "step": 945 + }, + { + "epoch": 0.2, + "learning_rate": 9.733096079942603e-06, + "loss": 2.6391, + "step": 946 + }, + { + "epoch": 0.2, + "learning_rate": 9.731908683689355e-06, + "loss": 2.6406, + "step": 947 + }, + { + "epoch": 0.2, + "learning_rate": 9.730718724818704e-06, + "loss": 2.6371, + "step": 948 + }, + { + "epoch": 0.2, + "learning_rate": 9.729526203975085e-06, + "loss": 2.7234, + "step": 949 + }, + { + "epoch": 0.2, + "learning_rate": 9.728331121804322e-06, + "loss": 2.6615, + "step": 950 + }, + { + "epoch": 0.2, + "learning_rate": 9.727133478953622e-06, + "loss": 2.6097, + "step": 951 + }, + { + "epoch": 0.2, + "learning_rate": 9.725933276071585e-06, + "loss": 2.6359, + "step": 952 + }, + { + "epoch": 0.2, + "learning_rate": 9.724730513808191e-06, + "loss": 2.551, + "step": 953 + }, + { + "epoch": 0.2, + "learning_rate": 9.723525192814814e-06, + "loss": 2.6603, + "step": 954 + }, + { + "epoch": 0.2, + "learning_rate": 9.722317313744204e-06, + "loss": 2.6211, + "step": 955 + }, + { + "epoch": 0.2, + "learning_rate": 9.721106877250501e-06, + "loss": 2.606, + "step": 956 + }, + { + "epoch": 0.2, + "learning_rate": 9.719893883989236e-06, + "loss": 2.5952, + "step": 957 + }, + { + "epoch": 0.2, + "learning_rate": 9.718678334617314e-06, + "loss": 2.6432, + "step": 958 + }, + { + "epoch": 0.2, + "learning_rate": 9.717460229793027e-06, + "loss": 2.8004, + "step": 959 + }, + { + "epoch": 0.2, + "learning_rate": 9.71623957017606e-06, + "loss": 2.7137, + "step": 960 + }, + { + "epoch": 0.2, + "learning_rate": 9.715016356427471e-06, + "loss": 2.6158, + "step": 961 + }, + { + "epoch": 0.2, + "learning_rate": 9.713790589209704e-06, + "loss": 2.6514, + "step": 962 + }, + { + "epoch": 0.2, + "learning_rate": 9.712562269186588e-06, + "loss": 2.6508, + "step": 963 + }, + { + "epoch": 0.2, + "learning_rate": 9.711331397023334e-06, + "loss": 2.6552, + "step": 964 + }, + { + "epoch": 0.2, + "learning_rate": 9.710097973386531e-06, + "loss": 2.6809, + "step": 965 + }, + { + "epoch": 0.2, + "learning_rate": 9.708861998944159e-06, + "loss": 2.6349, + "step": 966 + }, + { + "epoch": 0.2, + "learning_rate": 9.707623474365569e-06, + "loss": 2.6194, + "step": 967 + }, + { + "epoch": 0.2, + "learning_rate": 9.7063824003215e-06, + "loss": 2.6417, + "step": 968 + }, + { + "epoch": 0.2, + "learning_rate": 9.705138777484067e-06, + "loss": 2.6492, + "step": 969 + }, + { + "epoch": 0.2, + "learning_rate": 9.70389260652677e-06, + "loss": 2.6428, + "step": 970 + }, + { + "epoch": 0.2, + "learning_rate": 9.702643888124484e-06, + "loss": 2.6378, + "step": 971 + }, + { + "epoch": 0.2, + "learning_rate": 9.701392622953469e-06, + "loss": 2.6411, + "step": 972 + }, + { + "epoch": 0.2, + "learning_rate": 9.700138811691361e-06, + "loss": 2.6946, + "step": 973 + }, + { + "epoch": 0.2, + "learning_rate": 9.698882455017175e-06, + "loss": 2.6012, + "step": 974 + }, + { + "epoch": 0.2, + "learning_rate": 9.697623553611303e-06, + "loss": 2.6225, + "step": 975 + }, + { + "epoch": 0.2, + "learning_rate": 9.69636210815552e-06, + "loss": 2.6765, + "step": 976 + }, + { + "epoch": 0.2, + "learning_rate": 9.695098119332972e-06, + "loss": 2.6765, + "step": 977 + }, + { + "epoch": 0.21, + "learning_rate": 9.69383158782819e-06, + "loss": 2.6324, + "step": 978 + }, + { + "epoch": 0.21, + "learning_rate": 9.692562514327078e-06, + "loss": 2.678, + "step": 979 + }, + { + "epoch": 0.21, + "learning_rate": 9.691290899516912e-06, + "loss": 2.6934, + "step": 980 + }, + { + "epoch": 0.21, + "learning_rate": 9.690016744086354e-06, + "loss": 2.6775, + "step": 981 + }, + { + "epoch": 0.21, + "learning_rate": 9.688740048725435e-06, + "loss": 2.6525, + "step": 982 + }, + { + "epoch": 0.21, + "learning_rate": 9.687460814125564e-06, + "loss": 2.6786, + "step": 983 + }, + { + "epoch": 0.21, + "learning_rate": 9.686179040979525e-06, + "loss": 2.6595, + "step": 984 + }, + { + "epoch": 0.21, + "learning_rate": 9.684894729981474e-06, + "loss": 2.6532, + "step": 985 + }, + { + "epoch": 0.21, + "learning_rate": 9.683607881826946e-06, + "loss": 2.6279, + "step": 986 + }, + { + "epoch": 0.21, + "learning_rate": 9.682318497212848e-06, + "loss": 2.6278, + "step": 987 + }, + { + "epoch": 0.21, + "learning_rate": 9.68102657683746e-06, + "loss": 2.673, + "step": 988 + }, + { + "epoch": 0.21, + "learning_rate": 9.679732121400435e-06, + "loss": 2.6144, + "step": 989 + }, + { + "epoch": 0.21, + "learning_rate": 9.678435131602802e-06, + "loss": 2.6223, + "step": 990 + }, + { + "epoch": 0.21, + "learning_rate": 9.677135608146954e-06, + "loss": 2.6275, + "step": 991 + }, + { + "epoch": 0.21, + "learning_rate": 9.67583355173667e-06, + "loss": 2.5902, + "step": 992 + }, + { + "epoch": 0.21, + "learning_rate": 9.67452896307709e-06, + "loss": 2.6245, + "step": 993 + }, + { + "epoch": 0.21, + "learning_rate": 9.673221842874727e-06, + "loss": 2.6001, + "step": 994 + }, + { + "epoch": 0.21, + "learning_rate": 9.671912191837468e-06, + "loss": 2.6758, + "step": 995 + }, + { + "epoch": 0.21, + "learning_rate": 9.670600010674567e-06, + "loss": 2.6735, + "step": 996 + }, + { + "epoch": 0.21, + "learning_rate": 9.669285300096653e-06, + "loss": 2.6534, + "step": 997 + }, + { + "epoch": 0.21, + "learning_rate": 9.667968060815721e-06, + "loss": 2.6685, + "step": 998 + }, + { + "epoch": 0.21, + "learning_rate": 9.666648293545134e-06, + "loss": 2.6358, + "step": 999 + }, + { + "epoch": 0.21, + "learning_rate": 9.66532599899963e-06, + "loss": 2.6212, + "step": 1000 + }, + { + "epoch": 0.21, + "learning_rate": 9.664001177895312e-06, + "loss": 2.6594, + "step": 1001 + }, + { + "epoch": 0.21, + "learning_rate": 9.66267383094965e-06, + "loss": 2.6737, + "step": 1002 + }, + { + "epoch": 0.21, + "learning_rate": 9.661343958881482e-06, + "loss": 2.6341, + "step": 1003 + }, + { + "epoch": 0.21, + "learning_rate": 9.660011562411018e-06, + "loss": 2.6759, + "step": 1004 + }, + { + "epoch": 0.21, + "learning_rate": 9.65867664225983e-06, + "loss": 2.6217, + "step": 1005 + }, + { + "epoch": 0.21, + "learning_rate": 9.657339199150861e-06, + "loss": 2.6255, + "step": 1006 + }, + { + "epoch": 0.21, + "learning_rate": 9.655999233808415e-06, + "loss": 2.6824, + "step": 1007 + }, + { + "epoch": 0.21, + "learning_rate": 9.654656746958168e-06, + "loss": 2.6043, + "step": 1008 + }, + { + "epoch": 0.21, + "learning_rate": 9.653311739327156e-06, + "loss": 2.6057, + "step": 1009 + }, + { + "epoch": 0.21, + "learning_rate": 9.651964211643784e-06, + "loss": 2.6592, + "step": 1010 + }, + { + "epoch": 0.21, + "learning_rate": 9.650614164637821e-06, + "loss": 2.6517, + "step": 1011 + }, + { + "epoch": 0.21, + "learning_rate": 9.649261599040399e-06, + "loss": 2.6272, + "step": 1012 + }, + { + "epoch": 0.21, + "learning_rate": 9.647906515584014e-06, + "loss": 2.6593, + "step": 1013 + }, + { + "epoch": 0.21, + "learning_rate": 9.64654891500253e-06, + "loss": 2.61, + "step": 1014 + }, + { + "epoch": 0.21, + "learning_rate": 9.645188798031167e-06, + "loss": 2.6386, + "step": 1015 + }, + { + "epoch": 0.21, + "learning_rate": 9.643826165406512e-06, + "loss": 2.6071, + "step": 1016 + }, + { + "epoch": 0.21, + "learning_rate": 9.642461017866513e-06, + "loss": 2.6336, + "step": 1017 + }, + { + "epoch": 0.21, + "learning_rate": 9.641093356150484e-06, + "loss": 2.6464, + "step": 1018 + }, + { + "epoch": 0.21, + "learning_rate": 9.639723180999094e-06, + "loss": 2.6804, + "step": 1019 + }, + { + "epoch": 0.21, + "learning_rate": 9.638350493154378e-06, + "loss": 2.6506, + "step": 1020 + }, + { + "epoch": 0.21, + "learning_rate": 9.63697529335973e-06, + "loss": 2.6151, + "step": 1021 + }, + { + "epoch": 0.21, + "learning_rate": 9.635597582359905e-06, + "loss": 2.663, + "step": 1022 + }, + { + "epoch": 0.21, + "learning_rate": 9.634217360901016e-06, + "loss": 2.6835, + "step": 1023 + }, + { + "epoch": 0.21, + "learning_rate": 9.63283462973054e-06, + "loss": 2.6072, + "step": 1024 + }, + { + "epoch": 0.21, + "learning_rate": 9.631449389597307e-06, + "loss": 2.6249, + "step": 1025 + }, + { + "epoch": 0.22, + "learning_rate": 9.630061641251513e-06, + "loss": 2.6085, + "step": 1026 + }, + { + "epoch": 0.22, + "learning_rate": 9.628671385444706e-06, + "loss": 2.6427, + "step": 1027 + }, + { + "epoch": 0.22, + "learning_rate": 9.627278622929791e-06, + "loss": 2.5994, + "step": 1028 + }, + { + "epoch": 0.22, + "learning_rate": 9.625883354461043e-06, + "loss": 2.6203, + "step": 1029 + }, + { + "epoch": 0.22, + "learning_rate": 9.624485580794076e-06, + "loss": 2.6116, + "step": 1030 + }, + { + "epoch": 0.22, + "learning_rate": 9.623085302685875e-06, + "loss": 2.6199, + "step": 1031 + }, + { + "epoch": 0.22, + "learning_rate": 9.621682520894773e-06, + "loss": 2.6425, + "step": 1032 + }, + { + "epoch": 0.22, + "learning_rate": 9.620277236180464e-06, + "loss": 2.6563, + "step": 1033 + }, + { + "epoch": 0.22, + "learning_rate": 9.618869449303996e-06, + "loss": 2.6275, + "step": 1034 + }, + { + "epoch": 0.22, + "learning_rate": 9.617459161027769e-06, + "loss": 2.6774, + "step": 1035 + }, + { + "epoch": 0.22, + "learning_rate": 9.616046372115541e-06, + "loss": 2.6297, + "step": 1036 + }, + { + "epoch": 0.22, + "learning_rate": 9.614631083332427e-06, + "loss": 2.648, + "step": 1037 + }, + { + "epoch": 0.22, + "learning_rate": 9.613213295444887e-06, + "loss": 2.6651, + "step": 1038 + }, + { + "epoch": 0.22, + "learning_rate": 9.611793009220742e-06, + "loss": 2.6883, + "step": 1039 + }, + { + "epoch": 0.22, + "learning_rate": 9.610370225429164e-06, + "loss": 2.626, + "step": 1040 + }, + { + "epoch": 0.22, + "learning_rate": 9.608944944840678e-06, + "loss": 2.7438, + "step": 1041 + }, + { + "epoch": 0.22, + "learning_rate": 9.607517168227158e-06, + "loss": 2.605, + "step": 1042 + }, + { + "epoch": 0.22, + "learning_rate": 9.606086896361835e-06, + "loss": 2.6488, + "step": 1043 + }, + { + "epoch": 0.22, + "learning_rate": 9.604654130019287e-06, + "loss": 2.6558, + "step": 1044 + }, + { + "epoch": 0.22, + "learning_rate": 9.603218869975444e-06, + "loss": 2.6588, + "step": 1045 + }, + { + "epoch": 0.22, + "learning_rate": 9.601781117007586e-06, + "loss": 2.6556, + "step": 1046 + }, + { + "epoch": 0.22, + "learning_rate": 9.600340871894346e-06, + "loss": 2.6821, + "step": 1047 + }, + { + "epoch": 0.22, + "learning_rate": 9.598898135415706e-06, + "loss": 2.6162, + "step": 1048 + }, + { + "epoch": 0.22, + "learning_rate": 9.597452908352994e-06, + "loss": 2.6218, + "step": 1049 + }, + { + "epoch": 0.22, + "learning_rate": 9.596005191488887e-06, + "loss": 2.6557, + "step": 1050 + }, + { + "epoch": 0.22, + "learning_rate": 9.594554985607414e-06, + "loss": 2.7025, + "step": 1051 + }, + { + "epoch": 0.22, + "learning_rate": 9.59310229149395e-06, + "loss": 2.6201, + "step": 1052 + }, + { + "epoch": 0.22, + "learning_rate": 9.591647109935217e-06, + "loss": 2.6297, + "step": 1053 + }, + { + "epoch": 0.22, + "learning_rate": 9.590189441719285e-06, + "loss": 2.6416, + "step": 1054 + }, + { + "epoch": 0.22, + "learning_rate": 9.588729287635571e-06, + "loss": 2.5855, + "step": 1055 + }, + { + "epoch": 0.22, + "learning_rate": 9.587266648474835e-06, + "loss": 2.6829, + "step": 1056 + }, + { + "epoch": 0.22, + "learning_rate": 9.58580152502919e-06, + "loss": 2.6751, + "step": 1057 + }, + { + "epoch": 0.22, + "learning_rate": 9.584333918092085e-06, + "loss": 2.6492, + "step": 1058 + }, + { + "epoch": 0.22, + "learning_rate": 9.582863828458323e-06, + "loss": 2.5884, + "step": 1059 + }, + { + "epoch": 0.22, + "learning_rate": 9.581391256924043e-06, + "loss": 2.6454, + "step": 1060 + }, + { + "epoch": 0.22, + "learning_rate": 9.579916204286734e-06, + "loss": 2.6532, + "step": 1061 + }, + { + "epoch": 0.22, + "learning_rate": 9.578438671345228e-06, + "loss": 2.6146, + "step": 1062 + }, + { + "epoch": 0.22, + "learning_rate": 9.576958658899698e-06, + "loss": 2.7177, + "step": 1063 + }, + { + "epoch": 0.22, + "learning_rate": 9.575476167751663e-06, + "loss": 2.6646, + "step": 1064 + }, + { + "epoch": 0.22, + "learning_rate": 9.57399119870398e-06, + "loss": 2.6721, + "step": 1065 + }, + { + "epoch": 0.22, + "learning_rate": 9.572503752560852e-06, + "loss": 2.6879, + "step": 1066 + }, + { + "epoch": 0.22, + "learning_rate": 9.571013830127822e-06, + "loss": 2.662, + "step": 1067 + }, + { + "epoch": 0.22, + "learning_rate": 9.569521432211773e-06, + "loss": 2.6513, + "step": 1068 + }, + { + "epoch": 0.22, + "learning_rate": 9.568026559620931e-06, + "loss": 2.7053, + "step": 1069 + }, + { + "epoch": 0.22, + "learning_rate": 9.566529213164859e-06, + "loss": 2.5893, + "step": 1070 + }, + { + "epoch": 0.22, + "learning_rate": 9.565029393654463e-06, + "loss": 2.6567, + "step": 1071 + }, + { + "epoch": 0.22, + "learning_rate": 9.563527101901985e-06, + "loss": 2.668, + "step": 1072 + }, + { + "epoch": 0.22, + "learning_rate": 9.56202233872101e-06, + "loss": 2.5901, + "step": 1073 + }, + { + "epoch": 0.23, + "learning_rate": 9.560515104926457e-06, + "loss": 2.688, + "step": 1074 + }, + { + "epoch": 0.23, + "learning_rate": 9.559005401334586e-06, + "loss": 2.6208, + "step": 1075 + }, + { + "epoch": 0.23, + "learning_rate": 9.557493228762995e-06, + "loss": 2.6646, + "step": 1076 + }, + { + "epoch": 0.23, + "learning_rate": 9.555978588030614e-06, + "loss": 2.5562, + "step": 1077 + }, + { + "epoch": 0.23, + "learning_rate": 9.554461479957718e-06, + "loss": 2.5935, + "step": 1078 + }, + { + "epoch": 0.23, + "learning_rate": 9.552941905365911e-06, + "loss": 2.7111, + "step": 1079 + }, + { + "epoch": 0.23, + "learning_rate": 9.551419865078136e-06, + "loss": 2.6339, + "step": 1080 + }, + { + "epoch": 0.23, + "learning_rate": 9.54989535991867e-06, + "loss": 2.6539, + "step": 1081 + }, + { + "epoch": 0.23, + "learning_rate": 9.548368390713126e-06, + "loss": 2.6711, + "step": 1082 + }, + { + "epoch": 0.23, + "learning_rate": 9.546838958288452e-06, + "loss": 2.634, + "step": 1083 + }, + { + "epoch": 0.23, + "learning_rate": 9.545307063472928e-06, + "loss": 2.6703, + "step": 1084 + }, + { + "epoch": 0.23, + "learning_rate": 9.543772707096169e-06, + "loss": 2.6307, + "step": 1085 + }, + { + "epoch": 0.23, + "learning_rate": 9.542235889989123e-06, + "loss": 2.6394, + "step": 1086 + }, + { + "epoch": 0.23, + "learning_rate": 9.540696612984068e-06, + "loss": 2.6671, + "step": 1087 + }, + { + "epoch": 0.23, + "learning_rate": 9.53915487691462e-06, + "loss": 2.7156, + "step": 1088 + }, + { + "epoch": 0.23, + "learning_rate": 9.53761068261572e-06, + "loss": 2.6767, + "step": 1089 + }, + { + "epoch": 0.23, + "learning_rate": 9.536064030923646e-06, + "loss": 2.603, + "step": 1090 + }, + { + "epoch": 0.23, + "learning_rate": 9.534514922676003e-06, + "loss": 2.6093, + "step": 1091 + }, + { + "epoch": 0.23, + "learning_rate": 9.532963358711727e-06, + "loss": 2.7255, + "step": 1092 + }, + { + "epoch": 0.23, + "learning_rate": 9.531409339871085e-06, + "loss": 2.6431, + "step": 1093 + }, + { + "epoch": 0.23, + "learning_rate": 9.529852866995676e-06, + "loss": 2.6174, + "step": 1094 + }, + { + "epoch": 0.23, + "learning_rate": 9.528293940928422e-06, + "loss": 2.6811, + "step": 1095 + }, + { + "epoch": 0.23, + "learning_rate": 9.526732562513575e-06, + "loss": 2.7033, + "step": 1096 + }, + { + "epoch": 0.23, + "learning_rate": 9.525168732596722e-06, + "loss": 2.7059, + "step": 1097 + }, + { + "epoch": 0.23, + "learning_rate": 9.523602452024765e-06, + "loss": 2.637, + "step": 1098 + }, + { + "epoch": 0.23, + "learning_rate": 9.522033721645948e-06, + "loss": 2.6081, + "step": 1099 + }, + { + "epoch": 0.23, + "learning_rate": 9.520462542309832e-06, + "loss": 2.6017, + "step": 1100 + }, + { + "epoch": 0.23, + "learning_rate": 9.518888914867303e-06, + "loss": 2.6414, + "step": 1101 + }, + { + "epoch": 0.23, + "learning_rate": 9.51731284017058e-06, + "loss": 2.6189, + "step": 1102 + }, + { + "epoch": 0.23, + "learning_rate": 9.515734319073204e-06, + "loss": 2.643, + "step": 1103 + }, + { + "epoch": 0.23, + "learning_rate": 9.514153352430037e-06, + "loss": 2.6261, + "step": 1104 + }, + { + "epoch": 0.23, + "learning_rate": 9.512569941097272e-06, + "loss": 2.6613, + "step": 1105 + }, + { + "epoch": 0.23, + "learning_rate": 9.510984085932421e-06, + "loss": 2.6597, + "step": 1106 + }, + { + "epoch": 0.23, + "learning_rate": 9.509395787794325e-06, + "loss": 2.6962, + "step": 1107 + }, + { + "epoch": 0.23, + "learning_rate": 9.507805047543139e-06, + "loss": 2.5964, + "step": 1108 + }, + { + "epoch": 0.23, + "learning_rate": 9.50621186604035e-06, + "loss": 2.6416, + "step": 1109 + }, + { + "epoch": 0.23, + "learning_rate": 9.50461624414876e-06, + "loss": 2.6587, + "step": 1110 + }, + { + "epoch": 0.23, + "learning_rate": 9.5030181827325e-06, + "loss": 2.6051, + "step": 1111 + }, + { + "epoch": 0.23, + "learning_rate": 9.501417682657015e-06, + "loss": 2.7114, + "step": 1112 + }, + { + "epoch": 0.23, + "learning_rate": 9.499814744789072e-06, + "loss": 2.6969, + "step": 1113 + }, + { + "epoch": 0.23, + "learning_rate": 9.498209369996763e-06, + "loss": 2.6394, + "step": 1114 + }, + { + "epoch": 0.23, + "learning_rate": 9.496601559149494e-06, + "loss": 2.6476, + "step": 1115 + }, + { + "epoch": 0.23, + "learning_rate": 9.494991313117992e-06, + "loss": 2.6277, + "step": 1116 + }, + { + "epoch": 0.23, + "learning_rate": 9.493378632774308e-06, + "loss": 2.6268, + "step": 1117 + }, + { + "epoch": 0.23, + "learning_rate": 9.491763518991803e-06, + "loss": 2.684, + "step": 1118 + }, + { + "epoch": 0.23, + "learning_rate": 9.490145972645159e-06, + "loss": 2.6216, + "step": 1119 + }, + { + "epoch": 0.23, + "learning_rate": 9.488525994610379e-06, + "loss": 2.6976, + "step": 1120 + }, + { + "epoch": 0.24, + "learning_rate": 9.486903585764778e-06, + "loss": 2.6812, + "step": 1121 + }, + { + "epoch": 0.24, + "learning_rate": 9.485278746986993e-06, + "loss": 2.6073, + "step": 1122 + }, + { + "epoch": 0.24, + "learning_rate": 9.483651479156966e-06, + "loss": 2.5872, + "step": 1123 + }, + { + "epoch": 0.24, + "learning_rate": 9.482021783155971e-06, + "loss": 2.6731, + "step": 1124 + }, + { + "epoch": 0.24, + "learning_rate": 9.48038965986658e-06, + "loss": 2.6014, + "step": 1125 + }, + { + "epoch": 0.24, + "learning_rate": 9.478755110172691e-06, + "loss": 2.6926, + "step": 1126 + }, + { + "epoch": 0.24, + "learning_rate": 9.477118134959513e-06, + "loss": 2.6432, + "step": 1127 + }, + { + "epoch": 0.24, + "learning_rate": 9.475478735113567e-06, + "loss": 2.6986, + "step": 1128 + }, + { + "epoch": 0.24, + "learning_rate": 9.473836911522688e-06, + "loss": 2.6258, + "step": 1129 + }, + { + "epoch": 0.24, + "learning_rate": 9.472192665076023e-06, + "loss": 2.6591, + "step": 1130 + }, + { + "epoch": 0.24, + "learning_rate": 9.470545996664033e-06, + "loss": 2.6841, + "step": 1131 + }, + { + "epoch": 0.24, + "learning_rate": 9.468896907178491e-06, + "loss": 2.6602, + "step": 1132 + }, + { + "epoch": 0.24, + "learning_rate": 9.467245397512475e-06, + "loss": 2.6485, + "step": 1133 + }, + { + "epoch": 0.24, + "learning_rate": 9.465591468560383e-06, + "loss": 2.6208, + "step": 1134 + }, + { + "epoch": 0.24, + "learning_rate": 9.463935121217916e-06, + "loss": 2.6957, + "step": 1135 + }, + { + "epoch": 0.24, + "learning_rate": 9.46227635638209e-06, + "loss": 2.6935, + "step": 1136 + }, + { + "epoch": 0.24, + "learning_rate": 9.460615174951221e-06, + "loss": 2.6209, + "step": 1137 + }, + { + "epoch": 0.24, + "learning_rate": 9.458951577824946e-06, + "loss": 2.6919, + "step": 1138 + }, + { + "epoch": 0.24, + "learning_rate": 9.457285565904204e-06, + "loss": 2.6565, + "step": 1139 + }, + { + "epoch": 0.24, + "learning_rate": 9.45561714009124e-06, + "loss": 2.7144, + "step": 1140 + }, + { + "epoch": 0.24, + "learning_rate": 9.45394630128961e-06, + "loss": 2.6472, + "step": 1141 + }, + { + "epoch": 0.24, + "learning_rate": 9.452273050404173e-06, + "loss": 2.6396, + "step": 1142 + }, + { + "epoch": 0.24, + "learning_rate": 9.450597388341099e-06, + "loss": 2.6454, + "step": 1143 + }, + { + "epoch": 0.24, + "learning_rate": 9.448919316007861e-06, + "loss": 2.7306, + "step": 1144 + }, + { + "epoch": 0.24, + "learning_rate": 9.447238834313235e-06, + "loss": 2.709, + "step": 1145 + }, + { + "epoch": 0.24, + "learning_rate": 9.445555944167308e-06, + "loss": 2.5913, + "step": 1146 + }, + { + "epoch": 0.24, + "learning_rate": 9.443870646481465e-06, + "loss": 2.6529, + "step": 1147 + }, + { + "epoch": 0.24, + "learning_rate": 9.442182942168398e-06, + "loss": 2.5819, + "step": 1148 + }, + { + "epoch": 0.24, + "learning_rate": 9.440492832142105e-06, + "loss": 2.7027, + "step": 1149 + }, + { + "epoch": 0.24, + "learning_rate": 9.43880031731788e-06, + "loss": 2.6302, + "step": 1150 + }, + { + "epoch": 0.24, + "learning_rate": 9.437105398612323e-06, + "loss": 2.6759, + "step": 1151 + }, + { + "epoch": 0.24, + "learning_rate": 9.43540807694334e-06, + "loss": 2.6162, + "step": 1152 + }, + { + "epoch": 0.24, + "learning_rate": 9.433708353230128e-06, + "loss": 2.6444, + "step": 1153 + }, + { + "epoch": 0.24, + "learning_rate": 9.432006228393198e-06, + "loss": 2.6632, + "step": 1154 + }, + { + "epoch": 0.24, + "learning_rate": 9.43030170335435e-06, + "loss": 2.712, + "step": 1155 + }, + { + "epoch": 0.24, + "learning_rate": 9.42859477903669e-06, + "loss": 2.6453, + "step": 1156 + }, + { + "epoch": 0.24, + "learning_rate": 9.426885456364622e-06, + "loss": 2.6077, + "step": 1157 + }, + { + "epoch": 0.24, + "learning_rate": 9.425173736263847e-06, + "loss": 2.6569, + "step": 1158 + }, + { + "epoch": 0.24, + "learning_rate": 9.423459619661367e-06, + "loss": 2.6245, + "step": 1159 + }, + { + "epoch": 0.24, + "learning_rate": 9.42174310748548e-06, + "loss": 2.6642, + "step": 1160 + }, + { + "epoch": 0.24, + "learning_rate": 9.420024200665783e-06, + "loss": 2.6666, + "step": 1161 + }, + { + "epoch": 0.24, + "learning_rate": 9.41830290013317e-06, + "loss": 2.6086, + "step": 1162 + }, + { + "epoch": 0.24, + "learning_rate": 9.416579206819828e-06, + "loss": 2.6267, + "step": 1163 + }, + { + "epoch": 0.24, + "learning_rate": 9.414853121659244e-06, + "loss": 2.6244, + "step": 1164 + }, + { + "epoch": 0.24, + "learning_rate": 9.413124645586195e-06, + "loss": 2.5915, + "step": 1165 + }, + { + "epoch": 0.24, + "learning_rate": 9.411393779536761e-06, + "loss": 2.6781, + "step": 1166 + }, + { + "epoch": 0.24, + "learning_rate": 9.40966052444831e-06, + "loss": 2.6769, + "step": 1167 + }, + { + "epoch": 0.24, + "learning_rate": 9.407924881259505e-06, + "loss": 2.7021, + "step": 1168 + }, + { + "epoch": 0.25, + "learning_rate": 9.406186850910301e-06, + "loss": 2.6302, + "step": 1169 + }, + { + "epoch": 0.25, + "learning_rate": 9.404446434341951e-06, + "loss": 2.6249, + "step": 1170 + }, + { + "epoch": 0.25, + "learning_rate": 9.402703632496995e-06, + "loss": 2.6589, + "step": 1171 + }, + { + "epoch": 0.25, + "learning_rate": 9.400958446319267e-06, + "loss": 2.6956, + "step": 1172 + }, + { + "epoch": 0.25, + "learning_rate": 9.39921087675389e-06, + "loss": 2.6579, + "step": 1173 + }, + { + "epoch": 0.25, + "learning_rate": 9.397460924747282e-06, + "loss": 2.7117, + "step": 1174 + }, + { + "epoch": 0.25, + "learning_rate": 9.395708591247148e-06, + "loss": 2.6764, + "step": 1175 + }, + { + "epoch": 0.25, + "learning_rate": 9.393953877202484e-06, + "loss": 2.6679, + "step": 1176 + }, + { + "epoch": 0.25, + "learning_rate": 9.392196783563573e-06, + "loss": 2.6461, + "step": 1177 + }, + { + "epoch": 0.25, + "learning_rate": 9.39043731128199e-06, + "loss": 2.6531, + "step": 1178 + }, + { + "epoch": 0.25, + "learning_rate": 9.388675461310596e-06, + "loss": 2.6294, + "step": 1179 + }, + { + "epoch": 0.25, + "learning_rate": 9.386911234603542e-06, + "loss": 2.6523, + "step": 1180 + }, + { + "epoch": 0.25, + "learning_rate": 9.385144632116263e-06, + "loss": 2.7144, + "step": 1181 + }, + { + "epoch": 0.25, + "learning_rate": 9.383375654805482e-06, + "loss": 2.6532, + "step": 1182 + }, + { + "epoch": 0.25, + "learning_rate": 9.381604303629208e-06, + "loss": 2.6196, + "step": 1183 + }, + { + "epoch": 0.25, + "learning_rate": 9.379830579546736e-06, + "loss": 2.6605, + "step": 1184 + }, + { + "epoch": 0.25, + "learning_rate": 9.378054483518648e-06, + "loss": 2.6201, + "step": 1185 + }, + { + "epoch": 0.25, + "learning_rate": 9.376276016506805e-06, + "loss": 2.6465, + "step": 1186 + }, + { + "epoch": 0.25, + "learning_rate": 9.374495179474356e-06, + "loss": 2.7001, + "step": 1187 + }, + { + "epoch": 0.25, + "learning_rate": 9.372711973385736e-06, + "loss": 2.5905, + "step": 1188 + }, + { + "epoch": 0.25, + "learning_rate": 9.370926399206657e-06, + "loss": 2.6505, + "step": 1189 + }, + { + "epoch": 0.25, + "learning_rate": 9.369138457904116e-06, + "loss": 2.6873, + "step": 1190 + }, + { + "epoch": 0.25, + "learning_rate": 9.367348150446396e-06, + "loss": 2.5855, + "step": 1191 + }, + { + "epoch": 0.25, + "learning_rate": 9.365555477803055e-06, + "loss": 2.6569, + "step": 1192 + }, + { + "epoch": 0.25, + "learning_rate": 9.363760440944933e-06, + "loss": 2.7236, + "step": 1193 + }, + { + "epoch": 0.25, + "learning_rate": 9.361963040844156e-06, + "loss": 2.6887, + "step": 1194 + }, + { + "epoch": 0.25, + "learning_rate": 9.360163278474124e-06, + "loss": 2.6999, + "step": 1195 + }, + { + "epoch": 0.25, + "learning_rate": 9.358361154809517e-06, + "loss": 2.6465, + "step": 1196 + }, + { + "epoch": 0.25, + "learning_rate": 9.356556670826297e-06, + "loss": 2.6302, + "step": 1197 + }, + { + "epoch": 0.25, + "learning_rate": 9.354749827501702e-06, + "loss": 2.6881, + "step": 1198 + }, + { + "epoch": 0.25, + "learning_rate": 9.352940625814244e-06, + "loss": 2.6373, + "step": 1199 + }, + { + "epoch": 0.25, + "learning_rate": 9.351129066743723e-06, + "loss": 2.7034, + "step": 1200 + }, + { + "epoch": 0.25, + "learning_rate": 9.349315151271204e-06, + "loss": 2.6598, + "step": 1201 + }, + { + "epoch": 0.25, + "learning_rate": 9.347498880379036e-06, + "loss": 2.5474, + "step": 1202 + }, + { + "epoch": 0.25, + "learning_rate": 9.345680255050836e-06, + "loss": 2.5871, + "step": 1203 + }, + { + "epoch": 0.25, + "learning_rate": 9.343859276271505e-06, + "loss": 2.6212, + "step": 1204 + }, + { + "epoch": 0.25, + "learning_rate": 9.342035945027213e-06, + "loss": 2.6249, + "step": 1205 + }, + { + "epoch": 0.25, + "learning_rate": 9.340210262305405e-06, + "loss": 2.6758, + "step": 1206 + }, + { + "epoch": 0.25, + "learning_rate": 9.338382229094798e-06, + "loss": 2.6247, + "step": 1207 + }, + { + "epoch": 0.25, + "learning_rate": 9.336551846385386e-06, + "loss": 2.7236, + "step": 1208 + }, + { + "epoch": 0.25, + "learning_rate": 9.334719115168431e-06, + "loss": 2.6548, + "step": 1209 + }, + { + "epoch": 0.25, + "learning_rate": 9.332884036436472e-06, + "loss": 2.6561, + "step": 1210 + }, + { + "epoch": 0.25, + "learning_rate": 9.331046611183311e-06, + "loss": 2.6367, + "step": 1211 + }, + { + "epoch": 0.25, + "learning_rate": 9.32920684040403e-06, + "loss": 2.6729, + "step": 1212 + }, + { + "epoch": 0.25, + "learning_rate": 9.327364725094977e-06, + "loss": 2.7049, + "step": 1213 + }, + { + "epoch": 0.25, + "learning_rate": 9.325520266253769e-06, + "loss": 2.6508, + "step": 1214 + }, + { + "epoch": 0.25, + "learning_rate": 9.323673464879293e-06, + "loss": 2.7611, + "step": 1215 + }, + { + "epoch": 0.25, + "learning_rate": 9.321824321971703e-06, + "loss": 2.6395, + "step": 1216 + }, + { + "epoch": 0.26, + "learning_rate": 9.319972838532425e-06, + "loss": 2.6209, + "step": 1217 + }, + { + "epoch": 0.26, + "learning_rate": 9.318119015564151e-06, + "loss": 2.6106, + "step": 1218 + }, + { + "epoch": 0.26, + "learning_rate": 9.316262854070836e-06, + "loss": 2.6577, + "step": 1219 + }, + { + "epoch": 0.26, + "learning_rate": 9.314404355057708e-06, + "loss": 2.731, + "step": 1220 + }, + { + "epoch": 0.26, + "learning_rate": 9.312543519531258e-06, + "loss": 2.6473, + "step": 1221 + }, + { + "epoch": 0.26, + "learning_rate": 9.310680348499239e-06, + "loss": 2.6725, + "step": 1222 + }, + { + "epoch": 0.26, + "learning_rate": 9.308814842970675e-06, + "loss": 2.6747, + "step": 1223 + }, + { + "epoch": 0.26, + "learning_rate": 9.306947003955848e-06, + "loss": 2.6659, + "step": 1224 + }, + { + "epoch": 0.26, + "learning_rate": 9.305076832466308e-06, + "loss": 2.6873, + "step": 1225 + }, + { + "epoch": 0.26, + "learning_rate": 9.303204329514868e-06, + "loss": 2.686, + "step": 1226 + }, + { + "epoch": 0.26, + "learning_rate": 9.301329496115601e-06, + "loss": 2.6557, + "step": 1227 + }, + { + "epoch": 0.26, + "learning_rate": 9.299452333283844e-06, + "loss": 2.6881, + "step": 1228 + }, + { + "epoch": 0.26, + "learning_rate": 9.297572842036199e-06, + "loss": 2.6246, + "step": 1229 + }, + { + "epoch": 0.26, + "learning_rate": 9.29569102339052e-06, + "loss": 2.6508, + "step": 1230 + }, + { + "epoch": 0.26, + "learning_rate": 9.29380687836593e-06, + "loss": 2.6801, + "step": 1231 + }, + { + "epoch": 0.26, + "learning_rate": 9.291920407982807e-06, + "loss": 2.6384, + "step": 1232 + }, + { + "epoch": 0.26, + "learning_rate": 9.290031613262791e-06, + "loss": 2.6635, + "step": 1233 + }, + { + "epoch": 0.26, + "learning_rate": 9.28814049522878e-06, + "loss": 2.6254, + "step": 1234 + }, + { + "epoch": 0.26, + "learning_rate": 9.286247054904926e-06, + "loss": 2.6831, + "step": 1235 + }, + { + "epoch": 0.26, + "learning_rate": 9.284351293316647e-06, + "loss": 2.6704, + "step": 1236 + }, + { + "epoch": 0.26, + "learning_rate": 9.282453211490611e-06, + "loss": 2.6848, + "step": 1237 + }, + { + "epoch": 0.26, + "learning_rate": 9.280552810454745e-06, + "loss": 2.7522, + "step": 1238 + }, + { + "epoch": 0.26, + "learning_rate": 9.278650091238237e-06, + "loss": 2.6251, + "step": 1239 + }, + { + "epoch": 0.26, + "learning_rate": 9.27674505487152e-06, + "loss": 2.6751, + "step": 1240 + }, + { + "epoch": 0.26, + "learning_rate": 9.274837702386287e-06, + "loss": 2.6446, + "step": 1241 + }, + { + "epoch": 0.26, + "learning_rate": 9.27292803481549e-06, + "loss": 2.6328, + "step": 1242 + }, + { + "epoch": 0.26, + "learning_rate": 9.27101605319333e-06, + "loss": 2.6572, + "step": 1243 + }, + { + "epoch": 0.26, + "learning_rate": 9.26910175855526e-06, + "loss": 2.6322, + "step": 1244 + }, + { + "epoch": 0.26, + "learning_rate": 9.267185151937987e-06, + "loss": 2.6355, + "step": 1245 + }, + { + "epoch": 0.26, + "learning_rate": 9.265266234379473e-06, + "loss": 2.671, + "step": 1246 + }, + { + "epoch": 0.26, + "learning_rate": 9.263345006918926e-06, + "loss": 2.5757, + "step": 1247 + }, + { + "epoch": 0.26, + "learning_rate": 9.26142147059681e-06, + "loss": 2.6282, + "step": 1248 + }, + { + "epoch": 0.26, + "learning_rate": 9.259495626454836e-06, + "loss": 2.7099, + "step": 1249 + }, + { + "epoch": 0.26, + "learning_rate": 9.257567475535966e-06, + "loss": 2.6381, + "step": 1250 + }, + { + "epoch": 0.26, + "learning_rate": 9.255637018884414e-06, + "loss": 2.6268, + "step": 1251 + }, + { + "epoch": 0.26, + "learning_rate": 9.253704257545636e-06, + "loss": 2.6845, + "step": 1252 + }, + { + "epoch": 0.26, + "learning_rate": 9.251769192566346e-06, + "loss": 2.6476, + "step": 1253 + }, + { + "epoch": 0.26, + "learning_rate": 9.249831824994496e-06, + "loss": 2.5969, + "step": 1254 + }, + { + "epoch": 0.26, + "learning_rate": 9.247892155879287e-06, + "loss": 2.68, + "step": 1255 + }, + { + "epoch": 0.26, + "learning_rate": 9.24595018627117e-06, + "loss": 2.6693, + "step": 1256 + }, + { + "epoch": 0.26, + "learning_rate": 9.244005917221843e-06, + "loss": 2.6307, + "step": 1257 + }, + { + "epoch": 0.26, + "learning_rate": 9.242059349784243e-06, + "loss": 2.6684, + "step": 1258 + }, + { + "epoch": 0.26, + "learning_rate": 9.240110485012557e-06, + "loss": 2.6602, + "step": 1259 + }, + { + "epoch": 0.26, + "learning_rate": 9.238159323962211e-06, + "loss": 2.6618, + "step": 1260 + }, + { + "epoch": 0.26, + "learning_rate": 9.236205867689882e-06, + "loss": 2.6357, + "step": 1261 + }, + { + "epoch": 0.26, + "learning_rate": 9.234250117253482e-06, + "loss": 2.6299, + "step": 1262 + }, + { + "epoch": 0.26, + "learning_rate": 9.232292073712173e-06, + "loss": 2.5922, + "step": 1263 + }, + { + "epoch": 0.27, + "learning_rate": 9.230331738126352e-06, + "loss": 2.6771, + "step": 1264 + }, + { + "epoch": 0.27, + "learning_rate": 9.228369111557663e-06, + "loss": 2.653, + "step": 1265 + }, + { + "epoch": 0.27, + "learning_rate": 9.226404195068985e-06, + "loss": 2.6426, + "step": 1266 + }, + { + "epoch": 0.27, + "learning_rate": 9.224436989724444e-06, + "loss": 2.6723, + "step": 1267 + }, + { + "epoch": 0.27, + "learning_rate": 9.222467496589398e-06, + "loss": 2.6092, + "step": 1268 + }, + { + "epoch": 0.27, + "learning_rate": 9.22049571673045e-06, + "loss": 2.6001, + "step": 1269 + }, + { + "epoch": 0.27, + "learning_rate": 9.21852165121544e-06, + "loss": 2.6784, + "step": 1270 + }, + { + "epoch": 0.27, + "learning_rate": 9.21654530111344e-06, + "loss": 2.6058, + "step": 1271 + }, + { + "epoch": 0.27, + "learning_rate": 9.214566667494772e-06, + "loss": 2.6768, + "step": 1272 + }, + { + "epoch": 0.27, + "learning_rate": 9.21258575143098e-06, + "loss": 2.6276, + "step": 1273 + }, + { + "epoch": 0.27, + "learning_rate": 9.210602553994854e-06, + "loss": 2.62, + "step": 1274 + }, + { + "epoch": 0.27, + "learning_rate": 9.208617076260416e-06, + "loss": 2.654, + "step": 1275 + }, + { + "epoch": 0.27, + "learning_rate": 9.206629319302924e-06, + "loss": 2.7128, + "step": 1276 + }, + { + "epoch": 0.27, + "learning_rate": 9.204639284198871e-06, + "loss": 2.6615, + "step": 1277 + }, + { + "epoch": 0.27, + "learning_rate": 9.202646972025979e-06, + "loss": 2.6119, + "step": 1278 + }, + { + "epoch": 0.27, + "learning_rate": 9.200652383863207e-06, + "loss": 2.6591, + "step": 1279 + }, + { + "epoch": 0.27, + "learning_rate": 9.19865552079075e-06, + "loss": 2.676, + "step": 1280 + }, + { + "epoch": 0.27, + "learning_rate": 9.196656383890029e-06, + "loss": 2.6612, + "step": 1281 + }, + { + "epoch": 0.27, + "learning_rate": 9.194654974243697e-06, + "loss": 2.6389, + "step": 1282 + }, + { + "epoch": 0.27, + "learning_rate": 9.192651292935642e-06, + "loss": 2.5745, + "step": 1283 + }, + { + "epoch": 0.27, + "learning_rate": 9.190645341050978e-06, + "loss": 2.6685, + "step": 1284 + }, + { + "epoch": 0.27, + "learning_rate": 9.188637119676053e-06, + "loss": 2.649, + "step": 1285 + }, + { + "epoch": 0.27, + "learning_rate": 9.186626629898439e-06, + "loss": 2.6971, + "step": 1286 + }, + { + "epoch": 0.27, + "learning_rate": 9.184613872806939e-06, + "loss": 2.6767, + "step": 1287 + }, + { + "epoch": 0.27, + "learning_rate": 9.182598849491584e-06, + "loss": 2.7113, + "step": 1288 + }, + { + "epoch": 0.27, + "learning_rate": 9.180581561043633e-06, + "loss": 2.5907, + "step": 1289 + }, + { + "epoch": 0.27, + "learning_rate": 9.178562008555571e-06, + "loss": 2.6143, + "step": 1290 + }, + { + "epoch": 0.27, + "learning_rate": 9.176540193121108e-06, + "loss": 2.6296, + "step": 1291 + }, + { + "epoch": 0.27, + "learning_rate": 9.174516115835181e-06, + "loss": 2.7127, + "step": 1292 + }, + { + "epoch": 0.27, + "learning_rate": 9.172489777793952e-06, + "loss": 2.639, + "step": 1293 + }, + { + "epoch": 0.27, + "learning_rate": 9.170461180094805e-06, + "loss": 2.7175, + "step": 1294 + }, + { + "epoch": 0.27, + "learning_rate": 9.168430323836351e-06, + "loss": 2.6829, + "step": 1295 + }, + { + "epoch": 0.27, + "learning_rate": 9.166397210118421e-06, + "loss": 2.6223, + "step": 1296 + }, + { + "epoch": 0.27, + "learning_rate": 9.164361840042072e-06, + "loss": 2.6983, + "step": 1297 + }, + { + "epoch": 0.27, + "learning_rate": 9.162324214709582e-06, + "loss": 2.6389, + "step": 1298 + }, + { + "epoch": 0.27, + "learning_rate": 9.160284335224445e-06, + "loss": 2.6831, + "step": 1299 + }, + { + "epoch": 0.27, + "learning_rate": 9.158242202691386e-06, + "loss": 2.6623, + "step": 1300 + }, + { + "epoch": 0.27, + "learning_rate": 9.15619781821634e-06, + "loss": 2.6369, + "step": 1301 + }, + { + "epoch": 0.27, + "learning_rate": 9.154151182906467e-06, + "loss": 2.6545, + "step": 1302 + }, + { + "epoch": 0.27, + "learning_rate": 9.152102297870146e-06, + "loss": 2.6291, + "step": 1303 + }, + { + "epoch": 0.27, + "learning_rate": 9.150051164216976e-06, + "loss": 2.6689, + "step": 1304 + }, + { + "epoch": 0.27, + "learning_rate": 9.147997783057765e-06, + "loss": 2.632, + "step": 1305 + }, + { + "epoch": 0.27, + "learning_rate": 9.145942155504548e-06, + "loss": 2.6359, + "step": 1306 + }, + { + "epoch": 0.27, + "learning_rate": 9.143884282670572e-06, + "loss": 2.6177, + "step": 1307 + }, + { + "epoch": 0.27, + "learning_rate": 9.141824165670298e-06, + "loss": 2.6667, + "step": 1308 + }, + { + "epoch": 0.27, + "learning_rate": 9.139761805619411e-06, + "loss": 2.6869, + "step": 1309 + }, + { + "epoch": 0.27, + "learning_rate": 9.1376972036348e-06, + "loss": 2.6149, + "step": 1310 + }, + { + "epoch": 0.27, + "learning_rate": 9.135630360834577e-06, + "loss": 2.6579, + "step": 1311 + }, + { + "epoch": 0.28, + "learning_rate": 9.13356127833806e-06, + "loss": 2.7006, + "step": 1312 + }, + { + "epoch": 0.28, + "learning_rate": 9.131489957265785e-06, + "loss": 2.6958, + "step": 1313 + }, + { + "epoch": 0.28, + "learning_rate": 9.1294163987395e-06, + "loss": 2.6511, + "step": 1314 + }, + { + "epoch": 0.28, + "learning_rate": 9.12734060388216e-06, + "loss": 2.7316, + "step": 1315 + }, + { + "epoch": 0.28, + "learning_rate": 9.125262573817937e-06, + "loss": 2.6654, + "step": 1316 + }, + { + "epoch": 0.28, + "learning_rate": 9.123182309672212e-06, + "loss": 2.645, + "step": 1317 + }, + { + "epoch": 0.28, + "learning_rate": 9.121099812571571e-06, + "loss": 2.6542, + "step": 1318 + }, + { + "epoch": 0.28, + "learning_rate": 9.119015083643819e-06, + "loss": 2.6173, + "step": 1319 + }, + { + "epoch": 0.28, + "learning_rate": 9.11692812401796e-06, + "loss": 2.676, + "step": 1320 + }, + { + "epoch": 0.28, + "learning_rate": 9.114838934824213e-06, + "loss": 2.6503, + "step": 1321 + }, + { + "epoch": 0.28, + "learning_rate": 9.112747517193998e-06, + "loss": 2.5862, + "step": 1322 + }, + { + "epoch": 0.28, + "learning_rate": 9.110653872259947e-06, + "loss": 2.7228, + "step": 1323 + }, + { + "epoch": 0.28, + "learning_rate": 9.108558001155896e-06, + "loss": 2.6788, + "step": 1324 + }, + { + "epoch": 0.28, + "learning_rate": 9.106459905016889e-06, + "loss": 2.7018, + "step": 1325 + }, + { + "epoch": 0.28, + "learning_rate": 9.104359584979173e-06, + "loss": 2.6272, + "step": 1326 + }, + { + "epoch": 0.28, + "learning_rate": 9.102257042180194e-06, + "loss": 2.6104, + "step": 1327 + }, + { + "epoch": 0.28, + "learning_rate": 9.100152277758616e-06, + "loss": 2.597, + "step": 1328 + }, + { + "epoch": 0.28, + "learning_rate": 9.098045292854289e-06, + "loss": 2.613, + "step": 1329 + }, + { + "epoch": 0.28, + "learning_rate": 9.095936088608282e-06, + "loss": 2.6775, + "step": 1330 + }, + { + "epoch": 0.28, + "learning_rate": 9.093824666162851e-06, + "loss": 2.6421, + "step": 1331 + }, + { + "epoch": 0.28, + "learning_rate": 9.091711026661463e-06, + "loss": 2.6442, + "step": 1332 + }, + { + "epoch": 0.28, + "learning_rate": 9.089595171248784e-06, + "loss": 2.7233, + "step": 1333 + }, + { + "epoch": 0.28, + "learning_rate": 9.087477101070676e-06, + "loss": 2.6146, + "step": 1334 + }, + { + "epoch": 0.28, + "learning_rate": 9.085356817274207e-06, + "loss": 2.6649, + "step": 1335 + }, + { + "epoch": 0.28, + "learning_rate": 9.083234321007636e-06, + "loss": 2.6804, + "step": 1336 + }, + { + "epoch": 0.28, + "learning_rate": 9.081109613420428e-06, + "loss": 2.6401, + "step": 1337 + }, + { + "epoch": 0.28, + "learning_rate": 9.078982695663238e-06, + "loss": 2.7124, + "step": 1338 + }, + { + "epoch": 0.28, + "learning_rate": 9.076853568887926e-06, + "loss": 2.6126, + "step": 1339 + }, + { + "epoch": 0.28, + "learning_rate": 9.07472223424754e-06, + "loss": 2.6546, + "step": 1340 + }, + { + "epoch": 0.28, + "learning_rate": 9.072588692896333e-06, + "loss": 2.6146, + "step": 1341 + }, + { + "epoch": 0.28, + "learning_rate": 9.070452945989742e-06, + "loss": 2.6718, + "step": 1342 + }, + { + "epoch": 0.28, + "learning_rate": 9.068314994684408e-06, + "loss": 2.6944, + "step": 1343 + }, + { + "epoch": 0.28, + "learning_rate": 9.066174840138162e-06, + "loss": 2.6596, + "step": 1344 + }, + { + "epoch": 0.28, + "learning_rate": 9.064032483510027e-06, + "loss": 2.7156, + "step": 1345 + }, + { + "epoch": 0.28, + "learning_rate": 9.061887925960219e-06, + "loss": 2.6388, + "step": 1346 + }, + { + "epoch": 0.28, + "learning_rate": 9.059741168650151e-06, + "loss": 2.6773, + "step": 1347 + }, + { + "epoch": 0.28, + "learning_rate": 9.057592212742419e-06, + "loss": 2.6012, + "step": 1348 + }, + { + "epoch": 0.28, + "learning_rate": 9.055441059400817e-06, + "loss": 2.6773, + "step": 1349 + }, + { + "epoch": 0.28, + "learning_rate": 9.05328770979032e-06, + "loss": 2.6654, + "step": 1350 + }, + { + "epoch": 0.28, + "learning_rate": 9.051132165077104e-06, + "loss": 2.6416, + "step": 1351 + }, + { + "epoch": 0.28, + "learning_rate": 9.048974426428527e-06, + "loss": 2.6274, + "step": 1352 + }, + { + "epoch": 0.28, + "learning_rate": 9.046814495013137e-06, + "loss": 2.6238, + "step": 1353 + }, + { + "epoch": 0.28, + "learning_rate": 9.044652372000664e-06, + "loss": 2.7003, + "step": 1354 + }, + { + "epoch": 0.28, + "learning_rate": 9.042488058562036e-06, + "loss": 2.6613, + "step": 1355 + }, + { + "epoch": 0.28, + "learning_rate": 9.040321555869356e-06, + "loss": 2.575, + "step": 1356 + }, + { + "epoch": 0.28, + "learning_rate": 9.038152865095918e-06, + "loss": 2.6072, + "step": 1357 + }, + { + "epoch": 0.28, + "learning_rate": 9.035981987416204e-06, + "loss": 2.6858, + "step": 1358 + }, + { + "epoch": 0.28, + "learning_rate": 9.033808924005872e-06, + "loss": 2.636, + "step": 1359 + }, + { + "epoch": 0.29, + "learning_rate": 9.031633676041773e-06, + "loss": 2.6808, + "step": 1360 + }, + { + "epoch": 0.29, + "learning_rate": 9.029456244701933e-06, + "loss": 2.618, + "step": 1361 + }, + { + "epoch": 0.29, + "learning_rate": 9.027276631165566e-06, + "loss": 2.6407, + "step": 1362 + }, + { + "epoch": 0.29, + "learning_rate": 9.025094836613066e-06, + "loss": 2.7212, + "step": 1363 + }, + { + "epoch": 0.29, + "learning_rate": 9.022910862226005e-06, + "loss": 2.6499, + "step": 1364 + }, + { + "epoch": 0.29, + "learning_rate": 9.020724709187143e-06, + "loss": 2.659, + "step": 1365 + }, + { + "epoch": 0.29, + "learning_rate": 9.018536378680411e-06, + "loss": 2.6446, + "step": 1366 + }, + { + "epoch": 0.29, + "learning_rate": 9.016345871890927e-06, + "loss": 2.7091, + "step": 1367 + }, + { + "epoch": 0.29, + "learning_rate": 9.014153190004981e-06, + "loss": 2.6742, + "step": 1368 + }, + { + "epoch": 0.29, + "learning_rate": 9.011958334210047e-06, + "loss": 2.6474, + "step": 1369 + }, + { + "epoch": 0.29, + "learning_rate": 9.009761305694771e-06, + "loss": 2.6689, + "step": 1370 + }, + { + "epoch": 0.29, + "learning_rate": 9.007562105648979e-06, + "loss": 2.5952, + "step": 1371 + }, + { + "epoch": 0.29, + "learning_rate": 9.005360735263673e-06, + "loss": 2.6022, + "step": 1372 + }, + { + "epoch": 0.29, + "learning_rate": 9.003157195731028e-06, + "loss": 2.6094, + "step": 1373 + }, + { + "epoch": 0.29, + "learning_rate": 9.000951488244395e-06, + "loss": 2.6686, + "step": 1374 + }, + { + "epoch": 0.29, + "learning_rate": 8.998743613998302e-06, + "loss": 2.6961, + "step": 1375 + }, + { + "epoch": 0.29, + "learning_rate": 8.996533574188446e-06, + "loss": 2.6665, + "step": 1376 + }, + { + "epoch": 0.29, + "learning_rate": 8.994321370011696e-06, + "loss": 2.6273, + "step": 1377 + }, + { + "epoch": 0.29, + "learning_rate": 8.9921070026661e-06, + "loss": 2.637, + "step": 1378 + }, + { + "epoch": 0.29, + "learning_rate": 8.989890473350869e-06, + "loss": 2.622, + "step": 1379 + }, + { + "epoch": 0.29, + "learning_rate": 8.987671783266392e-06, + "loss": 2.6088, + "step": 1380 + }, + { + "epoch": 0.29, + "learning_rate": 8.985450933614223e-06, + "loss": 2.6782, + "step": 1381 + }, + { + "epoch": 0.29, + "learning_rate": 8.983227925597089e-06, + "loss": 2.6136, + "step": 1382 + }, + { + "epoch": 0.29, + "learning_rate": 8.981002760418885e-06, + "loss": 2.6568, + "step": 1383 + }, + { + "epoch": 0.29, + "learning_rate": 8.978775439284672e-06, + "loss": 2.6737, + "step": 1384 + }, + { + "epoch": 0.29, + "learning_rate": 8.97654596340068e-06, + "loss": 2.632, + "step": 1385 + }, + { + "epoch": 0.29, + "learning_rate": 8.97431433397431e-06, + "loss": 2.6813, + "step": 1386 + }, + { + "epoch": 0.29, + "learning_rate": 8.972080552214121e-06, + "loss": 2.6562, + "step": 1387 + }, + { + "epoch": 0.29, + "learning_rate": 8.969844619329846e-06, + "loss": 2.6875, + "step": 1388 + }, + { + "epoch": 0.29, + "learning_rate": 8.967606536532377e-06, + "loss": 2.6314, + "step": 1389 + }, + { + "epoch": 0.29, + "learning_rate": 8.965366305033774e-06, + "loss": 2.673, + "step": 1390 + }, + { + "epoch": 0.29, + "learning_rate": 8.963123926047256e-06, + "loss": 2.6515, + "step": 1391 + }, + { + "epoch": 0.29, + "learning_rate": 8.96087940078721e-06, + "loss": 2.6342, + "step": 1392 + }, + { + "epoch": 0.29, + "learning_rate": 8.958632730469183e-06, + "loss": 2.7198, + "step": 1393 + }, + { + "epoch": 0.29, + "learning_rate": 8.956383916309888e-06, + "loss": 2.657, + "step": 1394 + }, + { + "epoch": 0.29, + "learning_rate": 8.954132959527188e-06, + "loss": 2.6422, + "step": 1395 + }, + { + "epoch": 0.29, + "learning_rate": 8.95187986134012e-06, + "loss": 2.654, + "step": 1396 + }, + { + "epoch": 0.29, + "learning_rate": 8.94962462296887e-06, + "loss": 2.7048, + "step": 1397 + }, + { + "epoch": 0.29, + "learning_rate": 8.947367245634789e-06, + "loss": 2.5822, + "step": 1398 + }, + { + "epoch": 0.29, + "learning_rate": 8.945107730560385e-06, + "loss": 2.7387, + "step": 1399 + }, + { + "epoch": 0.29, + "learning_rate": 8.942846078969323e-06, + "loss": 2.6411, + "step": 1400 + }, + { + "epoch": 0.29, + "learning_rate": 8.940582292086425e-06, + "loss": 2.6593, + "step": 1401 + }, + { + "epoch": 0.29, + "learning_rate": 8.938316371137668e-06, + "loss": 2.6515, + "step": 1402 + }, + { + "epoch": 0.29, + "learning_rate": 8.93604831735019e-06, + "loss": 2.6203, + "step": 1403 + }, + { + "epoch": 0.29, + "learning_rate": 8.933778131952277e-06, + "loss": 2.6757, + "step": 1404 + }, + { + "epoch": 0.29, + "learning_rate": 8.931505816173378e-06, + "loss": 2.6982, + "step": 1405 + }, + { + "epoch": 0.29, + "learning_rate": 8.929231371244087e-06, + "loss": 2.6071, + "step": 1406 + }, + { + "epoch": 0.3, + "learning_rate": 8.926954798396155e-06, + "loss": 2.6612, + "step": 1407 + }, + { + "epoch": 0.3, + "learning_rate": 8.924676098862484e-06, + "loss": 2.6409, + "step": 1408 + }, + { + "epoch": 0.3, + "learning_rate": 8.922395273877132e-06, + "loss": 2.6489, + "step": 1409 + }, + { + "epoch": 0.3, + "learning_rate": 8.920112324675304e-06, + "loss": 2.6976, + "step": 1410 + }, + { + "epoch": 0.3, + "learning_rate": 8.917827252493354e-06, + "loss": 2.6446, + "step": 1411 + }, + { + "epoch": 0.3, + "learning_rate": 8.915540058568792e-06, + "loss": 2.6547, + "step": 1412 + }, + { + "epoch": 0.3, + "learning_rate": 8.913250744140269e-06, + "loss": 2.6463, + "step": 1413 + }, + { + "epoch": 0.3, + "learning_rate": 8.910959310447592e-06, + "loss": 2.6879, + "step": 1414 + }, + { + "epoch": 0.3, + "learning_rate": 8.90866575873171e-06, + "loss": 2.7274, + "step": 1415 + }, + { + "epoch": 0.3, + "learning_rate": 8.906370090234721e-06, + "loss": 2.6208, + "step": 1416 + }, + { + "epoch": 0.3, + "learning_rate": 8.904072306199873e-06, + "loss": 2.6328, + "step": 1417 + }, + { + "epoch": 0.3, + "learning_rate": 8.901772407871553e-06, + "loss": 2.698, + "step": 1418 + }, + { + "epoch": 0.3, + "learning_rate": 8.899470396495297e-06, + "loss": 2.7189, + "step": 1419 + }, + { + "epoch": 0.3, + "learning_rate": 8.897166273317785e-06, + "loss": 2.6648, + "step": 1420 + }, + { + "epoch": 0.3, + "learning_rate": 8.894860039586841e-06, + "loss": 2.6979, + "step": 1421 + }, + { + "epoch": 0.3, + "learning_rate": 8.89255169655143e-06, + "loss": 2.5969, + "step": 1422 + }, + { + "epoch": 0.3, + "learning_rate": 8.890241245461661e-06, + "loss": 2.5711, + "step": 1423 + }, + { + "epoch": 0.3, + "learning_rate": 8.887928687568785e-06, + "loss": 2.6313, + "step": 1424 + }, + { + "epoch": 0.3, + "learning_rate": 8.885614024125192e-06, + "loss": 2.6522, + "step": 1425 + }, + { + "epoch": 0.3, + "learning_rate": 