{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9998920902125823, "eval_steps": 500, "global_step": 4633, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1.4388489208633095e-07, "loss": 0.3481, "step": 1 }, { "epoch": 0.0, "learning_rate": 2.877697841726619e-07, "loss": 0.548, "step": 2 }, { "epoch": 0.0, "learning_rate": 4.3165467625899287e-07, "loss": 0.3656, "step": 3 }, { "epoch": 0.0, "learning_rate": 5.755395683453238e-07, "loss": 0.3005, "step": 4 }, { "epoch": 0.0, "learning_rate": 7.194244604316547e-07, "loss": 0.2845, "step": 5 }, { "epoch": 0.0, "learning_rate": 8.633093525179857e-07, "loss": 0.3102, "step": 6 }, { "epoch": 0.0, "learning_rate": 1.0071942446043167e-06, "loss": 0.357, "step": 7 }, { "epoch": 0.0, "learning_rate": 1.1510791366906476e-06, "loss": 0.4875, "step": 8 }, { "epoch": 0.0, "learning_rate": 1.2949640287769785e-06, "loss": 0.3553, "step": 9 }, { "epoch": 0.0, "learning_rate": 1.4388489208633094e-06, "loss": 0.4387, "step": 10 }, { "epoch": 0.0, "learning_rate": 1.5827338129496403e-06, "loss": 0.2813, "step": 11 }, { "epoch": 0.0, "learning_rate": 1.7266187050359715e-06, "loss": 0.4425, "step": 12 }, { "epoch": 0.0, "learning_rate": 1.8705035971223024e-06, "loss": 0.3725, "step": 13 }, { "epoch": 0.0, "learning_rate": 2.0143884892086333e-06, "loss": 0.386, "step": 14 }, { "epoch": 0.0, "learning_rate": 2.158273381294964e-06, "loss": 0.4263, "step": 15 }, { "epoch": 0.0, "learning_rate": 2.302158273381295e-06, "loss": 0.3402, "step": 16 }, { "epoch": 0.0, "learning_rate": 2.4460431654676263e-06, "loss": 0.5059, "step": 17 }, { "epoch": 0.0, "learning_rate": 2.589928057553957e-06, "loss": 0.3659, "step": 18 }, { "epoch": 0.0, "learning_rate": 2.733812949640288e-06, "loss": 0.3581, "step": 19 }, { "epoch": 0.0, "learning_rate": 2.877697841726619e-06, "loss": 0.3458, "step": 20 }, { "epoch": 0.0, "learning_rate": 3.02158273381295e-06, "loss": 0.3778, "step": 21 }, { "epoch": 0.0, "learning_rate": 3.1654676258992807e-06, "loss": 0.3387, "step": 22 }, { "epoch": 0.0, "learning_rate": 3.309352517985612e-06, "loss": 0.3217, "step": 23 }, { "epoch": 0.01, "learning_rate": 3.453237410071943e-06, "loss": 0.3144, "step": 24 }, { "epoch": 0.01, "learning_rate": 3.5971223021582737e-06, "loss": 0.372, "step": 25 }, { "epoch": 0.01, "learning_rate": 3.741007194244605e-06, "loss": 0.4241, "step": 26 }, { "epoch": 0.01, "learning_rate": 3.884892086330936e-06, "loss": 0.3704, "step": 27 }, { "epoch": 0.01, "learning_rate": 4.028776978417267e-06, "loss": 0.3076, "step": 28 }, { "epoch": 0.01, "learning_rate": 4.172661870503597e-06, "loss": 0.3916, "step": 29 }, { "epoch": 0.01, "learning_rate": 4.316546762589928e-06, "loss": 0.3251, "step": 30 }, { "epoch": 0.01, "learning_rate": 4.46043165467626e-06, "loss": 0.3545, "step": 31 }, { "epoch": 0.01, "learning_rate": 4.60431654676259e-06, "loss": 0.2904, "step": 32 }, { "epoch": 0.01, "learning_rate": 4.748201438848921e-06, "loss": 0.3587, "step": 33 }, { "epoch": 0.01, "learning_rate": 4.892086330935253e-06, "loss": 0.3034, "step": 34 }, { "epoch": 0.01, "learning_rate": 5.035971223021583e-06, "loss": 0.232, "step": 35 }, { "epoch": 0.01, "learning_rate": 5.179856115107914e-06, "loss": 0.3345, "step": 36 }, { "epoch": 0.01, "learning_rate": 5.3237410071942456e-06, "loss": 0.343, "step": 37 }, { "epoch": 0.01, "learning_rate": 5.467625899280576e-06, "loss": 0.2894, "step": 38 }, { "epoch": 0.01, "learning_rate": 5.611510791366906e-06, "loss": 0.296, "step": 39 }, { "epoch": 0.01, "learning_rate": 5.755395683453238e-06, "loss": 0.3097, "step": 40 }, { "epoch": 0.01, "learning_rate": 5.899280575539568e-06, "loss": 0.3908, "step": 41 }, { "epoch": 0.01, "learning_rate": 6.0431654676259e-06, "loss": 0.2902, "step": 42 }, { "epoch": 0.01, "learning_rate": 6.1870503597122315e-06, "loss": 0.2441, "step": 43 }, { "epoch": 0.01, "learning_rate": 6.330935251798561e-06, "loss": 0.2881, "step": 44 }, { "epoch": 0.01, "learning_rate": 6.474820143884892e-06, "loss": 0.3347, "step": 45 }, { "epoch": 0.01, "learning_rate": 6.618705035971224e-06, "loss": 0.3842, "step": 46 }, { "epoch": 0.01, "learning_rate": 6.762589928057554e-06, "loss": 0.3708, "step": 47 }, { "epoch": 0.01, "learning_rate": 6.906474820143886e-06, "loss": 0.3765, "step": 48 }, { "epoch": 0.01, "learning_rate": 7.050359712230216e-06, "loss": 0.298, "step": 49 }, { "epoch": 0.01, "learning_rate": 7.194244604316547e-06, "loss": 0.2632, "step": 50 }, { "epoch": 0.01, "learning_rate": 7.338129496402878e-06, "loss": 0.2742, "step": 51 }, { "epoch": 0.01, "learning_rate": 7.48201438848921e-06, "loss": 0.3282, "step": 52 }, { "epoch": 0.01, "learning_rate": 7.62589928057554e-06, "loss": 0.3265, "step": 53 }, { "epoch": 0.01, "learning_rate": 7.769784172661872e-06, "loss": 0.2786, "step": 54 }, { "epoch": 0.01, "learning_rate": 7.913669064748202e-06, "loss": 0.3088, "step": 55 }, { "epoch": 0.01, "learning_rate": 8.057553956834533e-06, "loss": 0.4103, "step": 56 }, { "epoch": 0.01, "learning_rate": 8.201438848920865e-06, "loss": 0.4103, "step": 57 }, { "epoch": 0.01, "learning_rate": 8.345323741007195e-06, "loss": 0.312, "step": 58 }, { "epoch": 0.01, "learning_rate": 8.489208633093526e-06, "loss": 0.3002, "step": 59 }, { "epoch": 0.01, "learning_rate": 8.633093525179856e-06, "loss": 0.2914, "step": 60 }, { "epoch": 0.01, "learning_rate": 8.776978417266188e-06, "loss": 0.2993, "step": 61 }, { "epoch": 0.01, "learning_rate": 8.92086330935252e-06, "loss": 0.2902, "step": 62 }, { "epoch": 0.01, "learning_rate": 9.064748201438849e-06, "loss": 0.248, "step": 63 }, { "epoch": 0.01, "learning_rate": 9.20863309352518e-06, "loss": 0.3459, "step": 64 }, { "epoch": 0.01, "learning_rate": 9.35251798561151e-06, "loss": 0.2549, "step": 65 }, { "epoch": 0.01, "learning_rate": 9.496402877697842e-06, "loss": 0.4329, "step": 66 }, { "epoch": 0.01, "learning_rate": 9.640287769784174e-06, "loss": 0.3276, "step": 67 }, { "epoch": 0.01, "learning_rate": 9.784172661870505e-06, "loss": 0.3322, "step": 68 }, { "epoch": 0.01, "learning_rate": 9.928057553956835e-06, "loss": 0.2805, "step": 69 }, { "epoch": 0.02, "learning_rate": 1.0071942446043167e-05, "loss": 0.2356, "step": 70 }, { "epoch": 0.02, "learning_rate": 1.0215827338129498e-05, "loss": 0.3065, "step": 71 }, { "epoch": 0.02, "learning_rate": 1.0359712230215828e-05, "loss": 0.3549, "step": 72 }, { "epoch": 0.02, "learning_rate": 1.0503597122302158e-05, "loss": 0.2795, "step": 73 }, { "epoch": 0.02, "learning_rate": 1.0647482014388491e-05, "loss": 0.2579, "step": 74 }, { "epoch": 0.02, "learning_rate": 1.0791366906474821e-05, "loss": 0.3898, "step": 75 }, { "epoch": 0.02, "learning_rate": 1.0935251798561153e-05, "loss": 0.3386, "step": 76 }, { "epoch": 0.02, "learning_rate": 1.1079136690647482e-05, "loss": 0.3308, "step": 77 }, { "epoch": 0.02, "learning_rate": 1.1223021582733812e-05, "loss": 0.2544, "step": 78 }, { "epoch": 0.02, "learning_rate": 1.1366906474820146e-05, "loss": 0.2912, "step": 79 }, { "epoch": 0.02, "learning_rate": 1.1510791366906475e-05, "loss": 0.3413, "step": 80 }, { "epoch": 0.02, "learning_rate": 1.1654676258992807e-05, "loss": 0.2746, "step": 81 }, { "epoch": 0.02, "learning_rate": 1.1798561151079137e-05, "loss": 0.3055, "step": 82 }, { "epoch": 0.02, "learning_rate": 1.1942446043165468e-05, "loss": 0.2437, "step": 83 }, { "epoch": 0.02, "learning_rate": 1.20863309352518e-05, "loss": 0.2986, "step": 84 }, { "epoch": 0.02, "learning_rate": 1.223021582733813e-05, "loss": 0.2492, "step": 85 }, { "epoch": 0.02, "learning_rate": 1.2374100719424463e-05, "loss": 0.3267, "step": 86 }, { "epoch": 0.02, "learning_rate": 1.2517985611510793e-05, "loss": 0.3066, "step": 87 }, { "epoch": 0.02, "learning_rate": 1.2661870503597123e-05, "loss": 0.2828, "step": 88 }, { "epoch": 0.02, "learning_rate": 1.2805755395683454e-05, "loss": 0.3049, "step": 89 }, { "epoch": 0.02, "learning_rate": 1.2949640287769784e-05, "loss": 0.2745, "step": 90 }, { "epoch": 0.02, "learning_rate": 1.3093525179856117e-05, "loss": 0.276, "step": 91 }, { "epoch": 0.02, "learning_rate": 1.3237410071942447e-05, "loss": 0.3648, "step": 92 }, { "epoch": 0.02, "learning_rate": 1.3381294964028777e-05, "loss": 0.3976, "step": 93 }, { "epoch": 0.02, "learning_rate": 1.3525179856115109e-05, "loss": 0.3339, "step": 94 }, { "epoch": 0.02, "learning_rate": 1.3669064748201439e-05, "loss": 0.2789, "step": 95 }, { "epoch": 0.02, "learning_rate": 1.3812949640287772e-05, "loss": 0.3109, "step": 96 }, { "epoch": 0.02, "learning_rate": 1.3956834532374102e-05, "loss": 0.2732, "step": 97 }, { "epoch": 0.02, "learning_rate": 1.4100719424460432e-05, "loss": 0.2913, "step": 98 }, { "epoch": 0.02, "learning_rate": 1.4244604316546765e-05, "loss": 0.2877, "step": 99 }, { "epoch": 0.02, "learning_rate": 1.4388489208633095e-05, "loss": 0.3096, "step": 100 }, { "epoch": 0.02, "learning_rate": 1.4532374100719426e-05, "loss": 0.2783, "step": 101 }, { "epoch": 0.02, "learning_rate": 1.4676258992805756e-05, "loss": 0.2843, "step": 102 }, { "epoch": 0.02, "learning_rate": 1.4820143884892086e-05, "loss": 0.3157, "step": 103 }, { "epoch": 0.02, "learning_rate": 1.496402877697842e-05, "loss": 0.338, "step": 104 }, { "epoch": 0.02, "learning_rate": 1.5107913669064749e-05, "loss": 0.3258, "step": 105 }, { "epoch": 0.02, "learning_rate": 1.525179856115108e-05, "loss": 0.2684, "step": 106 }, { "epoch": 0.02, "learning_rate": 1.5395683453237412e-05, "loss": 0.3098, "step": 107 }, { "epoch": 0.02, "learning_rate": 1.5539568345323744e-05, "loss": 0.3562, "step": 108 }, { "epoch": 0.02, "learning_rate": 1.5683453237410072e-05, "loss": 0.3254, "step": 109 }, { "epoch": 0.02, "learning_rate": 1.5827338129496403e-05, "loss": 0.3375, "step": 110 }, { "epoch": 0.02, "learning_rate": 1.5971223021582735e-05, "loss": 0.2774, "step": 111 }, { "epoch": 0.02, "learning_rate": 1.6115107913669067e-05, "loss": 0.3432, "step": 112 }, { "epoch": 0.02, "learning_rate": 1.6258992805755398e-05, "loss": 0.2941, "step": 113 }, { "epoch": 0.02, "learning_rate": 1.640287769784173e-05, "loss": 0.3065, "step": 114 }, { "epoch": 0.02, "learning_rate": 1.6546762589928058e-05, "loss": 0.3716, "step": 115 }, { "epoch": 0.03, "learning_rate": 1.669064748201439e-05, "loss": 0.3381, "step": 116 }, { "epoch": 0.03, "learning_rate": 1.683453237410072e-05, "loss": 0.3122, "step": 117 }, { "epoch": 0.03, "learning_rate": 1.6978417266187053e-05, "loss": 0.3846, "step": 118 }, { "epoch": 0.03, "learning_rate": 1.7122302158273384e-05, "loss": 0.3031, "step": 119 }, { "epoch": 0.03, "learning_rate": 1.7266187050359712e-05, "loss": 0.3116, "step": 120 }, { "epoch": 0.03, "learning_rate": 1.7410071942446044e-05, "loss": 0.3207, "step": 121 }, { "epoch": 0.03, "learning_rate": 1.7553956834532375e-05, "loss": 0.3291, "step": 122 }, { "epoch": 0.03, "learning_rate": 1.7697841726618707e-05, "loss": 0.3009, "step": 123 }, { "epoch": 0.03, "learning_rate": 1.784172661870504e-05, "loss": 0.292, "step": 124 }, { "epoch": 0.03, "learning_rate": 1.7985611510791367e-05, "loss": 0.2692, "step": 125 }, { "epoch": 0.03, "learning_rate": 1.8129496402877698e-05, "loss": 0.283, "step": 126 }, { "epoch": 0.03, "learning_rate": 1.827338129496403e-05, "loss": 0.2615, "step": 127 }, { "epoch": 0.03, "learning_rate": 1.841726618705036e-05, "loss": 0.2727, "step": 128 }, { "epoch": 0.03, "learning_rate": 1.8561151079136693e-05, "loss": 0.3593, "step": 129 }, { "epoch": 0.03, "learning_rate": 1.870503597122302e-05, "loss": 0.2984, "step": 130 }, { "epoch": 0.03, "learning_rate": 1.8848920863309356e-05, "loss": 0.2871, "step": 131 }, { "epoch": 0.03, "learning_rate": 1.8992805755395684e-05, "loss": 0.3194, "step": 132 }, { "epoch": 0.03, "learning_rate": 1.9136690647482016e-05, "loss": 0.2399, "step": 133 }, { "epoch": 0.03, "learning_rate": 1.9280575539568347e-05, "loss": 0.3168, "step": 134 }, { "epoch": 0.03, "learning_rate": 1.9424460431654675e-05, "loss": 0.2432, "step": 135 }, { "epoch": 0.03, "learning_rate": 1.956834532374101e-05, "loss": 0.2847, "step": 136 }, { "epoch": 0.03, "learning_rate": 1.971223021582734e-05, "loss": 0.295, "step": 137 }, { "epoch": 0.03, "learning_rate": 1.985611510791367e-05, "loss": 0.3222, "step": 138 }, { "epoch": 0.03, "learning_rate": 2e-05, "loss": 0.3235, "step": 139 }, { "epoch": 0.03, "learning_rate": 1.999999755654922e-05, "loss": 0.3101, "step": 140 }, { "epoch": 0.03, "learning_rate": 1.999999022619806e-05, "loss": 0.3261, "step": 141 }, { "epoch": 0.03, "learning_rate": 1.999997800895011e-05, "loss": 0.2863, "step": 142 }, { "epoch": 0.03, "learning_rate": 1.999996090481134e-05, "loss": 0.3476, "step": 143 }, { "epoch": 0.03, "learning_rate": 1.999993891379011e-05, "loss": 0.3763, "step": 144 }, { "epoch": 0.03, "learning_rate": 1.999991203589716e-05, "loss": 0.3152, "step": 145 }, { "epoch": 0.03, "learning_rate": 1.9999880271145634e-05, "loss": 0.2629, "step": 146 }, { "epoch": 0.03, "learning_rate": 1.9999843619551047e-05, "loss": 0.3358, "step": 147 }, { "epoch": 0.03, "learning_rate": 1.9999802081131313e-05, "loss": 0.3169, "step": 148 }, { "epoch": 0.03, "learning_rate": 1.9999755655906737e-05, "loss": 0.2555, "step": 149 }, { "epoch": 0.03, "learning_rate": 1.9999704343899997e-05, "loss": 0.308, "step": 150 }, { "epoch": 0.03, "learning_rate": 1.9999648145136174e-05, "loss": 0.2919, "step": 151 }, { "epoch": 0.03, "learning_rate": 1.9999587059642732e-05, "loss": 0.3691, "step": 152 }, { "epoch": 0.03, "learning_rate": 1.9999521087449523e-05, "loss": 0.2601, "step": 153 }, { "epoch": 0.03, "learning_rate": 1.9999450228588785e-05, "loss": 0.3128, "step": 154 }, { "epoch": 0.03, "learning_rate": 1.9999374483095147e-05, "loss": 0.3109, "step": 155 }, { "epoch": 0.03, "learning_rate": 1.9999293851005628e-05, "loss": 0.3062, "step": 156 }, { "epoch": 0.03, "learning_rate": 1.9999208332359627e-05, "loss": 0.3443, "step": 157 }, { "epoch": 0.03, "learning_rate": 1.999911792719894e-05, "loss": 0.284, "step": 158 }, { "epoch": 0.03, "learning_rate": 1.9999022635567743e-05, "loss": 0.2725, "step": 159 }, { "epoch": 0.03, "learning_rate": 1.9998922457512608e-05, "loss": 0.2891, "step": 160 }, { "epoch": 0.03, "learning_rate": 1.9998817393082487e-05, "loss": 0.3032, "step": 161 }, { "epoch": 0.03, "learning_rate": 1.9998707442328728e-05, "loss": 0.2461, "step": 162 }, { "epoch": 0.04, "learning_rate": 1.9998592605305065e-05, "loss": 0.3392, "step": 163 }, { "epoch": 0.04, "learning_rate": 1.999847288206761e-05, "loss": 0.3212, "step": 164 }, { "epoch": 0.04, "learning_rate": 1.9998348272674877e-05, "loss": 0.2992, "step": 165 }, { "epoch": 0.04, "learning_rate": 1.999821877718776e-05, "loss": 0.3023, "step": 166 }, { "epoch": 0.04, "learning_rate": 1.9998084395669537e-05, "loss": 0.4263, "step": 167 }, { "epoch": 0.04, "learning_rate": 1.9997945128185884e-05, "loss": 0.2757, "step": 168 }, { "epoch": 0.04, "learning_rate": 1.9997800974804862e-05, "loss": 0.3063, "step": 169 }, { "epoch": 0.04, "learning_rate": 1.9997651935596912e-05, "loss": 0.3287, "step": 170 }, { "epoch": 0.04, "learning_rate": 1.9997498010634872e-05, "loss": 0.2971, "step": 171 }, { "epoch": 0.04, "learning_rate": 1.9997339199993958e-05, "loss": 0.3109, "step": 172 }, { "epoch": 0.04, "learning_rate": 1.9997175503751785e-05, "loss": 0.2764, "step": 173 }, { "epoch": 0.04, "learning_rate": 1.999700692198835e-05, "loss": 0.3299, "step": 174 }, { "epoch": 0.04, "learning_rate": 1.9996833454786035e-05, "loss": 0.3069, "step": 175 }, { "epoch": 0.04, "learning_rate": 1.9996655102229607e-05, "loss": 0.2684, "step": 176 }, { "epoch": 0.04, "learning_rate": 1.9996471864406235e-05, "loss": 0.336, "step": 177 }, { "epoch": 0.04, "learning_rate": 1.999628374140546e-05, "loss": 0.3295, "step": 178 }, { "epoch": 0.04, "learning_rate": 1.999609073331921e-05, "loss": 0.2741, "step": 179 }, { "epoch": 0.04, "learning_rate": 1.999589284024182e-05, "loss": 0.3292, "step": 180 }, { "epoch": 0.04, "learning_rate": 1.9995690062269985e-05, "loss": 0.3573, "step": 181 }, { "epoch": 0.04, "learning_rate": 1.999548239950281e-05, "loss": 0.2879, "step": 182 }, { "epoch": 0.04, "learning_rate": 1.999526985204177e-05, "loss": 0.2985, "step": 183 }, { "epoch": 0.04, "learning_rate": 1.9995052419990743e-05, "loss": 0.3159, "step": 184 }, { "epoch": 0.04, "learning_rate": 1.999483010345598e-05, "loss": 0.3431, "step": 185 }, { "epoch": 0.04, "learning_rate": 1.999460290254613e-05, "loss": 0.3251, "step": 186 }, { "epoch": 0.04, "learning_rate": 1.9994370817372216e-05, "loss": 0.2861, "step": 187 }, { "epoch": 0.04, "learning_rate": 1.9994133848047665e-05, "loss": 0.2976, "step": 188 }, { "epoch": 0.04, "learning_rate": 1.9993891994688273e-05, "loss": 0.3514, "step": 189 }, { "epoch": 0.04, "learning_rate": 1.9993645257412236e-05, "loss": 0.2876, "step": 190 }, { "epoch": 0.04, "learning_rate": 1.9993393636340134e-05, "loss": 0.2704, "step": 191 }, { "epoch": 0.04, "learning_rate": 1.9993137131594925e-05, "loss": 0.3066, "step": 192 }, { "epoch": 0.04, "learning_rate": 1.9992875743301967e-05, "loss": 0.2611, "step": 193 }, { "epoch": 0.04, "learning_rate": 1.9992609471588995e-05, "loss": 0.3127, "step": 194 }, { "epoch": 0.04, "learning_rate": 1.9992338316586132e-05, "loss": 0.3347, "step": 195 }, { "epoch": 0.04, "learning_rate": 1.9992062278425893e-05, "loss": 0.3081, "step": 196 }, { "epoch": 0.04, "learning_rate": 1.999178135724317e-05, "loss": 0.2932, "step": 197 }, { "epoch": 0.04, "learning_rate": 1.9991495553175252e-05, "loss": 0.3082, "step": 198 }, { "epoch": 0.04, "learning_rate": 1.9991204866361806e-05, "loss": 0.3226, "step": 199 }, { "epoch": 0.04, "learning_rate": 1.9990909296944886e-05, "loss": 0.2781, "step": 200 }, { "epoch": 0.04, "learning_rate": 1.9990608845068932e-05, "loss": 0.3409, "step": 201 }, { "epoch": 0.04, "learning_rate": 1.999030351088078e-05, "loss": 0.3041, "step": 202 }, { "epoch": 0.04, "learning_rate": 1.9989993294529637e-05, "loss": 0.3122, "step": 203 }, { "epoch": 0.04, "learning_rate": 1.9989678196167107e-05, "loss": 0.2649, "step": 204 }, { "epoch": 0.04, "learning_rate": 1.9989358215947168e-05, "loss": 0.2322, "step": 205 }, { "epoch": 0.04, "learning_rate": 1.99890333540262e-05, "loss": 0.3047, "step": 206 }, { "epoch": 0.04, "learning_rate": 1.9988703610562953e-05, "loss": 0.2912, "step": 207 }, { "epoch": 0.04, "learning_rate": 1.998836898571858e-05, "loss": 0.3431, "step": 208 }, { "epoch": 0.05, "learning_rate": 1.9988029479656596e-05, "loss": 0.2759, "step": 209 }, { "epoch": 0.05, "learning_rate": 1.9987685092542917e-05, "loss": 0.3185, "step": 210 }, { "epoch": 0.05, "learning_rate": 1.998733582454585e-05, "loss": 0.3383, "step": 211 }, { "epoch": 0.05, "learning_rate": 1.9986981675836067e-05, "loss": 0.2999, "step": 212 }, { "epoch": 0.05, "learning_rate": 1.998662264658665e-05, "loss": 0.2549, "step": 213 }, { "epoch": 0.05, "learning_rate": 1.998625873697304e-05, "loss": 0.2815, "step": 214 }, { "epoch": 0.05, "learning_rate": 1.9985889947173087e-05, "loss": 0.3049, "step": 215 }, { "epoch": 0.05, "learning_rate": 1.9985516277367004e-05, "loss": 0.2931, "step": 216 }, { "epoch": 0.05, "learning_rate": 1.998513772773741e-05, "loss": 0.2894, "step": 217 }, { "epoch": 0.05, "learning_rate": 1.9984754298469297e-05, "loss": 0.3394, "step": 218 }, { "epoch": 0.05, "learning_rate": 1.998436598975004e-05, "loss": 0.3436, "step": 219 }, { "epoch": 0.05, "learning_rate": 1.9983972801769398e-05, "loss": 0.3581, "step": 220 }, { "epoch": 0.05, "learning_rate": 1.9983574734719524e-05, "loss": 0.3081, "step": 221 }, { "epoch": 0.05, "learning_rate": 1.9983171788794946e-05, "loss": 0.3511, "step": 222 }, { "epoch": 0.05, "learning_rate": 1.9982763964192586e-05, "loss": 0.3282, "step": 223 }, { "epoch": 0.05, "learning_rate": 1.9982351261111736e-05, "loss": 0.3442, "step": 224 }, { "epoch": 0.05, "learning_rate": 1.998193367975408e-05, "loss": 0.3406, "step": 225 }, { "epoch": 0.05, "learning_rate": 1.9981511220323693e-05, "loss": 0.312, "step": 226 }, { "epoch": 0.05, "learning_rate": 1.9981083883027024e-05, "loss": 0.3046, "step": 227 }, { "epoch": 0.05, "learning_rate": 1.9980651668072906e-05, "loss": 0.249, "step": 228 }, { "epoch": 0.05, "learning_rate": 1.998021457567256e-05, "loss": 0.3861, "step": 229 }, { "epoch": 0.05, "learning_rate": 1.9979772606039587e-05, "loss": 0.2955, "step": 230 }, { "epoch": 0.05, "learning_rate": 1.9979325759389976e-05, "loss": 0.3257, "step": 231 }, { "epoch": 0.05, "learning_rate": 1.9978874035942097e-05, "loss": 0.261, "step": 232 }, { "epoch": 0.05, "learning_rate": 1.9978417435916695e-05, "loss": 0.3236, "step": 233 }, { "epoch": 0.05, "learning_rate": 1.9977955959536918e-05, "loss": 0.2779, "step": 234 }, { "epoch": 0.05, "learning_rate": 1.9977489607028272e-05, "loss": 0.3054, "step": 235 }, { "epoch": 0.05, "learning_rate": 1.9977018378618674e-05, "loss": 0.3106, "step": 236 }, { "epoch": 0.05, "learning_rate": 1.9976542274538394e-05, "loss": 0.3399, "step": 237 }, { "epoch": 0.05, "learning_rate": 1.997606129502011e-05, "loss": 0.3612, "step": 238 }, { "epoch": 0.05, "learning_rate": 1.9975575440298863e-05, "loss": 0.3116, "step": 239 }, { "epoch": 0.05, "learning_rate": 1.997508471061209e-05, "loss": 0.3044, "step": 240 }, { "epoch": 0.05, "learning_rate": 1.9974589106199612e-05, "loss": 0.2651, "step": 241 }, { "epoch": 0.05, "learning_rate": 1.9974088627303615e-05, "loss": 0.2874, "step": 242 }, { "epoch": 0.05, "learning_rate": 1.9973583274168686e-05, "loss": 0.3029, "step": 243 }, { "epoch": 0.05, "learning_rate": 1.997307304704178e-05, "loss": 0.2996, "step": 244 }, { "epoch": 0.05, "learning_rate": 1.9972557946172246e-05, "loss": 0.3032, "step": 245 }, { "epoch": 0.05, "learning_rate": 1.9972037971811802e-05, "loss": 0.3297, "step": 246 }, { "epoch": 0.05, "learning_rate": 1.9971513124214562e-05, "loss": 0.3194, "step": 247 }, { "epoch": 0.05, "learning_rate": 1.9970983403637006e-05, "loss": 0.3105, "step": 248 }, { "epoch": 0.05, "learning_rate": 1.997044881033801e-05, "loss": 0.3127, "step": 249 }, { "epoch": 0.05, "learning_rate": 1.996990934457882e-05, "loss": 0.3306, "step": 250 }, { "epoch": 0.05, "learning_rate": 1.9969365006623072e-05, "loss": 0.3201, "step": 251 }, { "epoch": 0.05, "learning_rate": 1.9968815796736777e-05, "loss": 0.2953, "step": 252 }, { "epoch": 0.05, "learning_rate": 1.9968261715188325e-05, "loss": 0.2289, "step": 253 }, { "epoch": 0.05, "learning_rate": 1.996770276224849e-05, "loss": 0.2628, "step": 254 }, { "epoch": 0.06, "learning_rate": 1.9967138938190434e-05, "loss": 0.2969, "step": 255 }, { "epoch": 0.06, "learning_rate": 1.9966570243289686e-05, "loss": 0.3655, "step": 256 }, { "epoch": 0.06, "learning_rate": 1.996599667782416e-05, "loss": 0.2957, "step": 257 }, { "epoch": 0.06, "learning_rate": 1.996541824207416e-05, "loss": 0.2881, "step": 258 }, { "epoch": 0.06, "learning_rate": 1.9964834936322358e-05, "loss": 0.3375, "step": 259 }, { "epoch": 0.06, "learning_rate": 1.9964246760853803e-05, "loss": 0.3332, "step": 260 }, { "epoch": 0.06, "learning_rate": 1.996365371595594e-05, "loss": 0.3062, "step": 261 }, { "epoch": 0.06, "learning_rate": 1.996305580191858e-05, "loss": 0.3056, "step": 262 }, { "epoch": 0.06, "learning_rate": 1.9962453019033917e-05, "loss": 0.2479, "step": 263 }, { "epoch": 0.06, "learning_rate": 1.9961845367596523e-05, "loss": 0.272, "step": 264 }, { "epoch": 0.06, "learning_rate": 1.996123284790336e-05, "loss": 0.2701, "step": 265 }, { "epoch": 0.06, "learning_rate": 1.9960615460253754e-05, "loss": 0.2762, "step": 266 }, { "epoch": 0.06, "learning_rate": 1.9959993204949416e-05, "loss": 0.3722, "step": 267 }, { "epoch": 0.06, "learning_rate": 1.995936608229444e-05, "loss": 0.3243, "step": 268 }, { "epoch": 0.06, "learning_rate": 1.9958734092595286e-05, "loss": 0.3147, "step": 269 }, { "epoch": 0.06, "learning_rate": 1.995809723616081e-05, "loss": 0.3065, "step": 270 }, { "epoch": 0.06, "learning_rate": 1.9957455513302234e-05, "loss": 0.3623, "step": 271 }, { "epoch": 0.06, "learning_rate": 1.9956808924333164e-05, "loss": 0.3088, "step": 272 }, { "epoch": 0.06, "learning_rate": 1.9956157469569577e-05, "loss": 0.276, "step": 273 }, { "epoch": 0.06, "learning_rate": 1.9955501149329837e-05, "loss": 0.2901, "step": 274 }, { "epoch": 0.06, "learning_rate": 1.995483996393468e-05, "loss": 0.3005, "step": 275 }, { "epoch": 0.06, "learning_rate": 1.995417391370722e-05, "loss": 0.3127, "step": 276 }, { "epoch": 0.06, "learning_rate": 1.995350299897295e-05, "loss": 0.3091, "step": 277 }, { "epoch": 0.06, "learning_rate": 1.9952827220059733e-05, "loss": 0.2328, "step": 278 }, { "epoch": 0.06, "learning_rate": 1.9952146577297827e-05, "loss": 0.2766, "step": 279 }, { "epoch": 0.06, "learning_rate": 1.995146107101985e-05, "loss": 0.2686, "step": 280 }, { "epoch": 0.06, "learning_rate": 1.9950770701560802e-05, "loss": 0.3656, "step": 281 }, { "epoch": 0.06, "learning_rate": 1.9950075469258058e-05, "loss": 0.3132, "step": 282 }, { "epoch": 0.06, "learning_rate": 1.9949375374451376e-05, "loss": 0.3416, "step": 283 }, { "epoch": 0.06, "learning_rate": 1.9948670417482877e-05, "loss": 0.277, "step": 284 }, { "epoch": 0.06, "learning_rate": 1.9947960598697077e-05, "loss": 0.3003, "step": 285 }, { "epoch": 0.06, "learning_rate": 1.994724591844085e-05, "loss": 0.3523, "step": 286 }, { "epoch": 0.06, "learning_rate": 1.9946526377063456e-05, "loss": 0.3252, "step": 287 }, { "epoch": 0.06, "learning_rate": 1.994580197491653e-05, "loss": 0.332, "step": 288 }, { "epoch": 0.06, "learning_rate": 1.9945072712354075e-05, "loss": 0.296, "step": 289 }, { "epoch": 0.06, "learning_rate": 1.994433858973248e-05, "loss": 0.32, "step": 290 }, { "epoch": 0.06, "learning_rate": 1.994359960741049e-05, "loss": 0.2783, "step": 291 }, { "epoch": 0.06, "learning_rate": 1.9942855765749262e-05, "loss": 0.3008, "step": 292 }, { "epoch": 0.06, "learning_rate": 1.9942107065112286e-05, "loss": 0.3111, "step": 293 }, { "epoch": 0.06, "learning_rate": 1.9941353505865447e-05, "loss": 0.2981, "step": 294 }, { "epoch": 0.06, "learning_rate": 1.9940595088377005e-05, "loss": 0.3571, "step": 295 }, { "epoch": 0.06, "learning_rate": 1.9939831813017595e-05, "loss": 0.2436, "step": 296 }, { "epoch": 0.06, "learning_rate": 1.9939063680160215e-05, "loss": 0.2746, "step": 297 }, { "epoch": 0.06, "learning_rate": 1.9938290690180248e-05, "loss": 0.2865, "step": 298 }, { "epoch": 0.06, "learning_rate": 1.9937512843455446e-05, "loss": 0.2282, "step": 299 }, { "epoch": 0.06, "learning_rate": 1.9936730140365933e-05, "loss": 0.3268, "step": 300 }, { "epoch": 0.06, "learning_rate": 1.993594258129421e-05, "loss": 0.3765, "step": 301 }, { "epoch": 0.07, "learning_rate": 1.993515016662515e-05, "loss": 0.2881, "step": 302 }, { "epoch": 0.07, "learning_rate": 1.9934352896745997e-05, "loss": 0.3308, "step": 303 }, { "epoch": 0.07, "learning_rate": 1.993355077204637e-05, "loss": 0.2479, "step": 304 }, { "epoch": 0.07, "learning_rate": 1.9932743792918256e-05, "loss": 0.2697, "step": 305 }, { "epoch": 0.07, "learning_rate": 1.9931931959756024e-05, "loss": 0.3239, "step": 306 }, { "epoch": 0.07, "learning_rate": 1.9931115272956405e-05, "loss": 0.3201, "step": 307 }, { "epoch": 0.07, "learning_rate": 1.9930293732918506e-05, "loss": 0.3074, "step": 308 }, { "epoch": 0.07, "learning_rate": 1.9929467340043805e-05, "loss": 0.3207, "step": 309 }, { "epoch": 0.07, "learning_rate": 1.9928636094736152e-05, "loss": 0.3233, "step": 310 }, { "epoch": 0.07, "learning_rate": 1.992779999740177e-05, "loss": 0.2995, "step": 311 }, { "epoch": 0.07, "learning_rate": 1.9926959048449253e-05, "loss": 0.292, "step": 312 }, { "epoch": 0.07, "learning_rate": 1.992611324828956e-05, "loss": 0.2626, "step": 313 }, { "epoch": 0.07, "learning_rate": 1.992526259733603e-05, "loss": 0.3194, "step": 314 }, { "epoch": 0.07, "learning_rate": 1.9924407096004362e-05, "loss": 0.2815, "step": 315 }, { "epoch": 0.07, "learning_rate": 1.9923546744712637e-05, "loss": 0.3065, "step": 316 }, { "epoch": 0.07, "learning_rate": 1.9922681543881295e-05, "loss": 0.2968, "step": 317 }, { "epoch": 0.07, "learning_rate": 1.9921811493933158e-05, "loss": 0.253, "step": 318 }, { "epoch": 0.07, "learning_rate": 1.99209365952934e-05, "loss": 0.2989, "step": 319 }, { "epoch": 0.07, "learning_rate": 1.9920056848389585e-05, "loss": 0.2689, "step": 320 }, { "epoch": 0.07, "learning_rate": 1.9919172253651637e-05, "loss": 0.2816, "step": 321 }, { "epoch": 0.07, "learning_rate": 1.9918282811511837e-05, "loss": 0.2931, "step": 322 }, { "epoch": 0.07, "learning_rate": 1.991738852240486e-05, "loss": 0.3408, "step": 323 }, { "epoch": 0.07, "learning_rate": 1.9916489386767734e-05, "loss": 0.3525, "step": 324 }, { "epoch": 0.07, "learning_rate": 1.991558540503985e-05, "loss": 0.367, "step": 325 }, { "epoch": 0.07, "learning_rate": 1.991467657766298e-05, "loss": 0.2978, "step": 326 }, { "epoch": 0.07, "learning_rate": 1.9913762905081262e-05, "loss": 0.3316, "step": 327 }, { "epoch": 0.07, "learning_rate": 1.9912844387741194e-05, "loss": 0.3401, "step": 328 }, { "epoch": 0.07, "learning_rate": 1.9911921026091646e-05, "loss": 0.2511, "step": 329 }, { "epoch": 0.07, "learning_rate": 1.9910992820583863e-05, "loss": 0.3217, "step": 330 }, { "epoch": 0.07, "learning_rate": 1.9910059771671442e-05, "loss": 0.367, "step": 331 }, { "epoch": 0.07, "learning_rate": 1.9909121879810357e-05, "loss": 0.3158, "step": 332 }, { "epoch": 0.07, "learning_rate": 1.9908179145458945e-05, "loss": 0.3063, "step": 333 }, { "epoch": 0.07, "learning_rate": 1.9907231569077916e-05, "loss": 0.3479, "step": 334 }, { "epoch": 0.07, "learning_rate": 1.9906279151130338e-05, "loss": 0.2677, "step": 335 }, { "epoch": 0.07, "learning_rate": 1.9905321892081647e-05, "loss": 0.3064, "step": 336 }, { "epoch": 0.07, "learning_rate": 1.990435979239965e-05, "loss": 0.3021, "step": 337 }, { "epoch": 0.07, "learning_rate": 1.990339285255451e-05, "loss": 0.3117, "step": 338 }, { "epoch": 0.07, "learning_rate": 1.990242107301877e-05, "loss": 0.3438, "step": 339 }, { "epoch": 0.07, "learning_rate": 1.9901444454267316e-05, "loss": 0.3099, "step": 340 }, { "epoch": 0.07, "learning_rate": 1.990046299677742e-05, "loss": 0.2786, "step": 341 }, { "epoch": 0.07, "learning_rate": 1.9899476701028715e-05, "loss": 0.2556, "step": 342 }, { "epoch": 0.07, "learning_rate": 1.9898485567503186e-05, "loss": 0.262, "step": 343 }, { "epoch": 0.07, "learning_rate": 1.9897489596685194e-05, "loss": 0.261, "step": 344 }, { "epoch": 0.07, "learning_rate": 1.9896488789061462e-05, "loss": 0.3777, "step": 345 }, { "epoch": 0.07, "learning_rate": 1.9895483145121067e-05, "loss": 0.3437, "step": 346 }, { "epoch": 0.07, "learning_rate": 1.9894472665355463e-05, "loss": 0.3108, "step": 347 }, { "epoch": 0.08, "learning_rate": 1.9893457350258462e-05, "loss": 0.2528, "step": 348 }, { "epoch": 0.08, "learning_rate": 1.989243720032624e-05, "loss": 0.3221, "step": 349 }, { "epoch": 0.08, "learning_rate": 1.989141221605733e-05, "loss": 0.3046, "step": 350 }, { "epoch": 0.08, "learning_rate": 1.9890382397952632e-05, "loss": 0.3146, "step": 351 }, { "epoch": 0.08, "learning_rate": 1.988934774651541e-05, "loss": 0.3141, "step": 352 }, { "epoch": 0.08, "learning_rate": 1.9888308262251286e-05, "loss": 0.3119, "step": 353 }, { "epoch": 0.08, "learning_rate": 1.988726394566825e-05, "loss": 0.2929, "step": 354 }, { "epoch": 0.08, "learning_rate": 1.9886214797276644e-05, "loss": 0.3635, "step": 355 }, { "epoch": 0.08, "learning_rate": 1.988516081758918e-05, "loss": 0.2712, "step": 356 }, { "epoch": 0.08, "learning_rate": 1.9884102007120925e-05, "loss": 0.2971, "step": 357 }, { "epoch": 0.08, "learning_rate": 1.9883038366389312e-05, "loss": 0.3057, "step": 358 }, { "epoch": 0.08, "learning_rate": 1.988196989591413e-05, "loss": 0.3361, "step": 359 }, { "epoch": 0.08, "learning_rate": 1.988089659621753e-05, "loss": 0.3233, "step": 360 }, { "epoch": 0.08, "learning_rate": 1.987981846782402e-05, "loss": 0.2468, "step": 361 }, { "epoch": 0.08, "learning_rate": 1.987873551126048e-05, "loss": 0.3105, "step": 362 }, { "epoch": 0.08, "learning_rate": 1.987764772705613e-05, "loss": 0.3179, "step": 363 }, { "epoch": 0.08, "learning_rate": 1.9876555115742565e-05, "loss": 0.3127, "step": 364 }, { "epoch": 0.08, "learning_rate": 1.9875457677853734e-05, "loss": 0.2636, "step": 365 }, { "epoch": 0.08, "learning_rate": 1.987435541392594e-05, "loss": 0.2729, "step": 366 }, { "epoch": 0.08, "learning_rate": 1.987324832449785e-05, "loss": 0.2984, "step": 367 }, { "epoch": 0.08, "learning_rate": 1.987213641011049e-05, "loss": 0.3453, "step": 368 }, { "epoch": 0.08, "learning_rate": 1.987101967130724e-05, "loss": 0.2931, "step": 369 }, { "epoch": 0.08, "learning_rate": 1.9869898108633834e-05, "loss": 0.3418, "step": 370 }, { "epoch": 0.08, "learning_rate": 1.9868771722638382e-05, "loss": 0.3086, "step": 371 }, { "epoch": 0.08, "learning_rate": 1.9867640513871323e-05, "loss": 0.3346, "step": 372 }, { "epoch": 0.08, "learning_rate": 1.9866504482885482e-05, "loss": 0.4087, "step": 373 }, { "epoch": 0.08, "learning_rate": 1.9865363630236013e-05, "loss": 0.3708, "step": 374 }, { "epoch": 0.08, "learning_rate": 1.9864217956480445e-05, "loss": 0.2891, "step": 375 }, { "epoch": 0.08, "learning_rate": 1.9863067462178658e-05, "loss": 0.2933, "step": 376 }, { "epoch": 0.08, "learning_rate": 1.9861912147892884e-05, "loss": 0.3076, "step": 377 }, { "epoch": 0.08, "learning_rate": 1.986075201418772e-05, "loss": 0.2899, "step": 378 }, { "epoch": 0.08, "learning_rate": 1.9859587061630107e-05, "loss": 0.3029, "step": 379 }, { "epoch": 0.08, "learning_rate": 1.9858417290789348e-05, "loss": 0.3449, "step": 380 }, { "epoch": 0.08, "learning_rate": 1.9857242702237095e-05, "loss": 0.3063, "step": 381 }, { "epoch": 0.08, "learning_rate": 1.9856063296547363e-05, "loss": 0.3163, "step": 382 }, { "epoch": 0.08, "learning_rate": 1.985487907429651e-05, "loss": 0.3847, "step": 383 }, { "epoch": 0.08, "learning_rate": 1.9853690036063263e-05, "loss": 0.3203, "step": 384 }, { "epoch": 0.08, "learning_rate": 1.9852496182428683e-05, "loss": 0.301, "step": 385 }, { "epoch": 0.08, "learning_rate": 1.98512975139762e-05, "loss": 0.3623, "step": 386 }, { "epoch": 0.08, "learning_rate": 1.985009403129159e-05, "loss": 0.3209, "step": 387 }, { "epoch": 0.08, "learning_rate": 1.9848885734962985e-05, "loss": 0.2817, "step": 388 }, { "epoch": 0.08, "learning_rate": 1.9847672625580865e-05, "loss": 0.3477, "step": 389 }, { "epoch": 0.08, "learning_rate": 1.9846454703738064e-05, "loss": 0.3142, "step": 390 }, { "epoch": 0.08, "learning_rate": 1.9845231970029774e-05, "loss": 0.284, "step": 391 }, { "epoch": 0.08, "learning_rate": 1.9844004425053526e-05, "loss": 0.2632, "step": 392 }, { "epoch": 0.08, "learning_rate": 1.9842772069409213e-05, "loss": 0.357, "step": 393 }, { "epoch": 0.09, "learning_rate": 1.9841534903699075e-05, "loss": 0.3134, "step": 394 }, { "epoch": 0.09, "learning_rate": 1.9840292928527703e-05, "loss": 0.2645, "step": 395 }, { "epoch": 0.09, "learning_rate": 1.9839046144502032e-05, "loss": 0.2769, "step": 396 }, { "epoch": 0.09, "learning_rate": 1.9837794552231365e-05, "loss": 0.3138, "step": 397 }, { "epoch": 0.09, "learning_rate": 1.983653815232733e-05, "loss": 0.3157, "step": 398 }, { "epoch": 0.09, "learning_rate": 1.9835276945403926e-05, "loss": 0.3008, "step": 399 }, { "epoch": 0.09, "learning_rate": 1.983401093207749e-05, "loss": 0.3768, "step": 400 }, { "epoch": 0.09, "learning_rate": 1.9832740112966707e-05, "loss": 0.3722, "step": 401 }, { "epoch": 0.09, "learning_rate": 1.983146448869262e-05, "loss": 0.2672, "step": 402 }, { "epoch": 0.09, "learning_rate": 1.9830184059878607e-05, "loss": 0.2971, "step": 403 }, { "epoch": 0.09, "learning_rate": 1.9828898827150408e-05, "loss": 0.2841, "step": 404 }, { "epoch": 0.09, "learning_rate": 1.98276087911361e-05, "loss": 0.2678, "step": 405 }, { "epoch": 0.09, "learning_rate": 1.982631395246611e-05, "loss": 0.2769, "step": 406 }, { "epoch": 0.09, "learning_rate": 1.9825014311773213e-05, "loss": 0.2349, "step": 407 }, { "epoch": 0.09, "learning_rate": 1.982370986969253e-05, "loss": 0.3255, "step": 408 }, { "epoch": 0.09, "learning_rate": 1.9822400626861534e-05, "loss": 0.2998, "step": 409 }, { "epoch": 0.09, "learning_rate": 1.982108658392004e-05, "loss": 0.3864, "step": 410 }, { "epoch": 0.09, "learning_rate": 1.9819767741510195e-05, "loss": 0.2991, "step": 411 }, { "epoch": 0.09, "learning_rate": 1.9818444100276517e-05, "loss": 0.3083, "step": 412 }, { "epoch": 0.09, "learning_rate": 1.981711566086585e-05, "loss": 0.2792, "step": 413 }, { "epoch": 0.09, "learning_rate": 1.9815782423927398e-05, "loss": 0.2749, "step": 414 }, { "epoch": 0.09, "learning_rate": 1.9814444390112687e-05, "loss": 0.3157, "step": 415 }, { "epoch": 0.09, "learning_rate": 1.9813101560075608e-05, "loss": 0.3191, "step": 416 }, { "epoch": 0.09, "learning_rate": 1.9811753934472396e-05, "loss": 0.3031, "step": 417 }, { "epoch": 0.09, "learning_rate": 1.981040151396161e-05, "loss": 0.3261, "step": 418 }, { "epoch": 0.09, "learning_rate": 1.9809044299204173e-05, "loss": 0.3495, "step": 419 }, { "epoch": 0.09, "learning_rate": 1.9807682290863336e-05, "loss": 0.278, "step": 420 }, { "epoch": 0.09, "learning_rate": 1.9806315489604706e-05, "loss": 0.3089, "step": 421 }, { "epoch": 0.09, "learning_rate": 1.980494389609622e-05, "loss": 0.3321, "step": 422 }, { "epoch": 0.09, "learning_rate": 1.9803567511008165e-05, "loss": 0.266, "step": 423 }, { "epoch": 0.09, "learning_rate": 1.9802186335013165e-05, "loss": 0.3356, "step": 424 }, { "epoch": 0.09, "learning_rate": 1.980080036878619e-05, "loss": 0.3178, "step": 425 }, { "epoch": 0.09, "learning_rate": 1.9799409613004544e-05, "loss": 0.359, "step": 426 }, { "epoch": 0.09, "learning_rate": 1.9798014068347882e-05, "loss": 0.32, "step": 427 }, { "epoch": 0.09, "learning_rate": 1.9796613735498185e-05, "loss": 0.3265, "step": 428 }, { "epoch": 0.09, "learning_rate": 1.979520861513979e-05, "loss": 0.294, "step": 429 }, { "epoch": 0.09, "learning_rate": 1.9793798707959358e-05, "loss": 0.3749, "step": 430 }, { "epoch": 0.09, "learning_rate": 1.9792384014645898e-05, "loss": 0.2625, "step": 431 }, { "epoch": 0.09, "learning_rate": 1.9790964535890758e-05, "loss": 0.3285, "step": 432 }, { "epoch": 0.09, "learning_rate": 1.978954027238763e-05, "loss": 0.3002, "step": 433 }, { "epoch": 0.09, "learning_rate": 1.9788111224832527e-05, "loss": 0.2626, "step": 434 }, { "epoch": 0.09, "learning_rate": 1.9786677393923818e-05, "loss": 0.2875, "step": 435 }, { "epoch": 0.09, "learning_rate": 1.9785238780362193e-05, "loss": 0.2544, "step": 436 }, { "epoch": 0.09, "learning_rate": 1.97837953848507e-05, "loss": 0.309, "step": 437 }, { "epoch": 0.09, "learning_rate": 1.9782347208094705e-05, "loss": 0.2879, "step": 438 }, { "epoch": 0.09, "learning_rate": 1.9780894250801918e-05, "loss": 0.3019, "step": 439 }, { "epoch": 0.09, "learning_rate": 1.977943651368239e-05, "loss": 0.3359, "step": 440 }, { "epoch": 0.1, "learning_rate": 1.9777973997448494e-05, "loss": 0.2885, "step": 441 }, { "epoch": 0.1, "learning_rate": 1.9776506702814955e-05, "loss": 0.2877, "step": 442 }, { "epoch": 0.1, "learning_rate": 1.9775034630498823e-05, "loss": 0.3173, "step": 443 }, { "epoch": 0.1, "learning_rate": 1.9773557781219482e-05, "loss": 0.3195, "step": 444 }, { "epoch": 0.1, "learning_rate": 1.977207615569866e-05, "loss": 0.3167, "step": 445 }, { "epoch": 0.1, "learning_rate": 1.9770589754660406e-05, "loss": 0.2861, "step": 446 }, { "epoch": 0.1, "learning_rate": 1.9769098578831113e-05, "loss": 0.3203, "step": 447 }, { "epoch": 0.1, "learning_rate": 1.976760262893951e-05, "loss": 0.2857, "step": 448 }, { "epoch": 0.1, "learning_rate": 1.976610190571664e-05, "loss": 0.3745, "step": 449 }, { "epoch": 0.1, "learning_rate": 1.9764596409895902e-05, "loss": 0.365, "step": 450 }, { "epoch": 0.1, "learning_rate": 1.9763086142213012e-05, "loss": 0.258, "step": 451 }, { "epoch": 0.1, "learning_rate": 1.9761571103406027e-05, "loss": 0.4052, "step": 452 }, { "epoch": 0.1, "learning_rate": 1.9760051294215326e-05, "loss": 0.334, "step": 453 }, { "epoch": 0.1, "learning_rate": 1.975852671538363e-05, "loss": 0.3277, "step": 454 }, { "epoch": 0.1, "learning_rate": 1.975699736765598e-05, "loss": 0.301, "step": 455 }, { "epoch": 0.1, "learning_rate": 1.9755463251779764e-05, "loss": 0.3267, "step": 456 }, { "epoch": 0.1, "learning_rate": 1.9753924368504675e-05, "loss": 0.3489, "step": 457 }, { "epoch": 0.1, "learning_rate": 1.975238071858276e-05, "loss": 0.3023, "step": 458 }, { "epoch": 0.1, "learning_rate": 1.9750832302768385e-05, "loss": 0.2566, "step": 459 }, { "epoch": 0.1, "learning_rate": 1.9749279121818235e-05, "loss": 0.3347, "step": 460 }, { "epoch": 0.1, "learning_rate": 1.974772117649135e-05, "loss": 0.3218, "step": 461 }, { "epoch": 0.1, "learning_rate": 1.9746158467549074e-05, "loss": 0.246, "step": 462 }, { "epoch": 0.1, "learning_rate": 1.9744590995755087e-05, "loss": 0.2879, "step": 463 }, { "epoch": 0.1, "learning_rate": 1.97430187618754e-05, "loss": 0.3343, "step": 464 }, { "epoch": 0.1, "learning_rate": 1.9741441766678346e-05, "loss": 0.2991, "step": 465 }, { "epoch": 0.1, "learning_rate": 1.9739860010934587e-05, "loss": 0.3531, "step": 466 }, { "epoch": 0.1, "learning_rate": 1.9738273495417113e-05, "loss": 0.2651, "step": 467 }, { "epoch": 0.1, "learning_rate": 1.9736682220901237e-05, "loss": 0.4027, "step": 468 }, { "epoch": 0.1, "learning_rate": 1.97350861881646e-05, "loss": 0.3191, "step": 469 }, { "epoch": 0.1, "learning_rate": 1.9733485397987167e-05, "loss": 0.3076, "step": 470 }, { "epoch": 0.1, "learning_rate": 1.973187985115123e-05, "loss": 0.2791, "step": 471 }, { "epoch": 0.1, "learning_rate": 1.9730269548441404e-05, "loss": 0.2787, "step": 472 }, { "epoch": 0.1, "learning_rate": 1.9728654490644625e-05, "loss": 0.2942, "step": 473 }, { "epoch": 0.1, "learning_rate": 1.9727034678550158e-05, "loss": 0.3076, "step": 474 }, { "epoch": 0.1, "learning_rate": 1.972541011294959e-05, "loss": 0.287, "step": 475 }, { "epoch": 0.1, "learning_rate": 1.972378079463683e-05, "loss": 0.299, "step": 476 }, { "epoch": 0.1, "learning_rate": 1.9722146724408104e-05, "loss": 0.3094, "step": 477 }, { "epoch": 0.1, "learning_rate": 1.9720507903061976e-05, "loss": 0.3234, "step": 478 }, { "epoch": 0.1, "learning_rate": 1.9718864331399316e-05, "loss": 0.284, "step": 479 }, { "epoch": 0.1, "learning_rate": 1.971721601022332e-05, "loss": 0.2997, "step": 480 }, { "epoch": 0.1, "learning_rate": 1.971556294033951e-05, "loss": 0.3082, "step": 481 }, { "epoch": 0.1, "learning_rate": 1.9713905122555725e-05, "loss": 0.3733, "step": 482 }, { "epoch": 0.1, "learning_rate": 1.971224255768212e-05, "loss": 0.3343, "step": 483 }, { "epoch": 0.1, "learning_rate": 1.9710575246531178e-05, "loss": 0.3155, "step": 484 }, { "epoch": 0.1, "learning_rate": 1.9708903189917697e-05, "loss": 0.2966, "step": 485 }, { "epoch": 0.1, "learning_rate": 1.9707226388658792e-05, "loss": 0.3035, "step": 486 }, { "epoch": 0.11, "learning_rate": 1.9705544843573902e-05, "loss": 0.2941, "step": 487 }, { "epoch": 0.11, "learning_rate": 1.9703858555484784e-05, "loss": 0.3835, "step": 488 }, { "epoch": 0.11, "learning_rate": 1.9702167525215504e-05, "loss": 0.2957, "step": 489 }, { "epoch": 0.11, "learning_rate": 1.9700471753592453e-05, "loss": 0.3175, "step": 490 }, { "epoch": 0.11, "learning_rate": 1.9698771241444342e-05, "loss": 0.2958, "step": 491 }, { "epoch": 0.11, "learning_rate": 1.9697065989602195e-05, "loss": 0.3236, "step": 492 }, { "epoch": 0.11, "learning_rate": 1.9695355998899346e-05, "loss": 0.2862, "step": 493 }, { "epoch": 0.11, "learning_rate": 1.9693641270171455e-05, "loss": 0.3091, "step": 494 }, { "epoch": 0.11, "learning_rate": 1.9691921804256494e-05, "loss": 0.3649, "step": 495 }, { "epoch": 0.11, "learning_rate": 1.969019760199474e-05, "loss": 0.2989, "step": 496 }, { "epoch": 0.11, "learning_rate": 1.968846866422881e-05, "loss": 0.3123, "step": 497 }, { "epoch": 0.11, "learning_rate": 1.9686734991803602e-05, "loss": 0.3089, "step": 498 }, { "epoch": 0.11, "learning_rate": 1.9684996585566356e-05, "loss": 0.3769, "step": 499 }, { "epoch": 0.11, "learning_rate": 1.9683253446366607e-05, "loss": 0.2762, "step": 500 }, { "epoch": 0.11, "learning_rate": 1.968150557505622e-05, "loss": 0.3434, "step": 501 }, { "epoch": 0.11, "learning_rate": 1.967975297248935e-05, "loss": 0.294, "step": 502 }, { "epoch": 0.11, "learning_rate": 1.9677995639522482e-05, "loss": 0.309, "step": 503 }, { "epoch": 0.11, "learning_rate": 1.967623357701441e-05, "loss": 0.3071, "step": 504 }, { "epoch": 0.11, "learning_rate": 1.9674466785826234e-05, "loss": 0.3631, "step": 505 }, { "epoch": 0.11, "learning_rate": 1.9672695266821367e-05, "loss": 0.2751, "step": 506 }, { "epoch": 0.11, "learning_rate": 1.9670919020865532e-05, "loss": 0.3041, "step": 507 }, { "epoch": 0.11, "learning_rate": 1.9669138048826765e-05, "loss": 0.265, "step": 508 }, { "epoch": 0.11, "learning_rate": 1.966735235157541e-05, "loss": 0.3201, "step": 509 }, { "epoch": 0.11, "learning_rate": 1.9665561929984117e-05, "loss": 0.2397, "step": 510 }, { "epoch": 0.11, "learning_rate": 1.9663766784927855e-05, "loss": 0.3116, "step": 511 }, { "epoch": 0.11, "learning_rate": 1.966196691728388e-05, "loss": 0.3166, "step": 512 }, { "epoch": 0.11, "learning_rate": 1.9660162327931785e-05, "loss": 0.2972, "step": 513 }, { "epoch": 0.11, "learning_rate": 1.965835301775344e-05, "loss": 0.303, "step": 514 }, { "epoch": 0.11, "learning_rate": 1.965653898763305e-05, "loss": 0.3474, "step": 515 }, { "epoch": 0.11, "learning_rate": 1.9654720238457106e-05, "loss": 0.3224, "step": 516 }, { "epoch": 0.11, "learning_rate": 1.9652896771114416e-05, "loss": 0.4267, "step": 517 }, { "epoch": 0.11, "learning_rate": 1.9651068586496087e-05, "loss": 0.3403, "step": 518 }, { "epoch": 0.11, "learning_rate": 1.9649235685495542e-05, "loss": 0.3186, "step": 519 }, { "epoch": 0.11, "learning_rate": 1.964739806900849e-05, "loss": 0.3037, "step": 520 }, { "epoch": 0.11, "learning_rate": 1.964555573793297e-05, "loss": 0.3357, "step": 521 }, { "epoch": 0.11, "learning_rate": 1.96437086931693e-05, "loss": 0.2996, "step": 522 }, { "epoch": 0.11, "learning_rate": 1.9641856935620117e-05, "loss": 0.293, "step": 523 }, { "epoch": 0.11, "learning_rate": 1.9640000466190358e-05, "loss": 0.3132, "step": 524 }, { "epoch": 0.11, "learning_rate": 1.963813928578726e-05, "loss": 0.2594, "step": 525 }, { "epoch": 0.11, "learning_rate": 1.9636273395320363e-05, "loss": 0.2775, "step": 526 }, { "epoch": 0.11, "learning_rate": 1.963440279570151e-05, "loss": 0.2576, "step": 527 }, { "epoch": 0.11, "learning_rate": 1.9632527487844844e-05, "loss": 0.3163, "step": 528 }, { "epoch": 0.11, "learning_rate": 1.963064747266681e-05, "loss": 0.3037, "step": 529 }, { "epoch": 0.11, "learning_rate": 1.962876275108615e-05, "loss": 0.2874, "step": 530 }, { "epoch": 0.11, "learning_rate": 1.9626873324023915e-05, "loss": 0.3051, "step": 531 }, { "epoch": 0.11, "learning_rate": 1.9624979192403444e-05, "loss": 0.337, "step": 532 }, { "epoch": 0.12, "learning_rate": 1.9623080357150386e-05, "loss": 0.3711, "step": 533 }, { "epoch": 0.12, "learning_rate": 1.9621176819192674e-05, "loss": 0.3209, "step": 534 }, { "epoch": 0.12, "learning_rate": 1.9619268579460558e-05, "loss": 0.2553, "step": 535 }, { "epoch": 0.12, "learning_rate": 1.9617355638886574e-05, "loss": 0.3563, "step": 536 }, { "epoch": 0.12, "learning_rate": 1.961543799840555e-05, "loss": 0.2952, "step": 537 }, { "epoch": 0.12, "learning_rate": 1.9613515658954627e-05, "loss": 0.3003, "step": 538 }, { "epoch": 0.12, "learning_rate": 1.9611588621473224e-05, "loss": 0.3638, "step": 539 }, { "epoch": 0.12, "learning_rate": 1.9609656886903075e-05, "loss": 0.3386, "step": 540 }, { "epoch": 0.12, "learning_rate": 1.9607720456188195e-05, "loss": 0.3199, "step": 541 }, { "epoch": 0.12, "learning_rate": 1.9605779330274898e-05, "loss": 0.2803, "step": 542 }, { "epoch": 0.12, "learning_rate": 1.9603833510111793e-05, "loss": 0.2542, "step": 543 }, { "epoch": 0.12, "learning_rate": 1.9601882996649784e-05, "loss": 0.3616, "step": 544 }, { "epoch": 0.12, "learning_rate": 1.959992779084207e-05, "loss": 0.2653, "step": 545 }, { "epoch": 0.12, "learning_rate": 1.9597967893644138e-05, "loss": 0.2943, "step": 546 }, { "epoch": 0.12, "learning_rate": 1.9596003306013774e-05, "loss": 0.3203, "step": 547 }, { "epoch": 0.12, "learning_rate": 1.9594034028911046e-05, "loss": 0.2953, "step": 548 }, { "epoch": 0.12, "learning_rate": 1.9592060063298327e-05, "loss": 0.28, "step": 549 }, { "epoch": 0.12, "learning_rate": 1.959008141014027e-05, "loss": 0.2656, "step": 550 }, { "epoch": 0.12, "learning_rate": 1.9588098070403828e-05, "loss": 0.3242, "step": 551 }, { "epoch": 0.12, "learning_rate": 1.9586110045058232e-05, "loss": 0.3299, "step": 552 }, { "epoch": 0.12, "learning_rate": 1.958411733507502e-05, "loss": 0.3247, "step": 553 }, { "epoch": 0.12, "learning_rate": 1.9582119941428002e-05, "loss": 0.3124, "step": 554 }, { "epoch": 0.12, "learning_rate": 1.9580117865093288e-05, "loss": 0.3352, "step": 555 }, { "epoch": 0.12, "learning_rate": 1.9578111107049272e-05, "loss": 0.2778, "step": 556 }, { "epoch": 0.12, "learning_rate": 1.957609966827664e-05, "loss": 0.2903, "step": 557 }, { "epoch": 0.12, "learning_rate": 1.9574083549758355e-05, "loss": 0.2869, "step": 558 }, { "epoch": 0.12, "learning_rate": 1.9572062752479684e-05, "loss": 0.3775, "step": 559 }, { "epoch": 0.12, "learning_rate": 1.9570037277428164e-05, "loss": 0.3528, "step": 560 }, { "epoch": 0.12, "learning_rate": 1.9568007125593626e-05, "loss": 0.3285, "step": 561 }, { "epoch": 0.12, "learning_rate": 1.9565972297968185e-05, "loss": 0.3434, "step": 562 }, { "epoch": 0.12, "learning_rate": 1.9563932795546244e-05, "loss": 0.3319, "step": 563 }, { "epoch": 0.12, "learning_rate": 1.9561888619324484e-05, "loss": 0.2905, "step": 564 }, { "epoch": 0.12, "learning_rate": 1.9559839770301878e-05, "loss": 0.2963, "step": 565 }, { "epoch": 0.12, "learning_rate": 1.9557786249479673e-05, "loss": 0.333, "step": 566 }, { "epoch": 0.12, "learning_rate": 1.955572805786141e-05, "loss": 0.3064, "step": 567 }, { "epoch": 0.12, "learning_rate": 1.9553665196452903e-05, "loss": 0.2818, "step": 568 }, { "epoch": 0.12, "learning_rate": 1.955159766626225e-05, "loss": 0.271, "step": 569 }, { "epoch": 0.12, "learning_rate": 1.9549525468299842e-05, "loss": 0.337, "step": 570 }, { "epoch": 0.12, "learning_rate": 1.954744860357833e-05, "loss": 0.3277, "step": 571 }, { "epoch": 0.12, "learning_rate": 1.9545367073112666e-05, "loss": 0.3383, "step": 572 }, { "epoch": 0.12, "learning_rate": 1.9543280877920073e-05, "loss": 0.3387, "step": 573 }, { "epoch": 0.12, "learning_rate": 1.9541190019020048e-05, "loss": 0.288, "step": 574 }, { "epoch": 0.12, "learning_rate": 1.9539094497434377e-05, "loss": 0.2868, "step": 575 }, { "epoch": 0.12, "learning_rate": 1.9536994314187122e-05, "loss": 0.3135, "step": 576 }, { "epoch": 0.12, "learning_rate": 1.953488947030462e-05, "loss": 0.345, "step": 577 }, { "epoch": 0.12, "learning_rate": 1.9532779966815484e-05, "loss": 0.3246, "step": 578 }, { "epoch": 0.12, "learning_rate": 1.9530665804750618e-05, "loss": 0.2693, "step": 579 }, { "epoch": 0.13, "learning_rate": 1.952854698514318e-05, "loss": 0.3252, "step": 580 }, { "epoch": 0.13, "learning_rate": 1.9526423509028627e-05, "loss": 0.3345, "step": 581 }, { "epoch": 0.13, "learning_rate": 1.9524295377444673e-05, "loss": 0.2843, "step": 582 }, { "epoch": 0.13, "learning_rate": 1.952216259143132e-05, "loss": 0.3349, "step": 583 }, { "epoch": 0.13, "learning_rate": 1.952002515203083e-05, "loss": 0.3201, "step": 