{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 945, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 6.896551724137931e-07, "loss": 1.0054, "step": 1 }, { "epoch": 0.0, "learning_rate": 1.3793103448275862e-06, "loss": 1.0132, "step": 2 }, { "epoch": 0.0, "learning_rate": 2.0689655172413796e-06, "loss": 1.0171, "step": 3 }, { "epoch": 0.0, "learning_rate": 2.7586206896551725e-06, "loss": 1.0454, "step": 4 }, { "epoch": 0.01, "learning_rate": 3.448275862068966e-06, "loss": 1.0244, "step": 5 }, { "epoch": 0.01, "learning_rate": 4.137931034482759e-06, "loss": 0.7725, "step": 6 }, { "epoch": 0.01, "learning_rate": 4.8275862068965525e-06, "loss": 0.6753, "step": 7 }, { "epoch": 0.01, "learning_rate": 5.517241379310345e-06, "loss": 0.5906, "step": 8 }, { "epoch": 0.01, "learning_rate": 6.206896551724138e-06, "loss": 0.4258, "step": 9 }, { "epoch": 0.01, "learning_rate": 6.896551724137932e-06, "loss": 0.4697, "step": 10 }, { "epoch": 0.01, "learning_rate": 7.586206896551724e-06, "loss": 0.4119, "step": 11 }, { "epoch": 0.01, "learning_rate": 8.275862068965518e-06, "loss": 0.3669, "step": 12 }, { "epoch": 0.01, "learning_rate": 8.965517241379312e-06, "loss": 0.3694, "step": 13 }, { "epoch": 0.01, "learning_rate": 9.655172413793105e-06, "loss": 0.3057, "step": 14 }, { "epoch": 0.02, "learning_rate": 1.0344827586206898e-05, "loss": 0.306, "step": 15 }, { "epoch": 0.02, "learning_rate": 1.103448275862069e-05, "loss": 0.386, "step": 16 }, { "epoch": 0.02, "learning_rate": 1.1724137931034483e-05, "loss": 0.4041, "step": 17 }, { "epoch": 0.02, "learning_rate": 1.2413793103448277e-05, "loss": 0.3486, "step": 18 }, { "epoch": 0.02, "learning_rate": 1.310344827586207e-05, "loss": 0.3163, "step": 19 }, { "epoch": 0.02, "learning_rate": 1.3793103448275863e-05, "loss": 0.3091, "step": 20 }, { "epoch": 0.02, "learning_rate": 1.4482758620689657e-05, "loss": 0.292, "step": 21 }, { "epoch": 0.02, "learning_rate": 1.5172413793103448e-05, "loss": 0.3126, "step": 22 }, { "epoch": 0.02, "learning_rate": 1.586206896551724e-05, "loss": 0.3416, "step": 23 }, { "epoch": 0.03, "learning_rate": 1.6551724137931037e-05, "loss": 0.3394, "step": 24 }, { "epoch": 0.03, "learning_rate": 1.7241379310344828e-05, "loss": 0.3389, "step": 25 }, { "epoch": 0.03, "learning_rate": 1.7931034482758623e-05, "loss": 0.2695, "step": 26 }, { "epoch": 0.03, "learning_rate": 1.8620689655172415e-05, "loss": 0.2732, "step": 27 }, { "epoch": 0.03, "learning_rate": 1.931034482758621e-05, "loss": 0.2657, "step": 28 }, { "epoch": 0.03, "learning_rate": 2e-05, "loss": 0.2874, "step": 29 }, { "epoch": 0.03, "learning_rate": 1.9999941186316965e-05, "loss": 0.2688, "step": 30 }, { "epoch": 0.03, "learning_rate": 1.999976474595967e-05, "loss": 0.2732, "step": 31 }, { "epoch": 0.03, "learning_rate": 1.999947068100353e-05, "loss": 0.254, "step": 32 }, { "epoch": 0.03, "learning_rate": 1.9999058994907564e-05, "loss": 0.2668, "step": 33 }, { "epoch": 0.04, "learning_rate": 1.9998529692514318e-05, "loss": 0.2683, "step": 34 }, { "epoch": 0.04, "learning_rate": 1.9997882780049847e-05, "loss": 0.2405, "step": 35 }, { "epoch": 0.04, "learning_rate": 1.9997118265123597e-05, "loss": 0.2194, "step": 36 }, { "epoch": 0.04, "learning_rate": 1.999623615672837e-05, "loss": 0.2686, "step": 37 }, { "epoch": 0.04, "learning_rate": 1.9995236465240168e-05, "loss": 0.2915, "step": 38 }, { "epoch": 0.04, "learning_rate": 1.99941192024181e-05, "loss": 0.2526, "step": 39 }, { "epoch": 0.04, "learning_rate": 1.999288438140423e-05, "loss": 0.2572, "step": 40 }, { "epoch": 0.04, "learning_rate": 1.999153201672344e-05, "loss": 0.2291, "step": 41 }, { "epoch": 0.04, "learning_rate": 1.9990062124283233e-05, "loss": 0.2449, "step": 42 }, { "epoch": 0.05, "learning_rate": 1.998847472137357e-05, "loss": 0.274, "step": 43 }, { "epoch": 0.05, "learning_rate": 1.998676982666665e-05, "loss": 0.2616, "step": 44 }, { "epoch": 0.05, "learning_rate": 1.9984947460216708e-05, "loss": 0.2395, "step": 45 }, { "epoch": 0.05, "learning_rate": 1.9983007643459757e-05, "loss": 0.2493, "step": 46 }, { "epoch": 0.05, "learning_rate": 1.9980950399213344e-05, "loss": 0.2405, "step": 47 }, { "epoch": 0.05, "learning_rate": 1.99787757516763e-05, "loss": 0.2522, "step": 48 }, { "epoch": 0.05, "learning_rate": 1.9976483726428423e-05, "loss": 0.2219, "step": 49 }, { "epoch": 0.05, "learning_rate": 1.997407435043021e-05, "loss": 0.2367, "step": 50 }, { "epoch": 0.05, "learning_rate": 1.9971547652022512e-05, "loss": 0.2212, "step": 51 }, { "epoch": 0.06, "learning_rate": 1.996890366092622e-05, "loss": 0.2114, "step": 52 }, { "epoch": 0.06, "learning_rate": 1.99661424082419e-05, "loss": 0.2485, "step": 53 }, { "epoch": 0.06, "learning_rate": 1.9963263926449446e-05, "loss": 0.2455, "step": 54 }, { "epoch": 0.06, "learning_rate": 1.9960268249407674e-05, "loss": 0.2487, "step": 55 }, { "epoch": 0.06, "learning_rate": 1.995715541235395e-05, "loss": 0.2037, "step": 56 }, { "epoch": 0.06, "learning_rate": 1.9953925451903757e-05, "loss": 0.2269, "step": 57 }, { "epoch": 0.06, "learning_rate": 1.9950578406050263e-05, "loss": 0.2332, "step": 58 }, { "epoch": 0.06, "learning_rate": 1.9947114314163892e-05, "loss": 0.2271, "step": 59 }, { "epoch": 0.06, "learning_rate": 1.9943533216991843e-05, "loss": 0.2351, "step": 60 }, { "epoch": 0.06, "learning_rate": 1.9939835156657616e-05, "loss": 0.2352, "step": 61 }, { "epoch": 0.07, "learning_rate": 1.9936020176660524e-05, "loss": 0.2251, "step": 62 }, { "epoch": 0.07, "learning_rate": 1.9932088321875175e-05, "loss": 0.2469, "step": 63 }, { "epoch": 0.07, "learning_rate": 1.9928039638550932e-05, "loss": 0.2332, "step": 64 }, { "epoch": 0.07, "learning_rate": 1.9923874174311394e-05, "loss": 0.2192, "step": 65 }, { "epoch": 0.07, "learning_rate": 1.9919591978153824e-05, "loss": 0.2198, "step": 66 }, { "epoch": 0.07, "learning_rate": 1.991519310044857e-05, "loss": 0.2253, "step": 67 }, { "epoch": 0.07, "learning_rate": 1.9910677592938458e-05, "loss": 0.2036, "step": 68 }, { "epoch": 0.07, "learning_rate": 1.990604550873823e-05, "loss": 0.2185, "step": 69 }, { "epoch": 0.07, "learning_rate": 1.990129690233386e-05, "loss": 0.2292, "step": 70 }, { "epoch": 0.08, "learning_rate": 1.989643182958196e-05, "loss": 0.2478, "step": 71 }, { "epoch": 0.08, "learning_rate": 1.98914503477091e-05, "loss": 0.2175, "step": 72 }, { "epoch": 0.08, "learning_rate": 1.9886352515311134e-05, "loss": 0.2115, "step": 73 }, { "epoch": 0.08, "learning_rate": 1.9881138392352528e-05, "loss": 0.1942, "step": 74 }, { "epoch": 0.08, "learning_rate": 1.987580804016563e-05, "loss": 0.2463, "step": 75 }, { "epoch": 0.08, "learning_rate": 1.9870361521449978e-05, "loss": 0.1982, "step": 76 }, { "epoch": 0.08, "learning_rate": 1.986479890027153e-05, "loss": 0.1976, "step": 77 }, { "epoch": 0.08, "learning_rate": 1.985912024206194e-05, "loss": 0.2089, "step": 78 }, { "epoch": 0.08, "learning_rate": 1.985332561361776e-05, "loss": 0.2252, "step": 79 }, { "epoch": 0.08, "learning_rate": 1.9847415083099685e-05, "loss": 0.2255, "step": 80 }, { "epoch": 0.09, "learning_rate": 1.9841388720031727e-05, "loss": 0.2118, "step": 81 }, { "epoch": 0.09, "learning_rate": 1.9835246595300405e-05, "loss": 0.2319, "step": 82 }, { "epoch": 0.09, "learning_rate": 1.9828988781153916e-05, "loss": 0.2213, "step": 83 }, { "epoch": 0.09, "learning_rate": 1.9822615351201283e-05, "loss": 0.2305, "step": 84 }, { "epoch": 0.09, "learning_rate": 1.9816126380411478e-05, "loss": 0.2321, "step": 85 }, { "epoch": 0.09, "learning_rate": 1.980952194511256e-05, "loss": 0.2576, "step": 86 }, { "epoch": 0.09, "learning_rate": 1.9802802122990758e-05, "loss": 0.1926, "step": 87 }, { "epoch": 0.09, "learning_rate": 1.9795966993089574e-05, "loss": 0.23, "step": 88 }, { "epoch": 0.09, "learning_rate": 1.9789016635808836e-05, "loss": 0.2072, "step": 89 }, { "epoch": 0.1, "learning_rate": 1.978195113290377e-05, "loss": 0.2241, "step": 90 }, { "epoch": 0.1, "learning_rate": 1.977477056748402e-05, "loss": 0.2291, "step": 91 }, { "epoch": 0.1, "learning_rate": 1.9767475024012698e-05, "loss": 0.215, "step": 92 }, { "epoch": 0.1, "learning_rate": 1.9760064588305347e-05, "loss": 0.239, "step": 93 }, { "epoch": 0.1, "learning_rate": 1.9752539347528973e-05, "loss": 0.2006, "step": 94 }, { "epoch": 0.1, "learning_rate": 1.9744899390201006e-05, "loss": 0.1998, "step": 95 }, { "epoch": 0.1, "learning_rate": 1.9737144806188248e-05, "loss": 0.2152, "step": 96 }, { "epoch": 0.1, "learning_rate": 1.9729275686705832e-05, "loss": 0.202, "step": 97 }, { "epoch": 0.1, "learning_rate": 1.972129212431613e-05, "loss": 0.226, "step": 98 }, { "epoch": 0.1, "learning_rate": 1.9713194212927694e-05, "loss": 0.2236, "step": 99 }, { "epoch": 0.11, "learning_rate": 1.9704982047794118e-05, "loss": 0.1976, "step": 100 }, { "epoch": 0.11, "learning_rate": 1.9696655725512933e-05, "loss": 0.2054, "step": 101 }, { "epoch": 0.11, "learning_rate": 1.968821534402448e-05, "loss": 0.1965, "step": 102 }, { "epoch": 0.11, "learning_rate": 1.9679661002610743e-05, "loss": 0.1757, "step": 103 }, { "epoch": 0.11, "learning_rate": 1.9670992801894185e-05, "loss": 0.2014, "step": 104 }, { "epoch": 0.11, "learning_rate": 1.9662210843836574e-05, "loss": 0.1932, "step": 105 }, { "epoch": 0.11, "learning_rate": 1.965331523173776e-05, "loss": 0.189, "step": 106 }, { "epoch": 0.11, "learning_rate": 1.9644306070234492e-05, "loss": 0.185, "step": 107 }, { "epoch": 0.11, "learning_rate": 1.9635183465299157e-05, "loss": 0.1849, "step": 108 }, { "epoch": 0.12, "learning_rate": 1.9625947524238564e-05, "loss": 0.1971, "step": 109 }, { "epoch": 0.12, "learning_rate": 1.9616598355692644e-05, "loss": 0.1665, "step": 110 }, { "epoch": 0.12, "learning_rate": 1.9607136069633212e-05, "loss": 0.1899, "step": 111 }, { "epoch": 0.12, "learning_rate": 1.9597560777362645e-05, "loss": 0.1633, "step": 112 }, { "epoch": 0.12, "learning_rate": 1.9587872591512583e-05, "loss": 0.1892, "step": 113 }, { "epoch": 0.12, "learning_rate": 1.9578071626042602e-05, "loss": 0.1857, "step": 114 }, { "epoch": 0.12, "learning_rate": 1.9568157996238884e-05, "loss": 0.2025, "step": 115 }, { "epoch": 0.12, "learning_rate": 1.9558131818712838e-05, "loss": 0.1909, "step": 116 }, { "epoch": 0.12, "learning_rate": 1.9547993211399753e-05, "loss": 0.2058, "step": 117 }, { "epoch": 0.12, "learning_rate": 1.9537742293557397e-05, "loss": 0.2028, "step": 118 }, { "epoch": 0.13, "learning_rate": 1.9527379185764613e-05, "loss": 0.1567, "step": 