diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,31686 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0, + "eval_steps": 500, + "global_step": 5276, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 1.2578616352201258e-07, + "loss": 1.2468, + "step": 1 + }, + { + "epoch": 0.0, + "learning_rate": 2.5157232704402517e-07, + "loss": 1.3521, + "step": 2 + }, + { + "epoch": 0.0, + "learning_rate": 3.773584905660378e-07, + "loss": 1.2954, + "step": 3 + }, + { + "epoch": 0.0, + "learning_rate": 5.031446540880503e-07, + "loss": 1.2968, + "step": 4 + }, + { + "epoch": 0.0, + "learning_rate": 6.28930817610063e-07, + "loss": 1.3373, + "step": 5 + }, + { + "epoch": 0.0, + "learning_rate": 7.547169811320755e-07, + "loss": 1.2575, + "step": 6 + }, + { + "epoch": 0.0, + "learning_rate": 8.805031446540881e-07, + "loss": 1.3151, + "step": 7 + }, + { + "epoch": 0.0, + "learning_rate": 1.0062893081761007e-06, + "loss": 1.256, + "step": 8 + }, + { + "epoch": 0.0, + "learning_rate": 1.1320754716981133e-06, + "loss": 1.2682, + "step": 9 + }, + { + "epoch": 0.0, + "learning_rate": 1.257861635220126e-06, + "loss": 1.2259, + "step": 10 + }, + { + "epoch": 0.0, + "learning_rate": 1.3836477987421384e-06, + "loss": 0.2806, + "step": 11 + }, + { + "epoch": 0.0, + "learning_rate": 1.509433962264151e-06, + "loss": 1.2461, + "step": 12 + }, + { + "epoch": 0.0, + "learning_rate": 1.6352201257861635e-06, + "loss": 0.2414, + "step": 13 + }, + { + "epoch": 0.0, + "learning_rate": 1.7610062893081762e-06, + "loss": 1.1702, + "step": 14 + }, + { + "epoch": 0.0, + "learning_rate": 1.8867924528301889e-06, + "loss": 1.224, + "step": 15 + }, + { + "epoch": 0.0, + "learning_rate": 2.0125786163522013e-06, + "loss": 1.146, + "step": 16 + }, + { + "epoch": 0.0, + "learning_rate": 2.138364779874214e-06, + "loss": 1.1212, + "step": 17 + }, + { + "epoch": 0.0, + "learning_rate": 2.2641509433962266e-06, + "loss": 1.0892, + "step": 18 + }, + { + "epoch": 0.0, + "learning_rate": 2.389937106918239e-06, + "loss": 1.0945, + "step": 19 + }, + { + "epoch": 0.0, + "learning_rate": 2.515723270440252e-06, + "loss": 1.0964, + "step": 20 + }, + { + "epoch": 0.0, + "learning_rate": 2.6415094339622644e-06, + "loss": 1.0913, + "step": 21 + }, + { + "epoch": 0.0, + "learning_rate": 2.767295597484277e-06, + "loss": 1.0595, + "step": 22 + }, + { + "epoch": 0.0, + "learning_rate": 2.8930817610062893e-06, + "loss": 1.1281, + "step": 23 + }, + { + "epoch": 0.0, + "learning_rate": 3.018867924528302e-06, + "loss": 1.0541, + "step": 24 + }, + { + "epoch": 0.0, + "learning_rate": 3.1446540880503146e-06, + "loss": 1.052, + "step": 25 + }, + { + "epoch": 0.0, + "learning_rate": 3.270440251572327e-06, + "loss": 1.0685, + "step": 26 + }, + { + "epoch": 0.01, + "learning_rate": 3.3962264150943395e-06, + "loss": 0.9898, + "step": 27 + }, + { + "epoch": 0.01, + "learning_rate": 3.5220125786163524e-06, + "loss": 0.9717, + "step": 28 + }, + { + "epoch": 0.01, + "learning_rate": 3.647798742138365e-06, + "loss": 0.9909, + "step": 29 + }, + { + "epoch": 0.01, + "learning_rate": 3.7735849056603777e-06, + "loss": 1.0362, + "step": 30 + }, + { + "epoch": 0.01, + "learning_rate": 3.89937106918239e-06, + "loss": 0.9585, + "step": 31 + }, + { + "epoch": 0.01, + "learning_rate": 4.025157232704403e-06, + "loss": 1.0337, + "step": 32 + }, + { + "epoch": 0.01, + "learning_rate": 4.150943396226416e-06, + "loss": 1.0617, + "step": 33 + }, + { + "epoch": 0.01, + "learning_rate": 4.276729559748428e-06, + "loss": 0.9499, + "step": 34 + }, + { + "epoch": 0.01, + "learning_rate": 4.402515723270441e-06, + "loss": 1.0051, + "step": 35 + }, + { + "epoch": 0.01, + "learning_rate": 4.528301886792453e-06, + "loss": 0.993, + "step": 36 + }, + { + "epoch": 0.01, + "learning_rate": 4.654088050314466e-06, + "loss": 0.9645, + "step": 37 + }, + { + "epoch": 0.01, + "learning_rate": 4.779874213836478e-06, + "loss": 0.99, + "step": 38 + }, + { + "epoch": 0.01, + "learning_rate": 4.905660377358491e-06, + "loss": 0.9783, + "step": 39 + }, + { + "epoch": 0.01, + "learning_rate": 5.031446540880504e-06, + "loss": 1.0076, + "step": 40 + }, + { + "epoch": 0.01, + "learning_rate": 5.157232704402516e-06, + "loss": 0.9925, + "step": 41 + }, + { + "epoch": 0.01, + "learning_rate": 5.283018867924529e-06, + "loss": 0.9366, + "step": 42 + }, + { + "epoch": 0.01, + "learning_rate": 5.408805031446541e-06, + "loss": 0.9185, + "step": 43 + }, + { + "epoch": 0.01, + "learning_rate": 5.534591194968554e-06, + "loss": 0.9468, + "step": 44 + }, + { + "epoch": 0.01, + "learning_rate": 5.660377358490566e-06, + "loss": 0.9773, + "step": 45 + }, + { + "epoch": 0.01, + "learning_rate": 5.786163522012579e-06, + "loss": 0.9529, + "step": 46 + }, + { + "epoch": 0.01, + "learning_rate": 5.911949685534591e-06, + "loss": 0.9357, + "step": 47 + }, + { + "epoch": 0.01, + "learning_rate": 6.037735849056604e-06, + "loss": 0.9366, + "step": 48 + }, + { + "epoch": 0.01, + "learning_rate": 6.163522012578617e-06, + "loss": 0.9503, + "step": 49 + }, + { + "epoch": 0.01, + "learning_rate": 6.289308176100629e-06, + "loss": 0.9052, + "step": 50 + }, + { + "epoch": 0.01, + "learning_rate": 6.415094339622642e-06, + "loss": 0.9921, + "step": 51 + }, + { + "epoch": 0.01, + "learning_rate": 6.540880503144654e-06, + "loss": 0.971, + "step": 52 + }, + { + "epoch": 0.01, + "learning_rate": 6.666666666666667e-06, + "loss": 0.9493, + "step": 53 + }, + { + "epoch": 0.01, + "learning_rate": 6.792452830188679e-06, + "loss": 0.9083, + "step": 54 + }, + { + "epoch": 0.01, + "learning_rate": 6.9182389937106915e-06, + "loss": 0.9482, + "step": 55 + }, + { + "epoch": 0.01, + "learning_rate": 7.044025157232705e-06, + "loss": 0.917, + "step": 56 + }, + { + "epoch": 0.01, + "learning_rate": 7.169811320754717e-06, + "loss": 0.9681, + "step": 57 + }, + { + "epoch": 0.01, + "learning_rate": 7.29559748427673e-06, + "loss": 0.9529, + "step": 58 + }, + { + "epoch": 0.01, + "learning_rate": 7.421383647798742e-06, + "loss": 0.9915, + "step": 59 + }, + { + "epoch": 0.01, + "learning_rate": 7.5471698113207555e-06, + "loss": 0.8782, + "step": 60 + }, + { + "epoch": 0.01, + "learning_rate": 7.672955974842768e-06, + "loss": 0.9703, + "step": 61 + }, + { + "epoch": 0.01, + "learning_rate": 7.79874213836478e-06, + "loss": 0.9383, + "step": 62 + }, + { + "epoch": 0.01, + "learning_rate": 7.924528301886793e-06, + "loss": 0.2508, + "step": 63 + }, + { + "epoch": 0.01, + "learning_rate": 8.050314465408805e-06, + "loss": 0.9228, + "step": 64 + }, + { + "epoch": 0.01, + "learning_rate": 8.17610062893082e-06, + "loss": 0.9546, + "step": 65 + }, + { + "epoch": 0.01, + "learning_rate": 8.301886792452832e-06, + "loss": 0.9585, + "step": 66 + }, + { + "epoch": 0.01, + "learning_rate": 8.427672955974844e-06, + "loss": 0.9162, + "step": 67 + }, + { + "epoch": 0.01, + "learning_rate": 8.553459119496857e-06, + "loss": 0.9361, + "step": 68 + }, + { + "epoch": 0.01, + "learning_rate": 8.67924528301887e-06, + "loss": 0.2163, + "step": 69 + }, + { + "epoch": 0.01, + "learning_rate": 8.805031446540882e-06, + "loss": 0.2366, + "step": 70 + }, + { + "epoch": 0.01, + "learning_rate": 8.930817610062894e-06, + "loss": 0.9556, + "step": 71 + }, + { + "epoch": 0.01, + "learning_rate": 9.056603773584907e-06, + "loss": 0.973, + "step": 72 + }, + { + "epoch": 0.01, + "learning_rate": 9.182389937106919e-06, + "loss": 0.9082, + "step": 73 + }, + { + "epoch": 0.01, + "learning_rate": 9.308176100628931e-06, + "loss": 0.2238, + "step": 74 + }, + { + "epoch": 0.01, + "learning_rate": 9.433962264150944e-06, + "loss": 0.8988, + "step": 75 + }, + { + "epoch": 0.01, + "learning_rate": 9.559748427672956e-06, + "loss": 0.9422, + "step": 76 + }, + { + "epoch": 0.01, + "learning_rate": 9.685534591194969e-06, + "loss": 0.986, + "step": 77 + }, + { + "epoch": 0.01, + "learning_rate": 9.811320754716981e-06, + "loss": 0.9315, + "step": 78 + }, + { + "epoch": 0.01, + "learning_rate": 9.937106918238994e-06, + "loss": 0.9037, + "step": 79 + }, + { + "epoch": 0.02, + "learning_rate": 1.0062893081761008e-05, + "loss": 0.9697, + "step": 80 + }, + { + "epoch": 0.02, + "learning_rate": 1.018867924528302e-05, + "loss": 0.9146, + "step": 81 + }, + { + "epoch": 0.02, + "learning_rate": 1.0314465408805033e-05, + "loss": 0.9449, + "step": 82 + }, + { + "epoch": 0.02, + "learning_rate": 1.0440251572327045e-05, + "loss": 0.9528, + "step": 83 + }, + { + "epoch": 0.02, + "learning_rate": 1.0566037735849058e-05, + "loss": 0.894, + "step": 84 + }, + { + "epoch": 0.02, + "learning_rate": 1.069182389937107e-05, + "loss": 0.9226, + "step": 85 + }, + { + "epoch": 0.02, + "learning_rate": 1.0817610062893083e-05, + "loss": 0.9313, + "step": 86 + }, + { + "epoch": 0.02, + "learning_rate": 1.0943396226415095e-05, + "loss": 0.97, + "step": 87 + }, + { + "epoch": 0.02, + "learning_rate": 1.1069182389937107e-05, + "loss": 0.9749, + "step": 88 + }, + { + "epoch": 0.02, + "learning_rate": 1.119496855345912e-05, + "loss": 0.9569, + "step": 89 + }, + { + "epoch": 0.02, + "learning_rate": 1.1320754716981132e-05, + "loss": 0.9357, + "step": 90 + }, + { + "epoch": 0.02, + "learning_rate": 1.1446540880503145e-05, + "loss": 0.8904, + "step": 91 + }, + { + "epoch": 0.02, + "learning_rate": 1.1572327044025157e-05, + "loss": 0.909, + "step": 92 + }, + { + "epoch": 0.02, + "learning_rate": 1.169811320754717e-05, + "loss": 0.2334, + "step": 93 + }, + { + "epoch": 0.02, + "learning_rate": 1.1823899371069182e-05, + "loss": 0.8795, + "step": 94 + }, + { + "epoch": 0.02, + "learning_rate": 1.1949685534591196e-05, + "loss": 0.9611, + "step": 95 + }, + { + "epoch": 0.02, + "learning_rate": 1.2075471698113209e-05, + "loss": 0.8727, + "step": 96 + }, + { + "epoch": 0.02, + "learning_rate": 1.2201257861635221e-05, + "loss": 0.9563, + "step": 97 + }, + { + "epoch": 0.02, + "learning_rate": 1.2327044025157234e-05, + "loss": 0.9687, + "step": 98 + }, + { + "epoch": 0.02, + "learning_rate": 1.2452830188679246e-05, + "loss": 0.8722, + "step": 99 + }, + { + "epoch": 0.02, + "learning_rate": 1.2578616352201259e-05, + "loss": 0.8696, + "step": 100 + }, + { + "epoch": 0.02, + "learning_rate": 1.2704402515723271e-05, + "loss": 0.9163, + "step": 101 + }, + { + "epoch": 0.02, + "learning_rate": 1.2830188679245283e-05, + "loss": 0.9538, + "step": 102 + }, + { + "epoch": 0.02, + "learning_rate": 1.2955974842767296e-05, + "loss": 0.9214, + "step": 103 + }, + { + "epoch": 0.02, + "learning_rate": 1.3081761006289308e-05, + "loss": 0.9384, + "step": 104 + }, + { + "epoch": 0.02, + "learning_rate": 1.320754716981132e-05, + "loss": 0.942, + "step": 105 + }, + { + "epoch": 0.02, + "learning_rate": 1.3333333333333333e-05, + "loss": 0.906, + "step": 106 + }, + { + "epoch": 0.02, + "learning_rate": 1.3459119496855346e-05, + "loss": 0.921, + "step": 107 + }, + { + "epoch": 0.02, + "learning_rate": 1.3584905660377358e-05, + "loss": 0.921, + "step": 108 + }, + { + "epoch": 0.02, + "learning_rate": 1.371069182389937e-05, + "loss": 0.96, + "step": 109 + }, + { + "epoch": 0.02, + "learning_rate": 1.3836477987421383e-05, + "loss": 0.9337, + "step": 110 + }, + { + "epoch": 0.02, + "learning_rate": 1.3962264150943397e-05, + "loss": 0.2299, + "step": 111 + }, + { + "epoch": 0.02, + "learning_rate": 1.408805031446541e-05, + "loss": 0.8846, + "step": 112 + }, + { + "epoch": 0.02, + "learning_rate": 1.4213836477987422e-05, + "loss": 0.9484, + "step": 113 + }, + { + "epoch": 0.02, + "learning_rate": 1.4339622641509435e-05, + "loss": 0.25, + "step": 114 + }, + { + "epoch": 0.02, + "learning_rate": 1.4465408805031447e-05, + "loss": 0.9006, + "step": 115 + }, + { + "epoch": 0.02, + "learning_rate": 1.459119496855346e-05, + "loss": 0.8915, + "step": 116 + }, + { + "epoch": 0.02, + "learning_rate": 1.4716981132075472e-05, + "loss": 0.94, + "step": 117 + }, + { + "epoch": 0.02, + "learning_rate": 1.4842767295597484e-05, + "loss": 0.8982, + "step": 118 + }, + { + "epoch": 0.02, + "learning_rate": 1.4968553459119497e-05, + "loss": 0.2128, + "step": 119 + }, + { + "epoch": 0.02, + "learning_rate": 1.5094339622641511e-05, + "loss": 0.8771, + "step": 120 + }, + { + "epoch": 0.02, + "learning_rate": 1.5220125786163525e-05, + "loss": 0.8914, + "step": 121 + }, + { + "epoch": 0.02, + "learning_rate": 1.5345911949685536e-05, + "loss": 0.9134, + "step": 122 + }, + { + "epoch": 0.02, + "learning_rate": 1.547169811320755e-05, + "loss": 0.9125, + "step": 123 + }, + { + "epoch": 0.02, + "learning_rate": 1.559748427672956e-05, + "loss": 0.8778, + "step": 124 + }, + { + "epoch": 0.02, + "learning_rate": 1.5723270440251575e-05, + "loss": 0.9545, + "step": 125 + }, + { + "epoch": 0.02, + "learning_rate": 1.5849056603773586e-05, + "loss": 0.9372, + "step": 126 + }, + { + "epoch": 0.02, + "learning_rate": 1.59748427672956e-05, + "loss": 0.8722, + "step": 127 + }, + { + "epoch": 0.02, + "learning_rate": 1.610062893081761e-05, + "loss": 0.8503, + "step": 128 + }, + { + "epoch": 0.02, + "learning_rate": 1.6226415094339625e-05, + "loss": 0.9891, + "step": 129 + }, + { + "epoch": 0.02, + "learning_rate": 1.635220125786164e-05, + "loss": 0.9127, + "step": 130 + }, + { + "epoch": 0.02, + "learning_rate": 1.647798742138365e-05, + "loss": 0.9218, + "step": 131 + }, + { + "epoch": 0.03, + "learning_rate": 1.6603773584905664e-05, + "loss": 0.9229, + "step": 132 + }, + { + "epoch": 0.03, + "learning_rate": 1.6729559748427675e-05, + "loss": 0.8741, + "step": 133 + }, + { + "epoch": 0.03, + "learning_rate": 1.685534591194969e-05, + "loss": 0.9295, + "step": 134 + }, + { + "epoch": 0.03, + "learning_rate": 1.69811320754717e-05, + "loss": 0.9082, + "step": 135 + }, + { + "epoch": 0.03, + "learning_rate": 1.7106918238993714e-05, + "loss": 0.8943, + "step": 136 + }, + { + "epoch": 0.03, + "learning_rate": 1.7232704402515724e-05, + "loss": 0.8687, + "step": 137 + }, + { + "epoch": 0.03, + "learning_rate": 1.735849056603774e-05, + "loss": 0.9303, + "step": 138 + }, + { + "epoch": 0.03, + "learning_rate": 1.748427672955975e-05, + "loss": 0.9487, + "step": 139 + }, + { + "epoch": 0.03, + "learning_rate": 1.7610062893081763e-05, + "loss": 0.2515, + "step": 140 + }, + { + "epoch": 0.03, + "learning_rate": 1.7735849056603774e-05, + "loss": 0.9346, + "step": 141 + }, + { + "epoch": 0.03, + "learning_rate": 1.7861635220125788e-05, + "loss": 0.9154, + "step": 142 + }, + { + "epoch": 0.03, + "learning_rate": 1.79874213836478e-05, + "loss": 0.9062, + "step": 143 + }, + { + "epoch": 0.03, + "learning_rate": 1.8113207547169813e-05, + "loss": 0.8482, + "step": 144 + }, + { + "epoch": 0.03, + "learning_rate": 1.8238993710691827e-05, + "loss": 0.9559, + "step": 145 + }, + { + "epoch": 0.03, + "learning_rate": 1.8364779874213838e-05, + "loss": 0.9193, + "step": 146 + }, + { + "epoch": 0.03, + "learning_rate": 1.8490566037735852e-05, + "loss": 0.9326, + "step": 147 + }, + { + "epoch": 0.03, + "learning_rate": 1.8616352201257863e-05, + "loss": 0.8773, + "step": 148 + }, + { + "epoch": 0.03, + "learning_rate": 1.8742138364779877e-05, + "loss": 0.9505, + "step": 149 + }, + { + "epoch": 0.03, + "learning_rate": 1.8867924528301888e-05, + "loss": 0.9435, + "step": 150 + }, + { + "epoch": 0.03, + "learning_rate": 1.8993710691823902e-05, + "loss": 0.9004, + "step": 151 + }, + { + "epoch": 0.03, + "learning_rate": 1.9119496855345913e-05, + "loss": 0.9069, + "step": 152 + }, + { + "epoch": 0.03, + "learning_rate": 1.9245283018867927e-05, + "loss": 0.9357, + "step": 153 + }, + { + "epoch": 0.03, + "learning_rate": 1.9371069182389938e-05, + "loss": 0.8625, + "step": 154 + }, + { + "epoch": 0.03, + "learning_rate": 1.9496855345911952e-05, + "loss": 0.2306, + "step": 155 + }, + { + "epoch": 0.03, + "learning_rate": 1.9622641509433963e-05, + "loss": 0.9701, + "step": 156 + }, + { + "epoch": 0.03, + "learning_rate": 1.9748427672955977e-05, + "loss": 0.8967, + "step": 157 + }, + { + "epoch": 0.03, + "learning_rate": 1.9874213836477987e-05, + "loss": 0.9492, + "step": 158 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.966, + "step": 159 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999998115314448e-05, + "loss": 0.907, + "step": 160 + }, + { + "epoch": 0.03, + "learning_rate": 1.999999246125849e-05, + "loss": 0.8955, + "step": 161 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999983037834263e-05, + "loss": 0.9339, + "step": 162 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999969845045317e-05, + "loss": 0.9065, + "step": 163 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999952882896627e-05, + "loss": 0.8834, + "step": 164 + }, + { + "epoch": 0.03, + "learning_rate": 1.999993215139459e-05, + "loss": 0.9336, + "step": 165 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999907650547006e-05, + "loss": 0.8868, + "step": 166 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999879380363127e-05, + "loss": 0.8952, + "step": 167 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999847340853603e-05, + "loss": 0.969, + "step": 168 + }, + { + "epoch": 0.03, + "learning_rate": 1.999981153203051e-05, + "loss": 0.9257, + "step": 169 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999771953907345e-05, + "loss": 0.9444, + "step": 170 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999728606499027e-05, + "loss": 0.9056, + "step": 171 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999681489821897e-05, + "loss": 0.8924, + "step": 172 + }, + { + "epoch": 0.03, + "learning_rate": 1.999963060389371e-05, + "loss": 0.9132, + "step": 173 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999575948733655e-05, + "loss": 0.8458, + "step": 174 + }, + { + "epoch": 0.03, + "learning_rate": 1.999951752436233e-05, + "loss": 0.8931, + "step": 175 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999455330801752e-05, + "loss": 0.233, + "step": 176 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999389368075372e-05, + "loss": 0.9683, + "step": 177 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999319636208048e-05, + "loss": 0.9261, + "step": 178 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999246135226068e-05, + "loss": 0.869, + "step": 179 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999168865157137e-05, + "loss": 0.8656, + "step": 180 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999087826030376e-05, + "loss": 0.8922, + "step": 181 + }, + { + "epoch": 0.03, + "learning_rate": 1.999900301787634e-05, + "loss": 0.9242, + "step": 182 + }, + { + "epoch": 0.03, + "learning_rate": 1.999891444072699e-05, + "loss": 0.9459, + "step": 183 + }, + { + "epoch": 0.03, + "learning_rate": 1.9998822094615717e-05, + "loss": 0.9344, + "step": 184 + }, + { + "epoch": 0.04, + "learning_rate": 1.9998725979577327e-05, + "loss": 0.9016, + "step": 185 + }, + { + "epoch": 0.04, + "learning_rate": 1.999862609564805e-05, + "loss": 0.9009, + "step": 186 + }, + { + "epoch": 0.04, + "learning_rate": 1.999852244286554e-05, + "loss": 0.9206, + "step": 187 + }, + { + "epoch": 0.04, + "learning_rate": 1.9998415021268864e-05, + "loss": 0.8998, + "step": 188 + }, + { + "epoch": 0.04, + "learning_rate": 1.9998303830898514e-05, + "loss": 0.9341, + "step": 189 + }, + { + "epoch": 0.04, + "learning_rate": 1.99981888717964e-05, + "loss": 0.8813, + "step": 190 + }, + { + "epoch": 0.04, + "learning_rate": 1.999807014400586e-05, + "loss": 0.9866, + "step": 191 + }, + { + "epoch": 0.04, + "learning_rate": 1.9997947647571636e-05, + "loss": 0.9207, + "step": 192 + }, + { + "epoch": 0.04, + "learning_rate": 1.9997821382539914e-05, + "loss": 0.9444, + "step": 193 + }, + { + "epoch": 0.04, + "learning_rate": 1.9997691348958278e-05, + "loss": 0.911, + "step": 194 + }, + { + "epoch": 0.04, + "learning_rate": 1.999755754687575e-05, + "loss": 0.9595, + "step": 195 + }, + { + "epoch": 0.04, + "learning_rate": 1.999741997634276e-05, + "loss": 0.9205, + "step": 196 + }, + { + "epoch": 0.04, + "learning_rate": 1.9997278637411167e-05, + "loss": 0.8934, + "step": 197 + }, + { + "epoch": 0.04, + "learning_rate": 1.9997133530134244e-05, + "loss": 0.9311, + "step": 198 + }, + { + "epoch": 0.04, + "learning_rate": 1.9996984654566683e-05, + "loss": 0.8642, + "step": 199 + }, + { + "epoch": 0.04, + "learning_rate": 1.9996832010764612e-05, + "loss": 0.9164, + "step": 200 + }, + { + "epoch": 0.04, + "learning_rate": 1.999667559878556e-05, + "loss": 0.8892, + "step": 201 + }, + { + "epoch": 0.04, + "learning_rate": 1.9996515418688493e-05, + "loss": 0.9139, + "step": 202 + }, + { + "epoch": 0.04, + "learning_rate": 1.999635147053377e-05, + "loss": 0.902, + "step": 203 + }, + { + "epoch": 0.04, + "learning_rate": 1.9996183754383213e-05, + "loss": 0.2382, + "step": 204 + }, + { + "epoch": 0.04, + "learning_rate": 1.9996012270300024e-05, + "loss": 0.9386, + "step": 205 + }, + { + "epoch": 0.04, + "learning_rate": 1.9995837018348848e-05, + "loss": 0.9395, + "step": 206 + }, + { + "epoch": 0.04, + "learning_rate": 1.9995657998595742e-05, + "loss": 0.9216, + "step": 207 + }, + { + "epoch": 0.04, + "learning_rate": 1.9995475211108183e-05, + "loss": 0.8944, + "step": 208 + }, + { + "epoch": 0.04, + "learning_rate": 1.999528865595508e-05, + "loss": 0.8998, + "step": 209 + }, + { + "epoch": 0.04, + "learning_rate": 1.999509833320674e-05, + "loss": 0.8965, + "step": 210 + }, + { + "epoch": 0.04, + "learning_rate": 1.9994904242934912e-05, + "loss": 0.2422, + "step": 211 + }, + { + "epoch": 0.04, + "learning_rate": 1.9994706385212753e-05, + "loss": 0.9421, + "step": 212 + }, + { + "epoch": 0.04, + "learning_rate": 1.9994504760114838e-05, + "loss": 0.9057, + "step": 213 + }, + { + "epoch": 0.04, + "learning_rate": 1.9994299367717175e-05, + "loss": 0.9083, + "step": 214 + }, + { + "epoch": 0.04, + "learning_rate": 1.9994090208097176e-05, + "loss": 0.2369, + "step": 215 + }, + { + "epoch": 0.04, + "learning_rate": 1.9993877281333688e-05, + "loss": 0.2354, + "step": 216 + }, + { + "epoch": 0.04, + "learning_rate": 1.999366058750697e-05, + "loss": 0.8934, + "step": 217 + }, + { + "epoch": 0.04, + "learning_rate": 1.99934401266987e-05, + "loss": 0.2472, + "step": 218 + }, + { + "epoch": 0.04, + "learning_rate": 1.999321589899198e-05, + "loss": 0.9532, + "step": 219 + }, + { + "epoch": 0.04, + "learning_rate": 1.9992987904471323e-05, + "loss": 0.9261, + "step": 220 + }, + { + "epoch": 0.04, + "learning_rate": 1.9992756143222672e-05, + "loss": 0.8919, + "step": 221 + }, + { + "epoch": 0.04, + "learning_rate": 1.9992520615333393e-05, + "loss": 0.946, + "step": 222 + }, + { + "epoch": 0.04, + "learning_rate": 1.999228132089226e-05, + "loss": 0.8636, + "step": 223 + }, + { + "epoch": 0.04, + "learning_rate": 1.9992038259989467e-05, + "loss": 0.9065, + "step": 224 + }, + { + "epoch": 0.04, + "learning_rate": 1.9991791432716643e-05, + "loss": 0.8612, + "step": 225 + }, + { + "epoch": 0.04, + "learning_rate": 1.999154083916682e-05, + "loss": 0.9714, + "step": 226 + }, + { + "epoch": 0.04, + "learning_rate": 1.9991286479434456e-05, + "loss": 0.9308, + "step": 227 + }, + { + "epoch": 0.04, + "learning_rate": 1.999102835361543e-05, + "loss": 0.9744, + "step": 228 + }, + { + "epoch": 0.04, + "learning_rate": 1.9990766461807037e-05, + "loss": 0.8742, + "step": 229 + }, + { + "epoch": 0.04, + "learning_rate": 1.9990500804108e-05, + "loss": 0.9286, + "step": 230 + }, + { + "epoch": 0.04, + "learning_rate": 1.9990231380618445e-05, + "loss": 0.9477, + "step": 231 + }, + { + "epoch": 0.04, + "learning_rate": 1.998995819143994e-05, + "loss": 0.9251, + "step": 232 + }, + { + "epoch": 0.04, + "learning_rate": 1.998968123667545e-05, + "loss": 0.8745, + "step": 233 + }, + { + "epoch": 0.04, + "learning_rate": 1.9989400516429373e-05, + "loss": 0.9005, + "step": 234 + }, + { + "epoch": 0.04, + "learning_rate": 1.9989116030807525e-05, + "loss": 0.8619, + "step": 235 + }, + { + "epoch": 0.04, + "learning_rate": 1.9988827779917138e-05, + "loss": 0.877, + "step": 236 + }, + { + "epoch": 0.04, + "learning_rate": 1.998853576386686e-05, + "loss": 0.9086, + "step": 237 + }, + { + "epoch": 0.05, + "learning_rate": 1.998823998276677e-05, + "loss": 0.8877, + "step": 238 + }, + { + "epoch": 0.05, + "learning_rate": 1.9987940436728356e-05, + "loss": 0.8807, + "step": 239 + }, + { + "epoch": 0.05, + "learning_rate": 1.9987637125864525e-05, + "loss": 0.8852, + "step": 240 + }, + { + "epoch": 0.05, + "learning_rate": 1.998733005028961e-05, + "loss": 0.931, + "step": 241 + }, + { + "epoch": 0.05, + "learning_rate": 1.9987019210119352e-05, + "loss": 0.8869, + "step": 242 + }, + { + "epoch": 0.05, + "learning_rate": 1.9986704605470932e-05, + "loss": 0.8763, + "step": 243 + }, + { + "epoch": 0.05, + "learning_rate": 1.9986386236462926e-05, + "loss": 0.9042, + "step": 244 + }, + { + "epoch": 0.05, + "learning_rate": 1.998606410321534e-05, + "loss": 0.8457, + "step": 245 + }, + { + "epoch": 0.05, + "learning_rate": 1.9985738205849597e-05, + "loss": 0.9112, + "step": 246 + }, + { + "epoch": 0.05, + "learning_rate": 1.9985408544488542e-05, + "loss": 0.953, + "step": 247 + }, + { + "epoch": 0.05, + "learning_rate": 1.998507511925644e-05, + "loss": 0.9526, + "step": 248 + }, + { + "epoch": 0.05, + "learning_rate": 1.9984737930278966e-05, + "loss": 0.8448, + "step": 249 + }, + { + "epoch": 0.05, + "learning_rate": 1.9984396977683223e-05, + "loss": 0.8602, + "step": 250 + }, + { + "epoch": 0.05, + "learning_rate": 1.9984052261597723e-05, + "loss": 0.8995, + "step": 251 + }, + { + "epoch": 0.05, + "learning_rate": 1.9983703782152407e-05, + "loss": 0.9472, + "step": 252 + }, + { + "epoch": 0.05, + "learning_rate": 1.998335153947863e-05, + "loss": 0.9141, + "step": 253 + }, + { + "epoch": 0.05, + "learning_rate": 1.9982995533709163e-05, + "loss": 0.9679, + "step": 254 + }, + { + "epoch": 0.05, + "learning_rate": 1.99826357649782e-05, + "loss": 0.8865, + "step": 255 + }, + { + "epoch": 0.05, + "learning_rate": 1.998227223342135e-05, + "loss": 0.9271, + "step": 256 + }, + { + "epoch": 0.05, + "learning_rate": 1.998190493917564e-05, + "loss": 0.9467, + "step": 257 + }, + { + "epoch": 0.05, + "learning_rate": 1.998153388237952e-05, + "loss": 0.9368, + "step": 258 + }, + { + "epoch": 0.05, + "learning_rate": 1.998115906317285e-05, + "loss": 0.8756, + "step": 259 + }, + { + "epoch": 0.05, + "learning_rate": 1.9980780481696923e-05, + "loss": 0.8875, + "step": 260 + }, + { + "epoch": 0.05, + "learning_rate": 1.998039813809443e-05, + "loss": 0.9167, + "step": 261 + }, + { + "epoch": 0.05, + "learning_rate": 1.9980012032509495e-05, + "loss": 0.8492, + "step": 262 + }, + { + "epoch": 0.05, + "learning_rate": 1.9979622165087654e-05, + "loss": 0.8706, + "step": 263 + }, + { + "epoch": 0.05, + "learning_rate": 1.9979228535975866e-05, + "loss": 0.8958, + "step": 264 + }, + { + "epoch": 0.05, + "learning_rate": 1.9978831145322503e-05, + "loss": 0.9408, + "step": 265 + }, + { + "epoch": 0.05, + "learning_rate": 1.9978429993277354e-05, + "loss": 0.9215, + "step": 266 + }, + { + "epoch": 0.05, + "learning_rate": 1.9978025079991626e-05, + "loss": 0.8707, + "step": 267 + }, + { + "epoch": 0.05, + "learning_rate": 1.9977616405617953e-05, + "loss": 0.879, + "step": 268 + }, + { + "epoch": 0.05, + "learning_rate": 1.9977203970310374e-05, + "loss": 0.9364, + "step": 269 + }, + { + "epoch": 0.05, + "learning_rate": 1.9976787774224356e-05, + "loss": 0.9536, + "step": 270 + }, + { + "epoch": 0.05, + "learning_rate": 1.9976367817516773e-05, + "loss": 0.8716, + "step": 271 + }, + { + "epoch": 0.05, + "learning_rate": 1.9975944100345926e-05, + "loss": 0.9235, + "step": 272 + }, + { + "epoch": 0.05, + "learning_rate": 1.9975516622871528e-05, + "loss": 0.2291, + "step": 273 + }, + { + "epoch": 0.05, + "learning_rate": 1.9975085385254715e-05, + "loss": 0.8907, + "step": 274 + }, + { + "epoch": 0.05, + "learning_rate": 1.997465038765803e-05, + "loss": 0.8698, + "step": 275 + }, + { + "epoch": 0.05, + "learning_rate": 1.9974211630245445e-05, + "loss": 0.968, + "step": 276 + }, + { + "epoch": 0.05, + "learning_rate": 1.997376911318234e-05, + "loss": 0.8882, + "step": 277 + }, + { + "epoch": 0.05, + "learning_rate": 1.9973322836635517e-05, + "loss": 0.92, + "step": 278 + }, + { + "epoch": 0.05, + "learning_rate": 1.9972872800773198e-05, + "loss": 0.9547, + "step": 279 + }, + { + "epoch": 0.05, + "learning_rate": 1.9972419005765015e-05, + "loss": 0.8363, + "step": 280 + }, + { + "epoch": 0.05, + "learning_rate": 1.9971961451782022e-05, + "loss": 0.9322, + "step": 281 + }, + { + "epoch": 0.05, + "learning_rate": 1.9971500138996687e-05, + "loss": 0.8981, + "step": 282 + }, + { + "epoch": 0.05, + "learning_rate": 1.9971035067582893e-05, + "loss": 0.9118, + "step": 283 + }, + { + "epoch": 0.05, + "learning_rate": 1.9970566237715946e-05, + "loss": 0.8853, + "step": 284 + }, + { + "epoch": 0.05, + "learning_rate": 1.9970093649572567e-05, + "loss": 0.8785, + "step": 285 + }, + { + "epoch": 0.05, + "learning_rate": 1.9969617303330887e-05, + "loss": 0.9535, + "step": 286 + }, + { + "epoch": 0.05, + "learning_rate": 1.9969137199170463e-05, + "loss": 0.893, + "step": 287 + }, + { + "epoch": 0.05, + "learning_rate": 1.9968653337272262e-05, + "loss": 0.8693, + "step": 288 + }, + { + "epoch": 0.05, + "learning_rate": 1.9968165717818668e-05, + "loss": 0.8984, + "step": 289 + }, + { + "epoch": 0.05, + "learning_rate": 1.9967674340993487e-05, + "loss": 0.8544, + "step": 290 + }, + { + "epoch": 0.06, + "learning_rate": 1.9967179206981937e-05, + "loss": 0.9098, + "step": 291 + }, + { + "epoch": 0.06, + "learning_rate": 1.9966680315970647e-05, + "loss": 0.8842, + "step": 292 + }, + { + "epoch": 0.06, + "learning_rate": 1.9966177668147672e-05, + "loss": 0.9207, + "step": 293 + }, + { + "epoch": 0.06, + "learning_rate": 1.9965671263702476e-05, + "loss": 0.9163, + "step": 294 + }, + { + "epoch": 0.06, + "learning_rate": 1.9965161102825944e-05, + "loss": 0.8915, + "step": 295 + }, + { + "epoch": 0.06, + "learning_rate": 1.9964647185710375e-05, + "loss": 0.905, + "step": 296 + }, + { + "epoch": 0.06, + "learning_rate": 1.9964129512549482e-05, + "loss": 0.9058, + "step": 297 + }, + { + "epoch": 0.06, + "learning_rate": 1.9963608083538396e-05, + "loss": 0.879, + "step": 298 + }, + { + "epoch": 0.06, + "learning_rate": 1.996308289887366e-05, + "loss": 0.8899, + "step": 299 + }, + { + "epoch": 0.06, + "learning_rate": 1.9962553958753238e-05, + "loss": 0.8838, + "step": 300 + }, + { + "epoch": 0.06, + "learning_rate": 1.996202126337651e-05, + "loss": 0.9122, + "step": 301 + }, + { + "epoch": 0.06, + "learning_rate": 1.9961484812944264e-05, + "loss": 0.9012, + "step": 302 + }, + { + "epoch": 0.06, + "learning_rate": 1.9960944607658706e-05, + "loss": 0.83, + "step": 303 + }, + { + "epoch": 0.06, + "learning_rate": 1.9960400647723472e-05, + "loss": 0.9012, + "step": 304 + }, + { + "epoch": 0.06, + "learning_rate": 1.9959852933343586e-05, + "loss": 0.8943, + "step": 305 + }, + { + "epoch": 0.06, + "learning_rate": 1.9959301464725507e-05, + "loss": 0.9418, + "step": 306 + }, + { + "epoch": 0.06, + "learning_rate": 1.9958746242077105e-05, + "loss": 0.8796, + "step": 307 + }, + { + "epoch": 0.06, + "learning_rate": 1.9958187265607664e-05, + "loss": 0.2333, + "step": 308 + }, + { + "epoch": 0.06, + "learning_rate": 1.9957624535527883e-05, + "loss": 0.8794, + "step": 309 + }, + { + "epoch": 0.06, + "learning_rate": 1.9957058052049877e-05, + "loss": 0.9116, + "step": 310 + }, + { + "epoch": 0.06, + "learning_rate": 1.9956487815387172e-05, + "loss": 0.8639, + "step": 311 + }, + { + "epoch": 0.06, + "learning_rate": 1.9955913825754713e-05, + "loss": 0.9148, + "step": 312 + }, + { + "epoch": 0.06, + "learning_rate": 1.995533608336886e-05, + "loss": 0.8899, + "step": 313 + }, + { + "epoch": 0.06, + "learning_rate": 1.9954754588447376e-05, + "loss": 0.889, + "step": 314 + }, + { + "epoch": 0.06, + "learning_rate": 1.9954169341209464e-05, + "loss": 0.9048, + "step": 315 + }, + { + "epoch": 0.06, + "learning_rate": 1.995358034187571e-05, + "loss": 0.9271, + "step": 316 + }, + { + "epoch": 0.06, + "learning_rate": 1.995298759066814e-05, + "loss": 0.965, + "step": 317 + }, + { + "epoch": 0.06, + "learning_rate": 1.995239108781018e-05, + "loss": 0.8977, + "step": 318 + }, + { + "epoch": 0.06, + "learning_rate": 1.9951790833526675e-05, + "loss": 0.8709, + "step": 319 + }, + { + "epoch": 0.06, + "learning_rate": 1.995118682804388e-05, + "loss": 0.9083, + "step": 320 + }, + { + "epoch": 0.06, + "learning_rate": 1.9950579071589467e-05, + "loss": 0.8877, + "step": 321 + }, + { + "epoch": 0.06, + "learning_rate": 1.994996756439253e-05, + "loss": 0.9014, + "step": 322 + }, + { + "epoch": 0.06, + "learning_rate": 1.994935230668356e-05, + "loss": 0.8423, + "step": 323 + }, + { + "epoch": 0.06, + "learning_rate": 1.9948733298694474e-05, + "loss": 0.8939, + "step": 324 + }, + { + "epoch": 0.06, + "learning_rate": 1.99481105406586e-05, + "loss": 0.8844, + "step": 325 + }, + { + "epoch": 0.06, + "learning_rate": 1.9947484032810676e-05, + "loss": 0.8791, + "step": 326 + }, + { + "epoch": 0.06, + "learning_rate": 1.9946853775386857e-05, + "loss": 0.9231, + "step": 327 + }, + { + "epoch": 0.06, + "learning_rate": 1.9946219768624714e-05, + "loss": 0.9058, + "step": 328 + }, + { + "epoch": 0.06, + "learning_rate": 1.994558201276322e-05, + "loss": 0.8413, + "step": 329 + }, + { + "epoch": 0.06, + "learning_rate": 1.9944940508042775e-05, + "loss": 0.8977, + "step": 330 + }, + { + "epoch": 0.06, + "learning_rate": 1.9944295254705187e-05, + "loss": 0.9014, + "step": 331 + }, + { + "epoch": 0.06, + "learning_rate": 1.994364625299367e-05, + "loss": 0.249, + "step": 332 + }, + { + "epoch": 0.06, + "learning_rate": 1.994299350315286e-05, + "loss": 0.8895, + "step": 333 + }, + { + "epoch": 0.06, + "learning_rate": 1.9942337005428805e-05, + "loss": 0.9167, + "step": 334 + }, + { + "epoch": 0.06, + "learning_rate": 1.994167676006896e-05, + "loss": 0.8829, + "step": 335 + }, + { + "epoch": 0.06, + "learning_rate": 1.9941012767322197e-05, + "loss": 0.9035, + "step": 336 + }, + { + "epoch": 0.06, + "learning_rate": 1.9940345027438798e-05, + "loss": 0.9095, + "step": 337 + }, + { + "epoch": 0.06, + "learning_rate": 1.9939673540670462e-05, + "loss": 0.9022, + "step": 338 + }, + { + "epoch": 0.06, + "learning_rate": 1.9938998307270293e-05, + "loss": 0.8987, + "step": 339 + }, + { + "epoch": 0.06, + "learning_rate": 1.9938319327492818e-05, + "loss": 0.8727, + "step": 340 + }, + { + "epoch": 0.06, + "learning_rate": 1.9937636601593965e-05, + "loss": 0.9424, + "step": 341 + }, + { + "epoch": 0.06, + "learning_rate": 1.9936950129831078e-05, + "loss": 0.868, + "step": 342 + }, + { + "epoch": 0.07, + "learning_rate": 1.9936259912462917e-05, + "loss": 0.8893, + "step": 343 + }, + { + "epoch": 0.07, + "learning_rate": 1.9935565949749648e-05, + "loss": 0.8811, + "step": 344 + }, + { + "epoch": 0.07, + "learning_rate": 1.9934868241952854e-05, + "loss": 0.8759, + "step": 345 + }, + { + "epoch": 0.07, + "learning_rate": 1.9934166789335526e-05, + "loss": 0.8962, + "step": 346 + }, + { + "epoch": 0.07, + "learning_rate": 1.9933461592162063e-05, + "loss": 0.8889, + "step": 347 + }, + { + "epoch": 0.07, + "learning_rate": 1.9932752650698285e-05, + "loss": 0.8397, + "step": 348 + }, + { + "epoch": 0.07, + "learning_rate": 1.993203996521142e-05, + "loss": 0.9095, + "step": 349 + }, + { + "epoch": 0.07, + "learning_rate": 1.99313235359701e-05, + "loss": 0.8561, + "step": 350 + }, + { + "epoch": 0.07, + "learning_rate": 1.9930603363244374e-05, + "loss": 0.8165, + "step": 351 + }, + { + "epoch": 0.07, + "learning_rate": 1.9929879447305705e-05, + "loss": 0.8698, + "step": 352 + }, + { + "epoch": 0.07, + "learning_rate": 1.9929151788426966e-05, + "loss": 0.8606, + "step": 353 + }, + { + "epoch": 0.07, + "learning_rate": 1.9928420386882436e-05, + "loss": 0.8816, + "step": 354 + }, + { + "epoch": 0.07, + "learning_rate": 1.9927685242947804e-05, + "loss": 0.8926, + "step": 355 + }, + { + "epoch": 0.07, + "learning_rate": 1.9926946356900178e-05, + "loss": 0.883, + "step": 356 + }, + { + "epoch": 0.07, + "learning_rate": 1.9926203729018067e-05, + "loss": 0.8929, + "step": 357 + }, + { + "epoch": 0.07, + "learning_rate": 1.99254573595814e-05, + "loss": 0.8403, + "step": 358 + }, + { + "epoch": 0.07, + "learning_rate": 1.992470724887151e-05, + "loss": 0.8923, + "step": 359 + }, + { + "epoch": 0.07, + "learning_rate": 1.992395339717114e-05, + "loss": 0.8595, + "step": 360 + }, + { + "epoch": 0.07, + "learning_rate": 1.992319580476444e-05, + "loss": 0.8671, + "step": 361 + }, + { + "epoch": 0.07, + "learning_rate": 1.9922434471936987e-05, + "loss": 0.9019, + "step": 362 + }, + { + "epoch": 0.07, + "learning_rate": 1.9921669398975745e-05, + "loss": 0.9247, + "step": 363 + }, + { + "epoch": 0.07, + "learning_rate": 1.9920900586169105e-05, + "loss": 0.9639, + "step": 364 + }, + { + "epoch": 0.07, + "learning_rate": 1.9920128033806853e-05, + "loss": 0.9076, + "step": 365 + }, + { + "epoch": 0.07, + "learning_rate": 1.99193517421802e-05, + "loss": 0.8586, + "step": 366 + }, + { + "epoch": 0.07, + "learning_rate": 1.9918571711581757e-05, + "loss": 0.897, + "step": 367 + }, + { + "epoch": 0.07, + "learning_rate": 1.9917787942305548e-05, + "loss": 0.8846, + "step": 368 + }, + { + "epoch": 0.07, + "learning_rate": 1.9917000434647e-05, + "loss": 0.8991, + "step": 369 + }, + { + "epoch": 0.07, + "learning_rate": 1.991620918890296e-05, + "loss": 0.8651, + "step": 370 + }, + { + "epoch": 0.07, + "learning_rate": 1.9915414205371674e-05, + "loss": 0.8832, + "step": 371 + }, + { + "epoch": 0.07, + "learning_rate": 1.9914615484352797e-05, + "loss": 0.8682, + "step": 372 + }, + { + "epoch": 0.07, + "learning_rate": 1.9913813026147407e-05, + "loss": 0.8935, + "step": 373 + }, + { + "epoch": 0.07, + "learning_rate": 1.9913006831057967e-05, + "loss": 0.9128, + "step": 374 + }, + { + "epoch": 0.07, + "learning_rate": 1.9912196899388377e-05, + "loss": 0.9318, + "step": 375 + }, + { + "epoch": 0.07, + "learning_rate": 1.991138323144392e-05, + "loss": 0.8427, + "step": 376 + }, + { + "epoch": 0.07, + "learning_rate": 1.9910565827531297e-05, + "loss": 0.9044, + "step": 377 + }, + { + "epoch": 0.07, + "learning_rate": 1.9909744687958625e-05, + "loss": 0.905, + "step": 378 + }, + { + "epoch": 0.07, + "learning_rate": 1.9908919813035417e-05, + "loss": 0.8649, + "step": 379 + }, + { + "epoch": 0.07, + "learning_rate": 1.99080912030726e-05, + "loss": 0.8857, + "step": 380 + }, + { + "epoch": 0.07, + "learning_rate": 1.9907258858382508e-05, + "loss": 0.8415, + "step": 381 + }, + { + "epoch": 0.07, + "learning_rate": 1.9906422779278883e-05, + "loss": 0.9042, + "step": 382 + }, + { + "epoch": 0.07, + "learning_rate": 1.990558296607687e-05, + "loss": 0.8859, + "step": 383 + }, + { + "epoch": 0.07, + "learning_rate": 1.9904739419093033e-05, + "loss": 0.874, + "step": 384 + }, + { + "epoch": 0.07, + "learning_rate": 1.9903892138645333e-05, + "loss": 0.9013, + "step": 385 + }, + { + "epoch": 0.07, + "learning_rate": 1.990304112505314e-05, + "loss": 0.9317, + "step": 386 + }, + { + "epoch": 0.07, + "learning_rate": 1.9902186378637235e-05, + "loss": 0.9432, + "step": 387 + }, + { + "epoch": 0.07, + "learning_rate": 1.99013278997198e-05, + "loss": 0.8942, + "step": 388 + }, + { + "epoch": 0.07, + "learning_rate": 1.990046568862443e-05, + "loss": 0.849, + "step": 389 + }, + { + "epoch": 0.07, + "learning_rate": 1.9899599745676123e-05, + "loss": 0.2344, + "step": 390 + }, + { + "epoch": 0.07, + "learning_rate": 1.989873007120129e-05, + "loss": 0.9159, + "step": 391 + }, + { + "epoch": 0.07, + "learning_rate": 1.9897856665527737e-05, + "loss": 0.9466, + "step": 392 + }, + { + "epoch": 0.07, + "learning_rate": 1.9896979528984688e-05, + "loss": 0.8782, + "step": 393 + }, + { + "epoch": 0.07, + "learning_rate": 1.9896098661902762e-05, + "loss": 0.8363, + "step": 394 + }, + { + "epoch": 0.07, + "learning_rate": 1.9895214064614e-05, + "loss": 0.9433, + "step": 395 + }, + { + "epoch": 0.08, + "learning_rate": 1.9894325737451833e-05, + "loss": 0.8836, + "step": 396 + }, + { + "epoch": 0.08, + "learning_rate": 1.9893433680751105e-05, + "loss": 0.9352, + "step": 397 + }, + { + "epoch": 0.08, + "learning_rate": 1.9892537894848063e-05, + "loss": 0.9308, + "step": 398 + }, + { + "epoch": 0.08, + "learning_rate": 1.989163838008037e-05, + "loss": 0.893, + "step": 399 + }, + { + "epoch": 0.08, + "learning_rate": 1.9890735136787074e-05, + "loss": 0.8636, + "step": 400 + }, + { + "epoch": 0.08, + "learning_rate": 1.9889828165308658e-05, + "loss": 0.8595, + "step": 401 + }, + { + "epoch": 0.08, + "learning_rate": 1.988891746598698e-05, + "loss": 0.9362, + "step": 402 + }, + { + "epoch": 0.08, + "learning_rate": 1.9888003039165314e-05, + "loss": 0.8902, + "step": 403 + }, + { + "epoch": 0.08, + "learning_rate": 1.9887084885188354e-05, + "loss": 0.8886, + "step": 404 + }, + { + "epoch": 0.08, + "learning_rate": 1.9886163004402178e-05, + "loss": 0.9256, + "step": 405 + }, + { + "epoch": 0.08, + "learning_rate": 1.988523739715428e-05, + "loss": 0.8609, + "step": 406 + }, + { + "epoch": 0.08, + "learning_rate": 1.9884308063793552e-05, + "loss": 0.8808, + "step": 407 + }, + { + "epoch": 0.08, + "learning_rate": 1.9883375004670297e-05, + "loss": 0.8747, + "step": 408 + }, + { + "epoch": 0.08, + "learning_rate": 1.988243822013622e-05, + "loss": 0.8903, + "step": 409 + }, + { + "epoch": 0.08, + "learning_rate": 1.9881497710544428e-05, + "loss": 0.8563, + "step": 410 + }, + { + "epoch": 0.08, + "learning_rate": 1.9880553476249437e-05, + "loss": 0.9235, + "step": 411 + }, + { + "epoch": 0.08, + "learning_rate": 1.9879605517607162e-05, + "loss": 0.9339, + "step": 412 + }, + { + "epoch": 0.08, + "learning_rate": 1.987865383497492e-05, + "loss": 0.881, + "step": 413 + }, + { + "epoch": 0.08, + "learning_rate": 1.9877698428711444e-05, + "loss": 0.8653, + "step": 414 + }, + { + "epoch": 0.08, + "learning_rate": 1.9876739299176854e-05, + "loss": 0.8786, + "step": 415 + }, + { + "epoch": 0.08, + "learning_rate": 1.9875776446732686e-05, + "loss": 0.9182, + "step": 416 + }, + { + "epoch": 0.08, + "learning_rate": 1.9874809871741877e-05, + "loss": 0.93, + "step": 417 + }, + { + "epoch": 0.08, + "learning_rate": 1.9873839574568756e-05, + "loss": 0.8681, + "step": 418 + }, + { + "epoch": 0.08, + "learning_rate": 1.9872865555579073e-05, + "loss": 0.9136, + "step": 419 + }, + { + "epoch": 0.08, + "learning_rate": 1.987188781513997e-05, + "loss": 0.8519, + "step": 420 + }, + { + "epoch": 0.08, + "learning_rate": 1.987090635361999e-05, + "loss": 0.9235, + "step": 421 + }, + { + "epoch": 0.08, + "learning_rate": 1.9869921171389085e-05, + "loss": 0.9285, + "step": 422 + }, + { + "epoch": 0.08, + "learning_rate": 1.9868932268818608e-05, + "loss": 0.8856, + "step": 423 + }, + { + "epoch": 0.08, + "learning_rate": 1.986793964628131e-05, + "loss": 0.8669, + "step": 424 + }, + { + "epoch": 0.08, + "learning_rate": 1.9866943304151346e-05, + "loss": 0.831, + "step": 425 + }, + { + "epoch": 0.08, + "learning_rate": 1.986594324280428e-05, + "loss": 0.2533, + "step": 426 + }, + { + "epoch": 0.08, + "learning_rate": 1.9864939462617066e-05, + "loss": 0.8653, + "step": 427 + }, + { + "epoch": 0.08, + "learning_rate": 1.9863931963968074e-05, + "loss": 0.8679, + "step": 428 + }, + { + "epoch": 0.08, + "learning_rate": 1.986292074723706e-05, + "loss": 0.8908, + "step": 429 + }, + { + "epoch": 0.08, + "learning_rate": 1.9861905812805197e-05, + "loss": 0.9127, + "step": 430 + }, + { + "epoch": 0.08, + "learning_rate": 1.986088716105504e-05, + "loss": 0.8684, + "step": 431 + }, + { + "epoch": 0.08, + "learning_rate": 1.9859864792370565e-05, + "loss": 0.8706, + "step": 432 + }, + { + "epoch": 0.08, + "learning_rate": 1.985883870713714e-05, + "loss": 0.8333, + "step": 433 + }, + { + "epoch": 0.08, + "learning_rate": 1.9857808905741533e-05, + "loss": 0.946, + "step": 434 + }, + { + "epoch": 0.08, + "learning_rate": 1.9856775388571914e-05, + "loss": 0.8954, + "step": 435 + }, + { + "epoch": 0.08, + "learning_rate": 1.9855738156017856e-05, + "loss": 0.8913, + "step": 436 + }, + { + "epoch": 0.08, + "learning_rate": 1.9854697208470326e-05, + "loss": 0.907, + "step": 437 + }, + { + "epoch": 0.08, + "learning_rate": 1.98536525463217e-05, + "loss": 0.2395, + "step": 438 + }, + { + "epoch": 0.08, + "learning_rate": 1.985260416996575e-05, + "loss": 0.8551, + "step": 439 + }, + { + "epoch": 0.08, + "learning_rate": 1.985155207979765e-05, + "loss": 0.8346, + "step": 440 + }, + { + "epoch": 0.08, + "learning_rate": 1.9850496276213963e-05, + "loss": 0.9049, + "step": 441 + }, + { + "epoch": 0.08, + "learning_rate": 1.9849436759612668e-05, + "loss": 0.8949, + "step": 442 + }, + { + "epoch": 0.08, + "learning_rate": 1.9848373530393133e-05, + "loss": 0.9195, + "step": 443 + }, + { + "epoch": 0.08, + "learning_rate": 1.9847306588956132e-05, + "loss": 0.2652, + "step": 444 + }, + { + "epoch": 0.08, + "learning_rate": 1.9846235935703835e-05, + "loss": 0.9439, + "step": 445 + }, + { + "epoch": 0.08, + "learning_rate": 1.9845161571039805e-05, + "loss": 0.88, + "step": 446 + }, + { + "epoch": 0.08, + "learning_rate": 1.9844083495369017e-05, + "loss": 0.9122, + "step": 447 + }, + { + "epoch": 0.08, + "learning_rate": 1.9843001709097832e-05, + "loss": 0.8865, + "step": 448 + }, + { + "epoch": 0.09, + "learning_rate": 1.9841916212634012e-05, + "loss": 0.8949, + "step": 449 + }, + { + "epoch": 0.09, + "learning_rate": 1.9840827006386732e-05, + "loss": 0.8688, + "step": 450 + }, + { + "epoch": 0.09, + "learning_rate": 1.9839734090766552e-05, + "loss": 0.9132, + "step": 451 + }, + { + "epoch": 0.09, + "learning_rate": 1.9838637466185428e-05, + "loss": 0.8889, + "step": 452 + }, + { + "epoch": 0.09, + "learning_rate": 1.983753713305672e-05, + "loss": 0.8761, + "step": 453 + }, + { + "epoch": 0.09, + "learning_rate": 1.983643309179518e-05, + "loss": 0.8116, + "step": 454 + }, + { + "epoch": 0.09, + "learning_rate": 1.9835325342816966e-05, + "loss": 0.8974, + "step": 455 + }, + { + "epoch": 0.09, + "learning_rate": 1.9834213886539634e-05, + "loss": 0.9075, + "step": 456 + }, + { + "epoch": 0.09, + "learning_rate": 1.983309872338213e-05, + "loss": 0.8454, + "step": 457 + }, + { + "epoch": 0.09, + "learning_rate": 1.9831979853764794e-05, + "loss": 0.8234, + "step": 458 + }, + { + "epoch": 0.09, + "learning_rate": 1.983085727810938e-05, + "loss": 0.8692, + "step": 459 + }, + { + "epoch": 0.09, + "learning_rate": 1.982973099683902e-05, + "loss": 0.8824, + "step": 460 + }, + { + "epoch": 0.09, + "learning_rate": 1.9828601010378256e-05, + "loss": 0.8423, + "step": 461 + }, + { + "epoch": 0.09, + "learning_rate": 1.9827467319153018e-05, + "loss": 0.8866, + "step": 462 + }, + { + "epoch": 0.09, + "learning_rate": 1.9826329923590644e-05, + "loss": 0.8961, + "step": 463 + }, + { + "epoch": 0.09, + "learning_rate": 1.982518882411985e-05, + "loss": 0.8807, + "step": 464 + }, + { + "epoch": 0.09, + "learning_rate": 1.982404402117077e-05, + "loss": 0.8898, + "step": 465 + }, + { + "epoch": 0.09, + "learning_rate": 1.982289551517491e-05, + "loss": 0.8597, + "step": 466 + }, + { + "epoch": 0.09, + "learning_rate": 1.98217433065652e-05, + "loss": 0.9195, + "step": 467 + }, + { + "epoch": 0.09, + "learning_rate": 1.9820587395775934e-05, + "loss": 0.8685, + "step": 468 + }, + { + "epoch": 0.09, + "learning_rate": 1.9819427783242827e-05, + "loss": 0.2376, + "step": 469 + }, + { + "epoch": 0.09, + "learning_rate": 1.9818264469402977e-05, + "loss": 0.8751, + "step": 470 + }, + { + "epoch": 0.09, + "learning_rate": 1.9817097454694884e-05, + "loss": 0.7916, + "step": 471 + }, + { + "epoch": 0.09, + "learning_rate": 1.9815926739558434e-05, + "loss": 0.9139, + "step": 472 + }, + { + "epoch": 0.09, + "learning_rate": 1.9814752324434917e-05, + "loss": 0.8772, + "step": 473 + }, + { + "epoch": 0.09, + "learning_rate": 1.9813574209767013e-05, + "loss": 0.8434, + "step": 474 + }, + { + "epoch": 0.09, + "learning_rate": 1.9812392395998797e-05, + "loss": 0.9324, + "step": 475 + }, + { + "epoch": 0.09, + "learning_rate": 1.9811206883575735e-05, + "loss": 0.871, + "step": 476 + }, + { + "epoch": 0.09, + "learning_rate": 1.9810017672944693e-05, + "loss": 0.8562, + "step": 477 + }, + { + "epoch": 0.09, + "learning_rate": 1.9808824764553932e-05, + "loss": 0.8636, + "step": 478 + }, + { + "epoch": 0.09, + "learning_rate": 1.9807628158853097e-05, + "loss": 0.8961, + "step": 479 + }, + { + "epoch": 0.09, + "learning_rate": 1.980642785629324e-05, + "loss": 0.8603, + "step": 480 + }, + { + "epoch": 0.09, + "learning_rate": 1.9805223857326794e-05, + "loss": 0.8702, + "step": 481 + }, + { + "epoch": 0.09, + "learning_rate": 1.9804016162407593e-05, + "loss": 0.9001, + "step": 482 + }, + { + "epoch": 0.09, + "learning_rate": 1.9802804771990863e-05, + "loss": 0.9024, + "step": 483 + }, + { + "epoch": 0.09, + "learning_rate": 1.980158968653322e-05, + "loss": 0.8559, + "step": 484 + }, + { + "epoch": 0.09, + "learning_rate": 1.9800370906492677e-05, + "loss": 0.9144, + "step": 485 + }, + { + "epoch": 0.09, + "learning_rate": 1.979914843232863e-05, + "loss": 0.897, + "step": 486 + }, + { + "epoch": 0.09, + "learning_rate": 1.979792226450189e-05, + "loss": 0.8815, + "step": 487 + }, + { + "epoch": 0.09, + "learning_rate": 1.9796692403474632e-05, + "loss": 0.8394, + "step": 488 + }, + { + "epoch": 0.09, + "learning_rate": 1.9795458849710444e-05, + "loss": 0.8562, + "step": 489 + }, + { + "epoch": 0.09, + "learning_rate": 1.979422160367429e-05, + "loss": 0.8745, + "step": 490 + }, + { + "epoch": 0.09, + "learning_rate": 1.9792980665832542e-05, + "loss": 0.902, + "step": 491 + }, + { + "epoch": 0.09, + "learning_rate": 1.979173603665295e-05, + "loss": 0.8788, + "step": 492 + }, + { + "epoch": 0.09, + "learning_rate": 1.9790487716604663e-05, + "loss": 0.2448, + "step": 493 + }, + { + "epoch": 0.09, + "learning_rate": 1.978923570615822e-05, + "loss": 0.8183, + "step": 494 + }, + { + "epoch": 0.09, + "learning_rate": 1.9787980005785553e-05, + "loss": 0.8824, + "step": 495 + }, + { + "epoch": 0.09, + "learning_rate": 1.9786720615959974e-05, + "loss": 0.825, + "step": 496 + }, + { + "epoch": 0.09, + "learning_rate": 1.9785457537156197e-05, + "loss": 0.8701, + "step": 497 + }, + { + "epoch": 0.09, + "learning_rate": 1.9784190769850328e-05, + "loss": 0.8688, + "step": 498 + }, + { + "epoch": 0.09, + "learning_rate": 1.978292031451986e-05, + "loss": 0.906, + "step": 499 + }, + { + "epoch": 0.09, + "learning_rate": 1.978164617164366e-05, + "loss": 0.919, + "step": 500 + }, + { + "epoch": 0.09, + "learning_rate": 1.9780368341702016e-05, + "loss": 0.8769, + "step": 501 + }, + { + "epoch": 0.1, + "learning_rate": 1.977908682517658e-05, + "loss": 0.902, + "step": 502 + }, + { + "epoch": 0.1, + "learning_rate": 1.977780162255041e-05, + "loss": 0.2461, + "step": 503 + }, + { + "epoch": 0.1, + "learning_rate": 1.977651273430794e-05, + "loss": 0.9234, + "step": 504 + }, + { + "epoch": 0.1, + "learning_rate": 1.9775220160935004e-05, + "loss": 0.8736, + "step": 505 + }, + { + "epoch": 0.1, + "learning_rate": 1.977392390291882e-05, + "loss": 0.872, + "step": 506 + }, + { + "epoch": 0.1, + "learning_rate": 1.9772623960747995e-05, + "loss": 0.8566, + "step": 507 + }, + { + "epoch": 0.1, + "learning_rate": 1.9771320334912524e-05, + "loss": 0.8911, + "step": 508 + }, + { + "epoch": 0.1, + "learning_rate": 1.9770013025903797e-05, + "loss": 0.8725, + "step": 509 + }, + { + "epoch": 0.1, + "learning_rate": 1.976870203421458e-05, + "loss": 0.8997, + "step": 510 + }, + { + "epoch": 0.1, + "learning_rate": 1.976738736033904e-05, + "loss": 0.8944, + "step": 511 + }, + { + "epoch": 0.1, + "learning_rate": 1.9766069004772728e-05, + "loss": 0.8546, + "step": 512 + }, + { + "epoch": 0.1, + "learning_rate": 1.9764746968012572e-05, + "loss": 0.2618, + "step": 513 + }, + { + "epoch": 0.1, + "learning_rate": 1.9763421250556908e-05, + "loss": 0.8785, + "step": 514 + }, + { + "epoch": 0.1, + "learning_rate": 1.9762091852905444e-05, + "loss": 0.8703, + "step": 515 + }, + { + "epoch": 0.1, + "learning_rate": 1.9760758775559275e-05, + "loss": 0.9521, + "step": 516 + }, + { + "epoch": 0.1, + "learning_rate": 1.975942201902089e-05, + "loss": 0.815, + "step": 517 + }, + { + "epoch": 0.1, + "learning_rate": 1.9758081583794163e-05, + "loss": 0.8013, + "step": 518 + }, + { + "epoch": 0.1, + "learning_rate": 1.9756737470384353e-05, + "loss": 0.8713, + "step": 519 + }, + { + "epoch": 0.1, + "learning_rate": 1.9755389679298107e-05, + "loss": 0.8337, + "step": 520 + }, + { + "epoch": 0.1, + "learning_rate": 1.9754038211043454e-05, + "loss": 0.8725, + "step": 521 + }, + { + "epoch": 0.1, + "learning_rate": 1.9752683066129822e-05, + "loss": 0.8549, + "step": 522 + }, + { + "epoch": 0.1, + "learning_rate": 1.9751324245068008e-05, + "loss": 0.8676, + "step": 523 + }, + { + "epoch": 0.1, + "learning_rate": 1.97499617483702e-05, + "loss": 0.9005, + "step": 524 + }, + { + "epoch": 0.1, + "learning_rate": 1.974859557654998e-05, + "loss": 0.8167, + "step": 525 + }, + { + "epoch": 0.1, + "learning_rate": 1.97472257301223e-05, + "loss": 0.8779, + "step": 526 + }, + { + "epoch": 0.1, + "learning_rate": 1.9745852209603516e-05, + "loss": 0.9008, + "step": 527 + }, + { + "epoch": 0.1, + "learning_rate": 1.9744475015511354e-05, + "loss": 0.909, + "step": 528 + }, + { + "epoch": 0.1, + "learning_rate": 1.9743094148364927e-05, + "loss": 0.8335, + "step": 529 + }, + { + "epoch": 0.1, + "learning_rate": 1.974170960868474e-05, + "loss": 0.8159, + "step": 530 + }, + { + "epoch": 0.1, + "learning_rate": 1.9740321396992676e-05, + "loss": 0.8825, + "step": 531 + }, + { + "epoch": 0.1, + "learning_rate": 1.9738929513812003e-05, + "loss": 0.9005, + "step": 532 + }, + { + "epoch": 0.1, + "learning_rate": 1.9737533959667373e-05, + "loss": 0.8806, + "step": 533 + }, + { + "epoch": 0.1, + "learning_rate": 1.9736134735084823e-05, + "loss": 0.8442, + "step": 534 + }, + { + "epoch": 0.1, + "learning_rate": 1.973473184059177e-05, + "loss": 0.9381, + "step": 535 + }, + { + "epoch": 0.1, + "learning_rate": 1.9733325276717022e-05, + "loss": 0.8944, + "step": 536 + }, + { + "epoch": 0.1, + "learning_rate": 1.973191504399076e-05, + "loss": 0.9162, + "step": 537 + }, + { + "epoch": 0.1, + "learning_rate": 1.9730501142944555e-05, + "loss": 0.891, + "step": 538 + }, + { + "epoch": 0.1, + "learning_rate": 1.972908357411136e-05, + "loss": 0.8907, + "step": 539 + }, + { + "epoch": 0.1, + "learning_rate": 1.972766233802551e-05, + "loss": 0.8427, + "step": 540 + }, + { + "epoch": 0.1, + "learning_rate": 1.972623743522272e-05, + "loss": 0.8849, + "step": 541 + }, + { + "epoch": 0.1, + "learning_rate": 1.9724808866240083e-05, + "loss": 0.9463, + "step": 542 + }, + { + "epoch": 0.1, + "learning_rate": 1.972337663161609e-05, + "loss": 0.8781, + "step": 543 + }, + { + "epoch": 0.1, + "learning_rate": 1.97219407318906e-05, + "loss": 0.8998, + "step": 544 + }, + { + "epoch": 0.1, + "learning_rate": 1.9720501167604854e-05, + "loss": 0.8728, + "step": 545 + }, + { + "epoch": 0.1, + "learning_rate": 1.9719057939301477e-05, + "loss": 0.8895, + "step": 546 + }, + { + "epoch": 0.1, + "learning_rate": 1.971761104752448e-05, + "loss": 0.9416, + "step": 547 + }, + { + "epoch": 0.1, + "learning_rate": 1.9716160492819246e-05, + "loss": 0.873, + "step": 548 + }, + { + "epoch": 0.1, + "learning_rate": 1.9714706275732544e-05, + "loss": 0.2789, + "step": 549 + }, + { + "epoch": 0.1, + "learning_rate": 1.9713248396812524e-05, + "loss": 0.8729, + "step": 550 + }, + { + "epoch": 0.1, + "learning_rate": 1.9711786856608714e-05, + "loss": 0.8548, + "step": 551 + }, + { + "epoch": 0.1, + "learning_rate": 1.971032165567202e-05, + "loss": 0.9017, + "step": 552 + }, + { + "epoch": 0.1, + "learning_rate": 1.9708852794554735e-05, + "loss": 0.278, + "step": 553 + }, + { + "epoch": 0.11, + "learning_rate": 1.9707380273810524e-05, + "loss": 0.8584, + "step": 554 + }, + { + "epoch": 0.11, + "learning_rate": 1.9705904093994434e-05, + "loss": 0.8371, + "step": 555 + }, + { + "epoch": 0.11, + "learning_rate": 1.97044242556629e-05, + "loss": 0.2667, + "step": 556 + }, + { + "epoch": 0.11, + "learning_rate": 1.970294075937371e-05, + "loss": 0.9182, + "step": 557 + }, + { + "epoch": 0.11, + "learning_rate": 1.970145360568607e-05, + "loss": 0.8941, + "step": 558 + }, + { + "epoch": 0.11, + "learning_rate": 1.969996279516053e-05, + "loss": 0.8906, + "step": 559 + }, + { + "epoch": 0.11, + "learning_rate": 1.9698468328359038e-05, + "loss": 0.8919, + "step": 560 + }, + { + "epoch": 0.11, + "learning_rate": 1.969697020584491e-05, + "loss": 0.8477, + "step": 561 + }, + { + "epoch": 0.11, + "learning_rate": 1.9695468428182843e-05, + "loss": 0.8794, + "step": 562 + }, + { + "epoch": 0.11, + "learning_rate": 1.969396299593892e-05, + "loss": 0.9033, + "step": 563 + }, + { + "epoch": 0.11, + "learning_rate": 1.9692453909680588e-05, + "loss": 0.9106, + "step": 564 + }, + { + "epoch": 0.11, + "learning_rate": 1.969094116997668e-05, + "loss": 0.247, + "step": 565 + }, + { + "epoch": 0.11, + "learning_rate": 1.96894247773974e-05, + "loss": 0.828, + "step": 566 + }, + { + "epoch": 0.11, + "learning_rate": 1.968790473251434e-05, + "loss": 0.867, + "step": 567 + }, + { + "epoch": 0.11, + "learning_rate": 1.9686381035900454e-05, + "loss": 0.8748, + "step": 568 + }, + { + "epoch": 0.11, + "learning_rate": 1.9684853688130085e-05, + "loss": 0.8759, + "step": 569 + }, + { + "epoch": 0.11, + "learning_rate": 1.9683322689778945e-05, + "loss": 0.8662, + "step": 570 + }, + { + "epoch": 0.11, + "learning_rate": 1.968178804142412e-05, + "loss": 0.8834, + "step": 571 + }, + { + "epoch": 0.11, + "learning_rate": 1.968024974364408e-05, + "loss": 0.8891, + "step": 572 + }, + { + "epoch": 0.11, + "learning_rate": 1.967870779701867e-05, + "loss": 0.8062, + "step": 573 + }, + { + "epoch": 0.11, + "learning_rate": 1.9677162202129107e-05, + "loss": 0.8524, + "step": 574 + }, + { + "epoch": 0.11, + "learning_rate": 1.9675612959557972e-05, + "loss": 0.8508, + "step": 575 + }, + { + "epoch": 0.11, + "learning_rate": 1.9674060069889243e-05, + "loss": 0.8355, + "step": 576 + }, + { + "epoch": 0.11, + "learning_rate": 1.9672503533708257e-05, + "loss": 0.93, + "step": 577 + }, + { + "epoch": 0.11, + "learning_rate": 1.967094335160173e-05, + "loss": 0.927, + "step": 578 + }, + { + "epoch": 0.11, + "learning_rate": 1.9669379524157755e-05, + "loss": 0.8517, + "step": 579 + }, + { + "epoch": 0.11, + "learning_rate": 1.9667812051965794e-05, + "loss": 0.8967, + "step": 580 + }, + { + "epoch": 0.11, + "learning_rate": 1.9666240935616687e-05, + "loss": 0.8161, + "step": 581 + }, + { + "epoch": 0.11, + "learning_rate": 1.966466617570265e-05, + "loss": 0.8958, + "step": 582 + }, + { + "epoch": 0.11, + "learning_rate": 1.966308777281726e-05, + "loss": 0.8357, + "step": 583 + }, + { + "epoch": 0.11, + "learning_rate": 1.9661505727555482e-05, + "loss": 0.916, + "step": 584 + }, + { + "epoch": 0.11, + "learning_rate": 1.9659920040513646e-05, + "loss": 0.8822, + "step": 585 + }, + { + "epoch": 0.11, + "learning_rate": 1.9658330712289456e-05, + "loss": 0.8359, + "step": 586 + }, + { + "epoch": 0.11, + "learning_rate": 1.965673774348199e-05, + "loss": 0.8416, + "step": 587 + }, + { + "epoch": 0.11, + "learning_rate": 1.9655141134691693e-05, + "loss": 0.9542, + "step": 588 + }, + { + "epoch": 0.11, + "learning_rate": 1.9653540886520387e-05, + "loss": 0.7987, + "step": 589 + }, + { + "epoch": 0.11, + "learning_rate": 1.965193699957127e-05, + "loss": 0.8967, + "step": 590 + }, + { + "epoch": 0.11, + "learning_rate": 1.9650329474448903e-05, + "loss": 0.9093, + "step": 591 + }, + { + "epoch": 0.11, + "learning_rate": 1.9648718311759224e-05, + "loss": 0.8614, + "step": 592 + }, + { + "epoch": 0.11, + "learning_rate": 1.9647103512109535e-05, + "loss": 0.876, + "step": 593 + }, + { + "epoch": 0.11, + "learning_rate": 1.964548507610852e-05, + "loss": 0.8621, + "step": 594 + }, + { + "epoch": 0.11, + "learning_rate": 1.964386300436622e-05, + "loss": 0.8144, + "step": 595 + }, + { + "epoch": 0.11, + "learning_rate": 1.9642237297494066e-05, + "loss": 0.2655, + "step": 596 + }, + { + "epoch": 0.11, + "learning_rate": 1.9640607956104833e-05, + "loss": 0.265, + "step": 597 + }, + { + "epoch": 0.11, + "learning_rate": 1.963897498081269e-05, + "loss": 0.8669, + "step": 598 + }, + { + "epoch": 0.11, + "learning_rate": 1.9637338372233165e-05, + "loss": 0.2357, + "step": 599 + }, + { + "epoch": 0.11, + "learning_rate": 1.9635698130983153e-05, + "loss": 0.8839, + "step": 600 + }, + { + "epoch": 0.11, + "learning_rate": 1.9634054257680924e-05, + "loss": 0.903, + "step": 601 + }, + { + "epoch": 0.11, + "learning_rate": 1.9632406752946115e-05, + "loss": 0.8583, + "step": 602 + }, + { + "epoch": 0.11, + "learning_rate": 1.963075561739973e-05, + "loss": 0.8873, + "step": 603 + }, + { + "epoch": 0.11, + "learning_rate": 1.9629100851664142e-05, + "loss": 0.8321, + "step": 604 + }, + { + "epoch": 0.11, + "learning_rate": 1.96274424563631e-05, + "loss": 0.8978, + "step": 605 + }, + { + "epoch": 0.11, + "learning_rate": 1.9625780432121708e-05, + "loss": 0.8633, + "step": 606 + }, + { + "epoch": 0.12, + "learning_rate": 1.962411477956645e-05, + "loss": 0.8791, + "step": 607 + }, + { + "epoch": 0.12, + "learning_rate": 1.9622445499325167e-05, + "loss": 0.8803, + "step": 608 + }, + { + "epoch": 0.12, + "learning_rate": 1.9620772592027075e-05, + "loss": 0.884, + "step": 609 + }, + { + "epoch": 0.12, + "learning_rate": 1.9619096058302757e-05, + "loss": 0.8372, + "step": 610 + }, + { + "epoch": 0.12, + "learning_rate": 1.9617415898784157e-05, + "loss": 0.8413, + "step": 611 + }, + { + "epoch": 0.12, + "learning_rate": 1.9615732114104594e-05, + "loss": 0.899, + "step": 612 + }, + { + "epoch": 0.12, + "learning_rate": 1.961404470489874e-05, + "loss": 0.9453, + "step": 613 + }, + { + "epoch": 0.12, + "learning_rate": 1.9612353671802658e-05, + "loss": 0.9124, + "step": 614 + }, + { + "epoch": 0.12, + "learning_rate": 1.9610659015453746e-05, + "loss": 0.8878, + "step": 615 + }, + { + "epoch": 0.12, + "learning_rate": 1.960896073649079e-05, + "loss": 0.9053, + "step": 616 + }, + { + "epoch": 0.12, + "learning_rate": 1.960725883555393e-05, + "loss": 0.8677, + "step": 617 + }, + { + "epoch": 0.12, + "learning_rate": 1.9605553313284682e-05, + "loss": 0.8727, + "step": 618 + }, + { + "epoch": 0.12, + "learning_rate": 1.9603844170325916e-05, + "loss": 0.8649, + "step": 619 + }, + { + "epoch": 0.12, + "learning_rate": 1.960213140732187e-05, + "loss": 0.8324, + "step": 620 + }, + { + "epoch": 0.12, + "learning_rate": 1.960041502491815e-05, + "loss": 0.8845, + "step": 621 + }, + { + "epoch": 0.12, + "learning_rate": 1.9598695023761726e-05, + "loss": 0.9044, + "step": 622 + }, + { + "epoch": 0.12, + "learning_rate": 1.9596971404500933e-05, + "loss": 0.8492, + "step": 623 + }, + { + "epoch": 0.12, + "learning_rate": 1.9595244167785455e-05, + "loss": 0.8993, + "step": 624 + }, + { + "epoch": 0.12, + "learning_rate": 1.9593513314266368e-05, + "loss": 0.8371, + "step": 625 + }, + { + "epoch": 0.12, + "learning_rate": 1.9591778844596082e-05, + "loss": 0.815, + "step": 626 + }, + { + "epoch": 0.12, + "learning_rate": 1.9590040759428385e-05, + "loss": 0.2934, + "step": 627 + }, + { + "epoch": 0.12, + "learning_rate": 1.9588299059418434e-05, + "loss": 0.946, + "step": 628 + }, + { + "epoch": 0.12, + "learning_rate": 1.9586553745222732e-05, + "loss": 0.8613, + "step": 629 + }, + { + "epoch": 0.12, + "learning_rate": 1.9584804817499155e-05, + "loss": 0.9307, + "step": 630 + }, + { + "epoch": 0.12, + "learning_rate": 1.9583052276906937e-05, + "loss": 0.867, + "step": 631 + }, + { + "epoch": 0.12, + "learning_rate": 1.9581296124106682e-05, + "loss": 0.8696, + "step": 632 + }, + { + "epoch": 0.12, + "learning_rate": 1.9579536359760344e-05, + "loss": 0.2896, + "step": 633 + }, + { + "epoch": 0.12, + "learning_rate": 1.9577772984531244e-05, + "loss": 0.8861, + "step": 634 + }, + { + "epoch": 0.12, + "learning_rate": 1.957600599908406e-05, + "loss": 0.8189, + "step": 635 + }, + { + "epoch": 0.12, + "learning_rate": 1.9574235404084842e-05, + "loss": 0.8756, + "step": 636 + }, + { + "epoch": 0.12, + "learning_rate": 1.957246120020099e-05, + "loss": 0.8544, + "step": 637 + }, + { + "epoch": 0.12, + "learning_rate": 1.9570683388101263e-05, + "loss": 0.8423, + "step": 638 + }, + { + "epoch": 0.12, + "learning_rate": 1.9568901968455792e-05, + "loss": 0.8568, + "step": 639 + }, + { + "epoch": 0.12, + "learning_rate": 1.956711694193605e-05, + "loss": 0.8601, + "step": 640 + }, + { + "epoch": 0.12, + "learning_rate": 1.9565328309214894e-05, + "loss": 0.9459, + "step": 641 + }, + { + "epoch": 0.12, + "learning_rate": 1.9563536070966513e-05, + "loss": 0.8617, + "step": 642 + }, + { + "epoch": 0.12, + "learning_rate": 1.956174022786647e-05, + "loss": 0.9361, + "step": 643 + }, + { + "epoch": 0.12, + "learning_rate": 1.9559940780591695e-05, + "loss": 0.8362, + "step": 644 + }, + { + "epoch": 0.12, + "learning_rate": 1.9558137729820456e-05, + "loss": 0.85, + "step": 645 + }, + { + "epoch": 0.12, + "learning_rate": 1.9556331076232392e-05, + "loss": 0.9018, + "step": 646 + }, + { + "epoch": 0.12, + "learning_rate": 1.95545208205085e-05, + "loss": 0.8211, + "step": 647 + }, + { + "epoch": 0.12, + "learning_rate": 1.9552706963331133e-05, + "loss": 0.8874, + "step": 648 + }, + { + "epoch": 0.12, + "learning_rate": 1.9550889505383996e-05, + "loss": 0.904, + "step": 649 + }, + { + "epoch": 0.12, + "learning_rate": 1.9549068447352165e-05, + "loss": 0.2563, + "step": 650 + }, + { + "epoch": 0.12, + "learning_rate": 1.9547243789922055e-05, + "loss": 0.8826, + "step": 651 + }, + { + "epoch": 0.12, + "learning_rate": 1.9545415533781453e-05, + "loss": 0.7974, + "step": 652 + }, + { + "epoch": 0.12, + "learning_rate": 1.9543583679619495e-05, + "loss": 0.8815, + "step": 653 + }, + { + "epoch": 0.12, + "learning_rate": 1.9541748228126672e-05, + "loss": 0.8611, + "step": 654 + }, + { + "epoch": 0.12, + "learning_rate": 1.9539909179994837e-05, + "loss": 0.8837, + "step": 655 + }, + { + "epoch": 0.12, + "learning_rate": 1.9538066535917196e-05, + "loss": 0.9113, + "step": 656 + }, + { + "epoch": 0.12, + "learning_rate": 1.9536220296588315e-05, + "loss": 0.7696, + "step": 657 + }, + { + "epoch": 0.12, + "learning_rate": 1.9534370462704096e-05, + "loss": 0.8894, + "step": 658 + }, + { + "epoch": 0.12, + "learning_rate": 1.953251703496182e-05, + "loss": 0.886, + "step": 659 + }, + { + "epoch": 0.13, + "learning_rate": 1.9530660014060113e-05, + "loss": 0.7786, + "step": 660 + }, + { + "epoch": 0.13, + "learning_rate": 1.9528799400698947e-05, + "loss": 0.8605, + "step": 661 + }, + { + "epoch": 0.13, + "learning_rate": 1.952693519557967e-05, + "loss": 0.84, + "step": 662 + }, + { + "epoch": 0.13, + "learning_rate": 1.952506739940496e-05, + "loss": 0.9031, + "step": 663 + }, + { + "epoch": 0.13, + "learning_rate": 1.9523196012878857e-05, + "loss": 