584 }, { "epoch": 0.13, "learning_rate": 1.9517883060287765e-05, "loss": 0.3221, "step": 585 }, { "epoch": 0.13, "learning_rate": 1.951573631724893e-05, "loss": 0.3569, "step": 586 }, { "epoch": 0.13, "learning_rate": 1.9513584923963426e-05, "loss": 0.3172, "step": 587 }, { "epoch": 0.13, "learning_rate": 1.951142888148261e-05, "loss": 0.2948, "step": 588 }, { "epoch": 0.13, "learning_rate": 1.9509268190860124e-05, "loss": 0.3048, "step": 589 }, { "epoch": 0.13, "learning_rate": 1.9507102853151876e-05, "loss": 0.2891, "step": 590 }, { "epoch": 0.13, "learning_rate": 1.9504932869416045e-05, "loss": 0.3412, "step": 591 }, { "epoch": 0.13, "learning_rate": 1.9502758240713076e-05, "loss": 0.3321, "step": 592 }, { "epoch": 0.13, "learning_rate": 1.9500578968105696e-05, "loss": 0.3037, "step": 593 }, { "epoch": 0.13, "learning_rate": 1.9498395052658892e-05, "loss": 0.304, "step": 594 }, { "epoch": 0.13, "learning_rate": 1.9496206495439916e-05, "loss": 0.2941, "step": 595 }, { "epoch": 0.13, "learning_rate": 1.9494013297518302e-05, "loss": 0.3213, "step": 596 }, { "epoch": 0.13, "learning_rate": 1.949181545996584e-05, "loss": 0.2875, "step": 597 }, { "epoch": 0.13, "learning_rate": 1.948961298385659e-05, "loss": 0.284, "step": 598 }, { "epoch": 0.13, "learning_rate": 1.9487405870266888e-05, "loss": 0.3226, "step": 599 }, { "epoch": 0.13, "learning_rate": 1.9485194120275322e-05, "loss": 0.2405, "step": 600 }, { "epoch": 0.13, "learning_rate": 1.9482977734962753e-05, "loss": 0.3844, "step": 601 }, { "epoch": 0.13, "learning_rate": 1.948075671541231e-05, "loss": 0.2623, "step": 602 }, { "epoch": 0.13, "learning_rate": 1.9478531062709384e-05, "loss": 0.3288, "step": 603 }, { "epoch": 0.13, "learning_rate": 1.947630077794162e-05, "loss": 0.4324, "step": 604 }, { "epoch": 0.13, "learning_rate": 1.9474065862198948e-05, "loss": 0.28, "step": 605 }, { "epoch": 0.13, "learning_rate": 1.9471826316573543e-05, "loss": 0.2863, "step": 606 }, { "epoch": 0.13, "learning_rate": 1.9469582142159852e-05, "loss": 0.2516, "step": 607 }, { "epoch": 0.13, "learning_rate": 1.9467333340054576e-05, "loss": 0.2932, "step": 608 }, { "epoch": 0.13, "learning_rate": 1.9465079911356687e-05, "loss": 0.2541, "step": 609 }, { "epoch": 0.13, "learning_rate": 1.9462821857167412e-05, "loss": 0.3046, "step": 610 }, { "epoch": 0.13, "learning_rate": 1.9460559178590242e-05, "loss": 0.3314, "step": 611 }, { "epoch": 0.13, "learning_rate": 1.945829187673092e-05, "loss": 0.3276, "step": 612 }, { "epoch": 0.13, "learning_rate": 1.945601995269746e-05, "loss": 0.3711, "step": 613 }, { "epoch": 0.13, "learning_rate": 1.945374340760013e-05, "loss": 0.317, "step": 614 }, { "epoch": 0.13, "learning_rate": 1.945146224255145e-05, "loss": 0.3369, "step": 615 }, { "epoch": 0.13, "learning_rate": 1.9449176458666204e-05, "loss": 0.3136, "step": 616 }, { "epoch": 0.13, "learning_rate": 1.9446886057061435e-05, "loss": 0.3808, "step": 617 }, { "epoch": 0.13, "learning_rate": 1.9444591038856433e-05, "loss": 0.285, "step": 618 }, { "epoch": 0.13, "learning_rate": 1.9442291405172756e-05, "loss": 0.2942, "step": 619 }, { "epoch": 0.13, "learning_rate": 1.943998715713422e-05, "loss": 0.2913, "step": 620 }, { "epoch": 0.13, "learning_rate": 1.9437678295866872e-05, "loss": 0.3546, "step": 621 }, { "epoch": 0.13, "learning_rate": 1.943536482249904e-05, "loss": 0.259, "step": 622 }, { "epoch": 0.13, "learning_rate": 1.9433046738161293e-05, "loss": 0.2888, "step": 623 }, { "epoch": 0.13, "learning_rate": 1.9430724043986457e-05, "loss": 0.305, "step": 624 }, { "epoch": 0.13, "learning_rate": 1.942839674110961e-05, "loss": 0.2715, "step": 625 }, { "epoch": 0.14, "learning_rate": 1.942606483066808e-05, "loss": 0.2549, "step": 626 }, { "epoch": 0.14, "learning_rate": 1.9423728313801453e-05, "loss": 0.3278, "step": 627 }, { "epoch": 0.14, "learning_rate": 1.942138719165156e-05, "loss": 0.3151, "step": 628 }, { "epoch": 0.14, "learning_rate": 1.9419041465362477e-05, "loss": 0.3572, "step": 629 }, { "epoch": 0.14, "learning_rate": 1.9416691136080548e-05, "loss": 0.3094, "step": 630 }, { "epoch": 0.14, "learning_rate": 1.9414336204954353e-05, "loss": 0.2799, "step": 631 }, { "epoch": 0.14, "learning_rate": 1.941197667313472e-05, "loss": 0.3114, "step": 632 }, { "epoch": 0.14, "learning_rate": 1.940961254177473e-05, "loss": 0.2615, "step": 633 }, { "epoch": 0.14, "learning_rate": 1.940724381202971e-05, "loss": 0.3978, "step": 634 }, { "epoch": 0.14, "learning_rate": 1.940487048505724e-05, "loss": 0.2923, "step": 635 }, { "epoch": 0.14, "learning_rate": 1.9402492562017136e-05, "loss": 0.3282, "step": 636 }, { "epoch": 0.14, "learning_rate": 1.9400110044071468e-05, "loss": 0.3168, "step": 637 }, { "epoch": 0.14, "learning_rate": 1.939772293238455e-05, "loss": 0.3942, "step": 638 }, { "epoch": 0.14, "learning_rate": 1.939533122812294e-05, "loss": 0.3065, "step": 639 }, { "epoch": 0.14, "learning_rate": 1.939293493245544e-05, "loss": 0.3452, "step": 640 }, { "epoch": 0.14, "learning_rate": 1.939053404655309e-05, "loss": 0.2664, "step": 641 }, { "epoch": 0.14, "learning_rate": 1.9388128571589188e-05, "loss": 0.3304, "step": 642 }, { "epoch": 0.14, "learning_rate": 1.9385718508739263e-05, "loss": 0.3091, "step": 643 }, { "epoch": 0.14, "learning_rate": 1.9383303859181086e-05, "loss": 0.2896, "step": 644 }, { "epoch": 0.14, "learning_rate": 1.9380884624094677e-05, "loss": 0.3059, "step": 645 }, { "epoch": 0.14, "learning_rate": 1.937846080466229e-05, "loss": 0.2951, "step": 646 }, { "epoch": 0.14, "learning_rate": 1.937603240206842e-05, "loss": 0.2927, "step": 647 }, { "epoch": 0.14, "learning_rate": 1.9373599417499806e-05, "loss": 0.3332, "step": 648 }, { "epoch": 0.14, "learning_rate": 1.9371161852145426e-05, "loss": 0.3725, "step": 649 }, { "epoch": 0.14, "learning_rate": 1.936871970719649e-05, "loss": 0.334, "step": 650 }, { "epoch": 0.14, "learning_rate": 1.936627298384645e-05, "loss": 0.3188, "step": 651 }, { "epoch": 0.14, "learning_rate": 1.9363821683291e-05, "loss": 0.2959, "step": 652 }, { "epoch": 0.14, "learning_rate": 1.936136580672806e-05, "loss": 0.3925, "step": 653 }, { "epoch": 0.14, "learning_rate": 1.9358905355357798e-05, "loss": 0.3014, "step": 654 }, { "epoch": 0.14, "learning_rate": 1.9356440330382612e-05, "loss": 0.3011, "step": 655 }, { "epoch": 0.14, "learning_rate": 1.9353970733007133e-05, "loss": 0.3373, "step": 656 }, { "epoch": 0.14, "learning_rate": 1.9351496564438228e-05, "loss": 0.331, "step": 657 }, { "epoch": 0.14, "learning_rate": 1.9349017825885e-05, "loss": 0.3041, "step": 658 }, { "epoch": 0.14, "learning_rate": 1.934653451855879e-05, "loss": 0.2979, "step": 659 }, { "epoch": 0.14, "learning_rate": 1.9344046643673156e-05, "loss": 0.3458, "step": 660 }, { "epoch": 0.14, "learning_rate": 1.9341554202443906e-05, "loss": 0.3239, "step": 661 }, { "epoch": 0.14, "learning_rate": 1.9339057196089067e-05, "loss": 0.3281, "step": 662 }, { "epoch": 0.14, "learning_rate": 1.93365556258289e-05, "loss": 0.3056, "step": 663 }, { "epoch": 0.14, "learning_rate": 1.9334049492885906e-05, "loss": 0.3057, "step": 664 }, { "epoch": 0.14, "learning_rate": 1.93315387984848e-05, "loss": 0.2731, "step": 665 }, { "epoch": 0.14, "learning_rate": 1.932902354385253e-05, "loss": 0.2849, "step": 666 }, { "epoch": 0.14, "learning_rate": 1.9326503730218288e-05, "loss": 0.3005, "step": 667 }, { "epoch": 0.14, "learning_rate": 1.9323979358813476e-05, "loss": 0.3043, "step": 668 }, { "epoch": 0.14, "learning_rate": 1.932145043087173e-05, "loss": 0.3065, "step": 669 }, { "epoch": 0.14, "learning_rate": 1.9318916947628907e-05, "loss": 0.313, "step": 670 }, { "epoch": 0.14, "learning_rate": 1.93163789103231e-05, "loss": 0.3023, "step": 671 }, { "epoch": 0.15, "learning_rate": 1.9313836320194625e-05, "loss": 0.29, "step": 672 }, { "epoch": 0.15, "learning_rate": 1.931128917848602e-05, "loss": 0.2672, "step": 673 }, { "epoch": 0.15, "learning_rate": 1.9308737486442045e-05, "loss": 0.2521, "step": 674 }, { "epoch": 0.15, "learning_rate": 1.930618124530969e-05, "loss": 0.2807, "step": 675 }, { "epoch": 0.15, "learning_rate": 1.9303620456338157e-05, "loss": 0.4042, "step": 676 }, { "epoch": 0.15, "learning_rate": 1.930105512077889e-05, "loss": 0.2846, "step": 677 }, { "epoch": 0.15, "learning_rate": 1.929848523988554e-05, "loss": 0.3192, "step": 678 }, { "epoch": 0.15, "learning_rate": 1.9295910814913973e-05, "loss": 0.3659, "step": 679 }, { "epoch": 0.15, "learning_rate": 1.9293331847122298e-05, "loss": 0.2652, "step": 680 }, { "epoch": 0.15, "learning_rate": 1.9290748337770823e-05, "loss": 0.2593, "step": 681 }, { "epoch": 0.15, "learning_rate": 1.9288160288122086e-05, "loss": 0.2888, "step": 682 }, { "epoch": 0.15, "learning_rate": 1.9285567699440845e-05, "loss": 0.2744, "step": 683 }, { "epoch": 0.15, "learning_rate": 1.9282970572994068e-05, "loss": 0.3198, "step": 684 }, { "epoch": 0.15, "learning_rate": 1.9280368910050943e-05, "loss": 0.3194, "step": 685 }, { "epoch": 0.15, "learning_rate": 1.9277762711882884e-05, "loss": 0.3091, "step": 686 }, { "epoch": 0.15, "learning_rate": 1.927515197976351e-05, "loss": 0.2786, "step": 687 }, { "epoch": 0.15, "learning_rate": 1.927253671496866e-05, "loss": 0.3112, "step": 688 }, { "epoch": 0.15, "learning_rate": 1.926991691877639e-05, "loss": 0.287, "step": 689 }, { "epoch": 0.15, "learning_rate": 1.9267292592466965e-05, "loss": 0.3412, "step": 690 }, { "epoch": 0.15, "learning_rate": 1.9264663737322872e-05, "loss": 0.3576, "step": 691 }, { "epoch": 0.15, "learning_rate": 1.9262030354628804e-05, "loss": 0.336, "step": 692 }, { "epoch": 0.15, "learning_rate": 1.9259392445671673e-05, "loss": 0.3182, "step": 693 }, { "epoch": 0.15, "learning_rate": 1.9256750011740593e-05, "loss": 0.3142, "step": 694 }, { "epoch": 0.15, "learning_rate": 1.9254103054126897e-05, "loss": 0.2603, "step": 695 }, { "epoch": 0.15, "learning_rate": 1.9251451574124132e-05, "loss": 0.3778, "step": 696 }, { "epoch": 0.15, "learning_rate": 1.924879557302805e-05, "loss": 0.3902, "step": 697 }, { "epoch": 0.15, "learning_rate": 1.9246135052136603e-05, "loss": 0.3073, "step": 698 }, { "epoch": 0.15, "learning_rate": 1.9243470012749968e-05, "loss": 0.3165, "step": 699 }, { "epoch": 0.15, "learning_rate": 1.9240800456170526e-05, "loss": 0.3116, "step": 700 }, { "epoch": 0.15, "learning_rate": 1.923812638370286e-05, "loss": 0.3252, "step": 701 }, { "epoch": 0.15, "learning_rate": 1.9235447796653763e-05, "loss": 0.414, "step": 702 }, { "epoch": 0.15, "learning_rate": 1.923276469633223e-05, "loss": 0.3048, "step": 703 }, { "epoch": 0.15, "learning_rate": 1.9230077084049478e-05, "loss": 0.3221, "step": 704 }, { "epoch": 0.15, "learning_rate": 1.92273849611189e-05, "loss": 0.297, "step": 705 }, { "epoch": 0.15, "learning_rate": 1.9224688328856127e-05, "loss": 0.3099, "step": 706 }, { "epoch": 0.15, "learning_rate": 1.9221987188578964e-05, "loss": 0.2868, "step": 707 }, { "epoch": 0.15, "learning_rate": 1.9219281541607438e-05, "loss": 0.2652, "step": 708 }, { "epoch": 0.15, "learning_rate": 1.9216571389263768e-05, "loss": 0.4071, "step": 709 }, { "epoch": 0.15, "learning_rate": 1.921385673287238e-05, "loss": 0.2411, "step": 710 }, { "epoch": 0.15, "learning_rate": 1.9211137573759902e-05, "loss": 0.2745, "step": 711 }, { "epoch": 0.15, "learning_rate": 1.9208413913255158e-05, "loss": 0.2259, "step": 712 }, { "epoch": 0.15, "learning_rate": 1.9205685752689178e-05, "loss": 0.3011, "step": 713 }, { "epoch": 0.15, "learning_rate": 1.9202953093395185e-05, "loss": 0.312, "step": 714 }, { "epoch": 0.15, "learning_rate": 1.92002159367086e-05, "loss": 0.3163, "step": 715 }, { "epoch": 0.15, "learning_rate": 1.9197474283967046e-05, "loss": 0.3051, "step": 716 }, { "epoch": 0.15, "learning_rate": 1.919472813651034e-05, "loss": 0.2637, "step": 717 }, { "epoch": 0.15, "learning_rate": 1.91919774956805e-05, "loss": 0.3339, "step": 718 }, { "epoch": 0.16, "learning_rate": 1.918922236282174e-05, "loss": 0.2952, "step": 719 }, { "epoch": 0.16, "learning_rate": 1.918646273928046e-05, "loss": 0.2908, "step": 720 }, { "epoch": 0.16, "learning_rate": 1.9183698626405264e-05, "loss": 0.3185, "step": 721 }, { "epoch": 0.16, "learning_rate": 1.9180930025546947e-05, "loss": 0.3256, "step": 722 }, { "epoch": 0.16, "learning_rate": 1.9178156938058498e-05, "loss": 0.3624, "step": 723 }, { "epoch": 0.16, "learning_rate": 1.9175379365295093e-05, "loss": 0.2886, "step": 724 }, { "epoch": 0.16, "learning_rate": 1.917259730861411e-05, "loss": 0.3065, "step": 725 }, { "epoch": 0.16, "learning_rate": 1.916981076937511e-05, "loss": 0.3528, "step": 726 }, { "epoch": 0.16, "learning_rate": 1.9167019748939847e-05, "loss": 0.283, "step": 727 }, { "epoch": 0.16, "learning_rate": 1.9164224248672266e-05, "loss": 0.2793, "step": 728 }, { "epoch": 0.16, "learning_rate": 1.91614242699385e-05, "loss": 0.3, "step": 729 }, { "epoch": 0.16, "learning_rate": 1.9158619814106868e-05, "loss": 0.3317, "step": 730 }, { "epoch": 0.16, "learning_rate": 1.9155810882547888e-05, "loss": 0.2693, "step": 731 }, { "epoch": 0.16, "learning_rate": 1.9152997476634248e-05, "loss": 0.3634, "step": 732 }, { "epoch": 0.16, "learning_rate": 1.9150179597740842e-05, "loss": 0.3247, "step": 733 }, { "epoch": 0.16, "learning_rate": 1.9147357247244728e-05, "loss": 0.3486, "step": 734 }, { "epoch": 0.16, "learning_rate": 1.9144530426525167e-05, "loss": 0.2975, "step": 735 }, { "epoch": 0.16, "learning_rate": 1.9141699136963602e-05, "loss": 0.3168, "step": 736 }, { "epoch": 0.16, "learning_rate": 1.913886337994365e-05, "loss": 0.2773, "step": 737 }, { "epoch": 0.16, "learning_rate": 1.913602315685112e-05, "loss": 0.2823, "step": 738 }, { "epoch": 0.16, "learning_rate": 1.9133178469074003e-05, "loss": 0.3316, "step": 739 }, { "epoch": 0.16, "learning_rate": 1.913032931800247e-05, "loss": 0.2902, "step": 740 }, { "epoch": 0.16, "learning_rate": 1.9127475705028864e-05, "loss": 0.3255, "step": 741 }, { "epoch": 0.16, "learning_rate": 1.9124617631547728e-05, "loss": 0.2757, "step": 742 }, { "epoch": 0.16, "learning_rate": 1.9121755098955773e-05, "loss": 0.3215, "step": 743 }, { "epoch": 0.16, "learning_rate": 1.911888810865189e-05, "loss": 0.3309, "step": 744 }, { "epoch": 0.16, "learning_rate": 1.911601666203714e-05, "loss": 0.3345, "step": 745 }, { "epoch": 0.16, "learning_rate": 1.9113140760514787e-05, "loss": 0.3009, "step": 746 }, { "epoch": 0.16, "learning_rate": 1.9110260405490243e-05, "loss": 0.3722, "step": 747 }, { "epoch": 0.16, "learning_rate": 1.9107375598371112e-05, "loss": 0.2889, "step": 748 }, { "epoch": 0.16, "learning_rate": 1.9104486340567176e-05, "loss": 0.2952, "step": 749 }, { "epoch": 0.16, "learning_rate": 1.9101592633490376e-05, "loss": 0.4081, "step": 750 }, { "epoch": 0.16, "learning_rate": 1.9098694478554853e-05, "loss": 0.3204, "step": 751 }, { "epoch": 0.16, "learning_rate": 1.9095791877176895e-05, "loss": 0.3306, "step": 752 }, { "epoch": 0.16, "learning_rate": 1.9092884830774978e-05, "loss": 0.3261, "step": 753 }, { "epoch": 0.16, "learning_rate": 1.9089973340769745e-05, "loss": 0.3127, "step": 754 }, { "epoch": 0.16, "learning_rate": 1.908705740858402e-05, "loss": 0.352, "step": 755 }, { "epoch": 0.16, "learning_rate": 1.908413703564278e-05, "loss": 0.2822, "step": 756 }, { "epoch": 0.16, "learning_rate": 1.9081212223373192e-05, "loss": 0.2864, "step": 757 }, { "epoch": 0.16, "learning_rate": 1.907828297320458e-05, "loss": 0.2787, "step": 758 }, { "epoch": 0.16, "learning_rate": 1.9075349286568435e-05, "loss": 0.2939, "step": 759 }, { "epoch": 0.16, "learning_rate": 1.9072411164898427e-05, "loss": 0.2996, "step": 760 }, { "epoch": 0.16, "learning_rate": 1.9069468609630385e-05, "loss": 0.3344, "step": 761 }, { "epoch": 0.16, "learning_rate": 1.9066521622202304e-05, "loss": 0.305, "step": 762 }, { "epoch": 0.16, "learning_rate": 1.906357020405435e-05, "loss": 0.2813, "step": 763 }, { "epoch": 0.16, "learning_rate": 1.9060614356628853e-05, "loss": 0.346, "step": 764 }, { "epoch": 0.17, "learning_rate": 1.90576540813703e-05, "loss": 0.3439, "step": 765 }, { "epoch": 0.17, "learning_rate": 1.905468937972536e-05, "loss": 0.2957, "step": 766 }, { "epoch": 0.17, "learning_rate": 1.9051720253142848e-05, "loss": 0.3338, "step": 767 }, { "epoch": 0.17, "learning_rate": 1.9048746703073747e-05, "loss": 0.2435, "step": 768 }, { "epoch": 0.17, "learning_rate": 1.9045768730971198e-05, "loss": 0.2856, "step": 769 }, { "epoch": 0.17, "learning_rate": 1.9042786338290507e-05, "loss": 0.3458, "step": 770 }, { "epoch": 0.17, "learning_rate": 1.903979952648915e-05, "loss": 0.3366, "step": 771 }, { "epoch": 0.17, "learning_rate": 1.9036808297026742e-05, "loss": 0.3153, "step": 772 }, { "epoch": 0.17, "learning_rate": 1.9033812651365067e-05, "loss": 0.3396, "step": 773 }, { "epoch": 0.17, "learning_rate": 1.9030812590968074e-05, "loss": 0.3322, "step": 774 }, { "epoch": 0.17, "learning_rate": 1.9027808117301862e-05, "loss": 0.2987, "step": 775 }, { "epoch": 0.17, "learning_rate": 1.9024799231834685e-05, "loss": 0.2798, "step": 776 }, { "epoch": 0.17, "learning_rate": 1.9021785936036954e-05, "loss": 0.2936, "step": 777 }, { "epoch": 0.17, "learning_rate": 1.901876823138124e-05, "loss": 0.3121, "step": 778 }, { "epoch": 0.17, "learning_rate": 1.9015746119342267e-05, "loss": 0.2978, "step": 779 }, { "epoch": 0.17, "learning_rate": 1.9012719601396906e-05, "loss": 0.2796, "step": 780 }, { "epoch": 0.17, "learning_rate": 1.900968867902419e-05, "loss": 0.2948, "step": 781 }, { "epoch": 0.17, "learning_rate": 1.9006653353705306e-05, "loss": 0.3539, "step": 782 }, { "epoch": 0.17, "learning_rate": 1.900361362692358e-05, "loss": 0.3188, "step": 783 }, { "epoch": 0.17, "learning_rate": 1.90005695001645e-05, "loss": 0.2866, "step": 784 }, { "epoch": 0.17, "learning_rate": 1.8997520974915695e-05, "loss": 0.335, "step": 785 }, { "epoch": 0.17, "learning_rate": 1.899446805266696e-05, "loss": 0.312, "step": 786 }, { "epoch": 0.17, "learning_rate": 1.899141073491022e-05, "loss": 0.3125, "step": 787 }, { "epoch": 0.17, "learning_rate": 1.8988349023139558e-05, "loss": 0.3517, "step": 788 }, { "epoch": 0.17, "learning_rate": 1.89852829188512e-05, "loss": 0.3143, "step": 789 }, { "epoch": 0.17, "learning_rate": 1.898221242354353e-05, "loss": 0.3537, "step": 790 }, { "epoch": 0.17, "learning_rate": 1.8979137538717058e-05, "loss": 0.28, "step": 791 }, { "epoch": 0.17, "learning_rate": 1.8976058265874458e-05, "loss": 0.3728, "step": 792 }, { "epoch": 0.17, "learning_rate": 1.8972974606520532e-05, "loss": 0.3218, "step": 793 }, { "epoch": 0.17, "learning_rate": 1.896988656216224e-05, "loss": 0.3126, "step": 794 }, { "epoch": 0.17, "learning_rate": 1.896679413430868e-05, "loss": 0.2596, "step": 795 }, { "epoch": 0.17, "learning_rate": 1.896369732447109e-05, "loss": 0.3175, "step": 796 }, { "epoch": 0.17, "learning_rate": 1.8960596134162845e-05, "loss": 0.3184, "step": 797 }, { "epoch": 0.17, "learning_rate": 1.895749056489947e-05, "loss": 0.2764, "step": 798 }, { "epoch": 0.17, "learning_rate": 1.8954380618198633e-05, "loss": 0.2818, "step": 799 }, { "epoch": 0.17, "learning_rate": 1.895126629558012e-05, "loss": 0.323, "step": 800 }, { "epoch": 0.17, "learning_rate": 1.8948147598565882e-05, "loss": 0.244, "step": 801 }, { "epoch": 0.17, "learning_rate": 1.894502452867999e-05, "loss": 0.3387, "step": 802 }, { "epoch": 0.17, "learning_rate": 1.8941897087448658e-05, "loss": 0.3299, "step": 803 }, { "epoch": 0.17, "learning_rate": 1.8938765276400237e-05, "loss": 0.3248, "step": 804 }, { "epoch": 0.17, "learning_rate": 1.893562909706521e-05, "loss": 0.2735, "step": 805 }, { "epoch": 0.17, "learning_rate": 1.8932488550976202e-05, "loss": 0.3279, "step": 806 }, { "epoch": 0.17, "learning_rate": 1.892934363966796e-05, "loss": 0.2666, "step": 807 }, { "epoch": 0.17, "learning_rate": 1.892619436467738e-05, "loss": 0.319, "step": 808 }, { "epoch": 0.17, "learning_rate": 1.8923040727543474e-05, "loss": 0.2982, "step": 809 }, { "epoch": 0.17, "learning_rate": 1.8919882729807395e-05, "loss": 0.3286, "step": 810 }, { "epoch": 0.18, "learning_rate": 1.8916720373012425e-05, "loss": 0.3056, "step": 811 }, { "epoch": 0.18, "learning_rate": 1.891355365870398e-05, "loss": 0.3132, "step": 812 }, { "epoch": 0.18, "learning_rate": 1.8910382588429603e-05, "loss": 0.3178, "step": 813 }, { "epoch": 0.18, "learning_rate": 1.8907207163738956e-05, "loss": 0.2959, "step": 814 }, { "epoch": 0.18, "learning_rate": 1.8904027386183847e-05, "loss": 0.3192, "step": 815 }, { "epoch": 0.18, "learning_rate": 1.89008432573182e-05, "loss": 0.2997, "step": 816 }, { "epoch": 0.18, "learning_rate": 1.8897654778698064e-05, "loss": 0.3542, "step": 817 }, { "epoch": 0.18, "learning_rate": 1.8894461951881616e-05, "loss": 0.2996, "step": 818 }, { "epoch": 0.18, "learning_rate": 1.8891264778429165e-05, "loss": 0.3192, "step": 819 }, { "epoch": 0.18, "learning_rate": 1.8888063259903137e-05, "loss": 0.2919, "step": 820 }, { "epoch": 0.18, "learning_rate": 1.888485739786808e-05, "loss": 0.3606, "step": 821 }, { "epoch": 0.18, "learning_rate": 1.8881647193890667e-05, "loss": 0.3349, "step": 822 }, { "epoch": 0.18, "learning_rate": 1.8878432649539697e-05, "loss": 0.2301, "step": 823 }, { "epoch": 0.18, "learning_rate": 1.8875213766386084e-05, "loss": 0.3175, "step": 824 }, { "epoch": 0.18, "learning_rate": 1.887199054600286e-05, "loss": 0.3118, "step": 825 }, { "epoch": 0.18, "learning_rate": 1.8868762989965187e-05, "loss": 0.3021, "step": 826 }, { "epoch": 0.18, "learning_rate": 1.886553109985034e-05, "loss": 0.2724, "step": 827 }, { "epoch": 0.18, "learning_rate": 1.886229487723771e-05, "loss": 0.2834, "step": 828 }, { "epoch": 0.18, "learning_rate": 1.8859054323708804e-05, "loss": 0.3316, "step": 829 }, { "epoch": 0.18, "learning_rate": 1.8855809440847253e-05, "loss": 0.3235, "step": 830 }, { "epoch": 0.18, "learning_rate": 1.88525602302388e-05, "loss": 0.2831, "step": 831 }, { "epoch": 0.18, "learning_rate": 1.8849306693471297e-05, "loss": 0.3192, "step": 832 }, { "epoch": 0.18, "learning_rate": 1.884604883213472e-05, "loss": 0.2656, "step": 833 }, { "epoch": 0.18, "learning_rate": 1.884278664782115e-05, "loss": 0.2618, "step": 834 }, { "epoch": 0.18, "learning_rate": 1.883952014212479e-05, "loss": 0.2951, "step": 835 }, { "epoch": 0.18, "learning_rate": 1.8836249316641944e-05, "loss": 0.2765, "step": 836 }, { "epoch": 0.18, "learning_rate": 1.8832974172971036e-05, "loss": 0.3108, "step": 837 }, { "epoch": 0.18, "learning_rate": 1.882969471271259e-05, "loss": 0.3422, "step": 838 }, { "epoch": 0.18, "learning_rate": 1.8826410937469256e-05, "loss": 0.3059, "step": 839 }, { "epoch": 0.18, "learning_rate": 1.8823122848845777e-05, "loss": 0.3088, "step": 840 }, { "epoch": 0.18, "learning_rate": 1.8819830448449005e-05, "loss": 0.3651, "step": 841 }, { "epoch": 0.18, "learning_rate": 1.8816533737887913e-05, "loss": 0.2912, "step": 842 }, { "epoch": 0.18, "learning_rate": 1.8813232718773564e-05, "loss": 0.2588, "step": 843 }, { "epoch": 0.18, "learning_rate": 1.8809927392719136e-05, "loss": 0.3156, "step": 844 }, { "epoch": 0.18, "learning_rate": 1.880661776133991e-05, "loss": 0.3234, "step": 845 }, { "epoch": 0.18, "learning_rate": 1.880330382625327e-05, "loss": 0.3957, "step": 846 }, { "epoch": 0.18, "learning_rate": 1.87999855890787e-05, "loss": 0.2575, "step": 847 }, { "epoch": 0.18, "learning_rate": 1.8796663051437795e-05, "loss": 0.3576, "step": 848 }, { "epoch": 0.18, "learning_rate": 1.8793336214954245e-05, "loss": 0.3515, "step": 849 }, { "epoch": 0.18, "learning_rate": 1.8790005081253843e-05, "loss": 0.3334, "step": 850 }, { "epoch": 0.18, "learning_rate": 1.8786669651964478e-05, "loss": 0.2831, "step": 851 }, { "epoch": 0.18, "learning_rate": 1.8783329928716142e-05, "loss": 0.299, "step": 852 }, { "epoch": 0.18, "learning_rate": 1.8779985913140927e-05, "loss": 0.3098, "step": 853 }, { "epoch": 0.18, "learning_rate": 1.877663760687302e-05, "loss": 0.2843, "step": 854 }, { "epoch": 0.18, "learning_rate": 1.8773285011548702e-05, "loss": 0.4289, "step": 855 }, { "epoch": 0.18, "learning_rate": 1.8769928128806357e-05, "loss": 0.3251, "step": 856 }, { "epoch": 0.18, "learning_rate": 1.8766566960286464e-05, "loss": 0.2825, "step": 857 }, { "epoch": 0.19, "learning_rate": 1.8763201507631585e-05, "loss": 0.2933, "step": 858 }, { "epoch": 0.19, "learning_rate": 1.875983177248639e-05, "loss": 0.3368, "step": 859 }, { "epoch": 0.19, "learning_rate": 1.875645775649763e-05, "loss": 0.3077, "step": 860 }, { "epoch": 0.19, "learning_rate": 1.8753079461314158e-05, "loss": 0.2838, "step": 861 }, { "epoch": 0.19, "learning_rate": 1.874969688858691e-05, "loss": 0.3134, "step": 862 }, { "epoch": 0.19, "learning_rate": 1.8746310039968918e-05, "loss": 0.3277, "step": 863 }, { "epoch": 0.19, "learning_rate": 1.8742918917115304e-05, "loss": 0.3266, "step": 864 }, { "epoch": 0.19, "learning_rate": 1.873952352168327e-05, "loss": 0.2682, "step": 865 }, { "epoch": 0.19, "learning_rate": 1.8736123855332116e-05, "loss": 0.3285, "step": 866 }, { "epoch": 0.19, "learning_rate": 1.873271991972323e-05, "loss": 0.2818, "step": 867 }, { "epoch": 0.19, "learning_rate": 1.8729311716520075e-05, "loss": 0.2956, "step": 868 }, { "epoch": 0.19, "learning_rate": 1.8725899247388207e-05, "loss": 0.3549, "step": 869 }, { "epoch": 0.19, "learning_rate": 1.872248251399527e-05, "loss": 0.3636, "step": 870 }, { "epoch": 0.19, "learning_rate": 1.8719061518010986e-05, "loss": 0.2811, "step": 871 }, { "epoch": 0.19, "learning_rate": 1.8715636261107158e-05, "loss": 0.3177, "step": 872 }, { "epoch": 0.19, "learning_rate": 1.8712206744957684e-05, "loss": 0.3019, "step": 873 }, { "epoch": 0.19, "learning_rate": 1.8708772971238528e-05, "loss": 0.3244, "step": 874 }, { "epoch": 0.19, "learning_rate": 1.8705334941627745e-05, "loss": 0.2925, "step": 875 }, { "epoch": 0.19, "learning_rate": 1.870189265780546e-05, "loss": 0.332, "step": 876 }, { "epoch": 0.19, "learning_rate": 1.869844612145389e-05, "loss": 0.2929, "step": 877 }, { "epoch": 0.19, "learning_rate": 1.8694995334257323e-05, "loss": 0.337, "step": 878 }, { "epoch": 0.19, "learning_rate": 1.8691540297902123e-05, "loss": 0.2795, "step": 879 }, { "epoch": 0.19, "learning_rate": 1.8688081014076728e-05, "loss": 0.2897, "step": 880 }, { "epoch": 0.19, "learning_rate": 1.8684617484471662e-05, "loss": 0.3523, "step": 881 }, { "epoch": 0.19, "learning_rate": 1.868114971077952e-05, "loss": 0.3462, "step": 882 }, { "epoch": 0.19, "learning_rate": 1.867767769469496e-05, "loss": 0.3359, "step": 883 }, { "epoch": 0.19, "learning_rate": 1.8674201437914727e-05, "loss": 0.2862, "step": 884 }, { "epoch": 0.19, "learning_rate": 1.8670720942137636e-05, "loss": 0.2682, "step": 885 }, { "epoch": 0.19, "learning_rate": 1.8667236209064565e-05, "loss": 0.302, "step": 886 }, { "epoch": 0.19, "learning_rate": 1.8663747240398478e-05, "loss": 0.3136, "step": 887 }, { "epoch": 0.19, "learning_rate": 1.866025403784439e-05, "loss": 0.2546, "step": 888 }, { "epoch": 0.19, "learning_rate": 1.8656756603109397e-05, "loss": 0.3197, "step": 889 }, { "epoch": 0.19, "learning_rate": 1.8653254937902666e-05, "loss": 0.3697, "step": 890 }, { "epoch": 0.19, "learning_rate": 1.864974904393542e-05, "loss": 0.2912, "step": 891 }, { "epoch": 0.19, "learning_rate": 1.8646238922920955e-05, "loss": 0.3335, "step": 892 }, { "epoch": 0.19, "learning_rate": 1.8642724576574642e-05, "loss": 0.3024, "step": 893 }, { "epoch": 0.19, "learning_rate": 1.8639206006613893e-05, "loss": 0.2973, "step": 894 }, { "epoch": 0.19, "learning_rate": 1.8635683214758213e-05, "loss": 0.2825, "step": 895 }, { "epoch": 0.19, "learning_rate": 1.8632156202729147e-05, "loss": 0.3506, "step": 896 }, { "epoch": 0.19, "learning_rate": 1.8628624972250315e-05, "loss": 0.3017, "step": 897 }, { "epoch": 0.19, "learning_rate": 1.8625089525047388e-05, "loss": 0.2705, "step": 898 }, { "epoch": 0.19, "learning_rate": 1.8621549862848108e-05, "loss": 0.3112, "step": 899 }, { "epoch": 0.19, "learning_rate": 1.861800598738228e-05, "loss": 0.3077, "step": 900 }, { "epoch": 0.19, "learning_rate": 1.8614457900381747e-05, "loss": 0.285, "step": 901 }, { "epoch": 0.19, "learning_rate": 1.861090560358044e-05, "loss": 0.2381, "step": 902 }, { "epoch": 0.19, "learning_rate": 1.860734909871432e-05, "loss": 0.2693, "step": 903 }, { "epoch": 0.2, "learning_rate": 1.860378838752142e-05, "loss": 0.2694, "step": 904 }, { "epoch": 0.2, "learning_rate": 1.860022347174182e-05, "loss": 0.2994, "step": 905 }, { "epoch": 0.2, "learning_rate": 1.859665435311767e-05, "loss": 0.3138, "step": 906 }, { "epoch": 0.2, "learning_rate": 1.859308103339315e-05, "loss": 0.2735, "step": 907 }, { "epoch": 0.2, "learning_rate": 1.8589503514314516e-05, "loss": 0.3023, "step": 908 }, { "epoch": 0.2, "learning_rate": 1.8585921797630064e-05, "loss": 0.3024, "step": 909 }, { "epoch": 0.2, "learning_rate": 1.858233588509014e-05, "loss": 0.2875, "step": 910 }, { "epoch": 0.2, "learning_rate": 1.857874577844715e-05, "loss": 0.3037, "step": 911 }, { "epoch": 0.2, "learning_rate": 1.8575151479455536e-05, "loss": 0.3316, "step": 912 }, { "epoch": 0.2, "learning_rate": 1.8571552989871805e-05, "loss": 0.243, "step": 913 }, { "epoch": 0.2, "learning_rate": 1.8567950311454498e-05, "loss": 0.3058, "step": 914 }, { "epoch": 0.2, "learning_rate": 1.856434344596421e-05, "loss": 0.2651, "step": 915 }, { "epoch": 0.2, "learning_rate": 1.8560732395163585e-05, "loss": 0.3077, "step": 916 }, { "epoch": 0.2, "learning_rate": 1.8557117160817298e-05, "loss": 0.2714, "step": 917 }, { "epoch": 0.2, "learning_rate": 1.8553497744692085e-05, "loss": 0.3268, "step": 918 }, { "epoch": 0.2, "learning_rate": 1.854987414855672e-05, "loss": 0.4109, "step": 919 }, { "epoch": 0.2, "learning_rate": 1.8546246374182013e-05, "loss": 0.282, "step": 920 }, { "epoch": 0.2, "learning_rate": 1.854261442334083e-05, "loss": 0.3568, "step": 921 }, { "epoch": 0.2, "learning_rate": 1.8538978297808063e-05, "loss": 0.339, "step": 922 }, { "epoch": 0.2, "learning_rate": 1.8535337999360655e-05, "loss": 0.3216, "step": 923 }, { "epoch": 0.2, "learning_rate": 1.853169352977758e-05, "loss": 0.3096, "step": 924 }, { "epoch": 0.2, "learning_rate": 1.8528044890839855e-05, "loss": 0.3177, "step": 925 }, { "epoch": 0.2, "learning_rate": 1.852439208433054e-05, "loss": 0.3631, "step": 926 }, { "epoch": 0.2, "learning_rate": 1.8520735112034716e-05, "loss": 0.2918, "step": 927 }, { "epoch": 0.2, "learning_rate": 1.8517073975739514e-05, "loss": 0.3386, "step": 928 }, { "epoch": 0.2, "learning_rate": 1.85134086772341e-05, "loss": 0.3247, "step": 929 }, { "epoch": 0.2, "learning_rate": 1.850973921830966e-05, "loss": 0.2628, "step": 930 }, { "epoch": 0.2, "learning_rate": 1.8506065600759432e-05, "loss": 0.319, "step": 931 }, { "epoch": 0.2, "learning_rate": 1.8502387826378668e-05, "loss": 0.2975, "step": 932 }, { "epoch": 0.2, "learning_rate": 1.8498705896964665e-05, "loss": 0.3611, "step": 933 }, { "epoch": 0.2, "learning_rate": 1.8495019814316747e-05, "loss": 0.2859, "step": 934 }, { "epoch": 0.2, "learning_rate": 1.8491329580236264e-05, "loss": 0.2938, "step": 935 }, { "epoch": 0.2, "learning_rate": 1.8487635196526594e-05, "loss": 0.3044, "step": 936 }, { "epoch": 0.2, "learning_rate": 1.8483936664993152e-05, "loss": 0.3072, "step": 937 }, { "epoch": 0.2, "learning_rate": 1.848023398744337e-05, "loss": 0.3048, "step": 938 }, { "epoch": 0.2, "learning_rate": 1.8476527165686708e-05, "loss": 0.2951, "step": 939 }, { "epoch": 0.2, "learning_rate": 1.8472816201534662e-05, "loss": 0.3594, "step": 940 }, { "epoch": 0.2, "learning_rate": 1.8469101096800733e-05, "loss": 0.2696, "step": 941 }, { "epoch": 0.2, "learning_rate": 1.8465381853300465e-05, "loss": 0.2972, "step": 942 }, { "epoch": 0.2, "learning_rate": 1.846165847285141e-05, "loss": 0.345, "step": 943 }, { "epoch": 0.2, "learning_rate": 1.845793095727315e-05, "loss": 0.3043, "step": 944 }, { "epoch": 0.2, "learning_rate": 1.8454199308387285e-05, "loss": 0.3051, "step": 945 }, { "epoch": 0.2, "learning_rate": 1.8450463528017433e-05, "loss": 0.2737, "step": 946 }, { "epoch": 0.2, "learning_rate": 1.8446723617989238e-05, "loss": 0.3466, "step": 947 }, { "epoch": 0.2, "learning_rate": 1.844297958013035e-05, "loss": 0.2863, "step": 948 }, { "epoch": 0.2, "learning_rate": 1.8439231416270452e-05, "loss": 0.3499, "step": 949 }, { "epoch": 0.21, "learning_rate": 1.8435479128241228e-05, "loss": 0.2442, "step": 950 }, { "epoch": 0.21, "learning_rate": 1.8431722717876383e-05, "loss": 0.2903, "step": 951 }, { "epoch": 0.21, "learning_rate": 1.8427962187011644e-05, "loss": 0.3504, "step": 952 }, { "epoch": 0.21, "learning_rate": 1.842419753748474e-05, "loss": 0.322, "step": 953 }, { "epoch": 0.21, "learning_rate": 1.8420428771135425e-05, "loss": 0.2622, "step": 954 }, { "epoch": 0.21, "learning_rate": 1.8416655889805448e-05, "loss": 0.2702, "step": 955 }, { "epoch": 0.21, "learning_rate": 1.841287889533859e-05, "loss": 0.3695, "step": 956 }, { "epoch": 0.21, "learning_rate": 1.8409097789580625e-05, "loss": 0.2785, "step": 957 }, { "epoch": 0.21, "learning_rate": 1.840531257437934e-05, "loss": 0.2996, "step": 958 }, { "epoch": 0.21, "learning_rate": 1.8401523251584536e-05, "loss": 0.3015, "step": 959 }, { "epoch": 0.21, "learning_rate": 1.8397729823048018e-05, "loss": 0.2584, "step": 960 }, { "epoch": 0.21, "learning_rate": 1.83939322906236e-05, "loss": 0.2573, "step": 961 }, { "epoch": 0.21, "learning_rate": 1.8390130656167087e-05, "loss": 0.2897, "step": 962 }, { "epoch": 0.21, "learning_rate": 1.838632492153631e-05, "loss": 0.3205, "step": 963 }, { "epoch": 0.21, "learning_rate": 1.8382515088591095e-05, "loss": 0.284, "step": 964 }, { "epoch": 0.21, "learning_rate": 1.837870115919327e-05, "loss": 0.297, "step": 965 }, { "epoch": 0.21, "learning_rate": 1.8374883135206654e-05, "loss": 0.3156, "step": 966 }, { "epoch": 0.21, "learning_rate": 1.8371061018497093e-05, "loss": 0.2604, "step": 967 }, { "epoch": 0.21, "learning_rate": 1.8367234810932405e-05, "loss": 0.2922, "step": 968 }, { "epoch": 0.21, "learning_rate": 1.8363404514382426e-05, "loss": 0.2641, "step": 969 }, { "epoch": 0.21, "learning_rate": 1.8359570130718987e-05, "loss": 0.3608, "step": 970 }, { "epoch": 0.21, "learning_rate": 1.835573166181591e-05, "loss": 0.3208, "step": 971 }, { "epoch": 0.21, "learning_rate": 1.8351889109549013e-05, "loss": 0.308, "step": 972 }, { "epoch": 0.21, "learning_rate": 1.8348042475796122e-05, "loss": 0.2862, "step": 973 }, { "epoch": 0.21, "learning_rate": 1.8344191762437048e-05, "loss": 0.2735, "step": 974 }, { "epoch": 0.21, "learning_rate": 1.834033697135359e-05, "loss": 0.3012, "step": 975 }, { "epoch": 0.21, "learning_rate": 1.8336478104429543e-05, "loss": 0.3301, "step": 976 }, { "epoch": 0.21, "learning_rate": 1.833261516355071e-05, "loss": 0.3153, "step": 977 }, { "epoch": 0.21, "learning_rate": 1.8328748150604868e-05, "loss": 0.3686, "step": 978 }, { "epoch": 0.21, "learning_rate": 1.8324877067481782e-05, "loss": 0.2991, "step": 979 }, { "epoch": 0.21, "learning_rate": 1.8321001916073223e-05, "loss": 0.3029, "step": 980 }, { "epoch": 0.21, "learning_rate": 1.8317122698272926e-05, "loss": 0.3072, "step": 981 }, { "epoch": 0.21, "learning_rate": 1.8313239415976636e-05, "loss": 0.3511, "step": 982 }, { "epoch": 0.21, "learning_rate": 1.8309352071082077e-05, "loss": 0.2979, "step": 983 }, { "epoch": 0.21, "learning_rate": 1.830546066548895e-05, "loss": 0.3298, "step": 984 }, { "epoch": 0.21, "learning_rate": 1.8301565201098942e-05, "loss": 0.3199, "step": 985 }, { "epoch": 0.21, "learning_rate": 1.8297665679815742e-05, "loss": 0.352, "step": 986 }, { "epoch": 0.21, "learning_rate": 1.8293762103544997e-05, "loss": 0.3026, "step": 987 }, { "epoch": 0.21, "learning_rate": 1.828985447419435e-05, "loss": 0.3647, "step": 988 }, { "epoch": 0.21, "learning_rate": 1.8285942793673418e-05, "loss": 0.3337, "step": 989 }, { "epoch": 0.21, "learning_rate": 1.8282027063893802e-05, "loss": 0.3165, "step": 990 }, { "epoch": 0.21, "learning_rate": 1.8278107286769086e-05, "loss": 0.3181, "step": 991 }, { "epoch": 0.21, "learning_rate": 1.827418346421482e-05, "loss": 0.3373, "step": 992 }, { "epoch": 0.21, "learning_rate": 1.8270255598148542e-05, "loss": 0.3018, "step": 993 }, { "epoch": 0.21, "learning_rate": 1.8266323690489757e-05, "loss": 0.2966, "step": 994 }, { "epoch": 0.21, "learning_rate": 1.826238774315995e-05, "loss": 0.3228, "step": 995 }, { "epoch": 0.21, "learning_rate": 1.825844775808258e-05, "loss": 0.3272, "step": 996 }, { "epoch": 0.22, "learning_rate": 1.825450373718309e-05, "loss": 0.3264, "step": 997 }, { "epoch": 0.22, "learning_rate": 1.8250555682388866e-05, "loss": 0.2979, "step": 998 }, { "epoch": 0.22, "learning_rate": 1.8246603595629297e-05, "loss": 0.3431, "step": 999 }, { "epoch": 0.22, "learning_rate": 1.8242647478835717e-05, "loss": 0.3082, "step": 1000 }, { "epoch": 0.22, "learning_rate": 1.8238687333941455e-05, "loss": 0.326, "step": 1001 }, { "epoch": 0.22, "learning_rate": 1.8234723162881786e-05, "loss": 0.3164, "step": 1002 }, { "epoch": 0.22, "learning_rate": 1.823075496759396e-05, "loss": 0.2846, "step": 1003 }, { "epoch": 0.22, "learning_rate": 1.8226782750017206e-05, "loss": 0.324, "step": 1004 }, { "epoch": 0.22, "learning_rate": 1.8222806512092695e-05, "loss": 0.3226, "step": 1005 }, { "epoch": 0.22, "learning_rate": 1.821882625576358e-05, "loss": 0.3111, "step": 1006 }, { "epoch": 0.22, "learning_rate": 1.8214841982974975e-05, "loss": 0.2801, "step": 1007 }, { "epoch": 0.22, "learning_rate": 1.821085369567395e-05, "loss": 0.3113, "step": 1008 }, { "epoch": 0.22, "learning_rate": 1.820686139580955e-05, "loss": 0.3002, "step": 1009 }, { "epoch": 0.22, "learning_rate": 1.8202865085332768e-05, "loss": 0.2829, "step": 1010 }, { "epoch": 0.22, "learning_rate": 1.819886476619656e-05, "loss": 0.3068, "step": 1011 }, { "epoch": 0.22, "learning_rate": 1.8194860440355842e-05, "loss": 0.3163, "step": 1012 }, { "epoch": 0.22, "learning_rate": 1.8190852109767492e-05, "loss": 0.2755, "step": 1013 }, { "epoch": 0.22, "learning_rate": 1.818683977639034e-05, "loss": 0.3011, "step": 1014 }, { "epoch": 0.22, "learning_rate": 1.8182823442185172e-05, "loss": 0.3056, "step": 1015 }, { "epoch": 0.22, "learning_rate": 1.8178803109114738e-05, "loss": 0.3185, "step": 1016 }, { "epoch": 0.22, "learning_rate": 1.8174778779143728e-05, "loss": 0.3142, "step": 1017 }, { "epoch": 0.22, "learning_rate": 1.8170750454238794e-05, "loss": 0.293, "step": 1018 }, { "epoch": 0.22, "learning_rate": 1.8166718136368543e-05, "loss": 0.258, "step": 1019 }, { "epoch": 0.22, "learning_rate": 1.8162681827503525e-05, "loss": 0.3335, "step": 1020 }, { "epoch": 0.22, "learning_rate": 1.815864152961624e-05, "loss": 0.3547, "step": 1021 }, { "epoch": 0.22, "learning_rate": 1.8154597244681153e-05, "loss": 0.3204, "step": 1022 }, { "epoch": 0.22, "learning_rate": 1.8150548974674653e-05, "loss": 0.301, "step": 1023 }, { "epoch": 0.22, "learning_rate": 1.8146496721575098e-05, "loss": 0.3273, "step": 1024 }, { "epoch": 0.22, "learning_rate": 1.814244048736279e-05, "loss": 0.3143, "step": 1025 }, { "epoch": 0.22, "learning_rate": 1.8138380274019957e-05, "loss": 0.348, "step": 1026 }, { "epoch": 0.22, "learning_rate": 1.8134316083530795e-05, "loss": 0.2811, "step": 1027 }, { "epoch": 0.22, "learning_rate": 1.8130247917881432e-05, "loss": 0.3545, "step": 1028 }, { "epoch": 0.22, "learning_rate": 1.8126175779059933e-05, "loss": 0.381, "step": 1029 }, { "epoch": 0.22, "learning_rate": 1.8122099669056324e-05, "loss": 0.2708, "step": 1030 }, { "epoch": 0.22, "learning_rate": 1.8118019589862554e-05, "loss": 0.2938, "step": 1031 }, { "epoch": 0.22, "learning_rate": 1.8113935543472516e-05, "loss": 0.3136, "step": 1032 }, { "epoch": 0.22, "learning_rate": 1.8109847531882046e-05, "loss": 0.3089, "step": 1033 }, { "epoch": 0.22, "learning_rate": 1.8105755557088916e-05, "loss": 0.2806, "step": 1034 }, { "epoch": 0.22, "learning_rate": 1.8101659621092832e-05, "loss": 0.3459, "step": 1035 }, { "epoch": 0.22, "learning_rate": 1.8097559725895434e-05, "loss": 0.2916, "step": 1036 }, { "epoch": 0.22, "learning_rate": 1.8093455873500307e-05, "loss": 0.3077, "step": 1037 }, { "epoch": 0.22, "learning_rate": 1.808934806591296e-05, "loss": 0.3421, "step": 1038 }, { "epoch": 0.22, "learning_rate": 1.8085236305140837e-05, "loss": 0.2577, "step": 1039 }, { "epoch": 0.22, "learning_rate": 1.8081120593193318e-05, "loss": 0.3149, "step": 1040 }, { "epoch": 0.22, "learning_rate": 1.8077000932081706e-05, "loss": 0.3705, "step": 1041 }, { "epoch": 0.22, "learning_rate": 1.8072877323819247e-05, "loss": 0.2941, "step": 1042 }, { "epoch": 0.23, "learning_rate": 1.80687497704211e-05, "loss": 0.3403, "step": 1043 }, { "epoch": 0.23, "learning_rate": 1.8064618273904366e-05, "loss": 0.3308, "step": 1044 }, { "epoch": 0.23, "learning_rate": 1.8060482836288055e-05, "loss": 0.3126, "step": 1045 }, { "epoch": 0.23, "learning_rate": 1.8056343459593134e-05, "loss": 0.2976, "step": 1046 }, { "epoch": 0.23, "learning_rate": 1.8052200145842458e-05, "loss": 0.3052, "step": 1047 }, { "epoch": 0.23, "learning_rate": 1.8048052897060827e-05, "loss": 0.3193, "step": 1048 }, { "epoch": 0.23, "learning_rate": 1.804390171527497e-05, "loss": 0.3111, "step": 1049 }, { "epoch": 0.23, "learning_rate": 1.803974660251352e-05, "loss": 0.3088, "step": 1050 }, { "epoch": 0.23, "learning_rate": 1.8035587560807045e-05, "loss": 0.3249, "step": 1051 }, { "epoch": 0.23, "learning_rate": 1.8031424592188025e-05, "loss": 0.2448, "step": 1052 }, { "epoch": 0.23, "learning_rate": 1.802725769869086e-05, "loss": 0.2743, "step": 1053 }, { "epoch": 0.23, "learning_rate": 1.8023086882351872e-05, "loss": 0.3025, "step": 1054 }, { "epoch": 0.23, "learning_rate": 1.80189121452093e-05, "loss": 0.3936, "step": 1055 }, { "epoch": 0.23, "learning_rate": 1.8014733489303292e-05, "loss": 0.3528, "step": 1056 }, { "epoch": 0.23, "learning_rate": 1.8010550916675918e-05, "loss": 0.2827, "step": 1057 }, { "epoch": 0.23, "learning_rate": 1.8006364429371164e-05, "loss": 0.2774, "step": 1058 }, { "epoch": 0.23, "learning_rate": 1.800217402943492e-05, "loss": 0.3463, "step": 1059 }, { "epoch": 0.23, "learning_rate": 1.7997979718914994e-05, "loss": 0.3275, "step": 1060 }, { "epoch": 0.23, "learning_rate": 1.7993781499861103e-05, "loss": 0.3386, "step": 1061 }, { "epoch": 0.23, "learning_rate": 1.7989579374324883e-05, "loss": 0.2742, "step": 1062 }, { "epoch": 0.23, "learning_rate": 1.798537334435986e-05, "loss": 0.3535, "step": 1063 }, { "epoch": 0.23, "learning_rate": 1.7981163412021493e-05, "loss": 0.3361, "step": 1064 }, { "epoch": 0.23, "learning_rate": 1.7976949579367122e-05, "loss": 0.35, "step": 1065 }, { "epoch": 0.23, "learning_rate": 1.797273184845601e-05, "loss": 0.2563, "step": 1066 }, { "epoch": 0.23, "learning_rate": 1.7968510221349323e-05, "loss": 0.2942, "step": 1067 }, { "epoch": 0.23, "learning_rate": 1.7964284700110125e-05, "loss": 0.2954, "step": 1068 }, { "epoch": 0.23, "learning_rate": 1.796005528680339e-05, "loss": 0.3787, "step": 1069 }, { "epoch": 0.23, "learning_rate": 1.7955821983495985e-05, "loss": 0.2731, "step": 1070 }, { "epoch": 0.23, "learning_rate": 1.7951584792256694e-05, "loss": 0.294, "step": 1071 }, { "epoch": 0.23, "learning_rate": 1.794734371515618e-05, "loss": 0.3001, "step": 1072 }, { "epoch": 0.23, "learning_rate": 1.7943098754267023e-05, "loss": 0.3459, "step": 1073 }, { "epoch": 0.23, "learning_rate": 1.7938849911663685e-05, "loss": 0.2962, "step": 1074 }, { "epoch": 0.23, "learning_rate": 1.793459718942254e-05, "loss": 0.3306, "step": 1075 }, { "epoch": 0.23, "learning_rate": 1.7930340589621857e-05, "loss": 0.329, "step": 1076 }, { "epoch": 0.23, "learning_rate": 1.792608011434178e-05, "loss": 0.3683, "step": 1077 }, { "epoch": 0.23, "learning_rate": 1.7921815765664373e-05, "loss": 0.3273, "step": 1078 }, { "epoch": 0.23, "learning_rate": 1.7917547545673578e-05, "loss": 0.266, "step": 1079 }, { "epoch": 0.23, "learning_rate": 1.7913275456455232e-05, "loss": 0.3344, "step": 1080 }, { "epoch": 0.23, "learning_rate": 1.7908999500097063e-05, "loss": 0.3973, "step": 1081 }, { "epoch": 0.23, "learning_rate": 1.7904719678688683e-05, "loss": 0.3459, "step": 1082 }, { "epoch": 0.23, "learning_rate": 1.7900435994321607e-05, "loss": 0.2786, "step": 1083 }, { "epoch": 0.23, "learning_rate": 1.789614844908923e-05, "loss": 0.2669, "step": 1084 }, { "epoch": 0.23, "learning_rate": 1.7891857045086825e-05, "loss": 0.3111, "step": 1085 }, { "epoch": 0.23, "learning_rate": 1.7887561784411566e-05, "loss": 0.274, "step": 1086 }, { "epoch": 0.23, "learning_rate": 1.78832626691625e-05, "loss": 0.3013, "step": 1087 }, { "epoch": 0.23, "learning_rate": 1.7878959701440564e-05, "loss": 0.2834, "step": 1088 }, { "epoch": 0.24, "learning_rate": 1.787465288334858e-05, "loss": 0.2647, "step": 1089 }, { "epoch": 0.24, "learning_rate": 1.787034221699124e-05, "loss": 0.307, "step": 1090 }, { "epoch": 0.24, "learning_rate": 1.786602770447513e-05, "loss": 0.237, "step": 1091 }, { "epoch": 0.24, "learning_rate": 1.7861709347908703e-05, "loss": 0.2824, "step": 1092 }, { "epoch": 0.24, "learning_rate": 1.785738714940231e-05, "loss": 0.3816, "step": 1093 }, { "epoch": 0.24, "learning_rate": 1.7853061111068154e-05, "loss": 0.3094, "step": 1094 }, { "epoch": 0.24, "learning_rate": 1.7848731235020332e-05, "loss": 0.3416, "step": 1095 }, { "epoch": 0.24, "learning_rate": 1.7844397523374817e-05, "loss": 0.3402, "step": 1096 }, { "epoch": 0.24, "learning_rate": 1.7840059978249444e-05, "loss": 0.2943, "step": 1097 }, { "epoch": 0.24, "learning_rate": 1.783571860176393e-05, "loss": 0.2887, "step": 1098 }, { "epoch": 0.24, "learning_rate": 1.7831373396039863e-05, "loss": 0.2789, "step": 1099 }, { "epoch": 0.24, "learning_rate": 1.7827024363200707e-05, "loss": 0.3738, "step": 1100 }, { "epoch": 0.24, "learning_rate": 1.7822671505371785e-05, "loss": 0.3296, "step": 1101 }, { "epoch": 0.24, "learning_rate": 1.78183148246803e-05, "loss": 0.2925, "step": 1102 }, { "epoch": 0.24, "learning_rate": 1.7813954323255318e-05, "loss": 0.3376, "step": 1103 }, { "epoch": 0.24, "learning_rate": 1.780959000322777e-05, "loss": 0.3022, "step": 1104 }, { "epoch": 0.24, "learning_rate": 1.780522186673046e-05, "loss": 0.3647, "step": 1105 }, { "epoch": 0.24, "learning_rate": 1.7800849915898055e-05, "loss": 0.323, "step": 1106 }, { "epoch": 0.24, "learning_rate": 1.779647415286708e-05, "loss": 0.3253, "step": 1107 }, { "epoch": 0.24, "learning_rate": 1.7792094579775928e-05, "loss": 0.3111, "step": 1108 }, { "epoch": 0.24, "learning_rate": 1.7787711198764852e-05, "loss": 0.3471, "step": 1109 }, { "epoch": 0.24, "learning_rate": 1.7783324011975974e-05, "loss": 0.2378, "step": 1110 }, { "epoch": 0.24, "learning_rate": 1.7778933021553262e-05, "loss": 0.3193, "step": 1111 }, { "epoch": 0.24, "learning_rate": 1.777453822964255e-05, "loss": 0.3878, "step": 1112 }, { "epoch": 0.24, "learning_rate": 1.7770139638391535e-05, "loss": 0.329, "step": 1113 }, { "epoch": 0.24, "learning_rate": 1.7765737249949763e-05, "loss": 0.3268, "step": 1114 }, { "epoch": 0.24, "learning_rate": 1.776133106646863e-05, "loss": 0.34, "step": 1115 }, { "epoch": 0.24, "learning_rate": 1.775692109010141e-05, "loss": 0.316, "step": 1116 }, { "epoch": 0.24, "learning_rate": 1.7752507323003203e-05, "loss": 0.3435, "step": 1117 }, { "epoch": 0.24, "learning_rate": 1.7748089767330972e-05, "loss": 0.2777, "step": 1118 }, { "epoch": 0.24, "learning_rate": 1.7743668425243547e-05, "loss": 0.3166, "step": 1119 }, { "epoch": 0.24, "learning_rate": 1.7739243298901578e-05, "loss": 0.274, "step": 1120 }, { "epoch": 0.24, "learning_rate": 1.773481439046759e-05, "loss": 0.2959, "step": 1121 }, { "epoch": 0.24, "learning_rate": 1.7730381702105946e-05, "loss": 0.3103, "step": 1122 }, { "epoch": 0.24, "learning_rate": 1.7725945235982856e-05, "loss": 0.2776, "step": 1123 }, { "epoch": 0.24, "learning_rate": 1.7721504994266376e-05, "loss": 0.3369, "step": 1124 }, { "epoch": 0.24, "learning_rate": 1.7717060979126408e-05, "loss": 0.3248, "step": 1125 }, { "epoch": 0.24, "learning_rate": 1.7712613192734703e-05, "loss": 0.331, "step": 1126 }, { "epoch": 0.24, "learning_rate": 1.770816163726485e-05, "loss": 0.3274, "step": 1127 }, { "epoch": 0.24, "learning_rate": 1.7703706314892275e-05, "loss": 0.2761, "step": 1128 }, { "epoch": 0.24, "learning_rate": 1.769924722779425e-05, "loss": 0.2856, "step": 1129 }, { "epoch": 0.24, "learning_rate": 1.7694784378149895e-05, "loss": 0.2976, "step": 1130 }, { "epoch": 0.24, "learning_rate": 1.7690317768140155e-05, "loss": 0.2624, "step": 1131 }, { "epoch": 0.24, "learning_rate": 1.768584739994782e-05, "loss": 0.2716, "step": 1132 }, { "epoch": 0.24, "learning_rate": 1.768137327575751e-05, "loss": 0.2966, "step": 1133 }, { "epoch": 0.24, "learning_rate": 1.7676895397755694e-05, "loss": 0.2999, "step": 1134 }, { "epoch": 0.24, "learning_rate": 1.7672413768130662e-05, "loss": 0.3178, "step": 1135 }, { "epoch": 0.25, "learning_rate": 1.766792838907254e-05, "loss": 0.3235, "step": 1136 }, { "epoch": 0.25, "learning_rate": 1.7663439262773294e-05, "loss": 0.3527, "step": 1137 }, { "epoch": 0.25, "learning_rate": 1.7658946391426712e-05, "loss": 0.3321, "step": 1138 }, { "epoch": 0.25, "learning_rate": 1.765444977722842e-05, "loss": 0.2858, "step": 1139 }, { "epoch": 0.25, "learning_rate": 1.764994942237586e-05, "loss": 0.2846, "step": 1140 }, { "epoch": 0.25, "learning_rate": 1.764544532906832e-05, "loss": 0.3076, "step": 1141 }, { "epoch": 0.25, "learning_rate": 1.7640937499506907e-05, "loss": 0.3454, "step": 1142 }, { "epoch": 0.25, "learning_rate": 1.7636425935894545e-05, "loss": 0.3021, "step": 1143 }, { "epoch": 0.25, "learning_rate": 1.7631910640435998e-05, "loss": 0.3386, "step": 1144 }, { "epoch": 0.25, "learning_rate": 1.7627391615337844e-05, "loss": 0.2373, "step": 1145 }, { "epoch": 0.25, "learning_rate": 1.7622868862808485e-05, "loss": 0.2829, "step": 1146 }, { "epoch": 0.25, "learning_rate": 1.7618342385058147e-05, "loss": 0.3652, "step": 1147 }, { "epoch": 0.25, "learning_rate": 1.7613812184298872e-05, "loss": 0.2649, "step": 1148 }, { "epoch": 0.25, "learning_rate": 1.7609278262744525e-05, "loss": 0.2817, "step": 1149 }, { "epoch": 0.25, "learning_rate": 1.7604740622610796e-05, "loss": 0.2582, "step": 1150 }, { "epoch": 0.25, "learning_rate": 1.7600199266115182e-05, "loss": 0.3746, "step": 1151 }, { "epoch": 0.25, "learning_rate": 1.759565419547699e-05, "loss": 0.3046, "step": 1152 }, { "epoch": 0.25, "learning_rate": 1.7591105412917363e-05, "loss": 0.302, "step": 1153 }, { "epoch": 0.25, "learning_rate": 1.758655292065924e-05, "loss": 0.2626, "step": 1154 }, { "epoch": 0.25, "learning_rate": 1.7581996720927378e-05, "loss": 0.306, "step": 1155 }, { "epoch": 0.25, "learning_rate": 1.7577436815948353e-05, "loss": 0.3377, "step": 1156 }, { "epoch": 0.25, "learning_rate": 1.7572873207950544e-05, "loss": 0.3561, "step": 1157 }, { "epoch": 0.25, "learning_rate": 1.756830589916414e-05, "loss": 0.3068, "step": 1158 }, { "epoch": 0.25, "learning_rate": 1.7563734891821135e-05, "loss": 0.3054, "step": 1159 }, { "epoch": 0.25, "learning_rate": 1.7559160188155342e-05, "loss": 0.2642, "step": 1160 }, { "epoch": 0.25, "learning_rate": 1.7554581790402372e-05, "loss": 0.2799, "step": 1161 }, { "epoch": 0.25, "learning_rate": 1.754999970079964e-05, "loss": 0.3331, "step": 1162 }, { "epoch": 0.25, "learning_rate": 1.7545413921586374e-05, "loss": 0.2876, "step": 1163 }, { "epoch": 0.25, "learning_rate": 1.7540824455003594e-05, "loss": 0.3022, "step": 1164 }, { "epoch": 0.25, "learning_rate": 1.7536231303294128e-05, "loss": 0.3271, "step": 1165 }, { "epoch": 0.25, "learning_rate": 1.7531634468702608e-05, "loss": 