{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.9935483870967743, "eval_steps": 500, "global_step": 696, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 5.714285714285715e-07, "loss": 1.7586, "step": 1 }, { "epoch": 0.01, "learning_rate": 1.142857142857143e-06, "loss": 1.6779, "step": 2 }, { "epoch": 0.01, "learning_rate": 1.7142857142857145e-06, "loss": 1.6773, "step": 3 }, { "epoch": 0.02, "learning_rate": 2.285714285714286e-06, "loss": 1.6802, "step": 4 }, { "epoch": 0.02, "learning_rate": 2.8571428571428573e-06, "loss": 1.6635, "step": 5 }, { "epoch": 0.03, "learning_rate": 3.428571428571429e-06, "loss": 1.6254, "step": 6 }, { "epoch": 0.03, "learning_rate": 4.000000000000001e-06, "loss": 1.5515, "step": 7 }, { "epoch": 0.03, "learning_rate": 4.571428571428572e-06, "loss": 1.5368, "step": 8 }, { "epoch": 0.04, "learning_rate": 5.142857142857142e-06, "loss": 1.5184, "step": 9 }, { "epoch": 0.04, "learning_rate": 5.7142857142857145e-06, "loss": 1.4823, "step": 10 }, { "epoch": 0.05, "learning_rate": 6.285714285714286e-06, "loss": 1.4535, "step": 11 }, { "epoch": 0.05, "learning_rate": 6.857142857142858e-06, "loss": 1.3807, "step": 12 }, { "epoch": 0.06, "learning_rate": 7.428571428571429e-06, "loss": 1.379, "step": 13 }, { "epoch": 0.06, "learning_rate": 8.000000000000001e-06, "loss": 1.3351, "step": 14 }, { "epoch": 0.06, "learning_rate": 8.571428571428571e-06, "loss": 1.3309, "step": 15 }, { "epoch": 0.07, "learning_rate": 9.142857142857144e-06, "loss": 1.2838, "step": 16 }, { "epoch": 0.07, "learning_rate": 9.714285714285715e-06, "loss": 1.2489, "step": 17 }, { "epoch": 0.08, "learning_rate": 1.0285714285714285e-05, "loss": 1.2665, "step": 18 }, { "epoch": 0.08, "learning_rate": 1.0857142857142858e-05, "loss": 1.2752, "step": 19 }, { "epoch": 0.09, "learning_rate": 1.1428571428571429e-05, "loss": 1.2046, "step": 20 }, { "epoch": 0.09, "learning_rate": 1.2e-05, "loss": 1.1956, "step": 21 }, { "epoch": 0.09, "learning_rate": 1.2571428571428572e-05, "loss": 1.1442, "step": 22 }, { "epoch": 0.1, "learning_rate": 1.3142857142857145e-05, "loss": 1.1825, "step": 23 }, { "epoch": 0.1, "learning_rate": 1.3714285714285716e-05, "loss": 1.1286, "step": 24 }, { "epoch": 0.11, "learning_rate": 1.4285714285714287e-05, "loss": 1.1387, "step": 25 }, { "epoch": 0.11, "learning_rate": 1.4857142857142858e-05, "loss": 1.13, "step": 26 }, { "epoch": 0.12, "learning_rate": 1.542857142857143e-05, "loss": 1.1327, "step": 27 }, { "epoch": 0.12, "learning_rate": 1.6000000000000003e-05, "loss": 1.0921, "step": 28 }, { "epoch": 0.12, "learning_rate": 1.6571428571428574e-05, "loss": 1.0775, "step": 29 }, { "epoch": 0.13, "learning_rate": 1.7142857142857142e-05, "loss": 1.0485, "step": 30 }, { "epoch": 0.13, "learning_rate": 1.7714285714285717e-05, "loss": 1.0582, "step": 31 }, { "epoch": 0.14, "learning_rate": 1.8285714285714288e-05, "loss": 1.0672, "step": 32 }, { "epoch": 0.14, "learning_rate": 1.885714285714286e-05, "loss": 1.0185, "step": 33 }, { "epoch": 0.15, "learning_rate": 1.942857142857143e-05, "loss": 1.0485, "step": 34 }, { "epoch": 0.15, "learning_rate": 2e-05, "loss": 1.0544, "step": 35 }, { "epoch": 0.15, "learning_rate": 1.999988705525916e-05, "loss": 1.0075, "step": 36 }, { "epoch": 0.16, "learning_rate": 1.9999548223587948e-05, "loss": 0.9891, "step": 37 }, { "epoch": 0.16, "learning_rate": 1.999898351264021e-05, "loss": 1.0142, "step": 38 }, { "epoch": 0.17, "learning_rate": 1.999819293517218e-05, "loss": 0.995, "step": 39 }, { "epoch": 0.17, "learning_rate": 1.999717650904216e-05, "loss": 0.9475, "step": 40 }, { "epoch": 0.18, "learning_rate": 1.9995934257210156e-05, "loss": 0.9887, "step": 41 }, { "epoch": 0.18, "learning_rate": 1.9994466207737327e-05, "loss": 0.9438, "step": 42 }, { "epoch": 0.18, "learning_rate": 1.9992772393785366e-05, "loss": 0.9707, "step": 43 }, { "epoch": 0.19, "learning_rate": 1.999085285361575e-05, "loss": 0.9634, "step": 44 }, { "epoch": 0.19, "learning_rate": 1.9988707630588876e-05, "loss": 0.9461, "step": 45 }, { "epoch": 0.2, "learning_rate": 1.9986336773163066e-05, "loss": 0.9447, "step": 46 }, { "epoch": 0.2, "learning_rate": 1.9983740334893503e-05, "loss": 0.8977, "step": 47 }, { "epoch": 0.21, "learning_rate": 1.9980918374430995e-05, "loss": 0.8928, "step": 48 }, { "epoch": 0.21, "learning_rate": 1.997787095552066e-05, "loss": 0.8917, "step": 49 }, { "epoch": 0.22, "learning_rate": 1.997459814700049e-05, "loss": 0.9113, "step": 50 }, { "epoch": 0.22, "learning_rate": 1.9971100022799778e-05, "loss": 0.9448, "step": 51 }, { "epoch": 0.22, "learning_rate": 1.996737666193748e-05, "loss": 0.9447, "step": 52 }, { "epoch": 0.23, "learning_rate": 1.9963428148520393e-05, "loss": 0.9315, "step": 53 }, { "epoch": 0.23, "learning_rate": 1.9959254571741288e-05, "loss": 0.9246, "step": 54 }, { "epoch": 0.24, "learning_rate": 1.9954856025876874e-05, "loss": 0.9176, "step": 55 }, { "epoch": 0.24, "learning_rate": 1.995023261028567e-05, "loss": 0.8972, "step": 56 }, { "epoch": 0.25, "learning_rate": 1.994538442940578e-05, "loss": 0.9092, "step": 57 }, { "epoch": 0.25, "learning_rate": 1.99403115927525e-05, "loss": 0.9203, "step": 58 }, { "epoch": 0.25, "learning_rate": 1.9935014214915884e-05, "loss": 0.9033, "step": 59 }, { "epoch": 0.26, "learning_rate": 1.992949241555812e-05, "loss": 0.8863, "step": 60 }, { "epoch": 0.26, "learning_rate": 1.9923746319410852e-05, "loss": 0.8928, "step": 61 }, { "epoch": 0.27, "learning_rate": 1.991777605627234e-05, "loss": 0.875, "step": 62 }, { "epoch": 0.27, "learning_rate": 1.9911581761004556e-05, "loss": 0.8642, "step": 63 }, { "epoch": 0.28, "learning_rate": 1.9905163573530114e-05, "loss": 0.8559, "step": 64 }, { "epoch": 0.28, "learning_rate": 1.9898521638829115e-05, "loss": 0.8661, "step": 65 }, { "epoch": 0.28, "learning_rate": 1.9891656106935878e-05, "loss": 0.8797, "step": 66 }, { "epoch": 0.29, "learning_rate": 1.9884567132935545e-05, "loss": 0.8574, "step": 67 }, { "epoch": 0.29, "learning_rate": 1.987725487696059e-05, "loss": 0.8368, "step": 68 }, { "epoch": 0.3, "learning_rate": 1.986971950418718e-05, "loss": 0.8877, "step": 69 }, { "epoch": 0.3, "learning_rate": 1.9861961184831456e-05, "loss": 0.8518, "step": 70 }, { "epoch": 0.31, "learning_rate": 1.9853980094145698e-05, "loss": 0.8793, "step": 71 }, { "epoch": 0.31, "learning_rate": 1.984577641241435e-05, "loss": 0.8272, "step": 72 }, { "epoch": 0.31, "learning_rate": 1.9837350324949947e-05, "loss": 0.8362, "step": 73 }, { "epoch": 0.32, "learning_rate": 1.9828702022088943e-05, "loss": 0.866, "step": 74 }, { "epoch": 0.32, "learning_rate": 1.981983169918741e-05, "loss": 0.8486, "step": 75 }, { "epoch": 0.33, "learning_rate": 1.981073955661661e-05, "loss": 0.8609, "step": 76 }, { "epoch": 0.33, "learning_rate": 1.9801425799758475e-05, "loss": 0.8226, "step": 77 }, { "epoch": 0.34, "learning_rate": 1.9791890639000976e-05, "loss": 0.8371, "step": 78 }, { "epoch": 0.34, "learning_rate": 1.9782134289733374e-05, "loss": 0.7946, "step": 79 }, { "epoch": 0.34, "learning_rate": 1.977215697234133e-05, "loss": 0.8845, "step": 80 }, { "epoch": 0.35, "learning_rate": 1.9761958912201948e-05, "loss": 0.7966, "step": 81 }, { "epoch": 0.35, "learning_rate": 1.9751540339678686e-05, "loss": 0.8466, "step": 82 }, { "epoch": 0.36, "learning_rate": 1.9740901490116132e-05, "loss": 0.828, "step": 83 }, { "epoch": 0.36, "learning_rate": 1.9730042603834715e-05, "loss": 0.794, "step": 84 }, { "epoch": 0.37, "learning_rate": 1.9718963926125246e-05, "loss": 0.8455, "step": 85 }, { "epoch": 0.37, "learning_rate": 1.9707665707243406e-05, "loss": 0.8334, "step": 86 }, { "epoch": 0.37, "learning_rate": 1.9696148202404073e-05, "loss": 0.7748, "step": 87 }, { "epoch": 0.38, "learning_rate": 