0.2396, + "step": 664 + }, + { + "epoch": 0.13, + "learning_rate": 1.9521321036706765e-05, + "loss": 0.8826, + "step": 665 + }, + { + "epoch": 0.13, + "learning_rate": 1.9519442471595422e-05, + "loss": 0.8918, + "step": 666 + }, + { + "epoch": 0.13, + "learning_rate": 1.9517560318252936e-05, + "loss": 0.869, + "step": 667 + }, + { + "epoch": 0.13, + "learning_rate": 1.951567457738876e-05, + "loss": 0.8482, + "step": 668 + }, + { + "epoch": 0.13, + "learning_rate": 1.9513785249713697e-05, + "loss": 0.8485, + "step": 669 + }, + { + "epoch": 0.13, + "learning_rate": 1.9511892335939904e-05, + "loss": 0.8134, + "step": 670 + }, + { + "epoch": 0.13, + "learning_rate": 1.9509995836780895e-05, + "loss": 0.9638, + "step": 671 + }, + { + "epoch": 0.13, + "learning_rate": 1.9508095752951527e-05, + "loss": 0.79, + "step": 672 + }, + { + "epoch": 0.13, + "learning_rate": 1.9506192085168015e-05, + "loss": 0.924, + "step": 673 + }, + { + "epoch": 0.13, + "learning_rate": 1.9504284834147914e-05, + "loss": 0.8678, + "step": 674 + }, + { + "epoch": 0.13, + "learning_rate": 1.9502374000610152e-05, + "loss": 0.8537, + "step": 675 + }, + { + "epoch": 0.13, + "learning_rate": 1.950045958527498e-05, + "loss": 0.8699, + "step": 676 + }, + { + "epoch": 0.13, + "learning_rate": 1.9498541588864022e-05, + "loss": 0.8642, + "step": 677 + }, + { + "epoch": 0.13, + "learning_rate": 1.9496620012100232e-05, + "loss": 0.8766, + "step": 678 + }, + { + "epoch": 0.13, + "learning_rate": 1.9494694855707932e-05, + "loss": 0.8947, + "step": 679 + }, + { + "epoch": 0.13, + "learning_rate": 1.9492766120412782e-05, + "loss": 0.8811, + "step": 680 + }, + { + "epoch": 0.13, + "learning_rate": 1.949083380694179e-05, + "loss": 0.8454, + "step": 681 + }, + { + "epoch": 0.13, + "learning_rate": 1.9488897916023328e-05, + "loss": 0.8874, + "step": 682 + }, + { + "epoch": 0.13, + "learning_rate": 1.948695844838709e-05, + "loss": 0.8707, + "step": 683 + }, + { + "epoch": 0.13, + "learning_rate": 1.948501540476414e-05, + "loss": 0.8406, + "step": 684 + }, + { + "epoch": 0.13, + "learning_rate": 1.948306878588689e-05, + "loss": 0.8828, + "step": 685 + }, + { + "epoch": 0.13, + "learning_rate": 1.9481118592489086e-05, + "loss": 0.8861, + "step": 686 + }, + { + "epoch": 0.13, + "learning_rate": 1.9479164825305827e-05, + "loss": 0.8653, + "step": 687 + }, + { + "epoch": 0.13, + "learning_rate": 1.9477207485073563e-05, + "loss": 0.8471, + "step": 688 + }, + { + "epoch": 0.13, + "learning_rate": 1.9475246572530086e-05, + "loss": 0.9114, + "step": 689 + }, + { + "epoch": 0.13, + "learning_rate": 1.9473282088414542e-05, + "loss": 0.8661, + "step": 690 + }, + { + "epoch": 0.13, + "learning_rate": 1.9471314033467413e-05, + "loss": 0.8921, + "step": 691 + }, + { + "epoch": 0.13, + "learning_rate": 1.9469342408430534e-05, + "loss": 0.896, + "step": 692 + }, + { + "epoch": 0.13, + "learning_rate": 1.946736721404708e-05, + "loss": 0.8552, + "step": 693 + }, + { + "epoch": 0.13, + "learning_rate": 1.946538845106158e-05, + "loss": 0.8454, + "step": 694 + }, + { + "epoch": 0.13, + "learning_rate": 1.94634061202199e-05, + "loss": 0.9049, + "step": 695 + }, + { + "epoch": 0.13, + "learning_rate": 1.946142022226926e-05, + "loss": 0.9121, + "step": 696 + }, + { + "epoch": 0.13, + "learning_rate": 1.945943075795821e-05, + "loss": 0.7852, + "step": 697 + }, + { + "epoch": 0.13, + "learning_rate": 1.945743772803666e-05, + "loss": 0.8358, + "step": 698 + }, + { + "epoch": 0.13, + "learning_rate": 1.945544113325585e-05, + "loss": 0.7756, + "step": 699 + }, + { + "epoch": 0.13, + "learning_rate": 1.9453440974368378e-05, + "loss": 0.889, + "step": 700 + }, + { + "epoch": 0.13, + "learning_rate": 1.9451437252128172e-05, + "loss": 0.8662, + "step": 701 + }, + { + "epoch": 0.13, + "learning_rate": 1.944942996729051e-05, + "loss": 0.9072, + "step": 702 + }, + { + "epoch": 0.13, + "learning_rate": 1.9447419120612018e-05, + "loss": 0.8754, + "step": 703 + }, + { + "epoch": 0.13, + "learning_rate": 1.9445404712850653e-05, + "loss": 0.7917, + "step": 704 + }, + { + "epoch": 0.13, + "learning_rate": 1.9443386744765726e-05, + "loss": 0.8636, + "step": 705 + }, + { + "epoch": 0.13, + "learning_rate": 1.9441365217117876e-05, + "loss": 0.8724, + "step": 706 + }, + { + "epoch": 0.13, + "learning_rate": 1.9439340130669094e-05, + "loss": 0.8024, + "step": 707 + }, + { + "epoch": 0.13, + "learning_rate": 1.9437311486182713e-05, + "loss": 0.8208, + "step": 708 + }, + { + "epoch": 0.13, + "learning_rate": 1.9435279284423404e-05, + "loss": 0.8771, + "step": 709 + }, + { + "epoch": 0.13, + "learning_rate": 1.943324352615718e-05, + "loss": 0.2646, + "step": 710 + }, + { + "epoch": 0.13, + "learning_rate": 1.9431204212151394e-05, + "loss": 0.9238, + "step": 711 + }, + { + "epoch": 0.13, + "learning_rate": 1.942916134317473e-05, + "loss": 0.8825, + "step": 712 + }, + { + "epoch": 0.14, + "learning_rate": 1.9427114919997237e-05, + "loss": 0.8587, + "step": 713 + }, + { + "epoch": 0.14, + "learning_rate": 1.9425064943390277e-05, + "loss": 0.8498, + "step": 714 + }, + { + "epoch": 0.14, + "learning_rate": 1.942301141412656e-05, + "loss": 0.857, + "step": 715 + }, + { + "epoch": 0.14, + "learning_rate": 1.9420954332980148e-05, + "loss": 0.8416, + "step": 716 + }, + { + "epoch": 0.14, + "learning_rate": 1.9418893700726422e-05, + "loss": 0.2621, + "step": 717 + }, + { + "epoch": 0.14, + "learning_rate": 1.941682951814212e-05, + "loss": 0.8769, + "step": 718 + }, + { + "epoch": 0.14, + "learning_rate": 1.9414761786005293e-05, + "loss": 0.9248, + "step": 719 + }, + { + "epoch": 0.14, + "learning_rate": 1.9412690505095363e-05, + "loss": 0.8927, + "step": 720 + }, + { + "epoch": 0.14, + "learning_rate": 1.9410615676193064e-05, + "loss": 0.9029, + "step": 721 + }, + { + "epoch": 0.14, + "learning_rate": 1.940853730008048e-05, + "loss": 0.8264, + "step": 722 + }, + { + "epoch": 0.14, + "learning_rate": 1.9406455377541024e-05, + "loss": 0.8891, + "step": 723 + }, + { + "epoch": 0.14, + "learning_rate": 1.9404369909359453e-05, + "loss": 0.8774, + "step": 724 + }, + { + "epoch": 0.14, + "learning_rate": 1.9402280896321853e-05, + "loss": 0.8326, + "step": 725 + }, + { + "epoch": 0.14, + "learning_rate": 1.9400188339215657e-05, + "loss": 0.8343, + "step": 726 + }, + { + "epoch": 0.14, + "learning_rate": 1.9398092238829625e-05, + "loss": 0.871, + "step": 727 + }, + { + "epoch": 0.14, + "learning_rate": 1.9395992595953852e-05, + "loss": 0.9176, + "step": 728 + }, + { + "epoch": 0.14, + "learning_rate": 1.9393889411379773e-05, + "loss": 0.9078, + "step": 729 + }, + { + "epoch": 0.14, + "learning_rate": 1.939178268590016e-05, + "loss": 0.2726, + "step": 730 + }, + { + "epoch": 0.14, + "learning_rate": 1.9389672420309107e-05, + "loss": 0.842, + "step": 731 + }, + { + "epoch": 0.14, + "learning_rate": 1.938755861540206e-05, + "loss": 0.8804, + "step": 732 + }, + { + "epoch": 0.14, + "learning_rate": 1.9385441271975786e-05, + "loss": 0.8803, + "step": 733 + }, + { + "epoch": 0.14, + "learning_rate": 1.9383320390828397e-05, + "loss": 0.8512, + "step": 734 + }, + { + "epoch": 0.14, + "learning_rate": 1.9381195972759325e-05, + "loss": 0.8494, + "step": 735 + }, + { + "epoch": 0.14, + "learning_rate": 1.9379068018569342e-05, + "loss": 0.8446, + "step": 736 + }, + { + "epoch": 0.14, + "learning_rate": 1.9376936529060556e-05, + "loss": 0.8439, + "step": 737 + }, + { + "epoch": 0.14, + "learning_rate": 1.9374801505036404e-05, + "loss": 0.897, + "step": 738 + }, + { + "epoch": 0.14, + "learning_rate": 1.937266294730165e-05, + "loss": 0.875, + "step": 739 + }, + { + "epoch": 0.14, + "learning_rate": 1.9370520856662406e-05, + "loss": 0.8757, + "step": 740 + }, + { + "epoch": 0.14, + "learning_rate": 1.93683752339261e-05, + "loss": 0.8774, + "step": 741 + }, + { + "epoch": 0.14, + "learning_rate": 1.9366226079901494e-05, + "loss": 0.8151, + "step": 742 + }, + { + "epoch": 0.14, + "learning_rate": 1.936407339539869e-05, + "loss": 0.8094, + "step": 743 + }, + { + "epoch": 0.14, + "learning_rate": 1.936191718122911e-05, + "loss": 0.9259, + "step": 744 + }, + { + "epoch": 0.14, + "learning_rate": 1.9359757438205512e-05, + "loss": 0.8667, + "step": 745 + }, + { + "epoch": 0.14, + "learning_rate": 1.9357594167141985e-05, + "loss": 0.2887, + "step": 746 + }, + { + "epoch": 0.14, + "learning_rate": 1.9355427368853946e-05, + "loss": 0.8885, + "step": 747 + }, + { + "epoch": 0.14, + "learning_rate": 1.935325704415814e-05, + "loss": 0.871, + "step": 748 + }, + { + "epoch": 0.14, + "learning_rate": 1.9351083193872647e-05, + "loss": 0.88, + "step": 749 + }, + { + "epoch": 0.14, + "learning_rate": 1.9348905818816864e-05, + "loss": 0.893, + "step": 750 + }, + { + "epoch": 0.14, + "learning_rate": 1.9346724919811534e-05, + "loss": 0.2843, + "step": 751 + }, + { + "epoch": 0.14, + "learning_rate": 1.934454049767871e-05, + "loss": 0.8498, + "step": 752 + }, + { + "epoch": 0.14, + "learning_rate": 1.9342352553241787e-05, + "loss": 0.8297, + "step": 753 + }, + { + "epoch": 0.14, + "learning_rate": 1.9340161087325483e-05, + "loss": 0.8843, + "step": 754 + }, + { + "epoch": 0.14, + "learning_rate": 1.9337966100755837e-05, + "loss": 0.8196, + "step": 755 + }, + { + "epoch": 0.14, + "learning_rate": 1.933576759436023e-05, + "loss": 0.2302, + "step": 756 + }, + { + "epoch": 0.14, + "learning_rate": 1.933356556896735e-05, + "loss": 0.8405, + "step": 757 + }, + { + "epoch": 0.14, + "learning_rate": 1.933136002540723e-05, + "loss": 0.8717, + "step": 758 + }, + { + "epoch": 0.14, + "learning_rate": 1.932915096451122e-05, + "loss": 0.9017, + "step": 759 + }, + { + "epoch": 0.14, + "learning_rate": 1.9326938387111997e-05, + "loss": 0.8534, + "step": 760 + }, + { + "epoch": 0.14, + "learning_rate": 1.932472229404356e-05, + "loss": 0.8572, + "step": 761 + }, + { + "epoch": 0.14, + "learning_rate": 1.9322502686141237e-05, + "loss": 0.8466, + "step": 762 + }, + { + "epoch": 0.14, + "learning_rate": 1.932027956424169e-05, + "loss": 0.8509, + "step": 763 + }, + { + "epoch": 0.14, + "learning_rate": 1.931805292918288e-05, + "loss": 0.8911, + "step": 764 + }, + { + "epoch": 0.14, + "learning_rate": 1.9315822781804125e-05, + "loss": 0.9105, + "step": 765 + }, + { + "epoch": 0.15, + "learning_rate": 1.931358912294604e-05, + "loss": 0.9218, + "step": 766 + }, + { + "epoch": 0.15, + "learning_rate": 1.9311351953450574e-05, + "loss": 0.8842, + "step": 767 + }, + { + "epoch": 0.15, + "learning_rate": 1.9309111274161005e-05, + "loss": 0.9352, + "step": 768 + }, + { + "epoch": 0.15, + "learning_rate": 1.9306867085921924e-05, + "loss": 0.849, + "step": 769 + }, + { + "epoch": 0.15, + "learning_rate": 1.9304619389579248e-05, + "loss": 0.8658, + "step": 770 + }, + { + "epoch": 0.15, + "learning_rate": 1.9302368185980218e-05, + "loss": 0.918, + "step": 771 + }, + { + "epoch": 0.15, + "learning_rate": 1.9300113475973398e-05, + "loss": 0.7954, + "step": 772 + }, + { + "epoch": 0.15, + "learning_rate": 1.9297855260408672e-05, + "loss": 0.8469, + "step": 773 + }, + { + "epoch": 0.15, + "learning_rate": 1.9295593540137247e-05, + "loss": 0.8412, + "step": 774 + }, + { + "epoch": 0.15, + "learning_rate": 1.9293328316011645e-05, + "loss": 0.8085, + "step": 775 + }, + { + "epoch": 0.15, + "learning_rate": 1.929105958888571e-05, + "loss": 0.2844, + "step": 776 + }, + { + "epoch": 0.15, + "learning_rate": 1.928878735961462e-05, + "loss": 0.8631, + "step": 777 + }, + { + "epoch": 0.15, + "learning_rate": 1.9286511629054852e-05, + "loss": 0.8212, + "step": 778 + }, + { + "epoch": 0.15, + "learning_rate": 1.928423239806422e-05, + "loss": 0.8795, + "step": 779 + }, + { + "epoch": 0.15, + "learning_rate": 1.928194966750185e-05, + "loss": 0.8737, + "step": 780 + }, + { + "epoch": 0.15, + "learning_rate": 1.9279663438228182e-05, + "loss": 0.236, + "step": 781 + }, + { + "epoch": 0.15, + "learning_rate": 1.927737371110499e-05, + "loss": 0.2356, + "step": 782 + }, + { + "epoch": 0.15, + "learning_rate": 1.927508048699535e-05, + "loss": 0.8255, + "step": 783 + }, + { + "epoch": 0.15, + "learning_rate": 1.9272783766763663e-05, + "loss": 0.8374, + "step": 784 + }, + { + "epoch": 0.15, + "learning_rate": 1.9270483551275654e-05, + "loss": 0.8383, + "step": 785 + }, + { + "epoch": 0.15, + "learning_rate": 1.9268179841398354e-05, + "loss": 0.8276, + "step": 786 + }, + { + "epoch": 0.15, + "learning_rate": 1.926587263800012e-05, + "loss": 0.8409, + "step": 787 + }, + { + "epoch": 0.15, + "learning_rate": 1.9263561941950622e-05, + "loss": 0.8748, + "step": 788 + }, + { + "epoch": 0.15, + "learning_rate": 1.9261247754120846e-05, + "loss": 0.8788, + "step": 789 + }, + { + "epoch": 0.15, + "learning_rate": 1.9258930075383095e-05, + "loss": 0.8728, + "step": 790 + }, + { + "epoch": 0.15, + "learning_rate": 1.9256608906610988e-05, + "loss": 0.8897, + "step": 791 + }, + { + "epoch": 0.15, + "learning_rate": 1.9254284248679464e-05, + "loss": 0.8805, + "step": 792 + }, + { + "epoch": 0.15, + "learning_rate": 1.925195610246477e-05, + "loss": 0.8716, + "step": 793 + }, + { + "epoch": 0.15, + "learning_rate": 1.9249624468844468e-05, + "loss": 0.8325, + "step": 794 + }, + { + "epoch": 0.15, + "learning_rate": 1.924728934869744e-05, + "loss": 0.9053, + "step": 795 + }, + { + "epoch": 0.15, + "learning_rate": 1.924495074290388e-05, + "loss": 0.8126, + "step": 796 + }, + { + "epoch": 0.15, + "learning_rate": 1.924260865234529e-05, + "loss": 0.8717, + "step": 797 + }, + { + "epoch": 0.15, + "learning_rate": 1.9240263077904498e-05, + "loss": 0.8501, + "step": 798 + }, + { + "epoch": 0.15, + "learning_rate": 1.9237914020465637e-05, + "loss": 0.8389, + "step": 799 + }, + { + "epoch": 0.15, + "learning_rate": 1.923556148091415e-05, + "loss": 0.2915, + "step": 800 + }, + { + "epoch": 0.15, + "learning_rate": 1.9233205460136796e-05, + "loss": 0.9075, + "step": 801 + }, + { + "epoch": 0.15, + "learning_rate": 1.9230845959021652e-05, + "loss": 0.8716, + "step": 802 + }, + { + "epoch": 0.15, + "learning_rate": 1.92284829784581e-05, + "loss": 0.8883, + "step": 803 + }, + { + "epoch": 0.15, + "learning_rate": 1.922611651933683e-05, + "loss": 0.874, + "step": 804 + }, + { + "epoch": 0.15, + "learning_rate": 1.9223746582549853e-05, + "loss": 0.8805, + "step": 805 + }, + { + "epoch": 0.15, + "learning_rate": 1.9221373168990488e-05, + "loss": 0.8151, + "step": 806 + }, + { + "epoch": 0.15, + "learning_rate": 1.9218996279553357e-05, + "loss": 0.2828, + "step": 807 + }, + { + "epoch": 0.15, + "learning_rate": 1.9216615915134402e-05, + "loss": 0.8899, + "step": 808 + }, + { + "epoch": 0.15, + "learning_rate": 1.9214232076630868e-05, + "loss": 0.8788, + "step": 809 + }, + { + "epoch": 0.15, + "learning_rate": 1.9211844764941318e-05, + "loss": 0.8587, + "step": 810 + }, + { + "epoch": 0.15, + "learning_rate": 1.9209453980965608e-05, + "loss": 0.8693, + "step": 811 + }, + { + "epoch": 0.15, + "learning_rate": 1.9207059725604922e-05, + "loss": 0.9028, + "step": 812 + }, + { + "epoch": 0.15, + "learning_rate": 1.920466199976174e-05, + "loss": 0.8523, + "step": 813 + }, + { + "epoch": 0.15, + "learning_rate": 1.9202260804339856e-05, + "loss": 0.9104, + "step": 814 + }, + { + "epoch": 0.15, + "learning_rate": 1.9199856140244365e-05, + "loss": 0.8292, + "step": 815 + }, + { + "epoch": 0.15, + "learning_rate": 1.919744800838168e-05, + "loss": 0.8771, + "step": 816 + }, + { + "epoch": 0.15, + "learning_rate": 1.919503640965951e-05, + "loss": 0.8613, + "step": 817 + }, + { + "epoch": 0.16, + "learning_rate": 1.919262134498688e-05, + "loss": 0.8236, + "step": 818 + }, + { + "epoch": 0.16, + "learning_rate": 1.9190202815274118e-05, + "loss": 0.866, + "step": 819 + }, + { + "epoch": 0.16, + "learning_rate": 1.918778082143286e-05, + "loss": 0.8547, + "step": 820 + }, + { + "epoch": 0.16, + "learning_rate": 1.918535536437603e-05, + "loss": 0.8704, + "step": 821 + }, + { + "epoch": 0.16, + "learning_rate": 1.9182926445017893e-05, + "loss": 0.8138, + "step": 822 + }, + { + "epoch": 0.16, + "learning_rate": 1.918049406427399e-05, + "loss": 0.2788, + "step": 823 + }, + { + "epoch": 0.16, + "learning_rate": 1.917805822306117e-05, + "loss": 0.7942, + "step": 824 + }, + { + "epoch": 0.16, + "learning_rate": 1.91756189222976e-05, + "loss": 0.8195, + "step": 825 + }, + { + "epoch": 0.16, + "learning_rate": 1.917317616290274e-05, + "loss": 0.8158, + "step": 826 + }, + { + "epoch": 0.16, + "learning_rate": 1.9170729945797357e-05, + "loss": 0.9198, + "step": 827 + }, + { + "epoch": 0.16, + "learning_rate": 1.916828027190352e-05, + "loss": 0.8592, + "step": 828 + }, + { + "epoch": 0.16, + "learning_rate": 1.9165827142144606e-05, + "loss": 0.8571, + "step": 829 + }, + { + "epoch": 0.16, + "learning_rate": 1.9163370557445286e-05, + "loss": 0.8459, + "step": 830 + }, + { + "epoch": 0.16, + "learning_rate": 1.916091051873154e-05, + "loss": 0.7494, + "step": 831 + }, + { + "epoch": 0.16, + "learning_rate": 1.9158447026930648e-05, + "loss": 0.8911, + "step": 832 + }, + { + "epoch": 0.16, + "learning_rate": 1.915598008297119e-05, + "loss": 0.8558, + "step": 833 + }, + { + "epoch": 0.16, + "learning_rate": 1.915350968778305e-05, + "loss": 0.8173, + "step": 834 + }, + { + "epoch": 0.16, + "learning_rate": 1.9151035842297412e-05, + "loss": 0.262, + "step": 835 + }, + { + "epoch": 0.16, + "learning_rate": 1.914855854744676e-05, + "loss": 0.8676, + "step": 836 + }, + { + "epoch": 0.16, + "learning_rate": 1.9146077804164874e-05, + "loss": 0.9038, + "step": 837 + }, + { + "epoch": 0.16, + "learning_rate": 1.9143593613386845e-05, + "loss": 0.8853, + "step": 838 + }, + { + "epoch": 0.16, + "learning_rate": 1.9141105976049054e-05, + "loss": 0.8745, + "step": 839 + }, + { + "epoch": 0.16, + "learning_rate": 1.9138614893089183e-05, + "loss": 0.8365, + "step": 840 + }, + { + "epoch": 0.16, + "learning_rate": 1.9136120365446215e-05, + "loss": 0.8944, + "step": 841 + }, + { + "epoch": 0.16, + "learning_rate": 1.9133622394060428e-05, + "loss": 0.861, + "step": 842 + }, + { + "epoch": 0.16, + "learning_rate": 1.9131120979873404e-05, + "loss": 0.8187, + "step": 843 + }, + { + "epoch": 0.16, + "learning_rate": 1.9128616123828013e-05, + "loss": 0.8897, + "step": 844 + }, + { + "epoch": 0.16, + "learning_rate": 1.9126107826868436e-05, + "loss": 0.826, + "step": 845 + }, + { + "epoch": 0.16, + "learning_rate": 1.9123596089940134e-05, + "loss": 0.8633, + "step": 846 + }, + { + "epoch": 0.16, + "learning_rate": 1.912108091398988e-05, + "loss": 0.8182, + "step": 847 + }, + { + "epoch": 0.16, + "learning_rate": 1.9118562299965734e-05, + "loss": 0.8719, + "step": 848 + }, + { + "epoch": 0.16, + "learning_rate": 1.9116040248817057e-05, + "loss": 0.9116, + "step": 849 + }, + { + "epoch": 0.16, + "learning_rate": 1.9113514761494506e-05, + "loss": 0.8109, + "step": 850 + }, + { + "epoch": 0.16, + "learning_rate": 1.9110985838950026e-05, + "loss": 0.868, + "step": 851 + }, + { + "epoch": 0.16, + "learning_rate": 1.9108453482136866e-05, + "loss": 0.8508, + "step": 852 + }, + { + "epoch": 0.16, + "learning_rate": 1.9105917692009563e-05, + "loss": 0.8997, + "step": 853 + }, + { + "epoch": 0.16, + "learning_rate": 1.9103378469523953e-05, + "loss": 0.863, + "step": 854 + }, + { + "epoch": 0.16, + "learning_rate": 1.9100835815637157e-05, + "loss": 0.8145, + "step": 855 + }, + { + "epoch": 0.16, + "learning_rate": 1.90982897313076e-05, + "loss": 0.8275, + "step": 856 + }, + { + "epoch": 0.16, + "learning_rate": 1.9095740217494995e-05, + "loss": 0.8213, + "step": 857 + }, + { + "epoch": 0.16, + "learning_rate": 1.909318727516035e-05, + "loss": 0.8959, + "step": 858 + }, + { + "epoch": 0.16, + "learning_rate": 1.9090630905265963e-05, + "loss": 0.8461, + "step": 859 + }, + { + "epoch": 0.16, + "learning_rate": 1.9088071108775424e-05, + "loss": 0.8569, + "step": 860 + }, + { + "epoch": 0.16, + "learning_rate": 1.9085507886653613e-05, + "loss": 0.9082, + "step": 861 + }, + { + "epoch": 0.16, + "learning_rate": 1.9082941239866705e-05, + "loss": 0.827, + "step": 862 + }, + { + "epoch": 0.16, + "learning_rate": 1.908037116938217e-05, + "loss": 0.7992, + "step": 863 + }, + { + "epoch": 0.16, + "learning_rate": 1.9077797676168754e-05, + "loss": 0.8697, + "step": 864 + }, + { + "epoch": 0.16, + "learning_rate": 1.907522076119651e-05, + "loss": 0.8265, + "step": 865 + }, + { + "epoch": 0.16, + "learning_rate": 1.9072640425436762e-05, + "loss": 0.9176, + "step": 866 + }, + { + "epoch": 0.16, + "learning_rate": 1.907005666986215e-05, + "loss": 0.8506, + "step": 867 + }, + { + "epoch": 0.16, + "learning_rate": 1.9067469495446574e-05, + "loss": 0.8409, + "step": 868 + }, + { + "epoch": 0.16, + "learning_rate": 1.9064878903165242e-05, + "loss": 0.8848, + "step": 869 + }, + { + "epoch": 0.16, + "learning_rate": 1.906228489399464e-05, + "loss": 0.805, + "step": 870 + }, + { + "epoch": 0.17, + "learning_rate": 1.9059687468912558e-05, + "loss": 0.2712, + "step": 871 + }, + { + "epoch": 0.17, + "learning_rate": 1.9057086628898046e-05, + "loss": 0.8291, + "step": 872 + }, + { + "epoch": 0.17, + "learning_rate": 1.905448237493147e-05, + "loss": 0.8614, + "step": 873 + }, + { + "epoch": 0.17, + "learning_rate": 1.905187470799446e-05, + "loss": 0.8002, + "step": 874 + }, + { + "epoch": 0.17, + "learning_rate": 1.904926362906995e-05, + "loss": 0.8709, + "step": 875 + }, + { + "epoch": 0.17, + "learning_rate": 1.9046649139142145e-05, + "loss": 0.7979, + "step": 876 + }, + { + "epoch": 0.17, + "learning_rate": 1.9044031239196552e-05, + "loss": 0.2445, + "step": 877 + }, + { + "epoch": 0.17, + "learning_rate": 1.904140993021995e-05, + "loss": 0.9019, + "step": 878 + }, + { + "epoch": 0.17, + "learning_rate": 1.903878521320041e-05, + "loss": 0.8435, + "step": 879 + }, + { + "epoch": 0.17, + "learning_rate": 1.9036157089127278e-05, + "loss": 0.9203, + "step": 880 + }, + { + "epoch": 0.17, + "learning_rate": 1.9033525558991203e-05, + "loss": 0.8141, + "step": 881 + }, + { + "epoch": 0.17, + "learning_rate": 1.9030890623784092e-05, + "loss": 0.8663, + "step": 882 + }, + { + "epoch": 0.17, + "learning_rate": 1.9028252284499164e-05, + "loss": 0.9249, + "step": 883 + }, + { + "epoch": 0.17, + "learning_rate": 1.9025610542130902e-05, + "loss": 0.8309, + "step": 884 + }, + { + "epoch": 0.17, + "learning_rate": 1.9022965397675075e-05, + "loss": 0.8748, + "step": 885 + }, + { + "epoch": 0.17, + "learning_rate": 1.9020316852128737e-05, + "loss": 0.8271, + "step": 886 + }, + { + "epoch": 0.17, + "learning_rate": 1.901766490649022e-05, + "loss": 0.8695, + "step": 887 + }, + { + "epoch": 0.17, + "learning_rate": 1.9015009561759147e-05, + "loss": 0.8516, + "step": 888 + }, + { + "epoch": 0.17, + "learning_rate": 1.901235081893641e-05, + "loss": 0.8933, + "step": 889 + }, + { + "epoch": 0.17, + "learning_rate": 1.900968867902419e-05, + "loss": 0.882, + "step": 890 + }, + { + "epoch": 0.17, + "learning_rate": 1.900702314302595e-05, + "loss": 0.7781, + "step": 891 + }, + { + "epoch": 0.17, + "learning_rate": 1.900435421194643e-05, + "loss": 0.8347, + "step": 892 + }, + { + "epoch": 0.17, + "learning_rate": 1.900168188679164e-05, + "loss": 0.8428, + "step": 893 + }, + { + "epoch": 0.17, + "learning_rate": 1.8999006168568883e-05, + "loss": 0.8445, + "step": 894 + }, + { + "epoch": 0.17, + "learning_rate": 1.899632705828674e-05, + "loss": 0.8197, + "step": 895 + }, + { + "epoch": 0.17, + "learning_rate": 1.8993644556955064e-05, + "loss": 0.8684, + "step": 896 + }, + { + "epoch": 0.17, + "learning_rate": 1.899095866558499e-05, + "loss": 0.8536, + "step": 897 + }, + { + "epoch": 0.17, + "learning_rate": 1.8988269385188932e-05, + "loss": 0.2727, + "step": 898 + }, + { + "epoch": 0.17, + "learning_rate": 1.8985576716780575e-05, + "loss": 0.8417, + "step": 899 + }, + { + "epoch": 0.17, + "learning_rate": 1.898288066137489e-05, + "loss": 0.8739, + "step": 900 + }, + { + "epoch": 0.17, + "learning_rate": 1.8980181219988117e-05, + "loss": 0.9077, + "step": 901 + }, + { + "epoch": 0.17, + "learning_rate": 1.897747839363778e-05, + "loss": 0.8079, + "step": 902 + }, + { + "epoch": 0.17, + "learning_rate": 1.8974772183342667e-05, + "loss": 0.8382, + "step": 903 + }, + { + "epoch": 0.17, + "learning_rate": 1.8972062590122857e-05, + "loss": 0.8451, + "step": 904 + }, + { + "epoch": 0.17, + "learning_rate": 1.8969349614999695e-05, + "loss": 0.9116, + "step": 905 + }, + { + "epoch": 0.17, + "learning_rate": 1.8966633258995796e-05, + "loss": 0.7854, + "step": 906 + }, + { + "epoch": 0.17, + "learning_rate": 1.8963913523135062e-05, + "loss": 0.8475, + "step": 907 + }, + { + "epoch": 0.17, + "learning_rate": 1.8961190408442662e-05, + "loss": 0.8735, + "step": 908 + }, + { + "epoch": 0.17, + "learning_rate": 1.8958463915945034e-05, + "loss": 0.9149, + "step": 909 + }, + { + "epoch": 0.17, + "learning_rate": 1.8955734046669894e-05, + "loss": 0.8549, + "step": 910 + }, + { + "epoch": 0.17, + "learning_rate": 1.895300080164624e-05, + "loss": 0.2612, + "step": 911 + }, + { + "epoch": 0.17, + "learning_rate": 1.895026418190433e-05, + "loss": 0.8071, + "step": 912 + }, + { + "epoch": 0.17, + "learning_rate": 1.8947524188475686e-05, + "loss": 0.862, + "step": 913 + }, + { + "epoch": 0.17, + "learning_rate": 1.894478082239313e-05, + "loss": 0.2616, + "step": 914 + }, + { + "epoch": 0.17, + "learning_rate": 1.8942034084690727e-05, + "loss": 0.8867, + "step": 915 + }, + { + "epoch": 0.17, + "learning_rate": 1.8939283976403827e-05, + "loss": 0.8074, + "step": 916 + }, + { + "epoch": 0.17, + "learning_rate": 1.8936530498569054e-05, + "loss": 0.901, + "step": 917 + }, + { + "epoch": 0.17, + "learning_rate": 1.8933773652224288e-05, + "loss": 0.8109, + "step": 918 + }, + { + "epoch": 0.17, + "learning_rate": 1.893101343840869e-05, + "loss": 0.8131, + "step": 919 + }, + { + "epoch": 0.17, + "learning_rate": 1.892824985816269e-05, + "loss": 0.8889, + "step": 920 + }, + { + "epoch": 0.17, + "learning_rate": 1.8925482912527975e-05, + "loss": 0.285, + "step": 921 + }, + { + "epoch": 0.17, + "learning_rate": 1.8922712602547516e-05, + "loss": 0.8246, + "step": 922 + }, + { + "epoch": 0.17, + "learning_rate": 1.8919938929265547e-05, + "loss": 0.8161, + "step": 923 + }, + { + "epoch": 0.18, + "learning_rate": 1.891716189372757e-05, + "loss": 0.8624, + "step": 924 + }, + { + "epoch": 0.18, + "learning_rate": 1.8914381496980344e-05, + "loss": 0.8258, + "step": 925 + }, + { + "epoch": 0.18, + "learning_rate": 1.891159774007191e-05, + "loss": 0.8693, + "step": 926 + }, + { + "epoch": 0.18, + "learning_rate": 1.890881062405157e-05, + "loss": 0.885, + "step": 927 + }, + { + "epoch": 0.18, + "learning_rate": 1.8906020149969887e-05, + "loss": 0.9158, + "step": 928 + }, + { + "epoch": 0.18, + "learning_rate": 1.89032263188787e-05, + "loss": 0.8797, + "step": 929 + }, + { + "epoch": 0.18, + "learning_rate": 1.89004291318311e-05, + "loss": 0.8801, + "step": 930 + }, + { + "epoch": 0.18, + "learning_rate": 1.889762858988146e-05, + "loss": 0.8565, + "step": 931 + }, + { + "epoch": 0.18, + "learning_rate": 1.8894824694085397e-05, + "loss": 0.8346, + "step": 932 + }, + { + "epoch": 0.18, + "learning_rate": 1.8892017445499812e-05, + "loss": 0.864, + "step": 933 + }, + { + "epoch": 0.18, + "learning_rate": 1.8889206845182863e-05, + "loss": 0.8623, + "step": 934 + }, + { + "epoch": 0.18, + "learning_rate": 1.888639289419396e-05, + "loss": 0.8593, + "step": 935 + }, + { + "epoch": 0.18, + "learning_rate": 1.8883575593593793e-05, + "loss": 0.8508, + "step": 936 + }, + { + "epoch": 0.18, + "learning_rate": 1.8880754944444302e-05, + "loss": 0.8325, + "step": 937 + }, + { + "epoch": 0.18, + "learning_rate": 1.8877930947808702e-05, + "loss": 0.9307, + "step": 938 + }, + { + "epoch": 0.18, + "learning_rate": 1.8875103604751458e-05, + "loss": 0.8999, + "step": 939 + }, + { + "epoch": 0.18, + "learning_rate": 1.8872272916338293e-05, + "loss": 0.8244, + "step": 940 + }, + { + "epoch": 0.18, + "learning_rate": 1.8869438883636212e-05, + "loss": 0.8495, + "step": 941 + }, + { + "epoch": 0.18, + "learning_rate": 1.886660150771346e-05, + "loss": 0.2575, + "step": 942 + }, + { + "epoch": 0.18, + "learning_rate": 1.8863760789639548e-05, + "loss": 0.8378, + "step": 943 + }, + { + "epoch": 0.18, + "learning_rate": 1.886091673048525e-05, + "loss": 0.8808, + "step": 944 + }, + { + "epoch": 0.18, + "learning_rate": 1.88580693313226e-05, + "loss": 0.8293, + "step": 945 + }, + { + "epoch": 0.18, + "learning_rate": 1.8855218593224882e-05, + "loss": 0.7812, + "step": 946 + }, + { + "epoch": 0.18, + "learning_rate": 1.885236451726665e-05, + "loss": 0.2988, + "step": 947 + }, + { + "epoch": 0.18, + "learning_rate": 1.8849507104523712e-05, + "loss": 0.8384, + "step": 948 + }, + { + "epoch": 0.18, + "learning_rate": 1.884664635607313e-05, + "loss": 0.8927, + "step": 949 + }, + { + "epoch": 0.18, + "learning_rate": 1.8843782272993225e-05, + "loss": 0.9059, + "step": 950 + }, + { + "epoch": 0.18, + "learning_rate": 1.884091485636358e-05, + "loss": 0.9002, + "step": 951 + }, + { + "epoch": 0.18, + "learning_rate": 1.8838044107265032e-05, + "loss": 0.9211, + "step": 952 + }, + { + "epoch": 0.18, + "learning_rate": 1.8835170026779665e-05, + "loss": 0.9365, + "step": 953 + }, + { + "epoch": 0.18, + "learning_rate": 1.8832292615990834e-05, + "loss": 0.8652, + "step": 954 + }, + { + "epoch": 0.18, + "learning_rate": 1.882941187598314e-05, + "loss": 0.8857, + "step": 955 + }, + { + "epoch": 0.18, + "learning_rate": 1.8826527807842438e-05, + "loss": 0.8691, + "step": 956 + }, + { + "epoch": 0.18, + "learning_rate": 1.8823640412655844e-05, + "loss": 0.2595, + "step": 957 + }, + { + "epoch": 0.18, + "learning_rate": 1.8820749691511723e-05, + "loss": 0.9202, + "step": 958 + }, + { + "epoch": 0.18, + "learning_rate": 1.8817855645499694e-05, + "loss": 0.8511, + "step": 959 + }, + { + "epoch": 0.18, + "learning_rate": 1.8814958275710634e-05, + "loss": 0.8099, + "step": 960 + }, + { + "epoch": 0.18, + "learning_rate": 1.8812057583236663e-05, + "loss": 0.807, + "step": 961 + }, + { + "epoch": 0.18, + "learning_rate": 1.8809153569171166e-05, + "loss": 0.8089, + "step": 962 + }, + { + "epoch": 0.18, + "learning_rate": 1.880624623460877e-05, + "loss": 0.8673, + "step": 963 + }, + { + "epoch": 0.18, + "learning_rate": 1.880333558064536e-05, + "loss": 0.8459, + "step": 964 + }, + { + "epoch": 0.18, + "learning_rate": 1.8800421608378065e-05, + "loss": 0.8674, + "step": 965 + }, + { + "epoch": 0.18, + "learning_rate": 1.879750431890527e-05, + "loss": 0.8345, + "step": 966 + }, + { + "epoch": 0.18, + "learning_rate": 1.8794583713326616e-05, + "loss": 0.9046, + "step": 967 + }, + { + "epoch": 0.18, + "learning_rate": 1.8791659792742983e-05, + "loss": 0.8531, + "step": 968 + }, + { + "epoch": 0.18, + "learning_rate": 1.87887325582565e-05, + "loss": 0.8419, + "step": 969 + }, + { + "epoch": 0.18, + "learning_rate": 1.8785802010970565e-05, + "loss": 0.8753, + "step": 970 + }, + { + "epoch": 0.18, + "learning_rate": 1.878286815198979e-05, + "loss": 0.8702, + "step": 971 + }, + { + "epoch": 0.18, + "learning_rate": 1.8779930982420076e-05, + "loss": 0.7975, + "step": 972 + }, + { + "epoch": 0.18, + "learning_rate": 1.877699050336854e-05, + "loss": 0.8531, + "step": 973 + }, + { + "epoch": 0.18, + "learning_rate": 1.8774046715943553e-05, + "loss": 0.2712, + "step": 974 + }, + { + "epoch": 0.18, + "learning_rate": 1.8771099621254748e-05, + "loss": 0.8662, + "step": 975 + }, + { + "epoch": 0.18, + "learning_rate": 1.876814922041299e-05, + "loss": 0.8687, + "step": 976 + }, + { + "epoch": 0.19, + "learning_rate": 1.876519551453039e-05, + "loss": 0.8969, + "step": 977 + }, + { + "epoch": 0.19, + "learning_rate": 1.876223850472032e-05, + "loss": 0.8448, + "step": 978 + }, + { + "epoch": 0.19, + "learning_rate": 1.8759278192097377e-05, + "loss": 0.8739, + "step": 979 + }, + { + "epoch": 0.19, + "learning_rate": 1.875631457777742e-05, + "loss": 0.2657, + "step": 980 + }, + { + "epoch": 0.19, + "learning_rate": 1.875334766287754e-05, + "loss": 0.8748, + "step": 981 + }, + { + "epoch": 0.19, + "learning_rate": 1.8750377448516078e-05, + "loss": 0.8592, + "step": 982 + }, + { + "epoch": 0.19, + "learning_rate": 1.874740393581262e-05, + "loss": 0.8945, + "step": 983 + }, + { + "epoch": 0.19, + "learning_rate": 1.8744427125887992e-05, + "loss": 0.8523, + "step": 984 + }, + { + "epoch": 0.19, + "learning_rate": 1.8741447019864263e-05, + "loss": 0.8241, + "step": 985 + }, + { + "epoch": 0.19, + "learning_rate": 1.873846361886475e-05, + "loss": 0.9449, + "step": 986 + }, + { + "epoch": 0.19, + "learning_rate": 1.8735476924014003e-05, + "loss": 0.8419, + "step": 987 + }, + { + "epoch": 0.19, + "learning_rate": 1.873248693643782e-05, + "loss": 0.8348, + "step": 988 + }, + { + "epoch": 0.19, + "learning_rate": 1.8729493657263237e-05, + "loss": 0.8238, + "step": 989 + }, + { + "epoch": 0.19, + "learning_rate": 1.8726497087618534e-05, + "loss": 0.7521, + "step": 990 + }, + { + "epoch": 0.19, + "learning_rate": 1.872349722863323e-05, + "loss": 0.818, + "step": 991 + }, + { + "epoch": 0.19, + "learning_rate": 1.872049408143808e-05, + "loss": 0.9314, + "step": 992 + }, + { + "epoch": 0.19, + "learning_rate": 1.871748764716508e-05, + "loss": 0.8822, + "step": 993 + }, + { + "epoch": 0.19, + "learning_rate": 1.8714477926947472e-05, + "loss": 0.8835, + "step": 994 + }, + { + "epoch": 0.19, + "learning_rate": 1.871146492191973e-05, + "loss": 0.8434, + "step": 995 + }, + { + "epoch": 0.19, + "learning_rate": 1.8708448633217566e-05, + "loss": 0.8135, + "step": 996 + }, + { + "epoch": 0.19, + "learning_rate": 1.870542906197793e-05, + "loss": 0.8479, + "step": 997 + }, + { + "epoch": 0.19, + "learning_rate": 1.870240620933901e-05, + "loss": 0.8263, + "step": 998 + }, + { + "epoch": 0.19, + "learning_rate": 1.8699380076440242e-05, + "loss": 0.8797, + "step": 999 + }, + { + "epoch": 0.19, + "learning_rate": 1.8696350664422272e-05, + "loss": 0.8637, + "step": 1000 + }, + { + "epoch": 0.19, + "learning_rate": 1.8693317974427007e-05, + "loss": 0.8959, + "step": 1001 + }, + { + "epoch": 0.19, + "learning_rate": 1.8690282007597578e-05, + "loss": 0.8824, + "step": 1002 + }, + { + "epoch": 0.19, + "learning_rate": 1.8687242765078354e-05, + "loss": 0.8557, + "step": 1003 + }, + { + "epoch": 0.19, + "learning_rate": 1.868420024801494e-05, + "loss": 0.8597, + "step": 1004 + }, + { + "epoch": 0.19, + "learning_rate": 1.8681154457554168e-05, + "loss": 0.8572, + "step": 1005 + }, + { + "epoch": 0.19, + "learning_rate": 1.8678105394844114e-05, + "loss": 0.8633, + "step": 1006 + }, + { + "epoch": 0.19, + "learning_rate": 1.8675053061034085e-05, + "loss": 0.8661, + "step": 1007 + }, + { + "epoch": 0.19, + "learning_rate": 1.8671997457274615e-05, + "loss": 0.8505, + "step": 1008 + }, + { + "epoch": 0.19, + "learning_rate": 1.8668938584717473e-05, + "loss": 0.8353, + "step": 1009 + }, + { + "epoch": 0.19, + "learning_rate": 1.866587644451567e-05, + "loss": 0.8824, + "step": 1010 + }, + { + "epoch": 0.19, + "learning_rate": 1.8662811037823427e-05, + "loss": 0.8794, + "step": 1011 + }, + { + "epoch": 0.19, + "learning_rate": 1.865974236579622e-05, + "loss": 0.9025, + "step": 1012 + }, + { + "epoch": 0.19, + "learning_rate": 1.8656670429590745e-05, + "loss": 0.8528, + "step": 1013 + }, + { + "epoch": 0.19, + "learning_rate": 1.865359523036492e-05, + "loss": 0.9099, + "step": 1014 + }, + { + "epoch": 0.19, + "learning_rate": 1.8650516769277912e-05, + "loss": 0.874, + "step": 1015 + }, + { + "epoch": 0.19, + "learning_rate": 1.8647435047490104e-05, + "loss": 0.8488, + "step": 1016 + }, + { + "epoch": 0.19, + "learning_rate": 1.864435006616311e-05, + "loss": 0.8015, + "step": 1017 + }, + { + "epoch": 0.19, + "learning_rate": 1.864126182645977e-05, + "loss": 0.887, + "step": 1018 + }, + { + "epoch": 0.19, + "learning_rate": 1.8638170329544164e-05, + "loss": 0.8645, + "step": 1019 + }, + { + "epoch": 0.19, + "learning_rate": 1.8635075576581587e-05, + "loss": 0.8718, + "step": 1020 + }, + { + "epoch": 0.19, + "learning_rate": 1.863197756873857e-05, + "loss": 0.8338, + "step": 1021 + }, + { + "epoch": 0.19, + "learning_rate": 1.8628876307182863e-05, + "loss": 0.8823, + "step": 1022 + }, + { + "epoch": 0.19, + "learning_rate": 1.8625771793083448e-05, + "loss": 0.8626, + "step": 1023 + }, + { + "epoch": 0.19, + "learning_rate": 1.862266402761053e-05, + "loss": 0.8525, + "step": 1024 + }, + { + "epoch": 0.19, + "learning_rate": 1.8619553011935547e-05, + "loss": 0.8703, + "step": 1025 + }, + { + "epoch": 0.19, + "learning_rate": 1.861643874723115e-05, + "loss": 0.8242, + "step": 1026 + }, + { + "epoch": 0.19, + "learning_rate": 1.861332123467122e-05, + "loss": 0.8542, + "step": 1027 + }, + { + "epoch": 0.19, + "learning_rate": 1.8610200475430872e-05, + "loss": 0.8572, + "step": 1028 + }, + { + "epoch": 0.2, + "learning_rate": 1.860707647068643e-05, + "loss": 0.7956, + "step": 1029 + }, + { + "epoch": 0.2, + "learning_rate": 1.860394922161544e-05, + "loss": 0.874, + "step": 1030 + }, + { + "epoch": 0.2, + "learning_rate": 1.8600818729396696e-05, + "loss": 0.8864, + "step": 1031 + }, + { + "epoch": 0.2, + "learning_rate": 1.859768499521018e-05, + "loss": 0.2655, + "step": 1032 + }, + { + "epoch": 0.2, + "learning_rate": 1.859454802023712e-05, + "loss": 0.8656, + "step": 1033 + }, + { + "epoch": 0.2, + "learning_rate": 1.859140780565996e-05, + "loss": 0.8688, + "step": 1034 + }, + { + "epoch": 0.2, + "learning_rate": 1.858826435266236e-05, + "loss": 0.8643, + "step": 1035 + }, + { + "epoch": 0.2, + "learning_rate": 1.8585117662429204e-05, + "loss": 0.8748, + "step": 1036 + }, + { + "epoch": 0.2, + "learning_rate": 1.85819677361466e-05, + "loss": 0.8871, + "step": 1037 + }, + { + "epoch": 0.2, + "learning_rate": 1.8578814575001866e-05, + "loss": 0.2619, + "step": 1038 + }, + { + "epoch": 0.2, + "learning_rate": 1.857565818018355e-05, + "loss": 0.8608, + "step": 1039 + }, + { + "epoch": 0.2, + "learning_rate": 1.8572498552881413e-05, + "loss": 0.8399, + "step": 1040 + }, + { + "epoch": 0.2, + "learning_rate": 1.856933569428644e-05, + "loss": 0.819, + "step": 1041 + }, + { + "epoch": 0.2, + "learning_rate": 1.8566169605590824e-05, + "loss": 0.8486, + "step": 1042 + }, + { + "epoch": 0.2, + "learning_rate": 1.856300028798798e-05, + "loss": 0.8941, + "step": 1043 + }, + { + "epoch": 0.2, + "learning_rate": 1.8559827742672547e-05, + "loss": 0.8009, + "step": 1044 + }, + { + "epoch": 0.2, + "learning_rate": 1.8556651970840372e-05, + "loss": 0.8743, + "step": 1045 + }, + { + "epoch": 0.2, + "learning_rate": 1.8553472973688525e-05, + "loss": 0.8172, + "step": 1046 + }, + { + "epoch": 0.2, + "learning_rate": 1.8550290752415278e-05, + "loss": 0.8994, + "step": 1047 + }, + { + "epoch": 0.2, + "learning_rate": 1.8547105308220142e-05, + "loss": 0.8587, + "step": 1048 + }, + { + "epoch": 0.2, + "learning_rate": 1.854391664230382e-05, + "loss": 0.8718, + "step": 1049 + }, + { + "epoch": 0.2, + "learning_rate": 1.854072475586824e-05, + "loss": 0.8862, + "step": 1050 + }, + { + "epoch": 0.2, + "learning_rate": 1.8537529650116543e-05, + "loss": 0.8642, + "step": 1051 + }, + { + "epoch": 0.2, + "learning_rate": 1.8534331326253084e-05, + "loss": 0.8174, + "step": 1052 + }, + { + "epoch": 0.2, + "learning_rate": 1.8531129785483432e-05, + "loss": 0.8886, + "step": 1053 + }, + { + "epoch": 0.2, + "learning_rate": 1.8527925029014362e-05, + "loss": 0.8545, + "step": 1054 + }, + { + "epoch": 0.2, + "learning_rate": 1.852471705805387e-05, + "loss": 0.8819, + "step": 1055 + }, + { + "epoch": 0.2, + "learning_rate": 1.852150587381115e-05, + "loss": 0.7699, + "step": 1056 + }, + { + "epoch": 0.2, + "learning_rate": 1.8518291477496627e-05, + "loss": 0.8842, + "step": 1057 + }, + { + "epoch": 0.2, + "learning_rate": 1.851507387032192e-05, + "loss": 0.8609, + "step": 1058 + }, + { + "epoch": 0.2, + "learning_rate": 1.851185305349987e-05, + "loss": 0.9348, + "step": 1059 + }, + { + "epoch": 0.2, + "learning_rate": 1.850862902824452e-05, + "loss": 0.8573, + "step": 1060 + }, + { + "epoch": 0.2, + "learning_rate": 1.850540179577112e-05, + "loss": 0.8796, + "step": 1061 + }, + { + "epoch": 0.2, + "learning_rate": 1.8502171357296144e-05, + "loss": 0.8679, + "step": 1062 + }, + { + "epoch": 0.2, + "learning_rate": 1.8498937714037253e-05, + "loss": 0.9008, + "step": 1063 + }, + { + "epoch": 0.2, + "learning_rate": 1.8495700867213334e-05, + "loss": 0.8846, + "step": 1064 + }, + { + "epoch": 0.2, + "learning_rate": 1.8492460818044475e-05, + "loss": 0.8282, + "step": 1065 + }, + { + "epoch": 0.2, + "learning_rate": 1.8489217567751963e-05, + "loss": 0.2503, + "step": 1066 + }, + { + "epoch": 0.2, + "learning_rate": 1.8485971117558313e-05, + "loss": 0.8344, + "step": 1067 + }, + { + "epoch": 0.2, + "learning_rate": 1.848272146868722e-05, + "loss": 0.8429, + "step": 1068 + }, + { + "epoch": 0.2, + "learning_rate": 1.84794686223636e-05, + "loss": 0.837, + "step": 1069 + }, + { + "epoch": 0.2, + "learning_rate": 1.8476212579813575e-05, + "loss": 0.8473, + "step": 1070 + }, + { + "epoch": 0.2, + "learning_rate": 1.8472953342264464e-05, + "loss": 0.8321, + "step": 1071 + }, + { + "epoch": 0.2, + "learning_rate": 1.84696909109448e-05, + "loss": 0.8711, + "step": 1072 + }, + { + "epoch": 0.2, + "learning_rate": 1.8466425287084316e-05, + "loss": 0.8571, + "step": 1073 + }, + { + "epoch": 0.2, + "learning_rate": 1.8463156471913937e-05, + "loss": 0.8744, + "step": 1074 + }, + { + "epoch": 0.2, + "learning_rate": 1.8459884466665808e-05, + "loss": 0.8393, + "step": 1075 + }, + { + "epoch": 0.2, + "learning_rate": 1.8456609272573268e-05, + "loss": 0.8647, + "step": 1076 + }, + { + "epoch": 0.2, + "learning_rate": 1.8453330890870857e-05, + "loss": 0.8341, + "step": 1077 + }, + { + "epoch": 0.2, + "learning_rate": 1.8450049322794317e-05, + "loss": 0.8647, + "step": 1078 + }, + { + "epoch": 0.2, + "learning_rate": 1.8446764569580603e-05, + "loss": 0.9102, + "step": 1079 + }, + { + "epoch": 0.2, + "learning_rate": 1.844347663246785e-05, + "loss": 0.9067, + "step": 1080 + }, + { + "epoch": 0.2, + "learning_rate": 1.844018551269541e-05, + "loss": 0.8603, + "step": 1081 + }, + { + "epoch": 0.21, + "learning_rate": 1.8436891211503824e-05, + "loss": 0.9009, + "step": 1082 + }, + { + "epoch": 0.21, + "learning_rate": 1.8433593730134835e-05, + "loss": 0.8444, + "step": 1083 + }, + { + "epoch": 0.21, + "learning_rate": 1.8430293069831388e-05, + "loss": 0.8385, + "step": 1084 + }, + { + "epoch": 0.21, + "learning_rate": 1.842698923183763e-05, + "loss": 0.8159, + "step": 1085 + }, + { + "epoch": 0.21, + "learning_rate": 1.8423682217398892e-05, + "loss": 0.891, + "step": 1086 + }, + { + "epoch": 0.21, + "learning_rate": 1.8420372027761712e-05, + "loss": 0.8525, + "step": 1087 + }, + { + "epoch": 0.21, + "learning_rate": 1.8417058664173827e-05, + "loss": 0.8342, + "step": 1088 + }, + { + "epoch": 0.21, + "learning_rate": 1.8413742127884165e-05, + "loss": 0.857, + "step": 1089 + }, + { + "epoch": 0.21, + "learning_rate": 1.841042242014285e-05, + "loss": 0.8919, + "step": 1090 + }, + { + "epoch": 0.21, + "learning_rate": 1.8407099542201208e-05, + "loss": 0.8406, + "step": 1091 + }, + { + "epoch": 0.21, + "learning_rate": 1.8403773495311746e-05, + "loss": 0.839, + "step": 1092 + }, + { + "epoch": 0.21, + "learning_rate": 1.8400444280728184e-05, + "loss": 0.8371, + "step": 1093 + }, + { + "epoch": 0.21, + "learning_rate": 1.839711189970542e-05, + "loss": 0.2674, + "step": 1094 + }, + { + "epoch": 0.21, + "learning_rate": 1.839377635349956e-05, + "loss": 0.8676, + "step": 1095 + }, + { + "epoch": 0.21, + "learning_rate": 1.839043764336788e-05, + "loss": 0.8197, + "step": 1096 + }, + { + "epoch": 0.21, + "learning_rate": 1.838709577056888e-05, + "loss": 0.9236, + "step": 1097 + }, + { + "epoch": 0.21, + "learning_rate": 1.8383750736362226e-05, + "loss": 0.8437, + "step": 1098 + }, + { + "epoch": 0.21, + "learning_rate": 1.838040254200879e-05, + "loss": 0.8698, + "step": 1099 + }, + { + "epoch": 0.21, + "learning_rate": 1.837705118877063e-05, + "loss": 0.8544, + "step": 1100 + }, + { + "epoch": 0.21, + "learning_rate": 1.8373696677910997e-05, + "loss": 0.8402, + "step": 1101 + }, + { + "epoch": 0.21, + "learning_rate": 1.8370339010694326e-05, + "loss": 0.7814, + "step": 1102 + }, + { + "epoch": 0.21, + "learning_rate": 1.836697818838625e-05, + "loss": 0.7925, + "step": 1103 + }, + { + "epoch": 0.21, + "learning_rate": 1.8363614212253585e-05, + "loss": 0.8958, + "step": 1104 + }, + { + "epoch": 0.21, + "learning_rate": 1.8360247083564343e-05, + "loss": 0.8628, + "step": 1105 + }, + { + "epoch": 0.21, + "learning_rate": 1.8356876803587714e-05, + "loss": 0.8253, + "step": 1106 + }, + { + "epoch": 0.21, + "learning_rate": 1.8353503373594084e-05, + "loss": 0.8158, + "step": 1107 + }, + { + "epoch": 0.21, + "learning_rate": 1.8350126794855027e-05, + "loss": 0.8359, + "step": 1108 + }, + { + "epoch": 0.21, + "learning_rate": 1.8346747068643296e-05, + "loss": 0.8411, + "step": 1109 + }, + { + "epoch": 0.21, + "learning_rate": 1.8343364196232837e-05, + "loss": 0.8979, + "step": 1110 + }, + { + "epoch": 0.21, + "learning_rate": 1.833997817889878e-05, + "loss": 0.8245, + "step": 1111 + }, + { + "epoch": 0.21, + "learning_rate": 1.8336589017917444e-05, + "loss": 0.8483, + "step": 1112 + }, + { + "epoch": 0.21, + "learning_rate": 1.8333196714566327e-05, + "loss": 0.8734, + "step": 1113 + }, + { + "epoch": 0.21, + "learning_rate": 1.832980127012411e-05, + "loss": 0.8352, + "step": 1114 + }, + { + "epoch": 0.21, + "learning_rate": 1.832640268587067e-05, + "loss": 0.8654, + "step": 1115 + }, + { + "epoch": 0.21, + "learning_rate": 1.8323000963087053e-05, + "loss": 0.7958, + "step": 1116 + }, + { + "epoch": 0.21, + "learning_rate": 1.8319596103055496e-05, + "loss": 0.2913, + "step": 1117 + }, + { + "epoch": 0.21, + "learning_rate": 1.8316188107059418e-05, + "loss": 0.8657, + "step": 1118 + }, + { + "epoch": 0.21, + "learning_rate": 1.831277697638342e-05, + "loss": 0.8799, + "step": 1119 + }, + { + "epoch": 0.21, + "learning_rate": 1.8309362712313283e-05, + "loss": 0.8649, + "step": 1120 + }, + { + "epoch": 0.21, + "learning_rate": 1.830594531613597e-05, + "loss": 0.8323, + "step": 1121 + }, + { + "epoch": 0.21, + "learning_rate": 1.8302524789139623e-05, + "loss": 0.8774, + "step": 1122 + }, + { + "epoch": 0.21, + "learning_rate": 1.8299101132613567e-05, + "loss": 0.7932, + "step": 1123 + }, + { + "epoch": 0.21, + "learning_rate": 1.8295674347848304e-05, + "loss": 0.8074, + "step": 1124 + }, + { + "epoch": 0.21, + "learning_rate": 1.8292244436135517e-05, + "loss": 0.8107, + "step": 1125 + }, + { + "epoch": 0.21, + "learning_rate": 1.828881139876807e-05, + "loss": 0.8559, + "step": 1126 + }, + { + "epoch": 0.21, + "learning_rate": 1.8285375237039996e-05, + "loss": 0.9781, + "step": 1127 + }, + { + "epoch": 0.21, + "learning_rate": 1.828193595224652e-05, + "loss": 0.8525, + "step": 1128 + }, + { + "epoch": 0.21, + "learning_rate": 1.8278493545684028e-05, + "loss": 0.839, + "step": 1129 + }, + { + "epoch": 0.21, + "learning_rate": 1.8275048018650095e-05, + "loss": 0.854, + "step": 1130 + }, + { + "epoch": 0.21, + "learning_rate": 1.8271599372443467e-05, + "loss": 0.8281, + "step": 1131 + }, + { + "epoch": 0.21, + "learning_rate": 1.8268147608364068e-05, + "loss": 0.8188, + "step": 1132 + }, + { + "epoch": 0.21, + "learning_rate": 1.826469272771299e-05, + "loss": 0.8005, + "step": 1133 + }, + { + "epoch": 0.21, + "learning_rate": 1.8261234731792516e-05, + "loss": 0.8492, + "step": 1134 + }, + { + "epoch": 0.22, + "learning_rate": 1.8257773621906087e-05, + "loss": 0.8366, + "step": 1135 + }, + { + "epoch": 0.22, + "learning_rate": 1.8254309399358322e-05, + "loss": 0.7937, + "step": 1136 + }, + { + "epoch": 0.22, + "learning_rate": 1.825084206545502e-05, + "loss": 0.8583, + "step": 1137 + }, + { + "epoch": 0.22, + "learning_rate": 1.8247371621503145e-05, + "loss": 0.8592, + "step": 1138 + }, + { + "epoch": 0.22, + "learning_rate": 1.8243898068810833e-05, + "loss": 0.8378, + "step": 1139 + }, + { + "epoch": 0.22, + "learning_rate": 1.82404214086874e-05, + "loss": 0.8462, + "step": 1140 + }, + { + "epoch": 0.22, + "learning_rate": 1.823694164244333e-05, + "loss": 0.781, + "step": 1141 + }, + { + "epoch": 0.22, + "learning_rate": 1.8233458771390265e-05, + "loss": 0.2907, + "step": 1142 + }, + { + "epoch": 0.22, + "learning_rate": 1.8229972796841044e-05, + "loss": 0.8731, + "step": 1143 + }, + { + "epoch": 0.22, + "learning_rate": 1.822648372010965e-05, + "loss": 0.2715, + "step": 1144 + }, + { + "epoch": 0.22, + "learning_rate": 1.8222991542511247e-05, + "loss": 0.8881, + "step": 1145 + }, + { + "epoch": 0.22, + "learning_rate": 1.8219496265362164e-05, + "loss": 0.8414, + "step": 1146 + }, + { + "epoch": 0.22, + "learning_rate": 1.8215997889979904e-05, + "loss": 0.8728, + "step": 1147 + }, + { + "epoch": 0.22, + "learning_rate": 1.8212496417683135e-05, + "loss": 0.8166, + "step": 1148 + }, + { + "epoch": 0.22, + "learning_rate": 1.8208991849791693e-05, + "loss": 0.9737, + "step": 1149 + }, + { + "epoch": 0.22, + "learning_rate": 1.8205484187626576e-05, + "loss": 0.8391, + "step": 1150 + }, + { + "epoch": 0.22, + "learning_rate": 1.8201973432509957e-05, + "loss": 0.8611, + "step": 1151 + }, + { + "epoch": 0.22, + "learning_rate": 1.8198459585765164e-05, + "loss": 0.8334, + "step": 1152 + }, + { + "epoch": 0.22, + "learning_rate": 1.81949426487167e-05, + "loss": 0.2796, + "step": 1153 + }, + { + "epoch": 0.22, + "learning_rate": 1.8191422622690224e-05, + "loss": 0.802, + "step": 1154 + }, + { + "epoch": 0.22, + "learning_rate": 1.818789950901257e-05, + "loss": 0.7681, + "step": 1155 + }, + { + "epoch": 0.22, + "learning_rate": 1.818437330901173e-05, + "loss": 0.8318, + "step": 1156 + }, + { + "epoch": 0.22, + "learning_rate": 1.8180844024016858e-05, + "loss": 0.8236, + "step": 1157 + }, + { + "epoch": 0.22, + "learning_rate": 1.817731165535827e-05, + "loss": 0.8146, + "step": 1158 + }, + { + "epoch": 0.22, + "learning_rate": 1.8173776204367454e-05, + "loss": 0.8735, + "step": 1159 + }, + { + "epoch": 0.22, + "learning_rate": 1.8170237672377046e-05, + "loss": 0.8407, + "step": 1160 + }, + { + "epoch": 0.22, + "learning_rate": 1.8166696060720854e-05, + "loss": 0.8749, + "step": 1161 + }, + { + "epoch": 0.22, + "learning_rate": 1.8163151370733838e-05, + "loss": 0.9348, + "step": 1162 + }, + { + "epoch": 0.22, + "learning_rate": 1.8159603603752128e-05, + "loss": 0.8002, + "step": 1163 + }, + { + "epoch": 0.22, + "learning_rate": 1.8156052761113007e-05, + "loss": 0.9159, + "step": 1164 + }, + { + "epoch": 0.22, + "learning_rate": 1.815249884415492e-05, + "loss": 0.8358, + "step": 1165 + }, + { + "epoch": 0.22, + "learning_rate": 1.814894185421747e-05, + "loss": 0.8129, + "step": 1166 + }, + { + "epoch": 0.22, + "learning_rate": 1.814538179264142e-05, + "loss": 0.8807, + "step": 1167 + }, + { + "epoch": 0.22, + "learning_rate": 1.8141818660768687e-05, + "loss": 0.8927, + "step": 1168 + }, + { + "epoch": 0.22, + "learning_rate": 1.8138252459942347e-05, + "loss": 0.7865, + "step": 1169 + }, + { + "epoch": 0.22, + "learning_rate": 1.8134683191506636e-05, + "loss": 0.8413, + "step": 1170 + }, + { + "epoch": 0.22, + "learning_rate": 1.8131110856806942e-05, + "loss": 0.7926, + "step": 1171 + }, + { + "epoch": 0.22, + "learning_rate": 1.8127535457189812e-05, + "loss": 0.8228, + "step": 1172 + }, + { + "epoch": 0.22, + "learning_rate": 1.8123956994002946e-05, + "loss": 0.793, + "step": 1173 + }, + { + "epoch": 0.22, + "learning_rate": 1.81203754685952e-05, + "loss": 0.8539, + "step": 1174 + }, + { + "epoch": 0.22, + "learning_rate": 1.811679088231658e-05, + "loss": 0.8724, + "step": 1175 + }, + { + "epoch": 0.22, + "learning_rate": 1.8113203236518256e-05, + "loss": 0.8441, + "step": 1176 + }, + { + "epoch": 0.22, + "learning_rate": 1.8109612532552545e-05, + "loss": 0.8286, + "step": 1177 + }, + { + "epoch": 0.22, + "learning_rate": 1.8106018771772906e-05, + "loss": 0.8846, + "step": 1178 + }, + { + "epoch": 0.22, + "learning_rate": 1.8102421955533974e-05, + "loss": 0.8149, + "step": 1179 + }, + { + "epoch": 0.22, + "learning_rate": 1.8098822085191515e-05, + "loss": 0.8493, + "step": 1180 + }, + { + "epoch": 0.22, + "learning_rate": 1.8095219162102453e-05, + "loss": 0.8174, + "step": 1181 + }, + { + "epoch": 0.22, + "learning_rate": 1.809161318762487e-05, + "loss": 0.8117, + "step": 1182 + }, + { + "epoch": 0.22, + "learning_rate": 1.8088004163117984e-05, + "loss": 0.2805, + "step": 1183 + }, + { + "epoch": 0.22, + "learning_rate": 1.8084392089942174e-05, + "loss": 0.89, + "step": 1184 + }, + { + "epoch": 0.22, + "learning_rate": 1.8080776969458965e-05, + "loss": 0.2803, + "step": 1185 + }, + { + "epoch": 0.22, + "learning_rate": 1.807715880303103e-05, + "loss": 0.8939, + "step": 1186 + }, + { + "epoch": 0.22, + "learning_rate": 1.807353759202219e-05, + "loss": 0.8533, + "step": 1187 + }, + { + "epoch": 0.23, + "learning_rate": 1.8069913337797414e-05, + "loss": 0.801, + "step": 1188 + }, + { + "epoch": 0.23, + "learning_rate": 1.8066286041722812e-05, + "loss": 0.8913, + "step": 1189 + }, + { + "epoch": 0.23, + "learning_rate": 1.8062655705165658e-05, + "loss": 0.8939, + "step": 1190 + }, + { + "epoch": 0.23, + "learning_rate": 1.805902232949435e-05, + "loss": 0.799, + "step": 1191 + }, + { + "epoch": 0.23, + "learning_rate": 1.805538591607845e-05, + "loss": 0.8306, + "step": 1192 + }, + { + "epoch": 0.23, + "learning_rate": 1.805174646628865e-05, + "loss": 0.8297, + "step": 1193 + }, + { + "epoch": 0.23, + "learning_rate": 1.8048103981496798e-05, + "loss": 0.7777, + "step": 1194 + }, + { + "epoch": 0.23, + "learning_rate": 1.804445846307588e-05, + "loss": 0.2711, + "step": 1195 + }, + { + "epoch": 0.23, + "learning_rate": 1.804080991240003e-05, + "loss": 0.8174, + "step": 1196 + }, + { + "epoch": 0.23, + "learning_rate": 1.803715833084452e-05, + "loss": 0.8584, + "step": 1197 + }, + { + "epoch": 0.23, + "learning_rate": 1.8033503719785767e-05, + "loss": 0.8459, + "step": 1198 + }, + { + "epoch": 0.23, + "learning_rate": 1.8029846080601328e-05, + "loss": 0.8737, + "step": 1199 + }, + { + "epoch": 0.23, + "learning_rate": 1.802618541466991e-05, + "loss": 0.8804, + "step": 1200 + }, + { + "epoch": 0.23, + "learning_rate": 1.802252172337134e-05, + "loss": 0.8216, + "step": 1201 + }, + { + "epoch": 0.23, + "learning_rate": 1.801885500808661e-05, + "loss": 0.8376, + "step": 1202 + }, + { + "epoch": 0.23, + "learning_rate": 1.801518527019784e-05, + "loss": 0.8017, + "step": 1203 + }, + { + "epoch": 0.23, + "learning_rate": 1.801151251108829e-05, + "loss": 0.8046, + "step": 1204 + }, + { + "epoch": 0.23, + "learning_rate": 1.8007836732142356e-05, + "loss": 0.8198, + "step": 1205 + }, + { + "epoch": 0.23, + "learning_rate": 1.800415793474558e-05, + "loss": 0.8041, + "step": 1206 + }, + { + "epoch": 0.23, + "learning_rate": 1.800047612028463e-05, + "loss": 0.8132, + "step": 1207 + }, + { + "epoch": 0.23, + "learning_rate": 1.7996791290147327e-05, + "loss": 0.8217, + "step": 1208 + }, + { + "epoch": 0.23, + "learning_rate": 1.7993103445722615e-05, + "loss": 0.8217, + "step": 1209 + }, + { + "epoch": 0.23, + "learning_rate": 1.7989412588400582e-05, + "loss": 0.7922, + "step": 1210 + }, + { + "epoch": 0.23, + "learning_rate": 1.7985718719572448e-05, + "loss": 0.8307, + "step": 1211 + }, + { + "epoch": 0.23, + "learning_rate": 1.7982021840630566e-05, + "loss": 0.8035, + "step": 1212 + }, + { + "epoch": 0.23, + "learning_rate": 1.7978321952968435e-05, + "loss": 0.842, + "step": 1213 + }, + { + "epoch": 0.23, + "learning_rate": 1.7974619057980677e-05, + "loss": 0.8326, + "step": 1214 + }, + { + "epoch": 0.23, + "learning_rate": 1.7970913157063042e-05, + "loss": 0.8819, + "step": 1215 + }, + { + "epoch": 0.23, + "learning_rate": 1.7967204251612432e-05, + "loss": 0.8351, + "step": 1216 + }, + { + "epoch": 0.23, + "learning_rate": 1.7963492343026866e-05, + "loss": 0.8438, + "step": 1217 + }, + { + "epoch": 0.23, + "learning_rate": 1.7959777432705505e-05, + "loss": 0.8284, + "step": 1218 + }, + { + "epoch": 0.23, + "learning_rate": 1.795605952204863e-05, + "loss": 0.8165, + "step": 1219 + }, + { + "epoch": 0.23, + "learning_rate": 1.7952338612457664e-05, + "loss": 0.8232, + "step": 1220 + }, + { + "epoch": 0.23, + "learning_rate": 1.794861470533515e-05, + "loss": 0.8665, + "step": 1221 + }, + { + "epoch": 0.23, + "learning_rate": 1.7944887802084777e-05, + "loss": 0.867, + "step": 1222 + }, + { + "epoch": 0.23, + "learning_rate": 1.7941157904111346e-05, + "loss": 0.7979, + "step": 1223 + }, + { + "epoch": 0.23, + "learning_rate": 1.7937425012820796e-05, + "loss": 0.2724, + "step": 1224 + }, + { + "epoch": 0.23, + "learning_rate": 1.793368912962019e-05, + "loss": 0.8101, + "step": 1225 + }, + { + "epoch": 0.23, + "learning_rate": 1.7929950255917723e-05, + "loss": 0.8544, + "step": 1226 + }, + { + "epoch": 0.23, + "learning_rate": 1.7926208393122712e-05, + "loss": 0.9194, + "step": 1227 + }, + { + "epoch": 0.23, + "learning_rate": 1.7922463542645612e-05, + "loss": 0.8031, + "step": 1228 + }, + { + "epoch": 0.23, + "learning_rate": 1.7918715705897986e-05, + "loss": 0.9085, + "step": 1229 + }, + { + "epoch": 0.23, + "learning_rate": 1.7914964884292543e-05, + "loss": 0.7578, + "step": 1230 + }, + { + "epoch": 0.23, + "learning_rate": 1.79112110792431e-05, + "loss": 0.8749, + "step": 1231 + }, + { + "epoch": 0.23, + "learning_rate": 1.7907454292164606e-05, + "loss": 0.8497, + "step": 1232 + }, + { + "epoch": 0.23, + "learning_rate": 1.7903694524473134e-05, + "loss": 0.8648, + "step": 1233 + }, + { + "epoch": 0.23, + "learning_rate": 1.789993177758588e-05, + "loss": 0.867, + "step": 1234 + }, + { + "epoch": 0.23, + "learning_rate": 1.7896166052921165e-05, + "loss": 0.8539, + "step": 1235 + }, + { + "epoch": 0.23, + "learning_rate": 1.789239735189843e-05, + "loss": 0.8197, + "step": 1236 + }, + { + "epoch": 0.23, + "learning_rate": 1.7888625675938237e-05, + "loss": 0.8552, + "step": 1237 + }, + { + "epoch": 0.23, + "learning_rate": 1.7884851026462266e-05, + "loss": 0.7848, + "step": 1238 + }, + { + "epoch": 0.23, + "learning_rate": 1.788107340489333e-05, + "loss": 0.8418, + "step": 1239 + }, + { + "epoch": 0.24, + "learning_rate": 1.7877292812655353e-05, + "loss": 0.8162, + "step": 1240 + }, + { + "epoch": 0.24, + "learning_rate": 1.787350925117338e-05, + "loss": 0.8657, + "step": 1241 + }, + { + "epoch": 0.24, + "learning_rate": 1.7869722721873573e-05, + "loss": 0.887, + "step": 1242 + }, + { + "epoch": 0.24, + "learning_rate": 1.786593322618322e-05, + "loss": 0.8948, + "step": 1243 + }, + { + "epoch": 0.24, + "learning_rate": 1.7862140765530718e-05, + "loss": 0.2576, + "step": 1244 + }, + { + "epoch": 0.24, + "learning_rate": 1.785834534134559e-05, + "loss": 0.8269, + "step": 1245 + }, + { + "epoch": 0.24, + "learning_rate": 1.785454695505847e-05, + "loss": 0.8198, + "step": 1246 + }, + { + "epoch": 0.24, + "learning_rate": 1.785074560810111e-05, + "loss": 0.2742, + "step": 1247 + }, + { + "epoch": 0.24, + "learning_rate": 1.784694130190638e-05, + "loss": 0.8338, + "step": 1248 + }, + { + "epoch": 0.24, + "learning_rate": 1.7843134037908267e-05, + "loss": 0.8831, + "step": 1249 + }, + { + "epoch": 0.24, + "learning_rate": 1.7839323817541866e-05, + "loss": 0.7937, + "step": 1250 + }, + { + "epoch": 0.24, + "learning_rate": 1.783551064224339e-05, + "loss": 0.819, + "step": 1251 + }, + { + "epoch": 0.24, + "learning_rate": 1.783169451345017e-05, + "loss": 0.8932, + "step": 1252 + }, + { + "epoch": 0.24, + "learning_rate": 1.7827875432600645e-05, + "loss": 0.8993, + "step": 1253 + }, + { + "epoch": 0.24, + "learning_rate": 1.7824053401134366e-05, + "loss": 0.2737, + "step": 1254 + }, + { + "epoch": 0.24, + "learning_rate": 1.7820228420492003e-05, + "loss": 0.873, + "step": 1255 + }, + { + "epoch": 0.24, + "learning_rate": 1.781640049211533e-05, + "loss": 0.7887, + "step": 1256 + }, + { + "epoch": 0.24, + "learning_rate": 1.7812569617447233e-05, + "loss": 0.8013, + "step": 1257 + }, + { + "epoch": 0.24, + "learning_rate": 1.7808735797931715e-05, + "loss": 0.812, + "step": 1258 + }, + { + "epoch": 0.24, + "learning_rate": 1.7804899035013886e-05, + "loss": 0.8391, + "step": 1259 + }, + { + "epoch": 0.24, + "learning_rate": 1.780105933013996e-05, + "loss": 0.8281, + "step": 1260 + }, + { + "epoch": 0.24, + "learning_rate": 1.7797216684757267e-05, + "loss": 0.8788, + "step": 1261 + }, + { + "epoch": 0.24, + "learning_rate": 1.779337110031424e-05, + "loss": 0.8887, + "step": 1262 + }, + { + "epoch": 0.24, + "learning_rate": 1.7789522578260428e-05, + "loss": 0.846, + "step": 1263 + }, + { + "epoch": 0.24, + "learning_rate": 1.7785671120046472e-05, + "loss": 0.8181, + "step": 1264 + }, + { + "epoch": 0.24, + "learning_rate": 1.7781816727124138e-05, + "loss": 0.8736, + "step": 1265 + }, + { + "epoch": 0.24, + "learning_rate": 1.777795940094629e-05, + "loss": 0.8023, + "step": 1266 + }, + { + "epoch": 0.24, + "learning_rate": 1.7774099142966893e-05, + "loss": 0.8689, + "step": 1267 + }, + { + "epoch": 0.24, + "learning_rate": 1.7770235954641023e-05, + "loss": 0.7686, + "step": 1268 + }, + { + "epoch": 0.24, + "learning_rate": 1.776636983742486e-05, + "loss": 0.8736, + "step": 1269 + }, + { + "epoch": 0.24, + "learning_rate": 1.7762500792775687e-05, + "loss": 0.8553, + "step": 1270 + }, + { + "epoch": 0.24, + "learning_rate": 1.775862882215189e-05, + "loss": 0.8528, + "step": 1271 + }, + { + "epoch": 0.24, + "learning_rate": 1.7754753927012955e-05, + "loss": 0.8183, + "step": 1272 + }, + { + "epoch": 0.24, + "learning_rate": 1.775087610881948e-05, + "loss": 0.8841, + "step": 1273 + }, + { + "epoch": 0.24, + "learning_rate": 1.7746995369033154e-05, + "loss": 0.8773, + "step": 1274 + }, + { + "epoch": 0.24, + "learning_rate": 1.7743111709116772e-05, + "loss": 0.8124, + "step": 1275 + }, + { + "epoch": 0.24, + "learning_rate": 1.7739225130534235e-05, + "loss": 0.8876, + "step": 1276 + }, + { + "epoch": 0.24, + "learning_rate": 1.773533563475053e-05, + "loss": 0.8516, + "step": 1277 + }, + { + "epoch": 0.24, + "learning_rate": 1.773144322323176e-05, + "loss": 0.8646, + "step": 1278 + }, + { + "epoch": 0.24, + "learning_rate": 1.7727547897445117e-05, + "loss": 0.8007, + "step": 1279 + }, + { + "epoch": 0.24, + "learning_rate": 1.7723649658858894e-05, + "loss": 0.8451, + "step": 1280 + }, + { + "epoch": 0.24, + "learning_rate": 1.771974850894248e-05, + "loss": 0.8471, + "step": 1281 + }, + { + "epoch": 0.24, + "learning_rate": 1.7715844449166364e-05, + "loss": 0.7847, + "step": 1282 + }, + { + "epoch": 0.24, + "learning_rate": 1.771193748100213e-05, + "loss": 0.8328, + "step": 1283 + }, + { + "epoch": 0.24, + "learning_rate": 1.770802760592246e-05, + "loss": 0.8101, + "step": 1284 + }, + { + "epoch": 0.24, + "learning_rate": 1.7704114825401132e-05, + "loss": 0.8421, + "step": 1285 + }, + { + "epoch": 0.24, + "learning_rate": 1.770019914091302e-05, + "loss": 0.7865, + "step": 1286 + }, + { + "epoch": 0.24, + "learning_rate": 1.7696280553934086e-05, + "loss": 0.8561, + "step": 1287 + }, + { + "epoch": 0.24, + "learning_rate": 1.7692359065941394e-05, + "loss": 0.7746, + "step": 1288 + }, + { + "epoch": 0.24, + "learning_rate": 1.76884346784131e-05, + "loss": 0.7947, + "step": 1289 + }, + { + "epoch": 0.24, + "learning_rate": 1.768450739282845e-05, + "loss": 0.8127, + "step": 1290 + }, + { + "epoch": 0.24, + "learning_rate": 1.768057721066778e-05, + "loss": 0.8582, + "step": 1291 + }, + { + "epoch": 0.24, + "learning_rate": 1.767664413341252e-05, + "loss": 0.88, + "step": 1292 + }, + { + "epoch": 0.25, + "learning_rate": 1.76727081625452e-05, + "loss": 0.8446, + "step": 1293 + }, + { + "epoch": 0.25, + "learning_rate": 1.7668769299549432e-05, + "loss": 0.8867, + "step": 1294 + }, + { + "epoch": 0.25, + "learning_rate": 1.766482754590992e-05, + "loss": 0.8342, + "step": 1295 + }, + { + "epoch": 0.25, + "learning_rate": 1.766088290311245e-05, + "loss": 0.8822, + "step": 1296 + }, + { + "epoch": 0.25, + "learning_rate": 1.7656935372643915e-05, + "loss": 0.852, + "step": 1297 + }, + { + "epoch": 0.25, + "learning_rate": 1.7652984955992275e-05, + "loss": 0.8547, + "step": 1298 + }, + { + "epoch": 0.25, + "learning_rate": 1.76490316546466e-05, + "loss": 0.8581, + "step": 1299 + }, + { + "epoch": 0.25, + "learning_rate": 1.7645075470097024e-05, + "loss": 0.8293, + "step": 1300 + }, + { + "epoch": 0.25, + "learning_rate": 1.764111640383479e-05, + "loss": 0.8912, + "step": 1301 + }, + { + "epoch": 0.25, + "learning_rate": 1.763715445735221e-05, + "loss": 0.2896, + "step": 1302 + }, + { + "epoch": 0.25, + "learning_rate": 1.7633189632142693e-05, + "loss": 0.8347, + "step": 1303 + }, + { + "epoch": 0.25, + "learning_rate": 1.7629221929700725e-05, + "loss": 0.8414, + "step": 1304 + }, + { + "epoch": 0.25, + "learning_rate": 1.7625251351521883e-05, + "loss": 0.8742, + "step": 1305 + }, + { + "epoch": 0.25, + "learning_rate": 1.7621277899102828e-05, + "loss": 0.8657, + "step": 1306 + }, + { + "epoch": 0.25, + "learning_rate": 1.7617301573941296e-05, + "loss": 0.813, + "step": 1307 + }, + { + "epoch": 0.25, + "learning_rate": 1.7613322377536117e-05, + "loss": 0.7997, + "step": 1308 + }, + { + "epoch": 0.25, + "learning_rate": 1.760934031138719e-05, + "loss": 0.8029, + "step": 1309 + }, + { + "epoch": 0.25, + "learning_rate": 1.760535537699551e-05, + "loss": 0.8873, + "step": 1310 + }, + { + "epoch": 0.25, + "learning_rate": 1.7601367575863146e-05, + "loss": 0.8791, + "step": 1311 + }, + { + "epoch": 0.25, + "learning_rate": 1.7597376909493244e-05, + "loss": 0.8793, + "step": 1312 + }, + { + "epoch": 0.25, + "learning_rate": 1.759338337939004e-05, + "loss": 0.8309, + "step": 1313 + }, + { + "epoch": 0.25, + "learning_rate": 1.758938698705884e-05, + "loss": 0.8732, + "step": 1314 + }, + { + "epoch": 0.25, + "learning_rate": 1.7585387734006033e-05, + "loss": 0.8834, + "step": 1315 + }, + { + "epoch": 0.25, + "learning_rate": 1.758138562173909e-05, + "loss": 0.8604, + "step": 1316 + }, + { + "epoch": 0.25, + "learning_rate": 1.7577380651766548e-05, + "loss": 0.836, + "step": 1317 + }, + { + "epoch": 0.25, + "learning_rate": 1.7573372825598037e-05, + "loss": 0.7953, + "step": 1318 + }, + { + "epoch": 0.25, + "learning_rate": 1.7569362144744247e-05, + "loss": 0.8302, + "step": 1319 + }, + { + "epoch": 0.25, + "learning_rate": 1.7565348610716963e-05, + "loss": 0.8495, + "step": 1320 + }, + { + "epoch": 0.25, + "learning_rate": 1.7561332225029022e-05, + "loss": 0.8666, + "step": 1321 + }, + { + "epoch": 0.25, + "learning_rate": 1.755731298919436e-05, + "loss": 0.8714, + "step": 1322 + }, + { + "epoch": 0.25, + "learning_rate": 1.755329090472797e-05, + "loss": 0.8369, + "step": 1323 + }, + { + "epoch": 0.25, + "learning_rate": 1.7549265973145927e-05, + "loss": 0.847, + "step": 1324 + }, + { + "epoch": 0.25, + "learning_rate": 1.754523819596538e-05, + "loss": 0.825, + "step": 