0.3104, "step": 1166 }, { "epoch": 0.25, "learning_rate": 1.7527033953475455e-05, "loss": 0.3454, "step": 1167 }, { "epoch": 0.25, "learning_rate": 1.7522429759860903e-05, "loss": 0.293, "step": 1168 }, { "epoch": 0.25, "learning_rate": 1.751782189010897e-05, "loss": 0.3274, "step": 1169 }, { "epoch": 0.25, "learning_rate": 1.751321034647148e-05, "loss": 0.3075, "step": 1170 }, { "epoch": 0.25, "learning_rate": 1.7508595131202046e-05, "loss": 0.3324, "step": 1171 }, { "epoch": 0.25, "learning_rate": 1.7503976246556083e-05, "loss": 0.3439, "step": 1172 }, { "epoch": 0.25, "learning_rate": 1.7499353694790788e-05, "loss": 0.3839, "step": 1173 }, { "epoch": 0.25, "learning_rate": 1.7494727478165163e-05, "loss": 0.3336, "step": 1174 }, { "epoch": 0.25, "learning_rate": 1.749009759893999e-05, "loss": 0.2593, "step": 1175 }, { "epoch": 0.25, "learning_rate": 1.748546405937785e-05, "loss": 0.3548, "step": 1176 }, { "epoch": 0.25, "learning_rate": 1.7480826861743102e-05, "loss": 0.3011, "step": 1177 }, { "epoch": 0.25, "learning_rate": 1.7476186008301904e-05, "loss": 0.3159, "step": 1178 }, { "epoch": 0.25, "learning_rate": 1.7471541501322192e-05, "loss": 0.3731, "step": 1179 }, { "epoch": 0.25, "learning_rate": 1.7466893343073688e-05, "loss": 0.3008, "step": 1180 }, { "epoch": 0.25, "learning_rate": 1.7462241535827912e-05, "loss": 0.2935, "step": 1181 }, { "epoch": 0.26, "learning_rate": 1.745758608185815e-05, "loss": 0.2829, "step": 1182 }, { "epoch": 0.26, "learning_rate": 1.7452926983439477e-05, "loss": 0.2889, "step": 1183 }, { "epoch": 0.26, "learning_rate": 1.7448264242848746e-05, "loss": 0.2686, "step": 1184 }, { "epoch": 0.26, "learning_rate": 1.744359786236459e-05, "loss": 0.2845, "step": 1185 }, { "epoch": 0.26, "learning_rate": 1.7438927844267436e-05, "loss": 0.3236, "step": 1186 }, { "epoch": 0.26, "learning_rate": 1.743425419083946e-05, "loss": 0.3642, "step": 1187 }, { "epoch": 0.26, "learning_rate": 1.742957690436464e-05, "loss": 0.2935, "step": 1188 }, { "epoch": 0.26, "learning_rate": 1.7424895987128723e-05, "loss": 0.3546, "step": 1189 }, { "epoch": 0.26, "learning_rate": 1.7420211441419216e-05, "loss": 0.3101, "step": 1190 }, { "epoch": 0.26, "learning_rate": 1.7415523269525422e-05, "loss": 0.3685, "step": 1191 }, { "epoch": 0.26, "learning_rate": 1.7410831473738393e-05, "loss": 0.3307, "step": 1192 }, { "epoch": 0.26, "learning_rate": 1.7406136056350972e-05, "loss": 0.298, "step": 1193 }, { "epoch": 0.26, "learning_rate": 1.740143701965776e-05, "loss": 0.3035, "step": 1194 }, { "epoch": 0.26, "learning_rate": 1.739673436595513e-05, "loss": 0.3332, "step": 1195 }, { "epoch": 0.26, "learning_rate": 1.7392028097541223e-05, "loss": 0.3282, "step": 1196 }, { "epoch": 0.26, "learning_rate": 1.7387318216715946e-05, "loss": 0.289, "step": 1197 }, { "epoch": 0.26, "learning_rate": 1.7382604725780973e-05, "loss": 0.3269, "step": 1198 }, { "epoch": 0.26, "learning_rate": 1.737788762703974e-05, "loss": 0.2915, "step": 1199 }, { "epoch": 0.26, "learning_rate": 1.737316692279744e-05, "loss": 0.2723, "step": 1200 }, { "epoch": 0.26, "learning_rate": 1.7368442615361047e-05, "loss": 0.3333, "step": 1201 }, { "epoch": 0.26, "learning_rate": 1.7363714707039275e-05, "loss": 0.3025, "step": 1202 }, { "epoch": 0.26, "learning_rate": 1.7358983200142608e-05, "loss": 0.2495, "step": 1203 }, { "epoch": 0.26, "learning_rate": 1.7354248096983286e-05, "loss": 0.3078, "step": 1204 }, { "epoch": 0.26, "learning_rate": 1.734950939987531e-05, "loss": 0.3118, "step": 1205 }, { "epoch": 0.26, "learning_rate": 1.7344767111134433e-05, "loss": 0.2708, "step": 1206 }, { "epoch": 0.26, "learning_rate": 1.7340021233078165e-05, "loss": 0.2996, "step": 1207 }, { "epoch": 0.26, "learning_rate": 1.7335271768025773e-05, "loss": 0.3794, "step": 1208 }, { "epoch": 0.26, "learning_rate": 1.7330518718298263e-05, "loss": 0.3412, "step": 1209 }, { "epoch": 0.26, "learning_rate": 1.7325762086218415e-05, "loss": 0.3298, "step": 1210 }, { "epoch": 0.26, "learning_rate": 1.732100187411075e-05, "loss": 0.2822, "step": 1211 }, { "epoch": 0.26, "learning_rate": 1.7316238084301526e-05, "loss": 0.3357, "step": 1212 }, { "epoch": 0.26, "learning_rate": 1.7311470719118765e-05, "loss": 0.2765, "step": 1213 }, { "epoch": 0.26, "learning_rate": 1.7306699780892233e-05, "loss": 0.3411, "step": 1214 }, { "epoch": 0.26, "learning_rate": 1.730192527195344e-05, "loss": 0.2958, "step": 1215 }, { "epoch": 0.26, "learning_rate": 1.7297147194635638e-05, "loss": 0.3222, "step": 1216 }, { "epoch": 0.26, "learning_rate": 1.7292365551273835e-05, "loss": 0.328, "step": 1217 }, { "epoch": 0.26, "learning_rate": 1.7287580344204763e-05, "loss": 0.2951, "step": 1218 }, { "epoch": 0.26, "learning_rate": 1.728279157576691e-05, "loss": 0.3277, "step": 1219 }, { "epoch": 0.26, "learning_rate": 1.7277999248300502e-05, "loss": 0.2746, "step": 1220 }, { "epoch": 0.26, "learning_rate": 1.72732033641475e-05, "loss": 0.3553, "step": 1221 }, { "epoch": 0.26, "learning_rate": 1.726840392565161e-05, "loss": 0.2983, "step": 1222 }, { "epoch": 0.26, "learning_rate": 1.726360093515826e-05, "loss": 0.3627, "step": 1223 }, { "epoch": 0.26, "learning_rate": 1.725879439501463e-05, "loss": 0.2964, "step": 1224 }, { "epoch": 0.26, "learning_rate": 1.7253984307569634e-05, "loss": 0.2959, "step": 1225 }, { "epoch": 0.26, "learning_rate": 1.7249170675173905e-05, "loss": 0.3, "step": 1226 }, { "epoch": 0.26, "learning_rate": 1.724435350017982e-05, "loss": 0.3693, "step": 1227 }, { "epoch": 0.27, "learning_rate": 1.7239532784941496e-05, "loss": 0.3033, "step": 1228 }, { "epoch": 0.27, "learning_rate": 1.7234708531814753e-05, "loss": 0.2939, "step": 1229 }, { "epoch": 0.27, "learning_rate": 1.7229880743157165e-05, "loss": 0.2872, "step": 1230 }, { "epoch": 0.27, "learning_rate": 1.7225049421328024e-05, "loss": 0.2289, "step": 1231 }, { "epoch": 0.27, "learning_rate": 1.7220214568688346e-05, "loss": 0.3191, "step": 1232 }, { "epoch": 0.27, "learning_rate": 1.7215376187600883e-05, "loss": 0.3276, "step": 1233 }, { "epoch": 0.27, "learning_rate": 1.72105342804301e-05, "loss": 0.3217, "step": 1234 }, { "epoch": 0.27, "learning_rate": 1.7205688849542185e-05, "loss": 0.325, "step": 1235 }, { "epoch": 0.27, "learning_rate": 1.7200839897305054e-05, "loss": 0.3397, "step": 1236 }, { "epoch": 0.27, "learning_rate": 1.719598742608835e-05, "loss": 0.3355, "step": 1237 }, { "epoch": 0.27, "learning_rate": 1.7191131438263423e-05, "loss": 0.2879, "step": 1238 }, { "epoch": 0.27, "learning_rate": 1.7186271936203344e-05, "loss": 0.253, "step": 1239 }, { "epoch": 0.27, "learning_rate": 1.7181408922282903e-05, "loss": 0.2744, "step": 1240 }, { "epoch": 0.27, "learning_rate": 1.717654239887861e-05, "loss": 0.3292, "step": 1241 }, { "epoch": 0.27, "learning_rate": 1.7171672368368693e-05, "loss": 0.325, "step": 1242 }, { "epoch": 0.27, "learning_rate": 1.716679883313308e-05, "loss": 0.2821, "step": 1243 }, { "epoch": 0.27, "learning_rate": 1.716192179555342e-05, "loss": 0.3603, "step": 1244 }, { "epoch": 0.27, "learning_rate": 1.7157041258013074e-05, "loss": 0.2772, "step": 1245 }, { "epoch": 0.27, "learning_rate": 1.7152157222897114e-05, "loss": 0.3385, "step": 1246 }, { "epoch": 0.27, "learning_rate": 1.714726969259232e-05, "loss": 0.3249, "step": 1247 }, { "epoch": 0.27, "learning_rate": 1.714237866948718e-05, "loss": 0.2833, "step": 1248 }, { "epoch": 0.27, "learning_rate": 1.713748415597189e-05, "loss": 0.3021, "step": 1249 }, { "epoch": 0.27, "learning_rate": 1.7132586154438346e-05, "loss": 0.3281, "step": 1250 }, { "epoch": 0.27, "learning_rate": 1.7127684667280157e-05, "loss": 0.3056, "step": 1251 }, { "epoch": 0.27, "learning_rate": 1.7122779696892628e-05, "loss": 0.3292, "step": 1252 }, { "epoch": 0.27, "learning_rate": 1.7117871245672774e-05, "loss": 0.309, "step": 1253 }, { "epoch": 0.27, "learning_rate": 1.7112959316019307e-05, "loss": 0.2835, "step": 1254 }, { "epoch": 0.27, "learning_rate": 1.7108043910332635e-05, "loss": 0.3044, "step": 1255 }, { "epoch": 0.27, "learning_rate": 1.7103125031014868e-05, "loss": 0.2876, "step": 1256 }, { "epoch": 0.27, "learning_rate": 1.7098202680469817e-05, "loss": 0.3338, "step": 1257 }, { "epoch": 0.27, "learning_rate": 1.709327686110299e-05, "loss": 0.2567, "step": 1258 }, { "epoch": 0.27, "learning_rate": 1.7088347575321575e-05, "loss": 0.273, "step": 1259 }, { "epoch": 0.27, "learning_rate": 1.7083414825534473e-05, "loss": 0.2893, "step": 1260 }, { "epoch": 0.27, "learning_rate": 1.7078478614152273e-05, "loss": 0.3323, "step": 1261 }, { "epoch": 0.27, "learning_rate": 1.707353894358725e-05, "loss": 0.2734, "step": 1262 }, { "epoch": 0.27, "learning_rate": 1.7068595816253367e-05, "loss": 0.3574, "step": 1263 }, { "epoch": 0.27, "learning_rate": 1.706364923456629e-05, "loss": 0.3056, "step": 1264 }, { "epoch": 0.27, "learning_rate": 1.705869920094336e-05, "loss": 0.3276, "step": 1265 }, { "epoch": 0.27, "learning_rate": 1.7053745717803613e-05, "loss": 0.3085, "step": 1266 }, { "epoch": 0.27, "learning_rate": 1.704878878756776e-05, "loss": 0.3211, "step": 1267 }, { "epoch": 0.27, "learning_rate": 1.7043828412658214e-05, "loss": 0.2342, "step": 1268 }, { "epoch": 0.27, "learning_rate": 1.7038864595499055e-05, "loss": 0.3392, "step": 1269 }, { "epoch": 0.27, "learning_rate": 1.703389733851605e-05, "loss": 0.3309, "step": 1270 }, { "epoch": 0.27, "learning_rate": 1.7028926644136657e-05, "loss": 0.3041, "step": 1271 }, { "epoch": 0.27, "learning_rate": 1.702395251479e-05, "loss": 0.3342, "step": 1272 }, { "epoch": 0.27, "learning_rate": 1.7018974952906885e-05, "loss": 0.3528, "step": 1273 }, { "epoch": 0.27, "learning_rate": 1.7013993960919797e-05, "loss": 0.3432, "step": 1274 }, { "epoch": 0.28, "learning_rate": 1.7009009541262907e-05, "loss": 0.3518, "step": 1275 }, { "epoch": 0.28, "learning_rate": 1.7004021696372043e-05, "loss": 0.2694, "step": 1276 }, { "epoch": 0.28, "learning_rate": 1.6999030428684716e-05, "loss": 0.2708, "step": 1277 }, { "epoch": 0.28, "learning_rate": 1.6994035740640113e-05, "loss": 0.344, "step": 1278 }, { "epoch": 0.28, "learning_rate": 1.6989037634679088e-05, "loss": 0.3538, "step": 1279 }, { "epoch": 0.28, "learning_rate": 1.6984036113244165e-05, "loss": 0.281, "step": 1280 }, { "epoch": 0.28, "learning_rate": 1.697903117877954e-05, "loss": 0.3204, "step": 1281 }, { "epoch": 0.28, "learning_rate": 1.697402283373107e-05, "loss": 0.3017, "step": 1282 }, { "epoch": 0.28, "learning_rate": 1.6969011080546294e-05, "loss": 0.3364, "step": 1283 }, { "epoch": 0.28, "learning_rate": 1.6963995921674397e-05, "loss": 0.3619, "step": 1284 }, { "epoch": 0.28, "learning_rate": 1.6958977359566244e-05, "loss": 0.2851, "step": 1285 }, { "epoch": 0.28, "learning_rate": 1.695395539667435e-05, "loss": 0.2638, "step": 1286 }, { "epoch": 0.28, "learning_rate": 1.6948930035452905e-05, "loss": 0.2981, "step": 1287 }, { "epoch": 0.28, "learning_rate": 1.694390127835775e-05, "loss": 0.2914, "step": 1288 }, { "epoch": 0.28, "learning_rate": 1.6938869127846393e-05, "loss": 0.3419, "step": 1289 }, { "epoch": 0.28, "learning_rate": 1.6933833586377993e-05, "loss": 0.3726, "step": 1290 }, { "epoch": 0.28, "learning_rate": 1.692879465641337e-05, "loss": 0.2976, "step": 1291 }, { "epoch": 0.28, "learning_rate": 1.6923752340415e-05, "loss": 0.2886, "step": 1292 }, { "epoch": 0.28, "learning_rate": 1.6918706640847012e-05, "loss": 0.3271, "step": 1293 }, { "epoch": 0.28, "learning_rate": 1.6913657560175197e-05, "loss": 0.2982, "step": 1294 }, { "epoch": 0.28, "learning_rate": 1.6908605100866977e-05, "loss": 0.2623, "step": 1295 }, { "epoch": 0.28, "learning_rate": 1.690354926539145e-05, "loss": 0.3047, "step": 1296 }, { "epoch": 0.28, "learning_rate": 1.6898490056219356e-05, "loss": 0.3133, "step": 1297 }, { "epoch": 0.28, "learning_rate": 1.689342747582307e-05, "loss": 0.3117, "step": 1298 }, { "epoch": 0.28, "learning_rate": 1.688836152667663e-05, "loss": 0.2848, "step": 1299 }, { "epoch": 0.28, "learning_rate": 1.6883292211255712e-05, "loss": 0.2781, "step": 1300 }, { "epoch": 0.28, "learning_rate": 1.687821953203765e-05, "loss": 0.32, "step": 1301 }, { "epoch": 0.28, "learning_rate": 1.6873143491501402e-05, "loss": 0.2936, "step": 1302 }, { "epoch": 0.28, "learning_rate": 1.6868064092127586e-05, "loss": 0.2949, "step": 1303 }, { "epoch": 0.28, "learning_rate": 1.6862981336398453e-05, "loss": 0.3006, "step": 1304 }, { "epoch": 0.28, "learning_rate": 1.6857895226797896e-05, "loss": 0.3162, "step": 1305 }, { "epoch": 0.28, "learning_rate": 1.6852805765811442e-05, "loss": 0.3013, "step": 1306 }, { "epoch": 0.28, "learning_rate": 1.6847712955926264e-05, "loss": 0.3581, "step": 1307 }, { "epoch": 0.28, "learning_rate": 1.684261679963117e-05, "loss": 0.3019, "step": 1308 }, { "epoch": 0.28, "learning_rate": 1.6837517299416598e-05, "loss": 0.3596, "step": 1309 }, { "epoch": 0.28, "learning_rate": 1.6832414457774626e-05, "loss": 0.3438, "step": 1310 }, { "epoch": 0.28, "learning_rate": 1.682730827719896e-05, "loss": 0.3219, "step": 1311 }, { "epoch": 0.28, "learning_rate": 1.6822198760184942e-05, "loss": 0.2738, "step": 1312 }, { "epoch": 0.28, "learning_rate": 1.6817085909229544e-05, "loss": 0.3308, "step": 1313 }, { "epoch": 0.28, "learning_rate": 1.6811969726831362e-05, "loss": 0.2746, "step": 1314 }, { "epoch": 0.28, "learning_rate": 1.680685021549063e-05, "loss": 0.2981, "step": 1315 }, { "epoch": 0.28, "learning_rate": 1.6801727377709195e-05, "loss": 0.3381, "step": 1316 }, { "epoch": 0.28, "learning_rate": 1.6796601215990542e-05, "loss": 0.3653, "step": 1317 }, { "epoch": 0.28, "learning_rate": 1.6791471732839778e-05, "loss": 0.3365, "step": 1318 }, { "epoch": 0.28, "learning_rate": 1.6786338930763624e-05, "loss": 0.3116, "step": 1319 }, { "epoch": 0.28, "learning_rate": 1.6781202812270437e-05, "loss": 0.2755, "step": 1320 }, { "epoch": 0.29, "learning_rate": 1.6776063379870183e-05, "loss": 0.3158, "step": 1321 }, { "epoch": 0.29, "learning_rate": 1.6770920636074456e-05, "loss": 0.2775, "step": 1322 }, { "epoch": 0.29, "learning_rate": 1.676577458339646e-05, "loss": 0.3194, "step": 1323 }, { "epoch": 0.29, "learning_rate": 1.676062522435102e-05, "loss": 0.3376, "step": 1324 }, { "epoch": 0.29, "learning_rate": 1.6755472561454582e-05, "loss": 0.3323, "step": 1325 }, { "epoch": 0.29, "learning_rate": 1.67503165972252e-05, "loss": 0.3002, "step": 1326 }, { "epoch": 0.29, "learning_rate": 1.6745157334182537e-05, "loss": 0.2971, "step": 1327 }, { "epoch": 0.29, "learning_rate": 1.6739994774847884e-05, "loss": 0.3863, "step": 1328 }, { "epoch": 0.29, "learning_rate": 1.6734828921744127e-05, "loss": 0.302, "step": 1329 }, { "epoch": 0.29, "learning_rate": 1.6729659777395764e-05, "loss": 0.2686, "step": 1330 }, { "epoch": 0.29, "learning_rate": 1.6724487344328912e-05, "loss": 0.2552, "step": 1331 }, { "epoch": 0.29, "learning_rate": 1.6719311625071285e-05, "loss": 0.3229, "step": 1332 }, { "epoch": 0.29, "learning_rate": 1.671413262215221e-05, "loss": 0.342, "step": 1333 }, { "epoch": 0.29, "learning_rate": 1.670895033810261e-05, "loss": 0.2916, "step": 1334 }, { "epoch": 0.29, "learning_rate": 1.670376477545501e-05, "loss": 0.3194, "step": 1335 }, { "epoch": 0.29, "learning_rate": 1.669857593674356e-05, "loss": 0.3246, "step": 1336 }, { "epoch": 0.29, "learning_rate": 1.669338382450398e-05, "loss": 0.3181, "step": 1337 }, { "epoch": 0.29, "learning_rate": 1.6688188441273612e-05, "loss": 0.334, "step": 1338 }, { "epoch": 0.29, "learning_rate": 1.6682989789591387e-05, "loss": 0.2837, "step": 1339 }, { "epoch": 0.29, "learning_rate": 1.667778787199783e-05, "loss": 0.3229, "step": 1340 }, { "epoch": 0.29, "learning_rate": 1.6672582691035077e-05, "loss": 0.2602, "step": 1341 }, { "epoch": 0.29, "learning_rate": 1.6667374249246836e-05, "loss": 0.3238, "step": 1342 }, { "epoch": 0.29, "learning_rate": 1.6662162549178433e-05, "loss": 0.3032, "step": 1343 }, { "epoch": 0.29, "learning_rate": 1.6656947593376766e-05, "loss": 0.2892, "step": 1344 }, { "epoch": 0.29, "learning_rate": 1.6651729384390336e-05, "loss": 0.2899, "step": 1345 }, { "epoch": 0.29, "learning_rate": 1.6646507924769233e-05, "loss": 0.2771, "step": 1346 }, { "epoch": 0.29, "learning_rate": 1.6641283217065125e-05, "loss": 0.2691, "step": 1347 }, { "epoch": 0.29, "learning_rate": 1.6636055263831286e-05, "loss": 0.2978, "step": 1348 }, { "epoch": 0.29, "learning_rate": 1.6630824067622556e-05, "loss": 0.3279, "step": 1349 }, { "epoch": 0.29, "learning_rate": 1.6625589630995377e-05, "loss": 0.3928, "step": 1350 }, { "epoch": 0.29, "learning_rate": 1.6620351956507758e-05, "loss": 0.3071, "step": 1351 }, { "epoch": 0.29, "learning_rate": 1.6615111046719308e-05, "loss": 0.3135, "step": 1352 }, { "epoch": 0.29, "learning_rate": 1.66098669041912e-05, "loss": 0.3757, "step": 1353 }, { "epoch": 0.29, "learning_rate": 1.66046195314862e-05, "loss": 0.3322, "step": 1354 }, { "epoch": 0.29, "learning_rate": 1.6599368931168643e-05, "loss": 0.3015, "step": 1355 }, { "epoch": 0.29, "learning_rate": 1.6594115105804453e-05, "loss": 0.3007, "step": 1356 }, { "epoch": 0.29, "learning_rate": 1.658885805796111e-05, "loss": 0.3041, "step": 1357 }, { "epoch": 0.29, "learning_rate": 1.6583597790207696e-05, "loss": 0.3282, "step": 1358 }, { "epoch": 0.29, "learning_rate": 1.6578334305114844e-05, "loss": 0.3194, "step": 1359 }, { "epoch": 0.29, "learning_rate": 1.657306760525477e-05, "loss": 0.2667, "step": 1360 }, { "epoch": 0.29, "learning_rate": 1.6567797693201257e-05, "loss": 0.2456, "step": 1361 }, { "epoch": 0.29, "learning_rate": 1.6562524571529662e-05, "loss": 0.2734, "step": 1362 }, { "epoch": 0.29, "learning_rate": 1.6557248242816905e-05, "loss": 0.2889, "step": 1363 }, { "epoch": 0.29, "learning_rate": 1.655196870964147e-05, "loss": 0.3035, "step": 1364 }, { "epoch": 0.29, "learning_rate": 1.6546685974583425e-05, "loss": 0.2739, "step": 1365 }, { "epoch": 0.29, "learning_rate": 1.6541400040224386e-05, "loss": 0.3102, "step": 1366 }, { "epoch": 0.3, "learning_rate": 1.653611090914753e-05, "loss": 0.2833, "step": 1367 }, { "epoch": 0.3, "learning_rate": 1.6530818583937612e-05, "loss": 0.2931, "step": 1368 }, { "epoch": 0.3, "learning_rate": 1.6525523067180937e-05, "loss": 0.2983, "step": 1369 }, { "epoch": 0.3, "learning_rate": 1.652022436146537e-05, "loss": 0.285, "step": 1370 }, { "epoch": 0.3, "learning_rate": 1.651492246938034e-05, "loss": 0.284, "step": 1371 }, { "epoch": 0.3, "learning_rate": 1.6509617393516825e-05, "loss": 0.3177, "step": 1372 }, { "epoch": 0.3, "learning_rate": 1.6504309136467364e-05, "loss": 0.2636, "step": 1373 }, { "epoch": 0.3, "learning_rate": 1.6498997700826052e-05, "loss": 0.3288, "step": 1374 }, { "epoch": 0.3, "learning_rate": 1.6493683089188535e-05, "loss": 0.2534, "step": 1375 }, { "epoch": 0.3, "learning_rate": 1.6488365304152015e-05, "loss": 0.3192, "step": 1376 }, { "epoch": 0.3, "learning_rate": 1.6483044348315233e-05, "loss": 0.2955, "step": 1377 }, { "epoch": 0.3, "learning_rate": 1.6477720224278493e-05, "loss": 0.3388, "step": 1378 }, { "epoch": 0.3, "learning_rate": 1.647239293464364e-05, "loss": 0.2416, "step": 1379 }, { "epoch": 0.3, "learning_rate": 1.646706248201407e-05, "loss": 0.3817, "step": 1380 }, { "epoch": 0.3, "learning_rate": 1.646172886899472e-05, "loss": 0.3008, "step": 1381 }, { "epoch": 0.3, "learning_rate": 1.645639209819208e-05, "loss": 0.2994, "step": 1382 }, { "epoch": 0.3, "learning_rate": 1.645105217221417e-05, "loss": 0.3401, "step": 1383 }, { "epoch": 0.3, "learning_rate": 1.6445709093670564e-05, "loss": 0.2682, "step": 1384 }, { "epoch": 0.3, "learning_rate": 1.6440362865172373e-05, "loss": 0.3234, "step": 1385 }, { "epoch": 0.3, "learning_rate": 1.6435013489332238e-05, "loss": 0.2684, "step": 1386 }, { "epoch": 0.3, "learning_rate": 1.6429660968764355e-05, "loss": 0.2974, "step": 1387 }, { "epoch": 0.3, "learning_rate": 1.6424305306084444e-05, "loss": 0.3625, "step": 1388 }, { "epoch": 0.3, "learning_rate": 1.641894650390977e-05, "loss": 0.3359, "step": 1389 }, { "epoch": 0.3, "learning_rate": 1.6413584564859118e-05, "loss": 0.291, "step": 1390 }, { "epoch": 0.3, "learning_rate": 1.6408219491552818e-05, "loss": 0.3424, "step": 1391 }, { "epoch": 0.3, "learning_rate": 1.6402851286612735e-05, "loss": 0.2949, "step": 1392 }, { "epoch": 0.3, "learning_rate": 1.639747995266225e-05, "loss": 0.2886, "step": 1393 }, { "epoch": 0.3, "learning_rate": 1.6392105492326288e-05, "loss": 0.2891, "step": 1394 }, { "epoch": 0.3, "learning_rate": 1.6386727908231285e-05, "loss": 0.3187, "step": 1395 }, { "epoch": 0.3, "learning_rate": 1.638134720300522e-05, "loss": 0.3412, "step": 1396 }, { "epoch": 0.3, "learning_rate": 1.6375963379277592e-05, "loss": 0.324, "step": 1397 }, { "epoch": 0.3, "learning_rate": 1.637057643967942e-05, "loss": 0.3133, "step": 1398 }, { "epoch": 0.3, "learning_rate": 1.636518638684325e-05, "loss": 0.278, "step": 1399 }, { "epoch": 0.3, "learning_rate": 1.6359793223403144e-05, "loss": 0.2846, "step": 1400 }, { "epoch": 0.3, "learning_rate": 1.6354396951994695e-05, "loss": 0.3866, "step": 1401 }, { "epoch": 0.3, "learning_rate": 1.6348997575255e-05, "loss": 0.2658, "step": 1402 }, { "epoch": 0.3, "learning_rate": 1.6343595095822683e-05, "loss": 0.3357, "step": 1403 }, { "epoch": 0.3, "learning_rate": 1.6338189516337887e-05, "loss": 0.3248, "step": 1404 }, { "epoch": 0.3, "learning_rate": 1.633278083944226e-05, "loss": 0.3303, "step": 1405 }, { "epoch": 0.3, "learning_rate": 1.6327369067778978e-05, "loss": 0.3377, "step": 1406 }, { "epoch": 0.3, "learning_rate": 1.6321954203992708e-05, "loss": 0.3144, "step": 1407 }, { "epoch": 0.3, "learning_rate": 1.631653625072965e-05, "loss": 0.2803, "step": 1408 }, { "epoch": 0.3, "learning_rate": 1.6311115210637502e-05, "loss": 0.317, "step": 1409 }, { "epoch": 0.3, "learning_rate": 1.6305691086365466e-05, "loss": 0.3128, "step": 1410 }, { "epoch": 0.3, "learning_rate": 1.630026388056427e-05, "loss": 0.2634, "step": 1411 }, { "epoch": 0.3, "learning_rate": 1.629483359588613e-05, "loss": 0.2924, "step": 1412 }, { "epoch": 0.3, "learning_rate": 1.628940023498477e-05, "loss": 0.2798, "step": 1413 }, { "epoch": 0.31, "learning_rate": 1.6283963800515427e-05, "loss": 0.2881, "step": 1414 }, { "epoch": 0.31, "learning_rate": 1.6278524295134826e-05, "loss": 0.2921, "step": 1415 }, { "epoch": 0.31, "learning_rate": 1.6273081721501207e-05, "loss": 0.4041, "step": 1416 }, { "epoch": 0.31, "learning_rate": 1.6267636082274294e-05, "loss": 0.2684, "step": 1417 }, { "epoch": 0.31, "learning_rate": 1.626218738011532e-05, "loss": 0.3047, "step": 1418 }, { "epoch": 0.31, "learning_rate": 1.625673561768702e-05, "loss": 0.2995, "step": 1419 }, { "epoch": 0.31, "learning_rate": 1.6251280797653606e-05, "loss": 0.3071, "step": 1420 }, { "epoch": 0.31, "learning_rate": 1.62458229226808e-05, "loss": 0.2605, "step": 1421 }, { "epoch": 0.31, "learning_rate": 1.6240361995435808e-05, "loss": 0.2924, "step": 1422 }, { "epoch": 0.31, "learning_rate": 1.6234898018587336e-05, "loss": 0.2942, "step": 1423 }, { "epoch": 0.31, "learning_rate": 1.6229430994805576e-05, "loss": 0.3391, "step": 1424 }, { "epoch": 0.31, "learning_rate": 1.6223960926762203e-05, "loss": 0.3369, "step": 1425 }, { "epoch": 0.31, "learning_rate": 1.6218487817130392e-05, "loss": 0.2958, "step": 1426 }, { "epoch": 0.31, "learning_rate": 1.621301166858479e-05, "loss": 0.3685, "step": 1427 }, { "epoch": 0.31, "learning_rate": 1.6207532483801544e-05, "loss": 0.3068, "step": 1428 }, { "epoch": 0.31, "learning_rate": 1.6202050265458278e-05, "loss": 0.3181, "step": 1429 }, { "epoch": 0.31, "learning_rate": 1.6196565016234092e-05, "loss": 0.2737, "step": 1430 }, { "epoch": 0.31, "learning_rate": 1.619107673880958e-05, "loss": 0.336, "step": 1431 }, { "epoch": 0.31, "learning_rate": 1.6185585435866802e-05, "loss": 0.3265, "step": 1432 }, { "epoch": 0.31, "learning_rate": 1.618009111008931e-05, "loss": 0.3413, "step": 1433 }, { "epoch": 0.31, "learning_rate": 1.6174593764162124e-05, "loss": 0.2558, "step": 1434 }, { "epoch": 0.31, "learning_rate": 1.6169093400771743e-05, "loss": 0.3032, "step": 1435 }, { "epoch": 0.31, "learning_rate": 1.6163590022606142e-05, "loss": 0.2861, "step": 1436 }, { "epoch": 0.31, "learning_rate": 1.6158083632354765e-05, "loss": 0.3005, "step": 1437 }, { "epoch": 0.31, "learning_rate": 1.6152574232708534e-05, "loss": 0.3098, "step": 1438 }, { "epoch": 0.31, "learning_rate": 1.6147061826359834e-05, "loss": 0.2916, "step": 1439 }, { "epoch": 0.31, "learning_rate": 1.6141546416002526e-05, "loss": 0.274, "step": 1440 }, { "epoch": 0.31, "learning_rate": 1.613602800433194e-05, "loss": 0.2319, "step": 1441 }, { "epoch": 0.31, "learning_rate": 1.6130506594044867e-05, "loss": 0.2835, "step": 1442 }, { "epoch": 0.31, "learning_rate": 1.612498218783956e-05, "loss": 0.3014, "step": 1443 }, { "epoch": 0.31, "learning_rate": 1.6119454788415755e-05, "loss": 0.3019, "step": 1444 }, { "epoch": 0.31, "learning_rate": 1.6113924398474626e-05, "loss": 0.265, "step": 1445 }, { "epoch": 0.31, "learning_rate": 1.6108391020718824e-05, "loss": 0.2925, "step": 1446 }, { "epoch": 0.31, "learning_rate": 1.6102854657852457e-05, "loss": 0.2674, "step": 1447 }, { "epoch": 0.31, "learning_rate": 1.609731531258109e-05, "loss": 0.2564, "step": 1448 }, { "epoch": 0.31, "learning_rate": 1.6091772987611743e-05, "loss": 0.2986, "step": 1449 }, { "epoch": 0.31, "learning_rate": 1.6086227685652905e-05, "loss": 0.2679, "step": 1450 }, { "epoch": 0.31, "learning_rate": 1.608067940941451e-05, "loss": 0.2879, "step": 1451 }, { "epoch": 0.31, "learning_rate": 1.607512816160793e-05, "loss": 0.3079, "step": 1452 }, { "epoch": 0.31, "learning_rate": 1.6069573944946025e-05, "loss": 0.3197, "step": 1453 }, { "epoch": 0.31, "learning_rate": 1.6064016762143073e-05, "loss": 0.3456, "step": 1454 }, { "epoch": 0.31, "learning_rate": 1.6058456615914815e-05, "loss": 0.3023, "step": 1455 }, { "epoch": 0.31, "learning_rate": 1.605289350897845e-05, "loss": 0.3148, "step": 1456 }, { "epoch": 0.31, "learning_rate": 1.6047327444052606e-05, "loss": 0.2668, "step": 1457 }, { "epoch": 0.31, "learning_rate": 1.6041758423857365e-05, "loss": 0.2927, "step": 1458 }, { "epoch": 0.31, "learning_rate": 1.603618645111425e-05, "loss": 0.2811, "step": 1459 }, { "epoch": 0.32, "learning_rate": 1.6030611528546237e-05, "loss": 0.3345, "step": 1460 }, { "epoch": 0.32, "learning_rate": 1.6025033658877728e-05, "loss": 0.4042, "step": 1461 }, { "epoch": 0.32, "learning_rate": 1.601945284483457e-05, "loss": 0.2524, "step": 1462 }, { "epoch": 0.32, "learning_rate": 1.6013869089144063e-05, "loss": 0.3355, "step": 1463 }, { "epoch": 0.32, "learning_rate": 1.600828239453493e-05, "loss": 0.3534, "step": 1464 }, { "epoch": 0.32, "learning_rate": 1.600269276373733e-05, "loss": 0.2554, "step": 1465 }, { "epoch": 0.32, "learning_rate": 1.599710019948286e-05, "loss": 0.3058, "step": 1466 }, { "epoch": 0.32, "learning_rate": 1.5991504704504555e-05, "loss": 0.308, "step": 1467 }, { "epoch": 0.32, "learning_rate": 1.5985906281536877e-05, "loss": 0.3113, "step": 1468 }, { "epoch": 0.32, "learning_rate": 1.598030493331572e-05, "loss": 0.2942, "step": 1469 }, { "epoch": 0.32, "learning_rate": 1.5974700662578406e-05, "loss": 0.3002, "step": 1470 }, { "epoch": 0.32, "learning_rate": 1.596909347206369e-05, "loss": 0.3145, "step": 1471 }, { "epoch": 0.32, "learning_rate": 1.5963483364511746e-05, "loss": 0.2526, "step": 1472 }, { "epoch": 0.32, "learning_rate": 1.5957870342664183e-05, "loss": 0.2788, "step": 1473 }, { "epoch": 0.32, "learning_rate": 1.5952254409264034e-05, "loss": 0.322, "step": 1474 }, { "epoch": 0.32, "learning_rate": 1.5946635567055737e-05, "loss": 0.2723, "step": 1475 }, { "epoch": 0.32, "learning_rate": 1.5941013818785175e-05, "loss": 0.2838, "step": 1476 }, { "epoch": 0.32, "learning_rate": 1.5935389167199638e-05, "loss": 0.2926, "step": 1477 }, { "epoch": 0.32, "learning_rate": 1.592976161504784e-05, "loss": 0.351, "step": 1478 }, { "epoch": 0.32, "learning_rate": 1.5924131165079908e-05, "loss": 0.2908, "step": 1479 }, { "epoch": 0.32, "learning_rate": 1.5918497820047388e-05, "loss": 0.3703, "step": 1480 }, { "epoch": 0.32, "learning_rate": 1.591286158270324e-05, "loss": 0.2869, "step": 1481 }, { "epoch": 0.32, "learning_rate": 1.590722245580184e-05, "loss": 0.2859, "step": 1482 }, { "epoch": 0.32, "learning_rate": 1.590158044209897e-05, "loss": 0.3117, "step": 1483 }, { "epoch": 0.32, "learning_rate": 1.5895935544351826e-05, "loss": 0.2916, "step": 1484 }, { "epoch": 0.32, "learning_rate": 1.5890287765319022e-05, "loss": 0.3431, "step": 1485 }, { "epoch": 0.32, "learning_rate": 1.5884637107760563e-05, "loss": 0.3121, "step": 1486 }, { "epoch": 0.32, "learning_rate": 1.5878983574437875e-05, "loss": 0.2917, "step": 1487 }, { "epoch": 0.32, "learning_rate": 1.587332716811378e-05, "loss": 0.2873, "step": 1488 }, { "epoch": 0.32, "learning_rate": 1.5867667891552515e-05, "loss": 0.2985, "step": 1489 }, { "epoch": 0.32, "learning_rate": 1.5862005747519704e-05, "loss": 0.2733, "step": 1490 }, { "epoch": 0.32, "learning_rate": 1.5856340738782386e-05, "loss": 0.3259, "step": 1491 }, { "epoch": 0.32, "learning_rate": 1.5850672868108994e-05, "loss": 0.2775, "step": 1492 }, { "epoch": 0.32, "learning_rate": 1.584500213826936e-05, "loss": 0.2429, "step": 1493 }, { "epoch": 0.32, "learning_rate": 1.5839328552034715e-05, "loss": 0.3087, "step": 1494 }, { "epoch": 0.32, "learning_rate": 1.5833652112177685e-05, "loss": 0.3723, "step": 1495 }, { "epoch": 0.32, "learning_rate": 1.582797282147229e-05, "loss": 0.2565, "step": 1496 }, { "epoch": 0.32, "learning_rate": 1.5822290682693944e-05, "loss": 0.2959, "step": 1497 }, { "epoch": 0.32, "learning_rate": 1.581660569861945e-05, "loss": 0.359, "step": 1498 }, { "epoch": 0.32, "learning_rate": 1.5810917872027007e-05, "loss": 0.3393, "step": 1499 }, { "epoch": 0.32, "learning_rate": 1.5805227205696198e-05, "loss": 0.3517, "step": 1500 }, { "epoch": 0.32, "learning_rate": 1.579953370240799e-05, "loss": 0.2785, "step": 1501 }, { "epoch": 0.32, "learning_rate": 1.5793837364944758e-05, "loss": 0.3007, "step": 1502 }, { "epoch": 0.32, "learning_rate": 1.5788138196090227e-05, "loss": 0.2806, "step": 1503 }, { "epoch": 0.32, "learning_rate": 1.578243619862954e-05, "loss": 0.2897, "step": 1504 }, { "epoch": 0.32, "learning_rate": 1.5776731375349195e-05, "loss": 0.2446, "step": 1505 }, { "epoch": 0.33, "learning_rate": 1.577102372903709e-05, "loss": 0.317, "step": 1506 }, { "epoch": 0.33, "learning_rate": 1.5765313262482495e-05, "loss": 0.357, "step": 1507 }, { "epoch": 0.33, "learning_rate": 1.5759599978476057e-05, "loss": 0.3514, "step": 1508 }, { "epoch": 0.33, "learning_rate": 1.5753883879809804e-05, "loss": 0.2648, "step": 1509 }, { "epoch": 0.33, "learning_rate": 1.5748164969277136e-05, "loss": 0.3111, "step": 1510 }, { "epoch": 0.33, "learning_rate": 1.574244324967283e-05, "loss": 0.2835, "step": 1511 }, { "epoch": 0.33, "learning_rate": 1.5736718723793022e-05, "loss": 0.2919, "step": 1512 }, { "epoch": 0.33, "learning_rate": 1.5730991394435252e-05, "loss": 0.2941, "step": 1513 }, { "epoch": 0.33, "learning_rate": 1.5725261264398394e-05, "loss": 0.2997, "step": 1514 }, { "epoch": 0.33, "learning_rate": 1.571952833648271e-05, "loss": 0.3285, "step": 1515 }, { "epoch": 0.33, "learning_rate": 1.5713792613489833e-05, "loss": 0.261, "step": 1516 }, { "epoch": 0.33, "learning_rate": 1.5708054098222744e-05, "loss": 0.2892, "step": 1517 }, { "epoch": 0.33, "learning_rate": 1.57023127934858e-05, "loss": 0.2835, "step": 1518 }, { "epoch": 0.33, "learning_rate": 1.5696568702084726e-05, "loss": 0.3224, "step": 1519 }, { "epoch": 0.33, "learning_rate": 1.56908218268266e-05, "loss": 0.3, "step": 1520 }, { "epoch": 0.33, "learning_rate": 1.5685072170519862e-05, "loss": 0.3236, "step": 1521 }, { "epoch": 0.33, "learning_rate": 1.5679319735974313e-05, "loss": 0.2565, "step": 1522 }, { "epoch": 0.33, "learning_rate": 1.5673564526001114e-05, "loss": 0.2745, "step": 1523 }, { "epoch": 0.33, "learning_rate": 1.5667806543412777e-05, "loss": 0.3277, "step": 1524 }, { "epoch": 0.33, "learning_rate": 1.566204579102317e-05, "loss": 0.2322, "step": 1525 }, { "epoch": 0.33, "learning_rate": 1.565628227164752e-05, "loss": 0.3198, "step": 1526 }, { "epoch": 0.33, "learning_rate": 1.5650515988102398e-05, "loss": 0.2438, "step": 1527 }, { "epoch": 0.33, "learning_rate": 1.5644746943205734e-05, "loss": 0.3384, "step": 1528 }, { "epoch": 0.33, "learning_rate": 1.56389751397768e-05, "loss": 0.3007, "step": 1529 }, { "epoch": 0.33, "learning_rate": 1.563320058063622e-05, "loss": 0.3626, "step": 1530 }, { "epoch": 0.33, "learning_rate": 1.5627423268605967e-05, "loss": 0.2703, "step": 1531 }, { "epoch": 0.33, "learning_rate": 1.5621643206509354e-05, "loss": 0.3188, "step": 1532 }, { "epoch": 0.33, "learning_rate": 1.561586039717104e-05, "loss": 0.3196, "step": 1533 }, { "epoch": 0.33, "learning_rate": 1.561007484341703e-05, "loss": 0.3432, "step": 1534 }, { "epoch": 0.33, "learning_rate": 1.5604286548074665e-05, "loss": 0.2962, "step": 1535 }, { "epoch": 0.33, "learning_rate": 1.5598495513972626e-05, "loss": 0.3812, "step": 1536 }, { "epoch": 0.33, "learning_rate": 1.5592701743940937e-05, "loss": 0.2921, "step": 1537 }, { "epoch": 0.33, "learning_rate": 1.558690524081096e-05, "loss": 0.2595, "step": 1538 }, { "epoch": 0.33, "learning_rate": 1.5581106007415382e-05, "loss": 0.256, "step": 1539 }, { "epoch": 0.33, "learning_rate": 1.557530404658823e-05, "loss": 0.286, "step": 1540 }, { "epoch": 0.33, "learning_rate": 1.5569499361164875e-05, "loss": 0.3281, "step": 1541 }, { "epoch": 0.33, "learning_rate": 1.5563691953982e-05, "loss": 0.3184, "step": 1542 }, { "epoch": 0.33, "learning_rate": 1.5557881827877633e-05, "loss": 0.2748, "step": 1543 }, { "epoch": 0.33, "learning_rate": 1.5552068985691128e-05, "loss": 0.2979, "step": 1544 }, { "epoch": 0.33, "learning_rate": 1.5546253430263157e-05, "loss": 0.299, "step": 1545 }, { "epoch": 0.33, "learning_rate": 1.5540435164435726e-05, "loss": 0.3073, "step": 1546 }, { "epoch": 0.33, "learning_rate": 1.5534614191052168e-05, "loss": 0.3126, "step": 1547 }, { "epoch": 0.33, "learning_rate": 1.552879051295713e-05, "loss": 0.2429, "step": 1548 }, { "epoch": 0.33, "learning_rate": 1.5522964132996595e-05, "loss": 0.2971, "step": 1549 }, { "epoch": 0.33, "learning_rate": 1.551713505401785e-05, "loss": 0.2694, "step": 1550 }, { "epoch": 0.33, "learning_rate": 1.5511303278869508e-05, "loss": 0.3155, "step": 1551 }, { "epoch": 0.33, "learning_rate": 1.5505468810401505e-05, "loss": 0.3032, "step": 1552 }, { "epoch": 0.34, "learning_rate": 1.5499631651465086e-05, "loss": 0.2839, "step": 1553 }, { "epoch": 0.34, "learning_rate": 1.5493791804912814e-05, "loss": 0.2697, "step": 1554 }, { "epoch": 0.34, "learning_rate": 1.548794927359856e-05, "loss": 0.2785, "step": 1555 }, { "epoch": 0.34, "learning_rate": 1.5482104060377518e-05, "loss": 0.3206, "step": 1556 }, { "epoch": 0.34, "learning_rate": 1.547625616810618e-05, "loss": 0.2722, "step": 1557 }, { "epoch": 0.34, "learning_rate": 1.547040559964236e-05, "loss": 0.3086, "step": 1558 }, { "epoch": 0.34, "learning_rate": 1.546455235784517e-05, "loss": 0.3551, "step": 1559 }, { "epoch": 0.34, "learning_rate": 1.5458696445575028e-05, "loss": 0.2953, "step": 1560 }, { "epoch": 0.34, "learning_rate": 1.5452837865693666e-05, "loss": 0.2879, "step": 1561 }, { "epoch": 0.34, "learning_rate": 1.5446976621064114e-05, "loss": 0.25, "step": 1562 }, { "epoch": 0.34, "learning_rate": 1.5441112714550702e-05, "loss": 0.2568, "step": 1563 }, { "epoch": 0.34, "learning_rate": 1.5435246149019062e-05, "loss": 0.2869, "step": 1564 }, { "epoch": 0.34, "learning_rate": 1.5429376927336132e-05, "loss": 0.3008, "step": 1565 }, { "epoch": 0.34, "learning_rate": 1.542350505237014e-05, "loss": 0.2995, "step": 1566 }, { "epoch": 0.34, "learning_rate": 1.5417630526990613e-05, "loss": 0.3356, "step": 1567 }, { "epoch": 0.34, "learning_rate": 1.5411753354068376e-05, "loss": 0.2978, "step": 1568 }, { "epoch": 0.34, "learning_rate": 1.540587353647554e-05, "loss": 0.2774, "step": 1569 }, { "epoch": 0.34, "learning_rate": 1.5399991077085525e-05, "loss": 0.3512, "step": 1570 }, { "epoch": 0.34, "learning_rate": 1.5394105978773018e-05, "loss": 0.3102, "step": 1571 }, { "epoch": 0.34, "learning_rate": 1.5388218244414015e-05, "loss": 0.2905, "step": 1572 }, { "epoch": 0.34, "learning_rate": 1.5382327876885797e-05, "loss": 0.3343, "step": 1573 }, { "epoch": 0.34, "learning_rate": 1.537643487906692e-05, "loss": 0.3184, "step": 1574 }, { "epoch": 0.34, "learning_rate": 1.5370539253837243e-05, "loss": 0.2825, "step": 1575 }, { "epoch": 0.34, "learning_rate": 1.5364641004077895e-05, "loss": 0.3481, "step": 1576 }, { "epoch": 0.34, "learning_rate": 1.5358740132671294e-05, "loss": 0.2272, "step": 1577 }, { "epoch": 0.34, "learning_rate": 1.5352836642501133e-05, "loss": 0.3062, "step": 1578 }, { "epoch": 0.34, "learning_rate": 1.53469305364524e-05, "loss": 0.2746, "step": 1579 }, { "epoch": 0.34, "learning_rate": 1.5341021817411343e-05, "loss": 0.2832, "step": 1580 }, { "epoch": 0.34, "learning_rate": 1.5335110488265497e-05, "loss": 0.2709, "step": 1581 }, { "epoch": 0.34, "learning_rate": 1.5329196551903668e-05, "loss": 0.3013, "step": 1582 }, { "epoch": 0.34, "learning_rate": 1.532328001121594e-05, "loss": 0.3738, "step": 1583 }, { "epoch": 0.34, "learning_rate": 1.5317360869093674e-05, "loss": 0.3333, "step": 1584 }, { "epoch": 0.34, "learning_rate": 1.5311439128429487e-05, "loss": 0.2766, "step": 1585 }, { "epoch": 0.34, "learning_rate": 1.5305514792117283e-05, "loss": 0.3084, "step": 1586 }, { "epoch": 0.34, "learning_rate": 1.529958786305222e-05, "loss": 0.273, "step": 1587 }, { "epoch": 0.34, "learning_rate": 1.5293658344130734e-05, "loss": 0.3078, "step": 1588 }, { "epoch": 0.34, "learning_rate": 1.528772623825052e-05, "loss": 0.3109, "step": 1589 }, { "epoch": 0.34, "learning_rate": 1.5281791548310545e-05, "loss": 0.3558, "step": 1590 }, { "epoch": 0.34, "learning_rate": 1.527585427721103e-05, "loss": 0.2736, "step": 1591 }, { "epoch": 0.34, "learning_rate": 1.526991442785346e-05, "loss": 0.3301, "step": 1592 }, { "epoch": 0.34, "learning_rate": 1.5263972003140583e-05, "loss": 0.2909, "step": 1593 }, { "epoch": 0.34, "learning_rate": 1.5258027005976401e-05, "loss": 0.3062, "step": 1594 }, { "epoch": 0.34, "learning_rate": 1.5252079439266179e-05, "loss": 0.3128, "step": 1595 }, { "epoch": 0.34, "learning_rate": 1.5246129305916429e-05, "loss": 0.3217, "step": 1596 }, { "epoch": 0.34, "learning_rate": 1.5240176608834927e-05, "loss": 0.3295, "step": 1597 }, { "epoch": 0.34, "learning_rate": 1.5234221350930698e-05, "loss": 0.3185, "step": 1598 }, { "epoch": 0.35, "learning_rate": 1.5228263535114018e-05, "loss": 0.3296, "step": 1599 }, { "epoch": 0.35, "learning_rate": 1.5222303164296407e-05, "loss": 0.2702, "step": 1600 }, { "epoch": 0.35, "learning_rate": 1.5216340241390647e-05, "loss": 0.3144, "step": 1601 }, { "epoch": 0.35, "learning_rate": 1.5210374769310759e-05, "loss": 0.3156, "step": 1602 }, { "epoch": 0.35, "learning_rate": 1.5204406750972e-05, "loss": 0.2833, "step": 1603 }, { "epoch": 0.35, "learning_rate": 1.5198436189290896e-05, "loss": 0.2522, "step": 1604 }, { "epoch": 0.35, "learning_rate": 1.5192463087185196e-05, "loss": 0.3423, "step": 1605 }, { "epoch": 0.35, "learning_rate": 1.5186487447573894e-05, "loss": 0.362, "step": 1606 }, { "epoch": 0.35, "learning_rate": 1.5180509273377228e-05, "loss": 0.3415, "step": 1607 }, { "epoch": 0.35, "learning_rate": 1.5174528567516676e-05, "loss": 0.3095, "step": 1608 }, { "epoch": 0.35, "learning_rate": 1.5168545332914942e-05, "loss": 0.2847, "step": 1609 }, { "epoch": 0.35, "learning_rate": 1.5162559572495983e-05, "loss": 0.2871, "step": 1610 }, { "epoch": 0.35, "learning_rate": 1.5156571289184975e-05, "loss": 0.3251, "step": 1611 }, { "epoch": 0.35, "learning_rate": 1.5150580485908337e-05, "loss": 0.3131, "step": 1612 }, { "epoch": 0.35, "learning_rate": 1.5144587165593713e-05, "loss": 0.2517, "step": 1613 }, { "epoch": 0.35, "learning_rate": 1.5138591331169979e-05, "loss": 0.2914, "step": 1614 }, { "epoch": 0.35, "learning_rate": 1.5132592985567242e-05, "loss": 0.2625, "step": 1615 }, { "epoch": 0.35, "learning_rate": 1.5126592131716837e-05, "loss": 0.3193, "step": 1616 }, { "epoch": 0.35, "learning_rate": 1.5120588772551317e-05, "loss": 0.3368, "step": 1617 }, { "epoch": 0.35, "learning_rate": 1.5114582911004467e-05, "loss": 0.3767, "step": 1618 }, { "epoch": 0.35, "learning_rate": 1.5108574550011295e-05, "loss": 0.3068, "step": 1619 }, { "epoch": 0.35, "learning_rate": 1.510256369250802e-05, "loss": 0.3216, "step": 1620 }, { "epoch": 0.35, "learning_rate": 1.5096550341432097e-05, "loss": 0.2484, "step": 1621 }, { "epoch": 0.35, "learning_rate": 1.5090534499722188e-05, "loss": 0.2691, "step": 1622 }, { "epoch": 0.35, "learning_rate": 1.5084516170318181e-05, "loss": 0.2659, "step": 1623 }, { "epoch": 0.35, "learning_rate": 1.5078495356161162e-05, "loss": 0.2958, "step": 1624 }, { "epoch": 0.35, "learning_rate": 1.5072472060193456e-05, "loss": 0.3179, "step": 1625 }, { "epoch": 0.35, "learning_rate": 1.5066446285358582e-05, "loss": 0.3485, "step": 1626 }, { "epoch": 0.35, "learning_rate": 1.506041803460128e-05, "loss": 0.3657, "step": 1627 }, { "epoch": 0.35, "learning_rate": 1.5054387310867496e-05, "loss": 0.2906, "step": 1628 }, { "epoch": 0.35, "learning_rate": 1.5048354117104383e-05, "loss": 0.3159, "step": 1629 }, { "epoch": 0.35, "learning_rate": 1.5042318456260305e-05, "loss": 0.2863, "step": 1630 }, { "epoch": 0.35, "learning_rate": 1.5036280331284827e-05, "loss": 0.2927, "step": 1631 }, { "epoch": 0.35, "learning_rate": 1.503023974512873e-05, "loss": 0.3481, "step": 1632 }, { "epoch": 0.35, "learning_rate": 1.502419670074398e-05, "loss": 0.3397, "step": 1633 }, { "epoch": 0.35, "learning_rate": 1.5018151201083757e-05, "loss": 0.2885, "step": 1634 }, { "epoch": 0.35, "learning_rate": 1.5012103249102437e-05, "loss": 0.3526, "step": 1635 }, { "epoch": 0.35, "learning_rate": 1.5006052847755594e-05, "loss": 0.3003, "step": 1636 }, { "epoch": 0.35, "learning_rate": 1.5000000000000002e-05, "loss": 0.3827, "step": 1637 }, { "epoch": 0.35, "learning_rate": 1.4993944708793624e-05, "loss": 0.3019, "step": 1638 }, { "epoch": 0.35, "learning_rate": 1.4987886977095623e-05, "loss": 0.3293, "step": 1639 }, { "epoch": 0.35, "learning_rate": 1.4981826807866356e-05, "loss": 0.2943, "step": 1640 }, { "epoch": 0.35, "learning_rate": 1.4975764204067364e-05, "loss": 0.2631, "step": 1641 }, { "epoch": 0.35, "learning_rate": 1.496969916866138e-05, "loss": 0.3213, "step": 1642 }, { "epoch": 0.35, "learning_rate": 1.4963631704612333e-05, "loss": 0.2323, "step": 1643 }, { "epoch": 0.35, "learning_rate": 1.495756181488533e-05, "loss": 0.2857, "step": 1644 }, { "epoch": 0.36, "learning_rate": 1.4951489502446665e-05, "loss": 0.2527, "step": 1645 }, { "epoch": 0.36, "learning_rate": 1.4945414770263821e-05, "loss": 0.2738, "step": 1646 }, { "epoch": 0.36, "learning_rate": 1.4939337621305456e-05, "loss": 0.3399, "step": 1647 }, { "epoch": 0.36, "learning_rate": 1.4933258058541415e-05, "loss": 0.2479, "step": 1648 }, { "epoch": 0.36, "learning_rate": 1.4927176084942718e-05, "loss": 0.2677, "step": 1649 }, { "epoch": 0.36, "learning_rate": 1.4921091703481572e-05, "loss": 0.2614, "step": 1650 }, { "epoch": 0.36, "learning_rate": 1.4915004917131345e-05, "loss": 0.393, "step": 1651 }, { "epoch": 0.36, "learning_rate": 1.4908915728866598e-05, "loss": 0.3097, "step": 1652 }, { "epoch": 0.36, "learning_rate": 1.4902824141663053e-05, "loss": 0.3356, "step": 1653 }, { "epoch": 0.36, "learning_rate": 1.489673015849761e-05, "loss": 0.3198, "step": 1654 }, { "epoch": 0.36, "learning_rate": 1.4890633782348336e-05, "loss": 0.3247, "step": 1655 }, { "epoch": 0.36, "learning_rate": 1.4884535016194472e-05, "loss": 0.3341, "step": 1656 }, { "epoch": 0.36, "learning_rate": 1.4878433863016427e-05, "loss": 0.328, "step": 1657 }, { "epoch": 0.36, "learning_rate": 1.4872330325795772e-05, "loss": 0.2975, "step": 1658 }, { "epoch": 0.36, "learning_rate": 1.4866224407515246e-05, "loss": 0.3897, "step": 1659 }, { "epoch": 0.36, "learning_rate": 1.4860116111158753e-05, "loss": 0.3146, "step": 1660 }, { "epoch": 0.36, "learning_rate": 1.4854005439711355e-05, "loss": 0.3141, "step": 1661 }, { "epoch": 0.36, "learning_rate": 1.4847892396159277e-05, "loss": 0.2951, "step": 1662 }, { "epoch": 0.36, "learning_rate": 1.4841776983489901e-05, "loss": 0.2198, "step": 1663 }, { "epoch": 0.36, "learning_rate": 1.4835659204691775e-05, "loss": 0.3435, "step": 1664 }, { "epoch": 0.36, "learning_rate": 1.4829539062754597e-05, "loss": 0.2862, "step": 1665 }, { "epoch": 0.36, "learning_rate": 1.4823416560669212e-05, "loss": 0.3088, "step": 1666 }, { "epoch": 0.36, "learning_rate": 1.4817291701427635e-05, "loss": 0.3284, "step": 1667 }, { "epoch": 0.36, "learning_rate": 1.481116448802302e-05, "loss": 0.3257, "step": 1668 }, { "epoch": 0.36, "learning_rate": 1.4805034923449677e-05, "loss": 0.2944, "step": 1669 }, { "epoch": 0.36, "learning_rate": 1.4798903010703066e-05, "loss": 0.3154, "step": 1670 }, { "epoch": 0.36, "learning_rate": 1.4792768752779787e-05, "loss": 0.2798, "step": 1671 }, { "epoch": 0.36, "learning_rate": 1.4786632152677596e-05, "loss": 0.3088, "step": 1672 }, { "epoch": 0.36, "learning_rate": 1.4780493213395387e-05, "loss": 0.3363, "step": 1673 }, { "epoch": 0.36, "learning_rate": 1.47743519379332e-05, "loss": 0.3286, "step": 1674 }, { "epoch": 0.36, "learning_rate": 1.4768208329292217e-05, "loss": 0.276, "step": 1675 }, { "epoch": 0.36, "learning_rate": 1.476206239047476e-05, "loss": 0.3185, "step": 1676 }, { "epoch": 0.36, "learning_rate": 1.4755914124484286e-05, "loss": 0.2777, "step": 1677 }, { "epoch": 0.36, "learning_rate": 1.4749763534325393e-05, "loss": 0.3059, "step": 1678 }, { "epoch": 0.36, "learning_rate": 1.474361062300381e-05, "loss": 0.3298, "step": 1679 }, { "epoch": 0.36, "learning_rate": 1.473745539352641e-05, "loss": 0.3084, "step": 1680 }, { "epoch": 0.36, "learning_rate": 1.4731297848901191e-05, "loss": 0.2593, "step": 1681 }, { "epoch": 0.36, "learning_rate": 1.4725137992137284e-05, "loss": 0.3148, "step": 1682 }, { "epoch": 0.36, "learning_rate": 1.4718975826244954e-05, "loss": 0.2886, "step": 1683 }, { "epoch": 0.36, "learning_rate": 1.4712811354235585e-05, "loss": 0.2786, "step": 1684 }, { "epoch": 0.36, "learning_rate": 1.4706644579121696e-05, "loss": 0.2876, "step": 1685 }, { "epoch": 0.36, "learning_rate": 1.4700475503916932e-05, "loss": 0.3753, "step": 1686 }, { "epoch": 0.36, "learning_rate": 1.4694304131636052e-05, "loss": 0.3233, "step": 1687 }, { "epoch": 0.36, "learning_rate": 1.4688130465294957e-05, "loss": 0.3299, "step": 1688 }, { "epoch": 0.36, "learning_rate": 1.4681954507910647e-05, "loss": 0.2792, "step": 1689 }, { "epoch": 0.36, "learning_rate": 1.4675776262501254e-05, "loss": 0.3026, "step": 1690 }, { "epoch": 0.36, "learning_rate": 1.4669595732086023e-05, "loss": 0.3115, "step": 1691 }, { "epoch": 0.37, "learning_rate": 1.4663412919685327e-05, "loss": 0.2828, "step": 1692 }, { "epoch": 0.37, "learning_rate": 1.4657227828320637e-05, "loss": 0.3613, "step": 1693 }, { "epoch": 0.37, "learning_rate": 1.465104046101455e-05, "loss": 0.2744, "step": 1694 }, { "epoch": 0.37, "learning_rate": 1.4644850820790773e-05, "loss": 0.3088, "step": 1695 }, { "epoch": 0.37, "learning_rate": 1.4638658910674117e-05, "loss": 0.2676, "step": 1696 }, { "epoch": 0.37, "learning_rate": 1.4632464733690514e-05, "loss": 0.3251, "step": 1697 }, { "epoch": 0.37, "learning_rate": 1.462626829286699e-05, "loss": 0.2471, "step": 1698 }, { "epoch": 0.37, "learning_rate": 1.4620069591231692e-05, "loss": 0.285, "step": 1699 }, { "epoch": 0.37, "learning_rate": 1.4613868631813859e-05, "loss": 0.3253, "step": 1700 }, { "epoch": 0.37, "learning_rate": 1.4607665417643842e-05, "loss": 0.2955, "step": 1701 }, { "epoch": 0.37, "learning_rate": 1.460145995175309e-05, "loss": 0.2896, "step": 1702 }, { "epoch": 0.37, "learning_rate": 1.4595252237174151e-05, "loss": 0.2795, "step": 1703 }, { "epoch": 0.37, "learning_rate": 1.4589042276940677e-05, "loss": 0.3487, "step": 1704 }, { "epoch": 0.37, "learning_rate": 1.4582830074087413e-05, "loss": 0.3402, "step": 1705 }, { "epoch": 0.37, "learning_rate": 1.4576615631650204e-05, "loss": 0.3368, "step": 1706 }, { "epoch": 0.37, "learning_rate": 1.4570398952665982e-05, "loss": 0.2753, "step": 1707 }, { "epoch": 0.37, "learning_rate": 1.456418004017278e-05, "loss": 0.2979, "step": 1708 }, { "epoch": 0.37, "learning_rate": 1.4557958897209716e-05, "loss": 0.2815, "step": 1709 }, { "epoch": 0.37, "learning_rate": 1.4551735526817008e-05, "loss": 0.3111, "step": 1710 }, { "epoch": 0.37, "learning_rate": 1.4545509932035949e-05, "loss": 0.3074, "step": 1711 }, { "epoch": 0.37, "learning_rate": 1.4539282115908931e-05, "loss": 0.2901, "step": 1712 }, { "epoch": 0.37, "learning_rate": 1.4533052081479421e-05, "loss": 0.3317, "step": 1713 }, { "epoch": 0.37, "learning_rate": 1.4526819831791983e-05, "loss": 0.3247, "step": 1714 }, { "epoch": 0.37, "learning_rate": 1.4520585369892246e-05, "loss": 0.3354, "step": 1715 }, { "epoch": 0.37, "learning_rate": 1.4514348698826939e-05, "loss": 0.2791, "step": 1716 }, { "epoch": 0.37, "learning_rate": 1.4508109821643859e-05, "loss": 0.3074, "step": 1717 }, { "epoch": 0.37, "learning_rate": 1.4501868741391882e-05, "loss": 0.3089, "step": 1718 }, { "epoch": 0.37, "learning_rate": 1.4495625461120966e-05, "loss": 0.3215, "step": 1719 }, { "epoch": 0.37, "learning_rate": 1.448937998388214e-05, "loss": 0.2769, "step": 1720 }, { "epoch": 0.37, "learning_rate": 1.4483132312727501e-05, "loss": 0.2849, "step": 1721 }, { "epoch": 0.37, "learning_rate": 1.447688245071023e-05, "loss": 0.2692, "step": 1722 }, { "epoch": 0.37, "learning_rate": 1.4470630400884575e-05, "loss": 0.3385, "step": 1723 }, { "epoch": 0.37, "learning_rate": 1.4464376166305847e-05, "loss": 0.2681, "step": 1724 }, { "epoch": 0.37, "learning_rate": 1.4458119750030433e-05, "loss": 0.2843, "step": 1725 }, { "epoch": 0.37, "learning_rate": 1.4451861155115774e-05, "loss": 0.3433, "step": 1726 }, { "epoch": 0.37, "learning_rate": 1.4445600384620393e-05, "loss": 0.2789, "step": 1727 }, { "epoch": 0.37, "learning_rate": 1.4439337441603863e-05, "loss": 0.2868, "step": 1728 }, { "epoch": 0.37, "learning_rate": 1.4433072329126821e-05, "loss": 0.3338, "step": 1729 }, { "epoch": 0.37, "learning_rate": 1.442680505025097e-05, "loss": 0.325, "step": 1730 }, { "epoch": 0.37, "learning_rate": 1.4420535608039063e-05, "loss": 0.3797, "step": 1731 }, { "epoch": 0.37, "learning_rate": 1.4414264005554918e-05, "loss": 0.2242, "step": 1732 }, { "epoch": 0.37, "learning_rate": 1.4407990245863401e-05, "loss": 0.2767, "step": 1733 }, { "epoch": 0.37, "learning_rate": 1.4401714332030444e-05, "loss": 0.2801, "step": 1734 }, { "epoch": 0.37, "learning_rate": 1.4395436267123017e-05, "loss": 0.2675, "step": 1735 }, { "epoch": 0.37, "learning_rate": 1.4389156054209152e-05, "loss": 0.3068, "step": 1736 }, { "epoch": 0.37, "learning_rate": 1.4382873696357926e-05, "loss": 0.3416, "step": 1737 }, { "epoch": 0.38, "learning_rate": 1.4376589196639468e-05, "loss": 0.2679, "step": 1738 }, { "epoch": 0.38, "learning_rate": 1.437030255812495e-05, "loss": 0.2443, "step": 1739 }, { "epoch": 0.38, "learning_rate": 1.4364013783886586e-05, "loss": 0.3642, "step": 1740 }, { "epoch": 0.38, "learning_rate": 1.4357722876997644e-05, "loss": 0.3268, "step": 1741 }, { "epoch": 0.38, "learning_rate": 1.4351429840532427e-05, "loss": 0.2987, "step": 1742 }, { "epoch": 