8.883297256384417e-06, + "loss": 2.6853, + "step": 1426 + }, + { + "epoch": 0.3, + "learning_rate": 8.880978385601127e-06, + "loss": 2.6069, + "step": 1427 + }, + { + "epoch": 0.3, + "learning_rate": 8.878657413031133e-06, + "loss": 2.6319, + "step": 1428 + }, + { + "epoch": 0.3, + "learning_rate": 8.876334339931384e-06, + "loss": 2.6684, + "step": 1429 + }, + { + "epoch": 0.3, + "learning_rate": 8.874009167559968e-06, + "loss": 2.5958, + "step": 1430 + }, + { + "epoch": 0.3, + "learning_rate": 8.871681897176103e-06, + "loss": 2.6348, + "step": 1431 + }, + { + "epoch": 0.3, + "learning_rate": 8.869352530040152e-06, + "loss": 2.5656, + "step": 1432 + }, + { + "epoch": 0.3, + "learning_rate": 8.867021067413608e-06, + "loss": 2.6799, + "step": 1433 + }, + { + "epoch": 0.3, + "learning_rate": 8.8646875105591e-06, + "loss": 2.5949, + "step": 1434 + }, + { + "epoch": 0.3, + "learning_rate": 8.862351860740393e-06, + "loss": 2.6107, + "step": 1435 + }, + { + "epoch": 0.3, + "learning_rate": 8.86001411922238e-06, + "loss": 2.6587, + "step": 1436 + }, + { + "epoch": 0.3, + "learning_rate": 8.857674287271096e-06, + "loss": 2.5936, + "step": 1437 + }, + { + "epoch": 0.3, + "learning_rate": 8.855332366153702e-06, + "loss": 2.609, + "step": 1438 + }, + { + "epoch": 0.3, + "learning_rate": 8.852988357138488e-06, + "loss": 2.5886, + "step": 1439 + }, + { + "epoch": 0.3, + "learning_rate": 8.85064226149488e-06, + "loss": 2.6121, + "step": 1440 + }, + { + "epoch": 0.3, + "learning_rate": 8.84829408049343e-06, + "loss": 2.6436, + "step": 1441 + }, + { + "epoch": 0.3, + "learning_rate": 8.845943815405827e-06, + "loss": 2.6654, + "step": 1442 + }, + { + "epoch": 0.3, + "learning_rate": 8.843591467504875e-06, + "loss": 2.6355, + "step": 1443 + }, + { + "epoch": 0.3, + "learning_rate": 8.84123703806452e-06, + "loss": 2.6752, + "step": 1444 + }, + { + "epoch": 0.3, + "learning_rate": 8.838880528359826e-06, + "loss": 2.6873, + "step": 1445 + }, + { + "epoch": 0.3, + "learning_rate": 8.83652193966699e-06, + "loss": 2.6495, + "step": 1446 + }, + { + "epoch": 0.3, + "learning_rate": 8.834161273263328e-06, + "loss": 2.6146, + "step": 1447 + }, + { + "epoch": 0.3, + "learning_rate": 8.831798530427289e-06, + "loss": 2.6398, + "step": 1448 + }, + { + "epoch": 0.3, + "learning_rate": 8.829433712438439e-06, + "loss": 2.6124, + "step": 1449 + }, + { + "epoch": 0.3, + "learning_rate": 8.827066820577472e-06, + "loss": 2.6467, + "step": 1450 + }, + { + "epoch": 0.3, + "learning_rate": 8.824697856126206e-06, + "loss": 2.6314, + "step": 1451 + }, + { + "epoch": 0.3, + "learning_rate": 8.822326820367579e-06, + "loss": 2.6295, + "step": 1452 + }, + { + "epoch": 0.3, + "learning_rate": 8.819953714585651e-06, + "loss": 2.6915, + "step": 1453 + }, + { + "epoch": 0.3, + "learning_rate": 8.817578540065605e-06, + "loss": 2.6244, + "step": 1454 + }, + { + "epoch": 0.31, + "learning_rate": 8.815201298093741e-06, + "loss": 2.6004, + "step": 1455 + }, + { + "epoch": 0.31, + "learning_rate": 8.812821989957482e-06, + "loss": 2.67, + "step": 1456 + }, + { + "epoch": 0.31, + "learning_rate": 8.81044061694537e-06, + "loss": 2.6423, + "step": 1457 + }, + { + "epoch": 0.31, + "learning_rate": 8.808057180347062e-06, + "loss": 2.6204, + "step": 1458 + }, + { + "epoch": 0.31, + "learning_rate": 8.805671681453334e-06, + "loss": 2.6793, + "step": 1459 + }, + { + "epoch": 0.31, + "learning_rate": 8.80328412155608e-06, + "loss": 2.6559, + "step": 1460 + }, + { + "epoch": 0.31, + "learning_rate": 8.800894501948309e-06, + "loss": 2.6681, + "step": 1461 + }, + { + "epoch": 0.31, + "learning_rate": 8.798502823924146e-06, + "loss": 2.6239, + "step": 1462 + }, + { + "epoch": 0.31, + "learning_rate": 8.796109088778831e-06, + "loss": 2.6717, + "step": 1463 + }, + { + "epoch": 0.31, + "learning_rate": 8.793713297808718e-06, + "loss": 2.6171, + "step": 1464 + }, + { + "epoch": 0.31, + "learning_rate": 8.791315452311273e-06, + "loss": 2.6897, + "step": 1465 + }, + { + "epoch": 0.31, + "learning_rate": 8.788915553585079e-06, + "loss": 2.616, + "step": 1466 + }, + { + "epoch": 0.31, + "learning_rate": 8.78651360292982e-06, + "loss": 2.5923, + "step": 1467 + }, + { + "epoch": 0.31, + "learning_rate": 8.784109601646309e-06, + "loss": 2.6569, + "step": 1468 + }, + { + "epoch": 0.31, + "learning_rate": 8.781703551036451e-06, + "loss": 2.6214, + "step": 1469 + }, + { + "epoch": 0.31, + "learning_rate": 8.779295452403274e-06, + "loss": 2.6654, + "step": 1470 + }, + { + "epoch": 0.31, + "learning_rate": 8.77688530705091e-06, + "loss": 2.6195, + "step": 1471 + }, + { + "epoch": 0.31, + "learning_rate": 8.774473116284598e-06, + "loss": 2.6471, + "step": 1472 + }, + { + "epoch": 0.31, + "learning_rate": 8.772058881410686e-06, + "loss": 2.7086, + "step": 1473 + }, + { + "epoch": 0.31, + "learning_rate": 8.769642603736633e-06, + "loss": 2.6997, + "step": 1474 + }, + { + "epoch": 0.31, + "learning_rate": 8.767224284570999e-06, + "loss": 2.6035, + "step": 1475 + }, + { + "epoch": 0.31, + "learning_rate": 8.764803925223449e-06, + "loss": 2.55, + "step": 1476 + }, + { + "epoch": 0.31, + "learning_rate": 8.76238152700476e-06, + "loss": 2.5944, + "step": 1477 + }, + { + "epoch": 0.31, + "learning_rate": 8.759957091226805e-06, + "loss": 2.7669, + "step": 1478 + }, + { + "epoch": 0.31, + "learning_rate": 8.757530619202566e-06, + "loss": 2.6519, + "step": 1479 + }, + { + "epoch": 0.31, + "learning_rate": 8.755102112246124e-06, + "loss": 2.676, + "step": 1480 + }, + { + "epoch": 0.31, + "learning_rate": 8.752671571672664e-06, + "loss": 2.6595, + "step": 1481 + }, + { + "epoch": 0.31, + "learning_rate": 8.750238998798473e-06, + "loss": 2.6274, + "step": 1482 + }, + { + "epoch": 0.31, + "learning_rate": 8.747804394940938e-06, + "loss": 2.645, + "step": 1483 + }, + { + "epoch": 0.31, + "learning_rate": 8.745367761418546e-06, + "loss": 2.6237, + "step": 1484 + }, + { + "epoch": 0.31, + "learning_rate": 8.74292909955088e-06, + "loss": 2.66, + "step": 1485 + }, + { + "epoch": 0.31, + "learning_rate": 8.740488410658626e-06, + "loss": 2.6766, + "step": 1486 + }, + { + "epoch": 0.31, + "learning_rate": 8.738045696063566e-06, + "loss": 2.6541, + "step": 1487 + }, + { + "epoch": 0.31, + "learning_rate": 8.735600957088582e-06, + "loss": 2.6868, + "step": 1488 + }, + { + "epoch": 0.31, + "learning_rate": 8.733154195057644e-06, + "loss": 2.6634, + "step": 1489 + }, + { + "epoch": 0.31, + "learning_rate": 8.730705411295826e-06, + "loss": 2.6624, + "step": 1490 + }, + { + "epoch": 0.31, + "learning_rate": 8.728254607129293e-06, + "loss": 2.6339, + "step": 1491 + }, + { + "epoch": 0.31, + "learning_rate": 8.725801783885306e-06, + "loss": 2.6615, + "step": 1492 + }, + { + "epoch": 0.31, + "learning_rate": 8.723346942892217e-06, + "loss": 2.7561, + "step": 1493 + }, + { + "epoch": 0.31, + "learning_rate": 8.720890085479474e-06, + "loss": 2.7048, + "step": 1494 + }, + { + "epoch": 0.31, + "learning_rate": 8.718431212977614e-06, + "loss": 2.6424, + "step": 1495 + }, + { + "epoch": 0.31, + "learning_rate": 8.715970326718269e-06, + "loss": 2.676, + "step": 1496 + }, + { + "epoch": 0.31, + "learning_rate": 8.713507428034154e-06, + "loss": 2.6844, + "step": 1497 + }, + { + "epoch": 0.31, + "learning_rate": 8.711042518259086e-06, + "loss": 2.6332, + "step": 1498 + }, + { + "epoch": 0.31, + "learning_rate": 8.708575598727958e-06, + "loss": 2.6477, + "step": 1499 + }, + { + "epoch": 0.31, + "learning_rate": 8.706106670776759e-06, + "loss": 2.7313, + "step": 1500 + }, + { + "epoch": 0.31, + "learning_rate": 8.703635735742566e-06, + "loss": 2.6328, + "step": 1501 + }, + { + "epoch": 0.31, + "learning_rate": 8.70116279496354e-06, + "loss": 2.6766, + "step": 1502 + }, + { + "epoch": 0.32, + "learning_rate": 8.69868784977893e-06, + "loss": 2.6499, + "step": 1503 + }, + { + "epoch": 0.32, + "learning_rate": 8.696210901529069e-06, + "loss": 2.6766, + "step": 1504 + }, + { + "epoch": 0.32, + "learning_rate": 8.693731951555376e-06, + "loss": 2.6739, + "step": 1505 + }, + { + "epoch": 0.32, + "learning_rate": 8.691251001200354e-06, + "loss": 2.6583, + "step": 1506 + }, + { + "epoch": 0.32, + "learning_rate": 8.688768051807589e-06, + "loss": 2.6317, + "step": 1507 + }, + { + "epoch": 0.32, + "learning_rate": 8.686283104721748e-06, + "loss": 2.6293, + "step": 1508 + }, + { + "epoch": 0.32, + "learning_rate": 8.683796161288587e-06, + "loss": 2.6978, + "step": 1509 + }, + { + "epoch": 0.32, + "learning_rate": 8.68130722285493e-06, + "loss": 2.6868, + "step": 1510 + }, + { + "epoch": 0.32, + "learning_rate": 8.678816290768695e-06, + "loss": 2.6477, + "step": 1511 + }, + { + "epoch": 0.32, + "learning_rate": 8.67632336637887e-06, + "loss": 2.695, + "step": 1512 + }, + { + "epoch": 0.32, + "learning_rate": 8.673828451035529e-06, + "loss": 2.6188, + "step": 1513 + }, + { + "epoch": 0.32, + "learning_rate": 8.671331546089818e-06, + "loss": 2.6315, + "step": 1514 + }, + { + "epoch": 0.32, + "learning_rate": 8.668832652893966e-06, + "loss": 2.613, + "step": 1515 + }, + { + "epoch": 0.32, + "learning_rate": 8.666331772801274e-06, + "loss": 2.6214, + "step": 1516 + }, + { + "epoch": 0.32, + "learning_rate": 8.663828907166123e-06, + "loss": 2.7011, + "step": 1517 + }, + { + "epoch": 0.32, + "learning_rate": 8.661324057343966e-06, + "loss": 2.6127, + "step": 1518 + }, + { + "epoch": 0.32, + "learning_rate": 8.658817224691335e-06, + "loss": 2.6699, + "step": 1519 + }, + { + "epoch": 0.32, + "learning_rate": 8.656308410565828e-06, + "loss": 2.6079, + "step": 1520 + }, + { + "epoch": 0.32, + "learning_rate": 8.653797616326125e-06, + "loss": 2.6814, + "step": 1521 + }, + { + "epoch": 0.32, + "learning_rate": 8.651284843331972e-06, + "loss": 2.601, + "step": 1522 + }, + { + "epoch": 0.32, + "learning_rate": 8.64877009294419e-06, + "loss": 2.6088, + "step": 1523 + }, + { + "epoch": 0.32, + "learning_rate": 8.646253366524672e-06, + "loss": 2.5697, + "step": 1524 + }, + { + "epoch": 0.32, + "learning_rate": 8.643734665436376e-06, + "loss": 2.632, + "step": 1525 + }, + { + "epoch": 0.32, + "learning_rate": 8.64121399104333e-06, + "loss": 2.6629, + "step": 1526 + }, + { + "epoch": 0.32, + "learning_rate": 8.638691344710642e-06, + "loss": 2.625, + "step": 1527 + }, + { + "epoch": 0.32, + "learning_rate": 8.636166727804468e-06, + "loss": 2.6209, + "step": 1528 + }, + { + "epoch": 0.32, + "learning_rate": 8.633640141692052e-06, + "loss": 2.5611, + "step": 1529 + }, + { + "epoch": 0.32, + "learning_rate": 8.631111587741687e-06, + "loss": 2.6887, + "step": 1530 + }, + { + "epoch": 0.32, + "learning_rate": 8.628581067322745e-06, + "loss": 2.581, + "step": 1531 + }, + { + "epoch": 0.32, + "learning_rate": 8.626048581805652e-06, + "loss": 2.6329, + "step": 1532 + }, + { + "epoch": 0.32, + "learning_rate": 8.623514132561908e-06, + "loss": 2.6849, + "step": 1533 + }, + { + "epoch": 0.32, + "learning_rate": 8.62097772096407e-06, + "loss": 2.6735, + "step": 1534 + }, + { + "epoch": 0.32, + "learning_rate": 8.61843934838576e-06, + "loss": 2.6874, + "step": 1535 + }, + { + "epoch": 0.32, + "learning_rate": 8.61589901620166e-06, + "loss": 2.6257, + "step": 1536 + }, + { + "epoch": 0.32, + "learning_rate": 8.613356725787517e-06, + "loss": 2.6718, + "step": 1537 + }, + { + "epoch": 0.32, + "learning_rate": 8.610812478520137e-06, + "loss": 2.6702, + "step": 1538 + }, + { + "epoch": 0.32, + "learning_rate": 8.608266275777384e-06, + "loss": 2.6712, + "step": 1539 + }, + { + "epoch": 0.32, + "learning_rate": 8.605718118938182e-06, + "loss": 2.6857, + "step": 1540 + }, + { + "epoch": 0.32, + "learning_rate": 8.603168009382513e-06, + "loss": 2.6447, + "step": 1541 + }, + { + "epoch": 0.32, + "learning_rate": 8.600615948491419e-06, + "loss": 2.6581, + "step": 1542 + }, + { + "epoch": 0.32, + "learning_rate": 8.598061937646995e-06, + "loss": 2.6726, + "step": 1543 + }, + { + "epoch": 0.32, + "learning_rate": 8.595505978232394e-06, + "loss": 2.653, + "step": 1544 + }, + { + "epoch": 0.32, + "learning_rate": 8.592948071631826e-06, + "loss": 2.6144, + "step": 1545 + }, + { + "epoch": 0.32, + "learning_rate": 8.59038821923055e-06, + "loss": 2.6777, + "step": 1546 + }, + { + "epoch": 0.32, + "learning_rate": 8.587826422414886e-06, + "loss": 2.7006, + "step": 1547 + }, + { + "epoch": 0.32, + "learning_rate": 8.5852626825722e-06, + "loss": 2.6712, + "step": 1548 + }, + { + "epoch": 0.32, + "learning_rate": 8.582697001090918e-06, + "loss": 2.6771, + "step": 1549 + }, + { + "epoch": 0.33, + "learning_rate": 8.580129379360508e-06, + "loss": 2.6728, + "step": 1550 + }, + { + "epoch": 0.33, + "learning_rate": 8.577559818771497e-06, + "loss": 2.6664, + "step": 1551 + }, + { + "epoch": 0.33, + "learning_rate": 8.574988320715458e-06, + "loss": 2.6584, + "step": 1552 + }, + { + "epoch": 0.33, + "learning_rate": 8.572414886585015e-06, + "loss": 2.6518, + "step": 1553 + }, + { + "epoch": 0.33, + "learning_rate": 8.56983951777384e-06, + "loss": 2.6014, + "step": 1554 + }, + { + "epoch": 0.33, + "learning_rate": 8.56726221567665e-06, + "loss": 2.6266, + "step": 1555 + }, + { + "epoch": 0.33, + "learning_rate": 8.564682981689214e-06, + "loss": 2.6292, + "step": 1556 + }, + { + "epoch": 0.33, + "learning_rate": 8.562101817208343e-06, + "loss": 2.7156, + "step": 1557 + }, + { + "epoch": 0.33, + "learning_rate": 8.559518723631894e-06, + "loss": 2.6591, + "step": 1558 + }, + { + "epoch": 0.33, + "learning_rate": 8.556933702358774e-06, + "loss": 2.646, + "step": 1559 + }, + { + "epoch": 0.33, + "learning_rate": 8.554346754788928e-06, + "loss": 2.7007, + "step": 1560 + }, + { + "epoch": 0.33, + "learning_rate": 8.551757882323344e-06, + "loss": 2.7317, + "step": 1561 + }, + { + "epoch": 0.33, + "learning_rate": 8.549167086364056e-06, + "loss": 2.6449, + "step": 1562 + }, + { + "epoch": 0.33, + "learning_rate": 8.546574368314138e-06, + "loss": 2.6734, + "step": 1563 + }, + { + "epoch": 0.33, + "learning_rate": 8.543979729577705e-06, + "loss": 2.6109, + "step": 1564 + }, + { + "epoch": 0.33, + "learning_rate": 8.541383171559911e-06, + "loss": 2.6372, + "step": 1565 + }, + { + "epoch": 0.33, + "learning_rate": 8.538784695666955e-06, + "loss": 2.6291, + "step": 1566 + }, + { + "epoch": 0.33, + "learning_rate": 8.536184303306065e-06, + "loss": 2.5846, + "step": 1567 + }, + { + "epoch": 0.33, + "learning_rate": 8.533581995885515e-06, + "loss": 2.6164, + "step": 1568 + }, + { + "epoch": 0.33, + "learning_rate": 8.530977774814616e-06, + "loss": 2.6091, + "step": 1569 + }, + { + "epoch": 0.33, + "learning_rate": 8.528371641503706e-06, + "loss": 2.6299, + "step": 1570 + }, + { + "epoch": 0.33, + "learning_rate": 8.525763597364171e-06, + "loss": 2.5869, + "step": 1571 + }, + { + "epoch": 0.33, + "learning_rate": 8.523153643808421e-06, + "loss": 2.7108, + "step": 1572 + }, + { + "epoch": 0.33, + "learning_rate": 8.520541782249912e-06, + "loss": 2.6098, + "step": 1573 + }, + { + "epoch": 0.33, + "learning_rate": 8.51792801410312e-06, + "loss": 2.697, + "step": 1574 + }, + { + "epoch": 0.33, + "learning_rate": 8.515312340783563e-06, + "loss": 2.6231, + "step": 1575 + }, + { + "epoch": 0.33, + "learning_rate": 8.512694763707788e-06, + "loss": 2.6168, + "step": 1576 + }, + { + "epoch": 0.33, + "learning_rate": 8.510075284293371e-06, + "loss": 2.6199, + "step": 1577 + }, + { + "epoch": 0.33, + "learning_rate": 8.507453903958924e-06, + "loss": 2.6912, + "step": 1578 + }, + { + "epoch": 0.33, + "learning_rate": 8.504830624124078e-06, + "loss": 2.6519, + "step": 1579 + }, + { + "epoch": 0.33, + "learning_rate": 8.502205446209506e-06, + "loss": 2.652, + "step": 1580 + }, + { + "epoch": 0.33, + "learning_rate": 8.499578371636895e-06, + "loss": 2.6552, + "step": 1581 + }, + { + "epoch": 0.33, + "learning_rate": 8.496949401828973e-06, + "loss": 2.6187, + "step": 1582 + }, + { + "epoch": 0.33, + "learning_rate": 8.494318538209485e-06, + "loss": 2.6132, + "step": 1583 + }, + { + "epoch": 0.33, + "learning_rate": 8.491685782203205e-06, + "loss": 2.6047, + "step": 1584 + }, + { + "epoch": 0.33, + "learning_rate": 8.489051135235928e-06, + "loss": 2.6665, + "step": 1585 + }, + { + "epoch": 0.33, + "learning_rate": 8.486414598734479e-06, + "loss": 2.7044, + "step": 1586 + }, + { + "epoch": 0.33, + "learning_rate": 8.483776174126702e-06, + "loss": 2.6687, + "step": 1587 + }, + { + "epoch": 0.33, + "learning_rate": 8.481135862841465e-06, + "loss": 2.6885, + "step": 1588 + }, + { + "epoch": 0.33, + "learning_rate": 8.47849366630866e-06, + "loss": 2.674, + "step": 1589 + }, + { + "epoch": 0.33, + "learning_rate": 8.475849585959195e-06, + "loss": 2.6467, + "step": 1590 + }, + { + "epoch": 0.33, + "learning_rate": 8.473203623225003e-06, + "loss": 2.6703, + "step": 1591 + }, + { + "epoch": 0.33, + "learning_rate": 8.470555779539034e-06, + "loss": 2.6393, + "step": 1592 + }, + { + "epoch": 0.33, + "learning_rate": 8.467906056335254e-06, + "loss": 2.6559, + "step": 1593 + }, + { + "epoch": 0.33, + "learning_rate": 8.465254455048653e-06, + "loss": 2.6695, + "step": 1594 + }, + { + "epoch": 0.33, + "learning_rate": 8.462600977115237e-06, + "loss": 2.6328, + "step": 1595 + }, + { + "epoch": 0.33, + "learning_rate": 8.459945623972021e-06, + "loss": 2.6207, + "step": 1596 + }, + { + "epoch": 0.33, + "learning_rate": 8.457288397057043e-06, + "loss": 2.65, + "step": 1597 + }, + { + "epoch": 0.34, + "learning_rate": 8.454629297809355e-06, + "loss": 2.6665, + "step": 1598 + }, + { + "epoch": 0.34, + "learning_rate": 8.45196832766902e-06, + "loss": 2.6635, + "step": 1599 + }, + { + "epoch": 0.34, + "learning_rate": 8.449305488077116e-06, + "loss": 2.5813, + "step": 1600 + }, + { + "epoch": 0.34, + "learning_rate": 8.446640780475735e-06, + "loss": 2.6691, + "step": 1601 + }, + { + "epoch": 0.34, + "learning_rate": 8.443974206307974e-06, + "loss": 2.6457, + "step": 1602 + }, + { + "epoch": 0.34, + "learning_rate": 8.441305767017951e-06, + "loss": 2.7296, + "step": 1603 + }, + { + "epoch": 0.34, + "learning_rate": 8.438635464050786e-06, + "loss": 2.6331, + "step": 1604 + }, + { + "epoch": 0.34, + "learning_rate": 8.435963298852613e-06, + "loss": 2.7468, + "step": 1605 + }, + { + "epoch": 0.34, + "learning_rate": 8.43328927287057e-06, + "loss": 2.6309, + "step": 1606 + }, + { + "epoch": 0.34, + "learning_rate": 8.430613387552809e-06, + "loss": 2.6562, + "step": 1607 + }, + { + "epoch": 0.34, + "learning_rate": 8.42793564434848e-06, + "loss": 2.6135, + "step": 1608 + }, + { + "epoch": 0.34, + "learning_rate": 8.425256044707748e-06, + "loss": 2.6831, + "step": 1609 + }, + { + "epoch": 0.34, + "learning_rate": 8.42257459008178e-06, + "loss": 2.631, + "step": 1610 + }, + { + "epoch": 0.34, + "learning_rate": 8.419891281922745e-06, + "loss": 2.6193, + "step": 1611 + }, + { + "epoch": 0.34, + "learning_rate": 8.41720612168382e-06, + "loss": 2.5947, + "step": 1612 + }, + { + "epoch": 0.34, + "learning_rate": 8.414519110819183e-06, + "loss": 2.6474, + "step": 1613 + }, + { + "epoch": 0.34, + "learning_rate": 8.411830250784014e-06, + "loss": 2.6732, + "step": 1614 + }, + { + "epoch": 0.34, + "learning_rate": 8.409139543034495e-06, + "loss": 2.6453, + "step": 1615 + }, + { + "epoch": 0.34, + "learning_rate": 8.40644698902781e-06, + "loss": 2.679, + "step": 1616 + }, + { + "epoch": 0.34, + "learning_rate": 8.403752590222139e-06, + "loss": 2.7627, + "step": 1617 + }, + { + "epoch": 0.34, + "learning_rate": 8.401056348076663e-06, + "loss": 2.6901, + "step": 1618 + }, + { + "epoch": 0.34, + "learning_rate": 8.398358264051563e-06, + "loss": 2.6828, + "step": 1619 + }, + { + "epoch": 0.34, + "learning_rate": 8.395658339608018e-06, + "loss": 2.7157, + "step": 1620 + }, + { + "epoch": 0.34, + "learning_rate": 8.392956576208198e-06, + "loss": 2.7009, + "step": 1621 + }, + { + "epoch": 0.34, + "learning_rate": 8.390252975315276e-06, + "loss": 2.6404, + "step": 1622 + }, + { + "epoch": 0.34, + "learning_rate": 8.387547538393417e-06, + "loss": 2.6819, + "step": 1623 + }, + { + "epoch": 0.34, + "learning_rate": 8.384840266907777e-06, + "loss": 2.686, + "step": 1624 + }, + { + "epoch": 0.34, + "learning_rate": 8.382131162324512e-06, + "loss": 2.6775, + "step": 1625 + }, + { + "epoch": 0.34, + "learning_rate": 8.379420226110766e-06, + "loss": 2.6431, + "step": 1626 + }, + { + "epoch": 0.34, + "learning_rate": 8.376707459734677e-06, + "loss": 2.6778, + "step": 1627 + }, + { + "epoch": 0.34, + "learning_rate": 8.373992864665374e-06, + "loss": 2.6834, + "step": 1628 + }, + { + "epoch": 0.34, + "learning_rate": 8.371276442372974e-06, + "loss": 2.6343, + "step": 1629 + }, + { + "epoch": 0.34, + "learning_rate": 8.368558194328587e-06, + "loss": 2.6474, + "step": 1630 + }, + { + "epoch": 0.34, + "learning_rate": 8.365838122004311e-06, + "loss": 2.6161, + "step": 1631 + }, + { + "epoch": 0.34, + "learning_rate": 8.36311622687323e-06, + "loss": 2.6513, + "step": 1632 + }, + { + "epoch": 0.34, + "learning_rate": 8.360392510409415e-06, + "loss": 2.6357, + "step": 1633 + }, + { + "epoch": 0.34, + "learning_rate": 8.357666974087928e-06, + "loss": 2.6513, + "step": 1634 + }, + { + "epoch": 0.34, + "learning_rate": 8.354939619384811e-06, + "loss": 2.5852, + "step": 1635 + }, + { + "epoch": 0.34, + "learning_rate": 8.352210447777093e-06, + "loss": 2.6419, + "step": 1636 + }, + { + "epoch": 0.34, + "learning_rate": 8.349479460742788e-06, + "loss": 2.5738, + "step": 1637 + }, + { + "epoch": 0.34, + "learning_rate": 8.346746659760893e-06, + "loss": 2.6293, + "step": 1638 + }, + { + "epoch": 0.34, + "learning_rate": 8.344012046311385e-06, + "loss": 2.5897, + "step": 1639 + }, + { + "epoch": 0.34, + "learning_rate": 8.341275621875224e-06, + "loss": 2.6062, + "step": 1640 + }, + { + "epoch": 0.34, + "learning_rate": 8.338537387934352e-06, + "loss": 2.5857, + "step": 1641 + }, + { + "epoch": 0.34, + "learning_rate": 8.33579734597169e-06, + "loss": 2.6536, + "step": 1642 + }, + { + "epoch": 0.34, + "learning_rate": 8.333055497471137e-06, + "loss": 2.6253, + "step": 1643 + }, + { + "epoch": 0.34, + "learning_rate": 8.330311843917571e-06, + "loss": 2.7212, + "step": 1644 + }, + { + "epoch": 0.34, + "learning_rate": 8.327566386796848e-06, + "loss": 2.6284, + "step": 1645 + }, + { + "epoch": 0.35, + "learning_rate": 8.324819127595802e-06, + "loss": 2.6158, + "step": 1646 + }, + { + "epoch": 0.35, + "learning_rate": 8.322070067802239e-06, + "loss": 2.6637, + "step": 1647 + }, + { + "epoch": 0.35, + "learning_rate": 8.319319208904946e-06, + "loss": 2.718, + "step": 1648 + }, + { + "epoch": 0.35, + "learning_rate": 8.31656655239368e-06, + "loss": 2.6864, + "step": 1649 + }, + { + "epoch": 0.35, + "learning_rate": 8.313812099759171e-06, + "loss": 2.6434, + "step": 1650 + }, + { + "epoch": 0.35, + "learning_rate": 8.311055852493124e-06, + "loss": 2.5855, + "step": 1651 + }, + { + "epoch": 0.35, + "learning_rate": 8.308297812088215e-06, + "loss": 2.6562, + "step": 1652 + }, + { + "epoch": 0.35, + "learning_rate": 8.305537980038092e-06, + "loss": 2.6275, + "step": 1653 + }, + { + "epoch": 0.35, + "learning_rate": 8.302776357837372e-06, + "loss": 2.6475, + "step": 1654 + }, + { + "epoch": 0.35, + "learning_rate": 8.30001294698164e-06, + "loss": 2.6673, + "step": 1655 + }, + { + "epoch": 0.35, + "learning_rate": 8.297247748967456e-06, + "loss": 2.6711, + "step": 1656 + }, + { + "epoch": 0.35, + "learning_rate": 8.29448076529234e-06, + "loss": 2.6336, + "step": 1657 + }, + { + "epoch": 0.35, + "learning_rate": 8.291711997454786e-06, + "loss": 2.5635, + "step": 1658 + }, + { + "epoch": 0.35, + "learning_rate": 8.288941446954246e-06, + "loss": 2.6879, + "step": 1659 + }, + { + "epoch": 0.35, + "learning_rate": 8.286169115291148e-06, + "loss": 2.6357, + "step": 1660 + }, + { + "epoch": 0.35, + "learning_rate": 8.283395003966873e-06, + "loss": 2.6986, + "step": 1661 + }, + { + "epoch": 0.35, + "learning_rate": 8.280619114483774e-06, + "loss": 2.6122, + "step": 1662 + }, + { + "epoch": 0.35, + "learning_rate": 8.277841448345167e-06, + "loss": 2.6422, + "step": 1663 + }, + { + "epoch": 0.35, + "learning_rate": 8.275062007055323e-06, + "loss": 2.6492, + "step": 1664 + }, + { + "epoch": 0.35, + "learning_rate": 8.272280792119482e-06, + "loss": 2.6545, + "step": 1665 + }, + { + "epoch": 0.35, + "learning_rate": 8.269497805043845e-06, + "loss": 2.6241, + "step": 1666 + }, + { + "epoch": 0.35, + "learning_rate": 8.266713047335563e-06, + "loss": 2.6745, + "step": 1667 + }, + { + "epoch": 0.35, + "learning_rate": 8.263926520502754e-06, + "loss": 2.6967, + "step": 1668 + }, + { + "epoch": 0.35, + "learning_rate": 8.261138226054495e-06, + "loss": 2.6978, + "step": 1669 + }, + { + "epoch": 0.35, + "learning_rate": 8.258348165500815e-06, + "loss": 2.684, + "step": 1670 + }, + { + "epoch": 0.35, + "learning_rate": 8.255556340352706e-06, + "loss": 2.5979, + "step": 1671 + }, + { + "epoch": 0.35, + "learning_rate": 8.252762752122107e-06, + "loss": 2.5855, + "step": 1672 + }, + { + "epoch": 0.35, + "learning_rate": 8.249967402321919e-06, + "loss": 2.5881, + "step": 1673 + }, + { + "epoch": 0.35, + "learning_rate": 8.247170292465993e-06, + "loss": 2.5937, + "step": 1674 + }, + { + "epoch": 0.35, + "learning_rate": 8.244371424069138e-06, + "loss": 2.6902, + "step": 1675 + }, + { + "epoch": 0.35, + "learning_rate": 8.241570798647107e-06, + "loss": 2.6854, + "step": 1676 + }, + { + "epoch": 0.35, + "learning_rate": 8.238768417716614e-06, + "loss": 2.6546, + "step": 1677 + }, + { + "epoch": 0.35, + "learning_rate": 8.23596428279532e-06, + "loss": 2.6503, + "step": 1678 + }, + { + "epoch": 0.35, + "learning_rate": 8.23315839540183e-06, + "loss": 2.7201, + "step": 1679 + }, + { + "epoch": 0.35, + "learning_rate": 8.230350757055707e-06, + "loss": 2.6087, + "step": 1680 + }, + { + "epoch": 0.35, + "learning_rate": 8.227541369277458e-06, + "loss": 2.6089, + "step": 1681 + }, + { + "epoch": 0.35, + "learning_rate": 8.224730233588539e-06, + "loss": 2.599, + "step": 1682 + }, + { + "epoch": 0.35, + "learning_rate": 8.221917351511351e-06, + "loss": 2.6682, + "step": 1683 + }, + { + "epoch": 0.35, + "learning_rate": 8.219102724569238e-06, + "loss": 2.6171, + "step": 1684 + }, + { + "epoch": 0.35, + "learning_rate": 8.216286354286499e-06, + "loss": 2.5556, + "step": 1685 + }, + { + "epoch": 0.35, + "learning_rate": 8.213468242188362e-06, + "loss": 2.6572, + "step": 1686 + }, + { + "epoch": 0.35, + "learning_rate": 8.210648389801015e-06, + "loss": 2.6797, + "step": 1687 + }, + { + "epoch": 0.35, + "learning_rate": 8.207826798651575e-06, + "loss": 2.5582, + "step": 1688 + }, + { + "epoch": 0.35, + "learning_rate": 8.205003470268107e-06, + "loss": 2.6684, + "step": 1689 + }, + { + "epoch": 0.35, + "learning_rate": 8.202178406179617e-06, + "loss": 2.5643, + "step": 1690 + }, + { + "epoch": 0.35, + "learning_rate": 8.199351607916048e-06, + "loss": 2.6851, + "step": 1691 + }, + { + "epoch": 0.35, + "learning_rate": 8.196523077008285e-06, + "loss": 2.6691, + "step": 1692 + }, + { + "epoch": 0.36, + "learning_rate": 8.19369281498815e-06, + "loss": 2.6511, + "step": 1693 + }, + { + "epoch": 0.36, + "learning_rate": 8.190860823388402e-06, + "loss": 2.7126, + "step": 1694 + }, + { + "epoch": 0.36, + "learning_rate": 8.188027103742738e-06, + "loss": 2.6338, + "step": 1695 + }, + { + "epoch": 0.36, + "learning_rate": 8.185191657585788e-06, + "loss": 2.6624, + "step": 1696 + }, + { + "epoch": 0.36, + "learning_rate": 8.182354486453123e-06, + "loss": 2.644, + "step": 1697 + }, + { + "epoch": 0.36, + "learning_rate": 8.179515591881241e-06, + "loss": 2.7187, + "step": 1698 + }, + { + "epoch": 0.36, + "learning_rate": 8.17667497540758e-06, + "loss": 2.6734, + "step": 1699 + }, + { + "epoch": 0.36, + "learning_rate": 8.173832638570503e-06, + "loss": 2.5815, + "step": 1700 + }, + { + "epoch": 0.36, + "learning_rate": 8.170988582909312e-06, + "loss": 2.6715, + "step": 1701 + }, + { + "epoch": 0.36, + "learning_rate": 8.168142809964235e-06, + "loss": 2.6301, + "step": 1702 + }, + { + "epoch": 0.36, + "learning_rate": 8.165295321276433e-06, + "loss": 2.6758, + "step": 1703 + }, + { + "epoch": 0.36, + "learning_rate": 8.162446118387997e-06, + "loss": 2.6073, + "step": 1704 + }, + { + "epoch": 0.36, + "learning_rate": 8.15959520284194e-06, + "loss": 2.6392, + "step": 1705 + }, + { + "epoch": 0.36, + "learning_rate": 8.156742576182208e-06, + "loss": 2.6353, + "step": 1706 + }, + { + "epoch": 0.36, + "learning_rate": 8.153888239953674e-06, + "loss": 2.6358, + "step": 1707 + }, + { + "epoch": 0.36, + "learning_rate": 8.151032195702136e-06, + "loss": 2.5957, + "step": 1708 + }, + { + "epoch": 0.36, + "learning_rate": 8.148174444974313e-06, + "loss": 2.6164, + "step": 1709 + }, + { + "epoch": 0.36, + "learning_rate": 8.145314989317853e-06, + "loss": 2.6701, + "step": 1710 + }, + { + "epoch": 0.36, + "learning_rate": 8.142453830281328e-06, + "loss": 2.7256, + "step": 1711 + }, + { + "epoch": 0.36, + "learning_rate": 8.139590969414224e-06, + "loss": 2.6267, + "step": 1712 + }, + { + "epoch": 0.36, + "learning_rate": 8.136726408266962e-06, + "loss": 2.682, + "step": 1713 + }, + { + "epoch": 0.36, + "learning_rate": 8.133860148390874e-06, + "loss": 2.6716, + "step": 1714 + }, + { + "epoch": 0.36, + "learning_rate": 8.130992191338216e-06, + "loss": 2.7215, + "step": 1715 + }, + { + "epoch": 0.36, + "learning_rate": 8.128122538662159e-06, + "loss": 2.6988, + "step": 1716 + }, + { + "epoch": 0.36, + "learning_rate": 8.125251191916798e-06, + "loss": 2.6229, + "step": 1717 + }, + { + "epoch": 0.36, + "learning_rate": 8.12237815265714e-06, + "loss": 2.583, + "step": 1718 + }, + { + "epoch": 0.36, + "learning_rate": 8.119503422439111e-06, + "loss": 2.6406, + "step": 1719 + }, + { + "epoch": 0.36, + "learning_rate": 8.116627002819554e-06, + "loss": 2.6472, + "step": 1720 + }, + { + "epoch": 0.36, + "learning_rate": 8.113748895356229e-06, + "loss": 2.6783, + "step": 1721 + }, + { + "epoch": 0.36, + "learning_rate": 8.110869101607801e-06, + "loss": 2.634, + "step": 1722 + }, + { + "epoch": 0.36, + "learning_rate": 8.107987623133859e-06, + "loss": 2.6705, + "step": 1723 + }, + { + "epoch": 0.36, + "learning_rate": 8.105104461494896e-06, + "loss": 2.682, + "step": 1724 + }, + { + "epoch": 0.36, + "learning_rate": 8.102219618252323e-06, + "loss": 2.6981, + "step": 1725 + }, + { + "epoch": 0.36, + "learning_rate": 8.099333094968456e-06, + "loss": 2.603, + "step": 1726 + }, + { + "epoch": 0.36, + "learning_rate": 8.096444893206524e-06, + "loss": 2.6551, + "step": 1727 + }, + { + "epoch": 0.36, + "learning_rate": 8.093555014530666e-06, + "loss": 2.6952, + "step": 1728 + }, + { + "epoch": 0.36, + "learning_rate": 8.090663460505927e-06, + "loss": 2.6456, + "step": 1729 + }, + { + "epoch": 0.36, + "learning_rate": 8.08777023269826e-06, + "loss": 2.683, + "step": 1730 + }, + { + "epoch": 0.36, + "learning_rate": 8.084875332674524e-06, + "loss": 2.6782, + "step": 1731 + }, + { + "epoch": 0.36, + "learning_rate": 8.081978762002484e-06, + "loss": 2.5897, + "step": 1732 + }, + { + "epoch": 0.36, + "learning_rate": 8.079080522250812e-06, + "loss": 2.6281, + "step": 1733 + }, + { + "epoch": 0.36, + "learning_rate": 8.07618061498908e-06, + "loss": 2.6428, + "step": 1734 + }, + { + "epoch": 0.36, + "learning_rate": 8.073279041787764e-06, + "loss": 2.6157, + "step": 1735 + }, + { + "epoch": 0.36, + "learning_rate": 8.070375804218244e-06, + "loss": 2.6251, + "step": 1736 + }, + { + "epoch": 0.36, + "learning_rate": 8.067470903852802e-06, + "loss": 2.7047, + "step": 1737 + }, + { + "epoch": 0.36, + "learning_rate": 8.064564342264615e-06, + "loss": 2.6423, + "step": 1738 + }, + { + "epoch": 0.36, + "learning_rate": 8.061656121027766e-06, + "loss": 2.6691, + "step": 1739 + }, + { + "epoch": 0.36, + "learning_rate": 8.058746241717235e-06, + "loss": 2.6962, + "step": 1740 + }, + { + "epoch": 0.37, + "learning_rate": 8.055834705908898e-06, + "loss": 2.5743, + "step": 1741 + }, + { + "epoch": 0.37, + "learning_rate": 8.052921515179528e-06, + "loss": 2.6897, + "step": 1742 + }, + { + "epoch": 0.37, + "learning_rate": 8.0500066711068e-06, + "loss": 2.6544, + "step": 1743 + }, + { + "epoch": 0.37, + "learning_rate": 8.047090175269276e-06, + "loss": 2.6405, + "step": 1744 + }, + { + "epoch": 0.37, + "learning_rate": 8.044172029246418e-06, + "loss": 2.574, + "step": 1745 + }, + { + "epoch": 0.37, + "learning_rate": 8.041252234618581e-06, + "loss": 2.6609, + "step": 1746 + }, + { + "epoch": 0.37, + "learning_rate": 8.03833079296701e-06, + "loss": 2.5913, + "step": 1747 + }, + { + "epoch": 0.37, + "learning_rate": 8.035407705873843e-06, + "loss": 2.6693, + "step": 1748 + }, + { + "epoch": 0.37, + "learning_rate": 8.032482974922116e-06, + "loss": 2.6389, + "step": 1749 + }, + { + "epoch": 0.37, + "learning_rate": 8.029556601695742e-06, + "loss": 2.6568, + "step": 1750 + }, + { + "epoch": 0.37, + "learning_rate": 8.026628587779537e-06, + "loss": 2.6245, + "step": 1751 + }, + { + "epoch": 0.37, + "learning_rate": 8.023698934759196e-06, + "loss": 2.6412, + "step": 1752 + }, + { + "epoch": 0.37, + "learning_rate": 8.020767644221304e-06, + "loss": 2.6666, + "step": 1753 + }, + { + "epoch": 0.37, + "learning_rate": 8.017834717753337e-06, + "loss": 2.6351, + "step": 1754 + }, + { + "epoch": 0.37, + "learning_rate": 8.01490015694365e-06, + "loss": 2.6588, + "step": 1755 + }, + { + "epoch": 0.37, + "learning_rate": 8.011963963381489e-06, + "loss": 2.6346, + "step": 1756 + }, + { + "epoch": 0.37, + "learning_rate": 8.009026138656983e-06, + "loss": 2.5969, + "step": 1757 + }, + { + "epoch": 0.37, + "learning_rate": 8.00608668436114e-06, + "loss": 2.6748, + "step": 1758 + }, + { + "epoch": 0.37, + "learning_rate": 8.003145602085858e-06, + "loss": 2.6614, + "step": 1759 + }, + { + "epoch": 0.37, + "learning_rate": 8.00020289342391e-06, + "loss": 2.6322, + "step": 1760 + }, + { + "epoch": 0.37, + "learning_rate": 7.997258559968954e-06, + "loss": 2.6854, + "step": 1761 + }, + { + "epoch": 0.37, + "learning_rate": 7.994312603315526e-06, + "loss": 2.7024, + "step": 1762 + }, + { + "epoch": 0.37, + "learning_rate": 7.99136502505904e-06, + "loss": 2.6542, + "step": 1763 + }, + { + "epoch": 0.37, + "learning_rate": 7.98841582679579e-06, + "loss": 2.6512, + "step": 1764 + }, + { + "epoch": 0.37, + "learning_rate": 7.985465010122946e-06, + "loss": 2.5305, + "step": 1765 + }, + { + "epoch": 0.37, + "learning_rate": 7.982512576638556e-06, + "loss": 2.6287, + "step": 1766 + }, + { + "epoch": 0.37, + "learning_rate": 7.979558527941547e-06, + "loss": 2.6994, + "step": 1767 + }, + { + "epoch": 0.37, + "learning_rate": 7.97660286563171e-06, + "loss": 2.6897, + "step": 1768 + }, + { + "epoch": 0.37, + "learning_rate": 7.973645591309722e-06, + "loss": 2.6387, + "step": 1769 + }, + { + "epoch": 0.37, + "learning_rate": 7.970686706577124e-06, + "loss": 2.6664, + "step": 1770 + }, + { + "epoch": 0.37, + "learning_rate": 7.967726213036336e-06, + "loss": 2.7064, + "step": 1771 + }, + { + "epoch": 0.37, + "learning_rate": 7.964764112290641e-06, + "loss": 2.6519, + "step": 1772 + }, + { + "epoch": 0.37, + "learning_rate": 7.961800405944201e-06, + "loss": 2.6391, + "step": 1773 + }, + { + "epoch": 0.37, + "learning_rate": 7.958835095602044e-06, + "loss": 2.6788, + "step": 1774 + }, + { + "epoch": 0.37, + "learning_rate": 7.955868182870067e-06, + "loss": 2.6697, + "step": 1775 + }, + { + "epoch": 0.37, + "learning_rate": 7.95289966935503e-06, + "loss": 2.6066, + "step": 1776 + }, + { + "epoch": 0.37, + "learning_rate": 7.94992955666457e-06, + "loss": 2.6583, + "step": 1777 + }, + { + "epoch": 0.37, + "learning_rate": 7.946957846407182e-06, + "loss": 2.6852, + "step": 1778 + }, + { + "epoch": 0.37, + "learning_rate": 7.943984540192228e-06, + "loss": 2.6829, + "step": 1779 + }, + { + "epoch": 0.37, + "learning_rate": 7.941009639629934e-06, + "loss": 2.6539, + "step": 1780 + }, + { + "epoch": 0.37, + "learning_rate": 7.938033146331392e-06, + "loss": 2.6594, + "step": 1781 + }, + { + "epoch": 0.37, + "learning_rate": 7.935055061908554e-06, + "loss": 2.6255, + "step": 1782 + }, + { + "epoch": 0.37, + "learning_rate": 7.932075387974236e-06, + "loss": 2.6879, + "step": 1783 + }, + { + "epoch": 0.37, + "learning_rate": 7.92909412614211e-06, + "loss": 2.6037, + "step": 1784 + }, + { + "epoch": 0.37, + "learning_rate": 7.926111278026714e-06, + "loss": 2.6352, + "step": 1785 + }, + { + "epoch": 0.37, + "learning_rate": 7.923126845243443e-06, + "loss": 2.6731, + "step": 1786 + }, + { + "epoch": 0.37, + "learning_rate": 7.920140829408546e-06, + "loss": 2.6162, + "step": 1787 + }, + { + "epoch": 0.37, + "learning_rate": 7.917153232139136e-06, + "loss": 2.5905, + "step": 1788 + }, + { + "epoch": 0.38, + "learning_rate": 7.91416405505318e-06, + "loss": 2.6826, + "step": 1789 + }, + { + "epoch": 0.38, + "learning_rate": 7.911173299769494e-06, + "loss": 2.6801, + "step": 1790 + }, + { + "epoch": 0.38, + "learning_rate": 7.908180967907761e-06, + "loss": 2.7062, + "step": 1791 + }, + { + "epoch": 0.38, + "learning_rate": 7.905187061088509e-06, + "loss": 2.6398, + "step": 1792 + }, + { + "epoch": 0.38, + "learning_rate": 7.902191580933123e-06, + "loss": 2.6785, + "step": 1793 + }, + { + "epoch": 0.38, + "learning_rate": 7.899194529063835e-06, + "loss": 2.6497, + "step": 1794 + }, + { + "epoch": 0.38, + "learning_rate": 7.896195907103733e-06, + "loss": 2.7048, + "step": 1795 + }, + { + "epoch": 0.38, + "learning_rate": 7.893195716676754e-06, + "loss": 2.6586, + "step": 1796 + }, + { + "epoch": 0.38, + "learning_rate": 7.890193959407686e-06, + "loss": 2.6214, + "step": 1797 + }, + { + "epoch": 0.38, + "learning_rate": 7.887190636922161e-06, + "loss": 2.7036, + "step": 1798 + }, + { + "epoch": 0.38, + "learning_rate": 7.884185750846663e-06, + "loss": 2.5866, + "step": 1799 + }, + { + "epoch": 0.38, + "learning_rate": 7.88117930280852e-06, + "loss": 2.6593, + "step": 1800 + }, + { + "epoch": 0.38, + "learning_rate": 7.878171294435909e-06, + "loss": 2.6789, + "step": 1801 + }, + { + "epoch": 0.38, + "learning_rate": 7.875161727357848e-06, + "loss": 2.6004, + "step": 1802 + }, + { + "epoch": 0.38, + "learning_rate": 7.872150603204202e-06, + "loss": 2.6237, + "step": 1803 + }, + { + "epoch": 0.38, + "learning_rate": 7.869137923605681e-06, + "loss": 2.6441, + "step": 1804 + }, + { + "epoch": 0.38, + "learning_rate": 7.866123690193832e-06, + "loss": 2.6354, + "step": 1805 + }, + { + "epoch": 0.38, + "learning_rate": 7.863107904601049e-06, + "loss": 2.6523, + "step": 1806 + }, + { + "epoch": 0.38, + "learning_rate": 7.860090568460559e-06, + "loss": 2.5684, + "step": 1807 + }, + { + "epoch": 0.38, + "learning_rate": 7.857071683406438e-06, + "loss": 2.6081, + "step": 1808 + }, + { + "epoch": 0.38, + "learning_rate": 7.854051251073597e-06, + "loss": 2.6807, + "step": 1809 + }, + { + "epoch": 0.38, + "learning_rate": 7.85102927309778e-06, + "loss": 2.6378, + "step": 1810 + }, + { + "epoch": 0.38, + "learning_rate": 7.848005751115579e-06, + "loss": 2.6458, + "step": 1811 + }, + { + "epoch": 0.38, + "learning_rate": 7.844980686764412e-06, + "loss": 2.7181, + "step": 1812 + }, + { + "epoch": 0.38, + "learning_rate": 7.841954081682534e-06, + "loss": 2.6311, + "step": 1813 + }, + { + "epoch": 0.38, + "learning_rate": 7.838925937509038e-06, + "loss": 2.7088, + "step": 1814 + }, + { + "epoch": 0.38, + "learning_rate": 7.835896255883854e-06, + "loss": 2.6965, + "step": 1815 + }, + { + "epoch": 0.38, + "learning_rate": 7.832865038447732e-06, + "loss": 2.6406, + "step": 1816 + }, + { + "epoch": 0.38, + "learning_rate": 7.829832286842265e-06, + "loss": 2.6988, + "step": 1817 + }, + { + "epoch": 0.38, + "learning_rate": 7.826798002709872e-06, + "loss": 2.6359, + "step": 1818 + }, + { + "epoch": 0.38, + "learning_rate": 7.823762187693804e-06, + "loss": 2.6956, + "step": 1819 + }, + { + "epoch": 0.38, + "learning_rate": 7.82072484343814e-06, + "loss": 2.6806, + "step": 1820 + }, + { + "epoch": 0.38, + "learning_rate": 7.817685971587786e-06, + "loss": 2.6568, + "step": 1821 + }, + { + "epoch": 0.38, + "learning_rate": 7.814645573788478e-06, + "loss": 2.697, + "step": 1822 + }, + { + "epoch": 0.38, + "learning_rate": 7.811603651686777e-06, + "loss": 2.6411, + "step": 1823 + }, + { + "epoch": 0.38, + "learning_rate": 7.80856020693007e-06, + "loss": 2.6586, + "step": 1824 + }, + { + "epoch": 0.38, + "learning_rate": 7.805515241166566e-06, + "loss": 2.6687, + "step": 1825 + }, + { + "epoch": 0.38, + "learning_rate": 7.802468756045301e-06, + "loss": 2.6537, + "step": 1826 + }, + { + "epoch": 0.38, + "learning_rate": 7.799420753216134e-06, + "loss": 2.6845, + "step": 1827 + }, + { + "epoch": 0.38, + "learning_rate": 7.796371234329743e-06, + "loss": 2.6234, + "step": 1828 + }, + { + "epoch": 0.38, + "learning_rate": 7.793320201037629e-06, + "loss": 2.6013, + "step": 1829 + }, + { + "epoch": 0.38, + "learning_rate": 7.790267654992115e-06, + "loss": 2.6535, + "step": 1830 + }, + { + "epoch": 0.38, + "learning_rate": 7.787213597846335e-06, + "loss": 2.5854, + "step": 1831 + }, + { + "epoch": 0.38, + "learning_rate": 7.784158031254251e-06, + "loss": 2.6471, + "step": 1832 + }, + { + "epoch": 0.38, + "learning_rate": 7.781100956870643e-06, + "loss": 2.6191, + "step": 1833 + }, + { + "epoch": 0.38, + "learning_rate": 7.778042376351096e-06, + "loss": 2.6369, + "step": 1834 + }, + { + "epoch": 0.38, + "learning_rate": 7.774982291352022e-06, + "loss": 2.6458, + "step": 1835 + }, + { + "epoch": 0.38, + "learning_rate": 7.771920703530647e-06, + "loss": 2.6632, + "step": 1836 + }, + { + "epoch": 0.39, + "learning_rate": 7.768857614545e-06, + "loss": 2.6153, + "step": 1837 + }, + { + "epoch": 0.39, + "learning_rate": 7.765793026053934e-06, + "loss": 2.6439, + "step": 1838 + }, + { + "epoch": 0.39, + "learning_rate": 7.762726939717114e-06, + "loss": 2.67, + "step": 1839 + }, + { + "epoch": 0.39, + "learning_rate": 7.75965935719501e-06, + "loss": 2.678, + "step": 1840 + }, + { + "epoch": 0.39, + "learning_rate": 7.756590280148904e-06, + "loss": 2.6678, + "step": 1841 + }, + { + "epoch": 0.39, + "learning_rate": 7.75351971024089e-06, + "loss": 2.6065, + "step": 1842 + }, + { + "epoch": 0.39, + "learning_rate": 7.750447649133872e-06, + "loss": 2.6704, + "step": 1843 + }, + { + "epoch": 0.39, + "learning_rate": 7.747374098491553e-06, + "loss": 2.5968, + "step": 1844 + }, + { + "epoch": 0.39, + "learning_rate": 7.744299059978452e-06, + "loss": 2.6094, + "step": 1845 + }, + { + "epoch": 0.39, + "learning_rate": 7.741222535259889e-06, + "loss": 2.6952, + "step": 1846 + }, + { + "epoch": 0.39, + "learning_rate": 7.73814452600199e-06, + "loss": 2.5886, + "step": 1847 + }, + { + "epoch": 0.39, + "learning_rate": 7.735065033871686e-06, + "loss": 2.6475, + "step": 1848 + }, + { + "epoch": 0.39, + "learning_rate": 7.731984060536707e-06, + "loss": 2.721, + "step": 1849 + }, + { + "epoch": 0.39, + "learning_rate": 7.728901607665591e-06, + "loss": 2.5898, + "step": 1850 + }, + { + "epoch": 0.39, + "learning_rate": 7.725817676927674e-06, + "loss": 2.5926, + "step": 1851 + }, + { + "epoch": 0.39, + "learning_rate": 7.722732269993093e-06, + "loss": 2.6803, + "step": 1852 + }, + { + "epoch": 0.39, + "learning_rate": 7.719645388532779e-06, + "loss": 2.6582, + "step": 1853 + }, + { + "epoch": 0.39, + "learning_rate": 7.716557034218475e-06, + "loss": 2.5928, + "step": 1854 + }, + { + "epoch": 0.39, + "learning_rate": 7.71346720872271e-06, + "loss": 2.6383, + "step": 1855 + }, + { + "epoch": 0.39, + "learning_rate": 7.71037591371881e-06, + "loss": 2.5785, + "step": 1856 + }, + { + "epoch": 0.39, + "learning_rate": 7.707283150880903e-06, + "loss": 2.6269, + "step": 1857 + }, + { + "epoch": 0.39, + "learning_rate": 7.704188921883911e-06, + "loss": 2.6241, + "step": 1858 + }, + { + "epoch": 0.39, + "learning_rate": 7.701093228403543e-06, + "loss": 2.6292, + "step": 1859 + }, + { + "epoch": 0.39, + "learning_rate": 7.697996072116308e-06, + "loss": 2.6398, + "step": 1860 + }, + { + "epoch": 0.39, + "learning_rate": 7.694897454699505e-06, + "loss": 2.6482, + "step": 1861 + }, + { + "epoch": 0.39, + "learning_rate": 7.691797377831226e-06, + "loss": 2.6552, + "step": 1862 + }, + { + "epoch": 0.39, + "learning_rate": 7.688695843190353e-06, + "loss": 2.6708, + "step": 1863 + }, + { + "epoch": 0.39, + "learning_rate": 7.685592852456553e-06, + "loss": 2.7286, + "step": 1864 + }, + { + "epoch": 0.39, + "learning_rate": 7.682488407310284e-06, + "loss": 2.6258, + "step": 1865 + }, + { + "epoch": 0.39, + "learning_rate": 7.679382509432798e-06, + "loss": 2.6461, + "step": 1866 + }, + { + "epoch": 0.39, + "learning_rate": 7.676275160506121e-06, + "loss": 2.5679, + "step": 1867 + }, + { + "epoch": 0.39, + "learning_rate": 7.673166362213077e-06, + "loss": 2.7018, + "step": 1868 + }, + { + "epoch": 0.39, + "learning_rate": 7.670056116237272e-06, + "loss": 2.6518, + "step": 1869 + }, + { + "epoch": 0.39, + "learning_rate": 7.666944424263089e-06, + "loss": 2.6998, + "step": 1870 + }, + { + "epoch": 0.39, + "learning_rate": 7.663831287975702e-06, + "loss": 2.6266, + "step": 1871 + }, + { + "epoch": 0.39, + "learning_rate": 7.660716709061065e-06, + "loss": 2.6554, + "step": 1872 + }, + { + "epoch": 0.39, + "learning_rate": 7.657600689205909e-06, + "loss": 2.6514, + "step": 1873 + }, + { + "epoch": 0.39, + "learning_rate": 7.654483230097752e-06, + "loss": 2.616, + "step": 1874 + }, + { + "epoch": 0.39, + "learning_rate": 7.651364333424886e-06, + "loss": 2.6222, + "step": 1875 + }, + { + "epoch": 0.39, + "learning_rate": 7.648244000876389e-06, + "loss": 2.6193, + "step": 1876 + }, + { + "epoch": 0.39, + "learning_rate": 7.645122234142103e-06, + "loss": 2.6707, + "step": 1877 + }, + { + "epoch": 0.39, + "learning_rate": 7.641999034912663e-06, + "loss": 2.62, + "step": 1878 + }, + { + "epoch": 0.39, + "learning_rate": 7.638874404879471e-06, + "loss": 2.6497, + "step": 1879 + }, + { + "epoch": 0.39, + "learning_rate": 7.635748345734702e-06, + "loss": 2.65, + "step": 1880 + }, + { + "epoch": 0.39, + "learning_rate": 7.63262085917131e-06, + "loss": 2.6713, + "step": 1881 + }, + { + "epoch": 0.39, + "learning_rate": 7.629491946883018e-06, + "loss": 2.6455, + "step": 1882 + }, + { + "epoch": 0.39, + "learning_rate": 7.626361610564325e-06, + "loss": 2.6865, + "step": 1883 + }, + { + "epoch": 0.4, + "learning_rate": 7.623229851910498e-06, + "loss": 2.6128, + "step": 1884 + }, + { + "epoch": 0.4, + "learning_rate": 7.620096672617575e-06, + "loss": 2.643, + "step": 1885 + }, + { + "epoch": 0.4, + "learning_rate": 7.616962074382364e-06, + "loss": 2.5977, + "step": 1886 + }, + { + "epoch": 0.4, + "learning_rate": 7.6138260589024434e-06, + "loss": 2.6216, + "step": 1887 + }, + { + "epoch": 0.4, + "learning_rate": 7.610688627876155e-06, + "loss": 2.6685, + "step": 1888 + }, + { + "epoch": 0.4, + "learning_rate": 7.607549783002608e-06, + "loss": 2.5946, + "step": 1889 + }, + { + "epoch": 0.4, + "learning_rate": 7.604409525981684e-06, + "loss": 2.6401, + "step": 1890 + }, + { + "epoch": 0.4, + "learning_rate": 7.6012678585140165e-06, + "loss": 2.6372, + "step": 1891 + }, + { + "epoch": 0.4, + "learning_rate": 7.598124782301015e-06, + "loss": 2.6902, + "step": 1892 + }, + { + "epoch": 0.4, + "learning_rate": 7.594980299044845e-06, + "loss": 2.6863, + "step": 1893 + }, + { + "epoch": 0.4, + "learning_rate": 7.5918344104484396e-06, + "loss": 2.6362, + "step": 1894 + }, + { + "epoch": 0.4, + "learning_rate": 7.588687118215485e-06, + "loss": 2.6646, + "step": 1895 + }, + { + "epoch": 0.4, + "learning_rate": 7.585538424050434e-06, + "loss": 2.6029, + "step": 1896 + }, + { + "epoch": 0.4, + "learning_rate": 7.582388329658498e-06, + "loss": 2.6756, + "step": 1897 + }, + { + "epoch": 0.4, + "learning_rate": 7.579236836745643e-06, + "loss": 2.6851, + "step": 1898 + }, + { + "epoch": 0.4, + "learning_rate": 7.576083947018595e-06, + "loss": 2.6403, + "step": 1899 + }, + { + "epoch": 0.4, + "learning_rate": 7.572929662184839e-06, + "loss": 2.638, + "step": 1900 + }, + { + "epoch": 0.4, + "learning_rate": 7.569773983952611e-06, + "loss": 2.6068, + "step": 1901 + }, + { + "epoch": 0.4, + "learning_rate": 7.566616914030902e-06, + "loss": 2.6581, + "step": 1902 + }, + { + "epoch": 0.4, + "learning_rate": 7.56345845412946e-06, + "loss": 2.6643, + "step": 1903 + }, + { + "epoch": 0.4, + "learning_rate": 7.560298605958782e-06, + "loss": 2.6457, + "step": 1904 + }, + { + "epoch": 0.4, + "learning_rate": 7.557137371230123e-06, + "loss": 2.6484, + "step": 1905 + }, + { + "epoch": 