1.968441167177557e-05, "loss": 0.7988, "step": 88 }, { "epoch": 0.38, "learning_rate": 1.9672456380473784e-05, "loss": 0.8018, "step": 89 }, { "epoch": 0.39, "learning_rate": 1.966028259855616e-05, "loss": 0.8104, "step": 90 }, { "epoch": 0.39, "learning_rate": 1.964789060101563e-05, "loss": 0.7908, "step": 91 }, { "epoch": 0.4, "learning_rate": 1.963528066777439e-05, "loss": 0.784, "step": 92 }, { "epoch": 0.4, "learning_rate": 1.962245308367756e-05, "loss": 0.7828, "step": 93 }, { "epoch": 0.4, "learning_rate": 1.9609408138486777e-05, "loss": 0.7801, "step": 94 }, { "epoch": 0.41, "learning_rate": 1.959614612687363e-05, "loss": 0.8136, "step": 95 }, { "epoch": 0.41, "learning_rate": 1.9582667348413014e-05, "loss": 0.7945, "step": 96 }, { "epoch": 0.42, "learning_rate": 1.9568972107576358e-05, "loss": 0.787, "step": 97 }, { "epoch": 0.42, "learning_rate": 1.955506071372474e-05, "loss": 0.8, "step": 98 }, { "epoch": 0.43, "learning_rate": 1.9540933481101925e-05, "loss": 0.7672, "step": 99 }, { "epoch": 0.43, "learning_rate": 1.9526590728827232e-05, "loss": 0.7956, "step": 100 }, { "epoch": 0.43, "learning_rate": 1.9512032780888347e-05, "loss": 0.793, "step": 101 }, { "epoch": 0.44, "learning_rate": 1.9497259966134008e-05, "loss": 0.7404, "step": 102 }, { "epoch": 0.44, "learning_rate": 1.9482272618266558e-05, "loss": 0.7726, "step": 103 }, { "epoch": 0.45, "learning_rate": 1.9467071075834422e-05, "loss": 0.7503, "step": 104 }, { "epoch": 0.45, "learning_rate": 1.945165568222445e-05, "loss": 0.7421, "step": 105 }, { "epoch": 0.46, "learning_rate": 1.9436026785654175e-05, "loss": 0.7826, "step": 106 }, { "epoch": 0.46, "learning_rate": 1.9420184739163932e-05, "loss": 0.7774, "step": 107 }, { "epoch": 0.46, "learning_rate": 1.9404129900608884e-05, "loss": 0.798, "step": 108 }, { "epoch": 0.47, "learning_rate": 1.9387862632650945e-05, "loss": 0.7328, "step": 109 }, { "epoch": 0.47, "learning_rate": 1.9371383302750594e-05, "loss": 0.7589, "step": 110 }, { "epoch": 0.48, "learning_rate": 1.9354692283158555e-05, "loss": 0.788, "step": 111 }, { "epoch": 0.48, "learning_rate": 1.933778995090741e-05, "loss": 0.7617, "step": 112 }, { "epoch": 0.49, "learning_rate": 1.932067668780306e-05, "loss": 0.7585, "step": 113 }, { "epoch": 0.49, "learning_rate": 1.930335288041612e-05, "loss": 0.744, "step": 114 }, { "epoch": 0.49, "learning_rate": 1.928581892007318e-05, "loss": 0.7634, "step": 115 }, { "epoch": 0.5, "learning_rate": 1.926807520284796e-05, "loss": 0.7431, "step": 116 }, { "epoch": 0.5, "learning_rate": 1.9250122129552365e-05, "loss": 0.7554, "step": 117 }, { "epoch": 0.51, "learning_rate": 1.923196010572744e-05, "loss": 0.7534, "step": 118 }, { "epoch": 0.51, "learning_rate": 1.9213589541634203e-05, "loss": 0.723, "step": 119 }, { "epoch": 0.52, "learning_rate": 1.9195010852244367e-05, "loss": 0.7624, "step": 120 }, { "epoch": 0.52, "learning_rate": 1.917622445723099e-05, "loss": 0.7538, "step": 121 }, { "epoch": 0.52, "learning_rate": 1.915723078095898e-05, "loss": 0.7327, "step": 122 }, { "epoch": 0.53, "learning_rate": 1.9138030252475487e-05, "loss": 0.7138, "step": 123 }, { "epoch": 0.53, "learning_rate": 1.911862330550027e-05, "loss": 0.7726, "step": 124 }, { "epoch": 0.54, "learning_rate": 1.9099010378415847e-05, "loss": 0.8036, "step": 125 }, { "epoch": 0.54, "learning_rate": 1.9079191914257606e-05, "loss": 0.7442, "step": 126 }, { "epoch": 0.55, "learning_rate": 1.9059168360703805e-05, "loss": 0.7361, "step": 127 }, { "epoch": 0.55, "learning_rate": 1.9038940170065467e-05, "loss": 0.7595, "step": 128 }, { "epoch": 0.55, "learning_rate": 1.9018507799276132e-05, "loss": 0.7403, "step": 129 }, { "epoch": 0.56, "learning_rate": 1.899787170988157e-05, "loss": 0.702, "step": 130 }, { "epoch": 0.56, "learning_rate": 1.8977032368029333e-05, "loss": 0.7371, "step": 131 }, { "epoch": 0.57, "learning_rate": 1.8955990244458234e-05, "loss": 0.7635, "step": 132 }, { "epoch": 0.57, "learning_rate": 1.8934745814487715e-05, "loss": 0.7208, "step": 133 }, { "epoch": 0.58, "learning_rate": 1.8913299558007096e-05, "loss": 0.6919, "step": 134 }, { "epoch": 0.58, "learning_rate": 1.889165195946476e-05, "loss": 0.7421, "step": 135 }, { "epoch": 0.58, "learning_rate": 1.8869803507857186e-05, "loss": 0.7129, "step": 136 }, { "epoch": 0.59, "learning_rate": 1.884775469671791e-05, "loss": 0.7216, "step": 137 }, { "epoch": 0.59, "learning_rate": 1.88255060241064e-05, "loss": 0.7125, "step": 138 }, { "epoch": 0.6, "learning_rate": 1.880305799259675e-05, "loss": 0.6914, "step": 139 }, { "epoch": 0.6, "learning_rate": 1.8780411109266387e-05, "loss": 0.7107, "step": 140 }, { "epoch": 0.61, "learning_rate": 1.8757565885684584e-05, "loss": 0.7369, "step": 141 }, { "epoch": 0.61, "learning_rate": 1.8734522837900917e-05, "loss": 0.7171, "step": 142 }, { "epoch": 0.62, "learning_rate": 1.8711282486433595e-05, "loss": 0.6836, "step": 143 }, { "epoch": 0.62, "learning_rate": 1.8687845356257708e-05, "loss": 0.6784, "step": 144 }, { "epoch": 0.62, "learning_rate": 1.8664211976793384e-05, "loss": 0.6925, "step": 145 }, { "epoch": 0.63, "learning_rate": 1.86403828818938e-05, "loss": 0.6724, "step": 146 }, { "epoch": 0.63, "learning_rate": 1.8616358609833146e-05, "loss": 0.6894, "step": 147 }, { "epoch": 0.64, "learning_rate": 1.859213970329446e-05, "loss": 0.7129, "step": 148 }, { "epoch": 0.64, "learning_rate": 1.8567726709357367e-05, "loss": 0.6725, "step": 149 }, { "epoch": 0.65, "learning_rate": 1.854312017948572e-05, "loss": 0.7012, "step": 150 }, { "epoch": 0.65, "learning_rate": 1.851832066951515e-05, "loss": 0.7045, "step": 151 }, { "epoch": 0.65, "learning_rate": 1.8493328739640497e-05, "loss": 0.684, "step": 152 }, { "epoch": 0.66, "learning_rate": 1.8468144954403176e-05, "loss": 0.6687, "step": 153 }, { "epoch": 0.66, "learning_rate": 1.84427698826784e-05, "loss": 0.6958, "step": 154 }, { "epoch": 0.67, "learning_rate": 1.841720409766235e-05, "loss": 0.6714, "step": 155 }, { "epoch": 0.67, "learning_rate": 1.8391448176859222e-05, "loss": 0.6661, "step": 156 }, { "epoch": 0.68, "learning_rate": 1.836550270206818e-05, "loss": 0.7179, "step": 157 }, { "epoch": 0.68, "learning_rate": 1.8339368259370198e-05, "loss": 0.6841, "step": 158 }, { "epoch": 0.68, "learning_rate": 1.8313045439114855e-05, "loss": 0.6818, "step": 159 }, { "epoch": 0.69, "learning_rate": 1.828653483590697e-05, "loss": 0.6982, "step": 160 }, { "epoch": 0.69, "learning_rate": 1.825983704859319e-05, "loss": 0.6723, "step": 161 }, { "epoch": 0.7, "learning_rate": 1.823295268024844e-05, "loss": 0.6741, "step": 162 }, { "epoch": 0.7, "learning_rate": 1.8205882338162334e-05, "loss": 0.6777, "step": 163 }, { "epoch": 0.71, "learning_rate": 1.8178626633825417e-05, "loss": 0.6828, "step": 164 }, { "epoch": 0.71, "learning_rate": 1.8151186182915384e-05, "loss": 0.6513, "step": 165 }, { "epoch": 0.71, "learning_rate": 1.8123561605283165e-05, "loss": 0.654, "step": 166 }, { "epoch": 0.72, "learning_rate": 1.8095753524938903e-05, "loss": 0.6938, "step": 167 }, { "epoch": 0.72, "learning_rate": 1.806776257003789e-05, "loss": 0.6823, "step": 168 }, { "epoch": 0.73, "learning_rate": 1.803958937286635e-05, "loss": 0.6744, "step": 169 }, { "epoch": 0.73, "learning_rate": 1.8011234569827173e-05, "loss": 0.679, "step": 170 }, { "epoch": 0.74, "learning_rate": 1.7982698801425543e-05, "loss": 0.698, "step": 171 }, { "epoch": 0.74, "learning_rate": 1.795398271225445e-05, "loss": 0.6836, "step": 172 }, { "epoch": 0.74, "learning_rate": 1.7925086950980137e-05, "loss": 0.6787, "step": 173 }, { "epoch": 0.75, "learning_rate": 1.789601217032747e-05, "loss": 0.6665, "step": 174 }, { "epoch": 0.75, "learning_rate": 1.786675902706515e-05, "loss": 0.66, "step": 175 }, { "epoch": 0.76, "learning_rate": 1.783732818199092e-05, "loss": 0.6881, "step": 176 }, { "epoch": 0.76, "learning_rate": 1.7807720299916617e-05, "loss": 0.6535, "step": 177 }, { "epoch": 0.77, "learning_rate": 1.777793604965315e-05, "loss": 0.6471, "step": 178 }, { "epoch": 0.77, "learning_rate": 1.7747976103995403e-05, "loss": 0.6646, "step": 179 }, { "epoch": 0.77, "learning_rate": 1.771784113970704e-05, "loss": 0.6314, "step": 180 }, { "epoch": 0.78, "learning_rate": 1.76875318375052e-05, "loss": 0.6686, "step": 181 }, { "epoch": 0.78, "learning_rate": 1.765704888204515e-05, "loss": 0.6157, "step": 182 }, { "epoch": 0.79, "learning_rate": 1.7626392961904787e-05, "loss": 0.7079, "step": 183 }, { "epoch": 0.79, "learning_rate": 1.7595564769569097e-05, "loss": 0.6686, "step": 184 }, { "epoch": 0.8, "learning_rate": 1.7564565001414523e-05, "loss": 0.6622, "step": 185 }, { "epoch": 0.8, "learning_rate": 1.7533394357693222e-05, "loss": 0.6542, "step": 186 }, { "epoch": 0.8, "learning_rate": 1.7502053542517245e-05, "loss": 0.6737, "step": 187 }, { "epoch": 0.81, "learning_rate": 1.7470543263842644e-05, "loss": 0.6702, "step": 188 }, { "epoch": 0.81, "learning_rate": 1.7438864233453476e-05, "loss": 0.6395, "step": 189 }, { "epoch": 0.82, "learning_rate": 1.740701716694571e-05, "loss": 0.6588, "step": 190 }, { "epoch": 0.82, "learning_rate": 1.737500278371108e-05, "loss": 0.6398, "step": 191 }, { "epoch": 0.83, "learning_rate": 1.734282180692083e-05, "loss": 0.6493, "step": 192 }, { "epoch": 0.83, "learning_rate": 1.731047496350938e-05, "loss": 0.6487, "step": 193 }, { "epoch": 0.83, "learning_rate": 1.72779629841579e-05, "loss": 0.6574, "step": 194 }, { "epoch": 0.84, "learning_rate": 1.7245286603277804e-05, "loss": 0.6306, "step": 195 }, { "epoch": 0.84, "learning_rate": 1.7212446558994163e-05, "loss": 0.6178, "step": 196 }, { "epoch": 0.85, "learning_rate": 1.7179443593129042e-05, "loss": 0.6436, "step": 197 }, { "epoch": 0.85, "learning_rate": 1.714627845118472e-05, "loss": 0.6632, "step": 198 }, { "epoch": 0.86, "learning_rate": 1.7112951882326872e-05, "loss": 0.6403, "step": 199 }, { "epoch": 0.86, "learning_rate": 1.7079464639367634e-05, "loss": 0.6807, "step": 200 }, { "epoch": 0.86, "learning_rate": 1.70458174787486e-05, "loss": 0.6446, "step": 201 }, { "epoch": 0.87, "learning_rate": 1.7012011160523744e-05, "loss": 0.634, "step": 202 }, { "epoch": 0.87, "learning_rate": 1.6978046448342228e-05, "loss": 0.6809, "step": 203 }, { "epoch": 0.88, "learning_rate": 1.694392410943118e-05, "loss": 0.6254, "step": 204 }, { "epoch": 0.88, "learning_rate": 1.690964491457834e-05, "loss": 0.6472, "step": 205 }, { "epoch": 0.89, "learning_rate": 1.687520963811467e-05, "loss": 0.6558, "step": 206 }, { "epoch": 0.89, "learning_rate": 1.6840619057896845e-05, "loss": 0.6424, "step": 207 }, { "epoch": 0.89, "learning_rate": 1.680587395528968e-05, "loss": 0.6443, "step": 208 }, { "epoch": 0.9, "learning_rate": 1.6770975115148506e-05, "loss": 0.6575, "step": 209 }, { "epoch": 0.9, "learning_rate": 1.6735923325801407e-05, "loss": 0.6402, "step": 210 }, { "epoch": 0.91, "learning_rate": 1.670071937903144e-05, "loss": 0.6604, "step": 211 }, { "epoch": 0.91, "learning_rate": 1.666536407005874e-05, "loss": 0.6457, "step": 212 }, { "epoch": 0.92, "learning_rate": 1.6629858197522537e-05, "loss": 0.6121, "step": 213 }, { "epoch": 0.92, "learning_rate": 1.6594202563463152e-05, "loss": 0.6222, "step": 214 }, { "epoch": 0.92, "learning_rate": 1.655839797330385e-05, "loss": 0.6632, "step": 215 }, { "epoch": 0.93, "learning_rate": 1.652244523583267e-05, "loss": 0.6305, "step": 216 }, { "epoch": 0.93, "learning_rate": 1.648634516318413e-05, "loss": 0.6426, "step": 217 }, { "epoch": 0.94, "learning_rate": 1.64500985708209e-05, "loss": 0.6438, "step": 218 }, { "epoch": 0.94, "learning_rate": 1.6413706277515375e-05, "loss": 0.6365, "step": 219 }, { "epoch": 0.95, "learning_rate": 1.6377169105331185e-05, "loss": 0.6433, "step": 220 }, { "epoch": 0.95, "learning_rate": 1.634048787960462e-05, "loss": 0.6328, "step": 221 }, { "epoch": 0.95, "learning_rate": 1.630366342892598e-05, "loss": 0.6161, "step": 222 }, { "epoch": 0.96, "learning_rate": 1.6266696585120884e-05, "loss": 0.6344, "step": 223 }, { "epoch": 0.96, "learning_rate": 1.6229588183231436e-05, "loss": 0.6234, "step": 224 }, { "epoch": 0.97, "learning_rate": 1.6192339061497415e-05, "loss": 0.6374, "step": 225 }, { "epoch": 0.97, "learning_rate": 1.6154950061337295e-05, "loss": 0.6472, "step": 226 }, { "epoch": 0.98, "learning_rate": 1.6117422027329267e-05, "loss": 0.6094, "step": 227 }, { "epoch": 0.98, "learning_rate": 1.6079755807192138e-05, "loss": 0.5943, "step": 228 }, { "epoch": 0.98, "learning_rate": 1.604195225176621e-05, "loss": 0.6191, "step": 229 }, { "epoch": 0.99, "learning_rate": 1.6004012214994036e-05, "loss": 0.6108, "step": 230 }, { "epoch": 0.99, "learning_rate": 1.5965936553901137e-05, "loss": 0.6543, "step": 231 }, { "epoch": 1.0, "learning_rate": 1.592772612857665e-05, "loss": 0.6439, "step": 232 }, { "epoch": 1.0, "learning_rate": 1.5889381802153898e-05, "loss": 0.6041, "step": 233 }, { "epoch": 1.01, "learning_rate": 1.5850904440790872e-05, "loss": 0.5625, "step": 234 }, { "epoch": 1.01, "learning_rate": 1.5812294913650697e-05, "loss": 0.5727, "step": 235 }, { "epoch": 1.02, "learning_rate": 1.5773554092881986e-05, "loss": 0.5793, "step": 236 }, { "epoch": 1.02, "learning_rate": 1.5734682853599124e-05, "loss": 0.572, "step": 237 }, { "epoch": 1.02, "learning_rate": 1.5695682073862527e-05, "loss": 0.5536, "step": 238 }, { "epoch": 1.03, "learning_rate": 1.5656552634658776e-05, "loss": 0.5855, "step": 239 }, { "epoch": 1.03, "learning_rate": 1.561729541988076e-05, "loss": 0.57, "step": 240 }, { "epoch": 1.04, "learning_rate": 1.5577911316307658e-05, "loss": 0.5916, "step": 241 }, { "epoch": 1.04, "learning_rate": 1.553840121358495e-05, "loss": 0.5818, "step": 242 }, { "epoch": 1.05, "learning_rate": 1.5498766004204294e-05, "loss": 0.5852, "step": 243 }, { "epoch": 1.05, "learning_rate": 1.545900658348338e-05, "loss": 0.5773, "step": 244 }, { "epoch": 1.05, "learning_rate": 1.541912384954571e-05, "loss": 0.5863, "step": 245 }, { "epoch": 1.06, "learning_rate": 1.5379118703300285e-05, "loss": 0.6045, "step": 246 }, { "epoch": 1.06, "learning_rate": 1.533899204842128e-05, "loss": 0.5633, "step": 247 }, { "epoch": 1.07, "learning_rate": 1.5298744791327632e-05, "loss": 0.5665, "step": 248 }, { "epoch": 1.07, "learning_rate": 1.5258377841162534e-05, "loss": 0.5866, "step": 249 }, { "epoch": 1.08, "learning_rate": 1.5217892109772936e-05, "loss": 0.5643, "step": 250 }, { "epoch": 1.08, "learning_rate": 1.5177288511688927e-05, "loss": 0.5773, "step": 251 }, { "epoch": 1.08, "learning_rate": 1.5136567964103077e-05, "loss": 0.5645, "step": 252 }, { "epoch": 1.09, "learning_rate": 1.5095731386849725e-05, "loss": 0.5798, "step": 253 }, { "epoch": 1.09, "learning_rate": 1.5054779702384199e-05, "loss": 0.5377, "step": 254 }, { "epoch": 1.1, "learning_rate": 1.5013713835761978e-05, "loss": 0.5762, "step": 255 }, { "epoch": 1.1, "learning_rate": 1.4972534714617792e-05, "loss": 0.5555, "step": 256 }, { "epoch": 1.11, "learning_rate": 1.4931243269144673e-05, "loss": 0.5759, "step": 257 }, { "epoch": 1.11, "learning_rate": 1.4889840432072947e-05, "loss": 0.5759, "step": 258 }, { "epoch": 1.11, "learning_rate": 1.4848327138649153e-05, "loss": 0.5947, "step": 259 }, { "epoch": 1.12, "learning_rate": 1.480670432661492e-05, "loss": 0.5476, "step": 260 }, { "epoch": 1.12, "learning_rate": 1.4764972936185798e-05, "loss": 0.5819, "step": 261 }, { "epoch": 1.13, "learning_rate": 1.4723133910029998e-05, "loss": 0.5923, "step": 262 }, { "epoch": 1.13, "learning_rate": 