0.38, "learning_rate": 1.4345134677566277e-05, "loss": 0.3258, "step": 1743 }, { "epoch": 0.38, "learning_rate": 1.4338837391175582e-05, "loss": 0.3087, "step": 1744 }, { "epoch": 0.38, "learning_rate": 1.4332537984437763e-05, "loss": 0.2738, "step": 1745 }, { "epoch": 0.38, "learning_rate": 1.4326236460431276e-05, "loss": 0.305, "step": 1746 }, { "epoch": 0.38, "learning_rate": 1.4319932822235611e-05, "loss": 0.3044, "step": 1747 }, { "epoch": 0.38, "learning_rate": 1.4313627072931304e-05, "loss": 0.3404, "step": 1748 }, { "epoch": 0.38, "learning_rate": 1.4307319215599904e-05, "loss": 0.2905, "step": 1749 }, { "epoch": 0.38, "learning_rate": 1.4301009253323999e-05, "loss": 0.358, "step": 1750 }, { "epoch": 0.38, "learning_rate": 1.429469718918721e-05, "loss": 0.363, "step": 1751 }, { "epoch": 0.38, "learning_rate": 1.4288383026274178e-05, "loss": 0.3344, "step": 1752 }, { "epoch": 0.38, "learning_rate": 1.4282066767670573e-05, "loss": 0.2868, "step": 1753 }, { "epoch": 0.38, "learning_rate": 1.4275748416463087e-05, "loss": 0.349, "step": 1754 }, { "epoch": 0.38, "learning_rate": 1.4269427975739439e-05, "loss": 0.269, "step": 1755 }, { "epoch": 0.38, "learning_rate": 1.4263105448588362e-05, "loss": 0.3096, "step": 1756 }, { "epoch": 0.38, "learning_rate": 1.4256780838099613e-05, "loss": 0.3295, "step": 1757 }, { "epoch": 0.38, "learning_rate": 1.4250454147363973e-05, "loss": 0.2958, "step": 1758 }, { "epoch": 0.38, "learning_rate": 1.4244125379473227e-05, "loss": 0.3378, "step": 1759 }, { "epoch": 0.38, "learning_rate": 1.4237794537520184e-05, "loss": 0.3415, "step": 1760 }, { "epoch": 0.38, "learning_rate": 1.4231461624598666e-05, "loss": 0.2657, "step": 1761 }, { "epoch": 0.38, "learning_rate": 1.4225126643803502e-05, "loss": 0.2708, "step": 1762 }, { "epoch": 0.38, "learning_rate": 1.4218789598230536e-05, "loss": 0.3159, "step": 1763 }, { "epoch": 0.38, "learning_rate": 1.4212450490976618e-05, "loss": 0.3346, "step": 1764 }, { "epoch": 0.38, "learning_rate": 1.4206109325139611e-05, "loss": 0.3896, "step": 1765 }, { "epoch": 0.38, "learning_rate": 1.4199766103818382e-05, "loss": 0.3316, "step": 1766 }, { "epoch": 0.38, "learning_rate": 1.4193420830112793e-05, "loss": 0.2994, "step": 1767 }, { "epoch": 0.38, "learning_rate": 1.418707350712372e-05, "loss": 0.3013, "step": 1768 }, { "epoch": 0.38, "learning_rate": 1.4180724137953041e-05, "loss": 0.3081, "step": 1769 }, { "epoch": 0.38, "learning_rate": 1.4174372725703627e-05, "loss": 0.2876, "step": 1770 }, { "epoch": 0.38, "learning_rate": 1.416801927347935e-05, "loss": 0.288, "step": 1771 }, { "epoch": 0.38, "learning_rate": 1.4161663784385082e-05, "loss": 0.3121, "step": 1772 }, { "epoch": 0.38, "learning_rate": 1.4155306261526689e-05, "loss": 0.3408, "step": 1773 }, { "epoch": 0.38, "learning_rate": 1.4148946708011024e-05, "loss": 0.3405, "step": 1774 }, { "epoch": 0.38, "learning_rate": 1.4142585126945942e-05, "loss": 0.3036, "step": 1775 }, { "epoch": 0.38, "learning_rate": 1.4136221521440285e-05, "loss": 0.3014, "step": 1776 }, { "epoch": 0.38, "learning_rate": 1.4129855894603885e-05, "loss": 0.3371, "step": 1777 }, { "epoch": 0.38, "learning_rate": 1.4123488249547561e-05, "loss": 0.261, "step": 1778 }, { "epoch": 0.38, "learning_rate": 1.4117118589383116e-05, "loss": 0.2888, "step": 1779 }, { "epoch": 0.38, "learning_rate": 1.4110746917223343e-05, "loss": 0.2579, "step": 1780 }, { "epoch": 0.38, "learning_rate": 1.4104373236182016e-05, "loss": 0.2644, "step": 1781 }, { "epoch": 0.38, "learning_rate": 1.4097997549373887e-05, "loss": 0.3031, "step": 1782 }, { "epoch": 0.38, "learning_rate": 1.4091619859914693e-05, "loss": 0.2687, "step": 1783 }, { "epoch": 0.39, "learning_rate": 1.4085240170921148e-05, "loss": 0.2907, "step": 1784 }, { "epoch": 0.39, "learning_rate": 1.4078858485510947e-05, "loss": 0.2658, "step": 1785 }, { "epoch": 0.39, "learning_rate": 1.407247480680275e-05, "loss": 0.2821, "step": 1786 }, { "epoch": 0.39, "learning_rate": 1.40660891379162e-05, "loss": 0.267, "step": 1787 }, { "epoch": 0.39, "learning_rate": 1.4059701481971914e-05, "loss": 0.2762, "step": 1788 }, { "epoch": 0.39, "learning_rate": 1.4053311842091472e-05, "loss": 0.3095, "step": 1789 }, { "epoch": 0.39, "learning_rate": 1.4046920221397435e-05, "loss": 0.3135, "step": 1790 }, { "epoch": 0.39, "learning_rate": 1.4040526623013317e-05, "loss": 0.2412, "step": 1791 }, { "epoch": 0.39, "learning_rate": 1.4034131050063611e-05, "loss": 0.2717, "step": 1792 }, { "epoch": 0.39, "learning_rate": 1.4027733505673768e-05, "loss": 0.343, "step": 1793 }, { "epoch": 0.39, "learning_rate": 1.402133399297021e-05, "loss": 0.3162, "step": 1794 }, { "epoch": 0.39, "learning_rate": 1.4014932515080308e-05, "loss": 0.2609, "step": 1795 }, { "epoch": 0.39, "learning_rate": 1.4008529075132411e-05, "loss": 0.3252, "step": 1796 }, { "epoch": 0.39, "learning_rate": 1.400212367625581e-05, "loss": 0.3199, "step": 1797 }, { "epoch": 0.39, "learning_rate": 1.399571632158076e-05, "loss": 0.2898, "step": 1798 }, { "epoch": 0.39, "learning_rate": 1.3989307014238474e-05, "loss": 0.2555, "step": 1799 }, { "epoch": 0.39, "learning_rate": 1.3982895757361119e-05, "loss": 0.2935, "step": 1800 }, { "epoch": 0.39, "learning_rate": 1.3976482554081813e-05, "loss": 0.3018, "step": 1801 }, { "epoch": 0.39, "learning_rate": 1.3970067407534624e-05, "loss": 0.3328, "step": 1802 }, { "epoch": 0.39, "learning_rate": 1.396365032085457e-05, "loss": 0.3334, "step": 1803 }, { "epoch": 0.39, "learning_rate": 1.3957231297177622e-05, "loss": 0.3288, "step": 1804 }, { "epoch": 0.39, "learning_rate": 1.3950810339640689e-05, "loss": 0.2726, "step": 1805 }, { "epoch": 0.39, "learning_rate": 1.394438745138163e-05, "loss": 0.2826, "step": 1806 }, { "epoch": 0.39, "learning_rate": 1.3937962635539251e-05, "loss": 0.2614, "step": 1807 }, { "epoch": 0.39, "learning_rate": 1.3931535895253296e-05, "loss": 0.3852, "step": 1808 }, { "epoch": 0.39, "learning_rate": 1.3925107233664446e-05, "loss": 0.3025, "step": 1809 }, { "epoch": 0.39, "learning_rate": 1.3918676653914325e-05, "loss": 0.2782, "step": 1810 }, { "epoch": 0.39, "learning_rate": 1.3912244159145498e-05, "loss": 0.3532, "step": 1811 }, { "epoch": 0.39, "learning_rate": 1.3905809752501458e-05, "loss": 0.3081, "step": 1812 }, { "epoch": 0.39, "learning_rate": 1.3899373437126637e-05, "loss": 0.3563, "step": 1813 }, { "epoch": 0.39, "learning_rate": 1.3892935216166404e-05, "loss": 0.3232, "step": 1814 }, { "epoch": 0.39, "learning_rate": 1.3886495092767045e-05, "loss": 0.2858, "step": 1815 }, { "epoch": 0.39, "learning_rate": 1.3880053070075794e-05, "loss": 0.285, "step": 1816 }, { "epoch": 0.39, "learning_rate": 1.3873609151240794e-05, "loss": 0.293, "step": 1817 }, { "epoch": 0.39, "learning_rate": 1.3867163339411135e-05, "loss": 0.2854, "step": 1818 }, { "epoch": 0.39, "learning_rate": 1.3860715637736817e-05, "loss": 0.3153, "step": 1819 }, { "epoch": 0.39, "learning_rate": 1.3854266049368769e-05, "loss": 0.2988, "step": 1820 }, { "epoch": 0.39, "learning_rate": 1.384781457745884e-05, "loss": 0.283, "step": 1821 }, { "epoch": 0.39, "learning_rate": 1.3841361225159803e-05, "loss": 0.3569, "step": 1822 }, { "epoch": 0.39, "learning_rate": 1.3834905995625346e-05, "loss": 0.2991, "step": 1823 }, { "epoch": 0.39, "learning_rate": 1.3828448892010076e-05, "loss": 0.2976, "step": 1824 }, { "epoch": 0.39, "learning_rate": 1.3821989917469516e-05, "loss": 0.3052, "step": 1825 }, { "epoch": 0.39, "learning_rate": 1.3815529075160106e-05, "loss": 0.3198, "step": 1826 }, { "epoch": 0.39, "learning_rate": 1.3809066368239195e-05, "loss": 0.3073, "step": 1827 }, { "epoch": 0.39, "learning_rate": 1.380260179986504e-05, "loss": 0.2742, "step": 1828 }, { "epoch": 0.39, "learning_rate": 1.3796135373196817e-05, "loss": 0.3341, "step": 1829 }, { "epoch": 0.39, "learning_rate": 1.3789667091394605e-05, "loss": 0.3363, "step": 1830 }, { "epoch": 0.4, "learning_rate": 1.3783196957619385e-05, "loss": 0.3211, "step": 1831 }, { "epoch": 0.4, "learning_rate": 1.3776724975033053e-05, "loss": 0.2592, "step": 1832 }, { "epoch": 0.4, "learning_rate": 1.3770251146798401e-05, "loss": 0.2937, "step": 1833 }, { "epoch": 0.4, "learning_rate": 1.3763775476079122e-05, "loss": 0.2537, "step": 1834 }, { "epoch": 0.4, "learning_rate": 1.3757297966039817e-05, "loss": 0.2765, "step": 1835 }, { "epoch": 0.4, "learning_rate": 1.375081861984598e-05, "loss": 0.2509, "step": 1836 }, { "epoch": 0.4, "learning_rate": 1.3744337440664005e-05, "loss": 0.3105, "step": 1837 }, { "epoch": 0.4, "learning_rate": 1.3737854431661177e-05, "loss": 0.3329, "step": 1838 }, { "epoch": 0.4, "learning_rate": 1.373136959600568e-05, "loss": 0.331, "step": 1839 }, { "epoch": 0.4, "learning_rate": 1.3724882936866596e-05, "loss": 0.2797, "step": 1840 }, { "epoch": 0.4, "learning_rate": 1.3718394457413878e-05, "loss": 0.2885, "step": 1841 }, { "epoch": 0.4, "learning_rate": 1.3711904160818393e-05, "loss": 0.2841, "step": 1842 }, { "epoch": 0.4, "learning_rate": 1.3705412050251883e-05, "loss": 0.2661, "step": 1843 }, { "epoch": 0.4, "learning_rate": 1.3698918128886976e-05, "loss": 0.2959, "step": 1844 }, { "epoch": 0.4, "learning_rate": 1.3692422399897187e-05, "loss": 0.3318, "step": 1845 }, { "epoch": 0.4, "learning_rate": 1.3685924866456918e-05, "loss": 0.3552, "step": 1846 }, { "epoch": 0.4, "learning_rate": 1.367942553174145e-05, "loss": 0.2323, "step": 1847 }, { "epoch": 0.4, "learning_rate": 1.367292439892694e-05, "loss": 0.3355, "step": 1848 }, { "epoch": 0.4, "learning_rate": 1.366642147119043e-05, "loss": 0.3868, "step": 1849 }, { "epoch": 0.4, "learning_rate": 1.3659916751709838e-05, "loss": 0.2926, "step": 1850 }, { "epoch": 0.4, "learning_rate": 1.3653410243663953e-05, "loss": 0.3153, "step": 1851 }, { "epoch": 0.4, "learning_rate": 1.3646901950232442e-05, "loss": 0.272, "step": 1852 }, { "epoch": 0.4, "learning_rate": 1.3640391874595846e-05, "loss": 0.3311, "step": 1853 }, { "epoch": 0.4, "learning_rate": 1.3633880019935575e-05, "loss": 0.3506, "step": 1854 }, { "epoch": 0.4, "learning_rate": 1.3627366389433907e-05, "loss": 0.3547, "step": 1855 }, { "epoch": 0.4, "learning_rate": 1.362085098627399e-05, "loss": 0.318, "step": 1856 }, { "epoch": 0.4, "learning_rate": 1.3614333813639837e-05, "loss": 0.3361, "step": 1857 }, { "epoch": 0.4, "learning_rate": 1.3607814874716326e-05, "loss": 0.3159, "step": 1858 }, { "epoch": 0.4, "learning_rate": 1.3601294172689196e-05, "loss": 0.2575, "step": 1859 }, { "epoch": 0.4, "learning_rate": 1.3594771710745054e-05, "loss": 0.2925, "step": 1860 }, { "epoch": 0.4, "learning_rate": 1.358824749207136e-05, "loss": 0.2784, "step": 1861 }, { "epoch": 0.4, "learning_rate": 1.3581721519856439e-05, "loss": 0.2984, "step": 1862 }, { "epoch": 0.4, "learning_rate": 1.3575193797289466e-05, "loss": 0.3254, "step": 1863 }, { "epoch": 0.4, "learning_rate": 1.3568664327560472e-05, "loss": 0.285, "step": 1864 }, { "epoch": 0.4, "learning_rate": 1.3562133113860351e-05, "loss": 0.3222, "step": 1865 }, { "epoch": 0.4, "learning_rate": 1.355560015938084e-05, "loss": 0.2713, "step": 1866 }, { "epoch": 0.4, "learning_rate": 1.3549065467314531e-05, "loss": 0.2808, "step": 1867 }, { "epoch": 0.4, "learning_rate": 1.3542529040854864e-05, "loss": 0.386, "step": 1868 }, { "epoch": 0.4, "learning_rate": 1.353599088319612e-05, "loss": 0.3048, "step": 1869 }, { "epoch": 0.4, "learning_rate": 1.3529450997533439e-05, "loss": 0.2453, "step": 1870 }, { "epoch": 0.4, "learning_rate": 1.3522909387062796e-05, "loss": 0.2805, "step": 1871 }, { "epoch": 0.4, "learning_rate": 1.3516366054981014e-05, "loss": 0.277, "step": 1872 }, { "epoch": 0.4, "learning_rate": 1.350982100448575e-05, "loss": 0.2682, "step": 1873 }, { "epoch": 0.4, "learning_rate": 1.3503274238775515e-05, "loss": 0.3046, "step": 1874 }, { "epoch": 0.4, "learning_rate": 1.3496725761049637e-05, "loss": 0.2708, "step": 1875 }, { "epoch": 0.4, "learning_rate": 1.3490175574508299e-05, "loss": 0.35, "step": 1876 }, { "epoch": 0.41, "learning_rate": 1.3483623682352516e-05, "loss": 0.3109, "step": 1877 }, { "epoch": 0.41, "learning_rate": 1.3477070087784128e-05, "loss": 0.2597, "step": 1878 }, { "epoch": 0.41, "learning_rate": 1.3470514794005814e-05, "loss": 0.2901, "step": 1879 }, { "epoch": 0.41, "learning_rate": 1.3463957804221079e-05, "loss": 0.3207, "step": 1880 }, { "epoch": 0.41, "learning_rate": 1.3457399121634265e-05, "loss": 0.2767, "step": 1881 }, { "epoch": 0.41, "learning_rate": 1.3450838749450529e-05, "loss": 0.3149, "step": 1882 }, { "epoch": 0.41, "learning_rate": 1.3444276690875864e-05, "loss": 0.3309, "step": 1883 }, { "epoch": 0.41, "learning_rate": 1.3437712949117083e-05, "loss": 0.2568, "step": 1884 }, { "epoch": 0.41, "learning_rate": 1.3431147527381825e-05, "loss": 0.3357, "step": 1885 }, { "epoch": 0.41, "learning_rate": 1.342458042887854e-05, "loss": 0.2906, "step": 1886 }, { "epoch": 0.41, "learning_rate": 1.341801165681651e-05, "loss": 0.2931, "step": 1887 }, { "epoch": 0.41, "learning_rate": 1.3411441214405829e-05, "loss": 0.2522, "step": 1888 }, { "epoch": 0.41, "learning_rate": 1.3404869104857405e-05, "loss": 0.2755, "step": 1889 }, { "epoch": 0.41, "learning_rate": 1.3398295331382962e-05, "loss": 0.3502, "step": 1890 }, { "epoch": 0.41, "learning_rate": 1.3391719897195041e-05, "loss": 0.3049, "step": 1891 }, { "epoch": 0.41, "learning_rate": 1.3385142805506996e-05, "loss": 0.3134, "step": 1892 }, { "epoch": 0.41, "learning_rate": 1.3378564059532981e-05, "loss": 0.3238, "step": 1893 }, { "epoch": 0.41, "learning_rate": 1.3371983662487963e-05, "loss": 0.3141, "step": 1894 }, { "epoch": 0.41, "learning_rate": 1.3365401617587723e-05, "loss": 0.3105, "step": 1895 }, { "epoch": 0.41, "learning_rate": 1.3358817928048838e-05, "loss": 0.294, "step": 1896 }, { "epoch": 0.41, "learning_rate": 1.3352232597088693e-05, "loss": 0.3228, "step": 1897 }, { "epoch": 0.41, "learning_rate": 1.3345645627925476e-05, "loss": 0.3238, "step": 1898 }, { "epoch": 0.41, "learning_rate": 1.3339057023778173e-05, "loss": 0.3511, "step": 1899 }, { "epoch": 0.41, "learning_rate": 1.3332466787866568e-05, "loss": 0.289, "step": 1900 }, { "epoch": 0.41, "learning_rate": 1.3325874923411244e-05, "loss": 0.3284, "step": 1901 }, { "epoch": 0.41, "learning_rate": 1.3319281433633584e-05, "loss": 0.2991, "step": 1902 }, { "epoch": 0.41, "learning_rate": 1.331268632175576e-05, "loss": 0.3062, "step": 1903 }, { "epoch": 0.41, "learning_rate": 1.330608959100074e-05, "loss": 0.3204, "step": 1904 }, { "epoch": 0.41, "learning_rate": 1.3299491244592277e-05, "loss": 0.281, "step": 1905 }, { "epoch": 0.41, "learning_rate": 1.3292891285754921e-05, "loss": 0.3242, "step": 1906 }, { "epoch": 0.41, "learning_rate": 1.3286289717714006e-05, "loss": 0.3505, "step": 1907 }, { "epoch": 0.41, "learning_rate": 1.3279686543695654e-05, "loss": 0.2665, "step": 1908 }, { "epoch": 0.41, "learning_rate": 1.327308176692677e-05, "loss": 0.3097, "step": 1909 }, { "epoch": 0.41, "learning_rate": 1.3266475390635047e-05, "loss": 0.3527, "step": 1910 }, { "epoch": 0.41, "learning_rate": 1.3259867418048953e-05, "loss": 0.3706, "step": 1911 }, { "epoch": 0.41, "learning_rate": 1.3253257852397733e-05, "loss": 0.3696, "step": 1912 }, { "epoch": 0.41, "learning_rate": 1.324664669691143e-05, "loss": 0.3292, "step": 1913 }, { "epoch": 0.41, "learning_rate": 1.3240033954820841e-05, "loss": 0.2792, "step": 1914 }, { "epoch": 0.41, "learning_rate": 1.323341962935755e-05, "loss": 0.279, "step": 1915 }, { "epoch": 0.41, "learning_rate": 1.3226803723753915e-05, "loss": 0.2668, "step": 1916 }, { "epoch": 0.41, "learning_rate": 1.3220186241243063e-05, "loss": 0.2619, "step": 1917 }, { "epoch": 0.41, "learning_rate": 1.3213567185058889e-05, "loss": 0.2529, "step": 1918 }, { "epoch": 0.41, "learning_rate": 1.3206946558436064e-05, "loss": 0.3248, "step": 1919 }, { "epoch": 0.41, "learning_rate": 1.3200324364610025e-05, "loss": 0.3577, "step": 1920 }, { "epoch": 0.41, "learning_rate": 1.3193700606816969e-05, "loss": 0.2709, "step": 1921 }, { "epoch": 0.41, "learning_rate": 1.3187075288293862e-05, "loss": 0.3719, "step": 1922 }, { "epoch": 0.42, "learning_rate": 1.3180448412278435e-05, "loss": 0.2753, "step": 1923 }, { "epoch": 0.42, "learning_rate": 1.3173819982009173e-05, "loss": 0.2994, "step": 1924 }, { "epoch": 0.42, "learning_rate": 1.3167190000725327e-05, "loss": 0.3712, "step": 1925 }, { "epoch": 0.42, "learning_rate": 1.3160558471666903e-05, "loss": 0.3346, "step": 1926 }, { "epoch": 0.42, "learning_rate": 1.3153925398074663e-05, "loss": 0.3086, "step": 1927 }, { "epoch": 0.42, "learning_rate": 1.3147290783190125e-05, "loss": 0.339, "step": 1928 }, { "epoch": 0.42, "learning_rate": 1.3140654630255563e-05, "loss": 0.2607, "step": 1929 }, { "epoch": 0.42, "learning_rate": 1.3134016942513998e-05, "loss": 0.332, "step": 1930 }, { "epoch": 0.42, "learning_rate": 1.31273777232092e-05, "loss": 0.2886, "step": 1931 }, { "epoch": 0.42, "learning_rate": 1.3120736975585692e-05, "loss": 0.258, "step": 1932 }, { "epoch": 0.42, "learning_rate": 1.3114094702888742e-05, "loss": 0.3035, "step": 1933 }, { "epoch": 0.42, "learning_rate": 1.3107450908364368e-05, "loss": 0.3172, "step": 1934 }, { "epoch": 0.42, "learning_rate": 1.3100805595259318e-05, "loss": 0.2556, "step": 1935 }, { "epoch": 0.42, "learning_rate": 1.3094158766821095e-05, "loss": 0.3144, "step": 1936 }, { "epoch": 0.42, "learning_rate": 1.308751042629794e-05, "loss": 0.3152, "step": 1937 }, { "epoch": 0.42, "learning_rate": 1.3080860576938829e-05, "loss": 0.2854, "step": 1938 }, { "epoch": 0.42, "learning_rate": 1.3074209221993484e-05, "loss": 0.3039, "step": 1939 }, { "epoch": 0.42, "learning_rate": 1.3067556364712351e-05, "loss": 0.2984, "step": 1940 }, { "epoch": 0.42, "learning_rate": 1.3060902008346616e-05, "loss": 0.2642, "step": 1941 }, { "epoch": 0.42, "learning_rate": 1.30542461561482e-05, "loss": 0.2797, "step": 1942 }, { "epoch": 0.42, "learning_rate": 1.3047588811369747e-05, "loss": 0.2735, "step": 1943 }, { "epoch": 0.42, "learning_rate": 1.3040929977264644e-05, "loss": 0.3614, "step": 1944 }, { "epoch": 0.42, "learning_rate": 1.3034269657086993e-05, "loss": 0.3403, "step": 1945 }, { "epoch": 0.42, "learning_rate": 1.3027607854091626e-05, "loss": 0.3045, "step": 1946 }, { "epoch": 0.42, "learning_rate": 1.3020944571534104e-05, "loss": 0.2543, "step": 1947 }, { "epoch": 0.42, "learning_rate": 1.3014279812670705e-05, "loss": 0.3293, "step": 1948 }, { "epoch": 0.42, "learning_rate": 1.3007613580758433e-05, "loss": 0.301, "step": 1949 }, { "epoch": 0.42, "learning_rate": 1.3000945879055005e-05, "loss": 0.2998, "step": 1950 }, { "epoch": 0.42, "learning_rate": 1.2994276710818866e-05, "loss": 0.2768, "step": 1951 }, { "epoch": 0.42, "learning_rate": 1.2987606079309178e-05, "loss": 0.288, "step": 1952 }, { "epoch": 0.42, "learning_rate": 1.2980933987785803e-05, "loss": 0.282, "step": 1953 }, { "epoch": 0.42, "learning_rate": 1.2974260439509326e-05, "loss": 0.321, "step": 1954 }, { "epoch": 0.42, "learning_rate": 1.296758543774105e-05, "loss": 0.2597, "step": 1955 }, { "epoch": 0.42, "learning_rate": 1.2960908985742983e-05, "loss": 0.3301, "step": 1956 }, { "epoch": 0.42, "learning_rate": 1.2954231086777838e-05, "loss": 0.3166, "step": 1957 }, { "epoch": 0.42, "learning_rate": 1.2947551744109044e-05, "loss": 0.29, "step": 1958 }, { "epoch": 0.42, "learning_rate": 1.2940870961000725e-05, "loss": 0.3369, "step": 1959 }, { "epoch": 0.42, "learning_rate": 1.2934188740717713e-05, "loss": 0.3763, "step": 1960 }, { "epoch": 0.42, "learning_rate": 1.2927505086525542e-05, "loss": 0.3647, "step": 1961 }, { "epoch": 0.42, "learning_rate": 1.2920820001690457e-05, "loss": 0.2426, "step": 1962 }, { "epoch": 0.42, "learning_rate": 1.2914133489479385e-05, "loss": 0.3651, "step": 1963 }, { "epoch": 0.42, "learning_rate": 1.2907445553159961e-05, "loss": 0.2956, "step": 1964 }, { "epoch": 0.42, "learning_rate": 1.2900756196000517e-05, "loss": 0.2832, "step": 1965 }, { "epoch": 0.42, "learning_rate": 1.289406542127007e-05, "loss": 0.2982, "step": 1966 }, { "epoch": 0.42, "learning_rate": 1.2887373232238339e-05, "loss": 0.3327, "step": 1967 }, { "epoch": 0.42, "learning_rate": 1.2880679632175728e-05, "loss": 0.3209, "step": 1968 }, { "epoch": 0.42, "learning_rate": 1.2873984624353337e-05, "loss": 0.3954, "step": 1969 }, { "epoch": 0.43, "learning_rate": 1.2867288212042954e-05, "loss": 0.3234, "step": 1970 }, { "epoch": 0.43, "learning_rate": 1.286059039851704e-05, "loss": 0.3225, "step": 1971 }, { "epoch": 0.43, "learning_rate": 1.2853891187048757e-05, "loss": 0.232, "step": 1972 }, { "epoch": 0.43, "learning_rate": 1.2847190580911942e-05, "loss": 0.3395, "step": 1973 }, { "epoch": 0.43, "learning_rate": 1.2840488583381117e-05, "loss": 0.2648, "step": 1974 }, { "epoch": 0.43, "learning_rate": 1.2833785197731476e-05, "loss": 0.3057, "step": 1975 }, { "epoch": 0.43, "learning_rate": 1.2827080427238912e-05, "loss": 0.3368, "step": 1976 }, { "epoch": 0.43, "learning_rate": 1.2820374275179963e-05, "loss": 0.3073, "step": 1977 }, { "epoch": 0.43, "learning_rate": 1.281366674483187e-05, "loss": 0.2647, "step": 1978 }, { "epoch": 0.43, "learning_rate": 1.2806957839472531e-05, "loss": 0.3026, "step": 1979 }, { "epoch": 0.43, "learning_rate": 1.280024756238053e-05, "loss": 0.2616, "step": 1980 }, { "epoch": 0.43, "learning_rate": 1.2793535916835107e-05, "loss": 0.2968, "step": 1981 }, { "epoch": 0.43, "learning_rate": 1.2786822906116178e-05, "loss": 0.3217, "step": 1982 }, { "epoch": 0.43, "learning_rate": 1.2780108533504327e-05, "loss": 0.315, "step": 1983 }, { "epoch": 0.43, "learning_rate": 1.2773392802280802e-05, "loss": 0.2791, "step": 1984 }, { "epoch": 0.43, "learning_rate": 1.2766675715727512e-05, "loss": 0.3347, "step": 1985 }, { "epoch": 0.43, "learning_rate": 1.2759957277127033e-05, "loss": 0.2416, "step": 1986 }, { "epoch": 0.43, "learning_rate": 1.27532374897626e-05, "loss": 0.2507, "step": 1987 }, { "epoch": 0.43, "learning_rate": 1.2746516356918106e-05, "loss": 0.3069, "step": 1988 }, { "epoch": 0.43, "learning_rate": 1.2739793881878103e-05, "loss": 0.3009, "step": 1989 }, { "epoch": 0.43, "learning_rate": 1.27330700679278e-05, "loss": 0.3116, "step": 1990 }, { "epoch": 0.43, "learning_rate": 1.2726344918353054e-05, "loss": 0.3452, "step": 1991 }, { "epoch": 0.43, "learning_rate": 1.2719618436440384e-05, "loss": 0.2621, "step": 1992 }, { "epoch": 0.43, "learning_rate": 1.2712890625476952e-05, "loss": 0.2814, "step": 1993 }, { "epoch": 0.43, "learning_rate": 1.270616148875058e-05, "loss": 0.2483, "step": 1994 }, { "epoch": 0.43, "learning_rate": 1.2699431029549722e-05, "loss": 0.313, "step": 1995 }, { "epoch": 0.43, "learning_rate": 1.2692699251163489e-05, "loss": 0.2788, "step": 1996 }, { "epoch": 0.43, "learning_rate": 1.2685966156881643e-05, "loss": 0.3191, "step": 1997 }, { "epoch": 0.43, "learning_rate": 1.267923174999457e-05, "loss": 0.3135, "step": 1998 }, { "epoch": 0.43, "learning_rate": 1.2672496033793311e-05, "loss": 0.3069, "step": 1999 }, { "epoch": 0.43, "learning_rate": 1.2665759011569551e-05, "loss": 0.2821, "step": 2000 }, { "epoch": 0.43, "learning_rate": 1.2659020686615602e-05, "loss": 0.3269, "step": 2001 }, { "epoch": 0.43, "learning_rate": 1.2652281062224415e-05, "loss": 0.2586, "step": 2002 }, { "epoch": 0.43, "learning_rate": 1.2645540141689576e-05, "loss": 0.3103, "step": 2003 }, { "epoch": 0.43, "learning_rate": 1.2638797928305315e-05, "loss": 0.3165, "step": 2004 }, { "epoch": 0.43, "learning_rate": 1.263205442536648e-05, "loss": 0.3495, "step": 2005 }, { "epoch": 0.43, "learning_rate": 1.2625309636168554e-05, "loss": 0.259, "step": 2006 }, { "epoch": 0.43, "learning_rate": 1.2618563564007654e-05, "loss": 0.3024, "step": 2007 }, { "epoch": 0.43, "learning_rate": 1.261181621218051e-05, "loss": 0.3206, "step": 2008 }, { "epoch": 0.43, "learning_rate": 1.2605067583984496e-05, "loss": 0.3196, "step": 2009 }, { "epoch": 0.43, "learning_rate": 1.2598317682717594e-05, "loss": 0.2906, "step": 2010 }, { "epoch": 0.43, "learning_rate": 1.2591566511678414e-05, "loss": 0.2502, "step": 2011 }, { "epoch": 0.43, "learning_rate": 1.2584814074166193e-05, "loss": 0.2944, "step": 2012 }, { "epoch": 0.43, "learning_rate": 1.2578060373480777e-05, "loss": 0.3429, "step": 2013 }, { "epoch": 0.43, "learning_rate": 1.2571305412922629e-05, "loss": 0.2501, "step": 2014 }, { "epoch": 0.43, "learning_rate": 1.2564549195792842e-05, "loss": 0.3051, "step": 2015 }, { "epoch": 0.44, "learning_rate": 1.2557791725393101e-05, "loss": 0.315, "step": 2016 }, { "epoch": 0.44, "learning_rate": 1.2551033005025718e-05, "loss": 0.3128, "step": 2017 }, { "epoch": 0.44, "learning_rate": 1.2544273037993625e-05, "loss": 0.2813, "step": 2018 }, { "epoch": 0.44, "learning_rate": 1.2537511827600338e-05, "loss": 0.3113, "step": 2019 }, { "epoch": 0.44, "learning_rate": 1.2530749377149993e-05, "loss": 0.3234, "step": 2020 }, { "epoch": 0.44, "learning_rate": 1.2523985689947344e-05, "loss": 0.312, "step": 2021 }, { "epoch": 0.44, "learning_rate": 1.2517220769297731e-05, "loss": 0.3272, "step": 2022 }, { "epoch": 0.44, "learning_rate": 1.2510454618507106e-05, "loss": 0.2467, "step": 2023 }, { "epoch": 0.44, "learning_rate": 1.250368724088202e-05, "loss": 0.3398, "step": 2024 }, { "epoch": 0.44, "learning_rate": 1.2496918639729622e-05, "loss": 0.2542, "step": 2025 }, { "epoch": 0.44, "learning_rate": 1.2490148818357666e-05, "loss": 0.2519, "step": 2026 }, { "epoch": 0.44, "learning_rate": 1.2483377780074488e-05, "loss": 0.3476, "step": 2027 }, { "epoch": 0.44, "learning_rate": 1.2476605528189038e-05, "loss": 0.3344, "step": 2028 }, { "epoch": 0.44, "learning_rate": 1.2469832066010843e-05, "loss": 0.2931, "step": 2029 }, { "epoch": 0.44, "learning_rate": 1.2463057396850027e-05, "loss": 0.3259, "step": 2030 }, { "epoch": 0.44, "learning_rate": 1.2456281524017308e-05, "loss": 0.2552, "step": 2031 }, { "epoch": 0.44, "learning_rate": 1.2449504450823983e-05, "loss": 0.3295, "step": 2032 }, { "epoch": 0.44, "learning_rate": 1.2442726180581947e-05, "loss": 0.278, "step": 2033 }, { "epoch": 0.44, "learning_rate": 1.2435946716603668e-05, "loss": 0.335, "step": 2034 }, { "epoch": 0.44, "learning_rate": 1.2429166062202207e-05, "loss": 0.2975, "step": 2035 }, { "epoch": 0.44, "learning_rate": 1.2422384220691207e-05, "loss": 0.2661, "step": 2036 }, { "epoch": 0.44, "learning_rate": 1.2415601195384876e-05, "loss": 0.3272, "step": 2037 }, { "epoch": 0.44, "learning_rate": 1.240881698959802e-05, "loss": 0.2737, "step": 2038 }, { "epoch": 0.44, "learning_rate": 1.2402031606646012e-05, "loss": 0.3003, "step": 2039 }, { "epoch": 0.44, "learning_rate": 1.2395245049844802e-05, "loss": 0.3236, "step": 2040 }, { "epoch": 0.44, "learning_rate": 1.2388457322510913e-05, "loss": 0.2815, "step": 2041 }, { "epoch": 0.44, "learning_rate": 1.238166842796144e-05, "loss": 0.3297, "step": 2042 }, { "epoch": 0.44, "learning_rate": 1.237487836951405e-05, "loss": 0.3043, "step": 2043 }, { "epoch": 0.44, "learning_rate": 1.2368087150486977e-05, "loss": 0.2706, "step": 2044 }, { "epoch": 0.44, "learning_rate": 1.2361294774199022e-05, "loss": 0.3697, "step": 2045 }, { "epoch": 0.44, "learning_rate": 1.2354501243969554e-05, "loss": 0.2901, "step": 2046 }, { "epoch": 0.44, "learning_rate": 1.2347706563118505e-05, "loss": 0.3111, "step": 2047 }, { "epoch": 0.44, "learning_rate": 1.2340910734966366e-05, "loss": 0.2485, "step": 2048 }, { "epoch": 0.44, "learning_rate": 1.2334113762834196e-05, "loss": 0.3185, "step": 2049 }, { "epoch": 0.44, "learning_rate": 1.2327315650043605e-05, "loss": 0.3407, "step": 2050 }, { "epoch": 0.44, "learning_rate": 1.2320516399916762e-05, "loss": 0.2616, "step": 2051 }, { "epoch": 0.44, "learning_rate": 1.2313716015776396e-05, "loss": 0.3617, "step": 2052 }, { "epoch": 0.44, "learning_rate": 1.2306914500945787e-05, "loss": 0.2674, "step": 2053 }, { "epoch": 0.44, "learning_rate": 1.2300111858748771e-05, "loss": 0.3081, "step": 2054 }, { "epoch": 0.44, "learning_rate": 1.2293308092509728e-05, "loss": 0.2823, "step": 2055 }, { "epoch": 0.44, "learning_rate": 1.2286503205553595e-05, "loss": 0.2567, "step": 2056 }, { "epoch": 0.44, "learning_rate": 1.2279697201205852e-05, "loss": 0.2912, "step": 2057 }, { "epoch": 0.44, "learning_rate": 1.2272890082792526e-05, "loss": 0.2945, "step": 2058 }, { "epoch": 0.44, "learning_rate": 1.2266081853640186e-05, "loss": 0.3289, "step": 2059 }, { "epoch": 0.44, "learning_rate": 1.2259272517075957e-05, "loss": 0.3044, "step": 2060 }, { "epoch": 0.44, "learning_rate": 1.2252462076427482e-05, "loss": 0.3735, "step": 2061 }, { "epoch": 0.45, "learning_rate": 1.224565053502296e-05, "loss": 0.2592, "step": 2062 }, { "epoch": 0.45, "learning_rate": 1.2238837896191131e-05, "loss": 0.3344, "step": 2063 }, { "epoch": 0.45, "learning_rate": 1.2232024163261258e-05, "loss": 0.3499, "step": 2064 }, { "epoch": 0.45, "learning_rate": 1.2225209339563144e-05, "loss": 0.316, "step": 2065 }, { "epoch": 0.45, "learning_rate": 1.2218393428427133e-05, "loss": 0.301, "step": 2066 }, { "epoch": 0.45, "learning_rate": 1.2211576433184087e-05, "loss": 0.3261, "step": 2067 }, { "epoch": 0.45, "learning_rate": 1.2204758357165409e-05, "loss": 0.2607, "step": 2068 }, { "epoch": 0.45, "learning_rate": 1.219793920370302e-05, "loss": 0.225, "step": 2069 }, { "epoch": 0.45, "learning_rate": 1.2191118976129381e-05, "loss": 0.2656, "step": 2070 }, { "epoch": 0.45, "learning_rate": 1.2184297677777463e-05, "loss": 0.2777, "step": 2071 }, { "epoch": 0.45, "learning_rate": 1.2177475311980776e-05, "loss": 0.289, "step": 2072 }, { "epoch": 0.45, "learning_rate": 1.2170651882073333e-05, "loss": 0.3183, "step": 2073 }, { "epoch": 0.45, "learning_rate": 1.216382739138968e-05, "loss": 0.294, "step": 2074 }, { "epoch": 0.45, "learning_rate": 1.2157001843264882e-05, "loss": 0.3297, "step": 2075 }, { "epoch": 0.45, "learning_rate": 1.2150175241034514e-05, "loss": 0.3089, "step": 2076 }, { "epoch": 0.45, "learning_rate": 1.2143347588034671e-05, "loss": 0.2871, "step": 2077 }, { "epoch": 0.45, "learning_rate": 1.2136518887601962e-05, "loss": 0.3155, "step": 2078 }, { "epoch": 0.45, "learning_rate": 1.2129689143073499e-05, "loss": 0.2588, "step": 2079 }, { "epoch": 0.45, "learning_rate": 1.2122858357786916e-05, "loss": 0.2711, "step": 2080 }, { "epoch": 0.45, "learning_rate": 1.2116026535080347e-05, "loss": 0.2922, "step": 2081 }, { "epoch": 0.45, "learning_rate": 1.2109193678292441e-05, "loss": 0.302, "step": 2082 }, { "epoch": 0.45, "learning_rate": 1.2102359790762349e-05, "loss": 0.2718, "step": 2083 }, { "epoch": 0.45, "learning_rate": 1.2095524875829716e-05, "loss": 0.3042, "step": 2084 }, { "epoch": 0.45, "learning_rate": 1.2088688936834705e-05, "loss": 0.2631, "step": 2085 }, { "epoch": 0.45, "learning_rate": 1.2081851977117971e-05, "loss": 0.3192, "step": 2086 }, { "epoch": 0.45, "learning_rate": 1.2075014000020665e-05, "loss": 0.2894, "step": 2087 }, { "epoch": 0.45, "learning_rate": 1.2068175008884445e-05, "loss": 0.3189, "step": 2088 }, { "epoch": 0.45, "learning_rate": 1.2061335007051453e-05, "loss": 0.3034, "step": 2089 }, { "epoch": 0.45, "learning_rate": 1.2054493997864337e-05, "loss": 0.2949, "step": 2090 }, { "epoch": 0.45, "learning_rate": 1.2047651984666224e-05, "loss": 0.2534, "step": 2091 }, { "epoch": 0.45, "learning_rate": 1.2040808970800741e-05, "loss": 0.3078, "step": 2092 }, { "epoch": 0.45, "learning_rate": 1.2033964959612003e-05, "loss": 0.3143, "step": 2093 }, { "epoch": 0.45, "learning_rate": 1.202711995444461e-05, "loss": 0.2867, "step": 2094 }, { "epoch": 0.45, "learning_rate": 1.2020273958643648e-05, "loss": 0.3065, "step": 2095 }, { "epoch": 0.45, "learning_rate": 1.2013426975554693e-05, "loss": 0.2847, "step": 2096 }, { "epoch": 0.45, "learning_rate": 1.200657900852379e-05, "loss": 0.2517, "step": 2097 }, { "epoch": 0.45, "learning_rate": 1.1999730060897476e-05, "loss": 0.3182, "step": 2098 }, { "epoch": 0.45, "learning_rate": 1.1992880136022766e-05, "loss": 0.3001, "step": 2099 }, { "epoch": 0.45, "learning_rate": 1.198602923724715e-05, "loss": 0.2482, "step": 2100 }, { "epoch": 0.45, "learning_rate": 1.1979177367918596e-05, "loss": 0.2629, "step": 2101 }, { "epoch": 0.45, "learning_rate": 1.1972324531385541e-05, "loss": 0.2993, "step": 2102 }, { "epoch": 0.45, "learning_rate": 1.1965470730996905e-05, "loss": 0.2997, "step": 2103 }, { "epoch": 0.45, "learning_rate": 1.1958615970102063e-05, "loss": 0.3483, "step": 2104 }, { "epoch": 0.45, "learning_rate": 1.1951760252050881e-05, "loss": 0.2386, "step": 2105 }, { "epoch": 0.45, "learning_rate": 1.1944903580193674e-05, "loss": 0.277, "step": 2106 }, { "epoch": 0.45, "learning_rate": 1.1938045957881233e-05, "loss": 0.3207, "step": 2107 }, { "epoch": 0.45, "learning_rate": 1.1931187388464807e-05, "loss": 0.2657, "step": 2108 }, { "epoch": 0.46, "learning_rate": 1.1924327875296113e-05, "loss": 0.2591, "step": 2109 }, { "epoch": 0.46, "learning_rate": 1.1917467421727329e-05, "loss": 0.3224, "step": 2110 }, { "epoch": 0.46, "learning_rate": 1.1910606031111089e-05, "loss": 0.3077, "step": 2111 }, { "epoch": 0.46, "learning_rate": 1.1903743706800488e-05, "loss": 0.3228, "step": 2112 }, { "epoch": 0.46, "learning_rate": 1.1896880452149077e-05, "loss": 0.2978, "step": 2113 }, { "epoch": 0.46, "learning_rate": 1.1890016270510865e-05, "loss": 0.2862, "step": 2114 }, { "epoch": 0.46, "learning_rate": 1.1883151165240298e-05, "loss": 0.2913, "step": 2115 }, { "epoch": 0.46, "learning_rate": 1.1876285139692298e-05, "loss": 0.3267, "step": 2116 }, { "epoch": 0.46, "learning_rate": 1.1869418197222217e-05, "loss": 0.3441, "step": 2117 }, { "epoch": 0.46, "learning_rate": 1.1862550341185863e-05, "loss": 0.2951, "step": 2118 }, { "epoch": 0.46, "learning_rate": 1.185568157493949e-05, "loss": 0.2886, "step": 2119 }, { "epoch": 0.46, "learning_rate": 1.1848811901839803e-05, "loss": 0.2455, "step": 2120 }, { "epoch": 0.46, "learning_rate": 1.1841941325243934e-05, "loss": 0.3067, "step": 2121 }, { "epoch": 0.46, "learning_rate": 1.1835069848509468e-05, "loss": 0.3319, "step": 2122 }, { "epoch": 0.46, "learning_rate": 1.1828197474994432e-05, "loss": 0.3214, "step": 2123 }, { "epoch": 0.46, "learning_rate": 1.1821324208057286e-05, "loss": 0.2914, "step": 2124 }, { "epoch": 0.46, "learning_rate": 1.1814450051056927e-05, "loss": 0.2569, "step": 2125 }, { "epoch": 0.46, "learning_rate": 1.1807575007352687e-05, "loss": 0.3466, "step": 2126 }, { "epoch": 0.46, "learning_rate": 1.1800699080304333e-05, "loss": 0.2932, "step": 2127 }, { "epoch": 0.46, "learning_rate": 1.1793822273272066e-05, "loss": 0.2737, "step": 2128 }, { "epoch": 0.46, "learning_rate": 1.1786944589616507e-05, "loss": 0.2948, "step": 2129 }, { "epoch": 0.46, "learning_rate": 1.1780066032698719e-05, "loss": 0.2778, "step": 2130 }, { "epoch": 0.46, "learning_rate": 1.1773186605880183e-05, "loss": 0.2793, "step": 2131 }, { "epoch": 0.46, "learning_rate": 1.1766306312522807e-05, "loss": 0.3275, "step": 2132 }, { "epoch": 0.46, "learning_rate": 1.175942515598892e-05, "loss": 0.2752, "step": 2133 }, { "epoch": 0.46, "learning_rate": 1.1752543139641276e-05, "loss": 0.3127, "step": 2134 }, { "epoch": 0.46, "learning_rate": 1.1745660266843055e-05, "loss": 0.2902, "step": 2135 }, { "epoch": 0.46, "learning_rate": 1.1738776540957839e-05, "loss": 0.287, "step": 2136 }, { "epoch": 0.46, "learning_rate": 1.1731891965349647e-05, "loss": 0.3026, "step": 2137 }, { "epoch": 0.46, "learning_rate": 1.17250065433829e-05, "loss": 0.3156, "step": 2138 }, { "epoch": 0.46, "learning_rate": 1.1718120278422429e-05, "loss": 0.2825, "step": 2139 }, { "epoch": 0.46, "learning_rate": 1.1711233173833495e-05, "loss": 0.2712, "step": 2140 }, { "epoch": 0.46, "learning_rate": 1.170434523298175e-05, "loss": 0.2809, "step": 2141 }, { "epoch": 0.46, "learning_rate": 1.169745645923327e-05, "loss": 0.2617, "step": 2142 }, { "epoch": 0.46, "learning_rate": 1.1690566855954523e-05, "loss": 0.3065, "step": 2143 }, { "epoch": 0.46, "learning_rate": 1.1683676426512396e-05, "loss": 0.2922, "step": 2144 }, { "epoch": 0.46, "learning_rate": 1.167678517427417e-05, "loss": 0.2663, "step": 2145 }, { "epoch": 0.46, "learning_rate": 1.1669893102607534e-05, "loss": 0.2209, "step": 2146 }, { "epoch": 0.46, "learning_rate": 1.1663000214880578e-05, "loss": 0.3271, "step": 2147 }, { "epoch": 0.46, "learning_rate": 1.1656106514461786e-05, "loss": 0.297, "step": 2148 }, { "epoch": 0.46, "learning_rate": 1.164921200472004e-05, "loss": 0.2277, "step": 2149 }, { "epoch": 0.46, "learning_rate": 1.1642316689024623e-05, "loss": 0.3176, "step": 2150 }, { "epoch": 0.46, "learning_rate": 1.1635420570745206e-05, "loss": 0.3005, "step": 2151 }, { "epoch": 0.46, "learning_rate": 1.1628523653251855e-05, "loss": 0.2949, "step": 2152 }, { "epoch": 0.46, "learning_rate": 1.162162593991502e-05, "loss": 0.3469, "step": 2153 }, { "epoch": 0.46, "learning_rate": 1.1614727434105552e-05, "loss": 0.3409, "step": 2154 }, { "epoch": 0.47, "learning_rate": 1.1607828139194683e-05, "loss": 0.2803, "step": 2155 }, { "epoch": 0.47, "learning_rate": 1.160092805855403e-05, "loss": 0.2917, "step": 2156 }, { "epoch": 0.47, "learning_rate": 1.1594027195555587e-05, "loss": 0.3022, "step": 2157 }, { "epoch": 0.47, "learning_rate": 1.158712555357175e-05, "loss": 0.3015, "step": 2158 }, { "epoch": 0.47, "learning_rate": 1.1580223135975274e-05, "loss": 0.2915, "step": 2159 }, { "epoch": 0.47, "learning_rate": 1.1573319946139306e-05, "loss": 0.341, "step": 2160 }, { "epoch": 0.47, "learning_rate": 1.1566415987437367e-05, "loss": 0.3267, "step": 2161 }, { "epoch": 0.47, "learning_rate": 1.1559511263243358e-05, "loss": 0.2948, "step": 2162 }, { "epoch": 0.47, "learning_rate": 1.155260577693154e-05, "loss": 0.2967, "step": 2163 }, { "epoch": 0.47, "learning_rate": 1.1545699531876563e-05, "loss": 0.3235, "step": 2164 }, { "epoch": 0.47, "learning_rate": 1.1538792531453438e-05, "loss": 0.253, "step": 2165 }, { "epoch": 0.47, "learning_rate": 1.153188477903755e-05, "loss": 0.2922, "step": 2166 }, { "epoch": 0.47, "learning_rate": 1.1524976278004651e-05, "loss": 0.3526, "step": 2167 }, { "epoch": 0.47, "learning_rate": 1.1518067031730856e-05, "loss": 0.2773, "step": 2168 }, { "epoch": 0.47, "learning_rate": 1.1511157043592642e-05, "loss": 0.2315, "step": 2169 }, { "epoch": 0.47, "learning_rate": 1.1504246316966859e-05, "loss": 0.286, "step": 2170 }, { "epoch": 0.47, "learning_rate": 1.1497334855230704e-05, "loss": 0.3415, "step": 2171 }, { "epoch": 0.47, "learning_rate": 1.1490422661761744e-05, "loss": 0.2909, "step": 2172 }, { "epoch": 0.47, "learning_rate": 1.1483509739937904e-05, "loss": 0.3141, "step": 2173 }, { "epoch": 0.47, "learning_rate": 1.1476596093137453e-05, "loss": 0.3673, "step": 2174 }, { "epoch": 0.47, "learning_rate": 1.1469681724739028e-05, "loss": 0.3069, "step": 2175 }, { "epoch": 0.47, "learning_rate": 1.146276663812161e-05, "loss": 0.3225, "step": 2176 }, { "epoch": 0.47, "learning_rate": 1.1455850836664532e-05, "loss": 0.3261, "step": 2177 }, { "epoch": 0.47, "learning_rate": 1.1448934323747481e-05, "loss": 0.2737, "step": 2178 }, { "epoch": 0.47, "learning_rate": 1.1442017102750488e-05, "loss": 0.255, "step": 2179 }, { "epoch": 0.47, "learning_rate": 1.1435099177053937e-05, "loss": 0.3157, "step": 2180 }, { "epoch": 0.47, "learning_rate": 1.1428180550038538e-05, "loss": 0.3014, "step": 2181 }, { "epoch": 0.47, "learning_rate": 1.142126122508536e-05, "loss": 0.2528, "step": 2182 }, { "epoch": 0.47, "learning_rate": 1.1414341205575817e-05, "loss": 0.3347, "step": 2183 }, { "epoch": 0.47, "learning_rate": 1.1407420494891646e-05, "loss": 0.3047, "step": 2184 }, { "epoch": 0.47, "learning_rate": 1.1400499096414932e-05, "loss": 0.2896, "step": 2185 }, { "epoch": 0.47, "learning_rate": 1.1393577013528096e-05, "loss": 0.3158, "step": 2186 }, { "epoch": 0.47, "learning_rate": 1.138665424961389e-05, "loss": 0.2456, "step": 2187 }, { "epoch": 0.47, "learning_rate": 1.13797308080554e-05, "loss": 0.2813, "step": 2188 }, { "epoch": 0.47, "learning_rate": 1.1372806692236049e-05, "loss": 0.3396, "step": 2189 }, { "epoch": 0.47, "learning_rate": 1.1365881905539577e-05, "loss": 0.3565, "step": 2190 }, { "epoch": 0.47, "learning_rate": 1.1358956451350065e-05, "loss": 0.3563, "step": 2191 }, { "epoch": 0.47, "learning_rate": 1.1352030333051911e-05, "loss": 0.258, "step": 2192 }, { "epoch": 0.47, "learning_rate": 1.1345103554029842e-05, "loss": 0.2968, "step": 2193 }, { "epoch": 0.47, "learning_rate": 1.1338176117668904e-05, "loss": 0.3018, "step": 2194 }, { "epoch": 0.47, "learning_rate": 1.1331248027354469e-05, "loss": 0.2675, "step": 2195 }, { "epoch": 0.47, "learning_rate": 1.132431928647223e-05, "loss": 0.2882, "step": 2196 }, { "epoch": 0.47, "learning_rate": 1.1317389898408188e-05, "loss": 0.3157, "step": 2197 }, { "epoch": 0.47, "learning_rate": 1.1310459866548675e-05, "loss": 0.3504, "step": 2198 }, { "epoch": 0.47, "learning_rate": 1.1303529194280319e-05, "loss": 0.37, "step": 2199 }, { "epoch": 0.47, "learning_rate": 1.1296597884990079e-05, "loss": 0.384, "step": 2200 }, { "epoch": 0.48, "learning_rate": 1.1289665942065215e-05, "loss": 0.3471, "step": 2201 }, { "epoch": 0.48, "learning_rate": 1.1282733368893297e-05, "loss": 0.3212, "step": 2202 }, { "epoch": 0.48, "learning_rate": 1.1275800168862215e-05, "loss": 0.3178, "step": 2203 }, { "epoch": 0.48, "learning_rate": 1.1268866345360144e-05, "loss": 0.3161, "step": 2204 }, { "epoch": 0.48, "learning_rate": 1.1261931901775578e-05, "loss": 0.3177, "step": 2205 }, { "epoch": 0.48, "learning_rate": 1.1254996841497311e-05, "loss": 0.2925, "step": 2206 }, { "epoch": 0.48, "learning_rate": 1.1248061167914446e-05, "loss": 0.2929, "step": 2207 }, { "epoch": 0.48, "learning_rate": 1.1241124884416372e-05, "loss": 0.2982, "step": 2208 }, { "epoch": 0.48, "learning_rate": 1.1234187994392785e-05, "loss": 0.2983, "step": 2209 }, { "epoch": 0.48, "learning_rate": 1.1227250501233671e-05, "loss": 0.3363, "step": 2210 }, { "epoch": 0.48, "learning_rate": 1.122031240832932e-05, "loss": 0.3048, "step": 2211 }, { "epoch": 0.48, "learning_rate": 1.1213373719070304e-05, "loss": 0.2876, "step": 2212 }, { "epoch": 0.48, "learning_rate": 1.1206434436847493e-05, "loss": 0.3427, "step": 2213 }, { "epoch": 0.48, "learning_rate": 1.119949456505205e-05, "loss": 0.3004, "step": 2214 }, { "epoch": 0.48, "learning_rate": 1.1192554107075421e-05, "loss": 0.3262, "step": 2215 }, { "epoch": 0.48, "learning_rate": 1.118561306630934e-05, "loss": 0.2674, "step": 2216 }, { "epoch": 0.48, "learning_rate": 1.1178671446145822e-05, "loss": 0.2746, "step": 2217 }, { "epoch": 0.48, "learning_rate": 1.1171729249977171e-05, "loss": 0.2466, "step": 2218 }, { "epoch": 0.48, "learning_rate": 1.1164786481195968e-05, "loss": 0.3329, "step": 2219 }, { "epoch": 0.48, "learning_rate": 1.1157843143195075e-05, "loss": 0.3022, "step": 2220 }, { "epoch": 0.48, "learning_rate": 1.1150899239367636e-05, "loss": 0.3217, "step": 2221 }, { "epoch": 0.48, "learning_rate": 1.114395477310707e-05, "loss": 0.3194, "step": 2222 }, { "epoch": 0.48, "learning_rate": 1.1137009747807064e-05, "loss": 0.2893, "step": 2223 }, { "epoch": 0.48, "learning_rate": 1.1130064166861588e-05, "loss": 0.305, "step": 2224 }, { "epoch": 0.48, "learning_rate": 1.1123118033664877e-05, "loss": 0.3235, "step": 2225 }, { "epoch": 0.48, "learning_rate": 1.1116171351611436e-05, "loss": 0.3189, "step": 2226 }, { "epoch": 0.48, "learning_rate": 1.1109224124096044e-05, "loss": 0.2977, "step": 2227 }, { "epoch": 0.48, "learning_rate": 1.1102276354513739e-05, "loss": 0.2684, "step": 2228 }, { "epoch": 0.48, "learning_rate": 1.1095328046259832e-05, "loss": 0.2975, "step": 2229 }, { "epoch": 0.48, "learning_rate": 1.1088379202729888e-05, "loss": 0.3272, "step": 2230 }, { "epoch": 0.48, "learning_rate": 1.108142982731974e-05, "loss": 0.3171, "step": 2231 }, { "epoch": 0.48, "learning_rate": 1.1074479923425483e-05, "loss": 0.2364, "step": 2232 }, { "epoch": 0.48, "learning_rate": 1.1067529494443463e-05, "loss": 0.2723, "step": 2233 }, { "epoch": 0.48, "learning_rate": 1.1060578543770287e-05, "loss": 0.3256, "step": 2234 }, { "epoch": 0.48, "learning_rate": 1.1053627074802815e-05, "loss": 0.3177, "step": 2235 }, { "epoch": 0.48, "learning_rate": 1.1046675090938166e-05, "loss": 0.2688, "step": 2236 }, { "epoch": 0.48, "learning_rate": 1.1039722595573699e-05, "loss": 0.2559, "step": 2237 }, { "epoch": 0.48, "learning_rate": 1.1032769592107034e-05, "loss": 0.3481, "step": 2238 }, { "epoch": 0.48, "learning_rate": 1.1025816083936036e-05, "loss": 0.295, "step": 2239 }, { "epoch": 0.48, "learning_rate": 1.1018862074458817e-05, "loss": 0.3275, "step": 2240 }, { "epoch": 0.48, "learning_rate": 1.1011907567073728e-05, "loss": 0.308, "step": 2241 }, { "epoch": 0.48, "learning_rate": 1.1004952565179372e-05, "loss": 0.304, "step": 2242 }, { "epoch": 0.48, "learning_rate": 1.099799707217459e-05, "loss": 0.3192, "step": 2243 }, { "epoch": 0.48, "learning_rate": 1.0991041091458463e-05, "loss": 0.2779, "step": 2244 }, { "epoch": 0.48, "learning_rate": 1.098408462643031e-05, "loss": 0.3696, "step": 2245 }, { "epoch": 0.48, "learning_rate": 1.0977127680489686e-05, "loss": 0.314, "step": 2246 }, { "epoch": 0.48, "learning_rate": 1.0970170257036381e-05, "loss": 0.3289, "step": 2247 }, { "epoch": 0.49, "learning_rate": 1.0963212359470422e-05, "loss": 0.3108, "step": 2248 }, { "epoch": 0.49, "learning_rate": 1.0956253991192063e-05, "loss": 0.333, "step": 2249 }, { "epoch": 0.49, "learning_rate": 1.0949295155601794e-05, "loss": 0.2568, "step": 2250 }, { "epoch": 0.49, "learning_rate": 1.0942335856100326e-05, "loss": 0.2697, "step": 2251 }, { "epoch": 0.49, "learning_rate": 1.09353760960886e-05, "loss": 0.2759, "step": 2252 }, { "epoch": 0.49, "learning_rate": 1.0928415878967781e-05, "loss": 0.3109, "step": 2253 }, { "epoch": 0.49, "learning_rate": 1.0921455208139262e-05, "loss": 0.2435, "step": 2254 }, { "epoch": 0.49, "learning_rate": 1.0914494087004652e-05, "loss": 0.2579, "step": 2255 }, { "epoch": 0.49, "learning_rate": 1.0907532518965785e-05, "loss": 0.307, "step": 2256 }, { "epoch": 0.49, "learning_rate": 1.0900570507424708e-05, "loss": 0.3375, "step": 2257 }, { "epoch": 0.49, "learning_rate": 1.0893608055783691e-05, "loss": 0.2578, "step": 2258 }, { "epoch": 0.49, "learning_rate": 1.0886645167445208e-05, "loss": 0.2928, "step": 2259 }, { "epoch": 0.49, "learning_rate": 1.0879681845811964e-05, "loss": 0.3309, "step": 2260 }, { "epoch": 0.49, "learning_rate": 1.0872718094286859e-05, "loss": 0.3032, "step": 2261 }, { "epoch": 0.49, "learning_rate": 1.086575391627301e-05, "loss": 0.3108, "step": 2262 }, { "epoch": 0.49, "learning_rate": 1.0858789315173746e-05, "loss": 0.3323, "step": 2263 }, { "epoch": 0.49, "learning_rate": 1.0851824294392598e-05, "loss": 0.3292, "step": 2264 }, { "epoch": 0.49, "learning_rate": 1.0844858857333302e-05, "loss": 0.3404, "step": 2265 }, { "epoch": 0.49, "learning_rate": 1.0837893007399796e-05, "loss": 0.3215, "step": 2266 }, { "epoch": 0.49, "learning_rate": 1.0830926747996225e-05, "loss": 0.3963, "step": 2267 }, { "epoch": 0.49, "learning_rate": 1.0823960082526932e-05, "loss": 0.317, "step": 2268 }, { "epoch": 0.49, "learning_rate": 1.0816993014396457e-05, "loss": 0.3393, "step": 2269 }, { "epoch": 0.49, "learning_rate": 1.0810025547009539e-05, "loss": 0.282, "step": 2270 }, { "epoch": 0.49, "learning_rate": 1.080305768377111e-05, "loss": 0.333, "step": 2271 }, { "epoch": 0.49, "learning_rate": 1.0796089428086294e-05, "loss": 0.3028, "step": 2272 }, { "epoch": 0.49, "learning_rate": 1.0789120783360408e-05, "loss": 0.2744, "step": 2273 }, { "epoch": 0.49, "learning_rate": 1.0782151752998965e-05, "loss": 0.286, "step": 2274 }, { "epoch": 0.49, "learning_rate": 1.077518234040766e-05, "loss": 0.262, "step": 2275 }, { "epoch": 0.49, "learning_rate": 1.0768212548992376e-05, "loss": 0.3574, "step": 2276 }, { "epoch": 0.49, "learning_rate": 1.076124238215918e-05, "loss": 0.2659, "step": 2277 }, { "epoch": 0.49, "learning_rate": 1.0754271843314326e-05, "loss": 0.3229, "step": 2278 }, { "epoch": 0.49, "learning_rate": 1.0747300935864245e-05, "loss": 0.2967, "step": 2279 }, { "epoch": 0.49, "learning_rate": 1.0740329663215551e-05, "loss": 0.2816, "step": 2280 }, { "epoch": 0.49, "learning_rate": 1.073335802877504e-05, "loss": 0.3011, "step": 2281 }, { "epoch": 0.49, "learning_rate": 1.072638603594968e-05, "loss": 0.3419, "step": 2282 }, { "epoch": 0.49, "learning_rate": 1.0719413688146612e-05, "loss": 0.3011, "step": 2283 }, { "epoch": 0.49, "learning_rate": 1.0712440988773159e-05, "loss": 0.3361, "step": 2284 }, { "epoch": 0.49, "learning_rate": 1.0705467941236805e-05, "loss": 0.3409, "step": 2285 }, { "epoch": 0.49, "learning_rate": 1.0698494548945214e-05, "loss": 0.2835, "step": 2286 }, { "epoch": 0.49, "learning_rate": 1.0691520815306209e-05, "loss": 0.3053, "step": 2287 }, { "epoch": 0.49, "learning_rate": 1.068454674372779e-05, "loss": 0.3151, "step": 2288 }, { "epoch": 0.49, "learning_rate": 1.0677572337618116e-05, "loss": 0.2799, "step": 2289 }, { "epoch": 0.49, "learning_rate": 1.0670597600385509e-05, "loss": 0.2816, "step": 2290 }, { "epoch": 0.49, "learning_rate": 1.0663622535438458e-05, "loss": 0.32, "step": 2291 }, { "epoch": 0.49, "learning_rate": 1.0656647146185604e-05, "loss": 0.3123, "step": 2292 }, { "epoch": 0.49, "learning_rate": 1.0649671436035754e-05, "loss": 0.3034, "step": 2293 }, { "epoch": 0.5, "learning_rate": 1.0642695408397868e-05, "loss": 0.3051, "step": 2294 }, { "epoch": 0.5, "learning_rate": 1.0635719066681064e-05, "loss": 0.2957, "step": 2295 }, { "epoch": 0.5, "learning_rate": 1.0628742414294606e-05, "loss": 0.2661, "step": 2296 }, { "epoch": 0.5, "learning_rate": 1.0621765454647919e-05, "loss": 0.2897, "step": 2297 }, { "epoch": 0.5, "learning_rate": 1.0614788191150575e-05, "loss": 0.3663, "step": 2298 }, { "epoch": 0.5, "learning_rate": 1.0607810627212296e-05, "loss": 0.37, "step": 2299 }, { "epoch": 0.5, "learning_rate": 1.0600832766242948e-05, "loss": 0.2898, "step": 2300 }, { "epoch": 0.5, "learning_rate": 1.0593854611652537e-05, "loss": 0.3165, "step": 2301 }, { "epoch": 0.5, "learning_rate": 1.0586876166851223e-05, "loss": 0.324, "step": 2302 }, { "epoch": 0.5, "learning_rate": 1.0579897435249305e-05, "loss": 0.3057, "step": 2303 }, { "epoch": 0.5, "learning_rate": 1.0572918420257214e-05, "loss": 0.4095, "step": 2304 }, { "epoch": 0.5, "learning_rate": 1.0565939125285536e-05, "loss": 0.296, "step": 2305 }, { "epoch": 0.5, "learning_rate": 1.0558959553744977e-05, "loss": 0.3073, "step": 2306 }, { "epoch": 0.5, "learning_rate": 1.0551979709046385e-05, "loss": 0.3144, "step": 2307 }, { "epoch": 0.5, "learning_rate": 1.054499959460074e-05, "loss": 0.2959, "step": 2308 }, { "epoch": 0.5, "learning_rate": 1.053801921381916e-05, "loss": 0.3074, "step": 2309 }, { "epoch": 0.5, "learning_rate": 1.0531038570112886e-05, "loss": 0.3284, "step": 2310 }, { "epoch": 0.5, "learning_rate": 1.052405766689329e-05, "loss": 0.2927, "step": 2311 }, { "epoch": 0.5, "learning_rate": 1.051707650757187e-05, "loss": 0.2713, "step": 2312 }, { "epoch": 0.5, "learning_rate": 1.0510095095560254e-05, "loss": 0.2814, "step": 2313 }, { "epoch": 0.5, "learning_rate": 1.0503113434270182e-05, "loss": 0.2419, "step": 2314 }, { "epoch": 0.5, "learning_rate": 1.0496131527113527e-05, "loss": 0.3275, "step": 2315 }, { "epoch": 0.5, "learning_rate": 1.0489149377502276e-05, "loss": 0.2894, "step": 2316 }, { "epoch": 0.5, "learning_rate": 1.0482166988848543e-05, "loss": 0.2777, "step": 2317 }, { "epoch": 0.5, "learning_rate": 1.0475184364564548e-05, "loss": 0.2726, "step": 2318 }, { "epoch": 0.5, "learning_rate": 