0.4, + "learning_rate": 7.553974751655479e-06, + "loss": 2.6349, + "step": 1906 + }, + { + "epoch": 0.4, + "learning_rate": 7.550810748947605e-06, + "loss": 2.6765, + "step": 1907 + }, + { + "epoch": 0.4, + "learning_rate": 7.547645364820001e-06, + "loss": 2.659, + "step": 1908 + }, + { + "epoch": 0.4, + "learning_rate": 7.544478600986915e-06, + "loss": 2.6897, + "step": 1909 + }, + { + "epoch": 0.4, + "learning_rate": 7.541310459163343e-06, + "loss": 2.6898, + "step": 1910 + }, + { + "epoch": 0.4, + "learning_rate": 7.538140941065027e-06, + "loss": 2.6413, + "step": 1911 + }, + { + "epoch": 0.4, + "learning_rate": 7.534970048408457e-06, + "loss": 2.6472, + "step": 1912 + }, + { + "epoch": 0.4, + "learning_rate": 7.5317977829108605e-06, + "loss": 2.5946, + "step": 1913 + }, + { + "epoch": 0.4, + "learning_rate": 7.528624146290217e-06, + "loss": 2.6628, + "step": 1914 + }, + { + "epoch": 0.4, + "learning_rate": 7.5254491402652415e-06, + "loss": 2.6076, + "step": 1915 + }, + { + "epoch": 0.4, + "learning_rate": 7.522272766555397e-06, + "loss": 2.6624, + "step": 1916 + }, + { + "epoch": 0.4, + "learning_rate": 7.5190950268808805e-06, + "loss": 2.7076, + "step": 1917 + }, + { + "epoch": 0.4, + "learning_rate": 7.515915922962633e-06, + "loss": 2.6247, + "step": 1918 + }, + { + "epoch": 0.4, + "learning_rate": 7.512735456522333e-06, + "loss": 2.6701, + "step": 1919 + }, + { + "epoch": 0.4, + "learning_rate": 7.5095536292824e-06, + "loss": 2.6257, + "step": 1920 + }, + { + "epoch": 0.4, + "learning_rate": 7.506370442965984e-06, + "loss": 2.655, + "step": 1921 + }, + { + "epoch": 0.4, + "learning_rate": 7.503185899296974e-06, + "loss": 2.5987, + "step": 1922 + }, + { + "epoch": 0.4, + "learning_rate": 7.500000000000001e-06, + "loss": 2.6388, + "step": 1923 + }, + { + "epoch": 0.4, + "learning_rate": 7.496812746800418e-06, + "loss": 2.5638, + "step": 1924 + }, + { + "epoch": 0.4, + "learning_rate": 7.4936241414243185e-06, + "loss": 2.6838, + "step": 1925 + }, + { + "epoch": 0.4, + "learning_rate": 7.490434185598527e-06, + "loss": 2.6493, + "step": 1926 + }, + { + "epoch": 0.4, + "learning_rate": 7.487242881050603e-06, + "loss": 2.6041, + "step": 1927 + }, + { + "epoch": 0.4, + "learning_rate": 7.484050229508826e-06, + "loss": 2.68, + "step": 1928 + }, + { + "epoch": 0.4, + "learning_rate": 7.480856232702217e-06, + "loss": 2.6706, + "step": 1929 + }, + { + "epoch": 0.4, + "learning_rate": 7.477660892360519e-06, + "loss": 2.6568, + "step": 1930 + }, + { + "epoch": 0.4, + "learning_rate": 7.474464210214202e-06, + "loss": 2.694, + "step": 1931 + }, + { + "epoch": 0.41, + "learning_rate": 7.471266187994466e-06, + "loss": 2.5751, + "step": 1932 + }, + { + "epoch": 0.41, + "learning_rate": 7.468066827433236e-06, + "loss": 2.6017, + "step": 1933 + }, + { + "epoch": 0.41, + "learning_rate": 7.464866130263159e-06, + "loss": 2.5713, + "step": 1934 + }, + { + "epoch": 0.41, + "learning_rate": 7.461664098217609e-06, + "loss": 2.5901, + "step": 1935 + }, + { + "epoch": 0.41, + "learning_rate": 7.4584607330306825e-06, + "loss": 2.6729, + "step": 1936 + }, + { + "epoch": 0.41, + "learning_rate": 7.4552560364371975e-06, + "loss": 2.6648, + "step": 1937 + }, + { + "epoch": 0.41, + "learning_rate": 7.4520500101726904e-06, + "loss": 2.6409, + "step": 1938 + }, + { + "epoch": 0.41, + "learning_rate": 7.448842655973423e-06, + "loss": 2.6931, + "step": 1939 + }, + { + "epoch": 0.41, + "learning_rate": 7.44563397557637e-06, + "loss": 2.5807, + "step": 1940 + }, + { + "epoch": 0.41, + "learning_rate": 7.442423970719232e-06, + "loss": 2.6118, + "step": 1941 + }, + { + "epoch": 0.41, + "learning_rate": 7.439212643140419e-06, + "loss": 2.6525, + "step": 1942 + }, + { + "epoch": 0.41, + "learning_rate": 7.435999994579062e-06, + "loss": 2.6457, + "step": 1943 + }, + { + "epoch": 0.41, + "learning_rate": 7.432786026775007e-06, + "loss": 2.6524, + "step": 1944 + }, + { + "epoch": 0.41, + "learning_rate": 7.429570741468815e-06, + "loss": 2.5665, + "step": 1945 + }, + { + "epoch": 0.41, + "learning_rate": 7.426354140401756e-06, + "loss": 2.6622, + "step": 1946 + }, + { + "epoch": 0.41, + "learning_rate": 7.423136225315818e-06, + "loss": 2.7331, + "step": 1947 + }, + { + "epoch": 0.41, + "learning_rate": 7.419916997953698e-06, + "loss": 2.6525, + "step": 1948 + }, + { + "epoch": 0.41, + "learning_rate": 7.4166964600588035e-06, + "loss": 2.5979, + "step": 1949 + }, + { + "epoch": 0.41, + "learning_rate": 7.413474613375252e-06, + "loss": 2.6128, + "step": 1950 + }, + { + "epoch": 0.41, + "learning_rate": 7.410251459647873e-06, + "loss": 2.6539, + "step": 1951 + }, + { + "epoch": 0.41, + "learning_rate": 7.4070270006221975e-06, + "loss": 2.6073, + "step": 1952 + }, + { + "epoch": 0.41, + "learning_rate": 7.403801238044469e-06, + "loss": 2.6662, + "step": 1953 + }, + { + "epoch": 0.41, + "learning_rate": 7.4005741736616345e-06, + "loss": 2.6356, + "step": 1954 + }, + { + "epoch": 0.41, + "learning_rate": 7.397345809221346e-06, + "loss": 2.6081, + "step": 1955 + }, + { + "epoch": 0.41, + "learning_rate": 7.39411614647196e-06, + "loss": 2.6054, + "step": 1956 + }, + { + "epoch": 0.41, + "learning_rate": 7.390885187162537e-06, + "loss": 2.6562, + "step": 1957 + }, + { + "epoch": 0.41, + "learning_rate": 7.387652933042835e-06, + "loss": 2.6372, + "step": 1958 + }, + { + "epoch": 0.41, + "learning_rate": 7.384419385863322e-06, + "loss": 2.6711, + "step": 1959 + }, + { + "epoch": 0.41, + "learning_rate": 7.381184547375159e-06, + "loss": 2.6589, + "step": 1960 + }, + { + "epoch": 0.41, + "learning_rate": 7.377948419330206e-06, + "loss": 2.636, + "step": 1961 + }, + { + "epoch": 0.41, + "learning_rate": 7.3747110034810276e-06, + "loss": 2.651, + "step": 1962 + }, + { + "epoch": 0.41, + "learning_rate": 7.371472301580882e-06, + "loss": 2.5825, + "step": 1963 + }, + { + "epoch": 0.41, + "learning_rate": 7.368232315383721e-06, + "loss": 2.6076, + "step": 1964 + }, + { + "epoch": 0.41, + "learning_rate": 7.364991046644198e-06, + "loss": 2.6868, + "step": 1965 + }, + { + "epoch": 0.41, + "learning_rate": 7.3617484971176565e-06, + "loss": 2.6737, + "step": 1966 + }, + { + "epoch": 0.41, + "learning_rate": 7.358504668560134e-06, + "loss": 2.6899, + "step": 1967 + }, + { + "epoch": 0.41, + "learning_rate": 7.355259562728363e-06, + "loss": 2.6705, + "step": 1968 + }, + { + "epoch": 0.41, + "learning_rate": 7.352013181379766e-06, + "loss": 2.6806, + "step": 1969 + }, + { + "epoch": 0.41, + "learning_rate": 7.348765526272457e-06, + "loss": 2.6218, + "step": 1970 + }, + { + "epoch": 0.41, + "learning_rate": 7.345516599165239e-06, + "loss": 2.6307, + "step": 1971 + }, + { + "epoch": 0.41, + "learning_rate": 7.342266401817605e-06, + "loss": 2.645, + "step": 1972 + }, + { + "epoch": 0.41, + "learning_rate": 7.339014935989734e-06, + "loss": 2.6894, + "step": 1973 + }, + { + "epoch": 0.41, + "learning_rate": 7.3357622034424954e-06, + "loss": 2.65, + "step": 1974 + }, + { + "epoch": 0.41, + "learning_rate": 7.332508205937441e-06, + "loss": 2.5951, + "step": 1975 + }, + { + "epoch": 0.41, + "learning_rate": 7.329252945236808e-06, + "loss": 2.6403, + "step": 1976 + }, + { + "epoch": 0.41, + "learning_rate": 7.325996423103521e-06, + "loss": 2.6924, + "step": 1977 + }, + { + "epoch": 0.41, + "learning_rate": 7.322738641301185e-06, + "loss": 2.6466, + "step": 1978 + }, + { + "epoch": 0.41, + "learning_rate": 7.319479601594085e-06, + "loss": 2.6971, + "step": 1979 + }, + { + "epoch": 0.42, + "learning_rate": 7.316219305747193e-06, + "loss": 2.6764, + "step": 1980 + }, + { + "epoch": 0.42, + "learning_rate": 7.31295775552616e-06, + "loss": 2.632, + "step": 1981 + }, + { + "epoch": 0.42, + "learning_rate": 7.309694952697308e-06, + "loss": 2.6122, + "step": 1982 + }, + { + "epoch": 0.42, + "learning_rate": 7.30643089902765e-06, + "loss": 2.647, + "step": 1983 + }, + { + "epoch": 0.42, + "learning_rate": 7.303165596284869e-06, + "loss": 2.6516, + "step": 1984 + }, + { + "epoch": 0.42, + "learning_rate": 7.299899046237323e-06, + "loss": 2.6437, + "step": 1985 + }, + { + "epoch": 0.42, + "learning_rate": 7.296631250654051e-06, + "loss": 2.5967, + "step": 1986 + }, + { + "epoch": 0.42, + "learning_rate": 7.293362211304763e-06, + "loss": 2.6289, + "step": 1987 + }, + { + "epoch": 0.42, + "learning_rate": 7.290091929959843e-06, + "loss": 2.6319, + "step": 1988 + }, + { + "epoch": 0.42, + "learning_rate": 7.286820408390348e-06, + "loss": 2.681, + "step": 1989 + }, + { + "epoch": 0.42, + "learning_rate": 7.283547648368008e-06, + "loss": 2.6594, + "step": 1990 + }, + { + "epoch": 0.42, + "learning_rate": 7.2802736516652205e-06, + "loss": 2.6278, + "step": 1991 + }, + { + "epoch": 0.42, + "learning_rate": 7.276998420055056e-06, + "loss": 2.6954, + "step": 1992 + }, + { + "epoch": 0.42, + "learning_rate": 7.27372195531125e-06, + "loss": 2.6032, + "step": 1993 + }, + { + "epoch": 0.42, + "learning_rate": 7.270444259208211e-06, + "loss": 2.6443, + "step": 1994 + }, + { + "epoch": 0.42, + "learning_rate": 7.267165333521012e-06, + "loss": 2.6955, + "step": 1995 + }, + { + "epoch": 0.42, + "learning_rate": 7.263885180025387e-06, + "loss": 2.6421, + "step": 1996 + }, + { + "epoch": 0.42, + "learning_rate": 7.2606038004977435e-06, + "loss": 2.6961, + "step": 1997 + }, + { + "epoch": 0.42, + "learning_rate": 7.257321196715146e-06, + "loss": 2.6021, + "step": 1998 + }, + { + "epoch": 0.42, + "learning_rate": 7.254037370455329e-06, + "loss": 2.6455, + "step": 1999 + }, + { + "epoch": 0.42, + "learning_rate": 7.250752323496679e-06, + "loss": 2.6036, + "step": 2000 + }, + { + "epoch": 0.42, + "learning_rate": 7.247466057618254e-06, + "loss": 2.6019, + "step": 2001 + }, + { + "epoch": 0.42, + "learning_rate": 7.2441785745997675e-06, + "loss": 2.636, + "step": 2002 + }, + { + "epoch": 0.42, + "learning_rate": 7.240889876221589e-06, + "loss": 2.6352, + "step": 2003 + }, + { + "epoch": 0.42, + "learning_rate": 7.237599964264752e-06, + "loss": 2.7035, + "step": 2004 + }, + { + "epoch": 0.42, + "learning_rate": 7.234308840510944e-06, + "loss": 2.7196, + "step": 2005 + }, + { + "epoch": 0.42, + "learning_rate": 7.23101650674251e-06, + "loss": 2.6471, + "step": 2006 + }, + { + "epoch": 0.42, + "learning_rate": 7.227722964742447e-06, + "loss": 2.5899, + "step": 2007 + }, + { + "epoch": 0.42, + "learning_rate": 7.224428216294413e-06, + "loss": 2.6279, + "step": 2008 + }, + { + "epoch": 0.42, + "learning_rate": 7.221132263182713e-06, + "loss": 2.6873, + "step": 2009 + }, + { + "epoch": 0.42, + "learning_rate": 7.217835107192308e-06, + "loss": 2.706, + "step": 2010 + }, + { + "epoch": 0.42, + "learning_rate": 7.214536750108808e-06, + "loss": 2.6719, + "step": 2011 + }, + { + "epoch": 0.42, + "learning_rate": 7.211237193718476e-06, + "loss": 2.6626, + "step": 2012 + }, + { + "epoch": 0.42, + "learning_rate": 7.207936439808222e-06, + "loss": 2.6181, + "step": 2013 + }, + { + "epoch": 0.42, + "learning_rate": 7.204634490165606e-06, + "loss": 2.6767, + "step": 2014 + }, + { + "epoch": 0.42, + "learning_rate": 7.201331346578836e-06, + "loss": 2.5918, + "step": 2015 + }, + { + "epoch": 0.42, + "learning_rate": 7.198027010836764e-06, + "loss": 2.6741, + "step": 2016 + }, + { + "epoch": 0.42, + "learning_rate": 7.1947214847288925e-06, + "loss": 2.6489, + "step": 2017 + }, + { + "epoch": 0.42, + "learning_rate": 7.191414770045364e-06, + "loss": 2.6604, + "step": 2018 + }, + { + "epoch": 0.42, + "learning_rate": 7.188106868576965e-06, + "loss": 2.671, + "step": 2019 + }, + { + "epoch": 0.42, + "learning_rate": 7.1847977821151304e-06, + "loss": 2.719, + "step": 2020 + }, + { + "epoch": 0.42, + "learning_rate": 7.181487512451927e-06, + "loss": 2.7163, + "step": 2021 + }, + { + "epoch": 0.42, + "learning_rate": 7.178176061380071e-06, + "loss": 2.6898, + "step": 2022 + }, + { + "epoch": 0.42, + "learning_rate": 7.1748634306929155e-06, + "loss": 2.6765, + "step": 2023 + }, + { + "epoch": 0.42, + "learning_rate": 7.17154962218445e-06, + "loss": 2.7247, + "step": 2024 + }, + { + "epoch": 0.42, + "learning_rate": 7.168234637649309e-06, + "loss": 2.6931, + "step": 2025 + }, + { + "epoch": 0.42, + "learning_rate": 7.164918478882753e-06, + "loss": 2.7147, + "step": 2026 + }, + { + "epoch": 0.43, + "learning_rate": 7.161601147680688e-06, + "loss": 2.6686, + "step": 2027 + }, + { + "epoch": 0.43, + "learning_rate": 7.1582826458396495e-06, + "loss": 2.7069, + "step": 2028 + }, + { + "epoch": 0.43, + "learning_rate": 7.15496297515681e-06, + "loss": 2.6391, + "step": 2029 + }, + { + "epoch": 0.43, + "learning_rate": 7.1516421374299735e-06, + "loss": 2.6757, + "step": 2030 + }, + { + "epoch": 0.43, + "learning_rate": 7.148320134457577e-06, + "loss": 2.7091, + "step": 2031 + }, + { + "epoch": 0.43, + "learning_rate": 7.144996968038687e-06, + "loss": 2.6129, + "step": 2032 + }, + { + "epoch": 0.43, + "learning_rate": 7.141672639973e-06, + "loss": 2.6406, + "step": 2033 + }, + { + "epoch": 0.43, + "learning_rate": 7.138347152060845e-06, + "loss": 2.6592, + "step": 2034 + }, + { + "epoch": 0.43, + "learning_rate": 7.135020506103176e-06, + "loss": 2.6768, + "step": 2035 + }, + { + "epoch": 0.43, + "learning_rate": 7.1316927039015736e-06, + "loss": 2.6787, + "step": 2036 + }, + { + "epoch": 0.43, + "learning_rate": 7.128363747258246e-06, + "loss": 2.6176, + "step": 2037 + }, + { + "epoch": 0.43, + "learning_rate": 7.125033637976029e-06, + "loss": 2.686, + "step": 2038 + }, + { + "epoch": 0.43, + "learning_rate": 7.121702377858375e-06, + "loss": 2.6013, + "step": 2039 + }, + { + "epoch": 0.43, + "learning_rate": 7.1183699687093696e-06, + "loss": 2.6221, + "step": 2040 + }, + { + "epoch": 0.43, + "learning_rate": 7.115036412333713e-06, + "loss": 2.6546, + "step": 2041 + }, + { + "epoch": 0.43, + "learning_rate": 7.111701710536732e-06, + "loss": 2.6738, + "step": 2042 + }, + { + "epoch": 0.43, + "learning_rate": 7.1083658651243696e-06, + "loss": 2.6407, + "step": 2043 + }, + { + "epoch": 0.43, + "learning_rate": 7.105028877903188e-06, + "loss": 2.7065, + "step": 2044 + }, + { + "epoch": 0.43, + "learning_rate": 7.101690750680373e-06, + "loss": 2.7064, + "step": 2045 + }, + { + "epoch": 0.43, + "learning_rate": 7.098351485263722e-06, + "loss": 2.6132, + "step": 2046 + }, + { + "epoch": 0.43, + "learning_rate": 7.095011083461651e-06, + "loss": 2.6106, + "step": 2047 + }, + { + "epoch": 0.43, + "learning_rate": 7.091669547083193e-06, + "loss": 2.6948, + "step": 2048 + }, + { + "epoch": 0.43, + "learning_rate": 7.088326877937992e-06, + "loss": 2.7127, + "step": 2049 + }, + { + "epoch": 0.43, + "learning_rate": 7.084983077836309e-06, + "loss": 2.6686, + "step": 2050 + }, + { + "epoch": 0.43, + "learning_rate": 7.081638148589015e-06, + "loss": 2.6679, + "step": 2051 + }, + { + "epoch": 0.43, + "learning_rate": 7.078292092007593e-06, + "loss": 2.6505, + "step": 2052 + }, + { + "epoch": 0.43, + "learning_rate": 7.074944909904138e-06, + "loss": 2.664, + "step": 2053 + }, + { + "epoch": 0.43, + "learning_rate": 7.071596604091353e-06, + "loss": 2.618, + "step": 2054 + }, + { + "epoch": 0.43, + "learning_rate": 7.068247176382549e-06, + "loss": 2.6752, + "step": 2055 + }, + { + "epoch": 0.43, + "learning_rate": 7.064896628591649e-06, + "loss": 2.56, + "step": 2056 + }, + { + "epoch": 0.43, + "learning_rate": 7.061544962533174e-06, + "loss": 2.6689, + "step": 2057 + }, + { + "epoch": 0.43, + "learning_rate": 7.058192180022259e-06, + "loss": 2.6268, + "step": 2058 + }, + { + "epoch": 0.43, + "learning_rate": 7.054838282874639e-06, + "loss": 2.6473, + "step": 2059 + }, + { + "epoch": 0.43, + "learning_rate": 7.051483272906656e-06, + "loss": 2.6456, + "step": 2060 + }, + { + "epoch": 0.43, + "learning_rate": 7.048127151935251e-06, + "loss": 2.6688, + "step": 2061 + }, + { + "epoch": 0.43, + "learning_rate": 7.04476992177797e-06, + "loss": 2.6681, + "step": 2062 + }, + { + "epoch": 0.43, + "learning_rate": 7.041411584252956e-06, + "loss": 2.621, + "step": 2063 + }, + { + "epoch": 0.43, + "learning_rate": 7.038052141178955e-06, + "loss": 2.7368, + "step": 2064 + }, + { + "epoch": 0.43, + "learning_rate": 7.0346915943753105e-06, + "loss": 2.6375, + "step": 2065 + }, + { + "epoch": 0.43, + "learning_rate": 7.0313299456619635e-06, + "loss": 2.5724, + "step": 2066 + }, + { + "epoch": 0.43, + "learning_rate": 7.027967196859451e-06, + "loss": 2.6083, + "step": 2067 + }, + { + "epoch": 0.43, + "learning_rate": 7.0246033497889085e-06, + "loss": 2.5904, + "step": 2068 + }, + { + "epoch": 0.43, + "learning_rate": 7.021238406272064e-06, + "loss": 2.7026, + "step": 2069 + }, + { + "epoch": 0.43, + "learning_rate": 7.0178723681312375e-06, + "loss": 2.6738, + "step": 2070 + }, + { + "epoch": 0.43, + "learning_rate": 7.014505237189349e-06, + "loss": 2.5987, + "step": 2071 + }, + { + "epoch": 0.43, + "learning_rate": 7.011137015269901e-06, + "loss": 2.7155, + "step": 2072 + }, + { + "epoch": 0.43, + "learning_rate": 7.0077677041969926e-06, + "loss": 2.6431, + "step": 2073 + }, + { + "epoch": 0.43, + "learning_rate": 7.004397305795313e-06, + "loss": 2.6177, + "step": 2074 + }, + { + "epoch": 0.44, + "learning_rate": 7.0010258218901375e-06, + "loss": 2.6698, + "step": 2075 + }, + { + "epoch": 0.44, + "learning_rate": 6.997653254307328e-06, + "loss": 2.678, + "step": 2076 + }, + { + "epoch": 0.44, + "learning_rate": 6.9942796048733395e-06, + "loss": 2.6401, + "step": 2077 + }, + { + "epoch": 0.44, + "learning_rate": 6.99090487541521e-06, + "loss": 2.646, + "step": 2078 + }, + { + "epoch": 0.44, + "learning_rate": 6.987529067760555e-06, + "loss": 2.6173, + "step": 2079 + }, + { + "epoch": 0.44, + "learning_rate": 6.984152183737587e-06, + "loss": 2.6364, + "step": 2080 + }, + { + "epoch": 0.44, + "learning_rate": 6.980774225175092e-06, + "loss": 2.5926, + "step": 2081 + }, + { + "epoch": 0.44, + "learning_rate": 6.977395193902441e-06, + "loss": 2.6647, + "step": 2082 + }, + { + "epoch": 0.44, + "learning_rate": 6.974015091749586e-06, + "loss": 2.6366, + "step": 2083 + }, + { + "epoch": 0.44, + "learning_rate": 6.970633920547059e-06, + "loss": 2.6355, + "step": 2084 + }, + { + "epoch": 0.44, + "learning_rate": 6.967251682125969e-06, + "loss": 2.6111, + "step": 2085 + }, + { + "epoch": 0.44, + "learning_rate": 6.9638683783180064e-06, + "loss": 2.6788, + "step": 2086 + }, + { + "epoch": 0.44, + "learning_rate": 6.960484010955436e-06, + "loss": 2.658, + "step": 2087 + }, + { + "epoch": 0.44, + "learning_rate": 6.957098581871099e-06, + "loss": 2.6874, + "step": 2088 + }, + { + "epoch": 0.44, + "learning_rate": 6.953712092898413e-06, + "loss": 2.6204, + "step": 2089 + }, + { + "epoch": 0.44, + "learning_rate": 6.950324545871367e-06, + "loss": 2.6602, + "step": 2090 + }, + { + "epoch": 0.44, + "learning_rate": 6.9469359426245255e-06, + "loss": 2.64, + "step": 2091 + }, + { + "epoch": 0.44, + "learning_rate": 6.943546284993026e-06, + "loss": 2.6173, + "step": 2092 + }, + { + "epoch": 0.44, + "learning_rate": 6.940155574812571e-06, + "loss": 2.6481, + "step": 2093 + }, + { + "epoch": 0.44, + "learning_rate": 6.936763813919436e-06, + "loss": 2.6107, + "step": 2094 + }, + { + "epoch": 0.44, + "learning_rate": 6.9333710041504735e-06, + "loss": 2.6268, + "step": 2095 + }, + { + "epoch": 0.44, + "learning_rate": 6.929977147343092e-06, + "loss": 2.686, + "step": 2096 + }, + { + "epoch": 0.44, + "learning_rate": 6.9265822453352715e-06, + "loss": 2.6315, + "step": 2097 + }, + { + "epoch": 0.44, + "learning_rate": 6.92318629996556e-06, + "loss": 2.6113, + "step": 2098 + }, + { + "epoch": 0.44, + "learning_rate": 6.919789313073072e-06, + "loss": 2.6789, + "step": 2099 + }, + { + "epoch": 0.44, + "learning_rate": 6.9163912864974806e-06, + "loss": 2.6665, + "step": 2100 + }, + { + "epoch": 0.44, + "learning_rate": 6.912992222079024e-06, + "loss": 2.6242, + "step": 2101 + }, + { + "epoch": 0.44, + "learning_rate": 6.909592121658504e-06, + "loss": 2.6536, + "step": 2102 + }, + { + "epoch": 0.44, + "learning_rate": 6.906190987077281e-06, + "loss": 2.6167, + "step": 2103 + }, + { + "epoch": 0.44, + "learning_rate": 6.9027888201772805e-06, + "loss": 2.6956, + "step": 2104 + }, + { + "epoch": 0.44, + "learning_rate": 6.899385622800981e-06, + "loss": 2.6451, + "step": 2105 + }, + { + "epoch": 0.44, + "learning_rate": 6.8959813967914206e-06, + "loss": 2.6155, + "step": 2106 + }, + { + "epoch": 0.44, + "learning_rate": 6.892576143992199e-06, + "loss": 2.6219, + "step": 2107 + }, + { + "epoch": 0.44, + "learning_rate": 6.889169866247466e-06, + "loss": 2.6618, + "step": 2108 + }, + { + "epoch": 0.44, + "learning_rate": 6.885762565401928e-06, + "loss": 2.7305, + "step": 2109 + }, + { + "epoch": 0.44, + "learning_rate": 6.88235424330085e-06, + "loss": 2.664, + "step": 2110 + }, + { + "epoch": 0.44, + "learning_rate": 6.8789449017900425e-06, + "loss": 2.5814, + "step": 2111 + }, + { + "epoch": 0.44, + "learning_rate": 6.875534542715874e-06, + "loss": 2.6792, + "step": 2112 + }, + { + "epoch": 0.44, + "learning_rate": 6.872123167925262e-06, + "loss": 2.6841, + "step": 2113 + }, + { + "epoch": 0.44, + "learning_rate": 6.868710779265675e-06, + "loss": 2.6867, + "step": 2114 + }, + { + "epoch": 0.44, + "learning_rate": 6.8652973785851276e-06, + "loss": 2.595, + "step": 2115 + }, + { + "epoch": 0.44, + "learning_rate": 6.8618829677321855e-06, + "loss": 2.6169, + "step": 2116 + }, + { + "epoch": 0.44, + "learning_rate": 6.858467548555963e-06, + "loss": 2.6305, + "step": 2117 + }, + { + "epoch": 0.44, + "learning_rate": 6.855051122906114e-06, + "loss": 2.7029, + "step": 2118 + }, + { + "epoch": 0.44, + "learning_rate": 6.851633692632844e-06, + "loss": 2.6666, + "step": 2119 + }, + { + "epoch": 0.44, + "learning_rate": 6.848215259586901e-06, + "loss": 2.7552, + "step": 2120 + }, + { + "epoch": 0.44, + "learning_rate": 6.844795825619572e-06, + "loss": 2.6117, + "step": 2121 + }, + { + "epoch": 0.44, + "learning_rate": 6.841375392582693e-06, + "loss": 2.6916, + "step": 2122 + }, + { + "epoch": 0.45, + "learning_rate": 6.837953962328635e-06, + "loss": 2.7179, + "step": 2123 + }, + { + "epoch": 0.45, + "learning_rate": 6.83453153671031e-06, + "loss": 2.6235, + "step": 2124 + }, + { + "epoch": 0.45, + "learning_rate": 6.831108117581174e-06, + "loss": 2.5453, + "step": 2125 + }, + { + "epoch": 0.45, + "learning_rate": 6.827683706795216e-06, + "loss": 2.6356, + "step": 2126 + }, + { + "epoch": 0.45, + "learning_rate": 6.824258306206961e-06, + "loss": 2.6464, + "step": 2127 + }, + { + "epoch": 0.45, + "learning_rate": 6.820831917671476e-06, + "loss": 2.6475, + "step": 2128 + }, + { + "epoch": 0.45, + "learning_rate": 6.817404543044358e-06, + "loss": 2.5999, + "step": 2129 + }, + { + "epoch": 0.45, + "learning_rate": 6.813976184181739e-06, + "loss": 2.6391, + "step": 2130 + }, + { + "epoch": 0.45, + "learning_rate": 6.810546842940284e-06, + "loss": 2.679, + "step": 2131 + }, + { + "epoch": 0.45, + "learning_rate": 6.807116521177195e-06, + "loss": 2.6346, + "step": 2132 + }, + { + "epoch": 0.45, + "learning_rate": 6.803685220750195e-06, + "loss": 2.6336, + "step": 2133 + }, + { + "epoch": 0.45, + "learning_rate": 6.8002529435175465e-06, + "loss": 2.6472, + "step": 2134 + }, + { + "epoch": 0.45, + "learning_rate": 6.796819691338035e-06, + "loss": 2.6281, + "step": 2135 + }, + { + "epoch": 0.45, + "learning_rate": 6.793385466070978e-06, + "loss": 2.5783, + "step": 2136 + }, + { + "epoch": 0.45, + "learning_rate": 6.7899502695762155e-06, + "loss": 2.6864, + "step": 2137 + }, + { + "epoch": 0.45, + "learning_rate": 6.786514103714119e-06, + "loss": 2.658, + "step": 2138 + }, + { + "epoch": 0.45, + "learning_rate": 6.78307697034558e-06, + "loss": 2.6354, + "step": 2139 + }, + { + "epoch": 0.45, + "learning_rate": 6.779638871332017e-06, + "loss": 2.6721, + "step": 2140 + }, + { + "epoch": 0.45, + "learning_rate": 6.776199808535371e-06, + "loss": 2.6541, + "step": 2141 + }, + { + "epoch": 0.45, + "learning_rate": 6.772759783818102e-06, + "loss": 2.5936, + "step": 2142 + }, + { + "epoch": 0.45, + "learning_rate": 6.769318799043196e-06, + "loss": 2.618, + "step": 2143 + }, + { + "epoch": 0.45, + "learning_rate": 6.765876856074156e-06, + "loss": 2.6615, + "step": 2144 + }, + { + "epoch": 0.45, + "learning_rate": 6.762433956775002e-06, + "loss": 2.5665, + "step": 2145 + }, + { + "epoch": 0.45, + "learning_rate": 6.758990103010278e-06, + "loss": 2.6437, + "step": 2146 + }, + { + "epoch": 0.45, + "learning_rate": 6.755545296645037e-06, + "loss": 2.6817, + "step": 2147 + }, + { + "epoch": 0.45, + "learning_rate": 6.752099539544853e-06, + "loss": 2.5761, + "step": 2148 + }, + { + "epoch": 0.45, + "learning_rate": 6.748652833575815e-06, + "loss": 2.597, + "step": 2149 + }, + { + "epoch": 0.45, + "learning_rate": 6.745205180604526e-06, + "loss": 2.629, + "step": 2150 + }, + { + "epoch": 0.45, + "learning_rate": 6.741756582498097e-06, + "loss": 2.6243, + "step": 2151 + }, + { + "epoch": 0.45, + "learning_rate": 6.738307041124157e-06, + "loss": 2.6603, + "step": 2152 + }, + { + "epoch": 0.45, + "learning_rate": 6.734856558350842e-06, + "loss": 2.6316, + "step": 2153 + }, + { + "epoch": 0.45, + "learning_rate": 6.731405136046798e-06, + "loss": 2.6366, + "step": 2154 + }, + { + "epoch": 0.45, + "learning_rate": 6.7279527760811835e-06, + "loss": 2.62, + "step": 2155 + }, + { + "epoch": 0.45, + "learning_rate": 6.724499480323662e-06, + "loss": 2.6079, + "step": 2156 + }, + { + "epoch": 0.45, + "learning_rate": 6.721045250644402e-06, + "loss": 2.564, + "step": 2157 + }, + { + "epoch": 0.45, + "learning_rate": 6.71759008891408e-06, + "loss": 2.6383, + "step": 2158 + }, + { + "epoch": 0.45, + "learning_rate": 6.714133997003878e-06, + "loss": 2.6211, + "step": 2159 + }, + { + "epoch": 0.45, + "learning_rate": 6.7106769767854785e-06, + "loss": 2.5684, + "step": 2160 + }, + { + "epoch": 0.45, + "learning_rate": 6.707219030131074e-06, + "loss": 2.608, + "step": 2161 + }, + { + "epoch": 0.45, + "learning_rate": 6.703760158913349e-06, + "loss": 2.6212, + "step": 2162 + }, + { + "epoch": 0.45, + "learning_rate": 6.700300365005494e-06, + "loss": 2.6279, + "step": 2163 + }, + { + "epoch": 0.45, + "learning_rate": 6.696839650281202e-06, + "loss": 2.6446, + "step": 2164 + }, + { + "epoch": 0.45, + "learning_rate": 6.693378016614657e-06, + "loss": 2.6528, + "step": 2165 + }, + { + "epoch": 0.45, + "learning_rate": 6.6899154658805456e-06, + "loss": 2.6895, + "step": 2166 + }, + { + "epoch": 0.45, + "learning_rate": 6.6864519999540525e-06, + "loss": 2.6243, + "step": 2167 + }, + { + "epoch": 0.45, + "learning_rate": 6.682987620710856e-06, + "loss": 2.6433, + "step": 2168 + }, + { + "epoch": 0.45, + "learning_rate": 6.679522330027126e-06, + "loss": 2.6376, + "step": 2169 + }, + { + "epoch": 0.46, + "learning_rate": 6.676056129779532e-06, + "loss": 2.6513, + "step": 2170 + }, + { + "epoch": 0.46, + "learning_rate": 6.6725890218452315e-06, + "loss": 2.6486, + "step": 2171 + }, + { + "epoch": 0.46, + "learning_rate": 6.669121008101875e-06, + "loss": 2.6711, + "step": 2172 + }, + { + "epoch": 0.46, + "learning_rate": 6.665652090427606e-06, + "loss": 2.6247, + "step": 2173 + }, + { + "epoch": 0.46, + "learning_rate": 6.662182270701051e-06, + "loss": 2.6393, + "step": 2174 + }, + { + "epoch": 0.46, + "learning_rate": 6.658711550801332e-06, + "loss": 2.6814, + "step": 2175 + }, + { + "epoch": 0.46, + "learning_rate": 6.655239932608055e-06, + "loss": 2.6909, + "step": 2176 + }, + { + "epoch": 0.46, + "learning_rate": 6.651767418001314e-06, + "loss": 2.5555, + "step": 2177 + }, + { + "epoch": 0.46, + "learning_rate": 6.648294008861686e-06, + "loss": 2.6288, + "step": 2178 + }, + { + "epoch": 0.46, + "learning_rate": 6.644819707070237e-06, + "loss": 2.6701, + "step": 2179 + }, + { + "epoch": 0.46, + "learning_rate": 6.64134451450851e-06, + "loss": 2.6486, + "step": 2180 + }, + { + "epoch": 0.46, + "learning_rate": 6.637868433058534e-06, + "loss": 2.5994, + "step": 2181 + }, + { + "epoch": 0.46, + "learning_rate": 6.634391464602821e-06, + "loss": 2.6515, + "step": 2182 + }, + { + "epoch": 0.46, + "learning_rate": 6.630913611024365e-06, + "loss": 2.673, + "step": 2183 + }, + { + "epoch": 0.46, + "learning_rate": 6.627434874206629e-06, + "loss": 2.5607, + "step": 2184 + }, + { + "epoch": 0.46, + "learning_rate": 6.623955256033565e-06, + "loss": 2.6789, + "step": 2185 + }, + { + "epoch": 0.46, + "learning_rate": 6.6204747583896e-06, + "loss": 2.6703, + "step": 2186 + }, + { + "epoch": 0.46, + "learning_rate": 6.616993383159632e-06, + "loss": 2.6271, + "step": 2187 + }, + { + "epoch": 0.46, + "learning_rate": 6.6135111322290415e-06, + "loss": 2.6143, + "step": 2188 + }, + { + "epoch": 0.46, + "learning_rate": 6.610028007483679e-06, + "loss": 2.6193, + "step": 2189 + }, + { + "epoch": 0.46, + "learning_rate": 6.60654401080987e-06, + "loss": 2.6887, + "step": 2190 + }, + { + "epoch": 0.46, + "learning_rate": 6.603059144094409e-06, + "loss": 2.6947, + "step": 2191 + }, + { + "epoch": 0.46, + "learning_rate": 6.599573409224567e-06, + "loss": 2.6958, + "step": 2192 + }, + { + "epoch": 0.46, + "learning_rate": 6.596086808088079e-06, + "loss": 2.6335, + "step": 2193 + }, + { + "epoch": 0.46, + "learning_rate": 6.5925993425731564e-06, + "loss": 2.749, + "step": 2194 + }, + { + "epoch": 0.46, + "learning_rate": 6.58911101456847e-06, + "loss": 2.6355, + "step": 2195 + }, + { + "epoch": 0.46, + "learning_rate": 6.585621825963165e-06, + "loss": 2.6475, + "step": 2196 + }, + { + "epoch": 0.46, + "learning_rate": 6.582131778646852e-06, + "loss": 2.6607, + "step": 2197 + }, + { + "epoch": 0.46, + "learning_rate": 6.578640874509599e-06, + "loss": 2.6322, + "step": 2198 + }, + { + "epoch": 0.46, + "learning_rate": 6.575149115441949e-06, + "loss": 2.6645, + "step": 2199 + }, + { + "epoch": 0.46, + "learning_rate": 6.571656503334901e-06, + "loss": 2.5765, + "step": 2200 + }, + { + "epoch": 0.46, + "learning_rate": 6.568163040079918e-06, + "loss": 2.5832, + "step": 2201 + }, + { + "epoch": 0.46, + "learning_rate": 6.564668727568921e-06, + "loss": 2.7309, + "step": 2202 + }, + { + "epoch": 0.46, + "learning_rate": 6.561173567694296e-06, + "loss": 2.6674, + "step": 2203 + }, + { + "epoch": 0.46, + "learning_rate": 6.557677562348887e-06, + "loss": 2.6325, + "step": 2204 + }, + { + "epoch": 0.46, + "learning_rate": 6.55418071342599e-06, + "loss": 2.6424, + "step": 2205 + }, + { + "epoch": 0.46, + "learning_rate": 6.550683022819367e-06, + "loss": 2.6652, + "step": 2206 + }, + { + "epoch": 0.46, + "learning_rate": 6.547184492423227e-06, + "loss": 2.5881, + "step": 2207 + }, + { + "epoch": 0.46, + "learning_rate": 6.543685124132238e-06, + "loss": 2.6686, + "step": 2208 + }, + { + "epoch": 0.46, + "learning_rate": 6.540184919841524e-06, + "loss": 2.5958, + "step": 2209 + }, + { + "epoch": 0.46, + "learning_rate": 6.536683881446658e-06, + "loss": 2.6679, + "step": 2210 + }, + { + "epoch": 0.46, + "learning_rate": 6.533182010843666e-06, + "loss": 2.6629, + "step": 2211 + }, + { + "epoch": 0.46, + "learning_rate": 6.529679309929025e-06, + "loss": 2.6455, + "step": 2212 + }, + { + "epoch": 0.46, + "learning_rate": 6.5261757805996605e-06, + "loss": 2.5629, + "step": 2213 + }, + { + "epoch": 0.46, + "learning_rate": 6.522671424752948e-06, + "loss": 2.6707, + "step": 2214 + }, + { + "epoch": 0.46, + "learning_rate": 6.51916624428671e-06, + "loss": 2.6194, + "step": 2215 + }, + { + "epoch": 0.46, + "learning_rate": 6.515660241099217e-06, + "loss": 2.5938, + "step": 2216 + }, + { + "epoch": 0.46, + "learning_rate": 6.512153417089182e-06, + "loss": 2.6068, + "step": 2217 + }, + { + "epoch": 0.47, + "learning_rate": 6.508645774155764e-06, + "loss": 2.6509, + "step": 2218 + }, + { + "epoch": 0.47, + "learning_rate": 6.5051373141985685e-06, + "loss": 2.6361, + "step": 2219 + }, + { + "epoch": 0.47, + "learning_rate": 6.501628039117637e-06, + "loss": 2.642, + "step": 2220 + }, + { + "epoch": 0.47, + "learning_rate": 6.498117950813458e-06, + "loss": 2.7302, + "step": 2221 + }, + { + "epoch": 0.47, + "learning_rate": 6.49460705118696e-06, + "loss": 2.6685, + "step": 2222 + }, + { + "epoch": 0.47, + "learning_rate": 6.491095342139507e-06, + "loss": 2.6104, + "step": 2223 + }, + { + "epoch": 0.47, + "learning_rate": 6.487582825572907e-06, + "loss": 2.6279, + "step": 2224 + }, + { + "epoch": 0.47, + "learning_rate": 6.484069503389398e-06, + "loss": 2.6519, + "step": 2225 + }, + { + "epoch": 0.47, + "learning_rate": 6.480555377491662e-06, + "loss": 2.6244, + "step": 2226 + }, + { + "epoch": 0.47, + "learning_rate": 6.47704044978281e-06, + "loss": 2.6744, + "step": 2227 + }, + { + "epoch": 0.47, + "learning_rate": 6.473524722166391e-06, + "loss": 2.6815, + "step": 2228 + }, + { + "epoch": 0.47, + "learning_rate": 6.470008196546387e-06, + "loss": 2.615, + "step": 2229 + }, + { + "epoch": 0.47, + "learning_rate": 6.46649087482721e-06, + "loss": 2.5807, + "step": 2230 + }, + { + "epoch": 0.47, + "learning_rate": 6.462972758913705e-06, + "loss": 2.6312, + "step": 2231 + }, + { + "epoch": 0.47, + "learning_rate": 6.459453850711147e-06, + "loss": 2.6846, + "step": 2232 + }, + { + "epoch": 0.47, + "learning_rate": 6.455934152125239e-06, + "loss": 2.6641, + "step": 2233 + }, + { + "epoch": 0.47, + "learning_rate": 6.452413665062111e-06, + "loss": 2.6451, + "step": 2234 + }, + { + "epoch": 0.47, + "learning_rate": 6.448892391428325e-06, + "loss": 2.636, + "step": 2235 + }, + { + "epoch": 0.47, + "learning_rate": 6.445370333130862e-06, + "loss": 2.6269, + "step": 2236 + }, + { + "epoch": 0.47, + "learning_rate": 6.4418474920771365e-06, + "loss": 2.6333, + "step": 2237 + }, + { + "epoch": 0.47, + "learning_rate": 6.438323870174979e-06, + "loss": 2.6839, + "step": 2238 + }, + { + "epoch": 0.47, + "learning_rate": 6.434799469332642e-06, + "loss": 2.6735, + "step": 2239 + }, + { + "epoch": 0.47, + "learning_rate": 6.431274291458811e-06, + "loss": 2.626, + "step": 2240 + }, + { + "epoch": 0.47, + "learning_rate": 6.427748338462579e-06, + "loss": 2.6423, + "step": 2241 + }, + { + "epoch": 0.47, + "learning_rate": 6.42422161225347e-06, + "loss": 2.6706, + "step": 2242 + }, + { + "epoch": 0.47, + "learning_rate": 6.420694114741417e-06, + "loss": 2.6367, + "step": 2243 + }, + { + "epoch": 0.47, + "learning_rate": 6.417165847836777e-06, + "loss": 2.6881, + "step": 2244 + }, + { + "epoch": 0.47, + "learning_rate": 6.413636813450324e-06, + "loss": 2.6012, + "step": 2245 + }, + { + "epoch": 0.47, + "learning_rate": 6.410107013493241e-06, + "loss": 2.6096, + "step": 2246 + }, + { + "epoch": 0.47, + "learning_rate": 6.406576449877134e-06, + "loss": 2.637, + "step": 2247 + }, + { + "epoch": 0.47, + "learning_rate": 6.403045124514018e-06, + "loss": 2.6629, + "step": 2248 + }, + { + "epoch": 0.47, + "learning_rate": 6.399513039316319e-06, + "loss": 2.7288, + "step": 2249 + }, + { + "epoch": 0.47, + "learning_rate": 6.39598019619688e-06, + "loss": 2.6998, + "step": 2250 + }, + { + "epoch": 0.47, + "learning_rate": 6.392446597068949e-06, + "loss": 2.6023, + "step": 2251 + }, + { + "epoch": 0.47, + "learning_rate": 6.388912243846186e-06, + "loss": 2.6886, + "step": 2252 + }, + { + "epoch": 0.47, + "learning_rate": 6.38537713844266e-06, + "loss": 2.6456, + "step": 2253 + }, + { + "epoch": 0.47, + "learning_rate": 6.381841282772844e-06, + "loss": 2.6857, + "step": 2254 + }, + { + "epoch": 0.47, + "learning_rate": 6.378304678751624e-06, + "loss": 2.601, + "step": 2255 + }, + { + "epoch": 0.47, + "learning_rate": 6.374767328294281e-06, + "loss": 2.6116, + "step": 2256 + }, + { + "epoch": 0.47, + "learning_rate": 6.3712292333165105e-06, + "loss": 2.6574, + "step": 2257 + }, + { + "epoch": 0.47, + "learning_rate": 6.367690395734407e-06, + "loss": 2.6923, + "step": 2258 + }, + { + "epoch": 0.47, + "learning_rate": 6.364150817464466e-06, + "loss": 2.6474, + "step": 2259 + }, + { + "epoch": 0.47, + "learning_rate": 6.360610500423583e-06, + "loss": 2.567, + "step": 2260 + }, + { + "epoch": 0.47, + "learning_rate": 6.35706944652906e-06, + "loss": 2.6285, + "step": 2261 + }, + { + "epoch": 0.47, + "learning_rate": 6.353527657698591e-06, + "loss": 2.6564, + "step": 2262 + }, + { + "epoch": 0.47, + "learning_rate": 6.349985135850272e-06, + "loss": 2.6295, + "step": 2263 + }, + { + "epoch": 0.47, + "learning_rate": 6.346441882902594e-06, + "loss": 2.5681, + "step": 2264 + }, + { + "epoch": 0.47, + "learning_rate": 6.342897900774446e-06, + "loss": 2.6479, + "step": 2265 + }, + { + "epoch": 0.48, + "learning_rate": 6.339353191385109e-06, + "loss": 2.6508, + "step": 2266 + }, + { + "epoch": 0.48, + "learning_rate": 6.335807756654262e-06, + "loss": 2.6878, + "step": 2267 + }, + { + "epoch": 0.48, + "learning_rate": 6.3322615985019705e-06, + "loss": 2.6546, + "step": 2268 + }, + { + "epoch": 0.48, + "learning_rate": 6.328714718848699e-06, + "loss": 2.688, + "step": 2269 + }, + { + "epoch": 0.48, + "learning_rate": 6.325167119615299e-06, + "loss": 2.6785, + "step": 2270 + }, + { + "epoch": 0.48, + "learning_rate": 6.3216188027230095e-06, + "loss": 2.6141, + "step": 2271 + }, + { + "epoch": 0.48, + "learning_rate": 6.318069770093462e-06, + "loss": 2.6785, + "step": 2272 + }, + { + "epoch": 0.48, + "learning_rate": 6.314520023648678e-06, + "loss": 2.6449, + "step": 2273 + }, + { + "epoch": 0.48, + "learning_rate": 6.310969565311056e-06, + "loss": 2.5765, + "step": 2274 + }, + { + "epoch": 0.48, + "learning_rate": 6.307418397003389e-06, + "loss": 2.633, + "step": 2275 + }, + { + "epoch": 0.48, + "learning_rate": 6.303866520648851e-06, + "loss": 2.6618, + "step": 2276 + }, + { + "epoch": 0.48, + "learning_rate": 6.300313938171e-06, + "loss": 2.6782, + "step": 2277 + }, + { + "epoch": 0.48, + "learning_rate": 6.296760651493775e-06, + "loss": 2.7006, + "step": 2278 + }, + { + "epoch": 0.48, + "learning_rate": 6.2932066625415e-06, + "loss": 2.615, + "step": 2279 + }, + { + "epoch": 0.48, + "learning_rate": 6.289651973238877e-06, + "loss": 2.584, + "step": 2280 + }, + { + "epoch": 0.48, + "learning_rate": 6.286096585510983e-06, + "loss": 2.5997, + "step": 2281 + }, + { + "epoch": 0.48, + "learning_rate": 6.2825405012832815e-06, + "loss": 2.6166, + "step": 2282 + }, + { + "epoch": 0.48, + "learning_rate": 6.278983722481609e-06, + "loss": 2.7019, + "step": 2283 + }, + { + "epoch": 0.48, + "learning_rate": 6.275426251032176e-06, + "loss": 2.6224, + "step": 2284 + }, + { + "epoch": 0.48, + "learning_rate": 6.2718680888615734e-06, + "loss": 2.5998, + "step": 2285 + }, + { + "epoch": 0.48, + "learning_rate": 6.2683092378967615e-06, + "loss": 2.6572, + "step": 2286 + }, + { + "epoch": 0.48, + "learning_rate": 6.264749700065075e-06, + "loss": 2.7156, + "step": 2287 + }, + { + "epoch": 0.48, + "learning_rate": 6.261189477294221e-06, + "loss": 2.6675, + "step": 2288 + }, + { + "epoch": 0.48, + "learning_rate": 6.25762857151228e-06, + "loss": 2.701, + "step": 2289 + }, + { + "epoch": 0.48, + "learning_rate": 6.254066984647694e-06, + "loss": 2.6482, + "step": 2290 + }, + { + "epoch": 0.48, + "learning_rate": 6.250504718629288e-06, + "loss": 2.6233, + "step": 2291 + }, + { + "epoch": 0.48, + "learning_rate": 6.246941775386239e-06, + "loss": 2.65, + "step": 2292 + }, + { + "epoch": 0.48, + "learning_rate": 6.243378156848101e-06, + "loss": 2.6678, + "step": 2293 + }, + { + "epoch": 0.48, + "learning_rate": 6.2398138649447935e-06, + "loss": 2.6639, + "step": 2294 + }, + { + "epoch": 0.48, + "learning_rate": 6.236248901606595e-06, + "loss": 2.6815, + "step": 2295 + }, + { + "epoch": 0.48, + "learning_rate": 6.23268326876415e-06, + "loss": 2.5607, + "step": 2296 + }, + { + "epoch": 0.48, + "learning_rate": 6.22911696834847e-06, + "loss": 2.636, + "step": 2297 + }, + { + "epoch": 0.48, + "learning_rate": 6.2255500022909235e-06, + "loss": 2.6555, + "step": 2298 + }, + { + "epoch": 0.48, + "learning_rate": 6.221982372523235e-06, + "loss": 2.669, + "step": 2299 + }, + { + "epoch": 0.48, + "learning_rate": 6.218414080977502e-06, + "loss": 2.6145, + "step": 2300 + }, + { + "epoch": 0.48, + "learning_rate": 6.214845129586167e-06, + "loss": 2.6776, + "step": 2301 + }, + { + "epoch": 0.48, + "learning_rate": 6.211275520282034e-06, + "loss": 2.6319, + "step": 2302 + }, + { + "epoch": 0.48, + "learning_rate": 6.207705254998269e-06, + "loss": 2.6204, + "step": 2303 + }, + { + "epoch": 0.48, + "learning_rate": 6.204134335668386e-06, + "loss": 2.6407, + "step": 2304 + }, + { + "epoch": 0.48, + "learning_rate": 6.200562764226256e-06, + "loss": 2.7073, + "step": 2305 + }, + { + "epoch": 0.48, + "learning_rate": 6.196990542606102e-06, + "loss": 2.604, + "step": 2306 + }, + { + "epoch": 0.48, + "learning_rate": 6.193417672742501e-06, + "loss": 2.7046, + "step": 2307 + }, + { + "epoch": 0.48, + "learning_rate": 6.189844156570379e-06, + "loss": 2.5907, + "step": 2308 + }, + { + "epoch": 0.48, + "learning_rate": 6.186269996025018e-06, + "loss": 2.6394, + "step": 2309 + }, + { + "epoch": 0.48, + "learning_rate": 6.182695193042039e-06, + "loss": 2.6788, + "step": 2310 + }, + { + "epoch": 0.48, + "learning_rate": 6.179119749557415e-06, + "loss": 2.7092, + "step": 2311 + }, + { + "epoch": 0.48, + "learning_rate": 6.175543667507472e-06, + "loss": 2.6214, + "step": 2312 + }, + { + "epoch": 0.49, + "learning_rate": 6.171966948828873e-06, + "loss": 2.6365, + "step": 2313 + }, + { + "epoch": 0.49, + "learning_rate": 6.168389595458629e-06, + "loss": 2.6773, + "step": 2314 + }, + { + "epoch": 0.49, + "learning_rate": 6.1648116093340985e-06, + "loss": 2.602, + "step": 2315 + }, + { + "epoch": 0.49, + "learning_rate": 6.161232992392978e-06, + "loss": 2.6095, + "step": 2316 + }, + { + "epoch": 0.49, + "learning_rate": 6.157653746573306e-06, + "loss": 2.5376, + "step": 2317 + }, + { + "epoch": 0.49, + "learning_rate": 6.154073873813463e-06, + "loss": 2.6542, + "step": 2318 + }, + { + "epoch": 0.49, + "learning_rate": 6.15049337605217e-06, + "loss": 2.6539, + "step": 2319 + }, + { + "epoch": 0.49, + "learning_rate": 6.146912255228483e-06, + "loss": 2.719, + "step": 2320 + }, + { + "epoch": 0.49, + "learning_rate": 6.143330513281799e-06, + "loss": 2.6203, + "step": 2321 + }, + { + "epoch": 0.49, + "learning_rate": 6.13974815215185e-06, + "loss": 2.7132, + "step": 2322 + }, + { + "epoch": 0.49, + "learning_rate": 6.136165173778702e-06, + "loss": 2.7159, + "step": 2323 + }, + { + "epoch": 0.49, + "learning_rate": 6.132581580102757e-06, + "loss": 2.625, + "step": 2324 + }, + { + "epoch": 0.49, + "learning_rate": 6.12899737306475e-06, + "loss": 2.5732, + "step": 2325 + }, + { + "epoch": 0.49, + "learning_rate": 6.125412554605747e-06, + "loss": 2.6256, + "step": 2326 + }, + { + "epoch": 0.49, + "learning_rate": 6.121827126667149e-06, + "loss": 2.6838, + "step": 2327 + }, + { + "epoch": 0.49, + "learning_rate": 6.118241091190681e-06, + "loss": 2.6937, + "step": 2328 + }, + { + "epoch": 0.49, + "learning_rate": 6.114654450118399e-06, + "loss": 2.6284, + "step": 2329 + }, + { + "epoch": 0.49, + "learning_rate": 6.111067205392693e-06, + "loss": 2.6143, + "step": 2330 + }, + { + "epoch": 0.49, + "learning_rate": 6.107479358956272e-06, + "loss": 2.6044, + "step": 2331 + }, + { + "epoch": 0.49, + "learning_rate": 6.103890912752171e-06, + "loss": 2.6682, + "step": 2332 + }, + { + "epoch": 0.49, + "learning_rate": 6.100301868723758e-06, + "loss": 2.7196, + "step": 2333 + }, + { + "epoch": 0.49, + "learning_rate": 6.096712228814718e-06, + "loss": 2.6663, + "step": 2334 + }, + { + "epoch": 0.49, + "learning_rate": 6.0931219949690575e-06, + "loss": 2.6031, + "step": 2335 + }, + { + "epoch": 0.49, + "learning_rate": 6.089531169131109e-06, + "loss": 2.6948, + "step": 2336 + }, + { + "epoch": 0.49, + "learning_rate": 6.0859397532455245e-06, + "loss": 2.5948, + "step": 2337 + }, + { + "epoch": 0.49, + "learning_rate": 6.082347749257275e-06, + "loss": 2.6345, + "step": 2338 + }, + { + "epoch": 0.49, + "learning_rate": 6.078755159111648e-06, + "loss": 2.6808, + "step": 2339 + }, + { + "epoch": 0.49, + "learning_rate": 6.075161984754252e-06, + "loss": 2.6835, + "step": 2340 + }, + { + "epoch": 0.49, + "learning_rate": 6.0715682281310106e-06, + "loss": 2.6279, + "step": 2341 + }, + { + "epoch": 0.49, + "learning_rate": 6.067973891188161e-06, + "loss": 2.5858, + "step": 2342 + }, + { + "epoch": 0.49, + "learning_rate": 6.064378975872257e-06, + "loss": 2.6425, + "step": 2343 + }, + { + "epoch": 0.49, + "learning_rate": 6.060783484130163e-06, + "loss": 2.7027, + "step": 2344 + }, + { + "epoch": 0.49, + "learning_rate": 6.057187417909061e-06, + "loss": 2.6346, + "step": 2345 + }, + { + "epoch": 0.49, + "learning_rate": 6.053590779156437e-06, + "loss": 2.6032, + "step": 2346 + }, + { + "epoch": 0.49, + "learning_rate": 6.049993569820091e-06, + "loss": 2.5936, + "step": 2347 + }, + { + "epoch": 0.49, + "learning_rate": 6.046395791848133e-06, + "loss": 2.5967, + "step": 2348 + }, + { + "epoch": 0.49, + "learning_rate": 6.042797447188977e-06, + "loss": 2.6506, + "step": 2349 + }, + { + "epoch": 0.49, + "learning_rate": 6.039198537791348e-06, + "loss": 2.6128, + "step": 2350 + }, + { + "epoch": 0.49, + "learning_rate": 6.035599065604275e-06, + "loss": 2.6566, + "step": 2351 + }, + { + "epoch": 0.49, + "learning_rate": 6.031999032577093e-06, + "loss": 2.6339, + "step": 2352 + }, + { + "epoch": 0.49, + "learning_rate": 6.0283984406594355e-06, + "loss": 2.6076, + "step": 2353 + }, + { + "epoch": 0.49, + "learning_rate": 6.024797291801247e-06, + "loss": 2.63, + "step": 2354 + }, + { + "epoch": 0.49, + "learning_rate": 6.021195587952768e-06, + "loss": 2.625, + "step": 2355 + }, + { + "epoch": 0.49, + "learning_rate": 6.017593331064542e-06, + "loss": 2.7059, + "step": 2356 + }, + { + "epoch": 0.49, + "learning_rate": 6.013990523087409e-06, + "loss": 2.625, + "step": 2357 + }, + { + "epoch": 0.49, + "learning_rate": 6.01038716597251e-06, + "loss": 2.6658, + "step": 2358 + }, + { + "epoch": 0.49, + "learning_rate": 6.006783261671284e-06, + "loss": 2.6439, + "step": 2359 + }, + { + "epoch": 0.49, + "learning_rate": 6.003178812135464e-06, + "loss": 2.6077, + "step": 2360 + }, + { + "epoch": 0.5, + "learning_rate": 5.99957381931708e-06, + "loss": 2.729, + "step": 2361 + }, + { + "epoch": 0.5, + "learning_rate": 5.995968285168455e-06, + "loss": 2.5947, + "step": 2362 + }, + { + "epoch": 0.5, + "learning_rate": 5.99236221164221e-06, + "loss": 2.6651, + "step": 2363 + }, + { + "epoch": 0.5, + "learning_rate": 5.9887556006912475e-06, + "loss": 2.6529, + "step": 2364 + }, + { + "epoch": 0.5, + "learning_rate": 5.985148454268769e-06, + "loss": 2.6212, + "step": 2365 + }, + { + "epoch": 0.5, + "learning_rate": 5.9815407743282694e-06, + "loss": 2.6336, + "step": 2366 + }, + { + "epoch": 0.5, + "learning_rate": 5.977932562823522e-06, + "loss": 2.6324, + "step": 2367 + }, + { + "epoch": 0.5, + "learning_rate": 5.974323821708594e-06, + "loss": 2.6959, + "step": 2368 + }, + { + "epoch": 0.5, + "learning_rate": 5.970714552937843e-06, + "loss": 2.6385, + "step": 2369 + }, + { + "epoch": 0.5, + "learning_rate": 5.967104758465907e-06, + "loss": 2.6654, + "step": 2370 + }, + { + "epoch": 0.5, + "learning_rate": 5.963494440247705e-06, + "loss": 2.6634, + "step": 2371 + }, + { + "epoch": 0.5, + "learning_rate": 5.959883600238452e-06, + "loss": 2.7035, + "step": 2372 + }, + { + "epoch": 0.5, + "learning_rate": 5.956272240393635e-06, + "loss": 2.6602, + "step": 2373 + }, + { + "epoch": 0.5, + "learning_rate": 5.9526603626690245e-06, + "loss": 2.6612, + "step": 2374 + }, + { + "epoch": 0.5, + "learning_rate": 5.949047969020676e-06, + "loss": 2.6312, + "step": 2375 + }, + { + "epoch": 0.5, + "learning_rate": 5.945435061404918e-06, + "loss": 2.657, + "step": 2376 + }, + { + "epoch": 0.5, + "learning_rate": 5.941821641778363e-06, + "loss": 2.6412, + "step": 2377 + }, + { + "epoch": 0.5, + "learning_rate": 5.938207712097895e-06, + "loss": 2.5864, + "step": 2378 + }, + { + "epoch": 0.5, + "learning_rate": 5.9345932743206814e-06, + "loss": 2.6905, + "step": 2379 + }, + { + "epoch": 0.5, + "learning_rate": 5.930978330404158e-06, + "loss": 2.6618, + "step": 2380 + }, + { + "epoch": 0.5, + "learning_rate": 5.927362882306039e-06, + "loss": 2.6348, + "step": 2381 + }, + { + "epoch": 0.5, + "learning_rate": 5.923746931984311e-06, + "loss": 2.5781, + "step": 2382 + }, + { + "epoch": 0.5, + "learning_rate": 5.920130481397229e-06, + "loss": 2.6075, + "step": 2383 + }, + { + "epoch": 0.5, + "learning_rate": 5.916513532503325e-06, + "loss": 2.6926, + "step": 2384 + }, + { + "epoch": 0.5, + "learning_rate": 