1325 + }, + { + "epoch": 0.25, + "learning_rate": 1.7541207574704536e-05, + "loss": 0.8207, + "step": 1326 + }, + { + "epoch": 0.25, + "learning_rate": 1.75371741108827e-05, + "loss": 0.846, + "step": 1327 + }, + { + "epoch": 0.25, + "learning_rate": 1.7533137806020226e-05, + "loss": 0.8259, + "step": 1328 + }, + { + "epoch": 0.25, + "learning_rate": 1.7529098661638552e-05, + "loss": 0.2766, + "step": 1329 + }, + { + "epoch": 0.25, + "learning_rate": 1.7525056679260178e-05, + "loss": 0.7984, + "step": 1330 + }, + { + "epoch": 0.25, + "learning_rate": 1.752101186040868e-05, + "loss": 0.906, + "step": 1331 + }, + { + "epoch": 0.25, + "learning_rate": 1.7516964206608697e-05, + "loss": 0.7941, + "step": 1332 + }, + { + "epoch": 0.25, + "learning_rate": 1.751291371938594e-05, + "loss": 0.8234, + "step": 1333 + }, + { + "epoch": 0.25, + "learning_rate": 1.750886040026719e-05, + "loss": 0.885, + "step": 1334 + }, + { + "epoch": 0.25, + "learning_rate": 1.7504804250780292e-05, + "loss": 0.879, + "step": 1335 + }, + { + "epoch": 0.25, + "learning_rate": 1.750074527245416e-05, + "loss": 0.8387, + "step": 1336 + }, + { + "epoch": 0.25, + "learning_rate": 1.7496683466818775e-05, + "loss": 0.8588, + "step": 1337 + }, + { + "epoch": 0.25, + "learning_rate": 1.749261883540518e-05, + "loss": 0.8703, + "step": 1338 + }, + { + "epoch": 0.25, + "learning_rate": 1.7488551379745487e-05, + "loss": 0.8489, + "step": 1339 + }, + { + "epoch": 0.25, + "learning_rate": 1.748448110137287e-05, + "loss": 0.8757, + "step": 1340 + }, + { + "epoch": 0.25, + "learning_rate": 1.7480408001821563e-05, + "loss": 0.7955, + "step": 1341 + }, + { + "epoch": 0.25, + "learning_rate": 1.747633208262688e-05, + "loss": 0.7946, + "step": 1342 + }, + { + "epoch": 0.25, + "learning_rate": 1.7472253345325175e-05, + "loss": 0.8491, + "step": 1343 + }, + { + "epoch": 0.25, + "learning_rate": 1.746817179145388e-05, + "loss": 0.8601, + "step": 1344 + }, + { + "epoch": 0.25, + "learning_rate": 1.746408742255149e-05, + "loss": 0.7915, + "step": 1345 + }, + { + "epoch": 0.26, + "learning_rate": 1.746000024015755e-05, + "loss": 0.8365, + "step": 1346 + }, + { + "epoch": 0.26, + "learning_rate": 1.7455910245812665e-05, + "loss": 0.882, + "step": 1347 + }, + { + "epoch": 0.26, + "learning_rate": 1.745181744105851e-05, + "loss": 0.8537, + "step": 1348 + }, + { + "epoch": 0.26, + "learning_rate": 1.744772182743782e-05, + "loss": 0.8, + "step": 1349 + }, + { + "epoch": 0.26, + "learning_rate": 1.7443623406494377e-05, + "loss": 0.8267, + "step": 1350 + }, + { + "epoch": 0.26, + "learning_rate": 1.743952217977303e-05, + "loss": 0.8568, + "step": 1351 + }, + { + "epoch": 0.26, + "learning_rate": 1.743541814881969e-05, + "loss": 0.8379, + "step": 1352 + }, + { + "epoch": 0.26, + "learning_rate": 1.743131131518131e-05, + "loss": 0.8172, + "step": 1353 + }, + { + "epoch": 0.26, + "learning_rate": 1.7427201680405907e-05, + "loss": 0.8455, + "step": 1354 + }, + { + "epoch": 0.26, + "learning_rate": 1.7423089246042562e-05, + "loss": 0.7773, + "step": 1355 + }, + { + "epoch": 0.26, + "learning_rate": 1.74189740136414e-05, + "loss": 0.8312, + "step": 1356 + }, + { + "epoch": 0.26, + "learning_rate": 1.7414855984753605e-05, + "loss": 0.2956, + "step": 1357 + }, + { + "epoch": 0.26, + "learning_rate": 1.7410735160931417e-05, + "loss": 0.2737, + "step": 1358 + }, + { + "epoch": 0.26, + "learning_rate": 1.7406611543728122e-05, + "loss": 0.2694, + "step": 1359 + }, + { + "epoch": 0.26, + "learning_rate": 1.7402485134698073e-05, + "loss": 0.8211, + "step": 1360 + }, + { + "epoch": 0.26, + "learning_rate": 1.739835593539666e-05, + "loss": 0.8712, + "step": 1361 + }, + { + "epoch": 0.26, + "learning_rate": 1.7394223947380334e-05, + "loss": 0.8768, + "step": 1362 + }, + { + "epoch": 0.26, + "learning_rate": 1.7390089172206594e-05, + "loss": 0.2742, + "step": 1363 + }, + { + "epoch": 0.26, + "learning_rate": 1.7385951611433987e-05, + "loss": 0.9349, + "step": 1364 + }, + { + "epoch": 0.26, + "learning_rate": 1.738181126662212e-05, + "loss": 0.8258, + "step": 1365 + }, + { + "epoch": 0.26, + "learning_rate": 1.7377668139331638e-05, + "loss": 0.8495, + "step": 1366 + }, + { + "epoch": 0.26, + "learning_rate": 1.737352223112424e-05, + "loss": 0.8295, + "step": 1367 + }, + { + "epoch": 0.26, + "learning_rate": 1.7369373543562675e-05, + "loss": 0.8436, + "step": 1368 + }, + { + "epoch": 0.26, + "learning_rate": 1.7365222078210734e-05, + "loss": 0.859, + "step": 1369 + }, + { + "epoch": 0.26, + "learning_rate": 1.736106783663326e-05, + "loss": 0.8512, + "step": 1370 + }, + { + "epoch": 0.26, + "learning_rate": 1.735691082039614e-05, + "loss": 0.8463, + "step": 1371 + }, + { + "epoch": 0.26, + "learning_rate": 1.735275103106631e-05, + "loss": 0.8324, + "step": 1372 + }, + { + "epoch": 0.26, + "learning_rate": 1.7348588470211746e-05, + "loss": 0.8515, + "step": 1373 + }, + { + "epoch": 0.26, + "learning_rate": 1.7344423139401473e-05, + "loss": 0.8775, + "step": 1374 + }, + { + "epoch": 0.26, + "learning_rate": 1.734025504020556e-05, + "loss": 0.8528, + "step": 1375 + }, + { + "epoch": 0.26, + "learning_rate": 1.733608417419512e-05, + "loss": 0.8606, + "step": 1376 + }, + { + "epoch": 0.26, + "learning_rate": 1.7331910542942298e-05, + "loss": 0.883, + "step": 1377 + }, + { + "epoch": 0.26, + "learning_rate": 1.7327734148020298e-05, + "loss": 0.8643, + "step": 1378 + }, + { + "epoch": 0.26, + "learning_rate": 1.732355499100336e-05, + "loss": 0.9069, + "step": 1379 + }, + { + "epoch": 0.26, + "learning_rate": 1.7319373073466757e-05, + "loss": 0.8601, + "step": 1380 + }, + { + "epoch": 0.26, + "learning_rate": 1.7315188396986813e-05, + "loss": 0.2791, + "step": 1381 + }, + { + "epoch": 0.26, + "learning_rate": 1.7311000963140887e-05, + "loss": 0.8091, + "step": 1382 + }, + { + "epoch": 0.26, + "learning_rate": 1.7306810773507376e-05, + "loss": 0.8344, + "step": 1383 + }, + { + "epoch": 0.26, + "learning_rate": 1.7302617829665725e-05, + "loss": 0.8395, + "step": 1384 + }, + { + "epoch": 0.26, + "learning_rate": 1.7298422133196406e-05, + "loss": 0.8267, + "step": 1385 + }, + { + "epoch": 0.26, + "learning_rate": 1.7294223685680927e-05, + "loss": 0.2755, + "step": 1386 + }, + { + "epoch": 0.26, + "learning_rate": 1.7290022488701845e-05, + "loss": 0.8548, + "step": 1387 + }, + { + "epoch": 0.26, + "learning_rate": 1.7285818543842748e-05, + "loss": 0.2732, + "step": 1388 + }, + { + "epoch": 0.26, + "learning_rate": 1.728161185268826e-05, + "loss": 0.8858, + "step": 1389 + }, + { + "epoch": 0.26, + "learning_rate": 1.7277402416824026e-05, + "loss": 0.9048, + "step": 1390 + }, + { + "epoch": 0.26, + "learning_rate": 1.7273190237836757e-05, + "loss": 0.8416, + "step": 1391 + }, + { + "epoch": 0.26, + "learning_rate": 1.7268975317314167e-05, + "loss": 0.8174, + "step": 1392 + }, + { + "epoch": 0.26, + "learning_rate": 1.7264757656845018e-05, + "loss": 0.86, + "step": 1393 + }, + { + "epoch": 0.26, + "learning_rate": 1.726053725801911e-05, + "loss": 0.8161, + "step": 1394 + }, + { + "epoch": 0.26, + "learning_rate": 1.7256314122427257e-05, + "loss": 0.8819, + "step": 1395 + }, + { + "epoch": 0.26, + "learning_rate": 1.7252088251661326e-05, + "loss": 0.7983, + "step": 1396 + }, + { + "epoch": 0.26, + "learning_rate": 1.7247859647314195e-05, + "loss": 0.8224, + "step": 1397 + }, + { + "epoch": 0.26, + "learning_rate": 1.7243628310979793e-05, + "loss": 0.7218, + "step": 1398 + }, + { + "epoch": 0.27, + "learning_rate": 1.7239394244253053e-05, + "loss": 0.8584, + "step": 1399 + }, + { + "epoch": 0.27, + "learning_rate": 1.7235157448729967e-05, + "loss": 0.8263, + "step": 1400 + }, + { + "epoch": 0.27, + "learning_rate": 1.723091792600753e-05, + "loss": 0.2566, + "step": 1401 + }, + { + "epoch": 0.27, + "learning_rate": 1.7226675677683777e-05, + "loss": 0.8907, + "step": 1402 + }, + { + "epoch": 0.27, + "learning_rate": 1.7222430705357774e-05, + "loss": 0.8883, + "step": 1403 + }, + { + "epoch": 0.27, + "learning_rate": 1.7218183010629604e-05, + "loss": 0.826, + "step": 1404 + }, + { + "epoch": 0.27, + "learning_rate": 1.7213932595100384e-05, + "loss": 0.9159, + "step": 1405 + }, + { + "epoch": 0.27, + "learning_rate": 1.720967946037225e-05, + "loss": 0.8857, + "step": 1406 + }, + { + "epoch": 0.27, + "learning_rate": 1.7205423608048367e-05, + "loss": 0.8784, + "step": 1407 + }, + { + "epoch": 0.27, + "learning_rate": 1.7201165039732928e-05, + "loss": 0.8344, + "step": 1408 + }, + { + "epoch": 0.27, + "learning_rate": 1.719690375703114e-05, + "loss": 0.2685, + "step": 1409 + }, + { + "epoch": 0.27, + "learning_rate": 1.719263976154924e-05, + "loss": 0.7863, + "step": 1410 + }, + { + "epoch": 0.27, + "learning_rate": 1.7188373054894487e-05, + "loss": 0.8195, + "step": 1411 + }, + { + "epoch": 0.27, + "learning_rate": 1.7184103638675157e-05, + "loss": 0.2596, + "step": 1412 + }, + { + "epoch": 0.27, + "learning_rate": 1.717983151450056e-05, + "loss": 0.8559, + "step": 1413 + }, + { + "epoch": 0.27, + "learning_rate": 1.7175556683981014e-05, + "loss": 0.8465, + "step": 1414 + }, + { + "epoch": 0.27, + "learning_rate": 1.7171279148727858e-05, + "loss": 0.8098, + "step": 1415 + }, + { + "epoch": 0.27, + "learning_rate": 1.716699891035346e-05, + "loss": 0.8032, + "step": 1416 + }, + { + "epoch": 0.27, + "learning_rate": 1.716271597047119e-05, + "loss": 0.8333, + "step": 1417 + }, + { + "epoch": 0.27, + "learning_rate": 1.715843033069546e-05, + "loss": 0.8986, + "step": 1418 + }, + { + "epoch": 0.27, + "learning_rate": 1.715414199264168e-05, + "loss": 0.8274, + "step": 1419 + }, + { + "epoch": 0.27, + "learning_rate": 1.714985095792628e-05, + "loss": 0.7852, + "step": 1420 + }, + { + "epoch": 0.27, + "learning_rate": 1.714555722816672e-05, + "loss": 0.8401, + "step": 1421 + }, + { + "epoch": 0.27, + "learning_rate": 1.714126080498146e-05, + "loss": 0.8601, + "step": 1422 + }, + { + "epoch": 0.27, + "learning_rate": 1.713696168998998e-05, + "loss": 0.8192, + "step": 1423 + }, + { + "epoch": 0.27, + "learning_rate": 1.7132659884812782e-05, + "loss": 0.258, + "step": 1424 + }, + { + "epoch": 0.27, + "learning_rate": 1.7128355391071368e-05, + "loss": 0.8421, + "step": 1425 + }, + { + "epoch": 0.27, + "learning_rate": 1.7124048210388268e-05, + "loss": 0.8271, + "step": 1426 + }, + { + "epoch": 0.27, + "learning_rate": 1.7119738344387014e-05, + "loss": 0.8293, + "step": 1427 + }, + { + "epoch": 0.27, + "learning_rate": 1.7115425794692156e-05, + "loss": 0.2366, + "step": 1428 + }, + { + "epoch": 0.27, + "learning_rate": 1.7111110562929257e-05, + "loss": 0.7464, + "step": 1429 + }, + { + "epoch": 0.27, + "learning_rate": 1.7106792650724885e-05, + "loss": 0.9358, + "step": 1430 + }, + { + "epoch": 0.27, + "learning_rate": 1.7102472059706618e-05, + "loss": 0.8245, + "step": 1431 + }, + { + "epoch": 0.27, + "learning_rate": 1.7098148791503052e-05, + "loss": 0.9056, + "step": 1432 + }, + { + "epoch": 0.27, + "learning_rate": 1.709382284774379e-05, + "loss": 0.7716, + "step": 1433 + }, + { + "epoch": 0.27, + "learning_rate": 1.7089494230059434e-05, + "loss": 0.9008, + "step": 1434 + }, + { + "epoch": 0.27, + "learning_rate": 1.7085162940081603e-05, + "loss": 0.7999, + "step": 1435 + }, + { + "epoch": 0.27, + "learning_rate": 1.708082897944292e-05, + "loss": 0.8683, + "step": 1436 + }, + { + "epoch": 0.27, + "learning_rate": 1.7076492349777017e-05, + "loss": 0.8381, + "step": 1437 + }, + { + "epoch": 0.27, + "learning_rate": 1.7072153052718535e-05, + "loss": 0.8389, + "step": 1438 + }, + { + "epoch": 0.27, + "learning_rate": 1.7067811089903108e-05, + "loss": 0.812, + "step": 1439 + }, + { + "epoch": 0.27, + "learning_rate": 1.706346646296739e-05, + "loss": 0.2396, + "step": 1440 + }, + { + "epoch": 0.27, + "learning_rate": 1.7059119173549024e-05, + "loss": 0.8618, + "step": 1441 + }, + { + "epoch": 0.27, + "learning_rate": 1.705476922328667e-05, + "loss": 0.8941, + "step": 1442 + }, + { + "epoch": 0.27, + "learning_rate": 1.705041661381999e-05, + "loss": 0.8462, + "step": 1443 + }, + { + "epoch": 0.27, + "learning_rate": 1.7046061346789633e-05, + "loss": 0.8609, + "step": 1444 + }, + { + "epoch": 0.27, + "learning_rate": 1.7041703423837267e-05, + "loss": 0.8485, + "step": 1445 + }, + { + "epoch": 0.27, + "learning_rate": 1.703734284660556e-05, + "loss": 0.8121, + "step": 1446 + }, + { + "epoch": 0.27, + "learning_rate": 1.7032979616738167e-05, + "loss": 0.8464, + "step": 1447 + }, + { + "epoch": 0.27, + "learning_rate": 1.7028613735879755e-05, + "loss": 0.8334, + "step": 1448 + }, + { + "epoch": 0.27, + "learning_rate": 1.7024245205675986e-05, + "loss": 0.8614, + "step": 1449 + }, + { + "epoch": 0.27, + "learning_rate": 1.701987402777352e-05, + "loss": 0.811, + "step": 1450 + }, + { + "epoch": 0.28, + "learning_rate": 1.701550020382002e-05, + "loss": 0.8229, + "step": 1451 + }, + { + "epoch": 0.28, + "learning_rate": 1.7011123735464136e-05, + "loss": 0.8374, + "step": 1452 + }, + { + "epoch": 0.28, + "learning_rate": 1.700674462435553e-05, + "loss": 0.9075, + "step": 1453 + }, + { + "epoch": 0.28, + "learning_rate": 1.7002362872144843e-05, + "loss": 0.2776, + "step": 1454 + }, + { + "epoch": 0.28, + "learning_rate": 1.6997978480483724e-05, + "loss": 0.8514, + "step": 1455 + }, + { + "epoch": 0.28, + "learning_rate": 1.6993591451024815e-05, + "loss": 0.846, + "step": 1456 + }, + { + "epoch": 0.28, + "learning_rate": 1.698920178542175e-05, + "loss": 0.8255, + "step": 1457 + }, + { + "epoch": 0.28, + "learning_rate": 1.6984809485329145e-05, + "loss": 0.815, + "step": 1458 + }, + { + "epoch": 0.28, + "learning_rate": 1.698041455240264e-05, + "loss": 0.8283, + "step": 1459 + }, + { + "epoch": 0.28, + "learning_rate": 1.6976016988298834e-05, + "loss": 0.8873, + "step": 1460 + }, + { + "epoch": 0.28, + "learning_rate": 1.697161679467534e-05, + "loss": 0.8772, + "step": 1461 + }, + { + "epoch": 0.28, + "learning_rate": 1.6967213973190747e-05, + "loss": 0.8612, + "step": 1462 + }, + { + "epoch": 0.28, + "learning_rate": 1.696280852550465e-05, + "loss": 0.2702, + "step": 1463 + }, + { + "epoch": 0.28, + "learning_rate": 1.695840045327762e-05, + "loss": 0.8057, + "step": 1464 + }, + { + "epoch": 0.28, + "learning_rate": 1.6953989758171226e-05, + "loss": 0.8907, + "step": 1465 + }, + { + "epoch": 0.28, + "learning_rate": 1.694957644184802e-05, + "loss": 0.8413, + "step": 1466 + }, + { + "epoch": 0.28, + "learning_rate": 1.6945160505971552e-05, + "loss": 0.7888, + "step": 1467 + }, + { + "epoch": 0.28, + "learning_rate": 1.6940741952206342e-05, + "loss": 0.7894, + "step": 1468 + }, + { + "epoch": 0.28, + "learning_rate": 1.693632078221791e-05, + "loss": 0.8659, + "step": 1469 + }, + { + "epoch": 0.28, + "learning_rate": 1.693189699767276e-05, + "loss": 0.8597, + "step": 1470 + }, + { + "epoch": 0.28, + "learning_rate": 1.6927470600238383e-05, + "loss": 0.8024, + "step": 1471 + }, + { + "epoch": 0.28, + "learning_rate": 1.692304159158325e-05, + "loss": 0.8713, + "step": 1472 + }, + { + "epoch": 0.28, + "learning_rate": 1.691860997337682e-05, + "loss": 0.7982, + "step": 1473 + }, + { + "epoch": 0.28, + "learning_rate": 1.6914175747289532e-05, + "loss": 0.8159, + "step": 1474 + }, + { + "epoch": 0.28, + "learning_rate": 1.6909738914992812e-05, + "loss": 0.8594, + "step": 1475 + }, + { + "epoch": 0.28, + "learning_rate": 1.6905299478159067e-05, + "loss": 0.8605, + "step": 1476 + }, + { + "epoch": 0.28, + "learning_rate": 1.690085743846168e-05, + "loss": 0.8371, + "step": 1477 + }, + { + "epoch": 0.28, + "learning_rate": 1.6896412797575035e-05, + "loss": 0.8544, + "step": 1478 + }, + { + "epoch": 0.28, + "learning_rate": 1.689196555717447e-05, + "loss": 0.766, + "step": 1479 + }, + { + "epoch": 0.28, + "learning_rate": 1.6887515718936312e-05, + "loss": 0.8272, + "step": 1480 + }, + { + "epoch": 0.28, + "learning_rate": 1.688306328453788e-05, + "loss": 0.8225, + "step": 1481 + }, + { + "epoch": 0.28, + "learning_rate": 1.6878608255657457e-05, + "loss": 0.7979, + "step": 1482 + }, + { + "epoch": 0.28, + "learning_rate": 1.687415063397431e-05, + "loss": 0.8562, + "step": 1483 + }, + { + "epoch": 0.28, + "learning_rate": 1.686969042116868e-05, + "loss": 0.8527, + "step": 1484 + }, + { + "epoch": 0.28, + "learning_rate": 1.686522761892179e-05, + "loss": 0.8782, + "step": 1485 + }, + { + "epoch": 0.28, + "learning_rate": 1.6860762228915833e-05, + "loss": 0.2858, + "step": 1486 + }, + { + "epoch": 0.28, + "learning_rate": 1.685629425283398e-05, + "loss": 0.8711, + "step": 1487 + }, + { + "epoch": 0.28, + "learning_rate": 1.6851823692360377e-05, + "loss": 0.7795, + "step": 1488 + }, + { + "epoch": 0.28, + "learning_rate": 1.6847350549180148e-05, + "loss": 0.7741, + "step": 1489 + }, + { + "epoch": 0.28, + "learning_rate": 1.6842874824979384e-05, + "loss": 0.8962, + "step": 1490 + }, + { + "epoch": 0.28, + "learning_rate": 1.683839652144515e-05, + "loss": 0.8849, + "step": 1491 + }, + { + "epoch": 0.28, + "learning_rate": 1.6833915640265485e-05, + "loss": 0.8712, + "step": 1492 + }, + { + "epoch": 0.28, + "learning_rate": 1.6829432183129407e-05, + "loss": 0.8211, + "step": 1493 + }, + { + "epoch": 0.28, + "learning_rate": 1.6824946151726885e-05, + "loss": 0.7761, + "step": 1494 + }, + { + "epoch": 0.28, + "learning_rate": 1.682045754774888e-05, + "loss": 0.838, + "step": 1495 + }, + { + "epoch": 0.28, + "learning_rate": 1.6815966372887305e-05, + "loss": 0.8125, + "step": 1496 + }, + { + "epoch": 0.28, + "learning_rate": 1.6811472628835058e-05, + "loss": 0.8488, + "step": 1497 + }, + { + "epoch": 0.28, + "learning_rate": 1.6806976317286e-05, + "loss": 0.8121, + "step": 1498 + }, + { + "epoch": 0.28, + "learning_rate": 1.680247743993495e-05, + "loss": 0.8922, + "step": 1499 + }, + { + "epoch": 0.28, + "learning_rate": 1.6797975998477704e-05, + "loss": 0.8288, + "step": 1500 + }, + { + "epoch": 0.28, + "learning_rate": 1.6793471994611025e-05, + "loss": 0.8507, + "step": 1501 + }, + { + "epoch": 0.28, + "learning_rate": 1.6788965430032636e-05, + "loss": 0.2552, + "step": 1502 + }, + { + "epoch": 0.28, + "learning_rate": 1.6784456306441234e-05, + "loss": 0.2951, + "step": 1503 + }, + { + "epoch": 0.29, + "learning_rate": 1.677994462553647e-05, + "loss": 0.8687, + "step": 1504 + }, + { + "epoch": 0.29, + "learning_rate": 1.677543038901896e-05, + "loss": 0.8582, + "step": 1505 + }, + { + "epoch": 0.29, + "learning_rate": 1.6770913598590297e-05, + "loss": 0.2853, + "step": 1506 + }, + { + "epoch": 0.29, + "learning_rate": 1.676639425595302e-05, + "loss": 0.8659, + "step": 1507 + }, + { + "epoch": 0.29, + "learning_rate": 1.6761872362810642e-05, + "loss": 0.859, + "step": 1508 + }, + { + "epoch": 0.29, + "learning_rate": 1.675734792086763e-05, + "loss": 0.8488, + "step": 1509 + }, + { + "epoch": 0.29, + "learning_rate": 1.675282093182941e-05, + "loss": 0.8896, + "step": 1510 + }, + { + "epoch": 0.29, + "learning_rate": 1.6748291397402376e-05, + "loss": 0.7855, + "step": 1511 + }, + { + "epoch": 0.29, + "learning_rate": 1.674375931929388e-05, + "loss": 0.8441, + "step": 1512 + }, + { + "epoch": 0.29, + "learning_rate": 1.6739224699212226e-05, + "loss": 0.7947, + "step": 1513 + }, + { + "epoch": 0.29, + "learning_rate": 1.6734687538866683e-05, + "loss": 0.8177, + "step": 1514 + }, + { + "epoch": 0.29, + "learning_rate": 1.6730147839967476e-05, + "loss": 0.8794, + "step": 1515 + }, + { + "epoch": 0.29, + "learning_rate": 1.672560560422578e-05, + "loss": 0.253, + "step": 1516 + }, + { + "epoch": 0.29, + "learning_rate": 1.672106083335374e-05, + "loss": 0.8454, + "step": 1517 + }, + { + "epoch": 0.29, + "learning_rate": 1.6716513529064443e-05, + "loss": 0.7621, + "step": 1518 + }, + { + "epoch": 0.29, + "learning_rate": 1.6711963693071943e-05, + "loss": 0.825, + "step": 1519 + }, + { + "epoch": 0.29, + "learning_rate": 1.6707411327091238e-05, + "loss": 0.8435, + "step": 1520 + }, + { + "epoch": 0.29, + "learning_rate": 1.6702856432838283e-05, + "loss": 0.8391, + "step": 1521 + }, + { + "epoch": 0.29, + "learning_rate": 1.6698299012029992e-05, + "loss": 0.8649, + "step": 1522 + }, + { + "epoch": 0.29, + "learning_rate": 1.6693739066384214e-05, + "loss": 0.8066, + "step": 1523 + }, + { + "epoch": 0.29, + "learning_rate": 1.6689176597619773e-05, + "loss": 0.8276, + "step": 1524 + }, + { + "epoch": 0.29, + "learning_rate": 1.668461160745643e-05, + "loss": 0.8339, + "step": 1525 + }, + { + "epoch": 0.29, + "learning_rate": 1.66800440976149e-05, + "loss": 0.8662, + "step": 1526 + }, + { + "epoch": 0.29, + "learning_rate": 1.667547406981684e-05, + "loss": 0.876, + "step": 1527 + }, + { + "epoch": 0.29, + "learning_rate": 1.6670901525784873e-05, + "loss": 0.8394, + "step": 1528 + }, + { + "epoch": 0.29, + "learning_rate": 1.6666326467242555e-05, + "loss": 0.8201, + "step": 1529 + }, + { + "epoch": 0.29, + "learning_rate": 1.6661748895914397e-05, + "loss": 0.8575, + "step": 1530 + }, + { + "epoch": 0.29, + "learning_rate": 1.6657168813525855e-05, + "loss": 0.879, + "step": 1531 + }, + { + "epoch": 0.29, + "learning_rate": 1.6652586221803328e-05, + "loss": 0.8541, + "step": 1532 + }, + { + "epoch": 0.29, + "learning_rate": 1.6648001122474174e-05, + "loss": 0.8556, + "step": 1533 + }, + { + "epoch": 0.29, + "learning_rate": 1.664341351726668e-05, + "loss": 0.8724, + "step": 1534 + }, + { + "epoch": 0.29, + "learning_rate": 1.6638823407910085e-05, + "loss": 0.784, + "step": 1535 + }, + { + "epoch": 0.29, + "learning_rate": 1.6634230796134578e-05, + "loss": 0.839, + "step": 1536 + }, + { + "epoch": 0.29, + "learning_rate": 1.6629635683671276e-05, + "loss": 0.8106, + "step": 1537 + }, + { + "epoch": 0.29, + "learning_rate": 1.662503807225225e-05, + "loss": 0.8475, + "step": 1538 + }, + { + "epoch": 0.29, + "learning_rate": 1.6620437963610514e-05, + "loss": 0.8668, + "step": 1539 + }, + { + "epoch": 0.29, + "learning_rate": 1.661583535948002e-05, + "loss": 0.8013, + "step": 1540 + }, + { + "epoch": 0.29, + "learning_rate": 1.6611230261595653e-05, + "loss": 0.8761, + "step": 1541 + }, + { + "epoch": 0.29, + "learning_rate": 1.6606622671693253e-05, + "loss": 0.8099, + "step": 1542 + }, + { + "epoch": 0.29, + "learning_rate": 1.6602012591509583e-05, + "loss": 0.8486, + "step": 1543 + }, + { + "epoch": 0.29, + "learning_rate": 1.6597400022782365e-05, + "loss": 0.8778, + "step": 1544 + }, + { + "epoch": 0.29, + "learning_rate": 1.659278496725024e-05, + "loss": 0.876, + "step": 1545 + }, + { + "epoch": 0.29, + "learning_rate": 1.6588167426652794e-05, + "loss": 0.8523, + "step": 1546 + }, + { + "epoch": 0.29, + "learning_rate": 1.658354740273055e-05, + "loss": 0.8579, + "step": 1547 + }, + { + "epoch": 0.29, + "learning_rate": 1.657892489722497e-05, + "loss": 0.8643, + "step": 1548 + }, + { + "epoch": 0.29, + "learning_rate": 1.657429991187844e-05, + "loss": 0.8296, + "step": 1549 + }, + { + "epoch": 0.29, + "learning_rate": 1.65696724484343e-05, + "loss": 0.827, + "step": 1550 + }, + { + "epoch": 0.29, + "learning_rate": 1.6565042508636803e-05, + "loss": 0.8535, + "step": 1551 + }, + { + "epoch": 0.29, + "learning_rate": 1.6560410094231144e-05, + "loss": 0.8364, + "step": 1552 + }, + { + "epoch": 0.29, + "learning_rate": 1.655577520696346e-05, + "loss": 0.8365, + "step": 1553 + }, + { + "epoch": 0.29, + "learning_rate": 1.6551137848580808e-05, + "loss": 0.8355, + "step": 1554 + }, + { + "epoch": 0.29, + "learning_rate": 1.654649802083118e-05, + "loss": 0.277, + "step": 1555 + }, + { + "epoch": 0.29, + "learning_rate": 1.65418557254635e-05, + "loss": 0.853, + "step": 1556 + }, + { + "epoch": 0.3, + "learning_rate": 1.6537210964227618e-05, + "loss": 0.8828, + "step": 1557 + }, + { + "epoch": 0.3, + "learning_rate": 1.6532563738874325e-05, + "loss": 0.8574, + "step": 1558 + }, + { + "epoch": 0.3, + "learning_rate": 1.6527914051155328e-05, + "loss": 0.8293, + "step": 1559 + }, + { + "epoch": 0.3, + "learning_rate": 1.6523261902823265e-05, + "loss": 0.8617, + "step": 1560 + }, + { + "epoch": 0.3, + "learning_rate": 1.6518607295631704e-05, + "loss": 0.8494, + "step": 1561 + }, + { + "epoch": 0.3, + "learning_rate": 1.651395023133514e-05, + "loss": 0.8133, + "step": 1562 + }, + { + "epoch": 0.3, + "learning_rate": 1.6509290711688996e-05, + "loss": 0.28, + "step": 1563 + }, + { + "epoch": 0.3, + "learning_rate": 1.6504628738449613e-05, + "loss": 0.8683, + "step": 1564 + }, + { + "epoch": 0.3, + "learning_rate": 1.6499964313374265e-05, + "loss": 0.2704, + "step": 1565 + }, + { + "epoch": 0.3, + "learning_rate": 1.6495297438221145e-05, + "loss": 0.8875, + "step": 1566 + }, + { + "epoch": 0.3, + "learning_rate": 1.6490628114749374e-05, + "loss": 0.8046, + "step": 1567 + }, + { + "epoch": 0.3, + "learning_rate": 1.6485956344718995e-05, + "loss": 0.8201, + "step": 1568 + }, + { + "epoch": 0.3, + "learning_rate": 1.6481282129890963e-05, + "loss": 0.7994, + "step": 1569 + }, + { + "epoch": 0.3, + "learning_rate": 1.6476605472027173e-05, + "loss": 0.8465, + "step": 1570 + }, + { + "epoch": 0.3, + "learning_rate": 1.647192637289042e-05, + "loss": 0.858, + "step": 1571 + }, + { + "epoch": 0.3, + "learning_rate": 1.646724483424444e-05, + "loss": 0.8585, + "step": 1572 + }, + { + "epoch": 0.3, + "learning_rate": 1.6462560857853876e-05, + "loss": 0.8625, + "step": 1573 + }, + { + "epoch": 0.3, + "learning_rate": 1.645787444548429e-05, + "loss": 0.7827, + "step": 1574 + }, + { + "epoch": 0.3, + "learning_rate": 1.6453185598902166e-05, + "loss": 0.2517, + "step": 1575 + }, + { + "epoch": 0.3, + "learning_rate": 1.6448494319874905e-05, + "loss": 0.7986, + "step": 1576 + }, + { + "epoch": 0.3, + "learning_rate": 1.6443800610170826e-05, + "loss": 0.8088, + "step": 1577 + }, + { + "epoch": 0.3, + "learning_rate": 1.6439104471559157e-05, + "loss": 0.831, + "step": 1578 + }, + { + "epoch": 0.3, + "learning_rate": 1.643440590581005e-05, + "loss": 0.8175, + "step": 1579 + }, + { + "epoch": 0.3, + "learning_rate": 1.6429704914694573e-05, + "loss": 0.8533, + "step": 1580 + }, + { + "epoch": 0.3, + "learning_rate": 1.64250014999847e-05, + "loss": 0.7929, + "step": 1581 + }, + { + "epoch": 0.3, + "learning_rate": 1.6420295663453317e-05, + "loss": 0.2819, + "step": 1582 + }, + { + "epoch": 0.3, + "learning_rate": 1.6415587406874235e-05, + "loss": 0.8227, + "step": 1583 + }, + { + "epoch": 0.3, + "learning_rate": 1.6410876732022173e-05, + "loss": 0.7564, + "step": 1584 + }, + { + "epoch": 0.3, + "learning_rate": 1.640616364067275e-05, + "loss": 0.8516, + "step": 1585 + }, + { + "epoch": 0.3, + "learning_rate": 1.6401448134602517e-05, + "loss": 0.2508, + "step": 1586 + }, + { + "epoch": 0.3, + "learning_rate": 1.6396730215588913e-05, + "loss": 0.8457, + "step": 1587 + }, + { + "epoch": 0.3, + "learning_rate": 1.6392009885410302e-05, + "loss": 0.8631, + "step": 1588 + }, + { + "epoch": 0.3, + "learning_rate": 1.638728714584595e-05, + "loss": 0.8465, + "step": 1589 + }, + { + "epoch": 0.3, + "learning_rate": 1.6382561998676037e-05, + "loss": 0.8922, + "step": 1590 + }, + { + "epoch": 0.3, + "learning_rate": 1.6377834445681636e-05, + "loss": 0.2784, + "step": 1591 + }, + { + "epoch": 0.3, + "learning_rate": 1.6373104488644747e-05, + "loss": 0.885, + "step": 1592 + }, + { + "epoch": 0.3, + "learning_rate": 1.636837212934826e-05, + "loss": 0.8873, + "step": 1593 + }, + { + "epoch": 0.3, + "learning_rate": 1.6363637369575984e-05, + "loss": 0.862, + "step": 1594 + }, + { + "epoch": 0.3, + "learning_rate": 1.6358900211112622e-05, + "loss": 0.8585, + "step": 1595 + }, + { + "epoch": 0.3, + "learning_rate": 1.635416065574378e-05, + "loss": 0.7975, + "step": 1596 + }, + { + "epoch": 0.3, + "learning_rate": 1.634941870525598e-05, + "loss": 0.8407, + "step": 1597 + }, + { + "epoch": 0.3, + "learning_rate": 1.6344674361436635e-05, + "loss": 0.8281, + "step": 1598 + }, + { + "epoch": 0.3, + "learning_rate": 1.6339927626074062e-05, + "loss": 0.8079, + "step": 1599 + }, + { + "epoch": 0.3, + "learning_rate": 1.6335178500957486e-05, + "loss": 0.7869, + "step": 1600 + }, + { + "epoch": 0.3, + "learning_rate": 1.633042698787703e-05, + "loss": 0.8016, + "step": 1601 + }, + { + "epoch": 0.3, + "learning_rate": 1.632567308862371e-05, + "loss": 0.8399, + "step": 1602 + }, + { + "epoch": 0.3, + "learning_rate": 1.632091680498945e-05, + "loss": 0.8403, + "step": 1603 + }, + { + "epoch": 0.3, + "learning_rate": 1.631615813876707e-05, + "loss": 0.8305, + "step": 1604 + }, + { + "epoch": 0.3, + "learning_rate": 1.6311397091750286e-05, + "loss": 0.8745, + "step": 1605 + }, + { + "epoch": 0.3, + "learning_rate": 1.630663366573372e-05, + "loss": 0.8303, + "step": 1606 + }, + { + "epoch": 0.3, + "learning_rate": 1.6301867862512874e-05, + "loss": 0.8248, + "step": 1607 + }, + { + "epoch": 0.3, + "learning_rate": 1.6297099683884163e-05, + "loss": 0.8763, + "step": 1608 + }, + { + "epoch": 0.3, + "learning_rate": 1.629232913164489e-05, + "loss": 0.8462, + "step": 1609 + }, + { + "epoch": 0.31, + "learning_rate": 1.628755620759325e-05, + "loss": 0.8389, + "step": 1610 + }, + { + "epoch": 0.31, + "learning_rate": 1.628278091352834e-05, + "loss": 0.8412, + "step": 1611 + }, + { + "epoch": 0.31, + "learning_rate": 1.6278003251250135e-05, + "loss": 0.829, + "step": 1612 + }, + { + "epoch": 0.31, + "learning_rate": 1.6273223222559528e-05, + "loss": 0.7905, + "step": 1613 + }, + { + "epoch": 0.31, + "learning_rate": 1.6268440829258277e-05, + "loss": 0.8288, + "step": 1614 + }, + { + "epoch": 0.31, + "learning_rate": 1.626365607314905e-05, + "loss": 0.7864, + "step": 1615 + }, + { + "epoch": 0.31, + "learning_rate": 1.6258868956035398e-05, + "loss": 0.8297, + "step": 1616 + }, + { + "epoch": 0.31, + "learning_rate": 1.625407947972176e-05, + "loss": 0.8581, + "step": 1617 + }, + { + "epoch": 0.31, + "learning_rate": 1.6249287646013474e-05, + "loss": 0.9098, + "step": 1618 + }, + { + "epoch": 0.31, + "learning_rate": 1.6244493456716754e-05, + "loss": 0.8287, + "step": 1619 + }, + { + "epoch": 0.31, + "learning_rate": 1.623969691363871e-05, + "loss": 0.7983, + "step": 1620 + }, + { + "epoch": 0.31, + "learning_rate": 1.6234898018587336e-05, + "loss": 0.8659, + "step": 1621 + }, + { + "epoch": 0.31, + "learning_rate": 1.6230096773371514e-05, + "loss": 0.7471, + "step": 1622 + }, + { + "epoch": 0.31, + "learning_rate": 1.6225293179801015e-05, + "loss": 0.2729, + "step": 1623 + }, + { + "epoch": 0.31, + "learning_rate": 1.6220487239686488e-05, + "loss": 0.9194, + "step": 1624 + }, + { + "epoch": 0.31, + "learning_rate": 1.621567895483947e-05, + "loss": 0.8472, + "step": 1625 + }, + { + "epoch": 0.31, + "learning_rate": 1.6210868327072384e-05, + "loss": 0.8731, + "step": 1626 + }, + { + "epoch": 0.31, + "learning_rate": 1.6206055358198528e-05, + "loss": 0.8163, + "step": 1627 + }, + { + "epoch": 0.31, + "learning_rate": 1.6201240050032098e-05, + "loss": 0.8267, + "step": 1628 + }, + { + "epoch": 0.31, + "learning_rate": 1.619642240438816e-05, + "loss": 0.8176, + "step": 1629 + }, + { + "epoch": 0.31, + "learning_rate": 1.619160242308266e-05, + "loss": 0.8863, + "step": 1630 + }, + { + "epoch": 0.31, + "learning_rate": 1.6186780107932425e-05, + "loss": 0.7653, + "step": 1631 + }, + { + "epoch": 0.31, + "learning_rate": 1.6181955460755175e-05, + "loss": 0.7957, + "step": 1632 + }, + { + "epoch": 0.31, + "learning_rate": 1.6177128483369487e-05, + "loss": 0.2774, + "step": 1633 + }, + { + "epoch": 0.31, + "learning_rate": 1.6172299177594835e-05, + "loss": 0.7961, + "step": 1634 + }, + { + "epoch": 0.31, + "learning_rate": 1.616746754525156e-05, + "loss": 0.9095, + "step": 1635 + }, + { + "epoch": 0.31, + "learning_rate": 1.616263358816089e-05, + "loss": 0.8267, + "step": 1636 + }, + { + "epoch": 0.31, + "learning_rate": 1.6157797308144916e-05, + "loss": 0.8953, + "step": 1637 + }, + { + "epoch": 0.31, + "learning_rate": 1.6152958707026615e-05, + "loss": 0.8661, + "step": 1638 + }, + { + "epoch": 0.31, + "learning_rate": 1.614811778662983e-05, + "loss": 0.8201, + "step": 1639 + }, + { + "epoch": 0.31, + "learning_rate": 1.614327454877929e-05, + "loss": 0.8201, + "step": 1640 + }, + { + "epoch": 0.31, + "learning_rate": 1.6138428995300587e-05, + "loss": 0.7952, + "step": 1641 + }, + { + "epoch": 0.31, + "learning_rate": 1.6133581128020193e-05, + "loss": 0.7788, + "step": 1642 + }, + { + "epoch": 0.31, + "learning_rate": 1.612873094876545e-05, + "loss": 0.8973, + "step": 1643 + }, + { + "epoch": 0.31, + "learning_rate": 1.6123878459364564e-05, + "loss": 0.7932, + "step": 1644 + }, + { + "epoch": 0.31, + "learning_rate": 1.6119023661646626e-05, + "loss": 0.8309, + "step": 1645 + }, + { + "epoch": 0.31, + "learning_rate": 1.611416655744158e-05, + "loss": 0.7771, + "step": 1646 + }, + { + "epoch": 0.31, + "learning_rate": 1.610930714858026e-05, + "loss": 0.8416, + "step": 1647 + }, + { + "epoch": 0.31, + "learning_rate": 1.610444543689435e-05, + "loss": 0.8634, + "step": 1648 + }, + { + "epoch": 0.31, + "learning_rate": 1.6099581424216413e-05, + "loss": 0.8609, + "step": 1649 + }, + { + "epoch": 0.31, + "learning_rate": 1.6094715112379874e-05, + "loss": 0.8823, + "step": 1650 + }, + { + "epoch": 0.31, + "learning_rate": 1.6089846503219027e-05, + "loss": 0.8077, + "step": 1651 + }, + { + "epoch": 0.31, + "learning_rate": 1.6084975598569033e-05, + "loss": 0.8839, + "step": 1652 + }, + { + "epoch": 0.31, + "learning_rate": 1.6080102400265915e-05, + "loss": 0.842, + "step": 1653 + }, + { + "epoch": 0.31, + "learning_rate": 1.607522691014656e-05, + "loss": 0.8683, + "step": 1654 + }, + { + "epoch": 0.31, + "learning_rate": 1.6070349130048724e-05, + "loss": 0.8213, + "step": 1655 + }, + { + "epoch": 0.31, + "learning_rate": 1.6065469061811026e-05, + "loss": 0.8598, + "step": 1656 + }, + { + "epoch": 0.31, + "learning_rate": 1.6060586707272943e-05, + "loss": 0.2715, + "step": 1657 + }, + { + "epoch": 0.31, + "learning_rate": 1.605570206827481e-05, + "loss": 0.8116, + "step": 1658 + }, + { + "epoch": 0.31, + "learning_rate": 1.605081514665784e-05, + "loss": 0.843, + "step": 1659 + }, + { + "epoch": 0.31, + "learning_rate": 1.6045925944264086e-05, + "loss": 0.8597, + "step": 1660 + }, + { + "epoch": 0.31, + "learning_rate": 1.604103446293647e-05, + "loss": 0.7636, + "step": 1661 + }, + { + "epoch": 0.32, + "learning_rate": 1.6036140704518773e-05, + "loss": 0.8577, + "step": 1662 + }, + { + "epoch": 0.32, + "learning_rate": 1.603124467085564e-05, + "loss": 0.8037, + "step": 1663 + }, + { + "epoch": 0.32, + "learning_rate": 1.6026346363792565e-05, + "loss": 0.861, + "step": 1664 + }, + { + "epoch": 0.32, + "learning_rate": 1.6021445785175898e-05, + "loss": 0.8127, + "step": 1665 + }, + { + "epoch": 0.32, + "learning_rate": 1.6016542936852856e-05, + "loss": 0.852, + "step": 1666 + }, + { + "epoch": 0.32, + "learning_rate": 1.6011637820671495e-05, + "loss": 0.8179, + "step": 1667 + }, + { + "epoch": 0.32, + "learning_rate": 1.6006730438480743e-05, + "loss": 0.8478, + "step": 1668 + }, + { + "epoch": 0.32, + "learning_rate": 1.6001820792130373e-05, + "loss": 0.7712, + "step": 1669 + }, + { + "epoch": 0.32, + "learning_rate": 1.599690888347101e-05, + "loss": 0.8017, + "step": 1670 + }, + { + "epoch": 0.32, + "learning_rate": 1.599199471435414e-05, + "loss": 0.8158, + "step": 1671 + }, + { + "epoch": 0.32, + "learning_rate": 1.5987078286632088e-05, + "loss": 0.8777, + "step": 1672 + }, + { + "epoch": 0.32, + "learning_rate": 1.5982159602158045e-05, + "loss": 0.8094, + "step": 1673 + }, + { + "epoch": 0.32, + "learning_rate": 1.5977238662786043e-05, + "loss": 0.2667, + "step": 1674 + }, + { + "epoch": 0.32, + "learning_rate": 1.5972315470370966e-05, + "loss": 0.8559, + "step": 1675 + }, + { + "epoch": 0.32, + "learning_rate": 1.5967390026768547e-05, + "loss": 0.8561, + "step": 1676 + }, + { + "epoch": 0.32, + "learning_rate": 1.5962462333835374e-05, + "loss": 0.8357, + "step": 1677 + }, + { + "epoch": 0.32, + "learning_rate": 1.5957532393428872e-05, + "loss": 0.895, + "step": 1678 + }, + { + "epoch": 0.32, + "learning_rate": 1.5952600207407317e-05, + "loss": 0.8396, + "step": 1679 + }, + { + "epoch": 0.32, + "learning_rate": 1.594766577762984e-05, + "loss": 0.859, + "step": 1680 + }, + { + "epoch": 0.32, + "learning_rate": 1.5942729105956403e-05, + "loss": 0.7892, + "step": 1681 + }, + { + "epoch": 0.32, + "learning_rate": 1.5937790194247827e-05, + "loss": 0.8633, + "step": 1682 + }, + { + "epoch": 0.32, + "learning_rate": 1.5932849044365768e-05, + "loss": 0.798, + "step": 1683 + }, + { + "epoch": 0.32, + "learning_rate": 1.5927905658172732e-05, + "loss": 0.8631, + "step": 1684 + }, + { + "epoch": 0.32, + "learning_rate": 1.5922960037532057e-05, + "loss": 0.8771, + "step": 1685 + }, + { + "epoch": 0.32, + "learning_rate": 1.5918012184307937e-05, + "loss": 0.8304, + "step": 1686 + }, + { + "epoch": 0.32, + "learning_rate": 1.5913062100365403e-05, + "loss": 0.8161, + "step": 1687 + }, + { + "epoch": 0.32, + "learning_rate": 1.5908109787570322e-05, + "loss": 0.8039, + "step": 1688 + }, + { + "epoch": 0.32, + "learning_rate": 1.5903155247789403e-05, + "loss": 0.7868, + "step": 1689 + }, + { + "epoch": 0.32, + "learning_rate": 1.58981984828902e-05, + "loss": 0.8391, + "step": 1690 + }, + { + "epoch": 0.32, + "learning_rate": 1.5893239494741098e-05, + "loss": 0.8624, + "step": 1691 + }, + { + "epoch": 0.32, + "learning_rate": 1.588827828521133e-05, + "loss": 0.872, + "step": 1692 + }, + { + "epoch": 0.32, + "learning_rate": 1.588331485617095e-05, + "loss": 0.8161, + "step": 1693 + }, + { + "epoch": 0.32, + "learning_rate": 1.587834920949087e-05, + "loss": 0.8027, + "step": 1694 + }, + { + "epoch": 0.32, + "learning_rate": 1.5873381347042816e-05, + "loss": 0.8108, + "step": 1695 + }, + { + "epoch": 0.32, + "learning_rate": 1.5868411270699365e-05, + "loss": 0.8644, + "step": 1696 + }, + { + "epoch": 0.32, + "learning_rate": 1.5863438982333926e-05, + "loss": 0.8706, + "step": 1697 + }, + { + "epoch": 0.32, + "learning_rate": 1.5858464483820736e-05, + "loss": 0.8442, + "step": 1698 + }, + { + "epoch": 0.32, + "learning_rate": 1.585348777703486e-05, + "loss": 0.7976, + "step": 1699 + }, + { + "epoch": 0.32, + "learning_rate": 1.5848508863852217e-05, + "loss": 0.831, + "step": 1700 + }, + { + "epoch": 0.32, + "learning_rate": 1.5843527746149537e-05, + "loss": 0.8637, + "step": 1701 + }, + { + "epoch": 0.32, + "learning_rate": 1.583854442580439e-05, + "loss": 0.8215, + "step": 1702 + }, + { + "epoch": 0.32, + "learning_rate": 1.5833558904695175e-05, + "loss": 0.8393, + "step": 1703 + }, + { + "epoch": 0.32, + "learning_rate": 1.5828571184701115e-05, + "loss": 0.8526, + "step": 1704 + }, + { + "epoch": 0.32, + "learning_rate": 1.5823581267702273e-05, + "loss": 0.8363, + "step": 1705 + }, + { + "epoch": 0.32, + "learning_rate": 1.581858915557953e-05, + "loss": 0.8217, + "step": 1706 + }, + { + "epoch": 0.32, + "learning_rate": 1.58135948502146e-05, + "loss": 0.8325, + "step": 1707 + }, + { + "epoch": 0.32, + "learning_rate": 1.5808598353490025e-05, + "loss": 0.266, + "step": 1708 + }, + { + "epoch": 0.32, + "learning_rate": 1.5803599667289164e-05, + "loss": 0.8051, + "step": 1709 + }, + { + "epoch": 0.32, + "learning_rate": 1.5798598793496205e-05, + "loss": 0.8548, + "step": 1710 + }, + { + "epoch": 0.32, + "learning_rate": 1.5793595733996174e-05, + "loss": 0.845, + "step": 1711 + }, + { + "epoch": 0.32, + "learning_rate": 1.5788590490674903e-05, + "loss": 0.8153, + "step": 1712 + }, + { + "epoch": 0.32, + "learning_rate": 1.5783583065419054e-05, + "loss": 0.8708, + "step": 1713 + }, + { + "epoch": 0.32, + "learning_rate": 1.5778573460116115e-05, + "loss": 0.2951, + "step": 1714 + }, + { + "epoch": 0.33, + "learning_rate": 1.5773561676654385e-05, + "loss": 0.7999, + "step": 1715 + }, + { + "epoch": 0.33, + "learning_rate": 1.5768547716923e-05, + "loss": 0.7696, + "step": 1716 + }, + { + "epoch": 0.33, + "learning_rate": 1.5763531582811903e-05, + "loss": 0.7865, + "step": 1717 + }, + { + "epoch": 0.33, + "learning_rate": 1.5758513276211854e-05, + "loss": 0.8316, + "step": 1718 + }, + { + "epoch": 0.33, + "learning_rate": 1.5753492799014455e-05, + "loss": 0.8058, + "step": 1719 + }, + { + "epoch": 0.33, + "learning_rate": 1.5748470153112093e-05, + "loss": 0.7642, + "step": 1720 + }, + { + "epoch": 0.33, + "learning_rate": 1.5743445340398003e-05, + "loss": 0.7741, + "step": 1721 + }, + { + "epoch": 0.33, + "learning_rate": 1.5738418362766218e-05, + "loss": 0.7896, + "step": 1722 + }, + { + "epoch": 0.33, + "learning_rate": 1.5733389222111592e-05, + "loss": 0.8798, + "step": 1723 + }, + { + "epoch": 0.33, + "learning_rate": 1.5728357920329797e-05, + "loss": 0.8379, + "step": 1724 + }, + { + "epoch": 0.33, + "learning_rate": 1.5723324459317315e-05, + "loss": 0.8479, + "step": 1725 + }, + { + "epoch": 0.33, + "learning_rate": 1.5718288840971446e-05, + "loss": 0.8085, + "step": 1726 + }, + { + "epoch": 0.33, + "learning_rate": 1.57132510671903e-05, + "loss": 0.8285, + "step": 1727 + }, + { + "epoch": 0.33, + "learning_rate": 1.5708211139872797e-05, + "loss": 0.7952, + "step": 1728 + }, + { + "epoch": 0.33, + "learning_rate": 1.570316906091868e-05, + "loss": 0.8659, + "step": 1729 + }, + { + "epoch": 0.33, + "learning_rate": 1.5698124832228496e-05, + "loss": 0.8579, + "step": 1730 + }, + { + "epoch": 0.33, + "learning_rate": 1.5693078455703594e-05, + "loss": 0.8764, + "step": 1731 + }, + { + "epoch": 0.33, + "learning_rate": 1.5688029933246148e-05, + "loss": 0.8467, + "step": 1732 + }, + { + "epoch": 0.33, + "learning_rate": 1.5682979266759133e-05, + "loss": 0.8558, + "step": 1733 + }, + { + "epoch": 0.33, + "learning_rate": 1.5677926458146327e-05, + "loss": 0.2869, + "step": 1734 + }, + { + "epoch": 0.33, + "learning_rate": 1.5672871509312323e-05, + "loss": 0.8317, + "step": 1735 + }, + { + "epoch": 0.33, + "learning_rate": 1.5667814422162523e-05, + "loss": 0.8137, + "step": 1736 + }, + { + "epoch": 0.33, + "learning_rate": 1.566275519860313e-05, + "loss": 0.7962, + "step": 1737 + }, + { + "epoch": 0.33, + "learning_rate": 1.565769384054115e-05, + "loss": 0.8457, + "step": 1738 + }, + { + "epoch": 0.33, + "learning_rate": 1.5652630349884393e-05, + "loss": 0.8507, + "step": 1739 + }, + { + "epoch": 0.33, + "learning_rate": 1.5647564728541485e-05, + "loss": 0.2648, + "step": 1740 + }, + { + "epoch": 0.33, + "learning_rate": 1.5642496978421842e-05, + "loss": 0.7983, + "step": 1741 + }, + { + "epoch": 0.33, + "learning_rate": 1.5637427101435693e-05, + "loss": 0.8526, + "step": 1742 + }, + { + "epoch": 0.33, + "learning_rate": 1.5632355099494052e-05, + "loss": 0.8226, + "step": 1743 + }, + { + "epoch": 0.33, + "learning_rate": 1.5627280974508753e-05, + "loss": 0.8254, + "step": 1744 + }, + { + "epoch": 0.33, + "learning_rate": 1.5622204728392422e-05, + "loss": 0.8644, + "step": 1745 + }, + { + "epoch": 0.33, + "learning_rate": 1.561712636305848e-05, + "loss": 0.8742, + "step": 1746 + }, + { + "epoch": 0.33, + "learning_rate": 1.5612045880421155e-05, + "loss": 0.8212, + "step": 1747 + }, + { + "epoch": 0.33, + "learning_rate": 1.560696328239547e-05, + "loss": 0.8856, + "step": 1748 + }, + { + "epoch": 0.33, + "learning_rate": 1.560187857089724e-05, + "loss": 0.7602, + "step": 1749 + }, + { + "epoch": 0.33, + "learning_rate": 1.5596791747843083e-05, + "loss": 0.7869, + "step": 1750 + }, + { + "epoch": 0.33, + "learning_rate": 1.5591702815150414e-05, + "loss": 0.8894, + "step": 1751 + }, + { + "epoch": 0.33, + "learning_rate": 1.5586611774737443e-05, + "loss": 0.8283, + "step": 1752 + }, + { + "epoch": 0.33, + "learning_rate": 1.5581518628523166e-05, + "loss": 0.2428, + "step": 1753 + }, + { + "epoch": 0.33, + "learning_rate": 1.557642337842738e-05, + "loss": 0.8203, + "step": 1754 + }, + { + "epoch": 0.33, + "learning_rate": 1.5571326026370676e-05, + "loss": 0.7852, + "step": 1755 + }, + { + "epoch": 0.33, + "learning_rate": 1.5566226574274435e-05, + "loss": 0.8541, + "step": 1756 + }, + { + "epoch": 0.33, + "learning_rate": 1.556112502406083e-05, + "loss": 0.8374, + "step": 1757 + }, + { + "epoch": 0.33, + "learning_rate": 1.555602137765282e-05, + "loss": 0.8297, + "step": 1758 + }, + { + "epoch": 0.33, + "learning_rate": 1.5550915636974165e-05, + "loss": 0.8374, + "step": 1759 + }, + { + "epoch": 0.33, + "learning_rate": 1.5545807803949404e-05, + "loss": 0.8269, + "step": 1760 + }, + { + "epoch": 0.33, + "learning_rate": 1.554069788050387e-05, + "loss": 0.8493, + "step": 1761 + }, + { + "epoch": 0.33, + "learning_rate": 1.5535585868563688e-05, + "loss": 0.8383, + "step": 1762 + }, + { + "epoch": 0.33, + "learning_rate": 1.5530471770055756e-05, + "loss": 0.8387, + "step": 1763 + }, + { + "epoch": 0.33, + "learning_rate": 1.552535558690777e-05, + "loss": 0.8449, + "step": 1764 + }, + { + "epoch": 0.33, + "learning_rate": 1.5520237321048213e-05, + "loss": 0.8186, + "step": 1765 + }, + { + "epoch": 0.33, + "learning_rate": 1.5515116974406347e-05, + "loss": 0.8196, + "step": 1766 + }, + { + "epoch": 0.33, + "learning_rate": 1.5509994548912223e-05, + "loss": 0.8235, + "step": 1767 + }, + { + "epoch": 0.34, + "learning_rate": 1.550487004649667e-05, + "loss": 0.8592, + "step": 1768 + }, + { + "epoch": 0.34, + "learning_rate": 1.5499743469091303e-05, + "loss": 0.8027, + "step": 1769 + }, + { + "epoch": 0.34, + "learning_rate": 1.5494614818628525e-05, + "loss": 0.783, + "step": 1770 + }, + { + "epoch": 0.34, + "learning_rate": 1.5489484097041506e-05, + "loss": 0.8276, + "step": 1771 + }, + { + "epoch": 0.34, + "learning_rate": 1.548435130626421e-05, + "loss": 0.8484, + "step": 1772 + }, + { + "epoch": 0.34, + "learning_rate": 1.5479216448231377e-05, + "loss": 0.3076, + "step": 1773 + }, + { + "epoch": 0.34, + "learning_rate": 1.5474079524878525e-05, + "loss": 0.8657, + "step": 1774 + }, + { + "epoch": 0.34, + "learning_rate": 1.546894053814195e-05, + "loss": 0.8686, + "step": 1775 + }, + { + "epoch": 0.34, + "learning_rate": 1.5463799489958727e-05, + "loss": 0.7901, + "step": 1776 + }, + { + "epoch": 0.34, + "learning_rate": 1.545865638226671e-05, + "loss": 0.3063, + "step": 1777 + }, + { + "epoch": 0.34, + "learning_rate": 1.5453511217004526e-05, + "loss": 0.8876, + "step": 1778 + }, + { + "epoch": 0.34, + "learning_rate": 1.5448363996111578e-05, + "loss": 0.8538, + "step": 1779 + }, + { + "epoch": 0.34, + "learning_rate": 1.544321472152804e-05, + "loss": 0.8532, + "step": 1780 + }, + { + "epoch": 0.34, + "learning_rate": 1.5438063395194873e-05, + "loss": 0.9101, + "step": 1781 + }, + { + "epoch": 0.34, + "learning_rate": 1.54329100190538e-05, + "loss": 0.8321, + "step": 1782 + }, + { + "epoch": 0.34, + "learning_rate": 1.542775459504732e-05, + "loss": 0.8736, + "step": 1783 + }, + { + "epoch": 0.34, + "learning_rate": 1.54225971251187e-05, + "loss": 0.859, + "step": 1784 + }, + { + "epoch": 0.34, + "learning_rate": 1.541743761121199e-05, + "loss": 0.8755, + "step": 1785 + }, + { + "epoch": 0.34, + "learning_rate": 1.5412276055271992e-05, + "loss": 0.7513, + "step": 1786 + }, + { + "epoch": 0.34, + "learning_rate": 1.5407112459244293e-05, + "loss": 0.8189, + "step": 1787 + }, + { + "epoch": 0.34, + "learning_rate": 1.5401946825075246e-05, + "loss": 0.2734, + "step": 1788 + }, + { + "epoch": 0.34, + "learning_rate": 1.5396779154711968e-05, + "loss": 0.8251, + "step": 1789 + }, + { + "epoch": 0.34, + "learning_rate": 1.5391609450102346e-05, + "loss": 0.823, + "step": 1790 + }, + { + "epoch": 0.34, + "learning_rate": 1.538643771319503e-05, + "loss": 0.8489, + "step": 1791 + }, + { + "epoch": 0.34, + "learning_rate": 1.5381263945939446e-05, + "loss": 0.7976, + "step": 1792 + }, + { + "epoch": 0.34, + "learning_rate": 1.5376088150285777e-05, + "loss": 0.8255, + "step": 1793 + }, + { + "epoch": 0.34, + "learning_rate": 1.5370910328184968e-05, + "loss": 0.824, + "step": 1794 + }, + { + "epoch": 0.34, + "learning_rate": 1.5365730481588737e-05, + "loss": 0.2674, + "step": 1795 + }, + { + "epoch": 0.34, + "learning_rate": 1.536054861244956e-05, + "loss": 0.8486, + "step": 1796 + }, + { + "epoch": 0.34, + "learning_rate": 1.5355364722720674e-05, + "loss": 0.8905, + "step": 1797 + }, + { + "epoch": 0.34, + "learning_rate": 1.535017881435608e-05, + "loss": 0.7707, + "step": 1798 + }, + { + "epoch": 0.34, + "learning_rate": 1.534499088931054e-05, + "loss": 0.8184, + "step": 1799 + }, + { + "epoch": 0.34, + "learning_rate": 1.5339800949539575e-05, + "loss": 0.8206, + "step": 1800 + }, + { + "epoch": 0.34, + "learning_rate": 1.5334608996999466e-05, + "loss": 0.8332, + "step": 1801 + }, + { + "epoch": 0.34, + "learning_rate": 1.5329415033647253e-05, + "loss": 0.8677, + "step": 1802 + }, + { + "epoch": 0.34, + "learning_rate": 1.532421906144073e-05, + "loss": 0.8459, + "step": 1803 + }, + { + "epoch": 0.34, + "learning_rate": 1.5319021082338458e-05, + "loss": 0.7718, + "step": 1804 + }, + { + "epoch": 0.34, + "learning_rate": 1.5313821098299745e-05, + "loss": 0.8955, + "step": 1805 + }, + { + "epoch": 0.34, + "learning_rate": 1.530861911128466e-05, + "loss": 0.851, + "step": 1806 + }, + { + "epoch": 0.34, + "learning_rate": 1.5303415123254018e-05, + "loss": 0.7602, + "step": 1807 + }, + { + "epoch": 0.34, + "learning_rate": 1.5298209136169403e-05, + "loss": 0.7776, + "step": 1808 + }, + { + "epoch": 0.34, + "learning_rate": 1.5293001151993144e-05, + "loss": 0.7599, + "step": 1809 + }, + { + "epoch": 0.34, + "learning_rate": 1.5287791172688315e-05, + "loss": 0.8753, + "step": 1810 + }, + { + "epoch": 0.34, + "learning_rate": 1.5282579200218762e-05, + "loss": 0.8588, + "step": 1811 + }, + { + "epoch": 0.34, + "learning_rate": 1.5277365236549063e-05, + "loss": 0.788, + "step": 1812 + }, + { + "epoch": 0.34, + "learning_rate": 1.527214928364456e-05, + "loss": 0.8336, + "step": 1813 + }, + { + "epoch": 0.34, + "learning_rate": 1.5266931343471334e-05, + "loss": 0.7653, + "step": 1814 + }, + { + "epoch": 0.34, + "learning_rate": 1.5261711417996224e-05, + "loss": 0.8054, + "step": 1815 + }, + { + "epoch": 0.34, + "learning_rate": 1.5256489509186814e-05, + "loss": 0.7873, + "step": 1816 + }, + { + "epoch": 0.34, + "learning_rate": 1.5251265619011432e-05, + "loss": 0.8106, + "step": 1817 + }, + { + "epoch": 0.34, + "learning_rate": 1.5246039749439159e-05, + "loss": 0.8438, + "step": 1818 + }, + { + "epoch": 0.34, + "learning_rate": 1.5240811902439813e-05, + "loss": 0.8042, + "step": 1819 + }, + { + "epoch": 0.34, + "learning_rate": 1.5235582079983975e-05, + "loss": 0.7788, + "step": 1820 + }, + { + "epoch": 0.35, + "learning_rate": 1.5230350284042947e-05, + "loss": 0.801, + "step": 1821 + }, + { + "epoch": 0.35, + "learning_rate": 1.5225116516588796e-05, + "loss": 0.7919, + "step": 1822 + }, + { + "epoch": 0.35, + "learning_rate": 1.5219880779594319e-05, + "loss": 0.8674, + "step": 1823 + }, + { + "epoch": 0.35, + "learning_rate": 1.521464307503306e-05, + "loss": 0.8522, + "step": 1824 + }, + { + "epoch": 0.35, + "learning_rate": 1.5209403404879305e-05, + "loss": 0.8726, + "step": 1825 + }, + { + "epoch": 0.35, + "learning_rate": 1.520416177110808e-05, + "loss": 0.8055, + "step": 1826 + }, + { + "epoch": 0.35, + "learning_rate": 1.519891817569515e-05, + "loss": 0.8928, + "step": 1827 + }, + { + "epoch": 0.35, + "learning_rate": 1.5193672620617022e-05, + "loss": 0.8738, + "step": 1828 + }, + { + "epoch": 0.35, + "learning_rate": 1.5188425107850942e-05, + "loss": 0.7935, + "step": 1829 + }, + { + "epoch": 0.35, + "learning_rate": 1.5183175639374886e-05, + "loss": 0.8376, + "step": 1830 + }, + { + "epoch": 0.35, + "learning_rate": 1.517792421716758e-05, + "loss": 0.7456, + "step": 1831 + }, + { + "epoch": 0.35, + "learning_rate": 1.5172670843208477e-05, + "loss": 0.7965, + "step": 1832 + }, + { + "epoch": 0.35, + "learning_rate": 1.516741551947777e-05, + "loss": 0.8209, + "step": 1833 + }, + { + "epoch": 0.35, + "learning_rate": 1.5162158247956388e-05, + "loss": 0.8262, + "step": 1834 + }, + { + "epoch": 0.35, + "learning_rate": 1.5156899030625987e-05, + "loss": 0.8097, + "step": 1835 + }, + { + "epoch": 0.35, + "learning_rate": 1.515163786946896e-05, + "loss": 0.293, + "step": 1836 + }, + { + "epoch": 0.35, + "learning_rate": 1.5146374766468442e-05, + "loss": 0.791, + "step": 1837 + }, + { + "epoch": 0.35, + "learning_rate": 1.5141109723608283e-05, + "loss": 0.8724, + "step": 1838 + }, + { + "epoch": 0.35, + "learning_rate": 1.5135842742873077e-05, + "loss": 0.3077, + "step": 1839 + }, + { + "epoch": 0.35, + "learning_rate": 1.5130573826248144e-05, + "loss": 0.2882, + "step": 1840 + }, + { + "epoch": 0.35, + "learning_rate": 1.5125302975719535e-05, + "loss": 0.7305, + "step": 1841 + }, + { + "epoch": 0.35, + "learning_rate": 1.512003019327403e-05, + "loss": 0.8046, + "step": 1842 + }, + { + "epoch": 0.35, + "learning_rate": 1.5114755480899132e-05, + "loss": 0.8774, + "step": 1843 + }, + { + "epoch": 0.35, + "learning_rate": 1.510947884058308e-05, + "loss": 0.8282, + "step": 1844 + }, + { + "epoch": 0.35, + "learning_rate": 1.5104200274314834e-05, + "loss": 0.8558, + "step": 1845 + }, + { + "epoch": 0.35, + "learning_rate": 1.5098919784084083e-05, + "loss": 0.8492, + "step": 1846 + }, + { + "epoch": 0.35, + "learning_rate": 1.5093637371881239e-05, + "loss": 0.7688, + "step": 1847 + }, + { + "epoch": 0.35, + "learning_rate": 1.5088353039697435e-05, + "loss": 0.7552, + "step": 1848 + }, + { + "epoch": 0.35, + "learning_rate": 1.5083066789524536e-05, + "loss": 0.8001, + "step": 1849 + }, + { + "epoch": 0.35, + "learning_rate": 1.5077778623355129e-05, + "loss": 0.8133, + "step": 1850 + }, + { + "epoch": 0.35, + "learning_rate": 1.5072488543182516e-05, + "loss": 0.8143, + "step": 1851 + }, + { + "epoch": 0.35, + "learning_rate": 1.5067196551000722e-05, + "loss": 0.8035, + "step": 1852 + }, + { + "epoch": 0.35, + "learning_rate": 1.5061902648804503e-05, + "loss": 0.8295, + "step": 1853 + }, + { + "epoch": 0.35, + "learning_rate": 1.505660683858932e-05, + "loss": 0.8054, + "step": 1854 + }, + { + "epoch": 0.35, + "learning_rate": 1.5051309122351364e-05, + "loss": 0.8692, + "step": 1855 + }, + { + "epoch": 0.35, + "learning_rate": 1.504600950208754e-05, + "loss": 0.8051, + "step": 1856 + }, + { + "epoch": 0.35, + "learning_rate": 1.504070797979547e-05, + "loss": 0.8403, + "step": 1857 + }, + { + "epoch": 0.35, + "learning_rate": 1.5035404557473497e-05, + "loss": 0.7935, + "step": 1858 + }, + { + "epoch": 0.35, + "learning_rate": 1.5030099237120674e-05, + "loss": 0.7024, + "step": 1859 + }, + { + "epoch": 0.35, + "learning_rate": 1.502479202073678e-05, + "loss": 0.8139, + "step": 1860 + }, + { + "epoch": 0.35, + "learning_rate": 1.5019482910322297e-05, + "loss": 0.8606, + "step": 1861 + }, + { + "epoch": 0.35, + "learning_rate": 1.5014171907878422e-05, + "loss": 0.8754, + "step": 1862 + }, + { + "epoch": 0.35, + "learning_rate": 1.5008859015407078e-05, + "loss": 0.8803, + "step": 1863 + }, + { + "epoch": 0.35, + "learning_rate": 1.5003544234910883e-05, + "loss": 0.8566, + "step": 1864 + }, + { + "epoch": 0.35, + "learning_rate": 1.499822756839318e-05, + "loss": 0.7143, + "step": 1865 + }, + { + "epoch": 0.35, + "learning_rate": 1.4992909017858017e-05, + "loss": 0.8236, + "step": 1866 + }, + { + "epoch": 0.35, + "learning_rate": 1.4987588585310154e-05, + "loss": 0.7972, + "step": 1867 + }, + { + "epoch": 0.35, + "learning_rate": 1.4982266272755052e-05, + "loss": 0.8512, + "step": 1868 + }, + { + "epoch": 0.35, + "learning_rate": 1.49769420821989e-05, + "loss": 0.764, + "step": 1869 + }, + { + "epoch": 0.35, + "learning_rate": 1.4971616015648576e-05, + "loss": 0.7864, + "step": 1870 + }, + { + "epoch": 0.35, + "learning_rate": 1.4966288075111673e-05, + "loss": 0.867, + "step": 1871 + }, + { + "epoch": 0.35, + "learning_rate": 1.496095826259649e-05, + "loss": 0.2777, + "step": 1872 + }, + { + "epoch": 0.36, + "learning_rate": 1.4955626580112031e-05, + "loss": 0.8636, + "step": 1873 + }, + { + "epoch": 0.36, + "learning_rate": 1.4950293029668004e-05, + "loss": 0.7841, + "step": 1874 + }, + { + "epoch": 0.36, + "learning_rate": 1.4944957613274823e-05, + "loss": 0.2939, + "step": 1875 + }, + { + "epoch": 0.36, + "learning_rate": 1.4939620332943605e-05, + "loss": 0.7793, + "step": 1876 + }, + { + "epoch": 0.36, + "learning_rate": 1.4934281190686168e-05, + "loss": 0.7812, + "step": 1877 + }, + { + "epoch": 0.36, + "learning_rate": 1.4928940188515032e-05, + "loss": 0.8357, + "step": 1878 + }, + { + "epoch": 0.36, + "learning_rate": 1.4923597328443423e-05, + "loss": 0.8672, + "step": 1879 + }, + { + "epoch": 0.36, + "learning_rate": 1.491825261248526e-05, + "loss": 0.8936, + "step": 1880 + }, + { + "epoch": 0.36, + "learning_rate": 1.4912906042655164e-05, + "loss": 0.8428, + "step": 1881 + }, + { + "epoch": 0.36, + "learning_rate": 1.4907557620968458e-05, + "loss": 0.7969, + "step": 1882 + }, + { + "epoch": 0.36, + "learning_rate": 1.4902207349441157e-05, + "loss": 0.849, + "step": 1883 + }, + { + "epoch": 0.36, + "learning_rate": 1.4896855230089985e-05, + "loss": 0.783, + "step": 1884 + }, + { + "epoch": 0.36, + "learning_rate": 1.4891501264932344e-05, + "loss": 0.7687, + "step": 1885 + }, + { + "epoch": 0.36, + "learning_rate": 1.488614545598635e-05, + "loss": 0.8472, + "step": 1886 + }, + { + "epoch": 0.36, + "learning_rate": 1.4880787805270802e-05, + "loss": 0.7269, + "step": 1887 + }, + { + "epoch": 0.36, + "learning_rate": 1.4875428314805195e-05, + "loss": 0.8672, + "step": 1888 + }, + { + "epoch": 0.36, + "learning_rate": 1.4870066986609729e-05, + "loss": 0.8155, + "step": 1889 + }, + { + "epoch": 0.36, + "learning_rate": 1.4864703822705277e-05, + "loss": 0.8153, + "step": 1890 + }, + { + "epoch": 0.36, + "learning_rate": 1.4859338825113423e-05, + "loss": 0.87, + "step": 1891 + }, + { + "epoch": 0.36, + "learning_rate": 1.4853971995856429e-05, + "loss": 0.8329, + "step": 1892 + }, + { + "epoch": 0.36, + "learning_rate": 1.4848603336957252e-05, + "loss": 0.8267, + "step": 1893 + }, + { + "epoch": 0.36, + "learning_rate": 1.4843232850439542e-05, + "loss": 0.8474, + "step": 1894 + }, + { + "epoch": 0.36, + "learning_rate": 1.483786053832763e-05, + "loss": 0.7466, + "step": 1895 + }, + { + "epoch": 0.36, + "learning_rate": 1.4832486402646546e-05, + "loss": 0.7825, + "step": 1896 + }, + { + "epoch": 0.36, + "learning_rate": 1.4827110445421997e-05, + "loss": 0.7992, + "step": 1897 + }, + { + "epoch": 0.36, + "learning_rate": 1.482173266868038e-05, + "loss": 0.796, + "step": 1898 + }, + { + "epoch": 0.36, + "learning_rate": 1.4816353074448781e-05, + "loss": 0.8302, + "step": 1899 + }, + { + "epoch": 0.36, + "learning_rate": 1.4810971664754967e-05, + "loss": 0.782, + "step": 1900 + }, + { + "epoch": 0.36, + "learning_rate": 1.4805588441627395e-05, + "loss": 0.8461, + "step": 1901 + }, + { + "epoch": 0.36, + "learning_rate": 1.4800203407095194e-05, + "loss": 0.8073, + "step": 1902 + }, + { + "epoch": 0.36, + "learning_rate": 1.479481656318819e-05, + "loss": 0.7631, + "step": 1903 + }, + { + "epoch": 0.36, + "learning_rate": 1.4789427911936882e-05, + "loss": 0.2648, + "step": 1904 + }, + { + "epoch": 0.36, + "learning_rate": 1.4784037455372452e-05, + "loss": 0.8131, + "step": 1905 + }, + { + "epoch": 0.36, + "learning_rate": 1.4778645195526764e-05, + "loss": 0.873, + "step": 1906 + }, + { + "epoch": 0.36, + "learning_rate": 1.477325113443236e-05, + "loss": 0.2615, + "step": 1907 + }, + { + "epoch": 0.36, + "learning_rate": 1.4767855274122464e-05, + "loss": 0.2495, + "step": 1908 + }, + { + "epoch": 0.36, + "learning_rate": 1.4762457616630972e-05, + "loss": 0.8246, + "step": 1909 + }, + { + "epoch": 0.36, + "learning_rate": 1.4757058163992466e-05, + "loss": 0.8477, + "step": 1910 + }, + { + "epoch": 0.36, + "learning_rate": 1.4751656918242198e-05, + "loss": 0.8427, + "step": 1911 + }, + { + "epoch": 0.36, + "learning_rate": 1.47462538814161e-05, + "loss": 0.8315, + "step": 1912 + }, + { + "epoch": 0.36, + "learning_rate": 1.4740849055550774e-05, + "loss": 0.2726, + "step": 1913 + }, + { + "epoch": 0.36, + "learning_rate": 1.47354424426835e-05, + "loss": 0.8271, + "step": 1914 + }, + { + "epoch": 0.36, + "learning_rate": 1.473003404485223e-05, + "loss": 0.867, + "step": 1915 + }, + { + "epoch": 0.36, + "learning_rate": 1.4724623864095595e-05, + "loss": 0.7989, + "step": 1916 + }, + { + "epoch": 0.36, + "learning_rate": 1.4719211902452888e-05, + "loss": 0.8509, + "step": 1917 + }, + { + "epoch": 0.36, + "learning_rate": 1.471379816196408e-05, + "loss": 0.8004, + "step": 1918 + }, + { + "epoch": 0.36, + "learning_rate": 1.4708382644669807e-05, + "loss": 0.8079, + "step": 1919 + }, + { + "epoch": 0.36, + "learning_rate": 1.4702965352611385e-05, + "loss": 0.7359, + "step": 1920 + }, + { + "epoch": 0.36, + "learning_rate": 1.4697546287830788e-05, + "loss": 0.7984, + "step": 1921 + }, + { + "epoch": 0.36, + "learning_rate": 1.4692125452370664e-05, + "loss": 0.7897, + "step": 1922 + }, + { + "epoch": 0.36, + "learning_rate": 1.4686702848274328e-05, + "loss": 0.2791, + "step": 1923 + }, + { + "epoch": 0.36, + "learning_rate": 1.4681278477585759e-05, + "loss": 0.8566, + "step": 1924 + }, + { + "epoch": 0.36, + "learning_rate": 1.46758523423496e-05, + "loss": 0.8513, + "step": 1925 + }, + { + "epoch": 0.37, + "learning_rate": 1.467042444461117e-05, + "loss": 0.8134, + "step": 1926 + }, + { + "epoch": 0.37, + "learning_rate": 1.4664994786416442e-05, + "loss": 0.7987, + "step": 1927 + }, + { + "epoch": 0.37, + "learning_rate": 1.4659563369812055e-05, + "loss": 0.8694, + "step": 1928 + }, + { + "epoch": 0.37, + "learning_rate": 1.4654130196845315e-05, + "loss": 0.2793, + "step": 1929 + }, + { + "epoch": 0.37, + "learning_rate": 1.4648695269564182e-05, + "loss": 0.8431, + "step": 1930 + }, + { + "epoch": 0.37, + "learning_rate": 1.4643258590017282e-05, + "loss": 0.8843, + "step": 1931 + }, + { + "epoch": 0.37, + "learning_rate": 1.4637820160253904e-05, + "loss": 0.8324, + "step": 1932 + }, + { + "epoch": 0.37, + "learning_rate": 1.4632379982323994e-05, + "loss": 0.8306, + "step": 1933 + }, + { + "epoch": 0.37, + "learning_rate": 1.4626938058278153e-05, + "loss": 0.8414, + "step": 1934 + }, + { + "epoch": 0.37, + "learning_rate": 1.4621494390167645e-05, + "loss": 0.8313, + "step": 1935 + }, + { + "epoch": 0.37, + "learning_rate": 1.4616048980044398e-05, + "loss": 0.8219, + "step": 1936 + }, + { + "epoch": 0.37, + "learning_rate": 1.461060182996098e-05, + "loss": 0.2309, + "step": 1937 + }, + { + "epoch": 0.37, + "learning_rate": 1.4605152941970628e-05, + "loss": 0.7947, + "step": 1938 + }, + { + "epoch": 0.37, + "learning_rate": 1.459970231812723e-05, + "loss": 0.8678, + "step": 1939 + }, + { + "epoch": 0.37, + "learning_rate": 1.4594249960485329e-05, + "loss": 0.8833, + "step": 1940 + }, + { + "epoch": 0.37, + "learning_rate": 1.458879587110012e-05, + "loss": 0.8563, + "step": 1941 + }, + { + "epoch": 0.37, + "learning_rate": 1.458334005202745e-05, + "loss": 0.8031, + "step": 1942 + }, + { + "epoch": 0.37, + "learning_rate": 1.4577882505323821e-05, + "loss": 0.7813, + "step": 1943 + }, + { + "epoch": 0.37, + "learning_rate": 1.4572423233046386e-05, + "loss": 0.8517, + "step": 1944 + }, + { + "epoch": 0.37, + "learning_rate": 1.4566962237252948e-05, + "loss": 0.8297, + "step": 1945 + }, + { + "epoch": 0.37, + "learning_rate": 1.4561499520001956e-05, + "loss": 0.8278, + "step": 1946 + }, + { + "epoch": 0.37, + "learning_rate": 1.4556035083352511e-05, + "loss": 0.8669, + "step": 1947 + }, + { + "epoch": 0.37, + "learning_rate": 1.4550568929364365e-05, + "loss": 0.8121, + "step": 1948 + }, + { + "epoch": 0.37, + "learning_rate": 1.4545101060097913e-05, + "loss": 0.8333, + "step": 1949 + }, + { + "epoch": 0.37, + "learning_rate": 1.4539631477614198e-05, + "loss": 0.843, + "step": 1950 + }, + { + "epoch": 0.37, + "learning_rate": 1.4534160183974908e-05, + "loss": 0.8242, + "step": 1951 + }, + { + "epoch": 0.37, + "learning_rate": 1.4528687181242375e-05, + "loss": 0.83, + "step": 1952 + }, + { + "epoch": 0.37, + "learning_rate": 1.452321247147958e-05, + "loss": 0.7759, + "step": 1953 + }, + { + "epoch": 0.37, + "learning_rate": 1.4517736056750142e-05, + "loss": 0.7392, + "step": 1954 + }, + { + "epoch": 0.37, + "learning_rate": 1.4512257939118328e-05, + "loss": 0.2788, + "step": 1955 + }, + { + "epoch": 0.37, + "learning_rate": 1.4506778120649038e-05, + "loss": 0.7573, + "step": 1956 + }, + { + "epoch": 0.37, + "learning_rate": 1.4501296603407827e-05, + "loss": 0.8282, + "step": 1957 + }, + { + "epoch": 0.37, + "learning_rate": 1.4495813389460875e-05, + "loss": 0.8503, + "step": 1958 + }, + { + "epoch": 0.37, + "learning_rate": 1.4490328480875011e-05, + "loss": 0.8263, + "step": 1959 + }, + { + "epoch": 0.37, + "learning_rate": 1.4484841879717703e-05, + "loss": 0.7976, + "step": 1960 + }, + { + "epoch": 0.37, + "learning_rate": 1.4479353588057052e-05, + "loss": 0.8933, + "step": 1961 + }, + { + "epoch": 0.37, + "learning_rate": 1.44738636079618e-05, + "loss": 0.8008, + "step": 1962 + }, + { + "epoch": 0.37, + "learning_rate": 1.4468371941501323e-05, + "loss": 0.834, + "step": 1963 + }, + { + "epoch": 0.37, + "learning_rate": 1.4462878590745634e-05, + "loss": 0.8457, + "step": 1964 + }, + { + "epoch": 0.37, + "learning_rate": 1.4457383557765385e-05, + "loss": 0.8533, + "step": 1965 + }, + { + "epoch": 0.37, + "learning_rate": 1.445188684463185e-05, + "loss": 0.8294, + "step": 1966 + }, + { + "epoch": 0.37, + "learning_rate": 1.444638845341695e-05, + "loss": 0.7975, + "step": 1967 + }, + { + "epoch": 0.37, + "learning_rate": 1.4440888386193235e-05, + "loss": 0.7994, + "step": 1968 + }, + { + "epoch": 0.37, + "learning_rate": 1.4435386645033878e-05, + "loss": 0.2645, + "step": 1969 + }, + { + "epoch": 0.37, + "learning_rate": 1.4429883232012691e-05, + "loss": 0.7789, + "step": 1970 + }, + { + "epoch": 0.37, + "learning_rate": 1.4424378149204117e-05, + "loss": 0.7765, + "step": 1971 + }, + { + "epoch": 0.37, + "learning_rate": 1.4418871398683227e-05, + "loss": 0.8157, + "step": 1972 + }, + { + "epoch": 0.37, + "learning_rate": 1.4413362982525714e-05, + "loss": 0.8556, + "step": 1973 + }, + { + "epoch": 0.37, + "learning_rate": 1.440785290280791e-05, + "loss": 0.7871, + "step": 1974 + }, + { + "epoch": 0.37, + "learning_rate": 1.4402341161606766e-05, + "loss": 0.8338, + "step": 1975 + }, + { + "epoch": 0.37, + "learning_rate": 1.4396827760999861e-05, + "loss": 0.8311, + "step": 1976 + }, + { + "epoch": 0.37, + "learning_rate": 1.4391312703065403e-05, + "loss": 0.8672, + "step": 1977 + }, + { + "epoch": 0.37, + "learning_rate": 1.4385795989882221e-05, + "loss": 0.8163, + "step": 1978 + }, + { + "epoch": 0.38, + "learning_rate": 1.4380277623529766e-05, + "loss": 0.8358, + "step": 1979 + }, + { + "epoch": 0.38, + "learning_rate": 1.4374757606088119e-05, + "loss": 0.8403, + "step": 1980 + }, + { + "epoch": 0.38, + "learning_rate": 1.4369235939637977e-05, + "loss": 0.8244, + "step": 1981 + }, + { + "epoch": 0.38, + "learning_rate": 1.4363712626260664e-05, + "loss": 0.8816, + "step": 1982 + }, + { + "epoch": 0.38, + "learning_rate": 