1.0468201508062632e-05, "loss": 0.3081, "step": 2319 }, { "epoch": 0.5, "learning_rate": 1.0461218422755248e-05, "loss": 0.3114, "step": 2320 }, { "epoch": 0.5, "learning_rate": 1.0454235112054957e-05, "loss": 0.3242, "step": 2321 }, { "epoch": 0.5, "learning_rate": 1.044725157937444e-05, "loss": 0.3189, "step": 2322 }, { "epoch": 0.5, "learning_rate": 1.0440267828126478e-05, "loss": 0.303, "step": 2323 }, { "epoch": 0.5, "learning_rate": 1.0433283861723967e-05, "loss": 0.2856, "step": 2324 }, { "epoch": 0.5, "learning_rate": 1.0426299683579891e-05, "loss": 0.2824, "step": 2325 }, { "epoch": 0.5, "learning_rate": 1.0419315297107358e-05, "loss": 0.3252, "step": 2326 }, { "epoch": 0.5, "learning_rate": 1.0412330705719566e-05, "loss": 0.3113, "step": 2327 }, { "epoch": 0.5, "learning_rate": 1.0405345912829818e-05, "loss": 0.3588, "step": 2328 }, { "epoch": 0.5, "learning_rate": 1.0398360921851514e-05, "loss": 0.2827, "step": 2329 }, { "epoch": 0.5, "learning_rate": 1.0391375736198146e-05, "loss": 0.289, "step": 2330 }, { "epoch": 0.5, "learning_rate": 1.038439035928331e-05, "loss": 0.2755, "step": 2331 }, { "epoch": 0.5, "learning_rate": 1.0377404794520685e-05, "loss": 0.3061, "step": 2332 }, { "epoch": 0.5, "learning_rate": 1.0370419045324056e-05, "loss": 0.2401, "step": 2333 }, { "epoch": 0.5, "learning_rate": 1.0363433115107285e-05, "loss": 0.2861, "step": 2334 }, { "epoch": 0.5, "learning_rate": 1.0356447007284327e-05, "loss": 0.3194, "step": 2335 }, { "epoch": 0.5, "learning_rate": 1.0349460725269226e-05, "loss": 0.2945, "step": 2336 }, { "epoch": 0.5, "learning_rate": 1.0342474272476108e-05, "loss": 0.2897, "step": 2337 }, { "epoch": 0.5, "learning_rate": 1.0335487652319182e-05, "loss": 0.2831, "step": 2338 }, { "epoch": 0.5, "learning_rate": 1.0328500868212745e-05, "loss": 0.3026, "step": 2339 }, { "epoch": 0.51, "learning_rate": 1.0321513923571166e-05, "loss": 0.2679, "step": 2340 }, { "epoch": 0.51, "learning_rate": 1.03145268218089e-05, "loss": 0.2663, "step": 2341 }, { "epoch": 0.51, "learning_rate": 1.0307539566340471e-05, "loss": 0.3051, "step": 2342 }, { "epoch": 0.51, "learning_rate": 1.0300552160580479e-05, "loss": 0.2722, "step": 2343 }, { "epoch": 0.51, "learning_rate": 1.0293564607943609e-05, "loss": 0.2584, "step": 2344 }, { "epoch": 0.51, "learning_rate": 1.0286576911844603e-05, "loss": 0.2935, "step": 2345 }, { "epoch": 0.51, "learning_rate": 1.027958907569828e-05, "loss": 0.2981, "step": 2346 }, { "epoch": 0.51, "learning_rate": 1.0272601102919533e-05, "loss": 0.3452, "step": 2347 }, { "epoch": 0.51, "learning_rate": 1.0265612996923306e-05, "loss": 0.3065, "step": 2348 }, { "epoch": 0.51, "learning_rate": 1.0258624761124622e-05, "loss": 0.2861, "step": 2349 }, { "epoch": 0.51, "learning_rate": 1.0251636398938563e-05, "loss": 0.292, "step": 2350 }, { "epoch": 0.51, "learning_rate": 1.0244647913780272e-05, "loss": 0.3465, "step": 2351 }, { "epoch": 0.51, "learning_rate": 1.0237659309064956e-05, "loss": 0.328, "step": 2352 }, { "epoch": 0.51, "learning_rate": 1.0230670588207874e-05, "loss": 0.3122, "step": 2353 }, { "epoch": 0.51, "learning_rate": 1.0223681754624346e-05, "loss": 0.3047, "step": 2354 }, { "epoch": 0.51, "learning_rate": 1.0216692811729747e-05, "loss": 0.3195, "step": 2355 }, { "epoch": 0.51, "learning_rate": 1.0209703762939502e-05, "loss": 0.3111, "step": 2356 }, { "epoch": 0.51, "learning_rate": 1.0202714611669091e-05, "loss": 0.2707, "step": 2357 }, { "epoch": 0.51, "learning_rate": 1.0195725361334048e-05, "loss": 0.2479, "step": 2358 }, { "epoch": 0.51, "learning_rate": 1.0188736015349948e-05, "loss": 0.267, "step": 2359 }, { "epoch": 0.51, "learning_rate": 1.0181746577132413e-05, "loss": 0.3203, "step": 2360 }, { "epoch": 0.51, "learning_rate": 1.0174757050097115e-05, "loss": 0.305, "step": 2361 }, { "epoch": 0.51, "learning_rate": 1.0167767437659768e-05, "loss": 0.2811, "step": 2362 }, { "epoch": 0.51, "learning_rate": 1.0160777743236124e-05, "loss": 0.3178, "step": 2363 }, { "epoch": 0.51, "learning_rate": 1.0153787970241977e-05, "loss": 0.3837, "step": 2364 }, { "epoch": 0.51, "learning_rate": 1.0146798122093167e-05, "loss": 0.3013, "step": 2365 }, { "epoch": 0.51, "learning_rate": 1.0139808202205561e-05, "loss": 0.2429, "step": 2366 }, { "epoch": 0.51, "learning_rate": 1.0132818213995058e-05, "loss": 0.2374, "step": 2367 }, { "epoch": 0.51, "learning_rate": 1.0125828160877605e-05, "loss": 0.2446, "step": 2368 }, { "epoch": 0.51, "learning_rate": 1.0118838046269166e-05, "loss": 0.2674, "step": 2369 }, { "epoch": 0.51, "learning_rate": 1.0111847873585743e-05, "loss": 0.3091, "step": 2370 }, { "epoch": 0.51, "learning_rate": 1.0104857646243366e-05, "loss": 0.3372, "step": 2371 }, { "epoch": 0.51, "learning_rate": 1.0097867367658091e-05, "loss": 0.3, "step": 2372 }, { "epoch": 0.51, "learning_rate": 1.0090877041245997e-05, "loss": 0.2676, "step": 2373 }, { "epoch": 0.51, "learning_rate": 1.0083886670423181e-05, "loss": 0.2777, "step": 2374 }, { "epoch": 0.51, "learning_rate": 1.0076896258605782e-05, "loss": 0.2972, "step": 2375 }, { "epoch": 0.51, "learning_rate": 1.0069905809209936e-05, "loss": 0.3623, "step": 2376 }, { "epoch": 0.51, "learning_rate": 1.0062915325651809e-05, "loss": 0.3376, "step": 2377 }, { "epoch": 0.51, "learning_rate": 1.0055924811347583e-05, "loss": 0.2799, "step": 2378 }, { "epoch": 0.51, "learning_rate": 1.004893426971345e-05, "loss": 0.2409, "step": 2379 }, { "epoch": 0.51, "learning_rate": 1.0041943704165625e-05, "loss": 0.2549, "step": 2380 }, { "epoch": 0.51, "learning_rate": 1.003495311812032e-05, "loss": 0.3157, "step": 2381 }, { "epoch": 0.51, "learning_rate": 1.0027962514993773e-05, "loss": 0.293, "step": 2382 }, { "epoch": 0.51, "learning_rate": 1.0020971898202218e-05, "loss": 0.2444, "step": 2383 }, { "epoch": 0.51, "learning_rate": 1.0013981271161905e-05, "loss": 0.2751, "step": 2384 }, { "epoch": 0.51, "learning_rate": 1.000699063728908e-05, "loss": 0.3654, "step": 2385 }, { "epoch": 0.51, "learning_rate": 1e-05, "loss": 0.2598, "step": 2386 }, { "epoch": 0.52, "learning_rate": 9.99300936271092e-06, "loss": 0.2914, "step": 2387 }, { "epoch": 0.52, "learning_rate": 9.9860187288381e-06, "loss": 0.3, "step": 2388 }, { "epoch": 0.52, "learning_rate": 9.979028101797785e-06, "loss": 0.3242, "step": 2389 }, { "epoch": 0.52, "learning_rate": 9.972037485006229e-06, "loss": 0.3078, "step": 2390 }, { "epoch": 0.52, "learning_rate": 9.965046881879684e-06, "loss": 0.3099, "step": 2391 }, { "epoch": 0.52, "learning_rate": 9.958056295834378e-06, "loss": 0.2807, "step": 2392 }, { "epoch": 0.52, "learning_rate": 9.951065730286553e-06, "loss": 0.2429, "step": 2393 }, { "epoch": 0.52, "learning_rate": 9.94407518865242e-06, "loss": 0.3071, "step": 2394 }, { "epoch": 0.52, "learning_rate": 9.937084674348193e-06, "loss": 0.2737, "step": 2395 }, { "epoch": 0.52, "learning_rate": 9.930094190790067e-06, "loss": 0.3026, "step": 2396 }, { "epoch": 0.52, "learning_rate": 9.923103741394222e-06, "loss": 0.271, "step": 2397 }, { "epoch": 0.52, "learning_rate": 9.916113329576817e-06, "loss": 0.2804, "step": 2398 }, { "epoch": 0.52, "learning_rate": 9.90912295875401e-06, "loss": 0.3211, "step": 2399 }, { "epoch": 0.52, "learning_rate": 9.902132632341912e-06, "loss": 0.3329, "step": 2400 }, { "epoch": 0.52, "learning_rate": 9.895142353756635e-06, "loss": 0.3057, "step": 2401 }, { "epoch": 0.52, "learning_rate": 9.888152126414262e-06, "loss": 0.2627, "step": 2402 }, { "epoch": 0.52, "learning_rate": 9.881161953730838e-06, "loss": 0.3362, "step": 2403 }, { "epoch": 0.52, "learning_rate": 9.874171839122397e-06, "loss": 0.3039, "step": 2404 }, { "epoch": 0.52, "learning_rate": 9.867181786004944e-06, "loss": 0.2536, "step": 2405 }, { "epoch": 0.52, "learning_rate": 9.860191797794442e-06, "loss": 0.249, "step": 2406 }, { "epoch": 0.52, "learning_rate": 9.853201877906836e-06, "loss": 0.3124, "step": 2407 }, { "epoch": 0.52, "learning_rate": 9.846212029758026e-06, "loss": 0.2707, "step": 2408 }, { "epoch": 0.52, "learning_rate": 9.839222256763878e-06, "loss": 0.3085, "step": 2409 }, { "epoch": 0.52, "learning_rate": 9.832232562340239e-06, "loss": 0.3123, "step": 2410 }, { "epoch": 0.52, "learning_rate": 9.825242949902888e-06, "loss": 0.3176, "step": 2411 }, { "epoch": 0.52, "learning_rate": 9.818253422867589e-06, "loss": 0.2249, "step": 2412 }, { "epoch": 0.52, "learning_rate": 9.811263984650056e-06, "loss": 0.2903, "step": 2413 }, { "epoch": 0.52, "learning_rate": 9.804274638665955e-06, "loss": 0.2781, "step": 2414 }, { "epoch": 0.52, "learning_rate": 9.797285388330909e-06, "loss": 0.2733, "step": 2415 }, { "epoch": 0.52, "learning_rate": 9.790296237060503e-06, "loss": 0.3054, "step": 2416 }, { "epoch": 0.52, "learning_rate": 9.783307188270256e-06, "loss": 0.265, "step": 2417 }, { "epoch": 0.52, "learning_rate": 9.776318245375657e-06, "loss": 0.2554, "step": 2418 }, { "epoch": 0.52, "learning_rate": 9.769329411792129e-06, "loss": 0.3521, "step": 2419 }, { "epoch": 0.52, "learning_rate": 9.762340690935047e-06, "loss": 0.3042, "step": 2420 }, { "epoch": 0.52, "learning_rate": 9.755352086219733e-06, "loss": 0.2801, "step": 2421 }, { "epoch": 0.52, "learning_rate": 9.74836360106144e-06, "loss": 0.2394, "step": 2422 }, { "epoch": 0.52, "learning_rate": 9.74137523887538e-06, "loss": 0.257, "step": 2423 }, { "epoch": 0.52, "learning_rate": 9.734387003076699e-06, "loss": 0.3321, "step": 2424 }, { "epoch": 0.52, "learning_rate": 9.72739889708047e-06, "loss": 0.2628, "step": 2425 }, { "epoch": 0.52, "learning_rate": 9.72041092430172e-06, "loss": 0.2883, "step": 2426 }, { "epoch": 0.52, "learning_rate": 9.7134230881554e-06, "loss": 0.3168, "step": 2427 }, { "epoch": 0.52, "learning_rate": 9.706435392056394e-06, "loss": 0.2743, "step": 2428 }, { "epoch": 0.52, "learning_rate": 9.69944783941952e-06, "loss": 0.2752, "step": 2429 }, { "epoch": 0.52, "learning_rate": 9.692460433659534e-06, "loss": 0.3107, "step": 2430 }, { "epoch": 0.52, "learning_rate": 9.685473178191106e-06, "loss": 0.2542, "step": 2431 }, { "epoch": 0.52, "learning_rate": 9.678486076428835e-06, "loss": 0.2975, "step": 2432 }, { "epoch": 0.53, "learning_rate": 9.671499131787259e-06, "loss": 0.3476, "step": 2433 }, { "epoch": 0.53, "learning_rate": 9.66451234768082e-06, "loss": 0.2753, "step": 2434 }, { "epoch": 0.53, "learning_rate": 9.657525727523897e-06, "loss": 0.2758, "step": 2435 }, { "epoch": 0.53, "learning_rate": 9.650539274730778e-06, "loss": 0.2862, "step": 2436 }, { "epoch": 0.53, "learning_rate": 9.643552992715676e-06, "loss": 0.3083, "step": 2437 }, { "epoch": 0.53, "learning_rate": 9.636566884892718e-06, "loss": 0.2975, "step": 2438 }, { "epoch": 0.53, "learning_rate": 9.629580954675948e-06, "loss": 0.2725, "step": 2439 }, { "epoch": 0.53, "learning_rate": 9.622595205479315e-06, "loss": 0.3198, "step": 2440 }, { "epoch": 0.53, "learning_rate": 9.615609640716695e-06, "loss": 0.3716, "step": 2441 }, { "epoch": 0.53, "learning_rate": 9.608624263801858e-06, "loss": 0.3422, "step": 2442 }, { "epoch": 0.53, "learning_rate": 9.60163907814849e-06, "loss": 0.2711, "step": 2443 }, { "epoch": 0.53, "learning_rate": 9.594654087170186e-06, "loss": 0.3173, "step": 2444 }, { "epoch": 0.53, "learning_rate": 9.587669294280435e-06, "loss": 0.289, "step": 2445 }, { "epoch": 0.53, "learning_rate": 9.580684702892643e-06, "loss": 0.2797, "step": 2446 }, { "epoch": 0.53, "learning_rate": 9.573700316420112e-06, "loss": 0.2983, "step": 2447 }, { "epoch": 0.53, "learning_rate": 9.566716138276036e-06, "loss": 0.2475, "step": 2448 }, { "epoch": 0.53, "learning_rate": 9.559732171873524e-06, "loss": 0.3113, "step": 2449 }, { "epoch": 0.53, "learning_rate": 9.552748420625563e-06, "loss": 0.2673, "step": 2450 }, { "epoch": 0.53, "learning_rate": 9.545764887945044e-06, "loss": 0.3293, "step": 2451 }, { "epoch": 0.53, "learning_rate": 9.538781577244758e-06, "loss": 0.3175, "step": 2452 }, { "epoch": 0.53, "learning_rate": 9.531798491937371e-06, "loss": 0.3273, "step": 2453 }, { "epoch": 0.53, "learning_rate": 9.524815635435452e-06, "loss": 0.3081, "step": 2454 }, { "epoch": 0.53, "learning_rate": 9.517833011151459e-06, "loss": 0.2728, "step": 2455 }, { "epoch": 0.53, "learning_rate": 9.510850622497726e-06, "loss": 0.2979, "step": 2456 }, { "epoch": 0.53, "learning_rate": 9.503868472886475e-06, "loss": 0.2787, "step": 2457 }, { "epoch": 0.53, "learning_rate": 9.496886565729823e-06, "loss": 0.3655, "step": 2458 }, { "epoch": 0.53, "learning_rate": 9.48990490443975e-06, "loss": 0.301, "step": 2459 }, { "epoch": 0.53, "learning_rate": 9.482923492428131e-06, "loss": 0.2926, "step": 2460 }, { "epoch": 0.53, "learning_rate": 9.475942333106713e-06, "loss": 0.326, "step": 2461 }, { "epoch": 0.53, "learning_rate": 9.468961429887115e-06, "loss": 0.2582, "step": 2462 }, { "epoch": 0.53, "learning_rate": 9.461980786180844e-06, "loss": 0.3148, "step": 2463 }, { "epoch": 0.53, "learning_rate": 9.455000405399262e-06, "loss": 0.2859, "step": 2464 }, { "epoch": 0.53, "learning_rate": 9.448020290953618e-06, "loss": 0.3707, "step": 2465 }, { "epoch": 0.53, "learning_rate": 9.441040446255028e-06, "loss": 0.2673, "step": 2466 }, { "epoch": 0.53, "learning_rate": 9.434060874714467e-06, "loss": 0.2388, "step": 2467 }, { "epoch": 0.53, "learning_rate": 9.427081579742786e-06, "loss": 0.3154, "step": 2468 }, { "epoch": 0.53, "learning_rate": 9.420102564750702e-06, "loss": 0.2688, "step": 2469 }, { "epoch": 0.53, "learning_rate": 9.41312383314878e-06, "loss": 0.327, "step": 2470 }, { "epoch": 0.53, "learning_rate": 9.406145388347465e-06, "loss": 0.2908, "step": 2471 }, { "epoch": 0.53, "learning_rate": 9.399167233757057e-06, "loss": 0.2717, "step": 2472 }, { "epoch": 0.53, "learning_rate": 9.392189372787707e-06, "loss": 0.3415, "step": 2473 }, { "epoch": 0.53, "learning_rate": 9.385211808849425e-06, "loss": 0.3188, "step": 2474 }, { "epoch": 0.53, "learning_rate": 9.378234545352084e-06, "loss": 0.2985, "step": 2475 }, { "epoch": 0.53, "learning_rate": 9.371257585705397e-06, "loss": 0.2601, "step": 2476 }, { "epoch": 0.53, "learning_rate": 9.364280933318943e-06, "loss": 0.329, "step": 2477 }, { "epoch": 0.53, "learning_rate": 9.357304591602134e-06, "loss": 0.2817, "step": 2478 }, { "epoch": 0.54, "learning_rate": 9.350328563964248e-06, "loss": 0.3135, "step": 2479 }, { "epoch": 0.54, "learning_rate": 9.343352853814401e-06, "loss": 0.2805, "step": 2480 }, { "epoch": 0.54, "learning_rate": 9.336377464561545e-06, "loss": 0.2738, "step": 2481 }, { "epoch": 0.54, "learning_rate": 9.329402399614491e-06, "loss": 0.2848, "step": 2482 }, { "epoch": 0.54, "learning_rate": 9.322427662381887e-06, "loss": 0.3443, "step": 2483 }, { "epoch": 0.54, "learning_rate": 9.31545325627221e-06, "loss": 0.2916, "step": 2484 }, { "epoch": 0.54, "learning_rate": 9.308479184693793e-06, "loss": 0.2838, "step": 2485 }, { "epoch": 0.54, "learning_rate": 9.301505451054793e-06, "loss": 0.292, "step": 2486 }, { "epoch": 0.54, "learning_rate": 9.294532058763198e-06, "loss": 0.2671, "step": 2487 }, { "epoch": 0.54, "learning_rate": 9.287559011226843e-06, "loss": 0.3106, "step": 2488 }, { "epoch": 0.54, "learning_rate": 9.28058631185339e-06, "loss": 0.3016, "step": 2489 }, { "epoch": 0.54, "learning_rate": 9.273613964050321e-06, "loss": 0.3326, "step": 2490 }, { "epoch": 0.54, "learning_rate": 9.266641971224963e-06, "loss": 0.2608, "step": 2491 }, { "epoch": 0.54, "learning_rate": 9.259670336784452e-06, "loss": 0.278, "step": 2492 }, { "epoch": 0.54, "learning_rate": 9.252699064135759e-06, "loss": 0.27, "step": 2493 }, { "epoch": 0.54, "learning_rate": 9.245728156685681e-06, "loss": 0.3441, "step": 2494 }, { "epoch": 0.54, "learning_rate": 9.238757617840824e-06, "loss": 0.2954, "step": 2495 }, { "epoch": 0.54, "learning_rate": 9.231787451007626e-06, "loss": 0.2572, "step": 2496 }, { "epoch": 0.54, "learning_rate": 9.224817659592342e-06, "loss": 0.3034, "step": 2497 }, { "epoch": 0.54, "learning_rate": 9.217848247001038e-06, "loss": 0.2912, "step": 2498 }, { "epoch": 0.54, "learning_rate": 9.210879216639592e-06, "loss": 0.33, "step": 2499 }, { "epoch": 0.54, "learning_rate": 9.203910571913713e-06, "loss": 0.3584, "step": 2500 }, { "epoch": 0.54, "learning_rate": 9.196942316228893e-06, "loss": 0.2743, "step": 2501 }, { "epoch": 0.54, "learning_rate": 9.189974452990463e-06, "loss": 0.2783, "step": 2502 }, { "epoch": 0.54, "learning_rate": 9.183006985603546e-06, "loss": 0.2758, "step": 2503 }, { "epoch": 0.54, "learning_rate": 9.176039917473071e-06, "loss": 0.3357, "step": 2504 }, { "epoch": 0.54, "learning_rate": 9.16907325200378e-06, "loss": 0.2853, "step": 2505 }, { "epoch": 0.54, "learning_rate": 9.162106992600207e-06, "loss": 0.3123, "step": 2506 }, { "epoch": 0.54, "learning_rate": 9.155141142666702e-06, "loss": 0.2882, "step": 2507 }, { "epoch": 0.54, "learning_rate": 9.148175705607407e-06, "loss": 0.3218, "step": 2508 }, { "epoch": 0.54, "learning_rate": 9.141210684826257e-06, "loss": 0.2823, "step": 2509 }, { "epoch": 0.54, "learning_rate": 9.134246083726992e-06, "loss": 0.3325, "step": 2510 }, { "epoch": 0.54, "learning_rate": 9.127281905713146e-06, "loss": 0.2894, "step": 2511 }, { "epoch": 0.54, "learning_rate": 9.120318154188039e-06, "loss": 0.283, "step": 2512 }, { "epoch": 0.54, "learning_rate": 9.11335483255479e-06, "loss": 0.2883, "step": 2513 }, { "epoch": 0.54, "learning_rate": 9.106391944216312e-06, "loss": 0.2876, "step": 2514 }, { "epoch": 0.54, "learning_rate": 9.099429492575294e-06, "loss": 0.3139, "step": 2515 }, { "epoch": 0.54, "learning_rate": 9.092467481034217e-06, "loss": 0.2564, "step": 2516 }, { "epoch": 0.54, "learning_rate": 9.08550591299535e-06, "loss": 0.2996, "step": 2517 }, { "epoch": 0.54, "learning_rate": 9.07854479186074e-06, "loss": 0.3329, "step": 2518 }, { "epoch": 0.54, "learning_rate": 9.071584121032224e-06, "loss": 0.2927, "step": 2519 }, { "epoch": 0.54, "learning_rate": 9.064623903911405e-06, "loss": 0.3183, "step": 2520 }, { "epoch": 0.54, "learning_rate": 9.057664143899677e-06, "loss": 0.3296, "step": 2521 }, { "epoch": 0.54, "learning_rate": 9.050704844398211e-06, "loss": 0.3134, "step": 2522 }, { "epoch": 0.54, "learning_rate": 9.043746008807939e-06, "loss": 0.3582, "step": 2523 }, { "epoch": 0.54, "learning_rate": 9.03678764052958e-06, "loss": 0.3271, "step": 2524 }, { "epoch": 0.54, "learning_rate": 9.029829742963622e-06, "loss": 0.3535, "step": 2525 }, { "epoch": 0.55, "learning_rate": 9.022872319510318e-06, "loss": 0.3303, "step": 2526 }, { "epoch": 0.55, "learning_rate": 9.015915373569694e-06, "loss": 0.2708, "step": 2527 }, { "epoch": 0.55, "learning_rate": 9.008958908541542e-06, "loss": 0.3273, "step": 2528 }, { "epoch": 0.55, "learning_rate": 9.002002927825411e-06, "loss": 0.3093, "step": 2529 }, { "epoch": 0.55, "learning_rate": 8.995047434820628e-06, "loss": 0.2736, "step": 2530 }, { "epoch": 0.55, "learning_rate": 8.988092432926275e-06, "loss": 0.2565, "step": 2531 }, { "epoch": 0.55, "learning_rate": 8.981137925541186e-06, "loss": 0.2647, "step": 2532 }, { "epoch": 0.55, "learning_rate": 8.974183916063967e-06, "loss": 0.3391, "step": 2533 }, { "epoch": 0.55, "learning_rate": 8.96723040789297e-06, "loss": 0.3041, "step": 2534 }, { "epoch": 0.55, "learning_rate": 8.960277404426303e-06, "loss": 0.2971, "step": 2535 }, { "epoch": 0.55, "learning_rate": 8.953324909061841e-06, "loss": 0.2958, "step": 2536 }, { "epoch": 0.55, "learning_rate": 8.946372925197186e-06, "loss": 0.299, "step": 2537 }, { "epoch": 0.55, "learning_rate": 8.939421456229715e-06, "loss": 0.255, "step": 2538 }, { "epoch": 0.55, "learning_rate": 8.93247050555654e-06, "loss": 0.3095, "step": 2539 }, { "epoch": 0.55, "learning_rate": 8.92552007657452e-06, "loss": 0.3292, "step": 2540 }, { "epoch": 0.55, "learning_rate": 8.91857017268026e-06, "loss": 0.3099, "step": 2541 }, { "epoch": 0.55, "learning_rate": 8.911620797270115e-06, "loss": 0.2972, "step": 2542 }, { "epoch": 0.55, "learning_rate": 8.904671953740171e-06, "loss": 0.2946, "step": 2543 }, { "epoch": 0.55, "learning_rate": 8.897723645486263e-06, "loss": 0.2877, "step": 2544 }, { "epoch": 0.55, "learning_rate": 8.890775875903958e-06, "loss": 0.2968, "step": 2545 }, { "epoch": 0.55, "learning_rate": 8.883828648388565e-06, "loss": 0.3028, "step": 2546 }, { "epoch": 0.55, "learning_rate": 8.876881966335128e-06, "loss": 0.2889, "step": 2547 }, { "epoch": 0.55, "learning_rate": 8.869935833138414e-06, "loss": 0.2542, "step": 2548 }, { "epoch": 0.55, "learning_rate": 8.862990252192936e-06, "loss": 0.2676, "step": 2549 }, { "epoch": 0.55, "learning_rate": 8.856045226892933e-06, "loss": 0.2768, "step": 2550 }, { "epoch": 0.55, "learning_rate": 8.849100760632366e-06, "loss": 0.3231, "step": 2551 }, { "epoch": 0.55, "learning_rate": 8.842156856804925e-06, "loss": 0.2999, "step": 2552 }, { "epoch": 0.55, "learning_rate": 8.835213518804037e-06, "loss": 0.2992, "step": 2553 }, { "epoch": 0.55, "learning_rate": 8.828270750022832e-06, "loss": 0.3218, "step": 2554 }, { "epoch": 0.55, "learning_rate": 8.821328553854178e-06, "loss": 0.2573, "step": 2555 }, { "epoch": 0.55, "learning_rate": 8.814386933690664e-06, "loss": 0.3182, "step": 2556 }, { "epoch": 0.55, "learning_rate": 8.80744589292458e-06, "loss": 0.3025, "step": 2557 }, { "epoch": 0.55, "learning_rate": 8.80050543494795e-06, "loss": 0.2819, "step": 2558 }, { "epoch": 0.55, "learning_rate": 8.793565563152509e-06, "loss": 0.2764, "step": 2559 }, { "epoch": 0.55, "learning_rate": 8.7866262809297e-06, "loss": 0.3299, "step": 2560 }, { "epoch": 0.55, "learning_rate": 8.779687591670687e-06, "loss": 0.2836, "step": 2561 }, { "epoch": 0.55, "learning_rate": 8.77274949876633e-06, "loss": 0.2785, "step": 2562 }, { "epoch": 0.55, "learning_rate": 8.765812005607219e-06, "loss": 0.2486, "step": 2563 }, { "epoch": 0.55, "learning_rate": 8.758875115583633e-06, "loss": 0.3179, "step": 2564 }, { "epoch": 0.55, "learning_rate": 8.751938832085556e-06, "loss": 0.2792, "step": 2565 }, { "epoch": 0.55, "learning_rate": 8.745003158502687e-06, "loss": 0.2677, "step": 2566 }, { "epoch": 0.55, "learning_rate": 8.738068098224427e-06, "loss": 0.2967, "step": 2567 }, { "epoch": 0.55, "learning_rate": 8.731133654639861e-06, "loss": 0.3297, "step": 2568 }, { "epoch": 0.55, "learning_rate": 8.724199831137789e-06, "loss": 0.2961, "step": 2569 }, { "epoch": 0.55, "learning_rate": 8.717266631106706e-06, "loss": 0.2523, "step": 2570 }, { "epoch": 0.55, "learning_rate": 8.710334057934789e-06, "loss": 0.3164, "step": 2571 }, { "epoch": 0.56, "learning_rate": 8.703402115009923e-06, "loss": 0.3032, "step": 2572 }, { "epoch": 0.56, "learning_rate": 8.696470805719684e-06, "loss": 0.2723, "step": 2573 }, { "epoch": 0.56, "learning_rate": 8.689540133451328e-06, "loss": 0.3395, "step": 2574 }, { "epoch": 0.56, "learning_rate": 8.682610101591813e-06, "loss": 0.2827, "step": 2575 }, { "epoch": 0.56, "learning_rate": 8.675680713527774e-06, "loss": 0.2813, "step": 2576 }, { "epoch": 0.56, "learning_rate": 8.668751972645531e-06, "loss": 0.2698, "step": 2577 }, { "epoch": 0.56, "learning_rate": 8.6618238823311e-06, "loss": 0.2899, "step": 2578 }, { "epoch": 0.56, "learning_rate": 8.654896445970162e-06, "loss": 0.3027, "step": 2579 }, { "epoch": 0.56, "learning_rate": 8.647969666948092e-06, "loss": 0.2682, "step": 2580 }, { "epoch": 0.56, "learning_rate": 8.641043548649938e-06, "loss": 0.3398, "step": 2581 }, { "epoch": 0.56, "learning_rate": 8.634118094460425e-06, "loss": 0.3157, "step": 2582 }, { "epoch": 0.56, "learning_rate": 8.627193307763953e-06, "loss": 0.3114, "step": 2583 }, { "epoch": 0.56, "learning_rate": 8.620269191944601e-06, "loss": 0.2732, "step": 2584 }, { "epoch": 0.56, "learning_rate": 8.613345750386111e-06, "loss": 0.2141, "step": 2585 }, { "epoch": 0.56, "learning_rate": 8.606422986471905e-06, "loss": 0.3127, "step": 2586 }, { "epoch": 0.56, "learning_rate": 8.59950090358507e-06, "loss": 0.2618, "step": 2587 }, { "epoch": 0.56, "learning_rate": 8.592579505108358e-06, "loss": 0.3035, "step": 2588 }, { "epoch": 0.56, "learning_rate": 8.585658794424188e-06, "loss": 0.3234, "step": 2589 }, { "epoch": 0.56, "learning_rate": 8.578738774914641e-06, "loss": 0.2944, "step": 2590 }, { "epoch": 0.56, "learning_rate": 8.571819449961464e-06, "loss": 0.3179, "step": 2591 }, { "epoch": 0.56, "learning_rate": 8.56490082294607e-06, "loss": 0.3117, "step": 2592 }, { "epoch": 0.56, "learning_rate": 8.557982897249513e-06, "loss": 0.3661, "step": 2593 }, { "epoch": 0.56, "learning_rate": 8.55106567625252e-06, "loss": 0.3114, "step": 2594 }, { "epoch": 0.56, "learning_rate": 8.544149163335473e-06, "loss": 0.2763, "step": 2595 }, { "epoch": 0.56, "learning_rate": 8.537233361878395e-06, "loss": 0.2764, "step": 2596 }, { "epoch": 0.56, "learning_rate": 8.530318275260974e-06, "loss": 0.3052, "step": 2597 }, { "epoch": 0.56, "learning_rate": 8.523403906862549e-06, "loss": 0.286, "step": 2598 }, { "epoch": 0.56, "learning_rate": 8.5164902600621e-06, "loss": 0.319, "step": 2599 }, { "epoch": 0.56, "learning_rate": 8.509577338238255e-06, "loss": 0.3173, "step": 2600 }, { "epoch": 0.56, "learning_rate": 8.5026651447693e-06, "loss": 0.2957, "step": 2601 }, { "epoch": 0.56, "learning_rate": 8.495753683033144e-06, "loss": 0.3437, "step": 2602 }, { "epoch": 0.56, "learning_rate": 8.488842956407361e-06, "loss": 0.229, "step": 2603 }, { "epoch": 0.56, "learning_rate": 8.48193296826915e-06, "loss": 0.2735, "step": 2604 }, { "epoch": 0.56, "learning_rate": 8.475023721995352e-06, "loss": 0.2847, "step": 2605 }, { "epoch": 0.56, "learning_rate": 8.468115220962455e-06, "loss": 0.317, "step": 2606 }, { "epoch": 0.56, "learning_rate": 8.461207468546565e-06, "loss": 0.3208, "step": 2607 }, { "epoch": 0.56, "learning_rate": 8.454300468123439e-06, "loss": 0.3176, "step": 2608 }, { "epoch": 0.56, "learning_rate": 8.447394223068465e-06, "loss": 0.2563, "step": 2609 }, { "epoch": 0.56, "learning_rate": 8.440488736756647e-06, "loss": 0.278, "step": 2610 }, { "epoch": 0.56, "learning_rate": 8.433584012562633e-06, "loss": 0.2762, "step": 2611 }, { "epoch": 0.56, "learning_rate": 8.426680053860698e-06, "loss": 0.268, "step": 2612 }, { "epoch": 0.56, "learning_rate": 8.41977686402473e-06, "loss": 0.3891, "step": 2613 }, { "epoch": 0.56, "learning_rate": 8.412874446428252e-06, "loss": 0.2853, "step": 2614 }, { "epoch": 0.56, "learning_rate": 8.405972804444414e-06, "loss": 0.2801, "step": 2615 }, { "epoch": 0.56, "learning_rate": 8.399071941445973e-06, "loss": 0.3131, "step": 2616 }, { "epoch": 0.56, "learning_rate": 8.39217186080532e-06, "loss": 0.3544, "step": 2617 }, { "epoch": 0.57, "learning_rate": 8.38527256589445e-06, "loss": 0.29, "step": 2618 }, { "epoch": 0.57, "learning_rate": 8.37837406008498e-06, "loss": 0.2665, "step": 2619 }, { "epoch": 0.57, "learning_rate": 8.37147634674815e-06, "loss": 0.3193, "step": 2620 }, { "epoch": 0.57, "learning_rate": 8.364579429254795e-06, "loss": 0.2629, "step": 2621 }, { "epoch": 0.57, "learning_rate": 8.357683310975378e-06, "loss": 0.3257, "step": 2622 }, { "epoch": 0.57, "learning_rate": 8.350787995279962e-06, "loss": 0.2858, "step": 2623 }, { "epoch": 0.57, "learning_rate": 8.343893485538217e-06, "loss": 0.2778, "step": 2624 }, { "epoch": 0.57, "learning_rate": 8.336999785119422e-06, "loss": 0.279, "step": 2625 }, { "epoch": 0.57, "learning_rate": 8.330106897392468e-06, "loss": 0.3047, "step": 2626 }, { "epoch": 0.57, "learning_rate": 8.323214825725832e-06, "loss": 0.2911, "step": 2627 }, { "epoch": 0.57, "learning_rate": 8.316323573487607e-06, "loss": 0.2659, "step": 2628 }, { "epoch": 0.57, "learning_rate": 8.30943314404548e-06, "loss": 0.2536, "step": 2629 }, { "epoch": 0.57, "learning_rate": 8.302543540766735e-06, "loss": 0.2599, "step": 2630 }, { "epoch": 0.57, "learning_rate": 8.295654767018254e-06, "loss": 0.2972, "step": 2631 }, { "epoch": 0.57, "learning_rate": 8.288766826166507e-06, "loss": 0.2719, "step": 2632 }, { "epoch": 0.57, "learning_rate": 8.28187972157757e-06, "loss": 0.2426, "step": 2633 }, { "epoch": 0.57, "learning_rate": 8.274993456617105e-06, "loss": 0.2588, "step": 2634 }, { "epoch": 0.57, "learning_rate": 8.268108034650356e-06, "loss": 0.267, "step": 2635 }, { "epoch": 0.57, "learning_rate": 8.261223459042161e-06, "loss": 0.2724, "step": 2636 }, { "epoch": 0.57, "learning_rate": 8.25433973315695e-06, "loss": 0.3098, "step": 2637 }, { "epoch": 0.57, "learning_rate": 8.247456860358725e-06, "loss": 0.2894, "step": 2638 }, { "epoch": 0.57, "learning_rate": 8.240574844011082e-06, "loss": 0.3022, "step": 2639 }, { "epoch": 0.57, "learning_rate": 8.233693687477198e-06, "loss": 0.2327, "step": 2640 }, { "epoch": 0.57, "learning_rate": 8.22681339411982e-06, "loss": 0.3542, "step": 2641 }, { "epoch": 0.57, "learning_rate": 8.219933967301281e-06, "loss": 0.2968, "step": 2642 }, { "epoch": 0.57, "learning_rate": 8.213055410383495e-06, "loss": 0.2776, "step": 2643 }, { "epoch": 0.57, "learning_rate": 8.206177726727936e-06, "loss": 0.2988, "step": 2644 }, { "epoch": 0.57, "learning_rate": 8.19930091969567e-06, "loss": 0.306, "step": 2645 }, { "epoch": 0.57, "learning_rate": 8.192424992647315e-06, "loss": 0.3034, "step": 2646 }, { "epoch": 0.57, "learning_rate": 8.185549948943076e-06, "loss": 0.3055, "step": 2647 }, { "epoch": 0.57, "learning_rate": 8.178675791942719e-06, "loss": 0.3402, "step": 2648 }, { "epoch": 0.57, "learning_rate": 8.17180252500557e-06, "loss": 0.3117, "step": 2649 }, { "epoch": 0.57, "learning_rate": 8.164930151490532e-06, "loss": 0.3231, "step": 2650 }, { "epoch": 0.57, "learning_rate": 8.15805867475607e-06, "loss": 0.3102, "step": 2651 }, { "epoch": 0.57, "learning_rate": 8.1511880981602e-06, "loss": 0.2904, "step": 2652 }, { "epoch": 0.57, "learning_rate": 8.144318425060508e-06, "loss": 0.2832, "step": 2653 }, { "epoch": 0.57, "learning_rate": 8.13744965881414e-06, "loss": 0.2914, "step": 2654 }, { "epoch": 0.57, "learning_rate": 8.130581802777786e-06, "loss": 0.2476, "step": 2655 }, { "epoch": 0.57, "learning_rate": 8.123714860307703e-06, "loss": 0.2883, "step": 2656 }, { "epoch": 0.57, "learning_rate": 8.116848834759704e-06, "loss": 0.2879, "step": 2657 }, { "epoch": 0.57, "learning_rate": 8.109983729489139e-06, "loss": 0.3239, "step": 2658 }, { "epoch": 0.57, "learning_rate": 8.103119547850924e-06, "loss": 0.3078, "step": 2659 }, { "epoch": 0.57, "learning_rate": 8.096256293199513e-06, "loss": 0.3477, "step": 2660 }, { "epoch": 0.57, "learning_rate": 8.089393968888913e-06, "loss": 0.3016, "step": 2661 }, { "epoch": 0.57, "learning_rate": 8.082532578272675e-06, "loss": 0.2556, "step": 2662 }, { "epoch": 0.57, "learning_rate": 8.07567212470389e-06, "loss": 0.3253, "step": 2663 }, { "epoch": 0.57, "learning_rate": 8.068812611535197e-06, "loss": 0.2876, "step": 2664 }, { "epoch": 0.58, "learning_rate": 8.061954042118772e-06, "loss": 0.3067, "step": 2665 }, { "epoch": 0.58, "learning_rate": 8.05509641980633e-06, "loss": 0.2618, "step": 2666 }, { "epoch": 0.58, "learning_rate": 8.04823974794912e-06, "loss": 0.3109, "step": 2667 }, { "epoch": 0.58, "learning_rate": 8.041384029897939e-06, "loss": 0.2862, "step": 2668 }, { "epoch": 0.58, "learning_rate": 8.0345292690031e-06, "loss": 0.2648, "step": 2669 }, { "epoch": 0.58, "learning_rate": 8.027675468614462e-06, "loss": 0.2945, "step": 2670 }, { "epoch": 0.58, "learning_rate": 8.02082263208141e-06, "loss": 0.3144, "step": 2671 }, { "epoch": 0.58, "learning_rate": 8.013970762752853e-06, "loss": 0.2996, "step": 2672 }, { "epoch": 0.58, "learning_rate": 8.00711986397724e-06, "loss": 0.3415, "step": 2673 }, { "epoch": 0.58, "learning_rate": 8.000269939102527e-06, "loss": 0.2627, "step": 2674 }, { "epoch": 0.58, "learning_rate": 7.993420991476212e-06, "loss": 0.2835, "step": 2675 }, { "epoch": 0.58, "learning_rate": 7.986573024445312e-06, "loss": 0.3226, "step": 2676 }, { "epoch": 0.58, "learning_rate": 7.979726041356353e-06, "loss": 0.2678, "step": 2677 }, { "epoch": 0.58, "learning_rate": 7.972880045555392e-06, "loss": 0.2759, "step": 2678 }, { "epoch": 0.58, "learning_rate": 7.966035040388e-06, "loss": 0.2647, "step": 2679 }, { "epoch": 0.58, "learning_rate": 7.95919102919926e-06, "loss": 0.3017, "step": 2680 }, { "epoch": 0.58, "learning_rate": 7.952348015333778e-06, "loss": 0.2592, "step": 2681 }, { "epoch": 0.58, "learning_rate": 7.945506002135668e-06, "loss": 0.2329, "step": 2682 }, { "epoch": 0.58, "learning_rate": 7.938664992948549e-06, "loss": 0.323, "step": 2683 }, { "epoch": 0.58, "learning_rate": 7.931824991115557e-06, "loss": 0.3175, "step": 2684 }, { "epoch": 0.58, "learning_rate": 7.924985999979337e-06, "loss": 0.2573, "step": 2685 }, { "epoch": 0.58, "learning_rate": 7.918148022882032e-06, "loss": 0.3113, "step": 2686 }, { "epoch": 0.58, "learning_rate": 7.911311063165298e-06, "loss": 0.2822, "step": 2687 }, { "epoch": 0.58, "learning_rate": 7.904475124170287e-06, "loss": 0.2896, "step": 2688 }, { "epoch": 0.58, "learning_rate": 7.897640209237656e-06, "loss": 0.292, "step": 2689 }, { "epoch": 0.58, "learning_rate": 7.890806321707562e-06, "loss": 0.2914, "step": 2690 }, { "epoch": 0.58, "learning_rate": 7.883973464919655e-06, "loss": 0.3396, "step": 2691 }, { "epoch": 0.58, "learning_rate": 7.877141642213086e-06, "loss": 0.2915, "step": 2692 }, { "epoch": 0.58, "learning_rate": 7.870310856926504e-06, "loss": 0.3006, "step": 2693 }, { "epoch": 0.58, "learning_rate": 7.863481112398041e-06, "loss": 0.255, "step": 2694 }, { "epoch": 0.58, "learning_rate": 7.85665241196533e-06, "loss": 0.2285, "step": 2695 }, { "epoch": 0.58, "learning_rate": 7.84982475896549e-06, "loss": 0.3304, "step": 2696 }, { "epoch": 0.58, "learning_rate": 7.84299815673512e-06, "loss": 0.333, "step": 2697 }, { "epoch": 0.58, "learning_rate": 7.836172608610321e-06, "loss": 0.3408, "step": 2698 }, { "epoch": 0.58, "learning_rate": 7.829348117926672e-06, "loss": 0.3305, "step": 2699 }, { "epoch": 0.58, "learning_rate": 7.822524688019227e-06, "loss": 0.3149, "step": 2700 }, { "epoch": 0.58, "learning_rate": 7.815702322222539e-06, "loss": 0.3222, "step": 2701 }, { "epoch": 0.58, "learning_rate": 7.808881023870622e-06, "loss": 0.3078, "step": 2702 }, { "epoch": 0.58, "learning_rate": 7.80206079629698e-06, "loss": 0.3159, "step": 2703 }, { "epoch": 0.58, "learning_rate": 7.795241642834596e-06, "loss": 0.2626, "step": 2704 }, { "epoch": 0.58, "learning_rate": 7.788423566815915e-06, "loss": 0.3537, "step": 2705 }, { "epoch": 0.58, "learning_rate": 7.78160657157287e-06, "loss": 0.272, "step": 2706 }, { "epoch": 0.58, "learning_rate": 7.774790660436857e-06, "loss": 0.2987, "step": 2707 }, { "epoch": 0.58, "learning_rate": 7.767975836738747e-06, "loss": 0.2803, "step": 2708 }, { "epoch": 0.58, "learning_rate": 7.76116210380887e-06, "loss": 0.3076, "step": 2709 }, { "epoch": 0.58, "learning_rate": 7.754349464977041e-06, "loss": 0.3004, "step": 2710 }, { "epoch": 0.59, "learning_rate": 7.74753792357252e-06, "loss": 0.2831, "step": 2711 }, { "epoch": 0.59, "learning_rate": 7.740727482924048e-06, "loss": 0.2569, "step": 2712 }, { "epoch": 0.59, "learning_rate": 7.733918146359816e-06, "loss": 0.3364, "step": 2713 }, { "epoch": 0.59, "learning_rate": 7.727109917207478e-06, "loss": 0.269, "step": 2714 }, { "epoch": 0.59, "learning_rate": 7.720302798794153e-06, "loss": 0.2839, "step": 2715 }, { "epoch": 0.59, "learning_rate": 7.713496794446408e-06, "loss": 0.2909, "step": 2716 }, { "epoch": 0.59, "learning_rate": 7.706691907490272e-06, "loss": 0.2956, "step": 2717 }, { "epoch": 0.59, "learning_rate": 7.699888141251232e-06, "loss": 0.2991, "step": 2718 }, { "epoch": 0.59, "learning_rate": 7.693085499054216e-06, "loss": 0.3315, "step": 2719 }, { "epoch": 0.59, "learning_rate": 7.686283984223606e-06, "loss": 0.241, "step": 2720 }, { "epoch": 0.59, "learning_rate": 7.679483600083243e-06, "loss": 0.2676, "step": 2721 }, { "epoch": 0.59, "learning_rate": 7.672684349956399e-06, "loss": 0.2765, "step": 2722 }, { "epoch": 0.59, "learning_rate": 7.665886237165804e-06, "loss": 0.3268, "step": 2723 }, { "epoch": 0.59, "learning_rate": 7.659089265033635e-06, "loss": 0.3079, "step": 2724 }, { "epoch": 0.59, "learning_rate": 7.652293436881497e-06, "loss": 0.2825, "step": 2725 }, { "epoch": 0.59, "learning_rate": 7.645498756030451e-06, "loss": 0.2765, "step": 2726 }, { "epoch": 0.59, "learning_rate": 7.638705225800981e-06, "loss": 0.2412, "step": 2727 }, { "epoch": 0.59, "learning_rate": 7.631912849513026e-06, "loss": 0.323, "step": 2728 }, { "epoch": 0.59, "learning_rate": 7.6251216304859555e-06, "loss": 0.3817, "step": 2729 }, { "epoch": 0.59, "learning_rate": 7.618331572038563e-06, "loss": 0.2786, "step": 2730 }, { "epoch": 0.59, "learning_rate": 7.61154267748909e-06, "loss": 0.3277, "step": 2731 }, { "epoch": 0.59, "learning_rate": 7.604754950155203e-06, "loss": 0.2223, "step": 2732 }, { "epoch": 0.59, "learning_rate": 7.597968393353992e-06, "loss": 0.3005, "step": 2733 }, { "epoch": 0.59, "learning_rate": 7.591183010401981e-06, "loss": 0.2508, "step": 2734 }, { "epoch": 0.59, "learning_rate": 7.584398804615127e-06, "loss": 0.2507, "step": 2735 }, { "epoch": 0.59, "learning_rate": 7.577615779308797e-06, "loss": 0.2733, "step": 2736 }, { "epoch": 0.59, "learning_rate": 7.5708339377977926e-06, "loss": 0.2978, "step": 2737 }, { "epoch": 0.59, "learning_rate": 7.564053283396335e-06, "loss": 0.2532, "step": 2738 }, { "epoch": 0.59, "learning_rate": 7.557273819418056e-06, "loss": 0.3406, "step": 2739 }, { "epoch": 0.59, "learning_rate": 7.55049554917602e-06, "loss": 0.2409, "step": 2740 }, { "epoch": 0.59, "learning_rate": 7.543718475982695e-06, "loss": 0.2777, "step": 2741 }, { "epoch": 0.59, "learning_rate": 7.536942603149975e-06, "loss": 0.298, "step": 2742 }, { "epoch": 0.59, "learning_rate": 7.530167933989161e-06, "loss": 0.2857, "step": 2743 }, { "epoch": 0.59, "learning_rate": 7.523394471810965e-06, "loss": 0.3085, "step": 2744 }, { "epoch": 0.59, "learning_rate": 7.516622219925513e-06, "loss": 0.2664, "step": 2745 }, { "epoch": 0.59, "learning_rate": 7.50985118164234e-06, "loss": 0.259, "step": 2746 }, { "epoch": 0.59, "learning_rate": 7.503081360270379e-06, "loss": 0.3268, "step": 2747 }, { "epoch": 0.59, "learning_rate": 7.496312759117983e-06, "loss": 0.3663, "step": 2748 }, { "epoch": 0.59, "learning_rate": 7.489545381492896e-06, "loss": 0.2954, "step": 2749 }, { "epoch": 0.59, "learning_rate": 7.482779230702272e-06, "loss": 0.2849, "step": 2750 }, { "epoch": 0.59, "learning_rate": 7.476014310052657e-06, "loss": 0.2709, "step": 2751 }, { "epoch": 0.59, "learning_rate": 7.469250622850008e-06, "loss": 0.3174, "step": 2752 }, { "epoch": 0.59, "learning_rate": 7.4624881723996666e-06, "loss": 0.3091, "step": 2753 }, { "epoch": 0.59, "learning_rate": 7.45572696200638e-06, "loss": 0.2491, "step": 2754 }, { "epoch": 0.59, "learning_rate": 7.448966994974282e-06, "loss": 0.3515, "step": 2755 }, { "epoch": 0.59, "learning_rate": 7.442208274606902e-06, "loss": 0.2772, "step": 2756 }, { "epoch": 0.6, "learning_rate": 7.435450804207165e-06, "loss": 0.3429, "step": 2757 }, { "epoch": 0.6, "learning_rate": 7.428694587077372e-06, "loss": 0.3079, "step": 2758 }, { "epoch": 0.6, "learning_rate": 7.421939626519226e-06, "loss": 0.2814, "step": 2759 }, { "epoch": 0.6, "learning_rate": 7.41518592583381e-06, "loss": 0.2504, "step": 2760 }, { "epoch": 0.6, "learning_rate": 7.408433488321588e-06, "loss": 0.2532, "step": 2761 }, { "epoch": 0.6, "learning_rate": 7.401682317282408e-06, "loss": 0.2754, "step": 2762 }, { "epoch": 0.6, "learning_rate": 7.39493241601551e-06, "loss": 0.2963, "step": 2763 }, { "epoch": 0.6, "learning_rate": 7.388183787819492e-06, "loss": 0.2637, "step": 2764 }, { "epoch": 0.6, "learning_rate": 7.3814364359923485e-06, "loss": 0.2714, "step": 2765 }, { "epoch": 0.6, "learning_rate": 7.374690363831448e-06, "loss": 0.2632, "step": 2766 }, { "epoch": 0.6, "learning_rate": 7.367945574633524e-06, "loss": 0.3273, "step": 2767 }, { "epoch": 0.6, "learning_rate": 7.361202071694691e-06, "loss": 0.2442, "step": 2768 }, { "epoch": 0.6, "learning_rate": 7.354459858310427e-06, "loss": 0.2994, "step": 2769 }, { "epoch": 0.6, "learning_rate": 7.3477189377755895e-06, "loss": 0.3229, "step": 2770 }, { "epoch": 0.6, "learning_rate": 7.340979313384404e-06, "loss": 0.2468, "step": 2771 }, { "epoch": 0.6, "learning_rate": 7.334240988430451e-06, "loss": 0.2326, "step": 2772 }, { "epoch": 0.6, "learning_rate": 7.32750396620669e-06, "loss": 0.2945, "step": 2773 }, { "epoch": 0.6, "learning_rate": 7.320768250005436e-06, "loss": 0.2392, "step": 2774 }, { "epoch": 0.6, "learning_rate": 7.3140338431183625e-06, "loss": 0.3484, "step": 2775 }, { "epoch": 0.6, "learning_rate": 7.30730074883651e-06, "loss": 0.2677, "step": 2776 }, { "epoch": 0.6, "learning_rate": 7.300568970450283e-06, "loss": 0.3242, "step": 2777 }, { "epoch": 0.6, "learning_rate": 7.293838511249422e-06, "loss": 0.2757, "step": 2778 }, { "epoch": 0.6, "learning_rate": 7.287109374523048e-06, "loss": 0.2901, "step": 2779 }, { "epoch": 0.6, "learning_rate": 7.28038156355962e-06, "loss": 0.2581, "step": 2780 }, { "epoch": 0.6, "learning_rate": 7.273655081646949e-06, "loss": 0.2329, "step": 2781 }, { "epoch": 0.6, "learning_rate": 7.266929932072207e-06, "loss": 0.2707, "step": 2782 }, { "epoch": 0.6, "learning_rate": 7.2602061181219e-06, "loss": 0.2864, "step": 2783 }, { "epoch": 0.6, "learning_rate": 7.253483643081896e-06, "loss": 0.3373, "step": 2784 }, { "epoch": 0.6, "learning_rate": 7.246762510237404e-06, "loss": 0.3126, "step": 2785 }, { "epoch": 0.6, "learning_rate": 7.240042722872971e-06, "loss": 0.3382, "step": 2786 }, { "epoch": 0.6, "learning_rate": 7.233324284272489e-06, "loss": 0.3307, "step": 2787 }, { "epoch": 0.6, "learning_rate": 7.226607197719202e-06, "loss": 0.3201, "step": 2788 }, { "epoch": 0.6, "learning_rate": 7.219891466495675e-06, "loss": 0.2812, "step": 2789 }, { "epoch": 0.6, "learning_rate": 7.2131770938838235e-06, "loss": 0.2471, "step": 2790 }, { "epoch": 0.6, "learning_rate": 7.206464083164894e-06, "loss": 0.3936, "step": 2791 }, { "epoch": 0.6, "learning_rate": 7.199752437619473e-06, "loss": 0.3184, "step": 2792 }, { "epoch": 0.6, "learning_rate": 7.193042160527468e-06, "loss": 0.3306, "step": 2793 }, { "epoch": 0.6, "learning_rate": 7.1863332551681345e-06, "loss": 0.2781, "step": 2794 }, { "epoch": 0.6, "learning_rate": 7.179625724820038e-06, "loss": 0.2731, "step": 2795 }, { "epoch": 0.6, "learning_rate": 7.1729195727610945e-06, "loss": 0.2833, "step": 2796 }, { "epoch": 0.6, "learning_rate": 7.166214802268525e-06, "loss": 0.3454, "step": 2797 }, { "epoch": 0.6, "learning_rate": 7.159511416618887e-06, "loss": 0.2484, "step": 2798 }, { "epoch": 0.6, "learning_rate": 7.1528094190880625e-06, "loss": 0.2932, "step": 2799 }, { "epoch": 0.6, "learning_rate": 7.146108812951247e-06, "loss": 0.2612, "step": 2800 }, { "epoch": 0.6, "learning_rate": 7.139409601482963e-06, "loss": 0.2764, "step": 2801 }, { "epoch": 0.6, "learning_rate": 7.1327117879570515e-06, "loss": 0.3184, "step": 2802 }, { "epoch": 0.6, "learning_rate": 7.126015375646665e-06, "loss": 0.2898, "step": 2803 }, { "epoch": 0.61, "learning_rate": 7.119320367824274e-06, "loss": 0.2943, "step": 2804 }, { "epoch": 0.61, "learning_rate": 7.112626767761668e-06, "loss": 0.2921, "step": 2805 }, { "epoch": 0.61, "learning_rate": 7.105934578729934e-06, "loss": 0.2853, "step": 2806 }, { "epoch": 0.61, "learning_rate": 7.099243803999485e-06, "loss": 0.278, "step": 2807 }, { "epoch": 0.61, "learning_rate": 7.092554446840041e-06, "loss": 0.2902, "step": 2808 }, { "epoch": 0.61, "learning_rate": 7.085866510520618e-06, "loss": 0.3333, "step": 2809 }, { "epoch": 0.61, "learning_rate": 7.079179998309549e-06, "loss": 0.2801, "step": 2810 }, { "epoch": 0.61, "learning_rate": 7.072494913474459e-06, "loss": 0.2426, "step": 2811 }, { "epoch": 0.61, "learning_rate": 7.06581125928229e-06, "loss": 0.2752, "step": 2812 }, { "epoch": 0.61, "learning_rate": 7.059129038999282e-06, "loss": 0.3065, "step": 2813 }, { "epoch": 0.61, "learning_rate": 7.052448255890958e-06, "loss": 0.3242, "step": 2814 }, { "epoch": 0.61, "learning_rate": 7.045768913222162e-06, "loss": 0.3389, "step": 2815 }, { "epoch": 0.61, "learning_rate": 7.03909101425702e-06, "loss": 0.2514, "step": 2816 }, { "epoch": 0.61, "learning_rate": 7.032414562258951e-06, "loss": 0.2716, "step": 2817 }, { "epoch": 0.61, "learning_rate": 7.025739560490674e-06, "loss": 0.3368, "step": 2818 }, { "epoch": 0.61, "learning_rate": 7.019066012214204e-06, "loss": 0.2817, "step": 2819 }, { "epoch": 0.61, "learning_rate": 7.012393920690826e-06, "loss": 0.3071, "step": 2820 }, { "epoch": 0.61, "learning_rate": 7.005723289181132e-06, "loss": 0.3089, "step": 2821 }, { "epoch": 0.61, "learning_rate": 6.999054120944997e-06, "loss": 0.3202, "step": 2822 }, { "epoch": 0.61, "learning_rate": 6.9923864192415705e-06, "loss": 0.3067, "step": 2823 }, { "epoch": 0.61, "learning_rate": 6.985720187329299e-06, "loss": 0.2765, "step": 2824 }, { "epoch": 0.61, "learning_rate": 6.979055428465898e-06, "loss": 0.291, "step": 2825 }, { "epoch": 0.61, "learning_rate": 6.972392145908375e-06, "loss": 0.2872, "step": 2826 }, { "epoch": 0.61, "learning_rate": 6.965730342913011e-06, "loss": 0.3837, "step": 2827 }, { "epoch": 0.61, "learning_rate": 6.959070022735359e-06, "loss": 0.2939, "step": 2828 }, { "epoch": 0.61, "learning_rate": 6.9524111886302536e-06, "loss": 0.277, "step": 2829 }, { "epoch": 0.61, "learning_rate": 6.945753843851806e-06, "loss": 0.2961, "step": 2830 }, { "epoch": 0.61, "learning_rate": 6.939097991653387e-06, "loss": 0.3011, "step": 2831 }, { "epoch": 0.61, "learning_rate": 6.932443635287652e-06, "loss": 0.3148, "step": 2832 }, { "epoch": 0.61, "learning_rate": 6.925790778006518e-06, "loss": 0.265, "step": 2833 }, { "epoch": 0.61, "learning_rate": 6.919139423061172e-06, "loss": 0.2545, "step": 2834 }, { "epoch": 0.61, "learning_rate": 6.912489573702061e-06, "loss": 0.336, "step": 2835 }, { "epoch": 0.61, "learning_rate": 6.9058412331789094e-06, "loss": 0.2897, "step": 2836 }, { "epoch": 0.61, "learning_rate": 6.899194404740686e-06, "loss": 0.3963, "step": 2837 }, { "epoch": 0.61, "learning_rate": 6.892549091635639e-06, "loss": 0.2976, "step": 2838 }, { "epoch": 0.61, "learning_rate": 6.885905297111261e-06, "loss": 0.2729, "step": 2839 }, { "epoch": 0.61, "learning_rate": 6.87926302441431e-06, "loss": 0.3389, "step": 2840 }, { "epoch": 0.61, "learning_rate": 6.872622276790804e-06, "loss": 0.291, "step": 2841 }, { "epoch": 0.61, "learning_rate": 6.865983057486005e-06, "loss": 0.3476, "step": 2842 }, { "epoch": 0.61, "learning_rate": 6.859345369744436e-06, "loss": 0.3286, "step": 2843 }, { "epoch": 0.61, "learning_rate": 6.852709216809876e-06, "loss": 0.3047, "step": 2844 }, { "epoch": 0.61, "learning_rate": 6.846074601925341e-06, "loss": 0.3536, "step": 2845 }, { "epoch": 0.61, "learning_rate": 6.839441528333099e-06, "loss": 0.3361, "step": 2846 }, { "epoch": 0.61, "learning_rate": 6.832809999274678e-06, "loss": 0.3987, "step": 2847 }, { "epoch": 0.61, "learning_rate": 6.8261800179908284e-06, "loss": 0.3442, "step": 2848 }, { "epoch": 0.61, "learning_rate": 6.819551587721567e-06, "loss": 0.306, "step": 2849 }, { "epoch": 0.62, "learning_rate": 6.81292471170614e-06, "loss": 0.2257, "step": 2850 }, { "epoch": 0.62, "learning_rate": 6.806299393183034e-06, "loss": 0.3042, "step": 2851 }, { "epoch": 0.62, "learning_rate": 6.799675635389981e-06, "loss": 0.372, "step": 2852 }, { "epoch": 0.62, "learning_rate": 6.7930534415639395e-06, "loss": 0.2649, "step": 2853 }, { "epoch": 0.62, "learning_rate": 6.786432814941113e-06, "loss": 0.3021, "step": 2854 }, { "epoch": 0.62, "learning_rate": 6.779813758756943e-06, "loss": 0.2588, "step": 2855 }, { "epoch": 0.62, "learning_rate": 6.773196276246088e-06, "loss": 0.2947, "step": 2856 }, { "epoch": 0.62, "learning_rate": 6.766580370642453e-06, "loss": 0.2645, "step": 2857 }, { "epoch": 0.62, "learning_rate": 6.759966045179165e-06, "loss": 0.2589, "step": 2858 }, { "epoch": 0.62, "learning_rate": 6.753353303088574e-06, "loss": 0.3302, "step": 2859 }, { "epoch": 0.62, "learning_rate": 6.7467421476022665e-06, "loss": 0.2886, "step": 2860 }, { "epoch": 0.62, "learning_rate": 6.740132581951053e-06, "loss": 0.2974, "step": 2861 }, { "epoch": 0.62, "learning_rate": 6.733524609364957e-06, "loss": 0.2671, "step": 2862 }, { "epoch": 0.62, "learning_rate": 6.7269182330732305e-06, "loss": 0.3254, "step": 2863 }, { "epoch": 0.62, "learning_rate": 6.72031345630435e-06, "loss": 0.3404, "step": 2864 }, { "epoch": 0.62, "learning_rate": 6.713710282285997e-06, "loss": 0.2891, "step": 2865 }, { "epoch": 0.62, "learning_rate": 6.707108714245083e-06, "loss": 0.2865, "step": 2866 }, { "epoch": 0.62, "learning_rate": 6.700508755407726e-06, "loss": 0.3536, "step": 2867 }, { "epoch": 0.62, "learning_rate": 6.6939104089992634e-06, "loss": 0.3438, "step": 2868 }, { "epoch": 0.62, "learning_rate": 6.687313678244243e-06, "loss": 0.3356, "step": 2869 }, { "epoch": 0.62, "learning_rate": 6.680718566366419e-06, "loss": 0.2758, "step": 2870 }, { "epoch": 0.62, "learning_rate": 6.6741250765887574e-06, "loss": 0.3015, "step": 2871 }, { "epoch": 0.62, "learning_rate": 6.667533212133437e-06, "loss": 0.2993, "step": 2872 }, { "epoch": 0.62, "learning_rate": 6.660942976221831e-06, "loss": 0.3049, "step": 2873 }, { "epoch": 0.62, "learning_rate": 6.6543543720745265e-06, "loss": 0.2884, "step": 2874 }, { "epoch": 0.62, "learning_rate": 6.647767402911311e-06, "loss": 0.2985, "step": 2875 }, { "epoch": 0.62, "learning_rate": 6.6411820719511645e-06, "loss": 0.2818, "step": 2876 }, { "epoch": 0.62, "learning_rate": 6.634598382412278e-06, "loss": 0.2852, "step": 2877 }, { "epoch": 0.62, "learning_rate": 6.62801633751204e-06, "loss": 0.2556, "step": 2878 }, { "epoch": 0.62, "learning_rate": 6.621435940467021e-06, "loss": 0.3077, "step": 2879 }, { "epoch": 0.62, "learning_rate": 6.614857194493007e-06, "loss": 0.2847, "step": 2880 }, { "epoch": 0.62, "learning_rate": 6.608280102804961e-06, "loss": 0.329, "step": 2881 }, { "epoch": 0.62, "learning_rate": 6.601704668617039e-06, "loss": 0.3122, "step": 2882 }, { "epoch": 0.62, "learning_rate": 6.595130895142601e-06, "loss": 0.2604, "step": 2883 }, { "epoch": 0.62, "learning_rate": 6.588558785594175e-06, "loss": 0.2737, "step": 2884 }, { "epoch": 0.62, "learning_rate": 6.58198834318349e-06, "loss": 0.2827, "step": 2885 }, { "epoch": 0.62, "learning_rate": 6.575419571121462e-06, "loss": 0.3058, "step": 2886 }, { "epoch": 0.62, "learning_rate": 6.568852472618179e-06, "loss": 0.3015, "step": 2887 }, { "epoch": 0.62, "learning_rate": 6.562287050882918e-06, "loss": 0.3442, "step": 2888 }, { "epoch": 0.62, "learning_rate": 6.555723309124139e-06, "loss": 0.2813, "step": 2889 }, { "epoch": 0.62, "learning_rate": 6.549161250549474e-06, "loss": 0.3426, "step": 2890 }, { "epoch": 0.62, "learning_rate": 6.542600878365737e-06, "loss": 0.3433, "step": 2891 }, { "epoch": 0.62, "learning_rate": 6.5360421957789235e-06, "loss": 0.3611, "step": 2892 }, { "epoch": 0.62, "learning_rate": 6.529485205994191e-06, "loss": 0.3092, "step": 2893 }, { "epoch": 0.62, "learning_rate": 6.522929912215878e-06, "loss": 0.3094, "step": 2894 }, { "epoch": 0.62, "learning_rate": 6.5163763176474884e-06, "loss": 0.2894, "step": 2895 }, { "epoch": 0.63, "learning_rate": 6.509824425491701e-06, "loss": 0.2827, "step": 2896 }, { "epoch": 0.63, "learning_rate": 6.5032742389503676e-06, "loss": 0.3095, "step": 2897 }, { "epoch": 0.63, "learning_rate": 6.49672576122449e-06, "loss": 0.3021, "step": 2898 }, { "epoch": 0.63, "learning_rate": 6.490178995514252e-06, "loss": 0.3247, "step": 2899 }, { "epoch": 0.63, "learning_rate": 6.483633945018991e-06, "loss": 0.2954, "step": 2900 }, { "epoch": 0.63, "learning_rate": 6.477090612937207e-06, "loss": 0.2931, "step": 2901 }, { "epoch": 0.63, "learning_rate": 6.470549002466562e-06, "loss": 0.2504, "step": 2902 }, { "epoch": 0.63, "learning_rate": 6.464009116803883e-06, "loss": 0.3111, "step": 2903 }, { "epoch": 0.63, "learning_rate": 6.45747095914514e-06, "loss": 0.2895, "step": 2904 }, { "epoch": 0.63, "learning_rate": 6.45093453268547e-06, "loss": 0.2914, "step": 2905 }, { "epoch": 0.63, "learning_rate": 6.444399840619162e-06, "loss": 0.3025, "step": 2906 }, { "epoch": 0.63, "learning_rate": 6.43786688613965e-06, "loss": 0.2846, "step": 2907 }, { "epoch": 0.63, "learning_rate": 6.431335672439532e-06, "loss": 0.3146, "step": 2908 }, { "epoch": 0.63, "learning_rate": 6.424806202710539e-06, "loss": 0.2585, "step": 2909 }, { "epoch": 0.63, "learning_rate": 6.418278480143565e-06, "loss": 0.2792, "step": 2910 }, { "epoch": 0.63, "learning_rate": 6.411752507928643e-06, "loss": 0.2655, "step": 2911 }, { "epoch": 0.63, "learning_rate": 6.40522828925495e-06, "loss": 0.3308, "step": 2912 }, { "epoch": 0.63, "learning_rate": 6.398705827310805e-06, "loss": 0.314, "step": 2913 }, { "epoch": 0.63, "learning_rate": 6.392185125283679e-06, "loss": 0.3117, "step": 2914 }, { "epoch": 0.63, "learning_rate": 6.385666186360166e-06, "loss": 0.3415, "step": 2915 }, { "epoch": 0.63, "learning_rate": 6.379149013726012e-06, "loss": 0.2418, "step": 2916 }, { "epoch": 0.63, "learning_rate": 6.372633610566097e-06, "loss": 0.271, "step": 2917 }, { "epoch": 0.63, "learning_rate": 6.366119980064427e-06, "loss": 0.2542, "step": 2918 }, { "epoch": 0.63, "learning_rate": 6.359608125404154e-06, "loss": 0.2538, "step": 2919 }, { "epoch": 0.63, "learning_rate": 6.353098049767561e-06, "loss": 0.3271, "step": 2920 }, { "epoch": 0.63, "learning_rate": 6.34658975633605e-06, "loss": 0.2442, "step": 2921 }, { "epoch": 0.63, "learning_rate": 6.340083248290167e-06, "loss": 0.314, "step": 2922 }, { "epoch": 0.63, "learning_rate": 6.3335785288095745e-06, "loss": 0.3351, "step": 2923 }, { "epoch": 0.63, "learning_rate": 6.327075601073062e-06, "loss": 0.2618, "step": 2924 }, { "epoch": 0.63, "learning_rate": 6.3205744682585545e-06, "loss": 0.2793, "step": 2925 }, { "epoch": 0.63, "learning_rate": 6.3140751335430825e-06, "loss": 0.3024, "step": 2926 }, { "epoch": 0.63, "learning_rate": 6.3075776001028125e-06, "loss": 0.2688, "step": 2927 }, { "epoch": 0.63, "learning_rate": 6.301081871113027e-06, "loss": 0.3077, "step": 2928 }, { "epoch": 0.63, "learning_rate": 6.294587949748121e-06, "loss": 0.3028, "step": 2929 }, { "epoch": 0.63, "learning_rate": 6.288095839181607e-06, "loss": 0.2343, "step": 2930 }, { "epoch": 0.63, "learning_rate": 6.281605542586124e-06, "loss": 0.3292, "step": 2931 }, { "epoch": 0.63, "learning_rate": 6.275117063133409e-06, "loss": 0.3039, "step": 2932 }, { "epoch": 0.63, "learning_rate": 6.268630403994318e-06, "loss": 0.3112, "step": 2933 }, { "epoch": 0.63, "learning_rate": 6.262145568338826e-06, "loss": 0.2513, "step": 2934 }, { "epoch": 0.63, "learning_rate": 6.255662559335999e-06, "loss": 0.3044, "step": 2935 }, { "epoch": 0.63, "learning_rate": 6.2491813801540245e-06, "loss": 0.3033, "step": 2936 }, { "epoch": 0.63, "learning_rate": 6.242702033960186e-06, "loss": 0.343, "step": 2937 }, { "epoch": 0.63, "learning_rate": 6.23622452392088e-06, "loss": 0.2607, "step": 2938 }, { "epoch": 0.63, "learning_rate": 6.229748853201605e-06, "loss": 0.293, "step": 2939 }, { "epoch": 0.63, "learning_rate": 6.223275024966949e-06, "loss": 0.2979, "step": 2940 }, { "epoch": 0.63, "learning_rate": 6.216803042380617e-06, "loss": 0.3028, "step": 2941 }, { "epoch": 0.63, "learning_rate": 6.210332908605401e-06, "loss": 0.2632, "step": 2942 }, { "epoch": 0.64, "learning_rate": 6.203864626803184e-06, "loss": 0.3135, "step": 2943 }, { "epoch": 0.64, "learning_rate": 6.197398200134961e-06, "loss": 0.3251, "step": 2944 }, { "epoch": 0.64, "learning_rate": 6.190933631760809e-06, "loss": 0.3168, "step": 2945 }, { "epoch": 0.64, "learning_rate": 6.184470924839897e-06, "loss": 0.3205, "step": 2946 }, { "epoch": 0.64, "learning_rate": 6.178010082530485e-06, "loss": 0.2788, "step": 2947 }, { "epoch": 0.64, "learning_rate": 6.171551107989928e-06, "loss": 0.2973, "step": 2948 }, { "epoch": 0.64, "learning_rate": 6.1650940043746566e-06, "loss": 0.3616, "step": 2949 }, { "epoch": 0.64, "learning_rate": 6.1586387748402025e-06, "loss": 0.2753, "step": 2950 }, { "epoch": 0.64, "learning_rate": 6.1521854225411626e-06, "loss": 0.2634, "step": 2951 }, { "epoch": 0.64, "learning_rate": 6.145733950631234e-06, "loss": 0.3332, "step": 2952 }, { "epoch": 0.64, "learning_rate": 6.139284362263185e-06, "loss": 0.298, "step": 2953 }, { "epoch": 0.64, "learning_rate": 6.132836660588868e-06, "loss": 0.2774, "step": 2954 }, { "epoch": 0.64, "learning_rate": 6.126390848759205e-06, "loss": 0.3063, "step": 2955 }, { "epoch": 0.64, "learning_rate": 6.119946929924212e-06, "loss": 0.3062, "step": 2956 }, { "epoch": 0.64, "learning_rate": 6.113504907232957e-06, "loss": 0.2741, "step": 2957 }, { "epoch": 0.64, "learning_rate": 6.107064783833599e-06, "loss": 0.2658, "step": 2958 }, { "epoch": 0.64, "learning_rate": 6.1006265628733664e-06, "loss": 0.2954, "step": 2959 }, { "epoch": 0.64, "learning_rate": 6.094190247498545e-06, "loss": 0.3187, "step": 2960 }, { "epoch": 0.64, "learning_rate": 6.087755840854503e-06, "loss": 0.2658, "step": 2961 }, { "epoch": 0.64, "learning_rate": 6.081323346085678e-06, "loss": 0.2815, "step": 2962 }, { "epoch": 0.64, "learning_rate": 6.074892766335556e-06, "loss": 0.2573, "step": 2963 }, { "epoch": 0.64, "learning_rate": 6.068464104746709e-06, "loss": 0.2954, "step": 2964 }, { "epoch": 0.64, "learning_rate": 6.062037364460752e-06, "loss": 0.261, "step": 2965 }, { "epoch": 0.64, "learning_rate": 6.05561254861837e-06, "loss": 0.2782, "step": 2966 }, { "epoch": 0.64, "learning_rate": 6.049189660359316e-06, "loss": 0.2705, "step": 2967 }, { "epoch": 0.64, "learning_rate": 6.042768702822381e-06, "loss": 0.2581, "step": 2968 }, { "epoch": 0.64, "learning_rate": 6.03634967914543e-06, "loss": 0.289, "step": 2969 }, { "epoch": 0.64, "learning_rate": 6.029932592465379e-06, "loss": 0.2534, "step": 2970 }, { "epoch": 0.64, "learning_rate": 6.02351744591819e-06, "loss": 0.3127, "step": 2971 }, { "epoch": 0.64, "learning_rate": 6.01710424263888e-06, "loss": 0.2795, "step": 2972 }, { "epoch": 0.64, "learning_rate": 6.0106929857615285e-06, "loss": 0.2388, "step": 2973 }, { "epoch": 0.64, "learning_rate": 6.004283678419243e-06, "loss": 0.2868, "step": 2974 }, { "epoch": 0.64, "learning_rate": 5.997876323744194e-06, "loss": 0.2883, "step": 2975 }, { "epoch": 0.64, "learning_rate": 5.991470924867593e-06, "loss": 0.2897, "step": 2976 }, { "epoch": 0.64, "learning_rate": 5.985067484919693e-06, "loss": 0.2842, "step": 2977 }, { "epoch": 0.64, "learning_rate": 5.978666007029796e-06, "loss": 0.2974, "step": 2978 }, { "epoch": 0.64, "learning_rate": 5.972266494326234e-06, "loss": 0.3071, "step": 2979 }, { "epoch": 0.64, "learning_rate": 5.9658689499363905e-06, "loss": 0.2781, "step": 2980 }, { "epoch": 0.64, "learning_rate": 5.959473376986686e-06, "loss": 0.3181, "step": 2981 }, { "epoch": 0.64, "learning_rate": 5.953079778602567e-06, "loss": 0.2895, "step": 2982 }, { "epoch": 0.64, "learning_rate": 5.946688157908528e-06, "loss": 0.308, "step": 2983 }, { "epoch": 0.64, "learning_rate": 5.940298518028089e-06, "loss": 0.3303, "step": 2984 }, { "epoch": 0.64, "learning_rate": 5.933910862083802e-06, "loss": 0.2935, "step": 2985 }, { "epoch": 0.64, "learning_rate": 5.927525193197251e-06, "loss": 0.2695, "step": 2986 }, { "epoch": 0.64, "learning_rate": 5.921141514489057e-06, "loss": 0.3391, "step": 2987 }, { "epoch": 0.64, "learning_rate": 5.914759829078853e-06, "loss": 0.2672, "step": 2988 }, { "epoch": 0.65, "learning_rate": 5.908380140085308e-06, "loss": 0.2562, "step": 2989 }, { "epoch": 0.65, "learning_rate": 5.902002450626116e-06, "loss": 0.2851, "step": 2990 }, { "epoch": 0.65, "learning_rate": 5.895626763817985e-06, "loss": 0.3408, "step": 2991 }, { "epoch": 0.65, "learning_rate": 5.88925308277666e-06, "loss": 0.3106, "step": 2992 }, { "epoch": 0.65, "learning_rate": 5.882881410616885e-06, "loss": 0.3154, "step": 2993 }, { "epoch": 0.65, "learning_rate": 5.876511750452443e-06, "loss": 0.3846, "step": 2994 }, { "epoch": 0.65, "learning_rate": 5.8701441053961185e-06, "loss": 0.2646, "step": 2995 }, { "epoch": 0.65, "learning_rate": 5.863778478559718e-06, "loss": 0.26, "step": 2996 }, { "epoch": 0.65, "learning_rate": 5.857414873054059e-06, "loss": 0.2703, "step": 2997 }, { "epoch": 0.65, "learning_rate": 5.85105329198898e-06, "loss": 0.3448, "step": 2998 }, { "epoch": 0.65, "learning_rate": 5.844693738473315e-06, "loss": 0.2679, "step": 2999 }, { "epoch": 0.65, "learning_rate": 5.838336215614917e-06, "loss": 0.2919, "step": 3000 }, { "epoch": 0.65, "learning_rate": 5.831980726520654e-06, "loss": 0.2833, "step": 3001 }, { "epoch": 0.65, "learning_rate": 5.825627274296374e-06, "loss": 0.2905, "step": 3002 }, { "epoch": 0.65, "learning_rate": 5.819275862046961e-06, "loss": 0.3368, "step": 3003 }, { "epoch": 0.65, "learning_rate": 5.812926492876285e-06, "loss": 0.2955, "step": 3004 }, { "epoch": 0.65, "learning_rate": 5.80657916988721e-06, "loss": 0.2581, "step": 3005 }, { "epoch": 0.65, "learning_rate": 5.800233896181626e-06, "loss": 0.2995, "step": 3006 }, { "epoch": 0.65, "learning_rate": 5.793890674860389e-06, "loss": 0.2632, "step": 3007 }, { "epoch": 0.65, "learning_rate": 5.787549509023382e-06, "loss": 0.3587, "step": 3008 }, { "epoch": 0.65, "learning_rate": 5.781210401769466e-06, "loss": 0.2661, "step": 3009 }, { "epoch": 0.65, "learning_rate": 5.7748733561965e-06, "loss": 0.3067, "step": 3010 }, { "epoch": 0.65, "learning_rate": 5.768538375401337e-06, "loss": 0.2584, "step": 3011 }, { "epoch": 0.65, "learning_rate": 5.762205462479817e-06, "loss": 0.3124, "step": 3012 }, { "epoch": 0.65, "learning_rate": 5.755874620526776e-06, "loss": 0.2653, "step": 3013 }, { "epoch": 0.65, "learning_rate": 5.749545852636027e-06, "loss": 0.2373, "step": 3014 }, { "epoch": 0.65, "learning_rate": 5.743219161900388e-06, "loss": 0.299, "step": 3015 }, { "epoch": 0.65, "learning_rate": 5.736894551411642e-06, "loss": 0.2654, "step": 3016 }, { "epoch": 0.65, "learning_rate": 5.730572024260563e-06, "loss": 0.3102, "step": 3017 }, { "epoch": 0.65, "learning_rate": 5.724251583536917e-06, "loss": 0.3354, "step": 3018 }, { "epoch": 0.65, "learning_rate": 5.717933232329428e-06, "loss": 0.2709, "step": 3019 }, { "epoch": 0.65, "learning_rate": 5.711616973725826e-06, "loss": 0.3312, "step": 3020 }, { "epoch": 0.65, "learning_rate": 5.705302810812791e-06, "loss": 0.2809, "step": 3021 }, { "epoch": 0.65, "learning_rate": 5.698990746676001e-06, "loss": 0.3108, "step": 3022 }, { "epoch": 0.65, "learning_rate": 5.692680784400102e-06, "loss": 0.2911, "step": 3023 }, { "epoch": 0.65, "learning_rate": 5.686372927068699e-06, "loss": 0.2606, "step": 3024 }, { "epoch": 0.65, "learning_rate": 5.680067177764389e-06, "loss": 0.2865, "step": 3025 }, { "epoch": 0.65, "learning_rate": 5.673763539568727e-06, "loss": 0.2798, "step": 3026 }, { "epoch": 0.65, "learning_rate": 5.6674620155622405e-06, "loss": 0.2713, "step": 3027 }, { "epoch": 0.65, "learning_rate": 5.66116260882442e-06, "loss": 0.3573, "step": 3028 }, { "epoch": 0.65, "learning_rate": 5.654865322433725e-06, "loss": 0.2736, "step": 3029 }, { "epoch": 0.65, "learning_rate": 5.648570159467576e-06, "loss": 0.262, "step": 3030 }, { "epoch": 0.65, "learning_rate": 5.642277123002358e-06, "loss": 0.2522, "step": 3031 }, { "epoch": 0.65, "learning_rate": 5.635986216113416e-06, "loss": 0.2501, "step": 3032 }, { "epoch": 0.65, "learning_rate": 5.629697441875055e-06, "loss": 0.3021, "step": 3033 }, { "epoch": 0.65, "learning_rate": 5.623410803360536e-06, "loss": 0.3368, "step": 3034 }, { "epoch": 0.66, "learning_rate": 5.6171263036420774e-06, "loss": 0.2915, "step": 3035 }, { "epoch": 0.66, "learning_rate": 5.6108439457908494e-06, "loss": 0.2356, "step": 3036 }, { "epoch": 0.66, "learning_rate": 5.604563732876989e-06, "loss": 0.2917, "step": 3037 }, { "epoch": 0.66, "learning_rate": 5.598285667969559e-06, "loss": 0.2661, "step": 3038 }, { "epoch": 0.66, "learning_rate": 5.592009754136599e-06, "loss": 0.305, "step": 3039 }, { "epoch": 0.66, "learning_rate": 5.585735994445088e-06, "loss": 0.2933, "step": 3040 }, { "epoch": 0.66, "learning_rate": 5.579464391960938e-06, "loss": 0.2477, "step": 3041 }, { "epoch": 0.66, "learning_rate": 5.5731949497490326e-06, "loss": 0.3148, "step": 3042 }, { "epoch": 0.66, "learning_rate": 5.56692767087318e-06, "loss": 0.2877, "step": 3043 }, { "epoch": 0.66, "learning_rate": 5.560662558396139e-06, "loss": 0.2833, "step": 3044 }, { "epoch": 0.66, "learning_rate": 5.5543996153796085e-06, "loss": 0.3495, "step": 3045 }, { "epoch": 0.66, "learning_rate": 5.548138844884228e-06, "loss": 0.3013, "step": 3046 }, { "epoch": 0.66, "learning_rate": 5.541880249969572e-06, "loss": 0.315, "step": 3047 }, { "epoch": 0.66, "learning_rate": 5.535623833694156e-06, "loss": 0.3157, "step": 3048 }, { "epoch": 0.66, "learning_rate": 5.529369599115428e-06, "loss": 0.3503, "step": 3049 }, { "epoch": 0.66, "learning_rate": 5.52311754928977e-06, "loss": 0.2535, "step": 3050 }, { "epoch": 0.66, "learning_rate": 5.516867687272504e-06, "loss": 0.2714, "step": 3051 }, { "epoch": 0.66, "learning_rate": 5.510620016117867e-06, "loss": 0.2741, "step": 3052 }, { "epoch": 0.66, "learning_rate": 5.504374538879035e-06, "loss": 0.254, "step": 3053 }, { "epoch": 0.66, "learning_rate": 5.498131258608122e-06, "loss": 0.2963, "step": 3054 }, { "epoch": 0.66, "learning_rate": 5.491890178356141e-06, "loss": 0.2288, "step": 3055 }, { "epoch": 0.66, "learning_rate": 5.485651301173062e-06, "loss": 0.3139, "step": 3056 }, { "epoch": 0.66, "learning_rate": 5.479414630107756e-06, "loss": 0.3184, "step": 3057 }, { "epoch": 0.66, "learning_rate": 5.473180168208021e-06, "loss": 0.2724, "step": 3058 }, { "epoch": 0.66, "learning_rate": 5.4669479185205844e-06, "loss": 0.3243, "step": 3059 }, { "epoch": 0.66, "learning_rate": 5.460717884091072e-06, "loss": 0.2936, "step": 3060 }, { "epoch": 0.66, "learning_rate": 5.454490067964054e-06, "loss": 0.2738, "step": 3061 }, { "epoch": 0.66, "learning_rate": 5.448264473182996e-06, "loss": 0.2757, "step": 3062 }, { "epoch": 0.66, "learning_rate": 5.442041102790285e-06, "loss": 0.3341, "step": 3063 }, { "epoch": 0.66, "learning_rate": 5.435819959827224e-06, "loss": 0.3052, "step": 3064 }, { "epoch": 0.66, "learning_rate": 5.429601047334022e-06, "loss": 0.3333, "step": 3065 }, { "epoch": 0.66, "learning_rate": 5.4233843683498e-06, "loss": 0.343, "step": 3066 }, { "epoch": 0.66, "learning_rate": 5.417169925912586e-06, "loss": 0.3161, "step": 3067 }, { "epoch": 0.66, "learning_rate": 5.410957723059327e-06, "loss": 0.2337, "step": 3068 }, { "epoch": 0.66, "learning_rate": 5.404747762825852e-06, "loss": 0.2698, "step": 3069 }, { "epoch": 0.66, "learning_rate": 5.398540048246912e-06, "loss": 0.2828, "step": 3070 }, { "epoch": 0.66, "learning_rate": 5.3923345823561624e-06, "loss": 0.2876, "step": 3071 }, { "epoch": 0.66, "learning_rate": 5.3861313681861425e-06, "loss": 0.2907, "step": 3072 }, { "epoch": 0.66, "learning_rate": 5.379930408768315e-06, "loss": 0.2783, "step": 3073 }, { "epoch": 0.66, "learning_rate": 5.373731707133012e-06, "loss": 0.3068, "step": 3074 }, { "epoch": 0.66, "learning_rate": 5.367535266309489e-06, "loss": 0.3491, "step": 3075 }, { "epoch": 0.66, "learning_rate": 5.361341089325884e-06, "loss": 0.2916, "step": 3076 }, { "epoch": 0.66, "learning_rate": 5.35514917920923e-06, "loss": 0.2368, "step": 3077 }, { "epoch": 0.66, "learning_rate": 5.348959538985452e-06, "loss": 0.2587, "step": 3078 }, { "epoch": 0.66, "learning_rate": 5.342772171679364e-06, "loss": 0.3245, "step": 3079 }, { "epoch": 0.66, "learning_rate": 5.3365870803146766e-06, "loss": 0.2664, "step": 3080 }, { "epoch": 0.66, "learning_rate": 5.3304042679139775e-06, "loss": 0.2957, "step": 3081 }, { "epoch": 0.67, "learning_rate": 5.324223737498751e-06, "loss": 0.306, "step": 3082 }, { "epoch": 0.67, "learning_rate": 5.318045492089359e-06, "loss": 0.3123, "step": 3083 }, { "epoch": 0.67, "learning_rate": 5.311869534705043e-06, "loss": 0.2997, "step": 3084 }, { "epoch": 0.67, "learning_rate": 5.3056958683639494e-06, "loss": 0.2798, "step": 3085 }, { "epoch": 0.67, "learning_rate": 5.299524496083069e-06, "loss": 0.3007, "step": 3086 }, { "epoch": 0.67, "learning_rate": 5.293355420878308e-06, "loss": 0.2671, "step": 3087 }, { "epoch": 0.67, "learning_rate": 5.28718864576442e-06, "loss": 0.344, "step": 3088 }, { "epoch": 0.67, "learning_rate": 5.281024173755047e-06, "loss": 0.3084, "step": 3089 }, { "epoch": 0.67, "learning_rate": 5.27486200786272e-06, "loss": 0.3028, "step": 3090 }, { "epoch": 0.67, "learning_rate": 5.26870215109881e-06, "loss": 0.3361, "step": 3091 }, { "epoch": 0.67, "learning_rate": 5.262544606473592e-06, "loss": 0.3159, "step": 3092 }, { "epoch": 0.67, "learning_rate": 5.256389376996192e-06, "loss": 0.2703, "step": 3093 }, { "epoch": 0.67, "learning_rate": 5.2502364656746115e-06, "loss": 0.3298, "step": 3094 }, { "epoch": 0.67, "learning_rate": 5.244085875515718e-06, "loss": 0.2948, "step": 3095 }, { "epoch": 0.67, "learning_rate": 5.237937609525243e-06, "loss": 0.2969, "step": 3096 }, { "epoch": 0.67, "learning_rate": 5.231791670707784e-06, "loss": 0.2578, "step": 3097 }, { "epoch": 0.67, "learning_rate": 5.225648062066802e-06, "loss": 0.3025, "step": 3098 }, { "epoch": 0.67, "learning_rate": 5.219506786604616e-06, "loss": 0.3312, "step": 3099 }, { "epoch": 0.67, "learning_rate": 5.213367847322408e-06, "loss": 0.2822, "step": 3100 }, { "epoch": 0.67, "learning_rate": 5.207231247220217e-06, "loss": 0.3535, "step": 3101 }, { "epoch": 0.67, "learning_rate": 5.201096989296941e-06, "loss": 0.2774, "step": 3102 }, { "epoch": 0.67, "learning_rate": 5.194965076550323e-06, "loss": 0.2669, "step": 3103 }, { "epoch": 0.67, "learning_rate": 5.188835511976985e-06, "loss": 0.291, "step": 3104 }, { "epoch": 0.67, "learning_rate": 5.182708298572366e-06, "loss": 0.319, "step": 3105 }, { "epoch": 0.67, "learning_rate": 5.176583439330789e-06, "loss": 0.2896, "step": 3106 }, { "epoch": 0.67, "learning_rate": 5.17046093724541e-06, "loss": 0.335, "step": 3107 }, { "epoch": 0.67, "learning_rate": 5.164340795308225e-06, "loss": 0.2431, "step": 3108 }, { "epoch": 0.67, "learning_rate": 5.158223016510099e-06, "loss": 0.2551, "step": 3109 }, { "epoch": 0.67, "learning_rate": 5.152107603840726e-06, "loss": 0.2856, "step": 3110 }, { "epoch": 0.67, "learning_rate": 5.145994560288649e-06, "loss": 0.2876, "step": 3111 }, { "epoch": 0.67, "learning_rate": 5.139883888841251e-06, "loss": 0.2551, "step": 3112 }, { "epoch": 0.67, "learning_rate": 5.1337755924847555e-06, "loss": 0.3264, "step": 3113 }, { "epoch": 0.67, "learning_rate": 5.12766967420423e-06, "loss": 0.2844, "step": 3114 }, { "epoch": 0.67, "learning_rate": 5.121566136983576e-06, "loss": 0.267, "step": 3115 }, { "epoch": 0.67, "learning_rate": 5.115464983805532e-06, "loss": 0.3, "step": 3116 }, { "epoch": 0.67, "learning_rate": 5.109366217651668e-06, "loss": 0.253, "step": 3117 }, { "epoch": 0.67, "learning_rate": 5.1032698415023965e-06, "loss": 0.2737, "step": 3118 }, { "epoch": 0.67, "learning_rate": 5.097175858336951e-06, "loss": 0.282, "step": 3119 }, { "epoch": 0.67, "learning_rate": 5.091084271133403e-06, "loss": 0.3205, "step": 3120 }, { "epoch": 0.67, "learning_rate": 5.084995082868658e-06, "loss": 0.2863, "step": 3121 }, { "epoch": 0.67, "learning_rate": 5.07890829651843e-06, "loss": 0.261, "step": 3122 }, { "epoch": 0.67, "learning_rate": 5.072823915057281e-06, "loss": 0.3302, "step": 3123 }, { "epoch": 0.67, "learning_rate": 5.066741941458591e-06, "loss": 0.3004, "step": 3124 }, { "epoch": 0.67, "learning_rate": 5.060662378694545e-06, "loss": 0.2862, "step": 3125 }, { "epoch": 0.67, "learning_rate": 5.054585229736182e-06, "loss": 0.2669, "step": 3126 }, { "epoch": 0.67, "learning_rate": 5.0485104975533364e-06, "loss": 0.3545, "step": 3127 }, { "epoch": 0.68, "learning_rate": 5.042438185114673e-06, "loss": 0.2866, "step": 3128 }, { "epoch": 0.68, "learning_rate": 5.036368295387669e-06, "loss": 0.2242, "step": 3129 }, { "epoch": 0.68, "learning_rate": 5.030300831338622e-06, "loss": 0.3244, "step": 3130 }, { "epoch": 0.68, "learning_rate": 5.0242357959326405e-06, "loss": 0.3142, "step": 3131 }, { "epoch": 0.68, "learning_rate": 5.018173192133649e-06, "loss": 0.3341, "step": 3132 }, { "epoch": 0.68, "learning_rate": 5.0121130229043795e-06, "loss": 0.3138, "step": 3133 }, { "epoch": 0.68, "learning_rate": 5.006055291206375e-06, "loss": 0.2683, "step": 3134 }, { "epoch": 0.68, "learning_rate": 5.000000000000003e-06, "loss": 0.2432, "step": 3135 }, { "epoch": 0.68, "learning_rate": 4.993947152244409e-06, "loss": 0.2887, "step": 3136 }, { "epoch": 0.68, "learning_rate": 4.987896750897563e-06, "loss": 0.2476, "step": 3137 }, { "epoch": 0.68, "learning_rate": 4.981848798916247e-06, "loss": 0.272, "step": 3138 }, { "epoch": 0.68, "learning_rate": 4.975803299256021e-06, "loss": 0.2507, "step": 3139 }, { "epoch": 0.68, "learning_rate": 4.969760254871272e-06, "loss": 0.3393, "step": 3140 }, { "epoch": 0.68, "learning_rate": 4.963719668715172e-06, "loss": 0.2734, "step": 3141 }, { "epoch": 0.68, "learning_rate": 4.957681543739697e-06, "loss": 0.2792, "step": 3142 }, { "epoch": 0.68, "learning_rate": 4.951645882895623e-06, "loss": 0.2845, "step": 3143 }, { "epoch": 0.68, "learning_rate": 4.945612689132508e-06, "loss": 0.2992, "step": 3144 }, { "epoch": 0.68, "learning_rate": 4.939581965398722e-06, "loss": 0.2845, "step": 3145 }, { "epoch": 0.68, "learning_rate": 4.933553714641419e-06, "loss": 0.3016, "step": 3146 }, { "epoch": 0.68, "learning_rate": 4.927527939806545e-06, "loss": 0.3285, "step": 3147 }, { "epoch": 0.68, "learning_rate": 4.921504643838839e-06, "loss": 0.297, "step": 3148 }, { "epoch": 0.68, "learning_rate": 4.9154838296818246e-06, "loss": 0.2412, "step": 3149 }, { "epoch": 0.68, "learning_rate": 4.909465500277813e-06, "loss": 0.247, "step": 3150 }, { "epoch": 0.68, "learning_rate": 4.9034496585679016e-06, "loss": 0.2525, "step": 3151 }, { "epoch": 0.68, "learning_rate": 4.8974363074919826e-06, "loss": 0.2701, "step": 3152 }, { "epoch": 0.68, "learning_rate": 4.891425449988712e-06, "loss": 0.3193, "step": 3153 }, { "epoch": 0.68, "learning_rate": 4.885417088995533e-06, "loss": 0.2969, "step": 3154 }, { "epoch": 0.68, "learning_rate": 4.879411227448688e-06, "loss": 0.2817, "step": 3155 }, { "epoch": 0.68, "learning_rate": 4.873407868283166e-06, "loss": 0.2609, "step": 3156 }, { "epoch": 0.68, "learning_rate": 4.867407014432762e-06, "loss": 0.2481, "step": 3157 }, { "epoch": 0.68, "learning_rate": 4.861408668830022e-06, "loss": 0.3286, "step": 3158 }, { "epoch": 0.68, "learning_rate": 4.85541283440629e-06, "loss": 0.2848, "step": 3159 }, { "epoch": 0.68, "learning_rate": 4.84941951409167e-06, "loss": 0.279, "step": 3160 }, { "epoch": 0.68, "learning_rate": 4.843428710815027e-06, "loss": 0.2917, "step": 3161 }, { "epoch": 0.68, "learning_rate": 4.83744042750402e-06, "loss": 0.3211, "step": 3162 }, { "epoch": 0.68, "learning_rate": 4.831454667085059e-06, "loss": 0.2885, "step": 3163 }, { "epoch": 0.68, "learning_rate": 4.825471432483328e-06, "loss": 0.2755, "step": 3164 }, { "epoch": 0.68, "learning_rate": 4.819490726622774e-06, "loss": 0.3147, "step": 3165 }, { "epoch": 0.68, "learning_rate": 4.81351255242611e-06, "loss": 0.3358, "step": 3166 }, { "epoch": 0.68, "learning_rate": 4.807536912814808e-06, "loss": 0.3138, "step": 3167 }, { "epoch": 0.68, "learning_rate": 4.801563810709103e-06, "loss": 0.3098, "step": 3168 }, { "epoch": 0.68, "learning_rate": 4.795593249028002e-06, "loss": 0.2941, "step": 3169 }, { "epoch": 0.68, "learning_rate": 4.789625230689244e-06, "loss": 0.3236, "step": 3170 }, { "epoch": 0.68, "learning_rate": 4.783659758609357e-06, "loss": 0.2924, "step": 3171 }, { "epoch": 0.68, "learning_rate": 4.777696835703597e-06, "loss": 0.286, "step": 3172 }, { "epoch": 0.68, "learning_rate": 4.771736464885984e-06, "loss": 0.349, "step": 3173 }, { "epoch": 0.69, "learning_rate": 4.765778649069306e-06, "loss": 0.2497, "step": 3174 }, { "epoch": 0.69, "learning_rate": 4.7598233911650736e-06, "loss": 0.3064, "step": 3175 }, { "epoch": 0.69, "learning_rate": 4.753870694083572e-06, "loss": 0.2821, "step": 3176 }, { "epoch": 0.69, "learning_rate": 4.747920560733825e-06, "loss": 0.2651, "step": 3177 }, { "epoch": 0.69, "learning_rate": 4.7419729940236015e-06, "loss": 0.233, "step": 3178 }, { "epoch": 0.69, "learning_rate": 4.73602799685942e-06, "loss": 0.3389, "step": 3179 }, { "epoch": 0.69, "learning_rate": 4.730085572146543e-06, "loss": 0.3225, "step": 3180 }, { "epoch": 0.69, "learning_rate": 4.724145722788973e-06, "loss": 0.2898, "step": 3181 }, { "epoch": 0.69, "learning_rate": 4.718208451689458e-06, "loss": 0.291, "step": 3182 }, { "epoch": 0.69, "learning_rate": 4.712273761749482e-06, "loss": 0.2589, "step": 3183 }, { "epoch": 0.69, "learning_rate": 4.7063416558692694e-06, "loss": 0.236, "step": 3184 }, { "epoch": 0.69, "learning_rate": 4.700412136947784e-06, "loss": 0.2846, "step": 3185 }, { "epoch": 0.69, "learning_rate": 4.694485207882723e-06, "loss": 0.323, "step": 3186 }, { "epoch": 0.69, "learning_rate": 4.688560871570513e-06, "loss": 0.2393, "step": 3187 }, { "epoch": 0.69, "learning_rate": 4.6826391309063305e-06, "loss": 0.3632, "step": 3188 }, { "epoch": 0.69, "learning_rate": 4.6767199887840575e-06, "loss": 0.2894, "step": 3189 }, { "epoch": 0.69, "learning_rate": 4.670803448096331e-06, "loss": 0.2936, "step": 3190 }, { "epoch": 0.69, "learning_rate": 4.664889511734509e-06, "loss": 0.3106, "step": 3191 }, { "epoch": 0.69, "learning_rate": 4.658978182588658e-06, "loss": 0.2897, "step": 3192 }, { "epoch": 0.69, "learning_rate": 4.6530694635476e-06, "loss": 0.2446, "step": 3193 }, { "epoch": 0.69, "learning_rate": 4.647163357498866e-06, "loss": 0.251, "step": 3194 }, { "epoch": 0.69, "learning_rate": 4.641259867328709e-06, "loss": 0.292, "step": 3195 }, { "epoch": 0.69, "learning_rate": 4.6353589959221065e-06, "loss": 0.296, "step": 3196 }, { "epoch": 0.69, "learning_rate": 4.629460746162759e-06, "loss": 0.2757, "step": 3197 }, { "epoch": 0.69, "learning_rate": 4.623565120933081e-06, "loss": 0.281, "step": 3198 }, { "epoch": 0.69, "learning_rate": 4.617672123114208e-06, "loss": 0.3052, "step": 3199 }, { "epoch": 0.69, "learning_rate": 4.611781755585986e-06, "loss": 0.2234, "step": 3200 }, { "epoch": 0.69, "learning_rate": 4.605894021226985e-06, "loss": 0.2918, "step": 3201 }, { "epoch": 0.69, "learning_rate": 4.6000089229144805e-06, "loss": 0.2667, "step": 3202 }, { "epoch": 0.69, "learning_rate": 4.5941264635244605e-06, "loss": 0.2564, "step": 3203 }, { "epoch": 0.69, "learning_rate": 4.5882466459316245e-06, "loss": 0.275, "step": 3204 }, { "epoch": 0.69, "learning_rate": 4.58236947300939e-06, "loss": 0.323, "step": 3205 }, { "epoch": 0.69, "learning_rate": 4.57649494762986e-06, "loss": 0.2958, "step": 3206 }, { "epoch": 0.69, "learning_rate": 4.570623072663869e-06, "loss": 0.2995, "step": 3207 }, { "epoch": 0.69, "learning_rate": 4.564753850980942e-06, "loss": 0.3186, "step": 3208 }, { "epoch": 0.69, "learning_rate": 4.5588872854493e-06, "loss": 0.3012, "step": 3209 }, { "epoch": 0.69, "learning_rate": 4.553023378935889e-06, "loss": 0.2799, "step": 3210 }, { "epoch": 0.69, "learning_rate": 4.547162134306337e-06, "loss": 0.2638, "step": 3211 }, { "epoch": 0.69, "learning_rate": 4.541303554424975e-06, "loss": 0.2826, "step": 3212 }, { "epoch": 0.69, "learning_rate": 4.5354476421548356e-06, "loss": 0.2392, "step": 3213 }, { "epoch": 0.69, "learning_rate": 4.529594400357644e-06, "loss": 0.2952, "step": 3214 }, { "epoch": 0.69, "learning_rate": 4.523743831893823e-06, "loss": 0.3221, "step": 3215 }, { "epoch": 0.69, "learning_rate": 4.517895939622486e-06, "loss": 0.3413, "step": 3216 }, { "epoch": 0.69, "learning_rate": 4.512050726401443e-06, "loss": 0.3216, "step": 3217 }, { "epoch": 0.69, "learning_rate": 4.506208195087188e-06, "loss": 0.3045, "step": 3218 }, { "epoch": 0.69, "learning_rate": 4.500368348534918e-06, "loss": 0.3149, "step": 3219 }, { "epoch": 0.69, "learning_rate": 4.4945311895984985e-06, "loss": 0.3188, "step": 3220 }, { "epoch": 0.7, "learning_rate": 4.488696721130491e-06, "loss": 0.3203, "step": 3221 }, { "epoch": 0.7, "learning_rate": 4.482864945982155e-06, "loss": 0.3454, "step": 3222 }, { "epoch": 0.7, "learning_rate": 4.477035867003405e-06, "loss": 0.3086, "step": 3223 }, { "epoch": 0.7, "learning_rate": 4.471209487042868e-06, "loss": 0.2961, "step": 3224 }, { "epoch": 0.7, "learning_rate": 4.465385808947833e-06, "loss": 0.258, "step": 3225 }, { "epoch": 0.7, "learning_rate": 4.459564835564275e-06, "loss": 0.2882, "step": 3226 }, { "epoch": 0.7, "learning_rate": 4.45374656973685e-06, "loss": 0.268, "step": 3227 }, { "epoch": 0.7, "learning_rate": 4.447931014308875e-06, "loss": 0.2993, "step": 3228 }, { "epoch": 0.7, "learning_rate": 4.442118172122367e-06, "loss": 0.2323, "step": 3229 }, { "epoch": 0.7, "learning_rate": 4.436308046018002e-06, "loss": 0.294, "step": 3230 }, { "epoch": 0.7, "learning_rate": 4.430500638835129e-06, "loss": 0.2533, "step": 3231 }, { "epoch": 0.7, "learning_rate": 4.4246959534117726e-06, "loss": 0.3228, "step": 3232 }, { "epoch": 0.7, "learning_rate": 4.418893992584624e-06, "loss": 0.3132, "step": 3233 }, { "epoch": 0.7, "learning_rate": 4.413094759189046e-06, "loss": 0.2441, "step": 3234 }, { "epoch": 0.7, "learning_rate": 4.4072982560590625e-06, "loss": 0.2585, "step": 3235 }, { "epoch": 0.7, "learning_rate": 4.401504486027378e-06, "loss": 0.2829, "step": 3236 }, { "epoch": 0.7, "learning_rate": 4.395713451925339e-06, "loss": 0.2917, "step": 3237 }, { "epoch": 0.7, "learning_rate": 4.389925156582971e-06, "loss": 0.3258, "step": 3238 }, { "epoch": 0.7, "learning_rate": 4.384139602828963e-06, "loss": 0.2822, "step": 3239 }, { "epoch": 0.7, "learning_rate": 4.378356793490647e-06, "loss": 0.2524, "step": 3240 }, { "epoch": 0.7, "learning_rate": 4.372576731394037e-06, "loss": 0.262, "step": 3241 }, { "epoch": 0.7, "learning_rate": 4.3667994193637794e-06, "loss": 0.2704, "step": 3242 }, { "epoch": 0.7, "learning_rate": 4.361024860223201e-06, "loss": 0.3098, "step": 3243 }, { "epoch": 0.7, "learning_rate": 4.355253056794271e-06, "loss": 0.298, "step": 3244 }, { "epoch": 0.7, "learning_rate": 4.349484011897603e-06, "loss": 0.2771, "step": 3245 }, { "epoch": 0.7, "learning_rate": 4.343717728352481e-06, "loss": 0.321, "step": 3246 }, { "epoch": 0.7, "learning_rate": 4.33795420897683e-06, "loss": 0.3016, "step": 3247 }, { "epoch": 0.7, "learning_rate": 4.332193456587225e-06, "loss": 0.3061, "step": 3248 }, { "epoch": 0.7, "learning_rate": 4.3264354739988865e-06, "loss": 0.3042, "step": 3249 }, { "epoch": 0.7, "learning_rate": 4.3206802640256875e-06, "loss": 0.2527, "step": 3250 }, { "epoch": 0.7, "learning_rate": 4.31492782948014e-06, "loss": 0.2787, "step": 3251 }, { "epoch": 0.7, "learning_rate": 4.309178173173399e-06, "loss": 0.2628, "step": 3252 }, { "epoch": 0.7, "learning_rate": 4.303431297915277e-06, "loss": 0.2753, "step": 3253 }, { "epoch": 0.7, "learning_rate": 4.2976872065142e-06, "loss": 0.2914, "step": 3254 }, { "epoch": 0.7, "learning_rate": 4.291945901777261e-06, "loss": 0.2746, "step": 3255 }, { "epoch": 0.7, "learning_rate": 4.286207386510173e-06, "loss": 0.2722, "step": 3256 }, { "epoch": 0.7, "learning_rate": 4.280471663517288e-06, "loss": 0.2701, "step": 3257 }, { "epoch": 0.7, "learning_rate": 4.27473873560161e-06, "loss": 0.2946, "step": 3258 }, { "epoch": 0.7, "learning_rate": 4.269008605564751e-06, "loss": 0.2953, "step": 3259 }, { "epoch": 0.7, "learning_rate": 4.2632812762069765e-06, "loss": 0.3281, "step": 3260 }, { "epoch": 0.7, "learning_rate": 4.257556750327176e-06, "loss": 0.3368, "step": 3261 }, { "epoch": 0.7, "learning_rate": 4.251835030722867e-06, "loss": 0.2795, "step": 3262 }, { "epoch": 0.7, "learning_rate": 4.246116120190198e-06, "loss": 0.2986, "step": 3263 }, { "epoch": 0.7, "learning_rate": 4.240400021523944e-06, "loss": 0.3452, "step": 3264 }, { "epoch": 0.7, "learning_rate": 4.2346867375175074e-06, "loss": 0.2788, "step": 3265 }, { "epoch": 0.7, "learning_rate": 4.228976270962912e-06, "loss": 0.3268, "step": 3266 }, { "epoch": 0.71, "learning_rate": 4.223268624650808e-06, "loss": 0.2692, "step": 3267 }, { "epoch": 0.71, "learning_rate": 4.2175638013704655e-06, "loss": 0.2995, "step": 3268 }, { "epoch": 0.71, "learning_rate": 4.211861803909776e-06, "loss": 0.3083, "step": 3269 }, { "epoch": 0.71, "learning_rate": 4.2061626350552475e-06, "loss": 0.3012, "step": 3270 }, { "epoch": 0.71, "learning_rate": 4.2004662975920065e-06, "loss": 0.2663, "step": 3271 }, { "epoch": 0.71, "learning_rate": 4.194772794303807e-06, "loss": 0.2706, "step": 3272 }, { "epoch": 0.71, "learning_rate": 4.189082127972997e-06, "loss": 0.2886, "step": 3273 }, { "epoch": 0.71, "learning_rate": 4.18339430138055e-06, "loss": 0.3401, "step": 3274 }, { "epoch": 0.71, "learning_rate": 4.17770931730606e-06, "loss": 0.2508, "step": 3275 }, { "epoch": 0.71, "learning_rate": 4.17202717852771e-06, "loss": 0.3404, "step": 3276 }, { "epoch": 0.71, "learning_rate": 4.166347887822315e-06, "loss": 0.2496, "step": 3277 }, { "epoch": 0.71, "learning_rate": 4.160671447965286e-06, "loss": 0.2538, "step": 3278 }, { "epoch": 0.71, "learning_rate": 4.154997861730642e-06, "loss": 0.2949, "step": 3279 }, { "epoch": 0.71, "learning_rate": 4.149327131891009e-06, "loss": 0.2287, "step": 3280 }, { "epoch": 0.71, "learning_rate": 4.143659261217617e-06, "loss": 0.3068, "step": 3281 }, { "epoch": 0.71, "learning_rate": 4.1379942524803e-06, "loss": 0.3229, "step": 3282 }, { "epoch": 0.71, "learning_rate": 4.13233210844749e-06, "loss": 0.2844, "step": 3283 }, { "epoch": 0.71, "learning_rate": 4.126672831886223e-06, "loss": 0.298, "step": 3284 }, { "epoch": 0.71, "learning_rate": 4.121016425562129e-06, "loss": 0.2795, "step": 3285 }, { "epoch": 0.71, "learning_rate": 4.115362892239441e-06, "loss": 0.3157, "step": 3286 }, { "epoch": 0.71, "learning_rate": 4.109712234680982e-06, "loss": 0.2836, "step": 3287 }, { "epoch": 0.71, "learning_rate": 4.104064455648172e-06, "loss": 0.3075, "step": 3288 }, { "epoch": 0.71, "learning_rate": 4.098419557901036e-06, "loss": 0.2582, "step": 3289 }, { "epoch": 0.71, "learning_rate": 4.092777544198162e-06, "loss": 0.2524, "step": 3290 }, { "epoch": 0.71, "learning_rate": 4.087138417296761e-06, "loss": 0.2671, "step": 3291 }, { "epoch": 0.71, "learning_rate": 4.081502179952617e-06, "loss": 0.2575, "step": 3292 }, { "epoch": 0.71, "learning_rate": 4.075868834920093e-06, "loss": 0.304, "step": 3293 }, { "epoch": 0.71, "learning_rate": 4.070238384952162e-06, "loss": 0.2807, "step": 3294 }, { "epoch": 0.71, "learning_rate": 4.0646108328003635e-06, "loss": 0.2754, "step": 3295 }, { "epoch": 0.71, "learning_rate": 4.058986181214826e-06, "loss": 0.3508, "step": 3296 }, { "epoch": 0.71, "learning_rate": 4.053364432944265e-06, "loss": 0.3362, "step": 3297 }, { "epoch": 0.71, "learning_rate": 4.047745590735972e-06, "loss": 0.3082, "step": 3298 }, { "epoch": 0.71, "learning_rate": 4.042129657335818e-06, "loss": 0.3348, "step": 3299 }, { "epoch": 0.71, "learning_rate": 4.036516635488257e-06, "loss": 0.291, "step": 3300 }, { "epoch": 0.71, "learning_rate": 4.030906527936315e-06, "loss": 0.2795, "step": 3301 }, { "epoch": 0.71, "learning_rate": 4.025299337421599e-06, "loss": 0.2599, "step": 3302 }, { "epoch": 0.71, "learning_rate": 4.019695066684285e-06, "loss": 0.2843, "step": 3303 }, { "epoch": 0.71, "learning_rate": 4.014093718463128e-06, "loss": 0.252, "step": 3304 }, { "epoch": 0.71, "learning_rate": 4.008495295495446e-06, "loss": 0.2684, "step": 3305 }, { "epoch": 0.71, "learning_rate": 4.002899800517144e-06, "loss": 0.2911, "step": 3306 }, { "epoch": 0.71, "learning_rate": 3.997307236262672e-06, "loss": 0.3157, "step": 3307 }, { "epoch": 0.71, "learning_rate": 3.9917176054650715e-06, "loss": 0.2662, "step": 3308 }, { "epoch": 0.71, "learning_rate": 3.9861309108559365e-06, "loss": 0.3582, "step": 3309 }, { "epoch": 0.71, "learning_rate": 3.980547155165429e-06, "loss": 0.2562, "step": 3310 }, { "epoch": 0.71, "learning_rate": 3.974966341122279e-06, "loss": 0.3622, "step": 3311 }, { "epoch": 0.71, "learning_rate": 3.969388471453766e-06, "loss": 0.2922, "step": 3312 }, { "epoch": 0.72, "learning_rate": 3.963813548885751e-06, "loss": 0.2523, "step": 3313 }, { "epoch": 0.72, "learning_rate": 3.958241576142638e-06, "loss": 0.2503, "step": 3314 }, { "epoch": 0.72, "learning_rate": 3.952672555947397e-06, "loss": 0.3083, "step": 3315 }, { "epoch": 0.72, "learning_rate": 3.947106491021552e-06, "loss": 0.2364, "step": 3316 }, { "epoch": 0.72, "learning_rate": 3.9415433840851845e-06, "loss": 0.3161, "step": 3317 }, { "epoch": 0.72, "learning_rate": 3.9359832378569315e-06, "loss": 0.3174, "step": 3318 }, { "epoch": 0.72, "learning_rate": 3.930426055053976e-06, "loss": 0.2625, "step": 3319 }, { "epoch": 0.72, "learning_rate": 3.924871838392071e-06, "loss": 0.2774, "step": 3320 }, { "epoch": 0.72, "learning_rate": 3.919320590585497e-06, "loss": 0.2595, "step": 3321 }, { "epoch": 0.72, "learning_rate": 3.913772314347093e-06, "loss": 0.3087, "step": 3322 }, { "epoch": 0.72, "learning_rate": 3.908227012388257e-06, "loss": 0.3141, "step": 3323 }, { "epoch": 0.72, "learning_rate": 3.902684687418912e-06, "loss": 0.2791, "step": 3324 }, { "epoch": 0.72, "learning_rate": 3.897145342147548e-06, "loss": 0.2484, "step": 3325 }, { "epoch": 0.72, "learning_rate": 3.891608979281178e-06, "loss": 0.2791, "step": 3326 }, { "epoch": 0.72, "learning_rate": 3.886075601525376e-06, "loss": 0.2689, "step": 3327 }, { "epoch": 0.72, "learning_rate": 3.880545211584251e-06, "loss": 0.2837, "step": 3328 }, { "epoch": 0.72, "learning_rate": 3.87501781216044e-06, "loss": 0.2694, "step": 3329 }, { "epoch": 0.72, "learning_rate": 3.869493405955137e-06, "loss": 0.2902, "step": 3330 }, { "epoch": 0.72, "learning_rate": 3.8639719956680624e-06, "loss": 0.3334, "step": 3331 }, { "epoch": 0.72, "learning_rate": 3.858453583997476e-06, "loss": 0.2797, "step": 3332 }, { "epoch": 0.72, "learning_rate": 3.852938173640169e-06, "loss": 0.2834, "step": 3333 }, { "epoch": 0.72, "learning_rate": 3.847425767291471e-06, "loss": 0.2962, "step": 3334 }, { "epoch": 0.72, "learning_rate": 3.8419163676452386e-06, "loss": 0.2933, "step": 3335 }, { "epoch": 0.72, "learning_rate": 3.836409977393859e-06, "loss": 0.2158, "step": 3336 }, { "epoch": 0.72, "learning_rate": 3.830906599228261e-06, "loss": 0.2688, "step": 3337 }, { "epoch": 0.72, "learning_rate": 3.825406235837876e-06, "loss": 0.3169, "step": 3338 }, { "epoch": 0.72, "learning_rate": 3.819908889910694e-06, "loss": 0.2948, "step": 3339 }, { "epoch": 0.72, "learning_rate": 3.814414564133202e-06, "loss": 0.229, "step": 3340 }, { "epoch": 0.72, "learning_rate": 3.8089232611904224e-06, "loss": 0.2836, "step": 3341 }, { "epoch": 0.72, "learning_rate": 3.8034349837659124e-06, "loss": 0.2128, "step": 3342 }, { "epoch": 0.72, "learning_rate": 3.7979497345417236e-06, "loss": 0.2507, "step": 3343 }, { "epoch": 0.72, "learning_rate": 3.792467516198456e-06, "loss": 0.2612, "step": 3344 }, { "epoch": 0.72, "learning_rate": 3.7869883314152114e-06, "loss": 0.3004, "step": 3345 }, { "epoch": 0.72, "learning_rate": 3.7815121828696124e-06, "loss": 0.2602, "step": 3346 }, { "epoch": 0.72, "learning_rate": 3.7760390732377995e-06, "loss": 0.3098, "step": 3347 }, { "epoch": 0.72, "learning_rate": 3.7705690051944276e-06, "loss": 0.2725, "step": 3348 }, { "epoch": 0.72, "learning_rate": 3.7651019814126656e-06, "loss": 0.3231, "step": 3349 }, { "epoch": 0.72, "learning_rate": 3.759638004564193e-06, "loss": 0.329, "step": 3350 }, { "epoch": 0.72, "learning_rate": 3.7541770773192045e-06, "loss": 0.3853, "step": 3351 }, { "epoch": 0.72, "learning_rate": 3.748719202346397e-06, "loss": 0.3024, "step": 3352 }, { "epoch": 0.72, "learning_rate": 3.7432643823129833e-06, "loss": 0.2569, "step": 3353 }, { "epoch": 0.72, "learning_rate": 3.73781261988468e-06, "loss": 0.2862, "step": 3354 }, { "epoch": 0.72, "learning_rate": 3.7323639177257056e-06, "loss": 0.2958, "step": 3355 }, { "epoch": 0.72, "learning_rate": 3.7269182784987977e-06, "loss": 0.2589, "step": 3356 }, { "epoch": 0.72, "learning_rate": 3.721475704865176e-06, "loss": 0.3177, "step": 3357 }, { "epoch": 0.72, "learning_rate": 3.7160361994845738e-06, "loss": 0.2629, "step": 3358 }, { "epoch": 0.72, "learning_rate": 3.7105997650152326e-06, "loss": 0.2956, "step": 3359 }, { "epoch": 0.73, "learning_rate": 3.705166404113871e-06, "loss": 0.2665, "step": 3360 }, { "epoch": 0.73, "learning_rate": 3.6997361194357304e-06, "loss": 0.29, "step": 3361 }, { "epoch": 0.73, "learning_rate": 3.6943089136345344e-06, "loss": 0.2897, "step": 3362 }, { "epoch": 0.73, "learning_rate": 3.6888847893625023e-06, "loss": 0.2709, "step": 3363 }, { "epoch": 0.73, "learning_rate": 3.683463749270356e-06, "loss": 0.3046, "step": 3364 }, { "epoch": 0.73, "learning_rate": 3.6780457960072937e-06, "loss": 0.3333, "step": 3365 }, { "epoch": 0.73, "learning_rate": 3.6726309322210262e-06, "loss": 0.2519, "step": 3366 }, { "epoch": 0.73, "learning_rate": 3.667219160557739e-06, "loss": 0.2782, "step": 3367 }, { "epoch": 0.73, "learning_rate": 3.661810483662115e-06, "loss": 0.341, "step": 3368 }, { "epoch": 0.73, "learning_rate": 3.6564049041773177e-06, "loss": 0.2576, "step": 3369 }, { "epoch": 0.73, "learning_rate": 3.6510024247450038e-06, "loss": 0.2778, "step": 3370 }, { "epoch": 0.73, "learning_rate": 3.6456030480053095e-06, "loss": 0.2267, "step": 3371 }, { "epoch": 0.73, "learning_rate": 3.640206776596854e-06, "loss": 0.2768, "step": 3372 }, { "epoch": 0.73, "learning_rate": 3.6348136131567537e-06, "loss": 0.2654, "step": 3373 }, { "epoch": 0.73, "learning_rate": 3.62942356032058e-06, "loss": 0.3125, "step": 3374 }, { "epoch": 0.73, "learning_rate": 3.6240366207224075e-06, "loss": 0.3015, "step": 3375 }, { "epoch": 0.73, "learning_rate": 3.6186527969947826e-06, "loss": 0.3003, "step": 3376 }, { "epoch": 0.73, "learning_rate": 3.613272091768717e-06, "loss": 0.2587, "step": 3377 }, { "epoch": 0.73, "learning_rate": 3.6078945076737193e-06, "loss": 0.313, "step": 3378 }, { "epoch": 0.73, "learning_rate": 3.602520047337751e-06, "loss": 0.2912, "step": 3379 }, { "epoch": 0.73, "learning_rate": 3.5971487133872675e-06, "loss": 0.2898, "step": 3380 }, { "epoch": 0.73, "learning_rate": 3.5917805084471822e-06, "loss": 0.277, "step": 3381 }, { "epoch": 0.73, "learning_rate": 3.586415435140885e-06, "loss": 0.3, "step": 3382 }, { "epoch": 0.73, "learning_rate": 3.581053496090234e-06, "loss": 0.3174, "step": 3383 }, { "epoch": 0.73, "learning_rate": 3.575694693915558e-06, "loss": 0.2853, "step": 3384 }, { "epoch": 0.73, "learning_rate": 3.5703390312356477e-06, "loss": 0.3173, "step": 3385 }, { "epoch": 0.73, "learning_rate": 3.564986510667765e-06, "loss": 0.3163, "step": 3386 }, { "epoch": 0.73, "learning_rate": 3.5596371348276325e-06, "loss": 0.2494, "step": 3387 }, { "epoch": 0.73, "learning_rate": 3.5542909063294385e-06, "loss": 0.3553, "step": 3388 }, { "epoch": 0.73, "learning_rate": 3.5489478277858293e-06, "loss": 0.3264, "step": 3389 }, { "epoch": 0.73, "learning_rate": 3.5436079018079238e-06, "loss": 0.2665, "step": 3390 }, { "epoch": 0.73, "learning_rate": 3.5382711310052788e-06, "loss": 0.3551, "step": 3391 }, { "epoch": 0.73, "learning_rate": 3.5329375179859338e-06, "loss": 0.3524, "step": 3392 }, { "epoch": 0.73, "learning_rate": 3.5276070653563633e-06, "loss": 0.2693, "step": 3393 }, { "epoch": 0.73, "learning_rate": 3.5222797757215087e-06, "loss": 0.3269, "step": 3394 }, { "epoch": 0.73, "learning_rate": 3.5169556516847725e-06, "loss": 0.3161, "step": 3395 }, { "epoch": 0.73, "learning_rate": 3.5116346958479873e-06, "loss": 0.2961, "step": 3396 }, { "epoch": 0.73, "learning_rate": 3.5063169108114647e-06, "loss": 0.2554, "step": 3397 }, { "epoch": 0.73, "learning_rate": 3.501002299173948e-06, "loss": 0.2885, "step": 3398 }, { "epoch": 0.73, "learning_rate": 3.495690863532637e-06, "loss": 0.282, "step": 3399 }, { "epoch": 0.73, "learning_rate": 3.490382606483178e-06, "loss": 0.2437, "step": 3400 }, { "epoch": 0.73, "learning_rate": 3.485077530619664e-06, "loss": 0.2756, "step": 3401 }, { "epoch": 0.73, "learning_rate": 3.4797756385346316e-06, "loss": 0.2888, "step": 3402 }, { "epoch": 0.73, "learning_rate": 3.474476932819062e-06, "loss": 0.2886, "step": 3403 }, { "epoch": 0.73, "learning_rate": 3.4691814160623904e-06, "loss": 0.2753, "step": 3404 }, { "epoch": 0.73, "learning_rate": 3.4638890908524726e-06, "loss": 0.3241, "step": 3405 }, { "epoch": 0.74, "learning_rate": 3.4585999597756194e-06, "loss": 0.3183, "step": 3406 }, { "epoch": 0.74, "learning_rate": 3.4533140254165776e-06, "loss": 0.2431, "step": 3407 }, { "epoch": 0.74, "learning_rate": 3.448031290358528e-06, "loss": 0.3133, "step": 3408 }, { "epoch": 0.74, "learning_rate": 3.4427517571831015e-06, "loss": 0.3052, "step": 3409 }, { "epoch": 0.74, "learning_rate": 3.4374754284703395e-06, "loss": 0.3098, "step": 3410 }, { "epoch": 0.74, "learning_rate": 3.432202306798742e-06, "loss": 0.2936, "step": 3411 }, { "epoch": 0.74, "learning_rate": 3.4269323947452324e-06, "loss": 0.2318, "step": 3412 }, { "epoch": 0.74, "learning_rate": 3.421665694885156e-06, "loss": 0.2828, "step": 3413 }, { "epoch": 0.74, "learning_rate": 3.4164022097923043e-06, "loss": 0.2665, "step": 3414 }, { "epoch": 0.74, "learning_rate": 3.4111419420388904e-06, "loss": 0.3216, "step": 3415 }, { "epoch": 0.74, "learning_rate": 3.4058848941955523e-06, "loss": 0.31, "step": 3416 }, { "epoch": 0.74, "learning_rate": 3.400631068831359e-06, "loss": 0.3374, "step": 3417 }, { "epoch": 0.74, "learning_rate": 3.3953804685138047e-06, "loss": 0.2618, "step": 3418 }, { "epoch": 0.74, "learning_rate": 3.390133095808804e-06, "loss": 0.2832, "step": 3419 }, { "epoch": 0.74, "learning_rate": 3.384888953280697e-06, "loss": 0.2651, "step": 3420 }, { "epoch": 0.74, "learning_rate": 3.3796480434922453e-06, "loss": 0.2479, "step": 3421 }, { "epoch": 0.74, "learning_rate": 3.374410369004625e-06, "loss": 0.252, "step": 3422 }, { "epoch": 0.74, "learning_rate": 3.3691759323774466e-06, "loss": 0.3295, "step": 3423 }, { "epoch": 0.74, "learning_rate": 3.363944736168718e-06, "loss": 0.2867, "step": 3424 }, { "epoch": 0.74, "learning_rate": 3.3587167829348743e-06, "loss": 0.2988, "step": 3425 }, { "epoch": 0.74, "learning_rate": 3.353492075230773e-06, "loss": 0.3211, "step": 3426 }, { "epoch": 0.74, "learning_rate": 3.3482706156096655e-06, "loss": 0.3097, "step": 3427 }, { "epoch": 0.74, "learning_rate": 3.343052406623236e-06, "loss": 0.2722, "step": 3428 }, { "epoch": 0.74, "learning_rate": 3.3378374508215704e-06, "loss": 0.2613, "step": 3429 }, { "epoch": 0.74, "learning_rate": 3.3326257507531657e-06, "loss": 0.2913, "step": 3430 }, { "epoch": 0.74, "learning_rate": 3.3274173089649274e-06, "loss": 0.303, "step": 3431 }, { "epoch": 0.74, "learning_rate": 3.3222121280021715e-06, "loss": 0.3352, "step": 3432 }, { "epoch": 0.74, "learning_rate": 3.3170102104086165e-06, "loss": 0.2801, "step": 3433 }, { "epoch": 0.74, "learning_rate": 3.311811558726391e-06, "loss": 0.3077, "step": 3434 }, { "epoch": 0.74, "learning_rate": 3.3066161754960222e-06, "loss": 0.2839, "step": 3435 }, { "epoch": 0.74, "learning_rate": 3.3014240632564443e-06, "loss": 0.3125, "step": 3436 }, { "epoch": 0.74, "learning_rate": 3.2962352245449913e-06, "loss": 0.3087, "step": 3437 }, { "epoch": 0.74, "learning_rate": 3.2910496618973975e-06, "loss": 0.2739, "step": 3438 }, { "epoch": 0.74, "learning_rate": 3.285867377847792e-06, "loss": 0.3465, "step": 3439 }, { "epoch": 0.74, "learning_rate": 3.2806883749287176e-06, "loss": 0.3374, "step": 3440 }, { "epoch": 0.74, "learning_rate": 3.2755126556710916e-06, "loss": 0.2756, "step": 3441 }, { "epoch": 0.74, "learning_rate": 3.2703402226042367e-06, "loss": 0.3195, "step": 3442 }, { "epoch": 0.74, "learning_rate": 3.2651710782558798e-06, "loss": 0.2814, "step": 3443 }, { "epoch": 0.74, "learning_rate": 3.2600052251521186e-06, "loss": 0.3377, "step": 3444 }, { "epoch": 0.74, "learning_rate": 3.2548426658174637e-06, "loss": 0.2862, "step": 3445 }, { "epoch": 0.74, "learning_rate": 3.2496834027748036e-06, "loss": 0.3517, "step": 3446 }, { "epoch": 0.74, "learning_rate": 3.2445274385454196e-06, "loss": 0.2598, "step": 3447 }, { "epoch": 0.74, "learning_rate": 3.2393747756489847e-06, "loss": 0.2824, "step": 3448 }, { "epoch": 0.74, "learning_rate": 3.2342254166035438e-06, "loss": 0.295, "step": 3449 }, { "epoch": 0.74, "learning_rate": 3.229079363925547e-06, "loss": 0.3213, "step": 3450 }, { "epoch": 0.74, "learning_rate": 3.2239366201298184e-06, "loss": 0.2821, "step": 3451 }, { "epoch": 0.75, "learning_rate": 3.218797187729565e-06, "loss": 0.2465, "step": 3452 }, { "epoch": 0.75, "learning_rate": 3.213661069236378e-06, "loss": 0.2945, "step": 3453 }, { "epoch": 0.75, "learning_rate": 3.2085282671602256e-06, "loss": 0.279, "step": 3454 }, { "epoch": 0.75, "learning_rate": 3.2033987840094604e-06, "loss": 0.2604, "step": 3455 }, { "epoch": 0.75, "learning_rate": 3.1982726222908046e-06, "loss": 0.2485, "step": 3456 }, { "epoch": 0.75, "learning_rate": 3.1931497845093753e-06, "loss": 0.3263, "step": 3457 }, { "epoch": 0.75, "learning_rate": 3.1880302731686373e-06, "loss": 0.2791, "step": 3458 }, { "epoch": 0.75, "learning_rate": 3.182914090770456e-06, "loss": 0.3542, "step": 3459 }, { "epoch": 0.75, "learning_rate": 3.1778012398150602e-06, "loss": 0.2898, "step": 3460 }, { "epoch": 0.75, "learning_rate": 3.1726917228010403e-06, "loss": 0.2738, "step": 3461 }, { "epoch": 0.75, "learning_rate": 3.1675855422253788e-06, "loss": 0.2607, "step": 3462 }, { "epoch": 0.75, "learning_rate": 3.1624827005834026e-06, "loss": 0.3008, "step": 3463 }, { "epoch": 0.75, "learning_rate": 3.157383200368832e-06, "loss": 0.2603, "step": 3464 }, { "epoch": 0.75, "learning_rate": 3.1522870440737362e-06, "loss": 0.3424, "step": 3465 }, { "epoch": 0.75, "learning_rate": 3.14719423418856e-06, "loss": 0.2781, "step": 3466 }, { "epoch": 0.75, "learning_rate": 3.142104773202107e-06, "loss": 0.296, "step": 3467 }, { "epoch": 0.75, "learning_rate": 3.1370186636015486e-06, "loss": 0.2961, "step": 3468 }, { "epoch": 0.75, "learning_rate": 3.1319359078724142e-06, "loss": 0.2569, "step": 3469 }, { "epoch": 0.75, "learning_rate": 3.126856508498599e-06, "loss": 0.3089, "step": 3470 }, { "epoch": 0.75, "learning_rate": 3.121780467962353e-06, "loss": 0.3209, "step": 3471 }, { "epoch": 0.75, "learning_rate": 3.11670778874429e-06, "loss": 0.2137, "step": 3472 }, { "epoch": 0.75, "learning_rate": 3.111638473323372e-06, "loss": 0.2816, "step": 3473 }, { "epoch": 0.75, "learning_rate": 3.106572524176936e-06, "loss": 0.2543, "step": 3474 }, { "epoch": 0.75, "learning_rate": 3.1015099437806464e-06, "loss": 0.2862, "step": 3475 }, { "epoch": 0.75, "learning_rate": 3.0964507346085514e-06, "loss": 0.3135, "step": 3476 }, { "epoch": 0.75, "learning_rate": 3.0913948991330257e-06, "loss": 0.3059, "step": 3477 }, { "epoch": 0.75, "learning_rate": 3.0863424398248067e-06, "loss": 0.2707, "step": 3478 }, { "epoch": 0.75, "learning_rate": 3.0812933591529904e-06, "loss": 0.322, "step": 3479 }, { "epoch": 0.75, "learning_rate": 3.0762476595850013e-06, "loss": 0.2979, "step": 3480 }, { "epoch": 