5.912896087261397e-06, + "loss": 2.6369, + "step": 2385 + }, + { + "epoch": 0.5, + "learning_rate": 5.909278147630509e-06, + "loss": 2.589, + "step": 2386 + }, + { + "epoch": 0.5, + "learning_rate": 5.90565971557e-06, + "loss": 2.7122, + "step": 2387 + }, + { + "epoch": 0.5, + "learning_rate": 5.902040793039472e-06, + "loss": 2.6395, + "step": 2388 + }, + { + "epoch": 0.5, + "learning_rate": 5.898421381998789e-06, + "loss": 2.6441, + "step": 2389 + }, + { + "epoch": 0.5, + "learning_rate": 5.894801484408086e-06, + "loss": 2.5857, + "step": 2390 + }, + { + "epoch": 0.5, + "learning_rate": 5.891181102227758e-06, + "loss": 2.6707, + "step": 2391 + }, + { + "epoch": 0.5, + "learning_rate": 5.887560237418463e-06, + "loss": 2.6423, + "step": 2392 + }, + { + "epoch": 0.5, + "learning_rate": 5.883938891941117e-06, + "loss": 2.6235, + "step": 2393 + }, + { + "epoch": 0.5, + "learning_rate": 5.880317067756906e-06, + "loss": 2.6055, + "step": 2394 + }, + { + "epoch": 0.5, + "learning_rate": 5.876694766827263e-06, + "loss": 2.6137, + "step": 2395 + }, + { + "epoch": 0.5, + "learning_rate": 5.873071991113889e-06, + "loss": 2.6241, + "step": 2396 + }, + { + "epoch": 0.5, + "learning_rate": 5.869448742578736e-06, + "loss": 2.6858, + "step": 2397 + }, + { + "epoch": 0.5, + "learning_rate": 5.865825023184015e-06, + "loss": 2.6469, + "step": 2398 + }, + { + "epoch": 0.5, + "learning_rate": 5.862200834892192e-06, + "loss": 2.658, + "step": 2399 + }, + { + "epoch": 0.5, + "learning_rate": 5.858576179665985e-06, + "loss": 2.6626, + "step": 2400 + }, + { + "epoch": 0.5, + "learning_rate": 5.8549510594683655e-06, + "loss": 2.6435, + "step": 2401 + }, + { + "epoch": 0.5, + "learning_rate": 5.851325476262558e-06, + "loss": 2.5661, + "step": 2402 + }, + { + "epoch": 0.5, + "learning_rate": 5.847699432012039e-06, + "loss": 2.6028, + "step": 2403 + }, + { + "epoch": 0.5, + "learning_rate": 5.84407292868053e-06, + "loss": 2.6299, + "step": 2404 + }, + { + "epoch": 0.5, + "learning_rate": 5.840445968232005e-06, + "loss": 2.6282, + "step": 2405 + }, + { + "epoch": 0.5, + "learning_rate": 5.8368185526306865e-06, + "loss": 2.595, + "step": 2406 + }, + { + "epoch": 0.5, + "learning_rate": 5.8331906838410365e-06, + "loss": 2.6455, + "step": 2407 + }, + { + "epoch": 0.5, + "learning_rate": 5.829562363827773e-06, + "loss": 2.601, + "step": 2408 + }, + { + "epoch": 0.51, + "learning_rate": 5.82593359455585e-06, + "loss": 2.6645, + "step": 2409 + }, + { + "epoch": 0.51, + "learning_rate": 5.822304377990467e-06, + "loss": 2.5835, + "step": 2410 + }, + { + "epoch": 0.51, + "learning_rate": 5.818674716097068e-06, + "loss": 2.6322, + "step": 2411 + }, + { + "epoch": 0.51, + "learning_rate": 5.815044610841336e-06, + "loss": 2.6716, + "step": 2412 + }, + { + "epoch": 0.51, + "learning_rate": 5.8114140641891934e-06, + "loss": 2.611, + "step": 2413 + }, + { + "epoch": 0.51, + "learning_rate": 5.8077830781068044e-06, + "loss": 2.7244, + "step": 2414 + }, + { + "epoch": 0.51, + "learning_rate": 5.804151654560569e-06, + "loss": 2.6323, + "step": 2415 + }, + { + "epoch": 0.51, + "learning_rate": 5.800519795517124e-06, + "loss": 2.59, + "step": 2416 + }, + { + "epoch": 0.51, + "learning_rate": 5.796887502943343e-06, + "loss": 2.6911, + "step": 2417 + }, + { + "epoch": 0.51, + "learning_rate": 5.793254778806336e-06, + "loss": 2.6323, + "step": 2418 + }, + { + "epoch": 0.51, + "learning_rate": 5.789621625073442e-06, + "loss": 2.5957, + "step": 2419 + }, + { + "epoch": 0.51, + "learning_rate": 5.785988043712239e-06, + "loss": 2.6339, + "step": 2420 + }, + { + "epoch": 0.51, + "learning_rate": 5.782354036690529e-06, + "loss": 2.6318, + "step": 2421 + }, + { + "epoch": 0.51, + "learning_rate": 5.77871960597635e-06, + "loss": 2.6798, + "step": 2422 + }, + { + "epoch": 0.51, + "learning_rate": 5.775084753537969e-06, + "loss": 2.5891, + "step": 2423 + }, + { + "epoch": 0.51, + "learning_rate": 5.771449481343884e-06, + "loss": 2.6406, + "step": 2424 + }, + { + "epoch": 0.51, + "learning_rate": 5.767813791362809e-06, + "loss": 2.6896, + "step": 2425 + }, + { + "epoch": 0.51, + "learning_rate": 5.764177685563698e-06, + "loss": 2.6398, + "step": 2426 + }, + { + "epoch": 0.51, + "learning_rate": 5.760541165915722e-06, + "loss": 2.6463, + "step": 2427 + }, + { + "epoch": 0.51, + "learning_rate": 5.756904234388279e-06, + "loss": 2.6145, + "step": 2428 + }, + { + "epoch": 0.51, + "learning_rate": 5.753266892950989e-06, + "loss": 2.7164, + "step": 2429 + }, + { + "epoch": 0.51, + "learning_rate": 5.749629143573694e-06, + "loss": 2.6348, + "step": 2430 + }, + { + "epoch": 0.51, + "learning_rate": 5.745990988226457e-06, + "loss": 2.7369, + "step": 2431 + }, + { + "epoch": 0.51, + "learning_rate": 5.742352428879565e-06, + "loss": 2.5706, + "step": 2432 + }, + { + "epoch": 0.51, + "learning_rate": 5.738713467503515e-06, + "loss": 2.659, + "step": 2433 + }, + { + "epoch": 0.51, + "learning_rate": 5.735074106069029e-06, + "loss": 2.642, + "step": 2434 + }, + { + "epoch": 0.51, + "learning_rate": 5.731434346547045e-06, + "loss": 2.6857, + "step": 2435 + }, + { + "epoch": 0.51, + "learning_rate": 5.727794190908712e-06, + "loss": 2.5799, + "step": 2436 + }, + { + "epoch": 0.51, + "learning_rate": 5.724153641125398e-06, + "loss": 2.679, + "step": 2437 + }, + { + "epoch": 0.51, + "learning_rate": 5.7205126991686825e-06, + "loss": 2.5499, + "step": 2438 + }, + { + "epoch": 0.51, + "learning_rate": 5.716871367010358e-06, + "loss": 2.6176, + "step": 2439 + }, + { + "epoch": 0.51, + "learning_rate": 5.713229646622428e-06, + "loss": 2.6543, + "step": 2440 + }, + { + "epoch": 0.51, + "learning_rate": 5.709587539977105e-06, + "loss": 2.6345, + "step": 2441 + }, + { + "epoch": 0.51, + "learning_rate": 5.705945049046817e-06, + "loss": 2.6389, + "step": 2442 + }, + { + "epoch": 0.51, + "learning_rate": 5.702302175804189e-06, + "loss": 2.6343, + "step": 2443 + }, + { + "epoch": 0.51, + "learning_rate": 5.698658922222062e-06, + "loss": 2.6712, + "step": 2444 + }, + { + "epoch": 0.51, + "learning_rate": 5.695015290273481e-06, + "loss": 2.6287, + "step": 2445 + }, + { + "epoch": 0.51, + "learning_rate": 5.6913712819316945e-06, + "loss": 2.6477, + "step": 2446 + }, + { + "epoch": 0.51, + "learning_rate": 5.687726899170155e-06, + "loss": 2.6042, + "step": 2447 + }, + { + "epoch": 0.51, + "learning_rate": 5.68408214396252e-06, + "loss": 2.5996, + "step": 2448 + }, + { + "epoch": 0.51, + "learning_rate": 5.680437018282643e-06, + "loss": 2.666, + "step": 2449 + }, + { + "epoch": 0.51, + "learning_rate": 5.6767915241045855e-06, + "loss": 2.6532, + "step": 2450 + }, + { + "epoch": 0.51, + "learning_rate": 5.673145663402605e-06, + "loss": 2.6632, + "step": 2451 + }, + { + "epoch": 0.51, + "learning_rate": 5.669499438151156e-06, + "loss": 2.6673, + "step": 2452 + }, + { + "epoch": 0.51, + "learning_rate": 5.665852850324893e-06, + "loss": 2.6246, + "step": 2453 + }, + { + "epoch": 0.51, + "learning_rate": 5.662205901898665e-06, + "loss": 2.7087, + "step": 2454 + }, + { + "epoch": 0.51, + "learning_rate": 5.65855859484752e-06, + "loss": 2.6647, + "step": 2455 + }, + { + "epoch": 0.51, + "learning_rate": 5.654910931146692e-06, + "loss": 2.5847, + "step": 2456 + }, + { + "epoch": 0.52, + "learning_rate": 5.6512629127716205e-06, + "loss": 2.6186, + "step": 2457 + }, + { + "epoch": 0.52, + "learning_rate": 5.647614541697925e-06, + "loss": 2.6399, + "step": 2458 + }, + { + "epoch": 0.52, + "learning_rate": 5.64396581990142e-06, + "loss": 2.6116, + "step": 2459 + }, + { + "epoch": 0.52, + "learning_rate": 5.640316749358118e-06, + "loss": 2.6092, + "step": 2460 + }, + { + "epoch": 0.52, + "learning_rate": 5.636667332044208e-06, + "loss": 2.6949, + "step": 2461 + }, + { + "epoch": 0.52, + "learning_rate": 5.633017569936071e-06, + "loss": 2.5703, + "step": 2462 + }, + { + "epoch": 0.52, + "learning_rate": 5.62936746501028e-06, + "loss": 2.6159, + "step": 2463 + }, + { + "epoch": 0.52, + "learning_rate": 5.625717019243589e-06, + "loss": 2.6119, + "step": 2464 + }, + { + "epoch": 0.52, + "learning_rate": 5.622066234612936e-06, + "loss": 2.5863, + "step": 2465 + }, + { + "epoch": 0.52, + "learning_rate": 5.618415113095443e-06, + "loss": 2.6541, + "step": 2466 + }, + { + "epoch": 0.52, + "learning_rate": 5.614763656668417e-06, + "loss": 2.6367, + "step": 2467 + }, + { + "epoch": 0.52, + "learning_rate": 5.611111867309344e-06, + "loss": 2.6824, + "step": 2468 + }, + { + "epoch": 0.52, + "learning_rate": 5.60745974699589e-06, + "loss": 2.681, + "step": 2469 + }, + { + "epoch": 0.52, + "learning_rate": 5.603807297705902e-06, + "loss": 2.577, + "step": 2470 + }, + { + "epoch": 0.52, + "learning_rate": 5.600154521417405e-06, + "loss": 2.5927, + "step": 2471 + }, + { + "epoch": 0.52, + "learning_rate": 5.5965014201085974e-06, + "loss": 2.6292, + "step": 2472 + }, + { + "epoch": 0.52, + "learning_rate": 5.59284799575786e-06, + "loss": 2.6953, + "step": 2473 + }, + { + "epoch": 0.52, + "learning_rate": 5.589194250343741e-06, + "loss": 2.6067, + "step": 2474 + }, + { + "epoch": 0.52, + "learning_rate": 5.585540185844972e-06, + "loss": 2.6833, + "step": 2475 + }, + { + "epoch": 0.52, + "learning_rate": 5.581885804240448e-06, + "loss": 2.5805, + "step": 2476 + }, + { + "epoch": 0.52, + "learning_rate": 5.57823110750924e-06, + "loss": 2.6857, + "step": 2477 + }, + { + "epoch": 0.52, + "learning_rate": 5.5745760976305914e-06, + "loss": 2.6521, + "step": 2478 + }, + { + "epoch": 0.52, + "learning_rate": 5.570920776583911e-06, + "loss": 2.6739, + "step": 2479 + }, + { + "epoch": 0.52, + "learning_rate": 5.567265146348779e-06, + "loss": 2.5827, + "step": 2480 + }, + { + "epoch": 0.52, + "learning_rate": 5.563609208904943e-06, + "loss": 2.6722, + "step": 2481 + }, + { + "epoch": 0.52, + "learning_rate": 5.559952966232317e-06, + "loss": 2.6209, + "step": 2482 + }, + { + "epoch": 0.52, + "learning_rate": 5.556296420310977e-06, + "loss": 2.629, + "step": 2483 + }, + { + "epoch": 0.52, + "learning_rate": 5.552639573121168e-06, + "loss": 2.6364, + "step": 2484 + }, + { + "epoch": 0.52, + "learning_rate": 5.5489824266432946e-06, + "loss": 2.5976, + "step": 2485 + }, + { + "epoch": 0.52, + "learning_rate": 5.545324982857926e-06, + "loss": 2.6442, + "step": 2486 + }, + { + "epoch": 0.52, + "learning_rate": 5.541667243745791e-06, + "loss": 2.5686, + "step": 2487 + }, + { + "epoch": 0.52, + "learning_rate": 5.538009211287778e-06, + "loss": 2.6727, + "step": 2488 + }, + { + "epoch": 0.52, + "learning_rate": 5.534350887464934e-06, + "loss": 2.6714, + "step": 2489 + }, + { + "epoch": 0.52, + "learning_rate": 5.530692274258468e-06, + "loss": 2.6478, + "step": 2490 + }, + { + "epoch": 0.52, + "learning_rate": 5.527033373649739e-06, + "loss": 2.673, + "step": 2491 + }, + { + "epoch": 0.52, + "learning_rate": 5.523374187620266e-06, + "loss": 2.6579, + "step": 2492 + }, + { + "epoch": 0.52, + "learning_rate": 5.519714718151724e-06, + "loss": 2.6901, + "step": 2493 + }, + { + "epoch": 0.52, + "learning_rate": 5.516054967225937e-06, + "loss": 2.6773, + "step": 2494 + }, + { + "epoch": 0.52, + "learning_rate": 5.512394936824881e-06, + "loss": 2.7005, + "step": 2495 + }, + { + "epoch": 0.52, + "learning_rate": 5.508734628930691e-06, + "loss": 2.6121, + "step": 2496 + }, + { + "epoch": 0.52, + "learning_rate": 5.505074045525644e-06, + "loss": 2.6181, + "step": 2497 + }, + { + "epoch": 0.52, + "learning_rate": 5.501413188592167e-06, + "loss": 2.6903, + "step": 2498 + }, + { + "epoch": 0.52, + "learning_rate": 5.497752060112843e-06, + "loss": 2.6088, + "step": 2499 + }, + { + "epoch": 0.52, + "learning_rate": 5.494090662070396e-06, + "loss": 2.6302, + "step": 2500 + }, + { + "epoch": 0.52, + "learning_rate": 5.4904289964476905e-06, + "loss": 2.6555, + "step": 2501 + }, + { + "epoch": 0.52, + "learning_rate": 5.486767065227749e-06, + "loss": 2.6441, + "step": 2502 + }, + { + "epoch": 0.52, + "learning_rate": 5.483104870393728e-06, + "loss": 2.6453, + "step": 2503 + }, + { + "epoch": 0.53, + "learning_rate": 5.479442413928927e-06, + "loss": 2.6113, + "step": 2504 + }, + { + "epoch": 0.53, + "learning_rate": 5.475779697816794e-06, + "loss": 2.6856, + "step": 2505 + }, + { + "epoch": 0.53, + "learning_rate": 5.4721167240409115e-06, + "loss": 2.6613, + "step": 2506 + }, + { + "epoch": 0.53, + "learning_rate": 5.468453494585002e-06, + "loss": 2.6456, + "step": 2507 + }, + { + "epoch": 0.53, + "learning_rate": 5.464790011432931e-06, + "loss": 2.6815, + "step": 2508 + }, + { + "epoch": 0.53, + "learning_rate": 5.4611262765686955e-06, + "loss": 2.6554, + "step": 2509 + }, + { + "epoch": 0.53, + "learning_rate": 5.457462291976432e-06, + "loss": 2.6407, + "step": 2510 + }, + { + "epoch": 0.53, + "learning_rate": 5.453798059640414e-06, + "loss": 2.6082, + "step": 2511 + }, + { + "epoch": 0.53, + "learning_rate": 5.450133581545045e-06, + "loss": 2.6094, + "step": 2512 + }, + { + "epoch": 0.53, + "learning_rate": 5.446468859674862e-06, + "loss": 2.677, + "step": 2513 + }, + { + "epoch": 0.53, + "learning_rate": 5.4428038960145405e-06, + "loss": 2.6082, + "step": 2514 + }, + { + "epoch": 0.53, + "learning_rate": 5.439138692548878e-06, + "loss": 2.6142, + "step": 2515 + }, + { + "epoch": 0.53, + "learning_rate": 5.435473251262805e-06, + "loss": 2.6197, + "step": 2516 + }, + { + "epoch": 0.53, + "learning_rate": 5.431807574141386e-06, + "loss": 2.6796, + "step": 2517 + }, + { + "epoch": 0.53, + "learning_rate": 5.428141663169807e-06, + "loss": 2.6332, + "step": 2518 + }, + { + "epoch": 0.53, + "learning_rate": 5.424475520333381e-06, + "loss": 2.624, + "step": 2519 + }, + { + "epoch": 0.53, + "learning_rate": 5.42080914761755e-06, + "loss": 2.6506, + "step": 2520 + }, + { + "epoch": 0.53, + "learning_rate": 5.417142547007879e-06, + "loss": 2.6321, + "step": 2521 + }, + { + "epoch": 0.53, + "learning_rate": 5.4134757204900525e-06, + "loss": 2.6706, + "step": 2522 + }, + { + "epoch": 0.53, + "learning_rate": 5.409808670049884e-06, + "loss": 2.6451, + "step": 2523 + }, + { + "epoch": 0.53, + "learning_rate": 5.4061413976733056e-06, + "loss": 2.7043, + "step": 2524 + }, + { + "epoch": 0.53, + "learning_rate": 5.402473905346368e-06, + "loss": 2.6479, + "step": 2525 + }, + { + "epoch": 0.53, + "learning_rate": 5.398806195055243e-06, + "loss": 2.6864, + "step": 2526 + }, + { + "epoch": 0.53, + "learning_rate": 5.395138268786218e-06, + "loss": 2.5929, + "step": 2527 + }, + { + "epoch": 0.53, + "learning_rate": 5.3914701285257e-06, + "loss": 2.7047, + "step": 2528 + }, + { + "epoch": 0.53, + "learning_rate": 5.387801776260212e-06, + "loss": 2.6497, + "step": 2529 + }, + { + "epoch": 0.53, + "learning_rate": 5.384133213976388e-06, + "loss": 2.5312, + "step": 2530 + }, + { + "epoch": 0.53, + "learning_rate": 5.38046444366098e-06, + "loss": 2.6884, + "step": 2531 + }, + { + "epoch": 0.53, + "learning_rate": 5.376795467300853e-06, + "loss": 2.6398, + "step": 2532 + }, + { + "epoch": 0.53, + "learning_rate": 5.373126286882977e-06, + "loss": 2.5862, + "step": 2533 + }, + { + "epoch": 0.53, + "learning_rate": 5.36945690439444e-06, + "loss": 2.5809, + "step": 2534 + }, + { + "epoch": 0.53, + "learning_rate": 5.365787321822438e-06, + "loss": 2.6474, + "step": 2535 + }, + { + "epoch": 0.53, + "learning_rate": 5.3621175411542725e-06, + "loss": 2.6637, + "step": 2536 + }, + { + "epoch": 0.53, + "learning_rate": 5.358447564377352e-06, + "loss": 2.6195, + "step": 2537 + }, + { + "epoch": 0.53, + "learning_rate": 5.354777393479195e-06, + "loss": 2.6523, + "step": 2538 + }, + { + "epoch": 0.53, + "learning_rate": 5.351107030447423e-06, + "loss": 2.6497, + "step": 2539 + }, + { + "epoch": 0.53, + "learning_rate": 5.34743647726976e-06, + "loss": 2.6529, + "step": 2540 + }, + { + "epoch": 0.53, + "learning_rate": 5.343765735934039e-06, + "loss": 2.5929, + "step": 2541 + }, + { + "epoch": 0.53, + "learning_rate": 5.340094808428184e-06, + "loss": 2.5728, + "step": 2542 + }, + { + "epoch": 0.53, + "learning_rate": 5.336423696740233e-06, + "loss": 2.6584, + "step": 2543 + }, + { + "epoch": 0.53, + "learning_rate": 5.332752402858312e-06, + "loss": 2.6022, + "step": 2544 + }, + { + "epoch": 0.53, + "learning_rate": 5.329080928770652e-06, + "loss": 2.6774, + "step": 2545 + }, + { + "epoch": 0.53, + "learning_rate": 5.325409276465581e-06, + "loss": 2.5887, + "step": 2546 + }, + { + "epoch": 0.53, + "learning_rate": 5.321737447931524e-06, + "loss": 2.6645, + "step": 2547 + }, + { + "epoch": 0.53, + "learning_rate": 5.318065445156999e-06, + "loss": 2.6939, + "step": 2548 + }, + { + "epoch": 0.53, + "learning_rate": 5.314393270130617e-06, + "loss": 2.6509, + "step": 2549 + }, + { + "epoch": 0.53, + "learning_rate": 5.310720924841092e-06, + "loss": 2.7073, + "step": 2550 + }, + { + "epoch": 0.53, + "learning_rate": 5.307048411277217e-06, + "loss": 2.6084, + "step": 2551 + }, + { + "epoch": 0.54, + "learning_rate": 5.303375731427882e-06, + "loss": 2.6479, + "step": 2552 + }, + { + "epoch": 0.54, + "learning_rate": 5.299702887282072e-06, + "loss": 2.6244, + "step": 2553 + }, + { + "epoch": 0.54, + "learning_rate": 5.296029880828854e-06, + "loss": 2.5832, + "step": 2554 + }, + { + "epoch": 0.54, + "learning_rate": 5.292356714057382e-06, + "loss": 2.5826, + "step": 2555 + }, + { + "epoch": 0.54, + "learning_rate": 5.288683388956904e-06, + "loss": 2.6446, + "step": 2556 + }, + { + "epoch": 0.54, + "learning_rate": 5.285009907516752e-06, + "loss": 2.5582, + "step": 2557 + }, + { + "epoch": 0.54, + "learning_rate": 5.281336271726333e-06, + "loss": 2.6452, + "step": 2558 + }, + { + "epoch": 0.54, + "learning_rate": 5.27766248357515e-06, + "loss": 2.6506, + "step": 2559 + }, + { + "epoch": 0.54, + "learning_rate": 5.273988545052784e-06, + "loss": 2.6735, + "step": 2560 + }, + { + "epoch": 0.54, + "learning_rate": 5.270314458148896e-06, + "loss": 2.6301, + "step": 2561 + }, + { + "epoch": 0.54, + "learning_rate": 5.2666402248532255e-06, + "loss": 2.7112, + "step": 2562 + }, + { + "epoch": 0.54, + "learning_rate": 5.262965847155599e-06, + "loss": 2.6683, + "step": 2563 + }, + { + "epoch": 0.54, + "learning_rate": 5.259291327045912e-06, + "loss": 2.6402, + "step": 2564 + }, + { + "epoch": 0.54, + "learning_rate": 5.255616666514146e-06, + "loss": 2.6096, + "step": 2565 + }, + { + "epoch": 0.54, + "learning_rate": 5.2519418675503495e-06, + "loss": 2.6493, + "step": 2566 + }, + { + "epoch": 0.54, + "learning_rate": 5.248266932144652e-06, + "loss": 2.5897, + "step": 2567 + }, + { + "epoch": 0.54, + "learning_rate": 5.244591862287259e-06, + "loss": 2.6827, + "step": 2568 + }, + { + "epoch": 0.54, + "learning_rate": 5.240916659968438e-06, + "loss": 2.6279, + "step": 2569 + }, + { + "epoch": 0.54, + "learning_rate": 5.23724132717854e-06, + "loss": 2.6363, + "step": 2570 + }, + { + "epoch": 0.54, + "learning_rate": 5.233565865907982e-06, + "loss": 2.5867, + "step": 2571 + }, + { + "epoch": 0.54, + "learning_rate": 5.229890278147252e-06, + "loss": 2.6115, + "step": 2572 + }, + { + "epoch": 0.54, + "learning_rate": 5.2262145658869005e-06, + "loss": 2.6322, + "step": 2573 + }, + { + "epoch": 0.54, + "learning_rate": 5.222538731117553e-06, + "loss": 2.6147, + "step": 2574 + }, + { + "epoch": 0.54, + "learning_rate": 5.2188627758298995e-06, + "loss": 2.6418, + "step": 2575 + }, + { + "epoch": 0.54, + "learning_rate": 5.215186702014692e-06, + "loss": 2.6707, + "step": 2576 + }, + { + "epoch": 0.54, + "learning_rate": 5.211510511662749e-06, + "loss": 2.5697, + "step": 2577 + }, + { + "epoch": 0.54, + "learning_rate": 5.207834206764953e-06, + "loss": 2.6145, + "step": 2578 + }, + { + "epoch": 0.54, + "learning_rate": 5.204157789312248e-06, + "loss": 2.6999, + "step": 2579 + }, + { + "epoch": 0.54, + "learning_rate": 5.200481261295637e-06, + "loss": 2.6791, + "step": 2580 + }, + { + "epoch": 0.54, + "learning_rate": 5.196804624706185e-06, + "loss": 2.6875, + "step": 2581 + }, + { + "epoch": 0.54, + "learning_rate": 5.193127881535015e-06, + "loss": 2.6544, + "step": 2582 + }, + { + "epoch": 0.54, + "learning_rate": 5.189451033773309e-06, + "loss": 2.5457, + "step": 2583 + }, + { + "epoch": 0.54, + "learning_rate": 5.185774083412301e-06, + "loss": 2.6331, + "step": 2584 + }, + { + "epoch": 0.54, + "learning_rate": 5.182097032443288e-06, + "loss": 2.6607, + "step": 2585 + }, + { + "epoch": 0.54, + "learning_rate": 5.178419882857617e-06, + "loss": 2.5994, + "step": 2586 + }, + { + "epoch": 0.54, + "learning_rate": 5.174742636646688e-06, + "loss": 2.7047, + "step": 2587 + }, + { + "epoch": 0.54, + "learning_rate": 5.1710652958019525e-06, + "loss": 2.6143, + "step": 2588 + }, + { + "epoch": 0.54, + "learning_rate": 5.1673878623149185e-06, + "loss": 2.5906, + "step": 2589 + }, + { + "epoch": 0.54, + "learning_rate": 5.16371033817714e-06, + "loss": 2.6978, + "step": 2590 + }, + { + "epoch": 0.54, + "learning_rate": 5.1600327253802184e-06, + "loss": 2.598, + "step": 2591 + }, + { + "epoch": 0.54, + "learning_rate": 5.156355025915808e-06, + "loss": 2.6673, + "step": 2592 + }, + { + "epoch": 0.54, + "learning_rate": 5.152677241775606e-06, + "loss": 2.5752, + "step": 2593 + }, + { + "epoch": 0.54, + "learning_rate": 5.1489993749513576e-06, + "loss": 2.6309, + "step": 2594 + }, + { + "epoch": 0.54, + "learning_rate": 5.145321427434853e-06, + "loss": 2.6171, + "step": 2595 + }, + { + "epoch": 0.54, + "learning_rate": 5.141643401217924e-06, + "loss": 2.6275, + "step": 2596 + }, + { + "epoch": 0.54, + "learning_rate": 5.1379652982924465e-06, + "loss": 2.6561, + "step": 2597 + }, + { + "epoch": 0.54, + "learning_rate": 5.134287120650338e-06, + "loss": 2.6717, + "step": 2598 + }, + { + "epoch": 0.54, + "learning_rate": 5.130608870283558e-06, + "loss": 2.6551, + "step": 2599 + }, + { + "epoch": 0.55, + "learning_rate": 5.1269305491841015e-06, + "loss": 2.6425, + "step": 2600 + }, + { + "epoch": 0.55, + "learning_rate": 5.123252159344003e-06, + "loss": 2.5902, + "step": 2601 + }, + { + "epoch": 0.55, + "learning_rate": 5.119573702755338e-06, + "loss": 2.6156, + "step": 2602 + }, + { + "epoch": 0.55, + "learning_rate": 5.115895181410213e-06, + "loss": 2.6504, + "step": 2603 + }, + { + "epoch": 0.55, + "learning_rate": 5.112216597300773e-06, + "loss": 2.6366, + "step": 2604 + }, + { + "epoch": 0.55, + "learning_rate": 5.108537952419195e-06, + "loss": 2.6201, + "step": 2605 + }, + { + "epoch": 0.55, + "learning_rate": 5.10485924875769e-06, + "loss": 2.6108, + "step": 2606 + }, + { + "epoch": 0.55, + "learning_rate": 5.1011804883085e-06, + "loss": 2.6453, + "step": 2607 + }, + { + "epoch": 0.55, + "learning_rate": 5.097501673063903e-06, + "loss": 2.6932, + "step": 2608 + }, + { + "epoch": 0.55, + "learning_rate": 5.093822805016194e-06, + "loss": 2.6402, + "step": 2609 + }, + { + "epoch": 0.55, + "learning_rate": 5.0901438861577115e-06, + "loss": 2.6509, + "step": 2610 + }, + { + "epoch": 0.55, + "learning_rate": 5.08646491848081e-06, + "loss": 2.6189, + "step": 2611 + }, + { + "epoch": 0.55, + "learning_rate": 5.0827859039778784e-06, + "loss": 2.7467, + "step": 2612 + }, + { + "epoch": 0.55, + "learning_rate": 5.079106844641326e-06, + "loss": 2.6668, + "step": 2613 + }, + { + "epoch": 0.55, + "learning_rate": 5.0754277424635885e-06, + "loss": 2.6845, + "step": 2614 + }, + { + "epoch": 0.55, + "learning_rate": 5.071748599437124e-06, + "loss": 2.6529, + "step": 2615 + }, + { + "epoch": 0.55, + "learning_rate": 5.068069417554413e-06, + "loss": 2.6907, + "step": 2616 + }, + { + "epoch": 0.55, + "learning_rate": 5.064390198807957e-06, + "loss": 2.62, + "step": 2617 + }, + { + "epoch": 0.55, + "learning_rate": 5.060710945190278e-06, + "loss": 2.6168, + "step": 2618 + }, + { + "epoch": 0.55, + "learning_rate": 5.057031658693916e-06, + "loss": 2.6478, + "step": 2619 + }, + { + "epoch": 0.55, + "learning_rate": 5.053352341311431e-06, + "loss": 2.6196, + "step": 2620 + }, + { + "epoch": 0.55, + "learning_rate": 5.049672995035394e-06, + "loss": 2.6501, + "step": 2621 + }, + { + "epoch": 0.55, + "learning_rate": 5.045993621858399e-06, + "loss": 2.679, + "step": 2622 + }, + { + "epoch": 0.55, + "learning_rate": 5.04231422377305e-06, + "loss": 2.6292, + "step": 2623 + }, + { + "epoch": 0.55, + "learning_rate": 5.038634802771966e-06, + "loss": 2.6192, + "step": 2624 + }, + { + "epoch": 0.55, + "learning_rate": 5.0349553608477766e-06, + "loss": 2.6169, + "step": 2625 + }, + { + "epoch": 0.55, + "learning_rate": 5.031275899993128e-06, + "loss": 2.7057, + "step": 2626 + }, + { + "epoch": 0.55, + "learning_rate": 5.027596422200668e-06, + "loss": 2.6379, + "step": 2627 + }, + { + "epoch": 0.55, + "learning_rate": 5.023916929463062e-06, + "loss": 2.6364, + "step": 2628 + }, + { + "epoch": 0.55, + "learning_rate": 5.020237423772979e-06, + "loss": 2.6285, + "step": 2629 + }, + { + "epoch": 0.55, + "learning_rate": 5.016557907123095e-06, + "loss": 2.6372, + "step": 2630 + }, + { + "epoch": 0.55, + "learning_rate": 5.012878381506095e-06, + "loss": 2.6538, + "step": 2631 + }, + { + "epoch": 0.55, + "learning_rate": 5.009198848914664e-06, + "loss": 2.6483, + "step": 2632 + }, + { + "epoch": 0.55, + "learning_rate": 5.005519311341495e-06, + "loss": 2.5946, + "step": 2633 + }, + { + "epoch": 0.55, + "learning_rate": 5.001839770779281e-06, + "loss": 2.6178, + "step": 2634 + }, + { + "epoch": 0.55, + "learning_rate": 4.99816022922072e-06, + "loss": 2.6244, + "step": 2635 + }, + { + "epoch": 0.55, + "learning_rate": 4.994480688658508e-06, + "loss": 2.6213, + "step": 2636 + }, + { + "epoch": 0.55, + "learning_rate": 4.990801151085337e-06, + "loss": 2.6231, + "step": 2637 + }, + { + "epoch": 0.55, + "learning_rate": 4.987121618493906e-06, + "loss": 2.5707, + "step": 2638 + }, + { + "epoch": 0.55, + "learning_rate": 4.983442092876906e-06, + "loss": 2.6517, + "step": 2639 + }, + { + "epoch": 0.55, + "learning_rate": 4.9797625762270225e-06, + "loss": 2.6613, + "step": 2640 + }, + { + "epoch": 0.55, + "learning_rate": 4.9760830705369396e-06, + "loss": 2.6642, + "step": 2641 + }, + { + "epoch": 0.55, + "learning_rate": 4.972403577799334e-06, + "loss": 2.6223, + "step": 2642 + }, + { + "epoch": 0.55, + "learning_rate": 4.968724100006873e-06, + "loss": 2.6738, + "step": 2643 + }, + { + "epoch": 0.55, + "learning_rate": 4.965044639152223e-06, + "loss": 2.6924, + "step": 2644 + }, + { + "epoch": 0.55, + "learning_rate": 4.961365197228035e-06, + "loss": 2.626, + "step": 2645 + }, + { + "epoch": 0.55, + "learning_rate": 4.957685776226952e-06, + "loss": 2.6126, + "step": 2646 + }, + { + "epoch": 0.56, + "learning_rate": 4.954006378141604e-06, + "loss": 2.6539, + "step": 2647 + }, + { + "epoch": 0.56, + "learning_rate": 4.950327004964607e-06, + "loss": 2.6296, + "step": 2648 + }, + { + "epoch": 0.56, + "learning_rate": 4.946647658688572e-06, + "loss": 2.6143, + "step": 2649 + }, + { + "epoch": 0.56, + "learning_rate": 4.942968341306085e-06, + "loss": 2.6832, + "step": 2650 + }, + { + "epoch": 0.56, + "learning_rate": 4.9392890548097235e-06, + "loss": 2.6492, + "step": 2651 + }, + { + "epoch": 0.56, + "learning_rate": 4.935609801192044e-06, + "loss": 2.5444, + "step": 2652 + }, + { + "epoch": 0.56, + "learning_rate": 4.93193058244559e-06, + "loss": 2.677, + "step": 2653 + }, + { + "epoch": 0.56, + "learning_rate": 4.928251400562878e-06, + "loss": 2.6983, + "step": 2654 + }, + { + "epoch": 0.56, + "learning_rate": 4.924572257536413e-06, + "loss": 2.6177, + "step": 2655 + }, + { + "epoch": 0.56, + "learning_rate": 4.920893155358676e-06, + "loss": 2.6705, + "step": 2656 + }, + { + "epoch": 0.56, + "learning_rate": 4.917214096022123e-06, + "loss": 2.6819, + "step": 2657 + }, + { + "epoch": 0.56, + "learning_rate": 4.91353508151919e-06, + "loss": 2.7093, + "step": 2658 + }, + { + "epoch": 0.56, + "learning_rate": 4.909856113842289e-06, + "loss": 2.6954, + "step": 2659 + }, + { + "epoch": 0.56, + "learning_rate": 4.906177194983807e-06, + "loss": 2.6387, + "step": 2660 + }, + { + "epoch": 0.56, + "learning_rate": 4.9024983269360994e-06, + "loss": 2.601, + "step": 2661 + }, + { + "epoch": 0.56, + "learning_rate": 4.8988195116915005e-06, + "loss": 2.6675, + "step": 2662 + }, + { + "epoch": 0.56, + "learning_rate": 4.8951407512423125e-06, + "loss": 2.6344, + "step": 2663 + }, + { + "epoch": 0.56, + "learning_rate": 4.891462047580807e-06, + "loss": 2.6099, + "step": 2664 + }, + { + "epoch": 0.56, + "learning_rate": 4.887783402699229e-06, + "loss": 2.6194, + "step": 2665 + }, + { + "epoch": 0.56, + "learning_rate": 4.884104818589788e-06, + "loss": 2.6732, + "step": 2666 + }, + { + "epoch": 0.56, + "learning_rate": 4.8804262972446645e-06, + "loss": 2.601, + "step": 2667 + }, + { + "epoch": 0.56, + "learning_rate": 4.876747840655999e-06, + "loss": 2.6706, + "step": 2668 + }, + { + "epoch": 0.56, + "learning_rate": 4.8730694508159e-06, + "loss": 2.6587, + "step": 2669 + }, + { + "epoch": 0.56, + "learning_rate": 4.869391129716443e-06, + "loss": 2.5815, + "step": 2670 + }, + { + "epoch": 0.56, + "learning_rate": 4.8657128793496625e-06, + "loss": 2.6313, + "step": 2671 + }, + { + "epoch": 0.56, + "learning_rate": 4.862034701707554e-06, + "loss": 2.6574, + "step": 2672 + }, + { + "epoch": 0.56, + "learning_rate": 4.858356598782076e-06, + "loss": 2.7188, + "step": 2673 + }, + { + "epoch": 0.56, + "learning_rate": 4.854678572565147e-06, + "loss": 2.6554, + "step": 2674 + }, + { + "epoch": 0.56, + "learning_rate": 4.851000625048643e-06, + "loss": 2.6532, + "step": 2675 + }, + { + "epoch": 0.56, + "learning_rate": 4.847322758224396e-06, + "loss": 2.5926, + "step": 2676 + }, + { + "epoch": 0.56, + "learning_rate": 4.843644974084194e-06, + "loss": 2.5808, + "step": 2677 + }, + { + "epoch": 0.56, + "learning_rate": 4.839967274619783e-06, + "loss": 2.5719, + "step": 2678 + }, + { + "epoch": 0.56, + "learning_rate": 4.8362896618228614e-06, + "loss": 2.5957, + "step": 2679 + }, + { + "epoch": 0.56, + "learning_rate": 4.8326121376850815e-06, + "loss": 2.6435, + "step": 2680 + }, + { + "epoch": 0.56, + "learning_rate": 4.8289347041980475e-06, + "loss": 2.635, + "step": 2681 + }, + { + "epoch": 0.56, + "learning_rate": 4.825257363353314e-06, + "loss": 2.666, + "step": 2682 + }, + { + "epoch": 0.56, + "learning_rate": 4.821580117142385e-06, + "loss": 2.6037, + "step": 2683 + }, + { + "epoch": 0.56, + "learning_rate": 4.817902967556714e-06, + "loss": 2.6008, + "step": 2684 + }, + { + "epoch": 0.56, + "learning_rate": 4.814225916587699e-06, + "loss": 2.6683, + "step": 2685 + }, + { + "epoch": 0.56, + "learning_rate": 4.810548966226693e-06, + "loss": 2.674, + "step": 2686 + }, + { + "epoch": 0.56, + "learning_rate": 4.806872118464987e-06, + "loss": 2.6337, + "step": 2687 + }, + { + "epoch": 0.56, + "learning_rate": 4.803195375293815e-06, + "loss": 2.5675, + "step": 2688 + }, + { + "epoch": 0.56, + "learning_rate": 4.799518738704366e-06, + "loss": 2.641, + "step": 2689 + }, + { + "epoch": 0.56, + "learning_rate": 4.795842210687754e-06, + "loss": 2.6769, + "step": 2690 + }, + { + "epoch": 0.56, + "learning_rate": 4.792165793235048e-06, + "loss": 2.694, + "step": 2691 + }, + { + "epoch": 0.56, + "learning_rate": 4.788489488337252e-06, + "loss": 2.6154, + "step": 2692 + }, + { + "epoch": 0.56, + "learning_rate": 4.78481329798531e-06, + "loss": 2.6172, + "step": 2693 + }, + { + "epoch": 0.56, + "learning_rate": 4.781137224170101e-06, + "loss": 2.6427, + "step": 2694 + }, + { + "epoch": 0.57, + "learning_rate": 4.7774612688824475e-06, + "loss": 2.7034, + "step": 2695 + }, + { + "epoch": 0.57, + "learning_rate": 4.773785434113101e-06, + "loss": 2.6881, + "step": 2696 + }, + { + "epoch": 0.57, + "learning_rate": 4.770109721852751e-06, + "loss": 2.5953, + "step": 2697 + }, + { + "epoch": 0.57, + "learning_rate": 4.766434134092019e-06, + "loss": 2.6672, + "step": 2698 + }, + { + "epoch": 0.57, + "learning_rate": 4.7627586728214606e-06, + "loss": 2.6434, + "step": 2699 + }, + { + "epoch": 0.57, + "learning_rate": 4.759083340031563e-06, + "loss": 2.6899, + "step": 2700 + }, + { + "epoch": 0.57, + "learning_rate": 4.755408137712744e-06, + "loss": 2.6115, + "step": 2701 + }, + { + "epoch": 0.57, + "learning_rate": 4.751733067855348e-06, + "loss": 2.6062, + "step": 2702 + }, + { + "epoch": 0.57, + "learning_rate": 4.7480581324496505e-06, + "loss": 2.6332, + "step": 2703 + }, + { + "epoch": 0.57, + "learning_rate": 4.744383333485857e-06, + "loss": 2.6098, + "step": 2704 + }, + { + "epoch": 0.57, + "learning_rate": 4.74070867295409e-06, + "loss": 2.7142, + "step": 2705 + }, + { + "epoch": 0.57, + "learning_rate": 4.737034152844403e-06, + "loss": 2.6297, + "step": 2706 + }, + { + "epoch": 0.57, + "learning_rate": 4.733359775146775e-06, + "loss": 2.6183, + "step": 2707 + }, + { + "epoch": 0.57, + "learning_rate": 4.729685541851107e-06, + "loss": 2.6359, + "step": 2708 + }, + { + "epoch": 0.57, + "learning_rate": 4.726011454947217e-06, + "loss": 2.6655, + "step": 2709 + }, + { + "epoch": 0.57, + "learning_rate": 4.72233751642485e-06, + "loss": 2.5718, + "step": 2710 + }, + { + "epoch": 0.57, + "learning_rate": 4.718663728273669e-06, + "loss": 2.6863, + "step": 2711 + }, + { + "epoch": 0.57, + "learning_rate": 4.714990092483252e-06, + "loss": 2.6085, + "step": 2712 + }, + { + "epoch": 0.57, + "learning_rate": 4.7113166110430965e-06, + "loss": 2.6069, + "step": 2713 + }, + { + "epoch": 0.57, + "learning_rate": 4.707643285942619e-06, + "loss": 2.6694, + "step": 2714 + }, + { + "epoch": 0.57, + "learning_rate": 4.703970119171148e-06, + "loss": 2.6342, + "step": 2715 + }, + { + "epoch": 0.57, + "learning_rate": 4.7002971127179284e-06, + "loss": 2.6267, + "step": 2716 + }, + { + "epoch": 0.57, + "learning_rate": 4.696624268572118e-06, + "loss": 2.6858, + "step": 2717 + }, + { + "epoch": 0.57, + "learning_rate": 4.692951588722786e-06, + "loss": 2.6371, + "step": 2718 + }, + { + "epoch": 0.57, + "learning_rate": 4.68927907515891e-06, + "loss": 2.6401, + "step": 2719 + }, + { + "epoch": 0.57, + "learning_rate": 4.6856067298693834e-06, + "loss": 2.71, + "step": 2720 + }, + { + "epoch": 0.57, + "learning_rate": 4.6819345548430026e-06, + "loss": 2.7189, + "step": 2721 + }, + { + "epoch": 0.57, + "learning_rate": 4.678262552068477e-06, + "loss": 2.6238, + "step": 2722 + }, + { + "epoch": 0.57, + "learning_rate": 4.674590723534419e-06, + "loss": 2.6708, + "step": 2723 + }, + { + "epoch": 0.57, + "learning_rate": 4.670919071229348e-06, + "loss": 2.5854, + "step": 2724 + }, + { + "epoch": 0.57, + "learning_rate": 4.6672475971416906e-06, + "loss": 2.6796, + "step": 2725 + }, + { + "epoch": 0.57, + "learning_rate": 4.6635763032597704e-06, + "loss": 2.6404, + "step": 2726 + }, + { + "epoch": 0.57, + "learning_rate": 4.659905191571817e-06, + "loss": 2.624, + "step": 2727 + }, + { + "epoch": 0.57, + "learning_rate": 4.656234264065963e-06, + "loss": 2.6504, + "step": 2728 + }, + { + "epoch": 0.57, + "learning_rate": 4.65256352273024e-06, + "loss": 2.5934, + "step": 2729 + }, + { + "epoch": 0.57, + "learning_rate": 4.648892969552578e-06, + "loss": 2.6744, + "step": 2730 + }, + { + "epoch": 0.57, + "learning_rate": 4.645222606520806e-06, + "loss": 2.6263, + "step": 2731 + }, + { + "epoch": 0.57, + "learning_rate": 4.641552435622651e-06, + "loss": 2.5915, + "step": 2732 + }, + { + "epoch": 0.57, + "learning_rate": 4.637882458845731e-06, + "loss": 2.6411, + "step": 2733 + }, + { + "epoch": 0.57, + "learning_rate": 4.6342126781775644e-06, + "loss": 2.67, + "step": 2734 + }, + { + "epoch": 0.57, + "learning_rate": 4.630543095605562e-06, + "loss": 2.6319, + "step": 2735 + }, + { + "epoch": 0.57, + "learning_rate": 4.626873713117025e-06, + "loss": 2.6217, + "step": 2736 + }, + { + "epoch": 0.57, + "learning_rate": 4.6232045326991494e-06, + "loss": 2.6368, + "step": 2737 + }, + { + "epoch": 0.57, + "learning_rate": 4.619535556339021e-06, + "loss": 2.6649, + "step": 2738 + }, + { + "epoch": 0.57, + "learning_rate": 4.615866786023613e-06, + "loss": 2.6696, + "step": 2739 + }, + { + "epoch": 0.57, + "learning_rate": 4.6121982237397915e-06, + "loss": 2.6889, + "step": 2740 + }, + { + "epoch": 0.57, + "learning_rate": 4.6085298714743025e-06, + "loss": 2.668, + "step": 2741 + }, + { + "epoch": 0.57, + "learning_rate": 4.6048617312137834e-06, + "loss": 2.5984, + "step": 2742 + }, + { + "epoch": 0.58, + "learning_rate": 4.601193804944759e-06, + "loss": 2.5899, + "step": 2743 + }, + { + "epoch": 0.58, + "learning_rate": 4.597526094653633e-06, + "loss": 2.6365, + "step": 2744 + }, + { + "epoch": 0.58, + "learning_rate": 4.593858602326694e-06, + "loss": 2.5973, + "step": 2745 + }, + { + "epoch": 0.58, + "learning_rate": 4.590191329950116e-06, + "loss": 2.6961, + "step": 2746 + }, + { + "epoch": 0.58, + "learning_rate": 4.58652427950995e-06, + "loss": 2.6349, + "step": 2747 + }, + { + "epoch": 0.58, + "learning_rate": 4.582857452992124e-06, + "loss": 2.6439, + "step": 2748 + }, + { + "epoch": 0.58, + "learning_rate": 4.579190852382452e-06, + "loss": 2.6593, + "step": 2749 + }, + { + "epoch": 0.58, + "learning_rate": 4.575524479666621e-06, + "loss": 2.5971, + "step": 2750 + }, + { + "epoch": 0.58, + "learning_rate": 4.571858336830194e-06, + "loss": 2.6283, + "step": 2751 + }, + { + "epoch": 0.58, + "learning_rate": 4.568192425858614e-06, + "loss": 2.6289, + "step": 2752 + }, + { + "epoch": 0.58, + "learning_rate": 4.564526748737195e-06, + "loss": 2.6593, + "step": 2753 + }, + { + "epoch": 0.58, + "learning_rate": 4.560861307451125e-06, + "loss": 2.6801, + "step": 2754 + }, + { + "epoch": 0.58, + "learning_rate": 4.557196103985462e-06, + "loss": 2.6767, + "step": 2755 + }, + { + "epoch": 0.58, + "learning_rate": 4.553531140325139e-06, + "loss": 2.6749, + "step": 2756 + }, + { + "epoch": 0.58, + "learning_rate": 4.549866418454957e-06, + "loss": 2.5995, + "step": 2757 + }, + { + "epoch": 0.58, + "learning_rate": 4.546201940359587e-06, + "loss": 2.6772, + "step": 2758 + }, + { + "epoch": 0.58, + "learning_rate": 4.542537708023569e-06, + "loss": 2.6988, + "step": 2759 + }, + { + "epoch": 0.58, + "learning_rate": 4.538873723431306e-06, + "loss": 2.6041, + "step": 2760 + }, + { + "epoch": 0.58, + "learning_rate": 4.535209988567072e-06, + "loss": 2.661, + "step": 2761 + }, + { + "epoch": 0.58, + "learning_rate": 4.531546505415e-06, + "loss": 2.6549, + "step": 2762 + }, + { + "epoch": 0.58, + "learning_rate": 4.52788327595909e-06, + "loss": 2.6817, + "step": 2763 + }, + { + "epoch": 0.58, + "learning_rate": 4.524220302183207e-06, + "loss": 2.7001, + "step": 2764 + }, + { + "epoch": 0.58, + "learning_rate": 4.520557586071074e-06, + "loss": 2.6744, + "step": 2765 + }, + { + "epoch": 0.58, + "learning_rate": 4.5168951296062735e-06, + "loss": 2.6724, + "step": 2766 + }, + { + "epoch": 0.58, + "learning_rate": 4.5132329347722515e-06, + "loss": 2.6985, + "step": 2767 + }, + { + "epoch": 0.58, + "learning_rate": 4.509571003552311e-06, + "loss": 2.6344, + "step": 2768 + }, + { + "epoch": 0.58, + "learning_rate": 4.5059093379296075e-06, + "loss": 2.6822, + "step": 2769 + }, + { + "epoch": 0.58, + "learning_rate": 4.502247939887158e-06, + "loss": 2.5892, + "step": 2770 + }, + { + "epoch": 0.58, + "learning_rate": 4.498586811407834e-06, + "loss": 2.692, + "step": 2771 + }, + { + "epoch": 0.58, + "learning_rate": 4.494925954474359e-06, + "loss": 2.5545, + "step": 2772 + }, + { + "epoch": 0.58, + "learning_rate": 4.491265371069311e-06, + "loss": 2.6767, + "step": 2773 + }, + { + "epoch": 0.58, + "learning_rate": 4.487605063175119e-06, + "loss": 2.6409, + "step": 2774 + }, + { + "epoch": 0.58, + "learning_rate": 4.4839450327740656e-06, + "loss": 2.6536, + "step": 2775 + }, + { + "epoch": 0.58, + "learning_rate": 4.480285281848278e-06, + "loss": 2.6638, + "step": 2776 + }, + { + "epoch": 0.58, + "learning_rate": 4.4766258123797355e-06, + "loss": 2.6973, + "step": 2777 + }, + { + "epoch": 0.58, + "learning_rate": 4.472966626350262e-06, + "loss": 2.6158, + "step": 2778 + }, + { + "epoch": 0.58, + "learning_rate": 4.469307725741534e-06, + "loss": 2.6329, + "step": 2779 + }, + { + "epoch": 0.58, + "learning_rate": 4.465649112535067e-06, + "loss": 2.6206, + "step": 2780 + }, + { + "epoch": 0.58, + "learning_rate": 4.461990788712224e-06, + "loss": 2.5997, + "step": 2781 + }, + { + "epoch": 0.58, + "learning_rate": 4.45833275625421e-06, + "loss": 2.6633, + "step": 2782 + }, + { + "epoch": 0.58, + "learning_rate": 4.4546750171420764e-06, + "loss": 2.6203, + "step": 2783 + }, + { + "epoch": 0.58, + "learning_rate": 4.451017573356707e-06, + "loss": 2.6502, + "step": 2784 + }, + { + "epoch": 0.58, + "learning_rate": 4.447360426878834e-06, + "loss": 2.6632, + "step": 2785 + }, + { + "epoch": 0.58, + "learning_rate": 4.443703579689025e-06, + "loss": 2.6654, + "step": 2786 + }, + { + "epoch": 0.58, + "learning_rate": 4.440047033767683e-06, + "loss": 2.6533, + "step": 2787 + }, + { + "epoch": 0.58, + "learning_rate": 4.436390791095057e-06, + "loss": 2.6754, + "step": 2788 + }, + { + "epoch": 0.58, + "learning_rate": 4.432734853651222e-06, + "loss": 2.7108, + "step": 2789 + }, + { + "epoch": 0.59, + "learning_rate": 4.429079223416092e-06, + "loss": 2.7025, + "step": 2790 + }, + { + "epoch": 0.59, + "learning_rate": 4.425423902369411e-06, + "loss": 2.5743, + "step": 2791 + }, + { + "epoch": 0.59, + "learning_rate": 4.421768892490762e-06, + "loss": 2.6451, + "step": 2792 + }, + { + "epoch": 0.59, + "learning_rate": 4.418114195759554e-06, + "loss": 2.5995, + "step": 2793 + }, + { + "epoch": 0.59, + "learning_rate": 4.414459814155029e-06, + "loss": 2.6059, + "step": 2794 + }, + { + "epoch": 0.59, + "learning_rate": 4.41080574965626e-06, + "loss": 2.5967, + "step": 2795 + }, + { + "epoch": 0.59, + "learning_rate": 4.407152004242142e-06, + "loss": 2.6723, + "step": 2796 + }, + { + "epoch": 0.59, + "learning_rate": 4.403498579891405e-06, + "loss": 2.695, + "step": 2797 + }, + { + "epoch": 0.59, + "learning_rate": 4.399845478582598e-06, + "loss": 2.6512, + "step": 2798 + }, + { + "epoch": 0.59, + "learning_rate": 4.396192702294099e-06, + "loss": 2.6432, + "step": 2799 + }, + { + "epoch": 0.59, + "learning_rate": 4.392540253004111e-06, + "loss": 2.6233, + "step": 2800 + }, + { + "epoch": 0.59, + "learning_rate": 4.388888132690657e-06, + "loss": 2.6339, + "step": 2801 + }, + { + "epoch": 0.59, + "learning_rate": 4.385236343331584e-06, + "loss": 2.6349, + "step": 2802 + }, + { + "epoch": 0.59, + "learning_rate": 4.381584886904557e-06, + "loss": 2.6037, + "step": 2803 + }, + { + "epoch": 0.59, + "learning_rate": 4.3779337653870666e-06, + "loss": 2.6632, + "step": 2804 + }, + { + "epoch": 0.59, + "learning_rate": 4.3742829807564126e-06, + "loss": 2.5838, + "step": 2805 + }, + { + "epoch": 0.59, + "learning_rate": 4.3706325349897205e-06, + "loss": 2.5973, + "step": 2806 + }, + { + "epoch": 0.59, + "learning_rate": 4.3669824300639305e-06, + "loss": 2.6716, + "step": 2807 + }, + { + "epoch": 0.59, + "learning_rate": 4.363332667955794e-06, + "loss": 2.6303, + "step": 2808 + }, + { + "epoch": 0.59, + "learning_rate": 4.359683250641883e-06, + "loss": 2.6032, + "step": 2809 + }, + { + "epoch": 0.59, + "learning_rate": 4.35603418009858e-06, + "loss": 2.6408, + "step": 2810 + }, + { + "epoch": 0.59, + "learning_rate": 4.352385458302077e-06, + "loss": 2.6991, + "step": 2811 + }, + { + "epoch": 0.59, + "learning_rate": 4.348737087228382e-06, + "loss": 2.6539, + "step": 2812 + }, + { + "epoch": 0.59, + "learning_rate": 4.345089068853309e-06, + "loss": 2.6959, + "step": 2813 + }, + { + "epoch": 0.59, + "learning_rate": 4.341441405152482e-06, + "loss": 2.6258, + "step": 2814 + }, + { + "epoch": 0.59, + "learning_rate": 4.337794098101335e-06, + "loss": 2.6963, + "step": 2815 + }, + { + "epoch": 0.59, + "learning_rate": 4.3341471496751085e-06, + "loss": 2.6444, + "step": 2816 + }, + { + "epoch": 0.59, + "learning_rate": 4.330500561848846e-06, + "loss": 2.5933, + "step": 2817 + }, + { + "epoch": 0.59, + "learning_rate": 4.326854336597395e-06, + "loss": 2.5926, + "step": 2818 + }, + { + "epoch": 0.59, + "learning_rate": 4.323208475895416e-06, + "loss": 2.626, + "step": 2819 + }, + { + "epoch": 0.59, + "learning_rate": 4.319562981717358e-06, + "loss": 2.7057, + "step": 2820 + }, + { + "epoch": 0.59, + "learning_rate": 4.315917856037482e-06, + "loss": 2.6523, + "step": 2821 + }, + { + "epoch": 0.59, + "learning_rate": 4.312273100829845e-06, + "loss": 2.6778, + "step": 2822 + }, + { + "epoch": 0.59, + "learning_rate": 4.308628718068306e-06, + "loss": 2.6946, + "step": 2823 + }, + { + "epoch": 0.59, + "learning_rate": 4.304984709726519e-06, + "loss": 2.5786, + "step": 2824 + }, + { + "epoch": 0.59, + "learning_rate": 4.3013410777779375e-06, + "loss": 2.6085, + "step": 2825 + }, + { + "epoch": 0.59, + "learning_rate": 4.297697824195813e-06, + "loss": 2.608, + "step": 2826 + }, + { + "epoch": 0.59, + "learning_rate": 4.294054950953186e-06, + "loss": 2.6065, + "step": 2827 + }, + { + "epoch": 0.59, + "learning_rate": 4.290412460022896e-06, + "loss": 2.692, + "step": 2828 + }, + { + "epoch": 0.59, + "learning_rate": 4.286770353377573e-06, + "loss": 2.5764, + "step": 2829 + }, + { + "epoch": 0.59, + "learning_rate": 4.283128632989644e-06, + "loss": 2.5918, + "step": 2830 + }, + { + "epoch": 0.59, + "learning_rate": 4.279487300831318e-06, + "loss": 2.6711, + "step": 2831 + }, + { + "epoch": 0.59, + "learning_rate": 4.2758463588746035e-06, + "loss": 2.633, + "step": 2832 + }, + { + "epoch": 0.59, + "learning_rate": 4.27220580909129e-06, + "loss": 2.6724, + "step": 2833 + }, + { + "epoch": 0.59, + "learning_rate": 4.2685656534529576e-06, + "loss": 2.6279, + "step": 2834 + }, + { + "epoch": 0.59, + "learning_rate": 4.264925893930972e-06, + "loss": 2.6313, + "step": 2835 + }, + { + "epoch": 0.59, + "learning_rate": 4.261286532496486e-06, + "loss": 2.6821, + "step": 2836 + }, + { + "epoch": 0.59, + "learning_rate": 4.257647571120437e-06, + "loss": 2.6261, + "step": 2837 + }, + { + "epoch": 0.6, + "learning_rate": 4.2540090117735435e-06, + "loss": 2.5842, + "step": 2838 + }, + { + "epoch": 0.6, + "learning_rate": 4.250370856426307e-06, + "loss": 2.6609, + "step": 2839 + }, + { + "epoch": 0.6, + "learning_rate": 4.246733107049012e-06, + "loss": 2.6274, + "step": 2840 + }, + { + "epoch": 0.6, + "learning_rate": 4.243095765611724e-06, + "loss": 2.6375, + "step": 2841 + }, + { + "epoch": 0.6, + "learning_rate": 4.23945883408428e-06, + "loss": 2.7103, + "step": 2842 + }, + { + "epoch": 0.6, + "learning_rate": 4.2358223144363046e-06, + "loss": 2.659, + "step": 2843 + }, + { + "epoch": 0.6, + "learning_rate": 4.232186208637193e-06, + "loss": 2.666, + "step": 2844 + }, + { + "epoch": 0.6, + "learning_rate": 4.228550518656118e-06, + "loss": 2.6516, + "step": 2845 + }, + { + "epoch": 0.6, + "learning_rate": 4.22491524646203e-06, + "loss": 2.6388, + "step": 2846 + }, + { + "epoch": 0.6, + "learning_rate": 4.22128039402365e-06, + "loss": 2.6437, + "step": 2847 + }, + { + "epoch": 0.6, + "learning_rate": 4.217645963309474e-06, + "loss": 2.6429, + "step": 2848 + }, + { + "epoch": 0.6, + "learning_rate": 4.214011956287765e-06, + "loss": 2.6522, + "step": 2849 + }, + { + "epoch": 0.6, + "learning_rate": 4.210378374926559e-06, + "loss": 2.676, + "step": 2850 + }, + { + "epoch": 0.6, + "learning_rate": 4.206745221193665e-06, + "loss": 2.6579, + "step": 2851 + }, + { + "epoch": 0.6, + "learning_rate": 4.2031124970566576e-06, + "loss": 2.6731, + "step": 2852 + }, + { + "epoch": 0.6, + "learning_rate": 4.199480204482877e-06, + "loss": 2.7069, + "step": 2853 + }, + { + "epoch": 0.6, + "learning_rate": 4.1958483454394314e-06, + "loss": 2.6244, + "step": 2854 + }, + { + "epoch": 0.6, + "learning_rate": 4.192216921893198e-06, + "loss": 2.6133, + "step": 2855 + }, + { + "epoch": 0.6, + "learning_rate": 4.188585935810807e-06, + "loss": 2.7361, + "step": 2856 + }, + { + "epoch": 0.6, + "learning_rate": 4.184955389158665e-06, + "loss": 2.5916, + "step": 2857 + }, + { + "epoch": 0.6, + "learning_rate": 4.1813252839029325e-06, + "loss": 2.611, + "step": 2858 + }, + { + "epoch": 0.6, + "learning_rate": 4.177695622009534e-06, + "loss": 2.6495, + "step": 2859 + }, + { + "epoch": 0.6, + "learning_rate": 4.174066405444151e-06, + "loss": 2.6426, + "step": 2860 + }, + { + "epoch": 0.6, + "learning_rate": 4.170437636172227e-06, + "loss": 2.6216, + "step": 2861 + }, + { + "epoch": 0.6, + "learning_rate": 4.166809316158964e-06, + "loss": 2.638, + "step": 2862 + }, + { + "epoch": 0.6, + "learning_rate": 4.163181447369316e-06, + "loss": 2.625, + "step": 2863 + }, + { + "epoch": 0.6, + "learning_rate": 