119 }, { "epoch": 0.13, "learning_rate": 1.951690400991991e-05, "loss": 0.1813, "step": 120 }, { "epoch": 0.13, "learning_rate": 1.9506316889240027e-05, "loss": 0.1808, "step": 121 }, { "epoch": 0.13, "learning_rate": 1.9495617948258473e-05, "loss": 0.1801, "step": 122 }, { "epoch": 0.13, "learning_rate": 1.9484807312824066e-05, "loss": 0.2242, "step": 123 }, { "epoch": 0.13, "learning_rate": 1.947388511009947e-05, "loss": 0.1932, "step": 124 }, { "epoch": 0.13, "learning_rate": 1.946285146855968e-05, "loss": 0.1923, "step": 125 }, { "epoch": 0.13, "learning_rate": 1.9451706517990516e-05, "loss": 0.1837, "step": 126 }, { "epoch": 0.13, "learning_rate": 1.944045038948709e-05, "loss": 0.1685, "step": 127 }, { "epoch": 0.14, "learning_rate": 1.9429083215452276e-05, "loss": 0.2152, "step": 128 }, { "epoch": 0.14, "learning_rate": 1.941760512959516e-05, "loss": 0.2018, "step": 129 }, { "epoch": 0.14, "learning_rate": 1.940601626692943e-05, "loss": 0.1931, "step": 130 }, { "epoch": 0.14, "learning_rate": 1.939431676377183e-05, "loss": 0.2258, "step": 131 }, { "epoch": 0.14, "learning_rate": 1.9382506757740532e-05, "loss": 0.194, "step": 132 }, { "epoch": 0.14, "learning_rate": 1.9370586387753532e-05, "loss": 0.2153, "step": 133 }, { "epoch": 0.14, "learning_rate": 1.9358555794027e-05, "loss": 0.212, "step": 134 }, { "epoch": 0.14, "learning_rate": 1.9346415118073634e-05, "loss": 0.2058, "step": 135 }, { "epoch": 0.14, "learning_rate": 1.9334164502701017e-05, "loss": 0.173, "step": 136 }, { "epoch": 0.14, "learning_rate": 1.932180409200991e-05, "loss": 0.1968, "step": 137 }, { "epoch": 0.15, "learning_rate": 1.930933403139256e-05, "loss": 0.23, "step": 138 }, { "epoch": 0.15, "learning_rate": 1.9296754467531015e-05, "loss": 0.1965, "step": 139 }, { "epoch": 0.15, "learning_rate": 1.9284065548395362e-05, "loss": 0.1716, "step": 140 }, { "epoch": 0.15, "learning_rate": 1.9271267423242028e-05, "loss": 0.2417, "step": 141 }, { "epoch": 0.15, "learning_rate": 1.9258360242611973e-05, "loss": 0.2018, "step": 142 }, { "epoch": 0.15, "learning_rate": 1.9245344158328975e-05, "loss": 0.1996, "step": 143 }, { "epoch": 0.15, "learning_rate": 1.9232219323497797e-05, "loss": 0.1869, "step": 144 }, { "epoch": 0.15, "learning_rate": 1.921898589250242e-05, "loss": 0.1926, "step": 145 }, { "epoch": 0.15, "learning_rate": 1.9205644021004203e-05, "loss": 0.2506, "step": 146 }, { "epoch": 0.16, "learning_rate": 1.919219386594007e-05, "loss": 0.2209, "step": 147 }, { "epoch": 0.16, "learning_rate": 1.9178635585520648e-05, "loss": 0.2094, "step": 148 }, { "epoch": 0.16, "learning_rate": 1.9164969339228424e-05, "loss": 0.2151, "step": 149 }, { "epoch": 0.16, "learning_rate": 1.915119528781585e-05, "loss": 0.2075, "step": 150 }, { "epoch": 0.16, "learning_rate": 1.9137313593303465e-05, "loss": 0.202, "step": 151 }, { "epoch": 0.16, "learning_rate": 1.9123324418977982e-05, "loss": 0.1809, "step": 152 }, { "epoch": 0.16, "learning_rate": 1.9109227929390378e-05, "loss": 0.2255, "step": 153 }, { "epoch": 0.16, "learning_rate": 1.9095024290353948e-05, "loss": 0.2092, "step": 154 }, { "epoch": 0.16, "learning_rate": 1.9080713668942356e-05, "loss": 0.2036, "step": 155 }, { "epoch": 0.17, "learning_rate": 1.906629623348767e-05, "loss": 0.1858, "step": 156 }, { "epoch": 0.17, "learning_rate": 1.905177215357839e-05, "loss": 0.1843, "step": 157 }, { "epoch": 0.17, "learning_rate": 1.903714160005744e-05, "loss": 0.2146, "step": 158 }, { "epoch": 0.17, "learning_rate": 1.9022404745020165e-05, "loss": 0.2113, "step": 159 }, { "epoch": 0.17, "learning_rate": 1.900756176181231e-05, "loss": 0.1903, "step": 160 }, { "epoch": 0.17, "learning_rate": 1.8992612825027978e-05, "loss": 0.1765, "step": 161 }, { "epoch": 0.17, "learning_rate": 1.897755811050758e-05, "loss": 0.191, "step": 162 }, { "epoch": 0.17, "learning_rate": 1.896239779533575e-05, "loss": 0.1844, "step": 163 }, { "epoch": 0.17, "learning_rate": 1.8947132057839287e-05, "loss": 0.1686, "step": 164 }, { "epoch": 0.17, "learning_rate": 1.8931761077585037e-05, "loss": 0.1933, "step": 165 }, { "epoch": 0.18, "learning_rate": 1.8916285035377794e-05, "loss": 0.1661, "step": 166 }, { "epoch": 0.18, "learning_rate": 1.8900704113258166e-05, "loss": 0.1726, "step": 167 }, { "epoch": 0.18, "learning_rate": 1.8885018494500436e-05, "loss": 0.176, "step": 168 }, { "epoch": 0.18, "learning_rate": 1.8869228363610406e-05, "loss": 0.189, "step": 169 }, { "epoch": 0.18, "learning_rate": 1.8853333906323226e-05, "loss": 0.1692, "step": 170 }, { "epoch": 0.18, "learning_rate": 1.8837335309601214e-05, "loss": 0.1749, "step": 171 }, { "epoch": 0.18, "learning_rate": 1.882123276163164e-05, "loss": 0.22, "step": 172 }, { "epoch": 0.18, "learning_rate": 1.8805026451824547e-05, "loss": 0.1686, "step": 173 }, { "epoch": 0.18, "learning_rate": 1.8788716570810478e-05, "loss": 0.1675, "step": 174 }, { "epoch": 0.19, "learning_rate": 1.8772303310438275e-05, "loss": 0.1842, "step": 175 }, { "epoch": 0.19, "learning_rate": 1.875578686377279e-05, "loss": 0.175, "step": 176 }, { "epoch": 0.19, "learning_rate": 1.8739167425092644e-05, "loss": 0.1985, "step": 177 }, { "epoch": 0.19, "learning_rate": 1.872244518988791e-05, "loss": 0.1766, "step": 178 }, { "epoch": 0.19, "learning_rate": 1.8705620354857833e-05, "loss": 0.1706, "step": 179 }, { "epoch": 0.19, "learning_rate": 1.868869311790852e-05, "loss": 0.2061, "step": 180 }, { "epoch": 0.19, "learning_rate": 1.8671663678150605e-05, "loss": 0.1982, "step": 181 }, { "epoch": 0.19, "learning_rate": 1.8654532235896897e-05, "loss": 0.1971, "step": 182 }, { "epoch": 0.19, "learning_rate": 1.8637298992660042e-05, "loss": 0.2109, "step": 183 }, { "epoch": 0.19, "learning_rate": 1.861996415115014e-05, "loss": 0.1826, "step": 184 }, { "epoch": 0.2, "learning_rate": 1.860252791527236e-05, "loss": 0.1697, "step": 185 }, { "epoch": 0.2, "learning_rate": 1.858499049012456e-05, "loss": 0.2126, "step": 186 }, { "epoch": 0.2, "learning_rate": 1.8567352081994852e-05, "loss": 0.1738, "step": 187 }, { "epoch": 0.2, "learning_rate": 1.8549612898359185e-05, "loss": 0.2177, "step": 188 }, { "epoch": 0.2, "learning_rate": 1.8531773147878895e-05, "loss": 0.189, "step": 189 }, { "epoch": 0.2, "learning_rate": 1.851383304039828e-05, "loss": 0.1908, "step": 190 }, { "epoch": 0.2, "learning_rate": 1.849579278694209e-05, "loss": 0.1669, "step": 191 }, { "epoch": 0.2, "learning_rate": 1.8477652599713082e-05, "loss": 0.1788, "step": 192 }, { "epoch": 0.2, "learning_rate": 1.8459412692089497e-05, "loss": 0.2174, "step": 193 }, { "epoch": 0.21, "learning_rate": 1.8441073278622562e-05, "loss": 0.188, "step": 194 }, { "epoch": 0.21, "learning_rate": 1.842263457503397e-05, "loss": 0.1992, "step": 195 }, { "epoch": 0.21, "learning_rate": 1.8404096798213334e-05, "loss": 0.1494, "step": 196 }, { "epoch": 0.21, "learning_rate": 1.838546016621564e-05, "loss": 0.1837, "step": 197 }, { "epoch": 0.21, "learning_rate": 1.8366724898258678e-05, "loss": 0.198, "step": 198 }, { "epoch": 0.21, "learning_rate": 1.8347891214720477e-05, "loss": 0.1876, "step": 199 }, { "epoch": 0.21, "learning_rate": 1.8328959337136692e-05, "loss": 0.1809, "step": 200 }, { "epoch": 0.21, "learning_rate": 1.8309929488198012e-05, "loss": 0.1947, "step": 201 }, { "epoch": 0.21, "learning_rate": 1.829080189174754e-05, "loss": 0.1687, "step": 202 }, { "epoch": 0.21, "learning_rate": 1.8271576772778154e-05, "loss": 0.1722, "step": 203 }, { "epoch": 0.22, "learning_rate": 1.825225435742986e-05, "loss": 0.1736, "step": 204 }, { "epoch": 0.22, "learning_rate": 1.8232834872987147e-05, "loss": 0.1716, "step": 205 }, { "epoch": 0.22, "learning_rate": 1.82133185478763e-05, "loss": 0.214, "step": 206 }, { "epoch": 0.22, "learning_rate": 1.8193705611662697e-05, "loss": 0.1625, "step": 207 }, { "epoch": 0.22, "learning_rate": 1.8173996295048147e-05, "loss": 0.2052, "step": 208 }, { "epoch": 0.22, "learning_rate": 1.8154190829868152e-05, "loss": 0.174, "step": 209 }, { "epoch": 0.22, "learning_rate": 1.8134289449089184e-05, "loss": 0.1893, "step": 210 }, { "epoch": 0.22, "learning_rate": 1.8114292386805935e-05, "loss": 0.1875, "step": 211 }, { "epoch": 0.22, "learning_rate": 1.809419987823859e-05, "loss": 0.2062, "step": 212 }, { "epoch": 0.23, "learning_rate": 1.8074012159730034e-05, "loss": 0.2076, "step": 213 }, { "epoch": 0.23, "learning_rate": 1.805372946874308e-05, "loss": 0.2224, "step": 214 }, { "epoch": 0.23, "learning_rate": 1.8033352043857677e-05, "loss": 0.204, "step": 215 }, { "epoch": 0.23, "learning_rate": 1.801288012476811e-05, "loss": 0.1863, "step": 216 }, { "epoch": 0.23, "learning_rate": 1.7992313952280175e-05, "loss": 0.1793, "step": 217 }, { "epoch": 0.23, "learning_rate": 1.7971653768308334e-05, "loss": 0.1749, "step": 218 }, { "epoch": 0.23, "learning_rate": 1.7950899815872894e-05, "loss": 0.2104, "step": 219 }, { "epoch": 0.23, "learning_rate": 1.793005233909713e-05, "loss": 0.2169, "step": 220 }, { "epoch": 0.23, "learning_rate": 1.790911158320442e-05, "loss": 0.1751, "step": 221 }, { "epoch": 0.23, "learning_rate": 1.7888077794515362e-05, "loss": 0.1779, "step": 222 }, { "epoch": 0.24, "learning_rate": 1.786695122044487e-05, "loss": 0.1709, "step": 223 }, { "epoch": 0.24, "learning_rate": 1.7845732109499275e-05, "loss": 0.2078, "step": 224 }, { "epoch": 0.24, "learning_rate": 1.782442071127338e-05, "loss": 0.1628, "step": 225 }, { "epoch": 0.24, "learning_rate": 1.7803017276447558e-05, "loss": 0.1854, "step": 226 }, { "epoch": 0.24, "learning_rate": 1.778152205678477e-05, "loss": 0.1862, "step": 227 }, { "epoch": 0.24, "learning_rate": 1.775993530512763e-05, "loss": 0.1888, "step": 228 }, { "epoch": 0.24, "learning_rate": 1.7738257275395404e-05, "loss": 0.212, "step": 229 }, { "epoch": 0.24, "learning_rate": 1.771648822258105e-05, "loss": 0.1711, "step": 230 }, { "epoch": 0.24, "learning_rate": 1.7694628402748203e-05, "loss": 0.1914, "step": 231 }, { "epoch": 0.25, "learning_rate": 1.7672678073028163e-05, "loss": 0.1647, "step": 232 }, { "epoch": 0.25, "learning_rate": 1.765063749161688e-05, "loss": 0.1597, "step": 233 }, { "epoch": 0.25, "learning_rate": 1.7628506917771906e-05, "loss": 0.1624, "step": 234 }, { "epoch": 0.25, "learning_rate": 1.7606286611809353e-05, "loss": 0.1767, "step": 235 }, { "epoch": 0.25, "learning_rate": 1.758397683510083e-05, "loss": 0.1908, "step": 236 }, { "epoch": 0.25, "learning_rate": 1.7561577850070355e-05, "loss": 0.2068, "step": 237 }, { "epoch": 0.25, "learning_rate": 