1.4358187668038118e-05, + "loss": 0.8557, + "step": 1983 + }, + { + "epoch": 0.38, + "learning_rate": 1.4352661067052902e-05, + "loss": 0.8412, + "step": 1984 + }, + { + "epoch": 0.38, + "learning_rate": 1.4347132825388199e-05, + "loss": 0.8174, + "step": 1985 + }, + { + "epoch": 0.38, + "learning_rate": 1.4341602945127806e-05, + "loss": 0.8449, + "step": 1986 + }, + { + "epoch": 0.38, + "learning_rate": 1.4336071428356143e-05, + "loss": 0.8024, + "step": 1987 + }, + { + "epoch": 0.38, + "learning_rate": 1.4330538277158239e-05, + "loss": 0.832, + "step": 1988 + }, + { + "epoch": 0.38, + "learning_rate": 1.4325003493619749e-05, + "loss": 0.8174, + "step": 1989 + }, + { + "epoch": 0.38, + "learning_rate": 1.4319467079826937e-05, + "loss": 0.8103, + "step": 1990 + }, + { + "epoch": 0.38, + "learning_rate": 1.4313929037866679e-05, + "loss": 0.8201, + "step": 1991 + }, + { + "epoch": 0.38, + "learning_rate": 1.4308389369826474e-05, + "loss": 0.8638, + "step": 1992 + }, + { + "epoch": 0.38, + "learning_rate": 1.4302848077794427e-05, + "loss": 0.84, + "step": 1993 + }, + { + "epoch": 0.38, + "learning_rate": 1.4297305163859254e-05, + "loss": 0.8407, + "step": 1994 + }, + { + "epoch": 0.38, + "learning_rate": 1.429176063011029e-05, + "loss": 0.8145, + "step": 1995 + }, + { + "epoch": 0.38, + "learning_rate": 1.4286214478637468e-05, + "loss": 0.8677, + "step": 1996 + }, + { + "epoch": 0.38, + "learning_rate": 1.4280666711531346e-05, + "loss": 0.8259, + "step": 1997 + }, + { + "epoch": 0.38, + "learning_rate": 1.427511733088308e-05, + "loss": 0.7857, + "step": 1998 + }, + { + "epoch": 0.38, + "learning_rate": 1.4269566338784436e-05, + "loss": 0.879, + "step": 1999 + }, + { + "epoch": 0.38, + "learning_rate": 1.426401373732779e-05, + "loss": 0.8352, + "step": 2000 + }, + { + "epoch": 0.38, + "learning_rate": 1.4258459528606124e-05, + "loss": 0.8387, + "step": 2001 + }, + { + "epoch": 0.38, + "learning_rate": 1.4252903714713026e-05, + "loss": 0.8248, + "step": 2002 + }, + { + "epoch": 0.38, + "learning_rate": 1.4247346297742688e-05, + "loss": 0.7715, + "step": 2003 + }, + { + "epoch": 0.38, + "learning_rate": 1.4241787279789903e-05, + "loss": 0.7846, + "step": 2004 + }, + { + "epoch": 0.38, + "learning_rate": 1.4236226662950076e-05, + "loss": 0.8624, + "step": 2005 + }, + { + "epoch": 0.38, + "learning_rate": 1.4230664449319212e-05, + "loss": 0.2443, + "step": 2006 + }, + { + "epoch": 0.38, + "learning_rate": 1.422510064099391e-05, + "loss": 0.8116, + "step": 2007 + }, + { + "epoch": 0.38, + "learning_rate": 1.4219535240071378e-05, + "loss": 0.8301, + "step": 2008 + }, + { + "epoch": 0.38, + "learning_rate": 1.4213968248649424e-05, + "loss": 0.858, + "step": 2009 + }, + { + "epoch": 0.38, + "learning_rate": 1.420839966882645e-05, + "loss": 0.7823, + "step": 2010 + }, + { + "epoch": 0.38, + "learning_rate": 1.4202829502701462e-05, + "loss": 0.8986, + "step": 2011 + }, + { + "epoch": 0.38, + "learning_rate": 1.4197257752374062e-05, + "loss": 0.7933, + "step": 2012 + }, + { + "epoch": 0.38, + "learning_rate": 1.4191684419944454e-05, + "loss": 0.8749, + "step": 2013 + }, + { + "epoch": 0.38, + "learning_rate": 1.4186109507513425e-05, + "loss": 0.8829, + "step": 2014 + }, + { + "epoch": 0.38, + "learning_rate": 1.4180533017182374e-05, + "loss": 0.8562, + "step": 2015 + }, + { + "epoch": 0.38, + "learning_rate": 1.4174954951053284e-05, + "loss": 0.8384, + "step": 2016 + }, + { + "epoch": 0.38, + "learning_rate": 1.4169375311228736e-05, + "loss": 0.8706, + "step": 2017 + }, + { + "epoch": 0.38, + "learning_rate": 1.4163794099811901e-05, + "loss": 0.8036, + "step": 2018 + }, + { + "epoch": 0.38, + "learning_rate": 1.4158211318906547e-05, + "loss": 0.8332, + "step": 2019 + }, + { + "epoch": 0.38, + "learning_rate": 1.4152626970617031e-05, + "loss": 0.8514, + "step": 2020 + }, + { + "epoch": 0.38, + "learning_rate": 1.4147041057048303e-05, + "loss": 0.8094, + "step": 2021 + }, + { + "epoch": 0.38, + "learning_rate": 1.4141453580305897e-05, + "loss": 0.8113, + "step": 2022 + }, + { + "epoch": 0.38, + "learning_rate": 1.4135864542495944e-05, + "loss": 0.8872, + "step": 2023 + }, + { + "epoch": 0.38, + "learning_rate": 1.4130273945725154e-05, + "loss": 0.8209, + "step": 2024 + }, + { + "epoch": 0.38, + "learning_rate": 1.4124681792100838e-05, + "loss": 0.7153, + "step": 2025 + }, + { + "epoch": 0.38, + "learning_rate": 1.4119088083730883e-05, + "loss": 0.8113, + "step": 2026 + }, + { + "epoch": 0.38, + "learning_rate": 1.4113492822723765e-05, + "loss": 0.8366, + "step": 2027 + }, + { + "epoch": 0.38, + "learning_rate": 1.4107896011188546e-05, + "loss": 0.7852, + "step": 2028 + }, + { + "epoch": 0.38, + "learning_rate": 1.410229765123487e-05, + "loss": 0.8472, + "step": 2029 + }, + { + "epoch": 0.38, + "learning_rate": 1.4096697744972972e-05, + "loss": 0.8154, + "step": 2030 + }, + { + "epoch": 0.38, + "learning_rate": 1.4091096294513658e-05, + "loss": 0.8343, + "step": 2031 + }, + { + "epoch": 0.39, + "learning_rate": 1.4085493301968325e-05, + "loss": 0.8871, + "step": 2032 + }, + { + "epoch": 0.39, + "learning_rate": 1.4079888769448949e-05, + "loss": 0.8322, + "step": 2033 + }, + { + "epoch": 0.39, + "learning_rate": 1.4074282699068086e-05, + "loss": 0.8497, + "step": 2034 + }, + { + "epoch": 0.39, + "learning_rate": 1.4068675092938872e-05, + "loss": 0.7895, + "step": 2035 + }, + { + "epoch": 0.39, + "learning_rate": 1.4063065953175026e-05, + "loss": 0.8066, + "step": 2036 + }, + { + "epoch": 0.39, + "learning_rate": 1.4057455281890834e-05, + "loss": 0.8739, + "step": 2037 + }, + { + "epoch": 0.39, + "learning_rate": 1.405184308120117e-05, + "loss": 0.7454, + "step": 2038 + }, + { + "epoch": 0.39, + "learning_rate": 1.404622935322148e-05, + "loss": 0.857, + "step": 2039 + }, + { + "epoch": 0.39, + "learning_rate": 1.4040614100067789e-05, + "loss": 0.81, + "step": 2040 + }, + { + "epoch": 0.39, + "learning_rate": 1.4034997323856687e-05, + "loss": 0.867, + "step": 2041 + }, + { + "epoch": 0.39, + "learning_rate": 1.4029379026705352e-05, + "loss": 0.8245, + "step": 2042 + }, + { + "epoch": 0.39, + "learning_rate": 1.4023759210731527e-05, + "loss": 0.7524, + "step": 2043 + }, + { + "epoch": 0.39, + "learning_rate": 1.401813787805353e-05, + "loss": 0.8321, + "step": 2044 + }, + { + "epoch": 0.39, + "learning_rate": 1.401251503079025e-05, + "loss": 0.8259, + "step": 2045 + }, + { + "epoch": 0.39, + "learning_rate": 1.4006890671061143e-05, + "loss": 0.8741, + "step": 2046 + }, + { + "epoch": 0.39, + "learning_rate": 1.4001264800986243e-05, + "loss": 0.8463, + "step": 2047 + }, + { + "epoch": 0.39, + "learning_rate": 1.3995637422686147e-05, + "loss": 0.8187, + "step": 2048 + }, + { + "epoch": 0.39, + "learning_rate": 1.3990008538282027e-05, + "loss": 0.8035, + "step": 2049 + }, + { + "epoch": 0.39, + "learning_rate": 1.3984378149895612e-05, + "loss": 0.7905, + "step": 2050 + }, + { + "epoch": 0.39, + "learning_rate": 1.397874625964921e-05, + "loss": 0.8433, + "step": 2051 + }, + { + "epoch": 0.39, + "learning_rate": 1.3973112869665686e-05, + "loss": 0.8089, + "step": 2052 + }, + { + "epoch": 0.39, + "learning_rate": 1.3967477982068473e-05, + "loss": 0.817, + "step": 2053 + }, + { + "epoch": 0.39, + "learning_rate": 1.396184159898157e-05, + "loss": 0.8468, + "step": 2054 + }, + { + "epoch": 0.39, + "learning_rate": 1.3956203722529545e-05, + "loss": 0.7923, + "step": 2055 + }, + { + "epoch": 0.39, + "learning_rate": 1.3950564354837512e-05, + "loss": 0.792, + "step": 2056 + }, + { + "epoch": 0.39, + "learning_rate": 1.3944923498031169e-05, + "loss": 0.3155, + "step": 2057 + }, + { + "epoch": 0.39, + "learning_rate": 1.3939281154236755e-05, + "loss": 0.8246, + "step": 2058 + }, + { + "epoch": 0.39, + "learning_rate": 1.3933637325581084e-05, + "loss": 0.8139, + "step": 2059 + }, + { + "epoch": 0.39, + "learning_rate": 1.3927992014191523e-05, + "loss": 0.3043, + "step": 2060 + }, + { + "epoch": 0.39, + "learning_rate": 1.3922345222195997e-05, + "loss": 0.7696, + "step": 2061 + }, + { + "epoch": 0.39, + "learning_rate": 1.3916696951722997e-05, + "loss": 0.8441, + "step": 2062 + }, + { + "epoch": 0.39, + "learning_rate": 1.391104720490156e-05, + "loss": 0.8466, + "step": 2063 + }, + { + "epoch": 0.39, + "learning_rate": 1.3905395983861291e-05, + "loss": 0.8678, + "step": 2064 + }, + { + "epoch": 0.39, + "learning_rate": 1.3899743290732342e-05, + "loss": 0.8199, + "step": 2065 + }, + { + "epoch": 0.39, + "learning_rate": 1.389408912764542e-05, + "loss": 0.7966, + "step": 2066 + }, + { + "epoch": 0.39, + "learning_rate": 1.3888433496731795e-05, + "loss": 0.7712, + "step": 2067 + }, + { + "epoch": 0.39, + "learning_rate": 1.3882776400123278e-05, + "loss": 0.8255, + "step": 2068 + }, + { + "epoch": 0.39, + "learning_rate": 1.3877117839952243e-05, + "loss": 0.8004, + "step": 2069 + }, + { + "epoch": 0.39, + "learning_rate": 1.387145781835161e-05, + "loss": 0.8333, + "step": 2070 + }, + { + "epoch": 0.39, + "learning_rate": 1.386579633745485e-05, + "loss": 0.8725, + "step": 2071 + }, + { + "epoch": 0.39, + "learning_rate": 1.3860133399395987e-05, + "loss": 0.8526, + "step": 2072 + }, + { + "epoch": 0.39, + "learning_rate": 1.3854469006309592e-05, + "loss": 0.8151, + "step": 2073 + }, + { + "epoch": 0.39, + "learning_rate": 1.3848803160330784e-05, + "loss": 0.8561, + "step": 2074 + }, + { + "epoch": 0.39, + "learning_rate": 1.384313586359523e-05, + "loss": 0.8217, + "step": 2075 + }, + { + "epoch": 0.39, + "learning_rate": 1.3837467118239147e-05, + "loss": 0.7972, + "step": 2076 + }, + { + "epoch": 0.39, + "learning_rate": 1.3831796926399295e-05, + "loss": 0.8266, + "step": 2077 + }, + { + "epoch": 0.39, + "learning_rate": 1.382612529021298e-05, + "loss": 0.8096, + "step": 2078 + }, + { + "epoch": 0.39, + "learning_rate": 1.3820452211818047e-05, + "loss": 0.2947, + "step": 2079 + }, + { + "epoch": 0.39, + "learning_rate": 1.3814777693352899e-05, + "loss": 0.8084, + "step": 2080 + }, + { + "epoch": 0.39, + "learning_rate": 1.3809101736956465e-05, + "loss": 0.7737, + "step": 2081 + }, + { + "epoch": 0.39, + "learning_rate": 1.3803424344768223e-05, + "loss": 0.8667, + "step": 2082 + }, + { + "epoch": 0.39, + "learning_rate": 1.3797745518928201e-05, + "loss": 0.8566, + "step": 2083 + }, + { + "epoch": 0.39, + "learning_rate": 1.3792065261576953e-05, + "loss": 0.8736, + "step": 2084 + }, + { + "epoch": 0.4, + "learning_rate": 1.378638357485558e-05, + "loss": 0.8134, + "step": 2085 + }, + { + "epoch": 0.4, + "learning_rate": 1.3780700460905721e-05, + "loss": 0.7788, + "step": 2086 + }, + { + "epoch": 0.4, + "learning_rate": 1.377501592186955e-05, + "loss": 0.7916, + "step": 2087 + }, + { + "epoch": 0.4, + "learning_rate": 1.3769329959889784e-05, + "loss": 0.8283, + "step": 2088 + }, + { + "epoch": 0.4, + "learning_rate": 1.3763642577109673e-05, + "loss": 0.2874, + "step": 2089 + }, + { + "epoch": 0.4, + "learning_rate": 1.3757953775672999e-05, + "loss": 0.8281, + "step": 2090 + }, + { + "epoch": 0.4, + "learning_rate": 1.3752263557724088e-05, + "loss": 0.8293, + "step": 2091 + }, + { + "epoch": 0.4, + "learning_rate": 1.374657192540779e-05, + "loss": 0.8292, + "step": 2092 + }, + { + "epoch": 0.4, + "learning_rate": 1.3740878880869493e-05, + "loss": 0.8723, + "step": 2093 + }, + { + "epoch": 0.4, + "learning_rate": 1.3735184426255117e-05, + "loss": 0.7562, + "step": 2094 + }, + { + "epoch": 0.4, + "learning_rate": 1.3729488563711116e-05, + "loss": 0.8693, + "step": 2095 + }, + { + "epoch": 0.4, + "learning_rate": 1.3723791295384467e-05, + "loss": 0.852, + "step": 2096 + }, + { + "epoch": 0.4, + "learning_rate": 1.3718092623422688e-05, + "loss": 0.8343, + "step": 2097 + }, + { + "epoch": 0.4, + "learning_rate": 1.3712392549973814e-05, + "loss": 0.8129, + "step": 2098 + }, + { + "epoch": 0.4, + "learning_rate": 1.3706691077186415e-05, + "loss": 0.8406, + "step": 2099 + }, + { + "epoch": 0.4, + "learning_rate": 1.3700988207209594e-05, + "loss": 0.8389, + "step": 2100 + }, + { + "epoch": 0.4, + "learning_rate": 1.3695283942192965e-05, + "loss": 0.8371, + "step": 2101 + }, + { + "epoch": 0.4, + "learning_rate": 1.3689578284286685e-05, + "loss": 0.7671, + "step": 2102 + }, + { + "epoch": 0.4, + "learning_rate": 1.3683871235641424e-05, + "loss": 0.8517, + "step": 2103 + }, + { + "epoch": 0.4, + "learning_rate": 1.367816279840838e-05, + "loss": 0.834, + "step": 2104 + }, + { + "epoch": 0.4, + "learning_rate": 1.3672452974739278e-05, + "loss": 0.8491, + "step": 2105 + }, + { + "epoch": 0.4, + "learning_rate": 1.3666741766786362e-05, + "loss": 0.8469, + "step": 2106 + }, + { + "epoch": 0.4, + "learning_rate": 1.3661029176702395e-05, + "loss": 0.865, + "step": 2107 + }, + { + "epoch": 0.4, + "learning_rate": 1.365531520664067e-05, + "loss": 0.7999, + "step": 2108 + }, + { + "epoch": 0.4, + "learning_rate": 1.364959985875499e-05, + "loss": 0.3046, + "step": 2109 + }, + { + "epoch": 0.4, + "learning_rate": 1.3643883135199677e-05, + "loss": 0.8403, + "step": 2110 + }, + { + "epoch": 0.4, + "learning_rate": 1.3638165038129586e-05, + "loss": 0.8275, + "step": 2111 + }, + { + "epoch": 0.4, + "learning_rate": 1.3632445569700078e-05, + "loss": 0.8497, + "step": 2112 + }, + { + "epoch": 0.4, + "learning_rate": 1.3626724732067027e-05, + "loss": 0.8919, + "step": 2113 + }, + { + "epoch": 0.4, + "learning_rate": 1.3621002527386832e-05, + "loss": 0.8221, + "step": 2114 + }, + { + "epoch": 0.4, + "learning_rate": 1.3615278957816406e-05, + "loss": 0.7897, + "step": 2115 + }, + { + "epoch": 0.4, + "learning_rate": 1.3609554025513175e-05, + "loss": 0.8232, + "step": 2116 + }, + { + "epoch": 0.4, + "learning_rate": 1.3603827732635075e-05, + "loss": 0.8097, + "step": 2117 + }, + { + "epoch": 0.4, + "learning_rate": 1.3598100081340562e-05, + "loss": 0.2571, + "step": 2118 + }, + { + "epoch": 0.4, + "learning_rate": 1.3592371073788595e-05, + "loss": 0.8411, + "step": 2119 + }, + { + "epoch": 0.4, + "learning_rate": 1.3586640712138655e-05, + "loss": 0.8771, + "step": 2120 + }, + { + "epoch": 0.4, + "learning_rate": 1.3580908998550725e-05, + "loss": 0.873, + "step": 2121 + }, + { + "epoch": 0.4, + "learning_rate": 1.3575175935185306e-05, + "loss": 0.8402, + "step": 2122 + }, + { + "epoch": 0.4, + "learning_rate": 1.3569441524203395e-05, + "loss": 0.8454, + "step": 2123 + }, + { + "epoch": 0.4, + "learning_rate": 1.3563705767766509e-05, + "loss": 0.8154, + "step": 2124 + }, + { + "epoch": 0.4, + "learning_rate": 1.3557968668036665e-05, + "loss": 0.8121, + "step": 2125 + }, + { + "epoch": 0.4, + "learning_rate": 1.355223022717639e-05, + "loss": 0.768, + "step": 2126 + }, + { + "epoch": 0.4, + "learning_rate": 1.3546490447348715e-05, + "loss": 0.7895, + "step": 2127 + }, + { + "epoch": 0.4, + "learning_rate": 1.3540749330717175e-05, + "loss": 0.8015, + "step": 2128 + }, + { + "epoch": 0.4, + "learning_rate": 1.3535006879445813e-05, + "loss": 0.812, + "step": 2129 + }, + { + "epoch": 0.4, + "learning_rate": 1.352926309569917e-05, + "loss": 0.7522, + "step": 2130 + }, + { + "epoch": 0.4, + "learning_rate": 1.3523517981642287e-05, + "loss": 0.7872, + "step": 2131 + }, + { + "epoch": 0.4, + "learning_rate": 1.3517771539440719e-05, + "loss": 0.8274, + "step": 2132 + }, + { + "epoch": 0.4, + "learning_rate": 1.3512023771260507e-05, + "loss": 0.8402, + "step": 2133 + }, + { + "epoch": 0.4, + "learning_rate": 1.3506274679268203e-05, + "loss": 0.8127, + "step": 2134 + }, + { + "epoch": 0.4, + "learning_rate": 1.3500524265630848e-05, + "loss": 0.8384, + "step": 2135 + }, + { + "epoch": 0.4, + "learning_rate": 1.349477253251599e-05, + "loss": 0.8032, + "step": 2136 + }, + { + "epoch": 0.41, + "learning_rate": 1.348901948209167e-05, + "loss": 0.2697, + "step": 2137 + }, + { + "epoch": 0.41, + "learning_rate": 1.3483265116526424e-05, + "loss": 0.8413, + "step": 2138 + }, + { + "epoch": 0.41, + "learning_rate": 1.3477509437989289e-05, + "loss": 0.8284, + "step": 2139 + }, + { + "epoch": 0.41, + "learning_rate": 1.347175244864979e-05, + "loss": 0.7551, + "step": 2140 + }, + { + "epoch": 0.41, + "learning_rate": 1.3465994150677952e-05, + "loss": 0.7775, + "step": 2141 + }, + { + "epoch": 0.41, + "learning_rate": 1.3460234546244295e-05, + "loss": 0.8165, + "step": 2142 + }, + { + "epoch": 0.41, + "learning_rate": 1.3454473637519825e-05, + "loss": 0.8314, + "step": 2143 + }, + { + "epoch": 0.41, + "learning_rate": 1.3448711426676038e-05, + "loss": 0.8324, + "step": 2144 + }, + { + "epoch": 0.41, + "learning_rate": 1.344294791588493e-05, + "loss": 0.8412, + "step": 2145 + }, + { + "epoch": 0.41, + "learning_rate": 1.343718310731898e-05, + "loss": 0.833, + "step": 2146 + }, + { + "epoch": 0.41, + "learning_rate": 1.3431417003151162e-05, + "loss": 0.8002, + "step": 2147 + }, + { + "epoch": 0.41, + "learning_rate": 1.342564960555493e-05, + "loss": 0.7739, + "step": 2148 + }, + { + "epoch": 0.41, + "learning_rate": 1.3419880916704232e-05, + "loss": 0.8429, + "step": 2149 + }, + { + "epoch": 0.41, + "learning_rate": 1.3414110938773498e-05, + "loss": 0.8, + "step": 2150 + }, + { + "epoch": 0.41, + "learning_rate": 1.3408339673937651e-05, + "loss": 0.8167, + "step": 2151 + }, + { + "epoch": 0.41, + "learning_rate": 1.3402567124372095e-05, + "loss": 0.8, + "step": 2152 + }, + { + "epoch": 0.41, + "learning_rate": 1.3396793292252717e-05, + "loss": 0.7901, + "step": 2153 + }, + { + "epoch": 0.41, + "learning_rate": 1.3391018179755886e-05, + "loss": 0.8387, + "step": 2154 + }, + { + "epoch": 0.41, + "learning_rate": 1.338524178905846e-05, + "loss": 0.8005, + "step": 2155 + }, + { + "epoch": 0.41, + "learning_rate": 1.3379464122337771e-05, + "loss": 0.8237, + "step": 2156 + }, + { + "epoch": 0.41, + "learning_rate": 1.337368518177164e-05, + "loss": 0.8583, + "step": 2157 + }, + { + "epoch": 0.41, + "learning_rate": 1.3367904969538364e-05, + "loss": 0.867, + "step": 2158 + }, + { + "epoch": 0.41, + "learning_rate": 1.3362123487816716e-05, + "loss": 0.2572, + "step": 2159 + }, + { + "epoch": 0.41, + "learning_rate": 1.3356340738785954e-05, + "loss": 0.7767, + "step": 2160 + }, + { + "epoch": 0.41, + "learning_rate": 1.3350556724625809e-05, + "loss": 0.8262, + "step": 2161 + }, + { + "epoch": 0.41, + "learning_rate": 1.3344771447516494e-05, + "loss": 0.8281, + "step": 2162 + }, + { + "epoch": 0.41, + "learning_rate": 1.333898490963869e-05, + "loss": 0.8432, + "step": 2163 + }, + { + "epoch": 0.41, + "learning_rate": 1.3333197113173561e-05, + "loss": 0.8048, + "step": 2164 + }, + { + "epoch": 0.41, + "learning_rate": 1.332740806030274e-05, + "loss": 0.7925, + "step": 2165 + }, + { + "epoch": 0.41, + "learning_rate": 1.3321617753208339e-05, + "loss": 0.8174, + "step": 2166 + }, + { + "epoch": 0.41, + "learning_rate": 1.3315826194072936e-05, + "loss": 0.7671, + "step": 2167 + }, + { + "epoch": 0.41, + "learning_rate": 1.3310033385079589e-05, + "loss": 0.7897, + "step": 2168 + }, + { + "epoch": 0.41, + "learning_rate": 1.3304239328411816e-05, + "loss": 0.8262, + "step": 2169 + }, + { + "epoch": 0.41, + "learning_rate": 1.3298444026253618e-05, + "loss": 0.7814, + "step": 2170 + }, + { + "epoch": 0.41, + "learning_rate": 1.3292647480789457e-05, + "loss": 0.7815, + "step": 2171 + }, + { + "epoch": 0.41, + "learning_rate": 1.3286849694204268e-05, + "loss": 0.8267, + "step": 2172 + }, + { + "epoch": 0.41, + "learning_rate": 1.328105066868345e-05, + "loss": 0.8454, + "step": 2173 + }, + { + "epoch": 0.41, + "learning_rate": 1.3275250406412872e-05, + "loss": 0.9033, + "step": 2174 + }, + { + "epoch": 0.41, + "learning_rate": 1.3269448909578866e-05, + "loss": 0.8108, + "step": 2175 + }, + { + "epoch": 0.41, + "learning_rate": 1.3263646180368233e-05, + "loss": 0.7855, + "step": 2176 + }, + { + "epoch": 0.41, + "learning_rate": 1.3257842220968236e-05, + "loss": 0.8256, + "step": 2177 + }, + { + "epoch": 0.41, + "learning_rate": 1.3252037033566605e-05, + "loss": 0.7888, + "step": 2178 + }, + { + "epoch": 0.41, + "learning_rate": 1.324623062035153e-05, + "loss": 0.8546, + "step": 2179 + }, + { + "epoch": 0.41, + "learning_rate": 1.324042298351166e-05, + "loss": 0.8029, + "step": 2180 + }, + { + "epoch": 0.41, + "learning_rate": 1.3234614125236114e-05, + "loss": 0.8492, + "step": 2181 + }, + { + "epoch": 0.41, + "learning_rate": 1.3228804047714462e-05, + "loss": 0.8561, + "step": 2182 + }, + { + "epoch": 0.41, + "learning_rate": 1.3222992753136743e-05, + "loss": 0.8261, + "step": 2183 + }, + { + "epoch": 0.41, + "learning_rate": 1.3217180243693445e-05, + "loss": 0.7549, + "step": 2184 + }, + { + "epoch": 0.41, + "learning_rate": 1.321136652157552e-05, + "loss": 0.8062, + "step": 2185 + }, + { + "epoch": 0.41, + "learning_rate": 1.3205551588974375e-05, + "loss": 0.8523, + "step": 2186 + }, + { + "epoch": 0.41, + "learning_rate": 1.3199735448081872e-05, + "loss": 0.8312, + "step": 2187 + }, + { + "epoch": 0.41, + "learning_rate": 1.3193918101090335e-05, + "loss": 0.8416, + "step": 2188 + }, + { + "epoch": 0.41, + "learning_rate": 1.3188099550192537e-05, + "loss": 0.8228, + "step": 2189 + }, + { + "epoch": 0.42, + "learning_rate": 1.3182279797581704e-05, + "loss": 0.8299, + "step": 2190 + }, + { + "epoch": 0.42, + "learning_rate": 1.3176458845451516e-05, + "loss": 0.7739, + "step": 2191 + }, + { + "epoch": 0.42, + "learning_rate": 1.3170636695996106e-05, + "loss": 0.7944, + "step": 2192 + }, + { + "epoch": 0.42, + "learning_rate": 1.3164813351410062e-05, + "loss": 0.2964, + "step": 2193 + }, + { + "epoch": 0.42, + "learning_rate": 1.3158988813888414e-05, + "loss": 0.2558, + "step": 2194 + }, + { + "epoch": 0.42, + "learning_rate": 1.315316308562665e-05, + "loss": 0.2827, + "step": 2195 + }, + { + "epoch": 0.42, + "learning_rate": 1.31473361688207e-05, + "loss": 0.8791, + "step": 2196 + }, + { + "epoch": 0.42, + "learning_rate": 1.3141508065666946e-05, + "loss": 0.8176, + "step": 2197 + }, + { + "epoch": 0.42, + "learning_rate": 1.313567877836222e-05, + "loss": 0.8498, + "step": 2198 + }, + { + "epoch": 0.42, + "learning_rate": 1.312984830910379e-05, + "loss": 0.7961, + "step": 2199 + }, + { + "epoch": 0.42, + "learning_rate": 1.3124016660089383e-05, + "loss": 0.9101, + "step": 2200 + }, + { + "epoch": 0.42, + "learning_rate": 1.311818383351716e-05, + "loss": 0.866, + "step": 2201 + }, + { + "epoch": 0.42, + "learning_rate": 1.3112349831585732e-05, + "loss": 0.9013, + "step": 2202 + }, + { + "epoch": 0.42, + "learning_rate": 1.3106514656494147e-05, + "loss": 0.7863, + "step": 2203 + }, + { + "epoch": 0.42, + "learning_rate": 1.3100678310441905e-05, + "loss": 0.8192, + "step": 2204 + }, + { + "epoch": 0.42, + "learning_rate": 1.3094840795628937e-05, + "loss": 0.801, + "step": 2205 + }, + { + "epoch": 0.42, + "learning_rate": 1.3089002114255619e-05, + "loss": 0.8151, + "step": 2206 + }, + { + "epoch": 0.42, + "learning_rate": 1.3083162268522766e-05, + "loss": 0.8367, + "step": 2207 + }, + { + "epoch": 0.42, + "learning_rate": 1.3077321260631636e-05, + "loss": 0.8012, + "step": 2208 + }, + { + "epoch": 0.42, + "learning_rate": 1.307147909278392e-05, + "loss": 0.8518, + "step": 2209 + }, + { + "epoch": 0.42, + "learning_rate": 1.3065635767181748e-05, + "loss": 0.8623, + "step": 2210 + }, + { + "epoch": 0.42, + "learning_rate": 1.3059791286027688e-05, + "loss": 0.838, + "step": 2211 + }, + { + "epoch": 0.42, + "learning_rate": 1.3053945651524735e-05, + "loss": 0.8553, + "step": 2212 + }, + { + "epoch": 0.42, + "learning_rate": 1.3048098865876334e-05, + "loss": 0.8366, + "step": 2213 + }, + { + "epoch": 0.42, + "learning_rate": 1.304225093128635e-05, + "loss": 0.8251, + "step": 2214 + }, + { + "epoch": 0.42, + "learning_rate": 1.303640184995909e-05, + "loss": 0.783, + "step": 2215 + }, + { + "epoch": 0.42, + "learning_rate": 1.3030551624099288e-05, + "loss": 0.819, + "step": 2216 + }, + { + "epoch": 0.42, + "learning_rate": 1.302470025591211e-05, + "loss": 0.8587, + "step": 2217 + }, + { + "epoch": 0.42, + "learning_rate": 1.3018847747603158e-05, + "loss": 0.7733, + "step": 2218 + }, + { + "epoch": 0.42, + "learning_rate": 1.3012994101378455e-05, + "loss": 0.8391, + "step": 2219 + }, + { + "epoch": 0.42, + "learning_rate": 1.3007139319444462e-05, + "loss": 0.8229, + "step": 2220 + }, + { + "epoch": 0.42, + "learning_rate": 1.3001283404008059e-05, + "loss": 0.7743, + "step": 2221 + }, + { + "epoch": 0.42, + "learning_rate": 1.2995426357276561e-05, + "loss": 0.8434, + "step": 2222 + }, + { + "epoch": 0.42, + "learning_rate": 1.2989568181457704e-05, + "loss": 0.7929, + "step": 2223 + }, + { + "epoch": 0.42, + "learning_rate": 1.2983708878759655e-05, + "loss": 0.781, + "step": 2224 + }, + { + "epoch": 0.42, + "learning_rate": 1.2977848451390999e-05, + "loss": 0.8231, + "step": 2225 + }, + { + "epoch": 0.42, + "learning_rate": 1.2971986901560751e-05, + "loss": 0.7944, + "step": 2226 + }, + { + "epoch": 0.42, + "learning_rate": 1.2966124231478348e-05, + "loss": 0.8422, + "step": 2227 + }, + { + "epoch": 0.42, + "learning_rate": 1.2960260443353642e-05, + "loss": 0.8044, + "step": 2228 + }, + { + "epoch": 0.42, + "learning_rate": 1.2954395539396917e-05, + "loss": 0.845, + "step": 2229 + }, + { + "epoch": 0.42, + "learning_rate": 1.2948529521818873e-05, + "loss": 0.2975, + "step": 2230 + }, + { + "epoch": 0.42, + "learning_rate": 1.2942662392830632e-05, + "loss": 0.8235, + "step": 2231 + }, + { + "epoch": 0.42, + "learning_rate": 1.2936794154643727e-05, + "loss": 0.8062, + "step": 2232 + }, + { + "epoch": 0.42, + "learning_rate": 1.2930924809470115e-05, + "loss": 0.7548, + "step": 2233 + }, + { + "epoch": 0.42, + "learning_rate": 1.2925054359522174e-05, + "loss": 0.8126, + "step": 2234 + }, + { + "epoch": 0.42, + "learning_rate": 1.2919182807012691e-05, + "loss": 0.7937, + "step": 2235 + }, + { + "epoch": 0.42, + "learning_rate": 1.2913310154154876e-05, + "loss": 0.8585, + "step": 2236 + }, + { + "epoch": 0.42, + "learning_rate": 1.2907436403162348e-05, + "loss": 0.8009, + "step": 2237 + }, + { + "epoch": 0.42, + "learning_rate": 1.290156155624914e-05, + "loss": 0.806, + "step": 2238 + }, + { + "epoch": 0.42, + "learning_rate": 1.28956856156297e-05, + "loss": 0.7986, + "step": 2239 + }, + { + "epoch": 0.42, + "learning_rate": 1.2889808583518892e-05, + "loss": 0.8172, + "step": 2240 + }, + { + "epoch": 0.42, + "learning_rate": 1.2883930462131983e-05, + "loss": 0.9006, + "step": 2241 + }, + { + "epoch": 0.42, + "learning_rate": 1.287805125368466e-05, + "loss": 0.255, + "step": 2242 + }, + { + "epoch": 0.43, + "learning_rate": 1.2872170960393006e-05, + "loss": 0.8745, + "step": 2243 + }, + { + "epoch": 0.43, + "learning_rate": 1.286628958447353e-05, + "loss": 0.7973, + "step": 2244 + }, + { + "epoch": 0.43, + "learning_rate": 1.286040712814314e-05, + "loss": 0.8352, + "step": 2245 + }, + { + "epoch": 0.43, + "learning_rate": 1.2854523593619148e-05, + "loss": 0.793, + "step": 2246 + }, + { + "epoch": 0.43, + "learning_rate": 1.2848638983119281e-05, + "loss": 0.8256, + "step": 2247 + }, + { + "epoch": 0.43, + "learning_rate": 1.2842753298861665e-05, + "loss": 0.7738, + "step": 2248 + }, + { + "epoch": 0.43, + "learning_rate": 1.2836866543064831e-05, + "loss": 0.8746, + "step": 2249 + }, + { + "epoch": 0.43, + "learning_rate": 1.283097871794772e-05, + "loss": 0.3, + "step": 2250 + }, + { + "epoch": 0.43, + "learning_rate": 1.2825089825729666e-05, + "loss": 0.8189, + "step": 2251 + }, + { + "epoch": 0.43, + "learning_rate": 1.2819199868630419e-05, + "loss": 0.8404, + "step": 2252 + }, + { + "epoch": 0.43, + "learning_rate": 1.2813308848870114e-05, + "loss": 0.8278, + "step": 2253 + }, + { + "epoch": 0.43, + "learning_rate": 1.28074167686693e-05, + "loss": 0.8398, + "step": 2254 + }, + { + "epoch": 0.43, + "learning_rate": 1.2801523630248917e-05, + "loss": 0.8647, + "step": 2255 + }, + { + "epoch": 0.43, + "learning_rate": 1.2795629435830313e-05, + "loss": 0.8364, + "step": 2256 + }, + { + "epoch": 0.43, + "learning_rate": 1.2789734187635223e-05, + "loss": 0.8039, + "step": 2257 + }, + { + "epoch": 0.43, + "learning_rate": 1.2783837887885787e-05, + "loss": 0.7997, + "step": 2258 + }, + { + "epoch": 0.43, + "learning_rate": 1.2777940538804545e-05, + "loss": 0.7885, + "step": 2259 + }, + { + "epoch": 0.43, + "learning_rate": 1.2772042142614418e-05, + "loss": 0.8106, + "step": 2260 + }, + { + "epoch": 0.43, + "learning_rate": 1.2766142701538733e-05, + "loss": 0.7874, + "step": 2261 + }, + { + "epoch": 0.43, + "learning_rate": 1.276024221780121e-05, + "loss": 0.8397, + "step": 2262 + }, + { + "epoch": 0.43, + "learning_rate": 1.275434069362596e-05, + "loss": 0.828, + "step": 2263 + }, + { + "epoch": 0.43, + "learning_rate": 1.2748438131237484e-05, + "loss": 0.7265, + "step": 2264 + }, + { + "epoch": 0.43, + "learning_rate": 1.274253453286068e-05, + "loss": 0.7715, + "step": 2265 + }, + { + "epoch": 0.43, + "learning_rate": 1.2736629900720832e-05, + "loss": 0.7836, + "step": 2266 + }, + { + "epoch": 0.43, + "learning_rate": 1.2730724237043614e-05, + "loss": 0.2634, + "step": 2267 + }, + { + "epoch": 0.43, + "learning_rate": 1.272481754405509e-05, + "loss": 0.7817, + "step": 2268 + }, + { + "epoch": 0.43, + "learning_rate": 1.2718909823981714e-05, + "loss": 0.813, + "step": 2269 + }, + { + "epoch": 0.43, + "learning_rate": 1.2713001079050325e-05, + "loss": 0.3149, + "step": 2270 + }, + { + "epoch": 0.43, + "learning_rate": 1.2707091311488143e-05, + "loss": 0.8482, + "step": 2271 + }, + { + "epoch": 0.43, + "learning_rate": 1.2701180523522782e-05, + "loss": 0.7905, + "step": 2272 + }, + { + "epoch": 0.43, + "learning_rate": 1.2695268717382242e-05, + "loss": 0.2608, + "step": 2273 + }, + { + "epoch": 0.43, + "learning_rate": 1.2689355895294895e-05, + "loss": 0.7722, + "step": 2274 + }, + { + "epoch": 0.43, + "learning_rate": 1.2683442059489503e-05, + "loss": 0.7963, + "step": 2275 + }, + { + "epoch": 0.43, + "learning_rate": 1.2677527212195217e-05, + "loss": 0.7784, + "step": 2276 + }, + { + "epoch": 0.43, + "learning_rate": 1.267161135564156e-05, + "loss": 0.8584, + "step": 2277 + }, + { + "epoch": 0.43, + "learning_rate": 1.2665694492058436e-05, + "loss": 0.8263, + "step": 2278 + }, + { + "epoch": 0.43, + "learning_rate": 1.2659776623676127e-05, + "loss": 0.8098, + "step": 2279 + }, + { + "epoch": 0.43, + "learning_rate": 1.2653857752725305e-05, + "loss": 0.8278, + "step": 2280 + }, + { + "epoch": 0.43, + "learning_rate": 1.2647937881437009e-05, + "loss": 0.8199, + "step": 2281 + }, + { + "epoch": 0.43, + "learning_rate": 1.2642017012042655e-05, + "loss": 0.8621, + "step": 2282 + }, + { + "epoch": 0.43, + "learning_rate": 1.263609514677404e-05, + "loss": 0.2962, + "step": 2283 + }, + { + "epoch": 0.43, + "learning_rate": 1.263017228786334e-05, + "loss": 0.8116, + "step": 2284 + }, + { + "epoch": 0.43, + "learning_rate": 1.2624248437543093e-05, + "loss": 0.8008, + "step": 2285 + }, + { + "epoch": 0.43, + "learning_rate": 1.261832359804622e-05, + "loss": 0.7725, + "step": 2286 + }, + { + "epoch": 0.43, + "learning_rate": 1.2612397771606015e-05, + "loss": 0.7122, + "step": 2287 + }, + { + "epoch": 0.43, + "learning_rate": 1.2606470960456143e-05, + "loss": 0.2417, + "step": 2288 + }, + { + "epoch": 0.43, + "learning_rate": 1.2600543166830635e-05, + "loss": 0.8218, + "step": 2289 + }, + { + "epoch": 0.43, + "learning_rate": 1.2594614392963898e-05, + "loss": 0.8488, + "step": 2290 + }, + { + "epoch": 0.43, + "learning_rate": 1.2588684641090708e-05, + "loss": 0.8584, + "step": 2291 + }, + { + "epoch": 0.43, + "learning_rate": 1.2582753913446204e-05, + "loss": 0.8663, + "step": 2292 + }, + { + "epoch": 0.43, + "learning_rate": 1.2576822212265904e-05, + "loss": 0.8451, + "step": 2293 + }, + { + "epoch": 0.43, + "learning_rate": 1.2570889539785683e-05, + "loss": 0.8171, + "step": 2294 + }, + { + "epoch": 0.43, + "learning_rate": 1.2564955898241787e-05, + "loss": 0.7794, + "step": 2295 + }, + { + "epoch": 0.44, + "learning_rate": 1.2559021289870823e-05, + "loss": 0.7891, + "step": 2296 + }, + { + "epoch": 0.44, + "learning_rate": 1.2553085716909768e-05, + "loss": 0.7964, + "step": 2297 + }, + { + "epoch": 0.44, + "learning_rate": 1.2547149181595957e-05, + "loss": 0.8322, + "step": 2298 + }, + { + "epoch": 0.44, + "learning_rate": 1.2541211686167089e-05, + "loss": 0.8123, + "step": 2299 + }, + { + "epoch": 0.44, + "learning_rate": 1.2535273232861234e-05, + "loss": 0.8335, + "step": 2300 + }, + { + "epoch": 0.44, + "learning_rate": 1.2529333823916807e-05, + "loss": 0.8077, + "step": 2301 + }, + { + "epoch": 0.44, + "learning_rate": 1.2523393461572599e-05, + "loss": 0.8026, + "step": 2302 + }, + { + "epoch": 0.44, + "learning_rate": 1.2517452148067748e-05, + "loss": 0.8871, + "step": 2303 + }, + { + "epoch": 0.44, + "learning_rate": 1.2511509885641754e-05, + "loss": 0.8219, + "step": 2304 + }, + { + "epoch": 0.44, + "learning_rate": 1.2505566676534481e-05, + "loss": 0.8341, + "step": 2305 + }, + { + "epoch": 0.44, + "learning_rate": 1.2499622522986146e-05, + "loss": 0.8664, + "step": 2306 + }, + { + "epoch": 0.44, + "learning_rate": 1.2493677427237316e-05, + "loss": 0.8278, + "step": 2307 + }, + { + "epoch": 0.44, + "learning_rate": 1.2487731391528919e-05, + "loss": 0.813, + "step": 2308 + }, + { + "epoch": 0.44, + "learning_rate": 1.248178441810224e-05, + "loss": 0.8017, + "step": 2309 + }, + { + "epoch": 0.44, + "learning_rate": 1.2475836509198912e-05, + "loss": 0.8546, + "step": 2310 + }, + { + "epoch": 0.44, + "learning_rate": 1.246988766706092e-05, + "loss": 0.78, + "step": 2311 + }, + { + "epoch": 0.44, + "learning_rate": 1.246393789393061e-05, + "loss": 0.8041, + "step": 2312 + }, + { + "epoch": 0.44, + "learning_rate": 1.2457987192050663e-05, + "loss": 0.8531, + "step": 2313 + }, + { + "epoch": 0.44, + "learning_rate": 1.2452035563664127e-05, + "loss": 0.82, + "step": 2314 + }, + { + "epoch": 0.44, + "learning_rate": 1.2446083011014389e-05, + "loss": 0.8135, + "step": 2315 + }, + { + "epoch": 0.44, + "learning_rate": 1.244012953634519e-05, + "loss": 0.8266, + "step": 2316 + }, + { + "epoch": 0.44, + "learning_rate": 1.243417514190061e-05, + "loss": 0.8176, + "step": 2317 + }, + { + "epoch": 0.44, + "learning_rate": 1.2428219829925083e-05, + "loss": 0.7928, + "step": 2318 + }, + { + "epoch": 0.44, + "learning_rate": 1.242226360266339e-05, + "loss": 0.7472, + "step": 2319 + }, + { + "epoch": 0.44, + "learning_rate": 1.2416306462360652e-05, + "loss": 0.7709, + "step": 2320 + }, + { + "epoch": 0.44, + "learning_rate": 1.2410348411262336e-05, + "loss": 0.8182, + "step": 2321 + }, + { + "epoch": 0.44, + "learning_rate": 1.2404389451614253e-05, + "loss": 0.7872, + "step": 2322 + }, + { + "epoch": 0.44, + "learning_rate": 1.2398429585662555e-05, + "loss": 0.7303, + "step": 2323 + }, + { + "epoch": 0.44, + "learning_rate": 1.239246881565374e-05, + "loss": 0.2672, + "step": 2324 + }, + { + "epoch": 0.44, + "learning_rate": 1.238650714383464e-05, + "loss": 0.8292, + "step": 2325 + }, + { + "epoch": 0.44, + "learning_rate": 1.2380544572452431e-05, + "loss": 0.8711, + "step": 2326 + }, + { + "epoch": 0.44, + "learning_rate": 1.2374581103754626e-05, + "loss": 0.847, + "step": 2327 + }, + { + "epoch": 0.44, + "learning_rate": 1.2368616739989082e-05, + "loss": 0.7805, + "step": 2328 + }, + { + "epoch": 0.44, + "learning_rate": 1.2362651483403985e-05, + "loss": 0.8357, + "step": 2329 + }, + { + "epoch": 0.44, + "learning_rate": 1.2356685336247864e-05, + "loss": 0.8006, + "step": 2330 + }, + { + "epoch": 0.44, + "learning_rate": 1.2350718300769581e-05, + "loss": 0.8336, + "step": 2331 + }, + { + "epoch": 0.44, + "learning_rate": 1.2344750379218333e-05, + "loss": 0.8264, + "step": 2332 + }, + { + "epoch": 0.44, + "learning_rate": 1.2338781573843649e-05, + "loss": 0.7881, + "step": 2333 + }, + { + "epoch": 0.44, + "learning_rate": 1.2332811886895395e-05, + "loss": 0.7952, + "step": 2334 + }, + { + "epoch": 0.44, + "learning_rate": 1.2326841320623769e-05, + "loss": 0.8601, + "step": 2335 + }, + { + "epoch": 0.44, + "learning_rate": 1.2320869877279297e-05, + "loss": 0.857, + "step": 2336 + }, + { + "epoch": 0.44, + "learning_rate": 1.2314897559112836e-05, + "loss": 0.8186, + "step": 2337 + }, + { + "epoch": 0.44, + "learning_rate": 1.2308924368375581e-05, + "loss": 0.8095, + "step": 2338 + }, + { + "epoch": 0.44, + "learning_rate": 1.2302950307319041e-05, + "loss": 0.8557, + "step": 2339 + }, + { + "epoch": 0.44, + "learning_rate": 1.2296975378195067e-05, + "loss": 0.8822, + "step": 2340 + }, + { + "epoch": 0.44, + "learning_rate": 1.2290999583255829e-05, + "loss": 0.7642, + "step": 2341 + }, + { + "epoch": 0.44, + "learning_rate": 1.2285022924753826e-05, + "loss": 0.8383, + "step": 2342 + }, + { + "epoch": 0.44, + "learning_rate": 1.2279045404941883e-05, + "loss": 0.2759, + "step": 2343 + }, + { + "epoch": 0.44, + "learning_rate": 1.2273067026073149e-05, + "loss": 0.7997, + "step": 2344 + }, + { + "epoch": 0.44, + "learning_rate": 1.2267087790401098e-05, + "loss": 0.7672, + "step": 2345 + }, + { + "epoch": 0.44, + "learning_rate": 1.2261107700179522e-05, + "loss": 0.7766, + "step": 2346 + }, + { + "epoch": 0.44, + "learning_rate": 1.2255126757662543e-05, + "loss": 0.8181, + "step": 2347 + }, + { + "epoch": 0.45, + "learning_rate": 1.2249144965104598e-05, + "loss": 0.7931, + "step": 2348 + }, + { + "epoch": 0.45, + "learning_rate": 1.224316232476045e-05, + "loss": 0.7674, + "step": 2349 + }, + { + "epoch": 0.45, + "learning_rate": 1.2237178838885168e-05, + "loss": 0.7651, + "step": 2350 + }, + { + "epoch": 0.45, + "learning_rate": 1.2231194509734163e-05, + "loss": 0.2616, + "step": 2351 + }, + { + "epoch": 0.45, + "learning_rate": 1.2225209339563144e-05, + "loss": 0.7614, + "step": 2352 + }, + { + "epoch": 0.45, + "learning_rate": 1.2219223330628146e-05, + "loss": 0.7947, + "step": 2353 + }, + { + "epoch": 0.45, + "learning_rate": 1.2213236485185516e-05, + "loss": 0.771, + "step": 2354 + }, + { + "epoch": 0.45, + "learning_rate": 1.2207248805491917e-05, + "loss": 0.7738, + "step": 2355 + }, + { + "epoch": 0.45, + "learning_rate": 1.2201260293804333e-05, + "loss": 0.803, + "step": 2356 + }, + { + "epoch": 0.45, + "learning_rate": 1.2195270952380052e-05, + "loss": 0.8218, + "step": 2357 + }, + { + "epoch": 0.45, + "learning_rate": 1.218928078347668e-05, + "loss": 0.8556, + "step": 2358 + }, + { + "epoch": 0.45, + "learning_rate": 1.2183289789352134e-05, + "loss": 0.775, + "step": 2359 + }, + { + "epoch": 0.45, + "learning_rate": 1.217729797226464e-05, + "loss": 0.7908, + "step": 2360 + }, + { + "epoch": 0.45, + "learning_rate": 1.217130533447274e-05, + "loss": 0.8235, + "step": 2361 + }, + { + "epoch": 0.45, + "learning_rate": 1.216531187823528e-05, + "loss": 0.8634, + "step": 2362 + }, + { + "epoch": 0.45, + "learning_rate": 1.2159317605811415e-05, + "loss": 0.8247, + "step": 2363 + }, + { + "epoch": 0.45, + "learning_rate": 1.215332251946061e-05, + "loss": 0.7951, + "step": 2364 + }, + { + "epoch": 0.45, + "learning_rate": 1.2147326621442636e-05, + "loss": 0.7529, + "step": 2365 + }, + { + "epoch": 0.45, + "learning_rate": 1.2141329914017567e-05, + "loss": 0.8007, + "step": 2366 + }, + { + "epoch": 0.45, + "learning_rate": 1.2135332399445787e-05, + "loss": 0.8026, + "step": 2367 + }, + { + "epoch": 0.45, + "learning_rate": 1.2129334079987981e-05, + "loss": 0.8054, + "step": 2368 + }, + { + "epoch": 0.45, + "learning_rate": 1.212333495790514e-05, + "loss": 0.7738, + "step": 2369 + }, + { + "epoch": 0.45, + "learning_rate": 1.2117335035458548e-05, + "loss": 0.8395, + "step": 2370 + }, + { + "epoch": 0.45, + "learning_rate": 1.2111334314909811e-05, + "loss": 0.8218, + "step": 2371 + }, + { + "epoch": 0.45, + "learning_rate": 1.2105332798520812e-05, + "loss": 0.8123, + "step": 2372 + }, + { + "epoch": 0.45, + "learning_rate": 1.2099330488553754e-05, + "loss": 0.7869, + "step": 2373 + }, + { + "epoch": 0.45, + "learning_rate": 1.2093327387271123e-05, + "loss": 0.8483, + "step": 2374 + }, + { + "epoch": 0.45, + "learning_rate": 1.2087323496935714e-05, + "loss": 0.8026, + "step": 2375 + }, + { + "epoch": 0.45, + "learning_rate": 1.208131881981062e-05, + "loss": 0.8431, + "step": 2376 + }, + { + "epoch": 0.45, + "learning_rate": 1.207531335815922e-05, + "loss": 0.8109, + "step": 2377 + }, + { + "epoch": 0.45, + "learning_rate": 1.2069307114245197e-05, + "loss": 0.2843, + "step": 2378 + }, + { + "epoch": 0.45, + "learning_rate": 1.2063300090332528e-05, + "loss": 0.8059, + "step": 2379 + }, + { + "epoch": 0.45, + "learning_rate": 1.2057292288685485e-05, + "loss": 0.8865, + "step": 2380 + }, + { + "epoch": 0.45, + "learning_rate": 1.2051283711568631e-05, + "loss": 0.7983, + "step": 2381 + }, + { + "epoch": 0.45, + "learning_rate": 1.2045274361246816e-05, + "loss": 0.8348, + "step": 2382 + }, + { + "epoch": 0.45, + "learning_rate": 1.2039264239985196e-05, + "loss": 0.7262, + "step": 2383 + }, + { + "epoch": 0.45, + "learning_rate": 1.2033253350049203e-05, + "loss": 0.2793, + "step": 2384 + }, + { + "epoch": 0.45, + "learning_rate": 1.2027241693704567e-05, + "loss": 0.2906, + "step": 2385 + }, + { + "epoch": 0.45, + "learning_rate": 1.2021229273217303e-05, + "loss": 0.7645, + "step": 2386 + }, + { + "epoch": 0.45, + "learning_rate": 1.2015216090853716e-05, + "loss": 0.7932, + "step": 2387 + }, + { + "epoch": 0.45, + "learning_rate": 1.2009202148880397e-05, + "loss": 0.7595, + "step": 2388 + }, + { + "epoch": 0.45, + "learning_rate": 1.2003187449564224e-05, + "loss": 0.8364, + "step": 2389 + }, + { + "epoch": 0.45, + "learning_rate": 1.199717199517236e-05, + "loss": 0.8523, + "step": 2390 + }, + { + "epoch": 0.45, + "learning_rate": 1.1991155787972254e-05, + "loss": 0.8211, + "step": 2391 + }, + { + "epoch": 0.45, + "learning_rate": 1.1985138830231638e-05, + "loss": 0.7886, + "step": 2392 + }, + { + "epoch": 0.45, + "learning_rate": 1.1979121124218525e-05, + "loss": 0.8651, + "step": 2393 + }, + { + "epoch": 0.45, + "learning_rate": 1.1973102672201216e-05, + "loss": 0.7731, + "step": 2394 + }, + { + "epoch": 0.45, + "learning_rate": 1.1967083476448282e-05, + "loss": 0.8269, + "step": 2395 + }, + { + "epoch": 0.45, + "learning_rate": 1.1961063539228586e-05, + "loss": 0.8289, + "step": 2396 + }, + { + "epoch": 0.45, + "learning_rate": 1.1955042862811265e-05, + "loss": 0.813, + "step": 2397 + }, + { + "epoch": 0.45, + "learning_rate": 1.1949021449465736e-05, + "loss": 0.786, + "step": 2398 + }, + { + "epoch": 0.45, + "learning_rate": 1.1942999301461694e-05, + "loss": 0.8412, + "step": 2399 + }, + { + "epoch": 0.45, + "learning_rate": 1.1936976421069104e-05, + "loss": 0.8889, + "step": 2400 + }, + { + "epoch": 0.46, + "learning_rate": 1.1930952810558221e-05, + "loss": 0.8088, + "step": 2401 + }, + { + "epoch": 0.46, + "learning_rate": 1.1924928472199562e-05, + "loss": 0.2555, + "step": 2402 + }, + { + "epoch": 0.46, + "learning_rate": 1.1918903408263924e-05, + "loss": 0.8175, + "step": 2403 + }, + { + "epoch": 0.46, + "learning_rate": 1.1912877621022383e-05, + "loss": 0.2845, + "step": 2404 + }, + { + "epoch": 0.46, + "learning_rate": 1.1906851112746274e-05, + "loss": 0.8079, + "step": 2405 + }, + { + "epoch": 0.46, + "learning_rate": 1.1900823885707216e-05, + "loss": 0.7728, + "step": 2406 + }, + { + "epoch": 0.46, + "learning_rate": 1.1894795942177092e-05, + "loss": 0.8381, + "step": 2407 + }, + { + "epoch": 0.46, + "learning_rate": 1.1888767284428062e-05, + "loss": 0.8583, + "step": 2408 + }, + { + "epoch": 0.46, + "learning_rate": 1.1882737914732547e-05, + "loss": 0.8443, + "step": 2409 + }, + { + "epoch": 0.46, + "learning_rate": 1.187670783536324e-05, + "loss": 0.8117, + "step": 2410 + }, + { + "epoch": 0.46, + "learning_rate": 1.1870677048593105e-05, + "loss": 0.7584, + "step": 2411 + }, + { + "epoch": 0.46, + "learning_rate": 1.1864645556695367e-05, + "loss": 0.7564, + "step": 2412 + }, + { + "epoch": 0.46, + "learning_rate": 1.1858613361943518e-05, + "loss": 0.8503, + "step": 2413 + }, + { + "epoch": 0.46, + "learning_rate": 1.185258046661132e-05, + "loss": 0.8066, + "step": 2414 + }, + { + "epoch": 0.46, + "learning_rate": 1.1846546872972793e-05, + "loss": 0.2416, + "step": 2415 + }, + { + "epoch": 0.46, + "learning_rate": 1.184051258330222e-05, + "loss": 0.7674, + "step": 2416 + }, + { + "epoch": 0.46, + "learning_rate": 1.1834477599874153e-05, + "loss": 0.8048, + "step": 2417 + }, + { + "epoch": 0.46, + "learning_rate": 1.1828441924963396e-05, + "loss": 0.7346, + "step": 2418 + }, + { + "epoch": 0.46, + "learning_rate": 1.1822405560845025e-05, + "loss": 0.2804, + "step": 2419 + }, + { + "epoch": 0.46, + "learning_rate": 1.1816368509794365e-05, + "loss": 0.8093, + "step": 2420 + }, + { + "epoch": 0.46, + "learning_rate": 1.1810330774087007e-05, + "loss": 0.7824, + "step": 2421 + }, + { + "epoch": 0.46, + "learning_rate": 1.1804292355998796e-05, + "loss": 0.8165, + "step": 2422 + }, + { + "epoch": 0.46, + "learning_rate": 1.1798253257805838e-05, + "loss": 0.273, + "step": 2423 + }, + { + "epoch": 0.46, + "learning_rate": 1.1792213481784491e-05, + "loss": 0.7882, + "step": 2424 + }, + { + "epoch": 0.46, + "learning_rate": 1.1786173030211372e-05, + "loss": 0.8444, + "step": 2425 + }, + { + "epoch": 0.46, + "learning_rate": 1.178013190536335e-05, + "loss": 0.869, + "step": 2426 + }, + { + "epoch": 0.46, + "learning_rate": 1.177409010951755e-05, + "loss": 0.7918, + "step": 2427 + }, + { + "epoch": 0.46, + "learning_rate": 1.1768047644951349e-05, + "loss": 0.8122, + "step": 2428 + }, + { + "epoch": 0.46, + "learning_rate": 1.1762004513942375e-05, + "loss": 0.8224, + "step": 2429 + }, + { + "epoch": 0.46, + "learning_rate": 1.1755960718768513e-05, + "loss": 0.801, + "step": 2430 + }, + { + "epoch": 0.46, + "learning_rate": 1.1749916261707888e-05, + "loss": 0.7551, + "step": 2431 + }, + { + "epoch": 0.46, + "learning_rate": 1.1743871145038883e-05, + "loss": 0.8052, + "step": 2432 + }, + { + "epoch": 0.46, + "learning_rate": 1.1737825371040128e-05, + "loss": 0.8326, + "step": 2433 + }, + { + "epoch": 0.46, + "learning_rate": 1.1731778941990497e-05, + "loss": 0.8079, + "step": 2434 + }, + { + "epoch": 0.46, + "learning_rate": 1.1725731860169114e-05, + "loss": 0.8234, + "step": 2435 + }, + { + "epoch": 0.46, + "learning_rate": 1.1719684127855348e-05, + "loss": 0.7919, + "step": 2436 + }, + { + "epoch": 0.46, + "learning_rate": 1.1713635747328817e-05, + "loss": 0.8253, + "step": 2437 + }, + { + "epoch": 0.46, + "learning_rate": 1.1707586720869375e-05, + "loss": 0.7765, + "step": 2438 + }, + { + "epoch": 0.46, + "learning_rate": 1.170153705075713e-05, + "loss": 0.8321, + "step": 2439 + }, + { + "epoch": 0.46, + "learning_rate": 1.1695486739272423e-05, + "loss": 0.8088, + "step": 2440 + }, + { + "epoch": 0.46, + "learning_rate": 1.1689435788695844e-05, + "loss": 0.2792, + "step": 2441 + }, + { + "epoch": 0.46, + "learning_rate": 1.1683384201308219e-05, + "loss": 0.9048, + "step": 2442 + }, + { + "epoch": 0.46, + "learning_rate": 1.1677331979390616e-05, + "loss": 0.7818, + "step": 2443 + }, + { + "epoch": 0.46, + "learning_rate": 1.1671279125224342e-05, + "loss": 0.811, + "step": 2444 + }, + { + "epoch": 0.46, + "learning_rate": 1.1665225641090943e-05, + "loss": 0.7823, + "step": 2445 + }, + { + "epoch": 0.46, + "learning_rate": 1.1659171529272202e-05, + "loss": 0.7575, + "step": 2446 + }, + { + "epoch": 0.46, + "learning_rate": 1.1653116792050137e-05, + "loss": 0.8129, + "step": 2447 + }, + { + "epoch": 0.46, + "learning_rate": 1.1647061431707e-05, + "loss": 0.7755, + "step": 2448 + }, + { + "epoch": 0.46, + "learning_rate": 1.164100545052529e-05, + "loss": 0.7525, + "step": 2449 + }, + { + "epoch": 0.46, + "learning_rate": 1.1634948850787726e-05, + "loss": 0.8281, + "step": 2450 + }, + { + "epoch": 0.46, + "learning_rate": 1.1628891634777263e-05, + "loss": 0.8614, + "step": 2451 + }, + { + "epoch": 0.46, + "learning_rate": 1.1622833804777094e-05, + "loss": 0.7478, + "step": 2452 + }, + { + "epoch": 0.46, + "learning_rate": 1.161677536307064e-05, + "loss": 0.7975, + "step": 2453 + }, + { + "epoch": 0.47, + "learning_rate": 1.1610716311941551e-05, + "loss": 0.8176, + "step": 2454 + }, + { + "epoch": 0.47, + "learning_rate": 1.1604656653673707e-05, + "loss": 0.7749, + "step": 2455 + }, + { + "epoch": 0.47, + "learning_rate": 1.159859639055122e-05, + "loss": 0.7813, + "step": 2456 + }, + { + "epoch": 0.47, + "learning_rate": 1.1592535524858428e-05, + "loss": 0.8352, + "step": 2457 + }, + { + "epoch": 0.47, + "learning_rate": 1.1586474058879895e-05, + "loss": 0.7666, + "step": 2458 + }, + { + "epoch": 0.47, + "learning_rate": 1.1580411994900416e-05, + "loss": 0.8219, + "step": 2459 + }, + { + "epoch": 0.47, + "learning_rate": 1.1574349335205003e-05, + "loss": 0.8183, + "step": 2460 + }, + { + "epoch": 0.47, + "learning_rate": 1.15682860820789e-05, + "loss": 0.7834, + "step": 2461 + }, + { + "epoch": 0.47, + "learning_rate": 1.156222223780757e-05, + "loss": 0.8385, + "step": 2462 + }, + { + "epoch": 0.47, + "learning_rate": 1.1556157804676706e-05, + "loss": 0.2635, + "step": 2463 + }, + { + "epoch": 0.47, + "learning_rate": 1.1550092784972212e-05, + "loss": 0.838, + "step": 2464 + }, + { + "epoch": 0.47, + "learning_rate": 1.1544027180980221e-05, + "loss": 0.8459, + "step": 2465 + }, + { + "epoch": 0.47, + "learning_rate": 1.1537960994987085e-05, + "loss": 0.8189, + "step": 2466 + }, + { + "epoch": 0.47, + "learning_rate": 1.1531894229279377e-05, + "loss": 0.7999, + "step": 2467 + }, + { + "epoch": 0.47, + "learning_rate": 1.1525826886143882e-05, + "loss": 0.8223, + "step": 2468 + }, + { + "epoch": 0.47, + "learning_rate": 1.1519758967867608e-05, + "loss": 0.7863, + "step": 2469 + }, + { + "epoch": 0.47, + "learning_rate": 1.151369047673778e-05, + "loss": 0.7356, + "step": 2470 + }, + { + "epoch": 0.47, + "learning_rate": 1.1507621415041837e-05, + "loss": 0.7535, + "step": 2471 + }, + { + "epoch": 0.47, + "learning_rate": 1.1501551785067431e-05, + "loss": 0.8423, + "step": 2472 + }, + { + "epoch": 0.47, + "learning_rate": 1.1495481589102435e-05, + "loss": 0.7457, + "step": 2473 + }, + { + "epoch": 0.47, + "learning_rate": 1.1489410829434928e-05, + "loss": 0.7747, + "step": 2474 + }, + { + "epoch": 0.47, + "learning_rate": 1.1483339508353206e-05, + "loss": 0.8114, + "step": 2475 + }, + { + "epoch": 0.47, + "learning_rate": 1.1477267628145777e-05, + "loss": 0.2922, + "step": 2476 + }, + { + "epoch": 0.47, + "learning_rate": 1.1471195191101352e-05, + "loss": 0.8603, + "step": 2477 + }, + { + "epoch": 0.47, + "learning_rate": 1.1465122199508865e-05, + "loss": 0.8091, + "step": 2478 + }, + { + "epoch": 0.47, + "learning_rate": 1.1459048655657445e-05, + "loss": 0.7531, + "step": 2479 + }, + { + "epoch": 0.47, + "learning_rate": 1.1452974561836443e-05, + "loss": 0.2507, + "step": 2480 + }, + { + "epoch": 0.47, + "learning_rate": 1.1446899920335407e-05, + "loss": 0.839, + "step": 2481 + }, + { + "epoch": 0.47, + "learning_rate": 1.1440824733444094e-05, + "loss": 0.8114, + "step": 2482 + }, + { + "epoch": 0.47, + "learning_rate": 1.1434749003452467e-05, + "loss": 0.8091, + "step": 2483 + }, + { + "epoch": 0.47, + "learning_rate": 1.1428672732650697e-05, + "loss": 0.8422, + "step": 2484 + }, + { + "epoch": 0.47, + "learning_rate": 1.1422595923329155e-05, + "loss": 0.8314, + "step": 2485 + }, + { + "epoch": 0.47, + "learning_rate": 1.1416518577778411e-05, + "loss": 0.7998, + "step": 2486 + }, + { + "epoch": 0.47, + "learning_rate": 1.1410440698289248e-05, + "loss": 0.8067, + "step": 2487 + }, + { + "epoch": 0.47, + "learning_rate": 1.1404362287152646e-05, + "loss": 0.8197, + "step": 2488 + }, + { + "epoch": 0.47, + "learning_rate": 1.1398283346659778e-05, + "loss": 0.8102, + "step": 2489 + }, + { + "epoch": 0.47, + "learning_rate": 1.1392203879102027e-05, + "loss": 0.795, + "step": 2490 + }, + { + "epoch": 0.47, + "learning_rate": 1.1386123886770965e-05, + "loss": 0.8095, + "step": 2491 + }, + { + "epoch": 0.47, + "learning_rate": 1.1380043371958371e-05, + "loss": 0.7355, + "step": 2492 + }, + { + "epoch": 0.47, + "learning_rate": 1.1373962336956214e-05, + "loss": 0.8351, + "step": 2493 + }, + { + "epoch": 0.47, + "learning_rate": 1.1367880784056661e-05, + "loss": 0.9094, + "step": 2494 + }, + { + "epoch": 0.47, + "learning_rate": 1.136179871555208e-05, + "loss": 0.8294, + "step": 2495 + }, + { + "epoch": 0.47, + "learning_rate": 1.1355716133735024e-05, + "loss": 0.809, + "step": 2496 + }, + { + "epoch": 0.47, + "learning_rate": 1.1349633040898246e-05, + "loss": 0.8322, + "step": 2497 + }, + { + "epoch": 0.47, + "learning_rate": 1.1343549439334688e-05, + "loss": 0.834, + "step": 2498 + }, + { + "epoch": 0.47, + "learning_rate": 1.1337465331337487e-05, + "loss": 0.7883, + "step": 2499 + }, + { + "epoch": 0.47, + "learning_rate": 1.1331380719199965e-05, + "loss": 0.8972, + "step": 2500 + }, + { + "epoch": 0.47, + "learning_rate": 1.1325295605215643e-05, + "loss": 0.738, + "step": 2501 + }, + { + "epoch": 0.47, + "learning_rate": 1.1319209991678224e-05, + "loss": 0.7706, + "step": 2502 + }, + { + "epoch": 0.47, + "learning_rate": 1.13131238808816e-05, + "loss": 0.8218, + "step": 2503 + }, + { + "epoch": 0.47, + "learning_rate": 1.1307037275119854e-05, + "loss": 0.8079, + "step": 2504 + }, + { + "epoch": 0.47, + "learning_rate": 1.1300950176687255e-05, + "loss": 0.7974, + "step": 2505 + }, + { + "epoch": 0.47, + "learning_rate": 1.1294862587878252e-05, + "loss": 0.8445, + "step": 2506 + }, + { + "epoch": 0.48, + "learning_rate": 1.1288774510987487e-05, + "loss": 0.8027, + "step": 2507 + }, + { + "epoch": 0.48, + "learning_rate": 1.1282685948309782e-05, + "loss": 0.8261, + "step": 2508 + }, + { + "epoch": 0.48, + "learning_rate": 1.1276596902140142e-05, + "loss": 0.8075, + "step": 2509 + }, + { + "epoch": 0.48, + "learning_rate": 1.1270507374773752e-05, + "loss": 0.8293, + "step": 2510 + }, + { + "epoch": 0.48, + "learning_rate": 1.1264417368505981e-05, + "loss": 0.8075, + "step": 2511 + }, + { + "epoch": 0.48, + "learning_rate": 1.1258326885632382e-05, + "loss": 0.8, + "step": 2512 + }, + { + "epoch": 0.48, + "learning_rate": 1.1252235928448683e-05, + "loss": 0.8298, + "step": 2513 + }, + { + "epoch": 0.48, + "learning_rate": 1.1246144499250788e-05, + "loss": 0.7803, + "step": 2514 + }, + { + "epoch": 0.48, + "learning_rate": 1.1240052600334787e-05, + "loss": 0.8015, + "step": 2515 + }, + { + "epoch": 0.48, + "learning_rate": 1.1233960233996941e-05, + "loss": 0.8293, + "step": 2516 + }, + { + "epoch": 0.48, + "learning_rate": 1.1227867402533688e-05, + "loss": 0.8051, + "step": 2517 + }, + { + "epoch": 0.48, + "learning_rate": 1.1221774108241646e-05, + "loss": 0.8373, + "step": 2518 + }, + { + "epoch": 0.48, + "learning_rate": 1.1215680353417598e-05, + "loss": 0.7805, + "step": 2519 + }, + { + "epoch": 0.48, + "learning_rate": 1.1209586140358511e-05, + "loss": 0.8174, + "step": 2520 + }, + { + "epoch": 0.48, + "learning_rate": 1.1203491471361518e-05, + "loss": 0.7463, + "step": 2521 + }, + { + "epoch": 0.48, + "learning_rate": 1.1197396348723923e-05, + "loss": 0.8238, + "step": 2522 + }, + { + "epoch": 0.48, + "learning_rate": 1.1191300774743212e-05, + "loss": 0.8413, + "step": 2523 + }, + { + "epoch": 0.48, + "learning_rate": 1.118520475171703e-05, + "loss": 0.8066, + "step": 2524 + }, + { + "epoch": 0.48, + "learning_rate": 1.117910828194319e-05, + "loss": 0.791, + "step": 2525 + }, + { + "epoch": 0.48, + "learning_rate": 1.1173011367719677e-05, + "loss": 0.7485, + "step": 2526 + }, + { + "epoch": 0.48, + "learning_rate": 1.1166914011344652e-05, + "loss": 0.7832, + "step": 2527 + }, + { + "epoch": 0.48, + "learning_rate": 1.116081621511643e-05, + "loss": 0.8325, + "step": 2528 + }, + { + "epoch": 0.48, + "learning_rate": 1.1154717981333499e-05, + "loss": 0.8038, + "step": 2529 + }, + { + "epoch": 0.48, + "learning_rate": 1.1148619312294504e-05, + "loss": 0.7747, + "step": 2530 + }, + { + "epoch": 0.48, + "learning_rate": 1.1142520210298267e-05, + "loss": 0.7732, + "step": 2531 + }, + { + "epoch": 0.48, + "learning_rate": 1.1136420677643763e-05, + "loss": 0.7441, + "step": 2532 + }, + { + "epoch": 0.48, + "learning_rate": 1.113032071663013e-05, + "loss": 0.7685, + "step": 2533 + }, + { + "epoch": 0.48, + "learning_rate": 1.112422032955667e-05, + "loss": 0.8814, + "step": 2534 + }, + { + "epoch": 0.48, + "learning_rate": 1.1118119518722848e-05, + "loss": 0.2792, + "step": 2535 + }, + { + "epoch": 0.48, + "learning_rate": 1.1112018286428286e-05, + "loss": 0.8164, + "step": 2536 + }, + { + "epoch": 0.48, + "learning_rate": 1.1105916634972762e-05, + "loss": 0.8005, + "step": 2537 + }, + { + "epoch": 0.48, + "learning_rate": 1.1099814566656217e-05, + "loss": 0.785, + "step": 2538 + }, + { + "epoch": 0.48, + "learning_rate": 1.1093712083778748e-05, + "loss": 0.8058, + "step": 2539 + }, + { + "epoch": 0.48, + "learning_rate": 1.1087609188640602e-05, + "loss": 0.7984, + "step": 2540 + }, + { + "epoch": 0.48, + "learning_rate": 1.1081505883542193e-05, + "loss": 0.8016, + "step": 2541 + }, + { + "epoch": 0.48, + "learning_rate": 1.107540217078408e-05, + "loss": 0.7276, + "step": 2542 + }, + { + "epoch": 0.48, + "learning_rate": 1.1069298052666978e-05, + "loss": 0.8235, + "step": 2543 + }, + { + "epoch": 0.48, + "learning_rate": 1.1063193531491758e-05, + "loss": 0.8302, + "step": 2544 + }, + { + "epoch": 0.48, + "learning_rate": 1.1057088609559439e-05, + "loss": 0.8182, + "step": 2545 + }, + { + "epoch": 0.48, + "learning_rate": 1.1050983289171195e-05, + "loss": 0.8219, + "step": 2546 + }, + { + "epoch": 0.48, + "learning_rate": 1.1044877572628345e-05, + "loss": 0.8454, + "step": 2547 + }, + { + "epoch": 0.48, + "learning_rate": 1.1038771462232363e-05, + "loss": 0.7722, + "step": 2548 + }, + { + "epoch": 0.48, + "learning_rate": 1.1032664960284864e-05, + "loss": 0.7814, + "step": 2549 + }, + { + "epoch": 0.48, + "learning_rate": 1.1026558069087618e-05, + "loss": 0.8395, + "step": 2550 + }, + { + "epoch": 0.48, + "learning_rate": 1.1020450790942541e-05, + "loss": 0.7346, + "step": 2551 + }, + { + "epoch": 0.48, + "learning_rate": 1.101434312815169e-05, + "loss": 0.7992, + "step": 2552 + }, + { + "epoch": 0.48, + "learning_rate": 1.1008235083017272e-05, + "loss": 0.8318, + "step": 2553 + }, + { + "epoch": 0.48, + "learning_rate": 1.1002126657841634e-05, + "loss": 0.8497, + "step": 2554 + }, + { + "epoch": 0.48, + "learning_rate": 1.0996017854927268e-05, + "loss": 0.8176, + "step": 2555 + }, + { + "epoch": 0.48, + "learning_rate": 1.0989908676576808e-05, + "loss": 0.8376, + "step": 2556 + }, + { + "epoch": 0.48, + "learning_rate": 1.0983799125093033e-05, + "loss": 0.827, + "step": 2557 + }, + { + "epoch": 0.48, + "learning_rate": 1.0977689202778856e-05, + "loss": 0.8007, + "step": 2558 + }, + { + "epoch": 0.49, + "learning_rate": 1.0971578911937338e-05, + "loss": 0.7868, + "step": 2559 + }, + { + "epoch": 0.49, + "learning_rate": 1.096546825487167e-05, + "loss": 0.7801, + "step": 2560 + }, + { + "epoch": 0.49, + "learning_rate": 1.0959357233885187e-05, + "loss": 0.8097, + "step": 2561 + }, + { + "epoch": 0.49, + "learning_rate": 1.095324585128136e-05, + "loss": 0.8247, + "step": 2562 + }, + { + "epoch": 0.49, + "learning_rate": 1.0947134109363795e-05, + "loss": 0.833, + "step": 2563 + }, + { + "epoch": 0.49, + "learning_rate": 1.0941022010436233e-05, + "loss": 0.8415, + "step": 2564 + }, + { + "epoch": 0.49, + "learning_rate": 1.0934909556802555e-05, + "loss": 0.7855, + "step": 2565 + }, + { + "epoch": 0.49, + "learning_rate": 1.092879675076677e-05, + "loss": 0.2767, + "step": 2566 + }, + { + "epoch": 0.49, + "learning_rate": 1.092268359463302e-05, + "loss": 0.7662, + "step": 2567 + }, + { + "epoch": 0.49, + "learning_rate": 1.0916570090705583e-05, + "loss": 0.7863, + "step": 2568 + }, + { + "epoch": 0.49, + "learning_rate": 1.0910456241288863e-05, + "loss": 0.2453, + "step": 2569 + }, + { + "epoch": 0.49, + "learning_rate": 1.0904342048687398e-05, + "loss": 0.8368, + "step": 2570 + }, + { + "epoch": 0.49, + "learning_rate": 1.0898227515205853e-05, + "loss": 0.8195, + "step": 2571 + }, + { + "epoch": 0.49, + "learning_rate": 1.0892112643149022e-05, + "loss": 0.8055, + "step": 2572 + }, + { + "epoch": 0.49, + "learning_rate": 1.0885997434821831e-05, + "loss": 0.8188, + "step": 2573 + }, + { + "epoch": 0.49, + "learning_rate": 1.0879881892529325e-05, + "loss": 0.8033, + "step": 2574 + }, + { + "epoch": 0.49, + "learning_rate": 1.087376601857668e-05, + "loss": 0.799, + "step": 2575 + }, + { + "epoch": 0.49, + "learning_rate": 1.0867649815269196e-05, + "loss": 0.7779, + "step": 2576 + }, + { + "epoch": 0.49, + "learning_rate": 1.0861533284912298e-05, + "loss": 0.8377, + "step": 2577 + }, + { + "epoch": 0.49, + "learning_rate": 1.0855416429811531e-05, + "loss": 0.8418, + "step": 2578 + }, + { + "epoch": 0.49, + "learning_rate": 1.0849299252272567e-05, + "loss": 0.8022, + "step": 2579 + }, + { + "epoch": 0.49, + "learning_rate": 1.0843181754601198e-05, + "loss": 0.7588, + "step": 2580 + }, + { + "epoch": 0.49, + "learning_rate": 1.0837063939103332e-05, + "loss": 0.8357, + "step": 2581 + }, + { + "epoch": 0.49, + "learning_rate": 1.0830945808085003e-05, + "loss": 0.8685, + "step": 2582 + }, + { + "epoch": 0.49, + "learning_rate": 1.0824827363852365e-05, + "loss": 0.7917, + "step": 2583 + }, + { + "epoch": 0.49, + "learning_rate": 1.0818708608711676e-05, + "loss": 0.853, + "step": 2584 + }, + { + "epoch": 0.49, + "learning_rate": 1.0812589544969333e-05, + "loss": 0.2602, + "step": 2585 + }, + { + "epoch": 0.49, + "learning_rate": 1.0806470174931836e-05, + "loss": 0.805, + "step": 2586 + }, + { + "epoch": 0.49, + "learning_rate": 1.08003505009058e-05, + "loss": 0.8212, + "step": 2587 + }, + { + "epoch": 0.49, + "learning_rate": 1.0794230525197959e-05, + "loss": 0.7273, + "step": 2588 + }, + { + "epoch": 0.49, + "learning_rate": 1.0788110250115158e-05, + "loss": 0.7971, + "step": 2589 + }, + { + "epoch": 0.49, + "learning_rate": 1.0781989677964355e-05, + "loss": 0.7962, + "step": 2590 + }, + { + "epoch": 0.49, + "learning_rate": 1.0775868811052623e-05, + "loss": 0.7799, + "step": 2591 + }, + { + "epoch": 0.49, + "learning_rate": 1.0769747651687144e-05, + "loss": 0.7653, + "step": 2592 + }, + { + "epoch": 0.49, + "learning_rate": 1.0763626202175205e-05, + "loss": 0.8327, + "step": 2593 + }, + { + "epoch": 0.49, + "learning_rate": 1.0757504464824213e-05, + "loss": 0.7835, + "step": 2594 + }, + { + "epoch": 0.49, + "learning_rate": 1.0751382441941677e-05, + "loss": 0.8054, + "step": 2595 + }, + { + "epoch": 0.49, + "learning_rate": 1.0745260135835215e-05, + "loss": 0.7771, + "step": 2596 + }, + { + "epoch": 0.49, + "learning_rate": 1.0739137548812548e-05, + "loss": 0.794, + "step": 2597 + }, + { + "epoch": 0.49, + "learning_rate": 1.073301468318151e-05, + "loss": 0.7672, + "step": 2598 + }, + { + "epoch": 0.49, + "learning_rate": 1.0726891541250036e-05, + "loss": 0.7704, + "step": 2599 + }, + { + "epoch": 0.49, + "learning_rate": 1.0720768125326161e-05, + "loss": 0.7681, + "step": 2600 + }, + { + "epoch": 0.49, + "learning_rate": 1.071464443771803e-05, + "loss": 0.8573, + "step": 2601 + }, + { + "epoch": 0.49, + "learning_rate": 1.0708520480733895e-05, + "loss": 0.7901, + "step": 2602 + }, + { + "epoch": 0.49, + "learning_rate": 1.0702396256682093e-05, + "loss": 0.8866, + "step": 2603 + }, + { + "epoch": 0.49, + "learning_rate": 1.0696271767871076e-05, + "loss": 0.8077, + "step": 2604 + }, + { + "epoch": 0.49, + "learning_rate": 1.0690147016609391e-05, + "loss": 0.8046, + "step": 2605 + }, + { + "epoch": 0.49, + "learning_rate": 1.0684022005205681e-05, + "loss": 0.8461, + "step": 2606 + }, + { + "epoch": 0.49, + "learning_rate": 1.0677896735968694e-05, + "loss": 0.783, + "step": 2607 + }, + { + "epoch": 0.49, + "learning_rate": 1.067177121120727e-05, + "loss": 0.794, + "step": 2608 + }, + { + "epoch": 0.49, + "learning_rate": 1.0665645433230345e-05, + "loss": 0.7701, + "step": 2609 + }, + { + "epoch": 0.49, + "learning_rate": 1.0659519404346955e-05, + "loss": 0.8267, + "step": 2610 + }, + { + "epoch": 0.49, + "learning_rate": 1.0653393126866224e-05, + "loss": 0.7707, + "step": 2611 + }, + { + "epoch": 0.5, + "learning_rate": 1.0647266603097376e-05, + "loss": 0.7842, + "step": 2612 + }, + { + "epoch": 0.5, + "learning_rate": 1.0641139835349726e-05, + "loss": 0.7678, + "step": 2613 + }, + { + "epoch": 0.5, + "learning_rate": 1.0635012825932677e-05, + "loss": 0.7501, + "step": 2614 + }, + { + "epoch": 0.5, + "learning_rate": 1.062888557715573e-05, + "loss": 0.8459, + "step": 2615 + }, + { + "epoch": 0.5, + "learning_rate": 1.0622758091328469e-05, + "loss": 0.8346, + "step": 2616 + }, + { + "epoch": 0.5, + "learning_rate": 1.0616630370760573e-05, + "loss": 0.8578, + "step": 2617 + }, + { + "epoch": 0.5, + "learning_rate": 1.0610502417761803e-05, + "loss": 0.8625, + "step": 2618 + }, + { + "epoch": 0.5, + "learning_rate": 1.0604374234642018e-05, + "loss": 0.8195, + "step": 2619 + }, + { + "epoch": 0.5, + "learning_rate": 1.0598245823711155e-05, + "loss": 0.8308, + "step": 2620 + }, + { + "epoch": 0.5, + "learning_rate": 1.0592117187279237e-05, + "loss": 0.7846, + "step": 2621 + }, + { + "epoch": 0.5, + "learning_rate": 1.0585988327656378e-05, + "loss": 0.7959, + "step": 2622 + }, + { + "epoch": 0.5, + "learning_rate": 1.0579859247152774e-05, + "loss": 0.8001, + "step": 2623 + }, + { + "epoch": 0.5, + "learning_rate": 1.0573729948078699e-05, + "loss": 0.8343, + "step": 2624 + }, + { + "epoch": 0.5, + "learning_rate": 1.0567600432744513e-05, + "loss": 0.8317, + "step": 2625 + }, + { + "epoch": 0.5, + "learning_rate": 1.0561470703460663e-05, + "loss": 0.26, + "step": 2626 + }, + { + "epoch": 0.5, + "learning_rate": 1.0555340762537667e-05, + "loss": 0.7814, + "step": 2627 + }, + { + "epoch": 0.5, + "learning_rate": 1.0549210612286127e-05, + "loss": 0.796, + "step": 2628 + }, + { + "epoch": 0.5, + "learning_rate": 1.0543080255016727e-05, + "loss": 0.8438, + "step": 2629 + }, + { + "epoch": 0.5, + "learning_rate": 1.0536949693040224e-05, + "loss": 0.7717, + "step": 2630 + }, + { + "epoch": 0.5, + "learning_rate": 1.0530818928667455e-05, + "loss": 0.7937, + "step": 2631 + }, + { + "epoch": 0.5, + "learning_rate": 1.052468796420933e-05, + "loss": 0.8077, + "step": 2632 + }, + { + "epoch": 0.5, + "learning_rate": 1.0518556801976843e-05, + "loss": 0.2785, + "step": 2633 + }, + { + "epoch": 0.5, + "learning_rate": 1.0512425444281049e-05, + "loss": 0.8208, + "step": 2634 + }, + { + "epoch": 0.5, + "learning_rate": 1.050629389343309e-05, + "loss": 0.784, + "step": 2635 + }, + { + "epoch": 0.5, + "learning_rate": 1.0500162151744175e-05, + "loss": 0.8123, + "step": 2636 + }, + { + "epoch": 0.5, + "learning_rate": 1.0494030221525582e-05, + "loss": 0.8033, + "step": 2637 + }, + { + "epoch": 0.5, + "learning_rate": 1.0487898105088665e-05, + "loss": 0.7756, + "step": 2638 + }, + { + "epoch": 0.5, + "learning_rate": 1.0481765804744847e-05, + "loss": 0.7694, + "step": 2639 + }, + { + "epoch": 