4.159554031767996e-06, + "loss": 2.6386, + "step": 2864 + }, + { + "epoch": 0.6, + "learning_rate": 4.155927071319472e-06, + "loss": 2.6745, + "step": 2865 + }, + { + "epoch": 0.6, + "learning_rate": 4.152300567987962e-06, + "loss": 2.648, + "step": 2866 + }, + { + "epoch": 0.6, + "learning_rate": 4.148674523737443e-06, + "loss": 2.657, + "step": 2867 + }, + { + "epoch": 0.6, + "learning_rate": 4.145048940531636e-06, + "loss": 2.6648, + "step": 2868 + }, + { + "epoch": 0.6, + "learning_rate": 4.141423820334018e-06, + "loss": 2.5864, + "step": 2869 + }, + { + "epoch": 0.6, + "learning_rate": 4.137799165107811e-06, + "loss": 2.6438, + "step": 2870 + }, + { + "epoch": 0.6, + "learning_rate": 4.134174976815986e-06, + "loss": 2.6795, + "step": 2871 + }, + { + "epoch": 0.6, + "learning_rate": 4.130551257421265e-06, + "loss": 2.6377, + "step": 2872 + }, + { + "epoch": 0.6, + "learning_rate": 4.126928008886112e-06, + "loss": 2.642, + "step": 2873 + }, + { + "epoch": 0.6, + "learning_rate": 4.123305233172738e-06, + "loss": 2.6674, + "step": 2874 + }, + { + "epoch": 0.6, + "learning_rate": 4.119682932243094e-06, + "loss": 2.6218, + "step": 2875 + }, + { + "epoch": 0.6, + "learning_rate": 4.116061108058882e-06, + "loss": 2.6848, + "step": 2876 + }, + { + "epoch": 0.6, + "learning_rate": 4.11243976258154e-06, + "loss": 2.5352, + "step": 2877 + }, + { + "epoch": 0.6, + "learning_rate": 4.108818897772243e-06, + "loss": 2.6648, + "step": 2878 + }, + { + "epoch": 0.6, + "learning_rate": 4.105198515591915e-06, + "loss": 2.6852, + "step": 2879 + }, + { + "epoch": 0.6, + "learning_rate": 4.101578618001212e-06, + "loss": 2.5782, + "step": 2880 + }, + { + "epoch": 0.6, + "learning_rate": 4.097959206960529e-06, + "loss": 2.7303, + "step": 2881 + }, + { + "epoch": 0.6, + "learning_rate": 4.0943402844300004e-06, + "loss": 2.5888, + "step": 2882 + }, + { + "epoch": 0.6, + "learning_rate": 4.090721852369492e-06, + "loss": 2.6598, + "step": 2883 + }, + { + "epoch": 0.6, + "learning_rate": 4.087103912738606e-06, + "loss": 2.5896, + "step": 2884 + }, + { + "epoch": 0.6, + "learning_rate": 4.0834864674966765e-06, + "loss": 2.652, + "step": 2885 + }, + { + "epoch": 0.61, + "learning_rate": 4.079869518602773e-06, + "loss": 2.61, + "step": 2886 + }, + { + "epoch": 0.61, + "learning_rate": 4.07625306801569e-06, + "loss": 2.6147, + "step": 2887 + }, + { + "epoch": 0.61, + "learning_rate": 4.072637117693962e-06, + "loss": 2.6666, + "step": 2888 + }, + { + "epoch": 0.61, + "learning_rate": 4.069021669595843e-06, + "loss": 2.6339, + "step": 2889 + }, + { + "epoch": 0.61, + "learning_rate": 4.065406725679319e-06, + "loss": 2.6562, + "step": 2890 + }, + { + "epoch": 0.61, + "learning_rate": 4.061792287902107e-06, + "loss": 2.6964, + "step": 2891 + }, + { + "epoch": 0.61, + "learning_rate": 4.05817835822164e-06, + "loss": 2.647, + "step": 2892 + }, + { + "epoch": 0.61, + "learning_rate": 4.054564938595084e-06, + "loss": 2.5872, + "step": 2893 + }, + { + "epoch": 0.61, + "learning_rate": 4.050952030979326e-06, + "loss": 2.6315, + "step": 2894 + }, + { + "epoch": 0.61, + "learning_rate": 4.047339637330976e-06, + "loss": 2.6625, + "step": 2895 + }, + { + "epoch": 0.61, + "learning_rate": 4.043727759606365e-06, + "loss": 2.6046, + "step": 2896 + }, + { + "epoch": 0.61, + "learning_rate": 4.040116399761547e-06, + "loss": 2.6182, + "step": 2897 + }, + { + "epoch": 0.61, + "learning_rate": 4.0365055597522955e-06, + "loss": 2.5623, + "step": 2898 + }, + { + "epoch": 0.61, + "learning_rate": 4.032895241534096e-06, + "loss": 2.6123, + "step": 2899 + }, + { + "epoch": 0.61, + "learning_rate": 4.029285447062159e-06, + "loss": 2.5969, + "step": 2900 + }, + { + "epoch": 0.61, + "learning_rate": 4.0256761782914065e-06, + "loss": 2.6855, + "step": 2901 + }, + { + "epoch": 0.61, + "learning_rate": 4.02206743717648e-06, + "loss": 2.6285, + "step": 2902 + }, + { + "epoch": 0.61, + "learning_rate": 4.018459225671732e-06, + "loss": 2.5735, + "step": 2903 + }, + { + "epoch": 0.61, + "learning_rate": 4.014851545731231e-06, + "loss": 2.707, + "step": 2904 + }, + { + "epoch": 0.61, + "learning_rate": 4.011244399308755e-06, + "loss": 2.6315, + "step": 2905 + }, + { + "epoch": 0.61, + "learning_rate": 4.007637788357793e-06, + "loss": 2.6508, + "step": 2906 + }, + { + "epoch": 0.61, + "learning_rate": 4.0040317148315456e-06, + "loss": 2.6072, + "step": 2907 + }, + { + "epoch": 0.61, + "learning_rate": 4.000426180682921e-06, + "loss": 2.5997, + "step": 2908 + }, + { + "epoch": 0.61, + "learning_rate": 3.996821187864537e-06, + "loss": 2.6514, + "step": 2909 + }, + { + "epoch": 0.61, + "learning_rate": 3.993216738328717e-06, + "loss": 2.6865, + "step": 2910 + }, + { + "epoch": 0.61, + "learning_rate": 3.98961283402749e-06, + "loss": 2.6112, + "step": 2911 + }, + { + "epoch": 0.61, + "learning_rate": 3.986009476912592e-06, + "loss": 2.6463, + "step": 2912 + }, + { + "epoch": 0.61, + "learning_rate": 3.98240666893546e-06, + "loss": 2.6126, + "step": 2913 + }, + { + "epoch": 0.61, + "learning_rate": 3.978804412047233e-06, + "loss": 2.6465, + "step": 2914 + }, + { + "epoch": 0.61, + "learning_rate": 3.975202708198754e-06, + "loss": 2.5869, + "step": 2915 + }, + { + "epoch": 0.61, + "learning_rate": 3.971601559340566e-06, + "loss": 2.6639, + "step": 2916 + }, + { + "epoch": 0.61, + "learning_rate": 3.968000967422909e-06, + "loss": 2.5709, + "step": 2917 + }, + { + "epoch": 0.61, + "learning_rate": 3.964400934395726e-06, + "loss": 2.6705, + "step": 2918 + }, + { + "epoch": 0.61, + "learning_rate": 3.960801462208653e-06, + "loss": 2.6802, + "step": 2919 + }, + { + "epoch": 0.61, + "learning_rate": 3.957202552811025e-06, + "loss": 2.6832, + "step": 2920 + }, + { + "epoch": 0.61, + "learning_rate": 3.95360420815187e-06, + "loss": 2.5851, + "step": 2921 + }, + { + "epoch": 0.61, + "learning_rate": 3.950006430179912e-06, + "loss": 2.6377, + "step": 2922 + }, + { + "epoch": 0.61, + "learning_rate": 3.946409220843564e-06, + "loss": 2.7031, + "step": 2923 + }, + { + "epoch": 0.61, + "learning_rate": 3.94281258209094e-06, + "loss": 2.6344, + "step": 2924 + }, + { + "epoch": 0.61, + "learning_rate": 3.939216515869838e-06, + "loss": 2.61, + "step": 2925 + }, + { + "epoch": 0.61, + "learning_rate": 3.935621024127744e-06, + "loss": 2.6775, + "step": 2926 + }, + { + "epoch": 0.61, + "learning_rate": 3.932026108811841e-06, + "loss": 2.6231, + "step": 2927 + }, + { + "epoch": 0.61, + "learning_rate": 3.928431771868993e-06, + "loss": 2.6691, + "step": 2928 + }, + { + "epoch": 0.61, + "learning_rate": 3.9248380152457484e-06, + "loss": 2.6525, + "step": 2929 + }, + { + "epoch": 0.61, + "learning_rate": 3.921244840888353e-06, + "loss": 2.6544, + "step": 2930 + }, + { + "epoch": 0.61, + "learning_rate": 3.917652250742727e-06, + "loss": 2.6102, + "step": 2931 + }, + { + "epoch": 0.61, + "learning_rate": 3.9140602467544755e-06, + "loss": 2.5943, + "step": 2932 + }, + { + "epoch": 0.62, + "learning_rate": 3.910468830868891e-06, + "loss": 2.6502, + "step": 2933 + }, + { + "epoch": 0.62, + "learning_rate": 3.906878005030944e-06, + "loss": 2.6539, + "step": 2934 + }, + { + "epoch": 0.62, + "learning_rate": 3.903287771185284e-06, + "loss": 2.6583, + "step": 2935 + }, + { + "epoch": 0.62, + "learning_rate": 3.899698131276243e-06, + "loss": 2.5955, + "step": 2936 + }, + { + "epoch": 0.62, + "learning_rate": 3.8961090872478294e-06, + "loss": 2.6377, + "step": 2937 + }, + { + "epoch": 0.62, + "learning_rate": 3.89252064104373e-06, + "loss": 2.6426, + "step": 2938 + }, + { + "epoch": 0.62, + "learning_rate": 3.888932794607308e-06, + "loss": 2.6213, + "step": 2939 + }, + { + "epoch": 0.62, + "learning_rate": 3.8853455498816e-06, + "loss": 2.7126, + "step": 2940 + }, + { + "epoch": 0.62, + "learning_rate": 3.8817589088093225e-06, + "loss": 2.6589, + "step": 2941 + }, + { + "epoch": 0.62, + "learning_rate": 3.878172873332854e-06, + "loss": 2.6003, + "step": 2942 + }, + { + "epoch": 0.62, + "learning_rate": 3.874587445394254e-06, + "loss": 2.6598, + "step": 2943 + }, + { + "epoch": 0.62, + "learning_rate": 3.871002626935251e-06, + "loss": 2.6096, + "step": 2944 + }, + { + "epoch": 0.62, + "learning_rate": 3.867418419897245e-06, + "loss": 2.6158, + "step": 2945 + }, + { + "epoch": 0.62, + "learning_rate": 3.8638348262213e-06, + "loss": 2.6177, + "step": 2946 + }, + { + "epoch": 0.62, + "learning_rate": 3.860251847848151e-06, + "loss": 2.6551, + "step": 2947 + }, + { + "epoch": 0.62, + "learning_rate": 3.856669486718201e-06, + "loss": 2.6545, + "step": 2948 + }, + { + "epoch": 0.62, + "learning_rate": 3.853087744771519e-06, + "loss": 2.6578, + "step": 2949 + }, + { + "epoch": 0.62, + "learning_rate": 3.849506623947832e-06, + "loss": 2.6096, + "step": 2950 + }, + { + "epoch": 0.62, + "learning_rate": 3.845926126186539e-06, + "loss": 2.6552, + "step": 2951 + }, + { + "epoch": 0.62, + "learning_rate": 3.842346253426695e-06, + "loss": 2.6409, + "step": 2952 + }, + { + "epoch": 0.62, + "learning_rate": 3.838767007607023e-06, + "loss": 2.6023, + "step": 2953 + }, + { + "epoch": 0.62, + "learning_rate": 3.8351883906659015e-06, + "loss": 2.6361, + "step": 2954 + }, + { + "epoch": 0.62, + "learning_rate": 3.831610404541371e-06, + "loss": 2.6649, + "step": 2955 + }, + { + "epoch": 0.62, + "learning_rate": 3.82803305117113e-06, + "loss": 2.6322, + "step": 2956 + }, + { + "epoch": 0.62, + "learning_rate": 3.824456332492531e-06, + "loss": 2.665, + "step": 2957 + }, + { + "epoch": 0.62, + "learning_rate": 3.820880250442587e-06, + "loss": 2.6594, + "step": 2958 + }, + { + "epoch": 0.62, + "learning_rate": 3.8173048069579635e-06, + "loss": 2.6175, + "step": 2959 + }, + { + "epoch": 0.62, + "learning_rate": 3.8137300039749837e-06, + "loss": 2.6243, + "step": 2960 + }, + { + "epoch": 0.62, + "learning_rate": 3.8101558434296206e-06, + "loss": 2.715, + "step": 2961 + }, + { + "epoch": 0.62, + "learning_rate": 3.8065823272574998e-06, + "loss": 2.6178, + "step": 2962 + }, + { + "epoch": 0.62, + "learning_rate": 3.803009457393901e-06, + "loss": 2.6528, + "step": 2963 + }, + { + "epoch": 0.62, + "learning_rate": 3.7994372357737473e-06, + "loss": 2.6016, + "step": 2964 + }, + { + "epoch": 0.62, + "learning_rate": 3.7958656643316154e-06, + "loss": 2.6239, + "step": 2965 + }, + { + "epoch": 0.62, + "learning_rate": 3.792294745001732e-06, + "loss": 2.5876, + "step": 2966 + }, + { + "epoch": 0.62, + "learning_rate": 3.788724479717967e-06, + "loss": 2.5865, + "step": 2967 + }, + { + "epoch": 0.62, + "learning_rate": 3.7851548704138346e-06, + "loss": 2.5779, + "step": 2968 + }, + { + "epoch": 0.62, + "learning_rate": 3.781585919022499e-06, + "loss": 2.61, + "step": 2969 + }, + { + "epoch": 0.62, + "learning_rate": 3.7780176274767666e-06, + "loss": 2.6161, + "step": 2970 + }, + { + "epoch": 0.62, + "learning_rate": 3.77444999770908e-06, + "loss": 2.6697, + "step": 2971 + }, + { + "epoch": 0.62, + "learning_rate": 3.770883031651531e-06, + "loss": 2.581, + "step": 2972 + }, + { + "epoch": 0.62, + "learning_rate": 3.7673167312358505e-06, + "loss": 2.6341, + "step": 2973 + }, + { + "epoch": 0.62, + "learning_rate": 3.7637510983934063e-06, + "loss": 2.6571, + "step": 2974 + }, + { + "epoch": 0.62, + "learning_rate": 3.7601861350552073e-06, + "loss": 2.6784, + "step": 2975 + }, + { + "epoch": 0.62, + "learning_rate": 3.7566218431518987e-06, + "loss": 2.6435, + "step": 2976 + }, + { + "epoch": 0.62, + "learning_rate": 3.753058224613762e-06, + "loss": 2.6524, + "step": 2977 + }, + { + "epoch": 0.62, + "learning_rate": 3.7494952813707154e-06, + "loss": 2.5995, + "step": 2978 + }, + { + "epoch": 0.62, + "learning_rate": 3.7459330153523064e-06, + "loss": 2.6505, + "step": 2979 + }, + { + "epoch": 0.62, + "learning_rate": 3.742371428487722e-06, + "loss": 2.5852, + "step": 2980 + }, + { + "epoch": 0.63, + "learning_rate": 3.7388105227057796e-06, + "loss": 2.6476, + "step": 2981 + }, + { + "epoch": 0.63, + "learning_rate": 3.7352502999349265e-06, + "loss": 2.663, + "step": 2982 + }, + { + "epoch": 0.63, + "learning_rate": 3.7316907621032405e-06, + "loss": 2.5974, + "step": 2983 + }, + { + "epoch": 0.63, + "learning_rate": 3.7281319111384274e-06, + "loss": 2.6525, + "step": 2984 + }, + { + "epoch": 0.63, + "learning_rate": 3.7245737489678257e-06, + "loss": 2.6985, + "step": 2985 + }, + { + "epoch": 0.63, + "learning_rate": 3.7210162775183927e-06, + "loss": 2.6791, + "step": 2986 + }, + { + "epoch": 0.63, + "learning_rate": 3.71745949871672e-06, + "loss": 2.6856, + "step": 2987 + }, + { + "epoch": 0.63, + "learning_rate": 3.7139034144890184e-06, + "loss": 2.611, + "step": 2988 + }, + { + "epoch": 0.63, + "learning_rate": 3.7103480267611246e-06, + "loss": 2.6149, + "step": 2989 + }, + { + "epoch": 0.63, + "learning_rate": 3.7067933374585003e-06, + "loss": 2.5752, + "step": 2990 + }, + { + "epoch": 0.63, + "learning_rate": 3.703239348506225e-06, + "loss": 2.6328, + "step": 2991 + }, + { + "epoch": 0.63, + "learning_rate": 3.6996860618290022e-06, + "loss": 2.6623, + "step": 2992 + }, + { + "epoch": 0.63, + "learning_rate": 3.696133479351151e-06, + "loss": 2.6807, + "step": 2993 + }, + { + "epoch": 0.63, + "learning_rate": 3.6925816029966134e-06, + "loss": 2.6489, + "step": 2994 + }, + { + "epoch": 0.63, + "learning_rate": 3.689030434688945e-06, + "loss": 2.6382, + "step": 2995 + }, + { + "epoch": 0.63, + "learning_rate": 3.6854799763513238e-06, + "loss": 2.614, + "step": 2996 + }, + { + "epoch": 0.63, + "learning_rate": 3.681930229906538e-06, + "loss": 2.6767, + "step": 2997 + }, + { + "epoch": 0.63, + "learning_rate": 3.6783811972769917e-06, + "loss": 2.6427, + "step": 2998 + }, + { + "epoch": 0.63, + "learning_rate": 3.6748328803847044e-06, + "loss": 2.6185, + "step": 2999 + }, + { + "epoch": 0.63, + "learning_rate": 3.671285281151303e-06, + "loss": 2.6971, + "step": 3000 + }, + { + "epoch": 0.63, + "learning_rate": 3.6677384014980307e-06, + "loss": 2.6645, + "step": 3001 + }, + { + "epoch": 0.63, + "learning_rate": 3.66419224334574e-06, + "loss": 2.638, + "step": 3002 + }, + { + "epoch": 0.63, + "learning_rate": 3.660646808614892e-06, + "loss": 2.6261, + "step": 3003 + }, + { + "epoch": 0.63, + "learning_rate": 3.6571020992255558e-06, + "loss": 2.6397, + "step": 3004 + }, + { + "epoch": 0.63, + "learning_rate": 3.6535581170974055e-06, + "loss": 2.6558, + "step": 3005 + }, + { + "epoch": 0.63, + "learning_rate": 3.6500148641497304e-06, + "loss": 2.6292, + "step": 3006 + }, + { + "epoch": 0.63, + "learning_rate": 3.6464723423014103e-06, + "loss": 2.6393, + "step": 3007 + }, + { + "epoch": 0.63, + "learning_rate": 3.6429305534709415e-06, + "loss": 2.6593, + "step": 3008 + }, + { + "epoch": 0.63, + "learning_rate": 3.6393894995764177e-06, + "loss": 2.6613, + "step": 3009 + }, + { + "epoch": 0.63, + "learning_rate": 3.6358491825355356e-06, + "loss": 2.631, + "step": 3010 + }, + { + "epoch": 0.63, + "learning_rate": 3.6323096042655936e-06, + "loss": 2.684, + "step": 3011 + }, + { + "epoch": 0.63, + "learning_rate": 3.62877076668349e-06, + "loss": 2.6691, + "step": 3012 + }, + { + "epoch": 0.63, + "learning_rate": 3.6252326717057195e-06, + "loss": 2.5968, + "step": 3013 + }, + { + "epoch": 0.63, + "learning_rate": 3.6216953212483796e-06, + "loss": 2.62, + "step": 3014 + }, + { + "epoch": 0.63, + "learning_rate": 3.618158717227158e-06, + "loss": 2.6867, + "step": 3015 + }, + { + "epoch": 0.63, + "learning_rate": 3.614622861557342e-06, + "loss": 2.6147, + "step": 3016 + }, + { + "epoch": 0.63, + "learning_rate": 3.611087756153815e-06, + "loss": 2.6088, + "step": 3017 + }, + { + "epoch": 0.63, + "learning_rate": 3.607553402931052e-06, + "loss": 2.6762, + "step": 3018 + }, + { + "epoch": 0.63, + "learning_rate": 3.6040198038031216e-06, + "loss": 2.7082, + "step": 3019 + }, + { + "epoch": 0.63, + "learning_rate": 3.6004869606836807e-06, + "loss": 2.5982, + "step": 3020 + }, + { + "epoch": 0.63, + "learning_rate": 3.596954875485985e-06, + "loss": 2.5795, + "step": 3021 + }, + { + "epoch": 0.63, + "learning_rate": 3.593423550122867e-06, + "loss": 2.715, + "step": 3022 + }, + { + "epoch": 0.63, + "learning_rate": 3.58989298650676e-06, + "loss": 2.6722, + "step": 3023 + }, + { + "epoch": 0.63, + "learning_rate": 3.5863631865496784e-06, + "loss": 2.6166, + "step": 3024 + }, + { + "epoch": 0.63, + "learning_rate": 3.5828341521632237e-06, + "loss": 2.6761, + "step": 3025 + }, + { + "epoch": 0.63, + "learning_rate": 3.5793058852585837e-06, + "loss": 2.6514, + "step": 3026 + }, + { + "epoch": 0.63, + "learning_rate": 3.5757783877465314e-06, + "loss": 2.6793, + "step": 3027 + }, + { + "epoch": 0.63, + "learning_rate": 3.572251661537422e-06, + "loss": 2.7005, + "step": 3028 + }, + { + "epoch": 0.64, + "learning_rate": 3.5687257085411913e-06, + "loss": 2.6358, + "step": 3029 + }, + { + "epoch": 0.64, + "learning_rate": 3.5652005306673593e-06, + "loss": 2.6003, + "step": 3030 + }, + { + "epoch": 0.64, + "learning_rate": 3.5616761298250235e-06, + "loss": 2.6042, + "step": 3031 + }, + { + "epoch": 0.64, + "learning_rate": 3.5581525079228647e-06, + "loss": 2.6217, + "step": 3032 + }, + { + "epoch": 0.64, + "learning_rate": 3.5546296668691376e-06, + "loss": 2.6376, + "step": 3033 + }, + { + "epoch": 0.64, + "learning_rate": 3.5511076085716768e-06, + "loss": 2.6746, + "step": 3034 + }, + { + "epoch": 0.64, + "learning_rate": 3.5475863349378907e-06, + "loss": 2.5958, + "step": 3035 + }, + { + "epoch": 0.64, + "learning_rate": 3.5440658478747643e-06, + "loss": 2.6343, + "step": 3036 + }, + { + "epoch": 0.64, + "learning_rate": 3.540546149288855e-06, + "loss": 2.6078, + "step": 3037 + }, + { + "epoch": 0.64, + "learning_rate": 3.537027241086296e-06, + "loss": 2.6249, + "step": 3038 + }, + { + "epoch": 0.64, + "learning_rate": 3.5335091251727914e-06, + "loss": 2.6092, + "step": 3039 + }, + { + "epoch": 0.64, + "learning_rate": 3.5299918034536146e-06, + "loss": 2.6828, + "step": 3040 + }, + { + "epoch": 0.64, + "learning_rate": 3.526475277833609e-06, + "loss": 2.7099, + "step": 3041 + }, + { + "epoch": 0.64, + "learning_rate": 3.5229595502171932e-06, + "loss": 2.6172, + "step": 3042 + }, + { + "epoch": 0.64, + "learning_rate": 3.51944462250834e-06, + "loss": 2.5947, + "step": 3043 + }, + { + "epoch": 0.64, + "learning_rate": 3.5159304966106034e-06, + "loss": 2.6614, + "step": 3044 + }, + { + "epoch": 0.64, + "learning_rate": 3.512417174427095e-06, + "loss": 2.686, + "step": 3045 + }, + { + "epoch": 0.64, + "learning_rate": 3.5089046578604934e-06, + "loss": 2.6141, + "step": 3046 + }, + { + "epoch": 0.64, + "learning_rate": 3.50539294881304e-06, + "loss": 2.6861, + "step": 3047 + }, + { + "epoch": 0.64, + "learning_rate": 3.5018820491865424e-06, + "loss": 2.5853, + "step": 3048 + }, + { + "epoch": 0.64, + "learning_rate": 3.498371960882364e-06, + "loss": 2.7052, + "step": 3049 + }, + { + "epoch": 0.64, + "learning_rate": 3.4948626858014345e-06, + "loss": 2.7031, + "step": 3050 + }, + { + "epoch": 0.64, + "learning_rate": 3.4913542258442378e-06, + "loss": 2.6962, + "step": 3051 + }, + { + "epoch": 0.64, + "learning_rate": 3.4878465829108197e-06, + "loss": 2.6206, + "step": 3052 + }, + { + "epoch": 0.64, + "learning_rate": 3.4843397589007842e-06, + "loss": 2.5453, + "step": 3053 + }, + { + "epoch": 0.64, + "learning_rate": 3.480833755713291e-06, + "loss": 2.6423, + "step": 3054 + }, + { + "epoch": 0.64, + "learning_rate": 3.477328575247053e-06, + "loss": 2.6532, + "step": 3055 + }, + { + "epoch": 0.64, + "learning_rate": 3.4738242194003403e-06, + "loss": 2.6631, + "step": 3056 + }, + { + "epoch": 0.64, + "learning_rate": 3.470320690070978e-06, + "loss": 2.6398, + "step": 3057 + }, + { + "epoch": 0.64, + "learning_rate": 3.466817989156336e-06, + "loss": 2.6685, + "step": 3058 + }, + { + "epoch": 0.64, + "learning_rate": 3.4633161185533435e-06, + "loss": 2.5996, + "step": 3059 + }, + { + "epoch": 0.64, + "learning_rate": 3.4598150801584773e-06, + "loss": 2.5876, + "step": 3060 + }, + { + "epoch": 0.64, + "learning_rate": 3.4563148758677634e-06, + "loss": 2.6696, + "step": 3061 + }, + { + "epoch": 0.64, + "learning_rate": 3.4528155075767746e-06, + "loss": 2.6506, + "step": 3062 + }, + { + "epoch": 0.64, + "learning_rate": 3.449316977180634e-06, + "loss": 2.7109, + "step": 3063 + }, + { + "epoch": 0.64, + "learning_rate": 3.4458192865740105e-06, + "loss": 2.5942, + "step": 3064 + }, + { + "epoch": 0.64, + "learning_rate": 3.4423224376511143e-06, + "loss": 2.6664, + "step": 3065 + }, + { + "epoch": 0.64, + "learning_rate": 3.438826432305704e-06, + "loss": 2.6004, + "step": 3066 + }, + { + "epoch": 0.64, + "learning_rate": 3.43533127243108e-06, + "loss": 2.6571, + "step": 3067 + }, + { + "epoch": 0.64, + "learning_rate": 3.431836959920083e-06, + "loss": 2.6498, + "step": 3068 + }, + { + "epoch": 0.64, + "learning_rate": 3.4283434966650997e-06, + "loss": 2.6387, + "step": 3069 + }, + { + "epoch": 0.64, + "learning_rate": 3.4248508845580513e-06, + "loss": 2.639, + "step": 3070 + }, + { + "epoch": 0.64, + "learning_rate": 3.4213591254904023e-06, + "loss": 2.6468, + "step": 3071 + }, + { + "epoch": 0.64, + "learning_rate": 3.417868221353151e-06, + "loss": 2.6134, + "step": 3072 + }, + { + "epoch": 0.64, + "learning_rate": 3.4143781740368355e-06, + "loss": 2.7175, + "step": 3073 + }, + { + "epoch": 0.64, + "learning_rate": 3.4108889854315315e-06, + "loss": 2.7128, + "step": 3074 + }, + { + "epoch": 0.64, + "learning_rate": 3.407400657426846e-06, + "loss": 2.5673, + "step": 3075 + }, + { + "epoch": 0.64, + "learning_rate": 3.403913191911922e-06, + "loss": 2.6009, + "step": 3076 + }, + { + "epoch": 0.65, + "learning_rate": 3.4004265907754343e-06, + "loss": 2.6381, + "step": 3077 + }, + { + "epoch": 0.65, + "learning_rate": 3.396940855905593e-06, + "loss": 2.5682, + "step": 3078 + }, + { + "epoch": 0.65, + "learning_rate": 3.393455989190132e-06, + "loss": 2.6481, + "step": 3079 + }, + { + "epoch": 0.65, + "learning_rate": 3.3899719925163223e-06, + "loss": 2.6271, + "step": 3080 + }, + { + "epoch": 0.65, + "learning_rate": 3.3864888677709597e-06, + "loss": 2.6471, + "step": 3081 + }, + { + "epoch": 0.65, + "learning_rate": 3.3830066168403697e-06, + "loss": 2.6363, + "step": 3082 + }, + { + "epoch": 0.65, + "learning_rate": 3.379525241610402e-06, + "loss": 2.6334, + "step": 3083 + }, + { + "epoch": 0.65, + "learning_rate": 3.3760447439664355e-06, + "loss": 2.6148, + "step": 3084 + }, + { + "epoch": 0.65, + "learning_rate": 3.3725651257933723e-06, + "loss": 2.7137, + "step": 3085 + }, + { + "epoch": 0.65, + "learning_rate": 3.3690863889756374e-06, + "loss": 2.5418, + "step": 3086 + }, + { + "epoch": 0.65, + "learning_rate": 3.365608535397179e-06, + "loss": 2.6557, + "step": 3087 + }, + { + "epoch": 0.65, + "learning_rate": 3.3621315669414677e-06, + "loss": 2.5664, + "step": 3088 + }, + { + "epoch": 0.65, + "learning_rate": 3.358655485491492e-06, + "loss": 2.5781, + "step": 3089 + }, + { + "epoch": 0.65, + "learning_rate": 3.355180292929765e-06, + "loss": 2.6313, + "step": 3090 + }, + { + "epoch": 0.65, + "learning_rate": 3.3517059911383144e-06, + "loss": 2.6575, + "step": 3091 + }, + { + "epoch": 0.65, + "learning_rate": 3.348232581998686e-06, + "loss": 2.6775, + "step": 3092 + }, + { + "epoch": 0.65, + "learning_rate": 3.3447600673919465e-06, + "loss": 2.6264, + "step": 3093 + }, + { + "epoch": 0.65, + "learning_rate": 3.3412884491986686e-06, + "loss": 2.605, + "step": 3094 + }, + { + "epoch": 0.65, + "learning_rate": 3.33781772929895e-06, + "loss": 2.6284, + "step": 3095 + }, + { + "epoch": 0.65, + "learning_rate": 3.334347909572396e-06, + "loss": 2.614, + "step": 3096 + }, + { + "epoch": 0.65, + "learning_rate": 3.3308789918981254e-06, + "loss": 2.6616, + "step": 3097 + }, + { + "epoch": 0.65, + "learning_rate": 3.3274109781547685e-06, + "loss": 2.6279, + "step": 3098 + }, + { + "epoch": 0.65, + "learning_rate": 3.3239438702204684e-06, + "loss": 2.6221, + "step": 3099 + }, + { + "epoch": 0.65, + "learning_rate": 3.3204776699728754e-06, + "loss": 2.6434, + "step": 3100 + }, + { + "epoch": 0.65, + "learning_rate": 3.317012379289146e-06, + "loss": 2.5968, + "step": 3101 + }, + { + "epoch": 0.65, + "learning_rate": 3.313548000045949e-06, + "loss": 2.6546, + "step": 3102 + }, + { + "epoch": 0.65, + "learning_rate": 3.3100845341194565e-06, + "loss": 2.5971, + "step": 3103 + }, + { + "epoch": 0.65, + "learning_rate": 3.3066219833853454e-06, + "loss": 2.6737, + "step": 3104 + }, + { + "epoch": 0.65, + "learning_rate": 3.3031603497188002e-06, + "loss": 2.6216, + "step": 3105 + }, + { + "epoch": 0.65, + "learning_rate": 3.2996996349945066e-06, + "loss": 2.6149, + "step": 3106 + }, + { + "epoch": 0.65, + "learning_rate": 3.2962398410866535e-06, + "loss": 2.6969, + "step": 3107 + }, + { + "epoch": 0.65, + "learning_rate": 3.2927809698689283e-06, + "loss": 2.6325, + "step": 3108 + }, + { + "epoch": 0.65, + "learning_rate": 3.2893230232145223e-06, + "loss": 2.6353, + "step": 3109 + }, + { + "epoch": 0.65, + "learning_rate": 3.285866002996124e-06, + "loss": 2.687, + "step": 3110 + }, + { + "epoch": 0.65, + "learning_rate": 3.282409911085922e-06, + "loss": 2.6519, + "step": 3111 + }, + { + "epoch": 0.65, + "learning_rate": 3.2789547493556e-06, + "loss": 2.6375, + "step": 3112 + }, + { + "epoch": 0.65, + "learning_rate": 3.275500519676339e-06, + "loss": 2.6637, + "step": 3113 + }, + { + "epoch": 0.65, + "learning_rate": 3.272047223918816e-06, + "loss": 2.6241, + "step": 3114 + }, + { + "epoch": 0.65, + "learning_rate": 3.2685948639532026e-06, + "loss": 2.6716, + "step": 3115 + }, + { + "epoch": 0.65, + "learning_rate": 3.26514344164916e-06, + "loss": 2.5342, + "step": 3116 + }, + { + "epoch": 0.65, + "learning_rate": 3.261692958875845e-06, + "loss": 2.6063, + "step": 3117 + }, + { + "epoch": 0.65, + "learning_rate": 3.2582434175019045e-06, + "loss": 2.6597, + "step": 3118 + }, + { + "epoch": 0.65, + "learning_rate": 3.2547948193954747e-06, + "loss": 2.6119, + "step": 3119 + }, + { + "epoch": 0.65, + "learning_rate": 3.2513471664241847e-06, + "loss": 2.6607, + "step": 3120 + }, + { + "epoch": 0.65, + "learning_rate": 3.2479004604551475e-06, + "loss": 2.6433, + "step": 3121 + }, + { + "epoch": 0.65, + "learning_rate": 3.2444547033549654e-06, + "loss": 2.6505, + "step": 3122 + }, + { + "epoch": 0.65, + "learning_rate": 3.2410098969897253e-06, + "loss": 2.6826, + "step": 3123 + }, + { + "epoch": 0.66, + "learning_rate": 3.2375660432249994e-06, + "loss": 2.643, + "step": 3124 + }, + { + "epoch": 0.66, + "learning_rate": 3.2341231439258454e-06, + "loss": 2.6245, + "step": 3125 + }, + { + "epoch": 0.66, + "learning_rate": 3.2306812009568044e-06, + "loss": 2.6547, + "step": 3126 + }, + { + "epoch": 0.66, + "learning_rate": 3.227240216181899e-06, + "loss": 2.6269, + "step": 3127 + }, + { + "epoch": 0.66, + "learning_rate": 3.22380019146463e-06, + "loss": 2.6352, + "step": 3128 + }, + { + "epoch": 0.66, + "learning_rate": 3.2203611286679847e-06, + "loss": 2.6159, + "step": 3129 + }, + { + "epoch": 0.66, + "learning_rate": 3.216923029654422e-06, + "loss": 2.7005, + "step": 3130 + }, + { + "epoch": 0.66, + "learning_rate": 3.2134858962858824e-06, + "loss": 2.6519, + "step": 3131 + }, + { + "epoch": 0.66, + "learning_rate": 3.2100497304237854e-06, + "loss": 2.618, + "step": 3132 + }, + { + "epoch": 0.66, + "learning_rate": 3.206614533929024e-06, + "loss": 2.5787, + "step": 3133 + }, + { + "epoch": 0.66, + "learning_rate": 3.203180308661965e-06, + "loss": 2.638, + "step": 3134 + }, + { + "epoch": 0.66, + "learning_rate": 3.199747056482454e-06, + "loss": 2.6638, + "step": 3135 + }, + { + "epoch": 0.66, + "learning_rate": 3.1963147792498063e-06, + "loss": 2.6588, + "step": 3136 + }, + { + "epoch": 0.66, + "learning_rate": 3.192883478822807e-06, + "loss": 2.6788, + "step": 3137 + }, + { + "epoch": 0.66, + "learning_rate": 3.1894531570597166e-06, + "loss": 2.5794, + "step": 3138 + }, + { + "epoch": 0.66, + "learning_rate": 3.1860238158182632e-06, + "loss": 2.6137, + "step": 3139 + }, + { + "epoch": 0.66, + "learning_rate": 3.182595456955644e-06, + "loss": 2.6547, + "step": 3140 + }, + { + "epoch": 0.66, + "learning_rate": 3.179168082328525e-06, + "loss": 2.6215, + "step": 3141 + }, + { + "epoch": 0.66, + "learning_rate": 3.1757416937930396e-06, + "loss": 2.608, + "step": 3142 + }, + { + "epoch": 0.66, + "learning_rate": 3.172316293204787e-06, + "loss": 2.6586, + "step": 3143 + }, + { + "epoch": 0.66, + "learning_rate": 3.1688918824188276e-06, + "loss": 2.6381, + "step": 3144 + }, + { + "epoch": 0.66, + "learning_rate": 3.1654684632896917e-06, + "loss": 2.6701, + "step": 3145 + }, + { + "epoch": 0.66, + "learning_rate": 3.1620460376713668e-06, + "loss": 2.6696, + "step": 3146 + }, + { + "epoch": 0.66, + "learning_rate": 3.1586246074173083e-06, + "loss": 2.6676, + "step": 3147 + }, + { + "epoch": 0.66, + "learning_rate": 3.1552041743804283e-06, + "loss": 2.6465, + "step": 3148 + }, + { + "epoch": 0.66, + "learning_rate": 3.1517847404131e-06, + "loss": 2.6085, + "step": 3149 + }, + { + "epoch": 0.66, + "learning_rate": 3.1483663073671554e-06, + "loss": 2.6719, + "step": 3150 + }, + { + "epoch": 0.66, + "learning_rate": 3.1449488770938884e-06, + "loss": 2.731, + "step": 3151 + }, + { + "epoch": 0.66, + "learning_rate": 3.1415324514440392e-06, + "loss": 2.6767, + "step": 3152 + }, + { + "epoch": 0.66, + "learning_rate": 3.1381170322678157e-06, + "loss": 2.5881, + "step": 3153 + }, + { + "epoch": 0.66, + "learning_rate": 3.1347026214148745e-06, + "loss": 2.5711, + "step": 3154 + }, + { + "epoch": 0.66, + "learning_rate": 3.131289220734327e-06, + "loss": 2.6418, + "step": 3155 + }, + { + "epoch": 0.66, + "learning_rate": 3.1278768320747388e-06, + "loss": 2.5887, + "step": 3156 + }, + { + "epoch": 0.66, + "learning_rate": 3.124465457284127e-06, + "loss": 2.6376, + "step": 3157 + }, + { + "epoch": 0.66, + "learning_rate": 3.1210550982099596e-06, + "loss": 2.6593, + "step": 3158 + }, + { + "epoch": 0.66, + "learning_rate": 3.1176457566991524e-06, + "loss": 2.6259, + "step": 3159 + }, + { + "epoch": 0.66, + "learning_rate": 3.1142374345980737e-06, + "loss": 2.6784, + "step": 3160 + }, + { + "epoch": 0.66, + "learning_rate": 3.110830133752536e-06, + "loss": 2.6208, + "step": 3161 + }, + { + "epoch": 0.66, + "learning_rate": 3.107423856007802e-06, + "loss": 2.64, + "step": 3162 + }, + { + "epoch": 0.66, + "learning_rate": 3.1040186032085794e-06, + "loss": 2.5831, + "step": 3163 + }, + { + "epoch": 0.66, + "learning_rate": 3.1006143771990205e-06, + "loss": 2.5487, + "step": 3164 + }, + { + "epoch": 0.66, + "learning_rate": 3.0972111798227224e-06, + "loss": 2.6405, + "step": 3165 + }, + { + "epoch": 0.66, + "learning_rate": 3.0938090129227204e-06, + "loss": 2.6291, + "step": 3166 + }, + { + "epoch": 0.66, + "learning_rate": 3.090407878341498e-06, + "loss": 2.6238, + "step": 3167 + }, + { + "epoch": 0.66, + "learning_rate": 3.0870077779209776e-06, + "loss": 2.5733, + "step": 3168 + }, + { + "epoch": 0.66, + "learning_rate": 3.0836087135025215e-06, + "loss": 2.6736, + "step": 3169 + }, + { + "epoch": 0.66, + "learning_rate": 3.080210686926928e-06, + "loss": 2.6006, + "step": 3170 + }, + { + "epoch": 0.66, + "learning_rate": 3.0768137000344387e-06, + "loss": 2.679, + "step": 3171 + }, + { + "epoch": 0.67, + "learning_rate": 3.0734177546647315e-06, + "loss": 2.5926, + "step": 3172 + }, + { + "epoch": 0.67, + "learning_rate": 3.070022852656911e-06, + "loss": 2.6409, + "step": 3173 + }, + { + "epoch": 0.67, + "learning_rate": 3.066628995849529e-06, + "loss": 2.5855, + "step": 3174 + }, + { + "epoch": 0.67, + "learning_rate": 3.063236186080565e-06, + "loss": 2.6652, + "step": 3175 + }, + { + "epoch": 0.67, + "learning_rate": 3.0598444251874315e-06, + "loss": 2.557, + "step": 3176 + }, + { + "epoch": 0.67, + "learning_rate": 3.0564537150069764e-06, + "loss": 2.7356, + "step": 3177 + }, + { + "epoch": 0.67, + "learning_rate": 3.0530640573754745e-06, + "loss": 2.6295, + "step": 3178 + }, + { + "epoch": 0.67, + "learning_rate": 3.0496754541286346e-06, + "loss": 2.6294, + "step": 3179 + }, + { + "epoch": 0.67, + "learning_rate": 3.0462879071015887e-06, + "loss": 2.6739, + "step": 3180 + }, + { + "epoch": 0.67, + "learning_rate": 3.042901418128903e-06, + "loss": 2.6136, + "step": 3181 + }, + { + "epoch": 0.67, + "learning_rate": 3.0395159890445647e-06, + "loss": 2.5937, + "step": 3182 + }, + { + "epoch": 0.67, + "learning_rate": 3.036131621681995e-06, + "loss": 2.5861, + "step": 3183 + }, + { + "epoch": 0.67, + "learning_rate": 3.0327483178740317e-06, + "loss": 2.6113, + "step": 3184 + }, + { + "epoch": 0.67, + "learning_rate": 3.029366079452943e-06, + "loss": 2.6419, + "step": 3185 + }, + { + "epoch": 0.67, + "learning_rate": 3.0259849082504145e-06, + "loss": 2.6266, + "step": 3186 + }, + { + "epoch": 0.67, + "learning_rate": 3.022604806097561e-06, + "loss": 2.6772, + "step": 3187 + }, + { + "epoch": 0.67, + "learning_rate": 3.0192257748249097e-06, + "loss": 2.7031, + "step": 3188 + }, + { + "epoch": 0.67, + "learning_rate": 3.015847816262415e-06, + "loss": 2.662, + "step": 3189 + }, + { + "epoch": 0.67, + "learning_rate": 3.0124709322394464e-06, + "loss": 2.6051, + "step": 3190 + }, + { + "epoch": 0.67, + "learning_rate": 3.009095124584792e-06, + "loss": 2.6753, + "step": 3191 + }, + { + "epoch": 0.67, + "learning_rate": 3.0057203951266605e-06, + "loss": 2.5631, + "step": 3192 + }, + { + "epoch": 0.67, + "learning_rate": 3.0023467456926723e-06, + "loss": 2.6165, + "step": 3193 + }, + { + "epoch": 0.67, + "learning_rate": 2.9989741781098654e-06, + "loss": 2.5995, + "step": 3194 + }, + { + "epoch": 0.67, + "learning_rate": 2.995602694204689e-06, + "loss": 2.6687, + "step": 3195 + }, + { + "epoch": 0.67, + "learning_rate": 2.992232295803009e-06, + "loss": 2.6649, + "step": 3196 + }, + { + "epoch": 0.67, + "learning_rate": 2.9888629847301e-06, + "loss": 2.6682, + "step": 3197 + }, + { + "epoch": 0.67, + "learning_rate": 2.9854947628106523e-06, + "loss": 2.618, + "step": 3198 + }, + { + "epoch": 0.67, + "learning_rate": 2.982127631868763e-06, + "loss": 2.5816, + "step": 3199 + }, + { + "epoch": 0.67, + "learning_rate": 2.978761593727938e-06, + "loss": 2.7133, + "step": 3200 + }, + { + "epoch": 0.67, + "learning_rate": 2.9753966502110944e-06, + "loss": 2.6364, + "step": 3201 + }, + { + "epoch": 0.67, + "learning_rate": 2.972032803140551e-06, + "loss": 2.6553, + "step": 3202 + }, + { + "epoch": 0.67, + "learning_rate": 2.9686700543380386e-06, + "loss": 2.5985, + "step": 3203 + }, + { + "epoch": 0.67, + "learning_rate": 2.965308405624691e-06, + "loss": 2.6604, + "step": 3204 + }, + { + "epoch": 0.67, + "learning_rate": 2.961947858821046e-06, + "loss": 2.6108, + "step": 3205 + }, + { + "epoch": 0.67, + "learning_rate": 2.9585884157470457e-06, + "loss": 2.6159, + "step": 3206 + }, + { + "epoch": 0.67, + "learning_rate": 2.9552300782220304e-06, + "loss": 2.6721, + "step": 3207 + }, + { + "epoch": 0.67, + "learning_rate": 2.951872848064751e-06, + "loss": 2.6498, + "step": 3208 + }, + { + "epoch": 0.67, + "learning_rate": 2.948516727093345e-06, + "loss": 2.6424, + "step": 3209 + }, + { + "epoch": 0.67, + "learning_rate": 2.9451617171253623e-06, + "loss": 2.6377, + "step": 3210 + }, + { + "epoch": 0.67, + "learning_rate": 2.9418078199777433e-06, + "loss": 2.6523, + "step": 3211 + }, + { + "epoch": 0.67, + "learning_rate": 2.9384550374668276e-06, + "loss": 2.6593, + "step": 3212 + }, + { + "epoch": 0.67, + "learning_rate": 2.935103371408353e-06, + "loss": 2.6177, + "step": 3213 + }, + { + "epoch": 0.67, + "learning_rate": 2.9317528236174515e-06, + "loss": 2.7047, + "step": 3214 + }, + { + "epoch": 0.67, + "learning_rate": 2.9284033959086494e-06, + "loss": 2.7097, + "step": 3215 + }, + { + "epoch": 0.67, + "learning_rate": 2.925055090095864e-06, + "loss": 2.6426, + "step": 3216 + }, + { + "epoch": 0.67, + "learning_rate": 2.9217079079924095e-06, + "loss": 2.6163, + "step": 3217 + }, + { + "epoch": 0.67, + "learning_rate": 2.918361851410987e-06, + "loss": 2.6284, + "step": 3218 + }, + { + "epoch": 0.67, + "learning_rate": 2.9150169221636927e-06, + "loss": 2.6021, + "step": 3219 + }, + { + "epoch": 0.68, + "learning_rate": 2.9116731220620093e-06, + "loss": 2.6056, + "step": 3220 + }, + { + "epoch": 0.68, + "learning_rate": 2.9083304529168087e-06, + "loss": 2.6479, + "step": 3221 + }, + { + "epoch": 0.68, + "learning_rate": 2.9049889165383497e-06, + "loss": 2.5646, + "step": 3222 + }, + { + "epoch": 0.68, + "learning_rate": 2.9016485147362795e-06, + "loss": 2.7021, + "step": 3223 + }, + { + "epoch": 0.68, + "learning_rate": 2.8983092493196286e-06, + "loss": 2.5689, + "step": 3224 + }, + { + "epoch": 0.68, + "learning_rate": 2.8949711220968124e-06, + "loss": 2.5388, + "step": 3225 + }, + { + "epoch": 0.68, + "learning_rate": 2.891634134875632e-06, + "loss": 2.6464, + "step": 3226 + }, + { + "epoch": 0.68, + "learning_rate": 2.8882982894632694e-06, + "loss": 2.6276, + "step": 3227 + }, + { + "epoch": 0.68, + "learning_rate": 2.884963587666287e-06, + "loss": 2.6405, + "step": 3228 + }, + { + "epoch": 0.68, + "learning_rate": 2.88163003129063e-06, + "loss": 2.603, + "step": 3229 + }, + { + "epoch": 0.68, + "learning_rate": 2.8782976221416265e-06, + "loss": 2.5966, + "step": 3230 + }, + { + "epoch": 0.68, + "learning_rate": 2.874966362023974e-06, + "loss": 2.638, + "step": 3231 + }, + { + "epoch": 0.68, + "learning_rate": 2.8716362527417565e-06, + "loss": 2.6658, + "step": 3232 + }, + { + "epoch": 0.68, + "learning_rate": 2.8683072960984294e-06, + "loss": 2.6094, + "step": 3233 + }, + { + "epoch": 0.68, + "learning_rate": 2.8649794938968268e-06, + "loss": 2.6587, + "step": 3234 + }, + { + "epoch": 0.68, + "learning_rate": 2.8616528479391548e-06, + "loss": 2.5874, + "step": 3235 + }, + { + "epoch": 0.68, + "learning_rate": 2.858327360027e-06, + "loss": 2.705, + "step": 3236 + }, + { + "epoch": 0.68, + "learning_rate": 2.8550030319613158e-06, + "loss": 2.6054, + "step": 3237 + }, + { + "epoch": 0.68, + "learning_rate": 2.851679865542426e-06, + "loss": 2.658, + "step": 3238 + }, + { + "epoch": 0.68, + "learning_rate": 2.8483578625700286e-06, + "loss": 2.6401, + "step": 3239 + }, + { + "epoch": 0.68, + "learning_rate": 2.8450370248431924e-06, + "loss": 2.6421, + "step": 3240 + }, + { + "epoch": 0.68, + "learning_rate": 2.8417173541603517e-06, + "loss": 2.6027, + "step": 3241 + }, + { + "epoch": 0.68, + "learning_rate": 2.838398852319313e-06, + "loss": 2.6136, + "step": 3242 + }, + { + "epoch": 0.68, + "learning_rate": 2.835081521117248e-06, + "loss": 2.6882, + "step": 3243 + }, + { + "epoch": 0.68, + "learning_rate": 2.8317653623506946e-06, + "loss": 2.5874, + "step": 3244 + }, + { + "epoch": 0.68, + "learning_rate": 2.8284503778155513e-06, + "loss": 2.6377, + "step": 3245 + }, + { + "epoch": 0.68, + "learning_rate": 2.825136569307087e-06, + "loss": 2.671, + "step": 3246 + }, + { + "epoch": 0.68, + "learning_rate": 2.821823938619929e-06, + "loss": 2.6045, + "step": 3247 + }, + { + "epoch": 0.68, + "learning_rate": 2.8185124875480742e-06, + "loss": 2.6777, + "step": 3248 + }, + { + "epoch": 0.68, + "learning_rate": 2.8152022178848717e-06, + "loss": 2.7056, + "step": 3249 + }, + { + "epoch": 0.68, + "learning_rate": 2.811893131423035e-06, + "loss": 2.6327, + "step": 3250 + }, + { + "epoch": 0.68, + "learning_rate": 2.808585229954637e-06, + "loss": 2.6825, + "step": 3251 + }, + { + "epoch": 0.68, + "learning_rate": 2.8052785152711104e-06, + "loss": 2.6701, + "step": 3252 + }, + { + "epoch": 0.68, + "learning_rate": 2.801972989163236e-06, + "loss": 2.6302, + "step": 3253 + }, + { + "epoch": 0.68, + "learning_rate": 2.7986686534211656e-06, + "loss": 2.5609, + "step": 3254 + }, + { + "epoch": 0.68, + "learning_rate": 2.7953655098343953e-06, + "loss": 2.6873, + "step": 3255 + }, + { + "epoch": 0.68, + "learning_rate": 2.7920635601917796e-06, + "loss": 2.6252, + "step": 3256 + }, + { + "epoch": 0.68, + "learning_rate": 2.7887628062815252e-06, + "loss": 2.6732, + "step": 3257 + }, + { + "epoch": 0.68, + "learning_rate": 2.785463249891193e-06, + "loss": 2.6257, + "step": 3258 + }, + { + "epoch": 0.68, + "learning_rate": 2.782164892807693e-06, + "loss": 2.6259, + "step": 3259 + }, + { + "epoch": 0.68, + "learning_rate": 2.7788677368172877e-06, + "loss": 2.5428, + "step": 3260 + }, + { + "epoch": 0.68, + "learning_rate": 2.7755717837055877e-06, + "loss": 2.6253, + "step": 3261 + }, + { + "epoch": 0.68, + "learning_rate": 2.7722770352575534e-06, + "loss": 2.7079, + "step": 3262 + }, + { + "epoch": 0.68, + "learning_rate": 2.7689834932574923e-06, + "loss": 2.593, + "step": 3263 + }, + { + "epoch": 0.68, + "learning_rate": 2.765691159489058e-06, + "loss": 2.6254, + "step": 3264 + }, + { + "epoch": 0.68, + "learning_rate": 2.76240003573525e-06, + "loss": 2.615, + "step": 3265 + }, + { + "epoch": 0.68, + "learning_rate": 2.7591101237784122e-06, + "loss": 2.614, + "step": 3266 + }, + { + "epoch": 0.69, + "learning_rate": 2.7558214254002346e-06, + "loss": 2.6292, + "step": 3267 + }, + { + "epoch": 0.69, + "learning_rate": 2.752533942381747e-06, + "loss": 2.6416, + "step": 3268 + }, + { + "epoch": 0.69, + "learning_rate": 2.7492476765033227e-06, + "loss": 2.6338, + "step": 3269 + }, + { + "epoch": 0.69, + "learning_rate": 2.745962629544674e-06, + "loss": 2.6786, + "step": 3270 + }, + { + "epoch": 0.69, + "learning_rate": 2.7426788032848533e-06, + "loss": 2.6875, + "step": 3271 + }, + { + "epoch": 0.69, + "learning_rate": 2.7393961995022565e-06, + "loss": 2.6663, + "step": 3272 + }, + { + "epoch": 0.69, + "learning_rate": 2.736114819974615e-06, + "loss": 2.6372, + "step": 3273 + }, + { + "epoch": 0.69, + "learning_rate": 2.7328346664789913e-06, + "loss": 2.6827, + "step": 3274 + }, + { + "epoch": 0.69, + "learning_rate": 2.7295557407917904e-06, + "loss": 2.6294, + "step": 3275 + }, + { + "epoch": 0.69, + "learning_rate": 2.7262780446887517e-06, + "loss": 2.6736, + "step": 3276 + }, + { + "epoch": 0.69, + "learning_rate": 2.723001579944945e-06, + "loss": 2.6584, + "step": 3277 + }, + { + "epoch": 0.69, + "learning_rate": 2.71972634833478e-06, + "loss": 2.625, + "step": 3278 + }, + { + "epoch": 0.69, + "learning_rate": 2.7164523516319923e-06, + "loss": 2.5821, + "step": 3279 + }, + { + "epoch": 0.69, + "learning_rate": 2.7131795916096536e-06, + "loss": 2.67, + "step": 3280 + }, + { + "epoch": 0.69, + "learning_rate": 2.709908070040159e-06, + "loss": 2.6595, + "step": 3281 + }, + { + "epoch": 0.69, + "learning_rate": 2.7066377886952395e-06, + "loss": 2.6904, + "step": 3282 + }, + { + "epoch": 0.69, + "learning_rate": 2.7033687493459493e-06, + "loss": 2.6109, + "step": 3283 + }, + { + "epoch": 0.69, + "learning_rate": 2.7001009537626775e-06, + "loss": 2.682, + "step": 3284 + }, + { + "epoch": 0.69, + "learning_rate": 2.6968344037151323e-06, + "loss": 2.6579, + "step": 3285 + }, + { + "epoch": 0.69, + "learning_rate": 2.6935691009723503e-06, + "loss": 2.5716, + "step": 3286 + }, + { + "epoch": 0.69, + "learning_rate": 2.690305047302692e-06, + "loss": 2.633, + "step": 3287 + }, + { + "epoch": 0.69, + "learning_rate": 2.6870422444738444e-06, + "loss": 2.682, + "step": 3288 + }, + { + "epoch": 0.69, + "learning_rate": 2.683780694252807e-06, + "loss": 2.5611, + "step": 3289 + }, + { + "epoch": 0.69, + "learning_rate": 2.6805203984059156e-06, + "loss": 2.6008, + "step": 3290 + }, + { + "epoch": 0.69, + "learning_rate": 2.6772613586988174e-06, + "loss": 2.6371, + "step": 3291 + }, + { + "epoch": 0.69, + "learning_rate": 2.67400357689648e-06, + "loss": 2.6947, + "step": 3292 + }, + { + "epoch": 0.69, + "learning_rate": 2.670747054763193e-06, + "loss": 2.5905, + "step": 3293 + }, + { + "epoch": 0.69, + "learning_rate": 2.667491794062561e-06, + "loss": 2.6238, + "step": 3294 + }, + { + "epoch": 0.69, + "learning_rate": 2.6642377965575062e-06, + "loss": 2.636, + "step": 3295 + }, + { + "epoch": 0.69, + "learning_rate": 2.6609850640102665e-06, + "loss": 2.684, + "step": 3296 + }, + { + "epoch": 0.69, + "learning_rate": 2.657733598182396e-06, + "loss": 2.6682, + "step": 3297 + }, + { + "epoch": 0.69, + "learning_rate": 2.654483400834762e-06, + "loss": 2.6219, + "step": 3298 + }, + { + "epoch": 0.69, + "learning_rate": 2.6512344737275443e-06, + "loss": 2.692, + "step": 3299 + }, + { + "epoch": 0.69, + "learning_rate": 2.647986818620235e-06, + "loss": 2.6211, + "step": 3300 + }, + { + "epoch": 0.69, + "learning_rate": 2.6447404372716383e-06, + "loss": 2.5803, + "step": 3301 + }, + { + "epoch": 0.69, + "learning_rate": 2.6414953314398673e-06, + "loss": 2.5687, + "step": 3302 + }, + { + "epoch": 0.69, + "learning_rate": 2.638251502882345e-06, + "loss": 2.6743, + "step": 3303 + }, + { + "epoch": 0.69, + "learning_rate": 2.6350089533558033e-06, + "loss": 2.6577, + "step": 3304 + }, + { + "epoch": 0.69, + "learning_rate": 2.63176768461628e-06, + "loss": 2.6257, + "step": 3305 + }, + { + "epoch": 0.69, + "learning_rate": 2.6285276984191197e-06, + "loss": 2.636, + "step": 3306 + }, + { + "epoch": 0.69, + "learning_rate": 2.6252889965189733e-06, + "loss": 2.6179, + "step": 3307 + }, + { + "epoch": 0.69, + "learning_rate": 2.6220515806697934e-06, + "loss": 2.6521, + "step": 3308 + }, + { + "epoch": 0.69, + "learning_rate": 2.618815452624844e-06, + "loss": 2.6578, + "step": 3309 + }, + { + "epoch": 0.69, + "learning_rate": 2.6155806141366804e-06, + "loss": 2.5893, + "step": 3310 + }, + { + "epoch": 0.69, + "learning_rate": 2.6123470669571665e-06, + "loss": 2.6285, + "step": 3311 + }, + { + "epoch": 0.69, + "learning_rate": 2.6091148128374666e-06, + "loss": 2.6017, + "step": 3312 + }, + { + "epoch": 0.69, + "learning_rate": 2.6058838535280404e-06, + "loss": 2.6153, + "step": 3313 + }, + { + "epoch": 0.69, + "learning_rate": 2.602654190778654e-06, + "loss": 2.6918, + "step": 3314 + }, + { + "epoch": 0.7, + "learning_rate": 2.5994258263383654e-06, + "loss": 2.6292, + "step": 3315 + }, + { + "epoch": 0.7, + "learning_rate": 2.596198761955533e-06, + "loss": 2.6618, + "step": 3316 + }, + { + "epoch": 0.7, + "learning_rate": 2.5929729993778046e-06, + "loss": 2.6109, + "step": 3317 + }, + { + "epoch": 0.7, + "learning_rate": 2.5897485403521292e-06, + "loss": 2.6111, + "step": 3318 + }, + { + "epoch": 0.7, + "learning_rate": 2.5865253866247475e-06, + "loss": 2.6379, + "step": 3319 + }, + { + "epoch": 0.7, + "learning_rate": 2.5833035399411977e-06, + "loss": 2.5355, + "step": 3320 + }, + { + "epoch": 0.7, + "learning_rate": 2.580083002046303e-06, + "loss": 2.6201, + "step": 3321 + }, + { + "epoch": 0.7, + "learning_rate": 2.5768637746841827e-06, + "loss": 2.6608, + "step": 3322 + }, + { + "epoch": 0.7, + "learning_rate": 2.573645859598245e-06, + "loss": 2.692, + "step": 3323 + }, + { + "epoch": 0.7, + "learning_rate": 2.570429258531188e-06, + "loss": 2.57, + "step": 3324 + }, + { + "epoch": 0.7, + "learning_rate": 2.567213973224992e-06, + "loss": 2.6286, + "step": 3325 + }, + { + "epoch": 0.7, + "learning_rate": 2.564000005420938e-06, + "loss": 2.6296, + "step": 3326 + }, + { + "epoch": 0.7, + "learning_rate": 2.560787356859582e-06, + "loss": 2.6231, + "step": 3327 + }, + { + "epoch": 0.7, + "learning_rate": 2.5575760292807695e-06, + "loss": 2.5841, + "step": 3328 + }, + { + "epoch": 0.7, + "learning_rate": 2.554366024423631e-06, + "loss": 2.7013, + "step": 3329 + }, + { + "epoch": 0.7, + "learning_rate": 2.551157344026579e-06, + "loss": 2.5843, + "step": 3330 + }, + { + "epoch": 0.7, + "learning_rate": 2.547949989827311e-06, + "loss": 2.5568, + "step": 3331 + }, + { + "epoch": 0.7, + "learning_rate": 2.5447439635628046e-06, + "loss": 2.6525, + "step": 3332 + }, + { + "epoch": 0.7, + "learning_rate": 2.5415392669693183e-06, + "loss": 2.6171, + "step": 3333 + }, + { + "epoch": 0.7, + "learning_rate": 2.5383359017823916e-06, + "loss": 2.7143, + "step": 3334 + }, + { + "epoch": 0.7, + "learning_rate": 2.535133869736842e-06, + "loss": 2.6246, + "step": 3335 + }, + { + "epoch": 0.7, + "learning_rate": 2.531933172566766e-06, + "loss": 2.6322, + "step": 3336 + }, + { + "epoch": 0.7, + "learning_rate": 2.5287338120055356e-06, + "loss": 2.6628, + "step": 3337 + }, + { + "epoch": 0.7, + "learning_rate": 2.5255357897857996e-06, + "loss": 2.6103, + "step": 3338 + }, + { + "epoch": 0.7, + "learning_rate": 2.5223391076394833e-06, + "loss": 2.6192, + "step": 3339 + }, + { + "epoch": 0.7, + "learning_rate": 2.519143767297785e-06, + "loss": 2.7111, + "step": 3340 + }, + { + "epoch": 0.7, + "learning_rate": 2.515949770491175e-06, + "loss": 