1.7539089920191298e-05, "loss": 0.174, "step": 238 }, { "epoch": 0.25, "learning_rate": 1.7516513309983253e-05, "loss": 0.186, "step": 239 }, { "epoch": 0.25, "learning_rate": 1.7493848285008938e-05, "loss": 0.1548, "step": 240 }, { "epoch": 0.26, "learning_rate": 1.7471095111871076e-05, "loss": 0.1749, "step": 241 }, { "epoch": 0.26, "learning_rate": 1.7448254058209244e-05, "loss": 0.1895, "step": 242 }, { "epoch": 0.26, "learning_rate": 1.7425325392696742e-05, "loss": 0.1668, "step": 243 }, { "epoch": 0.26, "learning_rate": 1.7402309385037423e-05, "loss": 0.1675, "step": 244 }, { "epoch": 0.26, "learning_rate": 1.7379206305962525e-05, "loss": 0.1688, "step": 245 }, { "epoch": 0.26, "learning_rate": 1.735601642722748e-05, "loss": 0.1871, "step": 246 }, { "epoch": 0.26, "learning_rate": 1.7332740021608722e-05, "loss": 0.1651, "step": 247 }, { "epoch": 0.26, "learning_rate": 1.7309377362900486e-05, "loss": 0.1945, "step": 248 }, { "epoch": 0.26, "learning_rate": 1.7285928725911562e-05, "loss": 0.1787, "step": 249 }, { "epoch": 0.26, "learning_rate": 1.7262394386462102e-05, "loss": 0.1766, "step": 250 }, { "epoch": 0.27, "learning_rate": 1.7238774621380335e-05, "loss": 0.1692, "step": 251 }, { "epoch": 0.27, "learning_rate": 1.721506970849934e-05, "loss": 0.1538, "step": 252 }, { "epoch": 0.27, "learning_rate": 1.719127992665376e-05, "loss": 0.1746, "step": 253 }, { "epoch": 0.27, "learning_rate": 1.7167405555676535e-05, "loss": 0.2094, "step": 254 }, { "epoch": 0.27, "learning_rate": 1.71434468763956e-05, "loss": 0.2239, "step": 255 }, { "epoch": 0.27, "learning_rate": 1.7119404170630594e-05, "loss": 0.1753, "step": 256 }, { "epoch": 0.27, "learning_rate": 1.709527772118953e-05, "loss": 0.1757, "step": 257 }, { "epoch": 0.27, "learning_rate": 1.7071067811865477e-05, "loss": 0.1679, "step": 258 }, { "epoch": 0.27, "learning_rate": 1.7046774727433223e-05, "loss": 0.1985, "step": 259 }, { "epoch": 0.28, "learning_rate": 1.7022398753645918e-05, "loss": 0.1835, "step": 260 }, { "epoch": 0.28, "learning_rate": 1.6997940177231722e-05, "loss": 0.212, "step": 261 }, { "epoch": 0.28, "learning_rate": 1.697339928589043e-05, "loss": 0.1696, "step": 262 }, { "epoch": 0.28, "learning_rate": 1.6948776368290085e-05, "loss": 0.1902, "step": 263 }, { "epoch": 0.28, "learning_rate": 1.6924071714063574e-05, "loss": 0.1707, "step": 264 }, { "epoch": 0.28, "learning_rate": 1.6899285613805246e-05, "loss": 0.16, "step": 265 }, { "epoch": 0.28, "learning_rate": 1.6874418359067463e-05, "loss": 0.179, "step": 266 }, { "epoch": 0.28, "learning_rate": 1.6849470242357197e-05, "loss": 0.1711, "step": 267 }, { "epoch": 0.28, "learning_rate": 1.682444155713257e-05, "loss": 0.1596, "step": 268 }, { "epoch": 0.28, "learning_rate": 1.6799332597799413e-05, "loss": 0.1937, "step": 269 }, { "epoch": 0.29, "learning_rate": 1.677414365970781e-05, "loss": 0.1669, "step": 270 }, { "epoch": 0.29, "learning_rate": 1.6748875039148592e-05, "loss": 0.1915, "step": 271 }, { "epoch": 0.29, "learning_rate": 1.67235270333499e-05, "loss": 0.1954, "step": 272 }, { "epoch": 0.29, "learning_rate": 1.6698099940473644e-05, "loss": 0.1753, "step": 273 }, { "epoch": 0.29, "learning_rate": 1.6672594059612018e-05, "loss": 0.1766, "step": 274 }, { "epoch": 0.29, "learning_rate": 1.664700969078398e-05, "loss": 0.1864, "step": 275 }, { "epoch": 0.29, "learning_rate": 1.662134713493173e-05, "loss": 0.163, "step": 276 }, { "epoch": 0.29, "learning_rate": 1.659560669391714e-05, "loss": 0.1931, "step": 277 }, { "epoch": 0.29, "learning_rate": 1.6569788670518253e-05, "loss": 0.1521, "step": 278 }, { "epoch": 0.3, "learning_rate": 1.6543893368425664e-05, "loss": 0.1662, "step": 279 }, { "epoch": 0.3, "learning_rate": 1.6517921092239002e-05, "loss": 0.1945, "step": 280 }, { "epoch": 0.3, "learning_rate": 1.6491872147463307e-05, "loss": 0.1652, "step": 281 }, { "epoch": 0.3, "learning_rate": 1.646574684050545e-05, "loss": 0.1874, "step": 282 }, { "epoch": 0.3, "learning_rate": 1.6439545478670543e-05, "loss": 0.2133, "step": 283 }, { "epoch": 0.3, "learning_rate": 1.64132683701583e-05, "loss": 0.1726, "step": 284 }, { "epoch": 0.3, "learning_rate": 1.6386915824059427e-05, "loss": 0.1853, "step": 285 }, { "epoch": 0.3, "learning_rate": 1.6360488150351984e-05, "loss": 0.1741, "step": 286 }, { "epoch": 0.3, "learning_rate": 1.6333985659897737e-05, "loss": 0.1936, "step": 287 }, { "epoch": 0.3, "learning_rate": 1.6307408664438496e-05, "loss": 0.1621, "step": 288 }, { "epoch": 0.31, "learning_rate": 1.6280757476592467e-05, "loss": 0.1691, "step": 289 }, { "epoch": 0.31, "learning_rate": 1.625403240985054e-05, "loss": 0.1737, "step": 290 }, { "epoch": 0.31, "learning_rate": 1.622723377857265e-05, "loss": 0.1847, "step": 291 }, { "epoch": 0.31, "learning_rate": 1.620036189798403e-05, "loss": 0.1832, "step": 292 }, { "epoch": 0.31, "learning_rate": 1.6173417084171537e-05, "loss": 0.1951, "step": 293 }, { "epoch": 0.31, "learning_rate": 1.6146399654079915e-05, "loss": 0.1627, "step": 294 }, { "epoch": 0.31, "learning_rate": 1.611930992550808e-05, "loss": 0.1633, "step": 295 }, { "epoch": 0.31, "learning_rate": 1.6092148217105372e-05, "loss": 0.1699, "step": 296 }, { "epoch": 0.31, "learning_rate": 1.6064914848367818e-05, "loss": 0.1613, "step": 297 }, { "epoch": 0.32, "learning_rate": 1.6037610139634358e-05, "loss": 0.1653, "step": 298 }, { "epoch": 0.32, "learning_rate": 1.601023441208309e-05, "loss": 0.1856, "step": 299 }, { "epoch": 0.32, "learning_rate": 1.598278798772748e-05, "loss": 0.1719, "step": 300 }, { "epoch": 0.32, "learning_rate": 1.5955271189412596e-05, "loss": 0.2009, "step": 301 }, { "epoch": 0.32, "learning_rate": 1.5927684340811292e-05, "loss": 0.1805, "step": 302 }, { "epoch": 0.32, "learning_rate": 1.5900027766420396e-05, "loss": 0.1934, "step": 303 }, { "epoch": 0.32, "learning_rate": 1.5872301791556904e-05, "loss": 0.2081, "step": 304 }, { "epoch": 0.32, "learning_rate": 1.5844506742354163e-05, "loss": 0.1702, "step": 305 }, { "epoch": 0.32, "learning_rate": 1.581664294575801e-05, "loss": 0.2058, "step": 306 }, { "epoch": 0.32, "learning_rate": 1.5788710729522953e-05, "loss": 0.2079, "step": 307 }, { "epoch": 0.33, "learning_rate": 1.576071042220829e-05, "loss": 0.1934, "step": 308 }, { "epoch": 0.33, "learning_rate": 1.5732642353174257e-05, "loss": 0.149, "step": 309 }, { "epoch": 0.33, "learning_rate": 1.5704506852578165e-05, "loss": 0.179, "step": 310 }, { "epoch": 0.33, "learning_rate": 1.567630425137049e-05, "loss": 0.1692, "step": 311 }, { "epoch": 0.33, "learning_rate": 1.5648034881291005e-05, "loss": 0.1747, "step": 312 }, { "epoch": 0.33, "learning_rate": 1.5619699074864864e-05, "loss": 0.1747, "step": 313 }, { "epoch": 0.33, "learning_rate": 1.5591297165398693e-05, "loss": 0.1605, "step": 314 }, { "epoch": 0.33, "learning_rate": 1.5562829486976675e-05, "loss": 0.2003, "step": 315 }, { "epoch": 0.33, "learning_rate": 1.553429637445661e-05, "loss": 0.1552, "step": 316 }, { "epoch": 0.34, "learning_rate": 1.5505698163465986e-05, "loss": 0.1979, "step": 317 }, { "epoch": 0.34, "learning_rate": 1.5477035190398028e-05, "loss": 0.1706, "step": 318 }, { "epoch": 0.34, "learning_rate": 1.5448307792407737e-05, "loss": 0.1963, "step": 319 }, { "epoch": 0.34, "learning_rate": 1.5419516307407925e-05, "loss": 0.1788, "step": 320 }, { "epoch": 0.34, "learning_rate": 1.5390661074065257e-05, "loss": 0.1719, "step": 321 }, { "epoch": 0.34, "learning_rate": 1.5361742431796235e-05, "loss": 0.1737, "step": 322 }, { "epoch": 0.34, "learning_rate": 1.5332760720763232e-05, "loss": 0.1971, "step": 323 }, { "epoch": 0.34, "learning_rate": 1.5303716281870485e-05, "loss": 0.1874, "step": 324 }, { "epoch": 0.34, "learning_rate": 1.5274609456760073e-05, "loss": 0.1752, "step": 325 }, { "epoch": 0.34, "learning_rate": 1.5245440587807917e-05, "loss": 0.1754, "step": 326 }, { "epoch": 0.35, "learning_rate": 1.5216210018119735e-05, "loss": 0.1836, "step": 327 }, { "epoch": 0.35, "learning_rate": 1.5186918091527025e-05, "loss": 0.1718, "step": 328 }, { "epoch": 0.35, "learning_rate": 1.5157565152583002e-05, "loss": 0.1783, "step": 329 }, { "epoch": 0.35, "learning_rate": 1.5128151546558554e-05, "loss": 0.1746, "step": 330 }, { "epoch": 0.35, "learning_rate": 1.5098677619438183e-05, "loss": 0.1828, "step": 331 }, { "epoch": 0.35, "learning_rate": 1.5069143717915928e-05, "loss": 0.1622, "step": 332 }, { "epoch": 0.35, "learning_rate": 1.50395501893913e-05, "loss": 0.1609, "step": 333 }, { "epoch": 0.35, "learning_rate": 1.5009897381965171e-05, "loss": 0.1647, "step": 334 }, { "epoch": 0.35, "learning_rate": 1.4980185644435712e-05, "loss": 0.161, "step": 335 }, { "epoch": 0.36, "learning_rate": 1.495041532629426e-05, "loss": 0.1456, "step": 336 }, { "epoch": 0.36, "learning_rate": 1.4920586777721231e-05, "loss": 0.1805, "step": 337 }, { "epoch": 0.36, "learning_rate": 1.4890700349581986e-05, "loss": 0.1694, "step": 338 }, { "epoch": 0.36, "learning_rate": 1.4860756393422699e-05, "loss": 0.1953, "step": 339 }, { "epoch": 0.36, "learning_rate": 1.4830755261466249e-05, "loss": 0.174, "step": 340 }, { "epoch": 0.36, "learning_rate": 1.4800697306608043e-05, "loss": 0.1779, "step": 341 }, { "epoch": 0.36, "learning_rate": 1.477058288241189e-05, "loss": 0.1992, "step": 342 }, { "epoch": 0.36, "learning_rate": 1.4740412343105828e-05, "loss": 0.1655, "step": 343 }, { "epoch": 0.36, "learning_rate": 1.4710186043577965e-05, "loss": 0.1842, "step": 344 }, { "epoch": 0.37, "learning_rate": 1.4679904339372301e-05, "loss": 0.179, "step": 345 }, { "epoch": 0.37, "learning_rate": 1.4649567586684548e-05, "loss": 0.1771, "step": 346 }, { "epoch": 0.37, "learning_rate": 1.4619176142357936e-05, "loss": 0.17, "step": 347 }, { "epoch": 0.37, "learning_rate": 1.4588730363879015e-05, "loss": 0.1736, "step": 348 }, { "epoch": 0.37, "learning_rate": 1.4558230609373469e-05, "loss": 0.1572, "step": 349 }, { "epoch": 0.37, "learning_rate": 1.4527677237601865e-05, "loss": 0.1661, "step": 350 }, { "epoch": 0.37, "learning_rate": 1.4497070607955477e-05, "loss": 0.1768, "step": 351 }, { "epoch": 0.37, "learning_rate": 1.4466411080452019e-05, "loss": 0.2008, "step": 352 }, { "epoch": 0.37, "learning_rate": 1.4435699015731449e-05, "loss": 0.151, "step": 353 }, { "epoch": 0.37, "learning_rate": 1.4404934775051682e-05, "loss": 0.1498, "step": 354 }, { "epoch": 0.38, "learning_rate": 1.4374118720284388e-05, "loss": 0.1884, "step": 355 }, { "epoch": 0.38, "learning_rate": 1.4343251213910698e-05, "loss": 