1.4681188193247118e-05, "loss": 0.5393, "step": 263 }, { "epoch": 1.14, "learning_rate": 1.4639136733346777e-05, "loss": 0.5497, "step": 264 }, { "epoch": 1.14, "learning_rate": 1.4596980480227223e-05, "loss": 0.5612, "step": 265 }, { "epoch": 1.14, "learning_rate": 1.4554720386153872e-05, "loss": 0.5657, "step": 266 }, { "epoch": 1.15, "learning_rate": 1.4512357405737798e-05, "loss": 0.5774, "step": 267 }, { "epoch": 1.15, "learning_rate": 1.4469892495914174e-05, "loss": 0.5703, "step": 268 }, { "epoch": 1.16, "learning_rate": 1.4427326615920642e-05, "loss": 0.566, "step": 269 }, { "epoch": 1.16, "learning_rate": 1.4384660727275664e-05, "loss": 0.57, "step": 270 }, { "epoch": 1.17, "learning_rate": 1.4341895793756784e-05, "loss": 0.5528, "step": 271 }, { "epoch": 1.17, "learning_rate": 1.4299032781378865e-05, "loss": 0.5695, "step": 272 }, { "epoch": 1.17, "learning_rate": 1.425607265837228e-05, "loss": 0.539, "step": 273 }, { "epoch": 1.18, "learning_rate": 1.4213016395161017e-05, "loss": 0.5318, "step": 274 }, { "epoch": 1.18, "learning_rate": 1.416986496434077e-05, "loss": 0.562, "step": 275 }, { "epoch": 1.19, "learning_rate": 1.4126619340656981e-05, "loss": 0.5444, "step": 276 }, { "epoch": 1.19, "learning_rate": 1.4083280500982797e-05, "loss": 0.551, "step": 277 }, { "epoch": 1.2, "learning_rate": 1.4039849424297023e-05, "loss": 0.5726, "step": 278 }, { "epoch": 1.2, "learning_rate": 1.3996327091661996e-05, "loss": 0.5391, "step": 279 }, { "epoch": 1.2, "learning_rate": 1.3952714486201436e-05, "loss": 0.5638, "step": 280 }, { "epoch": 1.21, "learning_rate": 1.3909012593078224e-05, "loss": 0.5458, "step": 281 }, { "epoch": 1.21, "learning_rate": 1.3865222399472155e-05, "loss": 0.5432, "step": 282 }, { "epoch": 1.22, "learning_rate": 1.3821344894557653e-05, "loss": 0.5435, "step": 283 }, { "epoch": 1.22, "learning_rate": 1.3777381069481397e-05, "loss": 0.526, "step": 284 }, { "epoch": 1.23, "learning_rate": 1.3733331917339952e-05, "loss": 0.5597, "step": 285 }, { "epoch": 1.23, "learning_rate": 1.3689198433157334e-05, "loss": 0.5641, "step": 286 }, { "epoch": 1.23, "learning_rate": 1.3644981613862524e-05, "loss": 0.5445, "step": 287 }, { "epoch": 1.24, "learning_rate": 1.3600682458266972e-05, "loss": 0.5428, "step": 288 }, { "epoch": 1.24, "learning_rate": 1.3556301967041999e-05, "loss": 0.5693, "step": 289 }, { "epoch": 1.25, "learning_rate": 1.3511841142696222e-05, "loss": 0.5643, "step": 290 }, { "epoch": 1.25, "learning_rate": 1.34673009895529e-05, "loss": 0.5534, "step": 291 }, { "epoch": 1.26, "learning_rate": 1.3422682513727246e-05, "loss": 0.568, "step": 292 }, { "epoch": 1.26, "learning_rate": 1.3377986723103694e-05, "loss": 0.5612, "step": 293 }, { "epoch": 1.26, "learning_rate": 1.333321462731314e-05, "loss": 0.5436, "step": 294 }, { "epoch": 1.27, "learning_rate": 1.328836723771014e-05, "loss": 0.5599, "step": 295 }, { "epoch": 1.27, "learning_rate": 1.3243445567350048e-05, "loss": 0.5338, "step": 296 }, { "epoch": 1.28, "learning_rate": 1.3198450630966153e-05, "loss": 0.5526, "step": 297 }, { "epoch": 1.28, "learning_rate": 1.3153383444946736e-05, "loss": 0.5484, "step": 298 }, { "epoch": 1.29, "learning_rate": 1.310824502731213e-05, "loss": 0.574, "step": 299 }, { "epoch": 1.29, "learning_rate": 1.3063036397691708e-05, "loss": 0.5819, "step": 300 }, { "epoch": 1.29, "learning_rate": 1.3017758577300863e-05, "loss": 0.5703, "step": 301 }, { "epoch": 1.3, "learning_rate": 1.2972412588917931e-05, "loss": 0.5207, "step": 302 }, { "epoch": 1.3, "learning_rate": 1.2926999456861099e-05, "loss": 0.5325, "step": 303 }, { "epoch": 1.31, "learning_rate": 1.2881520206965245e-05, "loss": 0.553, "step": 304 }, { "epoch": 1.31, "learning_rate": 1.2835975866558793e-05, "loss": 0.5561, "step": 305 }, { "epoch": 1.32, "learning_rate": 1.2790367464440484e-05, "loss": 0.5485, "step": 306 }, { "epoch": 1.32, "learning_rate": 1.2744696030856154e-05, "loss": 0.5312, "step": 307 }, { "epoch": 1.32, "learning_rate": 1.2698962597475446e-05, "loss": 0.5298, "step": 308 }, { "epoch": 1.33, "learning_rate": 1.265316819736852e-05, "loss": 0.5316, "step": 309 }, { "epoch": 1.33, "learning_rate": 1.2607313864982698e-05, "loss": 0.5483, "step": 310 }, { "epoch": 1.34, "learning_rate": 1.2561400636119126e-05, "loss": 0.5588, "step": 311 }, { "epoch": 1.34, "learning_rate": 1.2515429547909349e-05, "loss": 0.5388, "step": 312 }, { "epoch": 1.35, "learning_rate": 1.2469401638791892e-05, "loss": 0.5223, "step": 313 }, { "epoch": 1.35, "learning_rate": 1.2423317948488814e-05, "loss": 0.5238, "step": 314 }, { "epoch": 1.35, "learning_rate": 1.23771795179822e-05, "loss": 0.5034, "step": 315 }, { "epoch": 1.36, "learning_rate": 1.2330987389490671e-05, "loss": 0.52, "step": 316 }, { "epoch": 1.36, "learning_rate": 1.2284742606445818e-05, "loss": 0.5441, "step": 317 }, { "epoch": 1.37, "learning_rate": 1.2238446213468656e-05, "loss": 0.5618, "step": 318 }, { "epoch": 1.37, "learning_rate": 1.2192099256345999e-05, "loss": 0.5472, "step": 319 }, { "epoch": 1.38, "learning_rate": 1.2145702782006863e-05, "loss": 0.5502, "step": 320 }, { "epoch": 1.38, "learning_rate": 1.20992578384988e-05, "loss": 0.5259, "step": 321 }, { "epoch": 1.38, "learning_rate": 1.2052765474964232e-05, "loss": 0.5492, "step": 322 }, { "epoch": 1.39, "learning_rate": 1.200622674161675e-05, "loss": 0.5268, "step": 323 }, { "epoch": 1.39, "learning_rate": 1.195964268971739e-05, "loss": 0.5498, "step": 324 }, { "epoch": 1.4, "learning_rate": 1.1913014371550881e-05, "loss": 0.534, "step": 325 }, { "epoch": 1.4, "learning_rate": 1.1866342840401892e-05, "loss": 0.5294, "step": 326 }, { "epoch": 1.41, "learning_rate": 1.1819629150531218e-05, "loss": 0.5479, "step": 327 }, { "epoch": 1.41, "learning_rate": 1.177287435715198e-05, "loss": 0.5411, "step": 328 }, { "epoch": 1.42, "learning_rate": 1.1726079516405778e-05, "loss": 0.5548, "step": 329 }, { "epoch": 1.42, "learning_rate": 1.1679245685338847e-05, "loss": 0.5378, "step": 330 }, { "epoch": 1.42, "learning_rate": 1.1632373921878168e-05, "loss": 0.5503, "step": 331 }, { "epoch": 1.43, "learning_rate": 1.1585465284807575e-05, "loss": 0.5391, "step": 332 }, { "epoch": 1.43, "learning_rate": 1.1538520833743845e-05, "loss": 0.5496, "step": 333 }, { "epoch": 1.44, "learning_rate": 1.1491541629112746e-05, "loss": 0.5334, "step": 334 }, { "epoch": 1.44, "learning_rate": 1.1444528732125098e-05, "loss": 0.5329, "step": 335 }, { "epoch": 1.45, "learning_rate": 1.139748320475279e-05, "loss": 0.5368, "step": 336 }, { "epoch": 1.45, "learning_rate": 1.1350406109704806e-05, "loss": 0.5294, "step": 337 }, { "epoch": 1.45, "learning_rate": 1.1303298510403206e-05, "loss": 0.5149, "step": 338 }, { "epoch": 1.46, "learning_rate": 1.1256161470959107e-05, "loss": 0.5439, "step": 339 }, { "epoch": 1.46, "learning_rate": 1.1208996056148646e-05, "loss": 0.5588, "step": 340 }, { "epoch": 1.47, "learning_rate": 1.1161803331388942e-05, "loss": 0.555, "step": 341 }, { "epoch": 1.47, "learning_rate": 1.1114584362714006e-05, "loss": 0.5328, "step": 342 }, { "epoch": 1.48, "learning_rate": 1.1067340216750669e-05, "loss": 0.5317, "step": 343 }, { "epoch": 1.48, "learning_rate": 1.1020071960694498e-05, "loss": 0.5289, "step": 344 }, { "epoch": 1.48, "learning_rate": 1.0972780662285683e-05, "loss": 0.549, "step": 345 }, { "epoch": 1.49, "learning_rate": 1.0925467389784905e-05, "loss": 0.5388, "step": 346 }, { "epoch": 1.49, "learning_rate": 1.087813321194923e-05, "loss": 0.5412, "step": 347 }, { "epoch": 1.5, "learning_rate": 1.0830779198007943e-05, "loss": 0.5165, "step": 348 }, { "epoch": 1.5, "learning_rate": 1.0783406417638418e-05, "loss": 0.5354, "step": 349 }, { "epoch": 1.51, "learning_rate": 1.0736015940941928e-05, "loss": 