2.6059, + "step": 3341 + }, + { + "epoch": 0.7, + "learning_rate": 2.5127571189493994e-06, + "loss": 2.6962, + "step": 3342 + }, + { + "epoch": 0.7, + "learning_rate": 2.5095658144014732e-06, + "loss": 2.6902, + "step": 3343 + }, + { + "epoch": 0.7, + "learning_rate": 2.5063758585756814e-06, + "loss": 2.6545, + "step": 3344 + }, + { + "epoch": 0.7, + "learning_rate": 2.5031872531995837e-06, + "loss": 2.6867, + "step": 3345 + }, + { + "epoch": 0.7, + "learning_rate": 2.5000000000000015e-06, + "loss": 2.6567, + "step": 3346 + }, + { + "epoch": 0.7, + "learning_rate": 2.496814100703026e-06, + "loss": 2.6171, + "step": 3347 + }, + { + "epoch": 0.7, + "learning_rate": 2.4936295570340186e-06, + "loss": 2.5968, + "step": 3348 + }, + { + "epoch": 0.7, + "learning_rate": 2.490446370717603e-06, + "loss": 2.6057, + "step": 3349 + }, + { + "epoch": 0.7, + "learning_rate": 2.4872645434776666e-06, + "loss": 2.6744, + "step": 3350 + }, + { + "epoch": 0.7, + "learning_rate": 2.4840840770373675e-06, + "loss": 2.5979, + "step": 3351 + }, + { + "epoch": 0.7, + "learning_rate": 2.4809049731191203e-06, + "loss": 2.6594, + "step": 3352 + }, + { + "epoch": 0.7, + "learning_rate": 2.4777272334446055e-06, + "loss": 2.632, + "step": 3353 + }, + { + "epoch": 0.7, + "learning_rate": 2.47455085973476e-06, + "loss": 2.6233, + "step": 3354 + }, + { + "epoch": 0.7, + "learning_rate": 2.4713758537097837e-06, + "loss": 2.6614, + "step": 3355 + }, + { + "epoch": 0.7, + "learning_rate": 2.4682022170891403e-06, + "loss": 2.7282, + "step": 3356 + }, + { + "epoch": 0.7, + "learning_rate": 2.465029951591545e-06, + "loss": 2.5908, + "step": 3357 + }, + { + "epoch": 0.7, + "learning_rate": 2.461859058934974e-06, + "loss": 2.5746, + "step": 3358 + }, + { + "epoch": 0.7, + "learning_rate": 2.4586895408366585e-06, + "loss": 2.6622, + "step": 3359 + }, + { + "epoch": 0.7, + "learning_rate": 2.455521399013088e-06, + "loss": 2.6237, + "step": 3360 + }, + { + "epoch": 0.7, + "learning_rate": 2.4523546351800003e-06, + "loss": 2.676, + "step": 3361 + }, + { + "epoch": 0.7, + "learning_rate": 2.449189251052396e-06, + "loss": 2.7484, + "step": 3362 + }, + { + "epoch": 0.71, + "learning_rate": 2.446025248344522e-06, + "loss": 2.646, + "step": 3363 + }, + { + "epoch": 0.71, + "learning_rate": 2.4428626287698787e-06, + "loss": 2.6006, + "step": 3364 + }, + { + "epoch": 0.71, + "learning_rate": 2.4397013940412178e-06, + "loss": 2.6755, + "step": 3365 + }, + { + "epoch": 0.71, + "learning_rate": 2.436541545870541e-06, + "loss": 2.563, + "step": 3366 + }, + { + "epoch": 0.71, + "learning_rate": 2.4333830859690995e-06, + "loss": 2.6205, + "step": 3367 + }, + { + "epoch": 0.71, + "learning_rate": 2.4302260160473906e-06, + "loss": 2.5935, + "step": 3368 + }, + { + "epoch": 0.71, + "learning_rate": 2.4270703378151624e-06, + "loss": 2.6303, + "step": 3369 + }, + { + "epoch": 0.71, + "learning_rate": 2.4239160529814054e-06, + "loss": 2.6442, + "step": 3370 + }, + { + "epoch": 0.71, + "learning_rate": 2.420763163254359e-06, + "loss": 2.596, + "step": 3371 + }, + { + "epoch": 0.71, + "learning_rate": 2.4176116703415035e-06, + "loss": 2.63, + "step": 3372 + }, + { + "epoch": 0.71, + "learning_rate": 2.414461575949567e-06, + "loss": 2.6178, + "step": 3373 + }, + { + "epoch": 0.71, + "learning_rate": 2.4113128817845165e-06, + "loss": 2.6752, + "step": 3374 + }, + { + "epoch": 0.71, + "learning_rate": 2.408165589551562e-06, + "loss": 2.6442, + "step": 3375 + }, + { + "epoch": 0.71, + "learning_rate": 2.405019700955155e-06, + "loss": 2.6145, + "step": 3376 + }, + { + "epoch": 0.71, + "learning_rate": 2.4018752176989864e-06, + "loss": 2.6352, + "step": 3377 + }, + { + "epoch": 0.71, + "learning_rate": 2.3987321414859848e-06, + "loss": 2.6965, + "step": 3378 + }, + { + "epoch": 0.71, + "learning_rate": 2.3955904740183187e-06, + "loss": 2.6353, + "step": 3379 + }, + { + "epoch": 0.71, + "learning_rate": 2.392450216997391e-06, + "loss": 2.6467, + "step": 3380 + }, + { + "epoch": 0.71, + "learning_rate": 2.389311372123847e-06, + "loss": 2.681, + "step": 3381 + }, + { + "epoch": 0.71, + "learning_rate": 2.3861739410975586e-06, + "loss": 2.6213, + "step": 3382 + }, + { + "epoch": 0.71, + "learning_rate": 2.383037925617637e-06, + "loss": 2.5872, + "step": 3383 + }, + { + "epoch": 0.71, + "learning_rate": 2.3799033273824275e-06, + "loss": 2.5943, + "step": 3384 + }, + { + "epoch": 0.71, + "learning_rate": 2.376770148089505e-06, + "loss": 2.627, + "step": 3385 + }, + { + "epoch": 0.71, + "learning_rate": 2.373638389435676e-06, + "loss": 2.6031, + "step": 3386 + }, + { + "epoch": 0.71, + "learning_rate": 2.3705080531169823e-06, + "loss": 2.6902, + "step": 3387 + }, + { + "epoch": 0.71, + "learning_rate": 2.3673791408286906e-06, + "loss": 2.6557, + "step": 3388 + }, + { + "epoch": 0.71, + "learning_rate": 2.3642516542652993e-06, + "loss": 2.6765, + "step": 3389 + }, + { + "epoch": 0.71, + "learning_rate": 2.361125595120531e-06, + "loss": 2.6712, + "step": 3390 + }, + { + "epoch": 0.71, + "learning_rate": 2.3580009650873377e-06, + "loss": 2.6518, + "step": 3391 + }, + { + "epoch": 0.71, + "learning_rate": 2.3548777658578964e-06, + "loss": 2.6369, + "step": 3392 + }, + { + "epoch": 0.71, + "learning_rate": 2.3517559991236133e-06, + "loss": 2.6797, + "step": 3393 + }, + { + "epoch": 0.71, + "learning_rate": 2.348635666575114e-06, + "loss": 2.6559, + "step": 3394 + }, + { + "epoch": 0.71, + "learning_rate": 2.3455167699022497e-06, + "loss": 2.6675, + "step": 3395 + }, + { + "epoch": 0.71, + "learning_rate": 2.3423993107940947e-06, + "loss": 2.577, + "step": 3396 + }, + { + "epoch": 0.71, + "learning_rate": 2.3392832909389372e-06, + "loss": 2.6807, + "step": 3397 + }, + { + "epoch": 0.71, + "learning_rate": 2.3361687120242986e-06, + "loss": 2.646, + "step": 3398 + }, + { + "epoch": 0.71, + "learning_rate": 2.3330555757369115e-06, + "loss": 2.6629, + "step": 3399 + }, + { + "epoch": 0.71, + "learning_rate": 2.3299438837627286e-06, + "loss": 2.6561, + "step": 3400 + }, + { + "epoch": 0.71, + "learning_rate": 2.3268336377869222e-06, + "loss": 2.7195, + "step": 3401 + }, + { + "epoch": 0.71, + "learning_rate": 2.3237248394938794e-06, + "loss": 2.512, + "step": 3402 + }, + { + "epoch": 0.71, + "learning_rate": 2.320617490567204e-06, + "loss": 2.6326, + "step": 3403 + }, + { + "epoch": 0.71, + "learning_rate": 2.3175115926897164e-06, + "loss": 2.5709, + "step": 3404 + }, + { + "epoch": 0.71, + "learning_rate": 2.3144071475434487e-06, + "loss": 2.6573, + "step": 3405 + }, + { + "epoch": 0.71, + "learning_rate": 2.3113041568096483e-06, + "loss": 2.6695, + "step": 3406 + }, + { + "epoch": 0.71, + "learning_rate": 2.3082026221687736e-06, + "loss": 2.5491, + "step": 3407 + }, + { + "epoch": 0.71, + "learning_rate": 2.305102545300495e-06, + "loss": 2.5508, + "step": 3408 + }, + { + "epoch": 0.71, + "learning_rate": 2.3020039278836935e-06, + "loss": 2.6626, + "step": 3409 + }, + { + "epoch": 0.72, + "learning_rate": 2.2989067715964592e-06, + "loss": 2.6107, + "step": 3410 + }, + { + "epoch": 0.72, + "learning_rate": 2.295811078116092e-06, + "loss": 2.6512, + "step": 3411 + }, + { + "epoch": 0.72, + "learning_rate": 2.292716849119098e-06, + "loss": 2.7184, + "step": 3412 + }, + { + "epoch": 0.72, + "learning_rate": 2.289624086281192e-06, + "loss": 2.7054, + "step": 3413 + }, + { + "epoch": 0.72, + "learning_rate": 2.2865327912772933e-06, + "loss": 2.6263, + "step": 3414 + }, + { + "epoch": 0.72, + "learning_rate": 2.2834429657815267e-06, + "loss": 2.6309, + "step": 3415 + }, + { + "epoch": 0.72, + "learning_rate": 2.28035461146722e-06, + "loss": 2.6192, + "step": 3416 + }, + { + "epoch": 0.72, + "learning_rate": 2.2772677300069108e-06, + "loss": 2.6293, + "step": 3417 + }, + { + "epoch": 0.72, + "learning_rate": 2.2741823230723283e-06, + "loss": 2.6613, + "step": 3418 + }, + { + "epoch": 0.72, + "learning_rate": 2.2710983923344106e-06, + "loss": 2.6617, + "step": 3419 + }, + { + "epoch": 0.72, + "learning_rate": 2.2680159394632945e-06, + "loss": 2.7381, + "step": 3420 + }, + { + "epoch": 0.72, + "learning_rate": 2.2649349661283167e-06, + "loss": 2.6657, + "step": 3421 + }, + { + "epoch": 0.72, + "learning_rate": 2.26185547399801e-06, + "loss": 2.6326, + "step": 3422 + }, + { + "epoch": 0.72, + "learning_rate": 2.2587774647401106e-06, + "loss": 2.6422, + "step": 3423 + }, + { + "epoch": 0.72, + "learning_rate": 2.255700940021548e-06, + "loss": 2.6784, + "step": 3424 + }, + { + "epoch": 0.72, + "learning_rate": 2.252625901508449e-06, + "loss": 2.6801, + "step": 3425 + }, + { + "epoch": 0.72, + "learning_rate": 2.2495523508661303e-06, + "loss": 2.6679, + "step": 3426 + }, + { + "epoch": 0.72, + "learning_rate": 2.246480289759111e-06, + "loss": 2.6275, + "step": 3427 + }, + { + "epoch": 0.72, + "learning_rate": 2.2434097198510964e-06, + "loss": 2.5827, + "step": 3428 + }, + { + "epoch": 0.72, + "learning_rate": 2.240340642804991e-06, + "loss": 2.6642, + "step": 3429 + }, + { + "epoch": 0.72, + "learning_rate": 2.2372730602828864e-06, + "loss": 2.6484, + "step": 3430 + }, + { + "epoch": 0.72, + "learning_rate": 2.2342069739460654e-06, + "loss": 2.6459, + "step": 3431 + }, + { + "epoch": 0.72, + "learning_rate": 2.231142385455003e-06, + "loss": 2.6566, + "step": 3432 + }, + { + "epoch": 0.72, + "learning_rate": 2.2280792964693572e-06, + "loss": 2.6596, + "step": 3433 + }, + { + "epoch": 0.72, + "learning_rate": 2.2250177086479774e-06, + "loss": 2.5898, + "step": 3434 + }, + { + "epoch": 0.72, + "learning_rate": 2.2219576236489045e-06, + "loss": 2.6088, + "step": 3435 + }, + { + "epoch": 0.72, + "learning_rate": 2.218899043129359e-06, + "loss": 2.7225, + "step": 3436 + }, + { + "epoch": 0.72, + "learning_rate": 2.2158419687457484e-06, + "loss": 2.6754, + "step": 3437 + }, + { + "epoch": 0.72, + "learning_rate": 2.2127864021536664e-06, + "loss": 2.6957, + "step": 3438 + }, + { + "epoch": 0.72, + "learning_rate": 2.2097323450078877e-06, + "loss": 2.6651, + "step": 3439 + }, + { + "epoch": 0.72, + "learning_rate": 2.206679798962372e-06, + "loss": 2.6168, + "step": 3440 + }, + { + "epoch": 0.72, + "learning_rate": 2.2036287656702576e-06, + "loss": 2.6105, + "step": 3441 + }, + { + "epoch": 0.72, + "learning_rate": 2.200579246783867e-06, + "loss": 2.5895, + "step": 3442 + }, + { + "epoch": 0.72, + "learning_rate": 2.1975312439547e-06, + "loss": 2.6839, + "step": 3443 + }, + { + "epoch": 0.72, + "learning_rate": 2.194484758833435e-06, + "loss": 2.6572, + "step": 3444 + }, + { + "epoch": 0.72, + "learning_rate": 2.191439793069932e-06, + "loss": 2.6856, + "step": 3445 + }, + { + "epoch": 0.72, + "learning_rate": 2.1883963483132243e-06, + "loss": 2.686, + "step": 3446 + }, + { + "epoch": 0.72, + "learning_rate": 2.1853544262115235e-06, + "loss": 2.6454, + "step": 3447 + }, + { + "epoch": 0.72, + "learning_rate": 2.1823140284122157e-06, + "loss": 2.6565, + "step": 3448 + }, + { + "epoch": 0.72, + "learning_rate": 2.1792751565618625e-06, + "loss": 2.6106, + "step": 3449 + }, + { + "epoch": 0.72, + "learning_rate": 2.1762378123061986e-06, + "loss": 2.5835, + "step": 3450 + }, + { + "epoch": 0.72, + "learning_rate": 2.17320199729013e-06, + "loss": 2.6857, + "step": 3451 + }, + { + "epoch": 0.72, + "learning_rate": 2.170167713157736e-06, + "loss": 2.6512, + "step": 3452 + }, + { + "epoch": 0.72, + "learning_rate": 2.16713496155227e-06, + "loss": 2.6327, + "step": 3453 + }, + { + "epoch": 0.72, + "learning_rate": 2.1641037441161487e-06, + "loss": 2.6807, + "step": 3454 + }, + { + "epoch": 0.72, + "learning_rate": 2.161074062490962e-06, + "loss": 2.628, + "step": 3455 + }, + { + "epoch": 0.72, + "learning_rate": 2.1580459183174678e-06, + "loss": 2.6095, + "step": 3456 + }, + { + "epoch": 0.72, + "learning_rate": 2.1550193132355905e-06, + "loss": 2.6754, + "step": 3457 + }, + { + "epoch": 0.73, + "learning_rate": 2.1519942488844208e-06, + "loss": 2.6284, + "step": 3458 + }, + { + "epoch": 0.73, + "learning_rate": 2.148970726902219e-06, + "loss": 2.5596, + "step": 3459 + }, + { + "epoch": 0.73, + "learning_rate": 2.1459487489264036e-06, + "loss": 2.591, + "step": 3460 + }, + { + "epoch": 0.73, + "learning_rate": 2.142928316593563e-06, + "loss": 2.6542, + "step": 3461 + }, + { + "epoch": 0.73, + "learning_rate": 2.1399094315394427e-06, + "loss": 2.6664, + "step": 3462 + }, + { + "epoch": 0.73, + "learning_rate": 2.1368920953989543e-06, + "loss": 2.6253, + "step": 3463 + }, + { + "epoch": 0.73, + "learning_rate": 2.133876309806168e-06, + "loss": 2.6899, + "step": 3464 + }, + { + "epoch": 0.73, + "learning_rate": 2.1308620763943187e-06, + "loss": 2.6582, + "step": 3465 + }, + { + "epoch": 0.73, + "learning_rate": 2.127849396795797e-06, + "loss": 2.5929, + "step": 3466 + }, + { + "epoch": 0.73, + "learning_rate": 2.1248382726421525e-06, + "loss": 2.6104, + "step": 3467 + }, + { + "epoch": 0.73, + "learning_rate": 2.1218287055640935e-06, + "loss": 2.5915, + "step": 3468 + }, + { + "epoch": 0.73, + "learning_rate": 2.118820697191482e-06, + "loss": 2.6383, + "step": 3469 + }, + { + "epoch": 0.73, + "learning_rate": 2.1158142491533384e-06, + "loss": 2.6218, + "step": 3470 + }, + { + "epoch": 0.73, + "learning_rate": 2.1128093630778403e-06, + "loss": 2.6522, + "step": 3471 + }, + { + "epoch": 0.73, + "learning_rate": 2.109806040592315e-06, + "loss": 2.6095, + "step": 3472 + }, + { + "epoch": 0.73, + "learning_rate": 2.106804283323246e-06, + "loss": 2.6836, + "step": 3473 + }, + { + "epoch": 0.73, + "learning_rate": 2.1038040928962677e-06, + "loss": 2.6083, + "step": 3474 + }, + { + "epoch": 0.73, + "learning_rate": 2.100805470936168e-06, + "loss": 2.6191, + "step": 3475 + }, + { + "epoch": 0.73, + "learning_rate": 2.0978084190668785e-06, + "loss": 2.5748, + "step": 3476 + }, + { + "epoch": 0.73, + "learning_rate": 2.0948129389114913e-06, + "loss": 2.6282, + "step": 3477 + }, + { + "epoch": 0.73, + "learning_rate": 2.0918190320922394e-06, + "loss": 2.6675, + "step": 3478 + }, + { + "epoch": 0.73, + "learning_rate": 2.088826700230506e-06, + "loss": 2.5935, + "step": 3479 + }, + { + "epoch": 0.73, + "learning_rate": 2.085835944946823e-06, + "loss": 2.6206, + "step": 3480 + }, + { + "epoch": 0.73, + "learning_rate": 2.0828467678608655e-06, + "loss": 2.7017, + "step": 3481 + }, + { + "epoch": 0.73, + "learning_rate": 2.079859170591455e-06, + "loss": 2.6232, + "step": 3482 + }, + { + "epoch": 0.73, + "learning_rate": 2.076873154756559e-06, + "loss": 2.6141, + "step": 3483 + }, + { + "epoch": 0.73, + "learning_rate": 2.0738887219732872e-06, + "loss": 2.6795, + "step": 3484 + }, + { + "epoch": 0.73, + "learning_rate": 2.0709058738578915e-06, + "loss": 2.724, + "step": 3485 + }, + { + "epoch": 0.73, + "learning_rate": 2.0679246120257664e-06, + "loss": 2.6116, + "step": 3486 + }, + { + "epoch": 0.73, + "learning_rate": 2.064944938091447e-06, + "loss": 2.6055, + "step": 3487 + }, + { + "epoch": 0.73, + "learning_rate": 2.0619668536686095e-06, + "loss": 2.6245, + "step": 3488 + }, + { + "epoch": 0.73, + "learning_rate": 2.0589903603700656e-06, + "loss": 2.6513, + "step": 3489 + }, + { + "epoch": 0.73, + "learning_rate": 2.056015459807774e-06, + "loss": 2.6148, + "step": 3490 + }, + { + "epoch": 0.73, + "learning_rate": 2.0530421535928197e-06, + "loss": 2.6387, + "step": 3491 + }, + { + "epoch": 0.73, + "learning_rate": 2.050070443335431e-06, + "loss": 2.6477, + "step": 3492 + }, + { + "epoch": 0.73, + "learning_rate": 2.0471003306449707e-06, + "loss": 2.6491, + "step": 3493 + }, + { + "epoch": 0.73, + "learning_rate": 2.044131817129934e-06, + "loss": 2.5833, + "step": 3494 + }, + { + "epoch": 0.73, + "learning_rate": 2.0411649043979558e-06, + "loss": 2.5909, + "step": 3495 + }, + { + "epoch": 0.73, + "learning_rate": 2.0381995940557985e-06, + "loss": 2.6458, + "step": 3496 + }, + { + "epoch": 0.73, + "learning_rate": 2.0352358877093616e-06, + "loss": 2.6086, + "step": 3497 + }, + { + "epoch": 0.73, + "learning_rate": 2.0322737869636677e-06, + "loss": 2.6542, + "step": 3498 + }, + { + "epoch": 0.73, + "learning_rate": 2.0293132934228783e-06, + "loss": 2.6668, + "step": 3499 + }, + { + "epoch": 0.73, + "learning_rate": 2.0263544086902785e-06, + "loss": 2.6485, + "step": 3500 + }, + { + "epoch": 0.73, + "learning_rate": 2.0233971343682903e-06, + "loss": 2.6154, + "step": 3501 + }, + { + "epoch": 0.73, + "learning_rate": 2.020441472058454e-06, + "loss": 2.6689, + "step": 3502 + }, + { + "epoch": 0.73, + "learning_rate": 2.0174874233614433e-06, + "loss": 2.6054, + "step": 3503 + }, + { + "epoch": 0.73, + "learning_rate": 2.0145349898770566e-06, + "loss": 2.6497, + "step": 3504 + }, + { + "epoch": 0.73, + "learning_rate": 2.011584173204214e-06, + "loss": 2.5996, + "step": 3505 + }, + { + "epoch": 0.74, + "learning_rate": 2.008634974940962e-06, + "loss": 2.6241, + "step": 3506 + }, + { + "epoch": 0.74, + "learning_rate": 2.005687396684475e-06, + "loss": 2.6724, + "step": 3507 + }, + { + "epoch": 0.74, + "learning_rate": 2.002741440031046e-06, + "loss": 2.6305, + "step": 3508 + }, + { + "epoch": 0.74, + "learning_rate": 1.9997971065760897e-06, + "loss": 2.6723, + "step": 3509 + }, + { + "epoch": 0.74, + "learning_rate": 1.9968543979141425e-06, + "loss": 2.5916, + "step": 3510 + }, + { + "epoch": 0.74, + "learning_rate": 1.9939133156388617e-06, + "loss": 2.5927, + "step": 3511 + }, + { + "epoch": 0.74, + "learning_rate": 1.9909738613430187e-06, + "loss": 2.6117, + "step": 3512 + }, + { + "epoch": 0.74, + "learning_rate": 1.988036036618512e-06, + "loss": 2.6612, + "step": 3513 + }, + { + "epoch": 0.74, + "learning_rate": 1.9850998430563516e-06, + "loss": 2.6511, + "step": 3514 + }, + { + "epoch": 0.74, + "learning_rate": 1.982165282246665e-06, + "loss": 2.5899, + "step": 3515 + }, + { + "epoch": 0.74, + "learning_rate": 1.9792323557786975e-06, + "loss": 2.6504, + "step": 3516 + }, + { + "epoch": 0.74, + "learning_rate": 1.976301065240806e-06, + "loss": 2.6478, + "step": 3517 + }, + { + "epoch": 0.74, + "learning_rate": 1.9733714122204646e-06, + "loss": 2.5873, + "step": 3518 + }, + { + "epoch": 0.74, + "learning_rate": 1.970443398304258e-06, + "loss": 2.5796, + "step": 3519 + }, + { + "epoch": 0.74, + "learning_rate": 1.967517025077886e-06, + "loss": 2.6519, + "step": 3520 + }, + { + "epoch": 0.74, + "learning_rate": 1.9645922941261575e-06, + "loss": 2.6118, + "step": 3521 + }, + { + "epoch": 0.74, + "learning_rate": 1.9616692070329924e-06, + "loss": 2.6167, + "step": 3522 + }, + { + "epoch": 0.74, + "learning_rate": 1.9587477653814214e-06, + "loss": 2.6136, + "step": 3523 + }, + { + "epoch": 0.74, + "learning_rate": 1.955827970753583e-06, + "loss": 2.6348, + "step": 3524 + }, + { + "epoch": 0.74, + "learning_rate": 1.952909824730724e-06, + "loss": 2.598, + "step": 3525 + }, + { + "epoch": 0.74, + "learning_rate": 1.9499933288932017e-06, + "loss": 2.7106, + "step": 3526 + }, + { + "epoch": 0.74, + "learning_rate": 1.947078484820472e-06, + "loss": 2.6596, + "step": 3527 + }, + { + "epoch": 0.74, + "learning_rate": 1.944165294091104e-06, + "loss": 2.6417, + "step": 3528 + }, + { + "epoch": 0.74, + "learning_rate": 1.941253758282767e-06, + "loss": 2.663, + "step": 3529 + }, + { + "epoch": 0.74, + "learning_rate": 1.9383438789722353e-06, + "loss": 2.6623, + "step": 3530 + }, + { + "epoch": 0.74, + "learning_rate": 1.9354356577353857e-06, + "loss": 2.5704, + "step": 3531 + }, + { + "epoch": 0.74, + "learning_rate": 1.9325290961471993e-06, + "loss": 2.6312, + "step": 3532 + }, + { + "epoch": 0.74, + "learning_rate": 1.9296241957817575e-06, + "loss": 2.6635, + "step": 3533 + }, + { + "epoch": 0.74, + "learning_rate": 1.926720958212238e-06, + "loss": 2.6584, + "step": 3534 + }, + { + "epoch": 0.74, + "learning_rate": 1.9238193850109225e-06, + "loss": 2.5513, + "step": 3535 + }, + { + "epoch": 0.74, + "learning_rate": 1.9209194777491887e-06, + "loss": 2.5639, + "step": 3536 + }, + { + "epoch": 0.74, + "learning_rate": 1.9180212379975155e-06, + "loss": 2.6285, + "step": 3537 + }, + { + "epoch": 0.74, + "learning_rate": 1.9151246673254765e-06, + "loss": 2.6586, + "step": 3538 + }, + { + "epoch": 0.74, + "learning_rate": 1.912229767301741e-06, + "loss": 2.6302, + "step": 3539 + }, + { + "epoch": 0.74, + "learning_rate": 1.909336539494076e-06, + "loss": 2.6117, + "step": 3540 + }, + { + "epoch": 0.74, + "learning_rate": 1.9064449854693368e-06, + "loss": 2.6944, + "step": 3541 + }, + { + "epoch": 0.74, + "learning_rate": 1.903555106793477e-06, + "loss": 2.6566, + "step": 3542 + }, + { + "epoch": 0.74, + "learning_rate": 1.9006669050315458e-06, + "loss": 2.5892, + "step": 3543 + }, + { + "epoch": 0.74, + "learning_rate": 1.8977803817476787e-06, + "loss": 2.6156, + "step": 3544 + }, + { + "epoch": 0.74, + "learning_rate": 1.894895538505105e-06, + "loss": 2.5728, + "step": 3545 + }, + { + "epoch": 0.74, + "learning_rate": 1.8920123768661419e-06, + "loss": 2.6206, + "step": 3546 + }, + { + "epoch": 0.74, + "learning_rate": 1.8891308983922007e-06, + "loss": 2.6506, + "step": 3547 + }, + { + "epoch": 0.74, + "learning_rate": 1.886251104643772e-06, + "loss": 2.6968, + "step": 3548 + }, + { + "epoch": 0.74, + "learning_rate": 1.8833729971804454e-06, + "loss": 2.6971, + "step": 3549 + }, + { + "epoch": 0.74, + "learning_rate": 1.8804965775608896e-06, + "loss": 2.6351, + "step": 3550 + }, + { + "epoch": 0.74, + "learning_rate": 1.877621847342862e-06, + "loss": 2.6306, + "step": 3551 + }, + { + "epoch": 0.74, + "learning_rate": 1.8747488080832039e-06, + "loss": 2.6209, + "step": 3552 + }, + { + "epoch": 0.75, + "learning_rate": 1.8718774613378415e-06, + "loss": 2.592, + "step": 3553 + }, + { + "epoch": 0.75, + "learning_rate": 1.8690078086617847e-06, + "loss": 2.5672, + "step": 3554 + }, + { + "epoch": 0.75, + "learning_rate": 1.8661398516091256e-06, + "loss": 2.6552, + "step": 3555 + }, + { + "epoch": 0.75, + "learning_rate": 1.863273591733038e-06, + "loss": 2.703, + "step": 3556 + }, + { + "epoch": 0.75, + "learning_rate": 1.8604090305857757e-06, + "loss": 2.6648, + "step": 3557 + }, + { + "epoch": 0.75, + "learning_rate": 1.8575461697186754e-06, + "loss": 2.6309, + "step": 3558 + }, + { + "epoch": 0.75, + "learning_rate": 1.8546850106821485e-06, + "loss": 2.6097, + "step": 3559 + }, + { + "epoch": 0.75, + "learning_rate": 1.851825555025689e-06, + "loss": 2.6121, + "step": 3560 + }, + { + "epoch": 0.75, + "learning_rate": 1.8489678042978647e-06, + "loss": 2.6405, + "step": 3561 + }, + { + "epoch": 0.75, + "learning_rate": 1.8461117600463275e-06, + "loss": 2.6848, + "step": 3562 + }, + { + "epoch": 0.75, + "learning_rate": 1.843257423817793e-06, + "loss": 2.5727, + "step": 3563 + }, + { + "epoch": 0.75, + "learning_rate": 1.8404047971580624e-06, + "loss": 2.5965, + "step": 3564 + }, + { + "epoch": 0.75, + "learning_rate": 1.8375538816120053e-06, + "loss": 2.6179, + "step": 3565 + }, + { + "epoch": 0.75, + "learning_rate": 1.8347046787235677e-06, + "loss": 2.6618, + "step": 3566 + }, + { + "epoch": 0.75, + "learning_rate": 1.8318571900357646e-06, + "loss": 2.6263, + "step": 3567 + }, + { + "epoch": 0.75, + "learning_rate": 1.8290114170906886e-06, + "loss": 2.6556, + "step": 3568 + }, + { + "epoch": 0.75, + "learning_rate": 1.8261673614294996e-06, + "loss": 2.5798, + "step": 3569 + }, + { + "epoch": 0.75, + "learning_rate": 1.8233250245924234e-06, + "loss": 2.6177, + "step": 3570 + }, + { + "epoch": 0.75, + "learning_rate": 1.8204844081187612e-06, + "loss": 2.5891, + "step": 3571 + }, + { + "epoch": 0.75, + "learning_rate": 1.8176455135468796e-06, + "loss": 2.6454, + "step": 3572 + }, + { + "epoch": 0.75, + "learning_rate": 1.8148083424142122e-06, + "loss": 2.5902, + "step": 3573 + }, + { + "epoch": 0.75, + "learning_rate": 1.8119728962572635e-06, + "loss": 2.6804, + "step": 3574 + }, + { + "epoch": 0.75, + "learning_rate": 1.809139176611599e-06, + "loss": 2.6354, + "step": 3575 + }, + { + "epoch": 0.75, + "learning_rate": 1.8063071850118524e-06, + "loss": 2.6373, + "step": 3576 + }, + { + "epoch": 0.75, + "learning_rate": 1.803476922991717e-06, + "loss": 2.6442, + "step": 3577 + }, + { + "epoch": 0.75, + "learning_rate": 1.8006483920839524e-06, + "loss": 2.642, + "step": 3578 + }, + { + "epoch": 0.75, + "learning_rate": 1.7978215938203836e-06, + "loss": 2.5529, + "step": 3579 + }, + { + "epoch": 0.75, + "learning_rate": 1.7949965297318934e-06, + "loss": 2.707, + "step": 3580 + }, + { + "epoch": 0.75, + "learning_rate": 1.792173201348426e-06, + "loss": 2.6202, + "step": 3581 + }, + { + "epoch": 0.75, + "learning_rate": 1.789351610198986e-06, + "loss": 2.6143, + "step": 3582 + }, + { + "epoch": 0.75, + "learning_rate": 1.7865317578116397e-06, + "loss": 2.6621, + "step": 3583 + }, + { + "epoch": 0.75, + "learning_rate": 1.7837136457135035e-06, + "loss": 2.596, + "step": 3584 + }, + { + "epoch": 0.75, + "learning_rate": 1.7808972754307623e-06, + "loss": 2.5902, + "step": 3585 + }, + { + "epoch": 0.75, + "learning_rate": 1.7780826484886509e-06, + "loss": 2.6649, + "step": 3586 + }, + { + "epoch": 0.75, + "learning_rate": 1.7752697664114621e-06, + "loss": 2.592, + "step": 3587 + }, + { + "epoch": 0.75, + "learning_rate": 1.7724586307225422e-06, + "loss": 2.664, + "step": 3588 + }, + { + "epoch": 0.75, + "learning_rate": 1.769649242944294e-06, + "loss": 2.6078, + "step": 3589 + }, + { + "epoch": 0.75, + "learning_rate": 1.7668416045981712e-06, + "loss": 2.6072, + "step": 3590 + }, + { + "epoch": 0.75, + "learning_rate": 1.7640357172046824e-06, + "loss": 2.5965, + "step": 3591 + }, + { + "epoch": 0.75, + "learning_rate": 1.7612315822833865e-06, + "loss": 2.6442, + "step": 3592 + }, + { + "epoch": 0.75, + "learning_rate": 1.7584292013528935e-06, + "loss": 2.6625, + "step": 3593 + }, + { + "epoch": 0.75, + "learning_rate": 1.755628575930864e-06, + "loss": 2.6217, + "step": 3594 + }, + { + "epoch": 0.75, + "learning_rate": 1.7528297075340077e-06, + "loss": 2.5634, + "step": 3595 + }, + { + "epoch": 0.75, + "learning_rate": 1.7500325976780824e-06, + "loss": 2.675, + "step": 3596 + }, + { + "epoch": 0.75, + "learning_rate": 1.7472372478778926e-06, + "loss": 2.6133, + "step": 3597 + }, + { + "epoch": 0.75, + "learning_rate": 1.7444436596472959e-06, + "loss": 2.6444, + "step": 3598 + }, + { + "epoch": 0.75, + "learning_rate": 1.741651834499185e-06, + "loss": 2.6493, + "step": 3599 + }, + { + "epoch": 0.75, + "learning_rate": 1.738861773945506e-06, + "loss": 2.6013, + "step": 3600 + }, + { + "epoch": 0.76, + "learning_rate": 1.7360734794972473e-06, + "loss": 2.6538, + "step": 3601 + }, + { + "epoch": 0.76, + "learning_rate": 1.7332869526644396e-06, + "loss": 2.6077, + "step": 3602 + }, + { + "epoch": 0.76, + "learning_rate": 1.7305021949561563e-06, + "loss": 2.6691, + "step": 3603 + }, + { + "epoch": 0.76, + "learning_rate": 1.7277192078805167e-06, + "loss": 2.6458, + "step": 3604 + }, + { + "epoch": 0.76, + "learning_rate": 1.7249379929446786e-06, + "loss": 2.6536, + "step": 3605 + }, + { + "epoch": 0.76, + "learning_rate": 1.7221585516548362e-06, + "loss": 2.6216, + "step": 3606 + }, + { + "epoch": 0.76, + "learning_rate": 1.7193808855162274e-06, + "loss": 2.6223, + "step": 3607 + }, + { + "epoch": 0.76, + "learning_rate": 1.71660499603313e-06, + "loss": 2.6159, + "step": 3608 + }, + { + "epoch": 0.76, + "learning_rate": 1.7138308847088535e-06, + "loss": 2.61, + "step": 3609 + }, + { + "epoch": 0.76, + "learning_rate": 1.7110585530457535e-06, + "loss": 2.6659, + "step": 3610 + }, + { + "epoch": 0.76, + "learning_rate": 1.7082880025452147e-06, + "loss": 2.67, + "step": 3611 + }, + { + "epoch": 0.76, + "learning_rate": 1.705519234707661e-06, + "loss": 2.6383, + "step": 3612 + }, + { + "epoch": 0.76, + "learning_rate": 1.7027522510325461e-06, + "loss": 2.5956, + "step": 3613 + }, + { + "epoch": 0.76, + "learning_rate": 1.6999870530183615e-06, + "loss": 2.6667, + "step": 3614 + }, + { + "epoch": 0.76, + "learning_rate": 1.6972236421626297e-06, + "loss": 2.5969, + "step": 3615 + }, + { + "epoch": 0.76, + "learning_rate": 1.6944620199619095e-06, + "loss": 2.5848, + "step": 3616 + }, + { + "epoch": 0.76, + "learning_rate": 1.6917021879117861e-06, + "loss": 2.5921, + "step": 3617 + }, + { + "epoch": 0.76, + "learning_rate": 1.6889441475068774e-06, + "loss": 2.6613, + "step": 3618 + }, + { + "epoch": 0.76, + "learning_rate": 1.6861879002408322e-06, + "loss": 2.6522, + "step": 3619 + }, + { + "epoch": 0.76, + "learning_rate": 1.6834334476063214e-06, + "loss": 2.6588, + "step": 3620 + }, + { + "epoch": 0.76, + "learning_rate": 1.6806807910950546e-06, + "loss": 2.6219, + "step": 3621 + }, + { + "epoch": 0.76, + "learning_rate": 1.677929932197761e-06, + "loss": 2.661, + "step": 3622 + }, + { + "epoch": 0.76, + "learning_rate": 1.6751808724041996e-06, + "loss": 2.5871, + "step": 3623 + }, + { + "epoch": 0.76, + "learning_rate": 1.6724336132031532e-06, + "loss": 2.6939, + "step": 3624 + }, + { + "epoch": 0.76, + "learning_rate": 1.6696881560824308e-06, + "loss": 2.6159, + "step": 3625 + }, + { + "epoch": 0.76, + "learning_rate": 1.6669445025288649e-06, + "loss": 2.7021, + "step": 3626 + }, + { + "epoch": 0.76, + "learning_rate": 1.6642026540283112e-06, + "loss": 2.5752, + "step": 3627 + }, + { + "epoch": 0.76, + "learning_rate": 1.6614626120656484e-06, + "loss": 2.6396, + "step": 3628 + }, + { + "epoch": 0.76, + "learning_rate": 1.6587243781247764e-06, + "loss": 2.5942, + "step": 3629 + }, + { + "epoch": 0.76, + "learning_rate": 1.655987953688616e-06, + "loss": 2.6718, + "step": 3630 + }, + { + "epoch": 0.76, + "learning_rate": 1.6532533402391081e-06, + "loss": 2.6605, + "step": 3631 + }, + { + "epoch": 0.76, + "learning_rate": 1.6505205392572128e-06, + "loss": 2.666, + "step": 3632 + }, + { + "epoch": 0.76, + "learning_rate": 1.647789552222908e-06, + "loss": 2.6434, + "step": 3633 + }, + { + "epoch": 0.76, + "learning_rate": 1.6450603806151911e-06, + "loss": 2.6241, + "step": 3634 + }, + { + "epoch": 0.76, + "learning_rate": 1.642333025912074e-06, + "loss": 2.6921, + "step": 3635 + }, + { + "epoch": 0.76, + "learning_rate": 1.6396074895905866e-06, + "loss": 2.6721, + "step": 3636 + }, + { + "epoch": 0.76, + "learning_rate": 1.6368837731267727e-06, + "loss": 2.6548, + "step": 3637 + }, + { + "epoch": 0.76, + "learning_rate": 1.6341618779956913e-06, + "loss": 2.6162, + "step": 3638 + }, + { + "epoch": 0.76, + "learning_rate": 1.6314418056714133e-06, + "loss": 2.6732, + "step": 3639 + }, + { + "epoch": 0.76, + "learning_rate": 1.6287235576270267e-06, + "loss": 2.6125, + "step": 3640 + }, + { + "epoch": 0.76, + "learning_rate": 1.626007135334629e-06, + "loss": 2.6593, + "step": 3641 + }, + { + "epoch": 0.76, + "learning_rate": 1.6232925402653248e-06, + "loss": 2.6043, + "step": 3642 + }, + { + "epoch": 0.76, + "learning_rate": 1.620579773889236e-06, + "loss": 2.574, + "step": 3643 + }, + { + "epoch": 0.76, + "learning_rate": 1.6178688376754896e-06, + "loss": 2.6089, + "step": 3644 + }, + { + "epoch": 0.76, + "learning_rate": 1.615159733092223e-06, + "loss": 2.6198, + "step": 3645 + }, + { + "epoch": 0.76, + "learning_rate": 1.612452461606584e-06, + "loss": 2.6196, + "step": 3646 + }, + { + "epoch": 0.76, + "learning_rate": 1.6097470246847236e-06, + "loss": 2.6166, + "step": 3647 + }, + { + "epoch": 0.76, + "learning_rate": 1.6070434237918038e-06, + "loss": 2.6681, + "step": 3648 + }, + { + "epoch": 0.77, + "learning_rate": 1.6043416603919848e-06, + "loss": 2.6931, + "step": 3649 + }, + { + "epoch": 0.77, + "learning_rate": 1.6016417359484388e-06, + "loss": 2.6001, + "step": 3650 + }, + { + "epoch": 0.77, + "learning_rate": 1.598943651923338e-06, + "loss": 2.6074, + "step": 3651 + }, + { + "epoch": 0.77, + "learning_rate": 1.596247409777863e-06, + "loss": 2.6742, + "step": 3652 + }, + { + "epoch": 0.77, + "learning_rate": 1.5935530109721915e-06, + "loss": 2.6677, + "step": 3653 + }, + { + "epoch": 0.77, + "learning_rate": 1.5908604569655051e-06, + "loss": 2.6035, + "step": 3654 + }, + { + "epoch": 0.77, + "learning_rate": 1.588169749215988e-06, + "loss": 2.6502, + "step": 3655 + }, + { + "epoch": 0.77, + "learning_rate": 1.5854808891808192e-06, + "loss": 2.6199, + "step": 3656 + }, + { + "epoch": 0.77, + "learning_rate": 1.5827938783161806e-06, + "loss": 2.6648, + "step": 3657 + }, + { + "epoch": 0.77, + "learning_rate": 1.5801087180772562e-06, + "loss": 2.6604, + "step": 3658 + }, + { + "epoch": 0.77, + "learning_rate": 1.5774254099182217e-06, + "loss": 2.6629, + "step": 3659 + }, + { + "epoch": 0.77, + "learning_rate": 1.5747439552922529e-06, + "loss": 2.5854, + "step": 3660 + }, + { + "epoch": 0.77, + "learning_rate": 1.5720643556515214e-06, + "loss": 2.6913, + "step": 3661 + }, + { + "epoch": 0.77, + "learning_rate": 1.5693866124471935e-06, + "loss": 2.6157, + "step": 3662 + }, + { + "epoch": 0.77, + "learning_rate": 1.5667107271294307e-06, + "loss": 2.6573, + "step": 3663 + }, + { + "epoch": 0.77, + "learning_rate": 1.5640367011473884e-06, + "loss": 2.7327, + "step": 3664 + }, + { + "epoch": 0.77, + "learning_rate": 1.5613645359492141e-06, + "loss": 2.5841, + "step": 3665 + }, + { + "epoch": 0.77, + "learning_rate": 1.5586942329820498e-06, + "loss": 2.6402, + "step": 3666 + }, + { + "epoch": 0.77, + "learning_rate": 1.556025793692027e-06, + "loss": 2.67, + "step": 3667 + }, + { + "epoch": 0.77, + "learning_rate": 1.5533592195242674e-06, + "loss": 2.6807, + "step": 3668 + }, + { + "epoch": 0.77, + "learning_rate": 1.5506945119228855e-06, + "loss": 2.6066, + "step": 3669 + }, + { + "epoch": 0.77, + "learning_rate": 1.5480316723309819e-06, + "loss": 2.6032, + "step": 3670 + }, + { + "epoch": 0.77, + "learning_rate": 1.5453707021906467e-06, + "loss": 2.6059, + "step": 3671 + }, + { + "epoch": 0.77, + "learning_rate": 1.5427116029429584e-06, + "loss": 2.6693, + "step": 3672 + }, + { + "epoch": 0.77, + "learning_rate": 1.5400543760279813e-06, + "loss": 2.6788, + "step": 3673 + }, + { + "epoch": 0.77, + "learning_rate": 1.5373990228847657e-06, + "loss": 2.6429, + "step": 3674 + }, + { + "epoch": 0.77, + "learning_rate": 1.534745544951346e-06, + "loss": 2.6884, + "step": 3675 + }, + { + "epoch": 0.77, + "learning_rate": 1.5320939436647458e-06, + "loss": 2.6176, + "step": 3676 + }, + { + "epoch": 0.77, + "learning_rate": 1.529444220460969e-06, + "loss": 2.7074, + "step": 3677 + }, + { + "epoch": 0.77, + "learning_rate": 1.5267963767749989e-06, + "loss": 2.6732, + "step": 3678 + }, + { + "epoch": 0.77, + "learning_rate": 1.5241504140408065e-06, + "loss": 2.6725, + "step": 3679 + }, + { + "epoch": 0.77, + "learning_rate": 1.5215063336913421e-06, + "loss": 2.5953, + "step": 3680 + }, + { + "epoch": 0.77, + "learning_rate": 1.518864137158535e-06, + "loss": 2.5773, + "step": 3681 + }, + { + "epoch": 0.77, + "learning_rate": 1.5162238258732991e-06, + "loss": 2.5662, + "step": 3682 + }, + { + "epoch": 0.77, + "learning_rate": 1.5135854012655227e-06, + "loss": 2.6394, + "step": 3683 + }, + { + "epoch": 0.77, + "learning_rate": 1.5109488647640746e-06, + "loss": 2.5505, + "step": 3684 + }, + { + "epoch": 0.77, + "learning_rate": 1.5083142177967986e-06, + "loss": 2.6043, + "step": 3685 + }, + { + "epoch": 0.77, + "learning_rate": 1.5056814617905168e-06, + "loss": 2.6918, + "step": 3686 + }, + { + "epoch": 0.77, + "learning_rate": 1.5030505981710263e-06, + "loss": 2.6577, + "step": 3687 + }, + { + "epoch": 0.77, + "learning_rate": 1.500421628363104e-06, + "loss": 2.6117, + "step": 3688 + }, + { + "epoch": 0.77, + "learning_rate": 1.4977945537904953e-06, + "loss": 2.6303, + "step": 3689 + }, + { + "epoch": 0.77, + "learning_rate": 1.4951693758759217e-06, + "loss": 2.6552, + "step": 3690 + }, + { + "epoch": 0.77, + "learning_rate": 1.4925460960410792e-06, + "loss": 2.6516, + "step": 3691 + }, + { + "epoch": 0.77, + "learning_rate": 1.4899247157066303e-06, + "loss": 2.69, + "step": 3692 + }, + { + "epoch": 0.77, + "learning_rate": 1.4873052362922125e-06, + "loss": 2.5677, + "step": 3693 + }, + { + "epoch": 0.77, + "learning_rate": 1.4846876592164372e-06, + "loss": 2.6716, + "step": 3694 + }, + { + "epoch": 0.77, + "learning_rate": 1.4820719858968807e-06, + "loss": 2.6287, + "step": 3695 + }, + { + "epoch": 0.78, + "learning_rate": 1.4794582177500893e-06, + "loss": 2.6475, + "step": 3696 + }, + { + "epoch": 0.78, + "learning_rate": 1.4768463561915785e-06, + "loss": 2.5837, + "step": 3697 + }, + { + "epoch": 0.78, + "learning_rate": 1.4742364026358307e-06, + "loss": 2.6549, + "step": 3698 + }, + { + "epoch": 0.78, + "learning_rate": 1.4716283584962948e-06, + "loss": 2.6012, + "step": 3699 + }, + { + "epoch": 0.78, + "learning_rate": 1.4690222251853857e-06, + "loss": 2.6426, + "step": 3700 + }, + { + "epoch": 0.78, + "learning_rate": 1.4664180041144843e-06, + "loss": 2.6348, + "step": 3701 + }, + { + "epoch": 0.78, + "learning_rate": 1.4638156966939354e-06, + "loss": 2.6529, + "step": 3702 + }, + { + "epoch": 0.78, + "learning_rate": 1.4612153043330463e-06, + "loss": 2.6858, + "step": 3703 + }, + { + "epoch": 0.78, + "learning_rate": 1.4586168284400893e-06, + "loss": 2.6044, + "step": 3704 + }, + { + "epoch": 0.78, + "learning_rate": 1.456020270422297e-06, + "loss": 2.6205, + "step": 3705 + }, + { + "epoch": 0.78, + "learning_rate": 1.4534256316858642e-06, + "loss": 2.6596, + "step": 3706 + }, + { + "epoch": 0.78, + "learning_rate": 1.4508329136359462e-06, + "loss": 2.6117, + "step": 3707 + }, + { + "epoch": 0.78, + "learning_rate": 1.4482421176766582e-06, + "loss": 2.6397, + "step": 3708 + }, + { + "epoch": 0.78, + "learning_rate": 1.4456532452110744e-06, + "loss": 2.7371, + "step": 3709 + }, + { + "epoch": 0.78, + "learning_rate": 1.4430662976412268e-06, + "loss": 2.6146, + "step": 3710 + }, + { + "epoch": 0.78, + "learning_rate": 1.4404812763681064e-06, + "loss": 2.6998, + "step": 3711 + }, + { + "epoch": 0.78, + "learning_rate": 1.4378981827916577e-06, + "loss": 2.6303, + "step": 3712 + }, + { + "epoch": 0.78, + "learning_rate": 1.4353170183107884e-06, + "loss": 2.6379, + "step": 3713 + }, + { + "epoch": 0.78, + "learning_rate": 1.432737784323352e-06, + "loss": 2.6644, + "step": 3714 + }, + { + "epoch": 0.78, + "learning_rate": 1.4301604822261621e-06, + "loss": 2.6381, + "step": 3715 + }, + { + "epoch": 0.78, + "learning_rate": 1.4275851134149864e-06, + "loss": 2.5362, + "step": 3716 + }, + { + "epoch": 0.78, + "learning_rate": 1.425011679284542e-06, + "loss": 2.6457, + "step": 3717 + }, + { + "epoch": 0.78, + "learning_rate": 1.4224401812285033e-06, + "loss": 2.6561, + "step": 3718 + }, + { + "epoch": 0.78, + "learning_rate": 1.4198706206394924e-06, + "loss": 2.6711, + "step": 3719 + }, + { + "epoch": 0.78, + "learning_rate": 1.4173029989090847e-06, + "loss": 2.6294, + "step": 3720 + }, + { + "epoch": 0.78, + "learning_rate": 1.4147373174278011e-06, + "loss": 2.6064, + "step": 3721 + }, + { + "epoch": 0.78, + "learning_rate": 1.4121735775851164e-06, + "loss": 2.6907, + "step": 3722 + }, + { + "epoch": 0.78, + "learning_rate": 1.4096117807694504e-06, + "loss": 2.6333, + "step": 3723 + }, + { + "epoch": 0.78, + "learning_rate": 1.4070519283681749e-06, + "loss": 2.6036, + "step": 3724 + }, + { + "epoch": 0.78, + "learning_rate": 1.4044940217676061e-06, + "loss": 2.6511, + "step": 3725 + }, + { + "epoch": 0.78, + "learning_rate": 1.401938062353006e-06, + "loss": 2.6102, + "step": 3726 + }, + { + "epoch": 0.78, + "learning_rate": 1.3993840515085837e-06, + "loss": 2.6015, + "step": 3727 + }, + { + "epoch": 0.78, + "learning_rate": 1.3968319906174893e-06, + "loss": 2.642, + "step": 3728 + }, + { + "epoch": 0.78, + "learning_rate": 1.3942818810618198e-06, + "loss": 2.6427, + "step": 3729 + }, + { + "epoch": 0.78, + "learning_rate": 1.3917337242226175e-06, + "loss": 2.6924, + "step": 3730 + }, + { + "epoch": 0.78, + "learning_rate": 1.3891875214798644e-06, + "loss": 2.5507, + "step": 3731 + }, + { + "epoch": 0.78, + "learning_rate": 1.3866432742124836e-06, + "loss": 2.6707, + "step": 3732 + }, + { + "epoch": 0.78, + "learning_rate": 1.384100983798341e-06, + "loss": 2.6587, + "step": 3733 + }, + { + "epoch": 0.78, + "learning_rate": 1.3815606516142422e-06, + "loss": 2.6932, + "step": 3734 + }, + { + "epoch": 0.78, + "learning_rate": 1.3790222790359314e-06, + "loss": 2.6008, + "step": 3735 + }, + { + "epoch": 0.78, + "learning_rate": 1.3764858674380932e-06, + "loss": 2.5987, + "step": 3736 + }, + { + "epoch": 0.78, + "learning_rate": 1.3739514181943486e-06, + "loss": 2.6014, + "step": 3737 + }, + { + "epoch": 0.78, + "learning_rate": 1.3714189326772564e-06, + "loss": 2.6009, + "step": 3738 + }, + { + "epoch": 0.78, + "learning_rate": 1.368888412258313e-06, + "loss": 2.6602, + "step": 3739 + }, + { + "epoch": 0.78, + "learning_rate": 1.366359858307949e-06, + "loss": 2.6479, + "step": 3740 + }, + { + "epoch": 0.78, + "learning_rate": 1.3638332721955312e-06, + "loss": 2.6655, + "step": 3741 + }, + { + "epoch": 0.78, + "learning_rate": 1.3613086552893602e-06, + "loss": 2.6023, + "step": 3742 + }, + { + "epoch": 0.78, + "learning_rate": 1.358786008956669e-06, + "loss": 2.6574, + "step": 3743 + }, + { + "epoch": 0.79, + "learning_rate": 1.3562653345636262e-06, + "loss": 2.6863, + "step": 3744 + }, + { + "epoch": 0.79, + "learning_rate": 1.3537466334753296e-06, + "loss": 2.6558, + "step": 3745 + }, + { + "epoch": 0.79, + "learning_rate": 1.3512299070558104e-06, + "loss": 2.692, + "step": 3746 + }, + { + "epoch": 0.79, + "learning_rate": 1.3487151566680295e-06, + "loss": 2.5581, + "step": 3747 + }, + { + "epoch": 0.79, + "learning_rate": 1.3462023836738759e-06, + "loss": 2.6544, + "step": 3748 + }, + { + "epoch": 0.79, + "learning_rate": 1.343691589434174e-06, + "loss": 2.6981, + "step": 3749 + }, + { + "epoch": 0.79, + "learning_rate": 1.3411827753086682e-06, + "loss": 2.628, + "step": 3750 + }, + { + "epoch": 0.79, + "learning_rate": 1.3386759426560359e-06, + "loss": 2.6528, + "step": 3751 + }, + { + "epoch": 0.79, + "learning_rate": 1.336171092833879e-06, + "loss": 2.5667, + "step": 3752 + }, + { + "epoch": 0.79, + "learning_rate": 1.3336682271987278e-06, + "loss": 2.6786, + "step": 3753 + }, + { + "epoch": 0.79, + "learning_rate": 1.3311673471060349e-06, + "loss": 2.6037, + "step": 3754 + }, + { + "epoch": 0.79, + "learning_rate": 1.3286684539101823e-06, + "loss": 2.6302, + "step": 3755 + }, + { + "epoch": 0.79, + "learning_rate": 1.3261715489644732e-06, + "loss": 2.5864, + "step": 3756 + }, + { + "epoch": 0.79, + "learning_rate": 1.3236766336211316e-06, + "loss": 2.623, + "step": 3757 + }, + { + "epoch": 0.79, + "learning_rate": 1.3211837092313074e-06, + "loss": 2.6015, + "step": 3758 + }, + { + "epoch": 0.79, + "learning_rate": 1.3186927771450702e-06, + "loss": 2.665, + "step": 3759 + }, + { + "epoch": 0.79, + "learning_rate": 1.3162038387114147e-06, + "loss": 2.6737, + "step": 3760 + }, + { + "epoch": 0.79, + "learning_rate": 1.3137168952782514e-06, + "loss": 2.7065, + "step": 3761 + }, + { + "epoch": 0.79, + "learning_rate": 1.3112319481924119e-06, + "loss": 2.6264, + "step": 3762 + }, + { + "epoch": 0.79, + "learning_rate": 1.308748998799647e-06, + "loss": 2.6226, + "step": 3763 + }, + { + "epoch": 0.79, + "learning_rate": 1.3062680484446267e-06, + "loss": 2.5993, + "step": 3764 + }, + { + "epoch": 0.79, + "learning_rate": 1.3037890984709323e-06, + "loss": 2.6821, + "step": 3765 + }, + { + "epoch": 0.79, + "learning_rate": 1.3013121502210713e-06, + "loss": 2.6163, + "step": 3766 + }, + { + "epoch": 0.79, + "learning_rate": 1.298837205036461e-06, + "loss": 2.6987, + "step": 3767 + }, + { + "epoch": 0.79, + "learning_rate": 1.2963642642574353e-06, + "loss": 2.5715, + "step": 3768 + }, + { + "epoch": 0.79, + "learning_rate": 1.293893329223242e-06, + "loss": 2.6823, + "step": 3769 + }, + { + "epoch": 0.79, + "learning_rate": 1.291424401272044e-06, + "loss": 2.6397, + "step": 3770 + }, + { + "epoch": 0.79, + "learning_rate": 1.2889574817409161e-06, + "loss": 2.5884, + "step": 3771 + }, + { + "epoch": 0.79, + "learning_rate": 1.2864925719658462e-06, + "loss": 2.5862, + "step": 3772 + }, + { + "epoch": 0.79, + "learning_rate": 1.2840296732817332e-06, + "loss": 2.6102, + "step": 3773 + }, + { + "epoch": 0.79, + "learning_rate": 1.2815687870223863e-06, + "loss": 2.6386, + "step": 3774 + }, + { + "epoch": 0.79, + "learning_rate": 1.279109914520527e-06, + "loss": 2.6362, + "step": 3775 + }, + { + "epoch": 0.79, + "learning_rate": 1.276653057107784e-06, + "loss": 2.6306, + "step": 3776 + }, + { + "epoch": 0.79, + "learning_rate": 1.274198216114696e-06, + "loss": 2.6385, + "step": 3777 + }, + { + "epoch": 0.79, + "learning_rate": 1.2717453928707085e-06, + "loss": 2.6104, + "step": 3778 + }, + { + "epoch": 0.79, + "learning_rate": 1.2692945887041763e-06, + "loss": 2.6126, + "step": 3779 + }, + { + "epoch": 0.79, + "learning_rate": 1.2668458049423583e-06, + "loss": 2.6543, + "step": 3780 + }, + { + "epoch": 0.79, + "learning_rate": 1.2643990429114206e-06, + "loss": 2.6323, + "step": 3781 + }, + { + "epoch": 0.79, + "learning_rate": 1.261954303936434e-06, + "loss": 2.623, + "step": 3782 + }, + { + "epoch": 0.79, + "learning_rate": 1.259511589341375e-06, + "loss": 2.5682, + "step": 3783 + }, + { + "epoch": 0.79, + "learning_rate": 1.2570709004491205e-06, + "loss": 2.6385, + "step": 3784 + }, + { + "epoch": 0.79, + "learning_rate": 1.2546322385814564e-06, + "loss": 2.5413, + "step": 3785 + }, + { + "epoch": 0.79, + "learning_rate": 1.2521956050590633e-06, + "loss": 2.5956, + "step": 3786 + }, + { + "epoch": 0.79, + "learning_rate": 1.2497610012015282e-06, + "loss": 2.5804, + "step": 3787 + }, + { + "epoch": 0.79, + "learning_rate": 1.2473284283273373e-06, + "loss": 2.604, + "step": 3788 + }, + { + "epoch": 0.79, + "learning_rate": 1.2448978877538782e-06, + "loss": 2.5656, + "step": 3789 + }, + { + "epoch": 0.79, + "learning_rate": 1.2424693807974352e-06, + "loss": 2.632, + "step": 3790 + }, + { + "epoch": 0.79, + "learning_rate": 1.2400429087731952e-06, + "loss": 2.6691, + "step": 3791 + }, + { + "epoch": 0.8, + "learning_rate": 1.2376184729952422e-06, + "loss": 2.6259, + "step": 3792 + }, + { + "epoch": 0.8, + "learning_rate": 1.2351960747765524e-06, + "loss": 2.7161, + "step": 3793 + }, + { + "epoch": 0.8, + "learning_rate": 1.2327757154290037e-06, + "loss": 2.5999, + "step": 3794 + }, + { + "epoch": 0.8, + "learning_rate": 1.2303573962633692e-06, + "loss": 2.6209, + "step": 3795 + }, + { + "epoch": 0.8, + "learning_rate": 1.2279411185893136e-06, + "loss": 2.6388, + "step": 3796 + }, + { + "epoch": 0.8, + "learning_rate": 1.2255268837154034e-06, + "loss": 2.6299, + "step": 3797 + }, + { + "epoch": 0.8, + "learning_rate": 1.2231146929490912e-06, + "loss": 2.6454, + "step": 3798 + }, + { + "epoch": 0.8, + "learning_rate": 1.2207045475967261e-06, + "loss": 2.6576, + "step": 3799 + }, + { + "epoch": 0.8, + "learning_rate": 1.2182964489635502e-06, + "loss": 2.5872, + "step": 3800 + }, + { + "epoch": 0.8, + "learning_rate": 1.2158903983536924e-06, + "loss": 2.6275, + "step": 3801 + }, + { + "epoch": 0.8, + "learning_rate": 1.2134863970701794e-06, + "loss": 2.668, + "step": 3802 + }, + { + "epoch": 0.8, + "learning_rate": 1.211084446414923e-06, + "loss": 2.6542, + "step": 3803 + }, + { + "epoch": 0.8, + "learning_rate": 1.208684547688727e-06, + "loss": 2.6144, + "step": 3804 + }, + { + "epoch": 0.8, + "learning_rate": 1.2062867021912827e-06, + "loss": 2.6314, + "step": 3805 + }, + { + "epoch": 0.8, + "learning_rate": 1.20389091122117e-06, + "loss": 2.7312, + "step": 3806 + }, + { + "epoch": 0.8, + "learning_rate": 1.2014971760758547e-06, + "loss": 2.6973, + "step": 3807 + }, + { + "epoch": 0.8, + "learning_rate": 1.1991054980516924e-06, + "loss": 2.6695, + "step": 3808 + }, + { + "epoch": 0.8, + "learning_rate": 1.1967158784439214e-06, + "loss": 2.5584, + "step": 3809 + }, + { + "epoch": 0.8, + "learning_rate": 1.1943283185466676e-06, + "loss": 2.6109, + "step": 3810 + }, + { + "epoch": 0.8, + "learning_rate": 1.1919428196529391e-06, + "loss": 2.6093, + "step": 3811 + }, + { + "epoch": 0.8, + "learning_rate": 1.1895593830546308e-06, + "loss": 2.6193, + "step": 3812 + }, + { + "epoch": 0.8, + "learning_rate": 1.1871780100425178e-06, + "loss": 2.6368, + "step": 3813 + }, + { + "epoch": 0.8, + "learning_rate": 1.184798701906259e-06, + "loss": 2.664, + "step": 3814 + }, + { + "epoch": 0.8, + "learning_rate": 1.1824214599343958e-06, + "loss": 2.6349, + "step": 3815 + }, + { + "epoch": 0.8, + "learning_rate": 1.1800462854143497e-06, + "loss": 2.6064, + "step": 3816 + }, + { + "epoch": 0.8, + "learning_rate": 