0.1771, "step": 356 }, { "epoch": 0.38, "learning_rate": 1.4312332619016964e-05, "loss": 0.1902, "step": 357 }, { "epoch": 0.38, "learning_rate": 1.4281363299290466e-05, "loss": 0.1691, "step": 358 }, { "epoch": 0.38, "learning_rate": 1.4250343619015162e-05, "loss": 0.1592, "step": 359 }, { "epoch": 0.38, "learning_rate": 1.4219273943067375e-05, "loss": 0.1769, "step": 360 }, { "epoch": 0.38, "learning_rate": 1.4188154636911524e-05, "loss": 0.1571, "step": 361 }, { "epoch": 0.38, "learning_rate": 1.4156986066595807e-05, "loss": 0.1734, "step": 362 }, { "epoch": 0.38, "learning_rate": 1.412576859874791e-05, "loss": 0.1779, "step": 363 }, { "epoch": 0.39, "learning_rate": 1.4094502600570687e-05, "loss": 0.2015, "step": 364 }, { "epoch": 0.39, "learning_rate": 1.4063188439837831e-05, "loss": 0.1682, "step": 365 }, { "epoch": 0.39, "learning_rate": 1.4031826484889579e-05, "loss": 0.1337, "step": 366 }, { "epoch": 0.39, "learning_rate": 1.4000417104628332e-05, "loss": 0.1711, "step": 367 }, { "epoch": 0.39, "learning_rate": 1.396896066851437e-05, "loss": 0.1489, "step": 368 }, { "epoch": 0.39, "learning_rate": 1.393745754656146e-05, "loss": 0.1701, "step": 369 }, { "epoch": 0.39, "learning_rate": 1.3905908109332526e-05, "loss": 0.1627, "step": 370 }, { "epoch": 0.39, "learning_rate": 1.3874312727935293e-05, "loss": 0.159, "step": 371 }, { "epoch": 0.39, "learning_rate": 1.3842671774017905e-05, "loss": 0.1556, "step": 372 }, { "epoch": 0.39, "learning_rate": 1.3810985619764573e-05, "loss": 0.1837, "step": 373 }, { "epoch": 0.4, "learning_rate": 1.3779254637891181e-05, "loss": 0.1562, "step": 374 }, { "epoch": 0.4, "learning_rate": 1.3747479201640914e-05, "loss": 0.1582, "step": 375 }, { "epoch": 0.4, "learning_rate": 1.3715659684779857e-05, "loss": 0.153, "step": 376 }, { "epoch": 0.4, "learning_rate": 1.3683796461592604e-05, "loss": 0.1954, "step": 377 }, { "epoch": 0.4, "learning_rate": 1.3651889906877865e-05, "loss": 0.1959, "step": 378 }, { "epoch": 0.4, "learning_rate": 1.3619940395944027e-05, "loss": 0.1877, "step": 379 }, { "epoch": 0.4, "learning_rate": 1.3587948304604781e-05, "loss": 0.1806, "step": 380 }, { "epoch": 0.4, "learning_rate": 1.3555914009174665e-05, "loss": 0.1613, "step": 381 }, { "epoch": 0.4, "learning_rate": 1.3523837886464664e-05, "loss": 0.1575, "step": 382 }, { "epoch": 0.41, "learning_rate": 1.3491720313777756e-05, "loss": 0.1968, "step": 383 }, { "epoch": 0.41, "learning_rate": 1.3459561668904496e-05, "loss": 0.1797, "step": 384 }, { "epoch": 0.41, "learning_rate": 1.3427362330118542e-05, "loss": 0.1943, "step": 385 }, { "epoch": 0.41, "learning_rate": 1.3395122676172248e-05, "loss": 0.1663, "step": 386 }, { "epoch": 0.41, "learning_rate": 1.336284308629216e-05, "loss": 0.1837, "step": 387 }, { "epoch": 0.41, "learning_rate": 1.3330523940174603e-05, "loss": 0.1787, "step": 388 }, { "epoch": 0.41, "learning_rate": 1.3298165617981171e-05, "loss": 0.1735, "step": 389 }, { "epoch": 0.41, "learning_rate": 1.3265768500334286e-05, "loss": 0.1456, "step": 390 }, { "epoch": 0.41, "learning_rate": 1.3233332968312715e-05, "loss": 0.1782, "step": 391 }, { "epoch": 0.41, "learning_rate": 1.3200859403447072e-05, "loss": 0.1856, "step": 392 }, { "epoch": 0.42, "learning_rate": 1.3168348187715353e-05, "loss": 0.1901, "step": 393 }, { "epoch": 0.42, "learning_rate": 1.3135799703538418e-05, "loss": 0.1709, "step": 394 }, { "epoch": 0.42, "learning_rate": 1.3103214333775522e-05, "loss": 0.1805, "step": 395 }, { "epoch": 0.42, "learning_rate": 1.3070592461719778e-05, "loss": 0.1437, "step": 396 }, { "epoch": 0.42, "learning_rate": 1.3037934471093683e-05, "loss": 0.1498, "step": 397 }, { "epoch": 0.42, "learning_rate": 1.3005240746044572e-05, "loss": 0.2256, "step": 398 }, { "epoch": 0.42, "learning_rate": 1.2972511671140127e-05, "loss": 0.16, "step": 399 }, { "epoch": 0.42, "learning_rate": 1.293974763136383e-05, "loss": 0.1725, "step": 400 }, { "epoch": 0.42, "learning_rate": 1.2906949012110456e-05, "loss": 0.1752, "step": 401 }, { "epoch": 0.43, "learning_rate": 1.2874116199181522e-05, "loss": 0.1768, "step": 402 }, { "epoch": 0.43, "learning_rate": 1.2841249578780756e-05, "loss": 0.1611, "step": 403 }, { "epoch": 0.43, "learning_rate": 1.2808349537509564e-05, "loss": 0.1631, "step": 404 }, { "epoch": 0.43, "learning_rate": 1.2775416462362458e-05, "loss": 0.1418, "step": 405 }, { "epoch": 0.43, "learning_rate": 1.2742450740722532e-05, "loss": 0.1799, "step": 406 }, { "epoch": 0.43, "learning_rate": 1.2709452760356884e-05, "loss": 0.1442, "step": 407 }, { "epoch": 0.43, "learning_rate": 1.2676422909412067e-05, "loss": 0.1627, "step": 408 }, { "epoch": 0.43, "learning_rate": 1.2643361576409517e-05, "loss": 0.1718, "step": 409 }, { "epoch": 0.43, "learning_rate": 1.2610269150240985e-05, "loss": 0.1635, "step": 410 }, { "epoch": 0.43, "learning_rate": 1.2577146020163968e-05, "loss": 0.1553, "step": 411 }, { "epoch": 0.44, "learning_rate": 1.2543992575797118e-05, "loss": 0.1769, "step": 412 }, { "epoch": 0.44, "learning_rate": 1.2510809207115666e-05, "loss": 0.1843, "step": 413 }, { "epoch": 0.44, "learning_rate": 1.247759630444684e-05, "loss": 0.1547, "step": 414 }, { "epoch": 0.44, "learning_rate": 1.2444354258465268e-05, "loss": 0.1702, "step": 415 }, { "epoch": 0.44, "learning_rate": 1.2411083460188378e-05, "loss": 0.1331, "step": 416 }, { "epoch": 0.44, "learning_rate": 1.2377784300971807e-05, "loss": 0.1664, "step": 417 }, { "epoch": 0.44, "learning_rate": 1.2344457172504796e-05, "loss": 0.1741, "step": 418 }, { "epoch": 0.44, "learning_rate": 1.2311102466805578e-05, "loss": 0.173, "step": 419 }, { "epoch": 0.44, "learning_rate": 1.2277720576216772e-05, "loss": 0.1479, "step": 420 }, { "epoch": 0.45, "learning_rate": 1.2244311893400761e-05, "loss": 0.1768, "step": 421 }, { "epoch": 0.45, "learning_rate": 1.2210876811335089e-05, "loss": 0.1581, "step": 422 }, { "epoch": 0.45, "learning_rate": 1.2177415723307808e-05, "loss": 0.1713, "step": 423 }, { "epoch": 0.45, "learning_rate": 1.2143929022912895e-05, "loss": 0.1534, "step": 424 }, { "epoch": 0.45, "learning_rate": 1.2110417104045575e-05, "loss": 0.1492, "step": 425 }, { "epoch": 0.45, "learning_rate": 1.2076880360897737e-05, "loss": 0.173, "step": 426 }, { "epoch": 0.45, "learning_rate": 1.2043319187953242e-05, "loss": 0.1793, "step": 427 }, { "epoch": 0.45, "learning_rate": 1.2009733979983338e-05, "loss": 0.1532, "step": 428 }, { "epoch": 0.45, "learning_rate": 1.1976125132041974e-05, "loss": 0.2076, "step": 429 }, { "epoch": 0.46, "learning_rate": 1.1942493039461185e-05, "loss": 0.165, "step": 430 }, { "epoch": 0.46, "learning_rate": 1.1908838097846404e-05, "loss": 0.1684, "step": 431 }, { "epoch": 0.46, "learning_rate": 1.187516070307186e-05, "loss": 0.1588, "step": 432 }, { "epoch": 0.46, "learning_rate": 1.1841461251275868e-05, "loss": 0.166, "step": 433 }, { "epoch": 0.46, "learning_rate": 1.1807740138856203e-05, "loss": 0.1801, "step": 434 }, { "epoch": 0.46, "learning_rate": 1.177399776246543e-05, "loss": 0.1755, "step": 435 }, { "epoch": 0.46, "learning_rate": 1.1740234519006238e-05, "loss": 0.1556, "step": 436 }, { "epoch": 0.46, "learning_rate": 1.1706450805626762e-05, "loss": 0.1544, "step": 437 }, { "epoch": 0.46, "learning_rate": 1.1672647019715926e-05, "loss": 0.1785, "step": 438 }, { "epoch": 0.46, "learning_rate": 1.1638823558898762e-05, "loss": 0.1554, "step": 439 }, { "epoch": 0.47, "learning_rate": 1.1604980821031726e-05, "loss": 0.1469, "step": 440 }, { "epoch": 0.47, "learning_rate": 1.1571119204198038e-05, "loss": 0.157, "step": 441 }, { "epoch": 0.47, "learning_rate": 1.1537239106702968e-05, "loss": 0.1688, "step": 442 }, { "epoch": 0.47, "learning_rate": 1.1503340927069189e-05, "loss": 0.1732, "step": 443 }, { "epoch": 0.47, "learning_rate": 1.1469425064032051e-05, "loss": 0.189, "step": 444 }, { "epoch": 0.47, "learning_rate": 1.1435491916534919e-05, "loss": 0.1651, "step": 445 }, { "epoch": 0.47, "learning_rate": 1.1401541883724474e-05, "loss": 0.1724, "step": 446 }, { "epoch": 0.47, "learning_rate": 1.1367575364946006e-05, "loss": 0.1705, "step": 447 }, { "epoch": 0.47, "learning_rate": 1.133359275973873e-05, "loss": 0.1667, "step": 448 }, { "epoch": 0.48, "learning_rate": 1.1299594467831079e-05, "loss": 0.181, "step": 449 }, { "epoch": 0.48, "learning_rate": 1.1265580889136007e-05, "loss": 0.1566, "step": 450 }, { "epoch": 0.48, "learning_rate": 1.1231552423746284e-05, "loss": 0.1652, "step": 451 }, { "epoch": 0.48, "learning_rate": 1.1197509471929779e-05, "loss": 0.1476, "step": 452 }, { "epoch": 0.48, "learning_rate": 1.1163452434124773e-05, "loss": 0.1862, "step": 453 }, { "epoch": 0.48, "learning_rate": 1.1129381710935229e-05, "loss": 0.1567, "step": 454 }, { "epoch": 0.48, "learning_rate": 1.1095297703126094e-05, "loss": 0.1669, "step": 455 }, { "epoch": 0.48, "learning_rate": 1.1061200811618564e-05, "loss": 0.1525, "step": 456 }, { "epoch": 0.48, "learning_rate": 1.1027091437485404e-05, "loss": 0.169, "step": 457 }, { "epoch": 0.48, "learning_rate": 1.0992969981946192e-05, "loss": 0.1495, "step": 458 }, { "epoch": 0.49, "learning_rate": 1.0958836846362622e-05, "loss": 0.1311, "step": 459 }, { "epoch": 0.49, "learning_rate": 1.0924692432233781e-05, "loss": 0.1769, "step": 460 }, { "epoch": 0.49, "learning_rate": 1.0890537141191417e-05, "loss": 0.148, "step": 461 }, { "epoch": 0.49, "learning_rate": 1.0856371374995222e-05, "loss": 0.1639, "step": 462 }, { "epoch": 0.49, "learning_rate": 1.0822195535528106e-05, "loss": 0.1526, "step": 463 }, { "epoch": 0.49, "learning_rate": 1.0788010024791465e-05, "loss": 0.1486, "step": 464 }, { "epoch": 0.49, "learning_rate": 1.0753815244900459e-05, "loss": 0.1348, "step": 465 }, { "epoch": 0.49, "learning_rate": 1.0719611598079278e-05, "loss": 0.1766, "step": 466 }, { "epoch": 0.49, "learning_rate": 1.0685399486656407e-05, "loss": 0.1902, "step": 467 }, { "epoch": 0.5, "learning_rate": 1.0651179313059904e-05, "loss": 0.1559, "step": 468 }, { "epoch": 0.5, "learning_rate": 1.0616951479812658e-05, "loss": 0.1741, "step": 469 }, { "epoch": 0.5, "learning_rate": 1.0582716389527659e-05, "loss": 0.1879, "step": 470 }, { "epoch": 0.5, "learning_rate": 1.0548474444903247e-05, "loss": 0.1678, "step": 471 }, { "epoch": 0.5, "learning_rate": 1.051422604871841e-05, "loss": 0.1581, "step": 472 }, { "epoch": 0.5, "learning_rate": 1.0479971603828001e-05, "loss": 0.17, "step": 473 }, { "epoch": 0.5, "learning_rate": 1.0445711513158039e-05, "loss": 