0.5315, "step": 350 }, { "epoch": 1.51, "learning_rate": 1.0688608838419496e-05, "loss": 0.5309, "step": 351 }, { "epoch": 1.51, "learning_rate": 1.0641186180947709e-05, "loss": 0.5495, "step": 352 }, { "epoch": 1.52, "learning_rate": 1.0593749039754513e-05, "loss": 0.5096, "step": 353 }, { "epoch": 1.52, "learning_rate": 1.0546298486395033e-05, "loss": 0.5242, "step": 354 }, { "epoch": 1.53, "learning_rate": 1.0498835592727357e-05, "loss": 0.5238, "step": 355 }, { "epoch": 1.53, "learning_rate": 1.0451361430888336e-05, "loss": 0.5417, "step": 356 }, { "epoch": 1.54, "learning_rate": 1.0403877073269349e-05, "loss": 0.528, "step": 357 }, { "epoch": 1.54, "learning_rate": 1.0356383592492086e-05, "loss": 0.5333, "step": 358 }, { "epoch": 1.54, "learning_rate": 1.0308882061384325e-05, "loss": 0.5303, "step": 359 }, { "epoch": 1.55, "learning_rate": 1.0261373552955689e-05, "loss": 0.539, "step": 360 }, { "epoch": 1.55, "learning_rate": 1.0213859140373412e-05, "loss": 0.5355, "step": 361 }, { "epoch": 1.56, "learning_rate": 1.0166339896938099e-05, "loss": 0.5039, "step": 362 }, { "epoch": 1.56, "learning_rate": 1.0118816896059474e-05, "loss": 0.5284, "step": 363 }, { "epoch": 1.57, "learning_rate": 1.0071291211232143e-05, "loss": 0.5241, "step": 364 }, { "epoch": 1.57, "learning_rate": 1.0023763916011337e-05, "loss": 0.5059, "step": 365 }, { "epoch": 1.57, "learning_rate": 9.976236083988664e-06, "loss": 0.5217, "step": 366 }, { "epoch": 1.58, "learning_rate": 9.928708788767859e-06, "loss": 0.5185, "step": 367 }, { "epoch": 1.58, "learning_rate": 9.881183103940527e-06, "loss": 0.5423, "step": 368 }, { "epoch": 1.59, "learning_rate": 9.833660103061903e-06, "loss": 0.5428, "step": 369 }, { "epoch": 1.59, "learning_rate": 9.786140859626593e-06, "loss": 0.5288, "step": 370 }, { "epoch": 1.6, "learning_rate": 9.738626447044318e-06, "loss": 0.5138, "step": 371 }, { "epoch": 1.6, "learning_rate": 9.691117938615679e-06, "loss": 0.5244, "step": 372 }, { "epoch": 1.6, "learning_rate": 9.643616407507918e-06, "loss": 0.5294, "step": 373 }, { "epoch": 1.61, "learning_rate": 9.596122926730655e-06, "loss": 0.5318, "step": 374 }, { "epoch": 1.61, "learning_rate": 9.548638569111665e-06, "loss": 0.5053, "step": 375 }, { "epoch": 1.62, "learning_rate": 9.501164407272643e-06, "loss": 0.5206, "step": 376 }, { "epoch": 1.62, "learning_rate": 9.453701513604972e-06, "loss": 0.5063, "step": 377 }, { "epoch": 1.63, "learning_rate": 9.406250960245494e-06, "loss": 0.5282, "step": 378 }, { "epoch": 1.63, "learning_rate": 9.358813819052294e-06, "loss": 0.5103, "step": 379 }, { "epoch": 1.63, "learning_rate": 9.311391161580507e-06, "loss": 0.5304, "step": 380 }, { "epoch": 1.64, "learning_rate": 9.263984059058074e-06, "loss": 0.522, "step": 381 }, { "epoch": 1.64, "learning_rate": 9.216593582361585e-06, "loss": 0.5309, "step": 382 }, { "epoch": 1.65, "learning_rate": 9.169220801992057e-06, "loss": 0.5056, "step": 383 }, { "epoch": 1.65, "learning_rate": 9.121866788050772e-06, "loss": 0.518, "step": 384 }, { "epoch": 1.66, "learning_rate": 9.074532610215099e-06, "loss": 0.5185, "step": 385 }, { "epoch": 1.66, "learning_rate": 9.027219337714324e-06, "loss": 0.5071, "step": 386 }, { "epoch": 1.66, "learning_rate": 8.979928039305504e-06, "loss": 0.5475, "step": 387 }, { "epoch": 1.67, "learning_rate": 8.932659783249333e-06, "loss": 0.5228, "step": 388 }, { "epoch": 1.67, "learning_rate": 8.885415637285998e-06, "loss": 0.4934, "step": 389 }, { "epoch": 1.68, "learning_rate": 8.838196668611058e-06, "loss": 0.5281, "step": 390 }, { "epoch": 1.68, "learning_rate": 8.791003943851353e-06, "loss": 0.5234, "step": 391 }, { "epoch": 1.69, "learning_rate": 8.743838529040898e-06, "loss": 0.5131, "step": 392 }, { "epoch": 1.69, "learning_rate": 8.696701489596797e-06, "loss": 0.5246, "step": 393 }, { "epoch": 1.69, "learning_rate": 8.649593890295196e-06, "loss": 0.5053, "step": 394 }, { "epoch": 1.7, "learning_rate": 8.602516795247213e-06, "loss": 0.53, "step": 395 }, { "epoch": 1.7, "learning_rate": 8.555471267874905e-06, "loss": 0.4878, "step": 396 }, { "epoch": 1.71, "learning_rate": 8.508458370887255e-06, "loss": 0.5222, "step": 397 }, { "epoch": 1.71, "learning_rate": 8.461479166256157e-06, "loss": 0.503, "step": 398 }, { "epoch": 1.72, "learning_rate": 8.414534715192425e-06, "loss": 0.528, "step": 399 }, { "epoch": 1.72, "learning_rate": 8.367626078121837e-06, "loss": 0.5078, "step": 400 }, { "epoch": 1.72, "learning_rate": 8.320754314661158e-06, "loss": 0.5311, "step": 401 }, { "epoch": 1.73, "learning_rate": 8.273920483594225e-06, "loss": 0.5151, "step": 402 }, { "epoch": 1.73, "learning_rate": 8.227125642848023e-06, "loss": 0.5302, "step": 403 }, { "epoch": 1.74, "learning_rate": 8.180370849468784e-06, "loss": 0.4945, "step": 404 }, { "epoch": 1.74, "learning_rate": 8.133657159598108e-06, "loss": 0.5216, "step": 405 }, { "epoch": 1.75, "learning_rate": 8.086985628449119e-06, "loss": 0.5356, "step": 406 }, { "epoch": 1.75, "learning_rate": 8.040357310282615e-06, "loss": 0.5077, "step": 407 }, { "epoch": 1.75, "learning_rate": 7.993773258383251e-06, "loss": 0.5098, "step": 408 }, { "epoch": 1.76, "learning_rate": 7.947234525035772e-06, "loss": 0.5175, "step": 409 }, { "epoch": 1.76, "learning_rate": 7.900742161501204e-06, "loss": 0.4935, "step": 410 }, { "epoch": 1.77, "learning_rate": 7.854297217993138e-06, "loss": 0.5103, "step": 411 }, { "epoch": 1.77, "learning_rate": 7.807900743654003e-06, "loss": 0.5116, "step": 412 }, { "epoch": 1.78, "learning_rate": 7.761553786531346e-06, "loss": 0.5077, "step": 413 }, { "epoch": 1.78, "learning_rate": 7.71525739355418e-06, "loss": 0.4915, "step": 414 }, { "epoch": 1.78, "learning_rate": 7.669012610509334e-06, "loss": 0.4992, "step": 415 }, { "epoch": 1.79, "learning_rate": 7.622820482017804e-06, "loss": 0.4955, "step": 416 }, { "epoch": 1.79, "learning_rate": 7.57668205151119e-06, "loss": 0.5147, "step": 417 }, { "epoch": 1.8, "learning_rate": 7.53059836120811e-06, "loss": 0.5267, "step": 418 }, { "epoch": 1.8, "learning_rate": 7.484570452090655e-06, "loss": 0.5032, "step": 419 }, { "epoch": 1.81, "learning_rate": 7.438599363880874e-06, "loss": 0.5167, "step": 420 }, { "epoch": 1.81, "learning_rate": 7.392686135017302e-06, "loss": 0.525, "step": 421 }, { "epoch": 1.82, "learning_rate": 7.346831802631486e-06, "loss": 0.5262, "step": 422 }, { "epoch": 1.82, "learning_rate": 7.301037402524556e-06, "loss": 0.4958, "step": 423 }, { "epoch": 1.82, "learning_rate": 7.2553039691438475e-06, "loss": 0.5097, "step": 424 }, { "epoch": 1.83, "learning_rate": 7.209632535559519e-06, "loss": 0.5238, "step": 425 }, { "epoch": 1.83, "learning_rate": 7.16402413344121e-06, "loss": 0.5069, "step": 426 }, { "epoch": 1.84, "learning_rate": 7.118479793034758e-06, "loss": 0.497, "step": 427 }, { "epoch": 1.84, "learning_rate": 7.073000543138904e-06, "loss": 0.5294, "step": 428 }, { "epoch": 1.85, "learning_rate": 7.027587411082069e-06, "loss": 0.501, "step": 429 }, { "epoch": 1.85, "learning_rate": 6.9822414226991405e-06, "loss": 0.5157, "step": 430 }, { "epoch": 1.85, "learning_rate": 6.936963602308297e-06, "loss": 0.5141, "step": 431 }, { "epoch": 1.86, "learning_rate": 6.891754972687873e-06, "loss": 0.5122, "step": 432 }, { "epoch": 1.86, "learning_rate": 6.846616555053266e-06, "loss": 0.5152, "step": 433 }, { "epoch": 1.87, "learning_rate": 6.801549369033851e-06, "loss": 0.5131, "step": 434 }, { "epoch": 1.87, "learning_rate": 6.7565544326499535e-06, "loss": 0.5311, "step": 435 }, { "epoch": 1.88, "learning_rate": 6.7116327622898634e-06, "loss": 0.5177, "step": 436 }, { "epoch": 1.88, "learning_rate": 6.666785372686863e-06, "loss": 0.5123, "step": 437 }, { "epoch": 1.88, "learning_rate": 