1.177673179632422e-06, + "loss": 2.6439, + "step": 3817 + }, + { + "epoch": 0.8, + "learning_rate": 1.175302143873795e-06, + "loss": 2.6545, + "step": 3818 + }, + { + "epoch": 0.8, + "learning_rate": 1.172933179422529e-06, + "loss": 2.6159, + "step": 3819 + }, + { + "epoch": 0.8, + "learning_rate": 1.1705662875615615e-06, + "loss": 2.598, + "step": 3820 + }, + { + "epoch": 0.8, + "learning_rate": 1.1682014695727129e-06, + "loss": 2.6129, + "step": 3821 + }, + { + "epoch": 0.8, + "learning_rate": 1.1658387267366728e-06, + "loss": 2.5767, + "step": 3822 + }, + { + "epoch": 0.8, + "learning_rate": 1.163478060333012e-06, + "loss": 2.65, + "step": 3823 + }, + { + "epoch": 0.8, + "learning_rate": 1.1611194716401752e-06, + "loss": 2.646, + "step": 3824 + }, + { + "epoch": 0.8, + "learning_rate": 1.1587629619354823e-06, + "loss": 2.6018, + "step": 3825 + }, + { + "epoch": 0.8, + "learning_rate": 1.1564085324951252e-06, + "loss": 2.6263, + "step": 3826 + }, + { + "epoch": 0.8, + "learning_rate": 1.154056184594175e-06, + "loss": 2.6915, + "step": 3827 + }, + { + "epoch": 0.8, + "learning_rate": 1.1517059195065711e-06, + "loss": 2.6406, + "step": 3828 + }, + { + "epoch": 0.8, + "learning_rate": 1.1493577385051223e-06, + "loss": 2.5861, + "step": 3829 + }, + { + "epoch": 0.8, + "learning_rate": 1.1470116428615141e-06, + "loss": 2.5322, + "step": 3830 + }, + { + "epoch": 0.8, + "learning_rate": 1.1446676338463003e-06, + "loss": 2.6644, + "step": 3831 + }, + { + "epoch": 0.8, + "learning_rate": 1.1423257127289033e-06, + "loss": 2.6624, + "step": 3832 + }, + { + "epoch": 0.8, + "learning_rate": 1.1399858807776194e-06, + "loss": 2.6812, + "step": 3833 + }, + { + "epoch": 0.8, + "learning_rate": 1.1376481392596083e-06, + "loss": 2.6305, + "step": 3834 + }, + { + "epoch": 0.8, + "learning_rate": 1.1353124894409006e-06, + "loss": 2.5675, + "step": 3835 + }, + { + "epoch": 0.8, + "learning_rate": 1.132978932586395e-06, + "loss": 2.6573, + "step": 3836 + }, + { + "epoch": 0.8, + "learning_rate": 1.1306474699598507e-06, + "loss": 2.5876, + "step": 3837 + }, + { + "epoch": 0.8, + "learning_rate": 1.1283181028238982e-06, + "loss": 2.6291, + "step": 3838 + }, + { + "epoch": 0.8, + "learning_rate": 1.1259908324400343e-06, + "loss": 2.6532, + "step": 3839 + }, + { + "epoch": 0.81, + "learning_rate": 1.123665660068617e-06, + "loss": 2.6069, + "step": 3840 + }, + { + "epoch": 0.81, + "learning_rate": 1.1213425869688681e-06, + "loss": 2.6207, + "step": 3841 + }, + { + "epoch": 0.81, + "learning_rate": 1.1190216143988746e-06, + "loss": 2.6147, + "step": 3842 + }, + { + "epoch": 0.81, + "learning_rate": 1.1167027436155847e-06, + "loss": 2.6284, + "step": 3843 + }, + { + "epoch": 0.81, + "learning_rate": 1.1143859758748081e-06, + "loss": 2.6875, + "step": 3844 + }, + { + "epoch": 0.81, + "learning_rate": 1.112071312431216e-06, + "loss": 2.5999, + "step": 3845 + }, + { + "epoch": 0.81, + "learning_rate": 1.1097587545383399e-06, + "loss": 2.6387, + "step": 3846 + }, + { + "epoch": 0.81, + "learning_rate": 1.1074483034485717e-06, + "loss": 2.6906, + "step": 3847 + }, + { + "epoch": 0.81, + "learning_rate": 1.1051399604131601e-06, + "loss": 2.6475, + "step": 3848 + }, + { + "epoch": 0.81, + "learning_rate": 1.1028337266822158e-06, + "loss": 2.5929, + "step": 3849 + }, + { + "epoch": 0.81, + "learning_rate": 1.1005296035047042e-06, + "loss": 2.6723, + "step": 3850 + }, + { + "epoch": 0.81, + "learning_rate": 1.098227592128448e-06, + "loss": 2.6462, + "step": 3851 + }, + { + "epoch": 0.81, + "learning_rate": 1.0959276938001278e-06, + "loss": 2.6064, + "step": 3852 + }, + { + "epoch": 0.81, + "learning_rate": 1.0936299097652786e-06, + "loss": 2.6364, + "step": 3853 + }, + { + "epoch": 0.81, + "learning_rate": 1.091334241268291e-06, + "loss": 2.6679, + "step": 3854 + }, + { + "epoch": 0.81, + "learning_rate": 1.0890406895524092e-06, + "loss": 2.5685, + "step": 3855 + }, + { + "epoch": 0.81, + "learning_rate": 1.086749255859732e-06, + "loss": 2.6301, + "step": 3856 + }, + { + "epoch": 0.81, + "learning_rate": 1.08445994143121e-06, + "loss": 2.5634, + "step": 3857 + }, + { + "epoch": 0.81, + "learning_rate": 1.0821727475066467e-06, + "loss": 2.5988, + "step": 3858 + }, + { + "epoch": 0.81, + "learning_rate": 1.079887675324698e-06, + "loss": 2.725, + "step": 3859 + }, + { + "epoch": 0.81, + "learning_rate": 1.0776047261228694e-06, + "loss": 2.6667, + "step": 3860 + }, + { + "epoch": 0.81, + "learning_rate": 1.0753239011375172e-06, + "loss": 2.5362, + "step": 3861 + }, + { + "epoch": 0.81, + "learning_rate": 1.073045201603846e-06, + "loss": 2.6578, + "step": 3862 + }, + { + "epoch": 0.81, + "learning_rate": 1.070768628755914e-06, + "loss": 2.6085, + "step": 3863 + }, + { + "epoch": 0.81, + "learning_rate": 1.0684941838266222e-06, + "loss": 2.6684, + "step": 3864 + }, + { + "epoch": 0.81, + "learning_rate": 1.0662218680477237e-06, + "loss": 2.6304, + "step": 3865 + }, + { + "epoch": 0.81, + "learning_rate": 1.0639516826498125e-06, + "loss": 2.6074, + "step": 3866 + }, + { + "epoch": 0.81, + "learning_rate": 1.0616836288623338e-06, + "loss": 2.6187, + "step": 3867 + }, + { + "epoch": 0.81, + "learning_rate": 1.059417707913577e-06, + "loss": 2.677, + "step": 3868 + }, + { + "epoch": 0.81, + "learning_rate": 1.0571539210306785e-06, + "loss": 2.5967, + "step": 3869 + }, + { + "epoch": 0.81, + "learning_rate": 1.0548922694396157e-06, + "loss": 2.6706, + "step": 3870 + }, + { + "epoch": 0.81, + "learning_rate": 1.0526327543652115e-06, + "loss": 2.6138, + "step": 3871 + }, + { + "epoch": 0.81, + "learning_rate": 1.050375377031132e-06, + "loss": 2.6611, + "step": 3872 + }, + { + "epoch": 0.81, + "learning_rate": 1.0481201386598827e-06, + "loss": 2.6739, + "step": 3873 + }, + { + "epoch": 0.81, + "learning_rate": 1.0458670404728122e-06, + "loss": 2.5956, + "step": 3874 + }, + { + "epoch": 0.81, + "learning_rate": 1.0436160836901138e-06, + "loss": 2.6015, + "step": 3875 + }, + { + "epoch": 0.81, + "learning_rate": 1.041367269530817e-06, + "loss": 2.6859, + "step": 3876 + }, + { + "epoch": 0.81, + "learning_rate": 1.0391205992127912e-06, + "loss": 2.6246, + "step": 3877 + }, + { + "epoch": 0.81, + "learning_rate": 1.0368760739527455e-06, + "loss": 2.6503, + "step": 3878 + }, + { + "epoch": 0.81, + "learning_rate": 1.03463369496623e-06, + "loss": 2.6406, + "step": 3879 + }, + { + "epoch": 0.81, + "learning_rate": 1.0323934634676242e-06, + "loss": 2.6317, + "step": 3880 + }, + { + "epoch": 0.81, + "learning_rate": 1.0301553806701547e-06, + "loss": 2.733, + "step": 3881 + }, + { + "epoch": 0.81, + "learning_rate": 1.027919447785879e-06, + "loss": 2.637, + "step": 3882 + }, + { + "epoch": 0.81, + "learning_rate": 1.0256856660256914e-06, + "loss": 2.641, + "step": 3883 + }, + { + "epoch": 0.81, + "learning_rate": 1.02345403659932e-06, + "loss": 2.6681, + "step": 3884 + }, + { + "epoch": 0.81, + "learning_rate": 1.0212245607153293e-06, + "loss": 2.6134, + "step": 3885 + }, + { + "epoch": 0.81, + "learning_rate": 1.0189972395811164e-06, + "loss": 2.5787, + "step": 3886 + }, + { + "epoch": 0.82, + "learning_rate": 1.0167720744029118e-06, + "loss": 2.5789, + "step": 3887 + }, + { + "epoch": 0.82, + "learning_rate": 1.0145490663857783e-06, + "loss": 2.6801, + "step": 3888 + }, + { + "epoch": 0.82, + "learning_rate": 1.0123282167336096e-06, + "loss": 2.6063, + "step": 3889 + }, + { + "epoch": 0.82, + "learning_rate": 1.0101095266491323e-06, + "loss": 2.5955, + "step": 3890 + }, + { + "epoch": 0.82, + "learning_rate": 1.007892997333902e-06, + "loss": 2.6242, + "step": 3891 + }, + { + "epoch": 0.82, + "learning_rate": 1.0056786299883047e-06, + "loss": 2.6298, + "step": 3892 + }, + { + "epoch": 0.82, + "learning_rate": 1.0034664258115561e-06, + "loss": 2.6467, + "step": 3893 + }, + { + "epoch": 0.82, + "learning_rate": 1.0012563860016988e-06, + "loss": 2.6295, + "step": 3894 + }, + { + "epoch": 0.82, + "learning_rate": 9.990485117556053e-07, + "loss": 2.628, + "step": 3895 + }, + { + "epoch": 0.82, + "learning_rate": 9.968428042689738e-07, + "loss": 2.6176, + "step": 3896 + }, + { + "epoch": 0.82, + "learning_rate": 9.94639264736329e-07, + "loss": 2.6282, + "step": 3897 + }, + { + "epoch": 0.82, + "learning_rate": 9.924378943510215e-07, + "loss": 2.6745, + "step": 3898 + }, + { + "epoch": 0.82, + "learning_rate": 9.9023869430523e-07, + "loss": 2.6412, + "step": 3899 + }, + { + "epoch": 0.82, + "learning_rate": 9.880416657899545e-07, + "loss": 2.5925, + "step": 3900 + }, + { + "epoch": 0.82, + "learning_rate": 9.858468099950208e-07, + "loss": 2.6343, + "step": 3901 + }, + { + "epoch": 0.82, + "learning_rate": 9.836541281090757e-07, + "loss": 2.6568, + "step": 3902 + }, + { + "epoch": 0.82, + "learning_rate": 9.814636213195912e-07, + "loss": 2.6436, + "step": 3903 + }, + { + "epoch": 0.82, + "learning_rate": 9.792752908128584e-07, + "loss": 2.61, + "step": 3904 + }, + { + "epoch": 0.82, + "learning_rate": 9.77089137773995e-07, + "loss": 2.6741, + "step": 3905 + }, + { + "epoch": 0.82, + "learning_rate": 9.749051633869356e-07, + "loss": 2.5929, + "step": 3906 + }, + { + "epoch": 0.82, + "learning_rate": 9.727233688344345e-07, + "loss": 2.5744, + "step": 3907 + }, + { + "epoch": 0.82, + "learning_rate": 9.70543755298069e-07, + "loss": 2.6263, + "step": 3908 + }, + { + "epoch": 0.82, + "learning_rate": 9.683663239582291e-07, + "loss": 2.5816, + "step": 3909 + }, + { + "epoch": 0.82, + "learning_rate": 9.661910759941284e-07, + "loss": 2.5931, + "step": 3910 + }, + { + "epoch": 0.82, + "learning_rate": 9.640180125837972e-07, + "loss": 2.5697, + "step": 3911 + }, + { + "epoch": 0.82, + "learning_rate": 9.618471349040824e-07, + "loss": 2.6553, + "step": 3912 + }, + { + "epoch": 0.82, + "learning_rate": 9.596784441306456e-07, + "loss": 2.5801, + "step": 3913 + }, + { + "epoch": 0.82, + "learning_rate": 9.575119414379657e-07, + "loss": 2.6609, + "step": 3914 + }, + { + "epoch": 0.82, + "learning_rate": 9.553476279993378e-07, + "loss": 2.6228, + "step": 3915 + }, + { + "epoch": 0.82, + "learning_rate": 9.531855049868649e-07, + "loss": 2.6138, + "step": 3916 + }, + { + "epoch": 0.82, + "learning_rate": 9.510255735714735e-07, + "loss": 2.6656, + "step": 3917 + }, + { + "epoch": 0.82, + "learning_rate": 9.488678349228963e-07, + "loss": 2.6242, + "step": 3918 + }, + { + "epoch": 0.82, + "learning_rate": 9.467122902096809e-07, + "loss": 2.6524, + "step": 3919 + }, + { + "epoch": 0.82, + "learning_rate": 9.445589405991862e-07, + "loss": 2.5931, + "step": 3920 + }, + { + "epoch": 0.82, + "learning_rate": 9.424077872575821e-07, + "loss": 2.6234, + "step": 3921 + }, + { + "epoch": 0.82, + "learning_rate": 9.402588313498501e-07, + "loss": 2.5588, + "step": 3922 + }, + { + "epoch": 0.82, + "learning_rate": 9.381120740397809e-07, + "loss": 2.6689, + "step": 3923 + }, + { + "epoch": 0.82, + "learning_rate": 9.35967516489974e-07, + "loss": 2.6636, + "step": 3924 + }, + { + "epoch": 0.82, + "learning_rate": 9.338251598618391e-07, + "loss": 2.6655, + "step": 3925 + }, + { + "epoch": 0.82, + "learning_rate": 9.316850053155923e-07, + "loss": 2.6431, + "step": 3926 + }, + { + "epoch": 0.82, + "learning_rate": 9.295470540102586e-07, + "loss": 2.611, + "step": 3927 + }, + { + "epoch": 0.82, + "learning_rate": 9.274113071036683e-07, + "loss": 2.6636, + "step": 3928 + }, + { + "epoch": 0.82, + "learning_rate": 9.252777657524598e-07, + "loss": 2.5957, + "step": 3929 + }, + { + "epoch": 0.82, + "learning_rate": 9.231464311120752e-07, + "loss": 2.574, + "step": 3930 + }, + { + "epoch": 0.82, + "learning_rate": 9.210173043367627e-07, + "loss": 2.6374, + "step": 3931 + }, + { + "epoch": 0.82, + "learning_rate": 9.18890386579574e-07, + "loss": 2.5879, + "step": 3932 + }, + { + "epoch": 0.82, + "learning_rate": 9.167656789923651e-07, + "loss": 2.671, + "step": 3933 + }, + { + "epoch": 0.82, + "learning_rate": 9.146431827257952e-07, + "loss": 2.6607, + "step": 3934 + }, + { + "epoch": 0.83, + "learning_rate": 9.125228989293234e-07, + "loss": 2.6039, + "step": 3935 + }, + { + "epoch": 0.83, + "learning_rate": 9.104048287512168e-07, + "loss": 2.654, + "step": 3936 + }, + { + "epoch": 0.83, + "learning_rate": 9.082889733385386e-07, + "loss": 2.6112, + "step": 3937 + }, + { + "epoch": 0.83, + "learning_rate": 9.061753338371509e-07, + "loss": 2.6034, + "step": 3938 + }, + { + "epoch": 0.83, + "learning_rate": 9.04063911391721e-07, + "loss": 2.6527, + "step": 3939 + }, + { + "epoch": 0.83, + "learning_rate": 9.019547071457108e-07, + "loss": 2.6908, + "step": 3940 + }, + { + "epoch": 0.83, + "learning_rate": 8.998477222413854e-07, + "loss": 2.6462, + "step": 3941 + }, + { + "epoch": 0.83, + "learning_rate": 8.977429578198055e-07, + "loss": 2.6278, + "step": 3942 + }, + { + "epoch": 0.83, + "learning_rate": 8.956404150208292e-07, + "loss": 2.6961, + "step": 3943 + }, + { + "epoch": 0.83, + "learning_rate": 8.935400949831125e-07, + "loss": 2.7005, + "step": 3944 + }, + { + "epoch": 0.83, + "learning_rate": 8.914419988441053e-07, + "loss": 2.5772, + "step": 3945 + }, + { + "epoch": 0.83, + "learning_rate": 8.893461277400534e-07, + "loss": 2.6001, + "step": 3946 + }, + { + "epoch": 0.83, + "learning_rate": 8.87252482806003e-07, + "loss": 2.6528, + "step": 3947 + }, + { + "epoch": 0.83, + "learning_rate": 8.85161065175788e-07, + "loss": 2.6954, + "step": 3948 + }, + { + "epoch": 0.83, + "learning_rate": 8.8307187598204e-07, + "loss": 2.6097, + "step": 3949 + }, + { + "epoch": 0.83, + "learning_rate": 8.809849163561812e-07, + "loss": 2.59, + "step": 3950 + }, + { + "epoch": 0.83, + "learning_rate": 8.789001874284298e-07, + "loss": 2.6391, + "step": 3951 + }, + { + "epoch": 0.83, + "learning_rate": 8.768176903277897e-07, + "loss": 2.6767, + "step": 3952 + }, + { + "epoch": 0.83, + "learning_rate": 8.74737426182064e-07, + "loss": 2.6503, + "step": 3953 + }, + { + "epoch": 0.83, + "learning_rate": 8.726593961178415e-07, + "loss": 2.62, + "step": 3954 + }, + { + "epoch": 0.83, + "learning_rate": 8.705836012605018e-07, + "loss": 2.5973, + "step": 3955 + }, + { + "epoch": 0.83, + "learning_rate": 8.685100427342153e-07, + "loss": 2.6798, + "step": 3956 + }, + { + "epoch": 0.83, + "learning_rate": 8.664387216619402e-07, + "loss": 2.6295, + "step": 3957 + }, + { + "epoch": 0.83, + "learning_rate": 8.643696391654239e-07, + "loss": 2.631, + "step": 3958 + }, + { + "epoch": 0.83, + "learning_rate": 8.623027963651998e-07, + "loss": 2.6389, + "step": 3959 + }, + { + "epoch": 0.83, + "learning_rate": 8.602381943805899e-07, + "loss": 2.5902, + "step": 3960 + }, + { + "epoch": 0.83, + "learning_rate": 8.58175834329702e-07, + "loss": 2.54, + "step": 3961 + }, + { + "epoch": 0.83, + "learning_rate": 8.561157173294305e-07, + "loss": 2.6329, + "step": 3962 + }, + { + "epoch": 0.83, + "learning_rate": 8.540578444954545e-07, + "loss": 2.6346, + "step": 3963 + }, + { + "epoch": 0.83, + "learning_rate": 8.520022169422371e-07, + "loss": 2.6238, + "step": 3964 + }, + { + "epoch": 0.83, + "learning_rate": 8.499488357830266e-07, + "loss": 2.6396, + "step": 3965 + }, + { + "epoch": 0.83, + "learning_rate": 8.47897702129854e-07, + "loss": 2.6752, + "step": 3966 + }, + { + "epoch": 0.83, + "learning_rate": 8.458488170935336e-07, + "loss": 2.6528, + "step": 3967 + }, + { + "epoch": 0.83, + "learning_rate": 8.438021817836617e-07, + "loss": 2.5801, + "step": 3968 + }, + { + "epoch": 0.83, + "learning_rate": 8.417577973086161e-07, + "loss": 2.6034, + "step": 3969 + }, + { + "epoch": 0.83, + "learning_rate": 8.397156647755556e-07, + "loss": 2.6256, + "step": 3970 + }, + { + "epoch": 0.83, + "learning_rate": 8.376757852904194e-07, + "loss": 2.625, + "step": 3971 + }, + { + "epoch": 0.83, + "learning_rate": 8.356381599579277e-07, + "loss": 2.6466, + "step": 3972 + }, + { + "epoch": 0.83, + "learning_rate": 8.336027898815801e-07, + "loss": 2.6684, + "step": 3973 + }, + { + "epoch": 0.83, + "learning_rate": 8.31569676163651e-07, + "loss": 2.5796, + "step": 3974 + }, + { + "epoch": 0.83, + "learning_rate": 8.295388199051969e-07, + "loss": 2.6747, + "step": 3975 + }, + { + "epoch": 0.83, + "learning_rate": 8.275102222060499e-07, + "loss": 2.6367, + "step": 3976 + }, + { + "epoch": 0.83, + "learning_rate": 8.254838841648188e-07, + "loss": 2.6345, + "step": 3977 + }, + { + "epoch": 0.83, + "learning_rate": 8.234598068788918e-07, + "loss": 2.6398, + "step": 3978 + }, + { + "epoch": 0.83, + "learning_rate": 8.214379914444292e-07, + "loss": 2.6466, + "step": 3979 + }, + { + "epoch": 0.83, + "learning_rate": 8.194184389563681e-07, + "loss": 2.597, + "step": 3980 + }, + { + "epoch": 0.83, + "learning_rate": 8.174011505084173e-07, + "loss": 2.6415, + "step": 3981 + }, + { + "epoch": 0.83, + "learning_rate": 8.153861271930619e-07, + "loss": 2.6636, + "step": 3982 + }, + { + "epoch": 0.84, + "learning_rate": 8.133733701015623e-07, + "loss": 2.6473, + "step": 3983 + }, + { + "epoch": 0.84, + "learning_rate": 8.113628803239482e-07, + "loss": 2.6204, + "step": 3984 + }, + { + "epoch": 0.84, + "learning_rate": 8.09354658949022e-07, + "loss": 2.5548, + "step": 3985 + }, + { + "epoch": 0.84, + "learning_rate": 8.073487070643588e-07, + "loss": 2.6912, + "step": 3986 + }, + { + "epoch": 0.84, + "learning_rate": 8.05345025756305e-07, + "loss": 2.6181, + "step": 3987 + }, + { + "epoch": 0.84, + "learning_rate": 8.033436161099722e-07, + "loss": 2.6517, + "step": 3988 + }, + { + "epoch": 0.84, + "learning_rate": 8.013444792092506e-07, + "loss": 2.7062, + "step": 3989 + }, + { + "epoch": 0.84, + "learning_rate": 7.99347616136793e-07, + "loss": 2.5754, + "step": 3990 + }, + { + "epoch": 0.84, + "learning_rate": 7.973530279740227e-07, + "loss": 2.6169, + "step": 3991 + }, + { + "epoch": 0.84, + "learning_rate": 7.953607158011311e-07, + "loss": 2.635, + "step": 3992 + }, + { + "epoch": 0.84, + "learning_rate": 7.933706806970765e-07, + "loss": 2.6609, + "step": 3993 + }, + { + "epoch": 0.84, + "learning_rate": 7.913829237395849e-07, + "loss": 2.5953, + "step": 3994 + }, + { + "epoch": 0.84, + "learning_rate": 7.893974460051474e-07, + "loss": 2.652, + "step": 3995 + }, + { + "epoch": 0.84, + "learning_rate": 7.874142485690217e-07, + "loss": 2.5928, + "step": 3996 + }, + { + "epoch": 0.84, + "learning_rate": 7.854333325052305e-07, + "loss": 2.655, + "step": 3997 + }, + { + "epoch": 0.84, + "learning_rate": 7.834546988865605e-07, + "loss": 2.6222, + "step": 3998 + }, + { + "epoch": 0.84, + "learning_rate": 7.814783487845629e-07, + "loss": 2.641, + "step": 3999 + }, + { + "epoch": 0.84, + "learning_rate": 7.795042832695515e-07, + "loss": 2.6057, + "step": 4000 + }, + { + "epoch": 0.84, + "learning_rate": 7.775325034106024e-07, + "loss": 2.5903, + "step": 4001 + }, + { + "epoch": 0.84, + "learning_rate": 7.755630102755579e-07, + "loss": 2.67, + "step": 4002 + }, + { + "epoch": 0.84, + "learning_rate": 7.735958049310155e-07, + "loss": 2.6546, + "step": 4003 + }, + { + "epoch": 0.84, + "learning_rate": 7.716308884423385e-07, + "loss": 2.6165, + "step": 4004 + }, + { + "epoch": 0.84, + "learning_rate": 7.696682618736489e-07, + "loss": 2.653, + "step": 4005 + }, + { + "epoch": 0.84, + "learning_rate": 7.677079262878284e-07, + "loss": 2.6415, + "step": 4006 + }, + { + "epoch": 0.84, + "learning_rate": 7.657498827465176e-07, + "loss": 2.5952, + "step": 4007 + }, + { + "epoch": 0.84, + "learning_rate": 7.637941323101189e-07, + "loss": 2.5895, + "step": 4008 + }, + { + "epoch": 0.84, + "learning_rate": 7.6184067603779e-07, + "loss": 2.6069, + "step": 4009 + }, + { + "epoch": 0.84, + "learning_rate": 7.598895149874453e-07, + "loss": 2.7238, + "step": 4010 + }, + { + "epoch": 0.84, + "learning_rate": 7.579406502157583e-07, + "loss": 2.6273, + "step": 4011 + }, + { + "epoch": 0.84, + "learning_rate": 7.559940827781581e-07, + "loss": 2.6286, + "step": 4012 + }, + { + "epoch": 0.84, + "learning_rate": 7.540498137288294e-07, + "loss": 2.6897, + "step": 4013 + }, + { + "epoch": 0.84, + "learning_rate": 7.521078441207136e-07, + "loss": 2.6357, + "step": 4014 + }, + { + "epoch": 0.84, + "learning_rate": 7.501681750055056e-07, + "loss": 2.6147, + "step": 4015 + }, + { + "epoch": 0.84, + "learning_rate": 7.482308074336558e-07, + "loss": 2.6742, + "step": 4016 + }, + { + "epoch": 0.84, + "learning_rate": 7.462957424543643e-07, + "loss": 2.6079, + "step": 4017 + }, + { + "epoch": 0.84, + "learning_rate": 7.443629811155878e-07, + "loss": 2.5859, + "step": 4018 + }, + { + "epoch": 0.84, + "learning_rate": 7.42432524464034e-07, + "loss": 2.6123, + "step": 4019 + }, + { + "epoch": 0.84, + "learning_rate": 7.40504373545165e-07, + "loss": 2.6516, + "step": 4020 + }, + { + "epoch": 0.84, + "learning_rate": 7.385785294031917e-07, + "loss": 2.6657, + "step": 4021 + }, + { + "epoch": 0.84, + "learning_rate": 7.366549930810751e-07, + "loss": 2.6364, + "step": 4022 + }, + { + "epoch": 0.84, + "learning_rate": 7.347337656205295e-07, + "loss": 2.6396, + "step": 4023 + }, + { + "epoch": 0.84, + "learning_rate": 7.328148480620134e-07, + "loss": 2.624, + "step": 4024 + }, + { + "epoch": 0.84, + "learning_rate": 7.308982414447407e-07, + "loss": 2.6416, + "step": 4025 + }, + { + "epoch": 0.84, + "learning_rate": 7.289839468066706e-07, + "loss": 2.6924, + "step": 4026 + }, + { + "epoch": 0.84, + "learning_rate": 7.270719651845099e-07, + "loss": 2.6206, + "step": 4027 + }, + { + "epoch": 0.84, + "learning_rate": 7.251622976137129e-07, + "loss": 2.6246, + "step": 4028 + }, + { + "epoch": 0.84, + "learning_rate": 7.23254945128482e-07, + "loss": 2.6538, + "step": 4029 + }, + { + "epoch": 0.85, + "learning_rate": 7.213499087617648e-07, + "loss": 2.5952, + "step": 4030 + }, + { + "epoch": 0.85, + "learning_rate": 7.194471895452548e-07, + "loss": 2.5766, + "step": 4031 + }, + { + "epoch": 0.85, + "learning_rate": 7.175467885093906e-07, + "loss": 2.512, + "step": 4032 + }, + { + "epoch": 0.85, + "learning_rate": 7.156487066833551e-07, + "loss": 2.6337, + "step": 4033 + }, + { + "epoch": 0.85, + "learning_rate": 7.137529450950759e-07, + "loss": 2.6492, + "step": 4034 + }, + { + "epoch": 0.85, + "learning_rate": 7.118595047712229e-07, + "loss": 2.6363, + "step": 4035 + }, + { + "epoch": 0.85, + "learning_rate": 7.099683867372103e-07, + "loss": 2.6554, + "step": 4036 + }, + { + "epoch": 0.85, + "learning_rate": 7.080795920171934e-07, + "loss": 2.6497, + "step": 4037 + }, + { + "epoch": 0.85, + "learning_rate": 7.061931216340706e-07, + "loss": 2.6465, + "step": 4038 + }, + { + "epoch": 0.85, + "learning_rate": 7.043089766094802e-07, + "loss": 2.6586, + "step": 4039 + }, + { + "epoch": 0.85, + "learning_rate": 7.02427157963802e-07, + "loss": 2.6136, + "step": 4040 + }, + { + "epoch": 0.85, + "learning_rate": 7.005476667161559e-07, + "loss": 2.578, + "step": 4041 + }, + { + "epoch": 0.85, + "learning_rate": 6.986705038844005e-07, + "loss": 2.606, + "step": 4042 + }, + { + "epoch": 0.85, + "learning_rate": 6.96795670485133e-07, + "loss": 2.6455, + "step": 4043 + }, + { + "epoch": 0.85, + "learning_rate": 6.949231675336926e-07, + "loss": 2.6025, + "step": 4044 + }, + { + "epoch": 0.85, + "learning_rate": 6.93052996044154e-07, + "loss": 2.6015, + "step": 4045 + }, + { + "epoch": 0.85, + "learning_rate": 6.911851570293271e-07, + "loss": 2.659, + "step": 4046 + }, + { + "epoch": 0.85, + "learning_rate": 6.893196515007622e-07, + "loss": 2.6553, + "step": 4047 + }, + { + "epoch": 0.85, + "learning_rate": 6.874564804687434e-07, + "loss": 2.6566, + "step": 4048 + }, + { + "epoch": 0.85, + "learning_rate": 6.855956449422907e-07, + "loss": 2.6394, + "step": 4049 + }, + { + "epoch": 0.85, + "learning_rate": 6.83737145929163e-07, + "loss": 2.6461, + "step": 4050 + }, + { + "epoch": 0.85, + "learning_rate": 6.818809844358498e-07, + "loss": 2.6335, + "step": 4051 + }, + { + "epoch": 0.85, + "learning_rate": 6.800271614675763e-07, + "loss": 2.6124, + "step": 4052 + }, + { + "epoch": 0.85, + "learning_rate": 6.781756780282988e-07, + "loss": 2.6156, + "step": 4053 + }, + { + "epoch": 0.85, + "learning_rate": 6.763265351207099e-07, + "loss": 2.6126, + "step": 4054 + }, + { + "epoch": 0.85, + "learning_rate": 6.744797337462322e-07, + "loss": 2.6429, + "step": 4055 + }, + { + "epoch": 0.85, + "learning_rate": 6.726352749050236e-07, + "loss": 2.6013, + "step": 4056 + }, + { + "epoch": 0.85, + "learning_rate": 6.707931595959699e-07, + "loss": 2.5986, + "step": 4057 + }, + { + "epoch": 0.85, + "learning_rate": 6.689533888166893e-07, + "loss": 2.5765, + "step": 4058 + }, + { + "epoch": 0.85, + "learning_rate": 6.67115963563531e-07, + "loss": 2.6158, + "step": 4059 + }, + { + "epoch": 0.85, + "learning_rate": 6.652808848315706e-07, + "loss": 2.655, + "step": 4060 + }, + { + "epoch": 0.85, + "learning_rate": 6.634481536146153e-07, + "loss": 2.7138, + "step": 4061 + }, + { + "epoch": 0.85, + "learning_rate": 6.61617770905203e-07, + "loss": 2.6435, + "step": 4062 + }, + { + "epoch": 0.85, + "learning_rate": 6.597897376945966e-07, + "loss": 2.641, + "step": 4063 + }, + { + "epoch": 0.85, + "learning_rate": 6.579640549727884e-07, + "loss": 2.6202, + "step": 4064 + }, + { + "epoch": 0.85, + "learning_rate": 6.561407237284955e-07, + "loss": 2.6029, + "step": 4065 + }, + { + "epoch": 0.85, + "learning_rate": 6.543197449491646e-07, + "loss": 2.6755, + "step": 4066 + }, + { + "epoch": 0.85, + "learning_rate": 6.525011196209657e-07, + "loss": 2.5677, + "step": 4067 + }, + { + "epoch": 0.85, + "learning_rate": 6.506848487287964e-07, + "loss": 2.6295, + "step": 4068 + }, + { + "epoch": 0.85, + "learning_rate": 6.488709332562781e-07, + "loss": 2.6789, + "step": 4069 + }, + { + "epoch": 0.85, + "learning_rate": 6.470593741857562e-07, + "loss": 2.6115, + "step": 4070 + }, + { + "epoch": 0.85, + "learning_rate": 6.452501724983007e-07, + "loss": 2.6812, + "step": 4071 + }, + { + "epoch": 0.85, + "learning_rate": 6.434433291737046e-07, + "loss": 2.6347, + "step": 4072 + }, + { + "epoch": 0.85, + "learning_rate": 6.416388451904848e-07, + "loss": 2.6372, + "step": 4073 + }, + { + "epoch": 0.85, + "learning_rate": 6.398367215258783e-07, + "loss": 2.6388, + "step": 4074 + }, + { + "epoch": 0.85, + "learning_rate": 6.380369591558455e-07, + "loss": 2.5774, + "step": 4075 + }, + { + "epoch": 0.85, + "learning_rate": 6.362395590550685e-07, + "loss": 2.6498, + "step": 4076 + }, + { + "epoch": 0.85, + "learning_rate": 6.344445221969475e-07, + "loss": 2.6771, + "step": 4077 + }, + { + "epoch": 0.86, + "learning_rate": 6.32651849553606e-07, + "loss": 2.6169, + "step": 4078 + }, + { + "epoch": 0.86, + "learning_rate": 6.308615420958847e-07, + "loss": 2.5983, + "step": 4079 + }, + { + "epoch": 0.86, + "learning_rate": 6.29073600793344e-07, + "loss": 2.695, + "step": 4080 + }, + { + "epoch": 0.86, + "learning_rate": 6.272880266142662e-07, + "loss": 2.6056, + "step": 4081 + }, + { + "epoch": 0.86, + "learning_rate": 6.255048205256447e-07, + "loss": 2.6121, + "step": 4082 + }, + { + "epoch": 0.86, + "learning_rate": 6.237239834931975e-07, + "loss": 2.6976, + "step": 4083 + }, + { + "epoch": 0.86, + "learning_rate": 6.219455164813543e-07, + "loss": 2.6274, + "step": 4084 + }, + { + "epoch": 0.86, + "learning_rate": 6.201694204532638e-07, + "loss": 2.6272, + "step": 4085 + }, + { + "epoch": 0.86, + "learning_rate": 6.183956963707926e-07, + "loss": 2.6469, + "step": 4086 + }, + { + "epoch": 0.86, + "learning_rate": 6.166243451945186e-07, + "loss": 2.654, + "step": 4087 + }, + { + "epoch": 0.86, + "learning_rate": 6.148553678837388e-07, + "loss": 2.7021, + "step": 4088 + }, + { + "epoch": 0.86, + "learning_rate": 6.130887653964596e-07, + "loss": 2.6163, + "step": 4089 + }, + { + "epoch": 0.86, + "learning_rate": 6.113245386894051e-07, + "loss": 2.6344, + "step": 4090 + }, + { + "epoch": 0.86, + "learning_rate": 6.095626887180106e-07, + "loss": 2.6167, + "step": 4091 + }, + { + "epoch": 0.86, + "learning_rate": 6.078032164364278e-07, + "loss": 2.6352, + "step": 4092 + }, + { + "epoch": 0.86, + "learning_rate": 6.060461227975179e-07, + "loss": 2.6811, + "step": 4093 + }, + { + "epoch": 0.86, + "learning_rate": 6.042914087528529e-07, + "loss": 2.6651, + "step": 4094 + }, + { + "epoch": 0.86, + "learning_rate": 6.025390752527199e-07, + "loss": 2.6177, + "step": 4095 + }, + { + "epoch": 0.86, + "learning_rate": 6.007891232461122e-07, + "loss": 2.6601, + "step": 4096 + }, + { + "epoch": 0.86, + "learning_rate": 5.990415536807348e-07, + "loss": 2.5525, + "step": 4097 + }, + { + "epoch": 0.86, + "learning_rate": 5.972963675030058e-07, + "loss": 2.5817, + "step": 4098 + }, + { + "epoch": 0.86, + "learning_rate": 5.955535656580497e-07, + "loss": 2.6656, + "step": 4099 + }, + { + "epoch": 0.86, + "learning_rate": 5.938131490896992e-07, + "loss": 2.6402, + "step": 4100 + }, + { + "epoch": 0.86, + "learning_rate": 5.920751187404961e-07, + "loss": 2.6221, + "step": 4101 + }, + { + "epoch": 0.86, + "learning_rate": 5.903394755516916e-07, + "loss": 2.6339, + "step": 4102 + }, + { + "epoch": 0.86, + "learning_rate": 5.886062204632392e-07, + "loss": 2.6247, + "step": 4103 + }, + { + "epoch": 0.86, + "learning_rate": 5.86875354413805e-07, + "loss": 2.6131, + "step": 4104 + }, + { + "epoch": 0.86, + "learning_rate": 5.851468783407582e-07, + "loss": 2.6409, + "step": 4105 + }, + { + "epoch": 0.86, + "learning_rate": 5.834207931801733e-07, + "loss": 2.6316, + "step": 4106 + }, + { + "epoch": 0.86, + "learning_rate": 5.816970998668314e-07, + "loss": 2.6743, + "step": 4107 + }, + { + "epoch": 0.86, + "learning_rate": 5.799757993342175e-07, + "loss": 2.6527, + "step": 4108 + }, + { + "epoch": 0.86, + "learning_rate": 5.78256892514521e-07, + "loss": 2.6913, + "step": 4109 + }, + { + "epoch": 0.86, + "learning_rate": 5.765403803386344e-07, + "loss": 2.6601, + "step": 4110 + }, + { + "epoch": 0.86, + "learning_rate": 5.748262637361546e-07, + "loss": 2.6425, + "step": 4111 + }, + { + "epoch": 0.86, + "learning_rate": 5.731145436353796e-07, + "loss": 2.6654, + "step": 4112 + }, + { + "epoch": 0.86, + "learning_rate": 5.714052209633109e-07, + "loss": 2.6009, + "step": 4113 + }, + { + "epoch": 0.86, + "learning_rate": 5.696982966456511e-07, + "loss": 2.6547, + "step": 4114 + }, + { + "epoch": 0.86, + "learning_rate": 5.679937716068029e-07, + "loss": 2.6957, + "step": 4115 + }, + { + "epoch": 0.86, + "learning_rate": 5.66291646769871e-07, + "loss": 2.5891, + "step": 4116 + }, + { + "epoch": 0.86, + "learning_rate": 5.645919230566622e-07, + "loss": 2.6338, + "step": 4117 + }, + { + "epoch": 0.86, + "learning_rate": 5.628946013876779e-07, + "loss": 2.6388, + "step": 4118 + }, + { + "epoch": 0.86, + "learning_rate": 5.61199682682122e-07, + "loss": 2.72, + "step": 4119 + }, + { + "epoch": 0.86, + "learning_rate": 5.595071678578973e-07, + "loss": 2.5941, + "step": 4120 + }, + { + "epoch": 0.86, + "learning_rate": 5.578170578316017e-07, + "loss": 2.6423, + "step": 4121 + }, + { + "epoch": 0.86, + "learning_rate": 5.561293535185358e-07, + "loss": 2.6318, + "step": 4122 + }, + { + "epoch": 0.86, + "learning_rate": 5.54444055832693e-07, + "loss": 2.6251, + "step": 4123 + }, + { + "epoch": 0.86, + "learning_rate": 5.527611656867666e-07, + "loss": 2.6787, + "step": 4124 + }, + { + "epoch": 0.86, + "learning_rate": 5.510806839921418e-07, + "loss": 2.6363, + "step": 4125 + }, + { + "epoch": 0.87, + "learning_rate": 5.494026116589024e-07, + "loss": 2.5997, + "step": 4126 + }, + { + "epoch": 0.87, + "learning_rate": 5.477269495958276e-07, + "loss": 2.6736, + "step": 4127 + }, + { + "epoch": 0.87, + "learning_rate": 5.460536987103915e-07, + "loss": 2.6665, + "step": 4128 + }, + { + "epoch": 0.87, + "learning_rate": 5.443828599087608e-07, + "loss": 2.5942, + "step": 4129 + }, + { + "epoch": 0.87, + "learning_rate": 5.427144340957968e-07, + "loss": 2.6566, + "step": 4130 + }, + { + "epoch": 0.87, + "learning_rate": 5.41048422175055e-07, + "loss": 2.7151, + "step": 4131 + }, + { + "epoch": 0.87, + "learning_rate": 5.393848250487804e-07, + "loss": 2.6032, + "step": 4132 + }, + { + "epoch": 0.87, + "learning_rate": 5.377236436179123e-07, + "loss": 2.5933, + "step": 4133 + }, + { + "epoch": 0.87, + "learning_rate": 5.360648787820844e-07, + "loss": 2.6079, + "step": 4134 + }, + { + "epoch": 0.87, + "learning_rate": 5.344085314396175e-07, + "loss": 2.6489, + "step": 4135 + }, + { + "epoch": 0.87, + "learning_rate": 5.327546024875252e-07, + "loss": 2.7006, + "step": 4136 + }, + { + "epoch": 0.87, + "learning_rate": 5.311030928215106e-07, + "loss": 2.6008, + "step": 4137 + }, + { + "epoch": 0.87, + "learning_rate": 5.294540033359674e-07, + "loss": 2.6158, + "step": 4138 + }, + { + "epoch": 0.87, + "learning_rate": 5.278073349239776e-07, + "loss": 2.6165, + "step": 4139 + }, + { + "epoch": 0.87, + "learning_rate": 5.261630884773139e-07, + "loss": 2.6332, + "step": 4140 + }, + { + "epoch": 0.87, + "learning_rate": 5.245212648864346e-07, + "loss": 2.6232, + "step": 4141 + }, + { + "epoch": 0.87, + "learning_rate": 5.228818650404883e-07, + "loss": 2.5987, + "step": 4142 + }, + { + "epoch": 0.87, + "learning_rate": 5.212448898273104e-07, + "loss": 2.622, + "step": 4143 + }, + { + "epoch": 0.87, + "learning_rate": 5.196103401334219e-07, + "loss": 2.6361, + "step": 4144 + }, + { + "epoch": 0.87, + "learning_rate": 5.179782168440317e-07, + "loss": 2.5763, + "step": 4145 + }, + { + "epoch": 0.87, + "learning_rate": 5.163485208430347e-07, + "loss": 2.6444, + "step": 4146 + }, + { + "epoch": 0.87, + "learning_rate": 5.147212530130097e-07, + "loss": 2.6876, + "step": 4147 + }, + { + "epoch": 0.87, + "learning_rate": 5.130964142352223e-07, + "loss": 2.6243, + "step": 4148 + }, + { + "epoch": 0.87, + "learning_rate": 5.114740053896222e-07, + "loss": 2.6584, + "step": 4149 + }, + { + "epoch": 0.87, + "learning_rate": 5.098540273548419e-07, + "loss": 2.6254, + "step": 4150 + }, + { + "epoch": 0.87, + "learning_rate": 5.082364810081991e-07, + "loss": 2.6403, + "step": 4151 + }, + { + "epoch": 0.87, + "learning_rate": 5.066213672256931e-07, + "loss": 2.6046, + "step": 4152 + }, + { + "epoch": 0.87, + "learning_rate": 5.050086868820086e-07, + "loss": 2.6123, + "step": 4153 + }, + { + "epoch": 0.87, + "learning_rate": 5.033984408505083e-07, + "loss": 2.6842, + "step": 4154 + }, + { + "epoch": 0.87, + "learning_rate": 5.017906300032394e-07, + "loss": 2.6344, + "step": 4155 + }, + { + "epoch": 0.87, + "learning_rate": 5.001852552109293e-07, + "loss": 2.6421, + "step": 4156 + }, + { + "epoch": 0.87, + "learning_rate": 4.985823173429871e-07, + "loss": 2.6481, + "step": 4157 + }, + { + "epoch": 0.87, + "learning_rate": 4.969818172675001e-07, + "loss": 2.6579, + "step": 4158 + }, + { + "epoch": 0.87, + "learning_rate": 4.953837558512392e-07, + "loss": 2.6473, + "step": 4159 + }, + { + "epoch": 0.87, + "learning_rate": 4.937881339596518e-07, + "loss": 2.6227, + "step": 4160 + }, + { + "epoch": 0.87, + "learning_rate": 4.92194952456862e-07, + "loss": 2.6517, + "step": 4161 + }, + { + "epoch": 0.87, + "learning_rate": 4.906042122056775e-07, + "loss": 2.5776, + "step": 4162 + }, + { + "epoch": 0.87, + "learning_rate": 4.890159140675787e-07, + "loss": 2.6425, + "step": 4163 + }, + { + "epoch": 0.87, + "learning_rate": 4.874300589027286e-07, + "loss": 2.6522, + "step": 4164 + }, + { + "epoch": 0.87, + "learning_rate": 4.858466475699636e-07, + "loss": 2.5935, + "step": 4165 + }, + { + "epoch": 0.87, + "learning_rate": 4.842656809267976e-07, + "loss": 2.6057, + "step": 4166 + }, + { + "epoch": 0.87, + "learning_rate": 4.826871598294214e-07, + "loss": 2.5892, + "step": 4167 + }, + { + "epoch": 0.87, + "learning_rate": 4.811110851326983e-07, + "loss": 2.5699, + "step": 4168 + }, + { + "epoch": 0.87, + "learning_rate": 4.795374576901696e-07, + "loss": 2.6471, + "step": 4169 + }, + { + "epoch": 0.87, + "learning_rate": 4.779662783540523e-07, + "loss": 2.6005, + "step": 4170 + }, + { + "epoch": 0.87, + "learning_rate": 4.763975479752347e-07, + "loss": 2.6426, + "step": 4171 + }, + { + "epoch": 0.87, + "learning_rate": 4.7483126740328013e-07, + "loss": 2.5993, + "step": 4172 + }, + { + "epoch": 0.88, + "learning_rate": 4.732674374864255e-07, + "loss": 2.6552, + "step": 4173 + }, + { + "epoch": 0.88, + "learning_rate": 4.7170605907158e-07, + "loss": 2.5999, + "step": 4174 + }, + { + "epoch": 0.88, + "learning_rate": 4.7014713300432504e-07, + "loss": 2.5661, + "step": 4175 + }, + { + "epoch": 0.88, + "learning_rate": 4.6859066012891476e-07, + "loss": 2.6229, + "step": 4176 + }, + { + "epoch": 0.88, + "learning_rate": 4.6703664128827376e-07, + "loss": 2.5892, + "step": 4177 + }, + { + "epoch": 0.88, + "learning_rate": 4.6548507732399826e-07, + "loss": 2.6622, + "step": 4178 + }, + { + "epoch": 0.88, + "learning_rate": 4.639359690763551e-07, + "loss": 2.5956, + "step": 4179 + }, + { + "epoch": 0.88, + "learning_rate": 4.6238931738428086e-07, + "loss": 2.6411, + "step": 4180 + }, + { + "epoch": 0.88, + "learning_rate": 4.6084512308538165e-07, + "loss": 2.5882, + "step": 4181 + }, + { + "epoch": 0.88, + "learning_rate": 4.59303387015933e-07, + "loss": 2.6522, + "step": 4182 + }, + { + "epoch": 0.88, + "learning_rate": 4.5776411001087873e-07, + "loss": 2.6307, + "step": 4183 + }, + { + "epoch": 0.88, + "learning_rate": 4.562272929038325e-07, + "loss": 2.6486, + "step": 4184 + }, + { + "epoch": 0.88, + "learning_rate": 4.5469293652707304e-07, + "loss": 2.5915, + "step": 4185 + }, + { + "epoch": 0.88, + "learning_rate": 4.5316104171154953e-07, + "loss": 2.5783, + "step": 4186 + }, + { + "epoch": 0.88, + "learning_rate": 4.51631609286875e-07, + "loss": 2.5883, + "step": 4187 + }, + { + "epoch": 0.88, + "learning_rate": 4.501046400813308e-07, + "loss": 2.6496, + "step": 4188 + }, + { + "epoch": 0.88, + "learning_rate": 4.4858013492186605e-07, + "loss": 2.6061, + "step": 4189 + }, + { + "epoch": 0.88, + "learning_rate": 4.4705809463409077e-07, + "loss": 2.7045, + "step": 4190 + }, + { + "epoch": 0.88, + "learning_rate": 4.455385200422835e-07, + "loss": 2.5967, + "step": 4191 + }, + { + "epoch": 0.88, + "learning_rate": 4.44021411969387e-07, + "loss": 2.6508, + "step": 4192 + }, + { + "epoch": 0.88, + "learning_rate": 4.425067712370074e-07, + "loss": 2.6274, + "step": 4193 + }, + { + "epoch": 0.88, + "learning_rate": 4.409945986654146e-07, + "loss": 2.6747, + "step": 4194 + }, + { + "epoch": 0.88, + "learning_rate": 4.394848950735436e-07, + "loss": 2.6898, + "step": 4195 + }, + { + "epoch": 0.88, + "learning_rate": 4.379776612789921e-07, + "loss": 2.63, + "step": 4196 + }, + { + "epoch": 0.88, + "learning_rate": 4.364728980980165e-07, + "loss": 2.6352, + "step": 4197 + }, + { + "epoch": 0.88, + "learning_rate": 4.349706063455389e-07, + "loss": 2.6331, + "step": 4198 + }, + { + "epoch": 0.88, + "learning_rate": 4.334707868351423e-07, + "loss": 2.633, + "step": 4199 + }, + { + "epoch": 0.88, + "learning_rate": 4.3197344037906986e-07, + "loss": 2.5839, + "step": 4200 + }, + { + "epoch": 0.88, + "learning_rate": 4.304785677882273e-07, + "loss": 2.5926, + "step": 4201 + }, + { + "epoch": 0.88, + "learning_rate": 4.2898616987217866e-07, + "loss": 2.6573, + "step": 4202 + }, + { + "epoch": 0.88, + "learning_rate": 4.274962474391492e-07, + "loss": 2.5703, + "step": 4203 + }, + { + "epoch": 0.88, + "learning_rate": 4.2600880129602174e-07, + "loss": 2.5762, + "step": 4204 + }, + { + "epoch": 0.88, + "learning_rate": 4.245238322483386e-07, + "loss": 2.6512, + "step": 4205 + }, + { + "epoch": 0.88, + "learning_rate": 4.2304134110030316e-07, + "loss": 2.6347, + "step": 4206 + }, + { + "epoch": 0.88, + "learning_rate": 4.2156132865477375e-07, + "loss": 2.6004, + "step": 4207 + }, + { + "epoch": 0.88, + "learning_rate": 4.2008379571326753e-07, + "loss": 2.5626, + "step": 4208 + }, + { + "epoch": 0.88, + "learning_rate": 4.1860874307595943e-07, + "loss": 2.6325, + "step": 4209 + }, + { + "epoch": 0.88, + "learning_rate": 4.171361715416794e-07, + "loss": 2.6376, + "step": 4210 + }, + { + "epoch": 0.88, + "learning_rate": 4.156660819079156e-07, + "loss": 2.6685, + "step": 4211 + }, + { + "epoch": 0.88, + "learning_rate": 4.141984749708111e-07, + "loss": 2.5762, + "step": 4212 + }, + { + "epoch": 0.88, + "learning_rate": 4.127333515251647e-07, + "loss": 2.6323, + "step": 4213 + }, + { + "epoch": 0.88, + "learning_rate": 4.1127071236442993e-07, + "loss": 2.7301, + "step": 4214 + }, + { + "epoch": 0.88, + "learning_rate": 4.0981055828071545e-07, + "loss": 2.6706, + "step": 4215 + }, + { + "epoch": 0.88, + "learning_rate": 4.0835289006478376e-07, + "loss": 2.6585, + "step": 4216 + }, + { + "epoch": 0.88, + "learning_rate": 4.06897708506051e-07, + "loss": 2.6732, + "step": 4217 + }, + { + "epoch": 0.88, + "learning_rate": 4.0544501439258753e-07, + "loss": 2.6412, + "step": 4218 + }, + { + "epoch": 0.88, + "learning_rate": 4.039948085111145e-07, + "loss": 2.6538, + "step": 4219 + }, + { + "epoch": 0.88, + "learning_rate": 4.025470916470081e-07, + "loss": 2.6443, + "step": 4220 + }, + { + "epoch": 0.89, + "learning_rate": 4.0110186458429513e-07, + "loss": 2.6444, + "step": 4221 + }, + { + "epoch": 0.89, + "learning_rate": 3.996591281056539e-07, + "loss": 2.6104, + "step": 4222 + }, + { + "epoch": 0.89, + "learning_rate": 3.98218882992415e-07, + "loss": 2.6049, + "step": 4223 + }, + { + "epoch": 0.89, + "learning_rate": 3.967811300245572e-07, + "loss": 2.6758, + "step": 4224 + }, + { + "epoch": 0.89, + "learning_rate": 3.95345869980715e-07, + "loss": 2.6382, + "step": 4225 + }, + { + "epoch": 0.89, + "learning_rate": 3.939131036381666e-07, + "loss": 2.6004, + "step": 4226 + }, + { + "epoch": 0.89, + "learning_rate": 3.924828317728424e-07, + "loss": 2.6548, + "step": 4227 + }, + { + "epoch": 0.89, + "learning_rate": 3.910550551593234e-07, + "loss": 2.7088, + "step": 4228 + }, + { + "epoch": 0.89, + "learning_rate": 3.8962977457083663e-07, + "loss": 2.6334, + "step": 4229 + }, + { + "epoch": 0.89, + "learning_rate": 3.882069907792579e-07, + "loss": 2.5543, + "step": 4230 + }, + { + "epoch": 0.89, + "learning_rate": 3.8678670455511325e-07, + "loss": 2.602, + "step": 4231 + }, + { + "epoch": 0.89, + "learning_rate": 3.8536891666757446e-07, + "loss": 2.7042, + "step": 4232 + }, + { + "epoch": 0.89, + "learning_rate": 3.839536278844591e-07, + "loss": 2.5716, + "step": 4233 + }, + { + "epoch": 0.89, + "learning_rate": 3.8254083897223236e-07, + "loss": 2.6215, + "step": 4234 + }, + { + "epoch": 0.89, + "learning_rate": 3.8113055069600555e-07, + "loss": 2.6221, + "step": 4235 + }, + { + "epoch": 0.89, + "learning_rate": 3.797227638195361e-07, + "loss": 2.525, + "step": 4236 + }, + { + "epoch": 0.89, + "learning_rate": 3.783174791052274e-07, + "loss": 2.6655, + "step": 4237 + }, + { + "epoch": 0.89, + "learning_rate": 3.7691469731412635e-07, + "loss": 2.6838, + "step": 4238 + }, + { + "epoch": 0.89, + "learning_rate": 3.7551441920592567e-07, + "loss": 2.5649, + "step": 4239 + }, + { + "epoch": 0.89, + "learning_rate": 3.741166455389594e-07, + "loss": 2.7191, + "step": 4240 + }, + { + "epoch": 0.89, + "learning_rate": 3.7272137707020875e-07, + "loss": 2.6947, + "step": 4241 + }, + { + "epoch": 0.89, + "learning_rate": 3.713286145552958e-07, + "loss": 2.5306, + "step": 4242 + }, + { + "epoch": 0.89, + "learning_rate": 3.699383587484878e-07, + "loss": 2.6385, + "step": 4243 + }, + { + "epoch": 0.89, + "learning_rate": 3.685506104026931e-07, + "loss": 2.6432, + "step": 4244 + }, + { + "epoch": 0.89, + "learning_rate": 3.671653702694611e-07, + "loss": 2.6092, + "step": 4245 + }, + { + "epoch": 0.89, + "learning_rate": 3.6578263909898435e-07, + "loss": 2.6686, + "step": 4246 + }, + { + "epoch": 0.89, + "learning_rate": 3.644024176400962e-07, + "loss": 2.6614, + "step": 4247 + }, + { + "epoch": 0.89, + "learning_rate": 3.630247066402709e-07, + "loss": 2.5793, + "step": 4248 + }, + { + "epoch": 0.89, + "learning_rate": 3.616495068456233e-07, + "loss": 2.5913, + "step": 4249 + }, + { + "epoch": 0.89, + "learning_rate": 3.602768190009076e-07, + "loss": 2.5641, + "step": 4250 + }, + { + "epoch": 0.89, + "learning_rate": 3.589066438495176e-07, + "loss": 2.6589, + "step": 4251 + }, + { + "epoch": 0.89, + "learning_rate": 3.575389821334879e-07, + "loss": 2.5364, + "step": 4252 + }, + { + "epoch": 0.89, + "learning_rate": 3.561738345934901e-07, + "loss": 2.5857, + "step": 4253 + }, + { + "epoch": 0.89, + "learning_rate": 3.5481120196883544e-07, + "loss": 2.627, + "step": 4254 + }, + { + "epoch": 0.89, + "learning_rate": 3.5345108499747203e-07, + "loss": 2.6065, + "step": 4255 + }, + { + "epoch": 0.89, + "learning_rate": 3.5209348441598626e-07, + "loss": 2.5996, + "step": 4256 + }, + { + "epoch": 0.89, + "learning_rate": 3.5073840095960267e-07, + "loss": 2.5857, + "step": 4257 + }, + { + "epoch": 0.89, + "learning_rate": 3.493858353621804e-07, + "loss": 2.6342, + "step": 4258 + }, + { + "epoch": 0.89, + "learning_rate": 3.4803578835621685e-07, + "loss": 2.6508, + "step": 4259 + }, + { + "epoch": 0.89, + "learning_rate": 3.46688260672845e-07, + "loss": 2.6418, + "step": 4260 + }, + { + "epoch": 0.89, + "learning_rate": 3.453432530418338e-07, + "loss": 2.6056, + "step": 4261 + }, + { + "epoch": 0.89, + "learning_rate": 3.440007661915856e-07, + "loss": 2.623, + "step": 4262 + }, + { + "epoch": 0.89, + "learning_rate": 3.4266080084914054e-07, + "loss": 2.6299, + "step": 4263 + }, + { + "epoch": 0.89, + "learning_rate": 3.4132335774017036e-07, + "loss": 2.6176, + "step": 4264 + }, + { + "epoch": 0.89, + "learning_rate": 3.3998843758898336e-07, + "loss": 2.614, + "step": 4265 + }, + { + "epoch": 0.89, + "learning_rate": 3.3865604111851804e-07, + "loss": 2.7013, + "step": 4266 + }, + { + "epoch": 0.89, + "learning_rate": 3.3732616905035155e-07, + "loss": 2.5835, + "step": 4267 + }, + { + "epoch": 0.89, + "learning_rate": 3.3599882210468947e-07, + "loss": 2.5753, + "step": 4268 + }, + { + "epoch": 0.9, + "learning_rate": 3.346740010003707e-07, + "loss": 2.6324, + "step": 4269 + }, + { + "epoch": 0.9, + "learning_rate": 3.333517064548669e-07, + "loss": 2.6455, + "step": 4270 + }, + { + "epoch": 0.9, + "learning_rate": 3.320319391842813e-07, + "loss": 2.5915, + "step": 4271 + }, + { + "epoch": 0.9, + "learning_rate": 3.307146999033478e-07, + "loss": 2.7165, + "step": 4272 + }, + { + "epoch": 0.9, + "learning_rate": 3.293999893254335e-07, + "loss": 2.7077, + "step": 4273 + }, + { + "epoch": 0.9, + "learning_rate": 3.280878081625333e-07, + "loss": 2.66, + "step": 4274 + }, + { + "epoch": 0.9, + "learning_rate": 3.2677815712527374e-07, + "loss": 2.6003, + "step": 4275 + }, + { + "epoch": 0.9, + "learning_rate": 3.254710369229119e-07, + "loss": 2.6515, + "step": 4276 + }, + { + "epoch": 0.9, + "learning_rate": 3.241664482633311e-07, + "loss": 2.6744, + "step": 4277 + }, + { + "epoch": 0.9, + "learning_rate": 3.228643918530455e-07, + "loss": 2.6465, + "step": 4278 + }, + { + "epoch": 0.9, + "learning_rate": 3.215648683972e-07, + "loss": 2.674, + "step": 4279 + }, + { + "epoch": 0.9, + "learning_rate": 3.202678785995655e-07, + "loss": 2.6175, + "step": 4280 + }, + { + "epoch": 0.9, + "learning_rate": 3.189734231625408e-07, + "loss": 2.5446, + "step": 4281 + }, + { + "epoch": 0.9, + "learning_rate": 3.1768150278715213e-07, + "loss": 2.6171, + "step": 4282 + }, + { + "epoch": 0.9, + "learning_rate": 3.16392118173055e-07, + "loss": 2.6275, + "step": 4283 + }, + { + "epoch": 0.9, + "learning_rate": 3.151052700185264e-07, + "loss": 2.5496, + "step": 4284 + }, + { + "epoch": 0.9, + "learning_rate": 3.1382095902047604e-07, + "loss": 2.5886, + "step": 4285 + }, + { + "epoch": 0.9, + "learning_rate": 3.1253918587443645e-07, + "loss": 2.6466, + "step": 4286 + }, + { + "epoch": 0.9, + "learning_rate": 3.1125995127456507e-07, + "loss": 2.5953, + "step": 4287 + }, + { + "epoch": 0.9, + "learning_rate": 3.099832559136462e-07, + "loss": 2.5687, + "step": 4288 + }, + { + "epoch": 0.9, + "learning_rate": 3.0870910048308833e-07, + "loss": 2.6586, + "step": 4289 + }, + { + "epoch": 0.9, + "learning_rate": 3.07437485672924e-07, + "loss": 2.5965, + "step": 4290 + }, + { + "epoch": 0.9, + "learning_rate": 3.0616841217181005e-07, + "loss": 2.6481, + "step": 4291 + }, + { + "epoch": 0.9, + "learning_rate": 3.049018806670284e-07, + "loss": 2.6561, + "step": 4292 + }, + { + "epoch": 0.9, + "learning_rate": 3.0363789184448154e-07, + "loss": 2.6572, + "step": 4293 + }, + { + "epoch": 0.9, + "learning_rate": 3.0237644638869803e-07, + "loss": 2.6939, + "step": 4294 + }, + { + "epoch": 0.9, + "learning_rate": 3.0111754498282686e-07, + "loss": 2.6553, + "step": 4295 + }, + { + "epoch": 