0.1495, "step": 474 }, { "epoch": 0.5, "learning_rate": 1.0411446179700944e-05, "loss": 0.1721, "step": 475 }, { "epoch": 0.5, "learning_rate": 1.0377176006510812e-05, "loss": 0.1772, "step": 476 }, { "epoch": 0.5, "learning_rate": 1.0342901396698658e-05, "loss": 0.1719, "step": 477 }, { "epoch": 0.51, "learning_rate": 1.0308622753427692e-05, "loss": 0.1616, "step": 478 }, { "epoch": 0.51, "learning_rate": 1.0274340479908568e-05, "loss": 0.1441, "step": 479 }, { "epoch": 0.51, "learning_rate": 1.024005497939464e-05, "loss": 0.1609, "step": 480 }, { "epoch": 0.51, "learning_rate": 1.0205766655177217e-05, "loss": 0.1537, "step": 481 }, { "epoch": 0.51, "learning_rate": 1.0171475910580824e-05, "loss": 0.1652, "step": 482 }, { "epoch": 0.51, "learning_rate": 1.0137183148958462e-05, "loss": 0.1674, "step": 483 }, { "epoch": 0.51, "learning_rate": 1.0102888773686852e-05, "loss": 0.1942, "step": 484 }, { "epoch": 0.51, "learning_rate": 1.0068593188161698e-05, "loss": 0.1573, "step": 485 }, { "epoch": 0.51, "learning_rate": 1.0034296795792936e-05, "loss": 0.1552, "step": 486 }, { "epoch": 0.52, "learning_rate": 1e-05, "loss": 0.1719, "step": 487 }, { "epoch": 0.52, "learning_rate": 9.965703204207065e-06, "loss": 0.1786, "step": 488 }, { "epoch": 0.52, "learning_rate": 9.931406811838307e-06, "loss": 0.1584, "step": 489 }, { "epoch": 0.52, "learning_rate": 9.89711122631315e-06, "loss": 0.1495, "step": 490 }, { "epoch": 0.52, "learning_rate": 9.862816851041541e-06, "loss": 0.1563, "step": 491 }, { "epoch": 0.52, "learning_rate": 9.82852408941918e-06, "loss": 0.1406, "step": 492 }, { "epoch": 0.52, "learning_rate": 9.79423334482279e-06, "loss": 0.1956, "step": 493 }, { "epoch": 0.52, "learning_rate": 9.759945020605363e-06, "loss": 0.1658, "step": 494 }, { "epoch": 0.52, "learning_rate": 9.725659520091433e-06, "loss": 0.1519, "step": 495 }, { "epoch": 0.52, "learning_rate": 9.69137724657231e-06, "loss": 0.1797, "step": 496 }, { "epoch": 0.53, "learning_rate": 9.657098603301347e-06, "loss": 0.1713, "step": 497 }, { "epoch": 0.53, "learning_rate": 9.622823993489193e-06, "loss": 0.1469, "step": 498 }, { "epoch": 0.53, "learning_rate": 9.588553820299057e-06, "loss": 0.1526, "step": 499 }, { "epoch": 0.53, "learning_rate": 9.554288486841963e-06, "loss": 0.1721, "step": 500 }, { "epoch": 0.53, "learning_rate": 9.520028396172002e-06, "loss": 0.1497, "step": 501 }, { "epoch": 0.53, "learning_rate": 9.485773951281594e-06, "loss": 0.1597, "step": 502 }, { "epoch": 0.53, "learning_rate": 9.451525555096753e-06, "loss": 0.1603, "step": 503 }, { "epoch": 0.53, "learning_rate": 9.417283610472345e-06, "loss": 0.1558, "step": 504 }, { "epoch": 0.53, "learning_rate": 9.383048520187344e-06, "loss": 0.1417, "step": 505 }, { "epoch": 0.54, "learning_rate": 9.348820686940099e-06, "loss": 0.1705, "step": 506 }, { "epoch": 0.54, "learning_rate": 9.314600513343595e-06, "loss": 0.1592, "step": 507 }, { "epoch": 0.54, "learning_rate": 9.280388401920724e-06, "loss": 0.1918, "step": 508 }, { "epoch": 0.54, "learning_rate": 9.246184755099545e-06, "loss": 0.1534, "step": 509 }, { "epoch": 0.54, "learning_rate": 9.211989975208538e-06, "loss": 0.1559, "step": 510 }, { "epoch": 0.54, "learning_rate": 9.177804464471897e-06, "loss": 0.1515, "step": 511 }, { "epoch": 0.54, "learning_rate": 9.14362862500478e-06, "loss": 0.1519, "step": 512 }, { "epoch": 0.54, "learning_rate": 9.109462858808586e-06, "loss": 0.1731, "step": 513 }, { "epoch": 0.54, "learning_rate": 9.075307567766222e-06, "loss": 0.1337, "step": 514 }, { "epoch": 0.54, "learning_rate": 9.041163153637382e-06, "loss": 0.1588, "step": 515 }, { "epoch": 0.55, "learning_rate": 9.00703001805381e-06, "loss": 0.1674, "step": 516 }, { "epoch": 0.55, "learning_rate": 8.9729085625146e-06, "loss": 0.1519, "step": 517 }, { "epoch": 0.55, "learning_rate": 8.938799188381439e-06, "loss": 0.1472, "step": 518 }, { "epoch": 0.55, "learning_rate": 8.904702296873913e-06, "loss": 0.2063, "step": 519 }, { "epoch": 0.55, "learning_rate": 8.870618289064776e-06, "loss": 0.1545, "step": 520 }, { "epoch": 0.55, "learning_rate": 8.836547565875227e-06, "loss": 0.1587, "step": 521 }, { "epoch": 0.55, "learning_rate": 8.802490528070223e-06, "loss": 0.1602, "step": 522 }, { "epoch": 0.55, "learning_rate": 8.76844757625372e-06, "loss": 0.1548, "step": 523 }, { "epoch": 0.55, "learning_rate": 8.734419110863996e-06, "loss": 0.1608, "step": 524 }, { "epoch": 0.56, "learning_rate": 8.700405532168921e-06, "loss": 0.1613, "step": 525 }, { "epoch": 0.56, "learning_rate": 8.666407240261271e-06, "loss": 0.1769, "step": 526 }, { "epoch": 0.56, "learning_rate": 8.632424635053997e-06, "loss": 0.1738, "step": 527 }, { "epoch": 0.56, "learning_rate": 8.59845811627553e-06, "loss": 0.1274, "step": 528 }, { "epoch": 0.56, "learning_rate": 8.56450808346508e-06, "loss": 0.1793, "step": 529 }, { "epoch": 0.56, "learning_rate": 8.530574935967952e-06, "loss": 0.149, "step": 530 }, { "epoch": 0.56, "learning_rate": 8.496659072930815e-06, "loss": 0.2003, "step": 531 }, { "epoch": 0.56, "learning_rate": 8.462760893297033e-06, "loss": 0.1459, "step": 532 }, { "epoch": 0.56, "learning_rate": 8.428880795801965e-06, "loss": 0.1364, "step": 533 }, { "epoch": 0.57, "learning_rate": 8.395019178968274e-06, "loss": 0.1671, "step": 534 }, { "epoch": 0.57, "learning_rate": 8.36117644110124e-06, "loss": 0.1881, "step": 535 }, { "epoch": 0.57, "learning_rate": 8.327352980284076e-06, "loss": 0.1766, "step": 536 }, { "epoch": 0.57, "learning_rate": 8.293549194373243e-06, "loss": 0.1605, "step": 537 }, { "epoch": 0.57, "learning_rate": 8.259765480993767e-06, "loss": 0.1919, "step": 538 }, { "epoch": 0.57, "learning_rate": 8.226002237534574e-06, "loss": 0.1732, "step": 539 }, { "epoch": 0.57, "learning_rate": 8.1922598611438e-06, "loss": 0.1517, "step": 540 }, { "epoch": 0.57, "learning_rate": 8.158538748724139e-06, "loss": 0.1475, "step": 541 }, { "epoch": 0.57, "learning_rate": 8.124839296928144e-06, "loss": 0.1642, "step": 542 }, { "epoch": 0.57, "learning_rate": 8.091161902153594e-06, "loss": 0.1441, "step": 543 }, { "epoch": 0.58, "learning_rate": 8.057506960538818e-06, "loss": 0.1518, "step": 544 }, { "epoch": 0.58, "learning_rate": 8.023874867958027e-06, "loss": 0.1406, "step": 545 }, { "epoch": 0.58, "learning_rate": 7.990266020016667e-06, "loss": 0.1954, "step": 546 }, { "epoch": 0.58, "learning_rate": 7.95668081204676e-06, "loss": 0.1654, "step": 547 }, { "epoch": 0.58, "learning_rate": 7.923119639102268e-06, "loss": 0.1631, "step": 548 }, { "epoch": 0.58, "learning_rate": 7.889582895954427e-06, "loss": 0.1586, "step": 549 }, { "epoch": 0.58, "learning_rate": 7.85607097708711e-06, "loss": 0.1536, "step": 550 }, { "epoch": 0.58, "learning_rate": 7.822584276692192e-06, "loss": 0.1939, "step": 551 }, { "epoch": 0.58, "learning_rate": 7.789123188664914e-06, "loss": 0.1597, "step": 552 }, { "epoch": 0.59, "learning_rate": 7.75568810659924e-06, "loss": 0.1479, "step": 553 }, { "epoch": 0.59, "learning_rate": 7.722279423783231e-06, "loss": 0.1635, "step": 554 }, { "epoch": 0.59, "learning_rate": 7.688897533194423e-06, "loss": 0.1781, "step": 555 }, { "epoch": 0.59, "learning_rate": 7.655542827495205e-06, "loss": 0.1832, "step": 556 }, { "epoch": 0.59, "learning_rate": 7.622215699028196e-06, "loss": 0.1431, "step": 557 }, { "epoch": 0.59, "learning_rate": 7.588916539811626e-06, "loss": 0.1688, "step": 558 }, { "epoch": 0.59, "learning_rate": 7.555645741534736e-06, "loss": 0.1573, "step": 559 }, { "epoch": 0.59, "learning_rate": 7.52240369555316e-06, "loss": 0.1619, "step": 560 }, { "epoch": 0.59, "learning_rate": 7.489190792884338e-06, "loss": 0.1605, "step": 561 }, { "epoch": 0.59, "learning_rate": 7.456007424202886e-06, "loss": 0.1505, "step": 562 }, { "epoch": 0.6, "learning_rate": 7.422853979836035e-06, "loss": 0.1549, "step": 563 }, { "epoch": 0.6, "learning_rate": 7.389730849759018e-06, "loss": 0.1475, "step": 564 }, { "epoch": 0.6, "learning_rate": 7.3566384235904855e-06, "loss": 0.1652, "step": 565 }, { "epoch": 0.6, "learning_rate": 7.323577090587936e-06, "loss": 0.1417, "step": 566 }, { "epoch": 0.6, "learning_rate": 7.290547239643117e-06, "loss": 0.1606, "step": 567 }, { "epoch": 0.6, "learning_rate": 7.257549259277472e-06, "loss": 0.1558, "step": 568 }, { "epoch": 0.6, "learning_rate": 7.224583537637544e-06, "loss": 0.1742, "step": 569 }, { "epoch": 0.6, "learning_rate": 7.191650462490439e-06, "loss": 0.1355, "step": 570 }, { "epoch": 0.6, "learning_rate": 7.158750421219245e-06, "loss": 0.1612, "step": 571 }, { "epoch": 0.61, "learning_rate": 7.125883800818482e-06, "loss": 0.1493, "step": 572 }, { "epoch": 0.61, "learning_rate": 7.093050987889547e-06, "loss": 0.1343, "step": 573 }, { "epoch": 0.61, "learning_rate": 7.060252368636171e-06, "loss": 0.1364, "step": 574 }, { "epoch": 0.61, "learning_rate": 7.027488328859876e-06, "loss": 0.1659, "step": 575 }, { "epoch": 0.61, "learning_rate": 6.994759253955432e-06, "loss": 0.1554, "step": 576 }, { "epoch": 0.61, "learning_rate": 6.962065528906321e-06, "loss": 0.1582, "step": 577 }, { "epoch": 0.61, "learning_rate": 6.9294075382802215e-06, "loss": 0.1526, "step": 578 }, { "epoch": 0.61, "learning_rate": 6.896785666224482e-06, "loss": 0.1365, "step": 579 }, { "epoch": 0.61, "learning_rate": 6.864200296461585e-06, "loss": 0.1593, "step": 580 }, { "epoch": 0.61, "learning_rate": 6.831651812284652e-06, "loss": 0.1712, "step": 581 }, { "epoch": 0.62, "learning_rate": 6.799140596552929e-06, "loss": 0.1429, "step": 582 }, { "epoch": 0.62, "learning_rate": 6.766667031687286e-06, "loss": 0.1566, "step": 583 }, { "epoch": 0.62, "learning_rate": 6.734231499665717e-06, "loss": 0.1523, "step": 584 }, { "epoch": 0.62, "learning_rate": 6.7018343820188324e-06, "loss": 0.1514, "step": 585 }, { "epoch": 0.62, "learning_rate": 6.669476059825401e-06, "loss": 0.1373, "step": 586 }, { "epoch": 0.62, "learning_rate": 6.637156913707839e-06, "loss": 0.1738, "step": 587 }, { "epoch": 0.62, "learning_rate": 6.604877323827755e-06, "loss": 0.1703, "step": 588 }, { "epoch": 0.62, "learning_rate": 6.572637669881459e-06, "loss": 0.1828, "step": 589 }, { "epoch": 0.62, "learning_rate": 6.540438331095509e-06, "loss": 0.1624, "step": 590 }, { "epoch": 0.63, "learning_rate": 6.508279686222243e-06, "loss": 0.1625, "step": 591 }, { "epoch": 0.63, "learning_rate": 6.47616211353534e-06, "loss": 0.1369, "step": 592 }, { "epoch": 0.63, "learning_rate": 6.444085990825338e-06, "loss": 0.155, "step": 593 }, { "epoch": 0.63, "learning_rate": 6.412051695395225e-06, "loss": 0.1772, "step": 594 }, { "epoch": 0.63, "learning_rate": 6.380059604055975e-06, "loss": 0.1478, "step": 595 }, { "epoch": 0.63, "learning_rate": 6.34811009312214e-06, "loss": 0.1727, "step": 596 }, { "epoch": 0.63, "learning_rate": 6.316203538407397e-06, "loss": 0.1654, "step": 597 }, { "epoch": 0.63, "learning_rate": 6.284340315220146e-06, "loss": 0.1493, "step": 598 }, { "epoch": 0.63, "learning_rate": 6.252520798359092e-06, "loss": 0.1693, "step": 599 }, { "epoch": 0.63, "learning_rate": 6.22074536210882e-06, "loss": 0.1368, "step": 600 }, { "epoch": 0.64, "learning_rate": 6.18901438023543e-06, "loss": 0.1303, "step": 601 }, { "epoch": 0.64, "learning_rate": 6.157328225982098e-06, "loss": 0.1431, "step": 602 }, { "epoch": 0.64, "learning_rate": 6.125687272064713e-06, "loss": 0.2003, "step": 603 }, { "epoch": 0.64, "learning_rate": 6.094091890667473e-06, "loss": 0.1772, "step": 604 }, { "epoch": 0.64, "learning_rate": 6.0625424534385425e-06, "loss": 0.1752, "step": 605 }, { "epoch": 0.64, "learning_rate": 6.031039331485631e-06, "loss": 0.1427, "step": 606 }, { "epoch": 0.64, "learning_rate": 5.999582895371669e-06, "loss": 0.179, "step": 607 }, { "epoch": 0.64, "learning_rate": 5.968173515110427e-06, "loss": 0.1545, "step": 608 }, { "epoch": 0.64, "learning_rate": 5.936811560162169e-06, "loss": 0.1549, "step": 609 }, { "epoch": 0.65, "learning_rate": 5.905497399429316e-06, "loss": 0.1512, "step": 610 }, { "epoch": 0.65, "learning_rate": 5.874231401252091e-06, "loss": 0.1317, "step": 611 }, { "epoch": 0.65, "learning_rate": 5.843013933404197e-06, "loss": 0.1656, "step": 612 }, { "epoch": 0.65, "learning_rate": 5.811845363088477e-06, "loss": 0.1354, "step": 613 }, { "epoch": 0.65, "learning_rate": 5.78072605693263e-06, "loss": 0.176, "step": 614 }, { "epoch": 0.65, "learning_rate": 5.7496563809848434e-06, "loss": 0.1751, "step": 615 }, { "epoch": 0.65, "learning_rate": 5.718636700709538e-06, "loss": 0.161, "step": 616 }, { "epoch": 0.65, "learning_rate": 5.687667380983037e-06, "loss": 0.1559, "step": 617 }, { "epoch": 0.65, "learning_rate": 5.656748786089304e-06, "loss": 0.18, "step": 618 }, { "epoch": 0.66, "learning_rate": 5.625881279715615e-06, "loss": 0.1296, "step": 619 }, { "epoch": 0.66, "learning_rate": 5.59506522494832e-06, "loss": 0.1404, "step": 620 }, { "epoch": 0.66, "learning_rate": 5.564300984268556e-06, "loss": 0.1471, "step": 621 }, { "epoch": 0.66, "learning_rate": 5.533588919547979e-06, "loss": 0.1673, "step": 622 }, { "epoch": 0.66, "learning_rate": 5.502929392044528e-06, "loss": 0.1345, "step": 623 }, { "epoch": 0.66, "learning_rate": 5.472322762398139e-06, "loss": 0.1517, "step": 624 }, { "epoch": 0.66, "learning_rate": 5.441769390626537e-06, "loss": 0.1605, "step": 625 }, { "epoch": 0.66, "learning_rate": 5.411269636120984e-06, "loss": 0.1486, "step": 626 }, { "epoch": 0.66, "learning_rate": 5.3808238576420694e-06, "loss": 0.161, "step": 627 }, { "epoch": 0.66, "learning_rate": 5.350432413315455e-06, "loss": 0.1402, "step": 628 }, { "epoch": 0.67, "learning_rate": 5.3200956606277006e-06, "loss": 0.1523, "step": 629 }, { "epoch": 0.67, "learning_rate": 5.289813956422037e-06, "loss": 0.1819, "step": 630 }, { "epoch": 0.67, "learning_rate": 5.259587656894174e-06, "loss": 0.1451, "step": 631 }, { "epoch": 0.67, "learning_rate": 5.229417117588112e-06, "loss": 0.1317, "step": 632 }, { "epoch": 0.67, "learning_rate": 5.199302693391958e-06, "loss": 0.1534, "step": 633 }, { "epoch": 0.67, "learning_rate": 5.169244738533754e-06, "loss": 0.166, "step": 634 }, { "epoch": 0.67, "learning_rate": 5.139243606577302e-06, "loss": 0.1554, "step": 635 }, { "epoch": 0.67, "learning_rate": 5.109299650418018e-06, "loss": 0.1586, "step": 636 }, { "epoch": 0.67, "learning_rate": 5.07941322227877e-06, "loss": 0.1608, "step": 637 }, { "epoch": 0.68, "learning_rate": 5.049584673705742e-06, "loss": 0.1397, "step": 638 }, { "epoch": 0.68, "learning_rate": 5.019814355564292e-06, "loss": 0.15, "step": 639 }, { "epoch": 0.68, "learning_rate": 4.990102618034832e-06, "loss": 0.1719, "step": 640 }, { "epoch": 0.68, "learning_rate": 4.960449810608705e-06, "loss": 0.1709, "step": 641 }, { "epoch": 0.68, "learning_rate": 4.930856282084074e-06, "loss": 0.1503, "step": 642 }, { "epoch": 0.68, "learning_rate": 4.90132238056182e-06, "loss": 0.1611, "step": 643 }, { "epoch": 0.68, "learning_rate": 4.871848453441449e-06, "loss": 0.1637, "step": 644 }, { "epoch": 0.68, "learning_rate": 4.842434847417001e-06, "loss": 0.1622, "step": 645 }, { "epoch": 0.68, "learning_rate": 4.813081908472977e-06, "loss": 0.1268, "step": 646 }, { "epoch": 0.68, "learning_rate": 4.783789981880267e-06, "loss": 0.1566, "step": 647 }, { "epoch": 0.69, "learning_rate": 4.754559412192087e-06, "loss": 0.1494, "step": 648 }, { "epoch": 0.69, "learning_rate": 4.7253905432399295e-06, "loss": 0.1556, "step": 649 }, { "epoch": 0.69, "learning_rate": 4.696283718129517e-06, "loss": 0.1321, "step": 650 }, { "epoch": 0.69, "learning_rate": 4.667239279236768e-06, "loss": 0.1418, "step": 651 }, { "epoch": 0.69, "learning_rate": 4.6382575682037685e-06, "loss": 0.1356, "step": 652 }, { "epoch": 0.69, "learning_rate": 4.609338925934743e-06, "loss": 0.1297, "step": 653 }, { "epoch": 0.69, "learning_rate": 4.580483692592074e-06, "loss": 0.1531, "step": 654 }, { "epoch": 0.69, "learning_rate": 4.551692207592265e-06, "loss": 0.1638, "step": 655 }, { "epoch": 0.69, "learning_rate": 4.522964809601978e-06, "loss": 0.1347, "step": 656 }, { "epoch": 0.7, "learning_rate": 4.494301836534016e-06, "loss": 0.1639, "step": 657 }, { "epoch": 0.7, "learning_rate": 4.465703625543393e-06, "loss": 0.1619, "step": 658 }, { "epoch": 0.7, "learning_rate": 4.437170513023328e-06, "loss": 0.1426, "step": 659 }, { "epoch": 0.7, "learning_rate": 4.408702834601309e-06, "loss": 0.1478, "step": 660 }, { "epoch": 0.7, "learning_rate": 4.380300925135138e-06, "loss": 0.1368, "step": 661 }, { "epoch": 0.7, "learning_rate": 4.351965118708996e-06, "loss": 0.1509, "step": 662 }, { "epoch": 0.7, "learning_rate": 4.3236957486295115e-06, "loss": 0.1674, "step": 663 }, { "epoch": 0.7, "learning_rate": 4.295493147421836e-06, "loss": 0.1439, "step": 664 }, { "epoch": 0.7, "learning_rate": 4.267357646825746e-06, "loss": 0.1643, "step": 665 }, { "epoch": 0.7, "learning_rate": 4.239289577791712e-06, "loss": 0.1626, "step": 666 }, { "epoch": 0.71, "learning_rate": 4.211289270477047e-06, "loss": 0.1414, "step": 667 }, { "epoch": 0.71, "learning_rate": 4.183357054241989e-06, "loss": 0.1479, "step": 668 }, { "epoch": 0.71, "learning_rate": 4.155493257645842e-06, "loss": 0.135, "step": 669 }, { "epoch": 0.71, "learning_rate": 4.127698208443097e-06, "loss": 0.1652, "step": 670 }, { "epoch": 0.71, "learning_rate": 4.099972233579608e-06, "loss": 0.1674, "step": 671 }, { "epoch": 0.71, "learning_rate": 4.072315659188709e-06, "loss": 0.1458, "step": 672 }, { "epoch": 0.71, "learning_rate": 4.044728810587406e-06, "loss": 0.1427, "step": 673 }, { "epoch": 0.71, "learning_rate": 4.017212012272524e-06, "loss": 0.1591, "step": 674 }, { "epoch": 0.71, "learning_rate": 3.989765587916914e-06, "loss": 0.1525, "step": 675 }, { "epoch": 0.72, "learning_rate": 3.962389860365643e-06, "loss": 0.1439, "step": 676 }, { "epoch": 0.72, "learning_rate": 3.935085151632185e-06, "loss": 0.1467, "step": 677 }, { "epoch": 0.72, "learning_rate": 3.90785178289463e-06, "loss": 0.1445, "step": 678 }, { "epoch": 0.72, "learning_rate": 3.8806900744919205e-06, "loss": 0.1448, "step": 679 }, { "epoch": 0.72, "learning_rate": 3.853600345920087e-06, "loss": 0.1311, "step": 680 }, { "epoch": 0.72, "learning_rate": 3.826582915828468e-06, "loss": 0.145, "step": 681 }, { "epoch": 0.72, "learning_rate": 3.7996381020159733e-06, "loss": 0.1544, "step": 682 }, { "epoch": 0.72, "learning_rate": 3.7727662214273496e-06, "loss": 0.1776, "step": 683 }, { "epoch": 0.72, "learning_rate": 3.7459675901494587e-06, "loss": 0.1532, "step": 684 }, { "epoch": 0.72, "learning_rate": 3.719242523407539e-06, "loss": 0.1525, "step": 685 }, { "epoch": 0.73, "learning_rate": 3.6925913355615075e-06, "loss": 0.1479, "step": 686 }, { "epoch": 0.73, "learning_rate": 3.6660143401022687e-06, "loss": 0.1343, "step": 687 }, { "epoch": 0.73, "learning_rate": 3.6395118496480176e-06, "loss": 0.1765, "step": 688 }, { "epoch": 0.73, "learning_rate": 3.6130841759405776e-06, "loss": 0.1525, "step": 689 }, { "epoch": 0.73, "learning_rate": 3.5867316298417053e-06, "loss": 0.1207, "step": 690 }, { "epoch": 0.73, "learning_rate": 3.5604545213294616e-06, "loss": 0.1422, "step": 691 }, { "epoch": 0.73, "learning_rate": 3.5342531594945493e-06, "loss": 0.1448, "step": 692 }, { "epoch": 0.73, "learning_rate": 3.508127852536698e-06, "loss": 0.1448, "step": 693 }, { "epoch": 0.73, "learning_rate": 3.4820789077609997e-06, "loss": 0.1562, "step": 694 }, { "epoch": 0.74, "learning_rate": 3.4561066315743365e-06, "loss": 0.1523, "step": 695 }, { "epoch": 0.74, "learning_rate": 3.4302113294817505e-06, "loss": 0.1406, "step": 696 }, { "epoch": 0.74, "learning_rate": 3.4043933060828606e-06, "loss": 0.1373, "step": 697 }, { "epoch": 0.74, "learning_rate": 3.378652865068275e-06, "loss": 0.139, "step": 698 }, { "epoch": 0.74, "learning_rate": 3.352990309216022e-06, "loss": 0.1824, "step": 699 }, { "epoch": 0.74, "learning_rate": 3.3274059403879865e-06, "loss": 0.1463, "step": 700 }, { "epoch": 0.74, "learning_rate": 3.3019000595263573e-06, "loss": 0.1438, "step": 701 }, { "epoch": 0.74, "learning_rate": 3.2764729666501027e-06, "loss": 0.162, "step": 702 }, { "epoch": 0.74, "learning_rate": 3.2511249608514083e-06, "loss": 0.1421, "step": 703 }, { "epoch": 0.74, "learning_rate": 3.225856340292195e-06, "loss": 0.1581, "step": 704 }, { "epoch": 0.75, "learning_rate": 3.2006674022005857e-06, "loss": 0.1207, "step": 705 }, { "epoch": 0.75, "learning_rate": 3.1755584428674345e-06, "loss": 0.1577, "step": 706 }, { "epoch": 0.75, "learning_rate": 3.1505297576428074e-06, "loss": 0.139, "step": 707 }, { "epoch": 0.75, "learning_rate": 3.12558164093254e-06, "loss": 0.1411, "step": 708 }, { "epoch": 0.75, "learning_rate": 3.1007143861947575e-06, "loss": 0.1487, "step": 709 }, { "epoch": 0.75, "learning_rate": 3.075928285936427e-06, "loss": 0.1616, "step": 710 }, { "epoch": 0.75, "learning_rate": 3.0512236317099173e-06, "loss": 0.1791, "step": 711 }, { "epoch": 0.75, "learning_rate": 3.026600714109571e-06, "loss": 0.1325, "step": 712 }, { "epoch": 0.75, "learning_rate": 3.0020598227682794e-06, "loss": 0.1549, "step": 713 }, { "epoch": 0.76, "learning_rate": 2.977601246354086e-06, "loss": 0.1489, "step": 714 }, { "epoch": 0.76, "learning_rate": 2.953225272566782e-06, "loss": 0.1686, "step": 715 }, { "epoch": 0.76, "learning_rate": 2.9289321881345257e-06, "loss": 0.1417, "step": 716 }, { "epoch": 0.76, "learning_rate": 2.9047222788104712e-06, "loss": 0.1448, "step": 717 }, { "epoch": 0.76, "learning_rate": 2.880595829369406e-06, "loss": 0.1461, "step": 718 }, { "epoch": 0.76, "learning_rate": 2.8565531236043997e-06, "loss": 0.1566, "step": 719 }, { "epoch": 0.76, "learning_rate": 2.8325944443234674e-06, "loss": 0.124, "step": 720 }, { "epoch": 0.76, "learning_rate": 2.8087200733462427e-06, "loss": 0.1511, "step": 721 }, { "epoch": 0.76, "learning_rate": 2.7849302915006626e-06, "loss": 0.1261, "step": 722 }, { "epoch": 0.77, "learning_rate": 2.7612253786196663e-06, "loss": 0.153, "step": 723 }, { "epoch": 0.77, "learning_rate": 2.737605613537899e-06, "loss": 0.1627, "step": 724 }, { "epoch": 0.77, "learning_rate": 2.714071274088438e-06, "loss": 0.1552, "step": 725 }, { "epoch": 0.77, "learning_rate": 2.690622637099517e-06, "loss": 0.1519, "step": 726 }, { "epoch": 0.77, "learning_rate": 2.667259978391281e-06, "loss": 0.1546, "step": 727 }, { "epoch": 0.77, "learning_rate": 2.6439835727725217e-06, "loss": 0.1407, "step": 728 }, { "epoch": 0.77, "learning_rate": 2.6207936940374767e-06, "loss": 0.1183, "step": 729 }, { "epoch": 0.77, "learning_rate": 2.597690614962578e-06, "loss": 0.1262, "step": 730 }, { "epoch": 0.77, "learning_rate": 2.5746746073032624e-06, "loss": 0.1283, "step": 731 }, { "epoch": 0.77, "learning_rate": 2.5517459417907575e-06, "loss": 0.1537, "step": 732 }, { "epoch": 0.78, "learning_rate": 2.5289048881289256e-06, "loss": 0.1547, "step": 733 }, { "epoch": 0.78, "learning_rate": 2.506151714991062e-06, "loss": 0.1193, "step": 734 }, { "epoch": 0.78, "learning_rate": 2.4834866900167478e-06, "loss": 0.1649, "step": 735 }, { "epoch": 0.78, "learning_rate": 2.460910079808703e-06, "loss": 0.1245, "step": 736 }, { "epoch": 0.78, "learning_rate": 2.4384221499296466e-06, "loss": 0.1282, "step": 737 }, { "epoch": 0.78, "learning_rate": 2.416023164899175e-06, "loss": 0.1372, "step": 738 }, { "epoch": 0.78, "learning_rate": 2.3937133881906484e-06, "loss": 0.161, "step": 739 }, { "epoch": 0.78, "learning_rate": 2.371493082228098e-06, "loss": 0.1729, "step": 740 }, { "epoch": 0.78, "learning_rate": 2.3493625083831217e-06, "loss": 0.1481, "step": 741 }, { "epoch": 0.79, "learning_rate": 2.3273219269718384e-06, "loss": 0.1422, "step": 742 }, { "epoch": 0.79, "learning_rate": 2.3053715972517998e-06, "loss": 0.1421, "step": 743 }, { "epoch": 0.79, "learning_rate": 2.283511777418954e-06, "loss": 0.1369, "step": 744 }, { "epoch": 0.79, "learning_rate": 2.2617427246045976e-06, "loss": 0.13, "step": 745 }, { "epoch": 0.79, "learning_rate": 2.2400646948723724e-06, "loss": 0.1346, "step": 746 }, { "epoch": 0.79, "learning_rate": 2.218477943215229e-06, "loss": 0.1328, "step": 747 }, { "epoch": 0.79, "learning_rate": 2.1969827235524467e-06, "loss": 0.1382, "step": 748 }, { "epoch": 0.79, "learning_rate": 2.1755792887266236e-06, "loss": 0.1667, "step": 749 }, { "epoch": 0.79, "learning_rate": 2.1542678905007287e-06, "loss": 0.1215, "step": 750 }, { "epoch": 0.79, "learning_rate": 2.133048779555129e-06, "loss": 0.1579, "step": 751 }, { "epoch": 0.8, "learning_rate": 2.111922205484641e-06, "loss": 0.1506, "step": 752 }, { "epoch": 0.8, "learning_rate": 2.090888416795582e-06, "loss": 0.1332, "step": 753 }, { "epoch": 0.8, "learning_rate": 2.0699476609028714e-06, "loss": 0.1535, "step": 754 }, { "epoch": 0.8, "learning_rate": 2.0491001841271073e-06, "loss": 0.1409, "step": 755 }, { "epoch": 0.8, "learning_rate": 2.0283462316916712e-06, "loss": 0.1771, "step": 756 }, { "epoch": 0.8, "learning_rate": 2.007686047719831e-06, "loss": 0.1565, "step": 757 }, { "epoch": 0.8, "learning_rate": 1.987119875231891e-06, "loss": 0.145, "step": 758 }, { "epoch": 0.8, "learning_rate": 1.9666479561423247e-06, "loss": 0.165, "step": 759 }, { "epoch": 0.8, "learning_rate": 1.9462705312569265e-06, "loss": 0.1254, "step": 760 }, { "epoch": 0.81, "learning_rate": 1.9259878402699704e-06, "loss": 0.15, "step": 761 }, { "epoch": 0.81, "learning_rate": 1.9058001217614118e-06, "loss": 0.1693, "step": 762 }, { "epoch": 0.81, "learning_rate": 1.8857076131940643e-06, "loss": 0.1694, "step": 763 }, { "epoch": 0.81, "learning_rate": 1.8657105509108208e-06, "loss": 0.1464, "step": 764 }, { "epoch": 0.81, "learning_rate": 1.8458091701318504e-06, "loss": 0.1712, "step": 765 }, { "epoch": 0.81, "learning_rate": 1.8260037049518565e-06, "loss": 0.1452, "step": 766 }, { "epoch": 0.81, "learning_rate": 1.8062943883373052e-06, "loss": 0.1636, "step": 767 }, { "epoch": 0.81, "learning_rate": 1.7866814521237064e-06, "loss": 0.14, "step": 768 }, { "epoch": 0.81, "learning_rate": 1.7671651270128531e-06, "loss": 0.1506, "step": 769 }, { "epoch": 0.81, "learning_rate": 1.747745642570141e-06, "loss": 0.154, "step": 770 }, { "epoch": 0.82, "learning_rate": 1.7284232272218503e-06, "loss": 0.1264, "step": 771 }, { "epoch": 0.82, "learning_rate": 1.7091981082524633e-06, "loss": 0.1253, "step": 772 }, { "epoch": 0.82, "learning_rate": 1.69007051180199e-06, "loss": 0.1353, "step": 773 }, { "epoch": 0.82, "learning_rate": 1.6710406628633113e-06, "loss": 0.1415, "step": 774 }, { "epoch": 0.82, "learning_rate": 1.652108785279526e-06, "loss": 0.1536, "step": 775 }, { "epoch": 0.82, "learning_rate": 1.6332751017413218e-06, "loss": 0.1309, "step": 776 }, { "epoch": 0.82, "learning_rate": 1.6145398337843654e-06, "loss": 0.1349, "step": 777 }, { "epoch": 0.82, "learning_rate": 1.5959032017866705e-06, "loss": 0.1409, "step": 778 }, { "epoch": 0.82, "learning_rate": 1.5773654249660342e-06, "loss": 0.1288, "step": 779 }, { "epoch": 0.83, "learning_rate": 1.5589267213774394e-06, "loss": 0.1677, "step": 780 }, { "epoch": 0.83, "learning_rate": 1.5405873079105083e-06, "loss": 0.1105, "step": 781 }, { "epoch": 0.83, "learning_rate": 1.522347400286921e-06, "loss": 0.1428, "step": 782 }, { "epoch": 0.83, "learning_rate": 1.5042072130579122e-06, "loss": 0.1635, "step": 783 }, { "epoch": 0.83, "learning_rate": 1.4861669596017237e-06, "loss": 0.1631, "step": 784 }, { "epoch": 0.83, "learning_rate": 1.4682268521211075e-06, "loss": 0.149, "step": 785 }, { "epoch": 0.83, "learning_rate": 1.45038710164082e-06, "loss": 0.1434, "step": 786 }, { "epoch": 0.83, "learning_rate": 1.43264791800515e-06, "loss": 0.1626, "step": 787 }, { "epoch": 0.83, "learning_rate": 1.4150095098754402e-06, "loss": 0.1511, "step": 788 }, { "epoch": 0.83, "learning_rate": 1.3974720847276412e-06, "loss": 0.1329, "step": 789 }, { "epoch": 0.84, "learning_rate": 1.3800358488498645e-06, "loss": 0.1707, "step": 790 }, { "epoch": 0.84, "learning_rate": 1.3627010073399605e-06, "loss": 0.17, "step": 791 }, { "epoch": 0.84, "learning_rate": 1.3454677641031032e-06, "loss": 0.1479, "step": 792 }, { "epoch": 0.84, "learning_rate": 1.3283363218493962e-06, "loss": 0.1275, "step": 793 }, { "epoch": 0.84, "learning_rate": 1.3113068820914798e-06, "loss": 0.1429, "step": 794 }, { "epoch": 0.84, "learning_rate": 1.2943796451421686e-06, "loss": 0.1609, "step": 795 }, { "epoch": 0.84, "learning_rate": 1.2775548101120949e-06, "loss": 0.1356, "step": 796 }, { "epoch": 0.84, "learning_rate": 1.2608325749073591e-06, "loss": 0.1253, "step": 797 }, { "epoch": 0.84, "learning_rate": 1.2442131362272093e-06, "loss": 0.1437, "step": 798 }, { "epoch": 0.85, "learning_rate": 1.2276966895617271e-06, "loss": 0.1503, "step": 799 }, { "epoch": 0.85, "learning_rate": 1.2112834291895237e-06, "loss": 0.1301, "step": 800 }, { "epoch": 0.85, "learning_rate": 1.1949735481754565e-06, "loss": 0.1578, "step": 801 }, { "epoch": 0.85, "learning_rate": 1.1787672383683613e-06, "loss": 0.1425, "step": 802 }, { "epoch": 0.85, "learning_rate": 1.1626646903987904e-06, "loss": 0.1453, "step": 803 }, { "epoch": 0.85, "learning_rate": 1.146666093676776e-06, "loss": 0.1263, "step": 804 }, { "epoch": 0.85, "learning_rate": 1.130771636389596e-06, "loss": 0.1403, "step": 805 }, { "epoch": 0.85, "learning_rate": 1.1149815054995682e-06, "loss": 0.1499, "step": 806 }, { "epoch": 0.85, "learning_rate": 1.0992958867418358e-06, "loss": 0.1264, "step": 807 }, { "epoch": 0.86, "learning_rate": 1.083714964622208e-06, "loss": 0.1398, "step": 808 }, { "epoch": 0.86, "learning_rate": 1.0682389224149648e-06, "loss": 0.1495, "step": 809 }, { "epoch": 0.86, "learning_rate": 1.052867942160718e-06, "loss": 0.1531, "step": 810 }, { "epoch": 0.86, "learning_rate": 1.037602204664252e-06, "loss": 0.1473, "step": 811 }, { "epoch": 0.86, "learning_rate": 1.0224418894924226e-06, "loss": 0.1647, "step": 812 }, { "epoch": 0.86, "learning_rate": 1.0073871749720221e-06, "loss": 0.1373, "step": 813 }, { "epoch": 0.86, "learning_rate": 9.924382381876917e-07, "loss": 0.1507, "step": 814 }, { "epoch": 0.86, "learning_rate": 9.775952549798406e-07, "loss": 0.1407, "step": 815 }, { "epoch": 0.86, "learning_rate": 9.628583999425644e-07, "loss": 0.1623, "step": 816 }, { "epoch": 0.86, "learning_rate": 9.482278464216121e-07, "loss": 0.1476, "step": 817 }, { "epoch": 0.87, "learning_rate": 9.337037665123305e-07, "loss": 0.1503, "step": 818 }, { "epoch": 0.87, "learning_rate": 9.192863310576472e-07, "loss": 0.1356, "step": 819 }, { "epoch": 0.87, "learning_rate": 9.04975709646052e-07, "loss": 0.175, "step": 820 }, { "epoch": 0.87, "learning_rate": 8.907720706096223e-07, "loss": 0.1301, "step": 821 }, { "epoch": 0.87, "learning_rate": 8.766755810220195e-07, "loss": 0.1445, "step": 822 }, { "epoch": 0.87, "learning_rate": 8.626864066965401e-07, "loss": 0.1525, "step": 823 }, { "epoch": 0.87, "learning_rate": 8.488047121841525e-07, "loss": 0.1639, "step": 824 }, { "epoch": 0.87, "learning_rate": 8.350306607715774e-07, "loss": 0.1667, "step": 825 }, { "epoch": 0.87, "learning_rate": 8.21364414479352e-07, "loss": 0.1394, "step": 826 }, { "epoch": 0.88, "learning_rate": 8.07806134059933e-07, "loss": 0.142, "step": 827 }, { "epoch": 0.88, "learning_rate": 7.943559789957989e-07, "loss": 0.1265, "step": 828 }, { "epoch": 0.88, "learning_rate": 7.810141074975819e-07, "loss": 0.1558, "step": 829 }, { "epoch": 0.88, "learning_rate": 7.677806765022044e-07, "loss": 0.1281, "step": 830 }, { "epoch": 0.88, "learning_rate": 