0.75, "learning_rate": 3.0712053435866317e-06, "loss": 0.3556, "step": 3481 }, { "epoch": 0.75, "learning_rate": 3.0661664136220106e-06, "loss": 0.2736, "step": 3482 }, { "epoch": 0.75, "learning_rate": 3.0611308721536102e-06, "loss": 0.2892, "step": 3483 }, { "epoch": 0.75, "learning_rate": 3.0560987216422523e-06, "loss": 0.3166, "step": 3484 }, { "epoch": 0.75, "learning_rate": 3.0510699645470988e-06, "loss": 0.3435, "step": 3485 }, { "epoch": 0.75, "learning_rate": 3.046044603325654e-06, "loss": 0.2515, "step": 3486 }, { "epoch": 0.75, "learning_rate": 3.0410226404337585e-06, "loss": 0.3308, "step": 3487 }, { "epoch": 0.75, "learning_rate": 3.036004078325607e-06, "loss": 0.2861, "step": 3488 }, { "epoch": 0.75, "learning_rate": 3.0309889194537102e-06, "loss": 0.3042, "step": 3489 }, { "epoch": 0.75, "learning_rate": 3.025977166268932e-06, "loss": 0.3, "step": 3490 }, { "epoch": 0.75, "learning_rate": 3.0209688212204647e-06, "loss": 0.2597, "step": 3491 }, { "epoch": 0.75, "learning_rate": 3.015963886755836e-06, "loss": 0.2919, "step": 3492 }, { "epoch": 0.75, "learning_rate": 3.0109623653209164e-06, "loss": 0.2566, "step": 3493 }, { "epoch": 0.75, "learning_rate": 3.005964259359888e-06, "loss": 0.2898, "step": 3494 }, { "epoch": 0.75, "learning_rate": 3.000969571315285e-06, "loss": 0.2881, "step": 3495 }, { "epoch": 0.75, "learning_rate": 2.995978303627963e-06, "loss": 0.3118, "step": 3496 }, { "epoch": 0.75, "learning_rate": 2.9909904587370954e-06, "loss": 0.3409, "step": 3497 }, { "epoch": 0.75, "learning_rate": 2.9860060390802024e-06, "loss": 0.248, "step": 3498 }, { "epoch": 0.76, "learning_rate": 2.981025047093118e-06, "loss": 0.2531, "step": 3499 }, { "epoch": 0.76, "learning_rate": 2.976047485210004e-06, "loss": 0.2904, "step": 3500 }, { "epoch": 0.76, "learning_rate": 2.9710733558633454e-06, "loss": 0.3748, "step": 3501 }, { "epoch": 0.76, "learning_rate": 2.9661026614839507e-06, "loss": 0.243, "step": 3502 }, { "epoch": 0.76, "learning_rate": 2.961135404500949e-06, "loss": 0.3282, "step": 3503 }, { "epoch": 0.76, "learning_rate": 2.9561715873417906e-06, "loss": 0.2984, "step": 3504 }, { "epoch": 0.76, "learning_rate": 2.9512112124322424e-06, "loss": 0.2973, "step": 3505 }, { "epoch": 0.76, "learning_rate": 2.9462542821963922e-06, "loss": 0.2758, "step": 3506 }, { "epoch": 0.76, "learning_rate": 2.9413007990566434e-06, "loss": 0.2539, "step": 3507 }, { "epoch": 0.76, "learning_rate": 2.936350765433713e-06, "loss": 0.3042, "step": 3508 }, { "epoch": 0.76, "learning_rate": 2.9314041837466325e-06, "loss": 0.2576, "step": 3509 }, { "epoch": 0.76, "learning_rate": 2.926461056412755e-06, "loss": 0.2739, "step": 3510 }, { "epoch": 0.76, "learning_rate": 2.9215213858477265e-06, "loss": 0.2738, "step": 3511 }, { "epoch": 0.76, "learning_rate": 2.916585174465525e-06, "loss": 0.2818, "step": 3512 }, { "epoch": 0.76, "learning_rate": 2.911652424678425e-06, "loss": 0.2969, "step": 3513 }, { "epoch": 0.76, "learning_rate": 2.906723138897013e-06, "loss": 0.2641, "step": 3514 }, { "epoch": 0.76, "learning_rate": 2.901797319530183e-06, "loss": 0.2686, "step": 3515 }, { "epoch": 0.76, "learning_rate": 2.8968749689851328e-06, "loss": 0.3527, "step": 3516 }, { "epoch": 0.76, "learning_rate": 2.8919560896673682e-06, "loss": 0.3236, "step": 3517 }, { "epoch": 0.76, "learning_rate": 2.8870406839806964e-06, "loss": 0.3211, "step": 3518 }, { "epoch": 0.76, "learning_rate": 2.882128754327227e-06, "loss": 0.3295, "step": 3519 }, { "epoch": 0.76, "learning_rate": 2.877220303107373e-06, "loss": 0.2723, "step": 3520 }, { "epoch": 0.76, "learning_rate": 2.8723153327198473e-06, "loss": 0.2832, "step": 3521 }, { "epoch": 0.76, "learning_rate": 2.8674138455616574e-06, "loss": 0.2973, "step": 3522 }, { "epoch": 0.76, "learning_rate": 2.862515844028111e-06, "loss": 0.2538, "step": 3523 }, { "epoch": 0.76, "learning_rate": 2.857621330512822e-06, "loss": 0.2906, "step": 3524 }, { "epoch": 0.76, "learning_rate": 2.8527303074076818e-06, "loss": 0.2688, "step": 3525 }, { "epoch": 0.76, "learning_rate": 2.847842777102885e-06, "loss": 0.3229, "step": 3526 }, { "epoch": 0.76, "learning_rate": 2.8429587419869288e-06, "loss": 0.274, "step": 3527 }, { "epoch": 0.76, "learning_rate": 2.8380782044465817e-06, "loss": 0.3137, "step": 3528 }, { "epoch": 0.76, "learning_rate": 2.8332011668669213e-06, "loss": 0.3073, "step": 3529 }, { "epoch": 0.76, "learning_rate": 2.828327631631308e-06, "loss": 0.2751, "step": 3530 }, { "epoch": 0.76, "learning_rate": 2.823457601121388e-06, "loss": 0.252, "step": 3531 }, { "epoch": 0.76, "learning_rate": 2.8185910777171e-06, "loss": 0.3118, "step": 3532 }, { "epoch": 0.76, "learning_rate": 2.813728063796659e-06, "loss": 0.2798, "step": 3533 }, { "epoch": 0.76, "learning_rate": 2.808868561736581e-06, "loss": 0.251, "step": 3534 }, { "epoch": 0.76, "learning_rate": 2.8040125739116518e-06, "loss": 0.2915, "step": 3535 }, { "epoch": 0.76, "learning_rate": 2.7991601026949466e-06, "loss": 0.2696, "step": 3536 }, { "epoch": 0.76, "learning_rate": 2.7943111504578193e-06, "loss": 0.2764, "step": 3537 }, { "epoch": 0.76, "learning_rate": 2.7894657195699058e-06, "loss": 0.2951, "step": 3538 }, { "epoch": 0.76, "learning_rate": 2.7846238123991198e-06, "loss": 0.3169, "step": 3539 }, { "epoch": 0.76, "learning_rate": 2.7797854313116525e-06, "loss": 0.2736, "step": 3540 }, { "epoch": 0.76, "learning_rate": 2.77495057867198e-06, "loss": 0.3038, "step": 3541 }, { "epoch": 0.76, "learning_rate": 2.770119256842835e-06, "loss": 0.2649, "step": 3542 }, { "epoch": 0.76, "learning_rate": 2.765291468185248e-06, "loss": 0.2729, "step": 3543 }, { "epoch": 0.76, "learning_rate": 2.760467215058511e-06, "loss": 0.2585, "step": 3544 }, { "epoch": 0.77, "learning_rate": 2.7556464998201783e-06, "loss": 0.3146, "step": 3545 }, { "epoch": 0.77, "learning_rate": 2.7508293248261007e-06, "loss": 0.2247, "step": 3546 }, { "epoch": 0.77, "learning_rate": 2.7460156924303705e-06, "loss": 0.263, "step": 3547 }, { "epoch": 0.77, "learning_rate": 2.741205604985371e-06, "loss": 0.2809, "step": 3548 }, { "epoch": 0.77, "learning_rate": 2.7363990648417425e-06, "loss": 0.3144, "step": 3549 }, { "epoch": 0.77, "learning_rate": 2.731596074348395e-06, "loss": 0.3558, "step": 3550 }, { "epoch": 0.77, "learning_rate": 2.726796635852501e-06, "loss": 0.266, "step": 3551 }, { "epoch": 0.77, "learning_rate": 2.7220007516994995e-06, "loss": 0.3138, "step": 3552 }, { "epoch": 0.77, "learning_rate": 2.717208424233091e-06, "loss": 0.2156, "step": 3553 }, { "epoch": 0.77, "learning_rate": 2.7124196557952397e-06, "loss": 0.2207, "step": 3554 }, { "epoch": 0.77, "learning_rate": 2.7076344487261695e-06, "loss": 0.3258, "step": 3555 }, { "epoch": 0.77, "learning_rate": 2.7028528053643633e-06, "loss": 0.2724, "step": 3556 }, { "epoch": 0.77, "learning_rate": 2.6980747280465613e-06, "loss": 0.2585, "step": 3557 }, { "epoch": 0.77, "learning_rate": 2.6933002191077716e-06, "loss": 0.253, "step": 3558 }, { "epoch": 0.77, "learning_rate": 2.688529280881237e-06, "loss": 0.2913, "step": 3559 }, { "epoch": 0.77, "learning_rate": 2.6837619156984795e-06, "loss": 0.2862, "step": 3560 }, { "epoch": 0.77, "learning_rate": 2.6789981258892562e-06, "loss": 0.2915, "step": 3561 }, { "epoch": 0.77, "learning_rate": 2.674237913781583e-06, "loss": 0.2385, "step": 3562 }, { "epoch": 0.77, "learning_rate": 2.669481281701739e-06, "loss": 0.241, "step": 3563 }, { "epoch": 0.77, "learning_rate": 2.6647282319742307e-06, "loss": 0.2307, "step": 3564 }, { "epoch": 0.77, "learning_rate": 2.659978766921836e-06, "loss": 0.3067, "step": 3565 }, { "epoch": 0.77, "learning_rate": 2.6552328888655676e-06, "loss": 0.3373, "step": 3566 }, { "epoch": 0.77, "learning_rate": 2.650490600124691e-06, "loss": 0.302, "step": 3567 }, { "epoch": 0.77, "learning_rate": 2.6457519030167143e-06, "loss": 0.2645, "step": 3568 }, { "epoch": 0.77, "learning_rate": 2.6410167998573945e-06, "loss": 0.265, "step": 3569 }, { "epoch": 0.77, "learning_rate": 2.6362852929607273e-06, "loss": 0.2913, "step": 3570 }, { "epoch": 0.77, "learning_rate": 2.631557384638953e-06, "loss": 0.2996, "step": 3571 }, { "epoch": 0.77, "learning_rate": 2.6268330772025607e-06, "loss": 0.2698, "step": 3572 }, { "epoch": 0.77, "learning_rate": 2.622112372960265e-06, "loss": 0.2588, "step": 3573 }, { "epoch": 0.77, "learning_rate": 2.61739527421903e-06, "loss": 0.3414, "step": 3574 }, { "epoch": 0.77, "learning_rate": 2.612681783284057e-06, "loss": 0.308, "step": 3575 }, { "epoch": 0.77, "learning_rate": 2.607971902458777e-06, "loss": 0.3187, "step": 3576 }, { "epoch": 0.77, "learning_rate": 2.6032656340448735e-06, "loss": 0.2761, "step": 3577 }, { "epoch": 0.77, "learning_rate": 2.598562980342242e-06, "loss": 0.2697, "step": 3578 }, { "epoch": 0.77, "learning_rate": 2.5938639436490297e-06, "loss": 0.3342, "step": 3579 }, { "epoch": 0.77, "learning_rate": 2.5891685262616117e-06, "loss": 0.3045, "step": 3580 }, { "epoch": 0.77, "learning_rate": 2.584476730474582e-06, "loss": 0.3531, "step": 3581 }, { "epoch": 0.77, "learning_rate": 2.5797885585807847e-06, "loss": 0.3086, "step": 3582 }, { "epoch": 0.77, "learning_rate": 2.57510401287128e-06, "loss": 0.3182, "step": 3583 }, { "epoch": 0.77, "learning_rate": 2.5704230956353594e-06, "loss": 0.327, "step": 3584 }, { "epoch": 0.77, "learning_rate": 2.5657458091605414e-06, "loss": 0.2516, "step": 3585 }, { "epoch": 0.77, "learning_rate": 2.561072155732568e-06, "loss": 0.3246, "step": 3586 }, { "epoch": 0.77, "learning_rate": 2.556402137635411e-06, "loss": 0.2709, "step": 3587 }, { "epoch": 0.77, "learning_rate": 2.551735757151259e-06, "loss": 0.2976, "step": 3588 }, { "epoch": 0.77, "learning_rate": 2.5470730165605284e-06, "loss": 0.293, "step": 3589 }, { "epoch": 0.77, "learning_rate": 2.542413918141853e-06, "loss": 0.267, "step": 3590 }, { "epoch": 0.78, "learning_rate": 2.5377584641720897e-06, "loss": 0.2476, "step": 3591 }, { "epoch": 0.78, "learning_rate": 2.5331066569263118e-06, "loss": 0.3175, "step": 3592 }, { "epoch": 0.78, "learning_rate": 2.5284584986778106e-06, "loss": 0.2707, "step": 3593 }, { "epoch": 0.78, "learning_rate": 2.523813991698102e-06, "loss": 0.2509, "step": 3594 }, { "epoch": 0.78, "learning_rate": 2.5191731382569006e-06, "loss": 0.2906, "step": 3595 }, { "epoch": 0.78, "learning_rate": 2.5145359406221535e-06, "loss": 0.2439, "step": 3596 }, { "epoch": 0.78, "learning_rate": 2.5099024010600136e-06, "loss": 0.249, "step": 3597 }, { "epoch": 0.78, "learning_rate": 2.5052725218348396e-06, "loss": 0.3008, "step": 3598 }, { "epoch": 0.78, "learning_rate": 2.5006463052092133e-06, "loss": 0.3465, "step": 3599 }, { "epoch": 0.78, "learning_rate": 2.4960237534439202e-06, "loss": 0.3546, "step": 3600 }, { "epoch": 0.78, "learning_rate": 2.4914048687979553e-06, "loss": 0.2455, "step": 3601 }, { "epoch": 0.78, "learning_rate": 2.486789653528523e-06, "loss": 0.3296, "step": 3602 }, { "epoch": 0.78, "learning_rate": 2.4821781098910323e-06, "loss": 0.2461, "step": 3603 }, { "epoch": 0.78, "learning_rate": 2.4775702401391e-06, "loss": 0.2756, "step": 3604 }, { "epoch": 0.78, "learning_rate": 2.4729660465245465e-06, "loss": 0.3088, "step": 3605 }, { "epoch": 0.78, "learning_rate": 2.4683655312973953e-06, "loss": 0.2857, "step": 3606 }, { "epoch": 0.78, "learning_rate": 2.4637686967058706e-06, "loss": 0.3077, "step": 3607 }, { "epoch": 0.78, "learning_rate": 2.4591755449964084e-06, "loss": 0.2353, "step": 3608 }, { "epoch": 0.78, "learning_rate": 2.454586078413629e-06, "loss": 0.3672, "step": 3609 }, { "epoch": 0.78, "learning_rate": 2.450000299200359e-06, "loss": 0.2308, "step": 3610 }, { "epoch": 0.78, "learning_rate": 2.445418209597632e-06, "loss": 0.2934, "step": 3611 }, { "epoch": 0.78, "learning_rate": 2.4408398118446597e-06, "loss": 0.2857, "step": 3612 }, { "epoch": 0.78, "learning_rate": 2.436265108178867e-06, "loss": 0.2737, "step": 3613 }, { "epoch": 0.78, "learning_rate": 2.431694100835864e-06, "loss": 0.2849, "step": 3614 }, { "epoch": 0.78, "learning_rate": 2.427126792049458e-06, "loss": 0.2938, "step": 3615 }, { "epoch": 0.78, "learning_rate": 2.4225631840516506e-06, "loss": 0.2811, "step": 3616 }, { "epoch": 0.78, "learning_rate": 2.4180032790726226e-06, "loss": 0.3368, "step": 3617 }, { "epoch": 0.78, "learning_rate": 2.4134470793407637e-06, "loss": 0.2888, "step": 3618 }, { "epoch": 0.78, "learning_rate": 2.4088945870826406e-06, "loss": 0.2563, "step": 3619 }, { "epoch": 0.78, "learning_rate": 2.4043458045230116e-06, "loss": 0.359, "step": 3620 }, { "epoch": 0.78, "learning_rate": 2.3998007338848228e-06, "loss": 0.2801, "step": 3621 }, { "epoch": 0.78, "learning_rate": 2.3952593773892054e-06, "loss": 0.3131, "step": 3622 }, { "epoch": 0.78, "learning_rate": 2.3907217372554747e-06, "loss": 0.2384, "step": 3623 }, { "epoch": 0.78, "learning_rate": 2.3861878157011286e-06, "loss": 0.293, "step": 3624 }, { "epoch": 0.78, "learning_rate": 2.381657614941858e-06, "loss": 0.2574, "step": 3625 }, { "epoch": 0.78, "learning_rate": 2.377131137191516e-06, "loss": 0.2609, "step": 3626 }, { "epoch": 0.78, "learning_rate": 2.372608384662156e-06, "loss": 0.3082, "step": 3627 }, { "epoch": 0.78, "learning_rate": 2.3680893595640044e-06, "loss": 0.3052, "step": 3628 }, { "epoch": 0.78, "learning_rate": 2.363574064105454e-06, "loss": 0.2584, "step": 3629 }, { "epoch": 0.78, "learning_rate": 2.3590625004930968e-06, "loss": 0.2539, "step": 3630 }, { "epoch": 0.78, "learning_rate": 2.354554670931679e-06, "loss": 0.3262, "step": 3631 }, { "epoch": 0.78, "learning_rate": 2.35005057762414e-06, "loss": 0.2268, "step": 3632 }, { "epoch": 0.78, "learning_rate": 2.3455502227715844e-06, "loss": 0.298, "step": 3633 }, { "epoch": 0.78, "learning_rate": 2.3410536085732903e-06, "loss": 0.2809, "step": 3634 }, { "epoch": 0.78, "learning_rate": 2.3365607372267083e-06, "loss": 0.3216, "step": 3635 }, { "epoch": 0.78, "learning_rate": 2.332071610927461e-06, "loss": 0.2629, "step": 3636 }, { "epoch": 0.78, "learning_rate": 2.327586231869341e-06, "loss": 0.3402, "step": 3637 }, { "epoch": 0.79, "learning_rate": 2.3231046022443082e-06, "loss": 0.3052, "step": 3638 }, { "epoch": 0.79, "learning_rate": 2.318626724242491e-06, "loss": 0.2604, "step": 3639 }, { "epoch": 0.79, "learning_rate": 2.3141526000521842e-06, "loss": 0.2723, "step": 3640 }, { "epoch": 0.79, "learning_rate": 2.3096822318598444e-06, "loss": 0.2792, "step": 3641 }, { "epoch": 0.79, "learning_rate": 2.3052156218501075e-06, "loss": 0.2703, "step": 3642 }, { "epoch": 0.79, "learning_rate": 2.300752772205749e-06, "loss": 0.2778, "step": 3643 }, { "epoch": 0.79, "learning_rate": 2.29629368510773e-06, "loss": 0.2793, "step": 3644 }, { "epoch": 0.79, "learning_rate": 2.291838362735155e-06, "loss": 0.2719, "step": 3645 }, { "epoch": 0.79, "learning_rate": 2.287386807265297e-06, "loss": 0.2917, "step": 3646 }, { "epoch": 0.79, "learning_rate": 2.2829390208735945e-06, "loss": 0.3006, "step": 3647 }, { "epoch": 0.79, "learning_rate": 2.2784950057336265e-06, "loss": 0.3031, "step": 3648 }, { "epoch": 0.79, "learning_rate": 2.2740547640171463e-06, "loss": 0.3318, "step": 3649 }, { "epoch": 0.79, "learning_rate": 2.269618297894056e-06, "loss": 0.2671, "step": 3650 }, { "epoch": 0.79, "learning_rate": 2.265185609532411e-06, "loss": 0.2635, "step": 3651 }, { "epoch": 0.79, "learning_rate": 2.260756701098423e-06, "loss": 0.2904, "step": 3652 }, { "epoch": 0.79, "learning_rate": 2.2563315747564575e-06, "loss": 0.2605, "step": 3653 }, { "epoch": 0.79, "learning_rate": 2.251910232669028e-06, "loss": 0.283, "step": 3654 }, { "epoch": 0.79, "learning_rate": 2.2474926769967987e-06, "loss": 0.2978, "step": 3655 }, { "epoch": 0.79, "learning_rate": 2.2430789098985938e-06, "loss": 0.3204, "step": 3656 }, { "epoch": 0.79, "learning_rate": 2.2386689335313693e-06, "loss": 0.2803, "step": 3657 }, { "epoch": 0.79, "learning_rate": 2.234262750050241e-06, "loss": 0.2984, "step": 3658 }, { "epoch": 0.79, "learning_rate": 2.229860361608468e-06, "loss": 0.2553, "step": 3659 }, { "epoch": 0.79, "learning_rate": 2.225461770357449e-06, "loss": 0.3326, "step": 3660 }, { "epoch": 0.79, "learning_rate": 2.221066978446742e-06, "loss": 0.2642, "step": 3661 }, { "epoch": 0.79, "learning_rate": 2.2166759880240274e-06, "loss": 0.2802, "step": 3662 }, { "epoch": 0.79, "learning_rate": 2.2122888012351485e-06, "loss": 0.287, "step": 3663 }, { "epoch": 0.79, "learning_rate": 2.2079054202240777e-06, "loss": 0.2752, "step": 3664 }, { "epoch": 0.79, "learning_rate": 2.2035258471329236e-06, "loss": 0.2729, "step": 3665 }, { "epoch": 0.79, "learning_rate": 2.1991500841019474e-06, "loss": 0.3297, "step": 3666 }, { "epoch": 0.79, "learning_rate": 2.1947781332695406e-06, "loss": 0.3233, "step": 3667 }, { "epoch": 0.79, "learning_rate": 2.1904099967722315e-06, "loss": 0.28, "step": 3668 }, { "epoch": 0.79, "learning_rate": 2.186045676744686e-06, "loss": 0.2453, "step": 3669 }, { "epoch": 0.79, "learning_rate": 2.1816851753197023e-06, "loss": 0.254, "step": 3670 }, { "epoch": 0.79, "learning_rate": 2.177328494628217e-06, "loss": 0.251, "step": 3671 }, { "epoch": 0.79, "learning_rate": 2.172975636799296e-06, "loss": 0.274, "step": 3672 }, { "epoch": 0.79, "learning_rate": 2.1686266039601376e-06, "loss": 0.3041, "step": 3673 }, { "epoch": 0.79, "learning_rate": 2.1642813982360734e-06, "loss": 0.2769, "step": 3674 }, { "epoch": 0.79, "learning_rate": 2.1599400217505594e-06, "loss": 0.2581, "step": 3675 }, { "epoch": 0.79, "learning_rate": 2.1556024766251872e-06, "loss": 0.2913, "step": 3676 }, { "epoch": 0.79, "learning_rate": 2.151268764979666e-06, "loss": 0.268, "step": 3677 }, { "epoch": 0.79, "learning_rate": 2.1469388889318487e-06, "loss": 0.3364, "step": 3678 }, { "epoch": 0.79, "learning_rate": 2.1426128505976917e-06, "loss": 0.3186, "step": 3679 }, { "epoch": 0.79, "learning_rate": 2.138290652091295e-06, "loss": 0.2849, "step": 3680 }, { "epoch": 0.79, "learning_rate": 2.133972295524875e-06, "loss": 0.3169, "step": 3681 }, { "epoch": 0.79, "learning_rate": 2.1296577830087617e-06, "loss": 0.3354, "step": 3682 }, { "epoch": 0.79, "learning_rate": 2.125347116651426e-06, "loss": 0.2819, "step": 3683 }, { "epoch": 0.8, "learning_rate": 2.121040298559437e-06, "loss": 0.2712, "step": 3684 }, { "epoch": 0.8, "learning_rate": 2.116737330837502e-06, "loss": 0.2413, "step": 3685 }, { "epoch": 0.8, "learning_rate": 2.1124382155884373e-06, "loss": 0.2711, "step": 3686 }, { "epoch": 0.8, "learning_rate": 2.1081429549131773e-06, "loss": 0.314, "step": 3687 }, { "epoch": 0.8, "learning_rate": 2.1038515509107735e-06, "loss": 0.2863, "step": 3688 }, { "epoch": 0.8, "learning_rate": 2.0995640056783937e-06, "loss": 0.2771, "step": 3689 }, { "epoch": 0.8, "learning_rate": 2.0952803213113195e-06, "loss": 0.3382, "step": 3690 }, { "epoch": 0.8, "learning_rate": 2.09100049990294e-06, "loss": 0.291, "step": 3691 }, { "epoch": 0.8, "learning_rate": 2.0867245435447727e-06, "loss": 0.3068, "step": 3692 }, { "epoch": 0.8, "learning_rate": 2.082452454326426e-06, "loss": 0.337, "step": 3693 }, { "epoch": 0.8, "learning_rate": 2.078184234335627e-06, "loss": 0.2982, "step": 3694 }, { "epoch": 0.8, "learning_rate": 2.073919885658223e-06, "loss": 0.2278, "step": 3695 }, { "epoch": 0.8, "learning_rate": 2.0696594103781466e-06, "loss": 0.2538, "step": 3696 }, { "epoch": 0.8, "learning_rate": 2.0654028105774626e-06, "loss": 0.2723, "step": 3697 }, { "epoch": 0.8, "learning_rate": 2.0611500883363166e-06, "loss": 0.3579, "step": 3698 }, { "epoch": 0.8, "learning_rate": 2.0569012457329816e-06, "loss": 0.3538, "step": 3699 }, { "epoch": 0.8, "learning_rate": 2.0526562848438257e-06, "loss": 0.24, "step": 3700 }, { "epoch": 0.8, "learning_rate": 2.0484152077433095e-06, "loss": 0.2665, "step": 3701 }, { "epoch": 0.8, "learning_rate": 2.044178016504015e-06, "loss": 0.2883, "step": 3702 }, { "epoch": 0.8, "learning_rate": 2.0399447131966133e-06, "loss": 0.2843, "step": 3703 }, { "epoch": 0.8, "learning_rate": 2.0357152998898775e-06, "loss": 0.2654, "step": 3704 }, { "epoch": 0.8, "learning_rate": 2.0314897786506805e-06, "loss": 0.2807, "step": 3705 }, { "epoch": 0.8, "learning_rate": 2.0272681515439928e-06, "loss": 0.3108, "step": 3706 }, { "epoch": 0.8, "learning_rate": 2.0230504206328817e-06, "loss": 0.3281, "step": 3707 }, { "epoch": 0.8, "learning_rate": 2.018836587978509e-06, "loss": 0.2689, "step": 3708 }, { "epoch": 0.8, "learning_rate": 2.0146266556401405e-06, "loss": 0.2456, "step": 3709 }, { "epoch": 0.8, "learning_rate": 2.0104206256751213e-06, "loss": 0.3054, "step": 3710 }, { "epoch": 0.8, "learning_rate": 2.006218500138899e-06, "loss": 0.2848, "step": 3711 }, { "epoch": 0.8, "learning_rate": 2.002020281085011e-06, "loss": 0.3353, "step": 3712 }, { "epoch": 0.8, "learning_rate": 1.9978259705650825e-06, "loss": 0.2637, "step": 3713 }, { "epoch": 0.8, "learning_rate": 1.9936355706288414e-06, "loss": 0.2733, "step": 3714 }, { "epoch": 0.8, "learning_rate": 1.9894490833240822e-06, "loss": 0.2672, "step": 3715 }, { "epoch": 0.8, "learning_rate": 1.9852665106967093e-06, "loss": 0.2624, "step": 3716 }, { "epoch": 0.8, "learning_rate": 1.981087854790702e-06, "loss": 0.2989, "step": 3717 }, { "epoch": 0.8, "learning_rate": 1.976913117648128e-06, "loss": 0.2763, "step": 3718 }, { "epoch": 0.8, "learning_rate": 1.9727423013091406e-06, "loss": 0.2985, "step": 3719 }, { "epoch": 0.8, "learning_rate": 1.9685754078119767e-06, "loss": 0.2419, "step": 3720 }, { "epoch": 0.8, "learning_rate": 1.9644124391929555e-06, "loss": 0.289, "step": 3721 }, { "epoch": 0.8, "learning_rate": 1.960253397486479e-06, "loss": 0.3121, "step": 3722 }, { "epoch": 0.8, "learning_rate": 1.956098284725031e-06, "loss": 0.2917, "step": 3723 }, { "epoch": 0.8, "learning_rate": 1.9519471029391722e-06, "loss": 0.2788, "step": 3724 }, { "epoch": 0.8, "learning_rate": 1.947799854157547e-06, "loss": 0.2979, "step": 3725 }, { "epoch": 0.8, "learning_rate": 1.943656540406872e-06, "loss": 0.2417, "step": 3726 }, { "epoch": 0.8, "learning_rate": 1.939517163711944e-06, "loss": 0.2699, "step": 3727 }, { "epoch": 0.8, "learning_rate": 1.9353817260956407e-06, "loss": 0.2877, "step": 3728 }, { "epoch": 0.8, "learning_rate": 1.931250229578905e-06, "loss": 0.3352, "step": 3729 }, { "epoch": 0.81, "learning_rate": 1.927122676180756e-06, "loss": 0.2421, "step": 3730 }, { "epoch": 0.81, "learning_rate": 1.9229990679182974e-06, "loss": 0.2791, "step": 3731 }, { "epoch": 0.81, "learning_rate": 1.918879406806685e-06, "loss": 0.3349, "step": 3732 }, { "epoch": 0.81, "learning_rate": 1.914763694859165e-06, "loss": 0.2794, "step": 3733 }, { "epoch": 0.81, "learning_rate": 1.9106519340870433e-06, "loss": 0.2467, "step": 3734 }, { "epoch": 0.81, "learning_rate": 1.9065441264996963e-06, "loss": 0.2676, "step": 3735 }, { "epoch": 0.81, "learning_rate": 1.9024402741045679e-06, "loss": 0.2971, "step": 3736 }, { "epoch": 0.81, "learning_rate": 1.898340378907172e-06, "loss": 0.2698, "step": 3737 }, { "epoch": 0.81, "learning_rate": 1.8942444429110873e-06, "loss": 0.2691, "step": 3738 }, { "epoch": 0.81, "learning_rate": 1.8901524681179551e-06, "loss": 0.2871, "step": 3739 }, { "epoch": 0.81, "learning_rate": 1.8860644565274856e-06, "loss": 0.2904, "step": 3740 }, { "epoch": 0.81, "learning_rate": 1.8819804101374483e-06, "loss": 0.278, "step": 3741 }, { "epoch": 0.81, "learning_rate": 1.8779003309436783e-06, "loss": 0.2838, "step": 3742 }, { "epoch": 0.81, "learning_rate": 1.873824220940068e-06, "loss": 0.2836, "step": 3743 }, { "epoch": 0.81, "learning_rate": 1.869752082118571e-06, "loss": 0.2389, "step": 3744 }, { "epoch": 0.81, "learning_rate": 1.8656839164692087e-06, "loss": 0.2978, "step": 3745 }, { "epoch": 0.81, "learning_rate": 1.8616197259800427e-06, "loss": 0.303, "step": 3746 }, { "epoch": 0.81, "learning_rate": 1.8575595126372126e-06, "loss": 0.2637, "step": 3747 }, { "epoch": 0.81, "learning_rate": 1.853503278424903e-06, "loss": 0.2536, "step": 3748 }, { "epoch": 0.81, "learning_rate": 1.8494510253253472e-06, "loss": 0.2819, "step": 3749 }, { "epoch": 0.81, "learning_rate": 1.8454027553188515e-06, "loss": 0.2629, "step": 3750 }, { "epoch": 0.81, "learning_rate": 1.8413584703837618e-06, "loss": 0.2643, "step": 3751 }, { "epoch": 0.81, "learning_rate": 1.8373181724964795e-06, "loss": 0.3087, "step": 3752 }, { "epoch": 0.81, "learning_rate": 1.8332818636314597e-06, "loss": 0.262, "step": 3753 }, { "epoch": 0.81, "learning_rate": 1.829249545761207e-06, "loss": 0.2595, "step": 3754 }, { "epoch": 0.81, "learning_rate": 1.8252212208562737e-06, "loss": 0.2941, "step": 3755 }, { "epoch": 0.81, "learning_rate": 1.821196890885264e-06, "loss": 0.2758, "step": 3756 }, { "epoch": 0.81, "learning_rate": 1.8171765578148283e-06, "loss": 0.3016, "step": 3757 }, { "epoch": 0.81, "learning_rate": 1.8131602236096624e-06, "loss": 0.2786, "step": 3758 }, { "epoch": 0.81, "learning_rate": 1.809147890232511e-06, "loss": 0.3269, "step": 3759 }, { "epoch": 0.81, "learning_rate": 1.8051395596441613e-06, "loss": 0.3007, "step": 3760 }, { "epoch": 0.81, "learning_rate": 1.8011352338034427e-06, "loss": 0.2992, "step": 3761 }, { "epoch": 0.81, "learning_rate": 1.7971349146672368e-06, "loss": 0.3551, "step": 3762 }, { "epoch": 0.81, "learning_rate": 1.7931386041904508e-06, "loss": 0.2875, "step": 3763 }, { "epoch": 0.81, "learning_rate": 1.7891463043260492e-06, "loss": 0.2425, "step": 3764 }, { "epoch": 0.81, "learning_rate": 1.7851580170250304e-06, "loss": 0.2971, "step": 3765 }, { "epoch": 0.81, "learning_rate": 1.781173744236422e-06, "loss": 0.2704, "step": 3766 }, { "epoch": 0.81, "learning_rate": 1.7771934879073115e-06, "loss": 0.2822, "step": 3767 }, { "epoch": 0.81, "learning_rate": 1.7732172499827982e-06, "loss": 0.2847, "step": 3768 }, { "epoch": 0.81, "learning_rate": 1.7692450324060407e-06, "loss": 0.3201, "step": 3769 }, { "epoch": 0.81, "learning_rate": 1.7652768371182172e-06, "loss": 0.2415, "step": 3770 }, { "epoch": 0.81, "learning_rate": 1.7613126660585488e-06, "loss": 0.3027, "step": 3771 }, { "epoch": 0.81, "learning_rate": 1.7573525211642839e-06, "loss": 0.3136, "step": 3772 }, { "epoch": 0.81, "learning_rate": 1.7533964043707085e-06, "loss": 0.2958, "step": 3773 }, { "epoch": 0.81, "learning_rate": 1.7494443176111375e-06, "loss": 0.3303, "step": 3774 }, { "epoch": 0.81, "learning_rate": 1.7454962628169125e-06, "loss": 0.2905, "step": 3775 }, { "epoch": 0.81, "learning_rate": 1.7415522419174191e-06, "loss": 0.2866, "step": 3776 }, { "epoch": 0.82, "learning_rate": 1.7376122568400533e-06, "loss": 0.2475, "step": 3777 }, { "epoch": 0.82, "learning_rate": 1.733676309510245e-06, "loss": 0.2823, "step": 3778 }, { "epoch": 0.82, "learning_rate": 1.729744401851463e-06, "loss": 0.2788, "step": 3779 }, { "epoch": 0.82, "learning_rate": 1.725816535785181e-06, "loss": 0.3173, "step": 3780 }, { "epoch": 0.82, "learning_rate": 1.7218927132309182e-06, "loss": 0.2806, "step": 3781 }, { "epoch": 0.82, "learning_rate": 1.717972936106198e-06, "loss": 0.2822, "step": 3782 }, { "epoch": 0.82, "learning_rate": 1.7140572063265848e-06, "loss": 0.2467, "step": 3783 }, { "epoch": 0.82, "learning_rate": 1.7101455258056565e-06, "loss": 0.2648, "step": 3784 }, { "epoch": 0.82, "learning_rate": 1.7062378964550063e-06, "loss": 0.3051, "step": 3785 }, { "epoch": 0.82, "learning_rate": 1.7023343201842612e-06, "loss": 0.3023, "step": 3786 }, { "epoch": 0.82, "learning_rate": 1.6984347989010585e-06, "loss": 0.2641, "step": 3787 }, { "epoch": 0.82, "learning_rate": 1.694539334511055e-06, "loss": 0.2265, "step": 3788 }, { "epoch": 0.82, "learning_rate": 1.6906479289179268e-06, "loss": 0.2598, "step": 3789 }, { "epoch": 0.82, "learning_rate": 1.6867605840233635e-06, "loss": 0.2635, "step": 3790 }, { "epoch": 0.82, "learning_rate": 1.682877301727075e-06, "loss": 0.3501, "step": 3791 }, { "epoch": 0.82, "learning_rate": 1.6789980839267795e-06, "loss": 0.2606, "step": 3792 }, { "epoch": 0.82, "learning_rate": 1.6751229325182194e-06, "loss": 0.2532, "step": 3793 }, { "epoch": 0.82, "learning_rate": 1.6712518493951357e-06, "loss": 0.2901, "step": 3794 }, { "epoch": 0.82, "learning_rate": 1.6673848364492916e-06, "loss": 0.2729, "step": 3795 }, { "epoch": 0.82, "learning_rate": 1.663521895570459e-06, "loss": 0.2817, "step": 3796 }, { "epoch": 0.82, "learning_rate": 1.6596630286464154e-06, "loss": 0.274, "step": 3797 }, { "epoch": 0.82, "learning_rate": 1.655808237562958e-06, "loss": 0.3071, "step": 3798 }, { "epoch": 0.82, "learning_rate": 1.6519575242038778e-06, "loss": 0.2623, "step": 3799 }, { "epoch": 0.82, "learning_rate": 1.6481108904509858e-06, "loss": 0.2608, "step": 3800 }, { "epoch": 0.82, "learning_rate": 1.6442683381840941e-06, "loss": 0.2847, "step": 3801 }, { "epoch": 0.82, "learning_rate": 1.6404298692810127e-06, "loss": 0.2244, "step": 3802 }, { "epoch": 0.82, "learning_rate": 1.6365954856175725e-06, "loss": 0.3066, "step": 3803 }, { "epoch": 0.82, "learning_rate": 1.6327651890675954e-06, "loss": 0.2727, "step": 3804 }, { "epoch": 0.82, "learning_rate": 1.62893898150291e-06, "loss": 0.2907, "step": 3805 }, { "epoch": 0.82, "learning_rate": 1.6251168647933468e-06, "loss": 0.2662, "step": 3806 }, { "epoch": 0.82, "learning_rate": 1.6212988408067354e-06, "loss": 0.2864, "step": 3807 }, { "epoch": 0.82, "learning_rate": 1.6174849114089064e-06, "loss": 0.2866, "step": 3808 }, { "epoch": 0.82, "learning_rate": 1.6136750784636912e-06, "loss": 0.2764, "step": 3809 }, { "epoch": 0.82, "learning_rate": 1.609869343832916e-06, "loss": 0.2406, "step": 3810 }, { "epoch": 0.82, "learning_rate": 1.6060677093764043e-06, "loss": 0.3098, "step": 3811 }, { "epoch": 0.82, "learning_rate": 1.6022701769519855e-06, "loss": 0.2771, "step": 3812 }, { "epoch": 0.82, "learning_rate": 1.5984767484154673e-06, "loss": 0.2873, "step": 3813 }, { "epoch": 0.82, "learning_rate": 1.5946874256206612e-06, "loss": 0.2266, "step": 3814 }, { "epoch": 0.82, "learning_rate": 1.5909022104193805e-06, "loss": 0.2923, "step": 3815 }, { "epoch": 0.82, "learning_rate": 1.587121104661411e-06, "loss": 0.3222, "step": 3816 }, { "epoch": 0.82, "learning_rate": 1.5833441101945513e-06, "loss": 0.2697, "step": 3817 }, { "epoch": 0.82, "learning_rate": 1.579571228864578e-06, "loss": 0.3069, "step": 3818 }, { "epoch": 0.82, "learning_rate": 1.5758024625152601e-06, "loss": 0.2783, "step": 3819 }, { "epoch": 0.82, "learning_rate": 1.5720378129883584e-06, "loss": 0.2754, "step": 3820 }, { "epoch": 0.82, "learning_rate": 1.5682772821236192e-06, "loss": 0.247, "step": 3821 }, { "epoch": 0.82, "learning_rate": 1.5645208717587768e-06, "loss": 0.3028, "step": 3822 }, { "epoch": 0.83, "learning_rate": 1.5607685837295516e-06, "loss": 0.2916, "step": 3823 }, { "epoch": 0.83, "learning_rate": 1.5570204198696504e-06, "loss": 0.2385, "step": 3824 }, { "epoch": 0.83, "learning_rate": 1.5532763820107644e-06, "loss": 0.3104, "step": 3825 }, { "epoch": 0.83, "learning_rate": 1.5495364719825677e-06, "loss": 0.3027, "step": 3826 }, { "epoch": 0.83, "learning_rate": 1.5458006916127167e-06, "loss": 0.27, "step": 3827 }, { "epoch": 0.83, "learning_rate": 1.5420690427268493e-06, "loss": 0.3509, "step": 3828 }, { "epoch": 0.83, "learning_rate": 1.5383415271485913e-06, "loss": 0.2764, "step": 3829 }, { "epoch": 0.83, "learning_rate": 1.534618146699538e-06, "loss": 0.2954, "step": 3830 }, { "epoch": 0.83, "learning_rate": 1.530898903199266e-06, "loss": 0.2745, "step": 3831 }, { "epoch": 0.83, "learning_rate": 1.5271837984653426e-06, "loss": 0.3019, "step": 3832 }, { "epoch": 0.83, "learning_rate": 1.5234728343132921e-06, "loss": 0.2876, "step": 3833 }, { "epoch": 0.83, "learning_rate": 1.5197660125566328e-06, "loss": 0.2337, "step": 3834 }, { "epoch": 0.83, "learning_rate": 1.516063335006851e-06, "loss": 0.3084, "step": 3835 }, { "epoch": 0.83, "learning_rate": 1.5123648034734084e-06, "loss": 0.2939, "step": 3836 }, { "epoch": 0.83, "learning_rate": 1.5086704197637404e-06, "loss": 0.275, "step": 3837 }, { "epoch": 0.83, "learning_rate": 1.5049801856832557e-06, "loss": 0.2481, "step": 3838 }, { "epoch": 0.83, "learning_rate": 1.5012941030353368e-06, "loss": 0.2853, "step": 3839 }, { "epoch": 0.83, "learning_rate": 1.4976121736213345e-06, "loss": 0.3612, "step": 3840 }, { "epoch": 0.83, "learning_rate": 1.493934399240572e-06, "loss": 0.3113, "step": 3841 }, { "epoch": 0.83, "learning_rate": 1.4902607816903415e-06, "loss": 0.2984, "step": 3842 }, { "epoch": 0.83, "learning_rate": 1.4865913227659034e-06, "loss": 0.3463, "step": 3843 }, { "epoch": 0.83, "learning_rate": 1.482926024260487e-06, "loss": 0.221, "step": 3844 }, { "epoch": 0.83, "learning_rate": 1.4792648879652849e-06, "loss": 0.2831, "step": 3845 }, { "epoch": 0.83, "learning_rate": 1.4756079156694657e-06, "loss": 0.2471, "step": 3846 }, { "epoch": 0.83, "learning_rate": 1.471955109160146e-06, "loss": 0.2885, "step": 3847 }, { "epoch": 0.83, "learning_rate": 1.4683064702224225e-06, "loss": 0.2976, "step": 3848 }, { "epoch": 0.83, "learning_rate": 1.4646620006393497e-06, "loss": 0.3015, "step": 3849 }, { "epoch": 0.83, "learning_rate": 1.4610217021919381e-06, "loss": 0.2487, "step": 3850 }, { "epoch": 0.83, "learning_rate": 1.4573855766591737e-06, "loss": 0.3393, "step": 3851 }, { "epoch": 0.83, "learning_rate": 1.4537536258179873e-06, "loss": 0.2918, "step": 3852 }, { "epoch": 0.83, "learning_rate": 1.4501258514432836e-06, "loss": 0.2545, "step": 3853 }, { "epoch": 0.83, "learning_rate": 1.4465022553079167e-06, "loss": 0.2775, "step": 3854 }, { "epoch": 0.83, "learning_rate": 1.4428828391827055e-06, "loss": 0.317, "step": 3855 }, { "epoch": 0.83, "learning_rate": 1.4392676048364196e-06, "loss": 0.3536, "step": 3856 }, { "epoch": 0.83, "learning_rate": 1.4356565540357914e-06, "loss": 0.2716, "step": 3857 }, { "epoch": 0.83, "learning_rate": 1.4320496885455037e-06, "loss": 0.2512, "step": 3858 }, { "epoch": 0.83, "learning_rate": 1.4284470101281944e-06, "loss": 0.2698, "step": 3859 }, { "epoch": 0.83, "learning_rate": 1.424848520544465e-06, "loss": 0.3265, "step": 3860 }, { "epoch": 0.83, "learning_rate": 1.4212542215528534e-06, "loss": 0.3185, "step": 3861 }, { "epoch": 0.83, "learning_rate": 1.4176641149098591e-06, "loss": 0.2399, "step": 3862 }, { "epoch": 0.83, "learning_rate": 1.4140782023699396e-06, "loss": 0.2416, "step": 3863 }, { "epoch": 0.83, "learning_rate": 1.4104964856854842e-06, "loss": 0.2544, "step": 3864 }, { "epoch": 0.83, "learning_rate": 1.4069189666068517e-06, "loss": 0.2908, "step": 3865 }, { "epoch": 0.83, "learning_rate": 1.4033456468823326e-06, "loss": 0.3029, "step": 3866 }, { "epoch": 0.83, "learning_rate": 1.39977652825818e-06, "loss": 0.3218, "step": 3867 }, { "epoch": 0.83, "learning_rate": 1.3962116124785864e-06, "loss": 0.2578, "step": 3868 }, { "epoch": 0.84, "learning_rate": 1.3926509012856825e-06, "loss": 0.3038, "step": 3869 }, { "epoch": 0.84, "learning_rate": 1.3890943964195635e-06, "loss": 0.3117, "step": 3870 }, { "epoch": 0.84, "learning_rate": 1.3855420996182534e-06, "loss": 0.2816, "step": 3871 }, { "epoch": 0.84, "learning_rate": 1.3819940126177246e-06, "loss": 0.3195, "step": 3872 }, { "epoch": 0.84, "learning_rate": 1.3784501371518932e-06, "loss": 0.2662, "step": 3873 }, { "epoch": 0.84, "learning_rate": 1.3749104749526154e-06, "loss": 0.2933, "step": 3874 }, { "epoch": 0.84, "learning_rate": 1.3713750277496907e-06, "loss": 0.2611, "step": 3875 }, { "epoch": 0.84, "learning_rate": 1.3678437972708536e-06, "loss": 0.241, "step": 3876 }, { "epoch": 0.84, "learning_rate": 1.3643167852417894e-06, "loss": 0.2647, "step": 3877 }, { "epoch": 0.84, "learning_rate": 1.360793993386107e-06, "loss": 0.2662, "step": 3878 }, { "epoch": 0.84, "learning_rate": 1.3572754234253626e-06, "loss": 0.2815, "step": 3879 }, { "epoch": 0.84, "learning_rate": 1.3537610770790466e-06, "loss": 0.2881, "step": 3880 }, { "epoch": 0.84, "learning_rate": 1.3502509560645826e-06, "loss": 0.3298, "step": 3881 }, { "epoch": 0.84, "learning_rate": 1.34674506209734e-06, "loss": 0.2406, "step": 3882 }, { "epoch": 0.84, "learning_rate": 1.3432433968906045e-06, "loss": 0.3294, "step": 3883 }, { "epoch": 0.84, "learning_rate": 1.339745962155613e-06, "loss": 0.2246, "step": 3884 }, { "epoch": 0.84, "learning_rate": 1.3362527596015273e-06, "loss": 0.275, "step": 3885 }, { "epoch": 0.84, "learning_rate": 1.3327637909354351e-06, "loss": 0.286, "step": 3886 }, { "epoch": 0.84, "learning_rate": 1.329279057862366e-06, "loss": 0.2546, "step": 3887 }, { "epoch": 0.84, "learning_rate": 1.3257985620852741e-06, "loss": 0.2623, "step": 3888 }, { "epoch": 0.84, "learning_rate": 1.3223223053050428e-06, "loss": 0.2602, "step": 3889 }, { "epoch": 0.84, "learning_rate": 1.3188502892204846e-06, "loss": 0.3203, "step": 3890 }, { "epoch": 0.84, "learning_rate": 1.3153825155283395e-06, "loss": 0.2813, "step": 3891 }, { "epoch": 0.84, "learning_rate": 1.311918985923274e-06, "loss": 0.3549, "step": 3892 }, { "epoch": 0.84, "learning_rate": 1.3084597020978818e-06, "loss": 0.2066, "step": 3893 }, { "epoch": 0.84, "learning_rate": 1.3050046657426807e-06, "loss": 0.3184, "step": 3894 }, { "epoch": 0.84, "learning_rate": 1.3015538785461102e-06, "loss": 0.2674, "step": 3895 }, { "epoch": 0.84, "learning_rate": 1.2981073421945424e-06, "loss": 0.2941, "step": 3896 }, { "epoch": 0.84, "learning_rate": 1.2946650583722608e-06, "loss": 0.2788, "step": 3897 }, { "epoch": 0.84, "learning_rate": 1.2912270287614737e-06, "loss": 0.2682, "step": 3898 }, { "epoch": 0.84, "learning_rate": 1.2877932550423188e-06, "loss": 0.3075, "step": 3899 }, { "epoch": 0.84, "learning_rate": 1.284363738892841e-06, "loss": 0.2654, "step": 3900 }, { "epoch": 0.84, "learning_rate": 1.2809384819890158e-06, "loss": 0.3027, "step": 3901 }, { "epoch": 0.84, "learning_rate": 1.2775174860047313e-06, "loss": 0.2524, "step": 3902 }, { "epoch": 0.84, "learning_rate": 1.2741007526117932e-06, "loss": 0.3461, "step": 3903 }, { "epoch": 0.84, "learning_rate": 1.2706882834799273e-06, "loss": 0.3137, "step": 3904 }, { "epoch": 0.84, "learning_rate": 1.2672800802767715e-06, "loss": 0.2661, "step": 3905 }, { "epoch": 0.84, "learning_rate": 1.2638761446678827e-06, "loss": 0.2314, "step": 3906 }, { "epoch": 0.84, "learning_rate": 1.260476478316731e-06, "loss": 0.2984, "step": 3907 }, { "epoch": 0.84, "learning_rate": 1.257081082884699e-06, "loss": 0.2775, "step": 3908 }, { "epoch": 0.84, "learning_rate": 1.2536899600310837e-06, "loss": 0.2792, "step": 3909 }, { "epoch": 0.84, "learning_rate": 1.2503031114130925e-06, "loss": 0.2362, "step": 3910 }, { "epoch": 0.84, "learning_rate": 1.2469205386858452e-06, "loss": 0.2544, "step": 3911 }, { "epoch": 0.84, "learning_rate": 1.2435422435023715e-06, "loss": 0.2354, "step": 3912 }, { "epoch": 0.84, "learning_rate": 1.240168227513614e-06, "loss": 0.2527, "step": 3913 }, { "epoch": 0.84, "learning_rate": 1.2367984923684185e-06, "loss": 0.3351, "step": 3914 }, { "epoch": 0.84, "learning_rate": 1.2334330397135387e-06, "loss": 0.2738, "step": 3915 }, { "epoch": 0.85, "learning_rate": 1.2300718711936445e-06, "loss": 0.2671, "step": 3916 }, { "epoch": 0.85, "learning_rate": 1.2267149884512996e-06, "loss": 0.253, "step": 3917 }, { "epoch": 0.85, "learning_rate": 1.2233623931269834e-06, "loss": 0.2341, "step": 3918 }, { "epoch": 0.85, "learning_rate": 1.2200140868590759e-06, "loss": 0.2501, "step": 3919 }, { "epoch": 0.85, "learning_rate": 1.2166700712838607e-06, "loss": 0.3019, "step": 3920 }, { "epoch": 0.85, "learning_rate": 1.2133303480355263e-06, "loss": 0.2906, "step": 3921 }, { "epoch": 0.85, "learning_rate": 1.20999491874616e-06, "loss": 0.2552, "step": 3922 }, { "epoch": 0.85, "learning_rate": 1.2066637850457563e-06, "loss": 0.2574, "step": 3923 }, { "epoch": 0.85, "learning_rate": 1.203336948562206e-06, "loss": 0.3698, "step": 3924 }, { "epoch": 0.85, "learning_rate": 1.2000144109213007e-06, "loss": 0.3303, "step": 3925 }, { "epoch": 0.85, "learning_rate": 1.1966961737467331e-06, "loss": 0.2927, "step": 3926 }, { "epoch": 0.85, "learning_rate": 1.1933822386600924e-06, "loss": 0.3124, "step": 3927 }, { "epoch": 0.85, "learning_rate": 1.1900726072808667e-06, "loss": 0.2702, "step": 3928 }, { "epoch": 0.85, "learning_rate": 1.1867672812264375e-06, "loss": 0.2811, "step": 3929 }, { "epoch": 0.85, "learning_rate": 1.1834662621120906e-06, "loss": 0.2995, "step": 3930 }, { "epoch": 0.85, "learning_rate": 1.1801695515509949e-06, "loss": 0.321, "step": 3931 }, { "epoch": 0.85, "learning_rate": 1.1768771511542254e-06, "loss": 0.2804, "step": 3932 }, { "epoch": 0.85, "learning_rate": 1.1735890625307466e-06, "loss": 0.3037, "step": 3933 }, { "epoch": 0.85, "learning_rate": 1.1703052872874087e-06, "loss": 0.3771, "step": 3934 }, { "epoch": 0.85, "learning_rate": 1.1670258270289692e-06, "loss": 0.27, "step": 3935 }, { "epoch": 0.85, "learning_rate": 1.163750683358057e-06, "loss": 0.2588, "step": 3936 }, { "epoch": 0.85, "learning_rate": 1.1604798578752119e-06, "loss": 0.2773, "step": 3937 }, { "epoch": 0.85, "learning_rate": 1.1572133521788509e-06, "loss": 0.2793, "step": 3938 }, { "epoch": 0.85, "learning_rate": 1.1539511678652836e-06, "loss": 0.3187, "step": 3939 }, { "epoch": 0.85, "learning_rate": 1.1506933065287062e-06, "loss": 0.284, "step": 3940 }, { "epoch": 0.85, "learning_rate": 1.147439769761204e-06, "loss": 0.2756, "step": 3941 }, { "epoch": 0.85, "learning_rate": 1.1441905591527492e-06, "loss": 0.2589, "step": 3942 }, { "epoch": 0.85, "learning_rate": 1.1409456762911985e-06, "loss": 0.2618, "step": 3943 }, { "epoch": 0.85, "learning_rate": 1.1377051227622937e-06, "loss": 0.2944, "step": 3944 }, { "epoch": 0.85, "learning_rate": 1.1344689001496622e-06, "loss": 0.267, "step": 3945 }, { "epoch": 0.85, "learning_rate": 1.1312370100348125e-06, "loss": 0.2587, "step": 3946 }, { "epoch": 0.85, "learning_rate": 1.128009453997142e-06, "loss": 0.298, "step": 3947 }, { "epoch": 0.85, "learning_rate": 1.1247862336139182e-06, "loss": 0.2968, "step": 3948 }, { "epoch": 0.85, "learning_rate": 1.1215673504603054e-06, "loss": 0.285, "step": 3949 }, { "epoch": 0.85, "learning_rate": 1.118352806109332e-06, "loss": 0.2571, "step": 3950 }, { "epoch": 0.85, "learning_rate": 1.1151426021319212e-06, "loss": 0.2941, "step": 3951 }, { "epoch": 0.85, "learning_rate": 1.1119367400968662e-06, "loss": 0.2314, "step": 3952 }, { "epoch": 0.85, "learning_rate": 1.1087352215708346e-06, "loss": 0.2939, "step": 3953 }, { "epoch": 0.85, "learning_rate": 1.1055380481183853e-06, "loss": 0.3204, "step": 3954 }, { "epoch": 0.85, "learning_rate": 1.1023452213019404e-06, "loss": 0.3049, "step": 3955 }, { "epoch": 0.85, "learning_rate": 1.099156742681804e-06, "loss": 0.2677, "step": 3956 }, { "epoch": 0.85, "learning_rate": 1.0959726138161552e-06, "loss": 0.2876, "step": 3957 }, { "epoch": 0.85, "learning_rate": 1.0927928362610462e-06, "loss": 0.2346, "step": 3958 }, { "epoch": 0.85, "learning_rate": 1.0896174115704016e-06, "loss": 0.2491, "step": 3959 }, { "epoch": 0.85, "learning_rate": 1.0864463412960191e-06, "loss": 0.3343, "step": 3960 }, { "epoch": 0.85, "learning_rate": 1.0832796269875757e-06, "loss": 0.2948, "step": 3961 }, { "epoch": 0.86, "learning_rate": 1.0801172701926077e-06, "loss": 0.2824, "step": 3962 }, { "epoch": 0.86, "learning_rate": 1.0769592724565292e-06, "loss": 0.2687, "step": 3963 }, { "epoch": 0.86, "learning_rate": 1.0738056353226234e-06, "loss": 0.2961, "step": 3964 }, { "epoch": 0.86, "learning_rate": 1.0706563603320387e-06, "loss": 0.2513, "step": 3965 }, { "epoch": 0.86, "learning_rate": 1.0675114490238004e-06, "loss": 0.2952, "step": 3966 }, { "epoch": 0.86, "learning_rate": 1.0643709029347892e-06, "loss": 0.3099, "step": 3967 }, { "epoch": 0.86, "learning_rate": 1.0612347235997645e-06, "loss": 0.2841, "step": 3968 }, { "epoch": 0.86, "learning_rate": 1.0581029125513452e-06, "loss": 0.3571, "step": 3969 }, { "epoch": 0.86, "learning_rate": 1.0549754713200121e-06, "loss": 0.2989, "step": 3970 }, { "epoch": 0.86, "learning_rate": 1.0518524014341203e-06, "loss": 0.2753, "step": 3971 }, { "epoch": 0.86, "learning_rate": 1.048733704419882e-06, "loss": 0.276, "step": 3972 }, { "epoch": 0.86, "learning_rate": 1.0456193818013727e-06, "loss": 0.3027, "step": 3973 }, { "epoch": 0.86, "learning_rate": 1.0425094351005305e-06, "loss": 0.2751, "step": 3974 }, { "epoch": 0.86, "learning_rate": 1.0394038658371575e-06, "loss": 0.2937, "step": 3975 }, { "epoch": 0.86, "learning_rate": 1.0363026755289151e-06, "loss": 0.2877, "step": 3976 }, { "epoch": 0.86, "learning_rate": 1.0332058656913224e-06, "loss": 0.271, "step": 3977 }, { "epoch": 0.86, "learning_rate": 1.030113437837761e-06, "loss": 0.2307, "step": 3978 }, { "epoch": 0.86, "learning_rate": 1.0270253934794683e-06, "loss": 0.253, "step": 3979 }, { "epoch": 0.86, "learning_rate": 1.0239417341255464e-06, "loss": 0.2699, "step": 3980 }, { "epoch": 0.86, "learning_rate": 1.0208624612829443e-06, "loss": 0.298, "step": 3981 }, { "epoch": 0.86, "learning_rate": 1.0177875764564716e-06, "loss": 0.3022, "step": 3982 }, { "epoch": 0.86, "learning_rate": 1.0147170811488004e-06, "loss": 0.3026, "step": 3983 }, { "epoch": 0.86, "learning_rate": 1.0116509768604432e-06, "loss": 0.2659, "step": 3984 }, { "epoch": 0.86, "learning_rate": 1.0085892650897822e-06, "loss": 0.229, "step": 3985 }, { "epoch": 0.86, "learning_rate": 1.0055319473330416e-06, "loss": 0.2666, "step": 3986 }, { "epoch": 0.86, "learning_rate": 1.0024790250843054e-06, "loss": 0.2992, "step": 3987 }, { "epoch": 0.86, "learning_rate": 9.99430499835503e-07, "loss": 0.2867, "step": 3988 }, { "epoch": 0.86, "learning_rate": 9.963863730764222e-07, "loss": 0.3188, "step": 3989 }, { "epoch": 0.86, "learning_rate": 9.933466462946962e-07, "loss": 0.2653, "step": 3990 }, { "epoch": 0.86, "learning_rate": 9.903113209758098e-07, "loss": 0.2895, "step": 3991 }, { "epoch": 0.86, "learning_rate": 9.872803986030954e-07, "loss": 0.2649, "step": 3992 }, { "epoch": 0.86, "learning_rate": 9.842538806577372e-07, "loss": 0.2993, "step": 3993 }, { "epoch": 0.86, "learning_rate": 9.812317686187623e-07, "loss": 0.2828, "step": 3994 }, { "epoch": 0.86, "learning_rate": 9.782140639630499e-07, "loss": 0.3124, "step": 3995 }, { "epoch": 0.86, "learning_rate": 9.752007681653176e-07, "loss": 0.2545, "step": 3996 }, { "epoch": 0.86, "learning_rate": 9.721918826981413e-07, "loss": 0.2645, "step": 3997 }, { "epoch": 0.86, "learning_rate": 9.691874090319276e-07, "loss": 0.2809, "step": 3998 }, { "epoch": 0.86, "learning_rate": 9.661873486349327e-07, "loss": 0.3063, "step": 3999 }, { "epoch": 0.86, "learning_rate": 9.63191702973263e-07, "loss": 0.306, "step": 4000 }, { "epoch": 0.86, "learning_rate": 9.602004735108516e-07, "loss": 0.2606, "step": 4001 }, { "epoch": 0.86, "learning_rate": 9.572136617094928e-07, "loss": 0.2559, "step": 4002 }, { "epoch": 0.86, "learning_rate": 9.542312690288035e-07, "loss": 0.2808, "step": 4003 }, { "epoch": 0.86, "learning_rate": 9.512532969262556e-07, "loss": 0.2681, "step": 4004 }, { "epoch": 0.86, "learning_rate": 9.482797468571548e-07, "loss": 0.2767, "step": 4005 }, { "epoch": 0.86, "learning_rate": 9.453106202746387e-07, "loss": 0.2856, "step": 4006 }, { "epoch": 0.86, "learning_rate": 9.423459186296979e-07, "loss": 0.2962, "step": 4007 }, { "epoch": 0.87, "learning_rate": 9.393856433711501e-07, "loss": 0.2333, "step": 4008 }, { "epoch": 0.87, "learning_rate": 9.364297959456525e-07, "loss": 0.2938, "step": 4009 }, { "epoch": 0.87, "learning_rate": 9.334783777976997e-07, "loss": 0.2586, "step": 4010 }, { "epoch": 0.87, "learning_rate": 9.305313903696189e-07, "loss": 0.297, "step": 4011 }, { "epoch": 0.87, "learning_rate": 9.27588835101576e-07, "loss": 0.2989, "step": 4012 }, { "epoch": 0.87, "learning_rate": 9.246507134315641e-07, "loss": 0.2801, "step": 4013 }, { "epoch": 0.87, "learning_rate": 9.217170267954234e-07, "loss": 0.3003, "step": 4014 }, { "epoch": 0.87, "learning_rate": 9.187877766268071e-07, "loss": 0.2947, "step": 4015 }, { "epoch": 0.87, "learning_rate": 9.158629643572204e-07, "loss": 0.28, "step": 4016 }, { "epoch": 0.87, "learning_rate": 9.129425914159839e-07, "loss": 0.25, "step": 4017 }, { "epoch": 0.87, "learning_rate": 9.100266592302543e-07, "loss": 0.2744, "step": 4018 }, { "epoch": 0.87, "learning_rate": 9.071151692250268e-07, "loss": 0.2572, "step": 4019 }, { "epoch": 0.87, "learning_rate": 9.042081228231081e-07, "loss": 0.2889, "step": 4020 }, { "epoch": 0.87, "learning_rate": 9.013055214451505e-07, "loss": 0.2696, "step": 4021 }, { "epoch": 0.87, "learning_rate": 8.984073665096238e-07, "loss": 0.3374, "step": 4022 }, { "epoch": 0.87, "learning_rate": 8.955136594328284e-07, "loss": 0.2627, "step": 4023 }, { "epoch": 0.87, "learning_rate": 8.926244016288898e-07, "loss": 0.3083, "step": 4024 }, { "epoch": 0.87, "learning_rate": 8.897395945097598e-07, "loss": 0.2712, "step": 4025 }, { "epoch": 0.87, "learning_rate": 8.86859239485216e-07, "loss": 0.2513, "step": 4026 }, { "epoch": 0.87, "learning_rate": 8.839833379628592e-07, "loss": 0.2657, "step": 4027 }, { "epoch": 0.87, "learning_rate": 8.811118913481143e-07, "loss": 0.2682, "step": 4028 }, { "epoch": 0.87, "learning_rate": 8.782449010442296e-07, "loss": 0.3136, "step": 4029 }, { "epoch": 0.87, "learning_rate": 8.753823684522734e-07, "loss": 0.2349, "step": 4030 }, { "epoch": 0.87, "learning_rate": 8.725242949711376e-07, "loss": 0.25, "step": 4031 }, { "epoch": 0.87, "learning_rate": 8.696706819975343e-07, "loss": 0.3338, "step": 4032 }, { "epoch": 0.87, "learning_rate": 8.668215309259998e-07, "loss": 0.2832, "step": 4033 }, { "epoch": 0.87, "learning_rate": 8.639768431488816e-07, "loss": 0.2597, "step": 4034 }, { "epoch": 0.87, "learning_rate": 8.611366200563509e-07, "loss": 0.2869, "step": 4035 }, { "epoch": 0.87, "learning_rate": 8.58300863036402e-07, "loss": 0.3136, "step": 4036 }, { "epoch": 0.87, "learning_rate": 8.554695734748331e-07, "loss": 0.282, "step": 4037 }, { "epoch": 0.87, "learning_rate": 8.526427527552738e-07, "loss": 0.2706, "step": 4038 }, { "epoch": 0.87, "learning_rate": 8.498204022591628e-07, "loss": 0.3355, "step": 4039 }, { "epoch": 0.87, "learning_rate": 8.470025233657531e-07, "loss": 0.3224, "step": 4040 }, { "epoch": 0.87, "learning_rate": 8.441891174521155e-07, "loss": 0.3054, "step": 4041 }, { "epoch": 0.87, "learning_rate": 8.413801858931325e-07, "loss": 0.2666, "step": 4042 }, { "epoch": 0.87, "learning_rate": 8.385757300615039e-07, "loss": 0.2843, "step": 4043 }, { "epoch": 0.87, "learning_rate": 8.357757513277376e-07, "loss": 0.2822, "step": 4044 }, { "epoch": 0.87, "learning_rate": 8.329802510601559e-07, "loss": 0.2881, "step": 4045 }, { "epoch": 0.87, "learning_rate": 8.301892306248926e-07, "loss": 0.2633, "step": 4046 }, { "epoch": 0.87, "learning_rate": 8.274026913858924e-07, "loss": 0.3071, "step": 4047 }, { "epoch": 0.87, "learning_rate": 8.246206347049079e-07, "loss": 0.297, "step": 4048 }, { "epoch": 0.87, "learning_rate": 8.21843061941503e-07, "loss": 0.2522, "step": 4049 }, { "epoch": 0.87, "learning_rate": 8.190699744530539e-07, "loss": 0.3199, "step": 4050 }, { "epoch": 0.87, "learning_rate": 8.163013735947356e-07, "loss": 0.2492, "step": 4051 }, { "epoch": 0.87, "learning_rate": 8.135372607195402e-07, "loss": 0.217, "step": 4052 }, { "epoch": 0.87, "learning_rate": 8.107776371782628e-07, "loss": 0.2346, "step": 4053 }, { "epoch": 0.87, "learning_rate": 8.080225043195e-07, "loss": 0.289, "step": 4054 }, { "epoch": 0.88, "learning_rate": 8.052718634896616e-07, "loss": 0.2799, "step": 4055 }, { "epoch": 0.88, "learning_rate": 8.02525716032958e-07, "loss": 0.2279, "step": 4056 }, { "epoch": 0.88, "learning_rate": 7.997840632914033e-07, "loss": 0.309, "step": 4057 }, { "epoch": 0.88, "learning_rate": 7.970469066048181e-07, "loss": 