0.9, + "learning_rate": 2.9986118830863997e-07, + "loss": 2.6624, + "step": 4296 + }, + { + "epoch": 0.9, + "learning_rate": 2.986073770465314e-07, + "loss": 2.6675, + "step": 4297 + }, + { + "epoch": 0.9, + "learning_rate": 2.9735611187551696e-07, + "loss": 2.6473, + "step": 4298 + }, + { + "epoch": 0.9, + "learning_rate": 2.961073934732317e-07, + "loss": 2.571, + "step": 4299 + }, + { + "epoch": 0.9, + "learning_rate": 2.948612225159342e-07, + "loss": 2.6835, + "step": 4300 + }, + { + "epoch": 0.9, + "learning_rate": 2.936175996785018e-07, + "loss": 2.6099, + "step": 4301 + }, + { + "epoch": 0.9, + "learning_rate": 2.923765256344319e-07, + "loss": 2.6687, + "step": 4302 + }, + { + "epoch": 0.9, + "learning_rate": 2.9113800105584123e-07, + "loss": 2.6479, + "step": 4303 + }, + { + "epoch": 0.9, + "learning_rate": 2.8990202661346887e-07, + "loss": 2.5937, + "step": 4304 + }, + { + "epoch": 0.9, + "learning_rate": 2.8866860297666775e-07, + "loss": 2.6681, + "step": 4305 + }, + { + "epoch": 0.9, + "learning_rate": 2.8743773081341295e-07, + "loss": 2.641, + "step": 4306 + }, + { + "epoch": 0.9, + "learning_rate": 2.862094107902974e-07, + "loss": 2.6893, + "step": 4307 + }, + { + "epoch": 0.9, + "learning_rate": 2.8498364357253004e-07, + "loss": 2.6257, + "step": 4308 + }, + { + "epoch": 0.9, + "learning_rate": 2.8376042982394046e-07, + "loss": 2.6569, + "step": 4309 + }, + { + "epoch": 0.9, + "learning_rate": 2.8253977020697266e-07, + "loss": 2.5789, + "step": 4310 + }, + { + "epoch": 0.9, + "learning_rate": 2.8132166538268836e-07, + "loss": 2.7151, + "step": 4311 + }, + { + "epoch": 0.9, + "learning_rate": 2.8010611601076656e-07, + "loss": 2.695, + "step": 4312 + }, + { + "epoch": 0.9, + "learning_rate": 2.788931227494995e-07, + "loss": 2.6593, + "step": 4313 + }, + { + "epoch": 0.9, + "learning_rate": 2.7768268625579795e-07, + "loss": 2.6136, + "step": 4314 + }, + { + "epoch": 0.9, + "learning_rate": 2.764748071851875e-07, + "loss": 2.666, + "step": 4315 + }, + { + "epoch": 0.91, + "learning_rate": 2.752694861918087e-07, + "loss": 2.7014, + "step": 4316 + }, + { + "epoch": 0.91, + "learning_rate": 2.740667239284162e-07, + "loss": 2.751, + "step": 4317 + }, + { + "epoch": 0.91, + "learning_rate": 2.7286652104637877e-07, + "loss": 2.6481, + "step": 4318 + }, + { + "epoch": 0.91, + "learning_rate": 2.7166887819568055e-07, + "loss": 2.6008, + "step": 4319 + }, + { + "epoch": 0.91, + "learning_rate": 2.7047379602491597e-07, + "loss": 2.6547, + "step": 4320 + }, + { + "epoch": 0.91, + "learning_rate": 2.6928127518129677e-07, + "loss": 2.6225, + "step": 4321 + }, + { + "epoch": 0.91, + "learning_rate": 2.6809131631064634e-07, + "loss": 2.7006, + "step": 4322 + }, + { + "epoch": 0.91, + "learning_rate": 2.669039200573986e-07, + "loss": 2.6377, + "step": 4323 + }, + { + "epoch": 0.91, + "learning_rate": 2.6571908706460145e-07, + "loss": 2.5645, + "step": 4324 + }, + { + "epoch": 0.91, + "learning_rate": 2.645368179739155e-07, + "loss": 2.6841, + "step": 4325 + }, + { + "epoch": 0.91, + "learning_rate": 2.6335711342560966e-07, + "loss": 2.6644, + "step": 4326 + }, + { + "epoch": 0.91, + "learning_rate": 2.6217997405856735e-07, + "loss": 2.6288, + "step": 4327 + }, + { + "epoch": 0.91, + "learning_rate": 2.6100540051028136e-07, + "loss": 2.7087, + "step": 4328 + }, + { + "epoch": 0.91, + "learning_rate": 2.5983339341685455e-07, + "loss": 2.5667, + "step": 4329 + }, + { + "epoch": 0.91, + "learning_rate": 2.5866395341299977e-07, + "loss": 2.6006, + "step": 4330 + }, + { + "epoch": 0.91, + "learning_rate": 2.5749708113204097e-07, + "loss": 2.6553, + "step": 4331 + }, + { + "epoch": 0.91, + "learning_rate": 2.56332777205911e-07, + "loss": 2.6594, + "step": 4332 + }, + { + "epoch": 0.91, + "learning_rate": 2.551710422651504e-07, + "loss": 2.6795, + "step": 4333 + }, + { + "epoch": 0.91, + "learning_rate": 2.540118769389105e-07, + "loss": 2.6539, + "step": 4334 + }, + { + "epoch": 0.91, + "learning_rate": 2.528552818549501e-07, + "loss": 2.6556, + "step": 4335 + }, + { + "epoch": 0.91, + "learning_rate": 2.5170125763963547e-07, + "loss": 2.6066, + "step": 4336 + }, + { + "epoch": 0.91, + "learning_rate": 2.505498049179411e-07, + "loss": 2.6354, + "step": 4337 + }, + { + "epoch": 0.91, + "learning_rate": 2.494009243134493e-07, + "loss": 2.6548, + "step": 4338 + }, + { + "epoch": 0.91, + "learning_rate": 2.482546164483485e-07, + "loss": 2.5737, + "step": 4339 + }, + { + "epoch": 0.91, + "learning_rate": 2.471108819434359e-07, + "loss": 2.6037, + "step": 4340 + }, + { + "epoch": 0.91, + "learning_rate": 2.4596972141811226e-07, + "loss": 2.6831, + "step": 4341 + }, + { + "epoch": 0.91, + "learning_rate": 2.4483113549038596e-07, + "loss": 2.5919, + "step": 4342 + }, + { + "epoch": 0.91, + "learning_rate": 2.436951247768704e-07, + "loss": 2.6469, + "step": 4343 + }, + { + "epoch": 0.91, + "learning_rate": 2.4256168989278406e-07, + "loss": 2.5973, + "step": 4344 + }, + { + "epoch": 0.91, + "learning_rate": 2.4143083145195365e-07, + "loss": 2.6362, + "step": 4345 + }, + { + "epoch": 0.91, + "learning_rate": 2.40302550066806e-07, + "loss": 2.5876, + "step": 4346 + }, + { + "epoch": 0.91, + "learning_rate": 2.391768463483751e-07, + "loss": 2.5685, + "step": 4347 + }, + { + "epoch": 0.91, + "learning_rate": 2.380537209062983e-07, + "loss": 2.5853, + "step": 4348 + }, + { + "epoch": 0.91, + "learning_rate": 2.3693317434881623e-07, + "loss": 2.6166, + "step": 4349 + }, + { + "epoch": 0.91, + "learning_rate": 2.3581520728277185e-07, + "loss": 2.6356, + "step": 4350 + }, + { + "epoch": 0.91, + "learning_rate": 2.3469982031361526e-07, + "loss": 2.7012, + "step": 4351 + }, + { + "epoch": 0.91, + "learning_rate": 2.3358701404539552e-07, + "loss": 2.6693, + "step": 4352 + }, + { + "epoch": 0.91, + "learning_rate": 2.3247678908076443e-07, + "loss": 2.6656, + "step": 4353 + }, + { + "epoch": 0.91, + "learning_rate": 2.313691460209777e-07, + "loss": 2.708, + "step": 4354 + }, + { + "epoch": 0.91, + "learning_rate": 2.3026408546589162e-07, + "loss": 2.6141, + "step": 4355 + }, + { + "epoch": 0.91, + "learning_rate": 2.2916160801396191e-07, + "loss": 2.6279, + "step": 4356 + }, + { + "epoch": 0.91, + "learning_rate": 2.2806171426224989e-07, + "loss": 2.6906, + "step": 4357 + }, + { + "epoch": 0.91, + "learning_rate": 2.2696440480641401e-07, + "loss": 2.6125, + "step": 4358 + }, + { + "epoch": 0.91, + "learning_rate": 2.2586968024071454e-07, + "loss": 2.626, + "step": 4359 + }, + { + "epoch": 0.91, + "learning_rate": 2.247775411580111e-07, + "loss": 2.6625, + "step": 4360 + }, + { + "epoch": 0.91, + "learning_rate": 2.23687988149765e-07, + "loss": 2.6035, + "step": 4361 + }, + { + "epoch": 0.91, + "learning_rate": 2.2260102180603482e-07, + "loss": 2.5928, + "step": 4362 + }, + { + "epoch": 0.91, + "learning_rate": 2.2151664271547913e-07, + "loss": 2.6542, + "step": 4363 + }, + { + "epoch": 0.92, + "learning_rate": 2.2043485146535537e-07, + "loss": 2.6819, + "step": 4364 + }, + { + "epoch": 0.92, + "learning_rate": 2.1935564864151993e-07, + "loss": 2.6054, + "step": 4365 + }, + { + "epoch": 0.92, + "learning_rate": 2.182790348284264e-07, + "loss": 2.5494, + "step": 4366 + }, + { + "epoch": 0.92, + "learning_rate": 2.172050106091278e-07, + "loss": 2.6281, + "step": 4367 + }, + { + "epoch": 0.92, + "learning_rate": 2.1613357656527333e-07, + "loss": 2.6317, + "step": 4368 + }, + { + "epoch": 0.92, + "learning_rate": 2.150647332771094e-07, + "loss": 2.5881, + "step": 4369 + }, + { + "epoch": 0.92, + "learning_rate": 2.1399848132348078e-07, + "loss": 2.6299, + "step": 4370 + }, + { + "epoch": 0.92, + "learning_rate": 2.1293482128182775e-07, + "loss": 2.6367, + "step": 4371 + }, + { + "epoch": 0.92, + "learning_rate": 2.1187375372818675e-07, + "loss": 2.6738, + "step": 4372 + }, + { + "epoch": 0.92, + "learning_rate": 2.1081527923719035e-07, + "loss": 2.6561, + "step": 4373 + }, + { + "epoch": 0.92, + "learning_rate": 2.0975939838206838e-07, + "loss": 2.6102, + "step": 4374 + }, + { + "epoch": 0.92, + "learning_rate": 2.087061117346434e-07, + "loss": 2.598, + "step": 4375 + }, + { + "epoch": 0.92, + "learning_rate": 2.0765541986533577e-07, + "loss": 2.6221, + "step": 4376 + }, + { + "epoch": 0.92, + "learning_rate": 2.0660732334315814e-07, + "loss": 2.6589, + "step": 4377 + }, + { + "epoch": 0.92, + "learning_rate": 2.0556182273571923e-07, + "loss": 2.6141, + "step": 4378 + }, + { + "epoch": 0.92, + "learning_rate": 2.0451891860922167e-07, + "loss": 2.6542, + "step": 4379 + }, + { + "epoch": 0.92, + "learning_rate": 2.0347861152846093e-07, + "loss": 2.6811, + "step": 4380 + }, + { + "epoch": 0.92, + "learning_rate": 2.0244090205682688e-07, + "loss": 2.6221, + "step": 4381 + }, + { + "epoch": 0.92, + "learning_rate": 2.0140579075630384e-07, + "loss": 2.674, + "step": 4382 + }, + { + "epoch": 0.92, + "learning_rate": 2.0037327818746678e-07, + "loss": 2.5993, + "step": 4383 + }, + { + "epoch": 0.92, + "learning_rate": 1.9934336490948446e-07, + "loss": 2.5525, + "step": 4384 + }, + { + "epoch": 0.92, + "learning_rate": 1.9831605148011745e-07, + "loss": 2.6592, + "step": 4385 + }, + { + "epoch": 0.92, + "learning_rate": 1.9729133845571847e-07, + "loss": 2.6792, + "step": 4386 + }, + { + "epoch": 0.92, + "learning_rate": 1.9626922639123303e-07, + "loss": 2.7434, + "step": 4387 + }, + { + "epoch": 0.92, + "learning_rate": 1.9524971584019726e-07, + "loss": 2.6296, + "step": 4388 + }, + { + "epoch": 0.92, + "learning_rate": 1.9423280735473783e-07, + "loss": 2.6013, + "step": 4389 + }, + { + "epoch": 0.92, + "learning_rate": 1.9321850148557254e-07, + "loss": 2.6493, + "step": 4390 + }, + { + "epoch": 0.92, + "learning_rate": 1.9220679878201086e-07, + "loss": 2.5962, + "step": 4391 + }, + { + "epoch": 0.92, + "learning_rate": 1.9119769979194958e-07, + "loss": 2.6445, + "step": 4392 + }, + { + "epoch": 0.92, + "learning_rate": 1.9019120506187982e-07, + "loss": 2.5922, + "step": 4393 + }, + { + "epoch": 0.92, + "learning_rate": 1.8918731513687893e-07, + "loss": 2.6335, + "step": 4394 + }, + { + "epoch": 0.92, + "learning_rate": 1.8818603056061425e-07, + "loss": 2.5498, + "step": 4395 + }, + { + "epoch": 0.92, + "learning_rate": 1.8718735187534253e-07, + "loss": 2.638, + "step": 4396 + }, + { + "epoch": 0.92, + "learning_rate": 1.8619127962190952e-07, + "loss": 2.5941, + "step": 4397 + }, + { + "epoch": 0.92, + "learning_rate": 1.8519781433974926e-07, + "loss": 2.6245, + "step": 4398 + }, + { + "epoch": 0.92, + "learning_rate": 1.8420695656688413e-07, + "loss": 2.614, + "step": 4399 + }, + { + "epoch": 0.92, + "learning_rate": 1.8321870683992326e-07, + "loss": 2.626, + "step": 4400 + }, + { + "epoch": 0.92, + "learning_rate": 1.8223306569406463e-07, + "loss": 2.6097, + "step": 4401 + }, + { + "epoch": 0.92, + "learning_rate": 1.812500336630929e-07, + "loss": 2.651, + "step": 4402 + }, + { + "epoch": 0.92, + "learning_rate": 1.8026961127938059e-07, + "loss": 2.64, + "step": 4403 + }, + { + "epoch": 0.92, + "learning_rate": 1.7929179907388518e-07, + "loss": 2.6489, + "step": 4404 + }, + { + "epoch": 0.92, + "learning_rate": 1.783165975761525e-07, + "loss": 2.683, + "step": 4405 + }, + { + "epoch": 0.92, + "learning_rate": 1.7734400731431344e-07, + "loss": 2.6452, + "step": 4406 + }, + { + "epoch": 0.92, + "learning_rate": 1.7637402881508502e-07, + "loss": 2.5981, + "step": 4407 + }, + { + "epoch": 0.92, + "learning_rate": 1.7540666260376982e-07, + "loss": 2.6113, + "step": 4408 + }, + { + "epoch": 0.92, + "learning_rate": 1.744419092042554e-07, + "loss": 2.6462, + "step": 4409 + }, + { + "epoch": 0.92, + "learning_rate": 1.7347976913901555e-07, + "loss": 2.6547, + "step": 4410 + }, + { + "epoch": 0.92, + "learning_rate": 1.725202429291062e-07, + "loss": 2.5801, + "step": 4411 + }, + { + "epoch": 0.93, + "learning_rate": 1.7156333109417055e-07, + "loss": 2.597, + "step": 4412 + }, + { + "epoch": 0.93, + "learning_rate": 1.706090341524358e-07, + "loss": 2.648, + "step": 4413 + }, + { + "epoch": 0.93, + "learning_rate": 1.6965735262070958e-07, + "loss": 2.578, + "step": 4414 + }, + { + "epoch": 0.93, + "learning_rate": 1.687082870143869e-07, + "loss": 2.6936, + "step": 4415 + }, + { + "epoch": 0.93, + "learning_rate": 1.6776183784744493e-07, + "loss": 2.5966, + "step": 4416 + }, + { + "epoch": 0.93, + "learning_rate": 1.66818005632442e-07, + "loss": 2.5857, + "step": 4417 + }, + { + "epoch": 0.93, + "learning_rate": 1.6587679088052365e-07, + "loss": 2.685, + "step": 4418 + }, + { + "epoch": 0.93, + "learning_rate": 1.6493819410141277e-07, + "loss": 2.5924, + "step": 4419 + }, + { + "epoch": 0.93, + "learning_rate": 1.640022158034188e-07, + "loss": 2.6713, + "step": 4420 + }, + { + "epoch": 0.93, + "learning_rate": 1.6306885649342906e-07, + "loss": 2.6793, + "step": 4421 + }, + { + "epoch": 0.93, + "learning_rate": 1.6213811667691537e-07, + "loss": 2.5952, + "step": 4422 + }, + { + "epoch": 0.93, + "learning_rate": 1.6120999685793004e-07, + "loss": 2.5855, + "step": 4423 + }, + { + "epoch": 0.93, + "learning_rate": 1.6028449753910768e-07, + "loss": 2.6583, + "step": 4424 + }, + { + "epoch": 0.93, + "learning_rate": 1.5936161922166126e-07, + "loss": 2.6404, + "step": 4425 + }, + { + "epoch": 0.93, + "learning_rate": 1.5844136240538655e-07, + "loss": 2.6467, + "step": 4426 + }, + { + "epoch": 0.93, + "learning_rate": 1.575237275886593e-07, + "loss": 2.6941, + "step": 4427 + }, + { + "epoch": 0.93, + "learning_rate": 1.5660871526843257e-07, + "loss": 2.6364, + "step": 4428 + }, + { + "epoch": 0.93, + "learning_rate": 1.5569632594024388e-07, + "loss": 2.6553, + "step": 4429 + }, + { + "epoch": 0.93, + "learning_rate": 1.5478656009820626e-07, + "loss": 2.616, + "step": 4430 + }, + { + "epoch": 0.93, + "learning_rate": 1.5387941823501396e-07, + "loss": 2.6036, + "step": 4431 + }, + { + "epoch": 0.93, + "learning_rate": 1.5297490084194012e-07, + "loss": 2.6231, + "step": 4432 + }, + { + "epoch": 0.93, + "learning_rate": 1.520730084088351e-07, + "loss": 2.5963, + "step": 4433 + }, + { + "epoch": 0.93, + "learning_rate": 1.5117374142412932e-07, + "loss": 2.6189, + "step": 4434 + }, + { + "epoch": 0.93, + "learning_rate": 1.5027710037483101e-07, + "loss": 2.7117, + "step": 4435 + }, + { + "epoch": 0.93, + "learning_rate": 1.4938308574652505e-07, + "loss": 2.6733, + "step": 4436 + }, + { + "epoch": 0.93, + "learning_rate": 1.4849169802337638e-07, + "loss": 2.596, + "step": 4437 + }, + { + "epoch": 0.93, + "learning_rate": 1.476029376881244e-07, + "loss": 2.6654, + "step": 4438 + }, + { + "epoch": 0.93, + "learning_rate": 1.4671680522208797e-07, + "loss": 2.5887, + "step": 4439 + }, + { + "epoch": 0.93, + "learning_rate": 1.458333011051616e-07, + "loss": 2.5912, + "step": 4440 + }, + { + "epoch": 0.93, + "learning_rate": 1.4495242581581692e-07, + "loss": 2.7178, + "step": 4441 + }, + { + "epoch": 0.93, + "learning_rate": 1.4407417983110127e-07, + "loss": 2.6658, + "step": 4442 + }, + { + "epoch": 0.93, + "learning_rate": 1.4319856362663864e-07, + "loss": 2.662, + "step": 4443 + }, + { + "epoch": 0.93, + "learning_rate": 1.4232557767662914e-07, + "loss": 2.6499, + "step": 4444 + }, + { + "epoch": 0.93, + "learning_rate": 1.4145522245384735e-07, + "loss": 2.719, + "step": 4445 + }, + { + "epoch": 0.93, + "learning_rate": 1.4058749842964404e-07, + "loss": 2.623, + "step": 4446 + }, + { + "epoch": 0.93, + "learning_rate": 1.397224060739444e-07, + "loss": 2.6408, + "step": 4447 + }, + { + "epoch": 0.93, + "learning_rate": 1.388599458552492e-07, + "loss": 2.6011, + "step": 4448 + }, + { + "epoch": 0.93, + "learning_rate": 1.380001182406332e-07, + "loss": 2.6066, + "step": 4449 + }, + { + "epoch": 0.93, + "learning_rate": 1.3714292369574557e-07, + "loss": 2.6414, + "step": 4450 + }, + { + "epoch": 0.93, + "learning_rate": 1.3628836268480883e-07, + "loss": 2.6438, + "step": 4451 + }, + { + "epoch": 0.93, + "learning_rate": 1.3543643567062114e-07, + "loss": 2.6325, + "step": 4452 + }, + { + "epoch": 0.93, + "learning_rate": 1.3458714311455123e-07, + "loss": 2.6523, + "step": 4453 + }, + { + "epoch": 0.93, + "learning_rate": 1.337404854765445e-07, + "loss": 2.6471, + "step": 4454 + }, + { + "epoch": 0.93, + "learning_rate": 1.3289646321511695e-07, + "loss": 2.6463, + "step": 4455 + }, + { + "epoch": 0.93, + "learning_rate": 1.3205507678735962e-07, + "loss": 2.6637, + "step": 4456 + }, + { + "epoch": 0.93, + "learning_rate": 1.3121632664893192e-07, + "loss": 2.7257, + "step": 4457 + }, + { + "epoch": 0.93, + "learning_rate": 1.3038021325407046e-07, + "loss": 2.6975, + "step": 4458 + }, + { + "epoch": 0.93, + "learning_rate": 1.295467370555803e-07, + "loss": 2.6656, + "step": 4459 + }, + { + "epoch": 0.94, + "learning_rate": 1.2871589850484034e-07, + "loss": 2.6434, + "step": 4460 + }, + { + "epoch": 0.94, + "learning_rate": 1.2788769805180067e-07, + "loss": 2.7207, + "step": 4461 + }, + { + "epoch": 0.94, + "learning_rate": 1.2706213614498196e-07, + "loss": 2.6663, + "step": 4462 + }, + { + "epoch": 0.94, + "learning_rate": 1.2623921323147714e-07, + "loss": 2.6223, + "step": 4463 + }, + { + "epoch": 0.94, + "learning_rate": 1.2541892975694803e-07, + "loss": 2.648, + "step": 4464 + }, + { + "epoch": 0.94, + "learning_rate": 1.2460128616562816e-07, + "loss": 2.6346, + "step": 4465 + }, + { + "epoch": 0.94, + "learning_rate": 1.237862829003228e-07, + "loss": 2.6651, + "step": 4466 + }, + { + "epoch": 0.94, + "learning_rate": 1.2297392040240442e-07, + "loss": 2.5904, + "step": 4467 + }, + { + "epoch": 0.94, + "learning_rate": 1.2216419911181832e-07, + "loss": 2.609, + "step": 4468 + }, + { + "epoch": 0.94, + "learning_rate": 1.2135711946707708e-07, + "loss": 2.5639, + "step": 4469 + }, + { + "epoch": 0.94, + "learning_rate": 1.205526819052638e-07, + "loss": 2.5331, + "step": 4470 + }, + { + "epoch": 0.94, + "learning_rate": 1.1975088686203063e-07, + "loss": 2.5849, + "step": 4471 + }, + { + "epoch": 0.94, + "learning_rate": 1.1895173477159849e-07, + "loss": 2.7046, + "step": 4472 + }, + { + "epoch": 0.94, + "learning_rate": 1.1815522606675733e-07, + "loss": 2.6425, + "step": 4473 + }, + { + "epoch": 0.94, + "learning_rate": 1.1736136117886432e-07, + "loss": 2.5985, + "step": 4474 + }, + { + "epoch": 0.94, + "learning_rate": 1.1657014053784666e-07, + "loss": 2.6454, + "step": 4475 + }, + { + "epoch": 0.94, + "learning_rate": 1.1578156457219825e-07, + "loss": 2.6097, + "step": 4476 + }, + { + "epoch": 0.94, + "learning_rate": 1.1499563370898137e-07, + "loss": 2.6449, + "step": 4477 + }, + { + "epoch": 0.94, + "learning_rate": 1.14212348373825e-07, + "loss": 2.6691, + "step": 4478 + }, + { + "epoch": 0.94, + "learning_rate": 1.1343170899092704e-07, + "loss": 2.6163, + "step": 4479 + }, + { + "epoch": 0.94, + "learning_rate": 1.1265371598304986e-07, + "loss": 2.6774, + "step": 4480 + }, + { + "epoch": 0.94, + "learning_rate": 1.1187836977152533e-07, + "loss": 2.6257, + "step": 4481 + }, + { + "epoch": 0.94, + "learning_rate": 1.1110567077625034e-07, + "loss": 2.58, + "step": 4482 + }, + { + "epoch": 0.94, + "learning_rate": 1.1033561941568848e-07, + "loss": 2.6011, + "step": 4483 + }, + { + "epoch": 0.94, + "learning_rate": 1.0956821610686952e-07, + "loss": 2.6758, + "step": 4484 + }, + { + "epoch": 0.94, + "learning_rate": 1.0880346126538988e-07, + "loss": 2.6206, + "step": 4485 + }, + { + "epoch": 0.94, + "learning_rate": 1.0804135530541049e-07, + "loss": 2.57, + "step": 4486 + }, + { + "epoch": 0.94, + "learning_rate": 1.0728189863965788e-07, + "loss": 2.6201, + "step": 4487 + }, + { + "epoch": 0.94, + "learning_rate": 1.0652509167942471e-07, + "loss": 2.5598, + "step": 4488 + }, + { + "epoch": 0.94, + "learning_rate": 1.0577093483456757e-07, + "loss": 2.6546, + "step": 4489 + }, + { + "epoch": 0.94, + "learning_rate": 1.0501942851350921e-07, + "loss": 2.6352, + "step": 4490 + }, + { + "epoch": 0.94, + "learning_rate": 1.0427057312323574e-07, + "loss": 2.6319, + "step": 4491 + }, + { + "epoch": 0.94, + "learning_rate": 1.0352436906929886e-07, + "loss": 2.6484, + "step": 4492 + }, + { + "epoch": 0.94, + "learning_rate": 1.0278081675581253e-07, + "loss": 2.6891, + "step": 4493 + }, + { + "epoch": 0.94, + "learning_rate": 1.0203991658545576e-07, + "loss": 2.6101, + "step": 4494 + }, + { + "epoch": 0.94, + "learning_rate": 1.0130166895947091e-07, + "loss": 2.6167, + "step": 4495 + }, + { + "epoch": 0.94, + "learning_rate": 1.005660742776654e-07, + "loss": 2.7033, + "step": 4496 + }, + { + "epoch": 0.94, + "learning_rate": 9.983313293840835e-08, + "loss": 2.6238, + "step": 4497 + }, + { + "epoch": 0.94, + "learning_rate": 9.910284533863168e-08, + "loss": 2.6294, + "step": 4498 + }, + { + "epoch": 0.94, + "learning_rate": 9.837521187383126e-08, + "loss": 2.6168, + "step": 4499 + }, + { + "epoch": 0.94, + "learning_rate": 9.765023293806464e-08, + "loss": 2.6927, + "step": 4500 + }, + { + "epoch": 0.94, + "learning_rate": 9.692790892395166e-08, + "loss": 2.665, + "step": 4501 + }, + { + "epoch": 0.94, + "learning_rate": 9.620824022267549e-08, + "loss": 2.7023, + "step": 4502 + }, + { + "epoch": 0.94, + "learning_rate": 9.549122722398108e-08, + "loss": 2.7193, + "step": 4503 + }, + { + "epoch": 0.94, + "learning_rate": 9.477687031617389e-08, + "loss": 2.6195, + "step": 4504 + }, + { + "epoch": 0.94, + "learning_rate": 9.40651698861228e-08, + "loss": 2.6084, + "step": 4505 + }, + { + "epoch": 0.94, + "learning_rate": 9.335612631925617e-08, + "loss": 2.6248, + "step": 4506 + }, + { + "epoch": 0.95, + "learning_rate": 9.264973999956406e-08, + "loss": 2.6837, + "step": 4507 + }, + { + "epoch": 0.95, + "learning_rate": 9.19460113095988e-08, + "loss": 2.6557, + "step": 4508 + }, + { + "epoch": 0.95, + "learning_rate": 9.124494063047162e-08, + "loss": 2.6184, + "step": 4509 + }, + { + "epoch": 0.95, + "learning_rate": 9.054652834185551e-08, + "loss": 2.624, + "step": 4510 + }, + { + "epoch": 0.95, + "learning_rate": 8.985077482198346e-08, + "loss": 2.5901, + "step": 4511 + }, + { + "epoch": 0.95, + "learning_rate": 8.91576804476485e-08, + "loss": 2.6033, + "step": 4512 + }, + { + "epoch": 0.95, + "learning_rate": 8.846724559420316e-08, + "loss": 2.6436, + "step": 4513 + }, + { + "epoch": 0.95, + "learning_rate": 8.777947063556002e-08, + "loss": 2.6347, + "step": 4514 + }, + { + "epoch": 0.95, + "learning_rate": 8.709435594419169e-08, + "loss": 2.6773, + "step": 4515 + }, + { + "epoch": 0.95, + "learning_rate": 8.641190189112913e-08, + "loss": 2.5966, + "step": 4516 + }, + { + "epoch": 0.95, + "learning_rate": 8.57321088459634e-08, + "loss": 2.5955, + "step": 4517 + }, + { + "epoch": 0.95, + "learning_rate": 8.50549771768433e-08, + "loss": 2.6625, + "step": 4518 + }, + { + "epoch": 0.95, + "learning_rate": 8.438050725047775e-08, + "loss": 2.6149, + "step": 4519 + }, + { + "epoch": 0.95, + "learning_rate": 8.370869943213178e-08, + "loss": 2.6132, + "step": 4520 + }, + { + "epoch": 0.95, + "learning_rate": 8.303955408563269e-08, + "loss": 2.6273, + "step": 4521 + }, + { + "epoch": 0.95, + "learning_rate": 8.237307157336172e-08, + "loss": 2.6973, + "step": 4522 + }, + { + "epoch": 0.95, + "learning_rate": 8.17092522562607e-08, + "loss": 2.6362, + "step": 4523 + }, + { + "epoch": 0.95, + "learning_rate": 8.104809649382761e-08, + "loss": 2.6105, + "step": 4524 + }, + { + "epoch": 0.95, + "learning_rate": 8.038960464411938e-08, + "loss": 2.6879, + "step": 4525 + }, + { + "epoch": 0.95, + "learning_rate": 7.973377706374852e-08, + "loss": 2.6388, + "step": 4526 + }, + { + "epoch": 0.95, + "learning_rate": 7.908061410788703e-08, + "loss": 2.6332, + "step": 4527 + }, + { + "epoch": 0.95, + "learning_rate": 7.843011613026197e-08, + "loss": 2.6589, + "step": 4528 + }, + { + "epoch": 0.95, + "learning_rate": 7.778228348315763e-08, + "loss": 2.6748, + "step": 4529 + }, + { + "epoch": 0.95, + "learning_rate": 7.713711651741507e-08, + "loss": 2.6739, + "step": 4530 + }, + { + "epoch": 0.95, + "learning_rate": 7.649461558243087e-08, + "loss": 2.6123, + "step": 4531 + }, + { + "epoch": 0.95, + "learning_rate": 7.585478102615951e-08, + "loss": 2.6158, + "step": 4532 + }, + { + "epoch": 0.95, + "learning_rate": 7.52176131951099e-08, + "loss": 2.6406, + "step": 4533 + }, + { + "epoch": 0.95, + "learning_rate": 7.458311243434712e-08, + "loss": 2.6617, + "step": 4534 + }, + { + "epoch": 0.95, + "learning_rate": 7.395127908749356e-08, + "loss": 2.6775, + "step": 4535 + }, + { + "epoch": 0.95, + "learning_rate": 7.332211349672324e-08, + "loss": 2.621, + "step": 4536 + }, + { + "epoch": 0.95, + "learning_rate": 7.26956160027692e-08, + "loss": 2.6392, + "step": 4537 + }, + { + "epoch": 0.95, + "learning_rate": 7.207178694491778e-08, + "loss": 2.6266, + "step": 4538 + }, + { + "epoch": 0.95, + "learning_rate": 7.145062666101043e-08, + "loss": 2.6362, + "step": 4539 + }, + { + "epoch": 0.95, + "learning_rate": 7.083213548744305e-08, + "loss": 2.6191, + "step": 4540 + }, + { + "epoch": 0.95, + "learning_rate": 7.021631375916716e-08, + "loss": 2.6423, + "step": 4541 + }, + { + "epoch": 0.95, + "learning_rate": 6.960316180968818e-08, + "loss": 2.6158, + "step": 4542 + }, + { + "epoch": 0.95, + "learning_rate": 6.89926799710633e-08, + "loss": 2.6585, + "step": 4543 + }, + { + "epoch": 0.95, + "learning_rate": 6.838486857390692e-08, + "loss": 2.6482, + "step": 4544 + }, + { + "epoch": 0.95, + "learning_rate": 6.777972794738629e-08, + "loss": 2.667, + "step": 4545 + }, + { + "epoch": 0.95, + "learning_rate": 6.717725841922207e-08, + "loss": 2.6323, + "step": 4546 + }, + { + "epoch": 0.95, + "learning_rate": 6.657746031568769e-08, + "loss": 2.6327, + "step": 4547 + }, + { + "epoch": 0.95, + "learning_rate": 6.598033396161051e-08, + "loss": 2.5644, + "step": 4548 + }, + { + "epoch": 0.95, + "learning_rate": 6.538587968037135e-08, + "loss": 2.6626, + "step": 4549 + }, + { + "epoch": 0.95, + "learning_rate": 6.479409779390267e-08, + "loss": 2.5762, + "step": 4550 + }, + { + "epoch": 0.95, + "learning_rate": 6.420498862269198e-08, + "loss": 2.5981, + "step": 4551 + }, + { + "epoch": 0.95, + "learning_rate": 6.36185524857763e-08, + "loss": 2.6854, + "step": 4552 + }, + { + "epoch": 0.95, + "learning_rate": 6.303478970074716e-08, + "loss": 2.6421, + "step": 4553 + }, + { + "epoch": 0.95, + "learning_rate": 6.245370058374777e-08, + "loss": 2.6377, + "step": 4554 + }, + { + "epoch": 0.96, + "learning_rate": 6.187528544947363e-08, + "loss": 2.6192, + "step": 4555 + }, + { + "epoch": 0.96, + "learning_rate": 6.129954461117083e-08, + "loss": 2.5816, + "step": 4556 + }, + { + "epoch": 0.96, + "learning_rate": 6.072647838063994e-08, + "loss": 2.592, + "step": 4557 + }, + { + "epoch": 0.96, + "learning_rate": 6.015608706822996e-08, + "loss": 2.5985, + "step": 4558 + }, + { + "epoch": 0.96, + "learning_rate": 5.95883709828432e-08, + "loss": 2.6681, + "step": 4559 + }, + { + "epoch": 0.96, + "learning_rate": 5.902333043193154e-08, + "loss": 2.6534, + "step": 4560 + }, + { + "epoch": 0.96, + "learning_rate": 5.846096572150017e-08, + "loss": 2.6276, + "step": 4561 + }, + { + "epoch": 0.96, + "learning_rate": 5.790127715610328e-08, + "loss": 2.6514, + "step": 4562 + }, + { + "epoch": 0.96, + "learning_rate": 5.7344265038846713e-08, + "loss": 2.6344, + "step": 4563 + }, + { + "epoch": 0.96, + "learning_rate": 5.678992967138697e-08, + "loss": 2.6417, + "step": 4564 + }, + { + "epoch": 0.96, + "learning_rate": 5.6238271353929455e-08, + "loss": 2.6534, + "step": 4565 + }, + { + "epoch": 0.96, + "learning_rate": 5.5689290385230746e-08, + "loss": 2.6004, + "step": 4566 + }, + { + "epoch": 0.96, + "learning_rate": 5.514298706259857e-08, + "loss": 2.6371, + "step": 4567 + }, + { + "epoch": 0.96, + "learning_rate": 5.459936168188906e-08, + "loss": 2.5724, + "step": 4568 + }, + { + "epoch": 0.96, + "learning_rate": 5.4058414537508374e-08, + "loss": 2.5331, + "step": 4569 + }, + { + "epoch": 0.96, + "learning_rate": 5.352014592241217e-08, + "loss": 2.6157, + "step": 4570 + }, + { + "epoch": 0.96, + "learning_rate": 5.2984556128107266e-08, + "loss": 2.6468, + "step": 4571 + }, + { + "epoch": 0.96, + "learning_rate": 5.2451645444646096e-08, + "loss": 2.6154, + "step": 4572 + }, + { + "epoch": 0.96, + "learning_rate": 5.1921414160633345e-08, + "loss": 2.5822, + "step": 4573 + }, + { + "epoch": 0.96, + "learning_rate": 5.13938625632221e-08, + "loss": 2.6648, + "step": 4574 + }, + { + "epoch": 0.96, + "learning_rate": 5.086899093811326e-08, + "loss": 2.6079, + "step": 4575 + }, + { + "epoch": 0.96, + "learning_rate": 5.0346799569556124e-08, + "loss": 2.658, + "step": 4576 + }, + { + "epoch": 0.96, + "learning_rate": 4.982728874035059e-08, + "loss": 2.6337, + "step": 4577 + }, + { + "epoch": 0.96, + "learning_rate": 4.931045873184326e-08, + "loss": 2.6358, + "step": 4578 + }, + { + "epoch": 0.96, + "learning_rate": 4.879630982392747e-08, + "loss": 2.6372, + "step": 4579 + }, + { + "epoch": 0.96, + "learning_rate": 4.8284842295048265e-08, + "loss": 2.6413, + "step": 4580 + }, + { + "epoch": 0.96, + "learning_rate": 4.777605642219574e-08, + "loss": 2.6551, + "step": 4581 + }, + { + "epoch": 0.96, + "learning_rate": 4.726995248090838e-08, + "loss": 2.7029, + "step": 4582 + }, + { + "epoch": 0.96, + "learning_rate": 4.676653074527249e-08, + "loss": 2.5525, + "step": 4583 + }, + { + "epoch": 0.96, + "learning_rate": 4.626579148792165e-08, + "loss": 2.617, + "step": 4584 + }, + { + "epoch": 0.96, + "learning_rate": 4.5767734980036725e-08, + "loss": 2.6033, + "step": 4585 + }, + { + "epoch": 0.96, + "learning_rate": 4.5272361491345286e-08, + "loss": 2.6493, + "step": 4586 + }, + { + "epoch": 0.96, + "learning_rate": 4.477967129012328e-08, + "loss": 2.5319, + "step": 4587 + }, + { + "epoch": 0.96, + "learning_rate": 4.428966464319173e-08, + "loss": 2.6687, + "step": 4588 + }, + { + "epoch": 0.96, + "learning_rate": 4.380234181592002e-08, + "loss": 2.6826, + "step": 4589 + }, + { + "epoch": 0.96, + "learning_rate": 4.331770307222205e-08, + "loss": 2.6237, + "step": 4590 + }, + { + "epoch": 0.96, + "learning_rate": 4.2835748674560085e-08, + "loss": 2.6171, + "step": 4591 + }, + { + "epoch": 0.96, + "learning_rate": 4.23564788839409e-08, + "loss": 2.6519, + "step": 4592 + }, + { + "epoch": 0.96, + "learning_rate": 4.18798939599202e-08, + "loss": 2.6503, + "step": 4593 + }, + { + "epoch": 0.96, + "learning_rate": 4.140599416059598e-08, + "loss": 2.6455, + "step": 4594 + }, + { + "epoch": 0.96, + "learning_rate": 4.0934779742615174e-08, + "loss": 2.5993, + "step": 4595 + }, + { + "epoch": 0.96, + "learning_rate": 4.0466250961168076e-08, + "loss": 2.5453, + "step": 4596 + }, + { + "epoch": 0.96, + "learning_rate": 4.000040806999228e-08, + "loss": 2.6756, + "step": 4597 + }, + { + "epoch": 0.96, + "learning_rate": 3.953725132136932e-08, + "loss": 2.643, + "step": 4598 + }, + { + "epoch": 0.96, + "learning_rate": 3.907678096612855e-08, + "loss": 2.6852, + "step": 4599 + }, + { + "epoch": 0.96, + "learning_rate": 3.86189972536416e-08, + "loss": 2.6411, + "step": 4600 + }, + { + "epoch": 0.96, + "learning_rate": 3.816390043182572e-08, + "loss": 2.5989, + "step": 4601 + }, + { + "epoch": 0.96, + "learning_rate": 3.771149074714486e-08, + "loss": 2.6712, + "step": 4602 + }, + { + "epoch": 0.97, + "learning_rate": 3.726176844460638e-08, + "loss": 2.647, + "step": 4603 + }, + { + "epoch": 0.97, + "learning_rate": 3.681473376776101e-08, + "loss": 2.6601, + "step": 4604 + }, + { + "epoch": 0.97, + "learning_rate": 3.637038695870676e-08, + "loss": 2.5877, + "step": 4605 + }, + { + "epoch": 0.97, + "learning_rate": 3.5928728258083914e-08, + "loss": 2.6459, + "step": 4606 + }, + { + "epoch": 0.97, + "learning_rate": 3.548975790507836e-08, + "loss": 2.6619, + "step": 4607 + }, + { + "epoch": 0.97, + "learning_rate": 3.505347613741772e-08, + "loss": 2.587, + "step": 4608 + }, + { + "epoch": 0.97, + "learning_rate": 3.461988319137688e-08, + "loss": 2.6684, + "step": 4609 + }, + { + "epoch": 0.97, + "learning_rate": 3.41889793017719e-08, + "loss": 2.6384, + "step": 4610 + }, + { + "epoch": 0.97, + "learning_rate": 3.3760764701963875e-08, + "loss": 2.663, + "step": 4611 + }, + { + "epoch": 0.97, + "learning_rate": 3.3335239623857316e-08, + "loss": 2.613, + "step": 4612 + }, + { + "epoch": 0.97, + "learning_rate": 3.291240429789955e-08, + "loss": 2.6064, + "step": 4613 + }, + { + "epoch": 0.97, + "learning_rate": 3.249225895308294e-08, + "loss": 2.6587, + "step": 4614 + }, + { + "epoch": 0.97, + "learning_rate": 3.2074803816939946e-08, + "loss": 2.6751, + "step": 4615 + }, + { + "epoch": 0.97, + "learning_rate": 3.166003911554916e-08, + "loss": 2.6384, + "step": 4616 + }, + { + "epoch": 0.97, + "learning_rate": 3.124796507353089e-08, + "loss": 2.6403, + "step": 4617 + }, + { + "epoch": 0.97, + "learning_rate": 3.083858191404831e-08, + "loss": 2.5682, + "step": 4618 + }, + { + "epoch": 0.97, + "learning_rate": 3.0431889858807405e-08, + "loss": 2.6993, + "step": 4619 + }, + { + "epoch": 0.97, + "learning_rate": 3.002788912805643e-08, + "loss": 2.6292, + "step": 4620 + }, + { + "epoch": 0.97, + "learning_rate": 2.962657994058704e-08, + "loss": 2.6145, + "step": 4621 + }, + { + "epoch": 0.97, + "learning_rate": 2.9227962513732057e-08, + "loss": 2.64, + "step": 4622 + }, + { + "epoch": 0.97, + "learning_rate": 2.8832037063367678e-08, + "loss": 2.6305, + "step": 4623 + }, + { + "epoch": 0.97, + "learning_rate": 2.8438803803911285e-08, + "loss": 2.6105, + "step": 4624 + }, + { + "epoch": 0.97, + "learning_rate": 2.804826294832308e-08, + "loss": 2.6408, + "step": 4625 + }, + { + "epoch": 0.97, + "learning_rate": 2.7660414708104433e-08, + "loss": 2.6233, + "step": 4626 + }, + { + "epoch": 0.97, + "learning_rate": 2.7275259293298996e-08, + "loss": 2.7589, + "step": 4627 + }, + { + "epoch": 0.97, + "learning_rate": 2.6892796912492136e-08, + "loss": 2.62, + "step": 4628 + }, + { + "epoch": 0.97, + "learning_rate": 2.6513027772810394e-08, + "loss": 2.6772, + "step": 4629 + }, + { + "epoch": 0.97, + "learning_rate": 2.6135952079921477e-08, + "loss": 2.6685, + "step": 4630 + }, + { + "epoch": 0.97, + "learning_rate": 2.5761570038035367e-08, + "loss": 2.5535, + "step": 4631 + }, + { + "epoch": 0.97, + "learning_rate": 2.5389881849903207e-08, + "loss": 2.6279, + "step": 4632 + }, + { + "epoch": 0.97, + "learning_rate": 2.502088771681621e-08, + "loss": 2.6789, + "step": 4633 + }, + { + "epoch": 0.97, + "learning_rate": 2.4654587838606748e-08, + "loss": 2.6342, + "step": 4634 + }, + { + "epoch": 0.97, + "learning_rate": 2.4290982413650022e-08, + "loss": 2.6244, + "step": 4635 + }, + { + "epoch": 0.97, + "learning_rate": 2.393007163885963e-08, + "loss": 2.6569, + "step": 4636 + }, + { + "epoch": 0.97, + "learning_rate": 2.3571855709690894e-08, + "loss": 2.6657, + "step": 4637 + }, + { + "epoch": 0.97, + "learning_rate": 2.3216334820139186e-08, + "loss": 2.6813, + "step": 4638 + }, + { + "epoch": 0.97, + "learning_rate": 2.2863509162741603e-08, + "loss": 2.621, + "step": 4639 + }, + { + "epoch": 0.97, + "learning_rate": 2.251337892857419e-08, + "loss": 2.6034, + "step": 4640 + }, + { + "epoch": 0.97, + "learning_rate": 2.216594430725416e-08, + "loss": 2.6856, + "step": 4641 + }, + { + "epoch": 0.97, + "learning_rate": 2.1821205486938224e-08, + "loss": 2.6069, + "step": 4642 + }, + { + "epoch": 0.97, + "learning_rate": 2.147916265432426e-08, + "loss": 2.5521, + "step": 4643 + }, + { + "epoch": 0.97, + "learning_rate": 2.113981599464854e-08, + "loss": 2.6482, + "step": 4644 + }, + { + "epoch": 0.97, + "learning_rate": 2.0803165691687943e-08, + "loss": 2.6678, + "step": 4645 + }, + { + "epoch": 0.97, + "learning_rate": 2.0469211927759413e-08, + "loss": 2.5911, + "step": 4646 + }, + { + "epoch": 0.97, + "learning_rate": 2.013795488371939e-08, + "loss": 2.6789, + "step": 4647 + }, + { + "epoch": 0.97, + "learning_rate": 1.980939473896437e-08, + "loss": 2.6434, + "step": 4648 + }, + { + "epoch": 0.97, + "learning_rate": 1.948353167142869e-08, + "loss": 2.603, + "step": 4649 + }, + { + "epoch": 0.98, + "learning_rate": 1.9160365857587293e-08, + "loss": 2.6083, + "step": 4650 + }, + { + "epoch": 0.98, + "learning_rate": 1.8839897472454628e-08, + "loss": 2.707, + "step": 4651 + }, + { + "epoch": 0.98, + "learning_rate": 1.852212668958353e-08, + "loss": 2.5778, + "step": 4652 + }, + { + "epoch": 0.98, + "learning_rate": 1.82070536810669e-08, + "loss": 2.6589, + "step": 4653 + }, + { + "epoch": 0.98, + "learning_rate": 1.789467861753491e-08, + "loss": 2.6316, + "step": 4654 + }, + { + "epoch": 0.98, + "learning_rate": 1.7585001668158907e-08, + "loss": 2.6146, + "step": 4655 + }, + { + "epoch": 0.98, + "learning_rate": 1.727802300064696e-08, + "loss": 2.6213, + "step": 4656 + }, + { + "epoch": 0.98, + "learning_rate": 1.6973742781247195e-08, + "loss": 2.6035, + "step": 4657 + }, + { + "epoch": 0.98, + "learning_rate": 1.667216117474557e-08, + "loss": 2.629, + "step": 4658 + }, + { + "epoch": 0.98, + "learning_rate": 1.6373278344467557e-08, + "loss": 2.7634, + "step": 4659 + }, + { + "epoch": 0.98, + "learning_rate": 1.6077094452275898e-08, + "loss": 2.5656, + "step": 4660 + }, + { + "epoch": 0.98, + "learning_rate": 1.5783609658572284e-08, + "loss": 2.6194, + "step": 4661 + }, + { + "epoch": 0.98, + "learning_rate": 1.5492824122296803e-08, + "loss": 2.6416, + "step": 4662 + }, + { + "epoch": 0.98, + "learning_rate": 1.520473800092792e-08, + "loss": 2.5823, + "step": 4663 + }, + { + "epoch": 0.98, + "learning_rate": 1.4919351450480847e-08, + "loss": 2.6512, + "step": 4664 + }, + { + "epoch": 0.98, + "learning_rate": 1.4636664625510832e-08, + "loss": 2.6168, + "step": 4665 + }, + { + "epoch": 0.98, + "learning_rate": 1.4356677679109865e-08, + "loss": 2.6133, + "step": 4666 + }, + { + "epoch": 0.98, + "learning_rate": 1.4079390762907763e-08, + "loss": 2.6389, + "step": 4667 + }, + { + "epoch": 0.98, + "learning_rate": 1.3804804027072183e-08, + "loss": 2.5868, + "step": 4668 + }, + { + "epoch": 0.98, + "learning_rate": 1.3532917620309171e-08, + "loss": 2.6258, + "step": 4669 + }, + { + "epoch": 0.98, + "learning_rate": 1.3263731689860949e-08, + "loss": 2.7017, + "step": 4670 + }, + { + "epoch": 0.98, + "learning_rate": 1.2997246381508676e-08, + "loss": 2.666, + "step": 4671 + }, + { + "epoch": 0.98, + "learning_rate": 1.273346183957025e-08, + "loss": 2.6306, + "step": 4672 + }, + { + "epoch": 0.98, + "learning_rate": 1.2472378206901392e-08, + "loss": 2.6471, + "step": 4673 + }, + { + "epoch": 0.98, + "learning_rate": 1.221399562489456e-08, + "loss": 2.6588, + "step": 4674 + }, + { + "epoch": 0.98, + "learning_rate": 1.1958314233479485e-08, + "loss": 2.5958, + "step": 4675 + }, + { + "epoch": 0.98, + "learning_rate": 1.1705334171123739e-08, + "loss": 2.6646, + "step": 4676 + }, + { + "epoch": 0.98, + "learning_rate": 1.1455055574831065e-08, + "loss": 2.6205, + "step": 4677 + }, + { + "epoch": 0.98, + "learning_rate": 1.1207478580143039e-08, + "loss": 2.5836, + "step": 4678 + }, + { + "epoch": 0.98, + "learning_rate": 1.0962603321137965e-08, + "loss": 2.6081, + "step": 4679 + }, + { + "epoch": 0.98, + "learning_rate": 1.0720429930429211e-08, + "loss": 2.6146, + "step": 4680 + }, + { + "epoch": 0.98, + "learning_rate": 1.0480958539170195e-08, + "loss": 2.6863, + "step": 4681 + }, + { + "epoch": 0.98, + "learning_rate": 1.0244189277048289e-08, + "loss": 2.6122, + "step": 4682 + }, + { + "epoch": 0.98, + "learning_rate": 1.0010122272288703e-08, + "loss": 2.7438, + "step": 4683 + }, + { + "epoch": 0.98, + "learning_rate": 9.778757651652815e-09, + "loss": 2.6493, + "step": 4684 + }, + { + "epoch": 0.98, + "learning_rate": 9.550095540439841e-09, + "loss": 2.6281, + "step": 4685 + }, + { + "epoch": 0.98, + "learning_rate": 9.324136062482391e-09, + "loss": 2.6028, + "step": 4686 + }, + { + "epoch": 0.98, + "learning_rate": 9.100879340153135e-09, + "loss": 2.6486, + "step": 4687 + }, + { + "epoch": 0.98, + "learning_rate": 8.880325494358132e-09, + "loss": 2.6764, + "step": 4688 + }, + { + "epoch": 0.98, + "learning_rate": 8.66247464454073e-09, + "loss": 2.6347, + "step": 4689 + }, + { + "epoch": 0.98, + "learning_rate": 8.447326908680997e-09, + "loss": 2.6263, + "step": 4690 + }, + { + "epoch": 0.98, + "learning_rate": 8.23488240329462e-09, + "loss": 2.626, + "step": 4691 + }, + { + "epoch": 0.98, + "learning_rate": 8.0251412434329e-09, + "loss": 2.6357, + "step": 4692 + }, + { + "epoch": 0.98, + "learning_rate": 7.818103542683308e-09, + "loss": 2.6464, + "step": 4693 + }, + { + "epoch": 0.98, + "learning_rate": 7.613769413169492e-09, + "loss": 2.6344, + "step": 4694 + }, + { + "epoch": 0.98, + "learning_rate": 7.412138965551264e-09, + "loss": 2.6382, + "step": 4695 + }, + { + "epoch": 0.98, + "learning_rate": 7.213212309023499e-09, + "loss": 2.6478, + "step": 4696 + }, + { + "epoch": 0.98, + "learning_rate": 7.016989551317244e-09, + "loss": 2.6409, + "step": 4697 + }, + { + "epoch": 0.99, + "learning_rate": 6.823470798699161e-09, + "loss": 2.6434, + "step": 4698 + }, + { + "epoch": 0.99, + "learning_rate": 6.632656155970974e-09, + "loss": 2.6295, + "step": 4699 + }, + { + "epoch": 0.99, + "learning_rate": 6.4445457264711295e-09, + "loss": 2.6547, + "step": 4700 + }, + { + "epoch": 0.99, + "learning_rate": 6.259139612072584e-09, + "loss": 2.6525, + "step": 4701 + }, + { + "epoch": 0.99, + "learning_rate": 6.076437913183908e-09, + "loss": 2.6508, + "step": 4702 + }, + { + "epoch": 0.99, + "learning_rate": 5.896440728749286e-09, + "loss": 2.5756, + "step": 4703 + }, + { + "epoch": 0.99, + "learning_rate": 5.719148156248522e-09, + "loss": 2.6911, + "step": 4704 + }, + { + "epoch": 0.99, + "learning_rate": 5.544560291696477e-09, + "loss": 2.6192, + "step": 4705 + }, + { + "epoch": 0.99, + "learning_rate": 5.37267722964252e-09, + "loss": 2.6717, + "step": 4706 + }, + { + "epoch": 0.99, + "learning_rate": 5.20349906317219e-09, + "loss": 2.6618, + "step": 4707 + }, + { + "epoch": 0.99, + "learning_rate": 5.0370258839055334e-09, + "loss": 2.6241, + "step": 4708 + }, + { + "epoch": 0.99, + "learning_rate": 4.8732577819982084e-09, + "loss": 2.6114, + "step": 4709 + }, + { + "epoch": 0.99, + "learning_rate": 4.712194846140383e-09, + "loss": 2.646, + "step": 4710 + }, + { + "epoch": 0.99, + "learning_rate": 4.553837163557839e-09, + "loss": 2.6748, + "step": 4711 + }, + { + "epoch": 0.99, + "learning_rate": 4.398184820010865e-09, + "loss": 2.7086, + "step": 4712 + }, + { + "epoch": 0.99, + "learning_rate": 4.245237899794252e-09, + "loss": 2.563, + "step": 4713 + }, + { + "epoch": 0.99, + "learning_rate": 4.094996485738411e-09, + "loss": 2.5851, + "step": 4714 + }, + { + "epoch": 0.99, + "learning_rate": 3.9474606592088125e-09, + "loss": 2.6557, + "step": 4715 + }, + { + "epoch": 0.99, + "learning_rate": 3.8026305001037656e-09, + "loss": 2.6551, + "step": 4716 + }, + { + "epoch": 0.99, + "learning_rate": 3.6605060868588614e-09, + "loss": 2.7141, + "step": 4717 + }, + { + "epoch": 0.99, + "learning_rate": 3.5210874964425323e-09, + "loss": 2.6031, + "step": 4718 + }, + { + "epoch": 0.99, + "learning_rate": 3.384374804358825e-09, + "loss": 2.6681, + "step": 4719 + }, + { + "epoch": 0.99, + "learning_rate": 3.2503680846457387e-09, + "loss": 2.6623, + "step": 4720 + }, + { + "epoch": 0.99, + "learning_rate": 3.1190674098757756e-09, + "loss": 2.6582, + "step": 4721 + }, + { + "epoch": 0.99, + "learning_rate": 2.9904728511570557e-09, + "loss": 2.6484, + "step": 4722 + }, + { + "epoch": 0.99, + "learning_rate": 2.864584478131094e-09, + "loss": 2.5511, + "step": 4723 + }, + { + "epoch": 0.99, + "learning_rate": 2.7414023589739104e-09, + "loss": 2.622, + "step": 4724 + }, + { + "epoch": 0.99, + "learning_rate": 2.620926560396031e-09, + "loss": 2.6236, + "step": 4725 + }, + { + "epoch": 0.99, + "learning_rate": 2.5031571476424875e-09, + "loss": 2.6438, + "step": 4726 + }, + { + "epoch": 0.99, + "learning_rate": 2.3880941844933727e-09, + "loss": 2.6717, + "step": 4727 + }, + { + "epoch": 0.99, + "learning_rate": 2.275737733261618e-09, + "loss": 2.653, + "step": 4728 + }, + { + "epoch": 0.99, + "learning_rate": 2.1660878547946627e-09, + "loss": 2.6646, + "step": 4729 + }, + { + "epoch": 0.99, + "learning_rate": 2.0591446084755608e-09, + "loss": 2.5875, + "step": 4730 + }, + { + "epoch": 0.99, + "learning_rate": 1.9549080522202057e-09, + "loss": 2.6218, + "step": 4731 + }, + { + "epoch": 0.99, + "learning_rate": 1.8533782424789981e-09, + "loss": 2.6166, + "step": 4732 + }, + { + "epoch": 0.99, + "learning_rate": 1.754555234236288e-09, + "loss": 2.6072, + "step": 4733 + }, + { + "epoch": 0.99, + "learning_rate": 1.658439081011487e-09, + "loss": 2.6215, + "step": 4734 + }, + { + "epoch": 0.99, + "learning_rate": 1.565029834856291e-09, + "loss": 2.6553, + "step": 4735 + }, + { + "epoch": 0.99, + "learning_rate": 1.4743275463585672e-09, + "loss": 2.6654, + "step": 4736 + }, + { + "epoch": 0.99, + "learning_rate": 1.386332264639023e-09, + "loss": 2.6859, + "step": 4737 + }, + { + "epoch": 0.99, + "learning_rate": 1.301044037351762e-09, + "loss": 2.6207, + "step": 4738 + }, + { + "epoch": 0.99, + "learning_rate": 1.2184629106859468e-09, + "loss": 2.6519, + "step": 4739 + }, + { + "epoch": 0.99, + "learning_rate": 1.1385889293646923e-09, + "loss": 2.5686, + "step": 4740 + }, + { + "epoch": 0.99, + "learning_rate": 1.0614221366439525e-09, + "loss": 2.6518, + "step": 4741 + }, + { + "epoch": 0.99, + "learning_rate": 9.869625743147426e-10, + "loss": 2.7043, + "step": 4742 + }, + { + "epoch": 0.99, + "learning_rate": 9.15210282700918e-10, + "loss": 2.628, + "step": 4743 + }, + { + "epoch": 0.99, + "learning_rate": 8.461653006613945e-10, + "loss": 2.5939, + "step": 4744 + }, + { + "epoch": 0.99, + "learning_rate": 7.798276655879289e-10, + "loss": 2.5747, + "step": 4745 + }, + { + "epoch": 1.0, + "learning_rate": 7.161974134062277e-10, + "loss": 2.5801, + "step": 4746 + }, + { + "epoch": 1.0, + "learning_rate": 6.552745785765035e-10, + "loss": 2.6502, + "step": 4747 + }, + { + "epoch": 1.0, + "learning_rate": 5.97059194091254e-10, + "loss": 2.6519, + "step": 4748 + }, + { + "epoch": 1.0, + "learning_rate": 5.415512914785925e-10, + "loss": 2.6317, + "step": 4749 + }, + { + "epoch": 1.0, + "learning_rate": 4.887509007994728e-10, + "loss": 2.6623, + "step": 4750 + }, + { + "epoch": 1.0, + "learning_rate": 4.3865805064768895e-10, + "loss": 2.6586, + "step": 4751 + }, + { + "epoch": 1.0, + "learning_rate": 3.912727681520956e-10, + "loss": 2.6522, + "step": 4752 + }, + { + "epoch": 1.0, + "learning_rate": 3.4659507897438773e-10, + "loss": 2.6631, + "step": 4753 + }, + { + "epoch": 1.0, + "learning_rate": 3.0462500731076595e-10, + "loss": 2.6252, + "step": 4754 + }, + { + "epoch": 1.0, + "learning_rate": 2.6536257589082626e-10, + "loss": 2.6275, + "step": 4755 + }, + { + "epoch": 1.0, + "learning_rate": 2.288078059764498e-10, + "loss": 2.656, + "step": 4756 + }, + { + "epoch": 1.0, + "learning_rate": 1.9496071736513356e-10, + "loss": 2.654, + "step": 4757 + }, + { + "epoch": 1.0, + "learning_rate": 1.6382132838665966e-10, + "loss": 2.7308, + "step": 4758 + }, + { + "epoch": 1.0, + "learning_rate": 1.3538965590587095e-10, + "loss": 2.6288, + "step": 4759 + }, + { + "epoch": 1.0, + "learning_rate": 1.0966571531878523e-10, + "loss": 2.6425, + "step": 4760 + }, + { + "epoch": 1.0, + "learning_rate": 8.664952055759123e-11, + "loss": 2.6344, + "step": 4761 + }, + { + "epoch": 1.0, + "learning_rate": 6.634108408676288e-11, + "loss": 2.6571, + "step": 4762 + }, + { + "epoch": 1.0, + "learning_rate": 4.874041690416942e-11, + "loss": 2.6473, + "step": 4763 + }, + { + "epoch": 1.0, + "learning_rate": 3.384752854218576e-11, + "loss": 2.615, + "step": 4764 + }, + { + "epoch": 1.0, + "learning_rate": 2.1662427065471948e-11, + "loss": 2.6817, + "step": 4765 + }, + { + "epoch": 1.0, + "learning_rate": 1.2185119073748753e-11, + "loss": 2.6722, + "step": 4766 + }, + { + "epoch": 1.0, + "learning_rate": 5.4156096990221065e-12, + "loss": 2.6384, + "step": 4767 + }, + { + "epoch": 1.0, + "learning_rate": 1.3539026083586593e-12, + "loss": 2.6514, + "step": 4768 + }, + { + "epoch": 1.0, + "learning_rate": 0.0, + "loss": 2.6362, + "step": 4769 + }, + { + "epoch": 1.0, + "step": 4769, + "total_flos": 1.579472545147519e+18, + "train_loss": 2.6475883937027, + "train_runtime": 12560.2462, + "train_samples_per_second": 48.599, + "train_steps_per_second": 0.38 + } + ], + "max_steps": 4769, + "num_train_epochs": 1, + "total_flos": 1.579472545147519e+18, + "trial_name": null, + "trial_params": null +}