7.546558416710292e-07, "loss": 0.1569, "step": 831 }, { "epoch": 0.88, "learning_rate": 7.416397573880296e-07, "loss": 0.1552, "step": 832 }, { "epoch": 0.88, "learning_rate": 7.287325767579756e-07, "loss": 0.1323, "step": 833 }, { "epoch": 0.88, "learning_rate": 7.159344516046373e-07, "loss": 0.1512, "step": 834 }, { "epoch": 0.88, "learning_rate": 7.032455324689902e-07, "loss": 0.1762, "step": 835 }, { "epoch": 0.88, "learning_rate": 6.906659686074436e-07, "loss": 0.1467, "step": 836 }, { "epoch": 0.89, "learning_rate": 6.781959079900958e-07, "loss": 0.1315, "step": 837 }, { "epoch": 0.89, "learning_rate": 6.658354972989833e-07, "loss": 0.1303, "step": 838 }, { "epoch": 0.89, "learning_rate": 6.535848819263679e-07, "loss": 0.1561, "step": 839 }, { "epoch": 0.89, "learning_rate": 6.414442059730052e-07, "loss": 0.1328, "step": 840 }, { "epoch": 0.89, "learning_rate": 6.294136122464701e-07, "loss": 0.1543, "step": 841 }, { "epoch": 0.89, "learning_rate": 6.17493242259467e-07, "loss": 0.1628, "step": 842 }, { "epoch": 0.89, "learning_rate": 6.056832362281728e-07, "loss": 0.1709, "step": 843 }, { "epoch": 0.89, "learning_rate": 5.939837330705733e-07, "loss": 0.1422, "step": 844 }, { "epoch": 0.89, "learning_rate": 5.823948704048443e-07, "loss": 0.1393, "step": 845 }, { "epoch": 0.9, "learning_rate": 5.709167845477225e-07, "loss": 0.1646, "step": 846 }, { "epoch": 0.9, "learning_rate": 5.595496105129139e-07, "loss": 0.1255, "step": 847 }, { "epoch": 0.9, "learning_rate": 5.482934820094876e-07, "loss": 0.1604, "step": 848 }, { "epoch": 0.9, "learning_rate": 5.371485314403202e-07, "loss": 0.1602, "step": 849 }, { "epoch": 0.9, "learning_rate": 5.261148899005297e-07, "loss": 0.1641, "step": 850 }, { "epoch": 0.9, "learning_rate": 5.15192687175935e-07, "loss": 0.1516, "step": 851 }, { "epoch": 0.9, "learning_rate": 5.043820517415321e-07, "loss": 0.1536, "step": 852 }, { "epoch": 0.9, "learning_rate": 4.936831107599749e-07, "loss": 0.1403, "step": 853 }, { "epoch": 0.9, "learning_rate": 4.830959900800902e-07, "loss": 0.1242, "step": 854 }, { "epoch": 0.9, "learning_rate": 4.7262081423538717e-07, "loss": 0.1349, "step": 855 }, { "epoch": 0.91, "learning_rate": 4.6225770644260595e-07, "loss": 0.1512, "step": 856 }, { "epoch": 0.91, "learning_rate": 4.520067886002488e-07, "loss": 0.135, "step": 857 }, { "epoch": 0.91, "learning_rate": 4.418681812871639e-07, "loss": 0.1371, "step": 858 }, { "epoch": 0.91, "learning_rate": 4.318420037611182e-07, "loss": 0.1512, "step": 859 }, { "epoch": 0.91, "learning_rate": 4.2192837395739874e-07, "loss": 0.1597, "step": 860 }, { "epoch": 0.91, "learning_rate": 4.121274084874194e-07, "loss": 0.1558, "step": 861 }, { "epoch": 0.91, "learning_rate": 4.024392226373563e-07, "loss": 0.1432, "step": 862 }, { "epoch": 0.91, "learning_rate": 3.9286393036678914e-07, "loss": 0.1299, "step": 863 }, { "epoch": 0.91, "learning_rate": 3.834016443073574e-07, "loss": 0.1615, "step": 864 }, { "epoch": 0.92, "learning_rate": 3.7405247576144055e-07, "loss": 0.1161, "step": 865 }, { "epoch": 0.92, "learning_rate": 3.648165347008448e-07, "loss": 0.1371, "step": 866 }, { "epoch": 0.92, "learning_rate": 3.556939297655115e-07, "loss": 0.1303, "step": 867 }, { "epoch": 0.92, "learning_rate": 3.4668476826224205e-07, "loss": 0.1763, "step": 868 }, { "epoch": 0.92, "learning_rate": 3.3778915616342947e-07, "loss": 0.1381, "step": 869 }, { "epoch": 0.92, "learning_rate": 3.2900719810581406e-07, "loss": 0.1483, "step": 870 }, { "epoch": 0.92, "learning_rate": 3.203389973892579e-07, "loss": 0.1479, "step": 871 }, { "epoch": 0.92, "learning_rate": 3.117846559755211e-07, "loss": 0.1419, "step": 872 }, { "epoch": 0.92, "learning_rate": 3.033442744870685e-07, "loss": 0.1454, "step": 873 }, { "epoch": 0.92, "learning_rate": 2.95017952205886e-07, "loss": 0.1531, "step": 874 }, { "epoch": 0.93, "learning_rate": 2.868057870723073e-07, "loss": 0.1406, "step": 875 }, { "epoch": 0.93, "learning_rate": 2.7870787568387016e-07, "loss": 0.1427, "step": 876 }, { "epoch": 0.93, "learning_rate": 2.707243132941717e-07, "loss": 0.1558, "step": 877 }, { "epoch": 0.93, "learning_rate": 2.6285519381175407e-07, "loss": 0.1375, "step": 878 }, { "epoch": 0.93, "learning_rate": 2.551006097989961e-07, "loss": 0.1409, "step": 879 }, { "epoch": 0.93, "learning_rate": 2.474606524710288e-07, "loss": 0.1691, "step": 880 }, { "epoch": 0.93, "learning_rate": 2.399354116946584e-07, "loss": 0.1522, "step": 881 }, { "epoch": 0.93, "learning_rate": 2.3252497598730606e-07, "loss": 0.152, "step": 882 }, { "epoch": 0.93, "learning_rate": 2.2522943251597874e-07, "loss": 0.1555, "step": 883 }, { "epoch": 0.94, "learning_rate": 2.180488670962322e-07, "loss": 0.1644, "step": 884 }, { "epoch": 0.94, "learning_rate": 2.1098336419116628e-07, "loss": 0.1636, "step": 885 }, { "epoch": 0.94, "learning_rate": 2.0403300691042906e-07, "loss": 0.1588, "step": 886 }, { "epoch": 0.94, "learning_rate": 1.9719787700924308e-07, "loss": 0.1331, "step": 887 }, { "epoch": 0.94, "learning_rate": 1.9047805488744188e-07, "loss": 0.1278, "step": 888 }, { "epoch": 0.94, "learning_rate": 1.8387361958852378e-07, "loss": 0.1511, "step": 889 }, { "epoch": 0.94, "learning_rate": 1.7738464879871963e-07, "loss": 0.1613, "step": 890 }, { "epoch": 0.94, "learning_rate": 1.710112188460844e-07, "loss": 0.1159, "step": 891 }, { "epoch": 0.94, "learning_rate": 1.6475340469959689e-07, "loss": 0.1403, "step": 892 }, { "epoch": 0.94, "learning_rate": 1.5861127996827597e-07, "loss": 0.1335, "step": 893 }, { "epoch": 0.95, "learning_rate": 1.5258491690031796e-07, "loss": 0.1329, "step": 894 }, { "epoch": 0.95, "learning_rate": 1.4667438638224064e-07, "loss": 0.1555, "step": 895 }, { "epoch": 0.95, "learning_rate": 1.4087975793806275e-07, "loss": 0.1179, "step": 896 }, { "epoch": 0.95, "learning_rate": 1.3520109972846918e-07, "loss": 0.156, "step": 897 }, { "epoch": 0.95, "learning_rate": 1.2963847855002264e-07, "loss": 0.1517, "step": 898 }, { "epoch": 0.95, "learning_rate": 1.2419195983436881e-07, "loss": 0.1242, "step": 899 }, { "epoch": 0.95, "learning_rate": 1.1886160764747356e-07, "loss": 0.1666, "step": 900 }, { "epoch": 0.95, "learning_rate": 1.1364748468886688e-07, "loss": 0.155, "step": 901 }, { "epoch": 0.95, "learning_rate": 1.0854965229090352e-07, "loss": 0.1435, "step": 902 }, { "epoch": 0.96, "learning_rate": 1.0356817041804246e-07, "loss": 0.1658, "step": 903 }, { "epoch": 0.96, "learning_rate": 9.870309766614184e-08, "loss": 0.1481, "step": 904 }, { "epoch": 0.96, "learning_rate": 9.395449126177291e-08, "loss": 0.1384, "step": 905 }, { "epoch": 0.96, "learning_rate": 8.932240706154171e-08, "loss": 0.1556, "step": 906 }, { "epoch": 0.96, "learning_rate": 8.480689955143395e-08, "loss": 0.1419, "step": 907 }, { "epoch": 0.96, "learning_rate": 8.040802184617446e-08, "loss": 0.1354, "step": 908 }, { "epoch": 0.96, "learning_rate": 7.61258256886055e-08, "loss": 0.1516, "step": 909 }, { "epoch": 0.96, "learning_rate": 7.196036144907159e-08, "loss": 0.1555, "step": 910 }, { "epoch": 0.96, "learning_rate": 6.791167812483013e-08, "loss": 0.1404, "step": 911 }, { "epoch": 0.97, "learning_rate": 6.397982333947617e-08, "loss": 0.1223, "step": 912 }, { "epoch": 0.97, "learning_rate": 6.016484334238515e-08, "loss": 0.1414, "step": 913 }, { "epoch": 0.97, "learning_rate": 5.646678300816e-08, "loss": 0.1628, "step": 914 }, { "epoch": 0.97, "learning_rate": 5.2885685836109316e-08, "loss": 0.1603, "step": 915 }, { "epoch": 0.97, "learning_rate": 4.942159394973778e-08, "loss": 0.1467, "step": 916 }, { "epoch": 0.97, "learning_rate": 4.6074548096244346e-08, "loss": 0.1398, "step": 917 }, { "epoch": 0.97, "learning_rate": 4.284458764604926e-08, "loss": 0.1389, "step": 918 }, { "epoch": 0.97, "learning_rate": 3.973175059232559e-08, "loss": 0.1497, "step": 919 }, { "epoch": 0.97, "learning_rate": 3.6736073550556195e-08, "loss": 0.1467, "step": 920 }, { "epoch": 0.97, "learning_rate": 3.385759175809966e-08, "loss": 0.1453, "step": 921 }, { "epoch": 0.98, "learning_rate": 3.109633907378173e-08, "loss": 0.1378, "step": 922 }, { "epoch": 0.98, "learning_rate": 2.845234797748897e-08, "loss": 0.1332, "step": 923 }, { "epoch": 0.98, "learning_rate": 2.5925649569791266e-08, "loss": 0.1415, "step": 924 }, { "epoch": 0.98, "learning_rate": 2.3516273571577708e-08, "loss": 0.1371, "step": 925 }, { "epoch": 0.98, "learning_rate": 2.1224248323703512e-08, "loss": 0.1329, "step": 926 }, { "epoch": 0.98, "learning_rate": 1.9049600786658073e-08, "loss": 0.1987, "step": 927 }, { "epoch": 0.98, "learning_rate": 1.6992356540247446e-08, "loss": 0.1353, "step": 928 }, { "epoch": 0.98, "learning_rate": 1.5052539783292353e-08, "loss": 0.1281, "step": 929 }, { "epoch": 0.98, "learning_rate": 1.3230173333348417e-08, "loss": 0.1431, "step": 930 }, { "epoch": 0.99, "learning_rate": 1.1525278626431935e-08, "loss": 0.1428, "step": 931 }, { "epoch": 0.99, "learning_rate": 9.937875716770074e-09, "loss": 0.1397, "step": 932 }, { "epoch": 0.99, "learning_rate": 8.467983276563285e-09, "loss": 0.1508, "step": 933 }, { "epoch": 0.99, "learning_rate": 7.115618595772145e-09, "loss": 0.1465, "step": 934 }, { "epoch": 0.99, "learning_rate": 5.880797581904185e-09, "loss": 0.1338, "step": 935 }, { "epoch": 0.99, "learning_rate": 4.763534759835153e-09, "loss": 0.1458, "step": 936 }, { "epoch": 0.99, "learning_rate": 3.763843271631373e-09, "loss": 0.1539, "step": 937 }, { "epoch": 0.99, "learning_rate": 2.881734876403197e-09, "loss": 0.1475, "step": 938 }, { "epoch": 0.99, "learning_rate": 2.1172199501573453e-09, "loss": 0.152, "step": 939 }, { "epoch": 0.99, "learning_rate": 1.4703074856814437e-09, "loss": 0.1341, "step": 940 }, { "epoch": 1.0, "learning_rate": 9.410050924374414e-10, "loss": 0.1522, "step": 941 }, { "epoch": 1.0, "learning_rate": 5.293189964694634e-10, "loss": 0.1389, "step": 942 }, { "epoch": 1.0, "learning_rate": 2.352540403327552e-10, "loss": 0.1511, "step": 943 }, { "epoch": 1.0, "learning_rate": 5.881368303595202e-11, "loss": 0.122, "step": 944 }, { "epoch": 1.0, "learning_rate": 0.0, "loss": 0.1313, "step": 945 }, { "epoch": 1.0, "step": 945, "total_flos": 267616018546688.0, "train_loss": 0.1789411918196098, "train_runtime": 5663.5843, "train_samples_per_second": 32.028, "train_steps_per_second": 0.167 } ], "logging_steps": 1.0, "max_steps": 945, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 800, "total_flos": 267616018546688.0, "train_batch_size": 24, "trial_name": null, "trial_params": null }