0.2824, "step": 4058 }, { "epoch": 0.88, "learning_rate": 7.943142473108234e-07, "loss": 0.2814, "step": 4059 }, { "epoch": 0.88, "learning_rate": 7.915860867448411e-07, "loss": 0.2859, "step": 4060 }, { "epoch": 0.88, "learning_rate": 7.888624262400991e-07, "loss": 0.2744, "step": 4061 }, { "epoch": 0.88, "learning_rate": 7.861432671276214e-07, "loss": 0.249, "step": 4062 }, { "epoch": 0.88, "learning_rate": 7.834286107362344e-07, "loss": 0.2261, "step": 4063 }, { "epoch": 0.88, "learning_rate": 7.807184583925653e-07, "loss": 0.3032, "step": 4064 }, { "epoch": 0.88, "learning_rate": 7.780128114210384e-07, "loss": 0.2724, "step": 4065 }, { "epoch": 0.88, "learning_rate": 7.753116711438746e-07, "loss": 0.2733, "step": 4066 }, { "epoch": 0.88, "learning_rate": 7.726150388810994e-07, "loss": 0.2752, "step": 4067 }, { "epoch": 0.88, "learning_rate": 7.699229159505251e-07, "loss": 0.2801, "step": 4068 }, { "epoch": 0.88, "learning_rate": 7.672353036677693e-07, "loss": 0.2937, "step": 4069 }, { "epoch": 0.88, "learning_rate": 7.645522033462393e-07, "loss": 0.2608, "step": 4070 }, { "epoch": 0.88, "learning_rate": 7.618736162971429e-07, "loss": 0.311, "step": 4071 }, { "epoch": 0.88, "learning_rate": 7.591995438294775e-07, "loss": 0.2327, "step": 4072 }, { "epoch": 0.88, "learning_rate": 7.565299872500331e-07, "loss": 0.2768, "step": 4073 }, { "epoch": 0.88, "learning_rate": 7.538649478633997e-07, "loss": 0.2458, "step": 4074 }, { "epoch": 0.88, "learning_rate": 7.512044269719553e-07, "loss": 0.2623, "step": 4075 }, { "epoch": 0.88, "learning_rate": 7.48548425875869e-07, "loss": 0.2675, "step": 4076 }, { "epoch": 0.88, "learning_rate": 7.458969458731035e-07, "loss": 0.2969, "step": 4077 }, { "epoch": 0.88, "learning_rate": 7.432499882594091e-07, "loss": 0.3238, "step": 4078 }, { "epoch": 0.88, "learning_rate": 7.406075543283298e-07, "loss": 0.279, "step": 4079 }, { "epoch": 0.88, "learning_rate": 7.379696453711949e-07, "loss": 0.2719, "step": 4080 }, { "epoch": 0.88, "learning_rate": 7.353362626771299e-07, "loss": 0.2668, "step": 4081 }, { "epoch": 0.88, "learning_rate": 7.327074075330365e-07, "loss": 0.2617, "step": 4082 }, { "epoch": 0.88, "learning_rate": 7.300830812236126e-07, "loss": 0.3201, "step": 4083 }, { "epoch": 0.88, "learning_rate": 7.274632850313434e-07, "loss": 0.2566, "step": 4084 }, { "epoch": 0.88, "learning_rate": 7.248480202364927e-07, "loss": 0.2571, "step": 4085 }, { "epoch": 0.88, "learning_rate": 7.222372881171191e-07, "loss": 0.2873, "step": 4086 }, { "epoch": 0.88, "learning_rate": 7.196310899490577e-07, "loss": 0.2374, "step": 4087 }, { "epoch": 0.88, "learning_rate": 7.170294270059353e-07, "loss": 0.2853, "step": 4088 }, { "epoch": 0.88, "learning_rate": 7.144323005591591e-07, "loss": 0.2684, "step": 4089 }, { "epoch": 0.88, "learning_rate": 7.118397118779141e-07, "loss": 0.3094, "step": 4090 }, { "epoch": 0.88, "learning_rate": 7.09251662229179e-07, "loss": 0.2836, "step": 4091 }, { "epoch": 0.88, "learning_rate": 7.066681528777053e-07, "loss": 0.267, "step": 4092 }, { "epoch": 0.88, "learning_rate": 7.040891850860287e-07, "loss": 0.2973, "step": 4093 }, { "epoch": 0.88, "learning_rate": 7.015147601144656e-07, "loss": 0.3025, "step": 4094 }, { "epoch": 0.88, "learning_rate": 6.989448792211118e-07, "loss": 0.3049, "step": 4095 }, { "epoch": 0.88, "learning_rate": 6.963795436618437e-07, "loss": 0.2498, "step": 4096 }, { "epoch": 0.88, "learning_rate": 6.938187546903142e-07, "loss": 0.2593, "step": 4097 }, { "epoch": 0.88, "learning_rate": 6.912625135579587e-07, "loss": 0.3538, "step": 4098 }, { "epoch": 0.88, "learning_rate": 6.887108215139815e-07, "loss": 0.2555, "step": 4099 }, { "epoch": 0.88, "learning_rate": 6.861636798053761e-07, "loss": 0.2746, "step": 4100 }, { "epoch": 0.89, "learning_rate": 6.836210896769014e-07, "loss": 0.3078, "step": 4101 }, { "epoch": 0.89, "learning_rate": 6.810830523710943e-07, "loss": 0.2972, "step": 4102 }, { "epoch": 0.89, "learning_rate": 6.785495691282751e-07, "loss": 0.2943, "step": 4103 }, { "epoch": 0.89, "learning_rate": 6.760206411865255e-07, "loss": 0.2588, "step": 4104 }, { "epoch": 0.89, "learning_rate": 6.734962697817116e-07, "loss": 0.2742, "step": 4105 }, { "epoch": 0.89, "learning_rate": 6.709764561474685e-07, "loss": 0.2951, "step": 4106 }, { "epoch": 0.89, "learning_rate": 6.684612015152037e-07, "loss": 0.2793, "step": 4107 }, { "epoch": 0.89, "learning_rate": 6.65950507114096e-07, "loss": 0.2815, "step": 4108 }, { "epoch": 0.89, "learning_rate": 6.634443741710994e-07, "loss": 0.2576, "step": 4109 }, { "epoch": 0.89, "learning_rate": 6.609428039109356e-07, "loss": 0.2503, "step": 4110 }, { "epoch": 0.89, "learning_rate": 6.584457975560965e-07, "loss": 0.3079, "step": 4111 }, { "epoch": 0.89, "learning_rate": 6.55953356326845e-07, "loss": 0.2848, "step": 4112 }, { "epoch": 0.89, "learning_rate": 6.534654814412133e-07, "loss": 0.2954, "step": 4113 }, { "epoch": 0.89, "learning_rate": 6.509821741150002e-07, "loss": 0.2259, "step": 4114 }, { "epoch": 0.89, "learning_rate": 6.485034355617748e-07, "loss": 0.3165, "step": 4115 }, { "epoch": 0.89, "learning_rate": 6.460292669928703e-07, "loss": 0.2994, "step": 4116 }, { "epoch": 0.89, "learning_rate": 6.435596696173918e-07, "loss": 0.2642, "step": 4117 }, { "epoch": 0.89, "learning_rate": 6.410946446422051e-07, "loss": 0.2864, "step": 4118 }, { "epoch": 0.89, "learning_rate": 6.386341932719408e-07, "loss": 0.2867, "step": 4119 }, { "epoch": 0.89, "learning_rate": 6.361783167090041e-07, "loss": 0.2962, "step": 4120 }, { "epoch": 0.89, "learning_rate": 6.337270161535513e-07, "loss": 0.2968, "step": 4121 }, { "epoch": 0.89, "learning_rate": 6.312802928035123e-07, "loss": 0.2603, "step": 4122 }, { "epoch": 0.89, "learning_rate": 6.288381478545758e-07, "loss": 0.2964, "step": 4123 }, { "epoch": 0.89, "learning_rate": 6.26400582500194e-07, "loss": 0.3589, "step": 4124 }, { "epoch": 0.89, "learning_rate": 6.239675979315807e-07, "loss": 0.2817, "step": 4125 }, { "epoch": 0.89, "learning_rate": 6.215391953377125e-07, "loss": 0.2859, "step": 4126 }, { "epoch": 0.89, "learning_rate": 6.191153759053259e-07, "loss": 0.2617, "step": 4127 }, { "epoch": 0.89, "learning_rate": 6.166961408189165e-07, "loss": 0.3127, "step": 4128 }, { "epoch": 0.89, "learning_rate": 6.142814912607409e-07, "loss": 0.2513, "step": 4129 }, { "epoch": 0.89, "learning_rate": 6.118714284108141e-07, "loss": 0.2746, "step": 4130 }, { "epoch": 0.89, "learning_rate": 6.094659534469116e-07, "loss": 0.304, "step": 4131 }, { "epoch": 0.89, "learning_rate": 6.070650675445655e-07, "loss": 0.2578, "step": 4132 }, { "epoch": 0.89, "learning_rate": 6.046687718770617e-07, "loss": 0.359, "step": 4133 }, { "epoch": 0.89, "learning_rate": 6.02277067615451e-07, "loss": 0.3083, "step": 4134 }, { "epoch": 0.89, "learning_rate": 5.998899559285321e-07, "loss": 0.3515, "step": 4135 }, { "epoch": 0.89, "learning_rate": 5.975074379828649e-07, "loss": 0.2696, "step": 4136 }, { "epoch": 0.89, "learning_rate": 5.951295149427638e-07, "loss": 0.3175, "step": 4137 }, { "epoch": 0.89, "learning_rate": 5.927561879702904e-07, "loss": 0.2334, "step": 4138 }, { "epoch": 0.89, "learning_rate": 5.903874582252733e-07, "loss": 0.295, "step": 4139 }, { "epoch": 0.89, "learning_rate": 5.880233268652835e-07, "loss": 0.2665, "step": 4140 }, { "epoch": 0.89, "learning_rate": 5.856637950456501e-07, "loss": 0.3054, "step": 4141 }, { "epoch": 0.89, "learning_rate": 5.833088639194528e-07, "loss": 0.3165, "step": 4142 }, { "epoch": 0.89, "learning_rate": 5.809585346375235e-07, "loss": 0.289, "step": 4143 }, { "epoch": 0.89, "learning_rate": 5.786128083484444e-07, "loss": 0.3003, "step": 4144 }, { "epoch": 0.89, "learning_rate": 5.762716861985484e-07, "loss": 0.2931, "step": 4145 }, { "epoch": 0.89, "learning_rate": 5.739351693319206e-07, "loss": 0.2249, "step": 4146 }, { "epoch": 0.9, "learning_rate": 5.71603258890392e-07, "loss": 0.2825, "step": 4147 }, { "epoch": 0.9, "learning_rate": 5.692759560135441e-07, "loss": 0.269, "step": 4148 }, { "epoch": 0.9, "learning_rate": 5.669532618387097e-07, "loss": 0.2989, "step": 4149 }, { "epoch": 0.9, "learning_rate": 5.646351775009617e-07, "loss": 0.335, "step": 4150 }, { "epoch": 0.9, "learning_rate": 5.623217041331319e-07, "loss": 0.3258, "step": 4151 }, { "epoch": 0.9, "learning_rate": 5.600128428657847e-07, "loss": 0.3104, "step": 4152 }, { "epoch": 0.9, "learning_rate": 5.577085948272432e-07, "loss": 0.2192, "step": 4153 }, { "epoch": 0.9, "learning_rate": 5.554089611435687e-07, "loss": 0.3182, "step": 4154 }, { "epoch": 0.9, "learning_rate": 5.531139429385691e-07, "loss": 0.301, "step": 4155 }, { "epoch": 0.9, "learning_rate": 5.508235413338004e-07, "loss": 0.2841, "step": 4156 }, { "epoch": 0.9, "learning_rate": 5.485377574485528e-07, "loss": 0.2626, "step": 4157 }, { "epoch": 0.9, "learning_rate": 5.462565923998719e-07, "loss": 0.3194, "step": 4158 }, { "epoch": 0.9, "learning_rate": 5.439800473025391e-07, "loss": 0.3254, "step": 4159 }, { "epoch": 0.9, "learning_rate": 5.417081232690802e-07, "loss": 0.3279, "step": 4160 }, { "epoch": 0.9, "learning_rate": 5.394408214097602e-07, "loss": 0.2821, "step": 4161 }, { "epoch": 0.9, "learning_rate": 5.371781428325895e-07, "loss": 0.2463, "step": 4162 }, { "epoch": 0.9, "learning_rate": 5.349200886433148e-07, "loss": 0.3304, "step": 4163 }, { "epoch": 0.9, "learning_rate": 5.326666599454255e-07, "loss": 0.256, "step": 4164 }, { "epoch": 0.9, "learning_rate": 5.30417857840153e-07, "loss": 0.2483, "step": 4165 }, { "epoch": 0.9, "learning_rate": 5.281736834264606e-07, "loss": 0.2593, "step": 4166 }, { "epoch": 0.9, "learning_rate": 5.259341378010541e-07, "loss": 0.2814, "step": 4167 }, { "epoch": 0.9, "learning_rate": 5.236992220583814e-07, "loss": 0.2714, "step": 4168 }, { "epoch": 0.9, "learning_rate": 5.214689372906201e-07, "loss": 0.2933, "step": 4169 }, { "epoch": 0.9, "learning_rate": 5.192432845876916e-07, "loss": 0.2898, "step": 4170 }, { "epoch": 0.9, "learning_rate": 5.17022265037247e-07, "loss": 0.3271, "step": 4171 }, { "epoch": 0.9, "learning_rate": 5.148058797246791e-07, "loss": 0.3011, "step": 4172 }, { "epoch": 0.9, "learning_rate": 5.125941297331149e-07, "loss": 0.2836, "step": 4173 }, { "epoch": 0.9, "learning_rate": 5.103870161434087e-07, "loss": 0.2785, "step": 4174 }, { "epoch": 0.9, "learning_rate": 5.081845400341623e-07, "loss": 0.3015, "step": 4175 }, { "epoch": 0.9, "learning_rate": 5.059867024817011e-07, "loss": 0.267, "step": 4176 }, { "epoch": 0.9, "learning_rate": 5.037935045600861e-07, "loss": 0.2548, "step": 4177 }, { "epoch": 0.9, "learning_rate": 5.016049473411122e-07, "loss": 0.3082, "step": 4178 }, { "epoch": 0.9, "learning_rate": 4.994210318943061e-07, "loss": 0.3115, "step": 4179 }, { "epoch": 0.9, "learning_rate": 4.972417592869249e-07, "loss": 0.2981, "step": 4180 }, { "epoch": 0.9, "learning_rate": 4.950671305839571e-07, "loss": 0.3083, "step": 4181 }, { "epoch": 0.9, "learning_rate": 4.928971468481258e-07, "loss": 0.2775, "step": 4182 }, { "epoch": 0.9, "learning_rate": 4.907318091398761e-07, "loss": 0.3307, "step": 4183 }, { "epoch": 0.9, "learning_rate": 4.885711185173914e-07, "loss": 0.2502, "step": 4184 }, { "epoch": 0.9, "learning_rate": 4.864150760365771e-07, "loss": 0.3404, "step": 4185 }, { "epoch": 0.9, "learning_rate": 4.842636827510693e-07, "loss": 0.2597, "step": 4186 }, { "epoch": 0.9, "learning_rate": 4.821169397122371e-07, "loss": 0.3349, "step": 4187 }, { "epoch": 0.9, "learning_rate": 4.79974847969168e-07, "loss": 0.2801, "step": 4188 }, { "epoch": 0.9, "learning_rate": 4.778374085686832e-07, "loss": 0.2916, "step": 4189 }, { "epoch": 0.9, "learning_rate": 4.7570462255532855e-07, "loss": 0.293, "step": 4190 }, { "epoch": 0.9, "learning_rate": 4.735764909713747e-07, "loss": 0.3231, "step": 4191 }, { "epoch": 0.9, "learning_rate": 4.714530148568197e-07, "loss": 0.3022, "step": 4192 }, { "epoch": 0.9, "learning_rate": 4.693341952493846e-07, "loss": 0.2546, "step": 4193 }, { "epoch": 0.91, "learning_rate": 4.672200331845156e-07, "loss": 0.2632, "step": 4194 }, { "epoch": 0.91, "learning_rate": 4.6511052969538284e-07, "loss": 0.2587, "step": 4195 }, { "epoch": 0.91, "learning_rate": 4.630056858128806e-07, "loss": 0.3216, "step": 4196 }, { "epoch": 0.91, "learning_rate": 4.6090550256562505e-07, "loss": 0.3138, "step": 4197 }, { "epoch": 0.91, "learning_rate": 4.588099809799551e-07, "loss": 0.2862, "step": 4198 }, { "epoch": 0.91, "learning_rate": 4.567191220799305e-07, "loss": 0.2733, "step": 4199 }, { "epoch": 0.91, "learning_rate": 4.5463292688733394e-07, "loss": 0.2568, "step": 4200 }, { "epoch": 0.91, "learning_rate": 4.52551396421671e-07, "loss": 0.2415, "step": 4201 }, { "epoch": 0.91, "learning_rate": 4.504745317001624e-07, "loss": 0.3269, "step": 4202 }, { "epoch": 0.91, "learning_rate": 4.4840233373774967e-07, "loss": 0.2818, "step": 4203 }, { "epoch": 0.91, "learning_rate": 4.4633480354710155e-07, "loss": 0.3084, "step": 4204 }, { "epoch": 0.91, "learning_rate": 4.4427194213859216e-07, "loss": 0.275, "step": 4205 }, { "epoch": 0.91, "learning_rate": 4.4221375052032835e-07, "loss": 0.3028, "step": 4206 }, { "epoch": 0.91, "learning_rate": 4.401602296981244e-07, "loss": 0.2422, "step": 4207 }, { "epoch": 0.91, "learning_rate": 4.3811138067551664e-07, "loss": 0.2749, "step": 4208 }, { "epoch": 0.91, "learning_rate": 4.360672044537573e-07, "loss": 0.3325, "step": 4209 }, { "epoch": 0.91, "learning_rate": 4.340277020318162e-07, "loss": 0.3209, "step": 4210 }, { "epoch": 0.91, "learning_rate": 4.3199287440637617e-07, "loss": 0.3225, "step": 4211 }, { "epoch": 0.91, "learning_rate": 4.2996272257183834e-07, "loss": 0.3309, "step": 4212 }, { "epoch": 0.91, "learning_rate": 4.2793724752031807e-07, "loss": 0.2857, "step": 4213 }, { "epoch": 0.91, "learning_rate": 4.259164502416457e-07, "loss": 0.266, "step": 4214 }, { "epoch": 0.91, "learning_rate": 4.2390033172336366e-07, "loss": 0.2986, "step": 4215 }, { "epoch": 0.91, "learning_rate": 4.218888929507303e-07, "loss": 0.2665, "step": 4216 }, { "epoch": 0.91, "learning_rate": 4.198821349067128e-07, "loss": 0.2636, "step": 4217 }, { "epoch": 0.91, "learning_rate": 4.178800585720011e-07, "loss": 0.2798, "step": 4218 }, { "epoch": 0.91, "learning_rate": 4.1588266492498255e-07, "loss": 0.3244, "step": 4219 }, { "epoch": 0.91, "learning_rate": 4.1388995494176855e-07, "loss": 0.2796, "step": 4220 }, { "epoch": 0.91, "learning_rate": 4.119019295961757e-07, "loss": 0.346, "step": 4221 }, { "epoch": 0.91, "learning_rate": 4.099185898597313e-07, "loss": 0.3125, "step": 4222 }, { "epoch": 0.91, "learning_rate": 4.0793993670167454e-07, "loss": 0.2964, "step": 4223 }, { "epoch": 0.91, "learning_rate": 4.059659710889541e-07, "loss": 0.2731, "step": 4224 }, { "epoch": 0.91, "learning_rate": 4.039966939862283e-07, "loss": 0.2968, "step": 4225 }, { "epoch": 0.91, "learning_rate": 4.0203210635586186e-07, "loss": 0.3248, "step": 4226 }, { "epoch": 0.91, "learning_rate": 4.000722091579301e-07, "loss": 0.3307, "step": 4227 }, { "epoch": 0.91, "learning_rate": 3.981170033502157e-07, "loss": 0.2853, "step": 4228 }, { "epoch": 0.91, "learning_rate": 3.961664898882089e-07, "loss": 0.2528, "step": 4229 }, { "epoch": 0.91, "learning_rate": 3.94220669725105e-07, "loss": 0.3007, "step": 4230 }, { "epoch": 0.91, "learning_rate": 3.9227954381180765e-07, "loss": 0.2627, "step": 4231 }, { "epoch": 0.91, "learning_rate": 3.903431130969271e-07, "loss": 0.2886, "step": 4232 }, { "epoch": 0.91, "learning_rate": 3.8841137852677735e-07, "loss": 0.2742, "step": 4233 }, { "epoch": 0.91, "learning_rate": 3.864843410453767e-07, "loss": 0.3162, "step": 4234 }, { "epoch": 0.91, "learning_rate": 3.845620015944518e-07, "loss": 0.2565, "step": 4235 }, { "epoch": 0.91, "learning_rate": 3.8264436111342897e-07, "loss": 0.3006, "step": 4236 }, { "epoch": 0.91, "learning_rate": 3.807314205394419e-07, "loss": 0.2768, "step": 4237 }, { "epoch": 0.91, "learning_rate": 3.7882318080732615e-07, "loss": 0.2644, "step": 4238 }, { "epoch": 0.91, "learning_rate": 3.769196428496158e-07, "loss": 0.2669, "step": 4239 }, { "epoch": 0.92, "learning_rate": 3.7502080759655667e-07, "loss": 0.2442, "step": 4240 }, { "epoch": 0.92, "learning_rate": 3.731266759760854e-07, "loss": 0.3383, "step": 4241 }, { "epoch": 0.92, "learning_rate": 3.712372489138494e-07, "loss": 0.2863, "step": 4242 }, { "epoch": 0.92, "learning_rate": 3.693525273331922e-07, "loss": 0.2422, "step": 4243 }, { "epoch": 0.92, "learning_rate": 3.67472512155157e-07, "loss": 0.2677, "step": 4244 }, { "epoch": 0.92, "learning_rate": 3.655972042984912e-07, "loss": 0.2795, "step": 4245 }, { "epoch": 0.92, "learning_rate": 3.637266046796384e-07, "loss": 0.2927, "step": 4246 }, { "epoch": 0.92, "learning_rate": 3.6186071421274173e-07, "loss": 0.2494, "step": 4247 }, { "epoch": 0.92, "learning_rate": 3.599995338096418e-07, "loss": 0.2452, "step": 4248 }, { "epoch": 0.92, "learning_rate": 3.581430643798833e-07, "loss": 0.2821, "step": 4249 }, { "epoch": 0.92, "learning_rate": 3.5629130683070167e-07, "loss": 0.2938, "step": 4250 }, { "epoch": 0.92, "learning_rate": 3.5444426206703296e-07, "loss": 0.3043, "step": 4251 }, { "epoch": 0.92, "learning_rate": 3.5260193099151076e-07, "loss": 0.3001, "step": 4252 }, { "epoch": 0.92, "learning_rate": 3.5076431450446257e-07, "loss": 0.3065, "step": 4253 }, { "epoch": 0.92, "learning_rate": 3.489314135039157e-07, "loss": 0.2987, "step": 4254 }, { "epoch": 0.92, "learning_rate": 3.471032288855869e-07, "loss": 0.2864, "step": 4255 }, { "epoch": 0.92, "learning_rate": 3.45279761542896e-07, "loss": 0.2832, "step": 4256 }, { "epoch": 0.92, "learning_rate": 3.434610123669535e-07, "loss": 0.2988, "step": 4257 }, { "epoch": 0.92, "learning_rate": 3.4164698224656065e-07, "loss": 0.294, "step": 4258 }, { "epoch": 0.92, "learning_rate": 3.3983767206821837e-07, "loss": 0.2971, "step": 4259 }, { "epoch": 0.92, "learning_rate": 3.3803308271612043e-07, "loss": 0.2996, "step": 4260 }, { "epoch": 0.92, "learning_rate": 3.362332150721481e-07, "loss": 0.2995, "step": 4261 }, { "epoch": 0.92, "learning_rate": 3.3443807001588225e-07, "loss": 0.2633, "step": 4262 }, { "epoch": 0.92, "learning_rate": 3.326476484245911e-07, "loss": 0.3006, "step": 4263 }, { "epoch": 0.92, "learning_rate": 3.3086195117323584e-07, "loss": 0.2832, "step": 4264 }, { "epoch": 0.92, "learning_rate": 3.290809791344696e-07, "loss": 0.2519, "step": 4265 }, { "epoch": 0.92, "learning_rate": 3.273047331786372e-07, "loss": 0.2816, "step": 4266 }, { "epoch": 0.92, "learning_rate": 3.255332141737699e-07, "loss": 0.3171, "step": 4267 }, { "epoch": 0.92, "learning_rate": 3.23766422985593e-07, "loss": 0.2885, "step": 4268 }, { "epoch": 0.92, "learning_rate": 3.2200436047752026e-07, "loss": 0.301, "step": 4269 }, { "epoch": 0.92, "learning_rate": 3.2024702751065286e-07, "loss": 0.2786, "step": 4270 }, { "epoch": 0.92, "learning_rate": 3.184944249437849e-07, "loss": 0.3057, "step": 4271 }, { "epoch": 0.92, "learning_rate": 3.1674655363339246e-07, "loss": 0.2621, "step": 4272 }, { "epoch": 0.92, "learning_rate": 3.1500341443364556e-07, "loss": 0.276, "step": 4273 }, { "epoch": 0.92, "learning_rate": 3.132650081963995e-07, "loss": 0.3049, "step": 4274 }, { "epoch": 0.92, "learning_rate": 3.1153133577119466e-07, "loss": 0.3102, "step": 4275 }, { "epoch": 0.92, "learning_rate": 3.0980239800526e-07, "loss": 0.2719, "step": 4276 }, { "epoch": 0.92, "learning_rate": 3.080781957435108e-07, "loss": 0.2764, "step": 4277 }, { "epoch": 0.92, "learning_rate": 3.063587298285475e-07, "loss": 0.2683, "step": 4278 }, { "epoch": 0.92, "learning_rate": 3.0464400110065575e-07, "loss": 0.2641, "step": 4279 }, { "epoch": 0.92, "learning_rate": 3.0293401039780846e-07, "loss": 0.2995, "step": 4280 }, { "epoch": 0.92, "learning_rate": 3.012287585556595e-07, "loss": 0.275, "step": 4281 }, { "epoch": 0.92, "learning_rate": 2.9952824640754887e-07, "loss": 0.2773, "step": 4282 }, { "epoch": 0.92, "learning_rate": 2.978324747844996e-07, "loss": 0.2995, "step": 4283 }, { "epoch": 0.92, "learning_rate": 2.9614144451521886e-07, "loss": 0.2893, "step": 4284 }, { "epoch": 0.92, "learning_rate": 2.944551564261e-07, "loss": 0.3025, "step": 4285 }, { "epoch": 0.93, "learning_rate": 2.927736113412105e-07, "loss": 0.2658, "step": 4286 }, { "epoch": 0.93, "learning_rate": 2.9109681008230527e-07, "loss": 0.2794, "step": 4287 }, { "epoch": 0.93, "learning_rate": 2.8942475346882436e-07, "loss": 0.3005, "step": 4288 }, { "epoch": 0.93, "learning_rate": 2.8775744231788195e-07, "loss": 0.314, "step": 4289 }, { "epoch": 0.93, "learning_rate": 2.8609487744427736e-07, "loss": 0.2822, "step": 4290 }, { "epoch": 0.93, "learning_rate": 2.844370596604917e-07, "loss": 0.3042, "step": 4291 }, { "epoch": 0.93, "learning_rate": 2.8278398977668153e-07, "loss": 0.322, "step": 4292 }, { "epoch": 0.93, "learning_rate": 2.811356686006872e-07, "loss": 0.3151, "step": 4293 }, { "epoch": 0.93, "learning_rate": 2.7949209693802546e-07, "loss": 0.3284, "step": 4294 }, { "epoch": 0.93, "learning_rate": 2.7785327559189613e-07, "loss": 0.2922, "step": 4295 }, { "epoch": 0.93, "learning_rate": 2.7621920536317405e-07, "loss": 0.2889, "step": 4296 }, { "epoch": 0.93, "learning_rate": 2.745898870504116e-07, "loss": 0.2429, "step": 4297 }, { "epoch": 0.93, "learning_rate": 2.729653214498429e-07, "loss": 0.2243, "step": 4298 }, { "epoch": 0.93, "learning_rate": 2.713455093553763e-07, "loss": 0.2882, "step": 4299 }, { "epoch": 0.93, "learning_rate": 2.6973045155859855e-07, "loss": 0.3318, "step": 4300 }, { "epoch": 0.93, "learning_rate": 2.6812014884876947e-07, "loss": 0.2995, "step": 4301 }, { "epoch": 0.93, "learning_rate": 2.6651460201283397e-07, "loss": 0.2783, "step": 4302 }, { "epoch": 0.93, "learning_rate": 2.649138118354011e-07, "loss": 0.2658, "step": 4303 }, { "epoch": 0.93, "learning_rate": 2.63317779098764e-07, "loss": 0.221, "step": 4304 }, { "epoch": 0.93, "learning_rate": 2.617265045828898e-07, "loss": 0.2509, "step": 4305 }, { "epoch": 0.93, "learning_rate": 2.6013998906541547e-07, "loss": 0.2905, "step": 4306 }, { "epoch": 0.93, "learning_rate": 2.585582333216563e-07, "loss": 0.3212, "step": 4307 }, { "epoch": 0.93, "learning_rate": 2.5698123812460284e-07, "loss": 0.3132, "step": 4308 }, { "epoch": 0.93, "learning_rate": 2.554090042449142e-07, "loss": 0.2487, "step": 4309 }, { "epoch": 0.93, "learning_rate": 2.53841532450928e-07, "loss": 0.2772, "step": 4310 }, { "epoch": 0.93, "learning_rate": 2.5227882350865154e-07, "loss": 0.2796, "step": 4311 }, { "epoch": 0.93, "learning_rate": 2.507208781817638e-07, "loss": 0.2847, "step": 4312 }, { "epoch": 0.93, "learning_rate": 2.4916769723161926e-07, "loss": 0.2788, "step": 4313 }, { "epoch": 0.93, "learning_rate": 2.476192814172418e-07, "loss": 0.3003, "step": 4314 }, { "epoch": 0.93, "learning_rate": 2.460756314953272e-07, "loss": 0.2505, "step": 4315 }, { "epoch": 0.93, "learning_rate": 2.445367482202399e-07, "loss": 0.2785, "step": 4316 }, { "epoch": 0.93, "learning_rate": 2.430026323440193e-07, "loss": 0.2804, "step": 4317 }, { "epoch": 0.93, "learning_rate": 2.4147328461637143e-07, "loss": 0.2878, "step": 4318 }, { "epoch": 0.93, "learning_rate": 2.399487057846761e-07, "loss": 0.2642, "step": 4319 }, { "epoch": 0.93, "learning_rate": 2.3842889659397515e-07, "loss": 0.2581, "step": 4320 }, { "epoch": 0.93, "learning_rate": 2.3691385778699005e-07, "loss": 0.3016, "step": 4321 }, { "epoch": 0.93, "learning_rate": 2.3540359010410075e-07, "loss": 0.2583, "step": 4322 }, { "epoch": 0.93, "learning_rate": 2.338980942833602e-07, "loss": 0.2986, "step": 4323 }, { "epoch": 0.93, "learning_rate": 2.323973710604943e-07, "loss": 0.3131, "step": 4324 }, { "epoch": 0.93, "learning_rate": 2.309014211688865e-07, "loss": 0.3115, "step": 4325 }, { "epoch": 0.93, "learning_rate": 2.294102453395952e-07, "loss": 0.2937, "step": 4326 }, { "epoch": 0.93, "learning_rate": 2.2792384430134318e-07, "loss": 0.2555, "step": 4327 }, { "epoch": 0.93, "learning_rate": 2.264422187805193e-07, "loss": 0.2775, "step": 4328 }, { "epoch": 0.93, "learning_rate": 2.2496536950117996e-07, "loss": 0.2903, "step": 4329 }, { "epoch": 0.93, "learning_rate": 2.2349329718504676e-07, "loss": 0.2901, "step": 4330 }, { "epoch": 0.93, "learning_rate": 2.2202600255150753e-07, "loss": 0.2791, "step": 4331 }, { "epoch": 0.93, "learning_rate": 2.2056348631761314e-07, "loss": 0.3116, "step": 4332 }, { "epoch": 0.94, "learning_rate": 2.1910574919808304e-07, "loss": 0.2657, "step": 4333 }, { "epoch": 0.94, "learning_rate": 2.176527919052973e-07, "loss": 0.2595, "step": 4334 }, { "epoch": 0.94, "learning_rate": 2.1620461514930136e-07, "loss": 0.3228, "step": 4335 }, { "epoch": 0.94, "learning_rate": 2.1476121963780684e-07, "loss": 0.2528, "step": 4336 }, { "epoch": 0.94, "learning_rate": 2.1332260607618504e-07, "loss": 0.3129, "step": 4337 }, { "epoch": 0.94, "learning_rate": 2.1188877516747475e-07, "loss": 0.2715, "step": 4338 }, { "epoch": 0.94, "learning_rate": 2.104597276123721e-07, "loss": 0.2656, "step": 4339 }, { "epoch": 0.94, "learning_rate": 2.0903546410924068e-07, "loss": 0.2649, "step": 4340 }, { "epoch": 0.94, "learning_rate": 2.0761598535410487e-07, "loss": 0.3182, "step": 4341 }, { "epoch": 0.94, "learning_rate": 2.062012920406453e-07, "loss": 0.316, "step": 4342 }, { "epoch": 0.94, "learning_rate": 2.0479138486021343e-07, "loss": 0.319, "step": 4343 }, { "epoch": 0.94, "learning_rate": 2.0338626450181588e-07, "loss": 0.2439, "step": 4344 }, { "epoch": 0.94, "learning_rate": 2.0198593165212e-07, "loss": 0.3077, "step": 4345 }, { "epoch": 0.94, "learning_rate": 2.0059038699545509e-07, "loss": 0.2727, "step": 4346 }, { "epoch": 0.94, "learning_rate": 1.9919963121381115e-07, "loss": 0.27, "step": 4347 }, { "epoch": 0.94, "learning_rate": 1.9781366498683452e-07, "loss": 0.2342, "step": 4348 }, { "epoch": 0.94, "learning_rate": 1.9643248899183566e-07, "loss": 0.2642, "step": 4349 }, { "epoch": 0.94, "learning_rate": 1.9505610390378127e-07, "loss": 0.2924, "step": 4350 }, { "epoch": 0.94, "learning_rate": 1.9368451039529556e-07, "loss": 0.2576, "step": 4351 }, { "epoch": 0.94, "learning_rate": 1.9231770913666458e-07, "loss": 0.2653, "step": 4352 }, { "epoch": 0.94, "learning_rate": 1.909557007958307e-07, "loss": 0.2859, "step": 4353 }, { "epoch": 0.94, "learning_rate": 1.895984860383915e-07, "loss": 0.3467, "step": 4354 }, { "epoch": 0.94, "learning_rate": 1.8824606552760748e-07, "loss": 0.292, "step": 4355 }, { "epoch": 0.94, "learning_rate": 1.8689843992439116e-07, "loss": 0.254, "step": 4356 }, { "epoch": 0.94, "learning_rate": 1.855556098873146e-07, "loss": 0.2755, "step": 4357 }, { "epoch": 0.94, "learning_rate": 1.8421757607260614e-07, "loss": 0.2754, "step": 4358 }, { "epoch": 0.94, "learning_rate": 1.828843391341495e-07, "loss": 0.2656, "step": 4359 }, { "epoch": 0.94, "learning_rate": 1.8155589972348453e-07, "loss": 0.2916, "step": 4360 }, { "epoch": 0.94, "learning_rate": 1.8023225848980642e-07, "loss": 0.219, "step": 4361 }, { "epoch": 0.94, "learning_rate": 1.7891341607996437e-07, "loss": 0.2902, "step": 4362 }, { "epoch": 0.94, "learning_rate": 1.775993731384662e-07, "loss": 0.264, "step": 4363 }, { "epoch": 0.94, "learning_rate": 1.7629013030746933e-07, "loss": 0.2775, "step": 4364 }, { "epoch": 0.94, "learning_rate": 1.7498568822678863e-07, "loss": 0.314, "step": 4365 }, { "epoch": 0.94, "learning_rate": 1.7368604753389307e-07, "loss": 0.2757, "step": 4366 }, { "epoch": 0.94, "learning_rate": 1.7239120886390347e-07, "loss": 0.2651, "step": 4367 }, { "epoch": 0.94, "learning_rate": 1.7110117284959372e-07, "loss": 0.2555, "step": 4368 }, { "epoch": 0.94, "learning_rate": 1.6981594012139391e-07, "loss": 0.2823, "step": 4369 }, { "epoch": 0.94, "learning_rate": 1.6853551130738388e-07, "loss": 0.2592, "step": 4370 }, { "epoch": 0.94, "learning_rate": 1.672598870332942e-07, "loss": 0.2747, "step": 4371 }, { "epoch": 0.94, "learning_rate": 1.65989067922514e-07, "loss": 0.3274, "step": 4372 }, { "epoch": 0.94, "learning_rate": 1.647230545960765e-07, "loss": 0.31, "step": 4373 }, { "epoch": 0.94, "learning_rate": 1.6346184767267125e-07, "loss": 0.2197, "step": 4374 }, { "epoch": 0.94, "learning_rate": 1.6220544776863856e-07, "loss": 0.2765, "step": 4375 }, { "epoch": 0.94, "learning_rate": 1.6095385549796837e-07, "loss": 0.2782, "step": 4376 }, { "epoch": 0.94, "learning_rate": 1.5970707147230148e-07, "loss": 0.3334, "step": 4377 }, { "epoch": 0.94, "learning_rate": 1.584650963009271e-07, "loss": 0.2876, "step": 4378 }, { "epoch": 0.95, "learning_rate": 1.5722793059078867e-07, "loss": 0.2434, "step": 4379 }, { "epoch": 0.95, "learning_rate": 1.559955749464759e-07, "loss": 0.2762, "step": 4380 }, { "epoch": 0.95, "learning_rate": 1.5476802997022812e-07, "loss": 0.2871, "step": 4381 }, { "epoch": 0.95, "learning_rate": 1.535452962619366e-07, "loss": 0.2816, "step": 4382 }, { "epoch": 0.95, "learning_rate": 1.523273744191367e-07, "loss": 0.2428, "step": 4383 }, { "epoch": 0.95, "learning_rate": 1.5111426503701676e-07, "loss": 0.2676, "step": 4384 }, { "epoch": 0.95, "learning_rate": 1.4990596870841035e-07, "loss": 0.2737, "step": 4385 }, { "epoch": 0.95, "learning_rate": 1.4870248602380178e-07, "loss": 0.2678, "step": 4386 }, { "epoch": 0.95, "learning_rate": 1.4750381757131837e-07, "loss": 0.3342, "step": 4387 }, { "epoch": 0.95, "learning_rate": 1.4630996393673936e-07, "loss": 0.2757, "step": 4388 }, { "epoch": 0.95, "learning_rate": 1.4512092570348912e-07, "loss": 0.2739, "step": 4389 }, { "epoch": 0.95, "learning_rate": 1.4393670345263843e-07, "loss": 0.2897, "step": 4390 }, { "epoch": 0.95, "learning_rate": 1.427572977629066e-07, "loss": 0.3006, "step": 4391 }, { "epoch": 0.95, "learning_rate": 1.4158270921065363e-07, "loss": 0.3044, "step": 4392 }, { "epoch": 0.95, "learning_rate": 1.4041293836989377e-07, "loss": 0.3197, "step": 4393 }, { "epoch": 0.95, "learning_rate": 1.3924798581228083e-07, "loss": 0.2711, "step": 4394 }, { "epoch": 0.95, "learning_rate": 1.3808785210711606e-07, "loss": 0.2616, "step": 4395 }, { "epoch": 0.95, "learning_rate": 1.3693253782134486e-07, "loss": 0.2991, "step": 4396 }, { "epoch": 0.95, "learning_rate": 1.3578204351955782e-07, "loss": 0.2706, "step": 4397 }, { "epoch": 0.95, "learning_rate": 1.3463636976398963e-07, "loss": 0.2903, "step": 4398 }, { "epoch": 0.95, "learning_rate": 1.3349551711452247e-07, "loss": 0.285, "step": 4399 }, { "epoch": 0.95, "learning_rate": 1.323594861286759e-07, "loss": 0.3347, "step": 4400 }, { "epoch": 0.95, "learning_rate": 1.3122827736162024e-07, "loss": 0.2938, "step": 4401 }, { "epoch": 0.95, "learning_rate": 1.3010189136616447e-07, "loss": 0.3086, "step": 4402 }, { "epoch": 0.95, "learning_rate": 1.2898032869276377e-07, "loss": 0.286, "step": 4403 }, { "epoch": 0.95, "learning_rate": 1.2786358988951197e-07, "loss": 0.2566, "step": 4404 }, { "epoch": 0.95, "learning_rate": 1.267516755021525e-07, "loss": 0.3009, "step": 4405 }, { "epoch": 0.95, "learning_rate": 1.2564458607406294e-07, "loss": 0.3318, "step": 4406 }, { "epoch": 0.95, "learning_rate": 1.2454232214626828e-07, "loss": 0.2777, "step": 4407 }, { "epoch": 0.95, "learning_rate": 1.234448842574365e-07, "loss": 0.278, "step": 4408 }, { "epoch": 0.95, "learning_rate": 1.2235227294387085e-07, "loss": 0.3362, "step": 4409 }, { "epoch": 0.95, "learning_rate": 1.21264488739522e-07, "loss": 0.2525, "step": 4410 }, { "epoch": 0.95, "learning_rate": 1.2018153217597917e-07, "loss": 0.27, "step": 4411 }, { "epoch": 0.95, "learning_rate": 1.1910340378247232e-07, "loss": 0.2953, "step": 4412 }, { "epoch": 0.95, "learning_rate": 1.1803010408587224e-07, "loss": 0.2535, "step": 4413 }, { "epoch": 0.95, "learning_rate": 1.1696163361069046e-07, "loss": 0.3117, "step": 4414 }, { "epoch": 0.95, "learning_rate": 1.15897992879076e-07, "loss": 0.316, "step": 4415 }, { "epoch": 0.95, "learning_rate": 1.1483918241082192e-07, "loss": 0.279, "step": 4416 }, { "epoch": 0.95, "learning_rate": 1.137852027233577e-07, "loss": 0.2322, "step": 4417 }, { "epoch": 0.95, "learning_rate": 1.1273605433175239e-07, "loss": 0.2983, "step": 4418 }, { "epoch": 0.95, "learning_rate": 1.1169173774871478e-07, "loss": 0.2992, "step": 4419 }, { "epoch": 0.95, "learning_rate": 1.1065225348459219e-07, "loss": 0.3055, "step": 4420 }, { "epoch": 0.95, "learning_rate": 1.0961760204737048e-07, "loss": 0.2733, "step": 4421 }, { "epoch": 0.95, "learning_rate": 1.08587783942673e-07, "loss": 0.2878, "step": 4422 }, { "epoch": 0.95, "learning_rate": 1.075627996737627e-07, "loss": 0.2617, "step": 4423 }, { "epoch": 0.95, "learning_rate": 1.0654264974153783e-07, "loss": 0.3362, "step": 4424 }, { "epoch": 0.96, "learning_rate": 1.0552733464453845e-07, "loss": 0.225, "step": 4425 }, { "epoch": 0.96, "learning_rate": 1.0451685487893437e-07, "loss": 0.2852, "step": 4426 }, { "epoch": 0.96, "learning_rate": 1.0351121093854166e-07, "loss": 0.3023, "step": 4427 }, { "epoch": 0.96, "learning_rate": 1.0251040331480722e-07, "loss": 0.2671, "step": 4428 }, { "epoch": 0.96, "learning_rate": 1.0151443249681426e-07, "loss": 0.2909, "step": 4429 }, { "epoch": 0.96, "learning_rate": 1.0052329897128565e-07, "loss": 0.3388, "step": 4430 }, { "epoch": 0.96, "learning_rate": 9.953700322257842e-08, "loss": 0.32, "step": 4431 }, { "epoch": 0.96, "learning_rate": 9.855554573268588e-08, "loss": 0.3104, "step": 4432 }, { "epoch": 0.96, "learning_rate": 9.757892698123439e-08, "loss": 0.3724, "step": 4433 }, { "epoch": 0.96, "learning_rate": 9.660714744549104e-08, "loss": 0.35, "step": 4434 }, { "epoch": 0.96, "learning_rate": 9.564020760035264e-08, "loss": 0.2706, "step": 4435 }, { "epoch": 0.96, "learning_rate": 9.467810791835452e-08, "loss": 0.2717, "step": 4436 }, { "epoch": 0.96, "learning_rate": 9.372084886966392e-08, "loss": 0.2969, "step": 4437 }, { "epoch": 0.96, "learning_rate": 9.276843092208554e-08, "loss": 0.3061, "step": 4438 }, { "epoch": 0.96, "learning_rate": 9.182085454105594e-08, "loss": 0.245, "step": 4439 }, { "epoch": 0.96, "learning_rate": 9.087812018964582e-08, "loss": 0.2545, "step": 4440 }, { "epoch": 0.96, "learning_rate": 8.994022832855998e-08, "loss": 0.2613, "step": 4441 }, { "epoch": 0.96, "learning_rate": 8.900717941613957e-08, "loss": 0.3013, "step": 4442 }, { "epoch": 0.96, "learning_rate": 8.807897390835318e-08, "loss": 0.2744, "step": 4443 }, { "epoch": 0.96, "learning_rate": 8.715561225880687e-08, "loss": 0.2915, "step": 4444 }, { "epoch": 0.96, "learning_rate": 8.623709491873966e-08, "loss": 0.2994, "step": 4445 }, { "epoch": 0.96, "learning_rate": 8.532342233702029e-08, "loss": 0.3053, "step": 4446 }, { "epoch": 0.96, "learning_rate": 8.441459496015159e-08, "loss": 0.2353, "step": 4447 }, { "epoch": 0.96, "learning_rate": 8.351061323226939e-08, "loss": 0.2466, "step": 4448 }, { "epoch": 0.96, "learning_rate": 8.261147759514032e-08, "loss": 0.2429, "step": 4449 }, { "epoch": 0.96, "learning_rate": 8.17171884881629e-08, "loss": 0.3061, "step": 4450 }, { "epoch": 0.96, "learning_rate": 8.082774634836754e-08, "loss": 0.2671, "step": 4451 }, { "epoch": 0.96, "learning_rate": 7.994315161041655e-08, "loss": 0.3563, "step": 4452 }, { "epoch": 0.96, "learning_rate": 7.90634047066019e-08, "loss": 0.3107, "step": 4453 }, { "epoch": 0.96, "learning_rate": 7.818850606684746e-08, "loss": 0.274, "step": 4454 }, { "epoch": 0.96, "learning_rate": 7.731845611870681e-08, "loss": 0.2567, "step": 4455 }, { "epoch": 0.96, "learning_rate": 7.645325528736646e-08, "loss": 0.3133, "step": 4456 }, { "epoch": 0.96, "learning_rate": 7.559290399563935e-08, "loss": 0.2876, "step": 4457 }, { "epoch": 0.96, "learning_rate": 7.473740266397244e-08, "loss": 0.2532, "step": 4458 }, { "epoch": 0.96, "learning_rate": 7.388675171044024e-08, "loss": 0.2566, "step": 4459 }, { "epoch": 0.96, "learning_rate": 7.304095155074798e-08, "loss": 0.2496, "step": 4460 }, { "epoch": 0.96, "learning_rate": 7.220000259823056e-08, "loss": 0.2551, "step": 4461 }, { "epoch": 0.96, "learning_rate": 7.136390526384817e-08, "loss": 0.2682, "step": 4462 }, { "epoch": 0.96, "learning_rate": 7.053265995619618e-08, "loss": 0.3202, "step": 4463 }, { "epoch": 0.96, "learning_rate": 6.97062670814963e-08, "loss": 0.3461, "step": 4464 }, { "epoch": 0.96, "learning_rate": 6.888472704359661e-08, "loss": 0.3088, "step": 4465 }, { "epoch": 0.96, "learning_rate": 6.806804024397706e-08, "loss": 0.2781, "step": 4466 }, { "epoch": 0.96, "learning_rate": 6.725620708174396e-08, "loss": 0.2945, "step": 4467 }, { "epoch": 0.96, "learning_rate": 6.644922795363218e-08, "loss": 0.2771, "step": 4468 }, { "epoch": 0.96, "learning_rate": 6.564710325400403e-08, "loss": 0.2416, "step": 4469 }, { "epoch": 0.96, "learning_rate": 6.484983337485151e-08, "loss": 0.2721, "step": 4470 }, { "epoch": 0.96, "learning_rate": 6.405741870579185e-08, "loss": 0.3069, "step": 4471 }, { "epoch": 0.97, "learning_rate": 6.326985963406862e-08, "loss": 0.2801, "step": 4472 }, { "epoch": 0.97, "learning_rate": 6.248715654455618e-08, "loss": 0.2525, "step": 4473 }, { "epoch": 0.97, "learning_rate": 6.170930981975298e-08, "loss": 0.2438, "step": 4474 }, { "epoch": 0.97, "learning_rate": 6.09363198397861e-08, "loss": 0.2816, "step": 4475 }, { "epoch": 0.97, "learning_rate": 6.016818698240668e-08, "loss": 0.3179, "step": 4476 }, { "epoch": 0.97, "learning_rate": 5.940491162299444e-08, "loss": 0.2725, "step": 4477 }, { "epoch": 0.97, "learning_rate": 5.8646494134554366e-08, "loss": 0.2746, "step": 4478 }, { "epoch": 0.97, "learning_rate": 5.7892934887717746e-08, "loss": 0.2936, "step": 4479 }, { "epoch": 0.97, "learning_rate": 5.7144234250741116e-08, "loss": 0.2659, "step": 4480 }, { "epoch": 0.97, "learning_rate": 5.640039258950736e-08, "loss": 0.241, "step": 4481 }, { "epoch": 0.97, "learning_rate": 5.56614102675257e-08, "loss": 0.2543, "step": 4482 }, { "epoch": 0.97, "learning_rate": 5.492728764592725e-08, "loss": 0.2513, "step": 4483 }, { "epoch": 0.97, "learning_rate": 5.4198025083472825e-08, "loss": 0.2739, "step": 4484 }, { "epoch": 0.97, "learning_rate": 5.3473622936545124e-08, "loss": 0.2536, "step": 4485 }, { "epoch": 0.97, "learning_rate": 5.275408155914985e-08, "loss": 0.276, "step": 4486 }, { "epoch": 0.97, "learning_rate": 5.203940130292462e-08, "loss": 0.3032, "step": 4487 }, { "epoch": 0.97, "learning_rate": 5.132958251712228e-08, "loss": 0.2428, "step": 4488 }, { "epoch": 0.97, "learning_rate": 5.062462554862757e-08, "loss": 0.2916, "step": 4489 }, { "epoch": 0.97, "learning_rate": 4.99245307419427e-08, "loss": 0.3517, "step": 4490 }, { "epoch": 0.97, "learning_rate": 4.9229298439199546e-08, "loss": 0.2815, "step": 4491 }, { "epoch": 0.97, "learning_rate": 4.853892898015078e-08, "loss": 0.3044, "step": 4492 }, { "epoch": 0.97, "learning_rate": 4.785342270217319e-08, "loss": 0.2921, "step": 4493 }, { "epoch": 0.97, "learning_rate": 4.7172779940266586e-08, "loss": 0.2934, "step": 4494 }, { "epoch": 0.97, "learning_rate": 4.649700102705379e-08, "loss": 0.2283, "step": 4495 }, { "epoch": 0.97, "learning_rate": 4.5826086292782845e-08, "loss": 0.2785, "step": 4496 }, { "epoch": 0.97, "learning_rate": 4.51600360653226e-08, "loss": 0.2728, "step": 4497 }, { "epoch": 0.97, "learning_rate": 4.44988506701649e-08, "loss": 0.2774, "step": 4498 }, { "epoch": 0.97, "learning_rate": 4.3842530430424636e-08, "loss": 0.2905, "step": 4499 }, { "epoch": 0.97, "learning_rate": 4.319107566683855e-08, "loss": 0.2883, "step": 4500 }, { "epoch": 0.97, "learning_rate": 4.254448669776756e-08, "loss": 0.2617, "step": 4501 }, { "epoch": 0.97, "learning_rate": 4.190276383919223e-08, "loss": 0.2695, "step": 4502 }, { "epoch": 0.97, "learning_rate": 4.126590740471614e-08, "loss": 0.2812, "step": 4503 }, { "epoch": 0.97, "learning_rate": 4.0633917705564795e-08, "loss": 0.3184, "step": 4504 }, { "epoch": 0.97, "learning_rate": 4.000679505058558e-08, "loss": 0.2135, "step": 4505 }, { "epoch": 0.97, "learning_rate": 3.93845397462489e-08, "loss": 0.2938, "step": 4506 }, { "epoch": 0.97, "learning_rate": 3.8767152096641504e-08, "loss": 0.2722, "step": 4507 }, { "epoch": 0.97, "learning_rate": 3.8154632403476496e-08, "loss": 0.2918, "step": 4508 }, { "epoch": 0.97, "learning_rate": 3.754698096608666e-08, "loss": 0.3189, "step": 4509 }, { "epoch": 0.97, "learning_rate": 3.694419808142335e-08, "loss": 0.3174, "step": 4510 }, { "epoch": 0.97, "learning_rate": 3.6346284044063155e-08, "loss": 0.3238, "step": 4511 }, { "epoch": 0.97, "learning_rate": 3.5753239146198994e-08, "loss": 0.2759, "step": 4512 }, { "epoch": 0.97, "learning_rate": 3.5165063677645715e-08, "loss": 0.3096, "step": 4513 }, { "epoch": 0.97, "learning_rate": 3.4581757925841174e-08, "loss": 0.293, "step": 4514 }, { "epoch": 0.97, "learning_rate": 3.400332217583957e-08, "loss": 0.3019, "step": 4515 }, { "epoch": 0.97, "learning_rate": 3.34297567103159e-08, "loss": 0.2392, "step": 4516 }, { "epoch": 0.97, "learning_rate": 3.286106180956816e-08, "loss": 0.3346, "step": 4517 }, { "epoch": 0.98, "learning_rate": 3.2297237751509615e-08, "loss": 0.3198, "step": 4518 }, { "epoch": 0.98, "learning_rate": 3.173828481167762e-08, "loss": 0.3179, "step": 4519 }, { "epoch": 0.98, "learning_rate": 3.118420326322591e-08, "loss": 0.3265, "step": 4520 }, { "epoch": 0.98, "learning_rate": 3.063499337692788e-08, "loss": 0.2605, "step": 4521 }, { "epoch": 0.98, "learning_rate": 3.0090655421178835e-08, "loss": 0.2689, "step": 4522 }, { "epoch": 0.98, "learning_rate": 2.9551189661990442e-08, "loss": 0.2634, "step": 4523 }, { "epoch": 0.98, "learning_rate": 2.9016596362994032e-08, "loss": 0.2394, "step": 4524 }, { "epoch": 0.98, "learning_rate": 2.8486875785439516e-08, "loss": 0.2903, "step": 4525 }, { "epoch": 0.98, "learning_rate": 2.796202818819871e-08, "loss": 0.302, "step": 4526 }, { "epoch": 0.98, "learning_rate": 2.744205382775644e-08, "loss": 0.251, "step": 4527 }, { "epoch": 0.98, "learning_rate": 2.6926952958221673e-08, "loss": 0.2836, "step": 4528 }, { "epoch": 0.98, "learning_rate": 2.6416725831316383e-08, "loss": 0.2644, "step": 4529 }, { "epoch": 0.98, "learning_rate": 2.591137269638555e-08, "loss": 0.291, "step": 4530 }, { "epoch": 0.98, "learning_rate": 2.5410893800390523e-08, "loss": 0.282, "step": 4531 }, { "epoch": 0.98, "learning_rate": 2.4915289387908993e-08, "loss": 0.334, "step": 4532 }, { "epoch": 0.98, "learning_rate": 2.442455970113833e-08, "loss": 0.2581, "step": 4533 }, { "epoch": 0.98, "learning_rate": 2.3938704979894477e-08, "loss": 0.2959, "step": 4534 }, { "epoch": 0.98, "learning_rate": 2.3457725461607518e-08, "loss": 0.2371, "step": 4535 }, { "epoch": 0.98, "learning_rate": 2.298162138132942e-08, "loss": 0.3659, "step": 4536 }, { "epoch": 0.98, "learning_rate": 2.2510392971727414e-08, "loss": 0.2597, "step": 4537 }, { "epoch": 0.98, "learning_rate": 2.204404046308506e-08, "loss": 0.3057, "step": 4538 }, { "epoch": 0.98, "learning_rate": 2.1582564083305613e-08, "loss": 0.3128, "step": 4539 }, { "epoch": 0.98, "learning_rate": 2.1125964057906456e-08, "loss": 0.2802, "step": 4540 }, { "epoch": 0.98, "learning_rate": 2.0674240610024652e-08, "loss": 0.285, "step": 4541 }, { "epoch": 0.98, "learning_rate": 2.022739396041362e-08, "loss": 0.2629, "step": 4542 }, { "epoch": 0.98, "learning_rate": 1.9785424327440906e-08, "loss": 0.2757, "step": 4543 }, { "epoch": 0.98, "learning_rate": 1.9348331927094844e-08, "loss": 0.2759, "step": 4544 }, { "epoch": 0.98, "learning_rate": 1.8916116972976793e-08, "loss": 0.3244, "step": 4545 }, { "epoch": 0.98, "learning_rate": 1.848877967630669e-08, "loss": 0.2803, "step": 4546 }, { "epoch": 0.98, "learning_rate": 1.806632024591859e-08, "loss": 0.2479, "step": 4547 }, { "epoch": 0.98, "learning_rate": 1.7648738888266237e-08, "loss": 0.2929, "step": 4548 }, { "epoch": 0.98, "learning_rate": 1.7236035807416397e-08, "loss": 0.2799, "step": 4549 }, { "epoch": 0.98, "learning_rate": 1.6828211205054402e-08, "loss": 0.2948, "step": 4550 }, { "epoch": 0.98, "learning_rate": 1.64252652804775e-08, "loss": 0.2965, "step": 4551 }, { "epoch": 0.98, "learning_rate": 1.6027198230603723e-08, "loss": 0.3127, "step": 4552 }, { "epoch": 0.98, "learning_rate": 1.5634010249964137e-08, "loss": 0.2956, "step": 4553 }, { "epoch": 0.98, "learning_rate": 1.5245701530705036e-08, "loss": 0.2487, "step": 4554 }, { "epoch": 0.98, "learning_rate": 1.4862272262589073e-08, "loss": 0.2763, "step": 4555 }, { "epoch": 0.98, "learning_rate": 1.4483722632995245e-08, "loss": 0.2306, "step": 4556 }, { "epoch": 0.98, "learning_rate": 1.4110052826915578e-08, "loss": 0.2948, "step": 4557 }, { "epoch": 0.98, "learning_rate": 1.374126302696066e-08, "loss": 0.3257, "step": 4558 }, { "epoch": 0.98, "learning_rate": 1.3377353413354111e-08, "loss": 0.3051, "step": 4559 }, { "epoch": 0.98, "learning_rate": 1.3018324163932561e-08, "loss": 0.2741, "step": 4560 }, { "epoch": 0.98, "learning_rate": 1.2664175454153438e-08, "loss": 0.2677, "step": 4561 }, { "epoch": 0.98, "learning_rate": 1.2314907457083858e-08, "loss": 0.298, "step": 4562 }, { "epoch": 0.98, "learning_rate": 1.1970520343408398e-08, "loss": 0.2757, "step": 4563 }, { "epoch": 0.99, "learning_rate": 1.1631014281424658e-08, "loss": 0.2676, "step": 4564 }, { "epoch": 0.99, "learning_rate": 1.1296389437046585e-08, "loss": 0.2937, "step": 4565 }, { "epoch": 0.99, "learning_rate": 1.096664597380226e-08, "loss": 0.2559, "step": 4566 }, { "epoch": 0.99, "learning_rate": 1.0641784052832783e-08, "loss": 0.2526, "step": 4567 }, { "epoch": 0.99, "learning_rate": 1.0321803832897825e-08, "loss": 0.3216, "step": 4568 }, { "epoch": 0.99, "learning_rate": 1.0006705470365641e-08, "loss": 0.2728, "step": 4569 }, { "epoch": 0.99, "learning_rate": 9.696489119221942e-09, "loss": 0.2539, "step": 4570 }, { "epoch": 0.99, "learning_rate": 9.39115493106879e-09, "loss": 0.2787, "step": 4571 }, { "epoch": 0.99, "learning_rate": 9.090703055117944e-09, "loss": 0.3001, "step": 4572 }, { "epoch": 0.99, "learning_rate": 8.795133638197507e-09, "loss": 0.2299, "step": 4573 }, { "epoch": 0.99, "learning_rate": 8.504446824749712e-09, "loss": 0.2518, "step": 4574 }, { "epoch": 0.99, "learning_rate": 8.21864275682982e-09, "loss": 0.2684, "step": 4575 }, { "epoch": 0.99, "learning_rate": 7.937721574108326e-09, "loss": 0.2757, "step": 4576 }, { "epoch": 0.99, "learning_rate": 7.661683413868748e-09, "loss": 0.251, "step": 4577 }, { "epoch": 0.99, "learning_rate": 7.390528411007625e-09, "loss": 0.2408, "step": 4578 }, { "epoch": 0.99, "learning_rate": 7.124256698034515e-09, "loss": 0.3196, "step": 4579 }, { "epoch": 0.99, "learning_rate": 6.862868405076439e-09, "loss": 0.265, "step": 4580 }, { "epoch": 0.99, "learning_rate": 6.6063636598689974e-09, "loss": 0.2603, "step": 4581 }, { "epoch": 0.99, "learning_rate": 6.3547425877652505e-09, "loss": 0.3537, "step": 4582 }, { "epoch": 0.99, "learning_rate": 6.10800531172906e-09, "loss": 0.2845, "step": 4583 }, { "epoch": 0.99, "learning_rate": 5.866151952338417e-09, "loss": 0.284, "step": 4584 }, { "epoch": 0.99, "learning_rate": 5.629182627784336e-09, "loss": 0.2852, "step": 4585 }, { "epoch": 0.99, "learning_rate": 5.397097453873068e-09, "loss": 0.2455, "step": 4586 }, { "epoch": 0.99, "learning_rate": 5.169896544020558e-09, "loss": 0.2809, "step": 4587 }, { "epoch": 0.99, "learning_rate": 4.947580009257991e-09, "loss": 0.2538, "step": 4588 }, { "epoch": 0.99, "learning_rate": 4.7301479582295695e-09, "loss": 0.276, "step": 4589 }, { "epoch": 0.99, "learning_rate": 4.51760049719252e-09, "loss": 0.2472, "step": 4590 }, { "epoch": 0.99, "learning_rate": 4.309937730015978e-09, "loss": 0.2956, "step": 4591 }, { "epoch": 0.99, "learning_rate": 4.10715975818321e-09, "loss": 0.2697, "step": 4592 }, { "epoch": 0.99, "learning_rate": 3.909266680790502e-09, "loss": 0.2921, "step": 4593 }, { "epoch": 0.99, "learning_rate": 3.716258594543831e-09, "loss": 0.2933, "step": 4594 }, { "epoch": 0.99, "learning_rate": 3.5281355937677454e-09, "loss": 0.2637, "step": 4595 }, { "epoch": 0.99, "learning_rate": 3.3448977703931516e-09, "loss": 0.2756, "step": 4596 }, { "epoch": 0.99, "learning_rate": 3.166545213968419e-09, "loss": 0.3188, "step": 4597 }, { "epoch": 0.99, "learning_rate": 2.993078011651607e-09, "loss": 0.2566, "step": 4598 }, { "epoch": 0.99, "learning_rate": 2.824496248214903e-09, "loss": 0.333, "step": 4599 }, { "epoch": 0.99, "learning_rate": 2.6608000060424077e-09, "loss": 0.3048, "step": 4600 }, { "epoch": 0.99, "learning_rate": 2.5019893651301307e-09, "loss": 0.2676, "step": 4601 }, { "epoch": 0.99, "learning_rate": 2.3480644030893227e-09, "loss": 0.3525, "step": 4602 }, { "epoch": 0.99, "learning_rate": 2.1990251951398144e-09, "loss": 0.2669, "step": 4603 }, { "epoch": 0.99, "learning_rate": 2.054871814116677e-09, "loss": 0.2861, "step": 4604 }, { "epoch": 0.99, "learning_rate": 1.915604330464671e-09, "loss": 0.2941, "step": 4605 }, { "epoch": 0.99, "learning_rate": 1.781222812243799e-09, "loss": 0.2419, "step": 4606 }, { "epoch": 0.99, "learning_rate": 1.6517273251248633e-09, "loss": 0.2695, "step": 4607 }, { "epoch": 0.99, "learning_rate": 1.5271179323916863e-09, "loss": 0.3212, "step": 4608 }, { "epoch": 0.99, "learning_rate": 1.4073946949377803e-09, "loss": 0.2917, "step": 4609 }, { "epoch": 0.99, "learning_rate": 1.292557671271899e-09, "loss": 0.3196, "step": 4610 }, { "epoch": 1.0, "learning_rate": 1.1826069175135956e-09, "loss": 0.3667, "step": 4611 }, { "epoch": 1.0, "learning_rate": 1.077542487394334e-09, "loss": 0.2819, "step": 4612 }, { "epoch": 1.0, "learning_rate": 9.773644322597087e-10, "loss": 0.3049, "step": 4613 }, { "epoch": 1.0, "learning_rate": 8.820728010627833e-10, "loss": 0.3015, "step": 4614 }, { "epoch": 1.0, "learning_rate": 7.916676403751933e-10, "loss": 0.3493, "step": 4615 }, { "epoch": 1.0, "learning_rate": 7.061489943738231e-10, "loss": 0.3, "step": 4616 }, { "epoch": 1.0, "learning_rate": 6.255169048530185e-10, "loss": 0.2587, "step": 4617 }, { "epoch": 1.0, "learning_rate": 5.497714112157049e-10, "loss": 0.2352, "step": 4618 }, { "epoch": 1.0, "learning_rate": 4.789125504778281e-10, "loss": 0.3113, "step": 4619 }, { "epoch": 1.0, "learning_rate": 4.1294035726835434e-10, "loss": 0.3099, "step": 4620 }, { "epoch": 1.0, "learning_rate": 3.5185486382705023e-10, "loss": 0.3191, "step": 4621 }, { "epoch": 1.0, "learning_rate": 2.9565610000448217e-10, "loss": 0.3009, "step": 4622 }, { "epoch": 1.0, "learning_rate": 2.443440932664576e-10, "loss": 0.328, "step": 4623 }, { "epoch": 1.0, "learning_rate": 1.9791886868736343e-10, "loss": 0.3277, "step": 4624 }, { "epoch": 1.0, "learning_rate": 1.5638044895571746e-10, "loss": 0.266, "step": 4625 }, { "epoch": 1.0, "learning_rate": 1.1972885436972725e-10, "loss": 0.3, "step": 4626 }, { "epoch": 1.0, "learning_rate": 8.796410284062085e-11, "loss": 0.2974, "step": 4627 }, { "epoch": 1.0, "learning_rate": 6.10862098926468e-11, "loss": 0.2877, "step": 4628 }, { "epoch": 1.0, "learning_rate": 3.9095188660853714e-11, "loss": 0.3086, "step": 4629 }, { "epoch": 1.0, "learning_rate": 2.199104988998002e-11, "loss": 0.3109, "step": 4630 }, { "epoch": 1.0, "learning_rate": 9.773801941115324e-12, "loss": 0.2658, "step": 4631 }, { "epoch": 1.0, "learning_rate": 2.4434507839288247e-12, "loss": 0.2896, "step": 4632 }, { "epoch": 1.0, "learning_rate": 0.0, "loss": 0.3102, "step": 4633 }, { "epoch": 1.0, "step": 4633, "total_flos": 2.651199508553138e+19, "train_loss": 0.2987559914762609, "train_runtime": 61654.9581, "train_samples_per_second": 19.238, "train_steps_per_second": 0.075 } ], "logging_steps": 1.0, "max_steps": 4633, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 232, "total_flos": 2.651199508553138e+19, "train_batch_size": 16, "trial_name": null, "trial_params": null }