0.5, + "learning_rate": 1.0475633322805619e-05, + "loss": 0.88, + "step": 2640 + }, + { + "epoch": 0.5, + "learning_rate": 1.0469500661582536e-05, + "loss": 0.8751, + "step": 2641 + }, + { + "epoch": 0.5, + "learning_rate": 1.0463367823387232e-05, + "loss": 0.85, + "step": 2642 + }, + { + "epoch": 0.5, + "learning_rate": 1.04572348105314e-05, + "loss": 0.8252, + "step": 2643 + }, + { + "epoch": 0.5, + "learning_rate": 1.0451101625326798e-05, + "loss": 0.824, + "step": 2644 + }, + { + "epoch": 0.5, + "learning_rate": 1.0444968270085254e-05, + "loss": 0.8536, + "step": 2645 + }, + { + "epoch": 0.5, + "learning_rate": 1.0438834747118656e-05, + "loss": 0.8342, + "step": 2646 + }, + { + "epoch": 0.5, + "learning_rate": 1.0432701058738957e-05, + "loss": 0.8107, + "step": 2647 + }, + { + "epoch": 0.5, + "learning_rate": 1.0426567207258168e-05, + "loss": 0.7888, + "step": 2648 + }, + { + "epoch": 0.5, + "learning_rate": 1.0420433194988372e-05, + "loss": 0.8331, + "step": 2649 + }, + { + "epoch": 0.5, + "learning_rate": 1.0414299024241699e-05, + "loss": 0.7801, + "step": 2650 + }, + { + "epoch": 0.5, + "learning_rate": 1.0408164697330348e-05, + "loss": 0.8204, + "step": 2651 + }, + { + "epoch": 0.5, + "learning_rate": 1.0402030216566577e-05, + "loss": 0.8115, + "step": 2652 + }, + { + "epoch": 0.5, + "learning_rate": 1.0395895584262696e-05, + "loss": 0.7814, + "step": 2653 + }, + { + "epoch": 0.5, + "learning_rate": 1.038976080273108e-05, + "loss": 0.8032, + "step": 2654 + }, + { + "epoch": 0.5, + "learning_rate": 1.038362587428415e-05, + "loss": 0.8157, + "step": 2655 + }, + { + "epoch": 0.5, + "learning_rate": 1.0377490801234394e-05, + "loss": 0.804, + "step": 2656 + }, + { + "epoch": 0.5, + "learning_rate": 1.0371355585894342e-05, + "loss": 0.7943, + "step": 2657 + }, + { + "epoch": 0.5, + "learning_rate": 1.0365220230576592e-05, + "loss": 0.8244, + "step": 2658 + }, + { + "epoch": 0.5, + "learning_rate": 1.0359084737593778e-05, + "loss": 0.8298, + "step": 2659 + }, + { + "epoch": 0.5, + "learning_rate": 1.0352949109258605e-05, + "loss": 0.8375, + "step": 2660 + }, + { + "epoch": 0.5, + "learning_rate": 1.0346813347883815e-05, + "loss": 0.8062, + "step": 2661 + }, + { + "epoch": 0.5, + "learning_rate": 1.03406774557822e-05, + "loss": 0.8147, + "step": 2662 + }, + { + "epoch": 0.5, + "learning_rate": 1.0334541435266609e-05, + "loss": 0.8273, + "step": 2663 + }, + { + "epoch": 0.5, + "learning_rate": 1.032840528864994e-05, + "loss": 0.8423, + "step": 2664 + }, + { + "epoch": 0.51, + "learning_rate": 1.0322269018245128e-05, + "loss": 0.7504, + "step": 2665 + }, + { + "epoch": 0.51, + "learning_rate": 1.0316132626365162e-05, + "loss": 0.7883, + "step": 2666 + }, + { + "epoch": 0.51, + "learning_rate": 1.030999611532308e-05, + "loss": 0.7901, + "step": 2667 + }, + { + "epoch": 0.51, + "learning_rate": 1.0303859487431956e-05, + "loss": 0.7734, + "step": 2668 + }, + { + "epoch": 0.51, + "learning_rate": 1.0297722745004915e-05, + "loss": 0.8071, + "step": 2669 + }, + { + "epoch": 0.51, + "learning_rate": 1.0291585890355122e-05, + "loss": 0.7803, + "step": 2670 + }, + { + "epoch": 0.51, + "learning_rate": 1.0285448925795788e-05, + "loss": 0.7834, + "step": 2671 + }, + { + "epoch": 0.51, + "learning_rate": 1.0279311853640157e-05, + "loss": 0.8024, + "step": 2672 + }, + { + "epoch": 0.51, + "learning_rate": 1.0273174676201525e-05, + "loss": 0.8625, + "step": 2673 + }, + { + "epoch": 0.51, + "learning_rate": 1.026703739579322e-05, + "loss": 0.796, + "step": 2674 + }, + { + "epoch": 0.51, + "learning_rate": 1.0260900014728609e-05, + "loss": 0.8297, + "step": 2675 + }, + { + "epoch": 0.51, + "learning_rate": 1.02547625353211e-05, + "loss": 0.7617, + "step": 2676 + }, + { + "epoch": 0.51, + "learning_rate": 1.0248624959884139e-05, + "loss": 0.8542, + "step": 2677 + }, + { + "epoch": 0.51, + "learning_rate": 1.0242487290731203e-05, + "loss": 0.8392, + "step": 2678 + }, + { + "epoch": 0.51, + "learning_rate": 1.0236349530175807e-05, + "loss": 0.7563, + "step": 2679 + }, + { + "epoch": 0.51, + "learning_rate": 1.02302116805315e-05, + "loss": 0.8052, + "step": 2680 + }, + { + "epoch": 0.51, + "learning_rate": 1.0224073744111872e-05, + "loss": 0.8093, + "step": 2681 + }, + { + "epoch": 0.51, + "learning_rate": 1.0217935723230532e-05, + "loss": 0.8592, + "step": 2682 + }, + { + "epoch": 0.51, + "learning_rate": 1.021179762020113e-05, + "loss": 0.7997, + "step": 2683 + }, + { + "epoch": 0.51, + "learning_rate": 1.0205659437337345e-05, + "loss": 0.8277, + "step": 2684 + }, + { + "epoch": 0.51, + "learning_rate": 1.0199521176952884e-05, + "loss": 0.8187, + "step": 2685 + }, + { + "epoch": 0.51, + "learning_rate": 1.019338284136149e-05, + "loss": 0.8354, + "step": 2686 + }, + { + "epoch": 0.51, + "learning_rate": 1.0187244432876926e-05, + "loss": 0.8178, + "step": 2687 + }, + { + "epoch": 0.51, + "learning_rate": 1.0181105953812987e-05, + "loss": 0.8429, + "step": 2688 + }, + { + "epoch": 0.51, + "learning_rate": 1.017496740648349e-05, + "loss": 0.8011, + "step": 2689 + }, + { + "epoch": 0.51, + "learning_rate": 1.0168828793202284e-05, + "loss": 0.7687, + "step": 2690 + }, + { + "epoch": 0.51, + "learning_rate": 1.0162690116283242e-05, + "loss": 0.7931, + "step": 2691 + }, + { + "epoch": 0.51, + "learning_rate": 1.0156551378040259e-05, + "loss": 0.7667, + "step": 2692 + }, + { + "epoch": 0.51, + "learning_rate": 1.015041258078725e-05, + "loss": 0.8017, + "step": 2693 + }, + { + "epoch": 0.51, + "learning_rate": 1.0144273726838159e-05, + "loss": 0.7836, + "step": 2694 + }, + { + "epoch": 0.51, + "learning_rate": 1.0138134818506944e-05, + "loss": 0.7941, + "step": 2695 + }, + { + "epoch": 0.51, + "learning_rate": 1.013199585810759e-05, + "loss": 0.2485, + "step": 2696 + }, + { + "epoch": 0.51, + "learning_rate": 1.01258568479541e-05, + "loss": 0.7761, + "step": 2697 + }, + { + "epoch": 0.51, + "learning_rate": 1.0119717790360492e-05, + "loss": 0.7825, + "step": 2698 + }, + { + "epoch": 0.51, + "learning_rate": 1.0113578687640804e-05, + "loss": 0.8332, + "step": 2699 + }, + { + "epoch": 0.51, + "learning_rate": 1.0107439542109097e-05, + "loss": 0.7871, + "step": 2700 + }, + { + "epoch": 0.51, + "learning_rate": 1.0101300356079438e-05, + "loss": 0.828, + "step": 2701 + }, + { + "epoch": 0.51, + "learning_rate": 1.0095161131865916e-05, + "loss": 0.8011, + "step": 2702 + }, + { + "epoch": 0.51, + "learning_rate": 1.0089021871782632e-05, + "loss": 0.7866, + "step": 2703 + }, + { + "epoch": 0.51, + "learning_rate": 1.00828825781437e-05, + "loss": 0.2442, + "step": 2704 + }, + { + "epoch": 0.51, + "learning_rate": 1.007674325326325e-05, + "loss": 0.8463, + "step": 2705 + }, + { + "epoch": 0.51, + "learning_rate": 1.0070603899455418e-05, + "loss": 0.2389, + "step": 2706 + }, + { + "epoch": 0.51, + "learning_rate": 1.0064464519034358e-05, + "loss": 0.7951, + "step": 2707 + }, + { + "epoch": 0.51, + "learning_rate": 1.0058325114314227e-05, + "loss": 0.8367, + "step": 2708 + }, + { + "epoch": 0.51, + "learning_rate": 1.0052185687609197e-05, + "loss": 0.8517, + "step": 2709 + }, + { + "epoch": 0.51, + "learning_rate": 1.0046046241233443e-05, + "loss": 0.8599, + "step": 2710 + }, + { + "epoch": 0.51, + "learning_rate": 1.0039906777501154e-05, + "loss": 0.7254, + "step": 2711 + }, + { + "epoch": 0.51, + "learning_rate": 1.0033767298726518e-05, + "loss": 0.7848, + "step": 2712 + }, + { + "epoch": 0.51, + "learning_rate": 1.0027627807223735e-05, + "loss": 0.7894, + "step": 2713 + }, + { + "epoch": 0.51, + "learning_rate": 1.0021488305307003e-05, + "loss": 0.7629, + "step": 2714 + }, + { + "epoch": 0.51, + "learning_rate": 1.0015348795290535e-05, + "loss": 0.792, + "step": 2715 + }, + { + "epoch": 0.51, + "learning_rate": 1.0009209279488535e-05, + "loss": 0.7695, + "step": 2716 + }, + { + "epoch": 0.51, + "learning_rate": 1.0003069760215214e-05, + "loss": 0.7865, + "step": 2717 + }, + { + "epoch": 0.52, + "learning_rate": 9.99693023978479e-06, + "loss": 0.7823, + "step": 2718 + }, + { + "epoch": 0.52, + "learning_rate": 9.99079072051147e-06, + "loss": 0.818, + "step": 2719 + }, + { + "epoch": 0.52, + "learning_rate": 9.98465120470947e-06, + "loss": 0.2745, + "step": 2720 + }, + { + "epoch": 0.52, + "learning_rate": 9.978511694692999e-06, + "loss": 0.7786, + "step": 2721 + }, + { + "epoch": 0.52, + "learning_rate": 9.972372192776272e-06, + "loss": 0.8206, + "step": 2722 + }, + { + "epoch": 0.52, + "learning_rate": 9.966232701273485e-06, + "loss": 0.784, + "step": 2723 + }, + { + "epoch": 0.52, + "learning_rate": 9.960093222498852e-06, + "loss": 0.8314, + "step": 2724 + }, + { + "epoch": 0.52, + "learning_rate": 9.953953758766559e-06, + "loss": 0.8247, + "step": 2725 + }, + { + "epoch": 0.52, + "learning_rate": 9.94781431239081e-06, + "loss": 0.7817, + "step": 2726 + }, + { + "epoch": 0.52, + "learning_rate": 9.941674885685775e-06, + "loss": 0.7875, + "step": 2727 + }, + { + "epoch": 0.52, + "learning_rate": 9.935535480965647e-06, + "loss": 0.8581, + "step": 2728 + }, + { + "epoch": 0.52, + "learning_rate": 9.929396100544584e-06, + "loss": 0.8243, + "step": 2729 + }, + { + "epoch": 0.52, + "learning_rate": 9.923256746736755e-06, + "loss": 0.8147, + "step": 2730 + }, + { + "epoch": 0.52, + "learning_rate": 9.917117421856303e-06, + "loss": 0.7825, + "step": 2731 + }, + { + "epoch": 0.52, + "learning_rate": 9.91097812821737e-06, + "loss": 0.7547, + "step": 2732 + }, + { + "epoch": 0.52, + "learning_rate": 9.904838868134087e-06, + "loss": 0.8012, + "step": 2733 + }, + { + "epoch": 0.52, + "learning_rate": 9.898699643920562e-06, + "loss": 0.7884, + "step": 2734 + }, + { + "epoch": 0.52, + "learning_rate": 9.892560457890907e-06, + "loss": 0.7791, + "step": 2735 + }, + { + "epoch": 0.52, + "learning_rate": 9.886421312359196e-06, + "loss": 0.8544, + "step": 2736 + }, + { + "epoch": 0.52, + "learning_rate": 9.880282209639511e-06, + "loss": 0.8126, + "step": 2737 + }, + { + "epoch": 0.52, + "learning_rate": 9.874143152045902e-06, + "loss": 0.7861, + "step": 2738 + }, + { + "epoch": 0.52, + "learning_rate": 9.868004141892412e-06, + "loss": 0.7476, + "step": 2739 + }, + { + "epoch": 0.52, + "learning_rate": 9.861865181493057e-06, + "loss": 0.7953, + "step": 2740 + }, + { + "epoch": 0.52, + "learning_rate": 9.855726273161845e-06, + "loss": 0.8549, + "step": 2741 + }, + { + "epoch": 0.52, + "learning_rate": 9.849587419212751e-06, + "loss": 0.8041, + "step": 2742 + }, + { + "epoch": 0.52, + "learning_rate": 9.843448621959745e-06, + "loss": 0.8033, + "step": 2743 + }, + { + "epoch": 0.52, + "learning_rate": 9.83730988371676e-06, + "loss": 0.8152, + "step": 2744 + }, + { + "epoch": 0.52, + "learning_rate": 9.83117120679772e-06, + "loss": 0.8142, + "step": 2745 + }, + { + "epoch": 0.52, + "learning_rate": 9.825032593516512e-06, + "loss": 0.8101, + "step": 2746 + }, + { + "epoch": 0.52, + "learning_rate": 9.81889404618702e-06, + "loss": 0.825, + "step": 2747 + }, + { + "epoch": 0.52, + "learning_rate": 9.812755567123077e-06, + "loss": 0.8126, + "step": 2748 + }, + { + "epoch": 0.52, + "learning_rate": 9.806617158638515e-06, + "loss": 0.8822, + "step": 2749 + }, + { + "epoch": 0.52, + "learning_rate": 9.800478823047117e-06, + "loss": 0.785, + "step": 2750 + }, + { + "epoch": 0.52, + "learning_rate": 9.79434056266266e-06, + "loss": 0.2534, + "step": 2751 + }, + { + "epoch": 0.52, + "learning_rate": 9.788202379798873e-06, + "loss": 0.8425, + "step": 2752 + }, + { + "epoch": 0.52, + "learning_rate": 9.782064276769471e-06, + "loss": 0.8592, + "step": 2753 + }, + { + "epoch": 0.52, + "learning_rate": 9.775926255888132e-06, + "loss": 0.8198, + "step": 2754 + }, + { + "epoch": 0.52, + "learning_rate": 9.769788319468501e-06, + "loss": 0.7671, + "step": 2755 + }, + { + "epoch": 0.52, + "learning_rate": 9.763650469824198e-06, + "loss": 0.8315, + "step": 2756 + }, + { + "epoch": 0.52, + "learning_rate": 9.757512709268802e-06, + "loss": 0.8206, + "step": 2757 + }, + { + "epoch": 0.52, + "learning_rate": 9.751375040115864e-06, + "loss": 0.7652, + "step": 2758 + }, + { + "epoch": 0.52, + "learning_rate": 9.745237464678904e-06, + "loss": 0.7672, + "step": 2759 + }, + { + "epoch": 0.52, + "learning_rate": 9.739099985271394e-06, + "loss": 0.8059, + "step": 2760 + }, + { + "epoch": 0.52, + "learning_rate": 9.732962604206781e-06, + "loss": 0.8195, + "step": 2761 + }, + { + "epoch": 0.52, + "learning_rate": 9.726825323798478e-06, + "loss": 0.8549, + "step": 2762 + }, + { + "epoch": 0.52, + "learning_rate": 9.720688146359843e-06, + "loss": 0.8032, + "step": 2763 + }, + { + "epoch": 0.52, + "learning_rate": 9.714551074204216e-06, + "loss": 0.7789, + "step": 2764 + }, + { + "epoch": 0.52, + "learning_rate": 9.708414109644878e-06, + "loss": 0.8157, + "step": 2765 + }, + { + "epoch": 0.52, + "learning_rate": 9.702277254995089e-06, + "loss": 0.7932, + "step": 2766 + }, + { + "epoch": 0.52, + "learning_rate": 9.696140512568045e-06, + "loss": 0.7832, + "step": 2767 + }, + { + "epoch": 0.52, + "learning_rate": 9.690003884676925e-06, + "loss": 0.7946, + "step": 2768 + }, + { + "epoch": 0.52, + "learning_rate": 9.683867373634837e-06, + "loss": 0.8237, + "step": 2769 + }, + { + "epoch": 0.53, + "learning_rate": 9.677730981754875e-06, + "loss": 0.7239, + "step": 2770 + }, + { + "epoch": 0.53, + "learning_rate": 9.671594711350062e-06, + "loss": 0.8215, + "step": 2771 + }, + { + "epoch": 0.53, + "learning_rate": 9.665458564733393e-06, + "loss": 0.6921, + "step": 2772 + }, + { + "epoch": 0.53, + "learning_rate": 9.659322544217802e-06, + "loss": 0.2681, + "step": 2773 + }, + { + "epoch": 0.53, + "learning_rate": 9.653186652116188e-06, + "loss": 0.815, + "step": 2774 + }, + { + "epoch": 0.53, + "learning_rate": 9.647050890741397e-06, + "loss": 0.834, + "step": 2775 + }, + { + "epoch": 0.53, + "learning_rate": 9.640915262406223e-06, + "loss": 0.7583, + "step": 2776 + }, + { + "epoch": 0.53, + "learning_rate": 9.634779769423412e-06, + "loss": 0.7382, + "step": 2777 + }, + { + "epoch": 0.53, + "learning_rate": 9.628644414105663e-06, + "loss": 0.7803, + "step": 2778 + }, + { + "epoch": 0.53, + "learning_rate": 9.62250919876561e-06, + "loss": 0.8112, + "step": 2779 + }, + { + "epoch": 0.53, + "learning_rate": 9.616374125715855e-06, + "loss": 0.7753, + "step": 2780 + }, + { + "epoch": 0.53, + "learning_rate": 9.610239197268923e-06, + "loss": 0.7672, + "step": 2781 + }, + { + "epoch": 0.53, + "learning_rate": 9.604104415737309e-06, + "loss": 0.7562, + "step": 2782 + }, + { + "epoch": 0.53, + "learning_rate": 9.597969783433426e-06, + "loss": 0.8067, + "step": 2783 + }, + { + "epoch": 0.53, + "learning_rate": 9.591835302669657e-06, + "loss": 0.7994, + "step": 2784 + }, + { + "epoch": 0.53, + "learning_rate": 9.585700975758305e-06, + "loss": 0.7648, + "step": 2785 + }, + { + "epoch": 0.53, + "learning_rate": 9.579566805011635e-06, + "loss": 0.7908, + "step": 2786 + }, + { + "epoch": 0.53, + "learning_rate": 9.573432792741835e-06, + "loss": 0.8369, + "step": 2787 + }, + { + "epoch": 0.53, + "learning_rate": 9.567298941261045e-06, + "loss": 0.8564, + "step": 2788 + }, + { + "epoch": 0.53, + "learning_rate": 9.561165252881348e-06, + "loss": 0.6794, + "step": 2789 + }, + { + "epoch": 0.53, + "learning_rate": 9.555031729914746e-06, + "loss": 0.7838, + "step": 2790 + }, + { + "epoch": 0.53, + "learning_rate": 9.548898374673205e-06, + "loss": 0.7672, + "step": 2791 + }, + { + "epoch": 0.53, + "learning_rate": 9.5427651894686e-06, + "loss": 0.776, + "step": 2792 + }, + { + "epoch": 0.53, + "learning_rate": 9.536632176612772e-06, + "loss": 0.7567, + "step": 2793 + }, + { + "epoch": 0.53, + "learning_rate": 9.530499338417466e-06, + "loss": 0.7687, + "step": 2794 + }, + { + "epoch": 0.53, + "learning_rate": 9.524366677194385e-06, + "loss": 0.7917, + "step": 2795 + }, + { + "epoch": 0.53, + "learning_rate": 9.518234195255156e-06, + "loss": 0.8039, + "step": 2796 + }, + { + "epoch": 0.53, + "learning_rate": 9.512101894911337e-06, + "loss": 0.8052, + "step": 2797 + }, + { + "epoch": 0.53, + "learning_rate": 9.505969778474418e-06, + "loss": 0.7999, + "step": 2798 + }, + { + "epoch": 0.53, + "learning_rate": 9.499837848255829e-06, + "loss": 0.7761, + "step": 2799 + }, + { + "epoch": 0.53, + "learning_rate": 9.493706106566911e-06, + "loss": 0.8032, + "step": 2800 + }, + { + "epoch": 0.53, + "learning_rate": 9.487574555718955e-06, + "loss": 0.808, + "step": 2801 + }, + { + "epoch": 0.53, + "learning_rate": 9.48144319802316e-06, + "loss": 0.8243, + "step": 2802 + }, + { + "epoch": 0.53, + "learning_rate": 9.475312035790674e-06, + "loss": 0.8343, + "step": 2803 + }, + { + "epoch": 0.53, + "learning_rate": 9.46918107133255e-06, + "loss": 0.7897, + "step": 2804 + }, + { + "epoch": 0.53, + "learning_rate": 9.463050306959782e-06, + "loss": 0.837, + "step": 2805 + }, + { + "epoch": 0.53, + "learning_rate": 9.456919744983276e-06, + "loss": 0.8315, + "step": 2806 + }, + { + "epoch": 0.53, + "learning_rate": 9.450789387713878e-06, + "loss": 0.8774, + "step": 2807 + }, + { + "epoch": 0.53, + "learning_rate": 9.444659237462337e-06, + "loss": 0.789, + "step": 2808 + }, + { + "epoch": 0.53, + "learning_rate": 9.438529296539342e-06, + "loss": 0.768, + "step": 2809 + }, + { + "epoch": 0.53, + "learning_rate": 9.432399567255488e-06, + "loss": 0.7722, + "step": 2810 + }, + { + "epoch": 0.53, + "learning_rate": 9.426270051921305e-06, + "loss": 0.8069, + "step": 2811 + }, + { + "epoch": 0.53, + "learning_rate": 9.42014075284723e-06, + "loss": 0.7654, + "step": 2812 + }, + { + "epoch": 0.53, + "learning_rate": 9.414011672343623e-06, + "loss": 0.8337, + "step": 2813 + }, + { + "epoch": 0.53, + "learning_rate": 9.407882812720765e-06, + "loss": 0.7564, + "step": 2814 + }, + { + "epoch": 0.53, + "learning_rate": 9.40175417628885e-06, + "loss": 0.8046, + "step": 2815 + }, + { + "epoch": 0.53, + "learning_rate": 9.395625765357984e-06, + "loss": 0.7922, + "step": 2816 + }, + { + "epoch": 0.53, + "learning_rate": 9.389497582238199e-06, + "loss": 0.806, + "step": 2817 + }, + { + "epoch": 0.53, + "learning_rate": 9.383369629239432e-06, + "loss": 0.789, + "step": 2818 + }, + { + "epoch": 0.53, + "learning_rate": 9.377241908671533e-06, + "loss": 0.8074, + "step": 2819 + }, + { + "epoch": 0.53, + "learning_rate": 9.371114422844272e-06, + "loss": 0.7797, + "step": 2820 + }, + { + "epoch": 0.53, + "learning_rate": 9.364987174067323e-06, + "loss": 0.833, + "step": 2821 + }, + { + "epoch": 0.53, + "learning_rate": 9.358860164650276e-06, + "loss": 0.8682, + "step": 2822 + }, + { + "epoch": 0.54, + "learning_rate": 9.352733396902624e-06, + "loss": 0.8127, + "step": 2823 + }, + { + "epoch": 0.54, + "learning_rate": 9.346606873133777e-06, + "loss": 0.7552, + "step": 2824 + }, + { + "epoch": 0.54, + "learning_rate": 9.340480595653047e-06, + "loss": 0.832, + "step": 2825 + }, + { + "epoch": 0.54, + "learning_rate": 9.334354566769658e-06, + "loss": 0.7695, + "step": 2826 + }, + { + "epoch": 0.54, + "learning_rate": 9.328228788792732e-06, + "loss": 0.7796, + "step": 2827 + }, + { + "epoch": 0.54, + "learning_rate": 9.32210326403131e-06, + "loss": 0.7956, + "step": 2828 + }, + { + "epoch": 0.54, + "learning_rate": 9.31597799479432e-06, + "loss": 0.8494, + "step": 2829 + }, + { + "epoch": 0.54, + "learning_rate": 9.309852983390614e-06, + "loss": 0.7879, + "step": 2830 + }, + { + "epoch": 0.54, + "learning_rate": 9.303728232128925e-06, + "loss": 0.8159, + "step": 2831 + }, + { + "epoch": 0.54, + "learning_rate": 9.297603743317909e-06, + "loss": 0.7809, + "step": 2832 + }, + { + "epoch": 0.54, + "learning_rate": 9.291479519266108e-06, + "loss": 0.7932, + "step": 2833 + }, + { + "epoch": 0.54, + "learning_rate": 9.285355562281971e-06, + "loss": 0.7895, + "step": 2834 + }, + { + "epoch": 0.54, + "learning_rate": 9.27923187467384e-06, + "loss": 0.8236, + "step": 2835 + }, + { + "epoch": 0.54, + "learning_rate": 9.273108458749969e-06, + "loss": 0.813, + "step": 2836 + }, + { + "epoch": 0.54, + "learning_rate": 9.266985316818491e-06, + "loss": 0.7738, + "step": 2837 + }, + { + "epoch": 0.54, + "learning_rate": 9.260862451187455e-06, + "loss": 0.7825, + "step": 2838 + }, + { + "epoch": 0.54, + "learning_rate": 9.254739864164788e-06, + "loss": 0.8165, + "step": 2839 + }, + { + "epoch": 0.54, + "learning_rate": 9.248617558058328e-06, + "loss": 0.7704, + "step": 2840 + }, + { + "epoch": 0.54, + "learning_rate": 9.242495535175789e-06, + "loss": 0.7355, + "step": 2841 + }, + { + "epoch": 0.54, + "learning_rate": 9.2363737978248e-06, + "loss": 0.8004, + "step": 2842 + }, + { + "epoch": 0.54, + "learning_rate": 9.23025234831286e-06, + "loss": 0.8151, + "step": 2843 + }, + { + "epoch": 0.54, + "learning_rate": 9.224131188947377e-06, + "loss": 0.7934, + "step": 2844 + }, + { + "epoch": 0.54, + "learning_rate": 9.218010322035648e-06, + "loss": 0.759, + "step": 2845 + }, + { + "epoch": 0.54, + "learning_rate": 9.211889749884842e-06, + "loss": 0.7739, + "step": 2846 + }, + { + "epoch": 0.54, + "learning_rate": 9.205769474802045e-06, + "loss": 0.777, + "step": 2847 + }, + { + "epoch": 0.54, + "learning_rate": 9.1996494990942e-06, + "loss": 0.8445, + "step": 2848 + }, + { + "epoch": 0.54, + "learning_rate": 9.193529825068165e-06, + "loss": 0.8248, + "step": 2849 + }, + { + "epoch": 0.54, + "learning_rate": 9.187410455030667e-06, + "loss": 0.8172, + "step": 2850 + }, + { + "epoch": 0.54, + "learning_rate": 9.181291391288327e-06, + "loss": 0.7829, + "step": 2851 + }, + { + "epoch": 0.54, + "learning_rate": 9.17517263614764e-06, + "loss": 0.7489, + "step": 2852 + }, + { + "epoch": 0.54, + "learning_rate": 9.169054191914998e-06, + "loss": 0.8274, + "step": 2853 + }, + { + "epoch": 0.54, + "learning_rate": 9.162936060896672e-06, + "loss": 0.7709, + "step": 2854 + }, + { + "epoch": 0.54, + "learning_rate": 9.156818245398807e-06, + "loss": 0.8241, + "step": 2855 + }, + { + "epoch": 0.54, + "learning_rate": 9.150700747727433e-06, + "loss": 0.7759, + "step": 2856 + }, + { + "epoch": 0.54, + "learning_rate": 9.144583570188472e-06, + "loss": 0.8537, + "step": 2857 + }, + { + "epoch": 0.54, + "learning_rate": 9.138466715087704e-06, + "loss": 0.7998, + "step": 2858 + }, + { + "epoch": 0.54, + "learning_rate": 9.132350184730809e-06, + "loss": 0.7553, + "step": 2859 + }, + { + "epoch": 0.54, + "learning_rate": 9.126233981423323e-06, + "loss": 0.7946, + "step": 2860 + }, + { + "epoch": 0.54, + "learning_rate": 9.12011810747068e-06, + "loss": 0.8054, + "step": 2861 + }, + { + "epoch": 0.54, + "learning_rate": 9.114002565178172e-06, + "loss": 0.7961, + "step": 2862 + }, + { + "epoch": 0.54, + "learning_rate": 9.107887356850982e-06, + "loss": 0.8454, + "step": 2863 + }, + { + "epoch": 0.54, + "learning_rate": 9.10177248479415e-06, + "loss": 0.7676, + "step": 2864 + }, + { + "epoch": 0.54, + "learning_rate": 9.095657951312607e-06, + "loss": 0.8294, + "step": 2865 + }, + { + "epoch": 0.54, + "learning_rate": 9.08954375871114e-06, + "loss": 0.7583, + "step": 2866 + }, + { + "epoch": 0.54, + "learning_rate": 9.083429909294422e-06, + "loss": 0.8743, + "step": 2867 + }, + { + "epoch": 0.54, + "learning_rate": 9.07731640536698e-06, + "loss": 0.7653, + "step": 2868 + }, + { + "epoch": 0.54, + "learning_rate": 9.071203249233233e-06, + "loss": 0.7701, + "step": 2869 + }, + { + "epoch": 0.54, + "learning_rate": 9.065090443197447e-06, + "loss": 0.8437, + "step": 2870 + }, + { + "epoch": 0.54, + "learning_rate": 9.05897798956377e-06, + "loss": 0.7668, + "step": 2871 + }, + { + "epoch": 0.54, + "learning_rate": 9.052865890636208e-06, + "loss": 0.7941, + "step": 2872 + }, + { + "epoch": 0.54, + "learning_rate": 9.046754148718644e-06, + "loss": 0.7817, + "step": 2873 + }, + { + "epoch": 0.54, + "learning_rate": 9.040642766114817e-06, + "loss": 0.8029, + "step": 2874 + }, + { + "epoch": 0.54, + "learning_rate": 9.034531745128334e-06, + "loss": 0.8053, + "step": 2875 + }, + { + "epoch": 0.55, + "learning_rate": 9.028421088062665e-06, + "loss": 0.7359, + "step": 2876 + }, + { + "epoch": 0.55, + "learning_rate": 9.022310797221143e-06, + "loss": 0.7722, + "step": 2877 + }, + { + "epoch": 0.55, + "learning_rate": 9.01620087490697e-06, + "loss": 0.7878, + "step": 2878 + }, + { + "epoch": 0.55, + "learning_rate": 9.010091323423192e-06, + "loss": 0.8392, + "step": 2879 + }, + { + "epoch": 0.55, + "learning_rate": 9.003982145072737e-06, + "loss": 0.7183, + "step": 2880 + }, + { + "epoch": 0.55, + "learning_rate": 8.997873342158368e-06, + "loss": 0.7838, + "step": 2881 + }, + { + "epoch": 0.55, + "learning_rate": 8.991764916982731e-06, + "loss": 0.8294, + "step": 2882 + }, + { + "epoch": 0.55, + "learning_rate": 8.985656871848311e-06, + "loss": 0.7886, + "step": 2883 + }, + { + "epoch": 0.55, + "learning_rate": 8.979549209057462e-06, + "loss": 0.7329, + "step": 2884 + }, + { + "epoch": 0.55, + "learning_rate": 8.973441930912382e-06, + "loss": 0.8326, + "step": 2885 + }, + { + "epoch": 0.55, + "learning_rate": 8.967335039715141e-06, + "loss": 0.8263, + "step": 2886 + }, + { + "epoch": 0.55, + "learning_rate": 8.961228537767642e-06, + "loss": 0.8097, + "step": 2887 + }, + { + "epoch": 0.55, + "learning_rate": 8.95512242737166e-06, + "loss": 0.8423, + "step": 2888 + }, + { + "epoch": 0.55, + "learning_rate": 8.949016710828808e-06, + "loss": 0.7499, + "step": 2889 + }, + { + "epoch": 0.55, + "learning_rate": 8.942911390440563e-06, + "loss": 0.8287, + "step": 2890 + }, + { + "epoch": 0.55, + "learning_rate": 8.936806468508245e-06, + "loss": 0.7199, + "step": 2891 + }, + { + "epoch": 0.55, + "learning_rate": 8.930701947333026e-06, + "loss": 0.7964, + "step": 2892 + }, + { + "epoch": 0.55, + "learning_rate": 8.924597829215924e-06, + "loss": 0.7887, + "step": 2893 + }, + { + "epoch": 0.55, + "learning_rate": 8.918494116457812e-06, + "loss": 0.813, + "step": 2894 + }, + { + "epoch": 0.55, + "learning_rate": 8.9123908113594e-06, + "loss": 0.7684, + "step": 2895 + }, + { + "epoch": 0.55, + "learning_rate": 8.906287916221259e-06, + "loss": 0.7964, + "step": 2896 + }, + { + "epoch": 0.55, + "learning_rate": 8.900185433343786e-06, + "loss": 0.7723, + "step": 2897 + }, + { + "epoch": 0.55, + "learning_rate": 8.894083365027243e-06, + "loss": 0.7884, + "step": 2898 + }, + { + "epoch": 0.55, + "learning_rate": 8.887981713571717e-06, + "loss": 0.8394, + "step": 2899 + }, + { + "epoch": 0.55, + "learning_rate": 8.881880481277152e-06, + "loss": 0.8058, + "step": 2900 + }, + { + "epoch": 0.55, + "learning_rate": 8.875779670443334e-06, + "loss": 0.774, + "step": 2901 + }, + { + "epoch": 0.55, + "learning_rate": 8.869679283369872e-06, + "loss": 0.7635, + "step": 2902 + }, + { + "epoch": 0.55, + "learning_rate": 8.863579322356242e-06, + "loss": 0.2591, + "step": 2903 + }, + { + "epoch": 0.55, + "learning_rate": 8.857479789701734e-06, + "loss": 0.802, + "step": 2904 + }, + { + "epoch": 0.55, + "learning_rate": 8.851380687705498e-06, + "loss": 0.7538, + "step": 2905 + }, + { + "epoch": 0.55, + "learning_rate": 8.845282018666503e-06, + "loss": 0.7594, + "step": 2906 + }, + { + "epoch": 0.55, + "learning_rate": 8.839183784883572e-06, + "loss": 0.811, + "step": 2907 + }, + { + "epoch": 0.55, + "learning_rate": 8.833085988655349e-06, + "loss": 0.8127, + "step": 2908 + }, + { + "epoch": 0.55, + "learning_rate": 8.826988632280326e-06, + "loss": 0.777, + "step": 2909 + }, + { + "epoch": 0.55, + "learning_rate": 8.820891718056815e-06, + "loss": 0.8115, + "step": 2910 + }, + { + "epoch": 0.55, + "learning_rate": 8.814795248282974e-06, + "loss": 0.7827, + "step": 2911 + }, + { + "epoch": 0.55, + "learning_rate": 8.80869922525679e-06, + "loss": 0.7717, + "step": 2912 + }, + { + "epoch": 0.55, + "learning_rate": 8.802603651276079e-06, + "loss": 0.813, + "step": 2913 + }, + { + "epoch": 0.55, + "learning_rate": 8.796508528638485e-06, + "loss": 0.8075, + "step": 2914 + }, + { + "epoch": 0.55, + "learning_rate": 8.790413859641494e-06, + "loss": 0.7806, + "step": 2915 + }, + { + "epoch": 0.55, + "learning_rate": 8.784319646582406e-06, + "loss": 0.796, + "step": 2916 + }, + { + "epoch": 0.55, + "learning_rate": 8.77822589175836e-06, + "loss": 0.8221, + "step": 2917 + }, + { + "epoch": 0.55, + "learning_rate": 8.772132597466313e-06, + "loss": 0.8155, + "step": 2918 + }, + { + "epoch": 0.55, + "learning_rate": 8.766039766003065e-06, + "loss": 0.7301, + "step": 2919 + }, + { + "epoch": 0.55, + "learning_rate": 8.759947399665216e-06, + "loss": 0.8151, + "step": 2920 + }, + { + "epoch": 0.55, + "learning_rate": 8.753855500749217e-06, + "loss": 0.7963, + "step": 2921 + }, + { + "epoch": 0.55, + "learning_rate": 8.74776407155132e-06, + "loss": 0.7999, + "step": 2922 + }, + { + "epoch": 0.55, + "learning_rate": 8.741673114367621e-06, + "loss": 0.7647, + "step": 2923 + }, + { + "epoch": 0.55, + "learning_rate": 8.73558263149402e-06, + "loss": 0.7751, + "step": 2924 + }, + { + "epoch": 0.55, + "learning_rate": 8.729492625226253e-06, + "loss": 0.8319, + "step": 2925 + }, + { + "epoch": 0.55, + "learning_rate": 8.723403097859861e-06, + "loss": 0.7898, + "step": 2926 + }, + { + "epoch": 0.55, + "learning_rate": 8.71731405169022e-06, + "loss": 0.7936, + "step": 2927 + }, + { + "epoch": 0.55, + "learning_rate": 8.711225489012514e-06, + "loss": 0.8219, + "step": 2928 + }, + { + "epoch": 0.56, + "learning_rate": 8.705137412121748e-06, + "loss": 0.8499, + "step": 2929 + }, + { + "epoch": 0.56, + "learning_rate": 8.699049823312747e-06, + "loss": 0.7992, + "step": 2930 + }, + { + "epoch": 0.56, + "learning_rate": 8.692962724880148e-06, + "loss": 0.7749, + "step": 2931 + }, + { + "epoch": 0.56, + "learning_rate": 8.686876119118402e-06, + "loss": 0.7843, + "step": 2932 + }, + { + "epoch": 0.56, + "learning_rate": 8.68079000832178e-06, + "loss": 0.8033, + "step": 2933 + }, + { + "epoch": 0.56, + "learning_rate": 8.67470439478436e-06, + "loss": 0.7809, + "step": 2934 + }, + { + "epoch": 0.56, + "learning_rate": 8.668619280800035e-06, + "loss": 0.805, + "step": 2935 + }, + { + "epoch": 0.56, + "learning_rate": 8.662534668662517e-06, + "loss": 0.8206, + "step": 2936 + }, + { + "epoch": 0.56, + "learning_rate": 8.656450560665313e-06, + "loss": 0.8237, + "step": 2937 + }, + { + "epoch": 0.56, + "learning_rate": 8.650366959101757e-06, + "loss": 0.8326, + "step": 2938 + }, + { + "epoch": 0.56, + "learning_rate": 8.644283866264976e-06, + "loss": 0.8237, + "step": 2939 + }, + { + "epoch": 0.56, + "learning_rate": 8.638201284447922e-06, + "loss": 0.8212, + "step": 2940 + }, + { + "epoch": 0.56, + "learning_rate": 8.632119215943339e-06, + "loss": 0.7704, + "step": 2941 + }, + { + "epoch": 0.56, + "learning_rate": 8.62603766304379e-06, + "loss": 0.8144, + "step": 2942 + }, + { + "epoch": 0.56, + "learning_rate": 8.619956628041632e-06, + "loss": 0.81, + "step": 2943 + }, + { + "epoch": 0.56, + "learning_rate": 8.61387611322904e-06, + "loss": 0.8014, + "step": 2944 + }, + { + "epoch": 0.56, + "learning_rate": 8.607796120897978e-06, + "loss": 0.8133, + "step": 2945 + }, + { + "epoch": 0.56, + "learning_rate": 8.601716653340226e-06, + "loss": 0.7601, + "step": 2946 + }, + { + "epoch": 0.56, + "learning_rate": 8.595637712847357e-06, + "loss": 0.8142, + "step": 2947 + }, + { + "epoch": 0.56, + "learning_rate": 8.589559301710753e-06, + "loss": 0.8087, + "step": 2948 + }, + { + "epoch": 0.56, + "learning_rate": 8.583481422221592e-06, + "loss": 0.8556, + "step": 2949 + }, + { + "epoch": 0.56, + "learning_rate": 8.577404076670852e-06, + "loss": 0.8401, + "step": 2950 + }, + { + "epoch": 0.56, + "learning_rate": 8.571327267349307e-06, + "loss": 0.7975, + "step": 2951 + }, + { + "epoch": 0.56, + "learning_rate": 8.565250996547538e-06, + "loss": 0.7505, + "step": 2952 + }, + { + "epoch": 0.56, + "learning_rate": 8.55917526655591e-06, + "loss": 0.7773, + "step": 2953 + }, + { + "epoch": 0.56, + "learning_rate": 8.553100079664598e-06, + "loss": 0.8059, + "step": 2954 + }, + { + "epoch": 0.56, + "learning_rate": 8.54702543816356e-06, + "loss": 0.7456, + "step": 2955 + }, + { + "epoch": 0.56, + "learning_rate": 8.540951344342553e-06, + "loss": 0.2664, + "step": 2956 + }, + { + "epoch": 0.56, + "learning_rate": 8.534877800491139e-06, + "loss": 0.2662, + "step": 2957 + }, + { + "epoch": 0.56, + "learning_rate": 8.528804808898648e-06, + "loss": 0.8285, + "step": 2958 + }, + { + "epoch": 0.56, + "learning_rate": 8.522732371854228e-06, + "loss": 0.2754, + "step": 2959 + }, + { + "epoch": 0.56, + "learning_rate": 8.516660491646794e-06, + "loss": 0.8414, + "step": 2960 + }, + { + "epoch": 0.56, + "learning_rate": 8.510589170565074e-06, + "loss": 0.7927, + "step": 2961 + }, + { + "epoch": 0.56, + "learning_rate": 8.504518410897565e-06, + "loss": 0.753, + "step": 2962 + }, + { + "epoch": 0.56, + "learning_rate": 8.49844821493257e-06, + "loss": 0.7704, + "step": 2963 + }, + { + "epoch": 0.56, + "learning_rate": 8.492378584958165e-06, + "loss": 0.7828, + "step": 2964 + }, + { + "epoch": 0.56, + "learning_rate": 8.486309523262223e-06, + "loss": 0.7856, + "step": 2965 + }, + { + "epoch": 0.56, + "learning_rate": 8.480241032132394e-06, + "loss": 0.795, + "step": 2966 + }, + { + "epoch": 0.56, + "learning_rate": 8.47417311385612e-06, + "loss": 0.7797, + "step": 2967 + }, + { + "epoch": 0.56, + "learning_rate": 8.468105770720627e-06, + "loss": 0.8311, + "step": 2968 + }, + { + "epoch": 0.56, + "learning_rate": 8.462039005012917e-06, + "loss": 0.7861, + "step": 2969 + }, + { + "epoch": 0.56, + "learning_rate": 8.455972819019782e-06, + "loss": 0.7982, + "step": 2970 + }, + { + "epoch": 0.56, + "learning_rate": 8.449907215027793e-06, + "loss": 0.7985, + "step": 2971 + }, + { + "epoch": 0.56, + "learning_rate": 8.443842195323297e-06, + "loss": 0.791, + "step": 2972 + }, + { + "epoch": 0.56, + "learning_rate": 8.437777762192434e-06, + "loss": 0.8782, + "step": 2973 + }, + { + "epoch": 0.56, + "learning_rate": 8.431713917921104e-06, + "loss": 0.8019, + "step": 2974 + }, + { + "epoch": 0.56, + "learning_rate": 8.425650664795003e-06, + "loss": 0.8207, + "step": 2975 + }, + { + "epoch": 0.56, + "learning_rate": 8.419588005099588e-06, + "loss": 0.8011, + "step": 2976 + }, + { + "epoch": 0.56, + "learning_rate": 8.413525941120108e-06, + "loss": 0.8093, + "step": 2977 + }, + { + "epoch": 0.56, + "learning_rate": 8.407464475141575e-06, + "loss": 0.7919, + "step": 2978 + }, + { + "epoch": 0.56, + "learning_rate": 8.401403609448785e-06, + "loss": 0.8353, + "step": 2979 + }, + { + "epoch": 0.56, + "learning_rate": 8.395343346326295e-06, + "loss": 0.801, + "step": 2980 + }, + { + "epoch": 0.57, + "learning_rate": 8.389283688058454e-06, + "loss": 0.2522, + "step": 2981 + }, + { + "epoch": 0.57, + "learning_rate": 8.383224636929364e-06, + "loss": 0.797, + "step": 2982 + }, + { + "epoch": 0.57, + "learning_rate": 8.37716619522291e-06, + "loss": 0.8085, + "step": 2983 + }, + { + "epoch": 0.57, + "learning_rate": 8.37110836522274e-06, + "loss": 0.7471, + "step": 2984 + }, + { + "epoch": 0.57, + "learning_rate": 8.365051149212275e-06, + "loss": 0.7388, + "step": 2985 + }, + { + "epoch": 0.57, + "learning_rate": 8.358994549474712e-06, + "loss": 0.2792, + "step": 2986 + }, + { + "epoch": 0.57, + "learning_rate": 8.352938568293e-06, + "loss": 0.8641, + "step": 2987 + }, + { + "epoch": 0.57, + "learning_rate": 8.346883207949868e-06, + "loss": 0.7449, + "step": 2988 + }, + { + "epoch": 0.57, + "learning_rate": 8.340828470727801e-06, + "loss": 0.7111, + "step": 2989 + }, + { + "epoch": 0.57, + "learning_rate": 8.33477435890906e-06, + "loss": 0.752, + "step": 2990 + }, + { + "epoch": 0.57, + "learning_rate": 8.328720874775658e-06, + "loss": 0.2752, + "step": 2991 + }, + { + "epoch": 0.57, + "learning_rate": 8.322668020609387e-06, + "loss": 0.8133, + "step": 2992 + }, + { + "epoch": 0.57, + "learning_rate": 8.316615798691781e-06, + "loss": 0.8169, + "step": 2993 + }, + { + "epoch": 0.57, + "learning_rate": 8.310564211304159e-06, + "loss": 0.8021, + "step": 2994 + }, + { + "epoch": 0.57, + "learning_rate": 8.304513260727577e-06, + "loss": 0.8232, + "step": 2995 + }, + { + "epoch": 0.57, + "learning_rate": 8.298462949242873e-06, + "loss": 0.8239, + "step": 2996 + }, + { + "epoch": 0.57, + "learning_rate": 8.292413279130625e-06, + "loss": 0.7794, + "step": 2997 + }, + { + "epoch": 0.57, + "learning_rate": 8.286364252671188e-06, + "loss": 0.814, + "step": 2998 + }, + { + "epoch": 0.57, + "learning_rate": 8.280315872144654e-06, + "loss": 0.7627, + "step": 2999 + }, + { + "epoch": 0.57, + "learning_rate": 8.274268139830891e-06, + "loss": 0.832, + "step": 3000 + }, + { + "epoch": 0.57, + "learning_rate": 8.268221058009506e-06, + "loss": 0.7669, + "step": 3001 + }, + { + "epoch": 0.57, + "learning_rate": 8.262174628959877e-06, + "loss": 0.7207, + "step": 3002 + }, + { + "epoch": 0.57, + "learning_rate": 8.256128854961118e-06, + "loss": 0.7651, + "step": 3003 + }, + { + "epoch": 0.57, + "learning_rate": 8.250083738292116e-06, + "loss": 0.795, + "step": 3004 + }, + { + "epoch": 0.57, + "learning_rate": 8.24403928123149e-06, + "loss": 0.8108, + "step": 3005 + }, + { + "epoch": 0.57, + "learning_rate": 8.237995486057627e-06, + "loss": 0.8022, + "step": 3006 + }, + { + "epoch": 0.57, + "learning_rate": 8.231952355048656e-06, + "loss": 0.8422, + "step": 3007 + }, + { + "epoch": 0.57, + "learning_rate": 8.225909890482456e-06, + "loss": 0.7609, + "step": 3008 + }, + { + "epoch": 0.57, + "learning_rate": 8.219868094636653e-06, + "loss": 0.7798, + "step": 3009 + }, + { + "epoch": 0.57, + "learning_rate": 8.213826969788633e-06, + "loss": 0.7699, + "step": 3010 + }, + { + "epoch": 0.57, + "learning_rate": 8.207786518215512e-06, + "loss": 0.7899, + "step": 3011 + }, + { + "epoch": 0.57, + "learning_rate": 8.201746742194164e-06, + "loss": 0.8587, + "step": 3012 + }, + { + "epoch": 0.57, + "learning_rate": 8.195707644001206e-06, + "loss": 0.8142, + "step": 3013 + }, + { + "epoch": 0.57, + "learning_rate": 8.189669225912993e-06, + "loss": 0.8048, + "step": 3014 + }, + { + "epoch": 0.57, + "learning_rate": 8.183631490205636e-06, + "loss": 0.7573, + "step": 3015 + }, + { + "epoch": 0.57, + "learning_rate": 8.177594439154975e-06, + "loss": 0.8443, + "step": 3016 + }, + { + "epoch": 0.57, + "learning_rate": 8.171558075036606e-06, + "loss": 0.8073, + "step": 3017 + }, + { + "epoch": 0.57, + "learning_rate": 8.165522400125849e-06, + "loss": 0.7585, + "step": 3018 + }, + { + "epoch": 0.57, + "learning_rate": 8.159487416697782e-06, + "loss": 0.8024, + "step": 3019 + }, + { + "epoch": 0.57, + "learning_rate": 8.153453127027209e-06, + "loss": 0.8074, + "step": 3020 + }, + { + "epoch": 0.57, + "learning_rate": 8.147419533388683e-06, + "loss": 0.7967, + "step": 3021 + }, + { + "epoch": 0.57, + "learning_rate": 8.141386638056482e-06, + "loss": 0.8525, + "step": 3022 + }, + { + "epoch": 0.57, + "learning_rate": 8.135354443304638e-06, + "loss": 0.8068, + "step": 3023 + }, + { + "epoch": 0.57, + "learning_rate": 8.129322951406897e-06, + "loss": 0.8137, + "step": 3024 + }, + { + "epoch": 0.57, + "learning_rate": 8.123292164636761e-06, + "loss": 0.8423, + "step": 3025 + }, + { + "epoch": 0.57, + "learning_rate": 8.117262085267456e-06, + "loss": 0.8023, + "step": 3026 + }, + { + "epoch": 0.57, + "learning_rate": 8.111232715571942e-06, + "loss": 0.8275, + "step": 3027 + }, + { + "epoch": 0.57, + "learning_rate": 8.10520405782291e-06, + "loss": 0.7899, + "step": 3028 + }, + { + "epoch": 0.57, + "learning_rate": 8.09917611429279e-06, + "loss": 0.7895, + "step": 3029 + }, + { + "epoch": 0.57, + "learning_rate": 8.09314888725373e-06, + "loss": 0.8244, + "step": 3030 + }, + { + "epoch": 0.57, + "learning_rate": 8.087122378977624e-06, + "loss": 0.8259, + "step": 3031 + }, + { + "epoch": 0.57, + "learning_rate": 8.081096591736077e-06, + "loss": 0.7937, + "step": 3032 + }, + { + "epoch": 0.57, + "learning_rate": 8.075071527800445e-06, + "loss": 0.8589, + "step": 3033 + }, + { + "epoch": 0.58, + "learning_rate": 8.069047189441784e-06, + "loss": 0.8202, + "step": 3034 + }, + { + "epoch": 0.58, + "learning_rate": 8.0630235789309e-06, + "loss": 0.8178, + "step": 3035 + }, + { + "epoch": 0.58, + "learning_rate": 8.057000698538311e-06, + "loss": 0.7677, + "step": 3036 + }, + { + "epoch": 0.58, + "learning_rate": 8.05097855053427e-06, + "loss": 0.7854, + "step": 3037 + }, + { + "epoch": 0.58, + "learning_rate": 8.044957137188737e-06, + "loss": 0.784, + "step": 3038 + }, + { + "epoch": 0.58, + "learning_rate": 8.038936460771417e-06, + "loss": 0.8113, + "step": 3039 + }, + { + "epoch": 0.58, + "learning_rate": 8.03291652355172e-06, + "loss": 0.7905, + "step": 3040 + }, + { + "epoch": 0.58, + "learning_rate": 8.026897327798785e-06, + "loss": 0.8281, + "step": 3041 + }, + { + "epoch": 0.58, + "learning_rate": 8.020878875781477e-06, + "loss": 0.7452, + "step": 3042 + }, + { + "epoch": 0.58, + "learning_rate": 8.014861169768362e-06, + "loss": 0.7487, + "step": 3043 + }, + { + "epoch": 0.58, + "learning_rate": 8.00884421202775e-06, + "loss": 0.7736, + "step": 3044 + }, + { + "epoch": 0.58, + "learning_rate": 8.002828004827641e-06, + "loss": 0.8225, + "step": 3045 + }, + { + "epoch": 0.58, + "learning_rate": 7.99681255043578e-06, + "loss": 0.7588, + "step": 3046 + }, + { + "epoch": 0.58, + "learning_rate": 7.990797851119607e-06, + "loss": 0.7859, + "step": 3047 + }, + { + "epoch": 0.58, + "learning_rate": 7.984783909146287e-06, + "loss": 0.7773, + "step": 3048 + }, + { + "epoch": 0.58, + "learning_rate": 7.978770726782697e-06, + "loss": 0.7766, + "step": 3049 + }, + { + "epoch": 0.58, + "learning_rate": 7.972758306295436e-06, + "loss": 0.806, + "step": 3050 + }, + { + "epoch": 0.58, + "learning_rate": 7.966746649950797e-06, + "loss": 0.7782, + "step": 3051 + }, + { + "epoch": 0.58, + "learning_rate": 7.960735760014807e-06, + "loss": 0.7731, + "step": 3052 + }, + { + "epoch": 0.58, + "learning_rate": 7.954725638753184e-06, + "loss": 0.8001, + "step": 3053 + }, + { + "epoch": 0.58, + "learning_rate": 7.948716288431374e-06, + "loss": 0.7052, + "step": 3054 + }, + { + "epoch": 0.58, + "learning_rate": 7.942707711314517e-06, + "loss": 0.7273, + "step": 3055 + }, + { + "epoch": 0.58, + "learning_rate": 7.936699909667477e-06, + "loss": 0.8008, + "step": 3056 + }, + { + "epoch": 0.58, + "learning_rate": 7.930692885754806e-06, + "loss": 0.8419, + "step": 3057 + }, + { + "epoch": 0.58, + "learning_rate": 7.924686641840787e-06, + "loss": 0.8132, + "step": 3058 + }, + { + "epoch": 0.58, + "learning_rate": 7.918681180189384e-06, + "loss": 0.7972, + "step": 3059 + }, + { + "epoch": 0.58, + "learning_rate": 7.91267650306429e-06, + "loss": 0.8202, + "step": 3060 + }, + { + "epoch": 0.58, + "learning_rate": 7.906672612728879e-06, + "loss": 0.7909, + "step": 3061 + }, + { + "epoch": 0.58, + "learning_rate": 7.900669511446251e-06, + "loss": 0.8017, + "step": 3062 + }, + { + "epoch": 0.58, + "learning_rate": 7.894667201479191e-06, + "loss": 0.7849, + "step": 3063 + }, + { + "epoch": 0.58, + "learning_rate": 7.888665685090194e-06, + "loss": 0.7903, + "step": 3064 + }, + { + "epoch": 0.58, + "learning_rate": 7.882664964541453e-06, + "loss": 0.7349, + "step": 3065 + }, + { + "epoch": 0.58, + "learning_rate": 7.876665042094867e-06, + "loss": 0.8126, + "step": 3066 + }, + { + "epoch": 0.58, + "learning_rate": 7.870665920012022e-06, + "loss": 0.2938, + "step": 3067 + }, + { + "epoch": 0.58, + "learning_rate": 7.864667600554216e-06, + "loss": 0.2891, + "step": 3068 + }, + { + "epoch": 0.58, + "learning_rate": 7.858670085982434e-06, + "loss": 0.8199, + "step": 3069 + }, + { + "epoch": 0.58, + "learning_rate": 7.852673378557365e-06, + "loss": 0.8041, + "step": 3070 + }, + { + "epoch": 0.58, + "learning_rate": 7.846677480539392e-06, + "loss": 0.7603, + "step": 3071 + }, + { + "epoch": 0.58, + "learning_rate": 7.840682394188586e-06, + "loss": 0.8196, + "step": 3072 + }, + { + "epoch": 0.58, + "learning_rate": 7.834688121764724e-06, + "loss": 0.8051, + "step": 3073 + }, + { + "epoch": 0.58, + "learning_rate": 7.82869466552726e-06, + "loss": 0.7355, + "step": 3074 + }, + { + "epoch": 0.58, + "learning_rate": 7.822702027735363e-06, + "loss": 0.8026, + "step": 3075 + }, + { + "epoch": 0.58, + "learning_rate": 7.816710210647868e-06, + "loss": 0.7405, + "step": 3076 + }, + { + "epoch": 0.58, + "learning_rate": 7.810719216523324e-06, + "loss": 0.8381, + "step": 3077 + }, + { + "epoch": 0.58, + "learning_rate": 7.80472904761995e-06, + "loss": 0.8222, + "step": 3078 + }, + { + "epoch": 0.58, + "learning_rate": 7.79873970619567e-06, + "loss": 0.736, + "step": 3079 + }, + { + "epoch": 0.58, + "learning_rate": 7.792751194508082e-06, + "loss": 0.792, + "step": 3080 + }, + { + "epoch": 0.58, + "learning_rate": 7.786763514814488e-06, + "loss": 0.7813, + "step": 3081 + }, + { + "epoch": 0.58, + "learning_rate": 7.780776669371855e-06, + "loss": 0.7393, + "step": 3082 + }, + { + "epoch": 0.58, + "learning_rate": 7.774790660436857e-06, + "loss": 0.7546, + "step": 3083 + }, + { + "epoch": 0.58, + "learning_rate": 7.768805490265839e-06, + "loss": 0.7345, + "step": 3084 + }, + { + "epoch": 0.58, + "learning_rate": 7.762821161114834e-06, + "loss": 0.8169, + "step": 3085 + }, + { + "epoch": 0.58, + "learning_rate": 7.756837675239554e-06, + "loss": 0.7794, + "step": 3086 + }, + { + "epoch": 0.59, + "learning_rate": 7.750855034895405e-06, + "loss": 0.8379, + "step": 3087 + }, + { + "epoch": 0.59, + "learning_rate": 7.74487324233746e-06, + "loss": 0.7838, + "step": 3088 + }, + { + "epoch": 0.59, + "learning_rate": 7.738892299820483e-06, + "loss": 0.8183, + "step": 3089 + }, + { + "epoch": 0.59, + "learning_rate": 7.732912209598905e-06, + "loss": 0.8351, + "step": 3090 + }, + { + "epoch": 0.59, + "learning_rate": 7.726932973926855e-06, + "loss": 0.7555, + "step": 3091 + }, + { + "epoch": 0.59, + "learning_rate": 7.720954595058118e-06, + "loss": 0.7609, + "step": 3092 + }, + { + "epoch": 0.59, + "learning_rate": 7.714977075246179e-06, + "loss": 0.7931, + "step": 3093 + }, + { + "epoch": 0.59, + "learning_rate": 7.709000416744173e-06, + "loss": 0.8123, + "step": 3094 + }, + { + "epoch": 0.59, + "learning_rate": 7.703024621804933e-06, + "loss": 0.7678, + "step": 3095 + }, + { + "epoch": 0.59, + "learning_rate": 7.69704969268096e-06, + "loss": 0.7891, + "step": 3096 + }, + { + "epoch": 0.59, + "learning_rate": 7.69107563162442e-06, + "loss": 0.784, + "step": 3097 + }, + { + "epoch": 0.59, + "learning_rate": 7.685102440887165e-06, + "loss": 0.8015, + "step": 3098 + }, + { + "epoch": 0.59, + "learning_rate": 7.679130122720704e-06, + "loss": 0.8157, + "step": 3099 + }, + { + "epoch": 0.59, + "learning_rate": 7.673158679376233e-06, + "loss": 0.7937, + "step": 3100 + }, + { + "epoch": 0.59, + "learning_rate": 7.667188113104605e-06, + "loss": 0.7688, + "step": 3101 + }, + { + "epoch": 0.59, + "learning_rate": 7.661218426156353e-06, + "loss": 0.8785, + "step": 3102 + }, + { + "epoch": 0.59, + "learning_rate": 7.65524962078167e-06, + "loss": 0.8079, + "step": 3103 + }, + { + "epoch": 0.59, + "learning_rate": 7.64928169923042e-06, + "loss": 0.7622, + "step": 3104 + }, + { + "epoch": 0.59, + "learning_rate": 7.643314663752138e-06, + "loss": 0.8172, + "step": 3105 + }, + { + "epoch": 0.59, + "learning_rate": 7.637348516596016e-06, + "loss": 0.8086, + "step": 3106 + }, + { + "epoch": 0.59, + "learning_rate": 7.63138326001092e-06, + "loss": 0.8074, + "step": 3107 + }, + { + "epoch": 0.59, + "learning_rate": 7.625418896245378e-06, + "loss": 0.7742, + "step": 3108 + }, + { + "epoch": 0.59, + "learning_rate": 7.619455427547573e-06, + "loss": 0.8846, + "step": 3109 + }, + { + "epoch": 0.59, + "learning_rate": 7.613492856165366e-06, + "loss": 0.7845, + "step": 3110 + }, + { + "epoch": 0.59, + "learning_rate": 7.607531184346264e-06, + "loss": 0.7281, + "step": 3111 + }, + { + "epoch": 0.59, + "learning_rate": 7.601570414337449e-06, + "loss": 0.7532, + "step": 3112 + }, + { + "epoch": 0.59, + "learning_rate": 7.59561054838575e-06, + "loss": 0.8131, + "step": 3113 + }, + { + "epoch": 0.59, + "learning_rate": 7.58965158873767e-06, + "loss": 0.7748, + "step": 3114 + }, + { + "epoch": 0.59, + "learning_rate": 7.583693537639352e-06, + "loss": 0.8117, + "step": 3115 + }, + { + "epoch": 0.59, + "learning_rate": 7.577736397336614e-06, + "loss": 0.8437, + "step": 3116 + }, + { + "epoch": 0.59, + "learning_rate": 7.571780170074919e-06, + "loss": 0.7895, + "step": 3117 + }, + { + "epoch": 0.59, + "learning_rate": 7.565824858099397e-06, + "loss": 0.744, + "step": 3118 + }, + { + "epoch": 0.59, + "learning_rate": 7.559870463654814e-06, + "loss": 0.8257, + "step": 3119 + }, + { + "epoch": 0.59, + "learning_rate": 7.5539169889856135e-06, + "loss": 0.7674, + "step": 3120 + }, + { + "epoch": 0.59, + "learning_rate": 7.547964436335876e-06, + "loss": 0.7958, + "step": 3121 + }, + { + "epoch": 0.59, + "learning_rate": 7.5420128079493405e-06, + "loss": 0.7776, + "step": 3122 + }, + { + "epoch": 0.59, + "learning_rate": 7.536062106069394e-06, + "loss": 0.8029, + "step": 3123 + }, + { + "epoch": 0.59, + "learning_rate": 7.530112332939081e-06, + "loss": 0.8191, + "step": 3124 + }, + { + "epoch": 0.59, + "learning_rate": 7.524163490801091e-06, + "loss": 0.7956, + "step": 3125 + }, + { + "epoch": 0.59, + "learning_rate": 7.518215581897763e-06, + "loss": 0.8284, + "step": 3126 + }, + { + "epoch": 0.59, + "learning_rate": 7.512268608471083e-06, + "loss": 0.792, + "step": 3127 + }, + { + "epoch": 0.59, + "learning_rate": 7.506322572762685e-06, + "loss": 0.8371, + "step": 3128 + }, + { + "epoch": 0.59, + "learning_rate": 7.500377477013857e-06, + "loss": 0.7777, + "step": 3129 + }, + { + "epoch": 0.59, + "learning_rate": 7.494433323465518e-06, + "loss": 0.7313, + "step": 3130 + }, + { + "epoch": 0.59, + "learning_rate": 7.4884901143582465e-06, + "loss": 0.7348, + "step": 3131 + }, + { + "epoch": 0.59, + "learning_rate": 7.482547851932254e-06, + "loss": 0.826, + "step": 3132 + }, + { + "epoch": 0.59, + "learning_rate": 7.476606538427405e-06, + "loss": 0.7629, + "step": 3133 + }, + { + "epoch": 0.59, + "learning_rate": 7.470666176083193e-06, + "loss": 0.8148, + "step": 3134 + }, + { + "epoch": 0.59, + "learning_rate": 7.46472676713877e-06, + "loss": 0.2582, + "step": 3135 + }, + { + "epoch": 0.59, + "learning_rate": 7.458788313832912e-06, + "loss": 0.6847, + "step": 3136 + }, + { + "epoch": 0.59, + "learning_rate": 7.4528508184040495e-06, + "loss": 0.7677, + "step": 3137 + }, + { + "epoch": 0.59, + "learning_rate": 7.446914283090236e-06, + "loss": 0.7801, + "step": 3138 + }, + { + "epoch": 0.59, + "learning_rate": 7.440978710129183e-06, + "loss": 0.8109, + "step": 3139 + }, + { + "epoch": 0.6, + "learning_rate": 7.435044101758218e-06, + "loss": 0.8059, + "step": 3140 + }, + { + "epoch": 0.6, + "learning_rate": 7.42911046021432e-06, + "loss": 0.8152, + "step": 3141 + }, + { + "epoch": 0.6, + "learning_rate": 7.423177787734099e-06, + "loss": 0.7546, + "step": 3142 + }, + { + "epoch": 0.6, + "learning_rate": 7.4172460865538e-06, + "loss": 0.7801, + "step": 3143 + }, + { + "epoch": 0.6, + "learning_rate": 7.411315358909297e-06, + "loss": 0.8247, + "step": 3144 + }, + { + "epoch": 0.6, + "learning_rate": 7.405385607036108e-06, + "loss": 0.781, + "step": 3145 + }, + { + "epoch": 0.6, + "learning_rate": 7.399456833169368e-06, + "loss": 0.7936, + "step": 3146 + }, + { + "epoch": 0.6, + "learning_rate": 7.393529039543863e-06, + "loss": 0.8024, + "step": 3147 + }, + { + "epoch": 0.6, + "learning_rate": 7.387602228393987e-06, + "loss": 0.8579, + "step": 3148 + }, + { + "epoch": 0.6, + "learning_rate": 7.381676401953783e-06, + "loss": 0.7923, + "step": 3149 + }, + { + "epoch": 0.6, + "learning_rate": 7.37575156245691e-06, + "loss": 0.8389, + "step": 3150 + }, + { + "epoch": 0.6, + "learning_rate": 7.369827712136661e-06, + "loss": 0.7844, + "step": 3151 + }, + { + "epoch": 0.6, + "learning_rate": 7.363904853225961e-06, + "loss": 0.8022, + "step": 3152 + }, + { + "epoch": 0.6, + "learning_rate": 7.357982987957345e-06, + "loss": 0.7537, + "step": 3153 + }, + { + "epoch": 0.6, + "learning_rate": 7.352062118562995e-06, + "loss": 0.8153, + "step": 3154 + }, + { + "epoch": 0.6, + "learning_rate": 7.346142247274695e-06, + "loss": 0.7741, + "step": 3155 + }, + { + "epoch": 0.6, + "learning_rate": 7.340223376323874e-06, + "loss": 0.7655, + "step": 3156 + }, + { + "epoch": 0.6, + "learning_rate": 7.334305507941566e-06, + "loss": 0.7261, + "step": 3157 + }, + { + "epoch": 0.6, + "learning_rate": 7.328388644358443e-06, + "loss": 0.7847, + "step": 3158 + }, + { + "epoch": 0.6, + "learning_rate": 7.3224727878047815e-06, + "loss": 0.7704, + "step": 3159 + }, + { + "epoch": 0.6, + "learning_rate": 7.316557940510498e-06, + "loss": 0.7254, + "step": 3160 + }, + { + "epoch": 0.6, + "learning_rate": 7.310644104705108e-06, + "loss": 0.8033, + "step": 3161 + }, + { + "epoch": 0.6, + "learning_rate": 7.304731282617762e-06, + "loss": 0.8232, + "step": 3162 + }, + { + "epoch": 0.6, + "learning_rate": 7.29881947647722e-06, + "loss": 0.8238, + "step": 3163 + }, + { + "epoch": 0.6, + "learning_rate": 7.2929086885118614e-06, + "loss": 0.818, + "step": 3164 + }, + { + "epoch": 0.6, + "learning_rate": 7.286998920949679e-06, + "loss": 0.8354, + "step": 3165 + }, + { + "epoch": 0.6, + "learning_rate": 7.28109017601829e-06, + "loss": 0.7869, + "step": 3166 + }, + { + "epoch": 0.6, + "learning_rate": 7.275182455944912e-06, + "loss": 0.8246, + "step": 3167 + }, + { + "epoch": 0.6, + "learning_rate": 7.269275762956392e-06, + "loss": 0.7273, + "step": 3168 + }, + { + "epoch": 0.6, + "learning_rate": 7.263370099279173e-06, + "loss": 0.8052, + "step": 3169 + }, + { + "epoch": 0.6, + "learning_rate": 7.257465467139325e-06, + "loss": 0.7787, + "step": 3170 + }, + { + "epoch": 0.6, + "learning_rate": 7.251561868762519e-06, + "loss": 0.8655, + "step": 3171 + }, + { + "epoch": 0.6, + "learning_rate": 7.245659306374047e-06, + "loss": 0.7877, + "step": 3172 + }, + { + "epoch": 0.6, + "learning_rate": 7.239757782198793e-06, + "loss": 0.7768, + "step": 3173 + }, + { + "epoch": 0.6, + "learning_rate": 7.233857298461272e-06, + "loss": 0.8185, + "step": 3174 + }, + { + "epoch": 0.6, + "learning_rate": 7.2279578573855855e-06, + "loss": 0.7959, + "step": 3175 + }, + { + "epoch": 0.6, + "learning_rate": 7.2220594611954606e-06, + "loss": 0.7792, + "step": 3176 + }, + { + "epoch": 0.6, + "learning_rate": 7.216162112114213e-06, + "loss": 0.7873, + "step": 3177 + }, + { + "epoch": 0.6, + "learning_rate": 7.210265812364779e-06, + "loss": 0.7835, + "step": 3178 + }, + { + "epoch": 0.6, + "learning_rate": 7.2043705641696916e-06, + "loss": 0.804, + "step": 3179 + }, + { + "epoch": 0.6, + "learning_rate": 7.198476369751084e-06, + "loss": 0.7421, + "step": 3180 + }, + { + "epoch": 0.6, + "learning_rate": 7.192583231330703e-06, + "loss": 0.8061, + "step": 3181 + }, + { + "epoch": 0.6, + "learning_rate": 7.186691151129889e-06, + "loss": 0.8105, + "step": 3182 + }, + { + "epoch": 0.6, + "learning_rate": 7.1808001313695855e-06, + "loss": 0.6769, + "step": 3183 + }, + { + "epoch": 0.6, + "learning_rate": 7.174910174270335e-06, + "loss": 0.7996, + "step": 3184 + }, + { + "epoch": 0.6, + "learning_rate": 7.169021282052283e-06, + "loss": 0.7691, + "step": 3185 + }, + { + "epoch": 0.6, + "learning_rate": 7.163133456935169e-06, + "loss": 0.7916, + "step": 3186 + }, + { + "epoch": 0.6, + "learning_rate": 7.157246701138338e-06, + "loss": 0.7512, + "step": 3187 + }, + { + "epoch": 0.6, + "learning_rate": 7.151361016880719e-06, + "loss": 0.7275, + "step": 3188 + }, + { + "epoch": 0.6, + "learning_rate": 7.145476406380854e-06, + "loss": 0.7909, + "step": 3189 + }, + { + "epoch": 0.6, + "learning_rate": 7.1395928718568605e-06, + "loss": 0.8343, + "step": 3190 + }, + { + "epoch": 0.6, + "learning_rate": 7.133710415526471e-06, + "loss": 0.7482, + "step": 3191 + }, + { + "epoch": 0.61, + "learning_rate": 7.1278290396069946e-06, + "loss": 0.7713, + "step": 3192 + }, + { + "epoch": 0.61, + "learning_rate": 7.121948746315346e-06, + "loss": 0.8021, + "step": 3193 + }, + { + "epoch": 0.61, + "learning_rate": 7.116069537868018e-06, + "loss": 0.8596, + "step": 3194 + }, + { + "epoch": 0.61, + "learning_rate": 7.110191416481112e-06, + "loss": 0.7835, + "step": 3195 + }, + { + "epoch": 0.61, + "learning_rate": 7.104314384370301e-06, + "loss": 0.778, + "step": 3196 + }, + { + "epoch": 0.61, + "learning_rate": 7.098438443750865e-06, + "loss": 0.7583, + "step": 3197 + }, + { + "epoch": 0.61, + "learning_rate": 7.0925635968376565e-06, + "loss": 0.7288, + "step": 3198 + }, + { + "epoch": 0.61, + "learning_rate": 7.086689845845127e-06, + "loss": 0.8266, + "step": 3199 + }, + { + "epoch": 0.61, + "learning_rate": 7.0808171929873115e-06, + "loss": 0.7231, + "step": 3200 + }, + { + "epoch": 0.61, + "learning_rate": 7.07494564047783e-06, + "loss": 0.2635, + "step": 3201 + }, + { + "epoch": 0.61, + "learning_rate": 7.069075190529888e-06, + "loss": 0.7984, + "step": 3202 + }, + { + "epoch": 0.61, + "learning_rate": 7.06320584535628e-06, + "loss": 0.8296, + "step": 3203 + }, + { + "epoch": 0.61, + "learning_rate": 7.057337607169373e-06, + "loss": 0.8042, + "step": 3204 + }, + { + "epoch": 0.61, + "learning_rate": 7.05147047818113e-06, + "loss": 0.7633, + "step": 3205 + }, + { + "epoch": 0.61, + "learning_rate": 7.045604460603084e-06, + "loss": 0.738, + "step": 3206 + }, + { + "epoch": 0.61, + "learning_rate": 7.039739556646359e-06, + "loss": 0.8301, + "step": 3207 + }, + { + "epoch": 0.61, + "learning_rate": 7.033875768521656e-06, + "loss": 0.8382, + "step": 3208 + }, + { + "epoch": 0.61, + "learning_rate": 7.0280130984392494e-06, + "loss": 0.7376, + "step": 3209 + }, + { + "epoch": 0.61, + "learning_rate": 7.022151548609004e-06, + "loss": 0.8071, + "step": 3210 + }, + { + "epoch": 0.61, + "learning_rate": 7.016291121240346e-06, + "loss": 0.824, + "step": 3211 + }, + { + "epoch": 0.61, + "learning_rate": 7.010431818542298e-06, + "loss": 0.8155, + "step": 3212 + }, + { + "epoch": 0.61, + "learning_rate": 7.00457364272344e-06, + "loss": 0.7922, + "step": 3213 + }, + { + "epoch": 0.61, + "learning_rate": 6.998716595991943e-06, + "loss": 0.779, + "step": 3214 + }, + { + "epoch": 0.61, + "learning_rate": 6.99286068055554e-06, + "loss": 0.7626, + "step": 3215 + }, + { + "epoch": 0.61, + "learning_rate": 6.987005898621549e-06, + "loss": 0.7895, + "step": 3216 + }, + { + "epoch": 0.61, + "learning_rate": 6.981152252396844e-06, + "loss": 0.7889, + "step": 3217 + }, + { + "epoch": 0.61, + "learning_rate": 6.975299744087891e-06, + "loss": 0.8064, + "step": 3218 + }, + { + "epoch": 0.61, + "learning_rate": 6.969448375900715e-06, + "loss": 0.8035, + "step": 3219 + }, + { + "epoch": 0.61, + "learning_rate": 6.963598150040913e-06, + "loss": 0.7675, + "step": 3220 + }, + { + "epoch": 0.61, + "learning_rate": 6.957749068713653e-06, + "loss": 0.7893, + "step": 3221 + }, + { + "epoch": 0.61, + "learning_rate": 6.951901134123671e-06, + "loss": 0.7736, + "step": 3222 + }, + { + "epoch": 0.61, + "learning_rate": 6.946054348475266e-06, + "loss": 0.8127, + "step": 3223 + }, + { + "epoch": 0.61, + "learning_rate": 6.940208713972318e-06, + "loss": 0.2412, + "step": 3224 + }, + { + "epoch": 0.61, + "learning_rate": 6.934364232818254e-06, + "loss": 0.8246, + "step": 3225 + }, + { + "epoch": 0.61, + "learning_rate": 6.928520907216084e-06, + "loss": 0.8051, + "step": 3226 + }, + { + "epoch": 0.61, + "learning_rate": 6.922678739368367e-06, + "loss": 0.7576, + "step": 3227 + }, + { + "epoch": 0.61, + "learning_rate": 6.916837731477238e-06, + "loss": 0.7579, + "step": 3228 + }, + { + "epoch": 0.61, + "learning_rate": 6.910997885744386e-06, + "loss": 0.8364, + "step": 3229 + }, + { + "epoch": 0.61, + "learning_rate": 6.90515920437107e-06, + "loss": 0.8006, + "step": 3230 + }, + { + "epoch": 0.61, + "learning_rate": 6.8993216895580985e-06, + "loss": 0.7671, + "step": 3231 + }, + { + "epoch": 0.61, + "learning_rate": 6.8934853435058566e-06, + "loss": 0.8049, + "step": 3232 + }, + { + "epoch": 0.61, + "learning_rate": 6.887650168414272e-06, + "loss": 0.2594, + "step": 3233 + }, + { + "epoch": 0.61, + "learning_rate": 6.881816166482844e-06, + "loss": 0.7541, + "step": 3234 + }, + { + "epoch": 0.61, + "learning_rate": 6.875983339910619e-06, + "loss": 0.7832, + "step": 3235 + }, + { + "epoch": 0.61, + "learning_rate": 6.87015169089621e-06, + "loss": 0.8112, + "step": 3236 + }, + { + "epoch": 0.61, + "learning_rate": 6.864321221637784e-06, + "loss": 0.8093, + "step": 3237 + }, + { + "epoch": 0.61, + "learning_rate": 6.8584919343330544e-06, + "loss": 0.8178, + "step": 3238 + }, + { + "epoch": 0.61, + "learning_rate": 6.852663831179303e-06, + "loss": 0.8061, + "step": 3239 + }, + { + "epoch": 0.61, + "learning_rate": 6.846836914373353e-06, + "loss": 0.8352, + "step": 3240 + }, + { + "epoch": 0.61, + "learning_rate": 6.841011186111588e-06, + "loss": 0.8024, + "step": 3241 + }, + { + "epoch": 0.61, + "learning_rate": 6.835186648589939e-06, + "loss": 0.7644, + "step": 3242 + }, + { + "epoch": 0.61, + "learning_rate": 6.829363304003896e-06, + "loss": 0.7844, + "step": 3243 + }, + { + "epoch": 0.61, + "learning_rate": 6.823541154548485e-06, + "loss": 0.8444, + "step": 3244 + }, + { + "epoch": 0.62, + "learning_rate": 6.817720202418301e-06, + "loss": 0.7359, + "step": 3245 + }, + { + "epoch": 0.62, + "learning_rate": 6.811900449807465e-06, + "loss": 0.8269, + "step": 3246 + }, + { + "epoch": 0.62, + "learning_rate": 6.806081898909668e-06, + "loss": 0.7317, + "step": 3247 + }, + { + "epoch": 0.62, + "learning_rate": 6.8002645519181284e-06, + "loss": 0.791, + "step": 3248 + }, + { + "epoch": 0.62, + "learning_rate": 6.7944484110256304e-06, + "loss": 0.7797, + "step": 3249 + }, + { + "epoch": 0.62, + "learning_rate": 6.788633478424483e-06, + "loss": 0.8464, + "step": 3250 + }, + { + "epoch": 0.62, + "learning_rate": 6.7828197563065605e-06, + "loss": 0.8216, + "step": 3251 + }, + { + "epoch": 0.62, + "learning_rate": 6.7770072468632595e-06, + "loss": 0.764, + "step": 3252 + }, + { + "epoch": 0.62, + "learning_rate": 6.771195952285541e-06, + "loss": 0.7842, + "step": 3253 + }, + { + "epoch": 0.62, + "learning_rate": 6.765385874763888e-06, + "loss": 0.811, + "step": 3254 + }, + { + "epoch": 0.62, + "learning_rate": 6.759577016488343e-06, + "loss": 0.7861, + "step": 3255 + }, + { + "epoch": 0.62, + "learning_rate": 6.753769379648474e-06, + "loss": 0.8049, + "step": 3256 + }, + { + "epoch": 0.62, + "learning_rate": 6.747962966433399e-06, + "loss": 0.7955, + "step": 3257 + }, + { + "epoch": 0.62, + "learning_rate": 6.742157779031767e-06, + "loss": 0.7703, + "step": 3258 + }, + { + "epoch": 0.62, + "learning_rate": 6.736353819631771e-06, + "loss": 0.7864, + "step": 3259 + }, + { + "epoch": 0.62, + "learning_rate": 6.730551090421137e-06, + "loss": 0.7842, + "step": 3260 + }, + { + "epoch": 0.62, + "learning_rate": 6.724749593587134e-06, + "loss": 0.8339, + "step": 3261 + }, + { + "epoch": 0.62, + "learning_rate": 6.718949331316552e-06, + "loss": 0.8107, + "step": 3262 + }, + { + "epoch": 0.62, + "learning_rate": 6.713150305795732e-06, + "loss": 0.8258, + "step": 3263 + }, + { + "epoch": 0.62, + "learning_rate": 6.707352519210543e-06, + "loss": 0.7945, + "step": 3264 + }, + { + "epoch": 0.62, + "learning_rate": 6.701555973746381e-06, + "loss": 0.7798, + "step": 3265 + }, + { + "epoch": 0.62, + "learning_rate": 6.695760671588185e-06, + "loss": 0.8239, + "step": 3266 + }, + { + "epoch": 0.62, + "learning_rate": 6.689966614920414e-06, + "loss": 0.7324, + "step": 3267 + }, + { + "epoch": 0.62, + "learning_rate": 6.684173805927066e-06, + "loss": 0.2823, + "step": 3268 + }, + { + "epoch": 0.62, + "learning_rate": 6.678382246791661e-06, + "loss": 0.8583, + "step": 3269 + }, + { + "epoch": 0.62, + "learning_rate": 6.672591939697261e-06, + "loss": 0.7805, + "step": 3270 + }, + { + "epoch": 0.62, + "learning_rate": 6.66680288682644e-06, + "loss": 0.7821, + "step": 3271 + }, + { + "epoch": 0.62, + "learning_rate": 6.6610150903613135e-06, + "loss": 0.8156, + "step": 3272 + }, + { + "epoch": 0.62, + "learning_rate": 6.655228552483508e-06, + "loss": 0.822, + "step": 3273 + }, + { + "epoch": 0.62, + "learning_rate": 6.6494432753741935e-06, + "loss": 0.8357, + "step": 3274 + }, + { + "epoch": 0.62, + "learning_rate": 6.6436592612140486e-06, + "loss": 0.7943, + "step": 3275 + }, + { + "epoch": 0.62, + "learning_rate": 6.637876512183285e-06, + "loss": 0.7651, + "step": 3276 + }, + { + "epoch": 0.62, + "learning_rate": 6.632095030461639e-06, + "loss": 0.7629, + "step": 3277 + }, + { + "epoch": 0.62, + "learning_rate": 6.626314818228363e-06, + "loss": 0.79, + "step": 3278 + }, + { + "epoch": 0.62, + "learning_rate": 6.6205358776622315e-06, + "loss": 0.765, + "step": 3279 + }, + { + "epoch": 0.62, + "learning_rate": 6.614758210941545e-06, + "loss": 0.7104, + "step": 3280 + }, + { + "epoch": 0.62, + "learning_rate": 6.608981820244116e-06, + "loss": 0.8307, + "step": 3281 + }, + { + "epoch": 0.62, + "learning_rate": 6.6032067077472885e-06, + "loss": 0.8287, + "step": 3282 + }, + { + "epoch": 0.62, + "learning_rate": 6.597432875627907e-06, + "loss": 0.758, + "step": 3283 + }, + { + "epoch": 0.62, + "learning_rate": 6.591660326062352e-06, + "loss": 0.8137, + "step": 3284 + }, + { + "epoch": 0.62, + "learning_rate": 6.585889061226505e-06, + "loss": 0.2621, + "step": 3285 + }, + { + "epoch": 0.62, + "learning_rate": 6.580119083295775e-06, + "loss": 0.2544, + "step": 3286 + }, + { + "epoch": 0.62, + "learning_rate": 6.574350394445074e-06, + "loss": 0.7535, + "step": 3287 + }, + { + "epoch": 0.62, + "learning_rate": 6.568582996848844e-06, + "loss": 0.7468, + "step": 3288 + }, + { + "epoch": 0.62, + "learning_rate": 6.562816892681021e-06, + "loss": 0.8599, + "step": 3289 + }, + { + "epoch": 0.62, + "learning_rate": 6.557052084115074e-06, + "loss": 0.7661, + "step": 3290 + }, + { + "epoch": 0.62, + "learning_rate": 6.551288573323964e-06, + "loss": 0.7665, + "step": 3291 + }, + { + "epoch": 0.62, + "learning_rate": 6.545526362480177e-06, + "loss": 0.8075, + "step": 3292 + }, + { + "epoch": 0.62, + "learning_rate": 6.539765453755706e-06, + "loss": 0.755, + "step": 3293 + }, + { + "epoch": 0.62, + "learning_rate": 6.534005849322047e-06, + "loss": 0.7886, + "step": 3294 + }, + { + "epoch": 0.62, + "learning_rate": 6.528247551350213e-06, + "loss": 0.2671, + "step": 3295 + }, + { + "epoch": 0.62, + "learning_rate": 6.522490562010715e-06, + "loss": 0.7742, + "step": 3296 + }, + { + "epoch": 0.62, + "learning_rate": 6.516734883473579e-06, + "loss": 0.8269, + "step": 3297 + }, + { + "epoch": 0.63, + "learning_rate": 6.510980517908334e-06, + "loss": 0.8416, + "step": 3298 + }, + { + "epoch": 0.63, + "learning_rate": 6.505227467484012e-06, + "loss": 0.7675, + "step": 3299 + }, + { + "epoch": 0.63, + "learning_rate": 6.499475734369153e-06, + "loss": 0.8127, + "step": 3300 + }, + { + "epoch": 0.63, + "learning_rate": 6.493725320731801e-06, + "loss": 0.7826, + "step": 3301 + }, + { + "epoch": 