6.62201327689631e-06, "loss": 0.4966, "step": 438 }, { "epoch": 1.89, "learning_rate": 6.577317486272756e-06, "loss": 0.5268, "step": 439 }, { "epoch": 1.89, "learning_rate": 6.532699010447101e-06, "loss": 0.5075, "step": 440 }, { "epoch": 1.9, "learning_rate": 6.488158857303779e-06, "loss": 0.5256, "step": 441 }, { "epoch": 1.9, "learning_rate": 6.443698032958003e-06, "loss": 0.5096, "step": 442 }, { "epoch": 1.91, "learning_rate": 6.39931754173303e-06, "loss": 0.5072, "step": 443 }, { "epoch": 1.91, "learning_rate": 6.355018386137475e-06, "loss": 0.4937, "step": 444 }, { "epoch": 1.91, "learning_rate": 6.310801566842671e-06, "loss": 0.5039, "step": 445 }, { "epoch": 1.92, "learning_rate": 6.266668082660053e-06, "loss": 0.5048, "step": 446 }, { "epoch": 1.92, "learning_rate": 6.222618930518605e-06, "loss": 0.5063, "step": 447 }, { "epoch": 1.93, "learning_rate": 6.1786551054423485e-06, "loss": 0.521, "step": 448 }, { "epoch": 1.93, "learning_rate": 6.134777600527846e-06, "loss": 0.5088, "step": 449 }, { "epoch": 1.94, "learning_rate": 6.09098740692178e-06, "loss": 0.4949, "step": 450 }, { "epoch": 1.94, "learning_rate": 6.047285513798569e-06, "loss": 0.4994, "step": 451 }, { "epoch": 1.94, "learning_rate": 6.0036729083380095e-06, "loss": 0.5304, "step": 452 }, { "epoch": 1.95, "learning_rate": 5.960150575702981e-06, "loss": 0.4975, "step": 453 }, { "epoch": 1.95, "learning_rate": 5.916719499017206e-06, "loss": 0.5023, "step": 454 }, { "epoch": 1.96, "learning_rate": 5.873380659343022e-06, "loss": 0.5051, "step": 455 }, { "epoch": 1.96, "learning_rate": 5.83013503565923e-06, "loss": 0.5074, "step": 456 }, { "epoch": 1.97, "learning_rate": 5.786983604838984e-06, "loss": 0.5042, "step": 457 }, { "epoch": 1.97, "learning_rate": 5.743927341627722e-06, "loss": 0.515, "step": 458 }, { "epoch": 1.97, "learning_rate": 5.700967218621134e-06, "loss": 0.5167, "step": 459 }, { "epoch": 1.98, "learning_rate": 5.658104206243222e-06, "loss": 0.515, "step": 460 }, { "epoch": 1.98, "learning_rate": 5.615339272724338e-06, "loss": 0.4989, "step": 461 }, { "epoch": 1.99, "learning_rate": 5.572673384079361e-06, "loss": 0.4998, "step": 462 }, { "epoch": 1.99, "learning_rate": 5.53010750408583e-06, "loss": 0.5054, "step": 463 }, { "epoch": 2.0, "learning_rate": 5.487642594262203e-06, "loss": 0.4986, "step": 464 }, { "epoch": 2.0, "learning_rate": 5.445279613846132e-06, "loss": 0.4758, "step": 465 }, { "epoch": 2.0, "learning_rate": 5.403019519772781e-06, "loss": 0.4857, "step": 466 }, { "epoch": 2.01, "learning_rate": 5.360863266653228e-06, "loss": 0.4618, "step": 467 }, { "epoch": 2.01, "learning_rate": 5.318811806752884e-06, "loss": 0.4636, "step": 468 }, { "epoch": 2.02, "learning_rate": 5.2768660899700056e-06, "loss": 0.4645, "step": 469 }, { "epoch": 2.02, "learning_rate": 5.235027063814206e-06, "loss": 0.4767, "step": 470 }, { "epoch": 2.03, "learning_rate": 5.193295673385081e-06, "loss": 0.4747, "step": 471 }, { "epoch": 2.03, "learning_rate": 5.151672861350849e-06, "loss": 0.4849, "step": 472 }, { "epoch": 2.03, "learning_rate": 5.1101595679270566e-06, "loss": 0.451, "step": 473 }, { "epoch": 2.04, "learning_rate": 5.068756730855329e-06, "loss": 0.463, "step": 474 }, { "epoch": 2.04, "learning_rate": 5.027465285382213e-06, "loss": 0.4643, "step": 475 }, { "epoch": 2.05, "learning_rate": 4.986286164238026e-06, "loss": 0.4529, "step": 476 }, { "epoch": 2.05, "learning_rate": 4.945220297615806e-06, "loss": 0.4546, "step": 477 }, { "epoch": 2.06, "learning_rate": 4.904268613150278e-06, "loss": 0.4704, "step": 478 }, { "epoch": 2.06, "learning_rate": 4.863432035896924e-06, "loss": 0.4701, "step": 479 }, { "epoch": 2.06, "learning_rate": 4.822711488311077e-06, "loss": 0.459, "step": 480 }, { "epoch": 2.07, "learning_rate": 4.782107890227066e-06, "loss": 0.4901, "step": 481 }, { "epoch": 2.07, "learning_rate": 4.74162215883747e-06, "loss": 0.4505, "step": 482 }, { "epoch": 2.08, "learning_rate": 4.701255208672372e-06, "loss": 0.4671, "step": 483 }, { "epoch": 2.08, "learning_rate": 4.661007951578722e-06, "loss": 0.46, "step": 484 }, { "epoch": 2.09, "learning_rate": 4.620881296699719e-06, "loss": 0.4434, "step": 485 }, { "epoch": 2.09, "learning_rate": 4.580876150454292e-06, "loss": 0.4759, "step": 486 }, { "epoch": 2.09, "learning_rate": 4.540993416516618e-06, "loss": 0.4826, "step": 487 }, { "epoch": 2.1, "learning_rate": 4.501233995795709e-06, "loss": 0.4427, "step": 488 }, { "epoch": 2.1, "learning_rate": 4.461598786415052e-06, "loss": 0.4797, "step": 489 }, { "epoch": 2.11, "learning_rate": 4.4220886836923445e-06, "loss": 0.468, "step": 490 }, { "epoch": 2.11, "learning_rate": 4.382704580119242e-06, "loss": 0.4662, "step": 491 }, { "epoch": 2.12, "learning_rate": 4.343447365341226e-06, "loss": 0.4662, "step": 492 }, { "epoch": 2.12, "learning_rate": 4.304317926137478e-06, "loss": 0.4646, "step": 493 }, { "epoch": 2.12, "learning_rate": 4.265317146400877e-06, "loss": 0.4503, "step": 494 }, { "epoch": 2.13, "learning_rate": 4.2264459071180184e-06, "loss": 0.4462, "step": 495 }, { "epoch": 2.13, "learning_rate": 4.187705086349304e-06, "loss": 0.4583, "step": 496 }, { "epoch": 2.14, "learning_rate": 4.149095559209133e-06, "loss": 0.4626, "step": 497 }, { "epoch": 2.14, "learning_rate": 4.110618197846105e-06, "loss": 0.4765, "step": 498 }, { "epoch": 2.15, "learning_rate": 4.072273871423348e-06, "loss": 0.4756, "step": 499 }, { "epoch": 2.15, "learning_rate": 4.034063446098864e-06, "loss": 0.4585, "step": 500 }, { "epoch": 2.15, "learning_rate": 3.995987785005966e-06, "loss": 0.4522, "step": 501 }, { "epoch": 2.16, "learning_rate": 3.95804774823379e-06, "loss": 0.483, "step": 502 }, { "epoch": 2.16, "learning_rate": 3.920244192807865e-06, "loss": 0.4611, "step": 503 }, { "epoch": 2.17, "learning_rate": 3.882577972670737e-06, "loss": 0.4765, "step": 504 }, { "epoch": 2.17, "learning_rate": 3.845049938662709e-06, "loss": 0.4593, "step": 505 }, { "epoch": 2.18, "learning_rate": 3.8076609385025886e-06, "loss": 0.4595, "step": 506 }, { "epoch": 2.18, "learning_rate": 3.7704118167685676e-06, "loss": 0.4544, "step": 507 }, { "epoch": 2.18, "learning_rate": 3.7333034148791213e-06, "loss": 0.4599, "step": 508 }, { "epoch": 2.19, "learning_rate": 3.6963365710740195e-06, "loss": 0.4395, "step": 509 }, { "epoch": 2.19, "learning_rate": 3.659512120395384e-06, "loss": 0.4537, "step": 510 }, { "epoch": 2.2, "learning_rate": 3.6228308946688163e-06, "loss": 0.4688, "step": 511 }, { "epoch": 2.2, "learning_rate": 3.5862937224846284e-06, "loss": 0.4473, "step": 512 }, { "epoch": 2.21, "learning_rate": 3.5499014291791035e-06, "loss": 0.471, "step": 513 }, { "epoch": 2.21, "learning_rate": 3.513654836815872e-06, "loss": 0.4565, "step": 514 }, { "epoch": 2.22, "learning_rate": 3.4775547641673337e-06, "loss": 0.4836, "step": 515 }, { "epoch": 2.22, "learning_rate": 3.4416020266961503e-06, "loss": 0.4517, "step": 516 }, { "epoch": 2.22, "learning_rate": 3.4057974365368496e-06, "loss": 0.4636, "step": 517 }, { "epoch": 2.23, "learning_rate": 3.3701418024774656e-06, "loss": 0.4647, "step": 518 }, { "epoch": 2.23, "learning_rate": 3.3346359299412624e-06, "loss": 0.47, "step": 519 }, { "epoch": 2.24, "learning_rate": 3.29928062096856e-06, "loss": 0.46, "step": 520 }, { "epoch": 2.24, "learning_rate": 3.264076674198594e-06, "loss": 0.4835, "step": 521 }, { "epoch": 2.25, "learning_rate": 3.2290248848514993e-06, "loss": 0.445, "step": 522 }, { "epoch": 2.25, "learning_rate": 3.1941260447103228e-06, "loss": 0.4665, "step": 523 }, { "epoch": 2.25, "learning_rate": 3.1593809421031583e-06, "loss": 0.4562, "step": 524 }, { "epoch": 2.26, "learning_rate": 3.1247903618853326e-06, "loss": 0.4582, "step": 525 }, { "epoch": 2.26, "learning_rate": 