0.63, + "learning_rate": 6.487976228739493e-06, + "loss": 0.705, + "step": 3302 + }, + { + "epoch": 0.63, + "learning_rate": 6.482228460559285e-06, + "loss": 0.7338, + "step": 3303 + }, + { + "epoch": 0.63, + "learning_rate": 6.476482018357713e-06, + "loss": 0.8172, + "step": 3304 + }, + { + "epoch": 0.63, + "learning_rate": 6.470736904300836e-06, + "loss": 0.7845, + "step": 3305 + }, + { + "epoch": 0.63, + "learning_rate": 6.4649931205541895e-06, + "loss": 0.8201, + "step": 3306 + }, + { + "epoch": 0.63, + "learning_rate": 6.459250669282828e-06, + "loss": 0.8113, + "step": 3307 + }, + { + "epoch": 0.63, + "learning_rate": 6.453509552651287e-06, + "loss": 0.8376, + "step": 3308 + }, + { + "epoch": 0.63, + "learning_rate": 6.4477697728236146e-06, + "loss": 0.7583, + "step": 3309 + }, + { + "epoch": 0.63, + "learning_rate": 6.442031331963338e-06, + "loss": 0.7851, + "step": 3310 + }, + { + "epoch": 0.63, + "learning_rate": 6.436294232233496e-06, + "loss": 0.8051, + "step": 3311 + }, + { + "epoch": 0.63, + "learning_rate": 6.430558475796607e-06, + "loss": 0.7362, + "step": 3312 + }, + { + "epoch": 0.63, + "learning_rate": 6.4248240648146965e-06, + "loss": 0.7434, + "step": 3313 + }, + { + "epoch": 0.63, + "learning_rate": 6.4190910014492755e-06, + "loss": 0.7929, + "step": 3314 + }, + { + "epoch": 0.63, + "learning_rate": 6.413359287861348e-06, + "loss": 0.8433, + "step": 3315 + }, + { + "epoch": 0.63, + "learning_rate": 6.407628926211409e-06, + "loss": 0.767, + "step": 3316 + }, + { + "epoch": 0.63, + "learning_rate": 6.401899918659444e-06, + "loss": 0.8035, + "step": 3317 + }, + { + "epoch": 0.63, + "learning_rate": 6.396172267364928e-06, + "loss": 0.7947, + "step": 3318 + }, + { + "epoch": 0.63, + "learning_rate": 6.39044597448683e-06, + "loss": 0.8276, + "step": 3319 + }, + { + "epoch": 0.63, + "learning_rate": 6.384721042183597e-06, + "loss": 0.7708, + "step": 3320 + }, + { + "epoch": 0.63, + "learning_rate": 6.378997472613168e-06, + "loss": 0.8116, + "step": 3321 + }, + { + "epoch": 0.63, + "learning_rate": 6.373275267932977e-06, + "loss": 0.8314, + "step": 3322 + }, + { + "epoch": 0.63, + "learning_rate": 6.367554430299924e-06, + "loss": 0.8247, + "step": 3323 + }, + { + "epoch": 0.63, + "learning_rate": 6.361834961870414e-06, + "loss": 0.7645, + "step": 3324 + }, + { + "epoch": 0.63, + "learning_rate": 6.356116864800321e-06, + "loss": 0.7726, + "step": 3325 + }, + { + "epoch": 0.63, + "learning_rate": 6.350400141245014e-06, + "loss": 0.7697, + "step": 3326 + }, + { + "epoch": 0.63, + "learning_rate": 6.344684793359331e-06, + "loss": 0.8051, + "step": 3327 + }, + { + "epoch": 0.63, + "learning_rate": 6.338970823297605e-06, + "loss": 0.7186, + "step": 3328 + }, + { + "epoch": 0.63, + "learning_rate": 6.333258233213637e-06, + "loss": 0.7947, + "step": 3329 + }, + { + "epoch": 0.63, + "learning_rate": 6.327547025260723e-06, + "loss": 0.7822, + "step": 3330 + }, + { + "epoch": 0.63, + "learning_rate": 6.321837201591619e-06, + "loss": 0.8102, + "step": 3331 + }, + { + "epoch": 0.63, + "learning_rate": 6.31612876435858e-06, + "loss": 0.824, + "step": 3332 + }, + { + "epoch": 0.63, + "learning_rate": 6.310421715713318e-06, + "loss": 0.776, + "step": 3333 + }, + { + "epoch": 0.63, + "learning_rate": 6.304716057807036e-06, + "loss": 0.8132, + "step": 3334 + }, + { + "epoch": 0.63, + "learning_rate": 6.29901179279041e-06, + "loss": 0.8007, + "step": 3335 + }, + { + "epoch": 0.63, + "learning_rate": 6.293308922813588e-06, + "loss": 0.824, + "step": 3336 + }, + { + "epoch": 0.63, + "learning_rate": 6.287607450026189e-06, + "loss": 0.7791, + "step": 3337 + }, + { + "epoch": 0.63, + "learning_rate": 6.281907376577316e-06, + "loss": 0.7923, + "step": 3338 + }, + { + "epoch": 0.63, + "learning_rate": 6.276208704615534e-06, + "loss": 0.7577, + "step": 3339 + }, + { + "epoch": 0.63, + "learning_rate": 6.270511436288889e-06, + "loss": 0.8186, + "step": 3340 + }, + { + "epoch": 0.63, + "learning_rate": 6.264815573744884e-06, + "loss": 0.7961, + "step": 3341 + }, + { + "epoch": 0.63, + "learning_rate": 6.259121119130511e-06, + "loss": 0.7463, + "step": 3342 + }, + { + "epoch": 0.63, + "learning_rate": 6.253428074592214e-06, + "loss": 0.715, + "step": 3343 + }, + { + "epoch": 0.63, + "learning_rate": 6.247736442275918e-06, + "loss": 0.7969, + "step": 3344 + }, + { + "epoch": 0.63, + "learning_rate": 6.242046224327003e-06, + "loss": 0.7674, + "step": 3345 + }, + { + "epoch": 0.63, + "learning_rate": 6.236357422890333e-06, + "loss": 0.2364, + "step": 3346 + }, + { + "epoch": 0.63, + "learning_rate": 6.2306700401102185e-06, + "loss": 0.8069, + "step": 3347 + }, + { + "epoch": 0.63, + "learning_rate": 6.224984078130451e-06, + "loss": 0.8002, + "step": 3348 + }, + { + "epoch": 0.63, + "learning_rate": 6.219299539094283e-06, + "loss": 0.813, + "step": 3349 + }, + { + "epoch": 0.63, + "learning_rate": 6.213616425144422e-06, + "loss": 0.831, + "step": 3350 + }, + { + "epoch": 0.64, + "learning_rate": 6.2079347384230505e-06, + "loss": 0.7826, + "step": 3351 + }, + { + "epoch": 0.64, + "learning_rate": 6.2022544810718e-06, + "loss": 0.7148, + "step": 3352 + }, + { + "epoch": 0.64, + "learning_rate": 6.196575655231779e-06, + "loss": 0.7508, + "step": 3353 + }, + { + "epoch": 0.64, + "learning_rate": 6.190898263043539e-06, + "loss": 0.7556, + "step": 3354 + }, + { + "epoch": 0.64, + "learning_rate": 6.185222306647105e-06, + "loss": 0.8129, + "step": 3355 + }, + { + "epoch": 0.64, + "learning_rate": 6.1795477881819545e-06, + "loss": 0.7293, + "step": 3356 + }, + { + "epoch": 0.64, + "learning_rate": 6.173874709787025e-06, + "loss": 0.8037, + "step": 3357 + }, + { + "epoch": 0.64, + "learning_rate": 6.168203073600706e-06, + "loss": 0.8128, + "step": 3358 + }, + { + "epoch": 0.64, + "learning_rate": 6.162532881760857e-06, + "loss": 0.7947, + "step": 3359 + }, + { + "epoch": 0.64, + "learning_rate": 6.156864136404771e-06, + "loss": 0.7785, + "step": 3360 + }, + { + "epoch": 0.64, + "learning_rate": 6.15119683966922e-06, + "loss": 0.8412, + "step": 3361 + }, + { + "epoch": 0.64, + "learning_rate": 6.145530993690411e-06, + "loss": 0.8151, + "step": 3362 + }, + { + "epoch": 0.64, + "learning_rate": 6.139866600604018e-06, + "loss": 0.8115, + "step": 3363 + }, + { + "epoch": 0.64, + "learning_rate": 6.1342036625451525e-06, + "loss": 0.8386, + "step": 3364 + }, + { + "epoch": 0.64, + "learning_rate": 6.128542181648395e-06, + "loss": 0.7503, + "step": 3365 + }, + { + "epoch": 0.64, + "learning_rate": 6.12288216004776e-06, + "loss": 0.7442, + "step": 3366 + }, + { + "epoch": 0.64, + "learning_rate": 6.117223599876726e-06, + "loss": 0.7891, + "step": 3367 + }, + { + "epoch": 0.64, + "learning_rate": 6.111566503268208e-06, + "loss": 0.2519, + "step": 3368 + }, + { + "epoch": 0.64, + "learning_rate": 6.105910872354583e-06, + "loss": 0.7686, + "step": 3369 + }, + { + "epoch": 0.64, + "learning_rate": 6.100256709267661e-06, + "loss": 0.8251, + "step": 3370 + }, + { + "epoch": 0.64, + "learning_rate": 6.094604016138712e-06, + "loss": 0.7541, + "step": 3371 + }, + { + "epoch": 0.64, + "learning_rate": 6.088952795098442e-06, + "loss": 0.8135, + "step": 3372 + }, + { + "epoch": 0.64, + "learning_rate": 6.083303048277008e-06, + "loss": 0.7808, + "step": 3373 + }, + { + "epoch": 0.64, + "learning_rate": 6.077654777804006e-06, + "loss": 0.7995, + "step": 3374 + }, + { + "epoch": 0.64, + "learning_rate": 6.072007985808481e-06, + "loss": 0.7944, + "step": 3375 + }, + { + "epoch": 0.64, + "learning_rate": 6.06636267441892e-06, + "loss": 0.767, + "step": 3376 + }, + { + "epoch": 0.64, + "learning_rate": 6.060718845763249e-06, + "loss": 0.8363, + "step": 3377 + }, + { + "epoch": 0.64, + "learning_rate": 6.0550765019688355e-06, + "loss": 0.7522, + "step": 3378 + }, + { + "epoch": 0.64, + "learning_rate": 6.049435645162487e-06, + "loss": 0.8286, + "step": 3379 + }, + { + "epoch": 0.64, + "learning_rate": 6.0437962774704595e-06, + "loss": 0.8027, + "step": 3380 + }, + { + "epoch": 0.64, + "learning_rate": 6.038158401018428e-06, + "loss": 0.812, + "step": 3381 + }, + { + "epoch": 0.64, + "learning_rate": 6.03252201793153e-06, + "loss": 0.7718, + "step": 3382 + }, + { + "epoch": 0.64, + "learning_rate": 6.026887130334316e-06, + "loss": 0.7933, + "step": 3383 + }, + { + "epoch": 0.64, + "learning_rate": 6.021253740350793e-06, + "loss": 0.7886, + "step": 3384 + }, + { + "epoch": 0.64, + "learning_rate": 6.015621850104387e-06, + "loss": 0.8039, + "step": 3385 + }, + { + "epoch": 0.64, + "learning_rate": 6.009991461717977e-06, + "loss": 0.8234, + "step": 3386 + }, + { + "epoch": 0.64, + "learning_rate": 6.004362577313852e-06, + "loss": 0.8204, + "step": 3387 + }, + { + "epoch": 0.64, + "learning_rate": 5.9987351990137585e-06, + "loss": 0.7601, + "step": 3388 + }, + { + "epoch": 0.64, + "learning_rate": 5.9931093289388575e-06, + "loss": 0.7266, + "step": 3389 + }, + { + "epoch": 0.64, + "learning_rate": 5.987484969209755e-06, + "loss": 0.7813, + "step": 3390 + }, + { + "epoch": 0.64, + "learning_rate": 5.981862121946471e-06, + "loss": 0.7929, + "step": 3391 + }, + { + "epoch": 0.64, + "learning_rate": 5.976240789268475e-06, + "loss": 0.836, + "step": 3392 + }, + { + "epoch": 0.64, + "learning_rate": 5.9706209732946495e-06, + "loss": 0.7752, + "step": 3393 + }, + { + "epoch": 0.64, + "learning_rate": 5.965002676143317e-06, + "loss": 0.7371, + "step": 3394 + }, + { + "epoch": 0.64, + "learning_rate": 5.9593858999322155e-06, + "loss": 0.7308, + "step": 3395 + }, + { + "epoch": 0.64, + "learning_rate": 5.953770646778524e-06, + "loss": 0.7678, + "step": 3396 + }, + { + "epoch": 0.64, + "learning_rate": 5.9481569187988315e-06, + "loss": 0.816, + "step": 3397 + }, + { + "epoch": 0.64, + "learning_rate": 5.942544718109171e-06, + "loss": 0.7391, + "step": 3398 + }, + { + "epoch": 0.64, + "learning_rate": 5.936934046824977e-06, + "loss": 0.8336, + "step": 3399 + }, + { + "epoch": 0.64, + "learning_rate": 5.931324907061131e-06, + "loss": 0.7525, + "step": 3400 + }, + { + "epoch": 0.64, + "learning_rate": 5.925717300931917e-06, + "loss": 0.7893, + "step": 3401 + }, + { + "epoch": 0.64, + "learning_rate": 5.920111230551056e-06, + "loss": 0.7762, + "step": 3402 + }, + { + "epoch": 0.64, + "learning_rate": 5.914506698031679e-06, + "loss": 0.7823, + "step": 3403 + }, + { + "epoch": 0.65, + "learning_rate": 5.908903705486344e-06, + "loss": 0.8153, + "step": 3404 + }, + { + "epoch": 0.65, + "learning_rate": 5.903302255027032e-06, + "loss": 0.7323, + "step": 3405 + }, + { + "epoch": 0.65, + "learning_rate": 5.89770234876513e-06, + "loss": 0.234, + "step": 3406 + }, + { + "epoch": 0.65, + "learning_rate": 5.892103988811457e-06, + "loss": 0.8064, + "step": 3407 + }, + { + "epoch": 0.65, + "learning_rate": 5.886507177276235e-06, + "loss": 0.7405, + "step": 3408 + }, + { + "epoch": 0.65, + "learning_rate": 5.88091191626912e-06, + "loss": 0.7912, + "step": 3409 + }, + { + "epoch": 0.65, + "learning_rate": 5.8753182078991636e-06, + "loss": 0.759, + "step": 3410 + }, + { + "epoch": 0.65, + "learning_rate": 5.869726054274849e-06, + "loss": 0.8264, + "step": 3411 + }, + { + "epoch": 0.65, + "learning_rate": 5.86413545750406e-06, + "loss": 0.8151, + "step": 3412 + }, + { + "epoch": 0.65, + "learning_rate": 5.858546419694107e-06, + "loss": 0.7616, + "step": 3413 + }, + { + "epoch": 0.65, + "learning_rate": 5.852958942951701e-06, + "loss": 0.8284, + "step": 3414 + }, + { + "epoch": 0.65, + "learning_rate": 5.847373029382972e-06, + "loss": 0.7682, + "step": 3415 + }, + { + "epoch": 0.65, + "learning_rate": 5.8417886810934545e-06, + "loss": 0.7716, + "step": 3416 + }, + { + "epoch": 0.65, + "learning_rate": 5.836205900188106e-06, + "loss": 0.8344, + "step": 3417 + }, + { + "epoch": 0.65, + "learning_rate": 5.830624688771267e-06, + "loss": 0.8554, + "step": 3418 + }, + { + "epoch": 0.65, + "learning_rate": 5.825045048946723e-06, + "loss": 0.8217, + "step": 3419 + }, + { + "epoch": 0.65, + "learning_rate": 5.819466982817627e-06, + "loss": 0.7457, + "step": 3420 + }, + { + "epoch": 0.65, + "learning_rate": 5.8138904924865766e-06, + "loss": 0.8435, + "step": 3421 + }, + { + "epoch": 0.65, + "learning_rate": 5.80831558005555e-06, + "loss": 0.7905, + "step": 3422 + }, + { + "epoch": 0.65, + "learning_rate": 5.802742247625939e-06, + "loss": 0.7624, + "step": 3423 + }, + { + "epoch": 0.65, + "learning_rate": 5.797170497298541e-06, + "loss": 0.7667, + "step": 3424 + }, + { + "epoch": 0.65, + "learning_rate": 5.791600331173552e-06, + "loss": 0.8008, + "step": 3425 + }, + { + "epoch": 0.65, + "learning_rate": 5.786031751350579e-06, + "loss": 0.7682, + "step": 3426 + }, + { + "epoch": 0.65, + "learning_rate": 5.780464759928623e-06, + "loss": 0.778, + "step": 3427 + }, + { + "epoch": 0.65, + "learning_rate": 5.774899359006092e-06, + "loss": 0.7833, + "step": 3428 + }, + { + "epoch": 0.65, + "learning_rate": 5.76933555068079e-06, + "loss": 0.7876, + "step": 3429 + }, + { + "epoch": 0.65, + "learning_rate": 5.763773337049925e-06, + "loss": 0.7542, + "step": 3430 + }, + { + "epoch": 0.65, + "learning_rate": 5.758212720210094e-06, + "loss": 0.8159, + "step": 3431 + }, + { + "epoch": 0.65, + "learning_rate": 5.752653702257318e-06, + "loss": 0.2887, + "step": 3432 + }, + { + "epoch": 0.65, + "learning_rate": 5.747096285286975e-06, + "loss": 0.8059, + "step": 3433 + }, + { + "epoch": 0.65, + "learning_rate": 5.7415404713938804e-06, + "loss": 0.7791, + "step": 3434 + }, + { + "epoch": 0.65, + "learning_rate": 5.735986262672211e-06, + "loss": 0.7627, + "step": 3435 + }, + { + "epoch": 0.65, + "learning_rate": 5.73043366121557e-06, + "loss": 0.7773, + "step": 3436 + }, + { + "epoch": 0.65, + "learning_rate": 5.7248826691169225e-06, + "loss": 0.749, + "step": 3437 + }, + { + "epoch": 0.65, + "learning_rate": 5.719333288468659e-06, + "loss": 0.8256, + "step": 3438 + }, + { + "epoch": 0.65, + "learning_rate": 5.7137855213625315e-06, + "loss": 0.7596, + "step": 3439 + }, + { + "epoch": 0.65, + "learning_rate": 5.7082393698897164e-06, + "loss": 0.7247, + "step": 3440 + }, + { + "epoch": 0.65, + "learning_rate": 5.7026948361407455e-06, + "loss": 0.7439, + "step": 3441 + }, + { + "epoch": 0.65, + "learning_rate": 5.697151922205575e-06, + "loss": 0.8048, + "step": 3442 + }, + { + "epoch": 0.65, + "learning_rate": 5.691610630173526e-06, + "loss": 0.8167, + "step": 3443 + }, + { + "epoch": 0.65, + "learning_rate": 5.686070962133321e-06, + "loss": 0.7229, + "step": 3444 + }, + { + "epoch": 0.65, + "learning_rate": 5.680532920173067e-06, + "loss": 0.8519, + "step": 3445 + }, + { + "epoch": 0.65, + "learning_rate": 5.674996506380252e-06, + "loss": 0.7336, + "step": 3446 + }, + { + "epoch": 0.65, + "learning_rate": 5.669461722841762e-06, + "loss": 0.8704, + "step": 3447 + }, + { + "epoch": 0.65, + "learning_rate": 5.663928571643861e-06, + "loss": 0.2788, + "step": 3448 + }, + { + "epoch": 0.65, + "learning_rate": 5.658397054872197e-06, + "loss": 0.234, + "step": 3449 + }, + { + "epoch": 0.65, + "learning_rate": 5.6528671746118045e-06, + "loss": 0.7869, + "step": 3450 + }, + { + "epoch": 0.65, + "learning_rate": 5.647338932947102e-06, + "loss": 0.769, + "step": 3451 + }, + { + "epoch": 0.65, + "learning_rate": 5.641812331961887e-06, + "loss": 0.7819, + "step": 3452 + }, + { + "epoch": 0.65, + "learning_rate": 5.636287373739341e-06, + "loss": 0.7355, + "step": 3453 + }, + { + "epoch": 0.65, + "learning_rate": 5.630764060362026e-06, + "loss": 0.7873, + "step": 3454 + }, + { + "epoch": 0.65, + "learning_rate": 5.625242393911885e-06, + "loss": 0.773, + "step": 3455 + }, + { + "epoch": 0.66, + "learning_rate": 5.619722376470238e-06, + "loss": 0.825, + "step": 3456 + }, + { + "epoch": 0.66, + "learning_rate": 5.614204010117785e-06, + "loss": 0.6905, + "step": 3457 + }, + { + "epoch": 0.66, + "learning_rate": 5.608687296934602e-06, + "loss": 0.7464, + "step": 3458 + }, + { + "epoch": 0.66, + "learning_rate": 5.603172239000143e-06, + "loss": 0.7731, + "step": 3459 + }, + { + "epoch": 0.66, + "learning_rate": 5.597658838393235e-06, + "loss": 0.8447, + "step": 3460 + }, + { + "epoch": 0.66, + "learning_rate": 5.592147097192091e-06, + "loss": 0.7951, + "step": 3461 + }, + { + "epoch": 0.66, + "learning_rate": 5.586637017474286e-06, + "loss": 0.7967, + "step": 3462 + }, + { + "epoch": 0.66, + "learning_rate": 5.581128601316774e-06, + "loss": 0.2801, + "step": 3463 + }, + { + "epoch": 0.66, + "learning_rate": 5.575621850795882e-06, + "loss": 0.7516, + "step": 3464 + }, + { + "epoch": 0.66, + "learning_rate": 5.570116767987309e-06, + "loss": 0.7253, + "step": 3465 + }, + { + "epoch": 0.66, + "learning_rate": 5.564613354966124e-06, + "loss": 0.7428, + "step": 3466 + }, + { + "epoch": 0.66, + "learning_rate": 5.559111613806767e-06, + "loss": 0.7763, + "step": 3467 + }, + { + "epoch": 0.66, + "learning_rate": 5.553611546583049e-06, + "loss": 0.7971, + "step": 3468 + }, + { + "epoch": 0.66, + "learning_rate": 5.548113155368151e-06, + "loss": 0.7861, + "step": 3469 + }, + { + "epoch": 0.66, + "learning_rate": 5.542616442234618e-06, + "loss": 0.7746, + "step": 3470 + }, + { + "epoch": 0.66, + "learning_rate": 5.5371214092543665e-06, + "loss": 0.7806, + "step": 3471 + }, + { + "epoch": 0.66, + "learning_rate": 5.53162805849868e-06, + "loss": 0.8092, + "step": 3472 + }, + { + "epoch": 0.66, + "learning_rate": 5.526136392038204e-06, + "loss": 0.7838, + "step": 3473 + }, + { + "epoch": 0.66, + "learning_rate": 5.520646411942952e-06, + "loss": 0.7941, + "step": 3474 + }, + { + "epoch": 0.66, + "learning_rate": 5.5151581202823015e-06, + "loss": 0.7534, + "step": 3475 + }, + { + "epoch": 0.66, + "learning_rate": 5.509671519124993e-06, + "loss": 0.8044, + "step": 3476 + }, + { + "epoch": 0.66, + "learning_rate": 5.504186610539131e-06, + "loss": 0.7869, + "step": 3477 + }, + { + "epoch": 0.66, + "learning_rate": 5.498703396592179e-06, + "loss": 0.7971, + "step": 3478 + }, + { + "epoch": 0.66, + "learning_rate": 5.493221879350965e-06, + "loss": 0.8104, + "step": 3479 + }, + { + "epoch": 0.66, + "learning_rate": 5.487742060881678e-06, + "loss": 0.7956, + "step": 3480 + }, + { + "epoch": 0.66, + "learning_rate": 5.482263943249861e-06, + "loss": 0.8253, + "step": 3481 + }, + { + "epoch": 0.66, + "learning_rate": 5.47678752852042e-06, + "loss": 0.7675, + "step": 3482 + }, + { + "epoch": 0.66, + "learning_rate": 5.47131281875763e-06, + "loss": 0.7343, + "step": 3483 + }, + { + "epoch": 0.66, + "learning_rate": 5.465839816025093e-06, + "loss": 0.7275, + "step": 3484 + }, + { + "epoch": 0.66, + "learning_rate": 5.460368522385806e-06, + "loss": 0.7133, + "step": 3485 + }, + { + "epoch": 0.66, + "learning_rate": 5.454898939902087e-06, + "loss": 0.8172, + "step": 3486 + }, + { + "epoch": 0.66, + "learning_rate": 5.449431070635636e-06, + "loss": 0.7316, + "step": 3487 + }, + { + "epoch": 0.66, + "learning_rate": 5.443964916647488e-06, + "loss": 0.8444, + "step": 3488 + }, + { + "epoch": 0.66, + "learning_rate": 5.438500479998046e-06, + "loss": 0.8347, + "step": 3489 + }, + { + "epoch": 0.66, + "learning_rate": 5.433037762747054e-06, + "loss": 0.7879, + "step": 3490 + }, + { + "epoch": 0.66, + "learning_rate": 5.427576766953615e-06, + "loss": 0.8113, + "step": 3491 + }, + { + "epoch": 0.66, + "learning_rate": 5.42211749467618e-06, + "loss": 0.7376, + "step": 3492 + }, + { + "epoch": 0.66, + "learning_rate": 5.416659947972554e-06, + "loss": 0.7896, + "step": 3493 + }, + { + "epoch": 0.66, + "learning_rate": 5.411204128899883e-06, + "loss": 0.7782, + "step": 3494 + }, + { + "epoch": 0.66, + "learning_rate": 5.405750039514675e-06, + "loss": 0.8094, + "step": 3495 + }, + { + "epoch": 0.66, + "learning_rate": 5.400297681872773e-06, + "loss": 0.8062, + "step": 3496 + }, + { + "epoch": 0.66, + "learning_rate": 5.394847058029374e-06, + "loss": 0.7846, + "step": 3497 + }, + { + "epoch": 0.66, + "learning_rate": 5.3893981700390215e-06, + "loss": 0.7626, + "step": 3498 + }, + { + "epoch": 0.66, + "learning_rate": 5.3839510199556055e-06, + "loss": 0.7428, + "step": 3499 + }, + { + "epoch": 0.66, + "learning_rate": 5.378505609832356e-06, + "loss": 0.8032, + "step": 3500 + }, + { + "epoch": 0.66, + "learning_rate": 5.373061941721851e-06, + "loss": 0.774, + "step": 3501 + }, + { + "epoch": 0.66, + "learning_rate": 5.367620017676012e-06, + "loss": 0.7641, + "step": 3502 + }, + { + "epoch": 0.66, + "learning_rate": 5.362179839746096e-06, + "loss": 0.7504, + "step": 3503 + }, + { + "epoch": 0.66, + "learning_rate": 5.356741409982722e-06, + "loss": 0.8417, + "step": 3504 + }, + { + "epoch": 0.66, + "learning_rate": 5.35130473043582e-06, + "loss": 0.7285, + "step": 3505 + }, + { + "epoch": 0.66, + "learning_rate": 5.345869803154691e-06, + "loss": 0.788, + "step": 3506 + }, + { + "epoch": 0.66, + "learning_rate": 5.340436630187944e-06, + "loss": 0.771, + "step": 3507 + }, + { + "epoch": 0.66, + "learning_rate": 5.3350052135835616e-06, + "loss": 0.7915, + "step": 3508 + }, + { + "epoch": 0.67, + "learning_rate": 5.329575555388831e-06, + "loss": 0.7984, + "step": 3509 + }, + { + "epoch": 0.67, + "learning_rate": 5.324147657650404e-06, + "loss": 0.7996, + "step": 3510 + }, + { + "epoch": 0.67, + "learning_rate": 5.318721522414245e-06, + "loss": 0.7545, + "step": 3511 + }, + { + "epoch": 0.67, + "learning_rate": 5.313297151725679e-06, + "loss": 0.7616, + "step": 3512 + }, + { + "epoch": 0.67, + "learning_rate": 5.307874547629339e-06, + "loss": 0.7849, + "step": 3513 + }, + { + "epoch": 0.67, + "learning_rate": 5.302453712169218e-06, + "loss": 0.8301, + "step": 3514 + }, + { + "epoch": 0.67, + "learning_rate": 5.297034647388617e-06, + "loss": 0.7604, + "step": 3515 + }, + { + "epoch": 0.67, + "learning_rate": 5.2916173553301944e-06, + "loss": 0.7222, + "step": 3516 + }, + { + "epoch": 0.67, + "learning_rate": 5.286201838035925e-06, + "loss": 0.744, + "step": 3517 + }, + { + "epoch": 0.67, + "learning_rate": 5.280788097547116e-06, + "loss": 0.8284, + "step": 3518 + }, + { + "epoch": 0.67, + "learning_rate": 5.275376135904408e-06, + "loss": 0.7711, + "step": 3519 + }, + { + "epoch": 0.67, + "learning_rate": 5.2699659551477715e-06, + "loss": 0.7842, + "step": 3520 + }, + { + "epoch": 0.67, + "learning_rate": 5.264557557316503e-06, + "loss": 0.7616, + "step": 3521 + }, + { + "epoch": 0.67, + "learning_rate": 5.259150944449225e-06, + "loss": 0.7833, + "step": 3522 + }, + { + "epoch": 0.67, + "learning_rate": 5.253746118583902e-06, + "loss": 0.7634, + "step": 3523 + }, + { + "epoch": 0.67, + "learning_rate": 5.248343081757798e-06, + "loss": 0.8186, + "step": 3524 + }, + { + "epoch": 0.67, + "learning_rate": 5.242941836007536e-06, + "loss": 0.2843, + "step": 3525 + }, + { + "epoch": 0.67, + "learning_rate": 5.2375423833690255e-06, + "loss": 0.7625, + "step": 3526 + }, + { + "epoch": 0.67, + "learning_rate": 5.23214472587754e-06, + "loss": 0.8078, + "step": 3527 + }, + { + "epoch": 0.67, + "learning_rate": 5.22674886556764e-06, + "loss": 0.7704, + "step": 3528 + }, + { + "epoch": 0.67, + "learning_rate": 5.221354804473242e-06, + "loss": 0.7334, + "step": 3529 + }, + { + "epoch": 0.67, + "learning_rate": 5.215962544627551e-06, + "loss": 0.7995, + "step": 3530 + }, + { + "epoch": 0.67, + "learning_rate": 5.2105720880631235e-06, + "loss": 0.7285, + "step": 3531 + }, + { + "epoch": 0.67, + "learning_rate": 5.205183436811813e-06, + "loss": 0.8214, + "step": 3532 + }, + { + "epoch": 0.67, + "learning_rate": 5.1997965929048125e-06, + "loss": 0.7281, + "step": 3533 + }, + { + "epoch": 0.67, + "learning_rate": 5.194411558372609e-06, + "loss": 0.7546, + "step": 3534 + }, + { + "epoch": 0.67, + "learning_rate": 5.189028335245039e-06, + "loss": 0.7215, + "step": 3535 + }, + { + "epoch": 0.67, + "learning_rate": 5.183646925551221e-06, + "loss": 0.8033, + "step": 3536 + }, + { + "epoch": 0.67, + "learning_rate": 5.178267331319622e-06, + "loss": 0.7365, + "step": 3537 + }, + { + "epoch": 0.67, + "learning_rate": 5.172889554578005e-06, + "loss": 0.8009, + "step": 3538 + }, + { + "epoch": 0.67, + "learning_rate": 5.167513597353455e-06, + "loss": 0.8096, + "step": 3539 + }, + { + "epoch": 0.67, + "learning_rate": 5.1621394616723705e-06, + "loss": 0.8239, + "step": 3540 + }, + { + "epoch": 0.67, + "learning_rate": 5.15676714956046e-06, + "loss": 0.806, + "step": 3541 + }, + { + "epoch": 0.67, + "learning_rate": 5.1513966630427495e-06, + "loss": 0.7728, + "step": 3542 + }, + { + "epoch": 0.67, + "learning_rate": 5.14602800414357e-06, + "loss": 0.7728, + "step": 3543 + }, + { + "epoch": 0.67, + "learning_rate": 5.140661174886581e-06, + "loss": 0.727, + "step": 3544 + }, + { + "epoch": 0.67, + "learning_rate": 5.1352961772947215e-06, + "loss": 0.7759, + "step": 3545 + }, + { + "epoch": 0.67, + "learning_rate": 5.1299330133902755e-06, + "loss": 0.7878, + "step": 3546 + }, + { + "epoch": 0.67, + "learning_rate": 5.124571685194804e-06, + "loss": 0.7564, + "step": 3547 + }, + { + "epoch": 0.67, + "learning_rate": 5.119212194729204e-06, + "loss": 0.7494, + "step": 3548 + }, + { + "epoch": 0.67, + "learning_rate": 5.113854544013652e-06, + "loss": 0.7659, + "step": 3549 + }, + { + "epoch": 0.67, + "learning_rate": 5.10849873506766e-06, + "loss": 0.7399, + "step": 3550 + }, + { + "epoch": 0.67, + "learning_rate": 5.103144769910017e-06, + "loss": 0.744, + "step": 3551 + }, + { + "epoch": 0.67, + "learning_rate": 5.097792650558846e-06, + "loss": 0.7586, + "step": 3552 + }, + { + "epoch": 0.67, + "learning_rate": 5.092442379031544e-06, + "loss": 0.7886, + "step": 3553 + }, + { + "epoch": 0.67, + "learning_rate": 5.087093957344841e-06, + "loss": 0.7615, + "step": 3554 + }, + { + "epoch": 0.67, + "learning_rate": 5.081747387514742e-06, + "loss": 0.7924, + "step": 3555 + }, + { + "epoch": 0.67, + "learning_rate": 5.076402671556578e-06, + "loss": 0.7055, + "step": 3556 + }, + { + "epoch": 0.67, + "learning_rate": 5.071059811484968e-06, + "loss": 0.8011, + "step": 3557 + }, + { + "epoch": 0.67, + "learning_rate": 5.065718809313835e-06, + "loss": 0.3009, + "step": 3558 + }, + { + "epoch": 0.67, + "learning_rate": 5.060379667056399e-06, + "loss": 0.8181, + "step": 3559 + }, + { + "epoch": 0.67, + "learning_rate": 5.05504238672518e-06, + "loss": 0.8058, + "step": 3560 + }, + { + "epoch": 0.67, + "learning_rate": 5.049706970332e-06, + "loss": 0.7868, + "step": 3561 + }, + { + "epoch": 0.68, + "learning_rate": 5.044373419887973e-06, + "loss": 0.7869, + "step": 3562 + }, + { + "epoch": 0.68, + "learning_rate": 5.039041737403514e-06, + "loss": 0.7504, + "step": 3563 + }, + { + "epoch": 0.68, + "learning_rate": 5.03371192488833e-06, + "loss": 0.7953, + "step": 3564 + }, + { + "epoch": 0.68, + "learning_rate": 5.028383984351427e-06, + "loss": 0.8135, + "step": 3565 + }, + { + "epoch": 0.68, + "learning_rate": 5.023057917801102e-06, + "loss": 0.8104, + "step": 3566 + }, + { + "epoch": 0.68, + "learning_rate": 5.017733727244949e-06, + "loss": 0.7955, + "step": 3567 + }, + { + "epoch": 0.68, + "learning_rate": 5.0124114146898505e-06, + "loss": 0.7747, + "step": 3568 + }, + { + "epoch": 0.68, + "learning_rate": 5.007090982141987e-06, + "loss": 0.8319, + "step": 3569 + }, + { + "epoch": 0.68, + "learning_rate": 5.001772431606824e-06, + "loss": 0.8554, + "step": 3570 + }, + { + "epoch": 0.68, + "learning_rate": 4.996455765089121e-06, + "loss": 0.7538, + "step": 3571 + }, + { + "epoch": 0.68, + "learning_rate": 4.991140984592924e-06, + "loss": 0.257, + "step": 3572 + }, + { + "epoch": 0.68, + "learning_rate": 4.985828092121582e-06, + "loss": 0.8017, + "step": 3573 + }, + { + "epoch": 0.68, + "learning_rate": 4.980517089677706e-06, + "loss": 0.8407, + "step": 3574 + }, + { + "epoch": 0.68, + "learning_rate": 4.9752079792632244e-06, + "loss": 0.7183, + "step": 3575 + }, + { + "epoch": 0.68, + "learning_rate": 4.969900762879325e-06, + "loss": 0.7665, + "step": 3576 + }, + { + "epoch": 0.68, + "learning_rate": 4.964595442526506e-06, + "loss": 0.7718, + "step": 3577 + }, + { + "epoch": 0.68, + "learning_rate": 4.959292020204532e-06, + "loss": 0.7977, + "step": 3578 + }, + { + "epoch": 0.68, + "learning_rate": 4.953990497912463e-06, + "loss": 0.7119, + "step": 3579 + }, + { + "epoch": 0.68, + "learning_rate": 4.948690877648638e-06, + "loss": 0.7799, + "step": 3580 + }, + { + "epoch": 0.68, + "learning_rate": 4.9433931614106814e-06, + "loss": 0.7911, + "step": 3581 + }, + { + "epoch": 0.68, + "learning_rate": 4.938097351195499e-06, + "loss": 0.799, + "step": 3582 + }, + { + "epoch": 0.68, + "learning_rate": 4.932803448999278e-06, + "loss": 0.758, + "step": 3583 + }, + { + "epoch": 0.68, + "learning_rate": 4.927511456817487e-06, + "loss": 0.7013, + "step": 3584 + }, + { + "epoch": 0.68, + "learning_rate": 4.922221376644874e-06, + "loss": 0.7489, + "step": 3585 + }, + { + "epoch": 0.68, + "learning_rate": 4.916933210475465e-06, + "loss": 0.768, + "step": 3586 + }, + { + "epoch": 0.68, + "learning_rate": 4.9116469603025684e-06, + "loss": 0.7695, + "step": 3587 + }, + { + "epoch": 0.68, + "learning_rate": 4.906362628118767e-06, + "loss": 0.8117, + "step": 3588 + }, + { + "epoch": 0.68, + "learning_rate": 4.9010802159159224e-06, + "loss": 0.7257, + "step": 3589 + }, + { + "epoch": 0.68, + "learning_rate": 4.8957997256851696e-06, + "loss": 0.7914, + "step": 3590 + }, + { + "epoch": 0.68, + "learning_rate": 4.890521159416924e-06, + "loss": 0.7541, + "step": 3591 + }, + { + "epoch": 0.68, + "learning_rate": 4.885244519100873e-06, + "loss": 0.8067, + "step": 3592 + }, + { + "epoch": 0.68, + "learning_rate": 4.879969806725976e-06, + "loss": 0.8115, + "step": 3593 + }, + { + "epoch": 0.68, + "learning_rate": 4.874697024280469e-06, + "loss": 0.7477, + "step": 3594 + }, + { + "epoch": 0.68, + "learning_rate": 4.86942617375186e-06, + "loss": 0.7833, + "step": 3595 + }, + { + "epoch": 0.68, + "learning_rate": 4.864157257126928e-06, + "loss": 0.8056, + "step": 3596 + }, + { + "epoch": 0.68, + "learning_rate": 4.858890276391722e-06, + "loss": 0.8287, + "step": 3597 + }, + { + "epoch": 0.68, + "learning_rate": 4.8536252335315606e-06, + "loss": 0.797, + "step": 3598 + }, + { + "epoch": 0.68, + "learning_rate": 4.848362130531039e-06, + "loss": 0.7201, + "step": 3599 + }, + { + "epoch": 0.68, + "learning_rate": 4.843100969374015e-06, + "loss": 0.8604, + "step": 3600 + }, + { + "epoch": 0.68, + "learning_rate": 4.837841752043612e-06, + "loss": 0.2599, + "step": 3601 + }, + { + "epoch": 0.68, + "learning_rate": 4.8325844805222286e-06, + "loss": 0.7798, + "step": 3602 + }, + { + "epoch": 0.68, + "learning_rate": 4.8273291567915225e-06, + "loss": 0.7718, + "step": 3603 + }, + { + "epoch": 0.68, + "learning_rate": 4.822075782832421e-06, + "loss": 0.8218, + "step": 3604 + }, + { + "epoch": 0.68, + "learning_rate": 4.816824360625115e-06, + "loss": 0.7473, + "step": 3605 + }, + { + "epoch": 0.68, + "learning_rate": 4.811574892149062e-06, + "loss": 0.8044, + "step": 3606 + }, + { + "epoch": 0.68, + "learning_rate": 4.806327379382981e-06, + "loss": 0.8031, + "step": 3607 + }, + { + "epoch": 0.68, + "learning_rate": 4.801081824304853e-06, + "loss": 0.8156, + "step": 3608 + }, + { + "epoch": 0.68, + "learning_rate": 4.795838228891923e-06, + "loss": 0.7453, + "step": 3609 + }, + { + "epoch": 0.68, + "learning_rate": 4.790596595120699e-06, + "loss": 0.7699, + "step": 3610 + }, + { + "epoch": 0.68, + "learning_rate": 4.785356924966943e-06, + "loss": 0.8408, + "step": 3611 + }, + { + "epoch": 0.68, + "learning_rate": 4.780119220405685e-06, + "loss": 0.7639, + "step": 3612 + }, + { + "epoch": 0.68, + "learning_rate": 4.7748834834112075e-06, + "loss": 0.7893, + "step": 3613 + }, + { + "epoch": 0.68, + "learning_rate": 4.769649715957055e-06, + "loss": 0.7551, + "step": 3614 + }, + { + "epoch": 0.69, + "learning_rate": 4.76441792001603e-06, + "loss": 0.7802, + "step": 3615 + }, + { + "epoch": 0.69, + "learning_rate": 4.759188097560189e-06, + "loss": 0.7773, + "step": 3616 + }, + { + "epoch": 0.69, + "learning_rate": 4.753960250560843e-06, + "loss": 0.7657, + "step": 3617 + }, + { + "epoch": 0.69, + "learning_rate": 4.748734380988572e-06, + "loss": 0.7458, + "step": 3618 + }, + { + "epoch": 0.69, + "learning_rate": 4.7435104908131865e-06, + "loss": 0.8039, + "step": 3619 + }, + { + "epoch": 0.69, + "learning_rate": 4.73828858200378e-06, + "loss": 0.2779, + "step": 3620 + }, + { + "epoch": 0.69, + "learning_rate": 4.733068656528666e-06, + "loss": 0.7317, + "step": 3621 + }, + { + "epoch": 0.69, + "learning_rate": 4.727850716355444e-06, + "loss": 0.7237, + "step": 3622 + }, + { + "epoch": 0.69, + "learning_rate": 4.722634763450937e-06, + "loss": 0.7287, + "step": 3623 + }, + { + "epoch": 0.69, + "learning_rate": 4.7174207997812436e-06, + "loss": 0.787, + "step": 3624 + }, + { + "epoch": 0.69, + "learning_rate": 4.7122088273116864e-06, + "loss": 0.7506, + "step": 3625 + }, + { + "epoch": 0.69, + "learning_rate": 4.706998848006864e-06, + "loss": 0.802, + "step": 3626 + }, + { + "epoch": 0.69, + "learning_rate": 4.7017908638306e-06, + "loss": 0.8087, + "step": 3627 + }, + { + "epoch": 0.69, + "learning_rate": 4.696584876745984e-06, + "loss": 0.772, + "step": 3628 + }, + { + "epoch": 0.69, + "learning_rate": 4.6913808887153445e-06, + "loss": 0.8108, + "step": 3629 + }, + { + "epoch": 0.69, + "learning_rate": 4.686178901700257e-06, + "loss": 0.8215, + "step": 3630 + }, + { + "epoch": 0.69, + "learning_rate": 4.680978917661544e-06, + "loss": 0.7285, + "step": 3631 + }, + { + "epoch": 0.69, + "learning_rate": 4.675780938559272e-06, + "loss": 0.7002, + "step": 3632 + }, + { + "epoch": 0.69, + "learning_rate": 4.67058496635275e-06, + "loss": 0.7724, + "step": 3633 + }, + { + "epoch": 0.69, + "learning_rate": 4.665391003000537e-06, + "loss": 0.7418, + "step": 3634 + }, + { + "epoch": 0.69, + "learning_rate": 4.660199050460428e-06, + "loss": 0.797, + "step": 3635 + }, + { + "epoch": 0.69, + "learning_rate": 4.655009110689463e-06, + "loss": 0.2702, + "step": 3636 + }, + { + "epoch": 0.69, + "learning_rate": 4.649821185643924e-06, + "loss": 0.781, + "step": 3637 + }, + { + "epoch": 0.69, + "learning_rate": 4.6446352772793256e-06, + "loss": 0.7529, + "step": 3638 + }, + { + "epoch": 0.69, + "learning_rate": 4.6394513875504445e-06, + "loss": 0.8047, + "step": 3639 + }, + { + "epoch": 0.69, + "learning_rate": 4.634269518411263e-06, + "loss": 0.7637, + "step": 3640 + }, + { + "epoch": 0.69, + "learning_rate": 4.629089671815036e-06, + "loss": 0.7948, + "step": 3641 + }, + { + "epoch": 0.69, + "learning_rate": 4.623911849714226e-06, + "loss": 0.7527, + "step": 3642 + }, + { + "epoch": 0.69, + "learning_rate": 4.618736054060559e-06, + "loss": 0.8087, + "step": 3643 + }, + { + "epoch": 0.69, + "learning_rate": 4.61356228680497e-06, + "loss": 0.8604, + "step": 3644 + }, + { + "epoch": 0.69, + "learning_rate": 4.608390549897661e-06, + "loss": 0.7178, + "step": 3645 + }, + { + "epoch": 0.69, + "learning_rate": 4.603220845288035e-06, + "loss": 0.7712, + "step": 3646 + }, + { + "epoch": 0.69, + "learning_rate": 4.598053174924761e-06, + "loss": 0.8014, + "step": 3647 + }, + { + "epoch": 0.69, + "learning_rate": 4.592887540755708e-06, + "loss": 0.7823, + "step": 3648 + }, + { + "epoch": 0.69, + "learning_rate": 4.587723944728014e-06, + "loss": 0.7155, + "step": 3649 + }, + { + "epoch": 0.69, + "learning_rate": 4.5825623887880146e-06, + "loss": 0.8023, + "step": 3650 + }, + { + "epoch": 0.69, + "learning_rate": 4.5774028748813e-06, + "loss": 0.7622, + "step": 3651 + }, + { + "epoch": 0.69, + "learning_rate": 4.5722454049526825e-06, + "loss": 0.7589, + "step": 3652 + }, + { + "epoch": 0.69, + "learning_rate": 4.567089980946201e-06, + "loss": 0.2544, + "step": 3653 + }, + { + "epoch": 0.69, + "learning_rate": 4.561936604805128e-06, + "loss": 0.7775, + "step": 3654 + }, + { + "epoch": 0.69, + "learning_rate": 4.556785278471961e-06, + "loss": 0.8058, + "step": 3655 + }, + { + "epoch": 0.69, + "learning_rate": 4.551636003888427e-06, + "loss": 0.8252, + "step": 3656 + }, + { + "epoch": 0.69, + "learning_rate": 4.546488782995479e-06, + "loss": 0.792, + "step": 3657 + }, + { + "epoch": 0.69, + "learning_rate": 4.541343617733293e-06, + "loss": 0.7081, + "step": 3658 + }, + { + "epoch": 0.69, + "learning_rate": 4.536200510041271e-06, + "loss": 0.7647, + "step": 3659 + }, + { + "epoch": 0.69, + "learning_rate": 4.531059461858053e-06, + "loss": 0.7727, + "step": 3660 + }, + { + "epoch": 0.69, + "learning_rate": 4.525920475121475e-06, + "loss": 0.7925, + "step": 3661 + }, + { + "epoch": 0.69, + "learning_rate": 4.520783551768626e-06, + "loss": 0.8188, + "step": 3662 + }, + { + "epoch": 0.69, + "learning_rate": 4.515648693735789e-06, + "loss": 0.7675, + "step": 3663 + }, + { + "epoch": 0.69, + "learning_rate": 4.510515902958499e-06, + "loss": 0.8085, + "step": 3664 + }, + { + "epoch": 0.69, + "learning_rate": 4.505385181371477e-06, + "loss": 0.7711, + "step": 3665 + }, + { + "epoch": 0.69, + "learning_rate": 4.5002565309087e-06, + "loss": 0.7249, + "step": 3666 + }, + { + "epoch": 0.7, + "learning_rate": 4.495129953503331e-06, + "loss": 0.7902, + "step": 3667 + }, + { + "epoch": 0.7, + "learning_rate": 4.490005451087781e-06, + "loss": 0.8355, + "step": 3668 + }, + { + "epoch": 0.7, + "learning_rate": 4.484883025593653e-06, + "loss": 0.8287, + "step": 3669 + }, + { + "epoch": 0.7, + "learning_rate": 4.479762678951791e-06, + "loss": 0.804, + "step": 3670 + }, + { + "epoch": 0.7, + "learning_rate": 4.47464441309223e-06, + "loss": 0.7874, + "step": 3671 + }, + { + "epoch": 0.7, + "learning_rate": 4.469528229944247e-06, + "loss": 0.7831, + "step": 3672 + }, + { + "epoch": 0.7, + "learning_rate": 4.4644141314363165e-06, + "loss": 0.7737, + "step": 3673 + }, + { + "epoch": 0.7, + "learning_rate": 4.4593021194961305e-06, + "loss": 0.7724, + "step": 3674 + }, + { + "epoch": 0.7, + "learning_rate": 4.454192196050599e-06, + "loss": 0.768, + "step": 3675 + }, + { + "epoch": 0.7, + "learning_rate": 4.449084363025838e-06, + "loss": 0.7383, + "step": 3676 + }, + { + "epoch": 0.7, + "learning_rate": 4.4439786223471834e-06, + "loss": 0.8031, + "step": 3677 + }, + { + "epoch": 0.7, + "learning_rate": 4.438874975939176e-06, + "loss": 0.7789, + "step": 3678 + }, + { + "epoch": 0.7, + "learning_rate": 4.4337734257255695e-06, + "loss": 0.814, + "step": 3679 + }, + { + "epoch": 0.7, + "learning_rate": 4.4286739736293285e-06, + "loss": 0.7617, + "step": 3680 + }, + { + "epoch": 0.7, + "learning_rate": 4.423576621572624e-06, + "loss": 0.7285, + "step": 3681 + }, + { + "epoch": 0.7, + "learning_rate": 4.418481371476839e-06, + "loss": 0.8316, + "step": 3682 + }, + { + "epoch": 0.7, + "learning_rate": 4.4133882252625615e-06, + "loss": 0.8194, + "step": 3683 + }, + { + "epoch": 0.7, + "learning_rate": 4.408297184849585e-06, + "loss": 0.7463, + "step": 3684 + }, + { + "epoch": 0.7, + "learning_rate": 4.403208252156921e-06, + "loss": 0.8062, + "step": 3685 + }, + { + "epoch": 0.7, + "learning_rate": 4.398121429102762e-06, + "loss": 0.8261, + "step": 3686 + }, + { + "epoch": 0.7, + "learning_rate": 4.393036717604536e-06, + "loss": 0.8056, + "step": 3687 + }, + { + "epoch": 0.7, + "learning_rate": 4.387954119578846e-06, + "loss": 0.8162, + "step": 3688 + }, + { + "epoch": 0.7, + "learning_rate": 4.382873636941523e-06, + "loss": 0.8179, + "step": 3689 + }, + { + "epoch": 0.7, + "learning_rate": 4.3777952716075786e-06, + "loss": 0.8363, + "step": 3690 + }, + { + "epoch": 0.7, + "learning_rate": 4.37271902549125e-06, + "loss": 0.8155, + "step": 3691 + }, + { + "epoch": 0.7, + "learning_rate": 4.367644900505947e-06, + "loss": 0.7807, + "step": 3692 + }, + { + "epoch": 0.7, + "learning_rate": 4.36257289856431e-06, + "loss": 0.7807, + "step": 3693 + }, + { + "epoch": 0.7, + "learning_rate": 4.357503021578158e-06, + "loss": 0.7986, + "step": 3694 + }, + { + "epoch": 0.7, + "learning_rate": 4.3524352714585165e-06, + "loss": 0.7149, + "step": 3695 + }, + { + "epoch": 0.7, + "learning_rate": 4.347369650115608e-06, + "loss": 0.7329, + "step": 3696 + }, + { + "epoch": 0.7, + "learning_rate": 4.3423061594588554e-06, + "loss": 0.7443, + "step": 3697 + }, + { + "epoch": 0.7, + "learning_rate": 4.337244801396875e-06, + "loss": 0.82, + "step": 3698 + }, + { + "epoch": 0.7, + "learning_rate": 4.332185577837479e-06, + "loss": 0.8155, + "step": 3699 + }, + { + "epoch": 0.7, + "learning_rate": 4.327128490687678e-06, + "loss": 0.7628, + "step": 3700 + }, + { + "epoch": 0.7, + "learning_rate": 4.322073541853677e-06, + "loss": 0.771, + "step": 3701 + }, + { + "epoch": 0.7, + "learning_rate": 4.317020733240872e-06, + "loss": 0.8333, + "step": 3702 + }, + { + "epoch": 0.7, + "learning_rate": 4.311970066753854e-06, + "loss": 0.7857, + "step": 3703 + }, + { + "epoch": 0.7, + "learning_rate": 4.306921544296409e-06, + "loss": 0.8063, + "step": 3704 + }, + { + "epoch": 0.7, + "learning_rate": 4.301875167771509e-06, + "loss": 0.7466, + "step": 3705 + }, + { + "epoch": 0.7, + "learning_rate": 4.2968309390813226e-06, + "loss": 0.7009, + "step": 3706 + }, + { + "epoch": 0.7, + "learning_rate": 4.2917888601272065e-06, + "loss": 0.778, + "step": 3707 + }, + { + "epoch": 0.7, + "learning_rate": 4.286748932809707e-06, + "loss": 0.7761, + "step": 3708 + }, + { + "epoch": 0.7, + "learning_rate": 4.28171115902856e-06, + "loss": 0.7798, + "step": 3709 + }, + { + "epoch": 0.7, + "learning_rate": 4.27667554068269e-06, + "loss": 0.7529, + "step": 3710 + }, + { + "epoch": 0.7, + "learning_rate": 4.271642079670204e-06, + "loss": 0.7977, + "step": 3711 + }, + { + "epoch": 0.7, + "learning_rate": 4.266610777888407e-06, + "loss": 0.8165, + "step": 3712 + }, + { + "epoch": 0.7, + "learning_rate": 4.261581637233781e-06, + "loss": 0.7923, + "step": 3713 + }, + { + "epoch": 0.7, + "learning_rate": 4.2565546596019965e-06, + "loss": 0.7779, + "step": 3714 + }, + { + "epoch": 0.7, + "learning_rate": 4.2515298468879064e-06, + "loss": 0.7834, + "step": 3715 + }, + { + "epoch": 0.7, + "learning_rate": 4.246507200985548e-06, + "loss": 0.7722, + "step": 3716 + }, + { + "epoch": 0.7, + "learning_rate": 4.241486723788145e-06, + "loss": 0.7616, + "step": 3717 + }, + { + "epoch": 0.7, + "learning_rate": 4.236468417188101e-06, + "loss": 0.7899, + "step": 3718 + }, + { + "epoch": 0.7, + "learning_rate": 4.231452283077002e-06, + "loss": 0.7738, + "step": 3719 + }, + { + "epoch": 0.71, + "learning_rate": 4.226438323345615e-06, + "loss": 0.8535, + "step": 3720 + }, + { + "epoch": 0.71, + "learning_rate": 4.221426539883887e-06, + "loss": 0.7626, + "step": 3721 + }, + { + "epoch": 0.71, + "learning_rate": 4.216416934580947e-06, + "loss": 0.7715, + "step": 3722 + }, + { + "epoch": 0.71, + "learning_rate": 4.2114095093250986e-06, + "loss": 0.7434, + "step": 3723 + }, + { + "epoch": 0.71, + "learning_rate": 4.206404266003827e-06, + "loss": 0.8226, + "step": 3724 + }, + { + "epoch": 0.71, + "learning_rate": 4.201401206503795e-06, + "loss": 0.2291, + "step": 3725 + }, + { + "epoch": 0.71, + "learning_rate": 4.196400332710841e-06, + "loss": 0.7383, + "step": 3726 + }, + { + "epoch": 0.71, + "learning_rate": 4.19140164650998e-06, + "loss": 0.7735, + "step": 3727 + }, + { + "epoch": 0.71, + "learning_rate": 4.186405149785403e-06, + "loss": 0.8239, + "step": 3728 + }, + { + "epoch": 0.71, + "learning_rate": 4.181410844420473e-06, + "loss": 0.7771, + "step": 3729 + }, + { + "epoch": 0.71, + "learning_rate": 4.176418732297731e-06, + "loss": 0.7924, + "step": 3730 + }, + { + "epoch": 0.71, + "learning_rate": 4.171428815298889e-06, + "loss": 0.2448, + "step": 3731 + }, + { + "epoch": 0.71, + "learning_rate": 4.16644109530483e-06, + "loss": 0.7326, + "step": 3732 + }, + { + "epoch": 0.71, + "learning_rate": 4.161455574195611e-06, + "loss": 0.7854, + "step": 3733 + }, + { + "epoch": 0.71, + "learning_rate": 4.156472253850468e-06, + "loss": 0.7986, + "step": 3734 + }, + { + "epoch": 0.71, + "learning_rate": 4.1514911361477835e-06, + "loss": 0.7883, + "step": 3735 + }, + { + "epoch": 0.71, + "learning_rate": 4.146512222965144e-06, + "loss": 0.7855, + "step": 3736 + }, + { + "epoch": 0.71, + "learning_rate": 4.141535516179268e-06, + "loss": 0.7719, + "step": 3737 + }, + { + "epoch": 0.71, + "learning_rate": 4.136561017666079e-06, + "loss": 0.7953, + "step": 3738 + }, + { + "epoch": 0.71, + "learning_rate": 4.131588729300634e-06, + "loss": 0.7662, + "step": 3739 + }, + { + "epoch": 0.71, + "learning_rate": 4.1266186529571836e-06, + "loss": 0.8098, + "step": 3740 + }, + { + "epoch": 0.71, + "learning_rate": 4.1216507905091315e-06, + "loss": 0.7703, + "step": 3741 + }, + { + "epoch": 0.71, + "learning_rate": 4.116685143829049e-06, + "loss": 0.8006, + "step": 3742 + }, + { + "epoch": 0.71, + "learning_rate": 4.111721714788671e-06, + "loss": 0.7788, + "step": 3743 + }, + { + "epoch": 0.71, + "learning_rate": 4.106760505258901e-06, + "loss": 0.7556, + "step": 3744 + }, + { + "epoch": 0.71, + "learning_rate": 4.1018015171098015e-06, + "loss": 0.8246, + "step": 3745 + }, + { + "epoch": 0.71, + "learning_rate": 4.096844752210598e-06, + "loss": 0.7592, + "step": 3746 + }, + { + "epoch": 0.71, + "learning_rate": 4.091890212429681e-06, + "loss": 0.8183, + "step": 3747 + }, + { + "epoch": 0.71, + "learning_rate": 4.0869378996346e-06, + "loss": 0.7795, + "step": 3748 + }, + { + "epoch": 0.71, + "learning_rate": 4.0819878156920645e-06, + "loss": 0.7832, + "step": 3749 + }, + { + "epoch": 0.71, + "learning_rate": 4.077039962467946e-06, + "loss": 0.7422, + "step": 3750 + }, + { + "epoch": 0.71, + "learning_rate": 4.0720943418272726e-06, + "loss": 0.7942, + "step": 3751 + }, + { + "epoch": 0.71, + "learning_rate": 4.067150955634234e-06, + "loss": 0.795, + "step": 3752 + }, + { + "epoch": 0.71, + "learning_rate": 4.062209805752176e-06, + "loss": 0.7704, + "step": 3753 + }, + { + "epoch": 0.71, + "learning_rate": 4.057270894043596e-06, + "loss": 0.7879, + "step": 3754 + }, + { + "epoch": 0.71, + "learning_rate": 4.052334222370164e-06, + "loss": 0.8111, + "step": 3755 + }, + { + "epoch": 0.71, + "learning_rate": 4.047399792592682e-06, + "loss": 0.7882, + "step": 3756 + }, + { + "epoch": 0.71, + "learning_rate": 4.042467606571134e-06, + "loss": 0.8032, + "step": 3757 + }, + { + "epoch": 0.71, + "learning_rate": 4.037537666164628e-06, + "loss": 0.7975, + "step": 3758 + }, + { + "epoch": 0.71, + "learning_rate": 4.032609973231456e-06, + "loss": 0.736, + "step": 3759 + }, + { + "epoch": 0.71, + "learning_rate": 4.027684529629036e-06, + "loss": 0.7657, + "step": 3760 + }, + { + "epoch": 0.71, + "learning_rate": 4.022761337213963e-06, + "loss": 0.8291, + "step": 3761 + }, + { + "epoch": 0.71, + "learning_rate": 4.0178403978419565e-06, + "loss": 0.7566, + "step": 3762 + }, + { + "epoch": 0.71, + "learning_rate": 4.012921713367917e-06, + "loss": 0.7775, + "step": 3763 + }, + { + "epoch": 0.71, + "learning_rate": 4.008005285645863e-06, + "loss": 0.8285, + "step": 3764 + }, + { + "epoch": 0.71, + "learning_rate": 4.003091116528994e-06, + "loss": 0.7282, + "step": 3765 + }, + { + "epoch": 0.71, + "learning_rate": 3.998179207869629e-06, + "loss": 0.7985, + "step": 3766 + }, + { + "epoch": 0.71, + "learning_rate": 3.993269561519257e-06, + "loss": 0.744, + "step": 3767 + }, + { + "epoch": 0.71, + "learning_rate": 3.988362179328506e-06, + "loss": 0.7812, + "step": 3768 + }, + { + "epoch": 0.71, + "learning_rate": 3.983457063147148e-06, + "loss": 0.758, + "step": 3769 + }, + { + "epoch": 0.71, + "learning_rate": 3.978554214824103e-06, + "loss": 0.7744, + "step": 3770 + }, + { + "epoch": 0.71, + "learning_rate": 3.973653636207437e-06, + "loss": 0.8034, + "step": 3771 + }, + { + "epoch": 0.71, + "learning_rate": 3.968755329144362e-06, + "loss": 0.7641, + "step": 3772 + }, + { + "epoch": 0.72, + "learning_rate": 3.963859295481225e-06, + "loss": 0.7696, + "step": 3773 + }, + { + "epoch": 0.72, + "learning_rate": 3.958965537063535e-06, + "loss": 0.7244, + "step": 3774 + }, + { + "epoch": 0.72, + "learning_rate": 3.954074055735916e-06, + "loss": 0.7875, + "step": 3775 + }, + { + "epoch": 0.72, + "learning_rate": 3.949184853342165e-06, + "loss": 0.7814, + "step": 3776 + }, + { + "epoch": 0.72, + "learning_rate": 3.944297931725187e-06, + "loss": 0.8458, + "step": 3777 + }, + { + "epoch": 0.72, + "learning_rate": 3.939413292727061e-06, + "loss": 0.8084, + "step": 3778 + }, + { + "epoch": 0.72, + "learning_rate": 3.934530938188973e-06, + "loss": 0.8017, + "step": 3779 + }, + { + "epoch": 0.72, + "learning_rate": 3.929650869951278e-06, + "loss": 0.7263, + "step": 3780 + }, + { + "epoch": 0.72, + "learning_rate": 3.924773089853441e-06, + "loss": 0.7961, + "step": 3781 + }, + { + "epoch": 0.72, + "learning_rate": 3.919897599734092e-06, + "loss": 0.7897, + "step": 3782 + }, + { + "epoch": 0.72, + "learning_rate": 3.915024401430969e-06, + "loss": 0.782, + "step": 3783 + }, + { + "epoch": 0.72, + "learning_rate": 3.910153496780978e-06, + "loss": 0.7505, + "step": 3784 + }, + { + "epoch": 0.72, + "learning_rate": 3.9052848876201285e-06, + "loss": 0.7925, + "step": 3785 + }, + { + "epoch": 0.72, + "learning_rate": 3.900418575783594e-06, + "loss": 0.8032, + "step": 3786 + }, + { + "epoch": 0.72, + "learning_rate": 3.895554563105652e-06, + "loss": 0.7381, + "step": 3787 + }, + { + "epoch": 0.72, + "learning_rate": 3.890692851419743e-06, + "loss": 0.8155, + "step": 3788 + }, + { + "epoch": 0.72, + "learning_rate": 3.885833442558421e-06, + "loss": 0.7529, + "step": 3789 + }, + { + "epoch": 0.72, + "learning_rate": 3.88097633835338e-06, + "loss": 0.743, + "step": 3790 + }, + { + "epoch": 0.72, + "learning_rate": 3.876121540635439e-06, + "loss": 0.7694, + "step": 3791 + }, + { + "epoch": 0.72, + "learning_rate": 3.8712690512345555e-06, + "loss": 0.733, + "step": 3792 + }, + { + "epoch": 0.72, + "learning_rate": 3.866418871979809e-06, + "loss": 0.7631, + "step": 3793 + }, + { + "epoch": 0.72, + "learning_rate": 3.8615710046994144e-06, + "loss": 0.741, + "step": 3794 + }, + { + "epoch": 0.72, + "learning_rate": 3.856725451220713e-06, + "loss": 0.7653, + "step": 3795 + }, + { + "epoch": 0.72, + "learning_rate": 3.85188221337017e-06, + "loss": 0.2319, + "step": 3796 + }, + { + "epoch": 0.72, + "learning_rate": 3.84704129297339e-06, + "loss": 0.7807, + "step": 3797 + }, + { + "epoch": 0.72, + "learning_rate": 3.842202691855085e-06, + "loss": 0.7687, + "step": 3798 + }, + { + "epoch": 0.72, + "learning_rate": 3.837366411839114e-06, + "loss": 0.7693, + "step": 3799 + }, + { + "epoch": 0.72, + "learning_rate": 3.832532454748438e-06, + "loss": 0.84, + "step": 3800 + }, + { + "epoch": 0.72, + "learning_rate": 3.8277008224051685e-06, + "loss": 0.7984, + "step": 3801 + }, + { + "epoch": 0.72, + "learning_rate": 3.822871516630514e-06, + "loss": 0.7383, + "step": 3802 + }, + { + "epoch": 0.72, + "learning_rate": 3.818044539244832e-06, + "loss": 0.7147, + "step": 3803 + }, + { + "epoch": 0.72, + "learning_rate": 3.8132198920675745e-06, + "loss": 0.8186, + "step": 3804 + }, + { + "epoch": 0.72, + "learning_rate": 3.8083975769173463e-06, + "loss": 0.7472, + "step": 3805 + }, + { + "epoch": 0.72, + "learning_rate": 3.8035775956118416e-06, + "loss": 0.7582, + "step": 3806 + }, + { + "epoch": 0.72, + "learning_rate": 3.798759949967902e-06, + "loss": 0.2934, + "step": 3807 + }, + { + "epoch": 0.72, + "learning_rate": 3.7939446418014713e-06, + "loss": 0.7655, + "step": 3808 + }, + { + "epoch": 0.72, + "learning_rate": 3.7891316729276196e-06, + "loss": 0.7798, + "step": 3809 + }, + { + "epoch": 0.72, + "learning_rate": 3.7843210451605326e-06, + "loss": 0.7681, + "step": 3810 + }, + { + "epoch": 0.72, + "learning_rate": 3.779512760313514e-06, + "loss": 0.8468, + "step": 3811 + }, + { + "epoch": 0.72, + "learning_rate": 3.774706820198987e-06, + "loss": 0.8323, + "step": 3812 + }, + { + "epoch": 0.72, + "learning_rate": 3.7699032266284863e-06, + "loss": 0.7921, + "step": 3813 + }, + { + "epoch": 0.72, + "learning_rate": 3.7651019814126656e-06, + "loss": 0.7548, + "step": 3814 + }, + { + "epoch": 0.72, + "learning_rate": 3.760303086361292e-06, + "loss": 0.7663, + "step": 3815 + }, + { + "epoch": 0.72, + "learning_rate": 3.7555065432832482e-06, + "loss": 0.7139, + "step": 3816 + }, + { + "epoch": 0.72, + "learning_rate": 3.750712353986529e-06, + "loss": 0.7345, + "step": 3817 + }, + { + "epoch": 0.72, + "learning_rate": 3.7459205202782413e-06, + "loss": 0.2499, + "step": 3818 + }, + { + "epoch": 0.72, + "learning_rate": 3.7411310439646055e-06, + "loss": 0.7342, + "step": 3819 + }, + { + "epoch": 0.72, + "learning_rate": 3.736343926850954e-06, + "loss": 0.7345, + "step": 3820 + }, + { + "epoch": 0.72, + "learning_rate": 3.7315591707417263e-06, + "loss": 0.7161, + "step": 3821 + }, + { + "epoch": 0.72, + "learning_rate": 3.7267767774404783e-06, + "loss": 0.7674, + "step": 3822 + }, + { + "epoch": 0.72, + "learning_rate": 3.7219967487498645e-06, + "loss": 0.7969, + "step": 3823 + }, + { + "epoch": 0.72, + "learning_rate": 3.7172190864716675e-06, + "loss": 0.7634, + "step": 3824 + }, + { + "epoch": 0.72, + "learning_rate": 3.7124437924067514e-06, + "loss": 0.8128, + "step": 3825 + }, + { + "epoch": 0.73, + "learning_rate": 3.707670868355114e-06, + "loss": 0.7846, + "step": 3826 + }, + { + "epoch": 0.73, + "learning_rate": 3.702900316115836e-06, + "loss": 0.7668, + "step": 3827 + }, + { + "epoch": 0.73, + "learning_rate": 3.698132137487126e-06, + "loss": 0.7378, + "step": 3828 + }, + { + "epoch": 0.73, + "learning_rate": 3.693366334266282e-06, + "loss": 0.7303, + "step": 3829 + }, + { + "epoch": 0.73, + "learning_rate": 3.688602908249712e-06, + "loss": 0.8026, + "step": 3830 + }, + { + "epoch": 0.73, + "learning_rate": 3.6838418612329306e-06, + "loss": 0.7454, + "step": 3831 + }, + { + "epoch": 0.73, + "learning_rate": 3.679083195010551e-06, + "loss": 0.7507, + "step": 3832 + }, + { + "epoch": 0.73, + "learning_rate": 3.6743269113762923e-06, + "loss": 0.7337, + "step": 3833 + }, + { + "epoch": 0.73, + "learning_rate": 3.6695730121229734e-06, + "loss": 0.7896, + "step": 3834 + }, + { + "epoch": 0.73, + "learning_rate": 3.6648214990425158e-06, + "loss": 0.2822, + "step": 3835 + }, + { + "epoch": 0.73, + "learning_rate": 3.6600723739259404e-06, + "loss": 0.7084, + "step": 3836 + }, + { + "epoch": 0.73, + "learning_rate": 3.6553256385633697e-06, + "loss": 0.804, + "step": 3837 + }, + { + "epoch": 0.73, + "learning_rate": 3.6505812947440244e-06, + "loss": 0.7775, + "step": 3838 + }, + { + "epoch": 0.73, + "learning_rate": 3.6458393442562223e-06, + "loss": 0.2445, + "step": 3839 + }, + { + "epoch": 0.73, + "learning_rate": 3.6410997888873833e-06, + "loss": 0.8026, + "step": 3840 + }, + { + "epoch": 0.73, + "learning_rate": 3.6363626304240185e-06, + "loss": 0.7925, + "step": 3841 + }, + { + "epoch": 0.73, + "learning_rate": 3.6316278706517404e-06, + "loss": 0.8351, + "step": 3842 + }, + { + "epoch": 0.73, + "learning_rate": 3.626895511355256e-06, + "loss": 0.711, + "step": 3843 + }, + { + "epoch": 0.73, + "learning_rate": 3.6221655543183667e-06, + "loss": 0.7799, + "step": 3844 + }, + { + "epoch": 0.73, + "learning_rate": 3.6174380013239685e-06, + "loss": 0.7395, + "step": 3845 + }, + { + "epoch": 0.73, + "learning_rate": 3.6127128541540524e-06, + "loss": 0.7572, + "step": 3846 + }, + { + "epoch": 0.73, + "learning_rate": 3.6079901145897013e-06, + "loss": 0.8074, + "step": 3847 + }, + { + "epoch": 0.73, + "learning_rate": 3.6032697844110896e-06, + "loss": 0.7732, + "step": 3848 + }, + { + "epoch": 0.73, + "learning_rate": 3.5985518653974837e-06, + "loss": 0.7562, + "step": 3849 + }, + { + "epoch": 0.73, + "learning_rate": 3.5938363593272483e-06, + "loss": 0.7666, + "step": 3850 + }, + { + "epoch": 0.73, + "learning_rate": 3.5891232679778287e-06, + "loss": 0.7447, + "step": 3851 + }, + { + "epoch": 0.73, + "learning_rate": 3.5844125931257655e-06, + "loss": 0.7653, + "step": 3852 + }, + { + "epoch": 0.73, + "learning_rate": 3.5797043365466842e-06, + "loss": 0.7551, + "step": 3853 + }, + { + "epoch": 0.73, + "learning_rate": 3.5749985000153043e-06, + "loss": 0.8115, + "step": 3854 + }, + { + "epoch": 0.73, + "learning_rate": 3.5702950853054284e-06, + "loss": 0.8326, + "step": 3855 + }, + { + "epoch": 0.73, + "learning_rate": 3.565594094189949e-06, + "loss": 0.7284, + "step": 3856 + }, + { + "epoch": 0.73, + "learning_rate": 3.560895528440844e-06, + "loss": 0.8332, + "step": 3857 + }, + { + "epoch": 0.73, + "learning_rate": 3.556199389829177e-06, + "loss": 0.2431, + "step": 3858 + }, + { + "epoch": 0.73, + "learning_rate": 3.551505680125096e-06, + "loss": 0.8453, + "step": 3859 + }, + { + "epoch": 0.73, + "learning_rate": 3.5468144010978354e-06, + "loss": 0.8307, + "step": 3860 + }, + { + "epoch": 0.73, + "learning_rate": 3.542125554515713e-06, + "loss": 0.7447, + "step": 3861 + }, + { + "epoch": 0.73, + "learning_rate": 3.5374391421461273e-06, + "loss": 0.7606, + "step": 3862 + }, + { + "epoch": 0.73, + "learning_rate": 3.5327551657555625e-06, + "loss": 0.7885, + "step": 3863 + }, + { + "epoch": 0.73, + "learning_rate": 3.5280736271095827e-06, + "loss": 0.7705, + "step": 3864 + }, + { + "epoch": 0.73, + "learning_rate": 3.523394527972833e-06, + "loss": 0.8158, + "step": 3865 + }, + { + "epoch": 0.73, + "learning_rate": 3.5187178701090408e-06, + "loss": 0.7446, + "step": 3866 + }, + { + "epoch": 0.73, + "learning_rate": 3.514043655281011e-06, + "loss": 0.2792, + "step": 3867 + }, + { + "epoch": 0.73, + "learning_rate": 3.5093718852506254e-06, + "loss": 0.7212, + "step": 3868 + }, + { + "epoch": 0.73, + "learning_rate": 3.5047025617788578e-06, + "loss": 0.7512, + "step": 3869 + }, + { + "epoch": 0.73, + "learning_rate": 3.500035686625736e-06, + "loss": 0.6843, + "step": 3870 + }, + { + "epoch": 0.73, + "learning_rate": 3.495371261550392e-06, + "loss": 0.7184, + "step": 3871 + }, + { + "epoch": 0.73, + "learning_rate": 3.490709288311006e-06, + "loss": 0.7686, + "step": 3872 + }, + { + "epoch": 0.73, + "learning_rate": 3.4860497686648644e-06, + "loss": 0.8251, + "step": 3873 + }, + { + "epoch": 0.73, + "learning_rate": 3.4813927043682984e-06, + "loss": 0.8393, + "step": 3874 + }, + { + "epoch": 0.73, + "learning_rate": 3.4767380971767416e-06, + "loss": 0.7467, + "step": 3875 + }, + { + "epoch": 0.73, + "learning_rate": 3.4720859488446744e-06, + "loss": 0.7913, + "step": 3876 + }, + { + "epoch": 0.73, + "learning_rate": 3.4674362611256795e-06, + "loss": 0.7995, + "step": 3877 + }, + { + "epoch": 0.74, + "learning_rate": 3.4627890357723817e-06, + "loss": 0.8095, + "step": 3878 + }, + { + "epoch": 0.74, + "learning_rate": 3.4581442745365025e-06, + "loss": 0.8249, + "step": 3879 + }, + { + "epoch": 0.74, + "learning_rate": 3.4535019791688227e-06, + "loss": 0.8116, + "step": 3880 + }, + { + "epoch": 0.74, + "learning_rate": 3.4488621514191946e-06, + "loss": 0.7871, + "step": 3881 + }, + { + "epoch": 0.74, + "learning_rate": 3.4442247930365423e-06, + "loss": 0.8157, + "step": 3882 + }, + { + "epoch": 0.74, + "learning_rate": 3.4395899057688575e-06, + "loss": 0.8252, + "step": 3883 + }, + { + "epoch": 0.74, + "learning_rate": 3.4349574913632024e-06, + "loss": 0.7724, + "step": 3884 + }, + { + "epoch": 0.74, + "learning_rate": 3.430327551565704e-06, + "loss": 0.7498, + "step": 3885 + }, + { + "epoch": 0.74, + "learning_rate": 3.4257000881215606e-06, + "loss": 0.7168, + "step": 3886 + }, + { + "epoch": 0.74, + "learning_rate": 3.4210751027750333e-06, + "loss": 0.7933, + "step": 3887 + }, + { + "epoch": 0.74, + "learning_rate": 3.4164525972694506e-06, + "loss": 0.7801, + "step": 3888 + }, + { + "epoch": 0.74, + "learning_rate": 3.4118325733472047e-06, + "loss": 0.7844, + "step": 3889 + }, + { + "epoch": 0.74, + "learning_rate": 3.407215032749763e-06, + "loss": 0.7853, + "step": 3890 + }, + { + "epoch": 0.74, + "learning_rate": 3.402599977217634e-06, + "loss": 0.7908, + "step": 3891 + }, + { + "epoch": 0.74, + "learning_rate": 3.3979874084904185e-06, + "loss": 0.2587, + "step": 3892 + }, + { + "epoch": 0.74, + "learning_rate": 3.3933773283067495e-06, + "loss": 0.7456, + "step": 3893 + }, + { + "epoch": 0.74, + "learning_rate": 3.3887697384043525e-06, + "loss": 0.822, + "step": 3894 + }, + { + "epoch": 0.74, + "learning_rate": 3.3841646405199836e-06, + "loss": 0.7877, + "step": 3895 + }, + { + "epoch": 0.74, + "learning_rate": 3.379562036389491e-06, + "loss": 0.276, + "step": 3896 + }, + { + "epoch": 0.74, + "learning_rate": 3.374961927747751e-06, + "loss": 0.8299, + "step": 3897 + }, + { + "epoch": 0.74, + "learning_rate": 3.37036431632873e-06, + "loss": 0.7786, + "step": 3898 + }, + { + "epoch": 0.74, + "learning_rate": 3.3657692038654256e-06, + "loss": 0.2956, + "step": 3899 + }, + { + "epoch": 0.74, + "learning_rate": 3.361176592089919e-06, + "loss": 0.7911, + "step": 3900 + }, + { + "epoch": 0.74, + "learning_rate": 3.3565864827333227e-06, + "loss": 0.2498, + "step": 3901 + }, + { + "epoch": 0.74, + "learning_rate": 3.351998877525828e-06, + "loss": 0.2621, + "step": 3902 + }, + { + "epoch": 0.74, + "learning_rate": 3.347413778196673e-06, + "loss": 0.2684, + "step": 3903 + }, + { + "epoch": 0.74, + "learning_rate": 3.342831186474149e-06, + "loss": 0.8075, + "step": 3904 + }, + { + "epoch": 0.74, + "learning_rate": 3.3382511040856058e-06, + "loss": 0.7781, + "step": 3905 + }, + { + "epoch": 0.74, + "learning_rate": 3.333673532757448e-06, + "loss": 0.7816, + "step": 3906 + }, + { + "epoch": 0.74, + "learning_rate": 3.32909847421513e-06, + "loss": 0.7717, + "step": 3907 + }, + { + "epoch": 0.74, + "learning_rate": 3.324525930183161e-06, + "loss": 0.7471, + "step": 3908 + }, + { + "epoch": 0.74, + "learning_rate": 3.3199559023851047e-06, + "loss": 0.2296, + "step": 3909 + }, + { + "epoch": 0.74, + "learning_rate": 3.3153883925435693e-06, + "loss": 0.7755, + "step": 3910 + }, + { + "epoch": 0.74, + "learning_rate": 3.31082340238023e-06, + "loss": 0.7738, + "step": 3911 + }, + { + "epoch": 0.74, + "learning_rate": 3.306260933615786e-06, + "loss": 0.8061, + "step": 3912 + }, + { + "epoch": 0.74, + "learning_rate": 3.3017009879700145e-06, + "loss": 0.8316, + "step": 3913 + }, + { + "epoch": 0.74, + "learning_rate": 3.297143567161716e-06, + "loss": 0.7497, + "step": 3914 + }, + { + "epoch": 0.74, + "learning_rate": 3.2925886729087643e-06, + "loss": 0.7521, + "step": 3915 + }, + { + "epoch": 0.74, + "learning_rate": 3.2880363069280553e-06, + "loss": 0.8133, + "step": 3916 + }, + { + "epoch": 0.74, + "learning_rate": 3.2834864709355575e-06, + "loss": 0.7688, + "step": 3917 + }, + { + "epoch": 0.74, + "learning_rate": 3.27893916664626e-06, + "loss": 0.7709, + "step": 3918 + }, + { + "epoch": 0.74, + "learning_rate": 3.274394395774223e-06, + "loss": 0.2746, + "step": 3919 + }, + { + "epoch": 0.74, + "learning_rate": 3.2698521600325273e-06, + "loss": 0.7683, + "step": 3920 + }, + { + "epoch": 0.74, + "learning_rate": 3.265312461133322e-06, + "loss": 0.7676, + "step": 3921 + }, + { + "epoch": 0.74, + "learning_rate": 3.260775300787776e-06, + "loss": 0.7774, + "step": 3922 + }, + { + "epoch": 0.74, + "learning_rate": 3.256240680706123e-06, + "loss": 0.8056, + "step": 3923 + }, + { + "epoch": 0.74, + "learning_rate": 3.251708602597625e-06, + "loss": 0.8163, + "step": 3924 + }, + { + "epoch": 0.74, + "learning_rate": 3.2471790681705928e-06, + "loss": 0.8087, + "step": 3925 + }, + { + "epoch": 0.74, + "learning_rate": 3.2426520791323745e-06, + "loss": 0.7774, + "step": 3926 + }, + { + "epoch": 0.74, + "learning_rate": 3.23812763718936e-06, + "loss": 0.2691, + "step": 3927 + }, + { + "epoch": 0.74, + "learning_rate": 3.23360574404698e-06, + "loss": 0.2646, + "step": 3928 + }, + { + "epoch": 0.74, + "learning_rate": 3.2290864014097046e-06, + "loss": 0.7816, + "step": 3929 + }, + { + "epoch": 0.74, + "learning_rate": 3.224569610981041e-06, + "loss": 0.826, + "step": 3930 + }, + { + "epoch": 0.75, + "learning_rate": 3.2200553744635345e-06, + "loss": 0.7633, + "step": 3931 + }, + { + "epoch": 0.75, + "learning_rate": 3.215543693558769e-06, + "loss": 0.7602, + "step": 3932 + }, + { + "epoch": 0.75, + "learning_rate": 3.211034569967365e-06, + "loss": 0.7777, + "step": 3933 + }, + { + "epoch": 0.75, + "learning_rate": 3.206528005388977e-06, + "loss": 0.7283, + "step": 3934 + }, + { + "epoch": 0.75, + "learning_rate": 3.2020240015222946e-06, + "loss": 0.7217, + "step": 3935 + }, + { + "epoch": 0.75, + "learning_rate": 3.197522560065053e-06, + "loss": 0.8342, + "step": 3936 + }, + { + "epoch": 0.75, + "learning_rate": 3.1930236827140004e-06, + "loss": 0.7424, + "step": 3937 + }, + { + "epoch": 0.75, + "learning_rate": 3.1885273711649422e-06, + "loss": 0.7582, + "step": 3938 + }, + { + "epoch": 0.75, + "learning_rate": 3.1840336271126935e-06, + "loss": 0.7414, + "step": 3939 + }, + { + "epoch": 0.75, + "learning_rate": 3.1795424522511254e-06, + "loss": 0.7419, + "step": 3940 + }, + { + "epoch": 0.75, + "learning_rate": 3.1750538482731162e-06, + "loss": 0.7735, + "step": 3941 + }, + { + "epoch": 0.75, + "learning_rate": 3.170567816870599e-06, + "loss": 0.8072, + "step": 3942 + }, + { + "epoch": 0.75, + "learning_rate": 3.1660843597345137e-06, + "loss": 0.7852, + "step": 3943 + }, + { + "epoch": 0.75, + "learning_rate": 3.161603478554851e-06, + "loss": 0.2492, + "step": 3944 + }, + { + "epoch": 0.75, + "learning_rate": 3.157125175020618e-06, + "loss": 0.7528, + "step": 3945 + }, + { + "epoch": 0.75, + "learning_rate": 3.152649450819852e-06, + "loss": 0.7912, + "step": 3946 + }, + { + "epoch": 0.75, + "learning_rate": 3.1481763076396233e-06, + "loss": 0.7543, + "step": 3947 + }, + { + "epoch": 0.75, + "learning_rate": 3.1437057471660225e-06, + "loss": 0.7439, + "step": 3948 + }, + { + "epoch": 0.75, + "learning_rate": 3.1392377710841704e-06, + "loss": 0.7434, + "step": 3949 + }, + { + "epoch": 0.75, + "learning_rate": 3.134772381078214e-06, + "loss": 0.7737, + "step": 3950 + }, + { + "epoch": 0.75, + "learning_rate": 3.130309578831322e-06, + "loss": 0.2417, + "step": 3951 + }, + { + "epoch": 0.75, + "learning_rate": 3.125849366025694e-06, + "loss": 0.2886, + "step": 3952 + }, + { + "epoch": 0.75, + "learning_rate": 3.1213917443425456e-06, + "loss": 0.7752, + "step": 3953 + }, + { + "epoch": 0.75, + "learning_rate": 3.1169367154621234e-06, + "loss": 0.754, + "step": 3954 + }, + { + "epoch": 0.75, + "learning_rate": 3.1124842810636913e-06, + "loss": 0.791, + "step": 3955 + }, + { + "epoch": 0.75, + "learning_rate": 3.1080344428255373e-06, + "loss": 0.7949, + "step": 3956 + }, + { + "epoch": 0.75, + "learning_rate": 3.10358720242497e-06, + "loss": 0.8026, + "step": 3957 + }, + { + "epoch": 0.75, + "learning_rate": 3.0991425615383197e-06, + "loss": 0.7822, + "step": 3958 + }, + { + "epoch": 0.75, + "learning_rate": 3.0947005218409376e-06, + "loss": 0.7578, + "step": 3959 + }, + { + "epoch": 0.75, + "learning_rate": 3.0902610850071922e-06, + "loss": 0.8194, + "step": 3960 + }, + { + "epoch": 0.75, + "learning_rate": 3.085824252710473e-06, + "loss": 0.8128, + "step": 3961 + }, + { + "epoch": 0.75, + "learning_rate": 3.0813900266231823e-06, + "loss": 