3.0903550854216603e-06, "loss": 0.4762, "step": 526 }, { "epoch": 2.27, "learning_rate": 3.056075890568825e-06, "loss": 0.4547, "step": 527 }, { "epoch": 2.27, "learning_rate": 3.021953551657774e-06, "loss": 0.455, "step": 528 }, { "epoch": 2.28, "learning_rate": 2.987988839476258e-06, "loss": 0.4503, "step": 529 }, { "epoch": 2.28, "learning_rate": 2.9541825212514007e-06, "loss": 0.4707, "step": 530 }, { "epoch": 2.28, "learning_rate": 2.9205353606323683e-06, "loss": 0.4766, "step": 531 }, { "epoch": 2.29, "learning_rate": 2.887048117673129e-06, "loss": 0.4504, "step": 532 }, { "epoch": 2.29, "learning_rate": 2.8537215488152837e-06, "loss": 0.4499, "step": 533 }, { "epoch": 2.3, "learning_rate": 2.82055640687096e-06, "loss": 0.4515, "step": 534 }, { "epoch": 2.3, "learning_rate": 2.7875534410058392e-06, "loss": 0.4629, "step": 535 }, { "epoch": 2.31, "learning_rate": 2.7547133967221985e-06, "loss": 0.4722, "step": 536 }, { "epoch": 2.31, "learning_rate": 2.722037015842103e-06, "loss": 0.4527, "step": 537 }, { "epoch": 2.31, "learning_rate": 2.68952503649062e-06, "loss": 0.4759, "step": 538 }, { "epoch": 2.32, "learning_rate": 2.6571781930791706e-06, "loss": 0.4682, "step": 539 }, { "epoch": 2.32, "learning_rate": 2.624997216288925e-06, "loss": 0.454, "step": 540 }, { "epoch": 2.33, "learning_rate": 2.592982833054294e-06, "loss": 0.4675, "step": 541 }, { "epoch": 2.33, "learning_rate": 2.5611357665465254e-06, "loss": 0.4704, "step": 542 }, { "epoch": 2.34, "learning_rate": 2.529456736157356e-06, "loss": 0.4659, "step": 543 }, { "epoch": 2.34, "learning_rate": 2.497946457482756e-06, "loss": 0.48, "step": 544 }, { "epoch": 2.34, "learning_rate": 2.4666056423067832e-06, "loss": 0.4611, "step": 545 }, { "epoch": 2.35, "learning_rate": 2.43543499858548e-06, "loss": 0.4457, "step": 546 }, { "epoch": 2.35, "learning_rate": 2.404435230430905e-06, "loss": 0.4351, "step": 547 }, { "epoch": 2.36, "learning_rate": 2.373607038095217e-06, "loss": 0.4495, "step": 548 }, { "epoch": 2.36, "learning_rate": 2.3429511179548504e-06, "loss": 0.4407, "step": 549 }, { "epoch": 2.37, "learning_rate": 2.312468162494801e-06, "loss": 0.4444, "step": 550 }, { "epoch": 2.37, "learning_rate": 2.2821588602929635e-06, "loss": 0.4464, "step": 551 }, { "epoch": 2.37, "learning_rate": 2.2520238960046016e-06, "loss": 0.4663, "step": 552 }, { "epoch": 2.38, "learning_rate": 2.2220639503468542e-06, "loss": 0.4665, "step": 553 }, { "epoch": 2.38, "learning_rate": 2.192279700083385e-06, "loss": 0.4499, "step": 554 }, { "epoch": 2.39, "learning_rate": 2.1626718180090824e-06, "loss": 0.4459, "step": 555 }, { "epoch": 2.39, "learning_rate": 2.1332409729348526e-06, "loss": 0.4595, "step": 556 }, { "epoch": 2.4, "learning_rate": 2.1039878296725335e-06, "loss": 0.4736, "step": 557 }, { "epoch": 2.4, "learning_rate": 2.0749130490198655e-06, "loss": 0.4537, "step": 558 }, { "epoch": 2.4, "learning_rate": 2.0460172877455543e-06, "loss": 0.4678, "step": 559 }, { "epoch": 2.41, "learning_rate": 2.0173011985744608e-06, "loss": 0.45, "step": 560 }, { "epoch": 2.41, "learning_rate": 1.9887654301728287e-06, "loss": 0.4612, "step": 561 }, { "epoch": 2.42, "learning_rate": 1.960410627133654e-06, "loss": 0.4356, "step": 562 }, { "epoch": 2.42, "learning_rate": 1.932237429962116e-06, "loss": 0.4603, "step": 563 }, { "epoch": 2.43, "learning_rate": 1.9042464750610989e-06, "loss": 0.466, "step": 564 }, { "epoch": 2.43, "learning_rate": 1.8764383947168386e-06, "loss": 0.4742, "step": 565 }, { "epoch": 2.43, "learning_rate": 1.8488138170846148e-06, "loss": 0.4536, "step": 566 }, { "epoch": 2.44, "learning_rate": 1.8213733661745858e-06, "loss": 0.4578, "step": 567 }, { "epoch": 2.44, "learning_rate": 1.7941176618376687e-06, "loss": 0.4605, "step": 568 }, { "epoch": 2.45, "learning_rate": 1.7670473197515592e-06, "loss": 0.442, "step": 569 }, { "epoch": 2.45, "learning_rate": 1.740162951406812e-06, "loss": 0.4615, "step": 570 }, { "epoch": 2.46, "learning_rate": 1.7134651640930311e-06, "loss": 0.4639, "step": 571 }, { "epoch": 2.46, "learning_rate": 1.6869545608851467e-06, "loss": 0.4669, "step": 572 }, { "epoch": 2.46, "learning_rate": 1.6606317406298045e-06, "loss": 0.4464, "step": 573 }, { "epoch": 2.47, "learning_rate": 1.634497297931823e-06, "loss": 0.4569, "step": 574 }, { "epoch": 2.47, "learning_rate": 1.6085518231407781e-06, "loss": 0.4656, "step": 575 }, { "epoch": 2.48, "learning_rate": 1.5827959023376505e-06, "loss": 0.4674, "step": 576 }, { "epoch": 2.48, "learning_rate": 1.557230117321602e-06, "loss": 0.4457, "step": 577 }, { "epoch": 2.49, "learning_rate": 1.5318550455968285e-06, "loss": 0.4582, "step": 578 }, { "epoch": 2.49, "learning_rate": 1.5066712603595045e-06, "loss": 0.4576, "step": 579 }, { "epoch": 2.49, "learning_rate": 1.4816793304848543e-06, "loss": 0.4612, "step": 580 }, { "epoch": 2.5, "learning_rate": 1.456879820514282e-06, "loss": 0.4644, "step": 581 }, { "epoch": 2.5, "learning_rate": 1.4322732906426362e-06, "loss": 0.4727, "step": 582 }, { "epoch": 2.51, "learning_rate": 1.407860296705542e-06, "loss": 0.465, "step": 583 }, { "epoch": 2.51, "learning_rate": 1.3836413901668566e-06, "loss": 0.4584, "step": 584 }, { "epoch": 2.52, "learning_rate": 1.3596171181062024e-06, "loss": 0.4485, "step": 585 }, { "epoch": 2.52, "learning_rate": 1.335788023206619e-06, "loss": 0.4651, "step": 586 }, { "epoch": 2.52, "learning_rate": 1.3121546437422916e-06, "loss": 0.4701, "step": 587 }, { "epoch": 2.53, "learning_rate": 1.2887175135664088e-06, "loss": 0.4623, "step": 588 }, { "epoch": 2.53, "learning_rate": 1.2654771620990846e-06, "loss": 0.4527, "step": 589 }, { "epoch": 2.54, "learning_rate": 1.2424341143154173e-06, "loss": 0.4668, "step": 590 }, { "epoch": 2.54, "learning_rate": 1.2195888907336162e-06, "loss": 0.44, "step": 591 }, { "epoch": 2.55, "learning_rate": 1.1969420074032535e-06, "loss": 0.4495, "step": 592 }, { "epoch": 2.55, "learning_rate": 1.1744939758936047e-06, "loss": 0.4632, "step": 593 }, { "epoch": 2.55, "learning_rate": 1.1522453032820868e-06, "loss": 0.4676, "step": 594 }, { "epoch": 2.56, "learning_rate": 1.1301964921428165e-06, "loss": 0.4519, "step": 595 }, { "epoch": 2.56, "learning_rate": 1.108348040535242e-06, "loss": 0.4576, "step": 596 }, { "epoch": 2.57, "learning_rate": 1.0867004419929061e-06, "loss": 0.4588, "step": 597 }, { "epoch": 2.57, "learning_rate": 1.0652541855122888e-06, "loss": 0.4522, "step": 598 }, { "epoch": 2.58, "learning_rate": 1.044009755541766e-06, "loss": 0.457, "step": 599 }, { "epoch": 2.58, "learning_rate": 1.0229676319706671e-06, "loss": 0.4461, "step": 600 }, { "epoch": 2.58, "learning_rate": 1.0021282901184315e-06, "loss": 0.4458, "step": 601 }, { "epoch": 2.59, "learning_rate": 9.814922007238692e-07, "loss": 0.4668, "step": 602 }, { "epoch": 2.59, "learning_rate": 9.610598299345364e-07, "loss": 0.4501, "step": 603 }, { "epoch": 2.6, "learning_rate": 9.408316392961947e-07, "loss": 0.4722, "step": 604 }, { "epoch": 2.6, "learning_rate": 9.208080857423985e-07, "loss": 0.4508, "step": 605 }, { "epoch": 2.61, "learning_rate": 9.009896215841562e-07, "loss": 0.4447, "step": 606 }, { "epoch": 2.61, "learning_rate": 8.813766944997293e-07, "loss": 0.4468, "step": 607 }, { "epoch": 2.62, "learning_rate": 8.619697475245136e-07, "loss": 0.4481, "step": 608 }, { "epoch": 2.62, "learning_rate": 8.427692190410252e-07, "loss": 0.4752, "step": 609 }, { "epoch": 2.62, "learning_rate": 8.237755427690097e-07, "loss": 0.4443, "step": 610 }, { "epoch": 2.63, "learning_rate": 8.049891477556326e-07, "loss": 0.4465, "step": 611 }, { "epoch": 2.63, "learning_rate": 7.864104583657995e-07, "loss": 0.4449, "step": 612 }, { "epoch": 2.64, "learning_rate": 7.680398942725609e-07, "loss": 0.4734, "step": 613 }, { "epoch": 2.64, "learning_rate": 7.498778704476373e-07, "loss": 0.4624, "step": 614 }, { "epoch": 2.65, "learning_rate": 7.319247971520427e-07, "loss": 0.446, "step": 615 }, { "epoch": 2.65, "learning_rate": 7.141810799268222e-07, "loss": 0.4671, "step": 616 }, { "epoch": 2.65, "learning_rate": 6.966471195838808e-07, "loss": 0.4666, "step": 617 }, { "epoch": 2.66, "learning_rate": 6.793233121969422e-07, "loss": 0.4609, "step": 618 }, { "epoch": 2.66, "learning_rate": 6.62210049092592e-07, "loss": 0.446, "step": 619 }, { "epoch": 2.67, "learning_rate": 6.453077168414456e-07, "loss": 0.4537, "step": 620 }, { "epoch": 2.67, "learning_rate": 6.286166972494079e-07, "loss": 0.4599, "step": 621 }, { "epoch": 2.68, "learning_rate": 6.121373673490549e-07, "loss": 0.4911, "step": 622 }, { "epoch": 2.68, "learning_rate": 5.958700993911193e-07, "loss": 0.4722, "step": 623 }, { "epoch": 2.68, "learning_rate": 5.798152608360696e-07, "loss": 0.4431, "step": 624 }, { "epoch": 2.69, "learning_rate": 5.639732143458254e-07, "loss": 0.4388, "step": 625 }, { "epoch": 2.69, "learning_rate": 5.483443177755499e-07, "loss": 0.4504, "step": 626 }, { "epoch": 2.7, "learning_rate": 5.329289241655811e-07, "loss": 0.4663, "step": 627 }, { "epoch": 2.7, "learning_rate": 5.177273817334439e-07, "loss": 0.4538, "step": 628 }, { "epoch": 2.71, "learning_rate": 5.027400338659926e-07, "loss": 0.4784, "step": 629 }, { "epoch": 2.71, "learning_rate": 4.879672191116524e-07, "loss": 0.462, "step": 630 }, { "epoch": 2.71, "learning_rate": 4.734092711727711e-07, "loss": 0.4539, "step": 631 }, { "epoch": 2.72, "learning_rate": 4.5906651889807697e-07, "loss": 0.4435, "step": 632 }, { "epoch": 2.72, "learning_rate": 4.4493928627526105e-07, "loss": 0.47, "step": 633 }, { "epoch": 2.73, "learning_rate": 4.3102789242364553e-07, "loss": 0.4485, "step": 634 }, { "epoch": 2.73, "learning_rate": 4.17332651586988e-07, "loss": 0.4702, "step": 635 }, { "epoch": 2.74, "learning_rate": 4.0385387312637194e-07, "loss": 0.4593, "step": 636 }, { "epoch": 2.74, "learning_rate": 3.9059186151322534e-07, "loss": 0.4618, "step": 637 }, { "epoch": 2.74, "learning_rate": 3.7754691632244323e-07, "loss": 0.4635, "step": 638 }, { "epoch": 2.75, "learning_rate": 3.6471933222561375e-07, "loss": 0.474, "step": 639 }, { "epoch": 2.75, "learning_rate": 3.521093989843716e-07, "loss": 0.4544, "step": 640 }, { "epoch": 2.76, "learning_rate": 3.397174014438431e-07, "loss": 0.4794, "step": 641 }, { "epoch": 2.76, "learning_rate": 3.2754361952621936e-07, "loss": 0.4599, "step": 642 }, { "epoch": 2.77, "learning_rate": 3.1558832822442874e-07, "loss": 0.45, "step": 643 }, { "epoch": 2.77, "learning_rate": 3.0385179759592767e-07, "loss": 0.4594, "step": 644 }, { "epoch": 2.77, "learning_rate": 2.9233429275659643e-07, "loss": 0.4653, "step": 645 }, { "epoch": 2.78, "learning_rate": 2.8103607387475753e-07, "loss": 0.4578, "step": 646 }, { "epoch": 2.78, "learning_rate": 2.6995739616528905e-07, "loss": 0.4616, "step": 647 }, { "epoch": 2.79, "learning_rate": 2.5909850988386943e-07, "loss": 0.4624, "step": 648 }, { "epoch": 2.79, "learning_rate": 2.484596603213163e-07, "loss": 0.4433, "step": 649 }, { "epoch": 2.8, "learning_rate": 2.3804108779805325e-07, "loss": 0.4527, "step": 650 }, { "epoch": 2.8, "learning_rate": 2.2784302765867294e-07, "loss": 0.4542, "step": 651 }, { "epoch": 2.8, "learning_rate": 2.1786571026662705e-07, "loss": 0.4598, "step": 652 }, { "epoch": 2.81, "learning_rate": 2.0810936099902368e-07, "loss": 0.4448, "step": 653 }, { "epoch": 2.81, "learning_rate": 1.9857420024152808e-07, "loss": 0.4591, "step": 654 }, { "epoch": 2.82, "learning_rate": 1.8926044338339332e-07, "loss": 0.4621, "step": 655 }, { "epoch": 2.82, "learning_rate": 1.8016830081259095e-07, "loss": 0.458, "step": 656 }, { "epoch": 2.83, "learning_rate": 1.7129797791105686e-07, "loss": 0.4634, "step": 657 }, { "epoch": 2.83, "learning_rate": 1.6264967505005613e-07, "loss": 0.464, "step": 658 }, { "epoch": 2.83, "learning_rate": 1.5422358758565348e-07, "loss": 0.4584, "step": 659 }, { "epoch": 2.84, "learning_rate": 1.4601990585430214e-07, "loss": 0.4653, "step": 660 }, { "epoch": 2.84, "learning_rate": 1.3803881516854412e-07, "loss": 0.4502, "step": 661 }, { "epoch": 2.85, "learning_rate": 1.3028049581282231e-07, "loss": 0.4696, "step": 662 }, { "epoch": 2.85, "learning_rate": 1.2274512303941165e-07, "loss": 0.4537, "step": 663 }, { "epoch": 2.86, "learning_rate": 1.1543286706445556e-07, "loss": 0.4587, "step": 664 }, { "epoch": 2.86, "learning_rate": 1.0834389306412674e-07, "loss": 0.4499, "step": 665 }, { "epoch": 2.86, "learning_rate": 1.0147836117088916e-07, "loss": 0.4561, "step": 666 }, { "epoch": 2.87, "learning_rate": 9.483642646988978e-08, "loss": 0.4674, "step": 667 }, { "epoch": 2.87, "learning_rate": 8.841823899544577e-08, "loss": 0.4561, "step": 668 }, { "epoch": 2.88, "learning_rate": 8.222394372766173e-08, "loss": 0.4509, "step": 669 }, { "epoch": 2.88, "learning_rate": 7.625368058915228e-08, "loss": 0.4643, "step": 670 }, { "epoch": 2.89, "learning_rate": 7.050758444188122e-08, "loss": 0.4575, "step": 671 }, { "epoch": 2.89, "learning_rate": 6.498578508411734e-08, "loss": 0.4636, "step": 672 }, { "epoch": 2.89, "learning_rate": 5.968840724750013e-08, "loss": 0.4636, "step": 673 }, { "epoch": 2.9, "learning_rate": 5.4615570594223066e-08, "loss": 0.4661, "step": 674 }, { "epoch": 2.9, "learning_rate": 4.976738971433026e-08, "loss": 0.4538, "step": 675 }, { "epoch": 2.91, "learning_rate": 4.5143974123129655e-08, "loss": 0.4651, "step": 676 }, { "epoch": 2.91, "learning_rate": 4.074542825871275e-08, "loss": 0.4544, "step": 677 }, { "epoch": 2.92, "learning_rate": 3.6571851479607624e-08, "loss": 0.4608, "step": 678 }, { "epoch": 2.92, "learning_rate": 3.262333806252294e-08, "loss": 0.4706, "step": 679 }, { "epoch": 2.92, "learning_rate": 2.889997720022297e-08, "loss": 0.4496, "step": 680 }, { "epoch": 2.93, "learning_rate": 2.540185299951259e-08, "loss": 0.4655, "step": 681 }, { "epoch": 2.93, "learning_rate": 2.2129044479339833e-08, "loss": 0.4512, "step": 682 }, { "epoch": 2.94, "learning_rate": 1.908162556900628e-08, "loss": 0.453, "step": 683 }, { "epoch": 2.94, "learning_rate": 1.6259665106498346e-08, "loss": 0.4491, "step": 684 }, { "epoch": 2.95, "learning_rate": 1.3663226836936327e-08, "loss": 0.4554, "step": 685 }, { "epoch": 2.95, "learning_rate": 1.1292369411127768e-08, "loss": 0.4531, "step": 686 }, { "epoch": 2.95, "learning_rate": 9.147146384250739e-09, "loss": 0.4548, "step": 687 }, { "epoch": 2.96, "learning_rate": 7.227606214635918e-09, "loss": 0.4687, "step": 688 }, { "epoch": 2.96, "learning_rate": 5.5337922626752526e-09, "loss": 0.4641, "step": 689 }, { "epoch": 2.97, "learning_rate": 4.065742789846061e-09, "loss": 0.4636, "step": 690 }, { "epoch": 2.97, "learning_rate": 2.8234909578417347e-09, "loss": 0.4611, "step": 691 }, { "epoch": 2.98, "learning_rate": 1.807064827823446e-09, "loss": 0.4629, "step": 692 }, { "epoch": 2.98, "learning_rate": 1.016487359789542e-09, "loss": 0.4562, "step": 693 }, { "epoch": 2.98, "learning_rate": 4.5177641205262915e-10, "loss": 0.4709, "step": 694 }, { "epoch": 2.99, "learning_rate": 1.1294474083878293e-10, "loss": 0.4708, "step": 695 }, { "epoch": 2.99, "learning_rate": 0.0, "loss": 0.443, "step": 696 }, { "epoch": 2.99, "step": 696, "total_flos": 1196603910258688.0, "train_loss": 0.6090522945012854, "train_runtime": 52554.8084, "train_samples_per_second": 1.698, "train_steps_per_second": 0.013 } ], "logging_steps": 1.0, "max_steps": 696, "num_input_tokens_seen": 0, "num_train_epochs": 3, "save_steps": 400, "total_flos": 1196603910258688.0, "train_batch_size": 4, "trial_name": null, "trial_params": null }