0.7783, + "step": 3962 + }, + { + "epoch": 0.75, + "learning_rate": 3.07695840841675e-06, + "loss": 0.7695, + "step": 3963 + }, + { + "epoch": 0.75, + "learning_rate": 3.072529399761618e-06, + "loss": 0.7796, + "step": 3964 + }, + { + "epoch": 0.75, + "learning_rate": 3.0681030023272406e-06, + "loss": 0.7438, + "step": 3965 + }, + { + "epoch": 0.75, + "learning_rate": 3.0636792177820917e-06, + "loss": 0.798, + "step": 3966 + }, + { + "epoch": 0.75, + "learning_rate": 3.0592580477936606e-06, + "loss": 0.791, + "step": 3967 + }, + { + "epoch": 0.75, + "learning_rate": 3.054839494028451e-06, + "loss": 0.8155, + "step": 3968 + }, + { + "epoch": 0.75, + "learning_rate": 3.050423558151979e-06, + "loss": 0.7421, + "step": 3969 + }, + { + "epoch": 0.75, + "learning_rate": 3.0460102418287742e-06, + "loss": 0.7239, + "step": 3970 + }, + { + "epoch": 0.75, + "learning_rate": 3.0415995467223803e-06, + "loss": 0.7548, + "step": 3971 + }, + { + "epoch": 0.75, + "learning_rate": 3.0371914744953513e-06, + "loss": 0.7761, + "step": 3972 + }, + { + "epoch": 0.75, + "learning_rate": 3.0327860268092535e-06, + "loss": 0.8016, + "step": 3973 + }, + { + "epoch": 0.75, + "learning_rate": 3.0283832053246644e-06, + "loss": 0.7427, + "step": 3974 + }, + { + "epoch": 0.75, + "learning_rate": 3.023983011701168e-06, + "loss": 0.7678, + "step": 3975 + }, + { + "epoch": 0.75, + "learning_rate": 3.019585447597364e-06, + "loss": 0.789, + "step": 3976 + }, + { + "epoch": 0.75, + "learning_rate": 3.015190514670856e-06, + "loss": 0.7425, + "step": 3977 + }, + { + "epoch": 0.75, + "learning_rate": 3.0107982145782577e-06, + "loss": 0.7766, + "step": 3978 + }, + { + "epoch": 0.75, + "learning_rate": 3.006408548975188e-06, + "loss": 0.8001, + "step": 3979 + }, + { + "epoch": 0.75, + "learning_rate": 3.0020215195162785e-06, + "loss": 0.7749, + "step": 3980 + }, + { + "epoch": 0.75, + "learning_rate": 2.99763712785516e-06, + "loss": 0.7801, + "step": 3981 + }, + { + "epoch": 0.75, + "learning_rate": 2.993255375644475e-06, + "loss": 0.798, + "step": 3982 + }, + { + "epoch": 0.75, + "learning_rate": 2.988876264535866e-06, + "loss": 0.7981, + "step": 3983 + }, + { + "epoch": 0.76, + "learning_rate": 2.9844997961799814e-06, + "loss": 0.7743, + "step": 3984 + }, + { + "epoch": 0.76, + "learning_rate": 2.980125972226483e-06, + "loss": 0.7033, + "step": 3985 + }, + { + "epoch": 0.76, + "learning_rate": 2.975754794324015e-06, + "loss": 0.7652, + "step": 3986 + }, + { + "epoch": 0.76, + "learning_rate": 2.9713862641202484e-06, + "loss": 0.767, + "step": 3987 + }, + { + "epoch": 0.76, + "learning_rate": 2.967020383261834e-06, + "loss": 0.7888, + "step": 3988 + }, + { + "epoch": 0.76, + "learning_rate": 2.962657153394445e-06, + "loss": 0.7601, + "step": 3989 + }, + { + "epoch": 0.76, + "learning_rate": 2.9582965761627324e-06, + "loss": 0.8271, + "step": 3990 + }, + { + "epoch": 0.76, + "learning_rate": 2.953938653210369e-06, + "loss": 0.7128, + "step": 3991 + }, + { + "epoch": 0.76, + "learning_rate": 2.949583386180015e-06, + "loss": 0.8019, + "step": 3992 + }, + { + "epoch": 0.76, + "learning_rate": 2.9452307767133314e-06, + "loss": 0.8092, + "step": 3993 + }, + { + "epoch": 0.76, + "learning_rate": 2.940880826450979e-06, + "loss": 0.7373, + "step": 3994 + }, + { + "epoch": 0.76, + "learning_rate": 2.9365335370326143e-06, + "loss": 0.7615, + "step": 3995 + }, + { + "epoch": 0.76, + "learning_rate": 2.9321889100968937e-06, + "loss": 0.775, + "step": 3996 + }, + { + "epoch": 0.76, + "learning_rate": 2.9278469472814677e-06, + "loss": 0.7696, + "step": 3997 + }, + { + "epoch": 0.76, + "learning_rate": 2.923507650222983e-06, + "loss": 0.8074, + "step": 3998 + }, + { + "epoch": 0.76, + "learning_rate": 2.9191710205570824e-06, + "loss": 0.801, + "step": 3999 + }, + { + "epoch": 0.76, + "learning_rate": 2.9148370599184004e-06, + "loss": 0.7834, + "step": 4000 + }, + { + "epoch": 0.76, + "learning_rate": 2.9105057699405704e-06, + "loss": 0.727, + "step": 4001 + }, + { + "epoch": 0.76, + "learning_rate": 2.9061771522562143e-06, + "loss": 0.2708, + "step": 4002 + }, + { + "epoch": 0.76, + "learning_rate": 2.9018512084969497e-06, + "loss": 0.7944, + "step": 4003 + }, + { + "epoch": 0.76, + "learning_rate": 2.897527940293384e-06, + "loss": 0.7933, + "step": 4004 + }, + { + "epoch": 0.76, + "learning_rate": 2.8932073492751167e-06, + "loss": 0.7635, + "step": 4005 + }, + { + "epoch": 0.76, + "learning_rate": 2.8888894370707477e-06, + "loss": 0.7692, + "step": 4006 + }, + { + "epoch": 0.76, + "learning_rate": 2.884574205307844e-06, + "loss": 0.7896, + "step": 4007 + }, + { + "epoch": 0.76, + "learning_rate": 2.8802616556129902e-06, + "loss": 0.7971, + "step": 4008 + }, + { + "epoch": 0.76, + "learning_rate": 2.875951789611734e-06, + "loss": 0.7851, + "step": 4009 + }, + { + "epoch": 0.76, + "learning_rate": 2.8716446089286363e-06, + "loss": 0.8175, + "step": 4010 + }, + { + "epoch": 0.76, + "learning_rate": 2.8673401151872214e-06, + "loss": 0.7388, + "step": 4011 + }, + { + "epoch": 0.76, + "learning_rate": 2.8630383100100236e-06, + "loss": 0.7396, + "step": 4012 + }, + { + "epoch": 0.76, + "learning_rate": 2.858739195018542e-06, + "loss": 0.7351, + "step": 4013 + }, + { + "epoch": 0.76, + "learning_rate": 2.8544427718332847e-06, + "loss": 0.8106, + "step": 4014 + }, + { + "epoch": 0.76, + "learning_rate": 2.85014904207372e-06, + "loss": 0.7111, + "step": 4015 + }, + { + "epoch": 0.76, + "learning_rate": 2.8458580073583262e-06, + "loss": 0.7594, + "step": 4016 + }, + { + "epoch": 0.76, + "learning_rate": 2.8415696693045427e-06, + "loss": 0.298, + "step": 4017 + }, + { + "epoch": 0.76, + "learning_rate": 2.8372840295288106e-06, + "loss": 0.7898, + "step": 4018 + }, + { + "epoch": 0.76, + "learning_rate": 2.8330010896465456e-06, + "loss": 0.7364, + "step": 4019 + }, + { + "epoch": 0.76, + "learning_rate": 2.828720851272144e-06, + "loss": 0.7607, + "step": 4020 + }, + { + "epoch": 0.76, + "learning_rate": 2.82444331601899e-06, + "loss": 0.2446, + "step": 4021 + }, + { + "epoch": 0.76, + "learning_rate": 2.8201684854994416e-06, + "loss": 0.7859, + "step": 4022 + }, + { + "epoch": 0.76, + "learning_rate": 2.8158963613248437e-06, + "loss": 0.736, + "step": 4023 + }, + { + "epoch": 0.76, + "learning_rate": 2.811626945105518e-06, + "loss": 0.754, + "step": 4024 + }, + { + "epoch": 0.76, + "learning_rate": 2.8073602384507647e-06, + "loss": 0.2042, + "step": 4025 + }, + { + "epoch": 0.76, + "learning_rate": 2.8030962429688626e-06, + "loss": 0.7466, + "step": 4026 + }, + { + "epoch": 0.76, + "learning_rate": 2.798834960267077e-06, + "loss": 0.7907, + "step": 4027 + }, + { + "epoch": 0.76, + "learning_rate": 2.794576391951632e-06, + "loss": 0.7929, + "step": 4028 + }, + { + "epoch": 0.76, + "learning_rate": 2.7903205396277546e-06, + "loss": 0.7403, + "step": 4029 + }, + { + "epoch": 0.76, + "learning_rate": 2.7860674048996174e-06, + "loss": 0.7756, + "step": 4030 + }, + { + "epoch": 0.76, + "learning_rate": 2.7818169893703993e-06, + "loss": 0.8065, + "step": 4031 + }, + { + "epoch": 0.76, + "learning_rate": 2.7775692946422273e-06, + "loss": 0.7596, + "step": 4032 + }, + { + "epoch": 0.76, + "learning_rate": 2.7733243223162263e-06, + "loss": 0.7443, + "step": 4033 + }, + { + "epoch": 0.76, + "learning_rate": 2.7690820739924727e-06, + "loss": 0.8235, + "step": 4034 + }, + { + "epoch": 0.76, + "learning_rate": 2.7648425512700393e-06, + "loss": 0.783, + "step": 4035 + }, + { + "epoch": 0.76, + "learning_rate": 2.760605755746948e-06, + "loss": 0.8148, + "step": 4036 + }, + { + "epoch": 0.77, + "learning_rate": 2.756371689020214e-06, + "loss": 0.8309, + "step": 4037 + }, + { + "epoch": 0.77, + "learning_rate": 2.752140352685805e-06, + "loss": 0.761, + "step": 4038 + }, + { + "epoch": 0.77, + "learning_rate": 2.747911748338676e-06, + "loss": 0.7574, + "step": 4039 + }, + { + "epoch": 0.77, + "learning_rate": 2.7436858775727438e-06, + "loss": 0.7281, + "step": 4040 + }, + { + "epoch": 0.77, + "learning_rate": 2.739462741980893e-06, + "loss": 0.7699, + "step": 4041 + }, + { + "epoch": 0.77, + "learning_rate": 2.7352423431549825e-06, + "loss": 0.7401, + "step": 4042 + }, + { + "epoch": 0.77, + "learning_rate": 2.7310246826858355e-06, + "loss": 0.8323, + "step": 4043 + }, + { + "epoch": 0.77, + "learning_rate": 2.7268097621632473e-06, + "loss": 0.8284, + "step": 4044 + }, + { + "epoch": 0.77, + "learning_rate": 2.7225975831759743e-06, + "loss": 0.7458, + "step": 4045 + }, + { + "epoch": 0.77, + "learning_rate": 2.7183881473117466e-06, + "loss": 0.7481, + "step": 4046 + }, + { + "epoch": 0.77, + "learning_rate": 2.714181456157251e-06, + "loss": 0.745, + "step": 4047 + }, + { + "epoch": 0.77, + "learning_rate": 2.709977511298156e-06, + "loss": 0.2338, + "step": 4048 + }, + { + "epoch": 0.77, + "learning_rate": 2.705776314319073e-06, + "loss": 0.7355, + "step": 4049 + }, + { + "epoch": 0.77, + "learning_rate": 2.7015778668035997e-06, + "loss": 0.7862, + "step": 4050 + }, + { + "epoch": 0.77, + "learning_rate": 2.697382170334275e-06, + "loss": 0.7787, + "step": 4051 + }, + { + "epoch": 0.77, + "learning_rate": 2.693189226492625e-06, + "loss": 0.7974, + "step": 4052 + }, + { + "epoch": 0.77, + "learning_rate": 2.6889990368591145e-06, + "loss": 0.7848, + "step": 4053 + }, + { + "epoch": 0.77, + "learning_rate": 2.684811603013192e-06, + "loss": 0.8363, + "step": 4054 + }, + { + "epoch": 0.77, + "learning_rate": 2.680626926533245e-06, + "loss": 0.7892, + "step": 4055 + }, + { + "epoch": 0.77, + "learning_rate": 2.6764450089966466e-06, + "loss": 0.8083, + "step": 4056 + }, + { + "epoch": 0.77, + "learning_rate": 2.672265851979704e-06, + "loss": 0.7775, + "step": 4057 + }, + { + "epoch": 0.77, + "learning_rate": 2.6680894570577042e-06, + "loss": 0.7886, + "step": 4058 + }, + { + "epoch": 0.77, + "learning_rate": 2.6639158258048847e-06, + "loss": 0.8161, + "step": 4059 + }, + { + "epoch": 0.77, + "learning_rate": 2.6597449597944414e-06, + "loss": 0.7413, + "step": 4060 + }, + { + "epoch": 0.77, + "learning_rate": 2.655576860598528e-06, + "loss": 0.762, + "step": 4061 + }, + { + "epoch": 0.77, + "learning_rate": 2.6514115297882547e-06, + "loss": 0.8113, + "step": 4062 + }, + { + "epoch": 0.77, + "learning_rate": 2.647248968933691e-06, + "loss": 0.8081, + "step": 4063 + }, + { + "epoch": 0.77, + "learning_rate": 2.6430891796038603e-06, + "loss": 0.784, + "step": 4064 + }, + { + "epoch": 0.77, + "learning_rate": 2.638932163366742e-06, + "loss": 0.7885, + "step": 4065 + }, + { + "epoch": 0.77, + "learning_rate": 2.634777921789268e-06, + "loss": 0.7263, + "step": 4066 + }, + { + "epoch": 0.77, + "learning_rate": 2.6306264564373275e-06, + "loss": 0.6843, + "step": 4067 + }, + { + "epoch": 0.77, + "learning_rate": 2.6264777688757613e-06, + "loss": 0.7687, + "step": 4068 + }, + { + "epoch": 0.77, + "learning_rate": 2.6223318606683647e-06, + "loss": 0.7059, + "step": 4069 + }, + { + "epoch": 0.77, + "learning_rate": 2.618188733377882e-06, + "loss": 0.2615, + "step": 4070 + }, + { + "epoch": 0.77, + "learning_rate": 2.614048388566015e-06, + "loss": 0.7955, + "step": 4071 + }, + { + "epoch": 0.77, + "learning_rate": 2.6099108277934105e-06, + "loss": 0.7476, + "step": 4072 + }, + { + "epoch": 0.77, + "learning_rate": 2.6057760526196707e-06, + "loss": 0.7792, + "step": 4073 + }, + { + "epoch": 0.77, + "learning_rate": 2.6016440646033403e-06, + "loss": 0.7692, + "step": 4074 + }, + { + "epoch": 0.77, + "learning_rate": 2.597514865301931e-06, + "loss": 0.7641, + "step": 4075 + }, + { + "epoch": 0.77, + "learning_rate": 2.593388456271877e-06, + "loss": 0.7946, + "step": 4076 + }, + { + "epoch": 0.77, + "learning_rate": 2.589264839068588e-06, + "loss": 0.7781, + "step": 4077 + }, + { + "epoch": 0.77, + "learning_rate": 2.5851440152463957e-06, + "loss": 0.7406, + "step": 4078 + }, + { + "epoch": 0.77, + "learning_rate": 2.581025986358602e-06, + "loss": 0.766, + "step": 4079 + }, + { + "epoch": 0.77, + "learning_rate": 2.57691075395744e-06, + "loss": 0.8037, + "step": 4080 + }, + { + "epoch": 0.77, + "learning_rate": 2.572798319594094e-06, + "loss": 0.7916, + "step": 4081 + }, + { + "epoch": 0.77, + "learning_rate": 2.568688684818693e-06, + "loss": 0.7628, + "step": 4082 + }, + { + "epoch": 0.77, + "learning_rate": 2.564581851180312e-06, + "loss": 0.8026, + "step": 4083 + }, + { + "epoch": 0.77, + "learning_rate": 2.5604778202269687e-06, + "loss": 0.2481, + "step": 4084 + }, + { + "epoch": 0.77, + "learning_rate": 2.556376593505624e-06, + "loss": 0.7446, + "step": 4085 + }, + { + "epoch": 0.77, + "learning_rate": 2.5522781725621814e-06, + "loss": 0.8046, + "step": 4086 + }, + { + "epoch": 0.77, + "learning_rate": 2.548182558941491e-06, + "loss": 0.7379, + "step": 4087 + }, + { + "epoch": 0.77, + "learning_rate": 2.544089754187339e-06, + "loss": 0.8103, + "step": 4088 + }, + { + "epoch": 0.78, + "learning_rate": 2.5399997598424555e-06, + "loss": 0.7659, + "step": 4089 + }, + { + "epoch": 0.78, + "learning_rate": 2.5359125774485115e-06, + "loss": 0.7632, + "step": 4090 + }, + { + "epoch": 0.78, + "learning_rate": 2.531828208546119e-06, + "loss": 0.8107, + "step": 4091 + }, + { + "epoch": 0.78, + "learning_rate": 2.5277466546748273e-06, + "loss": 0.7579, + "step": 4092 + }, + { + "epoch": 0.78, + "learning_rate": 2.523667917373125e-06, + "loss": 0.7611, + "step": 4093 + }, + { + "epoch": 0.78, + "learning_rate": 2.519591998178439e-06, + "loss": 0.7553, + "step": 4094 + }, + { + "epoch": 0.78, + "learning_rate": 2.5155188986271373e-06, + "loss": 0.7752, + "step": 4095 + }, + { + "epoch": 0.78, + "learning_rate": 2.5114486202545186e-06, + "loss": 0.7939, + "step": 4096 + }, + { + "epoch": 0.78, + "learning_rate": 2.5073811645948253e-06, + "loss": 0.8057, + "step": 4097 + }, + { + "epoch": 0.78, + "learning_rate": 2.5033165331812303e-06, + "loss": 0.7745, + "step": 4098 + }, + { + "epoch": 0.78, + "learning_rate": 2.4992547275458447e-06, + "loss": 0.7836, + "step": 4099 + }, + { + "epoch": 0.78, + "learning_rate": 2.4951957492197097e-06, + "loss": 0.7213, + "step": 4100 + }, + { + "epoch": 0.78, + "learning_rate": 2.491139599732816e-06, + "loss": 0.7426, + "step": 4101 + }, + { + "epoch": 0.78, + "learning_rate": 2.487086280614063e-06, + "loss": 0.7747, + "step": 4102 + }, + { + "epoch": 0.78, + "learning_rate": 2.4830357933913063e-06, + "loss": 0.6864, + "step": 4103 + }, + { + "epoch": 0.78, + "learning_rate": 2.4789881395913228e-06, + "loss": 0.7263, + "step": 4104 + }, + { + "epoch": 0.78, + "learning_rate": 2.4749433207398234e-06, + "loss": 0.8111, + "step": 4105 + }, + { + "epoch": 0.78, + "learning_rate": 2.4709013383614488e-06, + "loss": 0.7566, + "step": 4106 + }, + { + "epoch": 0.78, + "learning_rate": 2.4668621939797745e-06, + "loss": 0.7953, + "step": 4107 + }, + { + "epoch": 0.78, + "learning_rate": 2.462825889117302e-06, + "loss": 0.7788, + "step": 4108 + }, + { + "epoch": 0.78, + "learning_rate": 2.4587924252954654e-06, + "loss": 0.7657, + "step": 4109 + }, + { + "epoch": 0.78, + "learning_rate": 2.4547618040346264e-06, + "loss": 0.7952, + "step": 4110 + }, + { + "epoch": 0.78, + "learning_rate": 2.4507340268540757e-06, + "loss": 0.7611, + "step": 4111 + }, + { + "epoch": 0.78, + "learning_rate": 2.446709095272033e-06, + "loss": 0.7602, + "step": 4112 + }, + { + "epoch": 0.78, + "learning_rate": 2.442687010805643e-06, + "loss": 0.7397, + "step": 4113 + }, + { + "epoch": 0.78, + "learning_rate": 2.438667774970981e-06, + "loss": 0.2459, + "step": 4114 + }, + { + "epoch": 0.78, + "learning_rate": 2.4346513892830427e-06, + "loss": 0.7558, + "step": 4115 + }, + { + "epoch": 0.78, + "learning_rate": 2.430637855255755e-06, + "loss": 0.7675, + "step": 4116 + }, + { + "epoch": 0.78, + "learning_rate": 2.4266271744019674e-06, + "loss": 0.6922, + "step": 4117 + }, + { + "epoch": 0.78, + "learning_rate": 2.4226193482334535e-06, + "loss": 0.2656, + "step": 4118 + }, + { + "epoch": 0.78, + "learning_rate": 2.418614378260914e-06, + "loss": 0.7739, + "step": 4119 + }, + { + "epoch": 0.78, + "learning_rate": 2.4146122659939687e-06, + "loss": 0.8227, + "step": 4120 + }, + { + "epoch": 0.78, + "learning_rate": 2.4106130129411608e-06, + "loss": 0.7615, + "step": 4121 + }, + { + "epoch": 0.78, + "learning_rate": 2.4066166206099638e-06, + "loss": 0.7527, + "step": 4122 + }, + { + "epoch": 0.78, + "learning_rate": 2.402623090506757e-06, + "loss": 0.7318, + "step": 4123 + }, + { + "epoch": 0.78, + "learning_rate": 2.39863242413686e-06, + "loss": 0.7494, + "step": 4124 + }, + { + "epoch": 0.78, + "learning_rate": 2.394644623004492e-06, + "loss": 0.756, + "step": 4125 + }, + { + "epoch": 0.78, + "learning_rate": 2.3906596886128153e-06, + "loss": 0.7827, + "step": 4126 + }, + { + "epoch": 0.78, + "learning_rate": 2.3866776224638864e-06, + "loss": 0.7869, + "step": 4127 + }, + { + "epoch": 0.78, + "learning_rate": 2.3826984260587084e-06, + "loss": 0.7651, + "step": 4128 + }, + { + "epoch": 0.78, + "learning_rate": 2.378722100897174e-06, + "loss": 0.7587, + "step": 4129 + }, + { + "epoch": 0.78, + "learning_rate": 2.374748648478117e-06, + "loss": 0.8356, + "step": 4130 + }, + { + "epoch": 0.78, + "learning_rate": 2.3707780702992755e-06, + "loss": 0.7816, + "step": 4131 + }, + { + "epoch": 0.78, + "learning_rate": 2.3668103678573095e-06, + "loss": 0.7879, + "step": 4132 + }, + { + "epoch": 0.78, + "learning_rate": 2.3628455426477925e-06, + "loss": 0.8145, + "step": 4133 + }, + { + "epoch": 0.78, + "learning_rate": 2.3588835961652135e-06, + "loss": 0.8177, + "step": 4134 + }, + { + "epoch": 0.78, + "learning_rate": 2.354924529902978e-06, + "loss": 0.7289, + "step": 4135 + }, + { + "epoch": 0.78, + "learning_rate": 2.350968345353405e-06, + "loss": 0.7285, + "step": 4136 + }, + { + "epoch": 0.78, + "learning_rate": 2.3470150440077267e-06, + "loss": 0.7277, + "step": 4137 + }, + { + "epoch": 0.78, + "learning_rate": 2.34306462735609e-06, + "loss": 0.7256, + "step": 4138 + }, + { + "epoch": 0.78, + "learning_rate": 2.339117096887552e-06, + "loss": 0.7386, + "step": 4139 + }, + { + "epoch": 0.78, + "learning_rate": 2.3351724540900823e-06, + "loss": 0.7806, + "step": 4140 + }, + { + "epoch": 0.78, + "learning_rate": 2.331230700450571e-06, + "loss": 0.7817, + "step": 4141 + }, + { + "epoch": 0.79, + "learning_rate": 2.327291837454799e-06, + "loss": 0.7149, + "step": 4142 + }, + { + "epoch": 0.79, + "learning_rate": 2.3233558665874833e-06, + "loss": 0.6998, + "step": 4143 + }, + { + "epoch": 0.79, + "learning_rate": 2.3194227893322243e-06, + "loss": 0.7307, + "step": 4144 + }, + { + "epoch": 0.79, + "learning_rate": 2.315492607171557e-06, + "loss": 0.7136, + "step": 4145 + }, + { + "epoch": 0.79, + "learning_rate": 2.3115653215869014e-06, + "loss": 0.8121, + "step": 4146 + }, + { + "epoch": 0.79, + "learning_rate": 2.307640934058608e-06, + "loss": 0.7786, + "step": 4147 + }, + { + "epoch": 0.79, + "learning_rate": 2.303719446065914e-06, + "loss": 0.7755, + "step": 4148 + }, + { + "epoch": 0.79, + "learning_rate": 2.2998008590869838e-06, + "loss": 0.7524, + "step": 4149 + }, + { + "epoch": 0.79, + "learning_rate": 2.2958851745988677e-06, + "loss": 0.7572, + "step": 4150 + }, + { + "epoch": 0.79, + "learning_rate": 2.2919723940775427e-06, + "loss": 0.7705, + "step": 4151 + }, + { + "epoch": 0.79, + "learning_rate": 2.288062518997872e-06, + "loss": 0.8283, + "step": 4152 + }, + { + "epoch": 0.79, + "learning_rate": 2.284155550833639e-06, + "loss": 0.7545, + "step": 4153 + }, + { + "epoch": 0.79, + "learning_rate": 2.2802514910575223e-06, + "loss": 0.7841, + "step": 4154 + }, + { + "epoch": 0.79, + "learning_rate": 2.276350341141108e-06, + "loss": 0.7736, + "step": 4155 + }, + { + "epoch": 0.79, + "learning_rate": 2.2724521025548828e-06, + "loss": 0.7624, + "step": 4156 + }, + { + "epoch": 0.79, + "learning_rate": 2.268556776768239e-06, + "loss": 0.7742, + "step": 4157 + }, + { + "epoch": 0.79, + "learning_rate": 2.2646643652494693e-06, + "loss": 0.7951, + "step": 4158 + }, + { + "epoch": 0.79, + "learning_rate": 2.260774869465767e-06, + "loss": 0.7775, + "step": 4159 + }, + { + "epoch": 0.79, + "learning_rate": 2.2568882908832292e-06, + "loss": 0.7992, + "step": 4160 + }, + { + "epoch": 0.79, + "learning_rate": 2.253004630966846e-06, + "loss": 0.7746, + "step": 4161 + }, + { + "epoch": 0.79, + "learning_rate": 2.2491238911805236e-06, + "loss": 0.2693, + "step": 4162 + }, + { + "epoch": 0.79, + "learning_rate": 2.245246072987045e-06, + "loss": 0.804, + "step": 4163 + }, + { + "epoch": 0.79, + "learning_rate": 2.2413711778481152e-06, + "loss": 0.2316, + "step": 4164 + }, + { + "epoch": 0.79, + "learning_rate": 2.237499207224314e-06, + "loss": 0.7382, + "step": 4165 + }, + { + "epoch": 0.79, + "learning_rate": 2.233630162575142e-06, + "loss": 0.6951, + "step": 4166 + }, + { + "epoch": 0.79, + "learning_rate": 2.2297640453589765e-06, + "loss": 0.2492, + "step": 4167 + }, + { + "epoch": 0.79, + "learning_rate": 2.22590085703311e-06, + "loss": 0.7948, + "step": 4168 + }, + { + "epoch": 0.79, + "learning_rate": 2.2220405990537096e-06, + "loss": 0.7724, + "step": 4169 + }, + { + "epoch": 0.79, + "learning_rate": 2.2181832728758635e-06, + "loss": 0.7718, + "step": 4170 + }, + { + "epoch": 0.79, + "learning_rate": 2.214328879953528e-06, + "loss": 0.8165, + "step": 4171 + }, + { + "epoch": 0.79, + "learning_rate": 2.2104774217395788e-06, + "loss": 0.7665, + "step": 4172 + }, + { + "epoch": 0.79, + "learning_rate": 2.206628899685761e-06, + "loss": 0.7845, + "step": 4173 + }, + { + "epoch": 0.79, + "learning_rate": 2.202783315242736e-06, + "loss": 0.736, + "step": 4174 + }, + { + "epoch": 0.79, + "learning_rate": 2.198940669860041e-06, + "loss": 0.7156, + "step": 4175 + }, + { + "epoch": 0.79, + "learning_rate": 2.1951009649861167e-06, + "loss": 0.7449, + "step": 4176 + }, + { + "epoch": 0.79, + "learning_rate": 2.191264202068286e-06, + "loss": 0.7081, + "step": 4177 + }, + { + "epoch": 0.79, + "learning_rate": 2.187430382552769e-06, + "loss": 0.7891, + "step": 4178 + }, + { + "epoch": 0.79, + "learning_rate": 2.183599507884674e-06, + "loss": 0.255, + "step": 4179 + }, + { + "epoch": 0.79, + "learning_rate": 2.1797715795079998e-06, + "loss": 0.8069, + "step": 4180 + }, + { + "epoch": 0.79, + "learning_rate": 2.175946598865636e-06, + "loss": 0.7729, + "step": 4181 + }, + { + "epoch": 0.79, + "learning_rate": 2.1721245673993585e-06, + "loss": 0.7147, + "step": 4182 + }, + { + "epoch": 0.79, + "learning_rate": 2.1683054865498322e-06, + "loss": 0.7864, + "step": 4183 + }, + { + "epoch": 0.79, + "learning_rate": 2.1644893577566118e-06, + "loss": 0.7461, + "step": 4184 + }, + { + "epoch": 0.79, + "learning_rate": 2.160676182458138e-06, + "loss": 0.7236, + "step": 4185 + }, + { + "epoch": 0.79, + "learning_rate": 2.156865962091734e-06, + "loss": 0.7694, + "step": 4186 + }, + { + "epoch": 0.79, + "learning_rate": 2.1530586980936218e-06, + "loss": 0.7853, + "step": 4187 + }, + { + "epoch": 0.79, + "learning_rate": 2.1492543918988906e-06, + "loss": 0.7664, + "step": 4188 + }, + { + "epoch": 0.79, + "learning_rate": 2.1454530449415347e-06, + "loss": 0.8139, + "step": 4189 + }, + { + "epoch": 0.79, + "learning_rate": 2.141654658654412e-06, + "loss": 0.7873, + "step": 4190 + }, + { + "epoch": 0.79, + "learning_rate": 2.137859234469286e-06, + "loss": 0.7787, + "step": 4191 + }, + { + "epoch": 0.79, + "learning_rate": 2.134066773816782e-06, + "loss": 0.7961, + "step": 4192 + }, + { + "epoch": 0.79, + "learning_rate": 2.1302772781264304e-06, + "loss": 0.8021, + "step": 4193 + }, + { + "epoch": 0.79, + "learning_rate": 2.1264907488266218e-06, + "loss": 0.7685, + "step": 4194 + }, + { + "epoch": 0.8, + "learning_rate": 2.122707187344648e-06, + "loss": 0.7018, + "step": 4195 + }, + { + "epoch": 0.8, + "learning_rate": 2.118926595106671e-06, + "loss": 0.7321, + "step": 4196 + }, + { + "epoch": 0.8, + "learning_rate": 2.115148973537735e-06, + "loss": 0.2297, + "step": 4197 + }, + { + "epoch": 0.8, + "learning_rate": 2.1113743240617668e-06, + "loss": 0.755, + "step": 4198 + }, + { + "epoch": 0.8, + "learning_rate": 2.107602648101573e-06, + "loss": 0.7738, + "step": 4199 + }, + { + "epoch": 0.8, + "learning_rate": 2.103833947078836e-06, + "loss": 0.7659, + "step": 4200 + }, + { + "epoch": 0.8, + "learning_rate": 2.100068222414121e-06, + "loss": 0.7781, + "step": 4201 + }, + { + "epoch": 0.8, + "learning_rate": 2.0963054755268687e-06, + "loss": 0.7736, + "step": 4202 + }, + { + "epoch": 0.8, + "learning_rate": 2.0925457078353974e-06, + "loss": 0.7921, + "step": 4203 + }, + { + "epoch": 0.8, + "learning_rate": 2.0887889207569034e-06, + "loss": 0.832, + "step": 4204 + }, + { + "epoch": 0.8, + "learning_rate": 2.08503511570746e-06, + "loss": 0.7472, + "step": 4205 + }, + { + "epoch": 0.8, + "learning_rate": 2.0812842941020152e-06, + "loss": 0.7569, + "step": 4206 + }, + { + "epoch": 0.8, + "learning_rate": 2.0775364573543923e-06, + "loss": 0.7377, + "step": 4207 + }, + { + "epoch": 0.8, + "learning_rate": 2.07379160687729e-06, + "loss": 0.764, + "step": 4208 + }, + { + "epoch": 0.8, + "learning_rate": 2.070049744082282e-06, + "loss": 0.8042, + "step": 4209 + }, + { + "epoch": 0.8, + "learning_rate": 2.0663108703798154e-06, + "loss": 0.7402, + "step": 4210 + }, + { + "epoch": 0.8, + "learning_rate": 2.062574987179209e-06, + "loss": 0.7676, + "step": 4211 + }, + { + "epoch": 0.8, + "learning_rate": 2.058842095888658e-06, + "loss": 0.7907, + "step": 4212 + }, + { + "epoch": 0.8, + "learning_rate": 2.055112197915227e-06, + "loss": 0.736, + "step": 4213 + }, + { + "epoch": 0.8, + "learning_rate": 2.051385294664848e-06, + "loss": 0.7869, + "step": 4214 + }, + { + "epoch": 0.8, + "learning_rate": 2.0476613875423392e-06, + "loss": 0.7519, + "step": 4215 + }, + { + "epoch": 0.8, + "learning_rate": 2.0439404779513715e-06, + "loss": 0.2724, + "step": 4216 + }, + { + "epoch": 0.8, + "learning_rate": 2.0402225672944975e-06, + "loss": 0.7373, + "step": 4217 + }, + { + "epoch": 0.8, + "learning_rate": 2.0365076569731347e-06, + "loss": 0.7123, + "step": 4218 + }, + { + "epoch": 0.8, + "learning_rate": 2.0327957483875693e-06, + "loss": 0.7464, + "step": 4219 + }, + { + "epoch": 0.8, + "learning_rate": 2.029086842936959e-06, + "loss": 0.7724, + "step": 4220 + }, + { + "epoch": 0.8, + "learning_rate": 2.025380942019327e-06, + "loss": 0.844, + "step": 4221 + }, + { + "epoch": 0.8, + "learning_rate": 2.0216780470315657e-06, + "loss": 0.784, + "step": 4222 + }, + { + "epoch": 0.8, + "learning_rate": 2.0179781593694335e-06, + "loss": 0.7605, + "step": 4223 + }, + { + "epoch": 0.8, + "learning_rate": 2.0142812804275545e-06, + "loss": 0.7586, + "step": 4224 + }, + { + "epoch": 0.8, + "learning_rate": 2.0105874115994207e-06, + "loss": 0.7621, + "step": 4225 + }, + { + "epoch": 0.8, + "learning_rate": 2.006896554277388e-06, + "loss": 0.7777, + "step": 4226 + }, + { + "epoch": 0.8, + "learning_rate": 2.0032087098526766e-06, + "loss": 0.7364, + "step": 4227 + }, + { + "epoch": 0.8, + "learning_rate": 1.9995238797153727e-06, + "loss": 0.743, + "step": 4228 + }, + { + "epoch": 0.8, + "learning_rate": 1.9958420652544254e-06, + "loss": 0.7915, + "step": 4229 + }, + { + "epoch": 0.8, + "learning_rate": 1.9921632678576476e-06, + "loss": 0.7869, + "step": 4230 + }, + { + "epoch": 0.8, + "learning_rate": 1.9884874889117145e-06, + "loss": 0.7456, + "step": 4231 + }, + { + "epoch": 0.8, + "learning_rate": 1.9848147298021626e-06, + "loss": 0.7958, + "step": 4232 + }, + { + "epoch": 0.8, + "learning_rate": 1.981144991913392e-06, + "loss": 0.799, + "step": 4233 + }, + { + "epoch": 0.8, + "learning_rate": 1.9774782766286637e-06, + "loss": 0.7847, + "step": 4234 + }, + { + "epoch": 0.8, + "learning_rate": 1.9738145853300937e-06, + "loss": 0.7973, + "step": 4235 + }, + { + "epoch": 0.8, + "learning_rate": 1.970153919398674e-06, + "loss": 0.7176, + "step": 4236 + }, + { + "epoch": 0.8, + "learning_rate": 1.966496280214234e-06, + "loss": 0.7377, + "step": 4237 + }, + { + "epoch": 0.8, + "learning_rate": 1.962841669155483e-06, + "loss": 0.7597, + "step": 4238 + }, + { + "epoch": 0.8, + "learning_rate": 1.9591900875999703e-06, + "loss": 0.2589, + "step": 4239 + }, + { + "epoch": 0.8, + "learning_rate": 1.9555415369241228e-06, + "loss": 0.7215, + "step": 4240 + }, + { + "epoch": 0.8, + "learning_rate": 1.951896018503204e-06, + "loss": 0.8098, + "step": 4241 + }, + { + "epoch": 0.8, + "learning_rate": 1.948253533711353e-06, + "loss": 0.8069, + "step": 4242 + }, + { + "epoch": 0.8, + "learning_rate": 1.944614083921553e-06, + "loss": 0.7456, + "step": 4243 + }, + { + "epoch": 0.8, + "learning_rate": 1.9409776705056514e-06, + "loss": 0.7205, + "step": 4244 + }, + { + "epoch": 0.8, + "learning_rate": 1.9373442948343444e-06, + "loss": 0.7509, + "step": 4245 + }, + { + "epoch": 0.8, + "learning_rate": 1.933713958277188e-06, + "loss": 0.6939, + "step": 4246 + }, + { + "epoch": 0.8, + "learning_rate": 1.930086662202589e-06, + "loss": 0.7779, + "step": 4247 + }, + { + "epoch": 0.81, + "learning_rate": 1.9264624079778117e-06, + "loss": 0.8148, + "step": 4248 + }, + { + "epoch": 0.81, + "learning_rate": 1.922841196968971e-06, + "loss": 0.8213, + "step": 4249 + }, + { + "epoch": 0.81, + "learning_rate": 1.919223030541035e-06, + "loss": 0.755, + "step": 4250 + }, + { + "epoch": 0.81, + "learning_rate": 1.9156079100578274e-06, + "loss": 0.7268, + "step": 4251 + }, + { + "epoch": 0.81, + "learning_rate": 1.9119958368820177e-06, + "loss": 0.716, + "step": 4252 + }, + { + "epoch": 0.81, + "learning_rate": 1.9083868123751338e-06, + "loss": 0.7462, + "step": 4253 + }, + { + "epoch": 0.81, + "learning_rate": 1.9047808378975485e-06, + "loss": 0.7751, + "step": 4254 + }, + { + "epoch": 0.81, + "learning_rate": 1.9011779148084885e-06, + "loss": 0.7934, + "step": 4255 + }, + { + "epoch": 0.81, + "learning_rate": 1.8975780444660274e-06, + "loss": 0.2551, + "step": 4256 + }, + { + "epoch": 0.81, + "learning_rate": 1.8939812282270964e-06, + "loss": 0.761, + "step": 4257 + }, + { + "epoch": 0.81, + "learning_rate": 1.8903874674474575e-06, + "loss": 0.7788, + "step": 4258 + }, + { + "epoch": 0.81, + "learning_rate": 1.8867967634817463e-06, + "loss": 0.7884, + "step": 4259 + }, + { + "epoch": 0.81, + "learning_rate": 1.8832091176834188e-06, + "loss": 0.7768, + "step": 4260 + }, + { + "epoch": 0.81, + "learning_rate": 1.8796245314048046e-06, + "loss": 0.7807, + "step": 4261 + }, + { + "epoch": 0.81, + "learning_rate": 1.8760430059970558e-06, + "loss": 0.8193, + "step": 4262 + }, + { + "epoch": 0.81, + "learning_rate": 1.8724645428101917e-06, + "loss": 0.7597, + "step": 4263 + }, + { + "epoch": 0.81, + "learning_rate": 1.8688891431930588e-06, + "loss": 0.7945, + "step": 4264 + }, + { + "epoch": 0.81, + "learning_rate": 1.8653168084933683e-06, + "loss": 0.7815, + "step": 4265 + }, + { + "epoch": 0.81, + "learning_rate": 1.8617475400576545e-06, + "loss": 0.7662, + "step": 4266 + }, + { + "epoch": 0.81, + "learning_rate": 1.8581813392313186e-06, + "loss": 0.8085, + "step": 4267 + }, + { + "epoch": 0.81, + "learning_rate": 1.8546182073585828e-06, + "loss": 0.806, + "step": 4268 + }, + { + "epoch": 0.81, + "learning_rate": 1.8510581457825317e-06, + "loss": 0.7779, + "step": 4269 + }, + { + "epoch": 0.81, + "learning_rate": 1.8475011558450818e-06, + "loss": 0.7439, + "step": 4270 + }, + { + "epoch": 0.81, + "learning_rate": 1.843947238886995e-06, + "loss": 0.8156, + "step": 4271 + }, + { + "epoch": 0.81, + "learning_rate": 1.8403963962478756e-06, + "loss": 0.7895, + "step": 4272 + }, + { + "epoch": 0.81, + "learning_rate": 1.8368486292661657e-06, + "loss": 0.7732, + "step": 4273 + }, + { + "epoch": 0.81, + "learning_rate": 1.8333039392791508e-06, + "loss": 0.723, + "step": 4274 + }, + { + "epoch": 0.81, + "learning_rate": 1.829762327622958e-06, + "loss": 0.7852, + "step": 4275 + }, + { + "epoch": 0.81, + "learning_rate": 1.8262237956325502e-06, + "loss": 0.7705, + "step": 4276 + }, + { + "epoch": 0.81, + "learning_rate": 1.8226883446417297e-06, + "loss": 0.7819, + "step": 4277 + }, + { + "epoch": 0.81, + "learning_rate": 1.8191559759831466e-06, + "loss": 0.7666, + "step": 4278 + }, + { + "epoch": 0.81, + "learning_rate": 1.8156266909882713e-06, + "loss": 0.8578, + "step": 4279 + }, + { + "epoch": 0.81, + "learning_rate": 1.812100490987434e-06, + "loss": 0.7358, + "step": 4280 + }, + { + "epoch": 0.81, + "learning_rate": 1.808577377309778e-06, + "loss": 0.8457, + "step": 4281 + }, + { + "epoch": 0.81, + "learning_rate": 1.805057351283307e-06, + "loss": 0.711, + "step": 4282 + }, + { + "epoch": 0.81, + "learning_rate": 1.8015404142348391e-06, + "loss": 0.7786, + "step": 4283 + }, + { + "epoch": 0.81, + "learning_rate": 1.7980265674900487e-06, + "loss": 0.768, + "step": 4284 + }, + { + "epoch": 0.81, + "learning_rate": 1.7945158123734241e-06, + "loss": 0.7746, + "step": 4285 + }, + { + "epoch": 0.81, + "learning_rate": 1.79100815020831e-06, + "loss": 0.7587, + "step": 4286 + }, + { + "epoch": 0.81, + "learning_rate": 1.7875035823168641e-06, + "loss": 0.7427, + "step": 4287 + }, + { + "epoch": 0.81, + "learning_rate": 1.7840021100200988e-06, + "loss": 0.7513, + "step": 4288 + }, + { + "epoch": 0.81, + "learning_rate": 1.7805037346378384e-06, + "loss": 0.7771, + "step": 4289 + }, + { + "epoch": 0.81, + "learning_rate": 1.777008457488757e-06, + "loss": 0.812, + "step": 4290 + }, + { + "epoch": 0.81, + "learning_rate": 1.7735162798903537e-06, + "loss": 0.7599, + "step": 4291 + }, + { + "epoch": 0.81, + "learning_rate": 1.7700272031589583e-06, + "loss": 0.8096, + "step": 4292 + }, + { + "epoch": 0.81, + "learning_rate": 1.766541228609734e-06, + "loss": 0.787, + "step": 4293 + }, + { + "epoch": 0.81, + "learning_rate": 1.763058357556674e-06, + "loss": 0.738, + "step": 4294 + }, + { + "epoch": 0.81, + "learning_rate": 1.7595785913126007e-06, + "loss": 0.8011, + "step": 4295 + }, + { + "epoch": 0.81, + "learning_rate": 1.756101931189169e-06, + "loss": 0.7336, + "step": 4296 + }, + { + "epoch": 0.81, + "learning_rate": 1.752628378496859e-06, + "loss": 0.7794, + "step": 4297 + }, + { + "epoch": 0.81, + "learning_rate": 1.7491579345449806e-06, + "loss": 0.7974, + "step": 4298 + }, + { + "epoch": 0.81, + "learning_rate": 1.7456906006416807e-06, + "loss": 0.7607, + "step": 4299 + }, + { + "epoch": 0.82, + "learning_rate": 1.7422263780939141e-06, + "loss": 0.758, + "step": 4300 + }, + { + "epoch": 0.82, + "learning_rate": 1.7387652682074873e-06, + "loss": 0.7758, + "step": 4301 + }, + { + "epoch": 0.82, + "learning_rate": 1.7353072722870091e-06, + "loss": 0.7711, + "step": 4302 + }, + { + "epoch": 0.82, + "learning_rate": 1.7318523916359376e-06, + "loss": 0.259, + "step": 4303 + }, + { + "epoch": 0.82, + "learning_rate": 1.7284006275565346e-06, + "loss": 0.7759, + "step": 4304 + }, + { + "epoch": 0.82, + "learning_rate": 1.7249519813499093e-06, + "loss": 0.7618, + "step": 4305 + }, + { + "epoch": 0.82, + "learning_rate": 1.7215064543159743e-06, + "loss": 0.7854, + "step": 4306 + }, + { + "epoch": 0.82, + "learning_rate": 1.718064047753485e-06, + "loss": 0.7149, + "step": 4307 + }, + { + "epoch": 0.82, + "learning_rate": 1.7146247629600044e-06, + "loss": 0.7636, + "step": 4308 + }, + { + "epoch": 0.82, + "learning_rate": 1.7111886012319323e-06, + "loss": 0.815, + "step": 4309 + }, + { + "epoch": 0.82, + "learning_rate": 1.7077555638644838e-06, + "loss": 0.7231, + "step": 4310 + }, + { + "epoch": 0.82, + "learning_rate": 1.7043256521516981e-06, + "loss": 0.2412, + "step": 4311 + }, + { + "epoch": 0.82, + "learning_rate": 1.7008988673864358e-06, + "loss": 0.7739, + "step": 4312 + }, + { + "epoch": 0.82, + "learning_rate": 1.6974752108603798e-06, + "loss": 0.7348, + "step": 4313 + }, + { + "epoch": 0.82, + "learning_rate": 1.6940546838640338e-06, + "loss": 0.6968, + "step": 4314 + }, + { + "epoch": 0.82, + "learning_rate": 1.6906372876867205e-06, + "loss": 0.7931, + "step": 4315 + }, + { + "epoch": 0.82, + "learning_rate": 1.6872230236165831e-06, + "loss": 0.7699, + "step": 4316 + }, + { + "epoch": 0.82, + "learning_rate": 1.6838118929405856e-06, + "loss": 0.7235, + "step": 4317 + }, + { + "epoch": 0.82, + "learning_rate": 1.680403896944508e-06, + "loss": 0.8607, + "step": 4318 + }, + { + "epoch": 0.82, + "learning_rate": 1.676999036912952e-06, + "loss": 0.7536, + "step": 4319 + }, + { + "epoch": 0.82, + "learning_rate": 1.6735973141293349e-06, + "loss": 0.7295, + "step": 4320 + }, + { + "epoch": 0.82, + "learning_rate": 1.6701987298758927e-06, + "loss": 0.7106, + "step": 4321 + }, + { + "epoch": 0.82, + "learning_rate": 1.666803285433678e-06, + "loss": 0.6944, + "step": 4322 + }, + { + "epoch": 0.82, + "learning_rate": 1.6634109820825595e-06, + "loss": 0.7097, + "step": 4323 + }, + { + "epoch": 0.82, + "learning_rate": 1.660021821101222e-06, + "loss": 0.2372, + "step": 4324 + }, + { + "epoch": 0.82, + "learning_rate": 1.656635803767167e-06, + "loss": 0.8008, + "step": 4325 + }, + { + "epoch": 0.82, + "learning_rate": 1.6532529313567092e-06, + "loss": 0.7382, + "step": 4326 + }, + { + "epoch": 0.82, + "learning_rate": 1.6498732051449762e-06, + "loss": 0.8015, + "step": 4327 + }, + { + "epoch": 0.82, + "learning_rate": 1.64649662640592e-06, + "loss": 0.7883, + "step": 4328 + }, + { + "epoch": 0.82, + "learning_rate": 1.6431231964122885e-06, + "loss": 0.7419, + "step": 4329 + }, + { + "epoch": 0.82, + "learning_rate": 1.6397529164356606e-06, + "loss": 0.7711, + "step": 4330 + }, + { + "epoch": 0.82, + "learning_rate": 1.6363857877464161e-06, + "loss": 0.7516, + "step": 4331 + }, + { + "epoch": 0.82, + "learning_rate": 1.633021811613752e-06, + "loss": 0.7446, + "step": 4332 + }, + { + "epoch": 0.82, + "learning_rate": 1.6296609893056759e-06, + "loss": 0.7762, + "step": 4333 + }, + { + "epoch": 0.82, + "learning_rate": 1.6263033220890056e-06, + "loss": 0.7332, + "step": 4334 + }, + { + "epoch": 0.82, + "learning_rate": 1.62294881122937e-06, + "loss": 0.7417, + "step": 4335 + }, + { + "epoch": 0.82, + "learning_rate": 1.61959745799121e-06, + "loss": 0.6797, + "step": 4336 + }, + { + "epoch": 0.82, + "learning_rate": 1.6162492636377747e-06, + "loss": 0.7385, + "step": 4337 + }, + { + "epoch": 0.82, + "learning_rate": 1.6129042294311227e-06, + "loss": 0.8104, + "step": 4338 + }, + { + "epoch": 0.82, + "learning_rate": 1.6095623566321217e-06, + "loss": 0.2701, + "step": 4339 + }, + { + "epoch": 0.82, + "learning_rate": 1.6062236465004465e-06, + "loss": 0.8049, + "step": 4340 + }, + { + "epoch": 0.82, + "learning_rate": 1.6028881002945817e-06, + "loss": 0.8014, + "step": 4341 + }, + { + "epoch": 0.82, + "learning_rate": 1.5995557192718191e-06, + "loss": 0.804, + "step": 4342 + }, + { + "epoch": 0.82, + "learning_rate": 1.5962265046882552e-06, + "loss": 0.8183, + "step": 4343 + }, + { + "epoch": 0.82, + "learning_rate": 1.592900457798796e-06, + "loss": 0.7636, + "step": 4344 + }, + { + "epoch": 0.82, + "learning_rate": 1.5895775798571523e-06, + "loss": 0.7643, + "step": 4345 + }, + { + "epoch": 0.82, + "learning_rate": 1.5862578721158383e-06, + "loss": 0.7237, + "step": 4346 + }, + { + "epoch": 0.82, + "learning_rate": 1.5829413358261759e-06, + "loss": 0.7353, + "step": 4347 + }, + { + "epoch": 0.82, + "learning_rate": 1.5796279722382912e-06, + "loss": 0.7403, + "step": 4348 + }, + { + "epoch": 0.82, + "learning_rate": 1.5763177826011134e-06, + "loss": 0.247, + "step": 4349 + }, + { + "epoch": 0.82, + "learning_rate": 1.5730107681623753e-06, + "loss": 0.771, + "step": 4350 + }, + { + "epoch": 0.82, + "learning_rate": 1.569706930168613e-06, + "loss": 0.7603, + "step": 4351 + }, + { + "epoch": 0.82, + "learning_rate": 1.5664062698651706e-06, + "loss": 0.738, + "step": 4352 + }, + { + "epoch": 0.83, + "learning_rate": 1.5631087884961805e-06, + "loss": 0.802, + "step": 4353 + }, + { + "epoch": 0.83, + "learning_rate": 1.5598144873045928e-06, + "loss": 0.7824, + "step": 4354 + }, + { + "epoch": 0.83, + "learning_rate": 1.5565233675321512e-06, + "loss": 0.8171, + "step": 4355 + }, + { + "epoch": 0.83, + "learning_rate": 1.5532354304193986e-06, + "loss": 0.7217, + "step": 4356 + }, + { + "epoch": 0.83, + "learning_rate": 1.5499506772056815e-06, + "loss": 0.7643, + "step": 4357 + }, + { + "epoch": 0.83, + "learning_rate": 1.5466691091291452e-06, + "loss": 0.7643, + "step": 4358 + }, + { + "epoch": 0.83, + "learning_rate": 1.5433907274267357e-06, + "loss": 0.8153, + "step": 4359 + }, + { + "epoch": 0.83, + "learning_rate": 1.5401155333341943e-06, + "loss": 0.7151, + "step": 4360 + }, + { + "epoch": 0.83, + "learning_rate": 1.5368435280860649e-06, + "loss": 0.2824, + "step": 4361 + }, + { + "epoch": 0.83, + "learning_rate": 1.5335747129156875e-06, + "loss": 0.7528, + "step": 4362 + }, + { + "epoch": 0.83, + "learning_rate": 1.5303090890551986e-06, + "loss": 0.7766, + "step": 4363 + }, + { + "epoch": 0.83, + "learning_rate": 1.5270466577355347e-06, + "loss": 0.7971, + "step": 4364 + }, + { + "epoch": 0.83, + "learning_rate": 1.5237874201864277e-06, + "loss": 0.7935, + "step": 4365 + }, + { + "epoch": 0.83, + "learning_rate": 1.5205313776364028e-06, + "loss": 0.7886, + "step": 4366 + }, + { + "epoch": 0.83, + "learning_rate": 1.5172785313127847e-06, + "loss": 0.7907, + "step": 4367 + }, + { + "epoch": 0.83, + "learning_rate": 1.5140288824416915e-06, + "loss": 0.7768, + "step": 4368 + }, + { + "epoch": 0.83, + "learning_rate": 1.510782432248038e-06, + "loss": 0.7442, + "step": 4369 + }, + { + "epoch": 0.83, + "learning_rate": 1.5075391819555297e-06, + "loss": 0.813, + "step": 4370 + }, + { + "epoch": 0.83, + "learning_rate": 1.5042991327866685e-06, + "loss": 0.8239, + "step": 4371 + }, + { + "epoch": 0.83, + "learning_rate": 1.5010622859627465e-06, + "loss": 0.7857, + "step": 4372 + }, + { + "epoch": 0.83, + "learning_rate": 1.4978286427038602e-06, + "loss": 0.7745, + "step": 4373 + }, + { + "epoch": 0.83, + "learning_rate": 1.4945982042288788e-06, + "loss": 0.6826, + "step": 4374 + }, + { + "epoch": 0.83, + "learning_rate": 1.4913709717554836e-06, + "loss": 0.8133, + "step": 4375 + }, + { + "epoch": 0.83, + "learning_rate": 1.4881469465001308e-06, + "loss": 0.804, + "step": 4376 + }, + { + "epoch": 0.83, + "learning_rate": 1.4849261296780827e-06, + "loss": 0.7547, + "step": 4377 + }, + { + "epoch": 0.83, + "learning_rate": 1.4817085225033755e-06, + "loss": 0.7693, + "step": 4378 + }, + { + "epoch": 0.83, + "learning_rate": 1.478494126188854e-06, + "loss": 0.7652, + "step": 4379 + }, + { + "epoch": 0.83, + "learning_rate": 1.4752829419461357e-06, + "loss": 0.7527, + "step": 4380 + }, + { + "epoch": 0.83, + "learning_rate": 1.472074970985643e-06, + "loss": 0.7955, + "step": 4381 + }, + { + "epoch": 0.83, + "learning_rate": 1.46887021451657e-06, + "loss": 0.7862, + "step": 4382 + }, + { + "epoch": 0.83, + "learning_rate": 1.4656686737469162e-06, + "loss": 0.7985, + "step": 4383 + }, + { + "epoch": 0.83, + "learning_rate": 1.462470349883458e-06, + "loss": 0.7549, + "step": 4384 + }, + { + "epoch": 0.83, + "learning_rate": 1.4592752441317614e-06, + "loss": 0.8104, + "step": 4385 + }, + { + "epoch": 0.83, + "learning_rate": 1.456083357696183e-06, + "loss": 0.7769, + "step": 4386 + }, + { + "epoch": 0.83, + "learning_rate": 1.4528946917798603e-06, + "loss": 0.7664, + "step": 4387 + }, + { + "epoch": 0.83, + "learning_rate": 1.449709247584722e-06, + "loss": 0.7929, + "step": 4388 + }, + { + "epoch": 0.83, + "learning_rate": 1.4465270263114795e-06, + "loss": 0.7366, + "step": 4389 + }, + { + "epoch": 0.83, + "learning_rate": 1.4433480291596302e-06, + "loss": 0.7656, + "step": 4390 + }, + { + "epoch": 0.83, + "learning_rate": 1.4401722573274535e-06, + "loss": 0.7893, + "step": 4391 + }, + { + "epoch": 0.83, + "learning_rate": 1.4369997120120228e-06, + "loss": 0.8024, + "step": 4392 + }, + { + "epoch": 0.83, + "learning_rate": 1.433830394409178e-06, + "loss": 0.8044, + "step": 4393 + }, + { + "epoch": 0.83, + "learning_rate": 1.4306643057135638e-06, + "loss": 0.7259, + "step": 4394 + }, + { + "epoch": 0.83, + "learning_rate": 1.4275014471185856e-06, + "loss": 0.8307, + "step": 4395 + }, + { + "epoch": 0.83, + "learning_rate": 1.4243418198164515e-06, + "loss": 0.78, + "step": 4396 + }, + { + "epoch": 0.83, + "learning_rate": 1.4211854249981348e-06, + "loss": 0.7981, + "step": 4397 + }, + { + "epoch": 0.83, + "learning_rate": 1.418032263853405e-06, + "loss": 0.7841, + "step": 4398 + }, + { + "epoch": 0.83, + "learning_rate": 1.4148823375707964e-06, + "loss": 0.8169, + "step": 4399 + }, + { + "epoch": 0.83, + "learning_rate": 1.4117356473376443e-06, + "loss": 0.793, + "step": 4400 + }, + { + "epoch": 0.83, + "learning_rate": 1.4085921943400416e-06, + "loss": 0.7704, + "step": 4401 + }, + { + "epoch": 0.83, + "learning_rate": 1.4054519797628829e-06, + "loss": 0.7707, + "step": 4402 + }, + { + "epoch": 0.83, + "learning_rate": 1.4023150047898215e-06, + "loss": 0.267, + "step": 4403 + }, + { + "epoch": 0.83, + "learning_rate": 1.399181270603308e-06, + "loss": 0.2449, + "step": 4404 + }, + { + "epoch": 0.83, + "learning_rate": 1.396050778384559e-06, + "loss": 0.7983, + "step": 4405 + }, + { + "epoch": 0.84, + "learning_rate": 1.3929235293135745e-06, + "loss": 0.7734, + "step": 4406 + }, + { + "epoch": 0.84, + "learning_rate": 1.389799524569131e-06, + "loss": 0.7989, + "step": 4407 + }, + { + "epoch": 0.84, + "learning_rate": 1.3866787653287804e-06, + "loss": 0.7319, + "step": 4408 + }, + { + "epoch": 0.84, + "learning_rate": 1.3835612527688535e-06, + "loss": 0.7486, + "step": 4409 + }, + { + "epoch": 0.84, + "learning_rate": 1.3804469880644566e-06, + "loss": 0.7813, + "step": 4410 + }, + { + "epoch": 0.84, + "learning_rate": 1.3773359723894719e-06, + "loss": 0.8269, + "step": 4411 + }, + { + "epoch": 0.84, + "learning_rate": 1.3742282069165535e-06, + "loss": 0.7595, + "step": 4412 + }, + { + "epoch": 0.84, + "learning_rate": 1.3711236928171412e-06, + "loss": 0.8003, + "step": 4413 + }, + { + "epoch": 0.84, + "learning_rate": 1.368022431261432e-06, + "loss": 0.8249, + "step": 4414 + }, + { + "epoch": 0.84, + "learning_rate": 1.3649244234184157e-06, + "loss": 0.7728, + "step": 4415 + }, + { + "epoch": 0.84, + "learning_rate": 1.3618296704558364e-06, + "loss": 0.7565, + "step": 4416 + }, + { + "epoch": 0.84, + "learning_rate": 1.3587381735402316e-06, + "loss": 0.7451, + "step": 4417 + }, + { + "epoch": 0.84, + "learning_rate": 1.3556499338368933e-06, + "loss": 0.7228, + "step": 4418 + }, + { + "epoch": 0.84, + "learning_rate": 1.3525649525098993e-06, + "loss": 0.7393, + "step": 4419 + }, + { + "epoch": 0.84, + "learning_rate": 1.3494832307220885e-06, + "loss": 0.7406, + "step": 4420 + }, + { + "epoch": 0.84, + "learning_rate": 1.3464047696350813e-06, + "loss": 0.7712, + "step": 4421 + }, + { + "epoch": 0.84, + "learning_rate": 1.3433295704092586e-06, + "loss": 0.7365, + "step": 4422 + }, + { + "epoch": 0.84, + "learning_rate": 1.3402576342037832e-06, + "loss": 0.7182, + "step": 4423 + }, + { + "epoch": 0.84, + "learning_rate": 1.3371889621765743e-06, + "loss": 0.7987, + "step": 4424 + }, + { + "epoch": 0.84, + "learning_rate": 1.3341235554843346e-06, + "loss": 0.7951, + "step": 4425 + }, + { + "epoch": 0.84, + "learning_rate": 1.3310614152825274e-06, + "loss": 0.774, + "step": 4426 + }, + { + "epoch": 0.84, + "learning_rate": 1.3280025427253884e-06, + "loss": 0.8226, + "step": 4427 + }, + { + "epoch": 0.84, + "learning_rate": 1.3249469389659174e-06, + "loss": 0.7619, + "step": 4428 + }, + { + "epoch": 0.84, + "learning_rate": 1.3218946051558867e-06, + "loss": 0.7614, + "step": 4429 + }, + { + "epoch": 0.84, + "learning_rate": 1.3188455424458345e-06, + "loss": 0.7646, + "step": 4430 + }, + { + "epoch": 0.84, + "learning_rate": 1.315799751985064e-06, + "loss": 0.7296, + "step": 4431 + }, + { + "epoch": 0.84, + "learning_rate": 1.312757234921649e-06, + "loss": 0.7704, + "step": 4432 + }, + { + "epoch": 0.84, + "learning_rate": 1.3097179924024251e-06, + "loss": 0.8408, + "step": 4433 + }, + { + "epoch": 0.84, + "learning_rate": 1.3066820255729962e-06, + "loss": 0.746, + "step": 4434 + }, + { + "epoch": 0.84, + "learning_rate": 1.3036493355777303e-06, + "loss": 0.2417, + "step": 4435 + }, + { + "epoch": 0.84, + "learning_rate": 1.3006199235597628e-06, + "loss": 0.818, + "step": 4436 + }, + { + "epoch": 0.84, + "learning_rate": 1.2975937906609891e-06, + "loss": 0.7209, + "step": 4437 + }, + { + "epoch": 0.84, + "learning_rate": 1.2945709380220717e-06, + "loss": 0.736, + "step": 4438 + }, + { + "epoch": 0.84, + "learning_rate": 1.2915513667824353e-06, + "loss": 0.249, + "step": 4439 + }, + { + "epoch": 0.84, + "learning_rate": 1.2885350780802731e-06, + "loss": 0.7928, + "step": 4440 + }, + { + "epoch": 0.84, + "learning_rate": 1.2855220730525287e-06, + "loss": 0.7622, + "step": 4441 + }, + { + "epoch": 0.84, + "learning_rate": 1.2825123528349226e-06, + "loss": 0.2367, + "step": 4442 + }, + { + "epoch": 0.84, + "learning_rate": 1.279505918561923e-06, + "loss": 0.8355, + "step": 4443 + }, + { + "epoch": 0.84, + "learning_rate": 1.2765027713667754e-06, + "loss": 0.7148, + "step": 4444 + }, + { + "epoch": 0.84, + "learning_rate": 1.273502912381467e-06, + "loss": 0.2494, + "step": 4445 + }, + { + "epoch": 0.84, + "learning_rate": 1.2705063427367637e-06, + "loss": 0.7727, + "step": 4446 + }, + { + "epoch": 0.84, + "learning_rate": 1.2675130635621813e-06, + "loss": 0.7469, + "step": 4447 + }, + { + "epoch": 0.84, + "learning_rate": 1.2645230759859983e-06, + "loss": 0.8214, + "step": 4448 + }, + { + "epoch": 0.84, + "learning_rate": 1.2615363811352522e-06, + "loss": 0.2354, + "step": 4449 + }, + { + "epoch": 0.84, + "learning_rate": 1.2585529801357377e-06, + "loss": 0.7879, + "step": 4450 + }, + { + "epoch": 0.84, + "learning_rate": 1.25557287411201e-06, + "loss": 0.8096, + "step": 4451 + }, + { + "epoch": 0.84, + "learning_rate": 1.2525960641873825e-06, + "loss": 0.7534, + "step": 4452 + }, + { + "epoch": 0.84, + "learning_rate": 1.2496225514839244e-06, + "loss": 0.7596, + "step": 4453 + }, + { + "epoch": 0.84, + "learning_rate": 1.2466523371224626e-06, + "loss": 0.7439, + "step": 4454 + }, + { + "epoch": 0.84, + "learning_rate": 1.2436854222225826e-06, + "loss": 0.735, + "step": 4455 + }, + { + "epoch": 0.84, + "learning_rate": 1.240721807902624e-06, + "loss": 0.7187, + "step": 4456 + }, + { + "epoch": 0.84, + "learning_rate": 1.2377614952796825e-06, + "loss": 0.7739, + "step": 4457 + }, + { + "epoch": 0.84, + "learning_rate": 1.2348044854696106e-06, + "loss": 0.7356, + "step": 4458 + }, + { + "epoch": 0.85, + "learning_rate": 1.2318507795870138e-06, + "loss": 0.7872, + "step": 4459 + }, + { + "epoch": 0.85, + "learning_rate": 1.228900378745256e-06, + "loss": 0.7959, + "step": 4460 + }, + { + "epoch": 0.85, + "learning_rate": 1.2259532840564492e-06, + "loss": 0.7653, + "step": 4461 + }, + { + "epoch": 0.85, + "learning_rate": 1.2230094966314653e-06, + "loss": 0.8349, + "step": 4462 + }, + { + "epoch": 0.85, + "learning_rate": 1.2200690175799267e-06, + "loss": 0.7992, + "step": 4463 + }, + { + "epoch": 0.85, + "learning_rate": 1.217131848010209e-06, + "loss": 0.7575, + "step": 4464 + }, + { + "epoch": 0.85, + "learning_rate": 1.2141979890294397e-06, + "loss": 0.7783, + "step": 4465 + }, + { + "epoch": 0.85, + "learning_rate": 1.2112674417434977e-06, + "loss": 0.7506, + "step": 4466 + }, + { + "epoch": 0.85, + "learning_rate": 1.2083402072570194e-06, + "loss": 0.7756, + "step": 4467 + }, + { + "epoch": 0.85, + "learning_rate": 1.2054162866733854e-06, + "loss": 0.8191, + "step": 4468 + }, + { + "epoch": 0.85, + "learning_rate": 1.2024956810947297e-06, + "loss": 0.2448, + "step": 4469 + }, + { + "epoch": 0.85, + "learning_rate": 1.1995783916219373e-06, + "loss": 0.7383, + "step": 4470 + }, + { + "epoch": 0.85, + "learning_rate": 1.196664419354644e-06, + "loss": 0.7079, + "step": 4471 + }, + { + "epoch": 0.85, + "learning_rate": 1.1937537653912322e-06, + "loss": 0.7654, + "step": 4472 + }, + { + "epoch": 0.85, + "learning_rate": 1.1908464308288359e-06, + "loss": 0.2633, + "step": 4473 + }, + { + "epoch": 0.85, + "learning_rate": 1.187942416763338e-06, + "loss": 0.7539, + "step": 4474 + }, + { + "epoch": 0.85, + "learning_rate": 1.1850417242893685e-06, + "loss": 0.7723, + "step": 4475 + }, + { + "epoch": 0.85, + "learning_rate": 1.1821443545003075e-06, + "loss": 0.7608, + "step": 4476 + }, + { + "epoch": 0.85, + "learning_rate": 1.179250308488279e-06, + "loss": 0.7579, + "step": 4477 + }, + { + "epoch": 0.85, + "learning_rate": 1.176359587344158e-06, + "loss": 0.7617, + "step": 4478 + }, + { + "epoch": 0.85, + "learning_rate": 1.1734721921575642e-06, + "loss": 0.784, + "step": 4479 + }, + { + "epoch": 0.85, + "learning_rate": 1.1705881240168638e-06, + "loss": 0.7084, + "step": 4480 + }, + { + "epoch": 0.85, + "learning_rate": 1.167707384009168e-06, + "loss": 0.7681, + "step": 4481 + }, + { + "epoch": 0.85, + "learning_rate": 1.1648299732203372e-06, + "loss": 0.7465, + "step": 4482 + }, + { + "epoch": 0.85, + "learning_rate": 1.1619558927349717e-06, + "loss": 0.7688, + "step": 4483 + }, + { + "epoch": 0.85, + "learning_rate": 1.1590851436364214e-06, + "loss": 0.7856, + "step": 4484 + }, + { + "epoch": 0.85, + "learning_rate": 1.1562177270067766e-06, + "loss": 0.7141, + "step": 4485 + }, + { + "epoch": 0.85, + "learning_rate": 1.153353643926871e-06, + "loss": 0.7922, + "step": 4486 + }, + { + "epoch": 0.85, + "learning_rate": 1.1504928954762906e-06, + "loss": 0.7178, + "step": 4487 + }, + { + "epoch": 0.85, + "learning_rate": 1.147635482733349e-06, + "loss": 0.7759, + "step": 4488 + }, + { + "epoch": 0.85, + "learning_rate": 1.1447814067751206e-06, + "loss": 0.7847, + "step": 4489 + }, + { + "epoch": 0.85, + "learning_rate": 1.1419306686774024e-06, + "loss": 0.7379, + "step": 4490 + }, + { + "epoch": 0.85, + "learning_rate": 1.1390832695147525e-06, + "loss": 0.7491, + "step": 4491 + }, + { + "epoch": 0.85, + "learning_rate": 1.1362392103604536e-06, + "loss": 0.6999, + "step": 4492 + }, + { + "epoch": 0.85, + "learning_rate": 1.1333984922865437e-06, + "loss": 0.7881, + "step": 4493 + }, + { + "epoch": 0.85, + "learning_rate": 1.1305611163637886e-06, + "loss": 0.7779, + "step": 4494 + }, + { + "epoch": 0.85, + "learning_rate": 1.127727083661706e-06, + "loss": 0.759, + "step": 4495 + }, + { + "epoch": 0.85, + "learning_rate": 1.124896395248546e-06, + "loss": 0.7694, + "step": 4496 + }, + { + "epoch": 0.85, + "learning_rate": 1.1220690521912992e-06, + "loss": 0.7352, + "step": 4497 + }, + { + "epoch": 0.85, + "learning_rate": 1.1192450555556977e-06, + "loss": 0.7412, + "step": 4498 + }, + { + "epoch": 0.85, + "learning_rate": 1.1164244064062101e-06, + "loss": 0.7609, + "step": 4499 + }, + { + "epoch": 0.85, + "learning_rate": 1.1136071058060428e-06, + "loss": 0.7954, + "step": 4500 + }, + { + "epoch": 0.85, + "learning_rate": 1.1107931548171413e-06, + "loss": 0.7863, + "step": 4501 + }, + { + "epoch": 0.85, + "learning_rate": 1.1079825545001887e-06, + "loss": 0.7857, + "step": 4502 + }, + { + "epoch": 0.85, + "learning_rate": 1.1051753059146053e-06, + "loss": 0.8094, + "step": 4503 + }, + { + "epoch": 0.85, + "learning_rate": 1.1023714101185457e-06, + "loss": 0.7762, + "step": 4504 + }, + { + "epoch": 0.85, + "learning_rate": 1.0995708681689033e-06, + "loss": 0.7849, + "step": 4505 + }, + { + "epoch": 0.85, + "learning_rate": 1.0967736811213048e-06, + "loss": 0.7839, + "step": 4506 + }, + { + "epoch": 0.85, + "learning_rate": 1.0939798500301135e-06, + "loss": 0.7535, + "step": 4507 + }, + { + "epoch": 0.85, + "learning_rate": 1.0911893759484337e-06, + "loss": 0.777, + "step": 4508 + }, + { + "epoch": 0.85, + "learning_rate": 1.0884022599280896e-06, + "loss": 0.7827, + "step": 4509 + }, + { + "epoch": 0.85, + "learning_rate": 1.0856185030196587e-06, + "loss": 0.2568, + "step": 4510 + }, + { + "epoch": 0.86, + "learning_rate": 1.0828381062724325e-06, + "loss": 0.7587, + "step": 4511 + }, + { + "epoch": 0.86, + "learning_rate": 1.0800610707344539e-06, + "loss": 0.7403, + "step": 4512 + }, + { + "epoch": 0.86, + "learning_rate": 1.0772873974524833e-06, + "loss": 0.7577, + "step": 4513 + }, + { + "epoch": 0.86, + "learning_rate": 1.0745170874720278e-06, + "loss": 0.7663, + "step": 4514 + }, + { + "epoch": 0.86, + "learning_rate": 1.0717501418373144e-06, + "loss": 0.7923, + "step": 4515 + }, + { + "epoch": 0.86, + "learning_rate": 1.0689865615913142e-06, + "loss": 0.7663, + "step": 4516 + }, + { + "epoch": 0.86, + "learning_rate": 1.066226347775714e-06, + "loss": 0.7732, + "step": 4517 + }, + { + "epoch": 0.86, + "learning_rate": 1.0634695014309503e-06, + "loss": 0.7683, + "step": 4518 + }, + { + "epoch": 0.86, + "learning_rate": 1.0607160235961734e-06, + "loss": 0.723, + "step": 4519 + }, + { + "epoch": 0.86, + "learning_rate": 1.0579659153092759e-06, + "loss": 0.7486, + "step": 4520 + }, + { + "epoch": 0.86, + "learning_rate": 1.0552191776068744e-06, + "loss": 0.7527, + "step": 4521 + }, + { + "epoch": 0.86, + "learning_rate": 1.052475811524315e-06, + "loss": 0.7508, + "step": 4522 + }, + { + "epoch": 0.86, + "learning_rate": 1.0497358180956762e-06, + "loss": 0.7599, + "step": 4523 + }, + { + "epoch": 0.86, + "learning_rate": 1.046999198353762e-06, + "loss": 0.7354, + "step": 4524 + }, + { + "epoch": 0.86, + "learning_rate": 1.0442659533301058e-06, + "loss": 0.2767, + "step": 4525 + }, + { + "epoch": 0.86, + "learning_rate": 1.0415360840549694e-06, + "loss": 0.8097, + "step": 4526 + }, + { + "epoch": 0.86, + "learning_rate": 1.0388095915573427e-06, + "loss": 0.8009, + "step": 4527 + }, + { + "epoch": 0.86, + "learning_rate": 1.0360864768649381e-06, + "loss": 0.7852, + "step": 4528 + }, + { + "epoch": 0.86, + "learning_rate": 1.0333667410042058e-06, + "loss": 0.7476, + "step": 4529 + }, + { + "epoch": 0.86, + "learning_rate": 1.0306503850003069e-06, + "loss": 0.7477, + "step": 4530 + }, + { + "epoch": 0.86, + "learning_rate": 1.0279374098771456e-06, + "loss": 0.7143, + "step": 4531 + }, + { + "epoch": 0.86, + "learning_rate": 1.0252278166573338e-06, + "loss": 0.7418, + "step": 4532 + }, + { + "epoch": 0.86, + "learning_rate": 1.0225216063622256e-06, + "loss": 0.7869, + "step": 4533 + }, + { + "epoch": 0.86, + "learning_rate": 1.0198187800118842e-06, + "loss": 0.7978, + "step": 4534 + }, + { + "epoch": 0.86, + "learning_rate": 1.0171193386251144e-06, + "loss": 0.8058, + "step": 4535 + }, + { + "epoch": 0.86, + "learning_rate": 1.0144232832194267e-06, + "loss": 0.761, + "step": 4536 + }, + { + "epoch": 0.86, + "learning_rate": 1.0117306148110729e-06, + "loss": 0.7623, + "step": 4537 + }, + { + "epoch": 0.86, + "learning_rate": 1.0090413344150108e-06, + "loss": 0.7642, + "step": 4538 + }, + { + "epoch": 0.86, + "learning_rate": 1.0063554430449396e-06, + "loss": 0.817, + "step": 4539 + }, + { + "epoch": 0.86, + "learning_rate": 1.003672941713262e-06, + "loss": 0.8501, + "step": 4540 + }, + { + "epoch": 0.86, + "learning_rate": 1.0009938314311186e-06, + "loss": 0.76, + "step": 4541 + }, + { + "epoch": 0.86, + "learning_rate": 9.983181132083631e-07, + "loss": 0.8238, + "step": 4542 + }, + { + "epoch": 0.86, + "learning_rate": 9.956457880535741e-07, + "loss": 0.7616, + "step": 4543 + }, + { + "epoch": 0.86, + "learning_rate": 9.929768569740506e-07, + "loss": 0.7549, + "step": 4544 + }, + { + "epoch": 0.86, + "learning_rate": 9.903113209758098e-07, + "loss": 0.7536, + "step": 4545 + }, + { + "epoch": 0.86, + "learning_rate": 9.876491810635913e-07, + "loss": 0.7517, + "step": 4546 + }, + { + "epoch": 0.86, + "learning_rate": 9.849904382408559e-07, + "loss": 0.7092, + "step": 4547 + }, + { + "epoch": 0.86, + "learning_rate": 9.82335093509782e-07, + "loss": 0.7994, + "step": 4548 + }, + { + "epoch": 0.86, + "learning_rate": 9.796831478712654e-07, + "loss": 0.775, + "step": 4549 + }, + { + "epoch": 0.86, + "learning_rate": 9.77034602324929e-07, + "loss": 0.753, + "step": 4550 + }, + { + "epoch": 0.86, + "learning_rate": 9.743894578690993e-07, + "loss": 0.7066, + "step": 4551 + }, + { + "epoch": 0.86, + "learning_rate": 9.71747715500837e-07, + "loss": 0.7921, + "step": 4552 + }, + { + "epoch": 0.86, + "learning_rate": 9.691093762159065e-07, + "loss": 0.7519, + "step": 4553 + }, + { + "epoch": 0.86, + "learning_rate": 9.664744410088023e-07, + "loss": 0.7805, + "step": 4554 + }, + { + "epoch": 0.86, + "learning_rate": 9.638429108727232e-07, + "loss": 0.7564, + "step": 4555 + }, + { + "epoch": 0.86, + "learning_rate": 9.612147867995958e-07, + "loss": 0.7463, + "step": 4556 + }, + { + "epoch": 0.86, + "learning_rate": 9.58590069780052e-07, + "loss": 0.7624, + "step": 4557 + }, + { + "epoch": 0.86, + "learning_rate": 9.559687608034508e-07, + "loss": 0.7703, + "step": 4558 + }, + { + "epoch": 0.86, + "learning_rate": 9.533508608578557e-07, + "loss": 0.7977, + "step": 4559 + }, + { + "epoch": 0.86, + "learning_rate": 9.507363709300533e-07, + "loss": 0.7624, + "step": 4560 + }, + { + "epoch": 0.86, + "learning_rate": 9.481252920055417e-07, + "loss": 0.7809, + "step": 4561 + }, + { + "epoch": 0.86, + "learning_rate": 9.455176250685338e-07, + "loss": 0.7625, + "step": 4562 + }, + { + "epoch": 0.86, + "learning_rate": 9.429133711019545e-07, + "loss": 0.7411, + "step": 4563 + }, + { + "epoch": 0.87, + "learning_rate": 9.40312531087445e-07, + "loss": 0.762, + "step": 4564 + }, + { + "epoch": 0.87, + "learning_rate": 9.37715106005358e-07, + "loss": 0.805, + "step": 4565 + }, + { + "epoch": 0.87, + "learning_rate": 9.351210968347601e-07, + "loss": 0.7432, + "step": 4566 + }, + { + "epoch": 0.87, + "learning_rate": 9.325305045534283e-07, + "loss": 0.7369, + "step": 4567 + }, + { + "epoch": 0.87, + "learning_rate": 9.299433301378536e-07, + "loss": 0.2329, + "step": 4568 + }, + { + "epoch": 0.87, + "learning_rate": 9.27359574563238e-07, + "loss": 0.776, + "step": 4569 + }, + { + "epoch": 0.87, + "learning_rate": 9.247792388034948e-07, + "loss": 0.7685, + "step": 4570 + }, + { + "epoch": 0.87, + "learning_rate": 9.222023238312483e-07, + "loss": 0.7015, + "step": 4571 + }, + { + "epoch": 0.87, + "learning_rate": 9.196288306178336e-07, + "loss": 0.7598, + "step": 4572 + }, + { + "epoch": 0.87, + "learning_rate": 9.170587601332959e-07, + "loss": 0.7951, + "step": 4573 + }, + { + "epoch": 0.87, + "learning_rate": 9.144921133463902e-07, + "loss": 0.7798, + "step": 4574 + }, + { + "epoch": 0.87, + "learning_rate": 9.119288912245805e-07, + "loss": 0.7482, + "step": 4575 + }, + { + "epoch": 0.87, + "learning_rate": 9.093690947340406e-07, + "loss": 0.7825, + "step": 4576 + }, + { + "epoch": 0.87, + "learning_rate": 9.068127248396518e-07, + "loss": 0.7331, + "step": 4577 + }, + { + "epoch": 0.87, + "learning_rate": 9.042597825050048e-07, + "loss": 0.8301, + "step": 4578 + }, + { + "epoch": 0.87, + "learning_rate": 9.017102686924028e-07, + "loss": 0.7652, + "step": 4579 + }, + { + "epoch": 0.87, + "learning_rate": 8.991641843628451e-07, + "loss": 0.7662, + "step": 4580 + }, + { + "epoch": 0.87, + "learning_rate": 8.966215304760506e-07, + "loss": 0.2407, + "step": 4581 + }, + { + "epoch": 0.87, + "learning_rate": 8.94082307990437e-07, + "loss": 0.7345, + "step": 4582 + }, + { + "epoch": 0.87, + "learning_rate": 8.915465178631344e-07, + "loss": 0.7308, + "step": 4583 + }, + { + "epoch": 0.87, + "learning_rate": 8.890141610499737e-07, + "loss": 0.7878, + "step": 4584 + }, + { + "epoch": 0.87, + "learning_rate": 8.864852385054945e-07, + "loss": 0.757, + "step": 4585 + }, + { + "epoch": 0.87, + "learning_rate": 8.839597511829423e-07, + "loss": 0.2699, + "step": 4586 + }, + { + "epoch": 0.87, + "learning_rate": 8.814377000342666e-07, + "loss": 0.7614, + "step": 4587 + }, + { + "epoch": 0.87, + "learning_rate": 8.789190860101226e-07, + "loss": 0.7442, + "step": 4588 + }, + { + "epoch": 0.87, + "learning_rate": 8.764039100598687e-07, + "loss": 0.7506, + "step": 4589 + }, + { + "epoch": 0.87, + "learning_rate": 8.738921731315686e-07, + "loss": 0.7767, + "step": 4590 + }, + { + "epoch": 0.87, + "learning_rate": 8.713838761719883e-07, + "loss": 0.756, + "step": 4591 + }, + { + "epoch": 0.87, + "learning_rate": 8.688790201265984e-07, + "loss": 0.8195, + "step": 4592 + }, + { + "epoch": 0.87, + "learning_rate": 8.663776059395723e-07, + "loss": 0.7865, + "step": 4593 + }, + { + "epoch": 0.87, + "learning_rate": 8.63879634553787e-07, + "loss": 0.793, + "step": 4594 + }, + { + "epoch": 0.87, + "learning_rate": 8.613851069108181e-07, + "loss": 0.7735, + "step": 4595 + }, + { + "epoch": 0.87, + "learning_rate": 8.58894023950948e-07, + "loss": 0.7727, + "step": 4596 + }, + { + "epoch": 0.87, + "learning_rate": 8.564063866131567e-07, + "loss": 0.8152, + "step": 4597 + }, + { + "epoch": 0.87, + "learning_rate": 8.539221958351274e-07, + "loss": 0.2499, + "step": 4598 + }, + { + "epoch": 0.87, + "learning_rate": 8.514414525532444e-07, + "loss": 0.7857, + "step": 4599 + }, + { + "epoch": 0.87, + "learning_rate": 8.489641577025909e-07, + "loss": 0.7522, + "step": 4600 + }, + { + "epoch": 0.87, + "learning_rate": 8.464903122169521e-07, + "loss": 0.8051, + "step": 4601 + }, + { + "epoch": 0.87, + "learning_rate": 8.44019917028811e-07, + "loss": 0.8, + "step": 4602 + }, + { + "epoch": 0.87, + "learning_rate": 8.415529730693551e-07, + "loss": 0.7212, + "step": 4603 + }, + { + "epoch": 0.87, + "learning_rate": 8.390894812684602e-07, + "loss": 0.8074, + "step": 4604 + }, + { + "epoch": 0.87, + "learning_rate": 8.366294425547139e-07, + "loss": 0.7983, + "step": 4605 + }, + { + "epoch": 0.87, + "learning_rate": 8.341728578553942e-07, + "loss": 0.7512, + "step": 4606 + }, + { + "epoch": 0.87, + "learning_rate": 8.317197280964795e-07, + "loss": 0.7671, + "step": 4607 + }, + { + "epoch": 0.87, + "learning_rate": 8.292700542026433e-07, + "loss": 0.7881, + "step": 4608 + }, + { + "epoch": 0.87, + "learning_rate": 8.268238370972614e-07, + "loss": 0.8168, + "step": 4609 + }, + { + "epoch": 0.87, + "learning_rate": 8.243810777024019e-07, + "loss": 0.7746, + "step": 4610 + }, + { + "epoch": 0.87, + "learning_rate": 8.219417769388316e-07, + "loss": 0.7793, + "step": 4611 + }, + { + "epoch": 0.87, + "learning_rate": 8.195059357260149e-07, + "loss": 0.786, + "step": 4612 + }, + { + "epoch": 0.87, + "learning_rate": 8.170735549821085e-07, + "loss": 0.246, + "step": 4613 + }, + { + "epoch": 0.87, + "learning_rate": 8.14644635623969e-07, + "loss": 0.7879, + "step": 4614 + }, + { + "epoch": 0.87, + "learning_rate": 8.122191785671451e-07, + "loss": 0.7648, + "step": 4615 + }, + { + "epoch": 0.87, + "learning_rate": 8.097971847258823e-07, + "loss": 0.8328, + "step": 4616 + }, + { + "epoch": 0.88, + "learning_rate": 8.073786550131202e-07, + "loss": 0.8048, + "step": 4617 + }, + { + "epoch": 0.88, + "learning_rate": 8.049635903404907e-07, + "loss": 0.6675, + "step": 4618 + }, + { + "epoch": 0.88, + "learning_rate": 8.025519916183233e-07, + "loss": 0.7458, + "step": 4619 + }, + { + "epoch": 0.88, + "learning_rate": 8.001438597556377e-07, + "loss": 0.7714, + "step": 4620 + }, + { + "epoch": 0.88, + "learning_rate": 7.977391956601488e-07, + "loss": 0.7858, + "step": 4621 + }, + { + "epoch": 0.88, + "learning_rate": 7.953380002382638e-07, + "loss": 0.7251, + "step": 4622 + }, + { + "epoch": 0.88, + "learning_rate": 7.929402743950798e-07, + "loss": 0.7641, + "step": 4623 + }, + { + "epoch": 0.88, + "learning_rate": 7.90546019034395e-07, + "loss": 0.7933, + "step": 4624 + }, + { + "epoch": 0.88, + "learning_rate": 7.881552350586863e-07, + "loss": 0.7345, + "step": 4625 + }, + { + "epoch": 0.88, + "learning_rate": 7.85767923369134e-07, + "loss": 0.2671, + "step": 4626 + }, + { + "epoch": 0.88, + "learning_rate": 7.833840848655993e-07, + "loss": 0.7794, + "step": 4627 + }, + { + "epoch": 0.88, + "learning_rate": 7.810037204466458e-07, + "loss": 0.755, + "step": 4628 + }, + { + "epoch": 0.88, + "learning_rate": 7.786268310095135e-07, + "loss": 0.7482, + "step": 4629 + }, + { + "epoch": 0.88, + "learning_rate": 7.762534174501491e-07, + "loss": 0.72, + "step": 4630 + }, + { + "epoch": 0.88, + "learning_rate": 7.738834806631712e-07, + "loss": 0.7733, + "step": 4631 + }, + { + "epoch": 0.88, + "learning_rate": 7.715170215419043e-07, + "loss": 0.7468, + "step": 4632 + }, + { + "epoch": 0.88, + "learning_rate": 7.691540409783493e-07, + "loss": 0.2687, + "step": 4633 + }, + { + "epoch": 0.88, + "learning_rate": 7.667945398632048e-07, + "loss": 0.7402, + "step": 4634 + }, + { + "epoch": 0.88, + "learning_rate": 7.644385190858539e-07, + "loss": 0.8073, + "step": 4635 + }, + { + "epoch": 0.88, + "learning_rate": 7.62085979534366e-07, + "loss": 0.7476, + "step": 4636 + }, + { + "epoch": 0.88, + "learning_rate": 7.597369220955031e-07, + "loss": 0.7521, + "step": 4637 + }, + { + "epoch": 0.88, + "learning_rate": 7.573913476547113e-07, + "loss": 0.7775, + "step": 4638 + }, + { + "epoch": 0.88, + "learning_rate": 7.550492570961243e-07, + "loss": 0.7625, + "step": 4639 + }, + { + "epoch": 0.88, + "learning_rate": 7.527106513025628e-07, + "loss": 0.7839, + "step": 4640 + }, + { + "epoch": 0.88, + "learning_rate": 7.50375531155535e-07, + "loss": 0.8167, + "step": 4641 + }, + { + "epoch": 0.88, + "learning_rate": 7.480438975352311e-07, + "loss": 0.2586, + "step": 4642 + }, + { + "epoch": 0.88, + "learning_rate": 7.45715751320537e-07, + "loss": 0.2854, + "step": 4643 + }, + { + "epoch": 0.88, + "learning_rate": 7.433910933890099e-07, + "loss": 0.7677, + "step": 4644 + }, + { + "epoch": 0.88, + "learning_rate": 7.410699246169062e-07, + "loss": 0.6929, + "step": 4645 + }, + { + "epoch": 0.88, + "learning_rate": 7.387522458791552e-07, + "loss": 0.6764, + "step": 4646 + }, + { + "epoch": 0.88, + "learning_rate": 7.364380580493813e-07, + "loss": 0.8057, + "step": 4647 + }, + { + "epoch": 0.88, + "learning_rate": 7.341273619998812e-07, + "loss": 0.7938, + "step": 4648 + }, + { + "epoch": 0.88, + "learning_rate": 7.318201586016482e-07, + "loss": 0.7892, + "step": 4649 + }, + { + "epoch": 0.88, + "learning_rate": 7.295164487243478e-07, + "loss": 0.8185, + "step": 4650 + }, + { + "epoch": 0.88, + "learning_rate": 7.272162332363397e-07, + "loss": 0.7735, + "step": 4651 + }, + { + "epoch": 0.88, + "learning_rate": 7.249195130046538e-07, + "loss": 0.7675, + "step": 4652 + }, + { + "epoch": 0.88, + "learning_rate": 7.226262888950153e-07, + "loss": 0.8046, + "step": 4653 + }, + { + "epoch": 0.88, + "learning_rate": 7.203365617718195e-07, + "loss": 0.7157, + "step": 4654 + }, + { + "epoch": 0.88, + "learning_rate": 7.180503324981536e-07, + "loss": 0.8342, + "step": 4655 + }, + { + "epoch": 0.88, + "learning_rate": 7.157676019357818e-07, + "loss": 0.7961, + "step": 4656 + }, + { + "epoch": 0.88, + "learning_rate": 7.134883709451502e-07, + "loss": 0.2564, + "step": 4657 + }, + { + "epoch": 0.88, + "learning_rate": 7.112126403853837e-07, + "loss": 0.788, + "step": 4658 + }, + { + "epoch": 0.88, + "learning_rate": 7.089404111142905e-07, + "loss": 0.7156, + "step": 4659 + }, + { + "epoch": 0.88, + "learning_rate": 7.066716839883592e-07, + "loss": 0.7226, + "step": 4660 + }, + { + "epoch": 0.88, + "learning_rate": 7.044064598627554e-07, + "loss": 0.7728, + "step": 4661 + }, + { + "epoch": 0.88, + "learning_rate": 7.021447395913283e-07, + "loss": 0.2496, + "step": 4662 + }, + { + "epoch": 0.88, + "learning_rate": 6.998865240266006e-07, + "loss": 0.7799, + "step": 4663 + }, + { + "epoch": 0.88, + "learning_rate": 6.976318140197835e-07, + "loss": 0.7418, + "step": 4664 + }, + { + "epoch": 0.88, + "learning_rate": 6.953806104207539e-07, + "loss": 0.8186, + "step": 4665 + }, + { + "epoch": 0.88, + "learning_rate": 6.931329140780807e-07, + "loss": 0.7699, + "step": 4666 + }, + { + "epoch": 0.88, + "learning_rate": 6.908887258389974e-07, + "loss": 0.7782, + "step": 4667 + }, + { + "epoch": 0.88, + "learning_rate": 6.886480465494283e-07, + "loss": 0.765, + "step": 4668 + }, + { + "epoch": 0.88, + "learning_rate": 6.864108770539623e-07, + "loss": 0.7383, + "step": 4669 + }, + { + "epoch": 0.89, + "learning_rate": 6.841772181958784e-07, + "loss": 0.7953, + "step": 4670 + }, + { + "epoch": 0.89, + "learning_rate": 6.819470708171183e-07, + "loss": 0.7777, + "step": 4671 + }, + { + "epoch": 0.89, + "learning_rate": 6.797204357583154e-07, + "loss": 0.8075, + "step": 4672 + }, + { + "epoch": 0.89, + "learning_rate": 6.774973138587615e-07, + "loss": 0.7732, + "step": 4673 + }, + { + "epoch": 0.89, + "learning_rate": 6.752777059564431e-07, + "loss": 0.7316, + "step": 4674 + }, + { + "epoch": 0.89, + "learning_rate": 6.730616128880052e-07, + "loss": 0.7747, + "step": 4675 + }, + { + "epoch": 0.89, + "learning_rate": 6.708490354887809e-07, + "loss": 0.7869, + "step": 4676 + }, + { + "epoch": 0.89, + "learning_rate": 6.686399745927707e-07, + "loss": 0.7703, + "step": 4677 + }, + { + "epoch": 0.89, + "learning_rate": 6.66434431032651e-07, + "loss": 0.8551, + "step": 4678 + }, + { + "epoch": 0.89, + "learning_rate": 6.642324056397731e-07, + "loss": 0.8023, + "step": 4679 + }, + { + "epoch": 0.89, + "learning_rate": 6.620338992441644e-07, + "loss": 0.8053, + "step": 4680 + }, + { + "epoch": 0.89, + "learning_rate": 6.598389126745209e-07, + "loss": 0.7535, + "step": 4681 + }, + { + "epoch": 0.89, + "learning_rate": 6.57647446758215e-07, + "loss": 0.748, + "step": 4682 + }, + { + "epoch": 0.89, + "learning_rate": 6.554595023212917e-07, + "loss": 0.7873, + "step": 4683 + }, + { + "epoch": 0.89, + "learning_rate": 6.53275080188469e-07, + "loss": 0.7621, + "step": 4684 + }, + { + "epoch": 0.89, + "learning_rate": 6.510941811831373e-07, + "loss": 0.7025, + "step": 4685 + }, + { + "epoch": 0.89, + "learning_rate": 6.489168061273565e-07, + "loss": 0.789, + "step": 4686 + }, + { + "epoch": 0.89, + "learning_rate": 6.467429558418614e-07, + "loss": 0.8244, + "step": 4687 + }, + { + "epoch": 0.89, + "learning_rate": 6.445726311460553e-07, + "loss": 0.7452, + "step": 4688 + }, + { + "epoch": 0.89, + "learning_rate": 6.424058328580163e-07, + "loss": 0.7614, + "step": 4689 + }, + { + "epoch": 0.89, + "learning_rate": 6.402425617944885e-07, + "loss": 0.7443, + "step": 4690 + }, + { + "epoch": 0.89, + "learning_rate": 6.380828187708932e-07, + "loss": 0.7589, + "step": 4691 + }, + { + "epoch": 0.89, + "learning_rate": 6.359266046013124e-07, + "loss": 0.722, + "step": 4692 + }, + { + "epoch": 0.89, + "learning_rate": 6.337739200985093e-07, + "loss": 0.786, + "step": 4693 + }, + { + "epoch": 0.89, + "learning_rate": 6.316247660739039e-07, + "loss": 0.243, + "step": 4694 + }, + { + "epoch": 0.89, + "learning_rate": 6.29479143337598e-07, + "loss": 0.7873, + "step": 4695 + }, + { + "epoch": 0.89, + "learning_rate": 6.273370526983513e-07, + "loss": 0.7367, + "step": 4696 + }, + { + "epoch": 0.89, + "learning_rate": 6.251984949636002e-07, + "loss": 0.7511, + "step": 4697 + }, + { + "epoch": 0.89, + "learning_rate": 6.230634709394479e-07, + "loss": 0.8217, + "step": 4698 + }, + { + "epoch": 0.89, + "learning_rate": 6.209319814306603e-07, + "loss": 0.8204, + "step": 4699 + }, + { + "epoch": 0.89, + "learning_rate": 6.188040272406782e-07, + "loss": 0.7756, + "step": 4700 + }, + { + "epoch": 0.89, + "learning_rate": 6.166796091716054e-07, + "loss": 0.8501, + "step": 4701 + }, + { + "epoch": 0.89, + "learning_rate": 6.145587280242138e-07, + "loss": 0.7152, + "step": 4702 + }, + { + "epoch": 0.89, + "learning_rate": 6.124413845979416e-07, + "loss": 0.7586, + "step": 4703 + }, + { + "epoch": 0.89, + "learning_rate": 6.103275796908947e-07, + "loss": 0.7432, + "step": 4704 + }, + { + "epoch": 0.89, + "learning_rate": 6.082173140998448e-07, + "loss": 0.7697, + "step": 4705 + }, + { + "epoch": 0.89, + "learning_rate": 6.06110588620229e-07, + "loss": 0.7161, + "step": 4706 + }, + { + "epoch": 0.89, + "learning_rate": 6.040074040461507e-07, + "loss": 0.7336, + "step": 4707 + }, + { + "epoch": 0.89, + "learning_rate": 6.019077611703783e-07, + "loss": 0.7334, + "step": 4708 + }, + { + "epoch": 0.89, + "learning_rate": 5.99811660784344e-07, + "loss": 0.2673, + "step": 4709 + }, + { + "epoch": 0.89, + "learning_rate": 5.977191036781471e-07, + "loss": 0.7193, + "step": 4710 + }, + { + "epoch": 0.89, + "learning_rate": 5.956300906405499e-07, + "loss": 0.7582, + "step": 4711 + }, + { + "epoch": 0.89, + "learning_rate": 5.93544622458978e-07, + "loss": 0.7206, + "step": 4712 + }, + { + "epoch": 0.89, + "learning_rate": 5.914626999195227e-07, + "loss": 0.7641, + "step": 4713 + }, + { + "epoch": 0.89, + "learning_rate": 5.893843238069363e-07, + "loss": 0.7918, + "step": 4714 + }, + { + "epoch": 0.89, + "learning_rate": 5.873094949046387e-07, + "loss": 0.7851, + "step": 4715 + }, + { + "epoch": 0.89, + "learning_rate": 5.852382139947077e-07, + "loss": 0.2815, + "step": 4716 + }, + { + "epoch": 0.89, + "learning_rate": 5.831704818578842e-07, + "loss": 0.836, + "step": 4717 + }, + { + "epoch": 0.89, + "learning_rate": 5.811062992735783e-07, + "loss": 0.7632, + "step": 4718 + }, + { + "epoch": 0.89, + "learning_rate": 5.790456670198541e-07, + "loss": 0.7877, + "step": 4719 + }, + { + "epoch": 0.89, + "learning_rate": 5.769885858734403e-07, + "loss": 0.6954, + "step": 4720 + }, + { + "epoch": 0.89, + "learning_rate": 5.749350566097267e-07, + "loss": 0.7433, + "step": 4721 + }, + { + "epoch": 0.89, + "learning_rate": 5.728850800027652e-07, + "loss": 0.7435, + "step": 4722 + }, + { + "epoch": 0.9, + "learning_rate": 5.708386568252688e-07, + "loss": 0.763, + "step": 4723 + }, + { + "epoch": 0.9, + "learning_rate": 5.687957878486106e-07, + "loss": 0.7993, + "step": 4724 + }, + { + "epoch": 0.9, + "learning_rate": 5.667564738428211e-07, + "loss": 0.7637, + "step": 4725 + }, + { + "epoch": 0.9, + "learning_rate": 5.647207155765965e-07, + "loss": 0.7849, + "step": 4726 + }, + { + "epoch": 0.9, + "learning_rate": 5.626885138172878e-07, + "loss": 0.7612, + "step": 4727 + }, + { + "epoch": 0.9, + "learning_rate": 5.606598693309073e-07, + "loss": 0.7458, + "step": 4728 + }, + { + "epoch": 0.9, + "learning_rate": 5.586347828821282e-07, + "loss": 0.7907, + "step": 4729 + }, + { + "epoch": 0.9, + "learning_rate": 5.566132552342784e-07, + "loss": 0.8435, + "step": 4730 + }, + { + "epoch": 0.9, + "learning_rate": 5.545952871493476e-07, + "loss": 0.7888, + "step": 4731 + }, + { + "epoch": 0.9, + "learning_rate": 5.525808793879839e-07, + "loss": 0.7566, + "step": 4732 + }, + { + "epoch": 0.9, + "learning_rate": 5.505700327094898e-07, + "loss": 0.7459, + "step": 4733 + }, + { + "epoch": 0.9, + "learning_rate": 5.485627478718314e-07, + "loss": 0.7439, + "step": 4734 + }, + { + "epoch": 0.9, + "learning_rate": 5.465590256316256e-07, + "loss": 0.2646, + "step": 4735 + }, + { + "epoch": 0.9, + "learning_rate": 5.445588667441526e-07, + "loss": 0.8014, + "step": 4736 + }, + { + "epoch": 0.9, + "learning_rate": 5.425622719633428e-07, + "loss": 0.8197, + "step": 4737 + }, + { + "epoch": 0.9, + "learning_rate": 5.405692420417918e-07, + "loss": 0.8062, + "step": 4738 + }, + { + "epoch": 0.9, + "learning_rate": 5.385797777307411e-07, + "loss": 0.7654, + "step": 4739 + }, + { + "epoch": 0.9, + "learning_rate": 5.365938797800996e-07, + "loss": 0.801, + "step": 4740 + }, + { + "epoch": 0.9, + "learning_rate": 5.346115489384196e-07, + "loss": 0.7919, + "step": 4741 + }, + { + "epoch": 0.9, + "learning_rate": 5.326327859529224e-07, + "loss": 0.7097, + "step": 4742 + }, + { + "epoch": 0.9, + "learning_rate": 5.306575915694689e-07, + "loss": 0.7939, + "step": 4743 + }, + { + "epoch": 0.9, + "learning_rate": 5.286859665325905e-07, + "loss": 0.2498, + "step": 4744 + }, + { + "epoch": 0.9, + "learning_rate": 5.267179115854593e-07, + "loss": 0.777, + "step": 4745 + }, + { + "epoch": 0.9, + "learning_rate": 5.24753427469914e-07, + "loss": 0.7491, + "step": 4746 + }, + { + "epoch": 0.9, + "learning_rate": 5.227925149264391e-07, + "loss": 0.7801, + "step": 4747 + }, + { + "epoch": 0.9, + "learning_rate": 5.208351746941753e-07, + "loss": 0.7636, + "step": 4748 + }, + { + "epoch": 0.9, + "learning_rate": 5.188814075109172e-07, + "loss": 0.7683, + "step": 4749 + }, + { + "epoch": 0.9, + "learning_rate": 5.169312141131111e-07, + "loss": 0.7386, + "step": 4750 + }, + { + "epoch": 0.9, + "learning_rate": 5.149845952358589e-07, + "loss": 0.7669, + "step": 4751 + }, + { + "epoch": 0.9, + "learning_rate": 5.130415516129128e-07, + "loss": 0.2492, + "step": 4752 + }, + { + "epoch": 0.9, + "learning_rate": 5.111020839766767e-07, + "loss": 0.7564, + "step": 4753 + }, + { + "epoch": 0.9, + "learning_rate": 5.091661930582103e-07, + "loss": 0.8085, + "step": 4754 + }, + { + "epoch": 0.9, + "learning_rate": 5.072338795872211e-07, + "loss": 0.7745, + "step": 4755 + }, + { + "epoch": 0.9, + "learning_rate": 5.053051442920697e-07, + "loss": 0.7352, + "step": 4756 + }, + { + "epoch": 0.9, + "learning_rate": 5.033799878997692e-07, + "loss": 0.7733, + "step": 4757 + }, + { + "epoch": 0.9, + "learning_rate": 5.014584111359811e-07, + "loss": 0.7147, + "step": 4758 + }, + { + "epoch": 0.9, + "learning_rate": 4.995404147250216e-07, + "loss": 0.7986, + "step": 4759 + }, + { + "epoch": 0.9, + "learning_rate": 4.976259993898503e-07, + "loss": 0.7647, + "step": 4760 + }, + { + "epoch": 0.9, + "learning_rate": 4.957151658520865e-07, + "loss": 0.7864, + "step": 4761 + }, + { + "epoch": 0.9, + "learning_rate": 4.938079148319885e-07, + "loss": 0.7186, + "step": 4762 + }, + { + "epoch": 0.9, + "learning_rate": 4.919042470484759e-07, + "loss": 0.7384, + "step": 4763 + }, + { + "epoch": 0.9, + "learning_rate": 4.900041632191066e-07, + "loss": 0.7579, + "step": 4764 + }, + { + "epoch": 0.9, + "learning_rate": 4.881076640600979e-07, + "loss": 0.754, + "step": 4765 + }, + { + "epoch": 0.9, + "learning_rate": 4.862147502863057e-07, + "loss": 0.7564, + "step": 4766 + }, + { + "epoch": 0.9, + "learning_rate": 4.843254226112437e-07, + "loss": 0.7797, + "step": 4767 + }, + { + "epoch": 0.9, + "learning_rate": 4.824396817470656e-07, + "loss": 0.7915, + "step": 4768 + }, + { + "epoch": 0.9, + "learning_rate": 4.805575284045816e-07, + "loss": 0.6785, + "step": 4769 + }, + { + "epoch": 0.9, + "learning_rate": 4.786789632932398e-07, + "loss": 0.745, + "step": 4770 + }, + { + "epoch": 0.9, + "learning_rate": 4.7680398712114497e-07, + "loss": 0.798, + "step": 4771 + }, + { + "epoch": 0.9, + "learning_rate": 4.7493260059504497e-07, + "loss": 0.7449, + "step": 4772 + }, + { + "epoch": 0.9, + "learning_rate": 4.730648044203334e-07, + "loss": 0.8007, + "step": 4773 + }, + { + "epoch": 0.9, + "learning_rate": 4.7120059930105265e-07, + "loss": 0.7837, + "step": 4774 + }, + { + "epoch": 0.91, + "learning_rate": 4.693399859398906e-07, + "loss": 0.7938, + "step": 4775 + }, + { + "epoch": 0.91, + "learning_rate": 4.674829650381818e-07, + "loss": 0.8046, + "step": 4776 + }, + { + "epoch": 0.91, + "learning_rate": 4.6562953729590745e-07, + "loss": 0.7472, + "step": 4777 + }, + { + "epoch": 0.91, + "learning_rate": 4.6377970341169086e-07, + "loss": 0.7528, + "step": 4778 + }, + { + "epoch": 0.91, + "learning_rate": 4.6193346408280216e-07, + "loss": 0.7424, + "step": 4779 + }, + { + "epoch": 0.91, + "learning_rate": 4.6009082000516346e-07, + "loss": 0.27, + "step": 4780 + }, + { + "epoch": 0.91, + "learning_rate": 4.5825177187332817e-07, + "loss": 0.7274, + "step": 4781 + }, + { + "epoch": 0.91, + "learning_rate": 4.5641632038050943e-07, + "loss": 0.7417, + "step": 4782 + }, + { + "epoch": 0.91, + "learning_rate": 4.5458446621854947e-07, + "loss": 0.7676, + "step": 4783 + }, + { + "epoch": 0.91, + "learning_rate": 4.527562100779481e-07, + "loss": 0.7683, + "step": 4784 + }, + { + "epoch": 0.91, + "learning_rate": 4.509315526478375e-07, + "loss": 0.8004, + "step": 4785 + }, + { + "epoch": 0.91, + "learning_rate": 4.491104946160052e-07, + "loss": 0.7297, + "step": 4786 + }, + { + "epoch": 0.91, + "learning_rate": 4.472930366688688e-07, + "loss": 0.8055, + "step": 4787 + }, + { + "epoch": 0.91, + "learning_rate": 4.4547917949150143e-07, + "loss": 0.8101, + "step": 4788 + }, + { + "epoch": 0.91, + "learning_rate": 4.4366892376760726e-07, + "loss": 0.7448, + "step": 4789 + }, + { + "epoch": 0.91, + "learning_rate": 4.418622701795461e-07, + "loss": 0.7347, + "step": 4790 + }, + { + "epoch": 0.91, + "learning_rate": 4.400592194083064e-07, + "loss": 0.7378, + "step": 4791 + }, + { + "epoch": 0.91, + "learning_rate": 4.38259772133528e-07, + "loss": 0.7547, + "step": 4792 + }, + { + "epoch": 0.91, + "learning_rate": 4.3646392903348823e-07, + "loss": 0.7259, + "step": 4793 + }, + { + "epoch": 0.91, + "learning_rate": 4.3467169078510897e-07, + "loss": 0.2626, + "step": 4794 + }, + { + "epoch": 0.91, + "learning_rate": 4.328830580639487e-07, + "loss": 0.7973, + "step": 4795 + }, + { + "epoch": 0.91, + "learning_rate": 4.3109803154421035e-07, + "loss": 0.8107, + "step": 4796 + }, + { + "epoch": 0.91, + "learning_rate": 4.29316611898738e-07, + "loss": 0.7886, + "step": 4797 + }, + { + "epoch": 0.91, + "learning_rate": 4.2753879979901323e-07, + "loss": 0.7726, + "step": 4798 + }, + { + "epoch": 0.91, + "learning_rate": 4.257645959151602e-07, + "loss": 0.7767, + "step": 4799 + }, + { + "epoch": 0.91, + "learning_rate": 4.2399400091594154e-07, + "loss": 0.2455, + "step": 4800 + }, + { + "epoch": 0.91, + "learning_rate": 4.222270154687602e-07, + "loss": 0.8014, + "step": 4801 + }, + { + "epoch": 0.91, + "learning_rate": 4.204636402396578e-07, + "loss": 0.7217, + "step": 4802 + }, + { + "epoch": 0.91, + "learning_rate": 4.187038758933204e-07, + "loss": 0.7535, + "step": 4803 + }, + { + "epoch": 0.91, + "learning_rate": 4.1694772309306185e-07, + "loss": 0.799, + "step": 4804 + }, + { + "epoch": 0.91, + "learning_rate": 4.151951825008471e-07, + "loss": 0.7549, + "step": 4805 + }, + { + "epoch": 0.91, + "learning_rate": 4.1344625477726994e-07, + "loss": 0.7508, + "step": 4806 + }, + { + "epoch": 0.91, + "learning_rate": 4.117009405815686e-07, + "loss": 0.8353, + "step": 4807 + }, + { + "epoch": 0.91, + "learning_rate": 4.099592405716135e-07, + "loss": 0.7393, + "step": 4808 + }, + { + "epoch": 0.91, + "learning_rate": 4.082211554039206e-07, + "loss": 0.8028, + "step": 4809 + }, + { + "epoch": 0.91, + "learning_rate": 4.064866857336347e-07, + "loss": 0.7443, + "step": 4810 + }, + { + "epoch": 0.91, + "learning_rate": 4.047558322145451e-07, + "loss": 0.7595, + "step": 4811 + }, + { + "epoch": 0.91, + "learning_rate": 4.030285954990698e-07, + "loss": 0.7733, + "step": 4812 + }, + { + "epoch": 0.91, + "learning_rate": 4.013049762382737e-07, + "loss": 0.8287, + "step": 4813 + }, + { + "epoch": 0.91, + "learning_rate": 3.9958497508185036e-07, + "loss": 0.2228, + "step": 4814 + }, + { + "epoch": 0.91, + "learning_rate": 3.978685926781323e-07, + "loss": 0.7382, + "step": 4815 + }, + { + "epoch": 0.91, + "learning_rate": 3.961558296740875e-07, + "loss": 0.7786, + "step": 4816 + }, + { + "epoch": 0.91, + "learning_rate": 3.9444668671532185e-07, + "loss": 0.7945, + "step": 4817 + }, + { + "epoch": 0.91, + "learning_rate": 3.92741164446071e-07, + "loss": 0.751, + "step": 4818 + }, + { + "epoch": 0.91, + "learning_rate": 3.910392635092131e-07, + "loss": 0.7616, + "step": 4819 + }, + { + "epoch": 0.91, + "learning_rate": 3.8934098454625593e-07, + "loss": 0.7658, + "step": 4820 + }, + { + "epoch": 0.91, + "learning_rate": 3.8764632819734526e-07, + "loss": 0.7573, + "step": 4821 + }, + { + "epoch": 0.91, + "learning_rate": 3.8595529510125795e-07, + "loss": 0.7276, + "step": 4822 + }, + { + "epoch": 0.91, + "learning_rate": 3.842678858954085e-07, + "loss": 0.8107, + "step": 4823 + }, + { + "epoch": 0.91, + "learning_rate": 3.8258410121584377e-07, + "loss": 0.7228, + "step": 4824 + }, + { + "epoch": 0.91, + "learning_rate": 3.809039416972449e-07, + "loss": 0.2509, + "step": 4825 + }, + { + "epoch": 0.91, + "learning_rate": 3.7922740797292635e-07, + "loss": 0.7225, + "step": 4826 + }, + { + "epoch": 0.91, + "learning_rate": 3.7755450067483603e-07, + "loss": 0.7444, + "step": 4827 + }, + { + "epoch": 0.92, + "learning_rate": 3.758852204335539e-07, + "loss": 0.7349, + "step": 4828 + }, + { + "epoch": 0.92, + "learning_rate": 3.742195678782934e-07, + "loss": 0.749, + "step": 4829 + }, + { + "epoch": 0.92, + "learning_rate": 3.725575436369033e-07, + "loss": 0.2498, + "step": 4830 + }, + { + "epoch": 0.92, + "learning_rate": 3.7089914833585814e-07, + "loss": 0.7407, + "step": 4831 + }, + { + "epoch": 0.92, + "learning_rate": 3.692443826002734e-07, + "loss": 0.6763, + "step": 4832 + }, + { + "epoch": 0.92, + "learning_rate": 3.6759324705388786e-07, + "loss": 0.7398, + "step": 4833 + }, + { + "epoch": 0.92, + "learning_rate": 3.659457423190782e-07, + "loss": 0.8285, + "step": 4834 + }, + { + "epoch": 0.92, + "learning_rate": 3.643018690168487e-07, + "loss": 0.7955, + "step": 4835 + }, + { + "epoch": 0.92, + "learning_rate": 3.626616277668371e-07, + "loss": 0.7573, + "step": 4836 + }, + { + "epoch": 0.92, + "learning_rate": 3.6102501918731105e-07, + "loss": 0.7172, + "step": 4837 + }, + { + "epoch": 0.92, + "learning_rate": 3.593920438951681e-07, + "loss": 0.7482, + "step": 4838 + }, + { + "epoch": 0.92, + "learning_rate": 3.5776270250593824e-07, + "loss": 0.8069, + "step": 4839 + }, + { + "epoch": 0.92, + "learning_rate": 3.5613699563378014e-07, + "loss": 0.7527, + "step": 4840 + }, + { + "epoch": 0.92, + "learning_rate": 3.545149238914836e-07, + "loss": 0.7184, + "step": 4841 + }, + { + "epoch": 0.92, + "learning_rate": 3.5289648789046616e-07, + "loss": 0.8011, + "step": 4842 + }, + { + "epoch": 0.92, + "learning_rate": 3.512816882407788e-07, + "loss": 0.7491, + "step": 4843 + }, + { + "epoch": 0.92, + "learning_rate": 3.4967052555109793e-07, + "loss": 0.7457, + "step": 4844 + }, + { + "epoch": 0.92, + "learning_rate": 3.480630004287311e-07, + "loss": 0.799, + "step": 4845 + }, + { + "epoch": 0.92, + "learning_rate": 3.4645911347961357e-07, + "loss": 0.7987, + "step": 4846 + }, + { + "epoch": 0.92, + "learning_rate": 3.448588653083107e-07, + "loss": 0.8186, + "step": 4847 + }, + { + "epoch": 0.92, + "learning_rate": 3.432622565180144e-07, + "loss": 0.7634, + "step": 4848 + }, + { + "epoch": 0.92, + "learning_rate": 3.4166928771054653e-07, + "loss": 0.832, + "step": 4849 + }, + { + "epoch": 0.92, + "learning_rate": 3.4007995948635686e-07, + "loss": 0.7638, + "step": 4850 + }, + { + "epoch": 0.92, + "learning_rate": 3.384942724445195e-07, + "loss": 0.7532, + "step": 4851 + }, + { + "epoch": 0.92, + "learning_rate": 3.3691222718274185e-07, + "loss": 0.77, + "step": 4852 + }, + { + "epoch": 0.92, + "learning_rate": 3.3533382429735253e-07, + "loss": 0.2683, + "step": 4853 + }, + { + "epoch": 0.92, + "learning_rate": 3.337590643833144e-07, + "loss": 0.7899, + "step": 4854 + }, + { + "epoch": 0.92, + "learning_rate": 3.321879480342072e-07, + "loss": 0.786, + "step": 4855 + }, + { + "epoch": 0.92, + "learning_rate": 3.3062047584224934e-07, + "loss": 0.7607, + "step": 4856 + }, + { + "epoch": 0.92, + "learning_rate": 3.290566483982727e-07, + "loss": 0.2412, + "step": 4857 + }, + { + "epoch": 0.92, + "learning_rate": 3.274964662917468e-07, + "loss": 0.7331, + "step": 4858 + }, + { + "epoch": 0.92, + "learning_rate": 3.259399301107602e-07, + "loss": 0.7263, + "step": 4859 + }, + { + "epoch": 0.92, + "learning_rate": 3.243870404420302e-07, + "loss": 0.7455, + "step": 4860 + }, + { + "epoch": 0.92, + "learning_rate": 3.228377978708974e-07, + "loss": 0.7912, + "step": 4861 + }, + { + "epoch": 0.92, + "learning_rate": 3.2129220298132926e-07, + "loss": 0.761, + "step": 4862 + }, + { + "epoch": 0.92, + "learning_rate": 3.197502563559185e-07, + "loss": 0.7568, + "step": 4863 + }, + { + "epoch": 0.92, + "learning_rate": 3.182119585758803e-07, + "loss": 0.7361, + "step": 4864 + }, + { + "epoch": 0.92, + "learning_rate": 3.1667731022105854e-07, + "loss": 0.7699, + "step": 4865 + }, + { + "epoch": 0.92, + "learning_rate": 3.1514631186991716e-07, + "loss": 0.7333, + "step": 4866 + }, + { + "epoch": 0.92, + "learning_rate": 3.1361896409954683e-07, + "loss": 0.7636, + "step": 4867 + }, + { + "epoch": 0.92, + "learning_rate": 3.1209526748566146e-07, + "loss": 0.7675, + "step": 4868 + }, + { + "epoch": 0.92, + "learning_rate": 3.105752226025993e-07, + "loss": 0.7424, + "step": 4869 + }, + { + "epoch": 0.92, + "learning_rate": 3.0905883002332213e-07, + "loss": 0.7336, + "step": 4870 + }, + { + "epoch": 0.92, + "learning_rate": 3.075460903194127e-07, + "loss": 0.7669, + "step": 4871 + }, + { + "epoch": 0.92, + "learning_rate": 3.060370040610816e-07, + "loss": 0.7371, + "step": 4872 + }, + { + "epoch": 0.92, + "learning_rate": 3.04531571817156e-07, + "loss": 0.8098, + "step": 4873 + }, + { + "epoch": 0.92, + "learning_rate": 3.030297941550919e-07, + "loss": 0.7004, + "step": 4874 + }, + { + "epoch": 0.92, + "learning_rate": 3.0153167164096444e-07, + "loss": 0.7473, + "step": 4875 + }, + { + "epoch": 0.92, + "learning_rate": 3.0003720483947064e-07, + "loss": 0.783, + "step": 4876 + }, + { + "epoch": 0.92, + "learning_rate": 2.985463943139322e-07, + "loss": 0.7097, + "step": 4877 + }, + { + "epoch": 0.92, + "learning_rate": 2.970592406262884e-07, + "loss": 0.8066, + "step": 4878 + }, + { + "epoch": 0.92, + "learning_rate": 2.9557574433710654e-07, + "loss": 0.7752, + "step": 4879 + }, + { + "epoch": 0.92, + "learning_rate": 2.94095906005567e-07, + "loss": 0.8314, + "step": 4880 + }, + { + "epoch": 0.93, + "learning_rate": 2.926197261894792e-07, + "loss": 0.736, + "step": 4881 + }, + { + "epoch": 0.93, + "learning_rate": 2.911472054452669e-07, + "loss": 0.7473, + "step": 4882 + }, + { + "epoch": 0.93, + "learning_rate": 2.896783443279816e-07, + "loss": 0.7985, + "step": 4883 + }, + { + "epoch": 0.93, + "learning_rate": 2.882131433912883e-07, + "loss": 0.8013, + "step": 4884 + }, + { + "epoch": 0.93, + "learning_rate": 2.867516031874773e-07, + "loss": 0.77, + "step": 4885 + }, + { + "epoch": 0.93, + "learning_rate": 2.8529372426745694e-07, + "loss": 0.7607, + "step": 4886 + }, + { + "epoch": 0.93, + "learning_rate": 2.838395071807554e-07, + "loss": 0.7406, + "step": 4887 + }, + { + "epoch": 0.93, + "learning_rate": 2.823889524755219e-07, + "loss": 0.766, + "step": 4888 + }, + { + "epoch": 0.93, + "learning_rate": 2.809420606985236e-07, + "loss": 0.7945, + "step": 4889 + }, + { + "epoch": 0.93, + "learning_rate": 2.7949883239514863e-07, + "loss": 0.7712, + "step": 4890 + }, + { + "epoch": 0.93, + "learning_rate": 2.7805926810940297e-07, + "loss": 0.7414, + "step": 4891 + }, + { + "epoch": 0.93, + "learning_rate": 2.766233683839115e-07, + "loss": 0.2316, + "step": 4892 + }, + { + "epoch": 0.93, + "learning_rate": 2.7519113375991667e-07, + "loss": 0.8202, + "step": 4893 + }, + { + "epoch": 0.93, + "learning_rate": 2.737625647772857e-07, + "loss": 0.734, + "step": 4894 + }, + { + "epoch": 0.93, + "learning_rate": 2.723376619744933e-07, + "loss": 0.7318, + "step": 4895 + }, + { + "epoch": 0.93, + "learning_rate": 2.709164258886421e-07, + "loss": 0.7297, + "step": 4896 + }, + { + "epoch": 0.93, + "learning_rate": 2.694988570554469e-07, + "loss": 0.7805, + "step": 4897 + }, + { + "epoch": 0.93, + "learning_rate": 2.6808495600924355e-07, + "loss": 0.8283, + "step": 4898 + }, + { + "epoch": 0.93, + "learning_rate": 2.666747232829814e-07, + "loss": 0.7883, + "step": 4899 + }, + { + "epoch": 0.93, + "learning_rate": 2.652681594082329e-07, + "loss": 0.7371, + "step": 4900 + }, + { + "epoch": 0.93, + "learning_rate": 2.638652649151796e-07, + "loss": 0.7901, + "step": 4901 + }, + { + "epoch": 0.93, + "learning_rate": 2.6246604033262955e-07, + "loss": 0.7211, + "step": 4902 + }, + { + "epoch": 0.93, + "learning_rate": 2.610704861879987e-07, + "loss": 0.7627, + "step": 4903 + }, + { + "epoch": 0.93, + "learning_rate": 2.5967860300732617e-07, + "loss": 0.741, + "step": 4904 + }, + { + "epoch": 0.93, + "learning_rate": 2.582903913152612e-07, + "loss": 0.7805, + "step": 4905 + }, + { + "epoch": 0.93, + "learning_rate": 2.569058516350742e-07, + "loss": 0.7858, + "step": 4906 + }, + { + "epoch": 0.93, + "learning_rate": 2.5552498448864873e-07, + "loss": 0.828, + "step": 4907 + }, + { + "epoch": 0.93, + "learning_rate": 2.541477903964862e-07, + "loss": 0.7819, + "step": 4908 + }, + { + "epoch": 0.93, + "learning_rate": 2.527742698777014e-07, + "loss": 0.741, + "step": 4909 + }, + { + "epoch": 0.93, + "learning_rate": 2.5140442345002457e-07, + "loss": 0.7736, + "step": 4910 + }, + { + "epoch": 0.93, + "learning_rate": 2.5003825162980276e-07, + "loss": 0.8044, + "step": 4911 + }, + { + "epoch": 0.93, + "learning_rate": 2.4867575493199515e-07, + "loss": 0.7656, + "step": 4912 + }, + { + "epoch": 0.93, + "learning_rate": 2.473169338701797e-07, + "loss": 0.2634, + "step": 4913 + }, + { + "epoch": 0.93, + "learning_rate": 2.4596178895654355e-07, + "loss": 0.7892, + "step": 4914 + }, + { + "epoch": 0.93, + "learning_rate": 2.446103207018946e-07, + "loss": 0.7428, + "step": 4915 + }, + { + "epoch": 0.93, + "learning_rate": 2.432625296156488e-07, + "loss": 0.2425, + "step": 4916 + }, + { + "epoch": 0.93, + "learning_rate": 2.4191841620583966e-07, + "loss": 0.7834, + "step": 4917 + }, + { + "epoch": 0.93, + "learning_rate": 2.40577980979112e-07, + "loss": 0.2486, + "step": 4918 + }, + { + "epoch": 0.93, + "learning_rate": 2.392412244407294e-07, + "loss": 0.7871, + "step": 4919 + }, + { + "epoch": 0.93, + "learning_rate": 2.379081470945599e-07, + "loss": 0.7339, + "step": 4920 + }, + { + "epoch": 0.93, + "learning_rate": 2.3657874944309377e-07, + "loss": 0.7791, + "step": 4921 + }, + { + "epoch": 0.93, + "learning_rate": 2.352530319874269e-07, + "loss": 0.797, + "step": 4922 + }, + { + "epoch": 0.93, + "learning_rate": 2.3393099522727502e-07, + "loss": 0.7685, + "step": 4923 + }, + { + "epoch": 0.93, + "learning_rate": 2.326126396609596e-07, + "loss": 0.7983, + "step": 4924 + }, + { + "epoch": 0.93, + "learning_rate": 2.3129796578542196e-07, + "loss": 0.7696, + "step": 4925 + }, + { + "epoch": 0.93, + "learning_rate": 2.2998697409620573e-07, + "loss": 0.7681, + "step": 4926 + }, + { + "epoch": 0.93, + "learning_rate": 2.2867966508747787e-07, + "loss": 0.7876, + "step": 4927 + }, + { + "epoch": 0.93, + "learning_rate": 2.2737603925200746e-07, + "loss": 0.8023, + "step": 4928 + }, + { + "epoch": 0.93, + "learning_rate": 2.260760970811826e-07, + "loss": 0.7555, + "step": 4929 + }, + { + "epoch": 0.93, + "learning_rate": 2.24779839064998e-07, + "loss": 0.7965, + "step": 4930 + }, + { + "epoch": 0.93, + "learning_rate": 2.2348726569206168e-07, + "loss": 0.7504, + "step": 4931 + }, + { + "epoch": 0.93, + "learning_rate": 2.2219837744959284e-07, + "loss": 0.7365, + "step": 4932 + }, + { + "epoch": 0.93, + "learning_rate": 2.2091317482342056e-07, + "loss": 0.7967, + "step": 4933 + }, + { + "epoch": 0.94, + "learning_rate": 2.1963165829798626e-07, + "loss": 0.7428, + "step": 4934 + }, + { + "epoch": 0.94, + "learning_rate": 2.1835382835634023e-07, + "loss": 0.7946, + "step": 4935 + }, + { + "epoch": 0.94, + "learning_rate": 2.1707968548014492e-07, + "loss": 0.7484, + "step": 4936 + }, + { + "epoch": 0.94, + "learning_rate": 2.158092301496717e-07, + "loss": 0.7862, + "step": 4937 + }, + { + "epoch": 0.94, + "learning_rate": 2.14542462843802e-07, + "loss": 0.2754, + "step": 4938 + }, + { + "epoch": 0.94, + "learning_rate": 2.1327938404002824e-07, + "loss": 0.8021, + "step": 4939 + }, + { + "epoch": 0.94, + "learning_rate": 2.1201999421445074e-07, + "loss": 0.7457, + "step": 4940 + }, + { + "epoch": 0.94, + "learning_rate": 2.1076429384177975e-07, + "loss": 0.7848, + "step": 4941 + }, + { + "epoch": 0.94, + "learning_rate": 2.095122833953378e-07, + "loss": 0.7919, + "step": 4942 + }, + { + "epoch": 0.94, + "learning_rate": 2.0826396334705067e-07, + "loss": 0.8011, + "step": 4943 + }, + { + "epoch": 0.94, + "learning_rate": 2.070193341674609e-07, + "loss": 0.8254, + "step": 4944 + }, + { + "epoch": 0.94, + "learning_rate": 2.0577839632571094e-07, + "loss": 0.2326, + "step": 4945 + }, + { + "epoch": 0.94, + "learning_rate": 2.0454115028956e-07, + "loss": 0.7694, + "step": 4946 + }, + { + "epoch": 0.94, + "learning_rate": 2.0330759652536835e-07, + "loss": 0.7255, + "step": 4947 + }, + { + "epoch": 0.94, + "learning_rate": 2.0207773549811182e-07, + "loss": 0.7772, + "step": 4948 + }, + { + "epoch": 0.94, + "learning_rate": 2.0085156767136847e-07, + "loss": 0.7643, + "step": 4949 + }, + { + "epoch": 0.94, + "learning_rate": 1.996290935073264e-07, + "loss": 0.7062, + "step": 4950 + }, + { + "epoch": 0.94, + "learning_rate": 1.9841031346678253e-07, + "loss": 0.7291, + "step": 4951 + }, + { + "epoch": 0.94, + "learning_rate": 1.9719522800914048e-07, + "loss": 0.7716, + "step": 4952 + }, + { + "epoch": 0.94, + "learning_rate": 1.9598383759240948e-07, + "loss": 0.7876, + "step": 4953 + }, + { + "epoch": 0.94, + "learning_rate": 1.9477614267320867e-07, + "loss": 0.7378, + "step": 4954 + }, + { + "epoch": 0.94, + "learning_rate": 1.935721437067628e-07, + "loss": 0.7601, + "step": 4955 + }, + { + "epoch": 0.94, + "learning_rate": 1.9237184114690445e-07, + "loss": 0.7946, + "step": 4956 + }, + { + "epoch": 0.94, + "learning_rate": 1.9117523544607052e-07, + "loss": 0.7744, + "step": 4957 + }, + { + "epoch": 0.94, + "learning_rate": 1.8998232705530806e-07, + "loss": 0.7858, + "step": 4958 + }, + { + "epoch": 0.94, + "learning_rate": 1.8879311642426735e-07, + "loss": 0.7536, + "step": 4959 + }, + { + "epoch": 0.94, + "learning_rate": 1.8760760400120648e-07, + "loss": 0.7985, + "step": 4960 + }, + { + "epoch": 0.94, + "learning_rate": 1.8642579023298913e-07, + "loss": 0.7736, + "step": 4961 + }, + { + "epoch": 0.94, + "learning_rate": 1.8524767556508448e-07, + "loss": 0.785, + "step": 4962 + }, + { + "epoch": 0.94, + "learning_rate": 1.8407326044156738e-07, + "loss": 0.7899, + "step": 4963 + }, + { + "epoch": 0.94, + "learning_rate": 1.8290254530511807e-07, + "loss": 0.7321, + "step": 4964 + }, + { + "epoch": 0.94, + "learning_rate": 1.8173553059702476e-07, + "loss": 0.7109, + "step": 4965 + }, + { + "epoch": 0.94, + "learning_rate": 1.8057221675717662e-07, + "loss": 0.7205, + "step": 4966 + }, + { + "epoch": 0.94, + "learning_rate": 1.7941260422406958e-07, + "loss": 0.7843, + "step": 4967 + }, + { + "epoch": 0.94, + "learning_rate": 1.7825669343480624e-07, + "loss": 0.7147, + "step": 4968 + }, + { + "epoch": 0.94, + "learning_rate": 1.7710448482509025e-07, + "loss": 0.7659, + "step": 4969 + }, + { + "epoch": 0.94, + "learning_rate": 1.759559788292331e-07, + "loss": 0.7154, + "step": 4970 + }, + { + "epoch": 0.94, + "learning_rate": 1.7481117588014962e-07, + "loss": 0.7473, + "step": 4971 + }, + { + "epoch": 0.94, + "learning_rate": 1.7367007640935795e-07, + "loss": 0.8246, + "step": 4972 + }, + { + "epoch": 0.94, + "learning_rate": 1.7253268084698184e-07, + "loss": 0.8061, + "step": 4973 + }, + { + "epoch": 0.94, + "learning_rate": 1.7139898962174606e-07, + "loss": 0.842, + "step": 4974 + }, + { + "epoch": 0.94, + "learning_rate": 1.7026900316098217e-07, + "loss": 0.786, + "step": 4975 + }, + { + "epoch": 0.94, + "learning_rate": 1.6914272189062387e-07, + "loss": 0.7412, + "step": 4976 + }, + { + "epoch": 0.94, + "learning_rate": 1.6802014623520823e-07, + "loss": 0.7587, + "step": 4977 + }, + { + "epoch": 0.94, + "learning_rate": 1.6690127661787458e-07, + "loss": 0.7808, + "step": 4978 + }, + { + "epoch": 0.94, + "learning_rate": 1.6578611346036888e-07, + "loss": 0.7657, + "step": 4979 + }, + { + "epoch": 0.94, + "learning_rate": 1.6467465718303488e-07, + "loss": 0.7964, + "step": 4980 + }, + { + "epoch": 0.94, + "learning_rate": 1.63566908204823e-07, + "loss": 0.7439, + "step": 4981 + }, + { + "epoch": 0.94, + "learning_rate": 1.6246286694328594e-07, + "loss": 0.7697, + "step": 4982 + }, + { + "epoch": 0.94, + "learning_rate": 1.6136253381457524e-07, + "loss": 0.7785, + "step": 4983 + }, + { + "epoch": 0.94, + "learning_rate": 1.602659092334502e-07, + "loss": 0.7637, + "step": 4984 + }, + { + "epoch": 0.94, + "learning_rate": 1.5917299361326688e-07, + "loss": 0.7744, + "step": 4985 + }, + { + "epoch": 0.95, + "learning_rate": 1.5808378736598684e-07, + "loss": 0.7967, + "step": 4986 + }, + { + "epoch": 0.95, + "learning_rate": 1.569982909021728e-07, + "loss": 0.7565, + "step": 4987 + }, + { + "epoch": 0.95, + "learning_rate": 1.559165046309863e-07, + "loss": 0.7471, + "step": 4988 + }, + { + "epoch": 0.95, + "learning_rate": 1.5483842896019675e-07, + "loss": 0.7595, + "step": 4989 + }, + { + "epoch": 0.95, + "learning_rate": 1.5376406429616686e-07, + "loss": 0.7872, + "step": 4990 + }, + { + "epoch": 0.95, + "learning_rate": 1.5269341104386824e-07, + "loss": 0.7712, + "step": 4991 + }, + { + "epoch": 0.95, + "learning_rate": 1.5162646960686588e-07, + "loss": 0.7105, + "step": 4992 + }, + { + "epoch": 0.95, + "learning_rate": 1.505632403873336e-07, + "loss": 0.7794, + "step": 4993 + }, + { + "epoch": 0.95, + "learning_rate": 1.4950372378603862e-07, + "loss": 0.7831, + "step": 4994 + }, + { + "epoch": 0.95, + "learning_rate": 1.4844792020235478e-07, + "loss": 0.792, + "step": 4995 + }, + { + "epoch": 0.95, + "learning_rate": 1.473958300342504e-07, + "loss": 0.7608, + "step": 4996 + }, + { + "epoch": 0.95, + "learning_rate": 1.463474536783005e-07, + "loss": 0.7396, + "step": 4997 + }, + { + "epoch": 0.95, + "learning_rate": 1.4530279152967564e-07, + "loss": 0.7501, + "step": 4998 + }, + { + "epoch": 0.95, + "learning_rate": 1.442618439821475e-07, + "loss": 0.7376, + "step": 4999 + }, + { + "epoch": 0.95, + "learning_rate": 1.4322461142808887e-07, + "loss": 0.2721, + "step": 5000 + }, + { + "epoch": 0.95, + "learning_rate": 1.4219109425847033e-07, + "loss": 0.7238, + "step": 5001 + }, + { + "epoch": 0.95, + "learning_rate": 1.4116129286286251e-07, + "loss": 0.8032, + "step": 5002 + }, + { + "epoch": 0.95, + "learning_rate": 1.401352076294371e-07, + "loss": 0.7784, + "step": 5003 + }, + { + "epoch": 0.95, + "learning_rate": 1.3911283894496253e-07, + "loss": 0.7217, + "step": 5004 + }, + { + "epoch": 0.95, + "learning_rate": 1.3809418719480716e-07, + "loss": 0.7217, + "step": 5005 + }, + { + "epoch": 0.95, + "learning_rate": 1.3707925276294054e-07, + "loss": 0.7641, + "step": 5006 + }, + { + "epoch": 0.95, + "learning_rate": 1.3606803603192776e-07, + "loss": 0.7391, + "step": 5007 + }, + { + "epoch": 0.95, + "learning_rate": 1.350605373829339e-07, + "loss": 0.2406, + "step": 5008 + }, + { + "epoch": 0.95, + "learning_rate": 1.3405675719572188e-07, + "loss": 0.7784, + "step": 5009 + }, + { + "epoch": 0.95, + "learning_rate": 1.3305669584865565e-07, + "loss": 0.7838, + "step": 5010 + }, + { + "epoch": 0.95, + "learning_rate": 1.3206035371869373e-07, + "loss": 0.6969, + "step": 5011 + }, + { + "epoch": 0.95, + "learning_rate": 1.3106773118139681e-07, + "loss": 0.8094, + "step": 5012 + }, + { + "epoch": 0.95, + "learning_rate": 1.3007882861091668e-07, + "loss": 0.7466, + "step": 5013 + }, + { + "epoch": 0.95, + "learning_rate": 1.2909364638001298e-07, + "loss": 0.7558, + "step": 5014 + }, + { + "epoch": 0.95, + "learning_rate": 1.2811218486003308e-07, + "loss": 0.8092, + "step": 5015 + }, + { + "epoch": 0.95, + "learning_rate": 1.2713444442092882e-07, + "loss": 0.8022, + "step": 5016 + }, + { + "epoch": 0.95, + "learning_rate": 1.261604254312454e-07, + "loss": 0.7497, + "step": 5017 + }, + { + "epoch": 0.95, + "learning_rate": 1.2519012825812804e-07, + "loss": 0.7762, + "step": 5018 + }, + { + "epoch": 0.95, + "learning_rate": 1.2422355326731527e-07, + "loss": 0.7998, + "step": 5019 + }, + { + "epoch": 0.95, + "learning_rate": 1.232607008231479e-07, + "loss": 0.7528, + "step": 5020 + }, + { + "epoch": 0.95, + "learning_rate": 1.223015712885589e-07, + "loss": 0.7918, + "step": 5021 + }, + { + "epoch": 0.95, + "learning_rate": 1.2134616502508133e-07, + "loss": 0.7409, + "step": 5022 + }, + { + "epoch": 0.95, + "learning_rate": 1.2039448239284156e-07, + "loss": 0.7563, + "step": 5023 + }, + { + "epoch": 0.95, + "learning_rate": 1.1944652375056597e-07, + "loss": 0.7722, + "step": 5024 + }, + { + "epoch": 0.95, + "learning_rate": 1.1850228945557429e-07, + "loss": 0.7222, + "step": 5025 + }, + { + "epoch": 0.95, + "learning_rate": 1.1756177986378293e-07, + "loss": 0.7871, + "step": 5026 + }, + { + "epoch": 0.95, + "learning_rate": 1.1662499532970495e-07, + "loss": 0.7753, + "step": 5027 + }, + { + "epoch": 0.95, + "learning_rate": 1.1569193620645125e-07, + "loss": 0.7829, + "step": 5028 + }, + { + "epoch": 0.95, + "learning_rate": 1.1476260284572382e-07, + "loss": 0.7856, + "step": 5029 + }, + { + "epoch": 0.95, + "learning_rate": 1.1383699559782357e-07, + "loss": 0.7529, + "step": 5030 + }, + { + "epoch": 0.95, + "learning_rate": 1.1291511481164807e-07, + "loss": 0.7488, + "step": 5031 + }, + { + "epoch": 0.95, + "learning_rate": 1.1199696083468492e-07, + "loss": 0.7559, + "step": 5032 + }, + { + "epoch": 0.95, + "learning_rate": 1.1108253401302393e-07, + "loss": 0.7449, + "step": 5033 + }, + { + "epoch": 0.95, + "learning_rate": 1.1017183469134497e-07, + "loss": 0.7812, + "step": 5034 + }, + { + "epoch": 0.95, + "learning_rate": 1.0926486321292451e-07, + "loss": 0.7595, + "step": 5035 + }, + { + "epoch": 0.95, + "learning_rate": 1.0836161991963245e-07, + "loss": 0.7814, + "step": 5036 + }, + { + "epoch": 0.95, + "learning_rate": 1.0746210515193755e-07, + "loss": 0.7282, + "step": 5037 + }, + { + "epoch": 0.95, + "learning_rate": 1.0656631924889749e-07, + "loss": 0.8089, + "step": 5038 + }, + { + "epoch": 0.96, + "learning_rate": 1.0567426254816993e-07, + "loss": 0.8058, + "step": 5039 + }, + { + "epoch": 0.96, + "learning_rate": 1.0478593538600146e-07, + "loss": 0.7804, + "step": 5040 + }, + { + "epoch": 0.96, + "learning_rate": 1.0390133809723758e-07, + "loss": 0.7822, + "step": 5041 + }, + { + "epoch": 0.96, + "learning_rate": 1.0302047101531487e-07, + "loss": 0.7838, + "step": 5042 + }, + { + "epoch": 0.96, + "learning_rate": 1.0214333447226443e-07, + "loss": 0.7723, + "step": 5043 + }, + { + "epoch": 0.96, + "learning_rate": 1.0126992879871289e-07, + "loss": 0.7799, + "step": 5044 + }, + { + "epoch": 0.96, + "learning_rate": 1.0040025432387801e-07, + "loss": 0.7248, + "step": 5045 + }, + { + "epoch": 0.96, + "learning_rate": 9.953431137557201e-08, + "loss": 0.7109, + "step": 5046 + }, + { + "epoch": 0.96, + "learning_rate": 9.86721002802038e-08, + "loss": 0.6918, + "step": 5047 + }, + { + "epoch": 0.96, + "learning_rate": 9.781362136276894e-08, + "loss": 0.7705, + "step": 5048 + }, + { + "epoch": 0.96, + "learning_rate": 9.695887494686307e-08, + "loss": 0.7124, + "step": 5049 + }, + { + "epoch": 0.96, + "learning_rate": 9.610786135466954e-08, + "loss": 0.7475, + "step": 5050 + }, + { + "epoch": 0.96, + "learning_rate": 9.526058090696844e-08, + "loss": 0.8148, + "step": 5051 + }, + { + "epoch": 0.96, + "learning_rate": 9.441703392313095e-08, + "loss": 0.7802, + "step": 5052 + }, + { + "epoch": 0.96, + "learning_rate": 9.35772207211194e-08, + "loss": 0.695, + "step": 5053 + }, + { + "epoch": 0.96, + "learning_rate": 9.274114161749392e-08, + "loss": 0.7456, + "step": 5054 + }, + { + "epoch": 0.96, + "learning_rate": 9.190879692740129e-08, + "loss": 0.7788, + "step": 5055 + }, + { + "epoch": 0.96, + "learning_rate": 9.108018696458498e-08, + "loss": 0.7747, + "step": 5056 + }, + { + "epoch": 0.96, + "learning_rate": 9.025531204137628e-08, + "loss": 0.8003, + "step": 5057 + }, + { + "epoch": 0.96, + "learning_rate": 8.943417246870312e-08, + "loss": 0.6997, + "step": 5058 + }, + { + "epoch": 0.96, + "learning_rate": 8.861676855608237e-08, + "loss": 0.7727, + "step": 5059 + }, + { + "epoch": 0.96, + "learning_rate": 8.780310061162534e-08, + "loss": 0.7236, + "step": 5060 + }, + { + "epoch": 0.96, + "learning_rate": 8.699316894203225e-08, + "loss": 0.6956, + "step": 5061 + }, + { + "epoch": 0.96, + "learning_rate": 8.618697385259778e-08, + "loss": 0.7617, + "step": 5062 + }, + { + "epoch": 0.96, + "learning_rate": 8.53845156472044e-08, + "loss": 0.7871, + "step": 5063 + }, + { + "epoch": 0.96, + "learning_rate": 8.458579462833016e-08, + "loss": 0.7688, + "step": 5064 + }, + { + "epoch": 0.96, + "learning_rate": 8.379081109704312e-08, + "loss": 0.7811, + "step": 5065 + }, + { + "epoch": 0.96, + "learning_rate": 8.299956535300135e-08, + "loss": 0.8042, + "step": 5066 + }, + { + "epoch": 0.96, + "learning_rate": 8.221205769445517e-08, + "loss": 0.7713, + "step": 5067 + }, + { + "epoch": 0.96, + "learning_rate": 8.142828841824379e-08, + "loss": 0.6865, + "step": 5068 + }, + { + "epoch": 0.96, + "learning_rate": 8.06482578198009e-08, + "loss": 0.7418, + "step": 5069 + }, + { + "epoch": 0.96, + "learning_rate": 7.987196619314797e-08, + "loss": 0.8123, + "step": 5070 + }, + { + "epoch": 0.96, + "learning_rate": 7.909941383089869e-08, + "loss": 0.7711, + "step": 5071 + }, + { + "epoch": 0.96, + "learning_rate": 7.833060102425682e-08, + "loss": 0.7709, + "step": 5072 + }, + { + "epoch": 0.96, + "learning_rate": 7.756552806301498e-08, + "loss": 0.7849, + "step": 5073 + }, + { + "epoch": 0.96, + "learning_rate": 7.680419523555916e-08, + "loss": 0.7436, + "step": 5074 + }, + { + "epoch": 0.96, + "learning_rate": 7.604660282886423e-08, + "loss": 0.7724, + "step": 5075 + }, + { + "epoch": 0.96, + "learning_rate": 7.529275112849288e-08, + "loss": 0.7571, + "step": 5076 + }, + { + "epoch": 0.96, + "learning_rate": 7.454264041860115e-08, + "loss": 0.8297, + "step": 5077 + }, + { + "epoch": 0.96, + "learning_rate": 7.379627098193398e-08, + "loss": 0.7541, + "step": 5078 + }, + { + "epoch": 0.96, + "learning_rate": 7.305364309982522e-08, + "loss": 0.7848, + "step": 5079 + }, + { + "epoch": 0.96, + "learning_rate": 7.23147570521987e-08, + "loss": 0.7389, + "step": 5080 + }, + { + "epoch": 0.96, + "learning_rate": 7.157961311756723e-08, + "loss": 0.7806, + "step": 5081 + }, + { + "epoch": 0.96, + "learning_rate": 7.084821157303467e-08, + "loss": 0.7419, + "step": 5082 + }, + { + "epoch": 0.96, + "learning_rate": 7.012055269429496e-08, + "loss": 0.816, + "step": 5083 + }, + { + "epoch": 0.96, + "learning_rate": 6.939663675562758e-08, + "loss": 0.7778, + "step": 5084 + }, + { + "epoch": 0.96, + "learning_rate": 6.867646402990425e-08, + "loss": 0.7469, + "step": 5085 + }, + { + "epoch": 0.96, + "learning_rate": 6.796003478858337e-08, + "loss": 0.7447, + "step": 5086 + }, + { + "epoch": 0.96, + "learning_rate": 6.724734930171561e-08, + "loss": 0.7178, + "step": 5087 + }, + { + "epoch": 0.96, + "learning_rate": 6.65384078379383e-08, + "loss": 0.7806, + "step": 5088 + }, + { + "epoch": 0.96, + "learning_rate": 6.583321066447657e-08, + "loss": 0.8163, + "step": 5089 + }, + { + "epoch": 0.96, + "learning_rate": 6.513175804714666e-08, + "loss": 0.7958, + "step": 5090 + }, + { + "epoch": 0.96, + "learning_rate": 6.443405025035155e-08, + "loss": 0.8033, + "step": 5091 + }, + { + "epoch": 0.97, + "learning_rate": 6.374008753708305e-08, + "loss": 0.7138, + "step": 5092 + }, + { + "epoch": 0.97, + "learning_rate": 6.304987016892194e-08, + "loss": 0.7777, + "step": 5093 + }, + { + "epoch": 0.97, + "learning_rate": 6.236339840603677e-08, + "loss": 0.7644, + "step": 5094 + }, + { + "epoch": 0.97, + "learning_rate": 6.16806725071828e-08, + "loss": 0.7973, + "step": 5095 + }, + { + "epoch": 0.97, + "learning_rate": 6.100169272970635e-08, + "loss": 0.7208, + "step": 5096 + }, + { + "epoch": 0.97, + "learning_rate": 6.032645932954052e-08, + "loss": 0.7621, + "step": 5097 + }, + { + "epoch": 0.97, + "learning_rate": 5.96549725612039e-08, + "loss": 0.759, + "step": 5098 + }, + { + "epoch": 0.97, + "learning_rate": 5.898723267780626e-08, + "loss": 0.7248, + "step": 5099 + }, + { + "epoch": 0.97, + "learning_rate": 5.832323993104294e-08, + "loss": 0.2758, + "step": 5100 + }, + { + "epoch": 0.97, + "learning_rate": 5.766299457119817e-08, + "loss": 0.7511, + "step": 5101 + }, + { + "epoch": 0.97, + "learning_rate": 5.7006496847141765e-08, + "loss": 0.7292, + "step": 5102 + }, + { + "epoch": 0.97, + "learning_rate": 5.635374700633245e-08, + "loss": 0.731, + "step": 5103 + }, + { + "epoch": 0.97, + "learning_rate": 5.5704745294815624e-08, + "loss": 0.7644, + "step": 5104 + }, + { + "epoch": 0.97, + "learning_rate": 5.505949195722671e-08, + "loss": 0.7524, + "step": 5105 + }, + { + "epoch": 0.97, + "learning_rate": 5.4417987236781156e-08, + "loss": 0.2552, + "step": 5106 + }, + { + "epoch": 0.97, + "learning_rate": 5.3780231375289984e-08, + "loss": 0.7835, + "step": 5107 + }, + { + "epoch": 0.97, + "learning_rate": 5.3146224613144225e-08, + "loss": 0.735, + "step": 5108 + }, + { + "epoch": 0.97, + "learning_rate": 5.2515967189326056e-08, + "loss": 0.7548, + "step": 5109 + }, + { + "epoch": 0.97, + "learning_rate": 5.18894593414021e-08, + "loss": 0.7446, + "step": 5110 + }, + { + "epoch": 0.97, + "learning_rate": 5.126670130552791e-08, + "loss": 0.8118, + "step": 5111 + }, + { + "epoch": 0.97, + "learning_rate": 5.064769331644126e-08, + "loss": 0.7975, + "step": 5112 + }, + { + "epoch": 0.97, + "learning_rate": 5.0032435607472175e-08, + "loss": 0.7835, + "step": 5113 + }, + { + "epoch": 0.97, + "learning_rate": 4.942092841053292e-08, + "loss": 0.7637, + "step": 5114 + }, + { + "epoch": 0.97, + "learning_rate": 4.8813171956123565e-08, + "loss": 0.7778, + "step": 5115 + }, + { + "epoch": 0.97, + "learning_rate": 4.820916647332863e-08, + "loss": 0.7593, + "step": 5116 + }, + { + "epoch": 0.97, + "learning_rate": 4.760891218982156e-08, + "loss": 0.8082, + "step": 5117 + }, + { + "epoch": 0.97, + "learning_rate": 4.7012409331861354e-08, + "loss": 0.7758, + "step": 5118 + }, + { + "epoch": 0.97, + "learning_rate": 4.641965812429039e-08, + "loss": 0.2929, + "step": 5119 + }, + { + "epoch": 0.97, + "learning_rate": 4.5830658790538826e-08, + "loss": 0.7228, + "step": 5120 + }, + { + "epoch": 0.97, + "learning_rate": 4.524541155262241e-08, + "loss": 0.7349, + "step": 5121 + }, + { + "epoch": 0.97, + "learning_rate": 4.4663916631143554e-08, + "loss": 0.7322, + "step": 5122 + }, + { + "epoch": 0.97, + "learning_rate": 4.408617424528805e-08, + "loss": 0.7245, + "step": 5123 + }, + { + "epoch": 0.97, + "learning_rate": 4.351218461282947e-08, + "loss": 0.7527, + "step": 5124 + }, + { + "epoch": 0.97, + "learning_rate": 4.294194795012363e-08, + "loss": 0.7009, + "step": 5125 + }, + { + "epoch": 0.97, + "learning_rate": 4.2375464472117487e-08, + "loss": 0.7963, + "step": 5126 + }, + { + "epoch": 0.97, + "learning_rate": 4.1812734392336906e-08, + "loss": 0.682, + "step": 5127 + }, + { + "epoch": 0.97, + "learning_rate": 4.125375792289665e-08, + "loss": 0.6815, + "step": 5128 + }, + { + "epoch": 0.97, + "learning_rate": 4.069853527449596e-08, + "loss": 0.7621, + "step": 5129 + }, + { + "epoch": 0.97, + "learning_rate": 4.014706665641854e-08, + "loss": 0.7596, + "step": 5130 + }, + { + "epoch": 0.97, + "learning_rate": 3.9599352276532556e-08, + "loss": 0.7786, + "step": 5131 + }, + { + "epoch": 0.97, + "learning_rate": 3.905539234129396e-08, + "loss": 0.7594, + "step": 5132 + }, + { + "epoch": 0.97, + "learning_rate": 3.851518705573876e-08, + "loss": 0.2578, + "step": 5133 + }, + { + "epoch": 0.97, + "learning_rate": 3.797873662349294e-08, + "loss": 0.7953, + "step": 5134 + }, + { + "epoch": 0.97, + "learning_rate": 3.744604124676254e-08, + "loss": 0.7593, + "step": 5135 + }, + { + "epoch": 0.97, + "learning_rate": 3.691710112634139e-08, + "loss": 0.7599, + "step": 5136 + }, + { + "epoch": 0.97, + "learning_rate": 3.639191646160667e-08, + "loss": 0.7503, + "step": 5137 + }, + { + "epoch": 0.97, + "learning_rate": 3.587048745051891e-08, + "loss": 0.8418, + "step": 5138 + }, + { + "epoch": 0.97, + "learning_rate": 3.535281428962534e-08, + "loss": 0.7524, + "step": 5139 + }, + { + "epoch": 0.97, + "learning_rate": 3.483889717405542e-08, + "loss": 0.7793, + "step": 5140 + }, + { + "epoch": 0.97, + "learning_rate": 3.43287362975242e-08, + "loss": 0.7552, + "step": 5141 + }, + { + "epoch": 0.97, + "learning_rate": 3.382233185233008e-08, + "loss": 0.7333, + "step": 5142 + }, + { + "epoch": 0.97, + "learning_rate": 3.3319684029354815e-08, + "loss": 0.7511, + "step": 5143 + }, + { + "epoch": 0.97, + "learning_rate": 3.2820793018065734e-08, + "loss": 0.7576, + "step": 5144 + }, + { + "epoch": 0.98, + "learning_rate": 3.232565900651352e-08, + "loss": 0.7546, + "step": 5145 + }, + { + "epoch": 0.98, + "learning_rate": 3.183428218133222e-08, + "loss": 0.8305, + "step": 5146 + }, + { + "epoch": 0.98, + "learning_rate": 3.134666272774034e-08, + "loss": 0.7607, + "step": 5147 + }, + { + "epoch": 0.98, + "learning_rate": 3.086280082953863e-08, + "loss": 0.7322, + "step": 5148 + }, + { + "epoch": 0.98, + "learning_rate": 3.0382696669114534e-08, + "loss": 0.7833, + "step": 5149 + }, + { + "epoch": 0.98, + "learning_rate": 2.9906350427435505e-08, + "loss": 0.7063, + "step": 5150 + }, + { + "epoch": 0.98, + "learning_rate": 2.94337622840557e-08, + "loss": 0.7108, + "step": 5151 + }, + { + "epoch": 0.98, + "learning_rate": 2.896493241710929e-08, + "loss": 0.7631, + "step": 5152 + }, + { + "epoch": 0.98, + "learning_rate": 2.8499861003317142e-08, + "loss": 0.8097, + "step": 5153 + }, + { + "epoch": 0.98, + "learning_rate": 2.8038548217980133e-08, + "loss": 0.7513, + "step": 5154 + }, + { + "epoch": 0.98, + "learning_rate": 2.7580994234985838e-08, + "loss": 0.7083, + "step": 5155 + }, + { + "epoch": 0.98, + "learning_rate": 2.7127199226802957e-08, + "loss": 0.7735, + "step": 5156 + }, + { + "epoch": 0.98, + "learning_rate": 2.667716336448356e-08, + "loss": 0.8057, + "step": 5157 + }, + { + "epoch": 0.98, + "learning_rate": 2.623088681766195e-08, + "loss": 0.7733, + "step": 5158 + }, + { + "epoch": 0.98, + "learning_rate": 2.5788369754558007e-08, + "loss": 0.7302, + "step": 5159 + }, + { + "epoch": 0.98, + "learning_rate": 2.534961234197275e-08, + "loss": 0.7919, + "step": 5160 + }, + { + "epoch": 0.98, + "learning_rate": 2.4914614745288333e-08, + "loss": 0.7378, + "step": 5161 + }, + { + "epoch": 0.98, + "learning_rate": 2.4483377128473596e-08, + "loss": 0.7412, + "step": 5162 + }, + { + "epoch": 0.98, + "learning_rate": 2.4055899654076286e-08, + "loss": 0.7497, + "step": 5163 + }, + { + "epoch": 0.98, + "learning_rate": 2.3632182483228628e-08, + "loss": 0.7495, + "step": 5164 + }, + { + "epoch": 0.98, + "learning_rate": 2.321222577564619e-08, + "loss": 0.7477, + "step": 5165 + }, + { + "epoch": 0.98, + "learning_rate": 2.2796029689626797e-08, + "loss": 0.7913, + "step": 5166 + }, + { + "epoch": 0.98, + "learning_rate": 2.238359438204829e-08, + "loss": 0.2537, + "step": 5167 + }, + { + "epoch": 0.98, + "learning_rate": 2.19749200083752e-08, + "loss": 0.6612, + "step": 5168 + }, + { + "epoch": 0.98, + "learning_rate": 2.157000672264986e-08, + "loss": 0.7446, + "step": 5169 + }, + { + "epoch": 0.98, + "learning_rate": 2.1168854677500182e-08, + "loss": 0.7337, + "step": 5170 + }, + { + "epoch": 0.98, + "learning_rate": 2.077146402413521e-08, + "loss": 0.8191, + "step": 5171 + }, + { + "epoch": 0.98, + "learning_rate": 2.037783491234735e-08, + "loss": 0.7632, + "step": 5172 + }, + { + "epoch": 0.98, + "learning_rate": 1.99879674905068e-08, + "loss": 0.7998, + "step": 5173 + }, + { + "epoch": 0.98, + "learning_rate": 1.9601861905572673e-08, + "loss": 0.288, + "step": 5174 + }, + { + "epoch": 0.98, + "learning_rate": 1.9219518303079666e-08, + "loss": 0.8337, + "step": 5175 + }, + { + "epoch": 0.98, + "learning_rate": 1.884093682714916e-08, + "loss": 0.7819, + "step": 5176 + }, + { + "epoch": 0.98, + "learning_rate": 1.846611762048256e-08, + "loss": 0.8051, + "step": 5177 + }, + { + "epoch": 0.98, + "learning_rate": 1.80950608243613e-08, + "loss": 0.7735, + "step": 5178 + }, + { + "epoch": 0.98, + "learning_rate": 1.7727766578652384e-08, + "loss": 0.2662, + "step": 5179 + }, + { + "epoch": 0.98, + "learning_rate": 1.7364235021801733e-08, + "loss": 0.7747, + "step": 5180 + }, + { + "epoch": 0.98, + "learning_rate": 1.7004466290837518e-08, + "loss": 0.7765, + "step": 5181 + }, + { + "epoch": 0.98, + "learning_rate": 1.6648460521371257e-08, + "loss": 0.8174, + "step": 5182 + }, + { + "epoch": 0.98, + "learning_rate": 1.6296217847593387e-08, + "loss": 0.7677, + "step": 5183 + }, + { + "epoch": 0.98, + "learning_rate": 1.594773840227881e-08, + "loss": 0.7589, + "step": 5184 + }, + { + "epoch": 0.98, + "learning_rate": 1.5603022316780235e-08, + "loss": 0.8075, + "step": 5185 + }, + { + "epoch": 0.98, + "learning_rate": 1.5262069721034834e-08, + "loss": 0.7096, + "step": 5186 + }, + { + "epoch": 0.98, + "learning_rate": 1.492488074356202e-08, + "loss": 0.7915, + "step": 5187 + }, + { + "epoch": 0.98, + "learning_rate": 1.4591455511457908e-08, + "loss": 0.723, + "step": 5188 + }, + { + "epoch": 0.98, + "learning_rate": 1.4261794150405294e-08, + "loss": 0.2774, + "step": 5189 + }, + { + "epoch": 0.98, + "learning_rate": 1.3935896784663671e-08, + "loss": 0.752, + "step": 5190 + }, + { + "epoch": 0.98, + "learning_rate": 1.3613763537078106e-08, + "loss": 0.2539, + "step": 5191 + }, + { + "epoch": 0.98, + "learning_rate": 1.329539452907036e-08, + "loss": 0.7527, + "step": 5192 + }, + { + "epoch": 0.98, + "learning_rate": 1.298078988064666e-08, + "loss": 0.7866, + "step": 5193 + }, + { + "epoch": 0.98, + "learning_rate": 1.2669949710393259e-08, + "loss": 0.8146, + "step": 5194 + }, + { + "epoch": 0.98, + "learning_rate": 1.2362874135477542e-08, + "loss": 0.7525, + "step": 5195 + }, + { + "epoch": 0.98, + "learning_rate": 1.2059563271646924e-08, + "loss": 0.7983, + "step": 5196 + }, + { + "epoch": 0.99, + "learning_rate": 1.1760017233232168e-08, + "loss": 0.7512, + "step": 5197 + }, + { + "epoch": 0.99, + "learning_rate": 1.1464236133140738e-08, + "loss": 0.8272, + "step": 5198 + }, + { + "epoch": 0.99, + "learning_rate": 1.117222008286456e-08, + "loss": 0.2371, + "step": 5199 + }, + { + "epoch": 0.99, + "learning_rate": 1.0883969192476695e-08, + "loss": 0.8321, + "step": 5200 + }, + { + "epoch": 0.99, + "learning_rate": 1.059948357062801e-08, + "loss": 0.2675, + "step": 5201 + }, + { + "epoch": 0.99, + "learning_rate": 1.0318763324551618e-08, + "loss": 0.7476, + "step": 5202 + }, + { + "epoch": 0.99, + "learning_rate": 1.0041808560062871e-08, + "loss": 0.2563, + "step": 5203 + }, + { + "epoch": 0.99, + "learning_rate": 9.768619381554934e-09, + "loss": 0.7958, + "step": 5204 + }, + { + "epoch": 0.99, + "learning_rate": 9.49919589200321e-09, + "loss": 0.7847, + "step": 5205 + }, + { + "epoch": 0.99, + "learning_rate": 9.233538192963132e-09, + "loss": 0.2445, + "step": 5206 + }, + { + "epoch": 0.99, + "learning_rate": 8.971646384572374e-09, + "loss": 0.6631, + "step": 5207 + }, + { + "epoch": 0.99, + "learning_rate": 8.713520565546418e-09, + "loss": 0.7585, + "step": 5208 + }, + { + "epoch": 0.99, + "learning_rate": 8.459160833182989e-09, + "loss": 0.7551, + "step": 5209 + }, + { + "epoch": 0.99, + "learning_rate": 8.208567283358726e-09, + "loss": 0.7304, + "step": 5210 + }, + { + "epoch": 0.99, + "learning_rate": 7.961740010532514e-09, + "loss": 0.7706, + "step": 5211 + }, + { + "epoch": 0.99, + "learning_rate": 7.718679107743266e-09, + "loss": 0.7419, + "step": 5212 + }, + { + "epoch": 0.99, + "learning_rate": 7.479384666608802e-09, + "loss": 0.7768, + "step": 5213 + }, + { + "epoch": 0.99, + "learning_rate": 7.2438567773280846e-09, + "loss": 0.8025, + "step": 5214 + }, + { + "epoch": 0.99, + "learning_rate": 7.012095528680096e-09, + "loss": 0.6736, + "step": 5215 + }, + { + "epoch": 0.99, + "learning_rate": 6.784101008024957e-09, + "loss": 0.7371, + "step": 5216 + }, + { + "epoch": 0.99, + "learning_rate": 6.5598733013017e-09, + "loss": 0.7674, + "step": 5217 + }, + { + "epoch": 0.99, + "learning_rate": 6.339412493031604e-09, + "loss": 0.8511, + "step": 5218 + }, + { + "epoch": 0.99, + "learning_rate": 6.122718666311534e-09, + "loss": 0.798, + "step": 5219 + }, + { + "epoch": 0.99, + "learning_rate": 5.909791902823925e-09, + "loss": 0.7577, + "step": 5220 + }, + { + "epoch": 0.99, + "learning_rate": 5.700632282827911e-09, + "loss": 0.7768, + "step": 5221 + }, + { + "epoch": 0.99, + "learning_rate": 5.4952398851637614e-09, + "loss": 0.7692, + "step": 5222 + }, + { + "epoch": 0.99, + "learning_rate": 5.293614787250656e-09, + "loss": 0.7926, + "step": 5223 + }, + { + "epoch": 0.99, + "learning_rate": 5.0957570650900236e-09, + "loss": 0.844, + "step": 5224 + }, + { + "epoch": 0.99, + "learning_rate": 4.901666793261095e-09, + "loss": 0.7838, + "step": 5225 + }, + { + "epoch": 0.99, + "learning_rate": 4.711344044923127e-09, + "loss": 0.7626, + "step": 5226 + }, + { + "epoch": 0.99, + "learning_rate": 4.524788891816512e-09, + "loss": 0.7325, + "step": 5227 + }, + { + "epoch": 0.99, + "learning_rate": 4.342001404260554e-09, + "loss": 0.7225, + "step": 5228 + }, + { + "epoch": 0.99, + "learning_rate": 4.162981651154585e-09, + "loss": 0.7194, + "step": 5229 + }, + { + "epoch": 0.99, + "learning_rate": 3.98772969997796e-09, + "loss": 0.766, + "step": 5230 + }, + { + "epoch": 0.99, + "learning_rate": 3.81624561678895e-09, + "loss": 0.6795, + "step": 5231 + }, + { + "epoch": 0.99, + "learning_rate": 3.648529466228068e-09, + "loss": 0.7953, + "step": 5232 + }, + { + "epoch": 0.99, + "learning_rate": 3.4845813115114147e-09, + "loss": 0.7791, + "step": 5233 + }, + { + "epoch": 0.99, + "learning_rate": 3.3244012144395545e-09, + "loss": 0.7666, + "step": 5234 + }, + { + "epoch": 0.99, + "learning_rate": 3.1679892353886354e-09, + "loss": 0.7075, + "step": 5235 + }, + { + "epoch": 0.99, + "learning_rate": 3.0153454333159417e-09, + "loss": 0.2496, + "step": 5236 + }, + { + "epoch": 0.99, + "learning_rate": 2.866469865759891e-09, + "loss": 0.7318, + "step": 5237 + }, + { + "epoch": 0.99, + "learning_rate": 2.7213625888355964e-09, + "loss": 0.7643, + "step": 5238 + }, + { + "epoch": 0.99, + "learning_rate": 2.580023657241526e-09, + "loss": 0.7857, + "step": 5239 + }, + { + "epoch": 0.99, + "learning_rate": 2.4424531242517313e-09, + "loss": 0.7994, + "step": 5240 + }, + { + "epoch": 0.99, + "learning_rate": 2.3086510417225093e-09, + "loss": 0.757, + "step": 5241 + }, + { + "epoch": 0.99, + "learning_rate": 2.178617460087962e-09, + "loss": 0.7795, + "step": 5242 + }, + { + "epoch": 0.99, + "learning_rate": 2.0523524283644345e-09, + "loss": 0.7452, + "step": 5243 + }, + { + "epoch": 0.99, + "learning_rate": 1.929855994143859e-09, + "loss": 0.7615, + "step": 5244 + }, + { + "epoch": 0.99, + "learning_rate": 1.8111282036004097e-09, + "loss": 0.7737, + "step": 5245 + }, + { + "epoch": 0.99, + "learning_rate": 1.6961691014871773e-09, + "loss": 0.7771, + "step": 5246 + }, + { + "epoch": 0.99, + "learning_rate": 1.5849787311361665e-09, + "loss": 0.7614, + "step": 5247 + }, + { + "epoch": 0.99, + "learning_rate": 1.4775571344605167e-09, + "loss": 0.7046, + "step": 5248 + }, + { + "epoch": 0.99, + "learning_rate": 1.3739043519489514e-09, + "loss": 0.7506, + "step": 5249 + }, + { + "epoch": 1.0, + "learning_rate": 1.2740204226746599e-09, + "loss": 0.7766, + "step": 5250 + }, + { + "epoch": 1.0, + "learning_rate": 1.177905384285305e-09, + "loss": 0.7316, + "step": 5251 + }, + { + "epoch": 1.0, + "learning_rate": 1.0855592730119046e-09, + "loss": 0.7325, + "step": 5252 + }, + { + "epoch": 1.0, + "learning_rate": 9.969821236621713e-10, + "loss": 0.7487, + "step": 5253 + }, + { + "epoch": 1.0, + "learning_rate": 9.12173969624952e-10, + "loss": 0.7467, + "step": 5254 + }, + { + "epoch": 1.0, + "learning_rate": 8.311348428657884e-10, + "loss": 0.793, + "step": 5255 + }, + { + "epoch": 1.0, + "learning_rate": 7.538647739335769e-10, + "loss": 0.7935, + "step": 5256 + }, + { + "epoch": 1.0, + "learning_rate": 6.803637919539086e-10, + "loss": 0.8005, + "step": 5257 + }, + { + "epoch": 1.0, + "learning_rate": 6.106319246301784e-10, + "loss": 0.7358, + "step": 5258 + }, + { + "epoch": 1.0, + "learning_rate": 5.44669198249137e-10, + "loss": 0.7436, + "step": 5259 + }, + { + "epoch": 1.0, + "learning_rate": 4.824756376731188e-10, + "loss": 0.774, + "step": 5260 + }, + { + "epoch": 1.0, + "learning_rate": 4.240512663455931e-10, + "loss": 0.7798, + "step": 5261 + }, + { + "epoch": 1.0, + "learning_rate": 3.6939610628894396e-10, + "loss": 0.7494, + "step": 5262 + }, + { + "epoch": 1.0, + "learning_rate": 3.1851017810557994e-10, + "loss": 0.7681, + "step": 5263 + }, + { + "epoch": 1.0, + "learning_rate": 2.713935009746038e-10, + "loss": 0.7919, + "step": 5264 + }, + { + "epoch": 1.0, + "learning_rate": 2.2804609265736356e-10, + "loss": 0.7917, + "step": 5265 + }, + { + "epoch": 1.0, + "learning_rate": 1.8846796949190116e-10, + "loss": 0.7557, + "step": 5266 + }, + { + "epoch": 1.0, + "learning_rate": 1.5265914639850388e-10, + "loss": 0.7681, + "step": 5267 + }, + { + "epoch": 1.0, + "learning_rate": 1.2061963687304278e-10, + "loss": 0.2689, + "step": 5268 + }, + { + "epoch": 1.0, + "learning_rate": 9.234945299363418e-11, + "loss": 0.7177, + "step": 5269 + }, + { + "epoch": 1.0, + "learning_rate": 6.784860541508841e-11, + "loss": 0.7812, + "step": 5270 + }, + { + "epoch": 1.0, + "learning_rate": 4.711710337446107e-11, + "loss": 0.7706, + "step": 5271 + }, + { + "epoch": 1.0, + "learning_rate": 3.0154954684391555e-11, + "loss": 0.7805, + "step": 5272 + }, + { + "epoch": 1.0, + "learning_rate": 1.6962165739764503e-11, + "loss": 0.7566, + "step": 5273 + }, + { + "epoch": 1.0, + "learning_rate": 7.538741513268833e-12, + "loss": 0.7416, + "step": 5274 + }, + { + "epoch": 1.0, + "learning_rate": 1.8846855553977804e-12, + "loss": 0.524, + "step": 5275 + }, + { + "epoch": 1.0, + "learning_rate": 0.0, + "loss": 0.3805, + "step": 5276 + }, + { + "epoch": 1.0, + "step": 5276, + "total_flos": 2179610069041152.0, + "train_loss": 0.7820551202401598, + "train_runtime": 66149.6769, + "train_samples_per_second": 10.209, + "train_steps_per_second": 0.08 + } + ], + "logging_steps": 1.0, + "max_steps": 5276, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 50000, + "total_flos": 2179610069041152.0, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +}