diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,54340 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 19.027484143763214,
+  "global_step": 18000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 8.7413,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 7.938,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-05,
+      "loss": 8.3007,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999471458773785e-05,
+      "loss": 7.1444,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998942917547569e-05,
+      "loss": 9.6766,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984143763213534e-05,
+      "loss": 5.474,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997885835095137e-05,
+      "loss": 4.2579,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9973572938689226e-05,
+      "loss": 3.893,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9968287526427065e-05,
+      "loss": 4.3634,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996300211416491e-05,
+      "loss": 4.1838,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995771670190275e-05,
+      "loss": 3.5253,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9952431289640597e-05,
+      "loss": 3.3387,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9947145877378436e-05,
+      "loss": 2.9198,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994186046511628e-05,
+      "loss": 2.8243,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.993657505285412e-05,
+      "loss": 2.5279,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.993128964059197e-05,
+      "loss": 2.3727,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992600422832981e-05,
+      "loss": 2.4728,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992071881606766e-05,
+      "loss": 2.4637,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99154334038055e-05,
+      "loss": 2.4645,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9910147991543345e-05,
+      "loss": 2.4394,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9904862579281184e-05,
+      "loss": 2.5194,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989957716701903e-05,
+      "loss": 2.9867,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989429175475687e-05,
+      "loss": 2.2137,
+      "step": 46
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9889006342494715e-05,
+      "loss": 2.6259,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9883720930232555e-05,
+      "loss": 2.768,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987843551797041e-05,
+      "loss": 1.9139,
+      "step": 52
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.987315010570825e-05,
+      "loss": 1.8541,
+      "step": 54
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.986786469344609e-05,
+      "loss": 1.9566,
+      "step": 56
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.986257928118393e-05,
+      "loss": 1.7323,
+      "step": 58
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.985729386892178e-05,
+      "loss": 1.8061,
+      "step": 60
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9852008456659624e-05,
+      "loss": 1.8183,
+      "step": 62
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9846723044397464e-05,
+      "loss": 1.5695,
+      "step": 64
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.984143763213531e-05,
+      "loss": 1.5084,
+      "step": 66
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.983615221987315e-05,
+      "loss": 1.5239,
+      "step": 68
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9830866807611e-05,
+      "loss": 1.1437,
+      "step": 70
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.982558139534884e-05,
+      "loss": 1.7891,
+      "step": 72
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.982029598308669e-05,
+      "loss": 1.8291,
+      "step": 74
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9815010570824526e-05,
+      "loss": 1.6036,
+      "step": 76
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.980972515856237e-05,
+      "loss": 1.2715,
+      "step": 78
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.980443974630021e-05,
+      "loss": 1.1591,
+      "step": 80
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.979915433403806e-05,
+      "loss": 1.6209,
+      "step": 82
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.97938689217759e-05,
+      "loss": 1.478,
+      "step": 84
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.978858350951374e-05,
+      "loss": 0.986,
+      "step": 86
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.978329809725159e-05,
+      "loss": 1.5824,
+      "step": 88
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9778012684989435e-05,
+      "loss": 1.3657,
+      "step": 90
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9772727272727275e-05,
+      "loss": 1.3425,
+      "step": 92
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.976744186046512e-05,
+      "loss": 1.1693,
+      "step": 94
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.976215644820296e-05,
+      "loss": 1.3152,
+      "step": 96
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9756871035940806e-05,
+      "loss": 0.9633,
+      "step": 98
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9751585623678645e-05,
+      "loss": 1.0875,
+      "step": 100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.974630021141649e-05,
+      "loss": 0.9981,
+      "step": 102
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.974101479915434e-05,
+      "loss": 1.003,
+      "step": 104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9735729386892183e-05,
+      "loss": 1.1275,
+      "step": 106
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.973044397463002e-05,
+      "loss": 1.2824,
+      "step": 108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.972515856236787e-05,
+      "loss": 1.2133,
+      "step": 110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.971987315010571e-05,
+      "loss": 1.1642,
+      "step": 112
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9714587737843554e-05,
+      "loss": 1.0921,
+      "step": 114
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.97093023255814e-05,
+      "loss": 1.3712,
+      "step": 116
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.970401691331924e-05,
+      "loss": 1.1114,
+      "step": 118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9698731501057085e-05,
+      "loss": 1.1718,
+      "step": 120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9693446088794925e-05,
+      "loss": 1.0484,
+      "step": 122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.968816067653278e-05,
+      "loss": 0.8344,
+      "step": 124
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.968287526427062e-05,
+      "loss": 1.4522,
+      "step": 126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.967758985200846e-05,
+      "loss": 2.1993,
+      "step": 128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.96723044397463e-05,
+      "loss": 1.0746,
+      "step": 130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.966701902748415e-05,
+      "loss": 0.9988,
+      "step": 132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.966173361522199e-05,
+      "loss": 1.04,
+      "step": 134
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9656448202959834e-05,
+      "loss": 1.434,
+      "step": 136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.965116279069767e-05,
+      "loss": 1.0161,
+      "step": 138
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.964587737843552e-05,
+      "loss": 1.3142,
+      "step": 140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9640591966173365e-05,
+      "loss": 1.1679,
+      "step": 142
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.963530655391121e-05,
+      "loss": 1.3562,
+      "step": 144
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.963002114164905e-05,
+      "loss": 1.2236,
+      "step": 146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9624735729386896e-05,
+      "loss": 1.2391,
+      "step": 148
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9619450317124736e-05,
+      "loss": 1.1668,
+      "step": 150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.961416490486258e-05,
+      "loss": 1.3492,
+      "step": 152
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.960887949260042e-05,
+      "loss": 0.8301,
+      "step": 154
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.960359408033827e-05,
+      "loss": 1.2563,
+      "step": 156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.959830866807611e-05,
+      "loss": 0.9544,
+      "step": 158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.959302325581396e-05,
+      "loss": 1.0004,
+      "step": 160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.95877378435518e-05,
+      "loss": 1.2484,
+      "step": 162
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9582452431289645e-05,
+      "loss": 1.2366,
+      "step": 164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.957716701902749e-05,
+      "loss": 0.9897,
+      "step": 166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.957188160676533e-05,
+      "loss": 1.0405,
+      "step": 168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9566596194503176e-05,
+      "loss": 0.9499,
+      "step": 170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9561310782241015e-05,
+      "loss": 1.1503,
+      "step": 172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.955602536997886e-05,
+      "loss": 0.9555,
+      "step": 174
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.95507399577167e-05,
+      "loss": 0.9089,
+      "step": 176
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9545454545454553e-05,
+      "loss": 1.3379,
+      "step": 178
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.954016913319239e-05,
+      "loss": 0.8154,
+      "step": 180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.953488372093024e-05,
+      "loss": 0.669,
+      "step": 182
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.952959830866808e-05,
+      "loss": 0.9411,
+      "step": 184
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9524312896405924e-05,
+      "loss": 1.1204,
+      "step": 186
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9519027484143763e-05,
+      "loss": 1.1171,
+      "step": 188
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.951374207188161e-05,
+      "loss": 1.167,
+      "step": 190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.950845665961945e-05,
+      "loss": 0.85,
+      "step": 192
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9503171247357295e-05,
+      "loss": 0.9862,
+      "step": 194
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.949788583509514e-05,
+      "loss": 1.0706,
+      "step": 196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.949260042283299e-05,
+      "loss": 0.9666,
+      "step": 198
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9487315010570826e-05,
+      "loss": 0.9068,
+      "step": 200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.948202959830867e-05,
+      "loss": 0.7309,
+      "step": 202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.947674418604651e-05,
+      "loss": 1.3316,
+      "step": 204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.947145877378436e-05,
+      "loss": 1.0254,
+      "step": 206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.94661733615222e-05,
+      "loss": 0.8078,
+      "step": 208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.946088794926004e-05,
+      "loss": 0.9885,
+      "step": 210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.945560253699789e-05,
+      "loss": 1.3213,
+      "step": 212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9450317124735735e-05,
+      "loss": 1.0121,
+      "step": 214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9445031712473574e-05,
+      "loss": 1.2225,
+      "step": 216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.943974630021142e-05,
+      "loss": 1.207,
+      "step": 218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9434460887949266e-05,
+      "loss": 0.6599,
+      "step": 220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9429175475687106e-05,
+      "loss": 0.9893,
+      "step": 222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.942389006342495e-05,
+      "loss": 0.8,
+      "step": 224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.941860465116279e-05,
+      "loss": 1.1942,
+      "step": 226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.941331923890064e-05,
+      "loss": 1.3698,
+      "step": 228
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9408033826638476e-05,
+      "loss": 1.0476,
+      "step": 230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.940274841437633e-05,
+      "loss": 1.1916,
+      "step": 232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.939746300211417e-05,
+      "loss": 1.2326,
+      "step": 234
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9392177589852015e-05,
+      "loss": 0.9399,
+      "step": 236
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9386892177589854e-05,
+      "loss": 1.2254,
+      "step": 238
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.93816067653277e-05,
+      "loss": 0.9529,
+      "step": 240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.937632135306554e-05,
+      "loss": 1.0299,
+      "step": 242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9371035940803385e-05,
+      "loss": 0.7292,
+      "step": 244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9365750528541225e-05,
+      "loss": 0.9186,
+      "step": 246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.936046511627907e-05,
+      "loss": 0.7604,
+      "step": 248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.935517970401692e-05,
+      "loss": 0.8169,
+      "step": 250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.934989429175476e-05,
+      "loss": 0.973,
+      "step": 252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.93446088794926e-05,
+      "loss": 0.8682,
+      "step": 254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.933932346723045e-05,
+      "loss": 0.9603,
+      "step": 256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.933403805496829e-05,
+      "loss": 1.1217,
+      "step": 258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9328752642706133e-05,
+      "loss": 1.0628,
+      "step": 260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.932346723044397e-05,
+      "loss": 1.0257,
+      "step": 262
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.931818181818182e-05,
+      "loss": 0.8398,
+      "step": 264
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9312896405919665e-05,
+      "loss": 0.9168,
+      "step": 266
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.930761099365751e-05,
+      "loss": 0.8925,
+      "step": 268
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.930232558139535e-05,
+      "loss": 0.7816,
+      "step": 270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9297040169133196e-05,
+      "loss": 1.0841,
+      "step": 272
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.929175475687104e-05,
+      "loss": 0.9992,
+      "step": 274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.928646934460888e-05,
+      "loss": 1.2984,
+      "step": 276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.928118393234673e-05,
+      "loss": 1.2014,
+      "step": 278
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.927589852008457e-05,
+      "loss": 0.8883,
+      "step": 280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.927061310782241e-05,
+      "loss": 1.1356,
+      "step": 282
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.926532769556025e-05,
+      "loss": 0.9082,
+      "step": 284
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9260042283298105e-05,
+      "loss": 0.7397,
+      "step": 286
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9254756871035944e-05,
+      "loss": 1.0028,
+      "step": 288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.924947145877379e-05,
+      "loss": 0.8124,
+      "step": 290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.924418604651163e-05,
+      "loss": 0.9205,
+      "step": 292
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9238900634249476e-05,
+      "loss": 1.2545,
+      "step": 294
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9233615221987315e-05,
+      "loss": 0.9391,
+      "step": 296
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.922832980972516e-05,
+      "loss": 0.9877,
+      "step": 298
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9223044397463e-05,
+      "loss": 0.7637,
+      "step": 300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9217758985200846e-05,
+      "loss": 1.0561,
+      "step": 302
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.921247357293869e-05,
+      "loss": 0.6914,
+      "step": 304
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.920718816067654e-05,
+      "loss": 0.933,
+      "step": 306
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.920190274841438e-05,
+      "loss": 0.916,
+      "step": 308
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9196617336152224e-05,
+      "loss": 0.9402,
+      "step": 310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.919133192389006e-05,
+      "loss": 0.8271,
+      "step": 312
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.918604651162791e-05,
+      "loss": 0.7607,
+      "step": 314
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.918076109936575e-05,
+      "loss": 0.7434,
+      "step": 316
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9175475687103595e-05,
+      "loss": 0.8802,
+      "step": 318
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.917019027484144e-05,
+      "loss": 1.3608,
+      "step": 320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.916490486257929e-05,
+      "loss": 1.0817,
+      "step": 322
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9159619450317126e-05,
+      "loss": 0.9358,
+      "step": 324
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.915433403805497e-05,
+      "loss": 0.6262,
+      "step": 326
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.914904862579282e-05,
+      "loss": 1.1558,
+      "step": 328
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.914376321353066e-05,
+      "loss": 0.9856,
+      "step": 330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9138477801268503e-05,
+      "loss": 0.9328,
+      "step": 332
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.913319238900634e-05,
+      "loss": 0.9733,
+      "step": 334
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.912790697674419e-05,
+      "loss": 0.9509,
+      "step": 336
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.912262156448203e-05,
+      "loss": 1.201,
+      "step": 338
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.911733615221988e-05,
+      "loss": 1.1361,
+      "step": 340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.911205073995772e-05,
+      "loss": 0.8874,
+      "step": 342
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9106765327695566e-05,
+      "loss": 0.7402,
+      "step": 344
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9101479915433406e-05,
+      "loss": 0.957,
+      "step": 346
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.909619450317125e-05,
+      "loss": 0.7774,
+      "step": 348
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.909090909090909e-05,
+      "loss": 0.8865,
+      "step": 350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.908562367864694e-05,
+      "loss": 0.8735,
+      "step": 352
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9080338266384776e-05,
+      "loss": 0.7073,
+      "step": 354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.907505285412262e-05,
+      "loss": 0.6819,
+      "step": 356
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.906976744186046e-05,
+      "loss": 1.3813,
+      "step": 358
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9064482029598314e-05,
+      "loss": 0.8954,
+      "step": 360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9059196617336154e-05,
+      "loss": 0.758,
+      "step": 362
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9053911205074e-05,
+      "loss": 1.0526,
+      "step": 364
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.904862579281184e-05,
+      "loss": 0.7349,
+      "step": 366
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9043340380549685e-05,
+      "loss": 0.8793,
+      "step": 368
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.903805496828753e-05,
+      "loss": 0.5134,
+      "step": 370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.903276955602537e-05,
+      "loss": 0.6381,
+      "step": 372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9027484143763217e-05,
+      "loss": 0.949,
+      "step": 374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9022198731501056e-05,
+      "loss": 0.8732,
+      "step": 376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.90169133192389e-05,
+      "loss": 0.6792,
+      "step": 378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.901162790697675e-05,
+      "loss": 0.8423,
+      "step": 380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9006342494714594e-05,
+      "loss": 0.9052,
+      "step": 382
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.900105708245243e-05,
+      "loss": 0.8757,
+      "step": 384
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.899577167019028e-05,
+      "loss": 1.0981,
+      "step": 386
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.899048625792812e-05,
+      "loss": 1.1139,
+      "step": 388
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8985200845665965e-05,
+      "loss": 0.8369,
+      "step": 390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8979915433403804e-05,
+      "loss": 0.8961,
+      "step": 392
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.897463002114165e-05,
+      "loss": 1.1247,
+      "step": 394
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8969344608879496e-05,
+      "loss": 0.7586,
+      "step": 396
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.896405919661734e-05,
+      "loss": 0.8276,
+      "step": 398
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.895877378435518e-05,
+      "loss": 0.7506,
+      "step": 400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.895348837209303e-05,
+      "loss": 0.6963,
+      "step": 402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.894820295983087e-05,
+      "loss": 0.7005,
+      "step": 404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.894291754756871e-05,
+      "loss": 0.9046,
+      "step": 406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.893763213530655e-05,
+      "loss": 0.8376,
+      "step": 408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.89323467230444e-05,
+      "loss": 0.7533,
+      "step": 410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.892706131078224e-05,
+      "loss": 0.7952,
+      "step": 412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.892177589852009e-05,
+      "loss": 0.8707,
+      "step": 414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.891649048625793e-05,
+      "loss": 0.9241,
+      "step": 416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8911205073995776e-05,
+      "loss": 0.8749,
+      "step": 418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8905919661733615e-05,
+      "loss": 0.8629,
+      "step": 420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.890063424947146e-05,
+      "loss": 1.1447,
+      "step": 422
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.889534883720931e-05,
+      "loss": 0.7138,
+      "step": 424
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.8890063424947146e-05,
+      "loss": 0.6883,
+      "step": 426
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.888477801268499e-05,
+      "loss": 0.7576,
+      "step": 428
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.887949260042283e-05,
+      "loss": 0.6091,
+      "step": 430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.8874207188160684e-05,
+      "loss": 0.7623,
+      "step": 432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.8868921775898524e-05,
+      "loss": 0.7938,
+      "step": 434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.886363636363637e-05,
+      "loss": 0.9892,
+      "step": 436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.885835095137421e-05,
+      "loss": 0.9199,
+      "step": 438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.8853065539112055e-05,
+      "loss": 1.0223,
+      "step": 440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.8847780126849894e-05,
+      "loss": 0.8448,
+      "step": 442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.884249471458774e-05,
+      "loss": 0.7734,
+      "step": 444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.883720930232558e-05,
+      "loss": 1.2691,
+      "step": 446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.8831923890063426e-05,
+      "loss": 0.9567,
+      "step": 448
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.882663847780127e-05,
+      "loss": 0.9065,
+      "step": 450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.882135306553912e-05,
+      "loss": 0.706,
+      "step": 452
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.881606765327696e-05,
+      "loss": 0.8382,
+      "step": 454
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.88107822410148e-05,
+      "loss": 0.8763,
+      "step": 456
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.880549682875264e-05,
+      "loss": 0.6008,
+      "step": 458
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.880021141649049e-05,
+      "loss": 0.9091,
+      "step": 460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.879492600422833e-05,
+      "loss": 1.0063,
+      "step": 462
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8789640591966174e-05,
+      "loss": 0.6134,
+      "step": 464
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.878435517970401e-05,
+      "loss": 0.9105,
+      "step": 466
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8779069767441866e-05,
+      "loss": 0.8555,
+      "step": 468
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8773784355179705e-05,
+      "loss": 1.0469,
+      "step": 470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.876849894291755e-05,
+      "loss": 1.0043,
+      "step": 472
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.876321353065539e-05,
+      "loss": 0.8886,
+      "step": 474
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.875792811839324e-05,
+      "loss": 0.8378,
+      "step": 476
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.875264270613108e-05,
+      "loss": 0.8491,
+      "step": 478
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.874735729386892e-05,
+      "loss": 0.8847,
+      "step": 480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.874207188160677e-05,
+      "loss": 0.8345,
+      "step": 482
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.873678646934461e-05,
+      "loss": 0.882,
+      "step": 484
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.873150105708246e-05,
+      "loss": 1.0436,
+      "step": 486
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.87262156448203e-05,
+      "loss": 0.9272,
+      "step": 488
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8720930232558146e-05,
+      "loss": 0.8267,
+      "step": 490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8715644820295985e-05,
+      "loss": 0.8139,
+      "step": 492
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.871035940803383e-05,
+      "loss": 0.9702,
+      "step": 494
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.870507399577167e-05,
+      "loss": 0.9904,
+      "step": 496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8699788583509516e-05,
+      "loss": 0.8216,
+      "step": 498
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8694503171247356e-05,
+      "loss": 0.8989,
+      "step": 500
+    },
+    {
+      "epoch": 0.53,
+      "eval_cer": 0.08070675406098604,
+      "eval_loss": 0.7687897086143494,
+      "eval_runtime": 128.8472,
+      "eval_samples_per_second": 6.527,
+      "eval_steps_per_second": 0.823,
+      "step": 500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.86892177589852e-05,
+      "loss": 1.0833,
+      "step": 502
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.868393234672305e-05,
+      "loss": 0.8538,
+      "step": 504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8678646934460894e-05,
+      "loss": 0.6459,
+      "step": 506
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.867336152219873e-05,
+      "loss": 0.6752,
+      "step": 508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.866807610993658e-05,
+      "loss": 0.7075,
+      "step": 510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.866279069767442e-05,
+      "loss": 0.7817,
+      "step": 512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8657505285412264e-05,
+      "loss": 0.8261,
+      "step": 514
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.8652219873150104e-05,
+      "loss": 0.8365,
+      "step": 516
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.864693446088795e-05,
+      "loss": 0.7488,
+      "step": 518
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.8641649048625796e-05,
+      "loss": 1.0128,
+      "step": 520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.863636363636364e-05,
+      "loss": 1.1424,
+      "step": 522
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.863107822410148e-05,
+      "loss": 0.8287,
+      "step": 524
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.862579281183933e-05,
+      "loss": 0.718,
+      "step": 526
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8620507399577167e-05,
+      "loss": 0.5674,
+      "step": 528
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.861522198731501e-05,
+      "loss": 0.8469,
+      "step": 530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.860993657505286e-05,
+      "loss": 1.1801,
+      "step": 532
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.86046511627907e-05,
+      "loss": 0.7818,
+      "step": 534
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.8599365750528544e-05,
+      "loss": 0.5959,
+      "step": 536
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.859408033826638e-05,
+      "loss": 0.8464,
+      "step": 538
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.8588794926004236e-05,
+      "loss": 0.711,
+      "step": 540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.8583509513742075e-05,
+      "loss": 0.5881,
+      "step": 542
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.857822410147992e-05,
+      "loss": 0.7966,
+      "step": 544
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.857293868921776e-05,
+      "loss": 0.542,
+      "step": 546
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.856765327695561e-05,
+      "loss": 0.8169,
+      "step": 548
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8562367864693446e-05,
+      "loss": 0.9783,
+      "step": 550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.855708245243129e-05,
+      "loss": 0.7638,
+      "step": 552
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.855179704016913e-05,
+      "loss": 0.8195,
+      "step": 554
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.854651162790698e-05,
+      "loss": 1.0137,
+      "step": 556
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8541226215644824e-05,
+      "loss": 1.0664,
+      "step": 558
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.853594080338267e-05,
+      "loss": 0.7189,
+      "step": 560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.853065539112051e-05,
+      "loss": 0.7912,
+      "step": 562
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8525369978858355e-05,
+      "loss": 0.7657,
+      "step": 564
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8520084566596194e-05,
+      "loss": 0.7469,
+      "step": 566
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.851479915433404e-05,
+      "loss": 0.6355,
+      "step": 568
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.850951374207188e-05,
+      "loss": 0.7281,
+      "step": 570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8504228329809726e-05,
+      "loss": 0.8989,
+      "step": 572
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.849894291754757e-05,
+      "loss": 0.837,
+      "step": 574
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.849365750528542e-05,
+      "loss": 1.067,
+      "step": 576
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.848837209302326e-05,
+      "loss": 0.7942,
+      "step": 578
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.84830866807611e-05,
+      "loss": 0.8782,
+      "step": 580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.847780126849894e-05,
+      "loss": 0.6836,
+      "step": 582
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.847251585623679e-05,
+      "loss": 0.8591,
+      "step": 584
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8467230443974635e-05,
+      "loss": 1.1122,
+      "step": 586
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8461945031712474e-05,
+      "loss": 1.2,
+      "step": 588
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.845665961945032e-05,
+      "loss": 0.6635,
+      "step": 590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.845137420718816e-05,
+      "loss": 0.7363,
+      "step": 592
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.844608879492601e-05,
+      "loss": 0.6953,
+      "step": 594
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.844080338266385e-05,
+      "loss": 0.8619,
+      "step": 596
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.84355179704017e-05,
+      "loss": 0.8771,
+      "step": 598
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.843023255813954e-05,
+      "loss": 0.83,
+      "step": 600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.842494714587738e-05,
+      "loss": 0.6184,
+      "step": 602
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.841966173361522e-05,
+      "loss": 0.6461,
+      "step": 604
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.841437632135307e-05,
+      "loss": 0.9024,
+      "step": 606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.840909090909091e-05,
+      "loss": 0.8576,
+      "step": 608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.840380549682875e-05,
+      "loss": 0.4654,
+      "step": 610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.83985200845666e-05,
+      "loss": 0.6721,
+      "step": 612
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8393234672304445e-05,
+      "loss": 0.7993,
+      "step": 614
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8387949260042285e-05,
+      "loss": 0.6995,
+      "step": 616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.838266384778013e-05,
+      "loss": 0.767,
+      "step": 618
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.837737843551797e-05,
+      "loss": 0.6822,
+      "step": 620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8372093023255816e-05,
+      "loss": 0.7364,
+      "step": 622
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8366807610993655e-05,
+      "loss": 0.8019,
+      "step": 624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.83615221987315e-05,
+      "loss": 0.9001,
+      "step": 626
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.835623678646935e-05,
+      "loss": 1.0015,
+      "step": 628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8350951374207194e-05,
+      "loss": 1.0959,
+      "step": 630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.834566596194503e-05,
+      "loss": 0.7717,
+      "step": 632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.834038054968288e-05,
+      "loss": 0.5083,
+      "step": 634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8335095137420725e-05,
+      "loss": 0.8368,
+      "step": 636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8329809725158564e-05,
+      "loss": 0.5508,
+      "step": 638
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.832452431289641e-05,
+      "loss": 0.7068,
+      "step": 640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.831923890063425e-05,
+      "loss": 0.6711,
+      "step": 642
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8313953488372096e-05,
+      "loss": 0.9935,
+      "step": 644
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8308668076109935e-05,
+      "loss": 0.609,
+      "step": 646
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.830338266384779e-05,
+      "loss": 0.7019,
+      "step": 648
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.829809725158563e-05,
+      "loss": 1.0631,
+      "step": 650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.829281183932347e-05,
+      "loss": 1.0473,
+      "step": 652
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.828752642706131e-05,
+      "loss": 1.1643,
+      "step": 654
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.828224101479916e-05,
+      "loss": 0.7424,
+      "step": 656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8276955602537e-05,
+      "loss": 0.8219,
+      "step": 658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8271670190274844e-05,
+      "loss": 0.8203,
+      "step": 660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.826638477801268e-05,
+      "loss": 0.8933,
+      "step": 662
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.826109936575053e-05,
+      "loss": 0.5901,
+      "step": 664
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8255813953488375e-05,
+      "loss": 0.5379,
+      "step": 666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.825052854122622e-05,
+      "loss": 0.7992,
+      "step": 668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.824524312896406e-05,
+      "loss": 0.8626,
+      "step": 670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.823995771670191e-05,
+      "loss": 0.7633,
+      "step": 672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8234672304439746e-05,
+      "loss": 1.0405,
+      "step": 674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.822938689217759e-05,
+      "loss": 0.6841,
+      "step": 676
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.822410147991543e-05,
+      "loss": 0.7965,
+      "step": 678
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.821881606765328e-05,
+      "loss": 0.7564,
+      "step": 680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8213530655391123e-05,
+      "loss": 0.6262,
+      "step": 682
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.820824524312897e-05,
+      "loss": 0.6553,
+      "step": 684
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.820295983086681e-05,
+      "loss": 0.6584,
+      "step": 686
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8197674418604655e-05,
+      "loss": 0.9452,
+      "step": 688
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.81923890063425e-05,
+      "loss": 0.574,
+      "step": 690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.818710359408034e-05,
+      "loss": 0.7583,
+      "step": 692
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8181818181818186e-05,
+      "loss": 0.7767,
+      "step": 694
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8176532769556026e-05,
+      "loss": 0.8234,
+      "step": 696
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.817124735729387e-05,
+      "loss": 1.0249,
+      "step": 698
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.816596194503171e-05,
+      "loss": 1.0602,
+      "step": 700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8160676532769564e-05,
+      "loss": 0.981,
+      "step": 702
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.81553911205074e-05,
+      "loss": 0.8407,
+      "step": 704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.815010570824525e-05,
+      "loss": 0.7522,
+      "step": 706
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.814482029598309e-05,
+      "loss": 0.7369,
+      "step": 708
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8139534883720934e-05,
+      "loss": 0.6982,
+      "step": 710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8134249471458774e-05,
+      "loss": 0.5641,
+      "step": 712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.812896405919662e-05,
+      "loss": 0.9017,
+      "step": 714
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.812367864693446e-05,
+      "loss": 0.9177,
+      "step": 716
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8118393234672305e-05,
+      "loss": 0.9034,
+      "step": 718
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.811310782241015e-05,
+      "loss": 0.7752,
+      "step": 720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8107822410148e-05,
+      "loss": 0.7535,
+      "step": 722
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8102536997885836e-05,
+      "loss": 0.7685,
+      "step": 724
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.809725158562368e-05,
+      "loss": 0.9903,
+      "step": 726
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.809196617336152e-05,
+      "loss": 0.9556,
+      "step": 728
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.808668076109937e-05,
+      "loss": 0.7408,
+      "step": 730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.808139534883721e-05,
+      "loss": 0.9776,
+      "step": 732
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.807610993657505e-05,
+      "loss": 0.6268,
+      "step": 734
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.80708245243129e-05,
+      "loss": 0.7483,
+      "step": 736
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8065539112050745e-05,
+      "loss": 0.5685,
+      "step": 738
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8060253699788585e-05,
+      "loss": 0.6607,
+      "step": 740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.805496828752643e-05,
+      "loss": 0.6576,
+      "step": 742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.804968287526428e-05,
+      "loss": 0.7757,
+      "step": 744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8044397463002116e-05,
+      "loss": 0.4771,
+      "step": 746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.803911205073996e-05,
+      "loss": 0.7673,
+      "step": 748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.80338266384778e-05,
+      "loss": 0.5479,
+      "step": 750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.802854122621565e-05,
+      "loss": 0.8368,
+      "step": 752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.802325581395349e-05,
+      "loss": 0.7432,
+      "step": 754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.801797040169134e-05,
+      "loss": 0.9759,
+      "step": 756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.801268498942918e-05,
+      "loss": 0.5834,
+      "step": 758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8007399577167025e-05,
+      "loss": 0.6704,
+      "step": 760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8002114164904864e-05,
+      "loss": 0.6211,
+      "step": 762
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.799682875264271e-05,
+      "loss": 0.7186,
+      "step": 764
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.799154334038055e-05,
+      "loss": 0.6583,
+      "step": 766
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7986257928118396e-05,
+      "loss": 0.5375,
+      "step": 768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7980972515856235e-05,
+      "loss": 0.7341,
+      "step": 770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.797568710359408e-05,
+      "loss": 0.5711,
+      "step": 772
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.797040169133193e-05,
+      "loss": 0.5584,
+      "step": 774
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.796511627906977e-05,
+      "loss": 0.5109,
+      "step": 776
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.795983086680761e-05,
+      "loss": 0.5721,
+      "step": 778
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.795454545454546e-05,
+      "loss": 0.6839,
+      "step": 780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.79492600422833e-05,
+      "loss": 0.5016,
+      "step": 782
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7943974630021144e-05,
+      "loss": 1.2874,
+      "step": 784
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.793868921775898e-05,
+      "loss": 0.6109,
+      "step": 786
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.793340380549683e-05,
+      "loss": 0.7601,
+      "step": 788
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7928118393234675e-05,
+      "loss": 0.9023,
+      "step": 790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.792283298097252e-05,
+      "loss": 0.6537,
+      "step": 792
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.791754756871036e-05,
+      "loss": 0.6189,
+      "step": 794
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7912262156448206e-05,
+      "loss": 0.9409,
+      "step": 796
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.790697674418605e-05,
+      "loss": 0.7473,
+      "step": 798
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.790169133192389e-05,
+      "loss": 0.6643,
+      "step": 800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.789640591966174e-05,
+      "loss": 0.6447,
+      "step": 802
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.789112050739958e-05,
+      "loss": 0.5678,
+      "step": 804
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.788583509513742e-05,
+      "loss": 0.5642,
+      "step": 806
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.788054968287526e-05,
+      "loss": 0.7571,
+      "step": 808
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7875264270613115e-05,
+      "loss": 0.7114,
+      "step": 810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7869978858350955e-05,
+      "loss": 0.7462,
+      "step": 812
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.78646934460888e-05,
+      "loss": 0.8739,
+      "step": 814
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.785940803382664e-05,
+      "loss": 0.9708,
+      "step": 816
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7854122621564486e-05,
+      "loss": 0.757,
+      "step": 818
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7848837209302325e-05,
+      "loss": 0.6415,
+      "step": 820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.784355179704017e-05,
+      "loss": 0.5284,
+      "step": 822
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.783826638477801e-05,
+      "loss": 0.9145,
+      "step": 824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.783298097251586e-05,
+      "loss": 0.6378,
+      "step": 826
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.78276955602537e-05,
+      "loss": 0.7644,
+      "step": 828
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.782241014799155e-05,
+      "loss": 0.7083,
+      "step": 830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.781712473572939e-05,
+      "loss": 0.5737,
+      "step": 832
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7811839323467234e-05,
+      "loss": 0.9134,
+      "step": 834
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7806553911205073e-05,
+      "loss": 0.7197,
+      "step": 836
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.780126849894292e-05,
+      "loss": 0.7081,
+      "step": 838
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7795983086680766e-05,
+      "loss": 0.6583,
+      "step": 840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7790697674418605e-05,
+      "loss": 0.5343,
+      "step": 842
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.778541226215645e-05,
+      "loss": 0.4942,
+      "step": 844
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.77801268498943e-05,
+      "loss": 0.6151,
+      "step": 846
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7774841437632136e-05,
+      "loss": 0.6031,
+      "step": 848
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.776955602536998e-05,
+      "loss": 0.8094,
+      "step": 850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.776427061310783e-05,
+      "loss": 0.6032,
+      "step": 852
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.775898520084567e-05,
+      "loss": 0.9748,
+      "step": 854
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7753699788583514e-05,
+      "loss": 0.8298,
+      "step": 856
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.774841437632135e-05,
+      "loss": 0.8524,
+      "step": 858
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.77431289640592e-05,
+      "loss": 0.671,
+      "step": 860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.773784355179704e-05,
+      "loss": 0.6802,
+      "step": 862
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.773255813953489e-05,
+      "loss": 0.8394,
+      "step": 864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.772727272727273e-05,
+      "loss": 0.6247,
+      "step": 866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.7721987315010577e-05,
+      "loss": 0.6984,
+      "step": 868
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.7716701902748416e-05,
+      "loss": 0.732,
+      "step": 870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.771141649048626e-05,
+      "loss": 0.6115,
+      "step": 872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.77061310782241e-05,
+      "loss": 0.7653,
+      "step": 874
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.770084566596195e-05,
+      "loss": 0.6636,
+      "step": 876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.7695560253699787e-05,
+      "loss": 0.9399,
+      "step": 878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.769027484143763e-05,
+      "loss": 0.4535,
+      "step": 880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.768498942917548e-05,
+      "loss": 0.5472,
+      "step": 882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.7679704016913325e-05,
+      "loss": 0.8799,
+      "step": 884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.7674418604651164e-05,
+      "loss": 0.694,
+      "step": 886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.766913319238901e-05,
+      "loss": 0.9077,
+      "step": 888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.766384778012685e-05,
+      "loss": 0.6647,
+      "step": 890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.7658562367864695e-05,
+      "loss": 0.7828,
+      "step": 892
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.765327695560254e-05,
+      "loss": 0.7554,
+      "step": 894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.764799154334038e-05,
+      "loss": 0.683,
+      "step": 896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.764270613107823e-05,
+      "loss": 0.6435,
+      "step": 898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.763742071881607e-05,
+      "loss": 0.8451,
+      "step": 900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.763213530655392e-05,
+      "loss": 0.9021,
+      "step": 902
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.762684989429176e-05,
+      "loss": 0.5666,
+      "step": 904
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7621564482029604e-05,
+      "loss": 0.7458,
+      "step": 906
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7616279069767444e-05,
+      "loss": 0.8112,
+      "step": 908
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.761099365750529e-05,
+      "loss": 0.7296,
+      "step": 910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.760570824524313e-05,
+      "loss": 0.6413,
+      "step": 912
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7600422832980975e-05,
+      "loss": 0.8478,
+      "step": 914
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7595137420718814e-05,
+      "loss": 0.7438,
+      "step": 916
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.758985200845667e-05,
+      "loss": 0.9292,
+      "step": 918
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7584566596194506e-05,
+      "loss": 0.6936,
+      "step": 920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.757928118393235e-05,
+      "loss": 0.6346,
+      "step": 922
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.757399577167019e-05,
+      "loss": 0.6186,
+      "step": 924
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.756871035940804e-05,
+      "loss": 0.6978,
+      "step": 926
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.756342494714588e-05,
+      "loss": 0.7146,
+      "step": 928
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.755813953488372e-05,
+      "loss": 0.655,
+      "step": 930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.755285412262156e-05,
+      "loss": 0.7053,
+      "step": 932
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.754756871035941e-05,
+      "loss": 0.5712,
+      "step": 934
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7542283298097254e-05,
+      "loss": 0.6625,
+      "step": 936
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.75369978858351e-05,
+      "loss": 0.7332,
+      "step": 938
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.753171247357294e-05,
+      "loss": 0.6703,
+      "step": 940
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7526427061310786e-05,
+      "loss": 0.6486,
+      "step": 942
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7521141649048625e-05,
+      "loss": 0.729,
+      "step": 944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.751585623678647e-05,
+      "loss": 0.7503,
+      "step": 946
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.751057082452432e-05,
+      "loss": 0.6571,
+      "step": 948
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7505285412262157e-05,
+      "loss": 0.5435,
+      "step": 950
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.75e-05,
+      "loss": 0.4819,
+      "step": 952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.749471458773785e-05,
+      "loss": 0.6185,
+      "step": 954
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7489429175475695e-05,
+      "loss": 0.8085,
+      "step": 956
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7484143763213534e-05,
+      "loss": 0.681,
+      "step": 958
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.747885835095138e-05,
+      "loss": 1.0466,
+      "step": 960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.747357293868922e-05,
+      "loss": 0.6033,
+      "step": 962
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7468287526427065e-05,
+      "loss": 0.7136,
+      "step": 964
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7463002114164905e-05,
+      "loss": 0.5318,
+      "step": 966
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.745771670190275e-05,
+      "loss": 0.4115,
+      "step": 968
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.745243128964059e-05,
+      "loss": 0.643,
+      "step": 970
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.744714587737844e-05,
+      "loss": 0.5568,
+      "step": 972
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.744186046511628e-05,
+      "loss": 0.8738,
+      "step": 974
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.743657505285413e-05,
+      "loss": 0.9258,
+      "step": 976
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.743128964059197e-05,
+      "loss": 0.583,
+      "step": 978
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.7426004228329814e-05,
+      "loss": 0.5308,
+      "step": 980
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.742071881606765e-05,
+      "loss": 0.9331,
+      "step": 982
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.74154334038055e-05,
+      "loss": 0.6287,
+      "step": 984
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.741014799154334e-05,
+      "loss": 0.4476,
+      "step": 986
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.7404862579281184e-05,
+      "loss": 0.5554,
+      "step": 988
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.739957716701903e-05,
+      "loss": 0.4947,
+      "step": 990
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.7394291754756876e-05,
+      "loss": 0.4979,
+      "step": 992
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.7389006342494716e-05,
+      "loss": 0.6792,
+      "step": 994
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.738372093023256e-05,
+      "loss": 0.8503,
+      "step": 996
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.73784355179704e-05,
+      "loss": 0.7678,
+      "step": 998
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.737315010570825e-05,
+      "loss": 0.6055,
+      "step": 1000
+    },
+    {
+      "epoch": 1.06,
+      "eval_cer": 0.08395554288971217,
+      "eval_loss": 0.5935352444648743,
+      "eval_runtime": 130.5287,
+      "eval_samples_per_second": 6.443,
+      "eval_steps_per_second": 0.812,
+      "step": 1000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.736786469344609e-05,
+      "loss": 0.6435,
+      "step": 1002
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.736257928118393e-05,
+      "loss": 0.4943,
+      "step": 1004
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.735729386892178e-05,
+      "loss": 0.6907,
+      "step": 1006
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.7352008456659625e-05,
+      "loss": 0.8161,
+      "step": 1008
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.734672304439747e-05,
+      "loss": 0.7885,
+      "step": 1010
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.734143763213531e-05,
+      "loss": 0.6831,
+      "step": 1012
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.7336152219873156e-05,
+      "loss": 0.5823,
+      "step": 1014
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.7330866807610995e-05,
+      "loss": 0.6312,
+      "step": 1016
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.732558139534884e-05,
+      "loss": 0.5614,
+      "step": 1018
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.732029598308668e-05,
+      "loss": 0.6951,
+      "step": 1020
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.7315010570824527e-05,
+      "loss": 0.7978,
+      "step": 1022
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.7309725158562366e-05,
+      "loss": 0.4999,
+      "step": 1024
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.730443974630021e-05,
+      "loss": 0.5096,
+      "step": 1026
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.729915433403806e-05,
+      "loss": 0.4208,
+      "step": 1028
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.7293868921775904e-05,
+      "loss": 0.6277,
+      "step": 1030
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.728858350951374e-05,
+      "loss": 0.7658,
+      "step": 1032
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.728329809725159e-05,
+      "loss": 0.7402,
+      "step": 1034
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.727801268498943e-05,
+      "loss": 0.5968,
+      "step": 1036
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.7272727272727275e-05,
+      "loss": 0.5667,
+      "step": 1038
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.7267441860465114e-05,
+      "loss": 0.5122,
+      "step": 1040
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.726215644820296e-05,
+      "loss": 0.5836,
+      "step": 1042
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.7256871035940806e-05,
+      "loss": 0.5438,
+      "step": 1044
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.725158562367865e-05,
+      "loss": 0.5334,
+      "step": 1046
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.724630021141649e-05,
+      "loss": 0.5546,
+      "step": 1048
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.724101479915434e-05,
+      "loss": 0.63,
+      "step": 1050
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.723572938689218e-05,
+      "loss": 0.4592,
+      "step": 1052
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.723044397463002e-05,
+      "loss": 0.3419,
+      "step": 1054
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.722515856236787e-05,
+      "loss": 0.5825,
+      "step": 1056
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.721987315010571e-05,
+      "loss": 0.5554,
+      "step": 1058
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.7214587737843554e-05,
+      "loss": 0.5521,
+      "step": 1060
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.7209302325581394e-05,
+      "loss": 0.6327,
+      "step": 1062
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.7204016913319246e-05,
+      "loss": 0.7493,
+      "step": 1064
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.7198731501057086e-05,
+      "loss": 0.6406,
+      "step": 1066
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.719344608879493e-05,
+      "loss": 0.6063,
+      "step": 1068
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.718816067653277e-05,
+      "loss": 0.7856,
+      "step": 1070
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.718287526427062e-05,
+      "loss": 0.9146,
+      "step": 1072
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.7177589852008456e-05,
+      "loss": 0.9914,
+      "step": 1074
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.71723044397463e-05,
+      "loss": 0.6502,
+      "step": 1076
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.716701902748414e-05,
+      "loss": 0.5753,
+      "step": 1078
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.716173361522199e-05,
+      "loss": 0.6573,
+      "step": 1080
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.7156448202959834e-05,
+      "loss": 0.6994,
+      "step": 1082
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.715116279069768e-05,
+      "loss": 0.4771,
+      "step": 1084
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.714587737843552e-05,
+      "loss": 0.4544,
+      "step": 1086
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.7140591966173365e-05,
+      "loss": 0.5647,
+      "step": 1088
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.7135306553911205e-05,
+      "loss": 0.6566,
+      "step": 1090
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.713002114164905e-05,
+      "loss": 0.5029,
+      "step": 1092
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.712473572938689e-05,
+      "loss": 0.6242,
+      "step": 1094
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.7119450317124736e-05,
+      "loss": 0.5858,
+      "step": 1096
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.711416490486258e-05,
+      "loss": 0.4173,
+      "step": 1098
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.710887949260043e-05,
+      "loss": 0.4745,
+      "step": 1100
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.710359408033827e-05,
+      "loss": 0.6373,
+      "step": 1102
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.7098308668076113e-05,
+      "loss": 0.6338,
+      "step": 1104
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.709302325581396e-05,
+      "loss": 0.502,
+      "step": 1106
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.70877378435518e-05,
+      "loss": 0.5438,
+      "step": 1108
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.7082452431289645e-05,
+      "loss": 0.4672,
+      "step": 1110
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.7077167019027484e-05,
+      "loss": 0.3756,
+      "step": 1112
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.707188160676533e-05,
+      "loss": 0.5381,
+      "step": 1114
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.706659619450317e-05,
+      "loss": 0.4322,
+      "step": 1116
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.706131078224102e-05,
+      "loss": 0.5286,
+      "step": 1118
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.705602536997886e-05,
+      "loss": 0.3846,
+      "step": 1120
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.705073995771671e-05,
+      "loss": 0.6808,
+      "step": 1122
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.704545454545455e-05,
+      "loss": 0.501,
+      "step": 1124
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.704016913319239e-05,
+      "loss": 0.514,
+      "step": 1126
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.703488372093023e-05,
+      "loss": 0.6151,
+      "step": 1128
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.702959830866808e-05,
+      "loss": 0.6041,
+      "step": 1130
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.702431289640592e-05,
+      "loss": 0.6237,
+      "step": 1132
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.7019027484143764e-05,
+      "loss": 0.4988,
+      "step": 1134
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.701374207188161e-05,
+      "loss": 0.5912,
+      "step": 1136
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.7008456659619456e-05,
+      "loss": 0.6301,
+      "step": 1138
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7003171247357295e-05,
+      "loss": 0.5094,
+      "step": 1140
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.699788583509514e-05,
+      "loss": 0.6045,
+      "step": 1142
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.699260042283298e-05,
+      "loss": 0.5534,
+      "step": 1144
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.6987315010570826e-05,
+      "loss": 0.649,
+      "step": 1146
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.6982029598308666e-05,
+      "loss": 0.4673,
+      "step": 1148
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.697674418604651e-05,
+      "loss": 0.6251,
+      "step": 1150
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.697145877378436e-05,
+      "loss": 0.6932,
+      "step": 1152
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.6966173361522204e-05,
+      "loss": 0.7142,
+      "step": 1154
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.696088794926004e-05,
+      "loss": 0.5804,
+      "step": 1156
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.695560253699789e-05,
+      "loss": 0.8723,
+      "step": 1158
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.6950317124735735e-05,
+      "loss": 0.4948,
+      "step": 1160
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.6945031712473575e-05,
+      "loss": 0.5645,
+      "step": 1162
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.693974630021142e-05,
+      "loss": 0.4606,
+      "step": 1164
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.693446088794926e-05,
+      "loss": 0.5359,
+      "step": 1166
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.6929175475687106e-05,
+      "loss": 0.4132,
+      "step": 1168
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.6923890063424945e-05,
+      "loss": 0.6268,
+      "step": 1170
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.69186046511628e-05,
+      "loss": 0.5611,
+      "step": 1172
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.691331923890064e-05,
+      "loss": 0.6392,
+      "step": 1174
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.6908033826638483e-05,
+      "loss": 0.8275,
+      "step": 1176
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.690274841437632e-05,
+      "loss": 0.7045,
+      "step": 1178
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.689746300211417e-05,
+      "loss": 0.7826,
+      "step": 1180
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.689217758985201e-05,
+      "loss": 0.7238,
+      "step": 1182
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.6886892177589854e-05,
+      "loss": 0.4435,
+      "step": 1184
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.6881606765327693e-05,
+      "loss": 0.5589,
+      "step": 1186
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.687632135306554e-05,
+      "loss": 0.5764,
+      "step": 1188
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.6871035940803386e-05,
+      "loss": 0.4692,
+      "step": 1190
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.686575052854123e-05,
+      "loss": 0.4907,
+      "step": 1192
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.686046511627907e-05,
+      "loss": 0.5897,
+      "step": 1194
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.685517970401692e-05,
+      "loss": 0.4332,
+      "step": 1196
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.6849894291754756e-05,
+      "loss": 0.3469,
+      "step": 1198
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.68446088794926e-05,
+      "loss": 0.592,
+      "step": 1200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.683932346723044e-05,
+      "loss": 0.6763,
+      "step": 1202
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.683403805496829e-05,
+      "loss": 0.9583,
+      "step": 1204
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.6828752642706134e-05,
+      "loss": 0.5109,
+      "step": 1206
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.682346723044398e-05,
+      "loss": 0.631,
+      "step": 1208
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.681818181818182e-05,
+      "loss": 0.7334,
+      "step": 1210
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.6812896405919665e-05,
+      "loss": 0.6342,
+      "step": 1212
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.680761099365751e-05,
+      "loss": 0.6196,
+      "step": 1214
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.680232558139535e-05,
+      "loss": 0.5315,
+      "step": 1216
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.6797040169133196e-05,
+      "loss": 0.5387,
+      "step": 1218
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.6791754756871036e-05,
+      "loss": 0.8501,
+      "step": 1220
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.678646934460888e-05,
+      "loss": 0.8125,
+      "step": 1222
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.678118393234672e-05,
+      "loss": 0.6331,
+      "step": 1224
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.6775898520084574e-05,
+      "loss": 0.5622,
+      "step": 1226
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.677061310782241e-05,
+      "loss": 0.5663,
+      "step": 1228
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.676532769556026e-05,
+      "loss": 0.5199,
+      "step": 1230
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.67600422832981e-05,
+      "loss": 0.8869,
+      "step": 1232
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.6754756871035945e-05,
+      "loss": 0.6617,
+      "step": 1234
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.6749471458773784e-05,
+      "loss": 0.4335,
+      "step": 1236
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.674418604651163e-05,
+      "loss": 0.8475,
+      "step": 1238
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.673890063424947e-05,
+      "loss": 0.5508,
+      "step": 1240
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.6733615221987315e-05,
+      "loss": 0.6348,
+      "step": 1242
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.672832980972516e-05,
+      "loss": 0.6159,
+      "step": 1244
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.672304439746301e-05,
+      "loss": 0.477,
+      "step": 1246
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.671775898520085e-05,
+      "loss": 0.5092,
+      "step": 1248
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.671247357293869e-05,
+      "loss": 0.5635,
+      "step": 1250
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.670718816067653e-05,
+      "loss": 0.8466,
+      "step": 1252
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.670190274841438e-05,
+      "loss": 0.5669,
+      "step": 1254
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.6696617336152224e-05,
+      "loss": 0.6561,
+      "step": 1256
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.6691331923890063e-05,
+      "loss": 0.7537,
+      "step": 1258
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.668604651162791e-05,
+      "loss": 0.5566,
+      "step": 1260
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.6680761099365756e-05,
+      "loss": 1.1829,
+      "step": 1262
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.6675475687103595e-05,
+      "loss": 0.8556,
+      "step": 1264
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.667019027484144e-05,
+      "loss": 0.6861,
+      "step": 1266
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.666490486257929e-05,
+      "loss": 0.634,
+      "step": 1268
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.6659619450317126e-05,
+      "loss": 0.6609,
+      "step": 1270
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.665433403805497e-05,
+      "loss": 0.6384,
+      "step": 1272
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.664904862579281e-05,
+      "loss": 0.4699,
+      "step": 1274
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.664376321353066e-05,
+      "loss": 0.6589,
+      "step": 1276
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.66384778012685e-05,
+      "loss": 0.6271,
+      "step": 1278
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.663319238900635e-05,
+      "loss": 0.7925,
+      "step": 1280
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.662790697674419e-05,
+      "loss": 0.7247,
+      "step": 1282
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.6622621564482035e-05,
+      "loss": 0.5185,
+      "step": 1284
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.6617336152219874e-05,
+      "loss": 0.5875,
+      "step": 1286
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.661205073995772e-05,
+      "loss": 0.6124,
+      "step": 1288
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.660676532769556e-05,
+      "loss": 0.7029,
+      "step": 1290
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6601479915433406e-05,
+      "loss": 0.5547,
+      "step": 1292
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6596194503171245e-05,
+      "loss": 0.8943,
+      "step": 1294
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.659090909090909e-05,
+      "loss": 0.5128,
+      "step": 1296
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.658562367864694e-05,
+      "loss": 0.5178,
+      "step": 1298
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.658033826638478e-05,
+      "loss": 0.4883,
+      "step": 1300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.657505285412262e-05,
+      "loss": 0.7211,
+      "step": 1302
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.656976744186047e-05,
+      "loss": 0.4854,
+      "step": 1304
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.656448202959831e-05,
+      "loss": 0.5831,
+      "step": 1306
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6559196617336154e-05,
+      "loss": 0.4782,
+      "step": 1308
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6553911205074e-05,
+      "loss": 0.6382,
+      "step": 1310
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.654862579281184e-05,
+      "loss": 0.3921,
+      "step": 1312
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.6543340380549685e-05,
+      "loss": 0.4697,
+      "step": 1314
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.653805496828753e-05,
+      "loss": 0.8628,
+      "step": 1316
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.653276955602537e-05,
+      "loss": 0.6685,
+      "step": 1318
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.652748414376322e-05,
+      "loss": 0.7783,
+      "step": 1320
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.652219873150106e-05,
+      "loss": 0.4967,
+      "step": 1322
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.65169133192389e-05,
+      "loss": 0.7527,
+      "step": 1324
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.651162790697675e-05,
+      "loss": 0.9232,
+      "step": 1326
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.650634249471459e-05,
+      "loss": 0.7172,
+      "step": 1328
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.6501057082452433e-05,
+      "loss": 0.7406,
+      "step": 1330
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.649577167019027e-05,
+      "loss": 0.5102,
+      "step": 1332
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.6490486257928126e-05,
+      "loss": 0.7775,
+      "step": 1334
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.6485200845665965e-05,
+      "loss": 0.6671,
+      "step": 1336
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.647991543340381e-05,
+      "loss": 0.726,
+      "step": 1338
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.647463002114165e-05,
+      "loss": 0.7379,
+      "step": 1340
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.6469344608879496e-05,
+      "loss": 0.5184,
+      "step": 1342
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.6464059196617336e-05,
+      "loss": 0.774,
+      "step": 1344
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.645877378435518e-05,
+      "loss": 0.7082,
+      "step": 1346
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.645348837209302e-05,
+      "loss": 0.655,
+      "step": 1348
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.644820295983087e-05,
+      "loss": 0.5688,
+      "step": 1350
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.644291754756871e-05,
+      "loss": 0.5803,
+      "step": 1352
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.643763213530656e-05,
+      "loss": 0.5136,
+      "step": 1354
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.64323467230444e-05,
+      "loss": 0.5327,
+      "step": 1356
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.6427061310782244e-05,
+      "loss": 0.3571,
+      "step": 1358
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.6421775898520084e-05,
+      "loss": 0.61,
+      "step": 1360
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.641649048625793e-05,
+      "loss": 0.6218,
+      "step": 1362
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.6411205073995776e-05,
+      "loss": 0.5032,
+      "step": 1364
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.6405919661733615e-05,
+      "loss": 0.6186,
+      "step": 1366
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.640063424947146e-05,
+      "loss": 0.5374,
+      "step": 1368
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.639534883720931e-05,
+      "loss": 0.5657,
+      "step": 1370
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.639006342494715e-05,
+      "loss": 0.6641,
+      "step": 1372
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.638477801268499e-05,
+      "loss": 0.6886,
+      "step": 1374
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.637949260042284e-05,
+      "loss": 0.7777,
+      "step": 1376
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.637420718816068e-05,
+      "loss": 0.5609,
+      "step": 1378
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.6368921775898524e-05,
+      "loss": 0.6644,
+      "step": 1380
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.636363636363636e-05,
+      "loss": 0.6055,
+      "step": 1382
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.635835095137421e-05,
+      "loss": 0.579,
+      "step": 1384
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.635306553911205e-05,
+      "loss": 0.6299,
+      "step": 1386
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.63477801268499e-05,
+      "loss": 0.3263,
+      "step": 1388
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.634249471458774e-05,
+      "loss": 0.4689,
+      "step": 1390
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.633720930232559e-05,
+      "loss": 0.596,
+      "step": 1392
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.6331923890063426e-05,
+      "loss": 0.6729,
+      "step": 1394
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.632663847780127e-05,
+      "loss": 0.6163,
+      "step": 1396
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.632135306553911e-05,
+      "loss": 0.5728,
+      "step": 1398
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.631606765327696e-05,
+      "loss": 0.5581,
+      "step": 1400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.63107822410148e-05,
+      "loss": 0.5471,
+      "step": 1402
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.630549682875264e-05,
+      "loss": 0.6324,
+      "step": 1404
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.630021141649049e-05,
+      "loss": 0.4537,
+      "step": 1406
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.6294926004228335e-05,
+      "loss": 0.5605,
+      "step": 1408
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.6289640591966174e-05,
+      "loss": 0.3762,
+      "step": 1410
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.628435517970402e-05,
+      "loss": 0.4534,
+      "step": 1412
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.627906976744186e-05,
+      "loss": 0.4939,
+      "step": 1414
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6273784355179706e-05,
+      "loss": 0.5225,
+      "step": 1416
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.626849894291755e-05,
+      "loss": 0.3952,
+      "step": 1418
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.626321353065539e-05,
+      "loss": 0.4838,
+      "step": 1420
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.625792811839324e-05,
+      "loss": 0.5493,
+      "step": 1422
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.625264270613108e-05,
+      "loss": 0.7322,
+      "step": 1424
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.624735729386893e-05,
+      "loss": 0.6912,
+      "step": 1426
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.624207188160677e-05,
+      "loss": 0.5979,
+      "step": 1428
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.6236786469344614e-05,
+      "loss": 0.5229,
+      "step": 1430
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.6231501057082454e-05,
+      "loss": 0.5489,
+      "step": 1432
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.62262156448203e-05,
+      "loss": 0.6757,
+      "step": 1434
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.622093023255814e-05,
+      "loss": 0.8744,
+      "step": 1436
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.6215644820295985e-05,
+      "loss": 0.6652,
+      "step": 1438
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.6210359408033824e-05,
+      "loss": 0.7747,
+      "step": 1440
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.620507399577168e-05,
+      "loss": 0.8047,
+      "step": 1442
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.6199788583509517e-05,
+      "loss": 0.6374,
+      "step": 1444
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.619450317124736e-05,
+      "loss": 0.8522,
+      "step": 1446
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.61892177589852e-05,
+      "loss": 1.1201,
+      "step": 1448
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.618393234672305e-05,
+      "loss": 0.7091,
+      "step": 1450
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.617864693446089e-05,
+      "loss": 0.6903,
+      "step": 1452
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.617336152219873e-05,
+      "loss": 0.7191,
+      "step": 1454
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.616807610993657e-05,
+      "loss": 0.6447,
+      "step": 1456
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.616279069767442e-05,
+      "loss": 0.5796,
+      "step": 1458
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.6157505285412265e-05,
+      "loss": 0.4474,
+      "step": 1460
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.615221987315011e-05,
+      "loss": 0.6297,
+      "step": 1462
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.614693446088795e-05,
+      "loss": 0.7815,
+      "step": 1464
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.6141649048625796e-05,
+      "loss": 0.4299,
+      "step": 1466
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.6136363636363635e-05,
+      "loss": 0.6843,
+      "step": 1468
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.613107822410148e-05,
+      "loss": 0.3767,
+      "step": 1470
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.612579281183933e-05,
+      "loss": 0.6761,
+      "step": 1472
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.612050739957717e-05,
+      "loss": 0.9297,
+      "step": 1474
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.611522198731501e-05,
+      "loss": 0.8386,
+      "step": 1476
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.610993657505286e-05,
+      "loss": 0.8722,
+      "step": 1478
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.6104651162790705e-05,
+      "loss": 0.8477,
+      "step": 1480
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.6099365750528544e-05,
+      "loss": 0.5865,
+      "step": 1482
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.609408033826639e-05,
+      "loss": 0.7178,
+      "step": 1484
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.608879492600423e-05,
+      "loss": 0.5507,
+      "step": 1486
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.6083509513742076e-05,
+      "loss": 0.5724,
+      "step": 1488
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.6078224101479915e-05,
+      "loss": 0.5513,
+      "step": 1490
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.607293868921776e-05,
+      "loss": 0.7183,
+      "step": 1492
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.60676532769556e-05,
+      "loss": 0.4628,
+      "step": 1494
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.606236786469345e-05,
+      "loss": 0.4151,
+      "step": 1496
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.605708245243129e-05,
+      "loss": 0.665,
+      "step": 1498
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.605179704016914e-05,
+      "loss": 0.4722,
+      "step": 1500
+    },
+    {
+      "epoch": 1.59,
+      "eval_cer": 0.06184098033627814,
+      "eval_loss": 0.5636065006256104,
+      "eval_runtime": 127.4465,
+      "eval_samples_per_second": 6.599,
+      "eval_steps_per_second": 0.832,
+      "step": 1500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.604651162790698e-05,
+      "loss": 0.4211,
+      "step": 1502
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.6041226215644824e-05,
+      "loss": 0.7062,
+      "step": 1504
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.603594080338266e-05,
+      "loss": 0.5417,
+      "step": 1506
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.603065539112051e-05,
+      "loss": 0.6403,
+      "step": 1508
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.602536997885835e-05,
+      "loss": 0.6087,
+      "step": 1510
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.6020084566596194e-05,
+      "loss": 0.5924,
+      "step": 1512
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.601479915433404e-05,
+      "loss": 0.457,
+      "step": 1514
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.600951374207189e-05,
+      "loss": 0.5073,
+      "step": 1516
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.6004228329809726e-05,
+      "loss": 0.4704,
+      "step": 1518
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.599894291754757e-05,
+      "loss": 0.6511,
+      "step": 1520
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.599365750528541e-05,
+      "loss": 0.5483,
+      "step": 1522
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.598837209302326e-05,
+      "loss": 0.5997,
+      "step": 1524
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.59830866807611e-05,
+      "loss": 0.4576,
+      "step": 1526
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.597780126849894e-05,
+      "loss": 0.6991,
+      "step": 1528
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.597251585623679e-05,
+      "loss": 0.7717,
+      "step": 1530
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.5967230443974635e-05,
+      "loss": 0.6123,
+      "step": 1532
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.596194503171248e-05,
+      "loss": 0.5343,
+      "step": 1534
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.595665961945032e-05,
+      "loss": 0.4572,
+      "step": 1536
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.5951374207188166e-05,
+      "loss": 0.8137,
+      "step": 1538
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.5946088794926005e-05,
+      "loss": 0.4088,
+      "step": 1540
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.594080338266385e-05,
+      "loss": 0.5861,
+      "step": 1542
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.593551797040169e-05,
+      "loss": 0.5733,
+      "step": 1544
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.593023255813954e-05,
+      "loss": 0.6435,
+      "step": 1546
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.5924947145877376e-05,
+      "loss": 0.4275,
+      "step": 1548
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.591966173361523e-05,
+      "loss": 0.4566,
+      "step": 1550
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.591437632135307e-05,
+      "loss": 0.5268,
+      "step": 1552
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.5909090909090914e-05,
+      "loss": 0.4623,
+      "step": 1554
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.5903805496828754e-05,
+      "loss": 0.5294,
+      "step": 1556
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.58985200845666e-05,
+      "loss": 0.3979,
+      "step": 1558
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.589323467230444e-05,
+      "loss": 0.5877,
+      "step": 1560
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.5887949260042285e-05,
+      "loss": 0.6169,
+      "step": 1562
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.5882663847780124e-05,
+      "loss": 0.6217,
+      "step": 1564
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.587737843551797e-05,
+      "loss": 0.5586,
+      "step": 1566
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.5872093023255816e-05,
+      "loss": 0.8468,
+      "step": 1568
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.586680761099366e-05,
+      "loss": 0.673,
+      "step": 1570
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.58615221987315e-05,
+      "loss": 0.6664,
+      "step": 1572
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.585623678646935e-05,
+      "loss": 0.644,
+      "step": 1574
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.5850951374207194e-05,
+      "loss": 0.7479,
+      "step": 1576
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.584566596194503e-05,
+      "loss": 0.6967,
+      "step": 1578
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.584038054968288e-05,
+      "loss": 0.4362,
+      "step": 1580
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.583509513742072e-05,
+      "loss": 0.6123,
+      "step": 1582
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.5829809725158565e-05,
+      "loss": 0.6258,
+      "step": 1584
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.582452431289641e-05,
+      "loss": 0.6673,
+      "step": 1586
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.581923890063426e-05,
+      "loss": 0.7197,
+      "step": 1588
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.5813953488372096e-05,
+      "loss": 0.6184,
+      "step": 1590
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.580866807610994e-05,
+      "loss": 0.4795,
+      "step": 1592
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.580338266384778e-05,
+      "loss": 0.5485,
+      "step": 1594
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.579809725158563e-05,
+      "loss": 0.6806,
+      "step": 1596
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.579281183932347e-05,
+      "loss": 0.735,
+      "step": 1598
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.578752642706131e-05,
+      "loss": 0.6628,
+      "step": 1600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.578224101479915e-05,
+      "loss": 0.7737,
+      "step": 1602
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.5776955602537005e-05,
+      "loss": 0.7288,
+      "step": 1604
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.5771670190274844e-05,
+      "loss": 0.4968,
+      "step": 1606
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.576638477801269e-05,
+      "loss": 0.4477,
+      "step": 1608
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.576109936575053e-05,
+      "loss": 0.5333,
+      "step": 1610
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.5755813953488375e-05,
+      "loss": 0.4593,
+      "step": 1612
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.5750528541226215e-05,
+      "loss": 0.516,
+      "step": 1614
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.574524312896406e-05,
+      "loss": 0.5752,
+      "step": 1616
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.57399577167019e-05,
+      "loss": 0.3994,
+      "step": 1618
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.5734672304439746e-05,
+      "loss": 0.4733,
+      "step": 1620
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.572938689217759e-05,
+      "loss": 0.5311,
+      "step": 1622
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.572410147991544e-05,
+      "loss": 0.7522,
+      "step": 1624
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.571881606765328e-05,
+      "loss": 0.5047,
+      "step": 1626
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.5713530655391124e-05,
+      "loss": 0.6271,
+      "step": 1628
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.570824524312897e-05,
+      "loss": 0.5289,
+      "step": 1630
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.570295983086681e-05,
+      "loss": 0.5299,
+      "step": 1632
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.5697674418604655e-05,
+      "loss": 0.6072,
+      "step": 1634
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.5692389006342494e-05,
+      "loss": 0.5663,
+      "step": 1636
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.568710359408034e-05,
+      "loss": 0.366,
+      "step": 1638
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.5681818181818186e-05,
+      "loss": 0.5508,
+      "step": 1640
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.567653276955603e-05,
+      "loss": 0.3564,
+      "step": 1642
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.567124735729387e-05,
+      "loss": 0.8547,
+      "step": 1644
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.566596194503172e-05,
+      "loss": 0.7649,
+      "step": 1646
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.566067653276956e-05,
+      "loss": 0.866,
+      "step": 1648
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.56553911205074e-05,
+      "loss": 0.3577,
+      "step": 1650
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.565010570824524e-05,
+      "loss": 0.4708,
+      "step": 1652
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.564482029598309e-05,
+      "loss": 0.5484,
+      "step": 1654
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.563953488372093e-05,
+      "loss": 0.5227,
+      "step": 1656
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.563424947145878e-05,
+      "loss": 0.4581,
+      "step": 1658
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.562896405919662e-05,
+      "loss": 0.4865,
+      "step": 1660
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.5623678646934466e-05,
+      "loss": 0.4035,
+      "step": 1662
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.5618393234672305e-05,
+      "loss": 0.445,
+      "step": 1664
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.561310782241015e-05,
+      "loss": 0.3442,
+      "step": 1666
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.560782241014799e-05,
+      "loss": 0.5735,
+      "step": 1668
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.560253699788584e-05,
+      "loss": 0.4627,
+      "step": 1670
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.5597251585623676e-05,
+      "loss": 0.7131,
+      "step": 1672
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.559196617336152e-05,
+      "loss": 0.4328,
+      "step": 1674
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.558668076109937e-05,
+      "loss": 0.5691,
+      "step": 1676
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.5581395348837214e-05,
+      "loss": 0.6523,
+      "step": 1678
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.5576109936575053e-05,
+      "loss": 0.6022,
+      "step": 1680
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.55708245243129e-05,
+      "loss": 0.7375,
+      "step": 1682
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.5565539112050746e-05,
+      "loss": 0.6585,
+      "step": 1684
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.5560253699788585e-05,
+      "loss": 0.8858,
+      "step": 1686
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.555496828752643e-05,
+      "loss": 0.67,
+      "step": 1688
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.554968287526427e-05,
+      "loss": 0.7282,
+      "step": 1690
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.5544397463002116e-05,
+      "loss": 0.6527,
+      "step": 1692
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.553911205073996e-05,
+      "loss": 0.4792,
+      "step": 1694
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.553382663847781e-05,
+      "loss": 0.745,
+      "step": 1696
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.552854122621565e-05,
+      "loss": 0.6185,
+      "step": 1698
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.5523255813953494e-05,
+      "loss": 0.446,
+      "step": 1700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.551797040169133e-05,
+      "loss": 0.5001,
+      "step": 1702
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.551268498942918e-05,
+      "loss": 0.375,
+      "step": 1704
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.550739957716702e-05,
+      "loss": 0.3736,
+      "step": 1706
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5502114164904864e-05,
+      "loss": 0.5857,
+      "step": 1708
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5496828752642704e-05,
+      "loss": 0.4895,
+      "step": 1710
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.549154334038055e-05,
+      "loss": 0.6071,
+      "step": 1712
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5486257928118396e-05,
+      "loss": 0.5467,
+      "step": 1714
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.548097251585624e-05,
+      "loss": 0.511,
+      "step": 1716
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.547568710359408e-05,
+      "loss": 0.8558,
+      "step": 1718
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.547040169133193e-05,
+      "loss": 0.8073,
+      "step": 1720
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.5465116279069766e-05,
+      "loss": 0.6995,
+      "step": 1722
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.545983086680761e-05,
+      "loss": 0.798,
+      "step": 1724
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 0.4806,
+      "step": 1726
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.54492600422833e-05,
+      "loss": 0.4374,
+      "step": 1728
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.5443974630021144e-05,
+      "loss": 0.7523,
+      "step": 1730
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.543868921775899e-05,
+      "loss": 0.5914,
+      "step": 1732
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.543340380549683e-05,
+      "loss": 0.4302,
+      "step": 1734
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.5428118393234675e-05,
+      "loss": 0.7145,
+      "step": 1736
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.542283298097252e-05,
+      "loss": 0.3885,
+      "step": 1738
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.541754756871036e-05,
+      "loss": 0.5144,
+      "step": 1740
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.541226215644821e-05,
+      "loss": 0.6347,
+      "step": 1742
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.5406976744186046e-05,
+      "loss": 0.6682,
+      "step": 1744
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.540169133192389e-05,
+      "loss": 0.422,
+      "step": 1746
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.539640591966173e-05,
+      "loss": 0.417,
+      "step": 1748
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.5391120507399584e-05,
+      "loss": 0.4943,
+      "step": 1750
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.5385835095137423e-05,
+      "loss": 0.6295,
+      "step": 1752
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.538054968287527e-05,
+      "loss": 0.6961,
+      "step": 1754
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.537526427061311e-05,
+      "loss": 0.6064,
+      "step": 1756
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.5369978858350955e-05,
+      "loss": 0.5825,
+      "step": 1758
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.5364693446088794e-05,
+      "loss": 0.5187,
+      "step": 1760
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.535940803382664e-05,
+      "loss": 0.7391,
+      "step": 1762
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.535412262156448e-05,
+      "loss": 0.7237,
+      "step": 1764
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.5348837209302326e-05,
+      "loss": 0.6358,
+      "step": 1766
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.534355179704017e-05,
+      "loss": 0.4581,
+      "step": 1768
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.533826638477802e-05,
+      "loss": 0.3613,
+      "step": 1770
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.533298097251586e-05,
+      "loss": 0.7856,
+      "step": 1772
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.53276955602537e-05,
+      "loss": 0.4416,
+      "step": 1774
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.532241014799154e-05,
+      "loss": 0.5197,
+      "step": 1776
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.531712473572939e-05,
+      "loss": 0.536,
+      "step": 1778
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.5311839323467234e-05,
+      "loss": 0.5705,
+      "step": 1780
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.5306553911205074e-05,
+      "loss": 0.4337,
+      "step": 1782
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.530126849894292e-05,
+      "loss": 0.4389,
+      "step": 1784
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.5295983086680766e-05,
+      "loss": 0.5438,
+      "step": 1786
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.5290697674418605e-05,
+      "loss": 0.4809,
+      "step": 1788
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.528541226215645e-05,
+      "loss": 0.4446,
+      "step": 1790
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.52801268498943e-05,
+      "loss": 0.7674,
+      "step": 1792
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.5274841437632136e-05,
+      "loss": 0.4956,
+      "step": 1794
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.526955602536998e-05,
+      "loss": 0.5586,
+      "step": 1796
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.526427061310782e-05,
+      "loss": 0.5171,
+      "step": 1798
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.525898520084567e-05,
+      "loss": 0.6569,
+      "step": 1800
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.525369978858351e-05,
+      "loss": 0.4957,
+      "step": 1802
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.524841437632136e-05,
+      "loss": 0.4242,
+      "step": 1804
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.52431289640592e-05,
+      "loss": 0.422,
+      "step": 1806
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.5237843551797045e-05,
+      "loss": 0.4738,
+      "step": 1808
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.5232558139534885e-05,
+      "loss": 0.4787,
+      "step": 1810
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.522727272727273e-05,
+      "loss": 0.4131,
+      "step": 1812
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.522198731501057e-05,
+      "loss": 0.5565,
+      "step": 1814
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.5216701902748416e-05,
+      "loss": 0.401,
+      "step": 1816
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.5211416490486255e-05,
+      "loss": 0.4694,
+      "step": 1818
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.52061310782241e-05,
+      "loss": 0.4531,
+      "step": 1820
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.520084566596195e-05,
+      "loss": 0.5779,
+      "step": 1822
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.5195560253699794e-05,
+      "loss": 0.5072,
+      "step": 1824
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.519027484143763e-05,
+      "loss": 0.7621,
+      "step": 1826
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.518498942917548e-05,
+      "loss": 0.45,
+      "step": 1828
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.517970401691332e-05,
+      "loss": 0.663,
+      "step": 1830
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.5174418604651164e-05,
+      "loss": 0.4742,
+      "step": 1832
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.516913319238901e-05,
+      "loss": 0.7337,
+      "step": 1834
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.516384778012685e-05,
+      "loss": 0.3277,
+      "step": 1836
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.5158562367864696e-05,
+      "loss": 0.4131,
+      "step": 1838
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.515327695560254e-05,
+      "loss": 0.5652,
+      "step": 1840
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.514799154334039e-05,
+      "loss": 0.6251,
+      "step": 1842
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.514270613107823e-05,
+      "loss": 0.428,
+      "step": 1844
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.513742071881607e-05,
+      "loss": 0.3634,
+      "step": 1846
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.513213530655391e-05,
+      "loss": 0.4997,
+      "step": 1848
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.512684989429176e-05,
+      "loss": 0.4841,
+      "step": 1850
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.51215644820296e-05,
+      "loss": 0.3301,
+      "step": 1852
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.5116279069767444e-05,
+      "loss": 0.4506,
+      "step": 1854
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.511099365750528e-05,
+      "loss": 0.4708,
+      "step": 1856
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.5105708245243136e-05,
+      "loss": 0.613,
+      "step": 1858
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.5100422832980975e-05,
+      "loss": 0.534,
+      "step": 1860
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.509513742071882e-05,
+      "loss": 0.6985,
+      "step": 1862
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.508985200845666e-05,
+      "loss": 0.6257,
+      "step": 1864
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.5084566596194507e-05,
+      "loss": 0.61,
+      "step": 1866
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.5079281183932346e-05,
+      "loss": 0.4835,
+      "step": 1868
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.507399577167019e-05,
+      "loss": 0.6178,
+      "step": 1870
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.506871035940803e-05,
+      "loss": 0.4642,
+      "step": 1872
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.506342494714588e-05,
+      "loss": 0.616,
+      "step": 1874
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.505813953488372e-05,
+      "loss": 0.6076,
+      "step": 1876
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.505285412262157e-05,
+      "loss": 0.5103,
+      "step": 1878
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.504756871035941e-05,
+      "loss": 0.5121,
+      "step": 1880
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.5042283298097255e-05,
+      "loss": 0.3769,
+      "step": 1882
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.5036997885835094e-05,
+      "loss": 0.3962,
+      "step": 1884
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.503171247357294e-05,
+      "loss": 0.3739,
+      "step": 1886
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.5026427061310786e-05,
+      "loss": 0.3313,
+      "step": 1888
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.5021141649048625e-05,
+      "loss": 0.463,
+      "step": 1890
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.501585623678647e-05,
+      "loss": 0.5002,
+      "step": 1892
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.501057082452432e-05,
+      "loss": 0.3315,
+      "step": 1894
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.5005285412262164e-05,
+      "loss": 0.339,
+      "step": 1896
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.5e-05,
+      "loss": 0.468,
+      "step": 1898
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.499471458773785e-05,
+      "loss": 0.4289,
+      "step": 1900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.498942917547569e-05,
+      "loss": 0.4292,
+      "step": 1902
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.4984143763213534e-05,
+      "loss": 0.609,
+      "step": 1904
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.4978858350951374e-05,
+      "loss": 0.4683,
+      "step": 1906
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.497357293868922e-05,
+      "loss": 0.4534,
+      "step": 1908
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.496828752642706e-05,
+      "loss": 0.5481,
+      "step": 1910
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.496300211416491e-05,
+      "loss": 0.4727,
+      "step": 1912
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.495771670190275e-05,
+      "loss": 0.4213,
+      "step": 1914
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.49524312896406e-05,
+      "loss": 0.4907,
+      "step": 1916
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.4947145877378436e-05,
+      "loss": 0.5339,
+      "step": 1918
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.494186046511628e-05,
+      "loss": 0.4394,
+      "step": 1920
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.493657505285412e-05,
+      "loss": 0.5539,
+      "step": 1922
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.493128964059197e-05,
+      "loss": 0.4241,
+      "step": 1924
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.492600422832981e-05,
+      "loss": 0.4557,
+      "step": 1926
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.492071881606765e-05,
+      "loss": 0.4038,
+      "step": 1928
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.49154334038055e-05,
+      "loss": 0.4505,
+      "step": 1930
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.4910147991543345e-05,
+      "loss": 0.4646,
+      "step": 1932
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.4904862579281184e-05,
+      "loss": 0.401,
+      "step": 1934
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.489957716701903e-05,
+      "loss": 0.3632,
+      "step": 1936
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.489429175475687e-05,
+      "loss": 0.3835,
+      "step": 1938
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.4889006342494716e-05,
+      "loss": 0.5795,
+      "step": 1940
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.488372093023256e-05,
+      "loss": 0.4528,
+      "step": 1942
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.48784355179704e-05,
+      "loss": 0.4788,
+      "step": 1944
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.487315010570825e-05,
+      "loss": 0.3118,
+      "step": 1946
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.486786469344609e-05,
+      "loss": 0.548,
+      "step": 1948
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.486257928118394e-05,
+      "loss": 0.5891,
+      "step": 1950
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.485729386892178e-05,
+      "loss": 0.5405,
+      "step": 1952
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.4852008456659625e-05,
+      "loss": 0.3774,
+      "step": 1954
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.4846723044397464e-05,
+      "loss": 0.6053,
+      "step": 1956
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.484143763213531e-05,
+      "loss": 0.557,
+      "step": 1958
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.483615221987315e-05,
+      "loss": 0.3202,
+      "step": 1960
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.4830866807610995e-05,
+      "loss": 0.4831,
+      "step": 1962
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.4825581395348835e-05,
+      "loss": 0.3305,
+      "step": 1964
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.482029598308669e-05,
+      "loss": 0.2836,
+      "step": 1966
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.481501057082453e-05,
+      "loss": 0.3986,
+      "step": 1968
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.480972515856237e-05,
+      "loss": 0.3521,
+      "step": 1970
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.480443974630021e-05,
+      "loss": 0.5026,
+      "step": 1972
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.479915433403806e-05,
+      "loss": 0.287,
+      "step": 1974
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.47938689217759e-05,
+      "loss": 0.3817,
+      "step": 1976
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4788583509513744e-05,
+      "loss": 0.4855,
+      "step": 1978
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.478329809725158e-05,
+      "loss": 0.4728,
+      "step": 1980
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.477801268498943e-05,
+      "loss": 0.3979,
+      "step": 1982
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.4772727272727275e-05,
+      "loss": 0.3166,
+      "step": 1984
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.476744186046512e-05,
+      "loss": 0.5586,
+      "step": 1986
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.476215644820296e-05,
+      "loss": 0.5657,
+      "step": 1988
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.4756871035940806e-05,
+      "loss": 0.4058,
+      "step": 1990
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.475158562367865e-05,
+      "loss": 0.33,
+      "step": 1992
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.474630021141649e-05,
+      "loss": 0.3553,
+      "step": 1994
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.474101479915434e-05,
+      "loss": 0.3877,
+      "step": 1996
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.473572938689218e-05,
+      "loss": 0.481,
+      "step": 1998
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.473044397463002e-05,
+      "loss": 0.4098,
+      "step": 2000
+    },
+    {
+      "epoch": 2.11,
+      "eval_cer": 0.054830436021658595,
+      "eval_loss": 0.4985881447792053,
+      "eval_runtime": 132.0888,
+      "eval_samples_per_second": 6.367,
+      "eval_steps_per_second": 0.802,
+      "step": 2000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.472515856236787e-05,
+      "loss": 0.482,
+      "step": 2002
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.4719873150105715e-05,
+      "loss": 0.3529,
+      "step": 2004
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.4714587737843555e-05,
+      "loss": 0.4144,
+      "step": 2006
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.47093023255814e-05,
+      "loss": 0.2902,
+      "step": 2008
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.470401691331924e-05,
+      "loss": 0.4057,
+      "step": 2010
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.4698731501057086e-05,
+      "loss": 0.4743,
+      "step": 2012
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.4693446088794925e-05,
+      "loss": 0.3409,
+      "step": 2014
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.468816067653277e-05,
+      "loss": 0.4337,
+      "step": 2016
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.468287526427061e-05,
+      "loss": 0.5333,
+      "step": 2018
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.467758985200846e-05,
+      "loss": 0.2719,
+      "step": 2020
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.46723044397463e-05,
+      "loss": 0.4625,
+      "step": 2022
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.466701902748415e-05,
+      "loss": 0.2435,
+      "step": 2024
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.466173361522199e-05,
+      "loss": 0.4023,
+      "step": 2026
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.4656448202959834e-05,
+      "loss": 0.4343,
+      "step": 2028
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.465116279069767e-05,
+      "loss": 0.3526,
+      "step": 2030
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.464587737843552e-05,
+      "loss": 0.4709,
+      "step": 2032
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.464059196617336e-05,
+      "loss": 0.5203,
+      "step": 2034
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.4635306553911205e-05,
+      "loss": 0.7966,
+      "step": 2036
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.463002114164905e-05,
+      "loss": 0.5275,
+      "step": 2038
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.46247357293869e-05,
+      "loss": 0.4834,
+      "step": 2040
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.4619450317124736e-05,
+      "loss": 0.6763,
+      "step": 2042
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.461416490486258e-05,
+      "loss": 0.5105,
+      "step": 2044
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.460887949260043e-05,
+      "loss": 0.3415,
+      "step": 2046
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.460359408033827e-05,
+      "loss": 0.5119,
+      "step": 2048
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.4598308668076114e-05,
+      "loss": 0.3535,
+      "step": 2050
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.459302325581395e-05,
+      "loss": 0.3661,
+      "step": 2052
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.45877378435518e-05,
+      "loss": 0.4584,
+      "step": 2054
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.4582452431289645e-05,
+      "loss": 0.5095,
+      "step": 2056
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.457716701902749e-05,
+      "loss": 0.7405,
+      "step": 2058
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.457188160676533e-05,
+      "loss": 0.412,
+      "step": 2060
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.4566596194503176e-05,
+      "loss": 0.3918,
+      "step": 2062
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.4561310782241016e-05,
+      "loss": 0.4374,
+      "step": 2064
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.455602536997886e-05,
+      "loss": 0.2783,
+      "step": 2066
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.45507399577167e-05,
+      "loss": 0.5863,
+      "step": 2068
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.454545454545455e-05,
+      "loss": 0.3459,
+      "step": 2070
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.4540169133192386e-05,
+      "loss": 0.3781,
+      "step": 2072
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.453488372093024e-05,
+      "loss": 0.2417,
+      "step": 2074
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.452959830866808e-05,
+      "loss": 0.3893,
+      "step": 2076
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.4524312896405925e-05,
+      "loss": 0.4426,
+      "step": 2078
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.4519027484143764e-05,
+      "loss": 0.372,
+      "step": 2080
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.451374207188161e-05,
+      "loss": 0.5169,
+      "step": 2082
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.450845665961945e-05,
+      "loss": 0.484,
+      "step": 2084
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.4503171247357295e-05,
+      "loss": 0.3746,
+      "step": 2086
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.4497885835095135e-05,
+      "loss": 0.3725,
+      "step": 2088
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.449260042283298e-05,
+      "loss": 0.4112,
+      "step": 2090
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.448731501057083e-05,
+      "loss": 0.3573,
+      "step": 2092
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.448202959830867e-05,
+      "loss": 0.6657,
+      "step": 2094
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.447674418604651e-05,
+      "loss": 0.4883,
+      "step": 2096
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.447145877378436e-05,
+      "loss": 0.4407,
+      "step": 2098
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.4466173361522204e-05,
+      "loss": 0.4828,
+      "step": 2100
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.4460887949260043e-05,
+      "loss": 0.4325,
+      "step": 2102
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.445560253699789e-05,
+      "loss": 0.4517,
+      "step": 2104
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.445031712473573e-05,
+      "loss": 0.3728,
+      "step": 2106
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.4445031712473575e-05,
+      "loss": 0.4571,
+      "step": 2108
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.443974630021142e-05,
+      "loss": 0.5384,
+      "step": 2110
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.443446088794927e-05,
+      "loss": 0.7312,
+      "step": 2112
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.4429175475687106e-05,
+      "loss": 0.4921,
+      "step": 2114
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.442389006342495e-05,
+      "loss": 0.3399,
+      "step": 2116
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.441860465116279e-05,
+      "loss": 0.4376,
+      "step": 2118
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.441331923890064e-05,
+      "loss": 0.3879,
+      "step": 2120
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.440803382663848e-05,
+      "loss": 0.272,
+      "step": 2122
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.440274841437632e-05,
+      "loss": 0.5572,
+      "step": 2124
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.439746300211416e-05,
+      "loss": 0.4251,
+      "step": 2126
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.4392177589852015e-05,
+      "loss": 0.4728,
+      "step": 2128
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.4386892177589854e-05,
+      "loss": 0.6148,
+      "step": 2130
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.43816067653277e-05,
+      "loss": 0.4623,
+      "step": 2132
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.437632135306554e-05,
+      "loss": 0.4297,
+      "step": 2134
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.4371035940803386e-05,
+      "loss": 0.2851,
+      "step": 2136
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.4365750528541225e-05,
+      "loss": 0.4021,
+      "step": 2138
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.436046511627907e-05,
+      "loss": 0.3805,
+      "step": 2140
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.435517970401691e-05,
+      "loss": 0.3594,
+      "step": 2142
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.4349894291754756e-05,
+      "loss": 0.2067,
+      "step": 2144
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.43446088794926e-05,
+      "loss": 0.4969,
+      "step": 2146
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.433932346723045e-05,
+      "loss": 0.3155,
+      "step": 2148
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.433403805496829e-05,
+      "loss": 0.4758,
+      "step": 2150
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.4328752642706134e-05,
+      "loss": 0.2722,
+      "step": 2152
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.432346723044398e-05,
+      "loss": 0.5713,
+      "step": 2154
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.431818181818182e-05,
+      "loss": 0.4699,
+      "step": 2156
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.4312896405919665e-05,
+      "loss": 0.4833,
+      "step": 2158
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.4307610993657505e-05,
+      "loss": 0.2738,
+      "step": 2160
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.430232558139535e-05,
+      "loss": 0.5175,
+      "step": 2162
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.42970401691332e-05,
+      "loss": 0.3853,
+      "step": 2164
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.429175475687104e-05,
+      "loss": 0.519,
+      "step": 2166
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.428646934460888e-05,
+      "loss": 0.4999,
+      "step": 2168
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.428118393234673e-05,
+      "loss": 0.2495,
+      "step": 2170
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.427589852008457e-05,
+      "loss": 0.4331,
+      "step": 2172
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.4270613107822413e-05,
+      "loss": 0.4553,
+      "step": 2174
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.426532769556025e-05,
+      "loss": 0.4201,
+      "step": 2176
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.42600422832981e-05,
+      "loss": 0.567,
+      "step": 2178
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.425475687103594e-05,
+      "loss": 0.2782,
+      "step": 2180
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.424947145877379e-05,
+      "loss": 0.4802,
+      "step": 2182
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.424418604651163e-05,
+      "loss": 0.3771,
+      "step": 2184
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.4238900634249476e-05,
+      "loss": 0.2445,
+      "step": 2186
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.4233615221987316e-05,
+      "loss": 0.4321,
+      "step": 2188
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.422832980972516e-05,
+      "loss": 0.3165,
+      "step": 2190
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.4223044397463e-05,
+      "loss": 0.4181,
+      "step": 2192
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.421775898520085e-05,
+      "loss": 0.3771,
+      "step": 2194
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.421247357293869e-05,
+      "loss": 0.4594,
+      "step": 2196
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.420718816067653e-05,
+      "loss": 0.357,
+      "step": 2198
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.420190274841438e-05,
+      "loss": 0.3837,
+      "step": 2200
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4196617336152224e-05,
+      "loss": 0.3,
+      "step": 2202
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4191331923890064e-05,
+      "loss": 0.5577,
+      "step": 2204
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.418604651162791e-05,
+      "loss": 0.7798,
+      "step": 2206
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4180761099365756e-05,
+      "loss": 0.3581,
+      "step": 2208
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4175475687103595e-05,
+      "loss": 0.5661,
+      "step": 2210
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.417019027484144e-05,
+      "loss": 0.4161,
+      "step": 2212
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.416490486257928e-05,
+      "loss": 0.3675,
+      "step": 2214
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4159619450317126e-05,
+      "loss": 0.5372,
+      "step": 2216
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.415433403805497e-05,
+      "loss": 0.5398,
+      "step": 2218
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.414904862579282e-05,
+      "loss": 0.3739,
+      "step": 2220
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.414376321353066e-05,
+      "loss": 0.5355,
+      "step": 2222
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.4138477801268504e-05,
+      "loss": 0.5392,
+      "step": 2224
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.413319238900634e-05,
+      "loss": 0.4832,
+      "step": 2226
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.412790697674419e-05,
+      "loss": 0.4194,
+      "step": 2228
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.412262156448203e-05,
+      "loss": 0.5088,
+      "step": 2230
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.4117336152219875e-05,
+      "loss": 0.401,
+      "step": 2232
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.4112050739957714e-05,
+      "loss": 0.5165,
+      "step": 2234
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.410676532769557e-05,
+      "loss": 0.4715,
+      "step": 2236
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.4101479915433406e-05,
+      "loss": 0.3201,
+      "step": 2238
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.409619450317125e-05,
+      "loss": 0.4222,
+      "step": 2240
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.409090909090909e-05,
+      "loss": 0.3206,
+      "step": 2242
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.408562367864694e-05,
+      "loss": 0.3141,
+      "step": 2244
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.408033826638478e-05,
+      "loss": 0.5056,
+      "step": 2246
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.407505285412262e-05,
+      "loss": 0.4578,
+      "step": 2248
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.406976744186047e-05,
+      "loss": 0.3295,
+      "step": 2250
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.406448202959831e-05,
+      "loss": 0.3376,
+      "step": 2252
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.4059196617336154e-05,
+      "loss": 0.4404,
+      "step": 2254
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.4053911205074e-05,
+      "loss": 0.3081,
+      "step": 2256
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.404862579281184e-05,
+      "loss": 0.2706,
+      "step": 2258
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.4043340380549686e-05,
+      "loss": 0.5501,
+      "step": 2260
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.403805496828753e-05,
+      "loss": 0.3977,
+      "step": 2262
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.403276955602537e-05,
+      "loss": 0.3854,
+      "step": 2264
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.402748414376322e-05,
+      "loss": 0.5107,
+      "step": 2266
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.4022198731501056e-05,
+      "loss": 0.3551,
+      "step": 2268
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.40169133192389e-05,
+      "loss": 0.376,
+      "step": 2270
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.401162790697675e-05,
+      "loss": 0.3767,
+      "step": 2272
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.4006342494714594e-05,
+      "loss": 0.3593,
+      "step": 2274
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.4001057082452434e-05,
+      "loss": 0.5396,
+      "step": 2276
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.399577167019028e-05,
+      "loss": 0.4324,
+      "step": 2278
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.399048625792812e-05,
+      "loss": 0.0967,
+      "step": 2280
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.3985200845665965e-05,
+      "loss": 0.4564,
+      "step": 2282
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.3979915433403804e-05,
+      "loss": 0.3536,
+      "step": 2284
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.397463002114165e-05,
+      "loss": 0.5021,
+      "step": 2286
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.396934460887949e-05,
+      "loss": 0.555,
+      "step": 2288
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.396405919661734e-05,
+      "loss": 0.5308,
+      "step": 2290
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.395877378435518e-05,
+      "loss": 0.4026,
+      "step": 2292
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.395348837209303e-05,
+      "loss": 0.5569,
+      "step": 2294
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.394820295983087e-05,
+      "loss": 0.5554,
+      "step": 2296
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.394291754756871e-05,
+      "loss": 0.5833,
+      "step": 2298
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.393763213530655e-05,
+      "loss": 0.5731,
+      "step": 2300
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.39323467230444e-05,
+      "loss": 0.2924,
+      "step": 2302
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.3927061310782245e-05,
+      "loss": 0.3258,
+      "step": 2304
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.3921775898520084e-05,
+      "loss": 0.4652,
+      "step": 2306
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.391649048625793e-05,
+      "loss": 0.4098,
+      "step": 2308
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.3911205073995776e-05,
+      "loss": 0.4637,
+      "step": 2310
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.390591966173362e-05,
+      "loss": 0.5208,
+      "step": 2312
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.390063424947146e-05,
+      "loss": 0.3701,
+      "step": 2314
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.389534883720931e-05,
+      "loss": 0.3613,
+      "step": 2316
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.389006342494715e-05,
+      "loss": 0.5958,
+      "step": 2318
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.388477801268499e-05,
+      "loss": 0.3932,
+      "step": 2320
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.387949260042283e-05,
+      "loss": 0.5768,
+      "step": 2322
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.387420718816068e-05,
+      "loss": 0.6017,
+      "step": 2324
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.3868921775898524e-05,
+      "loss": 0.4225,
+      "step": 2326
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.386363636363637e-05,
+      "loss": 0.5372,
+      "step": 2328
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.385835095137421e-05,
+      "loss": 0.4131,
+      "step": 2330
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.3853065539112056e-05,
+      "loss": 0.5359,
+      "step": 2332
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.3847780126849895e-05,
+      "loss": 0.2422,
+      "step": 2334
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.384249471458774e-05,
+      "loss": 0.3424,
+      "step": 2336
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.383720930232558e-05,
+      "loss": 0.4612,
+      "step": 2338
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.3831923890063426e-05,
+      "loss": 0.3877,
+      "step": 2340
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.3826638477801266e-05,
+      "loss": 0.7339,
+      "step": 2342
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.382135306553912e-05,
+      "loss": 0.4984,
+      "step": 2344
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.381606765327696e-05,
+      "loss": 0.6876,
+      "step": 2346
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.3810782241014804e-05,
+      "loss": 0.4758,
+      "step": 2348
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.380549682875264e-05,
+      "loss": 0.324,
+      "step": 2350
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.380021141649049e-05,
+      "loss": 0.4365,
+      "step": 2352
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.379492600422833e-05,
+      "loss": 0.3447,
+      "step": 2354
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.3789640591966174e-05,
+      "loss": 0.3975,
+      "step": 2356
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.378435517970402e-05,
+      "loss": 0.3833,
+      "step": 2358
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.377906976744186e-05,
+      "loss": 0.386,
+      "step": 2360
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.3773784355179706e-05,
+      "loss": 0.5437,
+      "step": 2362
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.376849894291755e-05,
+      "loss": 0.4142,
+      "step": 2364
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.37632135306554e-05,
+      "loss": 0.4588,
+      "step": 2366
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.375792811839324e-05,
+      "loss": 0.3383,
+      "step": 2368
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.375264270613108e-05,
+      "loss": 0.3192,
+      "step": 2370
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.374735729386892e-05,
+      "loss": 0.5349,
+      "step": 2372
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.374207188160677e-05,
+      "loss": 0.2711,
+      "step": 2374
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.373678646934461e-05,
+      "loss": 0.3692,
+      "step": 2376
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.3731501057082454e-05,
+      "loss": 0.2674,
+      "step": 2378
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.372621564482029e-05,
+      "loss": 0.3274,
+      "step": 2380
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.3720930232558146e-05,
+      "loss": 0.4498,
+      "step": 2382
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.3715644820295985e-05,
+      "loss": 0.4294,
+      "step": 2384
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.371035940803383e-05,
+      "loss": 0.4257,
+      "step": 2386
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.370507399577167e-05,
+      "loss": 0.4678,
+      "step": 2388
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.369978858350952e-05,
+      "loss": 0.3635,
+      "step": 2390
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.3694503171247356e-05,
+      "loss": 0.5947,
+      "step": 2392
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.36892177589852e-05,
+      "loss": 0.4308,
+      "step": 2394
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.368393234672304e-05,
+      "loss": 0.3585,
+      "step": 2396
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.367864693446089e-05,
+      "loss": 0.844,
+      "step": 2398
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.3673361522198734e-05,
+      "loss": 0.4748,
+      "step": 2400
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.366807610993658e-05,
+      "loss": 0.3876,
+      "step": 2402
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.366279069767442e-05,
+      "loss": 0.4596,
+      "step": 2404
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.3657505285412265e-05,
+      "loss": 0.2919,
+      "step": 2406
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.3652219873150104e-05,
+      "loss": 0.4279,
+      "step": 2408
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.364693446088795e-05,
+      "loss": 0.3172,
+      "step": 2410
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.3641649048625796e-05,
+      "loss": 0.3116,
+      "step": 2412
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.3636363636363636e-05,
+      "loss": 0.4824,
+      "step": 2414
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.363107822410148e-05,
+      "loss": 0.3576,
+      "step": 2416
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.362579281183933e-05,
+      "loss": 0.28,
+      "step": 2418
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.3620507399577174e-05,
+      "loss": 0.3368,
+      "step": 2420
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.361522198731501e-05,
+      "loss": 0.399,
+      "step": 2422
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.360993657505286e-05,
+      "loss": 0.3882,
+      "step": 2424
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.36046511627907e-05,
+      "loss": 0.3533,
+      "step": 2426
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.3599365750528544e-05,
+      "loss": 0.4115,
+      "step": 2428
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.3594080338266384e-05,
+      "loss": 0.4398,
+      "step": 2430
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.358879492600423e-05,
+      "loss": 0.3063,
+      "step": 2432
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.358350951374207e-05,
+      "loss": 0.3138,
+      "step": 2434
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.357822410147992e-05,
+      "loss": 0.3835,
+      "step": 2436
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.357293868921776e-05,
+      "loss": 0.493,
+      "step": 2438
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.356765327695561e-05,
+      "loss": 0.4659,
+      "step": 2440
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.3562367864693447e-05,
+      "loss": 0.4226,
+      "step": 2442
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.355708245243129e-05,
+      "loss": 0.3218,
+      "step": 2444
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.355179704016913e-05,
+      "loss": 0.271,
+      "step": 2446
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.354651162790698e-05,
+      "loss": 0.4356,
+      "step": 2448
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.354122621564482e-05,
+      "loss": 0.4688,
+      "step": 2450
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.353594080338266e-05,
+      "loss": 0.4016,
+      "step": 2452
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.353065539112051e-05,
+      "loss": 0.4853,
+      "step": 2454
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.3525369978858355e-05,
+      "loss": 0.4232,
+      "step": 2456
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.3520084566596195e-05,
+      "loss": 0.2393,
+      "step": 2458
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.351479915433404e-05,
+      "loss": 0.3287,
+      "step": 2460
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.350951374207189e-05,
+      "loss": 0.3494,
+      "step": 2462
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.3504228329809726e-05,
+      "loss": 0.5148,
+      "step": 2464
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.349894291754757e-05,
+      "loss": 0.4271,
+      "step": 2466
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.349365750528541e-05,
+      "loss": 0.2264,
+      "step": 2468
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.348837209302326e-05,
+      "loss": 0.3973,
+      "step": 2470
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.3483086680761104e-05,
+      "loss": 0.3798,
+      "step": 2472
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.347780126849895e-05,
+      "loss": 0.4355,
+      "step": 2474
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.347251585623679e-05,
+      "loss": 0.5765,
+      "step": 2476
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.3467230443974635e-05,
+      "loss": 0.4032,
+      "step": 2478
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.3461945031712474e-05,
+      "loss": 0.348,
+      "step": 2480
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.345665961945032e-05,
+      "loss": 0.2763,
+      "step": 2482
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.345137420718816e-05,
+      "loss": 0.4957,
+      "step": 2484
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.3446088794926006e-05,
+      "loss": 0.3058,
+      "step": 2486
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.3440803382663845e-05,
+      "loss": 0.4072,
+      "step": 2488
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.34355179704017e-05,
+      "loss": 0.3913,
+      "step": 2490
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.343023255813954e-05,
+      "loss": 0.2427,
+      "step": 2492
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.342494714587738e-05,
+      "loss": 0.3805,
+      "step": 2494
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.341966173361522e-05,
+      "loss": 0.3677,
+      "step": 2496
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.341437632135307e-05,
+      "loss": 0.3233,
+      "step": 2498
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.340909090909091e-05,
+      "loss": 0.3522,
+      "step": 2500
+    },
+    {
+      "epoch": 2.64,
+      "eval_cer": 0.0402393844400114,
+      "eval_loss": 0.47415897250175476,
+      "eval_runtime": 131.0076,
+      "eval_samples_per_second": 6.419,
+      "eval_steps_per_second": 0.809,
+      "step": 2500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.3403805496828754e-05,
+      "loss": 0.258,
+      "step": 2502
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.339852008456659e-05,
+      "loss": 0.3387,
+      "step": 2504
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.339323467230444e-05,
+      "loss": 0.3951,
+      "step": 2506
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.3387949260042285e-05,
+      "loss": 0.5066,
+      "step": 2508
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.338266384778013e-05,
+      "loss": 0.3595,
+      "step": 2510
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.337737843551797e-05,
+      "loss": 0.3599,
+      "step": 2512
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.337209302325582e-05,
+      "loss": 0.3667,
+      "step": 2514
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.336680761099366e-05,
+      "loss": 0.4277,
+      "step": 2516
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.33615221987315e-05,
+      "loss": 0.5447,
+      "step": 2518
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.335623678646935e-05,
+      "loss": 0.339,
+      "step": 2520
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.335095137420719e-05,
+      "loss": 0.4673,
+      "step": 2522
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.334566596194503e-05,
+      "loss": 0.408,
+      "step": 2524
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.334038054968288e-05,
+      "loss": 0.3587,
+      "step": 2526
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.3335095137420725e-05,
+      "loss": 0.464,
+      "step": 2528
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.3329809725158565e-05,
+      "loss": 0.3021,
+      "step": 2530
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.332452431289641e-05,
+      "loss": 0.4235,
+      "step": 2532
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.331923890063425e-05,
+      "loss": 0.413,
+      "step": 2534
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.3313953488372096e-05,
+      "loss": 0.3187,
+      "step": 2536
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.3308668076109935e-05,
+      "loss": 0.4356,
+      "step": 2538
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.330338266384778e-05,
+      "loss": 0.4303,
+      "step": 2540
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.329809725158562e-05,
+      "loss": 0.4085,
+      "step": 2542
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.3292811839323474e-05,
+      "loss": 0.4981,
+      "step": 2544
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.328752642706131e-05,
+      "loss": 0.6279,
+      "step": 2546
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.328224101479916e-05,
+      "loss": 0.5337,
+      "step": 2548
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.3276955602537e-05,
+      "loss": 0.3601,
+      "step": 2550
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.3271670190274844e-05,
+      "loss": 0.4367,
+      "step": 2552
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.3266384778012684e-05,
+      "loss": 0.276,
+      "step": 2554
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.326109936575053e-05,
+      "loss": 0.3276,
+      "step": 2556
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.325581395348837e-05,
+      "loss": 0.4869,
+      "step": 2558
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.3250528541226215e-05,
+      "loss": 0.468,
+      "step": 2560
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.324524312896406e-05,
+      "loss": 0.3476,
+      "step": 2562
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.323995771670191e-05,
+      "loss": 0.4213,
+      "step": 2564
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.3234672304439746e-05,
+      "loss": 0.2384,
+      "step": 2566
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.322938689217759e-05,
+      "loss": 0.4982,
+      "step": 2568
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.322410147991544e-05,
+      "loss": 0.372,
+      "step": 2570
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.321881606765328e-05,
+      "loss": 0.3857,
+      "step": 2572
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3213530655391124e-05,
+      "loss": 0.3346,
+      "step": 2574
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.320824524312896e-05,
+      "loss": 0.4945,
+      "step": 2576
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.320295983086681e-05,
+      "loss": 0.4013,
+      "step": 2578
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3197674418604655e-05,
+      "loss": 0.2965,
+      "step": 2580
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.31923890063425e-05,
+      "loss": 0.3815,
+      "step": 2582
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.318710359408034e-05,
+      "loss": 0.2204,
+      "step": 2584
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.318181818181819e-05,
+      "loss": 0.3671,
+      "step": 2586
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3176532769556026e-05,
+      "loss": 0.5415,
+      "step": 2588
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.317124735729387e-05,
+      "loss": 0.3768,
+      "step": 2590
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.316596194503171e-05,
+      "loss": 0.377,
+      "step": 2592
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.316067653276956e-05,
+      "loss": 0.5208,
+      "step": 2594
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.31553911205074e-05,
+      "loss": 0.3114,
+      "step": 2596
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.315010570824525e-05,
+      "loss": 0.2648,
+      "step": 2598
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.314482029598309e-05,
+      "loss": 0.2037,
+      "step": 2600
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.3139534883720935e-05,
+      "loss": 0.2728,
+      "step": 2602
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.3134249471458774e-05,
+      "loss": 0.5251,
+      "step": 2604
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.312896405919662e-05,
+      "loss": 0.4707,
+      "step": 2606
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.312367864693446e-05,
+      "loss": 0.2391,
+      "step": 2608
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.3118393234672305e-05,
+      "loss": 0.3286,
+      "step": 2610
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.3113107822410145e-05,
+      "loss": 0.2599,
+      "step": 2612
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.310782241014799e-05,
+      "loss": 0.5016,
+      "step": 2614
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.310253699788584e-05,
+      "loss": 0.4022,
+      "step": 2616
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.309725158562368e-05,
+      "loss": 0.4263,
+      "step": 2618
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.309196617336152e-05,
+      "loss": 0.4461,
+      "step": 2620
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.308668076109937e-05,
+      "loss": 0.3174,
+      "step": 2622
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.3081395348837214e-05,
+      "loss": 0.4512,
+      "step": 2624
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.3076109936575054e-05,
+      "loss": 0.3774,
+      "step": 2626
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.30708245243129e-05,
+      "loss": 0.3699,
+      "step": 2628
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.306553911205074e-05,
+      "loss": 0.4952,
+      "step": 2630
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.3060253699788585e-05,
+      "loss": 0.3127,
+      "step": 2632
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.305496828752643e-05,
+      "loss": 0.3005,
+      "step": 2634
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.304968287526428e-05,
+      "loss": 0.5216,
+      "step": 2636
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.3044397463002116e-05,
+      "loss": 0.1992,
+      "step": 2638
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.303911205073996e-05,
+      "loss": 0.3832,
+      "step": 2640
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.30338266384778e-05,
+      "loss": 0.3892,
+      "step": 2642
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.302854122621565e-05,
+      "loss": 0.4152,
+      "step": 2644
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.302325581395349e-05,
+      "loss": 0.3142,
+      "step": 2646
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.301797040169133e-05,
+      "loss": 0.381,
+      "step": 2648
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.301268498942917e-05,
+      "loss": 0.4858,
+      "step": 2650
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.3007399577167025e-05,
+      "loss": 0.4228,
+      "step": 2652
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.3002114164904865e-05,
+      "loss": 0.3344,
+      "step": 2654
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.299682875264271e-05,
+      "loss": 0.3139,
+      "step": 2656
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.299154334038055e-05,
+      "loss": 0.4021,
+      "step": 2658
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.2986257928118396e-05,
+      "loss": 0.2641,
+      "step": 2660
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.2980972515856235e-05,
+      "loss": 0.2865,
+      "step": 2662
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.297568710359408e-05,
+      "loss": 0.4632,
+      "step": 2664
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.297040169133193e-05,
+      "loss": 0.4016,
+      "step": 2666
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.296511627906977e-05,
+      "loss": 0.3713,
+      "step": 2668
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.295983086680761e-05,
+      "loss": 0.5221,
+      "step": 2670
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.295454545454546e-05,
+      "loss": 0.398,
+      "step": 2672
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.29492600422833e-05,
+      "loss": 0.4723,
+      "step": 2674
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.2943974630021144e-05,
+      "loss": 0.441,
+      "step": 2676
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.293868921775899e-05,
+      "loss": 0.3097,
+      "step": 2678
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.293340380549683e-05,
+      "loss": 0.4571,
+      "step": 2680
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.2928118393234676e-05,
+      "loss": 0.2796,
+      "step": 2682
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.2922832980972515e-05,
+      "loss": 0.2998,
+      "step": 2684
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.291754756871036e-05,
+      "loss": 0.3923,
+      "step": 2686
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.291226215644821e-05,
+      "loss": 0.2452,
+      "step": 2688
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.290697674418605e-05,
+      "loss": 0.4391,
+      "step": 2690
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.290169133192389e-05,
+      "loss": 0.4308,
+      "step": 2692
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.289640591966174e-05,
+      "loss": 0.2987,
+      "step": 2694
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.289112050739958e-05,
+      "loss": 0.6786,
+      "step": 2696
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.2885835095137424e-05,
+      "loss": 0.645,
+      "step": 2698
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.288054968287526e-05,
+      "loss": 0.6325,
+      "step": 2700
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.287526427061311e-05,
+      "loss": 0.3546,
+      "step": 2702
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.286997885835095e-05,
+      "loss": 0.3041,
+      "step": 2704
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.28646934460888e-05,
+      "loss": 0.3767,
+      "step": 2706
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.285940803382664e-05,
+      "loss": 0.3425,
+      "step": 2708
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.2854122621564486e-05,
+      "loss": 0.3014,
+      "step": 2710
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.2848837209302326e-05,
+      "loss": 0.3421,
+      "step": 2712
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.284355179704017e-05,
+      "loss": 0.3349,
+      "step": 2714
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.283826638477801e-05,
+      "loss": 0.2842,
+      "step": 2716
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.283298097251586e-05,
+      "loss": 0.4648,
+      "step": 2718
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.28276955602537e-05,
+      "loss": 0.2628,
+      "step": 2720
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.282241014799154e-05,
+      "loss": 0.3664,
+      "step": 2722
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.281712473572939e-05,
+      "loss": 0.4732,
+      "step": 2724
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.2811839323467235e-05,
+      "loss": 0.3712,
+      "step": 2726
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.2806553911205074e-05,
+      "loss": 0.4412,
+      "step": 2728
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.280126849894292e-05,
+      "loss": 0.2067,
+      "step": 2730
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.2795983086680766e-05,
+      "loss": 0.3256,
+      "step": 2732
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.2790697674418605e-05,
+      "loss": 0.2648,
+      "step": 2734
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.278541226215645e-05,
+      "loss": 0.2814,
+      "step": 2736
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.278012684989429e-05,
+      "loss": 0.4367,
+      "step": 2738
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.277484143763214e-05,
+      "loss": 0.3559,
+      "step": 2740
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.276955602536998e-05,
+      "loss": 0.6629,
+      "step": 2742
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.276427061310783e-05,
+      "loss": 0.5756,
+      "step": 2744
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.275898520084567e-05,
+      "loss": 0.337,
+      "step": 2746
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.2753699788583514e-05,
+      "loss": 0.577,
+      "step": 2748
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2748414376321353e-05,
+      "loss": 0.4437,
+      "step": 2750
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.27431289640592e-05,
+      "loss": 0.2032,
+      "step": 2752
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.273784355179704e-05,
+      "loss": 0.4406,
+      "step": 2754
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2732558139534885e-05,
+      "loss": 0.3424,
+      "step": 2756
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.2727272727272724e-05,
+      "loss": 0.4671,
+      "step": 2758
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.272198731501058e-05,
+      "loss": 0.3583,
+      "step": 2760
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.2716701902748416e-05,
+      "loss": 0.5143,
+      "step": 2762
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.271141649048626e-05,
+      "loss": 0.4685,
+      "step": 2764
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.27061310782241e-05,
+      "loss": 0.7,
+      "step": 2766
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.270084566596195e-05,
+      "loss": 0.5151,
+      "step": 2768
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.269556025369979e-05,
+      "loss": 0.4323,
+      "step": 2770
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.269027484143763e-05,
+      "loss": 0.4289,
+      "step": 2772
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.268498942917548e-05,
+      "loss": 0.2696,
+      "step": 2774
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.267970401691332e-05,
+      "loss": 0.649,
+      "step": 2776
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.2674418604651164e-05,
+      "loss": 0.3215,
+      "step": 2778
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.266913319238901e-05,
+      "loss": 0.3971,
+      "step": 2780
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.2663847780126857e-05,
+      "loss": 0.4493,
+      "step": 2782
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.2658562367864696e-05,
+      "loss": 0.523,
+      "step": 2784
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.265327695560254e-05,
+      "loss": 0.3435,
+      "step": 2786
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.264799154334038e-05,
+      "loss": 0.2757,
+      "step": 2788
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.264270613107823e-05,
+      "loss": 0.4671,
+      "step": 2790
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.2637420718816066e-05,
+      "loss": 0.3897,
+      "step": 2792
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.263213530655391e-05,
+      "loss": 0.2445,
+      "step": 2794
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.262684989429176e-05,
+      "loss": 0.3071,
+      "step": 2796
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.2621564482029605e-05,
+      "loss": 0.3959,
+      "step": 2798
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.2616279069767444e-05,
+      "loss": 0.3327,
+      "step": 2800
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.261099365750529e-05,
+      "loss": 0.3544,
+      "step": 2802
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.260570824524313e-05,
+      "loss": 0.4166,
+      "step": 2804
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.2600422832980975e-05,
+      "loss": 0.3214,
+      "step": 2806
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.2595137420718815e-05,
+      "loss": 0.5656,
+      "step": 2808
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.258985200845666e-05,
+      "loss": 0.3529,
+      "step": 2810
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.25845665961945e-05,
+      "loss": 0.5777,
+      "step": 2812
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.257928118393235e-05,
+      "loss": 0.3313,
+      "step": 2814
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.257399577167019e-05,
+      "loss": 0.3668,
+      "step": 2816
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.256871035940804e-05,
+      "loss": 0.3936,
+      "step": 2818
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.256342494714588e-05,
+      "loss": 0.3968,
+      "step": 2820
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.2558139534883724e-05,
+      "loss": 0.6001,
+      "step": 2822
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.255285412262156e-05,
+      "loss": 0.4649,
+      "step": 2824
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.254756871035941e-05,
+      "loss": 0.3453,
+      "step": 2826
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.2542283298097255e-05,
+      "loss": 0.4322,
+      "step": 2828
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.2536997885835094e-05,
+      "loss": 0.4257,
+      "step": 2830
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.253171247357294e-05,
+      "loss": 0.2827,
+      "step": 2832
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.2526427061310786e-05,
+      "loss": 0.4138,
+      "step": 2834
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.252114164904863e-05,
+      "loss": 0.3991,
+      "step": 2836
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.251585623678647e-05,
+      "loss": 0.3762,
+      "step": 2838
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.251057082452432e-05,
+      "loss": 0.3813,
+      "step": 2840
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.250528541226216e-05,
+      "loss": 0.4144,
+      "step": 2842
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.25e-05,
+      "loss": 0.3277,
+      "step": 2844
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.249471458773784e-05,
+      "loss": 0.4051,
+      "step": 2846
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.248942917547569e-05,
+      "loss": 0.4003,
+      "step": 2848
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.2484143763213534e-05,
+      "loss": 0.3624,
+      "step": 2850
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.247885835095138e-05,
+      "loss": 0.4119,
+      "step": 2852
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.247357293868922e-05,
+      "loss": 0.6358,
+      "step": 2854
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.2468287526427066e-05,
+      "loss": 0.5912,
+      "step": 2856
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.2463002114164905e-05,
+      "loss": 0.295,
+      "step": 2858
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.245771670190275e-05,
+      "loss": 0.3205,
+      "step": 2860
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.245243128964059e-05,
+      "loss": 0.3077,
+      "step": 2862
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.2447145877378437e-05,
+      "loss": 0.259,
+      "step": 2864
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.2441860465116276e-05,
+      "loss": 0.4562,
+      "step": 2866
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.243657505285413e-05,
+      "loss": 0.2321,
+      "step": 2868
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.243128964059197e-05,
+      "loss": 0.539,
+      "step": 2870
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.2426004228329814e-05,
+      "loss": 0.3658,
+      "step": 2872
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.242071881606765e-05,
+      "loss": 0.2908,
+      "step": 2874
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.24154334038055e-05,
+      "loss": 0.262,
+      "step": 2876
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.241014799154334e-05,
+      "loss": 0.3137,
+      "step": 2878
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.2404862579281185e-05,
+      "loss": 0.2519,
+      "step": 2880
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.239957716701903e-05,
+      "loss": 0.3046,
+      "step": 2882
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.239429175475687e-05,
+      "loss": 0.5082,
+      "step": 2884
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.2389006342494716e-05,
+      "loss": 0.2263,
+      "step": 2886
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.238372093023256e-05,
+      "loss": 0.4031,
+      "step": 2888
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.237843551797041e-05,
+      "loss": 0.3983,
+      "step": 2890
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.237315010570825e-05,
+      "loss": 0.2879,
+      "step": 2892
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.2367864693446094e-05,
+      "loss": 0.2856,
+      "step": 2894
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.236257928118393e-05,
+      "loss": 0.2928,
+      "step": 2896
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.235729386892178e-05,
+      "loss": 0.2686,
+      "step": 2898
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.235200845665962e-05,
+      "loss": 0.4004,
+      "step": 2900
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.2346723044397464e-05,
+      "loss": 0.455,
+      "step": 2902
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.234143763213531e-05,
+      "loss": 0.5315,
+      "step": 2904
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.2336152219873156e-05,
+      "loss": 0.3951,
+      "step": 2906
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.2330866807610996e-05,
+      "loss": 0.3007,
+      "step": 2908
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.232558139534884e-05,
+      "loss": 0.3168,
+      "step": 2910
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.232029598308668e-05,
+      "loss": 0.3028,
+      "step": 2912
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.231501057082453e-05,
+      "loss": 0.1548,
+      "step": 2914
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.2309725158562366e-05,
+      "loss": 0.2496,
+      "step": 2916
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.230443974630021e-05,
+      "loss": 0.4231,
+      "step": 2918
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.229915433403805e-05,
+      "loss": 0.2385,
+      "step": 2920
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.2293868921775904e-05,
+      "loss": 0.342,
+      "step": 2922
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.2288583509513744e-05,
+      "loss": 0.2755,
+      "step": 2924
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.228329809725159e-05,
+      "loss": 0.2918,
+      "step": 2926
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.227801268498943e-05,
+      "loss": 0.3105,
+      "step": 2928
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.2272727272727275e-05,
+      "loss": 0.1969,
+      "step": 2930
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.226744186046512e-05,
+      "loss": 0.3114,
+      "step": 2932
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.226215644820296e-05,
+      "loss": 0.2758,
+      "step": 2934
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.2256871035940807e-05,
+      "loss": 0.2677,
+      "step": 2936
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.2251585623678646e-05,
+      "loss": 0.4188,
+      "step": 2938
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.224630021141649e-05,
+      "loss": 0.2955,
+      "step": 2940
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.224101479915434e-05,
+      "loss": 0.3511,
+      "step": 2942
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.2235729386892184e-05,
+      "loss": 0.2193,
+      "step": 2944
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.223044397463002e-05,
+      "loss": 0.2371,
+      "step": 2946
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.222515856236787e-05,
+      "loss": 0.3929,
+      "step": 2948
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.221987315010571e-05,
+      "loss": 0.4396,
+      "step": 2950
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.2214587737843555e-05,
+      "loss": 0.5568,
+      "step": 2952
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.2209302325581394e-05,
+      "loss": 0.4471,
+      "step": 2954
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.220401691331924e-05,
+      "loss": 0.4105,
+      "step": 2956
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.2198731501057086e-05,
+      "loss": 0.5462,
+      "step": 2958
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.219344608879493e-05,
+      "loss": 0.5864,
+      "step": 2960
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.218816067653277e-05,
+      "loss": 0.5403,
+      "step": 2962
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.218287526427062e-05,
+      "loss": 0.611,
+      "step": 2964
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.217758985200846e-05,
+      "loss": 0.4517,
+      "step": 2966
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.21723044397463e-05,
+      "loss": 0.3851,
+      "step": 2968
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.216701902748414e-05,
+      "loss": 0.3653,
+      "step": 2970
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.216173361522199e-05,
+      "loss": 0.4052,
+      "step": 2972
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.215644820295983e-05,
+      "loss": 0.3847,
+      "step": 2974
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.215116279069768e-05,
+      "loss": 0.2637,
+      "step": 2976
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.214587737843552e-05,
+      "loss": 0.3769,
+      "step": 2978
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.2140591966173366e-05,
+      "loss": 0.3022,
+      "step": 2980
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.2135306553911205e-05,
+      "loss": 0.3418,
+      "step": 2982
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.213002114164905e-05,
+      "loss": 0.2043,
+      "step": 2984
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.21247357293869e-05,
+      "loss": 0.263,
+      "step": 2986
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.2119450317124736e-05,
+      "loss": 0.2247,
+      "step": 2988
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.211416490486258e-05,
+      "loss": 0.2779,
+      "step": 2990
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.210887949260042e-05,
+      "loss": 0.274,
+      "step": 2992
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.210359408033827e-05,
+      "loss": 0.2576,
+      "step": 2994
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.2098308668076114e-05,
+      "loss": 0.4279,
+      "step": 2996
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.209302325581396e-05,
+      "loss": 0.2541,
+      "step": 2998
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.20877378435518e-05,
+      "loss": 0.3729,
+      "step": 3000
+    },
+    {
+      "epoch": 3.17,
+      "eval_cer": 0.048674836135651184,
+      "eval_loss": 0.49270230531692505,
+      "eval_runtime": 130.1947,
+      "eval_samples_per_second": 6.46,
+      "eval_steps_per_second": 0.814,
+      "step": 3000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.2082452431289645e-05,
+      "loss": 0.3319,
+      "step": 3002
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.2077167019027485e-05,
+      "loss": 0.3086,
+      "step": 3004
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.207188160676533e-05,
+      "loss": 0.3868,
+      "step": 3006
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.206659619450317e-05,
+      "loss": 0.3581,
+      "step": 3008
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.2061310782241016e-05,
+      "loss": 0.2487,
+      "step": 3010
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.205602536997886e-05,
+      "loss": 0.38,
+      "step": 3012
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.205073995771671e-05,
+      "loss": 0.4104,
+      "step": 3014
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.204545454545455e-05,
+      "loss": 0.251,
+      "step": 3016
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.204016913319239e-05,
+      "loss": 0.2777,
+      "step": 3018
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.203488372093023e-05,
+      "loss": 0.2662,
+      "step": 3020
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.202959830866808e-05,
+      "loss": 0.2533,
+      "step": 3022
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.202431289640592e-05,
+      "loss": 0.3681,
+      "step": 3024
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.2019027484143764e-05,
+      "loss": 0.4595,
+      "step": 3026
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.20137420718816e-05,
+      "loss": 0.358,
+      "step": 3028
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.2008456659619456e-05,
+      "loss": 0.2452,
+      "step": 3030
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.2003171247357295e-05,
+      "loss": 0.5741,
+      "step": 3032
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.199788583509514e-05,
+      "loss": 0.652,
+      "step": 3034
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.199260042283298e-05,
+      "loss": 0.2136,
+      "step": 3036
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.198731501057083e-05,
+      "loss": 0.3405,
+      "step": 3038
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.198202959830867e-05,
+      "loss": 0.5211,
+      "step": 3040
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.197674418604651e-05,
+      "loss": 0.2128,
+      "step": 3042
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.197145877378436e-05,
+      "loss": 0.3436,
+      "step": 3044
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.19661733615222e-05,
+      "loss": 0.3668,
+      "step": 3046
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.1960887949260044e-05,
+      "loss": 0.2641,
+      "step": 3048
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.195560253699789e-05,
+      "loss": 0.2484,
+      "step": 3050
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.1950317124735736e-05,
+      "loss": 0.3322,
+      "step": 3052
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.1945031712473575e-05,
+      "loss": 0.3741,
+      "step": 3054
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.193974630021142e-05,
+      "loss": 0.5316,
+      "step": 3056
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.193446088794926e-05,
+      "loss": 0.2144,
+      "step": 3058
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.1929175475687106e-05,
+      "loss": 0.4243,
+      "step": 3060
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.1923890063424946e-05,
+      "loss": 0.3349,
+      "step": 3062
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.191860465116279e-05,
+      "loss": 0.3093,
+      "step": 3064
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.191331923890063e-05,
+      "loss": 0.2343,
+      "step": 3066
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.1908033826638484e-05,
+      "loss": 0.2864,
+      "step": 3068
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.190274841437632e-05,
+      "loss": 0.2337,
+      "step": 3070
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.189746300211417e-05,
+      "loss": 0.333,
+      "step": 3072
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.189217758985201e-05,
+      "loss": 0.2783,
+      "step": 3074
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.1886892177589855e-05,
+      "loss": 0.3432,
+      "step": 3076
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.1881606765327694e-05,
+      "loss": 0.3868,
+      "step": 3078
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.187632135306554e-05,
+      "loss": 0.3249,
+      "step": 3080
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.187103594080338e-05,
+      "loss": 0.3919,
+      "step": 3082
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.1865750528541225e-05,
+      "loss": 0.2175,
+      "step": 3084
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.186046511627907e-05,
+      "loss": 0.3022,
+      "step": 3086
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.185517970401692e-05,
+      "loss": 0.2926,
+      "step": 3088
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.184989429175476e-05,
+      "loss": 0.2434,
+      "step": 3090
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.18446088794926e-05,
+      "loss": 0.2615,
+      "step": 3092
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.183932346723045e-05,
+      "loss": 0.3064,
+      "step": 3094
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.183403805496829e-05,
+      "loss": 0.281,
+      "step": 3096
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.1828752642706134e-05,
+      "loss": 0.2042,
+      "step": 3098
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.182346723044397e-05,
+      "loss": 0.4044,
+      "step": 3100
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.181818181818182e-05,
+      "loss": 0.266,
+      "step": 3102
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.1812896405919666e-05,
+      "loss": 0.2419,
+      "step": 3104
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.180761099365751e-05,
+      "loss": 0.2929,
+      "step": 3106
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.180232558139535e-05,
+      "loss": 0.4756,
+      "step": 3108
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.17970401691332e-05,
+      "loss": 0.3084,
+      "step": 3110
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.1791754756871036e-05,
+      "loss": 0.3543,
+      "step": 3112
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.178646934460888e-05,
+      "loss": 0.393,
+      "step": 3114
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.178118393234672e-05,
+      "loss": 0.3009,
+      "step": 3116
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.177589852008457e-05,
+      "loss": 0.185,
+      "step": 3118
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.177061310782241e-05,
+      "loss": 0.2683,
+      "step": 3120
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.176532769556026e-05,
+      "loss": 0.5429,
+      "step": 3122
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.17600422832981e-05,
+      "loss": 0.3401,
+      "step": 3124
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.1754756871035945e-05,
+      "loss": 0.3918,
+      "step": 3126
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.1749471458773784e-05,
+      "loss": 0.3553,
+      "step": 3128
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.174418604651163e-05,
+      "loss": 0.2473,
+      "step": 3130
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.173890063424947e-05,
+      "loss": 0.4602,
+      "step": 3132
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.1733615221987316e-05,
+      "loss": 0.4051,
+      "step": 3134
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.172832980972516e-05,
+      "loss": 0.2363,
+      "step": 3136
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.1723044397463e-05,
+      "loss": 0.2591,
+      "step": 3138
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.171775898520085e-05,
+      "loss": 0.3065,
+      "step": 3140
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.171247357293869e-05,
+      "loss": 0.2056,
+      "step": 3142
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.170718816067653e-05,
+      "loss": 0.302,
+      "step": 3144
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.170190274841438e-05,
+      "loss": 0.3368,
+      "step": 3146
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.1696617336152225e-05,
+      "loss": 0.3408,
+      "step": 3148
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.1691331923890064e-05,
+      "loss": 0.3338,
+      "step": 3150
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.168604651162791e-05,
+      "loss": 0.4434,
+      "step": 3152
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.168076109936575e-05,
+      "loss": 0.4222,
+      "step": 3154
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.1675475687103595e-05,
+      "loss": 0.3405,
+      "step": 3156
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.167019027484144e-05,
+      "loss": 0.3749,
+      "step": 3158
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.166490486257929e-05,
+      "loss": 0.2047,
+      "step": 3160
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.165961945031713e-05,
+      "loss": 0.3287,
+      "step": 3162
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.165433403805497e-05,
+      "loss": 0.3589,
+      "step": 3164
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.164904862579281e-05,
+      "loss": 0.6135,
+      "step": 3166
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.164376321353066e-05,
+      "loss": 0.8384,
+      "step": 3168
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.16384778012685e-05,
+      "loss": 0.3048,
+      "step": 3170
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.1633192389006343e-05,
+      "loss": 0.4265,
+      "step": 3172
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.162790697674418e-05,
+      "loss": 0.4554,
+      "step": 3174
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.1622621564482036e-05,
+      "loss": 0.3637,
+      "step": 3176
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.1617336152219875e-05,
+      "loss": 0.5761,
+      "step": 3178
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.161205073995772e-05,
+      "loss": 0.2125,
+      "step": 3180
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.160676532769556e-05,
+      "loss": 0.2687,
+      "step": 3182
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.1601479915433406e-05,
+      "loss": 0.4152,
+      "step": 3184
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.1596194503171246e-05,
+      "loss": 0.3966,
+      "step": 3186
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.159090909090909e-05,
+      "loss": 0.3472,
+      "step": 3188
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.158562367864694e-05,
+      "loss": 0.2772,
+      "step": 3190
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.158033826638478e-05,
+      "loss": 0.2744,
+      "step": 3192
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.157505285412262e-05,
+      "loss": 0.3092,
+      "step": 3194
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.156976744186047e-05,
+      "loss": 0.4398,
+      "step": 3196
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.1564482029598315e-05,
+      "loss": 0.1722,
+      "step": 3198
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.1559196617336154e-05,
+      "loss": 0.4008,
+      "step": 3200
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.1553911205074e-05,
+      "loss": 0.3137,
+      "step": 3202
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.154862579281184e-05,
+      "loss": 0.4953,
+      "step": 3204
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.1543340380549686e-05,
+      "loss": 0.2774,
+      "step": 3206
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.1538054968287525e-05,
+      "loss": 0.3924,
+      "step": 3208
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.153276955602537e-05,
+      "loss": 0.3354,
+      "step": 3210
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.152748414376322e-05,
+      "loss": 0.2924,
+      "step": 3212
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.152219873150106e-05,
+      "loss": 0.4382,
+      "step": 3214
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.15169133192389e-05,
+      "loss": 0.2454,
+      "step": 3216
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.151162790697675e-05,
+      "loss": 0.3126,
+      "step": 3218
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.150634249471459e-05,
+      "loss": 0.5072,
+      "step": 3220
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.1501057082452434e-05,
+      "loss": 0.1802,
+      "step": 3222
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.149577167019027e-05,
+      "loss": 0.2585,
+      "step": 3224
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.149048625792812e-05,
+      "loss": 0.3149,
+      "step": 3226
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.148520084566596e-05,
+      "loss": 0.2631,
+      "step": 3228
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.147991543340381e-05,
+      "loss": 0.3179,
+      "step": 3230
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.147463002114165e-05,
+      "loss": 0.2586,
+      "step": 3232
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.14693446088795e-05,
+      "loss": 0.3983,
+      "step": 3234
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.1464059196617336e-05,
+      "loss": 0.2693,
+      "step": 3236
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.145877378435518e-05,
+      "loss": 0.1979,
+      "step": 3238
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.145348837209302e-05,
+      "loss": 0.2388,
+      "step": 3240
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.144820295983087e-05,
+      "loss": 0.361,
+      "step": 3242
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.1442917547568713e-05,
+      "loss": 0.2067,
+      "step": 3244
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.143763213530655e-05,
+      "loss": 0.2476,
+      "step": 3246
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.14323467230444e-05,
+      "loss": 0.4077,
+      "step": 3248
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.1427061310782245e-05,
+      "loss": 0.255,
+      "step": 3250
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.142177589852009e-05,
+      "loss": 0.3061,
+      "step": 3252
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.141649048625793e-05,
+      "loss": 0.6814,
+      "step": 3254
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.1411205073995776e-05,
+      "loss": 0.3628,
+      "step": 3256
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.1405919661733616e-05,
+      "loss": 0.2991,
+      "step": 3258
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.140063424947146e-05,
+      "loss": 0.3783,
+      "step": 3260
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.13953488372093e-05,
+      "loss": 0.2028,
+      "step": 3262
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.139006342494715e-05,
+      "loss": 0.2912,
+      "step": 3264
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.138477801268499e-05,
+      "loss": 0.4558,
+      "step": 3266
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.137949260042284e-05,
+      "loss": 0.4464,
+      "step": 3268
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.137420718816068e-05,
+      "loss": 0.5572,
+      "step": 3270
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.1368921775898524e-05,
+      "loss": 0.3313,
+      "step": 3272
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.1363636363636364e-05,
+      "loss": 0.27,
+      "step": 3274
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.135835095137421e-05,
+      "loss": 0.5401,
+      "step": 3276
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.135306553911205e-05,
+      "loss": 0.2565,
+      "step": 3278
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.1347780126849895e-05,
+      "loss": 0.3559,
+      "step": 3280
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.1342494714587734e-05,
+      "loss": 0.3827,
+      "step": 3282
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.133720930232559e-05,
+      "loss": 0.2983,
+      "step": 3284
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.1331923890063427e-05,
+      "loss": 0.4047,
+      "step": 3286
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.132663847780127e-05,
+      "loss": 0.4661,
+      "step": 3288
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.132135306553911e-05,
+      "loss": 0.623,
+      "step": 3290
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.131606765327696e-05,
+      "loss": 0.3762,
+      "step": 3292
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.13107822410148e-05,
+      "loss": 0.3287,
+      "step": 3294
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.130549682875264e-05,
+      "loss": 0.5006,
+      "step": 3296
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.130021141649049e-05,
+      "loss": 0.3024,
+      "step": 3298
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.129492600422833e-05,
+      "loss": 0.2122,
+      "step": 3300
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.1289640591966175e-05,
+      "loss": 0.7112,
+      "step": 3302
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.128435517970402e-05,
+      "loss": 0.1781,
+      "step": 3304
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.127906976744187e-05,
+      "loss": 0.3929,
+      "step": 3306
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.1273784355179706e-05,
+      "loss": 0.3355,
+      "step": 3308
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.126849894291755e-05,
+      "loss": 0.3383,
+      "step": 3310
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.126321353065539e-05,
+      "loss": 0.3993,
+      "step": 3312
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.125792811839324e-05,
+      "loss": 0.2688,
+      "step": 3314
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.125264270613108e-05,
+      "loss": 0.2217,
+      "step": 3316
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.124735729386892e-05,
+      "loss": 0.262,
+      "step": 3318
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.124207188160677e-05,
+      "loss": 0.3199,
+      "step": 3320
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.1236786469344615e-05,
+      "loss": 0.2936,
+      "step": 3322
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.1231501057082454e-05,
+      "loss": 0.2725,
+      "step": 3324
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.12262156448203e-05,
+      "loss": 0.2933,
+      "step": 3326
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.122093023255814e-05,
+      "loss": 0.392,
+      "step": 3328
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.1215644820295986e-05,
+      "loss": 0.2794,
+      "step": 3330
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.1210359408033825e-05,
+      "loss": 0.2913,
+      "step": 3332
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.120507399577167e-05,
+      "loss": 0.3144,
+      "step": 3334
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.119978858350951e-05,
+      "loss": 0.2582,
+      "step": 3336
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.119450317124736e-05,
+      "loss": 0.2666,
+      "step": 3338
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.11892177589852e-05,
+      "loss": 0.3842,
+      "step": 3340
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.118393234672305e-05,
+      "loss": 0.5303,
+      "step": 3342
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.117864693446089e-05,
+      "loss": 0.3226,
+      "step": 3344
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1173361522198734e-05,
+      "loss": 0.2795,
+      "step": 3346
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.116807610993657e-05,
+      "loss": 0.3647,
+      "step": 3348
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.116279069767442e-05,
+      "loss": 0.3534,
+      "step": 3350
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1157505285412265e-05,
+      "loss": 0.4055,
+      "step": 3352
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1152219873150104e-05,
+      "loss": 0.5529,
+      "step": 3354
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.114693446088795e-05,
+      "loss": 0.5072,
+      "step": 3356
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1141649048625797e-05,
+      "loss": 0.2841,
+      "step": 3358
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.113636363636364e-05,
+      "loss": 0.3891,
+      "step": 3360
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.113107822410148e-05,
+      "loss": 0.3782,
+      "step": 3362
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.112579281183933e-05,
+      "loss": 0.27,
+      "step": 3364
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.112050739957717e-05,
+      "loss": 0.2478,
+      "step": 3366
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.111522198731501e-05,
+      "loss": 0.3285,
+      "step": 3368
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.110993657505285e-05,
+      "loss": 0.3204,
+      "step": 3370
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.11046511627907e-05,
+      "loss": 0.3976,
+      "step": 3372
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.1099365750528545e-05,
+      "loss": 0.3353,
+      "step": 3374
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.109408033826639e-05,
+      "loss": 0.5545,
+      "step": 3376
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.108879492600423e-05,
+      "loss": 0.3212,
+      "step": 3378
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.1083509513742076e-05,
+      "loss": 0.4366,
+      "step": 3380
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.1078224101479915e-05,
+      "loss": 0.2358,
+      "step": 3382
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.107293868921776e-05,
+      "loss": 0.285,
+      "step": 3384
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.10676532769556e-05,
+      "loss": 0.2965,
+      "step": 3386
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.106236786469345e-05,
+      "loss": 0.2111,
+      "step": 3388
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.1057082452431286e-05,
+      "loss": 0.2055,
+      "step": 3390
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.105179704016914e-05,
+      "loss": 0.3141,
+      "step": 3392
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.104651162790698e-05,
+      "loss": 0.2548,
+      "step": 3394
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.1041226215644824e-05,
+      "loss": 0.1756,
+      "step": 3396
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.1035940803382664e-05,
+      "loss": 0.1975,
+      "step": 3398
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.103065539112051e-05,
+      "loss": 0.4897,
+      "step": 3400
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.1025369978858356e-05,
+      "loss": 0.243,
+      "step": 3402
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.1020084566596195e-05,
+      "loss": 0.2601,
+      "step": 3404
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.101479915433404e-05,
+      "loss": 0.3227,
+      "step": 3406
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.100951374207188e-05,
+      "loss": 0.3445,
+      "step": 3408
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.1004228329809726e-05,
+      "loss": 0.2215,
+      "step": 3410
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.099894291754757e-05,
+      "loss": 0.2449,
+      "step": 3412
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.099365750528542e-05,
+      "loss": 0.3551,
+      "step": 3414
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.098837209302326e-05,
+      "loss": 0.1306,
+      "step": 3416
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.0983086680761104e-05,
+      "loss": 0.2045,
+      "step": 3418
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.097780126849894e-05,
+      "loss": 0.1964,
+      "step": 3420
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.097251585623679e-05,
+      "loss": 0.3248,
+      "step": 3422
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.096723044397463e-05,
+      "loss": 0.4956,
+      "step": 3424
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.0961945031712474e-05,
+      "loss": 0.2128,
+      "step": 3426
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.095665961945032e-05,
+      "loss": 0.354,
+      "step": 3428
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.0951374207188167e-05,
+      "loss": 0.2535,
+      "step": 3430
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.0946088794926006e-05,
+      "loss": 0.3954,
+      "step": 3432
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.094080338266385e-05,
+      "loss": 0.4083,
+      "step": 3434
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.093551797040169e-05,
+      "loss": 0.2527,
+      "step": 3436
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.093023255813954e-05,
+      "loss": 0.2535,
+      "step": 3438
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.0924947145877377e-05,
+      "loss": 0.261,
+      "step": 3440
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.091966173361522e-05,
+      "loss": 0.4259,
+      "step": 3442
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.091437632135306e-05,
+      "loss": 0.4179,
+      "step": 3444
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.0909090909090915e-05,
+      "loss": 0.2675,
+      "step": 3446
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.0903805496828754e-05,
+      "loss": 0.4809,
+      "step": 3448
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.08985200845666e-05,
+      "loss": 0.2123,
+      "step": 3450
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.089323467230444e-05,
+      "loss": 0.1502,
+      "step": 3452
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.0887949260042285e-05,
+      "loss": 0.2021,
+      "step": 3454
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.088266384778013e-05,
+      "loss": 0.296,
+      "step": 3456
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.087737843551797e-05,
+      "loss": 0.3827,
+      "step": 3458
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.087209302325582e-05,
+      "loss": 0.3757,
+      "step": 3460
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.0866807610993656e-05,
+      "loss": 0.2955,
+      "step": 3462
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.08615221987315e-05,
+      "loss": 0.2052,
+      "step": 3464
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.085623678646935e-05,
+      "loss": 0.211,
+      "step": 3466
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.0850951374207194e-05,
+      "loss": 0.3555,
+      "step": 3468
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.0845665961945034e-05,
+      "loss": 0.3841,
+      "step": 3470
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.084038054968288e-05,
+      "loss": 0.3611,
+      "step": 3472
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.083509513742072e-05,
+      "loss": 0.4419,
+      "step": 3474
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.0829809725158565e-05,
+      "loss": 0.5077,
+      "step": 3476
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.0824524312896404e-05,
+      "loss": 0.2802,
+      "step": 3478
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.081923890063425e-05,
+      "loss": 0.5795,
+      "step": 3480
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.0813953488372096e-05,
+      "loss": 0.2565,
+      "step": 3482
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.080866807610994e-05,
+      "loss": 0.2815,
+      "step": 3484
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.080338266384778e-05,
+      "loss": 0.5439,
+      "step": 3486
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.079809725158563e-05,
+      "loss": 0.2885,
+      "step": 3488
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.079281183932347e-05,
+      "loss": 0.295,
+      "step": 3490
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.078752642706131e-05,
+      "loss": 0.2716,
+      "step": 3492
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.078224101479915e-05,
+      "loss": 0.423,
+      "step": 3494
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.0776955602537e-05,
+      "loss": 0.229,
+      "step": 3496
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.077167019027484e-05,
+      "loss": 0.3066,
+      "step": 3498
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.076638477801269e-05,
+      "loss": 0.2746,
+      "step": 3500
+    },
+    {
+      "epoch": 3.7,
+      "eval_cer": 0.031917925334853235,
+      "eval_loss": 0.4219053387641907,
+      "eval_runtime": 130.4096,
+      "eval_samples_per_second": 6.449,
+      "eval_steps_per_second": 0.813,
+      "step": 3500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.076109936575053e-05,
+      "loss": 0.4058,
+      "step": 3502
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.0755813953488376e-05,
+      "loss": 0.1374,
+      "step": 3504
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.0750528541226215e-05,
+      "loss": 0.1596,
+      "step": 3506
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.074524312896406e-05,
+      "loss": 0.3841,
+      "step": 3508
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.073995771670191e-05,
+      "loss": 0.326,
+      "step": 3510
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.073467230443975e-05,
+      "loss": 0.2823,
+      "step": 3512
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.072938689217759e-05,
+      "loss": 0.3303,
+      "step": 3514
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.072410147991543e-05,
+      "loss": 0.3175,
+      "step": 3516
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.0718816067653285e-05,
+      "loss": 0.2827,
+      "step": 3518
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.0713530655391124e-05,
+      "loss": 0.3925,
+      "step": 3520
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.070824524312897e-05,
+      "loss": 0.315,
+      "step": 3522
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.070295983086681e-05,
+      "loss": 0.318,
+      "step": 3524
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.0697674418604655e-05,
+      "loss": 0.3343,
+      "step": 3526
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.0692389006342495e-05,
+      "loss": 0.2611,
+      "step": 3528
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.068710359408034e-05,
+      "loss": 0.1885,
+      "step": 3530
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.068181818181818e-05,
+      "loss": 0.4439,
+      "step": 3532
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.0676532769556026e-05,
+      "loss": 0.1693,
+      "step": 3534
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.067124735729387e-05,
+      "loss": 0.2422,
+      "step": 3536
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.066596194503172e-05,
+      "loss": 0.2957,
+      "step": 3538
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.066067653276956e-05,
+      "loss": 0.4195,
+      "step": 3540
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.0655391120507404e-05,
+      "loss": 0.2689,
+      "step": 3542
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.065010570824524e-05,
+      "loss": 0.2917,
+      "step": 3544
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.064482029598309e-05,
+      "loss": 0.1926,
+      "step": 3546
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.063953488372093e-05,
+      "loss": 0.2167,
+      "step": 3548
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.0634249471458774e-05,
+      "loss": 0.2325,
+      "step": 3550
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.0628964059196614e-05,
+      "loss": 0.2004,
+      "step": 3552
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.0623678646934466e-05,
+      "loss": 0.2498,
+      "step": 3554
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.0618393234672306e-05,
+      "loss": 0.2831,
+      "step": 3556
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.061310782241015e-05,
+      "loss": 0.2611,
+      "step": 3558
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.060782241014799e-05,
+      "loss": 0.2673,
+      "step": 3560
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.060253699788584e-05,
+      "loss": 0.1925,
+      "step": 3562
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.059725158562368e-05,
+      "loss": 0.2026,
+      "step": 3564
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.059196617336152e-05,
+      "loss": 0.2491,
+      "step": 3566
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.058668076109937e-05,
+      "loss": 0.1843,
+      "step": 3568
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.058139534883721e-05,
+      "loss": 0.2792,
+      "step": 3570
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.057610993657506e-05,
+      "loss": 0.3438,
+      "step": 3572
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.05708245243129e-05,
+      "loss": 0.2346,
+      "step": 3574
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.0565539112050746e-05,
+      "loss": 0.2819,
+      "step": 3576
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.0560253699788585e-05,
+      "loss": 0.2978,
+      "step": 3578
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.055496828752643e-05,
+      "loss": 0.4547,
+      "step": 3580
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.054968287526427e-05,
+      "loss": 0.3207,
+      "step": 3582
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.054439746300212e-05,
+      "loss": 0.2616,
+      "step": 3584
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.0539112050739956e-05,
+      "loss": 0.3679,
+      "step": 3586
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.05338266384778e-05,
+      "loss": 0.2208,
+      "step": 3588
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.052854122621565e-05,
+      "loss": 0.2534,
+      "step": 3590
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.0523255813953494e-05,
+      "loss": 0.2538,
+      "step": 3592
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.0517970401691333e-05,
+      "loss": 0.3186,
+      "step": 3594
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.051268498942918e-05,
+      "loss": 0.216,
+      "step": 3596
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.050739957716702e-05,
+      "loss": 0.252,
+      "step": 3598
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0502114164904865e-05,
+      "loss": 0.3972,
+      "step": 3600
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0496828752642704e-05,
+      "loss": 0.3036,
+      "step": 3602
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.049154334038055e-05,
+      "loss": 0.3883,
+      "step": 3604
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0486257928118396e-05,
+      "loss": 0.2229,
+      "step": 3606
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.048097251585624e-05,
+      "loss": 0.359,
+      "step": 3608
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.047568710359408e-05,
+      "loss": 0.2468,
+      "step": 3610
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.047040169133193e-05,
+      "loss": 0.2458,
+      "step": 3612
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.046511627906977e-05,
+      "loss": 0.2324,
+      "step": 3614
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.045983086680761e-05,
+      "loss": 0.2602,
+      "step": 3616
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.045454545454546e-05,
+      "loss": 0.2921,
+      "step": 3618
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.04492600422833e-05,
+      "loss": 0.3894,
+      "step": 3620
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.0443974630021144e-05,
+      "loss": 0.3837,
+      "step": 3622
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.0438689217758984e-05,
+      "loss": 0.1566,
+      "step": 3624
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.0433403805496836e-05,
+      "loss": 0.2718,
+      "step": 3626
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.0428118393234676e-05,
+      "loss": 0.2338,
+      "step": 3628
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.042283298097252e-05,
+      "loss": 0.2583,
+      "step": 3630
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.041754756871036e-05,
+      "loss": 0.4098,
+      "step": 3632
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.041226215644821e-05,
+      "loss": 0.3291,
+      "step": 3634
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.0406976744186046e-05,
+      "loss": 0.3875,
+      "step": 3636
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.040169133192389e-05,
+      "loss": 0.3359,
+      "step": 3638
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.039640591966173e-05,
+      "loss": 0.545,
+      "step": 3640
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.039112050739958e-05,
+      "loss": 0.2262,
+      "step": 3642
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.0385835095137424e-05,
+      "loss": 0.2983,
+      "step": 3644
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.038054968287527e-05,
+      "loss": 0.3009,
+      "step": 3646
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.037526427061311e-05,
+      "loss": 0.3872,
+      "step": 3648
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.0369978858350955e-05,
+      "loss": 0.5378,
+      "step": 3650
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.0364693446088795e-05,
+      "loss": 0.3241,
+      "step": 3652
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.035940803382664e-05,
+      "loss": 0.2517,
+      "step": 3654
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.035412262156448e-05,
+      "loss": 0.3658,
+      "step": 3656
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.0348837209302326e-05,
+      "loss": 0.2965,
+      "step": 3658
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.034355179704017e-05,
+      "loss": 0.2556,
+      "step": 3660
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.033826638477802e-05,
+      "loss": 0.2858,
+      "step": 3662
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.033298097251586e-05,
+      "loss": 0.2957,
+      "step": 3664
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.0327695560253703e-05,
+      "loss": 0.2999,
+      "step": 3666
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.032241014799155e-05,
+      "loss": 0.3098,
+      "step": 3668
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.031712473572939e-05,
+      "loss": 0.2422,
+      "step": 3670
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.0311839323467235e-05,
+      "loss": 0.2735,
+      "step": 3672
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.0306553911205074e-05,
+      "loss": 0.3498,
+      "step": 3674
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.030126849894292e-05,
+      "loss": 0.2948,
+      "step": 3676
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.029598308668076e-05,
+      "loss": 0.2371,
+      "step": 3678
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.029069767441861e-05,
+      "loss": 0.3495,
+      "step": 3680
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.028541226215645e-05,
+      "loss": 0.2426,
+      "step": 3682
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.02801268498943e-05,
+      "loss": 0.4343,
+      "step": 3684
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.027484143763214e-05,
+      "loss": 0.3005,
+      "step": 3686
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.026955602536998e-05,
+      "loss": 0.233,
+      "step": 3688
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.026427061310782e-05,
+      "loss": 0.3673,
+      "step": 3690
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.025898520084567e-05,
+      "loss": 0.1899,
+      "step": 3692
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.025369978858351e-05,
+      "loss": 0.2641,
+      "step": 3694
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.0248414376321354e-05,
+      "loss": 0.2289,
+      "step": 3696
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.02431289640592e-05,
+      "loss": 0.1489,
+      "step": 3698
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.0237843551797046e-05,
+      "loss": 0.3439,
+      "step": 3700
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.0232558139534885e-05,
+      "loss": 0.2439,
+      "step": 3702
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.022727272727273e-05,
+      "loss": 0.3292,
+      "step": 3704
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.022198731501057e-05,
+      "loss": 0.1504,
+      "step": 3706
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.0216701902748416e-05,
+      "loss": 0.3573,
+      "step": 3708
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.0211416490486256e-05,
+      "loss": 0.2399,
+      "step": 3710
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.02061310782241e-05,
+      "loss": 0.2406,
+      "step": 3712
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.020084566596195e-05,
+      "loss": 0.2586,
+      "step": 3714
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.0195560253699794e-05,
+      "loss": 0.4283,
+      "step": 3716
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.019027484143763e-05,
+      "loss": 0.2388,
+      "step": 3718
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.018498942917548e-05,
+      "loss": 0.4452,
+      "step": 3720
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.0179704016913325e-05,
+      "loss": 0.2854,
+      "step": 3722
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.0174418604651165e-05,
+      "loss": 0.2051,
+      "step": 3724
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.016913319238901e-05,
+      "loss": 0.339,
+      "step": 3726
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.016384778012685e-05,
+      "loss": 0.1278,
+      "step": 3728
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.0158562367864696e-05,
+      "loss": 0.2469,
+      "step": 3730
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.0153276955602535e-05,
+      "loss": 0.2045,
+      "step": 3732
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.014799154334038e-05,
+      "loss": 0.2847,
+      "step": 3734
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.014270613107823e-05,
+      "loss": 0.1441,
+      "step": 3736
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.0137420718816073e-05,
+      "loss": 0.3645,
+      "step": 3738
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.013213530655391e-05,
+      "loss": 0.2791,
+      "step": 3740
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.012684989429176e-05,
+      "loss": 0.3352,
+      "step": 3742
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.01215644820296e-05,
+      "loss": 0.4004,
+      "step": 3744
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.0116279069767444e-05,
+      "loss": 0.2803,
+      "step": 3746
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.0110993657505283e-05,
+      "loss": 0.3235,
+      "step": 3748
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.010570824524313e-05,
+      "loss": 0.2786,
+      "step": 3750
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.010042283298097e-05,
+      "loss": 0.277,
+      "step": 3752
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.009513742071882e-05,
+      "loss": 0.3252,
+      "step": 3754
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.008985200845666e-05,
+      "loss": 0.361,
+      "step": 3756
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.008456659619451e-05,
+      "loss": 0.3615,
+      "step": 3758
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.0079281183932346e-05,
+      "loss": 0.3047,
+      "step": 3760
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.007399577167019e-05,
+      "loss": 0.3798,
+      "step": 3762
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.006871035940803e-05,
+      "loss": 0.3093,
+      "step": 3764
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.006342494714588e-05,
+      "loss": 0.2643,
+      "step": 3766
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.0058139534883724e-05,
+      "loss": 0.4188,
+      "step": 3768
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.005285412262156e-05,
+      "loss": 0.3424,
+      "step": 3770
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.004756871035941e-05,
+      "loss": 0.2737,
+      "step": 3772
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.0042283298097255e-05,
+      "loss": 0.2672,
+      "step": 3774
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.00369978858351e-05,
+      "loss": 0.2018,
+      "step": 3776
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.003171247357294e-05,
+      "loss": 0.1879,
+      "step": 3778
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.0026427061310787e-05,
+      "loss": 0.2752,
+      "step": 3780
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.0021141649048626e-05,
+      "loss": 0.304,
+      "step": 3782
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.001585623678647e-05,
+      "loss": 0.1757,
+      "step": 3784
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.001057082452431e-05,
+      "loss": 0.2514,
+      "step": 3786
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.000528541226216e-05,
+      "loss": 0.2022,
+      "step": 3788
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4e-05,
+      "loss": 0.2014,
+      "step": 3790
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.999471458773785e-05,
+      "loss": 0.1431,
+      "step": 3792
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.998942917547569e-05,
+      "loss": 0.2986,
+      "step": 3794
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.9984143763213535e-05,
+      "loss": 0.2922,
+      "step": 3796
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.9978858350951374e-05,
+      "loss": 0.323,
+      "step": 3798
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.997357293868922e-05,
+      "loss": 0.1974,
+      "step": 3800
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.996828752642706e-05,
+      "loss": 0.2474,
+      "step": 3802
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9963002114164905e-05,
+      "loss": 0.3273,
+      "step": 3804
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9957716701902745e-05,
+      "loss": 0.2557,
+      "step": 3806
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.99524312896406e-05,
+      "loss": 0.1653,
+      "step": 3808
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.994714587737844e-05,
+      "loss": 0.2223,
+      "step": 3810
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.994186046511628e-05,
+      "loss": 0.186,
+      "step": 3812
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.993657505285412e-05,
+      "loss": 0.1906,
+      "step": 3814
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.993128964059197e-05,
+      "loss": 0.2265,
+      "step": 3816
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.992600422832981e-05,
+      "loss": 0.2011,
+      "step": 3818
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.9920718816067654e-05,
+      "loss": 0.2358,
+      "step": 3820
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.99154334038055e-05,
+      "loss": 0.1286,
+      "step": 3822
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.991014799154334e-05,
+      "loss": 0.2227,
+      "step": 3824
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.9904862579281185e-05,
+      "loss": 0.1696,
+      "step": 3826
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.989957716701903e-05,
+      "loss": 0.2811,
+      "step": 3828
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.989429175475688e-05,
+      "loss": 0.2175,
+      "step": 3830
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.9889006342494716e-05,
+      "loss": 0.5743,
+      "step": 3832
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.988372093023256e-05,
+      "loss": 0.6482,
+      "step": 3834
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.98784355179704e-05,
+      "loss": 0.4804,
+      "step": 3836
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.987315010570825e-05,
+      "loss": 0.505,
+      "step": 3838
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.986786469344609e-05,
+      "loss": 0.6236,
+      "step": 3840
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.986257928118393e-05,
+      "loss": 0.3889,
+      "step": 3842
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.985729386892178e-05,
+      "loss": 0.1109,
+      "step": 3844
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.9852008456659625e-05,
+      "loss": 0.4023,
+      "step": 3846
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.9846723044397464e-05,
+      "loss": 0.4235,
+      "step": 3848
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.984143763213531e-05,
+      "loss": 0.3534,
+      "step": 3850
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.983615221987315e-05,
+      "loss": 0.3166,
+      "step": 3852
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.9830866807610996e-05,
+      "loss": 0.2571,
+      "step": 3854
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.9825581395348835e-05,
+      "loss": 0.4292,
+      "step": 3856
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.982029598308668e-05,
+      "loss": 0.3876,
+      "step": 3858
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.981501057082452e-05,
+      "loss": 0.4129,
+      "step": 3860
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.980972515856237e-05,
+      "loss": 0.1996,
+      "step": 3862
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.980443974630021e-05,
+      "loss": 0.2858,
+      "step": 3864
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.979915433403806e-05,
+      "loss": 0.33,
+      "step": 3866
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.97938689217759e-05,
+      "loss": 0.1376,
+      "step": 3868
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.9788583509513744e-05,
+      "loss": 0.1673,
+      "step": 3870
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.978329809725159e-05,
+      "loss": 0.172,
+      "step": 3872
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.977801268498943e-05,
+      "loss": 0.2426,
+      "step": 3874
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.9772727272727275e-05,
+      "loss": 0.3028,
+      "step": 3876
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.9767441860465115e-05,
+      "loss": 0.2281,
+      "step": 3878
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.976215644820296e-05,
+      "loss": 0.1447,
+      "step": 3880
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.975687103594081e-05,
+      "loss": 0.2487,
+      "step": 3882
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.975158562367865e-05,
+      "loss": 0.1818,
+      "step": 3884
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.974630021141649e-05,
+      "loss": 0.1965,
+      "step": 3886
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.974101479915434e-05,
+      "loss": 0.2904,
+      "step": 3888
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.973572938689218e-05,
+      "loss": 0.3242,
+      "step": 3890
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.9730443974630024e-05,
+      "loss": 0.263,
+      "step": 3892
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.972515856236786e-05,
+      "loss": 0.1676,
+      "step": 3894
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.971987315010571e-05,
+      "loss": 0.1705,
+      "step": 3896
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.9714587737843555e-05,
+      "loss": 0.2202,
+      "step": 3898
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.97093023255814e-05,
+      "loss": 0.2569,
+      "step": 3900
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.970401691331924e-05,
+      "loss": 0.1988,
+      "step": 3902
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.9698731501057086e-05,
+      "loss": 0.3049,
+      "step": 3904
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.9693446088794926e-05,
+      "loss": 0.1223,
+      "step": 3906
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.968816067653277e-05,
+      "loss": 0.2061,
+      "step": 3908
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.968287526427061e-05,
+      "loss": 0.1779,
+      "step": 3910
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.967758985200846e-05,
+      "loss": 0.2722,
+      "step": 3912
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.9672304439746296e-05,
+      "loss": 0.2599,
+      "step": 3914
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.966701902748415e-05,
+      "loss": 0.1343,
+      "step": 3916
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.966173361522199e-05,
+      "loss": 0.2221,
+      "step": 3918
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.9656448202959834e-05,
+      "loss": 0.2081,
+      "step": 3920
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.9651162790697674e-05,
+      "loss": 0.2064,
+      "step": 3922
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.964587737843552e-05,
+      "loss": 0.2185,
+      "step": 3924
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.9640591966173366e-05,
+      "loss": 0.2235,
+      "step": 3926
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.9635306553911205e-05,
+      "loss": 0.205,
+      "step": 3928
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.963002114164905e-05,
+      "loss": 0.1873,
+      "step": 3930
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.962473572938689e-05,
+      "loss": 0.1276,
+      "step": 3932
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.961945031712474e-05,
+      "loss": 0.2165,
+      "step": 3934
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.961416490486258e-05,
+      "loss": 0.1211,
+      "step": 3936
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.960887949260043e-05,
+      "loss": 0.1964,
+      "step": 3938
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.960359408033827e-05,
+      "loss": 0.3566,
+      "step": 3940
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.9598308668076114e-05,
+      "loss": 0.2372,
+      "step": 3942
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.959302325581395e-05,
+      "loss": 0.2184,
+      "step": 3944
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.95877378435518e-05,
+      "loss": 0.1324,
+      "step": 3946
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.958245243128964e-05,
+      "loss": 0.1911,
+      "step": 3948
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.9577167019027485e-05,
+      "loss": 0.3218,
+      "step": 3950
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.957188160676533e-05,
+      "loss": 0.1894,
+      "step": 3952
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.956659619450318e-05,
+      "loss": 0.323,
+      "step": 3954
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.9561310782241016e-05,
+      "loss": 0.3028,
+      "step": 3956
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.955602536997886e-05,
+      "loss": 0.2063,
+      "step": 3958
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.95507399577167e-05,
+      "loss": 0.3204,
+      "step": 3960
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.954545454545455e-05,
+      "loss": 0.1358,
+      "step": 3962
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.954016913319239e-05,
+      "loss": 0.3048,
+      "step": 3964
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.953488372093023e-05,
+      "loss": 0.2386,
+      "step": 3966
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.952959830866807e-05,
+      "loss": 0.2359,
+      "step": 3968
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.9524312896405925e-05,
+      "loss": 0.1386,
+      "step": 3970
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.9519027484143764e-05,
+      "loss": 0.1509,
+      "step": 3972
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.951374207188161e-05,
+      "loss": 0.1746,
+      "step": 3974
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.950845665961945e-05,
+      "loss": 0.2711,
+      "step": 3976
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.9503171247357296e-05,
+      "loss": 0.2591,
+      "step": 3978
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.949788583509514e-05,
+      "loss": 0.3,
+      "step": 3980
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.949260042283298e-05,
+      "loss": 0.2035,
+      "step": 3982
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.948731501057083e-05,
+      "loss": 0.2351,
+      "step": 3984
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.9482029598308666e-05,
+      "loss": 0.1863,
+      "step": 3986
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.947674418604652e-05,
+      "loss": 0.1715,
+      "step": 3988
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.947145877378436e-05,
+      "loss": 0.1884,
+      "step": 3990
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.9466173361522205e-05,
+      "loss": 0.4106,
+      "step": 3992
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.9460887949260044e-05,
+      "loss": 0.3041,
+      "step": 3994
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.945560253699789e-05,
+      "loss": 0.22,
+      "step": 3996
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.945031712473573e-05,
+      "loss": 0.2508,
+      "step": 3998
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.9445031712473575e-05,
+      "loss": 0.1456,
+      "step": 4000
+    },
+    {
+      "epoch": 4.23,
+      "eval_cer": 0.039897406668566546,
+      "eval_loss": 0.5804613828659058,
+      "eval_runtime": 129.4688,
+      "eval_samples_per_second": 6.496,
+      "eval_steps_per_second": 0.819,
+      "step": 4000
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.9439746300211415e-05,
+      "loss": 0.2934,
+      "step": 4002
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.943446088794926e-05,
+      "loss": 0.1868,
+      "step": 4004
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.942917547568711e-05,
+      "loss": 0.2808,
+      "step": 4006
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.942389006342495e-05,
+      "loss": 0.1501,
+      "step": 4008
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.941860465116279e-05,
+      "loss": 0.241,
+      "step": 4010
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.941331923890064e-05,
+      "loss": 0.1349,
+      "step": 4012
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.940803382663848e-05,
+      "loss": 0.3111,
+      "step": 4014
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.940274841437632e-05,
+      "loss": 0.2869,
+      "step": 4016
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.939746300211416e-05,
+      "loss": 0.2255,
+      "step": 4018
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.939217758985201e-05,
+      "loss": 0.2188,
+      "step": 4020
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.9386892177589855e-05,
+      "loss": 0.1902,
+      "step": 4022
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.93816067653277e-05,
+      "loss": 0.1701,
+      "step": 4024
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.937632135306554e-05,
+      "loss": 0.1584,
+      "step": 4026
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.9371035940803386e-05,
+      "loss": 0.2237,
+      "step": 4028
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.9365750528541225e-05,
+      "loss": 0.398,
+      "step": 4030
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.936046511627907e-05,
+      "loss": 0.1038,
+      "step": 4032
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.935517970401692e-05,
+      "loss": 0.1975,
+      "step": 4034
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.934989429175476e-05,
+      "loss": 0.2825,
+      "step": 4036
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.93446088794926e-05,
+      "loss": 0.1508,
+      "step": 4038
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.933932346723044e-05,
+      "loss": 0.1505,
+      "step": 4040
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.9334038054968295e-05,
+      "loss": 0.2344,
+      "step": 4042
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.9328752642706134e-05,
+      "loss": 0.1584,
+      "step": 4044
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.932346723044398e-05,
+      "loss": 0.2375,
+      "step": 4046
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.931818181818182e-05,
+      "loss": 0.3126,
+      "step": 4048
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.9312896405919666e-05,
+      "loss": 0.2596,
+      "step": 4050
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.9307610993657505e-05,
+      "loss": 0.1258,
+      "step": 4052
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.930232558139535e-05,
+      "loss": 0.2411,
+      "step": 4054
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.929704016913319e-05,
+      "loss": 0.2466,
+      "step": 4056
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.9291754756871036e-05,
+      "loss": 0.125,
+      "step": 4058
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.928646934460888e-05,
+      "loss": 0.1596,
+      "step": 4060
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.928118393234673e-05,
+      "loss": 0.2621,
+      "step": 4062
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.927589852008457e-05,
+      "loss": 0.208,
+      "step": 4064
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.9270613107822414e-05,
+      "loss": 0.2743,
+      "step": 4066
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.926532769556025e-05,
+      "loss": 0.3313,
+      "step": 4068
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.92600422832981e-05,
+      "loss": 0.1685,
+      "step": 4070
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.925475687103594e-05,
+      "loss": 0.2687,
+      "step": 4072
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.9249471458773785e-05,
+      "loss": 0.3284,
+      "step": 4074
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.924418604651163e-05,
+      "loss": 0.2718,
+      "step": 4076
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.923890063424948e-05,
+      "loss": 0.2013,
+      "step": 4078
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.9233615221987316e-05,
+      "loss": 0.2228,
+      "step": 4080
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.922832980972516e-05,
+      "loss": 0.5322,
+      "step": 4082
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.9223044397463e-05,
+      "loss": 0.36,
+      "step": 4084
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.921775898520085e-05,
+      "loss": 0.1814,
+      "step": 4086
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.9212473572938693e-05,
+      "loss": 0.2867,
+      "step": 4088
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.920718816067653e-05,
+      "loss": 0.2947,
+      "step": 4090
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.920190274841438e-05,
+      "loss": 0.3523,
+      "step": 4092
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.919661733615222e-05,
+      "loss": 0.1549,
+      "step": 4094
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.919133192389007e-05,
+      "loss": 0.3549,
+      "step": 4096
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.918604651162791e-05,
+      "loss": 0.3409,
+      "step": 4098
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.9180761099365756e-05,
+      "loss": 0.2373,
+      "step": 4100
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.9175475687103596e-05,
+      "loss": 0.2428,
+      "step": 4102
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.917019027484144e-05,
+      "loss": 0.2634,
+      "step": 4104
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.916490486257928e-05,
+      "loss": 0.1302,
+      "step": 4106
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.915961945031713e-05,
+      "loss": 0.26,
+      "step": 4108
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.9154334038054966e-05,
+      "loss": 0.217,
+      "step": 4110
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.914904862579281e-05,
+      "loss": 0.1351,
+      "step": 4112
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.914376321353066e-05,
+      "loss": 0.2058,
+      "step": 4114
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.9138477801268504e-05,
+      "loss": 0.1843,
+      "step": 4116
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.9133192389006344e-05,
+      "loss": 0.4963,
+      "step": 4118
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.912790697674419e-05,
+      "loss": 0.2972,
+      "step": 4120
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.912262156448203e-05,
+      "loss": 0.2613,
+      "step": 4122
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.9117336152219875e-05,
+      "loss": 0.1996,
+      "step": 4124
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.9112050739957714e-05,
+      "loss": 0.3973,
+      "step": 4126
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.910676532769556e-05,
+      "loss": 0.149,
+      "step": 4128
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.9101479915433406e-05,
+      "loss": 0.1527,
+      "step": 4130
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.909619450317125e-05,
+      "loss": 0.2652,
+      "step": 4132
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.909090909090909e-05,
+      "loss": 0.2499,
+      "step": 4134
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.908562367864694e-05,
+      "loss": 0.2321,
+      "step": 4136
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.9080338266384784e-05,
+      "loss": 0.2166,
+      "step": 4138
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.907505285412262e-05,
+      "loss": 0.2361,
+      "step": 4140
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.906976744186047e-05,
+      "loss": 0.3066,
+      "step": 4142
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.906448202959831e-05,
+      "loss": 0.1623,
+      "step": 4144
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.9059196617336155e-05,
+      "loss": 0.1691,
+      "step": 4146
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.9053911205073994e-05,
+      "loss": 0.2238,
+      "step": 4148
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.904862579281185e-05,
+      "loss": 0.2006,
+      "step": 4150
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.9043340380549686e-05,
+      "loss": 0.2972,
+      "step": 4152
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.903805496828753e-05,
+      "loss": 0.3729,
+      "step": 4154
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.903276955602537e-05,
+      "loss": 0.3234,
+      "step": 4156
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.902748414376322e-05,
+      "loss": 0.1333,
+      "step": 4158
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.902219873150106e-05,
+      "loss": 0.2893,
+      "step": 4160
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.90169133192389e-05,
+      "loss": 0.1391,
+      "step": 4162
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.901162790697674e-05,
+      "loss": 0.1938,
+      "step": 4164
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.900634249471459e-05,
+      "loss": 0.1412,
+      "step": 4166
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.9001057082452434e-05,
+      "loss": 0.3682,
+      "step": 4168
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.899577167019028e-05,
+      "loss": 0.1981,
+      "step": 4170
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.899048625792812e-05,
+      "loss": 0.1542,
+      "step": 4172
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.8985200845665966e-05,
+      "loss": 0.1952,
+      "step": 4174
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.8979915433403805e-05,
+      "loss": 0.234,
+      "step": 4176
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.897463002114165e-05,
+      "loss": 0.159,
+      "step": 4178
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.896934460887949e-05,
+      "loss": 0.2722,
+      "step": 4180
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.8964059196617336e-05,
+      "loss": 0.2511,
+      "step": 4182
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.895877378435518e-05,
+      "loss": 0.2582,
+      "step": 4184
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.895348837209303e-05,
+      "loss": 0.2339,
+      "step": 4186
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.894820295983087e-05,
+      "loss": 0.1998,
+      "step": 4188
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.8942917547568714e-05,
+      "loss": 0.1863,
+      "step": 4190
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.893763213530656e-05,
+      "loss": 0.3524,
+      "step": 4192
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.89323467230444e-05,
+      "loss": 0.3165,
+      "step": 4194
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.8927061310782245e-05,
+      "loss": 0.2742,
+      "step": 4196
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.8921775898520084e-05,
+      "loss": 0.1488,
+      "step": 4198
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.891649048625793e-05,
+      "loss": 0.3678,
+      "step": 4200
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.891120507399577e-05,
+      "loss": 0.2203,
+      "step": 4202
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.890591966173362e-05,
+      "loss": 0.2065,
+      "step": 4204
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.890063424947146e-05,
+      "loss": 0.184,
+      "step": 4206
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.889534883720931e-05,
+      "loss": 0.1449,
+      "step": 4208
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.889006342494715e-05,
+      "loss": 0.2047,
+      "step": 4210
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.888477801268499e-05,
+      "loss": 0.3436,
+      "step": 4212
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.887949260042283e-05,
+      "loss": 0.2107,
+      "step": 4214
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.887420718816068e-05,
+      "loss": 0.1966,
+      "step": 4216
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.886892177589852e-05,
+      "loss": 0.2369,
+      "step": 4218
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.8863636363636364e-05,
+      "loss": 0.2596,
+      "step": 4220
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.885835095137421e-05,
+      "loss": 0.2708,
+      "step": 4222
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.8853065539112056e-05,
+      "loss": 0.3761,
+      "step": 4224
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.8847780126849895e-05,
+      "loss": 0.3693,
+      "step": 4226
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.884249471458774e-05,
+      "loss": 0.3575,
+      "step": 4228
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.883720930232558e-05,
+      "loss": 0.4008,
+      "step": 4230
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.883192389006343e-05,
+      "loss": 0.1546,
+      "step": 4232
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.8826638477801266e-05,
+      "loss": 0.2317,
+      "step": 4234
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.882135306553911e-05,
+      "loss": 0.1259,
+      "step": 4236
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.881606765327696e-05,
+      "loss": 0.2638,
+      "step": 4238
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.8810782241014804e-05,
+      "loss": 0.3114,
+      "step": 4240
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.8805496828752643e-05,
+      "loss": 0.4307,
+      "step": 4242
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.880021141649049e-05,
+      "loss": 0.2683,
+      "step": 4244
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.8794926004228336e-05,
+      "loss": 0.2929,
+      "step": 4246
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.8789640591966175e-05,
+      "loss": 0.2354,
+      "step": 4248
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.878435517970402e-05,
+      "loss": 0.1789,
+      "step": 4250
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.877906976744186e-05,
+      "loss": 0.2649,
+      "step": 4252
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.8773784355179706e-05,
+      "loss": 0.234,
+      "step": 4254
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.8768498942917546e-05,
+      "loss": 0.2865,
+      "step": 4256
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.87632135306554e-05,
+      "loss": 0.2417,
+      "step": 4258
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.875792811839324e-05,
+      "loss": 0.1494,
+      "step": 4260
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.8752642706131084e-05,
+      "loss": 0.3175,
+      "step": 4262
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.874735729386892e-05,
+      "loss": 0.3808,
+      "step": 4264
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.874207188160677e-05,
+      "loss": 0.2946,
+      "step": 4266
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.873678646934461e-05,
+      "loss": 0.3407,
+      "step": 4268
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.8731501057082454e-05,
+      "loss": 0.2248,
+      "step": 4270
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.8726215644820294e-05,
+      "loss": 0.2135,
+      "step": 4272
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.872093023255814e-05,
+      "loss": 0.2039,
+      "step": 4274
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.8715644820295986e-05,
+      "loss": 0.2309,
+      "step": 4276
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.871035940803383e-05,
+      "loss": 0.3951,
+      "step": 4278
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.870507399577167e-05,
+      "loss": 0.296,
+      "step": 4280
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.869978858350952e-05,
+      "loss": 0.2853,
+      "step": 4282
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.8694503171247357e-05,
+      "loss": 0.1644,
+      "step": 4284
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.86892177589852e-05,
+      "loss": 0.2342,
+      "step": 4286
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.868393234672304e-05,
+      "loss": 0.1685,
+      "step": 4288
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.867864693446089e-05,
+      "loss": 0.1452,
+      "step": 4290
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.8673361522198734e-05,
+      "loss": 0.2645,
+      "step": 4292
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.866807610993658e-05,
+      "loss": 0.2565,
+      "step": 4294
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.866279069767442e-05,
+      "loss": 0.0879,
+      "step": 4296
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.8657505285412265e-05,
+      "loss": 0.2067,
+      "step": 4298
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.865221987315011e-05,
+      "loss": 0.163,
+      "step": 4300
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.864693446088795e-05,
+      "loss": 0.2396,
+      "step": 4302
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.86416490486258e-05,
+      "loss": 0.1791,
+      "step": 4304
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.8636363636363636e-05,
+      "loss": 0.263,
+      "step": 4306
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.863107822410148e-05,
+      "loss": 0.2548,
+      "step": 4308
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.862579281183932e-05,
+      "loss": 0.3144,
+      "step": 4310
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.8620507399577174e-05,
+      "loss": 0.1509,
+      "step": 4312
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.8615221987315014e-05,
+      "loss": 0.254,
+      "step": 4314
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.860993657505286e-05,
+      "loss": 0.2525,
+      "step": 4316
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.86046511627907e-05,
+      "loss": 0.3747,
+      "step": 4318
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8599365750528545e-05,
+      "loss": 0.3409,
+      "step": 4320
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8594080338266384e-05,
+      "loss": 0.1191,
+      "step": 4322
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.858879492600423e-05,
+      "loss": 0.2891,
+      "step": 4324
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.858350951374207e-05,
+      "loss": 0.1214,
+      "step": 4326
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.8578224101479916e-05,
+      "loss": 0.2835,
+      "step": 4328
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.857293868921776e-05,
+      "loss": 0.2172,
+      "step": 4330
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.856765327695561e-05,
+      "loss": 0.1076,
+      "step": 4332
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.856236786469345e-05,
+      "loss": 0.2779,
+      "step": 4334
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.855708245243129e-05,
+      "loss": 0.2579,
+      "step": 4336
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.855179704016913e-05,
+      "loss": 0.2079,
+      "step": 4338
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.854651162790698e-05,
+      "loss": 0.4571,
+      "step": 4340
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.8541226215644824e-05,
+      "loss": 0.304,
+      "step": 4342
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.8535940803382664e-05,
+      "loss": 0.2778,
+      "step": 4344
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.853065539112051e-05,
+      "loss": 0.1885,
+      "step": 4346
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.8525369978858356e-05,
+      "loss": 0.2225,
+      "step": 4348
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.8520084566596195e-05,
+      "loss": 0.2183,
+      "step": 4350
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.851479915433404e-05,
+      "loss": 0.2757,
+      "step": 4352
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.850951374207189e-05,
+      "loss": 0.4415,
+      "step": 4354
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.8504228329809727e-05,
+      "loss": 0.2638,
+      "step": 4356
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.849894291754757e-05,
+      "loss": 0.2034,
+      "step": 4358
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.849365750528541e-05,
+      "loss": 0.261,
+      "step": 4360
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.848837209302326e-05,
+      "loss": 0.2194,
+      "step": 4362
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.84830866807611e-05,
+      "loss": 0.2475,
+      "step": 4364
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.847780126849895e-05,
+      "loss": 0.289,
+      "step": 4366
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.847251585623679e-05,
+      "loss": 0.1766,
+      "step": 4368
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.8467230443974635e-05,
+      "loss": 0.2849,
+      "step": 4370
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.8461945031712475e-05,
+      "loss": 0.2087,
+      "step": 4372
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.845665961945032e-05,
+      "loss": 0.1763,
+      "step": 4374
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.845137420718816e-05,
+      "loss": 0.2629,
+      "step": 4376
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.8446088794926006e-05,
+      "loss": 0.2354,
+      "step": 4378
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.8440803382663845e-05,
+      "loss": 0.2312,
+      "step": 4380
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.843551797040169e-05,
+      "loss": 0.2316,
+      "step": 4382
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.843023255813954e-05,
+      "loss": 0.3674,
+      "step": 4384
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.8424947145877384e-05,
+      "loss": 0.1878,
+      "step": 4386
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.841966173361522e-05,
+      "loss": 0.1896,
+      "step": 4388
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.841437632135307e-05,
+      "loss": 0.3298,
+      "step": 4390
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.840909090909091e-05,
+      "loss": 0.1981,
+      "step": 4392
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.8403805496828754e-05,
+      "loss": 0.2604,
+      "step": 4394
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.83985200845666e-05,
+      "loss": 0.1946,
+      "step": 4396
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.839323467230444e-05,
+      "loss": 0.2386,
+      "step": 4398
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.8387949260042286e-05,
+      "loss": 0.1483,
+      "step": 4400
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.8382663847780125e-05,
+      "loss": 0.2855,
+      "step": 4402
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.837737843551798e-05,
+      "loss": 0.186,
+      "step": 4404
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.837209302325582e-05,
+      "loss": 0.215,
+      "step": 4406
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.836680761099366e-05,
+      "loss": 0.1128,
+      "step": 4408
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.83615221987315e-05,
+      "loss": 0.1597,
+      "step": 4410
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.835623678646935e-05,
+      "loss": 0.2013,
+      "step": 4412
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.835095137420719e-05,
+      "loss": 0.2183,
+      "step": 4414
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.8345665961945034e-05,
+      "loss": 0.3207,
+      "step": 4416
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.834038054968287e-05,
+      "loss": 0.2389,
+      "step": 4418
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.833509513742072e-05,
+      "loss": 0.3231,
+      "step": 4420
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.8329809725158565e-05,
+      "loss": 0.2078,
+      "step": 4422
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.832452431289641e-05,
+      "loss": 0.3902,
+      "step": 4424
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.831923890063425e-05,
+      "loss": 0.1898,
+      "step": 4426
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.8313953488372097e-05,
+      "loss": 0.2032,
+      "step": 4428
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.8308668076109936e-05,
+      "loss": 0.065,
+      "step": 4430
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.830338266384778e-05,
+      "loss": 0.197,
+      "step": 4432
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.829809725158562e-05,
+      "loss": 0.244,
+      "step": 4434
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.829281183932347e-05,
+      "loss": 0.3045,
+      "step": 4436
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.8287526427061307e-05,
+      "loss": 0.2582,
+      "step": 4438
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.828224101479916e-05,
+      "loss": 0.2491,
+      "step": 4440
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.8276955602537e-05,
+      "loss": 0.2893,
+      "step": 4442
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.8271670190274845e-05,
+      "loss": 0.2808,
+      "step": 4444
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.8266384778012684e-05,
+      "loss": 0.2402,
+      "step": 4446
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.826109936575053e-05,
+      "loss": 0.328,
+      "step": 4448
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.8255813953488376e-05,
+      "loss": 0.1825,
+      "step": 4450
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.8250528541226215e-05,
+      "loss": 0.2451,
+      "step": 4452
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.824524312896406e-05,
+      "loss": 0.2322,
+      "step": 4454
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.82399577167019e-05,
+      "loss": 0.239,
+      "step": 4456
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.8234672304439754e-05,
+      "loss": 0.4527,
+      "step": 4458
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.822938689217759e-05,
+      "loss": 0.3426,
+      "step": 4460
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.822410147991544e-05,
+      "loss": 0.1954,
+      "step": 4462
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.821881606765328e-05,
+      "loss": 0.2278,
+      "step": 4464
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.8213530655391124e-05,
+      "loss": 0.1666,
+      "step": 4466
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.8208245243128964e-05,
+      "loss": 0.137,
+      "step": 4468
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.820295983086681e-05,
+      "loss": 0.292,
+      "step": 4470
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.819767441860465e-05,
+      "loss": 0.2981,
+      "step": 4472
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.8192389006342495e-05,
+      "loss": 0.3711,
+      "step": 4474
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.818710359408034e-05,
+      "loss": 0.2092,
+      "step": 4476
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.818181818181819e-05,
+      "loss": 0.2499,
+      "step": 4478
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.8176532769556026e-05,
+      "loss": 0.2701,
+      "step": 4480
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.817124735729387e-05,
+      "loss": 0.1771,
+      "step": 4482
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.816596194503171e-05,
+      "loss": 0.2258,
+      "step": 4484
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.816067653276956e-05,
+      "loss": 0.4113,
+      "step": 4486
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.81553911205074e-05,
+      "loss": 0.1089,
+      "step": 4488
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.815010570824524e-05,
+      "loss": 0.1204,
+      "step": 4490
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.814482029598309e-05,
+      "loss": 0.1931,
+      "step": 4492
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.8139534883720935e-05,
+      "loss": 0.1632,
+      "step": 4494
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.8134249471458775e-05,
+      "loss": 0.2525,
+      "step": 4496
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.812896405919662e-05,
+      "loss": 0.1257,
+      "step": 4498
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.812367864693446e-05,
+      "loss": 0.3264,
+      "step": 4500
+    },
+    {
+      "epoch": 4.76,
+      "eval_cer": 0.030265032772869762,
+      "eval_loss": 0.3518345057964325,
+      "eval_runtime": 127.9997,
+      "eval_samples_per_second": 6.57,
+      "eval_steps_per_second": 0.828,
+      "step": 4500
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.8118393234672306e-05,
+      "loss": 0.1576,
+      "step": 4502
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.811310782241015e-05,
+      "loss": 0.1078,
+      "step": 4504
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.810782241014799e-05,
+      "loss": 0.2641,
+      "step": 4506
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.810253699788584e-05,
+      "loss": 0.2368,
+      "step": 4508
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.809725158562368e-05,
+      "loss": 0.218,
+      "step": 4510
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.809196617336153e-05,
+      "loss": 0.3123,
+      "step": 4512
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.808668076109937e-05,
+      "loss": 0.1478,
+      "step": 4514
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.8081395348837215e-05,
+      "loss": 0.2465,
+      "step": 4516
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.8076109936575054e-05,
+      "loss": 0.1555,
+      "step": 4518
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.80708245243129e-05,
+      "loss": 0.139,
+      "step": 4520
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.806553911205074e-05,
+      "loss": 0.2661,
+      "step": 4522
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.8060253699788585e-05,
+      "loss": 0.1964,
+      "step": 4524
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.8054968287526425e-05,
+      "loss": 0.1876,
+      "step": 4526
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.804968287526427e-05,
+      "loss": 0.2145,
+      "step": 4528
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.804439746300212e-05,
+      "loss": 0.2774,
+      "step": 4530
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.803911205073996e-05,
+      "loss": 0.1954,
+      "step": 4532
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.80338266384778e-05,
+      "loss": 0.2758,
+      "step": 4534
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.802854122621565e-05,
+      "loss": 0.2895,
+      "step": 4536
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.802325581395349e-05,
+      "loss": 0.2015,
+      "step": 4538
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.8017970401691334e-05,
+      "loss": 0.1902,
+      "step": 4540
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.801268498942917e-05,
+      "loss": 0.2945,
+      "step": 4542
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.800739957716702e-05,
+      "loss": 0.3151,
+      "step": 4544
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.8002114164904865e-05,
+      "loss": 0.2132,
+      "step": 4546
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.799682875264271e-05,
+      "loss": 0.1585,
+      "step": 4548
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.799154334038055e-05,
+      "loss": 0.3348,
+      "step": 4550
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.7986257928118396e-05,
+      "loss": 0.1911,
+      "step": 4552
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.7980972515856236e-05,
+      "loss": 0.2412,
+      "step": 4554
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.797568710359408e-05,
+      "loss": 0.2531,
+      "step": 4556
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.797040169133193e-05,
+      "loss": 0.2128,
+      "step": 4558
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.796511627906977e-05,
+      "loss": 0.288,
+      "step": 4560
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.795983086680761e-05,
+      "loss": 0.2211,
+      "step": 4562
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.795454545454545e-05,
+      "loss": 0.1853,
+      "step": 4564
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.7949260042283305e-05,
+      "loss": 0.2396,
+      "step": 4566
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.7943974630021145e-05,
+      "loss": 0.163,
+      "step": 4568
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.793868921775899e-05,
+      "loss": 0.2536,
+      "step": 4570
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.793340380549683e-05,
+      "loss": 0.2265,
+      "step": 4572
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.7928118393234676e-05,
+      "loss": 0.1727,
+      "step": 4574
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.7922832980972515e-05,
+      "loss": 0.1687,
+      "step": 4576
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.791754756871036e-05,
+      "loss": 0.3511,
+      "step": 4578
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.79122621564482e-05,
+      "loss": 0.1654,
+      "step": 4580
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.790697674418605e-05,
+      "loss": 0.2532,
+      "step": 4582
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.790169133192389e-05,
+      "loss": 0.2507,
+      "step": 4584
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.789640591966174e-05,
+      "loss": 0.1839,
+      "step": 4586
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.789112050739958e-05,
+      "loss": 0.2172,
+      "step": 4588
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.7885835095137424e-05,
+      "loss": 0.1444,
+      "step": 4590
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.7880549682875263e-05,
+      "loss": 0.1246,
+      "step": 4592
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.787526427061311e-05,
+      "loss": 0.1763,
+      "step": 4594
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.786997885835095e-05,
+      "loss": 0.2812,
+      "step": 4596
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.7864693446088795e-05,
+      "loss": 0.2248,
+      "step": 4598
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.785940803382664e-05,
+      "loss": 0.137,
+      "step": 4600
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.785412262156449e-05,
+      "loss": 0.3015,
+      "step": 4602
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7848837209302326e-05,
+      "loss": 0.1787,
+      "step": 4604
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.784355179704017e-05,
+      "loss": 0.2139,
+      "step": 4606
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.783826638477802e-05,
+      "loss": 0.1668,
+      "step": 4608
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.783298097251586e-05,
+      "loss": 0.215,
+      "step": 4610
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.7827695560253704e-05,
+      "loss": 0.1996,
+      "step": 4612
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.782241014799154e-05,
+      "loss": 0.208,
+      "step": 4614
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.781712473572939e-05,
+      "loss": 0.3617,
+      "step": 4616
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.781183932346723e-05,
+      "loss": 0.2004,
+      "step": 4618
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.780655391120508e-05,
+      "loss": 0.2921,
+      "step": 4620
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.780126849894292e-05,
+      "loss": 0.1529,
+      "step": 4622
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.7795983086680766e-05,
+      "loss": 0.1936,
+      "step": 4624
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.7790697674418606e-05,
+      "loss": 0.1113,
+      "step": 4626
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.778541226215645e-05,
+      "loss": 0.3297,
+      "step": 4628
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.778012684989429e-05,
+      "loss": 0.2202,
+      "step": 4630
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.777484143763214e-05,
+      "loss": 0.1229,
+      "step": 4632
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.7769556025369976e-05,
+      "loss": 0.1535,
+      "step": 4634
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.776427061310782e-05,
+      "loss": 0.2717,
+      "step": 4636
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.775898520084567e-05,
+      "loss": 0.2224,
+      "step": 4638
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.7753699788583515e-05,
+      "loss": 0.1491,
+      "step": 4640
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.7748414376321354e-05,
+      "loss": 0.3249,
+      "step": 4642
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.77431289640592e-05,
+      "loss": 0.1969,
+      "step": 4644
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.773784355179704e-05,
+      "loss": 0.1595,
+      "step": 4646
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.7732558139534885e-05,
+      "loss": 0.2788,
+      "step": 4648
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.7727272727272725e-05,
+      "loss": 0.1207,
+      "step": 4650
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.772198731501057e-05,
+      "loss": 0.1545,
+      "step": 4652
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.771670190274842e-05,
+      "loss": 0.3822,
+      "step": 4654
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.771141649048626e-05,
+      "loss": 0.2045,
+      "step": 4656
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.77061310782241e-05,
+      "loss": 0.1368,
+      "step": 4658
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.770084566596195e-05,
+      "loss": 0.1766,
+      "step": 4660
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.7695560253699794e-05,
+      "loss": 0.6979,
+      "step": 4662
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.7690274841437633e-05,
+      "loss": 0.1898,
+      "step": 4664
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.768498942917548e-05,
+      "loss": 0.205,
+      "step": 4666
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.767970401691332e-05,
+      "loss": 0.2457,
+      "step": 4668
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.7674418604651165e-05,
+      "loss": 0.2407,
+      "step": 4670
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.7669133192389004e-05,
+      "loss": 0.219,
+      "step": 4672
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.766384778012686e-05,
+      "loss": 0.1399,
+      "step": 4674
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.7658562367864696e-05,
+      "loss": 0.291,
+      "step": 4676
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.765327695560254e-05,
+      "loss": 0.1945,
+      "step": 4678
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.764799154334038e-05,
+      "loss": 0.2359,
+      "step": 4680
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.764270613107823e-05,
+      "loss": 0.2426,
+      "step": 4682
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.763742071881607e-05,
+      "loss": 0.2948,
+      "step": 4684
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.763213530655391e-05,
+      "loss": 0.2388,
+      "step": 4686
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.762684989429175e-05,
+      "loss": 0.0963,
+      "step": 4688
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.76215644820296e-05,
+      "loss": 0.25,
+      "step": 4690
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.7616279069767444e-05,
+      "loss": 0.1789,
+      "step": 4692
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.761099365750529e-05,
+      "loss": 0.2834,
+      "step": 4694
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.760570824524313e-05,
+      "loss": 0.1475,
+      "step": 4696
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.7600422832980976e-05,
+      "loss": 0.1781,
+      "step": 4698
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.7595137420718815e-05,
+      "loss": 0.2353,
+      "step": 4700
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.758985200845666e-05,
+      "loss": 0.4259,
+      "step": 4702
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.75845665961945e-05,
+      "loss": 0.2206,
+      "step": 4704
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.7579281183932346e-05,
+      "loss": 0.2497,
+      "step": 4706
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.757399577167019e-05,
+      "loss": 0.2344,
+      "step": 4708
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.756871035940804e-05,
+      "loss": 0.3204,
+      "step": 4710
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.756342494714588e-05,
+      "loss": 0.3352,
+      "step": 4712
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.7558139534883724e-05,
+      "loss": 0.2001,
+      "step": 4714
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.755285412262157e-05,
+      "loss": 0.1399,
+      "step": 4716
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.754756871035941e-05,
+      "loss": 0.1997,
+      "step": 4718
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.7542283298097255e-05,
+      "loss": 0.173,
+      "step": 4720
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.7536997885835095e-05,
+      "loss": 0.3078,
+      "step": 4722
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.753171247357294e-05,
+      "loss": 0.1478,
+      "step": 4724
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.752642706131078e-05,
+      "loss": 0.218,
+      "step": 4726
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.752114164904863e-05,
+      "loss": 0.2081,
+      "step": 4728
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.751585623678647e-05,
+      "loss": 0.1173,
+      "step": 4730
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.751057082452432e-05,
+      "loss": 0.2947,
+      "step": 4732
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.750528541226216e-05,
+      "loss": 0.1826,
+      "step": 4734
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.2061,
+      "step": 4736
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.749471458773784e-05,
+      "loss": 0.2811,
+      "step": 4738
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.748942917547569e-05,
+      "loss": 0.1638,
+      "step": 4740
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.748414376321353e-05,
+      "loss": 0.3567,
+      "step": 4742
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.7478858350951374e-05,
+      "loss": 0.2977,
+      "step": 4744
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.747357293868922e-05,
+      "loss": 0.1962,
+      "step": 4746
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.7468287526427066e-05,
+      "loss": 0.1857,
+      "step": 4748
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.7463002114164906e-05,
+      "loss": 0.2079,
+      "step": 4750
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.745771670190275e-05,
+      "loss": 0.1679,
+      "step": 4752
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.745243128964059e-05,
+      "loss": 0.299,
+      "step": 4754
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.744714587737844e-05,
+      "loss": 0.1429,
+      "step": 4756
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.7441860465116276e-05,
+      "loss": 0.1921,
+      "step": 4758
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.743657505285412e-05,
+      "loss": 0.2171,
+      "step": 4760
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.743128964059197e-05,
+      "loss": 0.1292,
+      "step": 4762
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.7426004228329814e-05,
+      "loss": 0.2783,
+      "step": 4764
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.7420718816067654e-05,
+      "loss": 0.2958,
+      "step": 4766
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.74154334038055e-05,
+      "loss": 0.146,
+      "step": 4768
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.7410147991543346e-05,
+      "loss": 0.1545,
+      "step": 4770
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.7404862579281185e-05,
+      "loss": 0.2194,
+      "step": 4772
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.739957716701903e-05,
+      "loss": 0.1789,
+      "step": 4774
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.739429175475687e-05,
+      "loss": 0.2369,
+      "step": 4776
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.7389006342494717e-05,
+      "loss": 0.1711,
+      "step": 4778
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.7383720930232556e-05,
+      "loss": 0.2223,
+      "step": 4780
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.737843551797041e-05,
+      "loss": 0.2188,
+      "step": 4782
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.737315010570825e-05,
+      "loss": 0.2343,
+      "step": 4784
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.7367864693446094e-05,
+      "loss": 0.109,
+      "step": 4786
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.736257928118393e-05,
+      "loss": 0.2162,
+      "step": 4788
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.735729386892178e-05,
+      "loss": 0.3718,
+      "step": 4790
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.735200845665962e-05,
+      "loss": 0.2146,
+      "step": 4792
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.7346723044397465e-05,
+      "loss": 0.1319,
+      "step": 4794
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.7341437632135304e-05,
+      "loss": 0.2113,
+      "step": 4796
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.733615221987315e-05,
+      "loss": 0.1171,
+      "step": 4798
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.7330866807610996e-05,
+      "loss": 0.2382,
+      "step": 4800
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.732558139534884e-05,
+      "loss": 0.1432,
+      "step": 4802
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.732029598308668e-05,
+      "loss": 0.2301,
+      "step": 4804
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.731501057082453e-05,
+      "loss": 0.1705,
+      "step": 4806
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.730972515856237e-05,
+      "loss": 0.0764,
+      "step": 4808
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.730443974630021e-05,
+      "loss": 0.123,
+      "step": 4810
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.729915433403806e-05,
+      "loss": 0.156,
+      "step": 4812
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.72938689217759e-05,
+      "loss": 0.2732,
+      "step": 4814
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.7288583509513744e-05,
+      "loss": 0.2014,
+      "step": 4816
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.728329809725159e-05,
+      "loss": 0.2538,
+      "step": 4818
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.727801268498943e-05,
+      "loss": 0.2327,
+      "step": 4820
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.7272727272727276e-05,
+      "loss": 0.2191,
+      "step": 4822
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.726744186046512e-05,
+      "loss": 0.2461,
+      "step": 4824
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.726215644820296e-05,
+      "loss": 0.1499,
+      "step": 4826
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.725687103594081e-05,
+      "loss": 0.1612,
+      "step": 4828
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.7251585623678646e-05,
+      "loss": 0.2029,
+      "step": 4830
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.724630021141649e-05,
+      "loss": 0.3797,
+      "step": 4832
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.724101479915433e-05,
+      "loss": 0.2086,
+      "step": 4834
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.7235729386892184e-05,
+      "loss": 0.1579,
+      "step": 4836
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.7230443974630024e-05,
+      "loss": 0.2036,
+      "step": 4838
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.722515856236787e-05,
+      "loss": 0.1914,
+      "step": 4840
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.721987315010571e-05,
+      "loss": 0.1965,
+      "step": 4842
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.7214587737843555e-05,
+      "loss": 0.109,
+      "step": 4844
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.7209302325581394e-05,
+      "loss": 0.1581,
+      "step": 4846
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.720401691331924e-05,
+      "loss": 0.3613,
+      "step": 4848
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.719873150105708e-05,
+      "loss": 0.2111,
+      "step": 4850
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.7193446088794926e-05,
+      "loss": 0.1572,
+      "step": 4852
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.718816067653277e-05,
+      "loss": 0.2681,
+      "step": 4854
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.718287526427062e-05,
+      "loss": 0.24,
+      "step": 4856
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.717758985200846e-05,
+      "loss": 0.1684,
+      "step": 4858
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.71723044397463e-05,
+      "loss": 0.1807,
+      "step": 4860
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.716701902748414e-05,
+      "loss": 0.171,
+      "step": 4862
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.716173361522199e-05,
+      "loss": 0.1212,
+      "step": 4864
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.7156448202959835e-05,
+      "loss": 0.2092,
+      "step": 4866
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.7151162790697674e-05,
+      "loss": 0.1592,
+      "step": 4868
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.714587737843552e-05,
+      "loss": 0.268,
+      "step": 4870
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.7140591966173366e-05,
+      "loss": 0.2267,
+      "step": 4872
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.713530655391121e-05,
+      "loss": 0.1299,
+      "step": 4874
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.713002114164905e-05,
+      "loss": 0.1087,
+      "step": 4876
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.71247357293869e-05,
+      "loss": 0.2027,
+      "step": 4878
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.711945031712474e-05,
+      "loss": 0.1678,
+      "step": 4880
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.711416490486258e-05,
+      "loss": 0.2345,
+      "step": 4882
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.710887949260042e-05,
+      "loss": 0.149,
+      "step": 4884
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.710359408033827e-05,
+      "loss": 0.1786,
+      "step": 4886
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.709830866807611e-05,
+      "loss": 0.2472,
+      "step": 4888
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.709302325581396e-05,
+      "loss": 0.1835,
+      "step": 4890
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.70877378435518e-05,
+      "loss": 0.0931,
+      "step": 4892
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.7082452431289646e-05,
+      "loss": 0.1802,
+      "step": 4894
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.7077167019027485e-05,
+      "loss": 0.2437,
+      "step": 4896
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.707188160676533e-05,
+      "loss": 0.1213,
+      "step": 4898
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.706659619450317e-05,
+      "loss": 0.1104,
+      "step": 4900
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.7061310782241016e-05,
+      "loss": 0.2072,
+      "step": 4902
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.7056025369978856e-05,
+      "loss": 0.1542,
+      "step": 4904
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.70507399577167e-05,
+      "loss": 0.1379,
+      "step": 4906
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.704545454545455e-05,
+      "loss": 0.1268,
+      "step": 4908
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.7040169133192394e-05,
+      "loss": 0.1603,
+      "step": 4910
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.703488372093023e-05,
+      "loss": 0.1306,
+      "step": 4912
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.702959830866808e-05,
+      "loss": 0.0702,
+      "step": 4914
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.702431289640592e-05,
+      "loss": 0.0872,
+      "step": 4916
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.7019027484143764e-05,
+      "loss": 0.4123,
+      "step": 4918
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.701374207188161e-05,
+      "loss": 0.2372,
+      "step": 4920
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.700845665961945e-05,
+      "loss": 0.2711,
+      "step": 4922
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.7003171247357296e-05,
+      "loss": 0.2139,
+      "step": 4924
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.699788583509514e-05,
+      "loss": 0.227,
+      "step": 4926
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.699260042283299e-05,
+      "loss": 0.0749,
+      "step": 4928
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.698731501057083e-05,
+      "loss": 0.1619,
+      "step": 4930
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.698202959830867e-05,
+      "loss": 0.3009,
+      "step": 4932
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.697674418604651e-05,
+      "loss": 0.2073,
+      "step": 4934
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.697145877378436e-05,
+      "loss": 0.326,
+      "step": 4936
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.69661733615222e-05,
+      "loss": 0.2313,
+      "step": 4938
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.6960887949260044e-05,
+      "loss": 0.1468,
+      "step": 4940
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.695560253699788e-05,
+      "loss": 0.1478,
+      "step": 4942
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.6950317124735736e-05,
+      "loss": 0.2012,
+      "step": 4944
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.6945031712473575e-05,
+      "loss": 0.1614,
+      "step": 4946
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.693974630021142e-05,
+      "loss": 0.1517,
+      "step": 4948
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.693446088794926e-05,
+      "loss": 0.1746,
+      "step": 4950
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.692917547568711e-05,
+      "loss": 0.1299,
+      "step": 4952
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.6923890063424946e-05,
+      "loss": 0.1044,
+      "step": 4954
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.691860465116279e-05,
+      "loss": 0.1396,
+      "step": 4956
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.691331923890063e-05,
+      "loss": 0.1941,
+      "step": 4958
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.690803382663848e-05,
+      "loss": 0.1695,
+      "step": 4960
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.6902748414376324e-05,
+      "loss": 0.2018,
+      "step": 4962
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.689746300211417e-05,
+      "loss": 0.1901,
+      "step": 4964
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.689217758985201e-05,
+      "loss": 0.1719,
+      "step": 4966
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.6886892177589855e-05,
+      "loss": 0.2726,
+      "step": 4968
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.6881606765327694e-05,
+      "loss": 0.2805,
+      "step": 4970
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.687632135306554e-05,
+      "loss": 0.2122,
+      "step": 4972
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.6871035940803386e-05,
+      "loss": 0.1963,
+      "step": 4974
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.6865750528541226e-05,
+      "loss": 0.2034,
+      "step": 4976
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.686046511627907e-05,
+      "loss": 0.1429,
+      "step": 4978
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.685517970401692e-05,
+      "loss": 0.2629,
+      "step": 4980
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.6849894291754764e-05,
+      "loss": 0.2621,
+      "step": 4982
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.68446088794926e-05,
+      "loss": 0.204,
+      "step": 4984
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.683932346723045e-05,
+      "loss": 0.0373,
+      "step": 4986
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.683403805496829e-05,
+      "loss": 0.2303,
+      "step": 4988
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.6828752642706135e-05,
+      "loss": 0.1849,
+      "step": 4990
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.6823467230443974e-05,
+      "loss": 0.2762,
+      "step": 4992
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.681818181818182e-05,
+      "loss": 0.2258,
+      "step": 4994
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.681289640591966e-05,
+      "loss": 0.1006,
+      "step": 4996
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.680761099365751e-05,
+      "loss": 0.2063,
+      "step": 4998
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.680232558139535e-05,
+      "loss": 0.1996,
+      "step": 5000
+    },
+    {
+      "epoch": 5.29,
+      "eval_cer": 0.04149330293530921,
+      "eval_loss": 0.4960121810436249,
+      "eval_runtime": 132.0838,
+      "eval_samples_per_second": 6.367,
+      "eval_steps_per_second": 0.803,
+      "step": 5000
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.67970401691332e-05,
+      "loss": 0.2156,
+      "step": 5002
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.679175475687104e-05,
+      "loss": 0.2322,
+      "step": 5004
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.678646934460888e-05,
+      "loss": 0.2876,
+      "step": 5006
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.678118393234672e-05,
+      "loss": 0.1763,
+      "step": 5008
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.677589852008457e-05,
+      "loss": 0.3529,
+      "step": 5010
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.677061310782241e-05,
+      "loss": 0.2029,
+      "step": 5012
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.676532769556025e-05,
+      "loss": 0.1825,
+      "step": 5014
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.67600422832981e-05,
+      "loss": 0.2479,
+      "step": 5016
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.6754756871035945e-05,
+      "loss": 0.166,
+      "step": 5018
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.6749471458773785e-05,
+      "loss": 0.1522,
+      "step": 5020
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.674418604651163e-05,
+      "loss": 0.2038,
+      "step": 5022
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.673890063424947e-05,
+      "loss": 0.1599,
+      "step": 5024
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.6733615221987316e-05,
+      "loss": 0.1344,
+      "step": 5026
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.672832980972516e-05,
+      "loss": 0.2085,
+      "step": 5028
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.6723044397463e-05,
+      "loss": 0.2214,
+      "step": 5030
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.671775898520085e-05,
+      "loss": 0.2809,
+      "step": 5032
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.6712473572938694e-05,
+      "loss": 0.3275,
+      "step": 5034
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.670718816067654e-05,
+      "loss": 0.1518,
+      "step": 5036
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.670190274841438e-05,
+      "loss": 0.137,
+      "step": 5038
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.6696617336152225e-05,
+      "loss": 0.1788,
+      "step": 5040
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.6691331923890064e-05,
+      "loss": 0.1478,
+      "step": 5042
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.668604651162791e-05,
+      "loss": 0.17,
+      "step": 5044
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.668076109936575e-05,
+      "loss": 0.1549,
+      "step": 5046
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.6675475687103596e-05,
+      "loss": 0.218,
+      "step": 5048
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.6670190274841435e-05,
+      "loss": 0.2056,
+      "step": 5050
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.666490486257929e-05,
+      "loss": 0.1911,
+      "step": 5052
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.665961945031713e-05,
+      "loss": 0.1425,
+      "step": 5054
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.665433403805497e-05,
+      "loss": 0.236,
+      "step": 5056
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.664904862579281e-05,
+      "loss": 0.2433,
+      "step": 5058
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.664376321353066e-05,
+      "loss": 0.1608,
+      "step": 5060
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.66384778012685e-05,
+      "loss": 0.1666,
+      "step": 5062
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.6633192389006344e-05,
+      "loss": 0.1748,
+      "step": 5064
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.662790697674418e-05,
+      "loss": 0.188,
+      "step": 5066
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.662262156448203e-05,
+      "loss": 0.1773,
+      "step": 5068
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.6617336152219875e-05,
+      "loss": 0.228,
+      "step": 5070
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.661205073995772e-05,
+      "loss": 0.1715,
+      "step": 5072
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.660676532769556e-05,
+      "loss": 0.2415,
+      "step": 5074
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.660147991543341e-05,
+      "loss": 0.1875,
+      "step": 5076
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.659619450317125e-05,
+      "loss": 0.2753,
+      "step": 5078
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.659090909090909e-05,
+      "loss": 0.2631,
+      "step": 5080
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.658562367864694e-05,
+      "loss": 0.1206,
+      "step": 5082
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.658033826638478e-05,
+      "loss": 0.2814,
+      "step": 5084
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.6575052854122623e-05,
+      "loss": 0.2033,
+      "step": 5086
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.656976744186046e-05,
+      "loss": 0.1643,
+      "step": 5088
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.6564482029598316e-05,
+      "loss": 0.1404,
+      "step": 5090
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.6559196617336155e-05,
+      "loss": 0.2004,
+      "step": 5092
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.6553911205074e-05,
+      "loss": 0.1059,
+      "step": 5094
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.654862579281184e-05,
+      "loss": 0.1295,
+      "step": 5096
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.6543340380549686e-05,
+      "loss": 0.3503,
+      "step": 5098
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.6538054968287526e-05,
+      "loss": 0.1633,
+      "step": 5100
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.653276955602537e-05,
+      "loss": 0.1093,
+      "step": 5102
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.652748414376321e-05,
+      "loss": 0.2215,
+      "step": 5104
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.652219873150106e-05,
+      "loss": 0.1032,
+      "step": 5106
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.65169133192389e-05,
+      "loss": 0.3177,
+      "step": 5108
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.651162790697675e-05,
+      "loss": 0.209,
+      "step": 5110
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.650634249471459e-05,
+      "loss": 0.1654,
+      "step": 5112
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.6501057082452434e-05,
+      "loss": 0.2999,
+      "step": 5114
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.6495771670190274e-05,
+      "loss": 0.1211,
+      "step": 5116
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.649048625792812e-05,
+      "loss": 0.2942,
+      "step": 5118
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.648520084566596e-05,
+      "loss": 0.2251,
+      "step": 5120
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.6479915433403805e-05,
+      "loss": 0.2525,
+      "step": 5122
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.647463002114165e-05,
+      "loss": 0.2324,
+      "step": 5124
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.64693446088795e-05,
+      "loss": 0.2722,
+      "step": 5126
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.6464059196617336e-05,
+      "loss": 0.1604,
+      "step": 5128
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.645877378435518e-05,
+      "loss": 0.155,
+      "step": 5130
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.645348837209303e-05,
+      "loss": 0.1968,
+      "step": 5132
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.644820295983087e-05,
+      "loss": 0.214,
+      "step": 5134
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.6442917547568714e-05,
+      "loss": 0.1837,
+      "step": 5136
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.643763213530655e-05,
+      "loss": 0.0795,
+      "step": 5138
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.64323467230444e-05,
+      "loss": 0.2023,
+      "step": 5140
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.642706131078224e-05,
+      "loss": 0.151,
+      "step": 5142
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.642177589852009e-05,
+      "loss": 0.1739,
+      "step": 5144
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.641649048625793e-05,
+      "loss": 0.0803,
+      "step": 5146
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.641120507399578e-05,
+      "loss": 0.2561,
+      "step": 5148
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.6405919661733616e-05,
+      "loss": 0.1125,
+      "step": 5150
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.640063424947146e-05,
+      "loss": 0.1583,
+      "step": 5152
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.63953488372093e-05,
+      "loss": 0.2679,
+      "step": 5154
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.639006342494715e-05,
+      "loss": 0.1561,
+      "step": 5156
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.638477801268499e-05,
+      "loss": 0.1881,
+      "step": 5158
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.637949260042283e-05,
+      "loss": 0.213,
+      "step": 5160
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.637420718816068e-05,
+      "loss": 0.0868,
+      "step": 5162
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.6368921775898525e-05,
+      "loss": 0.1728,
+      "step": 5164
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 0.1656,
+      "step": 5166
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.635835095137421e-05,
+      "loss": 0.4125,
+      "step": 5168
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.635306553911205e-05,
+      "loss": 0.2769,
+      "step": 5170
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.6347780126849896e-05,
+      "loss": 0.1373,
+      "step": 5172
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.6342494714587735e-05,
+      "loss": 0.3027,
+      "step": 5174
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.633720930232558e-05,
+      "loss": 0.2377,
+      "step": 5176
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.633192389006343e-05,
+      "loss": 0.1191,
+      "step": 5178
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.632663847780127e-05,
+      "loss": 0.1034,
+      "step": 5180
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.632135306553911e-05,
+      "loss": 0.1489,
+      "step": 5182
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.631606765327696e-05,
+      "loss": 0.1935,
+      "step": 5184
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.6310782241014804e-05,
+      "loss": 0.117,
+      "step": 5186
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.6305496828752644e-05,
+      "loss": 0.2481,
+      "step": 5188
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.630021141649049e-05,
+      "loss": 0.1525,
+      "step": 5190
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.629492600422833e-05,
+      "loss": 0.1148,
+      "step": 5192
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.6289640591966175e-05,
+      "loss": 0.1546,
+      "step": 5194
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.6284355179704014e-05,
+      "loss": 0.136,
+      "step": 5196
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.627906976744187e-05,
+      "loss": 0.1691,
+      "step": 5198
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.6273784355179706e-05,
+      "loss": 0.1914,
+      "step": 5200
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.626849894291755e-05,
+      "loss": 0.1867,
+      "step": 5202
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.626321353065539e-05,
+      "loss": 0.2652,
+      "step": 5204
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.625792811839324e-05,
+      "loss": 0.2976,
+      "step": 5206
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.625264270613108e-05,
+      "loss": 0.2308,
+      "step": 5208
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.624735729386892e-05,
+      "loss": 0.1615,
+      "step": 5210
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.624207188160676e-05,
+      "loss": 0.1069,
+      "step": 5212
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.623678646934461e-05,
+      "loss": 0.2488,
+      "step": 5214
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.6231501057082455e-05,
+      "loss": 0.2028,
+      "step": 5216
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.62262156448203e-05,
+      "loss": 0.4859,
+      "step": 5218
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.622093023255814e-05,
+      "loss": 0.1746,
+      "step": 5220
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.6215644820295986e-05,
+      "loss": 0.1445,
+      "step": 5222
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.6210359408033825e-05,
+      "loss": 0.1672,
+      "step": 5224
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.620507399577167e-05,
+      "loss": 0.1847,
+      "step": 5226
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.619978858350952e-05,
+      "loss": 0.3719,
+      "step": 5228
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.619450317124736e-05,
+      "loss": 0.4154,
+      "step": 5230
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.61892177589852e-05,
+      "loss": 0.2469,
+      "step": 5232
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.618393234672305e-05,
+      "loss": 0.1911,
+      "step": 5234
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.617864693446089e-05,
+      "loss": 0.177,
+      "step": 5236
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.6173361522198734e-05,
+      "loss": 0.1316,
+      "step": 5238
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.616807610993658e-05,
+      "loss": 0.3266,
+      "step": 5240
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.616279069767442e-05,
+      "loss": 0.2839,
+      "step": 5242
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.6157505285412266e-05,
+      "loss": 0.2419,
+      "step": 5244
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.6152219873150105e-05,
+      "loss": 0.2582,
+      "step": 5246
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.614693446088795e-05,
+      "loss": 0.2562,
+      "step": 5248
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.614164904862579e-05,
+      "loss": 0.3777,
+      "step": 5250
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.613636363636364e-05,
+      "loss": 0.3315,
+      "step": 5252
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.613107822410148e-05,
+      "loss": 0.1576,
+      "step": 5254
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.612579281183933e-05,
+      "loss": 0.1337,
+      "step": 5256
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.612050739957717e-05,
+      "loss": 0.1628,
+      "step": 5258
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.6115221987315014e-05,
+      "loss": 0.1957,
+      "step": 5260
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.610993657505285e-05,
+      "loss": 0.2507,
+      "step": 5262
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.61046511627907e-05,
+      "loss": 0.2434,
+      "step": 5264
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.609936575052854e-05,
+      "loss": 0.2171,
+      "step": 5266
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.6094080338266384e-05,
+      "loss": 0.3111,
+      "step": 5268
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.608879492600423e-05,
+      "loss": 0.153,
+      "step": 5270
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.6083509513742077e-05,
+      "loss": 0.1377,
+      "step": 5272
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.6078224101479916e-05,
+      "loss": 0.2569,
+      "step": 5274
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.607293868921776e-05,
+      "loss": 0.179,
+      "step": 5276
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.60676532769556e-05,
+      "loss": 0.2822,
+      "step": 5278
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.606236786469345e-05,
+      "loss": 0.1619,
+      "step": 5280
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.605708245243129e-05,
+      "loss": 0.2394,
+      "step": 5282
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.605179704016913e-05,
+      "loss": 0.2473,
+      "step": 5284
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.604651162790698e-05,
+      "loss": 0.2508,
+      "step": 5286
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.6041226215644825e-05,
+      "loss": 0.2127,
+      "step": 5288
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.6035940803382664e-05,
+      "loss": 0.1293,
+      "step": 5290
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.603065539112051e-05,
+      "loss": 0.4284,
+      "step": 5292
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.6025369978858356e-05,
+      "loss": 0.2449,
+      "step": 5294
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.6020084566596195e-05,
+      "loss": 0.2241,
+      "step": 5296
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.601479915433404e-05,
+      "loss": 0.1785,
+      "step": 5298
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.600951374207188e-05,
+      "loss": 0.1411,
+      "step": 5300
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.600422832980973e-05,
+      "loss": 0.0909,
+      "step": 5302
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.5998942917547566e-05,
+      "loss": 0.1268,
+      "step": 5304
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.599365750528542e-05,
+      "loss": 0.1075,
+      "step": 5306
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.598837209302326e-05,
+      "loss": 0.1462,
+      "step": 5308
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.5983086680761104e-05,
+      "loss": 0.2667,
+      "step": 5310
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.5977801268498944e-05,
+      "loss": 0.1675,
+      "step": 5312
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.597251585623679e-05,
+      "loss": 0.2495,
+      "step": 5314
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.596723044397463e-05,
+      "loss": 0.2061,
+      "step": 5316
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.5961945031712475e-05,
+      "loss": 0.2483,
+      "step": 5318
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.5956659619450314e-05,
+      "loss": 0.1289,
+      "step": 5320
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.595137420718816e-05,
+      "loss": 0.1895,
+      "step": 5322
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.5946088794926006e-05,
+      "loss": 0.2229,
+      "step": 5324
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.594080338266385e-05,
+      "loss": 0.2349,
+      "step": 5326
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.593551797040169e-05,
+      "loss": 0.0876,
+      "step": 5328
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.593023255813954e-05,
+      "loss": 0.1031,
+      "step": 5330
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.592494714587738e-05,
+      "loss": 0.1326,
+      "step": 5332
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.591966173361522e-05,
+      "loss": 0.2142,
+      "step": 5334
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.591437632135307e-05,
+      "loss": 0.2463,
+      "step": 5336
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.590909090909091e-05,
+      "loss": 0.2288,
+      "step": 5338
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.5903805496828754e-05,
+      "loss": 0.2073,
+      "step": 5340
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.58985200845666e-05,
+      "loss": 0.0884,
+      "step": 5342
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.5893234672304447e-05,
+      "loss": 0.2159,
+      "step": 5344
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.5887949260042286e-05,
+      "loss": 0.1365,
+      "step": 5346
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.588266384778013e-05,
+      "loss": 0.2142,
+      "step": 5348
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.587737843551797e-05,
+      "loss": 0.2101,
+      "step": 5350
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.587209302325582e-05,
+      "loss": 0.3025,
+      "step": 5352
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.5866807610993657e-05,
+      "loss": 0.1892,
+      "step": 5354
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.58615221987315e-05,
+      "loss": 0.1453,
+      "step": 5356
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.585623678646934e-05,
+      "loss": 0.1022,
+      "step": 5358
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.5850951374207195e-05,
+      "loss": 0.1156,
+      "step": 5360
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.5845665961945034e-05,
+      "loss": 0.2799,
+      "step": 5362
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.584038054968288e-05,
+      "loss": 0.2196,
+      "step": 5364
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.583509513742072e-05,
+      "loss": 0.277,
+      "step": 5366
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.5829809725158565e-05,
+      "loss": 0.1254,
+      "step": 5368
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.5824524312896405e-05,
+      "loss": 0.1106,
+      "step": 5370
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.581923890063425e-05,
+      "loss": 0.1766,
+      "step": 5372
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.581395348837209e-05,
+      "loss": 0.3162,
+      "step": 5374
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.5808668076109936e-05,
+      "loss": 0.2226,
+      "step": 5376
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.580338266384778e-05,
+      "loss": 0.1871,
+      "step": 5378
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.579809725158563e-05,
+      "loss": 0.3953,
+      "step": 5380
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.579281183932347e-05,
+      "loss": 0.2468,
+      "step": 5382
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.5787526427061314e-05,
+      "loss": 0.1599,
+      "step": 5384
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.578224101479915e-05,
+      "loss": 0.1013,
+      "step": 5386
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.5776955602537e-05,
+      "loss": 0.304,
+      "step": 5388
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.5771670190274845e-05,
+      "loss": 0.228,
+      "step": 5390
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.5766384778012684e-05,
+      "loss": 0.1226,
+      "step": 5392
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.576109936575053e-05,
+      "loss": 0.1913,
+      "step": 5394
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.5755813953488376e-05,
+      "loss": 0.1811,
+      "step": 5396
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.575052854122622e-05,
+      "loss": 0.2636,
+      "step": 5398
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.574524312896406e-05,
+      "loss": 0.1235,
+      "step": 5400
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.573995771670191e-05,
+      "loss": 0.2232,
+      "step": 5402
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.573467230443975e-05,
+      "loss": 0.1785,
+      "step": 5404
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.572938689217759e-05,
+      "loss": 0.1615,
+      "step": 5406
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.572410147991543e-05,
+      "loss": 0.101,
+      "step": 5408
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.571881606765328e-05,
+      "loss": 0.1539,
+      "step": 5410
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.571353065539112e-05,
+      "loss": 0.1954,
+      "step": 5412
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.570824524312897e-05,
+      "loss": 0.2136,
+      "step": 5414
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.570295983086681e-05,
+      "loss": 0.1079,
+      "step": 5416
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.5697674418604656e-05,
+      "loss": 0.2143,
+      "step": 5418
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.5692389006342495e-05,
+      "loss": 0.2804,
+      "step": 5420
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.568710359408034e-05,
+      "loss": 0.0697,
+      "step": 5422
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.568181818181818e-05,
+      "loss": 0.1546,
+      "step": 5424
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.5676532769556027e-05,
+      "loss": 0.2973,
+      "step": 5426
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.5671247357293866e-05,
+      "loss": 0.2149,
+      "step": 5428
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.566596194503171e-05,
+      "loss": 0.3598,
+      "step": 5430
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.566067653276956e-05,
+      "loss": 0.172,
+      "step": 5432
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.5655391120507404e-05,
+      "loss": 0.166,
+      "step": 5434
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.565010570824524e-05,
+      "loss": 0.1814,
+      "step": 5436
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.564482029598309e-05,
+      "loss": 0.1576,
+      "step": 5438
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.563953488372093e-05,
+      "loss": 0.1284,
+      "step": 5440
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.5634249471458775e-05,
+      "loss": 0.1752,
+      "step": 5442
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.562896405919662e-05,
+      "loss": 0.2017,
+      "step": 5444
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.562367864693446e-05,
+      "loss": 0.1809,
+      "step": 5446
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.5618393234672306e-05,
+      "loss": 0.1873,
+      "step": 5448
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.561310782241015e-05,
+      "loss": 0.1214,
+      "step": 5450
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.5607822410148e-05,
+      "loss": 0.0826,
+      "step": 5452
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.560253699788584e-05,
+      "loss": 0.2159,
+      "step": 5454
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.5597251585623684e-05,
+      "loss": 0.1289,
+      "step": 5456
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.559196617336152e-05,
+      "loss": 0.1642,
+      "step": 5458
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.558668076109937e-05,
+      "loss": 0.1882,
+      "step": 5460
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.558139534883721e-05,
+      "loss": 0.1427,
+      "step": 5462
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.5576109936575054e-05,
+      "loss": 0.104,
+      "step": 5464
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.5570824524312894e-05,
+      "loss": 0.11,
+      "step": 5466
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.5565539112050746e-05,
+      "loss": 0.097,
+      "step": 5468
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.5560253699788586e-05,
+      "loss": 0.1465,
+      "step": 5470
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.555496828752643e-05,
+      "loss": 0.2532,
+      "step": 5472
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.554968287526427e-05,
+      "loss": 0.1785,
+      "step": 5474
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.554439746300212e-05,
+      "loss": 0.2498,
+      "step": 5476
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.5539112050739956e-05,
+      "loss": 0.3845,
+      "step": 5478
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.55338266384778e-05,
+      "loss": 0.2038,
+      "step": 5480
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.552854122621564e-05,
+      "loss": 0.1827,
+      "step": 5482
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.552325581395349e-05,
+      "loss": 0.1429,
+      "step": 5484
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.5517970401691334e-05,
+      "loss": 0.1874,
+      "step": 5486
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.551268498942918e-05,
+      "loss": 0.1165,
+      "step": 5488
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.550739957716702e-05,
+      "loss": 0.1995,
+      "step": 5490
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.5502114164904865e-05,
+      "loss": 0.2121,
+      "step": 5492
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.5496828752642705e-05,
+      "loss": 0.1209,
+      "step": 5494
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.549154334038055e-05,
+      "loss": 0.1137,
+      "step": 5496
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.54862579281184e-05,
+      "loss": 0.1164,
+      "step": 5498
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.5480972515856236e-05,
+      "loss": 0.1558,
+      "step": 5500
+    },
+    {
+      "epoch": 5.81,
+      "eval_cer": 0.023254488458250212,
+      "eval_loss": 0.6493213176727295,
+      "eval_runtime": 128.6984,
+      "eval_samples_per_second": 6.535,
+      "eval_steps_per_second": 0.824,
+      "step": 5500
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.547568710359408e-05,
+      "loss": 0.1527,
+      "step": 5502
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.547040169133193e-05,
+      "loss": 0.1787,
+      "step": 5504
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.5465116279069774e-05,
+      "loss": 0.0842,
+      "step": 5506
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.5459830866807613e-05,
+      "loss": 0.1248,
+      "step": 5508
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.545454545454546e-05,
+      "loss": 0.2904,
+      "step": 5510
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.54492600422833e-05,
+      "loss": 0.1775,
+      "step": 5512
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.5443974630021145e-05,
+      "loss": 0.1804,
+      "step": 5514
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.5438689217758984e-05,
+      "loss": 0.1622,
+      "step": 5516
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.543340380549683e-05,
+      "loss": 0.1061,
+      "step": 5518
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.542811839323467e-05,
+      "loss": 0.2467,
+      "step": 5520
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.542283298097252e-05,
+      "loss": 0.1456,
+      "step": 5522
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.541754756871036e-05,
+      "loss": 0.0853,
+      "step": 5524
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.541226215644821e-05,
+      "loss": 0.2468,
+      "step": 5526
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.540697674418605e-05,
+      "loss": 0.1191,
+      "step": 5528
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.540169133192389e-05,
+      "loss": 0.1519,
+      "step": 5530
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.539640591966173e-05,
+      "loss": 0.1389,
+      "step": 5532
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.539112050739958e-05,
+      "loss": 0.1521,
+      "step": 5534
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.538583509513742e-05,
+      "loss": 0.1933,
+      "step": 5536
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.5380549682875264e-05,
+      "loss": 0.2494,
+      "step": 5538
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.537526427061311e-05,
+      "loss": 0.2218,
+      "step": 5540
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.5369978858350956e-05,
+      "loss": 0.193,
+      "step": 5542
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.5364693446088795e-05,
+      "loss": 0.149,
+      "step": 5544
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.535940803382664e-05,
+      "loss": 0.211,
+      "step": 5546
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.535412262156449e-05,
+      "loss": 0.2097,
+      "step": 5548
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.5348837209302326e-05,
+      "loss": 0.2378,
+      "step": 5550
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.534355179704017e-05,
+      "loss": 0.2757,
+      "step": 5552
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.533826638477801e-05,
+      "loss": 0.1784,
+      "step": 5554
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.533298097251586e-05,
+      "loss": 0.1451,
+      "step": 5556
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.5327695560253704e-05,
+      "loss": 0.1482,
+      "step": 5558
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.532241014799155e-05,
+      "loss": 0.1061,
+      "step": 5560
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.531712473572939e-05,
+      "loss": 0.0902,
+      "step": 5562
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.5311839323467235e-05,
+      "loss": 0.1869,
+      "step": 5564
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.5306553911205075e-05,
+      "loss": 0.2725,
+      "step": 5566
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.530126849894292e-05,
+      "loss": 0.2675,
+      "step": 5568
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.529598308668076e-05,
+      "loss": 0.2006,
+      "step": 5570
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.5290697674418606e-05,
+      "loss": 0.1599,
+      "step": 5572
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.5285412262156445e-05,
+      "loss": 0.1574,
+      "step": 5574
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.52801268498943e-05,
+      "loss": 0.103,
+      "step": 5576
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.527484143763214e-05,
+      "loss": 0.2207,
+      "step": 5578
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.5269556025369983e-05,
+      "loss": 0.1215,
+      "step": 5580
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.526427061310782e-05,
+      "loss": 0.3131,
+      "step": 5582
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.525898520084567e-05,
+      "loss": 0.2382,
+      "step": 5584
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.525369978858351e-05,
+      "loss": 0.2115,
+      "step": 5586
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.5248414376321354e-05,
+      "loss": 0.315,
+      "step": 5588
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.5243128964059193e-05,
+      "loss": 0.3125,
+      "step": 5590
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.523784355179704e-05,
+      "loss": 0.2226,
+      "step": 5592
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.5232558139534886e-05,
+      "loss": 0.1697,
+      "step": 5594
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.522727272727273e-05,
+      "loss": 0.199,
+      "step": 5596
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.522198731501057e-05,
+      "loss": 0.116,
+      "step": 5598
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.521670190274842e-05,
+      "loss": 0.3555,
+      "step": 5600
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.521141649048626e-05,
+      "loss": 0.1411,
+      "step": 5602
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.52061310782241e-05,
+      "loss": 0.1481,
+      "step": 5604
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.520084566596195e-05,
+      "loss": 0.2624,
+      "step": 5606
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.519556025369979e-05,
+      "loss": 0.2433,
+      "step": 5608
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.5190274841437634e-05,
+      "loss": 0.1429,
+      "step": 5610
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.518498942917548e-05,
+      "loss": 0.119,
+      "step": 5612
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.5179704016913326e-05,
+      "loss": 0.0767,
+      "step": 5614
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.5174418604651165e-05,
+      "loss": 0.2706,
+      "step": 5616
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.516913319238901e-05,
+      "loss": 0.1723,
+      "step": 5618
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.516384778012685e-05,
+      "loss": 0.1937,
+      "step": 5620
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.5158562367864696e-05,
+      "loss": 0.1043,
+      "step": 5622
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.5153276955602536e-05,
+      "loss": 0.2116,
+      "step": 5624
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.514799154334038e-05,
+      "loss": 0.1653,
+      "step": 5626
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.514270613107822e-05,
+      "loss": 0.1505,
+      "step": 5628
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.5137420718816074e-05,
+      "loss": 0.2293,
+      "step": 5630
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.513213530655391e-05,
+      "loss": 0.2016,
+      "step": 5632
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.512684989429176e-05,
+      "loss": 0.1426,
+      "step": 5634
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.51215644820296e-05,
+      "loss": 0.129,
+      "step": 5636
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.5116279069767445e-05,
+      "loss": 0.1259,
+      "step": 5638
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.5110993657505284e-05,
+      "loss": 0.1072,
+      "step": 5640
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.510570824524313e-05,
+      "loss": 0.2352,
+      "step": 5642
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.510042283298097e-05,
+      "loss": 0.3217,
+      "step": 5644
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.5095137420718815e-05,
+      "loss": 0.0973,
+      "step": 5646
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.508985200845666e-05,
+      "loss": 0.1055,
+      "step": 5648
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.508456659619451e-05,
+      "loss": 0.1537,
+      "step": 5650
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.507928118393235e-05,
+      "loss": 0.1017,
+      "step": 5652
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.507399577167019e-05,
+      "loss": 0.2179,
+      "step": 5654
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.506871035940804e-05,
+      "loss": 0.1378,
+      "step": 5656
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.506342494714588e-05,
+      "loss": 0.251,
+      "step": 5658
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.5058139534883724e-05,
+      "loss": 0.146,
+      "step": 5660
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.5052854122621563e-05,
+      "loss": 0.1565,
+      "step": 5662
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.504756871035941e-05,
+      "loss": 0.1334,
+      "step": 5664
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.5042283298097256e-05,
+      "loss": 0.1833,
+      "step": 5666
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.50369978858351e-05,
+      "loss": 0.1748,
+      "step": 5668
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.503171247357294e-05,
+      "loss": 0.1905,
+      "step": 5670
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.502642706131079e-05,
+      "loss": 0.1416,
+      "step": 5672
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.5021141649048626e-05,
+      "loss": 0.189,
+      "step": 5674
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.501585623678647e-05,
+      "loss": 0.1976,
+      "step": 5676
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.501057082452431e-05,
+      "loss": 0.1923,
+      "step": 5678
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.500528541226216e-05,
+      "loss": 0.2369,
+      "step": 5680
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.5e-05,
+      "loss": 0.0773,
+      "step": 5682
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.499471458773785e-05,
+      "loss": 0.1012,
+      "step": 5684
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.498942917547569e-05,
+      "loss": 0.1372,
+      "step": 5686
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.4984143763213535e-05,
+      "loss": 0.2137,
+      "step": 5688
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.4978858350951374e-05,
+      "loss": 0.213,
+      "step": 5690
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.497357293868922e-05,
+      "loss": 0.0937,
+      "step": 5692
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.496828752642706e-05,
+      "loss": 0.2398,
+      "step": 5694
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.4963002114164906e-05,
+      "loss": 0.1157,
+      "step": 5696
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.495771670190275e-05,
+      "loss": 0.0669,
+      "step": 5698
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.495243128964059e-05,
+      "loss": 0.2217,
+      "step": 5700
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.494714587737844e-05,
+      "loss": 0.2695,
+      "step": 5702
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.494186046511628e-05,
+      "loss": 0.2532,
+      "step": 5704
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.493657505285412e-05,
+      "loss": 0.1932,
+      "step": 5706
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.493128964059197e-05,
+      "loss": 0.2101,
+      "step": 5708
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.4926004228329815e-05,
+      "loss": 0.2146,
+      "step": 5710
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.4920718816067654e-05,
+      "loss": 0.1982,
+      "step": 5712
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.49154334038055e-05,
+      "loss": 0.0912,
+      "step": 5714
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.491014799154334e-05,
+      "loss": 0.1053,
+      "step": 5716
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.4904862579281185e-05,
+      "loss": 0.2478,
+      "step": 5718
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.489957716701903e-05,
+      "loss": 0.2633,
+      "step": 5720
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.489429175475688e-05,
+      "loss": 0.1436,
+      "step": 5722
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.488900634249472e-05,
+      "loss": 0.2109,
+      "step": 5724
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.488372093023256e-05,
+      "loss": 0.1513,
+      "step": 5726
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.48784355179704e-05,
+      "loss": 0.1171,
+      "step": 5728
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.487315010570825e-05,
+      "loss": 0.0842,
+      "step": 5730
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.486786469344609e-05,
+      "loss": 0.2574,
+      "step": 5732
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.4862579281183933e-05,
+      "loss": 0.0996,
+      "step": 5734
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.485729386892177e-05,
+      "loss": 0.0907,
+      "step": 5736
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.4852008456659626e-05,
+      "loss": 0.1762,
+      "step": 5738
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.4846723044397465e-05,
+      "loss": 0.116,
+      "step": 5740
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.484143763213531e-05,
+      "loss": 0.3131,
+      "step": 5742
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.483615221987315e-05,
+      "loss": 0.2087,
+      "step": 5744
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.4830866807610996e-05,
+      "loss": 0.2606,
+      "step": 5746
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.4825581395348836e-05,
+      "loss": 0.1987,
+      "step": 5748
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.482029598308668e-05,
+      "loss": 0.3575,
+      "step": 5750
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.481501057082453e-05,
+      "loss": 0.2289,
+      "step": 5752
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.480972515856237e-05,
+      "loss": 0.179,
+      "step": 5754
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.480443974630021e-05,
+      "loss": 0.1997,
+      "step": 5756
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.479915433403806e-05,
+      "loss": 0.1259,
+      "step": 5758
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.47938689217759e-05,
+      "loss": 0.1962,
+      "step": 5760
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.4788583509513744e-05,
+      "loss": 0.2209,
+      "step": 5762
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.478329809725159e-05,
+      "loss": 0.1538,
+      "step": 5764
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.477801268498943e-05,
+      "loss": 0.2426,
+      "step": 5766
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.4772727272727276e-05,
+      "loss": 0.3101,
+      "step": 5768
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.4767441860465115e-05,
+      "loss": 0.3042,
+      "step": 5770
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.476215644820296e-05,
+      "loss": 0.2477,
+      "step": 5772
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.47568710359408e-05,
+      "loss": 0.1398,
+      "step": 5774
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.475158562367865e-05,
+      "loss": 0.1257,
+      "step": 5776
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.474630021141649e-05,
+      "loss": 0.1497,
+      "step": 5778
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.474101479915434e-05,
+      "loss": 0.1602,
+      "step": 5780
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.473572938689218e-05,
+      "loss": 0.1371,
+      "step": 5782
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.4730443974630024e-05,
+      "loss": 0.1337,
+      "step": 5784
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.472515856236786e-05,
+      "loss": 0.139,
+      "step": 5786
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.471987315010571e-05,
+      "loss": 0.2151,
+      "step": 5788
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.471458773784355e-05,
+      "loss": 0.1326,
+      "step": 5790
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.4709302325581395e-05,
+      "loss": 0.1047,
+      "step": 5792
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.470401691331924e-05,
+      "loss": 0.2743,
+      "step": 5794
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.469873150105709e-05,
+      "loss": 0.1087,
+      "step": 5796
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.4693446088794926e-05,
+      "loss": 0.1329,
+      "step": 5798
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.468816067653277e-05,
+      "loss": 0.0915,
+      "step": 5800
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.468287526427061e-05,
+      "loss": 0.0456,
+      "step": 5802
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.467758985200846e-05,
+      "loss": 0.1955,
+      "step": 5804
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.4672304439746304e-05,
+      "loss": 0.1902,
+      "step": 5806
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.466701902748414e-05,
+      "loss": 0.1203,
+      "step": 5808
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.466173361522199e-05,
+      "loss": 0.1,
+      "step": 5810
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.4656448202959835e-05,
+      "loss": 0.112,
+      "step": 5812
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.465116279069768e-05,
+      "loss": 0.1416,
+      "step": 5814
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.464587737843552e-05,
+      "loss": 0.1604,
+      "step": 5816
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.4640591966173366e-05,
+      "loss": 0.1217,
+      "step": 5818
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.4635306553911206e-05,
+      "loss": 0.2795,
+      "step": 5820
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.463002114164905e-05,
+      "loss": 0.221,
+      "step": 5822
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.462473572938689e-05,
+      "loss": 0.1878,
+      "step": 5824
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.461945031712474e-05,
+      "loss": 0.1025,
+      "step": 5826
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.4614164904862576e-05,
+      "loss": 0.1729,
+      "step": 5828
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.460887949260043e-05,
+      "loss": 0.096,
+      "step": 5830
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.460359408033827e-05,
+      "loss": 0.158,
+      "step": 5832
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.4598308668076114e-05,
+      "loss": 0.1745,
+      "step": 5834
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.4593023255813954e-05,
+      "loss": 0.1705,
+      "step": 5836
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.459038054968288e-05,
+      "loss": 0.3037,
+      "step": 5838
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.458509513742072e-05,
+      "loss": 0.1047,
+      "step": 5840
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.457980972515856e-05,
+      "loss": 0.0526,
+      "step": 5842
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.457452431289641e-05,
+      "loss": 0.2037,
+      "step": 5844
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.456923890063425e-05,
+      "loss": 0.087,
+      "step": 5846
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.45639534883721e-05,
+      "loss": 0.1763,
+      "step": 5848
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.455866807610994e-05,
+      "loss": 0.0715,
+      "step": 5850
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.4553382663847786e-05,
+      "loss": 0.2581,
+      "step": 5852
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.4548097251585625e-05,
+      "loss": 0.1202,
+      "step": 5854
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.454281183932347e-05,
+      "loss": 0.0978,
+      "step": 5856
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.453752642706131e-05,
+      "loss": 0.1894,
+      "step": 5858
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.4532241014799156e-05,
+      "loss": 0.1545,
+      "step": 5860
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.4526955602536996e-05,
+      "loss": 0.1361,
+      "step": 5862
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.452167019027484e-05,
+      "loss": 0.2735,
+      "step": 5864
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.451638477801269e-05,
+      "loss": 0.211,
+      "step": 5866
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.4511099365750534e-05,
+      "loss": 0.107,
+      "step": 5868
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.450581395348837e-05,
+      "loss": 0.1412,
+      "step": 5870
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.450052854122622e-05,
+      "loss": 0.2465,
+      "step": 5872
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.449524312896406e-05,
+      "loss": 0.1566,
+      "step": 5874
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.4489957716701904e-05,
+      "loss": 0.1874,
+      "step": 5876
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.4484672304439744e-05,
+      "loss": 0.2111,
+      "step": 5878
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.447938689217759e-05,
+      "loss": 0.1737,
+      "step": 5880
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.4474101479915436e-05,
+      "loss": 0.0879,
+      "step": 5882
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.446881606765328e-05,
+      "loss": 0.158,
+      "step": 5884
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.446353065539112e-05,
+      "loss": 0.2724,
+      "step": 5886
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.445824524312897e-05,
+      "loss": 0.1235,
+      "step": 5888
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.445295983086681e-05,
+      "loss": 0.2047,
+      "step": 5890
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.444767441860465e-05,
+      "loss": 0.2076,
+      "step": 5892
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.44423890063425e-05,
+      "loss": 0.1676,
+      "step": 5894
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.443710359408034e-05,
+      "loss": 0.1896,
+      "step": 5896
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.4431818181818184e-05,
+      "loss": 0.145,
+      "step": 5898
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.442653276955602e-05,
+      "loss": 0.0989,
+      "step": 5900
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.4421247357293876e-05,
+      "loss": 0.1219,
+      "step": 5902
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.4415961945031715e-05,
+      "loss": 0.1201,
+      "step": 5904
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.441067653276956e-05,
+      "loss": 0.2582,
+      "step": 5906
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.44053911205074e-05,
+      "loss": 0.1738,
+      "step": 5908
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.440010570824525e-05,
+      "loss": 0.2368,
+      "step": 5910
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.4394820295983086e-05,
+      "loss": 0.1735,
+      "step": 5912
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.438953488372093e-05,
+      "loss": 0.1939,
+      "step": 5914
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.438424947145877e-05,
+      "loss": 0.1386,
+      "step": 5916
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.437896405919662e-05,
+      "loss": 0.0951,
+      "step": 5918
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.4373678646934464e-05,
+      "loss": 0.061,
+      "step": 5920
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.436839323467231e-05,
+      "loss": 0.1492,
+      "step": 5922
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.436310782241015e-05,
+      "loss": 0.1947,
+      "step": 5924
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.4357822410147995e-05,
+      "loss": 0.1067,
+      "step": 5926
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.4352536997885834e-05,
+      "loss": 0.1853,
+      "step": 5928
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.434725158562368e-05,
+      "loss": 0.2154,
+      "step": 5930
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.434196617336152e-05,
+      "loss": 0.2419,
+      "step": 5932
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.4336680761099366e-05,
+      "loss": 0.184,
+      "step": 5934
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.433139534883721e-05,
+      "loss": 0.1954,
+      "step": 5936
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.432610993657506e-05,
+      "loss": 0.2227,
+      "step": 5938
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.43208245243129e-05,
+      "loss": 0.2063,
+      "step": 5940
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.431553911205074e-05,
+      "loss": 0.1288,
+      "step": 5942
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.431025369978859e-05,
+      "loss": 0.1622,
+      "step": 5944
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.430496828752643e-05,
+      "loss": 0.1974,
+      "step": 5946
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.4299682875264275e-05,
+      "loss": 0.1642,
+      "step": 5948
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.4294397463002114e-05,
+      "loss": 0.1333,
+      "step": 5950
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.428911205073996e-05,
+      "loss": 0.1552,
+      "step": 5952
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.42838266384778e-05,
+      "loss": 0.165,
+      "step": 5954
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.427854122621565e-05,
+      "loss": 0.1734,
+      "step": 5956
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.427325581395349e-05,
+      "loss": 0.1146,
+      "step": 5958
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.426797040169134e-05,
+      "loss": 0.1115,
+      "step": 5960
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.426268498942918e-05,
+      "loss": 0.2913,
+      "step": 5962
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.425739957716702e-05,
+      "loss": 0.1059,
+      "step": 5964
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.425211416490486e-05,
+      "loss": 0.1041,
+      "step": 5966
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.424682875264271e-05,
+      "loss": 0.1611,
+      "step": 5968
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.424154334038055e-05,
+      "loss": 0.0764,
+      "step": 5970
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.423625792811839e-05,
+      "loss": 0.1817,
+      "step": 5972
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.423097251585624e-05,
+      "loss": 0.1004,
+      "step": 5974
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.4225687103594085e-05,
+      "loss": 0.0877,
+      "step": 5976
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.4220401691331925e-05,
+      "loss": 0.1462,
+      "step": 5978
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.421511627906977e-05,
+      "loss": 0.1329,
+      "step": 5980
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.420983086680761e-05,
+      "loss": 0.1345,
+      "step": 5982
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.4204545454545456e-05,
+      "loss": 0.1035,
+      "step": 5984
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.4199260042283295e-05,
+      "loss": 0.2515,
+      "step": 5986
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.419397463002114e-05,
+      "loss": 0.1297,
+      "step": 5988
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.418868921775899e-05,
+      "loss": 0.2148,
+      "step": 5990
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.4183403805496834e-05,
+      "loss": 0.1103,
+      "step": 5992
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.417811839323467e-05,
+      "loss": 0.114,
+      "step": 5994
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.417283298097252e-05,
+      "loss": 0.1915,
+      "step": 5996
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.4167547568710365e-05,
+      "loss": 0.1484,
+      "step": 5998
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.4162262156448204e-05,
+      "loss": 0.1164,
+      "step": 6000
+    },
+    {
+      "epoch": 6.34,
+      "eval_cer": 0.025876318039327443,
+      "eval_loss": 0.5018019676208496,
+      "eval_runtime": 128.3817,
+      "eval_samples_per_second": 6.551,
+      "eval_steps_per_second": 0.826,
+      "step": 6000
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.415697674418605e-05,
+      "loss": 0.0918,
+      "step": 6002
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.415169133192389e-05,
+      "loss": 0.1645,
+      "step": 6004
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.4146405919661736e-05,
+      "loss": 0.0649,
+      "step": 6006
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.4141120507399575e-05,
+      "loss": 0.158,
+      "step": 6008
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.413583509513743e-05,
+      "loss": 0.1728,
+      "step": 6010
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.413054968287527e-05,
+      "loss": 0.2382,
+      "step": 6012
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.412526427061311e-05,
+      "loss": 0.1304,
+      "step": 6014
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.411997885835095e-05,
+      "loss": 0.1578,
+      "step": 6016
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.41146934460888e-05,
+      "loss": 0.1018,
+      "step": 6018
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.410940803382664e-05,
+      "loss": 0.1749,
+      "step": 6020
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.4104122621564484e-05,
+      "loss": 0.0877,
+      "step": 6022
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.409883720930232e-05,
+      "loss": 0.1199,
+      "step": 6024
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.409355179704017e-05,
+      "loss": 0.1858,
+      "step": 6026
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.4088266384778015e-05,
+      "loss": 0.1073,
+      "step": 6028
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.408298097251586e-05,
+      "loss": 0.0823,
+      "step": 6030
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.40776955602537e-05,
+      "loss": 0.1692,
+      "step": 6032
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.407241014799155e-05,
+      "loss": 0.1037,
+      "step": 6034
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.4067124735729386e-05,
+      "loss": 0.0908,
+      "step": 6036
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.406183932346723e-05,
+      "loss": 0.1123,
+      "step": 6038
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.405655391120507e-05,
+      "loss": 0.1272,
+      "step": 6040
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.405126849894292e-05,
+      "loss": 0.1125,
+      "step": 6042
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.4045983086680763e-05,
+      "loss": 0.1315,
+      "step": 6044
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.404069767441861e-05,
+      "loss": 0.2304,
+      "step": 6046
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.403541226215645e-05,
+      "loss": 0.108,
+      "step": 6048
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.4030126849894295e-05,
+      "loss": 0.2105,
+      "step": 6050
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.402484143763214e-05,
+      "loss": 0.1575,
+      "step": 6052
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.401955602536998e-05,
+      "loss": 0.1972,
+      "step": 6054
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.4014270613107826e-05,
+      "loss": 0.1202,
+      "step": 6056
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.4008985200845665e-05,
+      "loss": 0.1426,
+      "step": 6058
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.400369978858351e-05,
+      "loss": 0.1163,
+      "step": 6060
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.399841437632135e-05,
+      "loss": 0.2422,
+      "step": 6062
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.3993128964059204e-05,
+      "loss": 0.1269,
+      "step": 6064
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.398784355179704e-05,
+      "loss": 0.1424,
+      "step": 6066
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.398255813953489e-05,
+      "loss": 0.1831,
+      "step": 6068
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.397727272727273e-05,
+      "loss": 0.2557,
+      "step": 6070
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.3971987315010574e-05,
+      "loss": 0.3715,
+      "step": 6072
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.3966701902748414e-05,
+      "loss": 0.3253,
+      "step": 6074
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.396141649048626e-05,
+      "loss": 0.0648,
+      "step": 6076
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.39561310782241e-05,
+      "loss": 0.2528,
+      "step": 6078
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.3950845665961945e-05,
+      "loss": 0.1489,
+      "step": 6080
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.394556025369979e-05,
+      "loss": 0.1198,
+      "step": 6082
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.394027484143764e-05,
+      "loss": 0.1566,
+      "step": 6084
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.3934989429175476e-05,
+      "loss": 0.2284,
+      "step": 6086
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.392970401691332e-05,
+      "loss": 0.1358,
+      "step": 6088
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.392441860465116e-05,
+      "loss": 0.0754,
+      "step": 6090
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.391913319238901e-05,
+      "loss": 0.0822,
+      "step": 6092
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.3913847780126854e-05,
+      "loss": 0.1326,
+      "step": 6094
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.390856236786469e-05,
+      "loss": 0.1727,
+      "step": 6096
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.390327695560254e-05,
+      "loss": 0.1164,
+      "step": 6098
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.389799154334038e-05,
+      "loss": 0.3431,
+      "step": 6100
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.3892706131078225e-05,
+      "loss": 0.1192,
+      "step": 6102
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.388742071881607e-05,
+      "loss": 0.0976,
+      "step": 6104
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.388213530655392e-05,
+      "loss": 0.1534,
+      "step": 6106
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.3876849894291756e-05,
+      "loss": 0.1809,
+      "step": 6108
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.38715644820296e-05,
+      "loss": 0.1546,
+      "step": 6110
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.386627906976744e-05,
+      "loss": 0.0665,
+      "step": 6112
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.386099365750529e-05,
+      "loss": 0.1246,
+      "step": 6114
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.385570824524313e-05,
+      "loss": 0.2174,
+      "step": 6116
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.385042283298097e-05,
+      "loss": 0.1845,
+      "step": 6118
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.384513742071882e-05,
+      "loss": 0.1134,
+      "step": 6120
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.3839852008456665e-05,
+      "loss": 0.2435,
+      "step": 6122
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.3834566596194504e-05,
+      "loss": 0.1558,
+      "step": 6124
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.382928118393235e-05,
+      "loss": 0.0459,
+      "step": 6126
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.382399577167019e-05,
+      "loss": 0.1331,
+      "step": 6128
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.3818710359408036e-05,
+      "loss": 0.1732,
+      "step": 6130
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.3813424947145875e-05,
+      "loss": 0.1099,
+      "step": 6132
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.380813953488372e-05,
+      "loss": 0.2,
+      "step": 6134
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.380285412262156e-05,
+      "loss": 0.1682,
+      "step": 6136
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.379756871035941e-05,
+      "loss": 0.1874,
+      "step": 6138
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.379228329809725e-05,
+      "loss": 0.1585,
+      "step": 6140
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.37869978858351e-05,
+      "loss": 0.2005,
+      "step": 6142
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.378171247357294e-05,
+      "loss": 0.1283,
+      "step": 6144
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.3776427061310784e-05,
+      "loss": 0.0849,
+      "step": 6146
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.377114164904863e-05,
+      "loss": 0.1139,
+      "step": 6148
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.376585623678647e-05,
+      "loss": 0.2118,
+      "step": 6150
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.3760570824524315e-05,
+      "loss": 0.2581,
+      "step": 6152
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.3755285412262154e-05,
+      "loss": 0.1161,
+      "step": 6154
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.375000000000001e-05,
+      "loss": 0.1572,
+      "step": 6156
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.3744714587737846e-05,
+      "loss": 0.1451,
+      "step": 6158
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.373942917547569e-05,
+      "loss": 0.1042,
+      "step": 6160
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.373414376321353e-05,
+      "loss": 0.143,
+      "step": 6162
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.372885835095138e-05,
+      "loss": 0.0863,
+      "step": 6164
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.372357293868922e-05,
+      "loss": 0.1268,
+      "step": 6166
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.371828752642706e-05,
+      "loss": 0.1119,
+      "step": 6168
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.37130021141649e-05,
+      "loss": 0.1199,
+      "step": 6170
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.370771670190275e-05,
+      "loss": 0.1628,
+      "step": 6172
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.3702431289640595e-05,
+      "loss": 0.2543,
+      "step": 6174
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.369714587737844e-05,
+      "loss": 0.1345,
+      "step": 6176
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.369186046511628e-05,
+      "loss": 0.0945,
+      "step": 6178
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.3686575052854126e-05,
+      "loss": 0.112,
+      "step": 6180
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.3681289640591965e-05,
+      "loss": 0.1513,
+      "step": 6182
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.367600422832981e-05,
+      "loss": 0.1097,
+      "step": 6184
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.367071881606765e-05,
+      "loss": 0.0726,
+      "step": 6186
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.36654334038055e-05,
+      "loss": 0.2753,
+      "step": 6188
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.3660147991543336e-05,
+      "loss": 0.3024,
+      "step": 6190
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.365486257928119e-05,
+      "loss": 0.4197,
+      "step": 6192
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.364957716701903e-05,
+      "loss": 0.093,
+      "step": 6194
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.3644291754756874e-05,
+      "loss": 0.1188,
+      "step": 6196
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.3639006342494713e-05,
+      "loss": 0.2256,
+      "step": 6198
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.363372093023256e-05,
+      "loss": 0.1508,
+      "step": 6200
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.3628435517970406e-05,
+      "loss": 0.1873,
+      "step": 6202
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.3623150105708245e-05,
+      "loss": 0.0964,
+      "step": 6204
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.361786469344609e-05,
+      "loss": 0.2517,
+      "step": 6206
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.361257928118393e-05,
+      "loss": 0.2191,
+      "step": 6208
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.360729386892178e-05,
+      "loss": 0.1957,
+      "step": 6210
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.360200845665962e-05,
+      "loss": 0.226,
+      "step": 6212
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.359672304439747e-05,
+      "loss": 0.1739,
+      "step": 6214
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.359143763213531e-05,
+      "loss": 0.1438,
+      "step": 6216
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.3586152219873154e-05,
+      "loss": 0.0768,
+      "step": 6218
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.358086680761099e-05,
+      "loss": 0.1309,
+      "step": 6220
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.357558139534884e-05,
+      "loss": 0.1162,
+      "step": 6222
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.357029598308668e-05,
+      "loss": 0.1296,
+      "step": 6224
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.3565010570824524e-05,
+      "loss": 0.1339,
+      "step": 6226
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.355972515856237e-05,
+      "loss": 0.1097,
+      "step": 6228
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.3554439746300217e-05,
+      "loss": 0.1484,
+      "step": 6230
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.3549154334038056e-05,
+      "loss": 0.1588,
+      "step": 6232
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.35438689217759e-05,
+      "loss": 0.3276,
+      "step": 6234
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.353858350951374e-05,
+      "loss": 0.0764,
+      "step": 6236
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.353329809725159e-05,
+      "loss": 0.1882,
+      "step": 6238
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.3528012684989427e-05,
+      "loss": 0.3437,
+      "step": 6240
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.352272727272727e-05,
+      "loss": 0.2354,
+      "step": 6242
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.351744186046512e-05,
+      "loss": 0.1915,
+      "step": 6244
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.3512156448202965e-05,
+      "loss": 0.1843,
+      "step": 6246
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.3506871035940804e-05,
+      "loss": 0.0724,
+      "step": 6248
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.350158562367865e-05,
+      "loss": 0.141,
+      "step": 6250
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.349630021141649e-05,
+      "loss": 0.1799,
+      "step": 6252
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.3491014799154335e-05,
+      "loss": 0.1416,
+      "step": 6254
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.348572938689218e-05,
+      "loss": 0.1412,
+      "step": 6256
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.348044397463002e-05,
+      "loss": 0.1163,
+      "step": 6258
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.347515856236787e-05,
+      "loss": 0.1239,
+      "step": 6260
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.3469873150105706e-05,
+      "loss": 0.2748,
+      "step": 6262
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.346458773784356e-05,
+      "loss": 0.13,
+      "step": 6264
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.34593023255814e-05,
+      "loss": 0.1479,
+      "step": 6266
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.3454016913319244e-05,
+      "loss": 0.162,
+      "step": 6268
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.3448731501057084e-05,
+      "loss": 0.1834,
+      "step": 6270
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.344344608879493e-05,
+      "loss": 0.1042,
+      "step": 6272
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.343816067653277e-05,
+      "loss": 0.1347,
+      "step": 6274
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.3432875264270615e-05,
+      "loss": 0.2021,
+      "step": 6276
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.3427589852008454e-05,
+      "loss": 0.1335,
+      "step": 6278
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.34223044397463e-05,
+      "loss": 0.1198,
+      "step": 6280
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.3417019027484146e-05,
+      "loss": 0.1227,
+      "step": 6282
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.341173361522199e-05,
+      "loss": 0.227,
+      "step": 6284
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.340644820295983e-05,
+      "loss": 0.1764,
+      "step": 6286
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.340116279069768e-05,
+      "loss": 0.2143,
+      "step": 6288
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.339587737843552e-05,
+      "loss": 0.1631,
+      "step": 6290
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.339059196617336e-05,
+      "loss": 0.3019,
+      "step": 6292
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.33853065539112e-05,
+      "loss": 0.1745,
+      "step": 6294
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.338002114164905e-05,
+      "loss": 0.189,
+      "step": 6296
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.3374735729386894e-05,
+      "loss": 0.1097,
+      "step": 6298
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.336945031712474e-05,
+      "loss": 0.1263,
+      "step": 6300
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.336416490486258e-05,
+      "loss": 0.1255,
+      "step": 6302
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.3358879492600426e-05,
+      "loss": 0.1996,
+      "step": 6304
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.3353594080338265e-05,
+      "loss": 0.2422,
+      "step": 6306
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.334830866807611e-05,
+      "loss": 0.111,
+      "step": 6308
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.334302325581396e-05,
+      "loss": 0.1312,
+      "step": 6310
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.3337737843551797e-05,
+      "loss": 0.2097,
+      "step": 6312
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.333245243128964e-05,
+      "loss": 0.2236,
+      "step": 6314
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.332716701902748e-05,
+      "loss": 0.2479,
+      "step": 6316
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.3321881606765335e-05,
+      "loss": 0.1397,
+      "step": 6318
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.3316596194503174e-05,
+      "loss": 0.1172,
+      "step": 6320
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.331131078224102e-05,
+      "loss": 0.1166,
+      "step": 6322
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.330602536997886e-05,
+      "loss": 0.1666,
+      "step": 6324
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.3300739957716705e-05,
+      "loss": 0.1743,
+      "step": 6326
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.3295454545454545e-05,
+      "loss": 0.1095,
+      "step": 6328
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.329016913319239e-05,
+      "loss": 0.1689,
+      "step": 6330
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.328488372093023e-05,
+      "loss": 0.1888,
+      "step": 6332
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.3279598308668076e-05,
+      "loss": 0.2615,
+      "step": 6334
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.327431289640592e-05,
+      "loss": 0.1308,
+      "step": 6336
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.326902748414377e-05,
+      "loss": 0.142,
+      "step": 6338
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.326374207188161e-05,
+      "loss": 0.3503,
+      "step": 6340
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.3258456659619454e-05,
+      "loss": 0.2034,
+      "step": 6342
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.325317124735729e-05,
+      "loss": 0.1894,
+      "step": 6344
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.324788583509514e-05,
+      "loss": 0.141,
+      "step": 6346
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.324260042283298e-05,
+      "loss": 0.1073,
+      "step": 6348
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.3237315010570824e-05,
+      "loss": 0.1851,
+      "step": 6350
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.323202959830867e-05,
+      "loss": 0.1882,
+      "step": 6352
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.3226744186046516e-05,
+      "loss": 0.1869,
+      "step": 6354
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.3221458773784356e-05,
+      "loss": 0.2836,
+      "step": 6356
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.32161733615222e-05,
+      "loss": 0.1976,
+      "step": 6358
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.321088794926005e-05,
+      "loss": 0.043,
+      "step": 6360
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.320560253699789e-05,
+      "loss": 0.0924,
+      "step": 6362
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.320031712473573e-05,
+      "loss": 0.2507,
+      "step": 6364
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.319503171247357e-05,
+      "loss": 0.232,
+      "step": 6366
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.318974630021142e-05,
+      "loss": 0.2588,
+      "step": 6368
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.318446088794926e-05,
+      "loss": 0.2609,
+      "step": 6370
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.317917547568711e-05,
+      "loss": 0.0929,
+      "step": 6372
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.317389006342495e-05,
+      "loss": 0.3943,
+      "step": 6374
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.3168604651162796e-05,
+      "loss": 0.1662,
+      "step": 6376
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.3163319238900635e-05,
+      "loss": 0.1248,
+      "step": 6378
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.315803382663848e-05,
+      "loss": 0.1521,
+      "step": 6380
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.315274841437632e-05,
+      "loss": 0.1616,
+      "step": 6382
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.3147463002114167e-05,
+      "loss": 0.1287,
+      "step": 6384
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.3142177589852006e-05,
+      "loss": 0.1336,
+      "step": 6386
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.313689217758985e-05,
+      "loss": 0.0605,
+      "step": 6388
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.31316067653277e-05,
+      "loss": 0.1347,
+      "step": 6390
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.3126321353065544e-05,
+      "loss": 0.2142,
+      "step": 6392
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.312103594080338e-05,
+      "loss": 0.1862,
+      "step": 6394
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.311575052854123e-05,
+      "loss": 0.1138,
+      "step": 6396
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.311046511627907e-05,
+      "loss": 0.1733,
+      "step": 6398
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.3105179704016915e-05,
+      "loss": 0.0542,
+      "step": 6400
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.3099894291754754e-05,
+      "loss": 0.1072,
+      "step": 6402
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.30946088794926e-05,
+      "loss": 0.2929,
+      "step": 6404
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.3089323467230446e-05,
+      "loss": 0.3078,
+      "step": 6406
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.308403805496829e-05,
+      "loss": 0.2085,
+      "step": 6408
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.307875264270613e-05,
+      "loss": 0.1387,
+      "step": 6410
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.307346723044398e-05,
+      "loss": 0.1443,
+      "step": 6412
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.3068181818181824e-05,
+      "loss": 0.2217,
+      "step": 6414
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.306289640591966e-05,
+      "loss": 0.1486,
+      "step": 6416
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.305761099365751e-05,
+      "loss": 0.1025,
+      "step": 6418
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.305232558139535e-05,
+      "loss": 0.1191,
+      "step": 6420
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.3047040169133194e-05,
+      "loss": 0.1424,
+      "step": 6422
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.3041754756871034e-05,
+      "loss": 0.2869,
+      "step": 6424
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.3036469344608886e-05,
+      "loss": 0.1608,
+      "step": 6426
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.3031183932346726e-05,
+      "loss": 0.1171,
+      "step": 6428
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.302589852008457e-05,
+      "loss": 0.0738,
+      "step": 6430
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.302061310782241e-05,
+      "loss": 0.2901,
+      "step": 6432
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.301532769556026e-05,
+      "loss": 0.2777,
+      "step": 6434
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.3010042283298096e-05,
+      "loss": 0.1199,
+      "step": 6436
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.300475687103594e-05,
+      "loss": 0.112,
+      "step": 6438
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.299947145877378e-05,
+      "loss": 0.1459,
+      "step": 6440
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.299418604651163e-05,
+      "loss": 0.0753,
+      "step": 6442
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.2988900634249474e-05,
+      "loss": 0.1264,
+      "step": 6444
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.298361522198732e-05,
+      "loss": 0.188,
+      "step": 6446
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.297832980972516e-05,
+      "loss": 0.1609,
+      "step": 6448
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.2973044397463005e-05,
+      "loss": 0.0904,
+      "step": 6450
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.2967758985200845e-05,
+      "loss": 0.1965,
+      "step": 6452
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.296247357293869e-05,
+      "loss": 0.1269,
+      "step": 6454
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.295718816067653e-05,
+      "loss": 0.1102,
+      "step": 6456
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.2951902748414376e-05,
+      "loss": 0.0769,
+      "step": 6458
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.294661733615222e-05,
+      "loss": 0.1696,
+      "step": 6460
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.294133192389007e-05,
+      "loss": 0.1866,
+      "step": 6462
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.293604651162791e-05,
+      "loss": 0.1734,
+      "step": 6464
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.293076109936575e-05,
+      "loss": 0.2824,
+      "step": 6466
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.29254756871036e-05,
+      "loss": 0.1817,
+      "step": 6468
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.292019027484144e-05,
+      "loss": 0.1357,
+      "step": 6470
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.2914904862579285e-05,
+      "loss": 0.2785,
+      "step": 6472
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.2909619450317124e-05,
+      "loss": 0.15,
+      "step": 6474
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.290433403805497e-05,
+      "loss": 0.1504,
+      "step": 6476
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.289904862579281e-05,
+      "loss": 0.3111,
+      "step": 6478
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.289376321353066e-05,
+      "loss": 0.1105,
+      "step": 6480
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.28884778012685e-05,
+      "loss": 0.0804,
+      "step": 6482
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.288319238900635e-05,
+      "loss": 0.23,
+      "step": 6484
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.287790697674419e-05,
+      "loss": 0.1211,
+      "step": 6486
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.287262156448203e-05,
+      "loss": 0.1741,
+      "step": 6488
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.286733615221987e-05,
+      "loss": 0.1537,
+      "step": 6490
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.286205073995772e-05,
+      "loss": 0.1826,
+      "step": 6492
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.285676532769556e-05,
+      "loss": 0.231,
+      "step": 6494
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.2851479915433404e-05,
+      "loss": 0.0593,
+      "step": 6496
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.284619450317125e-05,
+      "loss": 0.1864,
+      "step": 6498
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.2840909090909096e-05,
+      "loss": 0.1587,
+      "step": 6500
+    },
+    {
+      "epoch": 6.87,
+      "eval_cer": 0.018067825591336562,
+      "eval_loss": 0.6848969459533691,
+      "eval_runtime": 130.3001,
+      "eval_samples_per_second": 6.454,
+      "eval_steps_per_second": 0.814,
+      "step": 6500
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.2835623678646935e-05,
+      "loss": 0.0839,
+      "step": 6502
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.283033826638478e-05,
+      "loss": 0.2005,
+      "step": 6504
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.282505285412262e-05,
+      "loss": 0.086,
+      "step": 6506
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.2819767441860466e-05,
+      "loss": 0.0649,
+      "step": 6508
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.2814482029598306e-05,
+      "loss": 0.2194,
+      "step": 6510
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.280919661733615e-05,
+      "loss": 0.1811,
+      "step": 6512
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.2803911205074e-05,
+      "loss": 0.1294,
+      "step": 6514
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.2798625792811844e-05,
+      "loss": 0.1745,
+      "step": 6516
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.279334038054968e-05,
+      "loss": 0.2444,
+      "step": 6518
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.278805496828753e-05,
+      "loss": 0.1645,
+      "step": 6520
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.2782769556025375e-05,
+      "loss": 0.3193,
+      "step": 6522
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.2777484143763215e-05,
+      "loss": 0.262,
+      "step": 6524
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.277219873150106e-05,
+      "loss": 0.1257,
+      "step": 6526
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.27669133192389e-05,
+      "loss": 0.2004,
+      "step": 6528
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.2761627906976746e-05,
+      "loss": 0.1237,
+      "step": 6530
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.2756342494714585e-05,
+      "loss": 0.1606,
+      "step": 6532
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.275105708245244e-05,
+      "loss": 0.134,
+      "step": 6534
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.274577167019028e-05,
+      "loss": 0.2359,
+      "step": 6536
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.2740486257928123e-05,
+      "loss": 0.1267,
+      "step": 6538
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.273520084566596e-05,
+      "loss": 0.1402,
+      "step": 6540
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.272991543340381e-05,
+      "loss": 0.2543,
+      "step": 6542
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.272463002114165e-05,
+      "loss": 0.085,
+      "step": 6544
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.2719344608879494e-05,
+      "loss": 0.305,
+      "step": 6546
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.2714059196617333e-05,
+      "loss": 0.1374,
+      "step": 6548
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.270877378435518e-05,
+      "loss": 0.1432,
+      "step": 6550
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.2703488372093026e-05,
+      "loss": 0.1224,
+      "step": 6552
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.269820295983087e-05,
+      "loss": 0.1348,
+      "step": 6554
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.269291754756871e-05,
+      "loss": 0.1378,
+      "step": 6556
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.268763213530656e-05,
+      "loss": 0.1377,
+      "step": 6558
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.2682346723044396e-05,
+      "loss": 0.0546,
+      "step": 6560
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.267706131078224e-05,
+      "loss": 0.1603,
+      "step": 6562
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.267177589852009e-05,
+      "loss": 0.1801,
+      "step": 6564
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.266649048625793e-05,
+      "loss": 0.0764,
+      "step": 6566
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.2661205073995774e-05,
+      "loss": 0.1717,
+      "step": 6568
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.265591966173362e-05,
+      "loss": 0.1367,
+      "step": 6570
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.265063424947146e-05,
+      "loss": 0.2169,
+      "step": 6572
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.2645348837209305e-05,
+      "loss": 0.1314,
+      "step": 6574
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.264006342494715e-05,
+      "loss": 0.1091,
+      "step": 6576
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.263477801268499e-05,
+      "loss": 0.129,
+      "step": 6578
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.2629492600422836e-05,
+      "loss": 0.2033,
+      "step": 6580
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.2624207188160676e-05,
+      "loss": 0.1272,
+      "step": 6582
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.261892177589852e-05,
+      "loss": 0.211,
+      "step": 6584
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.261363636363636e-05,
+      "loss": 0.2411,
+      "step": 6586
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.2608350951374214e-05,
+      "loss": 0.1173,
+      "step": 6588
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.260306553911205e-05,
+      "loss": 0.1694,
+      "step": 6590
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.25977801268499e-05,
+      "loss": 0.15,
+      "step": 6592
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.259249471458774e-05,
+      "loss": 0.1222,
+      "step": 6594
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.2587209302325585e-05,
+      "loss": 0.0788,
+      "step": 6596
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.2581923890063424e-05,
+      "loss": 0.1771,
+      "step": 6598
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.257663847780127e-05,
+      "loss": 0.0798,
+      "step": 6600
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.257135306553911e-05,
+      "loss": 0.2348,
+      "step": 6602
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.2566067653276955e-05,
+      "loss": 0.1541,
+      "step": 6604
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.25607822410148e-05,
+      "loss": 0.0848,
+      "step": 6606
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.255549682875265e-05,
+      "loss": 0.0845,
+      "step": 6608
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.255021141649049e-05,
+      "loss": 0.1386,
+      "step": 6610
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.254492600422833e-05,
+      "loss": 0.07,
+      "step": 6612
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.253964059196617e-05,
+      "loss": 0.0768,
+      "step": 6614
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.253435517970402e-05,
+      "loss": 0.1933,
+      "step": 6616
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.2529069767441864e-05,
+      "loss": 0.1392,
+      "step": 6618
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.2523784355179703e-05,
+      "loss": 0.1438,
+      "step": 6620
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.251849894291755e-05,
+      "loss": 0.1893,
+      "step": 6622
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.2513213530655396e-05,
+      "loss": 0.1026,
+      "step": 6624
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.250792811839324e-05,
+      "loss": 0.1147,
+      "step": 6626
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.250264270613108e-05,
+      "loss": 0.0963,
+      "step": 6628
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.249735729386893e-05,
+      "loss": 0.0652,
+      "step": 6630
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.2492071881606766e-05,
+      "loss": 0.0809,
+      "step": 6632
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.248678646934461e-05,
+      "loss": 0.1304,
+      "step": 6634
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.248150105708245e-05,
+      "loss": 0.1327,
+      "step": 6636
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.24762156448203e-05,
+      "loss": 0.1143,
+      "step": 6638
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.247093023255814e-05,
+      "loss": 0.0977,
+      "step": 6640
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.246564482029599e-05,
+      "loss": 0.149,
+      "step": 6642
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.246035940803383e-05,
+      "loss": 0.1369,
+      "step": 6644
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.2455073995771675e-05,
+      "loss": 0.1006,
+      "step": 6646
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.2449788583509514e-05,
+      "loss": 0.1559,
+      "step": 6648
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.244450317124736e-05,
+      "loss": 0.0658,
+      "step": 6650
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.24392177589852e-05,
+      "loss": 0.2514,
+      "step": 6652
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.2433932346723046e-05,
+      "loss": 0.1672,
+      "step": 6654
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.2428646934460885e-05,
+      "loss": 0.1332,
+      "step": 6656
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.242336152219873e-05,
+      "loss": 0.1938,
+      "step": 6658
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.241807610993658e-05,
+      "loss": 0.1068,
+      "step": 6660
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.241279069767442e-05,
+      "loss": 0.1107,
+      "step": 6662
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.240750528541226e-05,
+      "loss": 0.0745,
+      "step": 6664
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.240221987315011e-05,
+      "loss": 0.0882,
+      "step": 6666
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.239693446088795e-05,
+      "loss": 0.0983,
+      "step": 6668
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.2391649048625794e-05,
+      "loss": 0.0925,
+      "step": 6670
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.238636363636364e-05,
+      "loss": 0.1857,
+      "step": 6672
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.238107822410148e-05,
+      "loss": 0.1183,
+      "step": 6674
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.2375792811839325e-05,
+      "loss": 0.067,
+      "step": 6676
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.237050739957717e-05,
+      "loss": 0.1301,
+      "step": 6678
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.236522198731502e-05,
+      "loss": 0.1827,
+      "step": 6680
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.235993657505286e-05,
+      "loss": 0.0783,
+      "step": 6682
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.23546511627907e-05,
+      "loss": 0.0923,
+      "step": 6684
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.234936575052854e-05,
+      "loss": 0.1305,
+      "step": 6686
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.234408033826639e-05,
+      "loss": 0.0894,
+      "step": 6688
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.233879492600423e-05,
+      "loss": 0.1987,
+      "step": 6690
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.2333509513742073e-05,
+      "loss": 0.1331,
+      "step": 6692
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.232822410147991e-05,
+      "loss": 0.1275,
+      "step": 6694
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.2322938689217766e-05,
+      "loss": 0.1744,
+      "step": 6696
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.2317653276955605e-05,
+      "loss": 0.0823,
+      "step": 6698
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.231236786469345e-05,
+      "loss": 0.1182,
+      "step": 6700
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.230708245243129e-05,
+      "loss": 0.1176,
+      "step": 6702
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.2301797040169136e-05,
+      "loss": 0.1704,
+      "step": 6704
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.2296511627906976e-05,
+      "loss": 0.1159,
+      "step": 6706
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.229122621564482e-05,
+      "loss": 0.1607,
+      "step": 6708
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.228594080338266e-05,
+      "loss": 0.0877,
+      "step": 6710
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.228065539112051e-05,
+      "loss": 0.1278,
+      "step": 6712
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.227536997885835e-05,
+      "loss": 0.1083,
+      "step": 6714
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.22700845665962e-05,
+      "loss": 0.1,
+      "step": 6716
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.226479915433404e-05,
+      "loss": 0.1898,
+      "step": 6718
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.2259513742071884e-05,
+      "loss": 0.0798,
+      "step": 6720
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.2254228329809724e-05,
+      "loss": 0.0928,
+      "step": 6722
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.224894291754757e-05,
+      "loss": 0.1268,
+      "step": 6724
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.2243657505285416e-05,
+      "loss": 0.1476,
+      "step": 6726
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.2238372093023255e-05,
+      "loss": 0.0909,
+      "step": 6728
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.22330866807611e-05,
+      "loss": 0.1367,
+      "step": 6730
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.222780126849895e-05,
+      "loss": 0.1678,
+      "step": 6732
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.222251585623679e-05,
+      "loss": 0.0876,
+      "step": 6734
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.221723044397463e-05,
+      "loss": 0.1025,
+      "step": 6736
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.221194503171248e-05,
+      "loss": 0.1406,
+      "step": 6738
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.220665961945032e-05,
+      "loss": 0.1127,
+      "step": 6740
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.2201374207188164e-05,
+      "loss": 0.0851,
+      "step": 6742
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.2196088794926e-05,
+      "loss": 0.1614,
+      "step": 6744
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.219080338266385e-05,
+      "loss": 0.0852,
+      "step": 6746
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.218551797040169e-05,
+      "loss": 0.0843,
+      "step": 6748
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.218023255813954e-05,
+      "loss": 0.1957,
+      "step": 6750
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.217494714587738e-05,
+      "loss": 0.2616,
+      "step": 6752
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.216966173361523e-05,
+      "loss": 0.203,
+      "step": 6754
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.2164376321353066e-05,
+      "loss": 0.115,
+      "step": 6756
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.215909090909091e-05,
+      "loss": 0.1622,
+      "step": 6758
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.215380549682875e-05,
+      "loss": 0.0653,
+      "step": 6760
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.21485200845666e-05,
+      "loss": 0.1484,
+      "step": 6762
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.214323467230444e-05,
+      "loss": 0.2066,
+      "step": 6764
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.213794926004228e-05,
+      "loss": 0.2578,
+      "step": 6766
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.213266384778013e-05,
+      "loss": 0.1562,
+      "step": 6768
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.2127378435517975e-05,
+      "loss": 0.2174,
+      "step": 6770
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.2122093023255814e-05,
+      "loss": 0.0996,
+      "step": 6772
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.211680761099366e-05,
+      "loss": 0.144,
+      "step": 6774
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.21115221987315e-05,
+      "loss": 0.2435,
+      "step": 6776
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.2106236786469346e-05,
+      "loss": 0.1747,
+      "step": 6778
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.210095137420719e-05,
+      "loss": 0.1178,
+      "step": 6780
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.209566596194503e-05,
+      "loss": 0.0761,
+      "step": 6782
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.209038054968288e-05,
+      "loss": 0.1601,
+      "step": 6784
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.2085095137420716e-05,
+      "loss": 0.1881,
+      "step": 6786
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.207980972515857e-05,
+      "loss": 0.0601,
+      "step": 6788
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.207452431289641e-05,
+      "loss": 0.0526,
+      "step": 6790
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.2069238900634254e-05,
+      "loss": 0.1336,
+      "step": 6792
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.2063953488372094e-05,
+      "loss": 0.3576,
+      "step": 6794
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.205866807610994e-05,
+      "loss": 0.1493,
+      "step": 6796
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.205338266384778e-05,
+      "loss": 0.1175,
+      "step": 6798
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.2048097251585625e-05,
+      "loss": 0.0731,
+      "step": 6800
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.2042811839323464e-05,
+      "loss": 0.1651,
+      "step": 6802
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.203752642706131e-05,
+      "loss": 0.1258,
+      "step": 6804
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.2032241014799157e-05,
+      "loss": 0.0815,
+      "step": 6806
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.2026955602537e-05,
+      "loss": 0.1047,
+      "step": 6808
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.202167019027484e-05,
+      "loss": 0.0978,
+      "step": 6810
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.201638477801269e-05,
+      "loss": 0.122,
+      "step": 6812
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.201109936575053e-05,
+      "loss": 0.2287,
+      "step": 6814
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.200581395348837e-05,
+      "loss": 0.1758,
+      "step": 6816
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.200052854122621e-05,
+      "loss": 0.1481,
+      "step": 6818
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.199524312896406e-05,
+      "loss": 0.1527,
+      "step": 6820
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.1989957716701905e-05,
+      "loss": 0.0681,
+      "step": 6822
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.198467230443975e-05,
+      "loss": 0.34,
+      "step": 6824
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.197938689217759e-05,
+      "loss": 0.0775,
+      "step": 6826
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.1974101479915436e-05,
+      "loss": 0.1494,
+      "step": 6828
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.196881606765328e-05,
+      "loss": 0.1659,
+      "step": 6830
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.196353065539112e-05,
+      "loss": 0.1885,
+      "step": 6832
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.195824524312897e-05,
+      "loss": 0.1345,
+      "step": 6834
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.195295983086681e-05,
+      "loss": 0.0972,
+      "step": 6836
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.194767441860465e-05,
+      "loss": 0.096,
+      "step": 6838
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.194238900634249e-05,
+      "loss": 0.2733,
+      "step": 6840
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.1937103594080345e-05,
+      "loss": 0.1251,
+      "step": 6842
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.1931818181818184e-05,
+      "loss": 0.1594,
+      "step": 6844
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.192653276955603e-05,
+      "loss": 0.1894,
+      "step": 6846
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.192124735729387e-05,
+      "loss": 0.125,
+      "step": 6848
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.1915961945031716e-05,
+      "loss": 0.0435,
+      "step": 6850
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.1910676532769555e-05,
+      "loss": 0.1681,
+      "step": 6852
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.19053911205074e-05,
+      "loss": 0.0633,
+      "step": 6854
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.190010570824524e-05,
+      "loss": 0.0713,
+      "step": 6856
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.1894820295983086e-05,
+      "loss": 0.0999,
+      "step": 6858
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.188953488372093e-05,
+      "loss": 0.1284,
+      "step": 6860
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.188424947145878e-05,
+      "loss": 0.0723,
+      "step": 6862
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.187896405919662e-05,
+      "loss": 0.0406,
+      "step": 6864
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.1873678646934464e-05,
+      "loss": 0.0977,
+      "step": 6866
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.18683932346723e-05,
+      "loss": 0.0727,
+      "step": 6868
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.186310782241015e-05,
+      "loss": 0.2983,
+      "step": 6870
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.185782241014799e-05,
+      "loss": 0.1568,
+      "step": 6872
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.1852536997885834e-05,
+      "loss": 0.306,
+      "step": 6874
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.184725158562368e-05,
+      "loss": 0.4258,
+      "step": 6876
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.1841966173361527e-05,
+      "loss": 0.1543,
+      "step": 6878
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.1836680761099366e-05,
+      "loss": 0.1296,
+      "step": 6880
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.183139534883721e-05,
+      "loss": 0.1369,
+      "step": 6882
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.182610993657506e-05,
+      "loss": 0.1389,
+      "step": 6884
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.18208245243129e-05,
+      "loss": 0.1247,
+      "step": 6886
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.181553911205074e-05,
+      "loss": 0.1223,
+      "step": 6888
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.181025369978858e-05,
+      "loss": 0.1556,
+      "step": 6890
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.180496828752643e-05,
+      "loss": 0.1862,
+      "step": 6892
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.179968287526427e-05,
+      "loss": 0.1612,
+      "step": 6894
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.179439746300212e-05,
+      "loss": 0.1985,
+      "step": 6896
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.178911205073996e-05,
+      "loss": 0.12,
+      "step": 6898
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.1783826638477806e-05,
+      "loss": 0.2225,
+      "step": 6900
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.1778541226215645e-05,
+      "loss": 0.1292,
+      "step": 6902
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.177325581395349e-05,
+      "loss": 0.0941,
+      "step": 6904
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.176797040169133e-05,
+      "loss": 0.318,
+      "step": 6906
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.176268498942918e-05,
+      "loss": 0.3259,
+      "step": 6908
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.1757399577167016e-05,
+      "loss": 0.1176,
+      "step": 6910
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.175211416490486e-05,
+      "loss": 0.1663,
+      "step": 6912
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.174682875264271e-05,
+      "loss": 0.171,
+      "step": 6914
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.1741543340380554e-05,
+      "loss": 0.2284,
+      "step": 6916
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.1736257928118394e-05,
+      "loss": 0.1861,
+      "step": 6918
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.173097251585624e-05,
+      "loss": 0.2835,
+      "step": 6920
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.172568710359408e-05,
+      "loss": 0.149,
+      "step": 6922
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.1720401691331925e-05,
+      "loss": 0.0834,
+      "step": 6924
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.1715116279069764e-05,
+      "loss": 0.1018,
+      "step": 6926
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.170983086680761e-05,
+      "loss": 0.057,
+      "step": 6928
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.1704545454545456e-05,
+      "loss": 0.1553,
+      "step": 6930
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.16992600422833e-05,
+      "loss": 0.0442,
+      "step": 6932
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.169397463002114e-05,
+      "loss": 0.0943,
+      "step": 6934
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.168868921775899e-05,
+      "loss": 0.1423,
+      "step": 6936
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.1683403805496834e-05,
+      "loss": 0.101,
+      "step": 6938
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.167811839323467e-05,
+      "loss": 0.0697,
+      "step": 6940
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.167283298097252e-05,
+      "loss": 0.0719,
+      "step": 6942
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.166754756871036e-05,
+      "loss": 0.2244,
+      "step": 6944
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.1662262156448205e-05,
+      "loss": 0.0793,
+      "step": 6946
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.1656976744186044e-05,
+      "loss": 0.0689,
+      "step": 6948
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.16516913319239e-05,
+      "loss": 0.1631,
+      "step": 6950
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.1646405919661736e-05,
+      "loss": 0.057,
+      "step": 6952
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.164112050739958e-05,
+      "loss": 0.1282,
+      "step": 6954
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.163583509513742e-05,
+      "loss": 0.183,
+      "step": 6956
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.163054968287527e-05,
+      "loss": 0.114,
+      "step": 6958
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.162526427061311e-05,
+      "loss": 0.1625,
+      "step": 6960
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.161997885835095e-05,
+      "loss": 0.1943,
+      "step": 6962
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.161469344608879e-05,
+      "loss": 0.1572,
+      "step": 6964
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.160940803382664e-05,
+      "loss": 0.1261,
+      "step": 6966
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.1604122621564484e-05,
+      "loss": 0.0798,
+      "step": 6968
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.159883720930233e-05,
+      "loss": 0.1337,
+      "step": 6970
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.159355179704017e-05,
+      "loss": 0.0814,
+      "step": 6972
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.1588266384778015e-05,
+      "loss": 0.0963,
+      "step": 6974
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.1582980972515855e-05,
+      "loss": 0.1317,
+      "step": 6976
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.15776955602537e-05,
+      "loss": 0.1232,
+      "step": 6978
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.157241014799155e-05,
+      "loss": 0.1589,
+      "step": 6980
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.1567124735729386e-05,
+      "loss": 0.1632,
+      "step": 6982
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.156183932346723e-05,
+      "loss": 0.0875,
+      "step": 6984
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.155655391120508e-05,
+      "loss": 0.1064,
+      "step": 6986
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.155126849894292e-05,
+      "loss": 0.134,
+      "step": 6988
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.1545983086680764e-05,
+      "loss": 0.0632,
+      "step": 6990
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.154069767441861e-05,
+      "loss": 0.1357,
+      "step": 6992
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.153541226215645e-05,
+      "loss": 0.0639,
+      "step": 6994
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.1530126849894295e-05,
+      "loss": 0.0779,
+      "step": 6996
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.1524841437632134e-05,
+      "loss": 0.1556,
+      "step": 6998
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.151955602536998e-05,
+      "loss": 0.2285,
+      "step": 7000
+    },
+    {
+      "epoch": 7.4,
+      "eval_cer": 0.016699914505557138,
+      "eval_loss": 0.7643967270851135,
+      "eval_runtime": 129.8037,
+      "eval_samples_per_second": 6.479,
+      "eval_steps_per_second": 0.817,
+      "step": 7000
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.151427061310782e-05,
+      "loss": 0.0917,
+      "step": 7002
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.150898520084567e-05,
+      "loss": 0.1798,
+      "step": 7004
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.150369978858351e-05,
+      "loss": 0.122,
+      "step": 7006
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.149841437632136e-05,
+      "loss": 0.0871,
+      "step": 7008
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.14931289640592e-05,
+      "loss": 0.0828,
+      "step": 7010
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.148784355179704e-05,
+      "loss": 0.1366,
+      "step": 7012
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.148255813953488e-05,
+      "loss": 0.1001,
+      "step": 7014
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.147727272727273e-05,
+      "loss": 0.1294,
+      "step": 7016
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.147198731501057e-05,
+      "loss": 0.116,
+      "step": 7018
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.1466701902748414e-05,
+      "loss": 0.0794,
+      "step": 7020
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.146141649048626e-05,
+      "loss": 0.0282,
+      "step": 7022
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.1456131078224106e-05,
+      "loss": 0.1102,
+      "step": 7024
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.1450845665961945e-05,
+      "loss": 0.0907,
+      "step": 7026
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.144556025369979e-05,
+      "loss": 0.097,
+      "step": 7028
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.144027484143763e-05,
+      "loss": 0.1432,
+      "step": 7030
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.143498942917548e-05,
+      "loss": 0.1493,
+      "step": 7032
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.142970401691332e-05,
+      "loss": 0.2088,
+      "step": 7034
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.142441860465116e-05,
+      "loss": 0.5473,
+      "step": 7036
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.141913319238901e-05,
+      "loss": 0.1131,
+      "step": 7038
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.1413847780126854e-05,
+      "loss": 0.0682,
+      "step": 7040
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.1408562367864693e-05,
+      "loss": 0.1884,
+      "step": 7042
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.140327695560254e-05,
+      "loss": 0.1398,
+      "step": 7044
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.1397991543340386e-05,
+      "loss": 0.2104,
+      "step": 7046
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.1392706131078225e-05,
+      "loss": 0.089,
+      "step": 7048
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.138742071881607e-05,
+      "loss": 0.2262,
+      "step": 7050
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.138213530655391e-05,
+      "loss": 0.3409,
+      "step": 7052
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.1376849894291756e-05,
+      "loss": 0.1543,
+      "step": 7054
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.1371564482029595e-05,
+      "loss": 0.079,
+      "step": 7056
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.136627906976745e-05,
+      "loss": 0.172,
+      "step": 7058
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.136099365750529e-05,
+      "loss": 0.1977,
+      "step": 7060
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.1355708245243134e-05,
+      "loss": 0.2035,
+      "step": 7062
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.135042283298097e-05,
+      "loss": 0.0555,
+      "step": 7064
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.134513742071882e-05,
+      "loss": 0.0926,
+      "step": 7066
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.133985200845666e-05,
+      "loss": 0.1551,
+      "step": 7068
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.1334566596194504e-05,
+      "loss": 0.1136,
+      "step": 7070
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.1329281183932344e-05,
+      "loss": 0.1401,
+      "step": 7072
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.132399577167019e-05,
+      "loss": 0.1895,
+      "step": 7074
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.1318710359408036e-05,
+      "loss": 0.1383,
+      "step": 7076
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.131342494714588e-05,
+      "loss": 0.1553,
+      "step": 7078
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.130813953488372e-05,
+      "loss": 0.0743,
+      "step": 7080
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.130285412262157e-05,
+      "loss": 0.0585,
+      "step": 7082
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.1297568710359406e-05,
+      "loss": 0.0266,
+      "step": 7084
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.129228329809725e-05,
+      "loss": 0.0958,
+      "step": 7086
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.12869978858351e-05,
+      "loss": 0.2182,
+      "step": 7088
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.128171247357294e-05,
+      "loss": 0.1488,
+      "step": 7090
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.1276427061310784e-05,
+      "loss": 0.2245,
+      "step": 7092
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.127114164904863e-05,
+      "loss": 0.1023,
+      "step": 7094
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.1265856236786476e-05,
+      "loss": 0.2489,
+      "step": 7096
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.1260570824524315e-05,
+      "loss": 0.2633,
+      "step": 7098
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.125528541226216e-05,
+      "loss": 0.1263,
+      "step": 7100
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.125e-05,
+      "loss": 0.1476,
+      "step": 7102
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.124471458773785e-05,
+      "loss": 0.1078,
+      "step": 7104
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.1239429175475686e-05,
+      "loss": 0.1322,
+      "step": 7106
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.123414376321353e-05,
+      "loss": 0.1396,
+      "step": 7108
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.122885835095137e-05,
+      "loss": 0.1505,
+      "step": 7110
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.1223572938689224e-05,
+      "loss": 0.2306,
+      "step": 7112
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.1218287526427063e-05,
+      "loss": 0.115,
+      "step": 7114
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.121300211416491e-05,
+      "loss": 0.0433,
+      "step": 7116
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.120771670190275e-05,
+      "loss": 0.059,
+      "step": 7118
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.1202431289640595e-05,
+      "loss": 0.0396,
+      "step": 7120
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.1197145877378434e-05,
+      "loss": 0.1071,
+      "step": 7122
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.119186046511628e-05,
+      "loss": 0.1492,
+      "step": 7124
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.118657505285412e-05,
+      "loss": 0.1364,
+      "step": 7126
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.1181289640591966e-05,
+      "loss": 0.365,
+      "step": 7128
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.117600422832981e-05,
+      "loss": 0.1789,
+      "step": 7130
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.117071881606766e-05,
+      "loss": 0.0941,
+      "step": 7132
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.11654334038055e-05,
+      "loss": 0.1209,
+      "step": 7134
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.116014799154334e-05,
+      "loss": 0.2109,
+      "step": 7136
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.115486257928118e-05,
+      "loss": 0.0642,
+      "step": 7138
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.114957716701903e-05,
+      "loss": 0.1333,
+      "step": 7140
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.1144291754756874e-05,
+      "loss": 0.0888,
+      "step": 7142
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.1139006342494714e-05,
+      "loss": 0.1243,
+      "step": 7144
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.113372093023256e-05,
+      "loss": 0.0929,
+      "step": 7146
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.1128435517970406e-05,
+      "loss": 0.1093,
+      "step": 7148
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.112315010570825e-05,
+      "loss": 0.1267,
+      "step": 7150
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.111786469344609e-05,
+      "loss": 0.0882,
+      "step": 7152
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.111257928118394e-05,
+      "loss": 0.2262,
+      "step": 7154
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.1107293868921776e-05,
+      "loss": 0.0967,
+      "step": 7156
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.110200845665962e-05,
+      "loss": 0.2711,
+      "step": 7158
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.109672304439746e-05,
+      "loss": 0.0687,
+      "step": 7160
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.109143763213531e-05,
+      "loss": 0.2071,
+      "step": 7162
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.108615221987315e-05,
+      "loss": 0.1116,
+      "step": 7164
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.1080866807611e-05,
+      "loss": 0.1037,
+      "step": 7166
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.107558139534884e-05,
+      "loss": 0.1094,
+      "step": 7168
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.1070295983086685e-05,
+      "loss": 0.123,
+      "step": 7170
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.1065010570824525e-05,
+      "loss": 0.1824,
+      "step": 7172
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.105972515856237e-05,
+      "loss": 0.0796,
+      "step": 7174
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.105443974630021e-05,
+      "loss": 0.0495,
+      "step": 7176
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.1049154334038056e-05,
+      "loss": 0.1828,
+      "step": 7178
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.1043868921775895e-05,
+      "loss": 0.0642,
+      "step": 7180
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.103858350951374e-05,
+      "loss": 0.1328,
+      "step": 7182
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.103329809725159e-05,
+      "loss": 0.1206,
+      "step": 7184
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.1028012684989433e-05,
+      "loss": 0.1608,
+      "step": 7186
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.102272727272727e-05,
+      "loss": 0.157,
+      "step": 7188
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.101744186046512e-05,
+      "loss": 0.0986,
+      "step": 7190
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.101215644820296e-05,
+      "loss": 0.1187,
+      "step": 7192
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.1006871035940804e-05,
+      "loss": 0.0329,
+      "step": 7194
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.100158562367865e-05,
+      "loss": 0.138,
+      "step": 7196
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.099630021141649e-05,
+      "loss": 0.1953,
+      "step": 7198
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.0991014799154336e-05,
+      "loss": 0.2044,
+      "step": 7200
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.098572938689218e-05,
+      "loss": 0.0648,
+      "step": 7202
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.098044397463003e-05,
+      "loss": 0.1554,
+      "step": 7204
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.097515856236787e-05,
+      "loss": 0.1202,
+      "step": 7206
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.096987315010571e-05,
+      "loss": 0.1143,
+      "step": 7208
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.096458773784355e-05,
+      "loss": 0.0999,
+      "step": 7210
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.09593023255814e-05,
+      "loss": 0.1562,
+      "step": 7212
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.095401691331924e-05,
+      "loss": 0.1498,
+      "step": 7214
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.0948731501057084e-05,
+      "loss": 0.1716,
+      "step": 7216
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.094344608879492e-05,
+      "loss": 0.1226,
+      "step": 7218
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.0938160676532776e-05,
+      "loss": 0.1101,
+      "step": 7220
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.0932875264270615e-05,
+      "loss": 0.0776,
+      "step": 7222
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.092758985200846e-05,
+      "loss": 0.1376,
+      "step": 7224
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.09223044397463e-05,
+      "loss": 0.1301,
+      "step": 7226
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.0917019027484147e-05,
+      "loss": 0.1084,
+      "step": 7228
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.0911733615221986e-05,
+      "loss": 0.1025,
+      "step": 7230
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.090644820295983e-05,
+      "loss": 0.0797,
+      "step": 7232
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.090116279069767e-05,
+      "loss": 0.0559,
+      "step": 7234
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.089587737843552e-05,
+      "loss": 0.1271,
+      "step": 7236
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.089059196617336e-05,
+      "loss": 0.1511,
+      "step": 7238
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.088530655391121e-05,
+      "loss": 0.0656,
+      "step": 7240
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.088002114164905e-05,
+      "loss": 0.3565,
+      "step": 7242
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.0874735729386895e-05,
+      "loss": 0.1438,
+      "step": 7244
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.0869450317124734e-05,
+      "loss": 0.1133,
+      "step": 7246
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.086416490486258e-05,
+      "loss": 0.1269,
+      "step": 7248
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.0858879492600426e-05,
+      "loss": 0.0913,
+      "step": 7250
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.0853594080338265e-05,
+      "loss": 0.0962,
+      "step": 7252
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.084830866807611e-05,
+      "loss": 0.2436,
+      "step": 7254
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.084302325581396e-05,
+      "loss": 0.1272,
+      "step": 7256
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.0837737843551804e-05,
+      "loss": 0.0837,
+      "step": 7258
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.083245243128964e-05,
+      "loss": 0.084,
+      "step": 7260
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.082716701902749e-05,
+      "loss": 0.1018,
+      "step": 7262
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.082188160676533e-05,
+      "loss": 0.0634,
+      "step": 7264
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.0816596194503174e-05,
+      "loss": 0.0727,
+      "step": 7266
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.0811310782241014e-05,
+      "loss": 0.0855,
+      "step": 7268
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.080602536997886e-05,
+      "loss": 0.1704,
+      "step": 7270
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.08007399577167e-05,
+      "loss": 0.1212,
+      "step": 7272
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.079545454545455e-05,
+      "loss": 0.1861,
+      "step": 7274
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.079016913319239e-05,
+      "loss": 0.1109,
+      "step": 7276
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.078488372093024e-05,
+      "loss": 0.148,
+      "step": 7278
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.0779598308668076e-05,
+      "loss": 0.107,
+      "step": 7280
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.077431289640592e-05,
+      "loss": 0.1321,
+      "step": 7282
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.076902748414376e-05,
+      "loss": 0.1371,
+      "step": 7284
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.076374207188161e-05,
+      "loss": 0.0779,
+      "step": 7286
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.075845665961945e-05,
+      "loss": 0.2148,
+      "step": 7288
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.075317124735729e-05,
+      "loss": 0.083,
+      "step": 7290
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.074788583509514e-05,
+      "loss": 0.1508,
+      "step": 7292
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.0742600422832985e-05,
+      "loss": 0.1394,
+      "step": 7294
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.0737315010570824e-05,
+      "loss": 0.221,
+      "step": 7296
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.073202959830867e-05,
+      "loss": 0.1494,
+      "step": 7298
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.0726744186046517e-05,
+      "loss": 0.121,
+      "step": 7300
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.0721458773784356e-05,
+      "loss": 0.1156,
+      "step": 7302
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.07161733615222e-05,
+      "loss": 0.0434,
+      "step": 7304
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.071088794926004e-05,
+      "loss": 0.138,
+      "step": 7306
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.070560253699789e-05,
+      "loss": 0.1409,
+      "step": 7308
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.070031712473573e-05,
+      "loss": 0.1047,
+      "step": 7310
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.069503171247358e-05,
+      "loss": 0.1325,
+      "step": 7312
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.068974630021142e-05,
+      "loss": 0.1291,
+      "step": 7314
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.0684460887949265e-05,
+      "loss": 0.0921,
+      "step": 7316
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.0679175475687104e-05,
+      "loss": 0.0725,
+      "step": 7318
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.067389006342495e-05,
+      "loss": 0.13,
+      "step": 7320
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.066860465116279e-05,
+      "loss": 0.0861,
+      "step": 7322
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.0663319238900635e-05,
+      "loss": 0.0812,
+      "step": 7324
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.0658033826638475e-05,
+      "loss": 0.2347,
+      "step": 7326
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.065274841437633e-05,
+      "loss": 0.0432,
+      "step": 7328
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.064746300211417e-05,
+      "loss": 0.168,
+      "step": 7330
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.064217758985201e-05,
+      "loss": 0.1131,
+      "step": 7332
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.063689217758985e-05,
+      "loss": 0.155,
+      "step": 7334
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.06316067653277e-05,
+      "loss": 0.1038,
+      "step": 7336
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.062632135306554e-05,
+      "loss": 0.1024,
+      "step": 7338
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.0621035940803384e-05,
+      "loss": 0.108,
+      "step": 7340
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.061575052854122e-05,
+      "loss": 0.0993,
+      "step": 7342
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.061046511627907e-05,
+      "loss": 0.1252,
+      "step": 7344
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.0605179704016915e-05,
+      "loss": 0.1117,
+      "step": 7346
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.059989429175476e-05,
+      "loss": 0.0714,
+      "step": 7348
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.05946088794926e-05,
+      "loss": 0.152,
+      "step": 7350
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.0589323467230446e-05,
+      "loss": 0.2915,
+      "step": 7352
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.058403805496829e-05,
+      "loss": 0.1396,
+      "step": 7354
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.057875264270613e-05,
+      "loss": 0.1844,
+      "step": 7356
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.057346723044398e-05,
+      "loss": 0.1751,
+      "step": 7358
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.056818181818182e-05,
+      "loss": 0.1955,
+      "step": 7360
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.056289640591966e-05,
+      "loss": 0.1451,
+      "step": 7362
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.055761099365751e-05,
+      "loss": 0.131,
+      "step": 7364
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.0552325581395355e-05,
+      "loss": 0.094,
+      "step": 7366
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.0547040169133195e-05,
+      "loss": 0.1367,
+      "step": 7368
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.054175475687104e-05,
+      "loss": 0.1617,
+      "step": 7370
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.053646934460888e-05,
+      "loss": 0.213,
+      "step": 7372
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.0531183932346726e-05,
+      "loss": 0.0715,
+      "step": 7374
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.0525898520084565e-05,
+      "loss": 0.1576,
+      "step": 7376
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.052061310782241e-05,
+      "loss": 0.1469,
+      "step": 7378
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.0515327695560254e-05,
+      "loss": 0.0214,
+      "step": 7380
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.05100422832981e-05,
+      "loss": 0.1703,
+      "step": 7382
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.0504756871035946e-05,
+      "loss": 0.0795,
+      "step": 7384
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.049947145877379e-05,
+      "loss": 0.0787,
+      "step": 7386
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.049418604651163e-05,
+      "loss": 0.1197,
+      "step": 7388
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.0488900634249474e-05,
+      "loss": 0.0709,
+      "step": 7390
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.0483615221987317e-05,
+      "loss": 0.0778,
+      "step": 7392
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.047832980972516e-05,
+      "loss": 0.2426,
+      "step": 7394
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.0473044397463002e-05,
+      "loss": 0.0691,
+      "step": 7396
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.0467758985200845e-05,
+      "loss": 0.1296,
+      "step": 7398
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.0462473572938694e-05,
+      "loss": 0.1927,
+      "step": 7400
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.0457188160676537e-05,
+      "loss": 0.1771,
+      "step": 7402
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.045190274841438e-05,
+      "loss": 0.1729,
+      "step": 7404
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.0446617336152222e-05,
+      "loss": 0.1181,
+      "step": 7406
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.0441331923890065e-05,
+      "loss": 0.1441,
+      "step": 7408
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.0436046511627908e-05,
+      "loss": 0.1117,
+      "step": 7410
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.043076109936575e-05,
+      "loss": 0.1002,
+      "step": 7412
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.0425475687103593e-05,
+      "loss": 0.1109,
+      "step": 7414
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.0420190274841436e-05,
+      "loss": 0.1843,
+      "step": 7416
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.0414904862579285e-05,
+      "loss": 0.2138,
+      "step": 7418
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.0409619450317128e-05,
+      "loss": 0.0987,
+      "step": 7420
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.040433403805497e-05,
+      "loss": 0.2392,
+      "step": 7422
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.0399048625792813e-05,
+      "loss": 0.2863,
+      "step": 7424
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.0393763213530656e-05,
+      "loss": 0.1728,
+      "step": 7426
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.0388477801268502e-05,
+      "loss": 0.21,
+      "step": 7428
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.0383192389006344e-05,
+      "loss": 0.1339,
+      "step": 7430
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.0377906976744187e-05,
+      "loss": 0.0803,
+      "step": 7432
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.037262156448203e-05,
+      "loss": 0.303,
+      "step": 7434
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.0367336152219876e-05,
+      "loss": 0.1453,
+      "step": 7436
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.0362050739957722e-05,
+      "loss": 0.0911,
+      "step": 7438
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.0356765327695565e-05,
+      "loss": 0.1804,
+      "step": 7440
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.0351479915433407e-05,
+      "loss": 0.0878,
+      "step": 7442
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.034619450317125e-05,
+      "loss": 0.1076,
+      "step": 7444
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.0340909090909093e-05,
+      "loss": 0.0919,
+      "step": 7446
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.0335623678646935e-05,
+      "loss": 0.1405,
+      "step": 7448
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.0330338266384778e-05,
+      "loss": 0.2786,
+      "step": 7450
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.032505285412262e-05,
+      "loss": 0.0523,
+      "step": 7452
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.0319767441860463e-05,
+      "loss": 0.1314,
+      "step": 7454
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.0314482029598313e-05,
+      "loss": 0.0958,
+      "step": 7456
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.0309196617336155e-05,
+      "loss": 0.1255,
+      "step": 7458
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.0303911205073998e-05,
+      "loss": 0.1244,
+      "step": 7460
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.029862579281184e-05,
+      "loss": 0.1663,
+      "step": 7462
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.0293340380549683e-05,
+      "loss": 0.092,
+      "step": 7464
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.0288054968287526e-05,
+      "loss": 0.1858,
+      "step": 7466
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.028276955602537e-05,
+      "loss": 0.1454,
+      "step": 7468
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.027748414376321e-05,
+      "loss": 0.1785,
+      "step": 7470
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.0272198731501057e-05,
+      "loss": 0.1131,
+      "step": 7472
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.0266913319238904e-05,
+      "loss": 0.1604,
+      "step": 7474
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.0261627906976746e-05,
+      "loss": 0.0588,
+      "step": 7476
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.025634249471459e-05,
+      "loss": 0.1184,
+      "step": 7478
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.025105708245243e-05,
+      "loss": 0.1357,
+      "step": 7480
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.0245771670190278e-05,
+      "loss": 0.0666,
+      "step": 7482
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.024048625792812e-05,
+      "loss": 0.0618,
+      "step": 7484
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.0235200845665963e-05,
+      "loss": 0.1297,
+      "step": 7486
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.0229915433403806e-05,
+      "loss": 0.0984,
+      "step": 7488
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.0224630021141648e-05,
+      "loss": 0.2079,
+      "step": 7490
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.0219344608879498e-05,
+      "loss": 0.0577,
+      "step": 7492
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.021405919661734e-05,
+      "loss": 0.1102,
+      "step": 7494
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.0208773784355183e-05,
+      "loss": 0.108,
+      "step": 7496
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.0203488372093026e-05,
+      "loss": 0.2446,
+      "step": 7498
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.019820295983087e-05,
+      "loss": 0.0924,
+      "step": 7500
+    },
+    {
+      "epoch": 7.93,
+      "eval_cer": 0.02718723282986606,
+      "eval_loss": 0.7875532507896423,
+      "eval_runtime": 127.8203,
+      "eval_samples_per_second": 6.58,
+      "eval_steps_per_second": 0.829,
+      "step": 7500
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.019291754756871e-05,
+      "loss": 0.1296,
+      "step": 7502
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.0187632135306554e-05,
+      "loss": 0.2015,
+      "step": 7504
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.0182346723044396e-05,
+      "loss": 0.0772,
+      "step": 7506
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.017706131078224e-05,
+      "loss": 0.1157,
+      "step": 7508
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.017177589852009e-05,
+      "loss": 0.0763,
+      "step": 7510
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.016649048625793e-05,
+      "loss": 0.275,
+      "step": 7512
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.0161205073995774e-05,
+      "loss": 0.1857,
+      "step": 7514
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.0155919661733617e-05,
+      "loss": 0.1416,
+      "step": 7516
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.015063424947146e-05,
+      "loss": 0.048,
+      "step": 7518
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.0145348837209302e-05,
+      "loss": 0.0743,
+      "step": 7520
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.0140063424947145e-05,
+      "loss": 0.1346,
+      "step": 7522
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.0134778012684987e-05,
+      "loss": 0.0731,
+      "step": 7524
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.0129492600422833e-05,
+      "loss": 0.0553,
+      "step": 7526
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.012420718816068e-05,
+      "loss": 0.0786,
+      "step": 7528
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.0118921775898522e-05,
+      "loss": 0.0682,
+      "step": 7530
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.0113636363636365e-05,
+      "loss": 0.0939,
+      "step": 7532
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.0108350951374207e-05,
+      "loss": 0.2162,
+      "step": 7534
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.0103065539112053e-05,
+      "loss": 0.2454,
+      "step": 7536
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.0097780126849896e-05,
+      "loss": 0.1349,
+      "step": 7538
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.009249471458774e-05,
+      "loss": 0.1636,
+      "step": 7540
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.008720930232558e-05,
+      "loss": 0.0599,
+      "step": 7542
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.0081923890063424e-05,
+      "loss": 0.1237,
+      "step": 7544
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.0076638477801274e-05,
+      "loss": 0.1353,
+      "step": 7546
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.0071353065539116e-05,
+      "loss": 0.162,
+      "step": 7548
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.006606765327696e-05,
+      "loss": 0.2098,
+      "step": 7550
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.00607822410148e-05,
+      "loss": 0.0433,
+      "step": 7552
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.0055496828752644e-05,
+      "loss": 0.0537,
+      "step": 7554
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.0050211416490487e-05,
+      "loss": 0.0926,
+      "step": 7556
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.004492600422833e-05,
+      "loss": 0.1546,
+      "step": 7558
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.0039640591966172e-05,
+      "loss": 0.1787,
+      "step": 7560
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.0034355179704015e-05,
+      "loss": 0.1037,
+      "step": 7562
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.0029069767441864e-05,
+      "loss": 0.1001,
+      "step": 7564
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.0023784355179707e-05,
+      "loss": 0.1114,
+      "step": 7566
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.001849894291755e-05,
+      "loss": 0.0804,
+      "step": 7568
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.0013213530655392e-05,
+      "loss": 0.0532,
+      "step": 7570
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.0007928118393235e-05,
+      "loss": 0.0644,
+      "step": 7572
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.0002642706131078e-05,
+      "loss": 0.0707,
+      "step": 7574
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.999735729386892e-05,
+      "loss": 0.1153,
+      "step": 7576
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.9992071881606763e-05,
+      "loss": 0.0249,
+      "step": 7578
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.998678646934461e-05,
+      "loss": 0.054,
+      "step": 7580
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.9981501057082455e-05,
+      "loss": 0.1267,
+      "step": 7582
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.9976215644820298e-05,
+      "loss": 0.0665,
+      "step": 7584
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.997093023255814e-05,
+      "loss": 0.1242,
+      "step": 7586
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.9965644820295987e-05,
+      "loss": 0.0634,
+      "step": 7588
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.996035940803383e-05,
+      "loss": 0.132,
+      "step": 7590
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.9955073995771672e-05,
+      "loss": 0.0788,
+      "step": 7592
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.9949788583509515e-05,
+      "loss": 0.1367,
+      "step": 7594
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.9944503171247357e-05,
+      "loss": 0.0911,
+      "step": 7596
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.99392177589852e-05,
+      "loss": 0.2105,
+      "step": 7598
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.993393234672305e-05,
+      "loss": 0.1738,
+      "step": 7600
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.9928646934460892e-05,
+      "loss": 0.22,
+      "step": 7602
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.9923361522198735e-05,
+      "loss": 0.1878,
+      "step": 7604
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.9918076109936577e-05,
+      "loss": 0.0618,
+      "step": 7606
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.991279069767442e-05,
+      "loss": 0.0667,
+      "step": 7608
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.9907505285412263e-05,
+      "loss": 0.0914,
+      "step": 7610
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.9902219873150105e-05,
+      "loss": 0.1477,
+      "step": 7612
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.9896934460887948e-05,
+      "loss": 0.16,
+      "step": 7614
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.989164904862579e-05,
+      "loss": 0.1531,
+      "step": 7616
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.988636363636364e-05,
+      "loss": 0.1038,
+      "step": 7618
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.9881078224101483e-05,
+      "loss": 0.0792,
+      "step": 7620
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.9875792811839326e-05,
+      "loss": 0.1897,
+      "step": 7622
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.9870507399577168e-05,
+      "loss": 0.1093,
+      "step": 7624
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.986522198731501e-05,
+      "loss": 0.1908,
+      "step": 7626
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.9859936575052854e-05,
+      "loss": 0.1275,
+      "step": 7628
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.9854651162790696e-05,
+      "loss": 0.0609,
+      "step": 7630
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.9849365750528542e-05,
+      "loss": 0.105,
+      "step": 7632
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.9844080338266385e-05,
+      "loss": 0.0822,
+      "step": 7634
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.983879492600423e-05,
+      "loss": 0.1121,
+      "step": 7636
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.9833509513742074e-05,
+      "loss": 0.0555,
+      "step": 7638
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.9828224101479916e-05,
+      "loss": 0.1721,
+      "step": 7640
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.9822938689217762e-05,
+      "loss": 0.0414,
+      "step": 7642
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.9817653276955605e-05,
+      "loss": 0.1267,
+      "step": 7644
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.9812367864693448e-05,
+      "loss": 0.116,
+      "step": 7646
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.980708245243129e-05,
+      "loss": 0.1334,
+      "step": 7648
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.9801797040169133e-05,
+      "loss": 0.062,
+      "step": 7650
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.9796511627906976e-05,
+      "loss": 0.2137,
+      "step": 7652
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.9791226215644825e-05,
+      "loss": 0.2643,
+      "step": 7654
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.9785940803382668e-05,
+      "loss": 0.159,
+      "step": 7656
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.978065539112051e-05,
+      "loss": 0.0571,
+      "step": 7658
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.9775369978858353e-05,
+      "loss": 0.1387,
+      "step": 7660
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.9770084566596196e-05,
+      "loss": 0.0776,
+      "step": 7662
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.976479915433404e-05,
+      "loss": 0.0535,
+      "step": 7664
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.975951374207188e-05,
+      "loss": 0.06,
+      "step": 7666
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.9754228329809724e-05,
+      "loss": 0.1233,
+      "step": 7668
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.9748942917547567e-05,
+      "loss": 0.1132,
+      "step": 7670
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.9743657505285416e-05,
+      "loss": 0.1112,
+      "step": 7672
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.973837209302326e-05,
+      "loss": 0.2012,
+      "step": 7674
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.97330866807611e-05,
+      "loss": 0.0808,
+      "step": 7676
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.9727801268498944e-05,
+      "loss": 0.063,
+      "step": 7678
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.9722515856236787e-05,
+      "loss": 0.0613,
+      "step": 7680
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.971723044397463e-05,
+      "loss": 0.1812,
+      "step": 7682
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.9711945031712472e-05,
+      "loss": 0.0844,
+      "step": 7684
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.9706659619450318e-05,
+      "loss": 0.2003,
+      "step": 7686
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.970137420718816e-05,
+      "loss": 0.1232,
+      "step": 7688
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.9696088794926007e-05,
+      "loss": 0.1256,
+      "step": 7690
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.969080338266385e-05,
+      "loss": 0.0865,
+      "step": 7692
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.9685517970401696e-05,
+      "loss": 0.102,
+      "step": 7694
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.9680232558139538e-05,
+      "loss": 0.1025,
+      "step": 7696
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.967494714587738e-05,
+      "loss": 0.0941,
+      "step": 7698
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.9669661733615224e-05,
+      "loss": 0.0641,
+      "step": 7700
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.9664376321353066e-05,
+      "loss": 0.0927,
+      "step": 7702
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.965909090909091e-05,
+      "loss": 0.1103,
+      "step": 7704
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.965380549682875e-05,
+      "loss": 0.2576,
+      "step": 7706
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.96485200845666e-05,
+      "loss": 0.0838,
+      "step": 7708
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.9643234672304444e-05,
+      "loss": 0.1668,
+      "step": 7710
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.9637949260042286e-05,
+      "loss": 0.1558,
+      "step": 7712
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.963266384778013e-05,
+      "loss": 0.0759,
+      "step": 7714
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.9627378435517972e-05,
+      "loss": 0.111,
+      "step": 7716
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.9622093023255814e-05,
+      "loss": 0.1078,
+      "step": 7718
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.9616807610993657e-05,
+      "loss": 0.1482,
+      "step": 7720
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.96115221987315e-05,
+      "loss": 0.1622,
+      "step": 7722
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.9606236786469342e-05,
+      "loss": 0.1325,
+      "step": 7724
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.9600951374207192e-05,
+      "loss": 0.1058,
+      "step": 7726
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.9595665961945035e-05,
+      "loss": 0.0843,
+      "step": 7728
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.9590380549682877e-05,
+      "loss": 0.1694,
+      "step": 7730
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.958509513742072e-05,
+      "loss": 0.0398,
+      "step": 7732
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.9579809725158563e-05,
+      "loss": 0.1988,
+      "step": 7734
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.9574524312896405e-05,
+      "loss": 0.1308,
+      "step": 7736
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.956923890063425e-05,
+      "loss": 0.053,
+      "step": 7738
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.9563953488372094e-05,
+      "loss": 0.1263,
+      "step": 7740
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.9558668076109937e-05,
+      "loss": 0.1169,
+      "step": 7742
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.9553382663847783e-05,
+      "loss": 0.02,
+      "step": 7744
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.9548097251585625e-05,
+      "loss": 0.083,
+      "step": 7746
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.954281183932347e-05,
+      "loss": 0.0907,
+      "step": 7748
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.9537526427061314e-05,
+      "loss": 0.0935,
+      "step": 7750
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.9532241014799157e-05,
+      "loss": 0.0604,
+      "step": 7752
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.9526955602537e-05,
+      "loss": 0.092,
+      "step": 7754
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.9521670190274842e-05,
+      "loss": 0.1189,
+      "step": 7756
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.9516384778012685e-05,
+      "loss": 0.1149,
+      "step": 7758
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.9511099365750527e-05,
+      "loss": 0.111,
+      "step": 7760
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.9505813953488377e-05,
+      "loss": 0.146,
+      "step": 7762
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.950052854122622e-05,
+      "loss": 0.0594,
+      "step": 7764
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.9495243128964062e-05,
+      "loss": 0.1223,
+      "step": 7766
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.9489957716701905e-05,
+      "loss": 0.1484,
+      "step": 7768
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.9484672304439748e-05,
+      "loss": 0.139,
+      "step": 7770
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.947938689217759e-05,
+      "loss": 0.2436,
+      "step": 7772
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.9474101479915433e-05,
+      "loss": 0.1065,
+      "step": 7774
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.9468816067653276e-05,
+      "loss": 0.1723,
+      "step": 7776
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.9463530655391118e-05,
+      "loss": 0.0795,
+      "step": 7778
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.9458245243128968e-05,
+      "loss": 0.0833,
+      "step": 7780
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.945295983086681e-05,
+      "loss": 0.0397,
+      "step": 7782
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.9447674418604653e-05,
+      "loss": 0.055,
+      "step": 7784
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.9442389006342496e-05,
+      "loss": 0.0573,
+      "step": 7786
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.943710359408034e-05,
+      "loss": 0.0732,
+      "step": 7788
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.943181818181818e-05,
+      "loss": 0.1044,
+      "step": 7790
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.9426532769556027e-05,
+      "loss": 0.0832,
+      "step": 7792
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.942124735729387e-05,
+      "loss": 0.0394,
+      "step": 7794
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.9415961945031712e-05,
+      "loss": 0.0308,
+      "step": 7796
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.941067653276956e-05,
+      "loss": 0.1011,
+      "step": 7798
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.94053911205074e-05,
+      "loss": 0.1278,
+      "step": 7800
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.9400105708245247e-05,
+      "loss": 0.1258,
+      "step": 7802
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.939482029598309e-05,
+      "loss": 0.1203,
+      "step": 7804
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.9389534883720933e-05,
+      "loss": 0.0701,
+      "step": 7806
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.9384249471458775e-05,
+      "loss": 0.1125,
+      "step": 7808
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.9378964059196618e-05,
+      "loss": 0.158,
+      "step": 7810
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.937367864693446e-05,
+      "loss": 0.1057,
+      "step": 7812
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.9368393234672303e-05,
+      "loss": 0.0738,
+      "step": 7814
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.9363107822410153e-05,
+      "loss": 0.1133,
+      "step": 7816
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.9357822410147995e-05,
+      "loss": 0.108,
+      "step": 7818
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.9352536997885838e-05,
+      "loss": 0.0945,
+      "step": 7820
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.934725158562368e-05,
+      "loss": 0.0498,
+      "step": 7822
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.9341966173361523e-05,
+      "loss": 0.0087,
+      "step": 7824
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.9336680761099366e-05,
+      "loss": 0.027,
+      "step": 7826
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.933139534883721e-05,
+      "loss": 0.0578,
+      "step": 7828
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.932610993657505e-05,
+      "loss": 0.118,
+      "step": 7830
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.9320824524312894e-05,
+      "loss": 0.1397,
+      "step": 7832
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.9315539112050744e-05,
+      "loss": 0.1436,
+      "step": 7834
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.9310253699788586e-05,
+      "loss": 0.0833,
+      "step": 7836
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.930496828752643e-05,
+      "loss": 0.2005,
+      "step": 7838
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.929968287526427e-05,
+      "loss": 0.0683,
+      "step": 7840
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.9294397463002114e-05,
+      "loss": 0.1512,
+      "step": 7842
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.9289112050739957e-05,
+      "loss": 0.1971,
+      "step": 7844
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.9283826638477803e-05,
+      "loss": 0.1115,
+      "step": 7846
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.9278541226215646e-05,
+      "loss": 0.1307,
+      "step": 7848
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.927325581395349e-05,
+      "loss": 0.1031,
+      "step": 7850
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.9267970401691334e-05,
+      "loss": 0.0409,
+      "step": 7852
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.926268498942918e-05,
+      "loss": 0.0485,
+      "step": 7854
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.9257399577167023e-05,
+      "loss": 0.0778,
+      "step": 7856
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.9252114164904866e-05,
+      "loss": 0.2528,
+      "step": 7858
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.924682875264271e-05,
+      "loss": 0.0568,
+      "step": 7860
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.924154334038055e-05,
+      "loss": 0.0842,
+      "step": 7862
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.9236257928118394e-05,
+      "loss": 0.1228,
+      "step": 7864
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.9230972515856236e-05,
+      "loss": 0.1569,
+      "step": 7866
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.922568710359408e-05,
+      "loss": 0.1621,
+      "step": 7868
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.922040169133193e-05,
+      "loss": 0.1528,
+      "step": 7870
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.921511627906977e-05,
+      "loss": 0.0645,
+      "step": 7872
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.9209830866807614e-05,
+      "loss": 0.1805,
+      "step": 7874
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.9204545454545457e-05,
+      "loss": 0.159,
+      "step": 7876
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.91992600422833e-05,
+      "loss": 0.2725,
+      "step": 7878
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.9193974630021142e-05,
+      "loss": 0.2843,
+      "step": 7880
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.9188689217758985e-05,
+      "loss": 0.0793,
+      "step": 7882
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.9183403805496827e-05,
+      "loss": 0.0768,
+      "step": 7884
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.917811839323467e-05,
+      "loss": 0.0909,
+      "step": 7886
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.917283298097252e-05,
+      "loss": 0.274,
+      "step": 7888
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.9167547568710362e-05,
+      "loss": 0.0688,
+      "step": 7890
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.9162262156448205e-05,
+      "loss": 0.1227,
+      "step": 7892
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.9156976744186047e-05,
+      "loss": 0.1165,
+      "step": 7894
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.915169133192389e-05,
+      "loss": 0.272,
+      "step": 7896
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.9146405919661736e-05,
+      "loss": 0.11,
+      "step": 7898
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.914112050739958e-05,
+      "loss": 0.1305,
+      "step": 7900
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.913583509513742e-05,
+      "loss": 0.0879,
+      "step": 7902
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.9130549682875264e-05,
+      "loss": 0.0813,
+      "step": 7904
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.912526427061311e-05,
+      "loss": 0.0774,
+      "step": 7906
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.9119978858350956e-05,
+      "loss": 0.0652,
+      "step": 7908
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.91146934460888e-05,
+      "loss": 0.1315,
+      "step": 7910
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.910940803382664e-05,
+      "loss": 0.1017,
+      "step": 7912
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.9104122621564484e-05,
+      "loss": 0.0486,
+      "step": 7914
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.9098837209302327e-05,
+      "loss": 0.0856,
+      "step": 7916
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.909355179704017e-05,
+      "loss": 0.0431,
+      "step": 7918
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.9088266384778012e-05,
+      "loss": 0.0312,
+      "step": 7920
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.9082980972515855e-05,
+      "loss": 0.1018,
+      "step": 7922
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.9077695560253704e-05,
+      "loss": 0.0721,
+      "step": 7924
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.9072410147991547e-05,
+      "loss": 0.0959,
+      "step": 7926
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.906712473572939e-05,
+      "loss": 0.0566,
+      "step": 7928
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.9061839323467232e-05,
+      "loss": 0.0306,
+      "step": 7930
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.9056553911205075e-05,
+      "loss": 0.2017,
+      "step": 7932
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.9051268498942918e-05,
+      "loss": 0.0892,
+      "step": 7934
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.904598308668076e-05,
+      "loss": 0.0562,
+      "step": 7936
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.9040697674418603e-05,
+      "loss": 0.0532,
+      "step": 7938
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.9035412262156446e-05,
+      "loss": 0.189,
+      "step": 7940
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.9030126849894295e-05,
+      "loss": 0.1269,
+      "step": 7942
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.9024841437632138e-05,
+      "loss": 0.0781,
+      "step": 7944
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.901955602536998e-05,
+      "loss": 0.0659,
+      "step": 7946
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.9014270613107823e-05,
+      "loss": 0.0893,
+      "step": 7948
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.9008985200845666e-05,
+      "loss": 0.1148,
+      "step": 7950
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.9003699788583512e-05,
+      "loss": 0.0789,
+      "step": 7952
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.8998414376321355e-05,
+      "loss": 0.1647,
+      "step": 7954
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.8993128964059197e-05,
+      "loss": 0.1093,
+      "step": 7956
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.898784355179704e-05,
+      "loss": 0.272,
+      "step": 7958
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.8982558139534886e-05,
+      "loss": 0.2171,
+      "step": 7960
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.8977272727272732e-05,
+      "loss": 0.0894,
+      "step": 7962
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.8971987315010575e-05,
+      "loss": 0.2088,
+      "step": 7964
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.8966701902748417e-05,
+      "loss": 0.0528,
+      "step": 7966
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.896141649048626e-05,
+      "loss": 0.1048,
+      "step": 7968
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.8956131078224103e-05,
+      "loss": 0.101,
+      "step": 7970
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.8950845665961945e-05,
+      "loss": 0.2322,
+      "step": 7972
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.8945560253699788e-05,
+      "loss": 0.2275,
+      "step": 7974
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.894027484143763e-05,
+      "loss": 0.1182,
+      "step": 7976
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.893498942917548e-05,
+      "loss": 0.0821,
+      "step": 7978
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.8929704016913323e-05,
+      "loss": 0.0642,
+      "step": 7980
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.8924418604651166e-05,
+      "loss": 0.0747,
+      "step": 7982
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.8919133192389008e-05,
+      "loss": 0.1306,
+      "step": 7984
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.891384778012685e-05,
+      "loss": 0.1336,
+      "step": 7986
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.8908562367864694e-05,
+      "loss": 0.074,
+      "step": 7988
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.8903276955602536e-05,
+      "loss": 0.2258,
+      "step": 7990
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.889799154334038e-05,
+      "loss": 0.0969,
+      "step": 7992
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.889270613107822e-05,
+      "loss": 0.0191,
+      "step": 7994
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.888742071881607e-05,
+      "loss": 0.0725,
+      "step": 7996
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.8882135306553914e-05,
+      "loss": 0.0552,
+      "step": 7998
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.8876849894291756e-05,
+      "loss": 0.0584,
+      "step": 8000
+    },
+    {
+      "epoch": 8.46,
+      "eval_cer": 0.023083499572527786,
+      "eval_loss": 0.834189772605896,
+      "eval_runtime": 129.0007,
+      "eval_samples_per_second": 6.519,
+      "eval_steps_per_second": 0.822,
+      "step": 8000
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.88715644820296e-05,
+      "loss": 0.0839,
+      "step": 8002
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.8866279069767442e-05,
+      "loss": 0.101,
+      "step": 8004
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.8860993657505288e-05,
+      "loss": 0.1871,
+      "step": 8006
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.885570824524313e-05,
+      "loss": 0.0541,
+      "step": 8008
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.8850422832980973e-05,
+      "loss": 0.071,
+      "step": 8010
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.8845137420718816e-05,
+      "loss": 0.0442,
+      "step": 8012
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.8839852008456665e-05,
+      "loss": 0.1274,
+      "step": 8014
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.8834566596194508e-05,
+      "loss": 0.1477,
+      "step": 8016
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.882928118393235e-05,
+      "loss": 0.2032,
+      "step": 8018
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.8823995771670193e-05,
+      "loss": 0.0629,
+      "step": 8020
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.8818710359408036e-05,
+      "loss": 0.1614,
+      "step": 8022
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.881342494714588e-05,
+      "loss": 0.2354,
+      "step": 8024
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.880813953488372e-05,
+      "loss": 0.0932,
+      "step": 8026
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.8802854122621564e-05,
+      "loss": 0.1353,
+      "step": 8028
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.8797568710359407e-05,
+      "loss": 0.1506,
+      "step": 8030
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.8792283298097256e-05,
+      "loss": 0.0269,
+      "step": 8032
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.87869978858351e-05,
+      "loss": 0.0624,
+      "step": 8034
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.878171247357294e-05,
+      "loss": 0.1213,
+      "step": 8036
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.8776427061310784e-05,
+      "loss": 0.1265,
+      "step": 8038
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.8771141649048627e-05,
+      "loss": 0.1232,
+      "step": 8040
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.876585623678647e-05,
+      "loss": 0.0704,
+      "step": 8042
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.8760570824524312e-05,
+      "loss": 0.0532,
+      "step": 8044
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.8755285412262155e-05,
+      "loss": 0.06,
+      "step": 8046
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.8749999999999997e-05,
+      "loss": 0.088,
+      "step": 8048
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.8744714587737847e-05,
+      "loss": 0.063,
+      "step": 8050
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.873942917547569e-05,
+      "loss": 0.0784,
+      "step": 8052
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.8734143763213532e-05,
+      "loss": 0.0938,
+      "step": 8054
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.8728858350951375e-05,
+      "loss": 0.1824,
+      "step": 8056
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.872357293868922e-05,
+      "loss": 0.2113,
+      "step": 8058
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.8718287526427064e-05,
+      "loss": 0.1183,
+      "step": 8060
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.8713002114164906e-05,
+      "loss": 0.0816,
+      "step": 8062
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.870771670190275e-05,
+      "loss": 0.1049,
+      "step": 8064
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.870243128964059e-05,
+      "loss": 0.1227,
+      "step": 8066
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.869714587737844e-05,
+      "loss": 0.1581,
+      "step": 8068
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.8691860465116284e-05,
+      "loss": 0.0702,
+      "step": 8070
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.8686575052854126e-05,
+      "loss": 0.2441,
+      "step": 8072
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.868128964059197e-05,
+      "loss": 0.1847,
+      "step": 8074
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.8676004228329812e-05,
+      "loss": 0.1233,
+      "step": 8076
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.8670718816067654e-05,
+      "loss": 0.3098,
+      "step": 8078
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.8665433403805497e-05,
+      "loss": 0.2001,
+      "step": 8080
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.866014799154334e-05,
+      "loss": 0.2392,
+      "step": 8082
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.8654862579281183e-05,
+      "loss": 0.1514,
+      "step": 8084
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.8649577167019032e-05,
+      "loss": 0.1599,
+      "step": 8086
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.8644291754756875e-05,
+      "loss": 0.1155,
+      "step": 8088
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.8639006342494717e-05,
+      "loss": 0.041,
+      "step": 8090
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.863372093023256e-05,
+      "loss": 0.1052,
+      "step": 8092
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.8628435517970403e-05,
+      "loss": 0.1127,
+      "step": 8094
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.8623150105708245e-05,
+      "loss": 0.1245,
+      "step": 8096
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.8617864693446088e-05,
+      "loss": 0.084,
+      "step": 8098
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.861257928118393e-05,
+      "loss": 0.1355,
+      "step": 8100
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.8607293868921777e-05,
+      "loss": 0.0758,
+      "step": 8102
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.8602008456659623e-05,
+      "loss": 0.1272,
+      "step": 8104
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.8596723044397465e-05,
+      "loss": 0.085,
+      "step": 8106
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.8591437632135308e-05,
+      "loss": 0.0379,
+      "step": 8108
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.858615221987315e-05,
+      "loss": 0.0785,
+      "step": 8110
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.8580866807610997e-05,
+      "loss": 0.072,
+      "step": 8112
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.857558139534884e-05,
+      "loss": 0.121,
+      "step": 8114
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.8570295983086682e-05,
+      "loss": 0.0658,
+      "step": 8116
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.8565010570824525e-05,
+      "loss": 0.0765,
+      "step": 8118
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.8559725158562368e-05,
+      "loss": 0.0394,
+      "step": 8120
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.855443974630021e-05,
+      "loss": 0.2187,
+      "step": 8122
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.854915433403806e-05,
+      "loss": 0.0988,
+      "step": 8124
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.8543868921775902e-05,
+      "loss": 0.07,
+      "step": 8126
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.8538583509513745e-05,
+      "loss": 0.0686,
+      "step": 8128
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.8533298097251588e-05,
+      "loss": 0.261,
+      "step": 8130
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.852801268498943e-05,
+      "loss": 0.1215,
+      "step": 8132
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.8522727272727273e-05,
+      "loss": 0.155,
+      "step": 8134
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.8517441860465116e-05,
+      "loss": 0.1924,
+      "step": 8136
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.851215644820296e-05,
+      "loss": 0.0767,
+      "step": 8138
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.85068710359408e-05,
+      "loss": 0.1421,
+      "step": 8140
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.850158562367865e-05,
+      "loss": 0.0753,
+      "step": 8142
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.8496300211416493e-05,
+      "loss": 0.1448,
+      "step": 8144
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.8491014799154336e-05,
+      "loss": 0.1149,
+      "step": 8146
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.848572938689218e-05,
+      "loss": 0.0598,
+      "step": 8148
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.848044397463002e-05,
+      "loss": 0.1086,
+      "step": 8150
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.8475158562367864e-05,
+      "loss": 0.0888,
+      "step": 8152
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.8469873150105706e-05,
+      "loss": 0.0574,
+      "step": 8154
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.8464587737843553e-05,
+      "loss": 0.0729,
+      "step": 8156
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.8459302325581395e-05,
+      "loss": 0.1762,
+      "step": 8158
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.845401691331924e-05,
+      "loss": 0.0921,
+      "step": 8160
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.8448731501057084e-05,
+      "loss": 0.0903,
+      "step": 8162
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.844344608879493e-05,
+      "loss": 0.1676,
+      "step": 8164
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.8438160676532773e-05,
+      "loss": 0.1132,
+      "step": 8166
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.8432875264270615e-05,
+      "loss": 0.1617,
+      "step": 8168
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.8427589852008458e-05,
+      "loss": 0.0692,
+      "step": 8170
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.84223044397463e-05,
+      "loss": 0.072,
+      "step": 8172
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.8417019027484143e-05,
+      "loss": 0.0783,
+      "step": 8174
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.8411733615221986e-05,
+      "loss": 0.0356,
+      "step": 8176
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.8406448202959835e-05,
+      "loss": 0.0552,
+      "step": 8178
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.8401162790697678e-05,
+      "loss": 0.1139,
+      "step": 8180
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.839587737843552e-05,
+      "loss": 0.1773,
+      "step": 8182
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.8390591966173363e-05,
+      "loss": 0.2647,
+      "step": 8184
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.8385306553911206e-05,
+      "loss": 0.1721,
+      "step": 8186
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.838002114164905e-05,
+      "loss": 0.0464,
+      "step": 8188
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.837473572938689e-05,
+      "loss": 0.0901,
+      "step": 8190
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.8369450317124734e-05,
+      "loss": 0.0405,
+      "step": 8192
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.8364164904862577e-05,
+      "loss": 0.0503,
+      "step": 8194
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.8358879492600426e-05,
+      "loss": 0.1983,
+      "step": 8196
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.835359408033827e-05,
+      "loss": 0.0998,
+      "step": 8198
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.834830866807611e-05,
+      "loss": 0.0986,
+      "step": 8200
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.8343023255813954e-05,
+      "loss": 0.1944,
+      "step": 8202
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.8337737843551797e-05,
+      "loss": 0.1027,
+      "step": 8204
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.833245243128964e-05,
+      "loss": 0.1601,
+      "step": 8206
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.8327167019027486e-05,
+      "loss": 0.0563,
+      "step": 8208
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.832188160676533e-05,
+      "loss": 0.0967,
+      "step": 8210
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.831659619450317e-05,
+      "loss": 0.0645,
+      "step": 8212
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.8311310782241017e-05,
+      "loss": 0.1535,
+      "step": 8214
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.830602536997886e-05,
+      "loss": 0.1287,
+      "step": 8216
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.8300739957716706e-05,
+      "loss": 0.0618,
+      "step": 8218
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.829545454545455e-05,
+      "loss": 0.1816,
+      "step": 8220
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.829016913319239e-05,
+      "loss": 0.1294,
+      "step": 8222
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.8284883720930234e-05,
+      "loss": 0.1351,
+      "step": 8224
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.8279598308668077e-05,
+      "loss": 0.1923,
+      "step": 8226
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.827431289640592e-05,
+      "loss": 0.0535,
+      "step": 8228
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.8269027484143762e-05,
+      "loss": 0.1522,
+      "step": 8230
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.826374207188161e-05,
+      "loss": 0.1228,
+      "step": 8232
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.8258456659619454e-05,
+      "loss": 0.2214,
+      "step": 8234
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.8253171247357297e-05,
+      "loss": 0.055,
+      "step": 8236
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.824788583509514e-05,
+      "loss": 0.0697,
+      "step": 8238
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.8242600422832982e-05,
+      "loss": 0.0846,
+      "step": 8240
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.8237315010570825e-05,
+      "loss": 0.1981,
+      "step": 8242
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.8232029598308667e-05,
+      "loss": 0.2626,
+      "step": 8244
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.822674418604651e-05,
+      "loss": 0.1027,
+      "step": 8246
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.8221458773784353e-05,
+      "loss": 0.072,
+      "step": 8248
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.8216173361522202e-05,
+      "loss": 0.0512,
+      "step": 8250
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.8210887949260045e-05,
+      "loss": 0.1911,
+      "step": 8252
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.8205602536997887e-05,
+      "loss": 0.1951,
+      "step": 8254
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.820031712473573e-05,
+      "loss": 0.1127,
+      "step": 8256
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.8195031712473573e-05,
+      "loss": 0.0594,
+      "step": 8258
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.8189746300211415e-05,
+      "loss": 0.0958,
+      "step": 8260
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.818446088794926e-05,
+      "loss": 0.2258,
+      "step": 8262
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.8179175475687104e-05,
+      "loss": 0.0807,
+      "step": 8264
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.8173890063424947e-05,
+      "loss": 0.1206,
+      "step": 8266
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.8168604651162793e-05,
+      "loss": 0.0782,
+      "step": 8268
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.8163319238900636e-05,
+      "loss": 0.067,
+      "step": 8270
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.815803382663848e-05,
+      "loss": 0.2475,
+      "step": 8272
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.8152748414376324e-05,
+      "loss": 0.2549,
+      "step": 8274
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.8147463002114167e-05,
+      "loss": 0.1755,
+      "step": 8276
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.814217758985201e-05,
+      "loss": 0.1862,
+      "step": 8278
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.8136892177589852e-05,
+      "loss": 0.1525,
+      "step": 8280
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.8131606765327695e-05,
+      "loss": 0.1334,
+      "step": 8282
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.8126321353065538e-05,
+      "loss": 0.0858,
+      "step": 8284
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.8121035940803387e-05,
+      "loss": 0.0971,
+      "step": 8286
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.811575052854123e-05,
+      "loss": 0.077,
+      "step": 8288
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.8110465116279073e-05,
+      "loss": 0.1299,
+      "step": 8290
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.8105179704016915e-05,
+      "loss": 0.1872,
+      "step": 8292
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.8099894291754758e-05,
+      "loss": 0.0986,
+      "step": 8294
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.80946088794926e-05,
+      "loss": 0.0461,
+      "step": 8296
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.8089323467230443e-05,
+      "loss": 0.1037,
+      "step": 8298
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.8084038054968286e-05,
+      "loss": 0.1127,
+      "step": 8300
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.807875264270613e-05,
+      "loss": 0.1026,
+      "step": 8302
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.8073467230443978e-05,
+      "loss": 0.0636,
+      "step": 8304
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.806818181818182e-05,
+      "loss": 0.1557,
+      "step": 8306
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.8062896405919663e-05,
+      "loss": 0.1203,
+      "step": 8308
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.8057610993657506e-05,
+      "loss": 0.0859,
+      "step": 8310
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.805232558139535e-05,
+      "loss": 0.1833,
+      "step": 8312
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.804704016913319e-05,
+      "loss": 0.2377,
+      "step": 8314
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.8041754756871037e-05,
+      "loss": 0.1107,
+      "step": 8316
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.803646934460888e-05,
+      "loss": 0.1154,
+      "step": 8318
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.8031183932346723e-05,
+      "loss": 0.1013,
+      "step": 8320
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.802589852008457e-05,
+      "loss": 0.0989,
+      "step": 8322
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.8020613107822415e-05,
+      "loss": 0.0776,
+      "step": 8324
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.8015327695560258e-05,
+      "loss": 0.0925,
+      "step": 8326
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.80100422832981e-05,
+      "loss": 0.1062,
+      "step": 8328
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.8004756871035943e-05,
+      "loss": 0.0799,
+      "step": 8330
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.7999471458773786e-05,
+      "loss": 0.1206,
+      "step": 8332
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.7994186046511628e-05,
+      "loss": 0.1726,
+      "step": 8334
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.798890063424947e-05,
+      "loss": 0.1599,
+      "step": 8336
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.7983615221987314e-05,
+      "loss": 0.1454,
+      "step": 8338
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.7978329809725163e-05,
+      "loss": 0.1246,
+      "step": 8340
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.7973044397463006e-05,
+      "loss": 0.0782,
+      "step": 8342
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.796775898520085e-05,
+      "loss": 0.1085,
+      "step": 8344
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.796247357293869e-05,
+      "loss": 0.0896,
+      "step": 8346
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.7957188160676534e-05,
+      "loss": 0.0475,
+      "step": 8348
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.7951902748414376e-05,
+      "loss": 0.1508,
+      "step": 8350
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.794661733615222e-05,
+      "loss": 0.0623,
+      "step": 8352
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.7941331923890062e-05,
+      "loss": 0.1372,
+      "step": 8354
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.7936046511627904e-05,
+      "loss": 0.0929,
+      "step": 8356
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.7930761099365754e-05,
+      "loss": 0.0702,
+      "step": 8358
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.7925475687103596e-05,
+      "loss": 0.107,
+      "step": 8360
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.792019027484144e-05,
+      "loss": 0.0428,
+      "step": 8362
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.7914904862579282e-05,
+      "loss": 0.1434,
+      "step": 8364
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.7909619450317125e-05,
+      "loss": 0.0995,
+      "step": 8366
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.790433403805497e-05,
+      "loss": 0.1524,
+      "step": 8368
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.7899048625792813e-05,
+      "loss": 0.0819,
+      "step": 8370
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.7893763213530656e-05,
+      "loss": 0.0574,
+      "step": 8372
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.78884778012685e-05,
+      "loss": 0.0799,
+      "step": 8374
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.7883192389006345e-05,
+      "loss": 0.1706,
+      "step": 8376
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.787790697674419e-05,
+      "loss": 0.0798,
+      "step": 8378
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.7872621564482033e-05,
+      "loss": 0.1162,
+      "step": 8380
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.7867336152219876e-05,
+      "loss": 0.0853,
+      "step": 8382
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.786205073995772e-05,
+      "loss": 0.0931,
+      "step": 8384
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.785676532769556e-05,
+      "loss": 0.0971,
+      "step": 8386
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.7851479915433404e-05,
+      "loss": 0.043,
+      "step": 8388
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.7846194503171247e-05,
+      "loss": 0.0413,
+      "step": 8390
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.784090909090909e-05,
+      "loss": 0.0894,
+      "step": 8392
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.783562367864694e-05,
+      "loss": 0.029,
+      "step": 8394
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.783033826638478e-05,
+      "loss": 0.0832,
+      "step": 8396
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.7825052854122624e-05,
+      "loss": 0.0876,
+      "step": 8398
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.7819767441860467e-05,
+      "loss": 0.1376,
+      "step": 8400
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.781448202959831e-05,
+      "loss": 0.1387,
+      "step": 8402
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.7809196617336152e-05,
+      "loss": 0.1245,
+      "step": 8404
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.7803911205073995e-05,
+      "loss": 0.1056,
+      "step": 8406
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.7798625792811838e-05,
+      "loss": 0.0913,
+      "step": 8408
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.779334038054968e-05,
+      "loss": 0.0532,
+      "step": 8410
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.778805496828753e-05,
+      "loss": 0.1707,
+      "step": 8412
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.7782769556025372e-05,
+      "loss": 0.1036,
+      "step": 8414
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.7777484143763215e-05,
+      "loss": 0.1387,
+      "step": 8416
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.7772198731501058e-05,
+      "loss": 0.091,
+      "step": 8418
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.77669133192389e-05,
+      "loss": 0.0763,
+      "step": 8420
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.7761627906976746e-05,
+      "loss": 0.092,
+      "step": 8422
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.775634249471459e-05,
+      "loss": 0.0948,
+      "step": 8424
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.7751057082452432e-05,
+      "loss": 0.0832,
+      "step": 8426
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.7745771670190274e-05,
+      "loss": 0.0764,
+      "step": 8428
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.774048625792812e-05,
+      "loss": 0.1405,
+      "step": 8430
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.7735200845665967e-05,
+      "loss": 0.0823,
+      "step": 8432
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.772991543340381e-05,
+      "loss": 0.1438,
+      "step": 8434
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.7724630021141652e-05,
+      "loss": 0.228,
+      "step": 8436
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.7719344608879495e-05,
+      "loss": 0.0858,
+      "step": 8438
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.7714059196617337e-05,
+      "loss": 0.0388,
+      "step": 8440
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.770877378435518e-05,
+      "loss": 0.0782,
+      "step": 8442
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.7703488372093023e-05,
+      "loss": 0.1014,
+      "step": 8444
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.7698202959830865e-05,
+      "loss": 0.1315,
+      "step": 8446
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.7692917547568715e-05,
+      "loss": 0.0287,
+      "step": 8448
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.7687632135306557e-05,
+      "loss": 0.0993,
+      "step": 8450
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.76823467230444e-05,
+      "loss": 0.1108,
+      "step": 8452
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.7677061310782243e-05,
+      "loss": 0.1189,
+      "step": 8454
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.7671775898520085e-05,
+      "loss": 0.0465,
+      "step": 8456
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.7666490486257928e-05,
+      "loss": 0.2443,
+      "step": 8458
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.766120507399577e-05,
+      "loss": 0.1708,
+      "step": 8460
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.7655919661733613e-05,
+      "loss": 0.2846,
+      "step": 8462
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.7650634249471456e-05,
+      "loss": 0.079,
+      "step": 8464
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.7645348837209305e-05,
+      "loss": 0.3057,
+      "step": 8466
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.7640063424947148e-05,
+      "loss": 0.0952,
+      "step": 8468
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.763477801268499e-05,
+      "loss": 0.0733,
+      "step": 8470
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.7629492600422834e-05,
+      "loss": 0.1229,
+      "step": 8472
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.7624207188160676e-05,
+      "loss": 0.1157,
+      "step": 8474
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.7618921775898522e-05,
+      "loss": 0.1169,
+      "step": 8476
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.7613636363636365e-05,
+      "loss": 0.0718,
+      "step": 8478
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.7608350951374208e-05,
+      "loss": 0.2532,
+      "step": 8480
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.760306553911205e-05,
+      "loss": 0.052,
+      "step": 8482
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.75977801268499e-05,
+      "loss": 0.1107,
+      "step": 8484
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.7592494714587742e-05,
+      "loss": 0.1298,
+      "step": 8486
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.7587209302325585e-05,
+      "loss": 0.0634,
+      "step": 8488
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.7581923890063428e-05,
+      "loss": 0.0778,
+      "step": 8490
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.757663847780127e-05,
+      "loss": 0.0882,
+      "step": 8492
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.7571353065539113e-05,
+      "loss": 0.0727,
+      "step": 8494
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.7566067653276956e-05,
+      "loss": 0.0334,
+      "step": 8496
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.75607822410148e-05,
+      "loss": 0.0673,
+      "step": 8498
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.755549682875264e-05,
+      "loss": 0.2054,
+      "step": 8500
+    },
+    {
+      "epoch": 8.99,
+      "eval_cer": 0.03425477343972642,
+      "eval_loss": 0.6548437476158142,
+      "eval_runtime": 125.5264,
+      "eval_samples_per_second": 6.7,
+      "eval_steps_per_second": 0.844,
+      "step": 8500
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.755021141649049e-05,
+      "loss": 0.0321,
+      "step": 8502
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.7544926004228333e-05,
+      "loss": 0.0655,
+      "step": 8504
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.7539640591966176e-05,
+      "loss": 0.1483,
+      "step": 8506
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.753435517970402e-05,
+      "loss": 0.0967,
+      "step": 8508
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.752906976744186e-05,
+      "loss": 0.1002,
+      "step": 8510
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.7523784355179704e-05,
+      "loss": 0.1041,
+      "step": 8512
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.7518498942917547e-05,
+      "loss": 0.1675,
+      "step": 8514
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.751321353065539e-05,
+      "loss": 0.0342,
+      "step": 8516
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.7507928118393232e-05,
+      "loss": 0.0659,
+      "step": 8518
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.750264270613108e-05,
+      "loss": 0.0966,
+      "step": 8520
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.7497357293868924e-05,
+      "loss": 0.0474,
+      "step": 8522
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.7492071881606767e-05,
+      "loss": 0.1123,
+      "step": 8524
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.748678646934461e-05,
+      "loss": 0.1332,
+      "step": 8526
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.7481501057082455e-05,
+      "loss": 0.2334,
+      "step": 8528
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.7476215644820298e-05,
+      "loss": 0.1152,
+      "step": 8530
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.747093023255814e-05,
+      "loss": 0.1294,
+      "step": 8532
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.7465644820295983e-05,
+      "loss": 0.051,
+      "step": 8534
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.7460359408033826e-05,
+      "loss": 0.0726,
+      "step": 8536
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.7455073995771676e-05,
+      "loss": 0.0495,
+      "step": 8538
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.7449788583509518e-05,
+      "loss": 0.134,
+      "step": 8540
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.744450317124736e-05,
+      "loss": 0.0087,
+      "step": 8542
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.7439217758985204e-05,
+      "loss": 0.1752,
+      "step": 8544
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.7433932346723046e-05,
+      "loss": 0.1692,
+      "step": 8546
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.742864693446089e-05,
+      "loss": 0.1097,
+      "step": 8548
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.742336152219873e-05,
+      "loss": 0.045,
+      "step": 8550
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.7418076109936574e-05,
+      "loss": 0.1581,
+      "step": 8552
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.7412790697674417e-05,
+      "loss": 0.0721,
+      "step": 8554
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.7407505285412266e-05,
+      "loss": 0.0617,
+      "step": 8556
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.740221987315011e-05,
+      "loss": 0.0526,
+      "step": 8558
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.7396934460887952e-05,
+      "loss": 0.0785,
+      "step": 8560
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.7391649048625794e-05,
+      "loss": 0.0973,
+      "step": 8562
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.7386363636363637e-05,
+      "loss": 0.0743,
+      "step": 8564
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.738107822410148e-05,
+      "loss": 0.1105,
+      "step": 8566
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.7375792811839322e-05,
+      "loss": 0.1266,
+      "step": 8568
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.7370507399577165e-05,
+      "loss": 0.0903,
+      "step": 8570
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.736522198731501e-05,
+      "loss": 0.2459,
+      "step": 8572
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.7359936575052857e-05,
+      "loss": 0.1202,
+      "step": 8574
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.73546511627907e-05,
+      "loss": 0.0555,
+      "step": 8576
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.7349365750528543e-05,
+      "loss": 0.218,
+      "step": 8578
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.7344080338266385e-05,
+      "loss": 0.0653,
+      "step": 8580
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.733879492600423e-05,
+      "loss": 0.0576,
+      "step": 8582
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.7333509513742074e-05,
+      "loss": 0.0617,
+      "step": 8584
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.7328224101479917e-05,
+      "loss": 0.0679,
+      "step": 8586
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.732293868921776e-05,
+      "loss": 0.0907,
+      "step": 8588
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.7317653276955602e-05,
+      "loss": 0.1537,
+      "step": 8590
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.731236786469345e-05,
+      "loss": 0.1437,
+      "step": 8592
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.7307082452431294e-05,
+      "loss": 0.0707,
+      "step": 8594
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.7301797040169137e-05,
+      "loss": 0.183,
+      "step": 8596
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.729651162790698e-05,
+      "loss": 0.0725,
+      "step": 8598
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.7291226215644822e-05,
+      "loss": 0.147,
+      "step": 8600
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.7285940803382665e-05,
+      "loss": 0.1171,
+      "step": 8602
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.7280655391120507e-05,
+      "loss": 0.0401,
+      "step": 8604
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.727536997885835e-05,
+      "loss": 0.0992,
+      "step": 8606
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.7270084566596193e-05,
+      "loss": 0.0551,
+      "step": 8608
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.7264799154334042e-05,
+      "loss": 0.1109,
+      "step": 8610
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.7259513742071885e-05,
+      "loss": 0.076,
+      "step": 8612
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.7254228329809728e-05,
+      "loss": 0.0863,
+      "step": 8614
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.724894291754757e-05,
+      "loss": 0.049,
+      "step": 8616
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.7243657505285413e-05,
+      "loss": 0.1014,
+      "step": 8618
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.7238372093023256e-05,
+      "loss": 0.0364,
+      "step": 8620
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.7233086680761098e-05,
+      "loss": 0.1735,
+      "step": 8622
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.722780126849894e-05,
+      "loss": 0.0667,
+      "step": 8624
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.7222515856236787e-05,
+      "loss": 0.0699,
+      "step": 8626
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.7217230443974633e-05,
+      "loss": 0.0562,
+      "step": 8628
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.7211945031712476e-05,
+      "loss": 0.0501,
+      "step": 8630
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.720665961945032e-05,
+      "loss": 0.0928,
+      "step": 8632
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.7201374207188164e-05,
+      "loss": 0.1008,
+      "step": 8634
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.7196088794926007e-05,
+      "loss": 0.0146,
+      "step": 8636
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.719080338266385e-05,
+      "loss": 0.0512,
+      "step": 8638
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.7185517970401692e-05,
+      "loss": 0.0517,
+      "step": 8640
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.7180232558139535e-05,
+      "loss": 0.0155,
+      "step": 8642
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.7174947145877378e-05,
+      "loss": 0.0762,
+      "step": 8644
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.7169661733615227e-05,
+      "loss": 0.0822,
+      "step": 8646
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.716437632135307e-05,
+      "loss": 0.0619,
+      "step": 8648
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.7159090909090913e-05,
+      "loss": 0.1611,
+      "step": 8650
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7153805496828755e-05,
+      "loss": 0.0875,
+      "step": 8652
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7148520084566598e-05,
+      "loss": 0.0389,
+      "step": 8654
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.714323467230444e-05,
+      "loss": 0.109,
+      "step": 8656
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7137949260042283e-05,
+      "loss": 0.0651,
+      "step": 8658
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7132663847780126e-05,
+      "loss": 0.1393,
+      "step": 8660
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.712737843551797e-05,
+      "loss": 0.1152,
+      "step": 8662
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7122093023255818e-05,
+      "loss": 0.0792,
+      "step": 8664
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.711680761099366e-05,
+      "loss": 0.0683,
+      "step": 8666
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7111522198731503e-05,
+      "loss": 0.1229,
+      "step": 8668
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7106236786469346e-05,
+      "loss": 0.1954,
+      "step": 8670
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.710095137420719e-05,
+      "loss": 0.1128,
+      "step": 8672
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.709566596194503e-05,
+      "loss": 0.0474,
+      "step": 8674
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7090380549682874e-05,
+      "loss": 0.152,
+      "step": 8676
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.708509513742072e-05,
+      "loss": 0.0323,
+      "step": 8678
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.7079809725158563e-05,
+      "loss": 0.0839,
+      "step": 8680
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.707452431289641e-05,
+      "loss": 0.1011,
+      "step": 8682
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.706923890063425e-05,
+      "loss": 0.0682,
+      "step": 8684
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.7063953488372094e-05,
+      "loss": 0.1056,
+      "step": 8686
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.705866807610994e-05,
+      "loss": 0.123,
+      "step": 8688
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.7053382663847783e-05,
+      "loss": 0.0388,
+      "step": 8690
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.7048097251585626e-05,
+      "loss": 0.0797,
+      "step": 8692
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.7042811839323468e-05,
+      "loss": 0.0899,
+      "step": 8694
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.703752642706131e-05,
+      "loss": 0.0955,
+      "step": 8696
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.7032241014799154e-05,
+      "loss": 0.1046,
+      "step": 8698
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.7026955602537003e-05,
+      "loss": 0.0889,
+      "step": 8700
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.7021670190274846e-05,
+      "loss": 0.0868,
+      "step": 8702
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.701638477801269e-05,
+      "loss": 0.1856,
+      "step": 8704
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.701109936575053e-05,
+      "loss": 0.1437,
+      "step": 8706
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.7005813953488374e-05,
+      "loss": 0.1376,
+      "step": 8708
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.7000528541226216e-05,
+      "loss": 0.0924,
+      "step": 8710
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.699524312896406e-05,
+      "loss": 0.0697,
+      "step": 8712
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.6989957716701902e-05,
+      "loss": 0.2081,
+      "step": 8714
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.6984672304439744e-05,
+      "loss": 0.0381,
+      "step": 8716
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.6979386892177594e-05,
+      "loss": 0.2289,
+      "step": 8718
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.6974101479915437e-05,
+      "loss": 0.0454,
+      "step": 8720
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.696881606765328e-05,
+      "loss": 0.1052,
+      "step": 8722
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.6963530655391122e-05,
+      "loss": 0.0191,
+      "step": 8724
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.6958245243128965e-05,
+      "loss": 0.035,
+      "step": 8726
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.6952959830866807e-05,
+      "loss": 0.1103,
+      "step": 8728
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.694767441860465e-05,
+      "loss": 0.0613,
+      "step": 8730
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.6942389006342496e-05,
+      "loss": 0.0965,
+      "step": 8732
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.693710359408034e-05,
+      "loss": 0.094,
+      "step": 8734
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.6931818181818185e-05,
+      "loss": 0.0707,
+      "step": 8736
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.6926532769556027e-05,
+      "loss": 0.0763,
+      "step": 8738
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.692124735729387e-05,
+      "loss": 0.2875,
+      "step": 8740
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.6915961945031716e-05,
+      "loss": 0.046,
+      "step": 8742
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.691067653276956e-05,
+      "loss": 0.0547,
+      "step": 8744
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.69053911205074e-05,
+      "loss": 0.0899,
+      "step": 8746
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.6900105708245244e-05,
+      "loss": 0.056,
+      "step": 8748
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.6894820295983087e-05,
+      "loss": 0.0906,
+      "step": 8750
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.688953488372093e-05,
+      "loss": 0.1307,
+      "step": 8752
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.688424947145878e-05,
+      "loss": 0.1062,
+      "step": 8754
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.687896405919662e-05,
+      "loss": 0.0706,
+      "step": 8756
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.6873678646934464e-05,
+      "loss": 0.0468,
+      "step": 8758
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.6868393234672307e-05,
+      "loss": 0.1303,
+      "step": 8760
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.686310782241015e-05,
+      "loss": 0.0624,
+      "step": 8762
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.6857822410147992e-05,
+      "loss": 0.0877,
+      "step": 8764
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.6852536997885835e-05,
+      "loss": 0.053,
+      "step": 8766
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.6847251585623678e-05,
+      "loss": 0.0406,
+      "step": 8768
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.684196617336152e-05,
+      "loss": 0.0733,
+      "step": 8770
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.683668076109937e-05,
+      "loss": 0.0395,
+      "step": 8772
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.6831395348837212e-05,
+      "loss": 0.04,
+      "step": 8774
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.6826109936575055e-05,
+      "loss": 0.0447,
+      "step": 8776
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.6820824524312898e-05,
+      "loss": 0.1337,
+      "step": 8778
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.681553911205074e-05,
+      "loss": 0.1441,
+      "step": 8780
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.6810253699788583e-05,
+      "loss": 0.0266,
+      "step": 8782
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.6804968287526426e-05,
+      "loss": 0.082,
+      "step": 8784
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.6799682875264272e-05,
+      "loss": 0.0276,
+      "step": 8786
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.6794397463002114e-05,
+      "loss": 0.0664,
+      "step": 8788
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.678911205073996e-05,
+      "loss": 0.0992,
+      "step": 8790
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.6783826638477803e-05,
+      "loss": 0.1348,
+      "step": 8792
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.677854122621565e-05,
+      "loss": 0.0989,
+      "step": 8794
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.6773255813953492e-05,
+      "loss": 0.0812,
+      "step": 8796
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.6767970401691335e-05,
+      "loss": 0.1295,
+      "step": 8798
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.6762684989429177e-05,
+      "loss": 0.1148,
+      "step": 8800
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.675739957716702e-05,
+      "loss": 0.0984,
+      "step": 8802
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.6752114164904863e-05,
+      "loss": 0.0716,
+      "step": 8804
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.6746828752642705e-05,
+      "loss": 0.0396,
+      "step": 8806
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.6741543340380548e-05,
+      "loss": 0.0483,
+      "step": 8808
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.6736257928118397e-05,
+      "loss": 0.0657,
+      "step": 8810
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.673097251585624e-05,
+      "loss": 0.2031,
+      "step": 8812
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.6725687103594083e-05,
+      "loss": 0.1448,
+      "step": 8814
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.6720401691331925e-05,
+      "loss": 0.0513,
+      "step": 8816
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.6715116279069768e-05,
+      "loss": 0.0472,
+      "step": 8818
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.670983086680761e-05,
+      "loss": 0.0745,
+      "step": 8820
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.6704545454545453e-05,
+      "loss": 0.0551,
+      "step": 8822
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.6699260042283296e-05,
+      "loss": 0.0293,
+      "step": 8824
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.669397463002114e-05,
+      "loss": 0.0633,
+      "step": 8826
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.6688689217758988e-05,
+      "loss": 0.0671,
+      "step": 8828
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.668340380549683e-05,
+      "loss": 0.1661,
+      "step": 8830
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.6678118393234674e-05,
+      "loss": 0.0222,
+      "step": 8832
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.6672832980972516e-05,
+      "loss": 0.1829,
+      "step": 8834
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.666754756871036e-05,
+      "loss": 0.2226,
+      "step": 8836
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.6662262156448205e-05,
+      "loss": 0.2149,
+      "step": 8838
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.6656976744186048e-05,
+      "loss": 0.1276,
+      "step": 8840
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.665169133192389e-05,
+      "loss": 0.1313,
+      "step": 8842
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.6646405919661733e-05,
+      "loss": 0.0533,
+      "step": 8844
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.664112050739958e-05,
+      "loss": 0.1269,
+      "step": 8846
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.6635835095137425e-05,
+      "loss": 0.153,
+      "step": 8848
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.6630549682875268e-05,
+      "loss": 0.1629,
+      "step": 8850
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.662526427061311e-05,
+      "loss": 0.1014,
+      "step": 8852
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.6619978858350953e-05,
+      "loss": 0.105,
+      "step": 8854
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.6614693446088796e-05,
+      "loss": 0.0713,
+      "step": 8856
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.660940803382664e-05,
+      "loss": 0.1546,
+      "step": 8858
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.660412262156448e-05,
+      "loss": 0.0783,
+      "step": 8860
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.6598837209302324e-05,
+      "loss": 0.0218,
+      "step": 8862
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.6593551797040173e-05,
+      "loss": 0.059,
+      "step": 8864
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.6588266384778016e-05,
+      "loss": 0.08,
+      "step": 8866
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.658298097251586e-05,
+      "loss": 0.0333,
+      "step": 8868
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.65776955602537e-05,
+      "loss": 0.1142,
+      "step": 8870
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.6572410147991544e-05,
+      "loss": 0.1063,
+      "step": 8872
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.6567124735729387e-05,
+      "loss": 0.0753,
+      "step": 8874
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.656183932346723e-05,
+      "loss": 0.0615,
+      "step": 8876
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.6556553911205072e-05,
+      "loss": 0.0701,
+      "step": 8878
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.6551268498942915e-05,
+      "loss": 0.0558,
+      "step": 8880
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.6545983086680764e-05,
+      "loss": 0.0647,
+      "step": 8882
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.6540697674418607e-05,
+      "loss": 0.1294,
+      "step": 8884
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.653541226215645e-05,
+      "loss": 0.145,
+      "step": 8886
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.6530126849894292e-05,
+      "loss": 0.1311,
+      "step": 8888
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.6524841437632135e-05,
+      "loss": 0.1293,
+      "step": 8890
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.651955602536998e-05,
+      "loss": 0.0762,
+      "step": 8892
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.6514270613107823e-05,
+      "loss": 0.029,
+      "step": 8894
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.6508985200845666e-05,
+      "loss": 0.086,
+      "step": 8896
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.650369978858351e-05,
+      "loss": 0.0596,
+      "step": 8898
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.6498414376321358e-05,
+      "loss": 0.0437,
+      "step": 8900
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.64931289640592e-05,
+      "loss": 0.1034,
+      "step": 8902
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.6487843551797044e-05,
+      "loss": 0.0791,
+      "step": 8904
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.6482558139534886e-05,
+      "loss": 0.1433,
+      "step": 8906
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.647727272727273e-05,
+      "loss": 0.0915,
+      "step": 8908
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.647198731501057e-05,
+      "loss": 0.052,
+      "step": 8910
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.6466701902748414e-05,
+      "loss": 0.1161,
+      "step": 8912
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.6461416490486257e-05,
+      "loss": 0.0793,
+      "step": 8914
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.64561310782241e-05,
+      "loss": 0.1089,
+      "step": 8916
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.645084566596195e-05,
+      "loss": 0.1494,
+      "step": 8918
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.6445560253699792e-05,
+      "loss": 0.128,
+      "step": 8920
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.6440274841437634e-05,
+      "loss": 0.0654,
+      "step": 8922
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.6434989429175477e-05,
+      "loss": 0.1304,
+      "step": 8924
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.642970401691332e-05,
+      "loss": 0.0682,
+      "step": 8926
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.6424418604651162e-05,
+      "loss": 0.0443,
+      "step": 8928
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.6419133192389005e-05,
+      "loss": 0.1399,
+      "step": 8930
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.6413847780126848e-05,
+      "loss": 0.0601,
+      "step": 8932
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.640856236786469e-05,
+      "loss": 0.0917,
+      "step": 8934
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.640327695560254e-05,
+      "loss": 0.0802,
+      "step": 8936
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.6397991543340383e-05,
+      "loss": 0.0652,
+      "step": 8938
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.6392706131078225e-05,
+      "loss": 0.0827,
+      "step": 8940
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.6387420718816068e-05,
+      "loss": 0.0792,
+      "step": 8942
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.6382135306553914e-05,
+      "loss": 0.0509,
+      "step": 8944
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.6376849894291757e-05,
+      "loss": 0.0269,
+      "step": 8946
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.63715644820296e-05,
+      "loss": 0.115,
+      "step": 8948
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.6366279069767442e-05,
+      "loss": 0.068,
+      "step": 8950
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.6360993657505285e-05,
+      "loss": 0.0246,
+      "step": 8952
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.6355708245243134e-05,
+      "loss": 0.0227,
+      "step": 8954
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.6350422832980977e-05,
+      "loss": 0.1095,
+      "step": 8956
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.634513742071882e-05,
+      "loss": 0.0427,
+      "step": 8958
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.6339852008456662e-05,
+      "loss": 0.1263,
+      "step": 8960
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.6334566596194505e-05,
+      "loss": 0.0586,
+      "step": 8962
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.6329281183932347e-05,
+      "loss": 0.2595,
+      "step": 8964
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.632399577167019e-05,
+      "loss": 0.0935,
+      "step": 8966
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.6318710359408033e-05,
+      "loss": 0.0694,
+      "step": 8968
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.6313424947145875e-05,
+      "loss": 0.0243,
+      "step": 8970
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.6308139534883725e-05,
+      "loss": 0.0722,
+      "step": 8972
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.6302854122621568e-05,
+      "loss": 0.0713,
+      "step": 8974
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.629756871035941e-05,
+      "loss": 0.0691,
+      "step": 8976
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.6292283298097253e-05,
+      "loss": 0.1,
+      "step": 8978
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.6286997885835096e-05,
+      "loss": 0.0964,
+      "step": 8980
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.6281712473572938e-05,
+      "loss": 0.0803,
+      "step": 8982
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.627642706131078e-05,
+      "loss": 0.0492,
+      "step": 8984
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.6271141649048624e-05,
+      "loss": 0.0719,
+      "step": 8986
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.626585623678647e-05,
+      "loss": 0.0741,
+      "step": 8988
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.6260570824524316e-05,
+      "loss": 0.0719,
+      "step": 8990
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.625528541226216e-05,
+      "loss": 0.0528,
+      "step": 8992
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.625e-05,
+      "loss": 0.0613,
+      "step": 8994
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.6244714587737844e-05,
+      "loss": 0.0499,
+      "step": 8996
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.623942917547569e-05,
+      "loss": 0.0338,
+      "step": 8998
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.6234143763213532e-05,
+      "loss": 0.1687,
+      "step": 9000
+    },
+    {
+      "epoch": 9.51,
+      "eval_cer": 0.03066400683955543,
+      "eval_loss": 0.7769936323165894,
+      "eval_runtime": 127.7044,
+      "eval_samples_per_second": 6.586,
+      "eval_steps_per_second": 0.83,
+      "step": 9000
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.6228858350951375e-05,
+      "loss": 0.0774,
+      "step": 9002
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.6223572938689218e-05,
+      "loss": 0.09,
+      "step": 9004
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.621828752642706e-05,
+      "loss": 0.0983,
+      "step": 9006
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.621300211416491e-05,
+      "loss": 0.0581,
+      "step": 9008
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.6207716701902753e-05,
+      "loss": 0.0721,
+      "step": 9010
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.6202431289640595e-05,
+      "loss": 0.0352,
+      "step": 9012
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.6197145877378438e-05,
+      "loss": 0.1807,
+      "step": 9014
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.619186046511628e-05,
+      "loss": 0.0639,
+      "step": 9016
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.6186575052854123e-05,
+      "loss": 0.0835,
+      "step": 9018
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.6181289640591966e-05,
+      "loss": 0.139,
+      "step": 9020
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.617600422832981e-05,
+      "loss": 0.0826,
+      "step": 9022
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.617071881606765e-05,
+      "loss": 0.1681,
+      "step": 9024
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.61654334038055e-05,
+      "loss": 0.0913,
+      "step": 9026
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.6160147991543343e-05,
+      "loss": 0.0909,
+      "step": 9028
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.6154862579281186e-05,
+      "loss": 0.0458,
+      "step": 9030
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.614957716701903e-05,
+      "loss": 0.0742,
+      "step": 9032
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.614429175475687e-05,
+      "loss": 0.1146,
+      "step": 9034
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.6139006342494714e-05,
+      "loss": 0.1539,
+      "step": 9036
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.6133720930232557e-05,
+      "loss": 0.0891,
+      "step": 9038
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.61284355179704e-05,
+      "loss": 0.2471,
+      "step": 9040
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.6123150105708246e-05,
+      "loss": 0.1032,
+      "step": 9042
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.611786469344609e-05,
+      "loss": 0.1223,
+      "step": 9044
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.6112579281183934e-05,
+      "loss": 0.1236,
+      "step": 9046
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.6107293868921777e-05,
+      "loss": 0.1142,
+      "step": 9048
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.610200845665962e-05,
+      "loss": 0.1083,
+      "step": 9050
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.6096723044397466e-05,
+      "loss": 0.0754,
+      "step": 9052
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.609143763213531e-05,
+      "loss": 0.0589,
+      "step": 9054
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.608615221987315e-05,
+      "loss": 0.0763,
+      "step": 9056
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.6080866807610994e-05,
+      "loss": 0.0907,
+      "step": 9058
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.6075581395348836e-05,
+      "loss": 0.0649,
+      "step": 9060
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.6070295983086686e-05,
+      "loss": 0.0344,
+      "step": 9062
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.606501057082453e-05,
+      "loss": 0.0958,
+      "step": 9064
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.605972515856237e-05,
+      "loss": 0.0438,
+      "step": 9066
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.6054439746300214e-05,
+      "loss": 0.0463,
+      "step": 9068
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.6049154334038056e-05,
+      "loss": 0.032,
+      "step": 9070
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.60438689217759e-05,
+      "loss": 0.0797,
+      "step": 9072
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.6038583509513742e-05,
+      "loss": 0.1319,
+      "step": 9074
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.6033298097251584e-05,
+      "loss": 0.0902,
+      "step": 9076
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.6028012684989427e-05,
+      "loss": 0.0814,
+      "step": 9078
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.6022727272727277e-05,
+      "loss": 0.1081,
+      "step": 9080
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.601744186046512e-05,
+      "loss": 0.0604,
+      "step": 9082
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.6012156448202962e-05,
+      "loss": 0.0702,
+      "step": 9084
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.6006871035940805e-05,
+      "loss": 0.0974,
+      "step": 9086
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 2.6001585623678647e-05,
+      "loss": 0.0582,
+      "step": 9088
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 2.599630021141649e-05,
+      "loss": 0.1291,
+      "step": 9090
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 2.5991014799154333e-05,
+      "loss": 0.0836,
+      "step": 9092
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 2.5985729386892175e-05,
+      "loss": 0.0978,
+      "step": 9094
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 2.598044397463002e-05,
+      "loss": 0.0746,
+      "step": 9096
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 2.5975158562367867e-05,
+      "loss": 0.0816,
+      "step": 9098
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 2.596987315010571e-05,
+      "loss": 0.0664,
+      "step": 9100
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 2.5964587737843553e-05,
+      "loss": 0.0668,
+      "step": 9102
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 2.59593023255814e-05,
+      "loss": 0.1026,
+      "step": 9104
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 2.595401691331924e-05,
+      "loss": 0.0647,
+      "step": 9106
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 2.5948731501057084e-05,
+      "loss": 0.0637,
+      "step": 9108
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 2.5943446088794927e-05,
+      "loss": 0.093,
+      "step": 9110
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 2.593816067653277e-05,
+      "loss": 0.0501,
+      "step": 9112
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 2.5932875264270612e-05,
+      "loss": 0.1349,
+      "step": 9114
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 2.592758985200846e-05,
+      "loss": 0.0521,
+      "step": 9116
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 2.5922304439746304e-05,
+      "loss": 0.0251,
+      "step": 9118
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 2.5917019027484147e-05,
+      "loss": 0.0449,
+      "step": 9120
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 2.591173361522199e-05,
+      "loss": 0.0847,
+      "step": 9122
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 2.5906448202959832e-05,
+      "loss": 0.1406,
+      "step": 9124
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 2.5901162790697675e-05,
+      "loss": 0.0304,
+      "step": 9126
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 2.5895877378435518e-05,
+      "loss": 0.0721,
+      "step": 9128
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 2.589059196617336e-05,
+      "loss": 0.061,
+      "step": 9130
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 2.5885306553911203e-05,
+      "loss": 0.1242,
+      "step": 9132
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 2.5880021141649052e-05,
+      "loss": 0.0477,
+      "step": 9134
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 2.5874735729386895e-05,
+      "loss": 0.0527,
+      "step": 9136
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 2.5869450317124738e-05,
+      "loss": 0.3248,
+      "step": 9138
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 2.586416490486258e-05,
+      "loss": 0.112,
+      "step": 9140
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 2.5858879492600423e-05,
+      "loss": 0.1445,
+      "step": 9142
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 2.5853594080338266e-05,
+      "loss": 0.131,
+      "step": 9144
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 2.584830866807611e-05,
+      "loss": 0.069,
+      "step": 9146
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 2.5843023255813955e-05,
+      "loss": 0.0555,
+      "step": 9148
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 2.5837737843551797e-05,
+      "loss": 0.1066,
+      "step": 9150
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 2.5832452431289643e-05,
+      "loss": 0.0949,
+      "step": 9152
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 2.5827167019027486e-05,
+      "loss": 0.1722,
+      "step": 9154
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 2.582188160676533e-05,
+      "loss": 0.0974,
+      "step": 9156
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 2.5816596194503175e-05,
+      "loss": 0.146,
+      "step": 9158
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 2.5811310782241017e-05,
+      "loss": 0.0641,
+      "step": 9160
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 2.580602536997886e-05,
+      "loss": 0.0692,
+      "step": 9162
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 2.5800739957716703e-05,
+      "loss": 0.0374,
+      "step": 9164
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 2.5795454545454545e-05,
+      "loss": 0.0374,
+      "step": 9166
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 2.5790169133192388e-05,
+      "loss": 0.0742,
+      "step": 9168
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 2.5784883720930237e-05,
+      "loss": 0.0503,
+      "step": 9170
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 2.577959830866808e-05,
+      "loss": 0.0705,
+      "step": 9172
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 2.5774312896405923e-05,
+      "loss": 0.1188,
+      "step": 9174
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 2.5769027484143765e-05,
+      "loss": 0.1263,
+      "step": 9176
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 2.5763742071881608e-05,
+      "loss": 0.0268,
+      "step": 9178
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 2.575845665961945e-05,
+      "loss": 0.0401,
+      "step": 9180
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 2.5753171247357293e-05,
+      "loss": 0.2058,
+      "step": 9182
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 2.5747885835095136e-05,
+      "loss": 0.0778,
+      "step": 9184
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 2.574260042283298e-05,
+      "loss": 0.0561,
+      "step": 9186
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 2.5737315010570828e-05,
+      "loss": 0.2252,
+      "step": 9188
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 2.573202959830867e-05,
+      "loss": 0.1165,
+      "step": 9190
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 2.5726744186046514e-05,
+      "loss": 0.1053,
+      "step": 9192
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 2.5721458773784356e-05,
+      "loss": 0.0465,
+      "step": 9194
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 2.57161733615222e-05,
+      "loss": 0.0784,
+      "step": 9196
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 2.571088794926004e-05,
+      "loss": 0.0708,
+      "step": 9198
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 2.5705602536997884e-05,
+      "loss": 0.2093,
+      "step": 9200
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 2.570031712473573e-05,
+      "loss": 0.2111,
+      "step": 9202
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 2.5695031712473573e-05,
+      "loss": 0.1183,
+      "step": 9204
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 2.568974630021142e-05,
+      "loss": 0.05,
+      "step": 9206
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 2.5684460887949262e-05,
+      "loss": 0.2306,
+      "step": 9208
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 2.5679175475687104e-05,
+      "loss": 0.0342,
+      "step": 9210
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 2.567389006342495e-05,
+      "loss": 0.1203,
+      "step": 9212
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 2.5668604651162793e-05,
+      "loss": 0.1217,
+      "step": 9214
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 2.5663319238900636e-05,
+      "loss": 0.0543,
+      "step": 9216
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 2.565803382663848e-05,
+      "loss": 0.1482,
+      "step": 9218
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 2.565274841437632e-05,
+      "loss": 0.1286,
+      "step": 9220
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 2.5647463002114164e-05,
+      "loss": 0.1369,
+      "step": 9222
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 2.5642177589852013e-05,
+      "loss": 0.0661,
+      "step": 9224
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 2.5636892177589856e-05,
+      "loss": 0.1009,
+      "step": 9226
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 2.56316067653277e-05,
+      "loss": 0.0759,
+      "step": 9228
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 2.562632135306554e-05,
+      "loss": 0.099,
+      "step": 9230
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 2.5621035940803384e-05,
+      "loss": 0.1197,
+      "step": 9232
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 2.5615750528541227e-05,
+      "loss": 0.1576,
+      "step": 9234
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 2.561046511627907e-05,
+      "loss": 0.1255,
+      "step": 9236
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 2.5605179704016912e-05,
+      "loss": 0.1013,
+      "step": 9238
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 2.5599894291754755e-05,
+      "loss": 0.0482,
+      "step": 9240
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 2.5594608879492604e-05,
+      "loss": 0.0474,
+      "step": 9242
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 2.5589323467230447e-05,
+      "loss": 0.0222,
+      "step": 9244
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 2.558403805496829e-05,
+      "loss": 0.0592,
+      "step": 9246
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.5578752642706132e-05,
+      "loss": 0.0322,
+      "step": 9248
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.5573467230443975e-05,
+      "loss": 0.1504,
+      "step": 9250
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.5568181818181817e-05,
+      "loss": 0.0378,
+      "step": 9252
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.556289640591966e-05,
+      "loss": 0.1115,
+      "step": 9254
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.5557610993657506e-05,
+      "loss": 0.1217,
+      "step": 9256
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.555232558139535e-05,
+      "loss": 0.1118,
+      "step": 9258
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.5547040169133195e-05,
+      "loss": 0.0188,
+      "step": 9260
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.5541754756871038e-05,
+      "loss": 0.092,
+      "step": 9262
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.5536469344608884e-05,
+      "loss": 0.203,
+      "step": 9264
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.5531183932346726e-05,
+      "loss": 0.0928,
+      "step": 9266
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.552589852008457e-05,
+      "loss": 0.0479,
+      "step": 9268
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.552061310782241e-05,
+      "loss": 0.0719,
+      "step": 9270
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.5515327695560254e-05,
+      "loss": 0.0685,
+      "step": 9272
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.5510042283298097e-05,
+      "loss": 0.0747,
+      "step": 9274
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 2.550475687103594e-05,
+      "loss": 0.1232,
+      "step": 9276
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 2.549947145877379e-05,
+      "loss": 0.0749,
+      "step": 9278
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 2.5494186046511632e-05,
+      "loss": 0.091,
+      "step": 9280
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 2.5488900634249474e-05,
+      "loss": 0.0926,
+      "step": 9282
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 2.5483615221987317e-05,
+      "loss": 0.0651,
+      "step": 9284
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 2.547832980972516e-05,
+      "loss": 0.0229,
+      "step": 9286
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 2.5473044397463003e-05,
+      "loss": 0.0882,
+      "step": 9288
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 2.5467758985200845e-05,
+      "loss": 0.151,
+      "step": 9290
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 2.5462473572938688e-05,
+      "loss": 0.0874,
+      "step": 9292
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 2.545718816067653e-05,
+      "loss": 0.1641,
+      "step": 9294
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 2.545190274841438e-05,
+      "loss": 0.1099,
+      "step": 9296
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 2.5446617336152223e-05,
+      "loss": 0.0902,
+      "step": 9298
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 2.5441331923890065e-05,
+      "loss": 0.0434,
+      "step": 9300
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 2.5436046511627908e-05,
+      "loss": 0.1326,
+      "step": 9302
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 2.543076109936575e-05,
+      "loss": 0.0813,
+      "step": 9304
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 2.5425475687103593e-05,
+      "loss": 0.0893,
+      "step": 9306
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 2.542019027484144e-05,
+      "loss": 0.0796,
+      "step": 9308
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 2.5414904862579282e-05,
+      "loss": 0.0507,
+      "step": 9310
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 2.5409619450317125e-05,
+      "loss": 0.05,
+      "step": 9312
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 2.540433403805497e-05,
+      "loss": 0.1088,
+      "step": 9314
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 2.5399048625792813e-05,
+      "loss": 0.108,
+      "step": 9316
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 2.539376321353066e-05,
+      "loss": 0.0851,
+      "step": 9318
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 2.5388477801268502e-05,
+      "loss": 0.0538,
+      "step": 9320
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 2.5383192389006345e-05,
+      "loss": 0.0998,
+      "step": 9322
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 2.5377906976744188e-05,
+      "loss": 0.0808,
+      "step": 9324
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 2.537262156448203e-05,
+      "loss": 0.086,
+      "step": 9326
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 2.5367336152219873e-05,
+      "loss": 0.0555,
+      "step": 9328
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 2.5362050739957716e-05,
+      "loss": 0.0468,
+      "step": 9330
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 2.5356765327695565e-05,
+      "loss": 0.0644,
+      "step": 9332
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 2.5351479915433408e-05,
+      "loss": 0.2785,
+      "step": 9334
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 2.534619450317125e-05,
+      "loss": 0.0736,
+      "step": 9336
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 2.5340909090909093e-05,
+      "loss": 0.0374,
+      "step": 9338
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 2.5335623678646936e-05,
+      "loss": 0.0782,
+      "step": 9340
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 2.533033826638478e-05,
+      "loss": 0.1181,
+      "step": 9342
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 2.532505285412262e-05,
+      "loss": 0.1358,
+      "step": 9344
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 2.5319767441860464e-05,
+      "loss": 0.0958,
+      "step": 9346
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 2.5314482029598306e-05,
+      "loss": 0.1822,
+      "step": 9348
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 2.5309196617336156e-05,
+      "loss": 0.1081,
+      "step": 9350
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 2.5303911205074e-05,
+      "loss": 0.0782,
+      "step": 9352
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 2.529862579281184e-05,
+      "loss": 0.1172,
+      "step": 9354
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 2.5293340380549684e-05,
+      "loss": 0.0593,
+      "step": 9356
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 2.5288054968287526e-05,
+      "loss": 0.0192,
+      "step": 9358
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 2.528276955602537e-05,
+      "loss": 0.073,
+      "step": 9360
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 2.5277484143763215e-05,
+      "loss": 0.0692,
+      "step": 9362
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 2.5272198731501058e-05,
+      "loss": 0.0479,
+      "step": 9364
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 2.52669133192389e-05,
+      "loss": 0.0936,
+      "step": 9366
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 2.5261627906976747e-05,
+      "loss": 0.06,
+      "step": 9368
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 2.5256342494714593e-05,
+      "loss": 0.194,
+      "step": 9370
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.5251057082452435e-05,
+      "loss": 0.1149,
+      "step": 9372
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.5245771670190278e-05,
+      "loss": 0.0893,
+      "step": 9374
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.524048625792812e-05,
+      "loss": 0.0446,
+      "step": 9376
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.5235200845665963e-05,
+      "loss": 0.1115,
+      "step": 9378
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.5229915433403806e-05,
+      "loss": 0.1539,
+      "step": 9380
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.522463002114165e-05,
+      "loss": 0.0537,
+      "step": 9382
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.521934460887949e-05,
+      "loss": 0.047,
+      "step": 9384
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.521405919661734e-05,
+      "loss": 0.0716,
+      "step": 9386
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.5208773784355183e-05,
+      "loss": 0.0348,
+      "step": 9388
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.5203488372093026e-05,
+      "loss": 0.1238,
+      "step": 9390
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.519820295983087e-05,
+      "loss": 0.016,
+      "step": 9392
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.519291754756871e-05,
+      "loss": 0.099,
+      "step": 9394
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.5187632135306554e-05,
+      "loss": 0.1096,
+      "step": 9396
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.5182346723044397e-05,
+      "loss": 0.1113,
+      "step": 9398
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.517706131078224e-05,
+      "loss": 0.038,
+      "step": 9400
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.5171775898520082e-05,
+      "loss": 0.0852,
+      "step": 9402
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.516649048625793e-05,
+      "loss": 0.1253,
+      "step": 9404
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.5161205073995774e-05,
+      "loss": 0.0963,
+      "step": 9406
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.5155919661733617e-05,
+      "loss": 0.1237,
+      "step": 9408
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.515063424947146e-05,
+      "loss": 0.0919,
+      "step": 9410
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.5145348837209302e-05,
+      "loss": 0.0494,
+      "step": 9412
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.514006342494715e-05,
+      "loss": 0.0852,
+      "step": 9414
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.513477801268499e-05,
+      "loss": 0.1492,
+      "step": 9416
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.5129492600422834e-05,
+      "loss": 0.0884,
+      "step": 9418
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.5124207188160676e-05,
+      "loss": 0.1411,
+      "step": 9420
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.5118921775898522e-05,
+      "loss": 0.0512,
+      "step": 9422
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.511363636363637e-05,
+      "loss": 0.1133,
+      "step": 9424
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.510835095137421e-05,
+      "loss": 0.0455,
+      "step": 9426
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 2.5103065539112054e-05,
+      "loss": 0.0743,
+      "step": 9428
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 2.5097780126849897e-05,
+      "loss": 0.1326,
+      "step": 9430
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 2.509249471458774e-05,
+      "loss": 0.0816,
+      "step": 9432
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 2.5087209302325582e-05,
+      "loss": 0.0581,
+      "step": 9434
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 2.5081923890063425e-05,
+      "loss": 0.1075,
+      "step": 9436
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 2.5076638477801267e-05,
+      "loss": 0.0341,
+      "step": 9438
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 2.5071353065539117e-05,
+      "loss": 0.0657,
+      "step": 9440
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 2.506606765327696e-05,
+      "loss": 0.0317,
+      "step": 9442
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 2.5060782241014802e-05,
+      "loss": 0.1027,
+      "step": 9444
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 2.5055496828752645e-05,
+      "loss": 0.1306,
+      "step": 9446
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 2.5050211416490487e-05,
+      "loss": 0.0308,
+      "step": 9448
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 2.504492600422833e-05,
+      "loss": 0.0878,
+      "step": 9450
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 2.5039640591966173e-05,
+      "loss": 0.0352,
+      "step": 9452
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 2.5034355179704015e-05,
+      "loss": 0.0678,
+      "step": 9454
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.5029069767441858e-05,
+      "loss": 0.1047,
+      "step": 9456
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.5023784355179707e-05,
+      "loss": 0.0423,
+      "step": 9458
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.501849894291755e-05,
+      "loss": 0.0521,
+      "step": 9460
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.5013213530655393e-05,
+      "loss": 0.0502,
+      "step": 9462
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.5007928118393235e-05,
+      "loss": 0.0516,
+      "step": 9464
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 2.5002642706131078e-05,
+      "loss": 0.1356,
+      "step": 9466
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 2.4997357293868924e-05,
+      "loss": 0.142,
+      "step": 9468
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 2.4992071881606767e-05,
+      "loss": 0.0707,
+      "step": 9470
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 2.4986786469344613e-05,
+      "loss": 0.0274,
+      "step": 9472
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 2.4981501057082456e-05,
+      "loss": 0.1028,
+      "step": 9474
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 2.4976215644820298e-05,
+      "loss": 0.0566,
+      "step": 9476
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 2.497093023255814e-05,
+      "loss": 0.2284,
+      "step": 9478
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 2.4965644820295984e-05,
+      "loss": 0.106,
+      "step": 9480
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 2.496035940803383e-05,
+      "loss": 0.076,
+      "step": 9482
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 2.4955073995771672e-05,
+      "loss": 0.1524,
+      "step": 9484
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 2.4949788583509515e-05,
+      "loss": 0.024,
+      "step": 9486
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 2.4944503171247358e-05,
+      "loss": 0.0783,
+      "step": 9488
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 2.4939217758985204e-05,
+      "loss": 0.0752,
+      "step": 9490
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 2.4933932346723046e-05,
+      "loss": 0.0556,
+      "step": 9492
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 2.492864693446089e-05,
+      "loss": 0.0667,
+      "step": 9494
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 2.4923361522198732e-05,
+      "loss": 0.0455,
+      "step": 9496
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 2.4918076109936574e-05,
+      "loss": 0.0423,
+      "step": 9498
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 2.491279069767442e-05,
+      "loss": 0.0616,
+      "step": 9500
+    },
+    {
+      "epoch": 10.04,
+      "eval_cer": 0.020860644058136223,
+      "eval_loss": 0.5472248196601868,
+      "eval_runtime": 127.8063,
+      "eval_samples_per_second": 6.58,
+      "eval_steps_per_second": 0.829,
+      "step": 9500
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 2.4907505285412263e-05,
+      "loss": 0.1153,
+      "step": 9502
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 2.4902219873150106e-05,
+      "loss": 0.0961,
+      "step": 9504
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 2.489693446088795e-05,
+      "loss": 0.0327,
+      "step": 9506
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 2.4891649048625795e-05,
+      "loss": 0.0733,
+      "step": 9508
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 2.4886363636363637e-05,
+      "loss": 0.0801,
+      "step": 9510
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 2.488107822410148e-05,
+      "loss": 0.0344,
+      "step": 9512
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 2.4875792811839323e-05,
+      "loss": 0.0918,
+      "step": 9514
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 2.487050739957717e-05,
+      "loss": 0.1247,
+      "step": 9516
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 2.486522198731501e-05,
+      "loss": 0.0697,
+      "step": 9518
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 2.4859936575052854e-05,
+      "loss": 0.0941,
+      "step": 9520
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 2.48546511627907e-05,
+      "loss": 0.0839,
+      "step": 9522
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 2.4849365750528543e-05,
+      "loss": 0.0163,
+      "step": 9524
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 2.484408033826639e-05,
+      "loss": 0.0625,
+      "step": 9526
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 2.483879492600423e-05,
+      "loss": 0.0795,
+      "step": 9528
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 2.4833509513742074e-05,
+      "loss": 0.0719,
+      "step": 9530
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 2.4828224101479917e-05,
+      "loss": 0.0796,
+      "step": 9532
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 2.482293868921776e-05,
+      "loss": 0.0175,
+      "step": 9534
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 2.4817653276955606e-05,
+      "loss": 0.0797,
+      "step": 9536
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 2.4812367864693448e-05,
+      "loss": 0.0956,
+      "step": 9538
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 2.480708245243129e-05,
+      "loss": 0.0526,
+      "step": 9540
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 2.4801797040169134e-05,
+      "loss": 0.0719,
+      "step": 9542
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 2.479651162790698e-05,
+      "loss": 0.0664,
+      "step": 9544
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 2.4791226215644822e-05,
+      "loss": 0.0272,
+      "step": 9546
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 2.4785940803382665e-05,
+      "loss": 0.043,
+      "step": 9548
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 2.4780655391120508e-05,
+      "loss": 0.1261,
+      "step": 9550
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 2.477536997885835e-05,
+      "loss": 0.2794,
+      "step": 9552
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 2.4770084566596196e-05,
+      "loss": 0.0698,
+      "step": 9554
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 2.476479915433404e-05,
+      "loss": 0.2251,
+      "step": 9556
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 2.4759513742071882e-05,
+      "loss": 0.0471,
+      "step": 9558
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 2.4754228329809724e-05,
+      "loss": 0.0455,
+      "step": 9560
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 2.474894291754757e-05,
+      "loss": 0.3214,
+      "step": 9562
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 2.4743657505285413e-05,
+      "loss": 0.2688,
+      "step": 9564
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 2.4738372093023256e-05,
+      "loss": 0.1333,
+      "step": 9566
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 2.47330866807611e-05,
+      "loss": 0.1161,
+      "step": 9568
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 2.4727801268498945e-05,
+      "loss": 0.0928,
+      "step": 9570
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 2.4722515856236787e-05,
+      "loss": 0.0991,
+      "step": 9572
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 2.4717230443974633e-05,
+      "loss": 0.0438,
+      "step": 9574
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 2.4711945031712476e-05,
+      "loss": 0.1875,
+      "step": 9576
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 2.470665961945032e-05,
+      "loss": 0.1174,
+      "step": 9578
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 2.4701374207188165e-05,
+      "loss": 0.0306,
+      "step": 9580
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 2.4696088794926007e-05,
+      "loss": 0.1353,
+      "step": 9582
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 2.469080338266385e-05,
+      "loss": 0.0791,
+      "step": 9584
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 2.4685517970401693e-05,
+      "loss": 0.0907,
+      "step": 9586
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 2.4680232558139535e-05,
+      "loss": 0.1941,
+      "step": 9588
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 2.467494714587738e-05,
+      "loss": 0.0339,
+      "step": 9590
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 2.4669661733615224e-05,
+      "loss": 0.0292,
+      "step": 9592
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 2.4664376321353067e-05,
+      "loss": 0.0474,
+      "step": 9594
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 2.465909090909091e-05,
+      "loss": 0.0832,
+      "step": 9596
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 2.4653805496828755e-05,
+      "loss": 0.0618,
+      "step": 9598
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 2.4648520084566598e-05,
+      "loss": 0.0498,
+      "step": 9600
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 2.464323467230444e-05,
+      "loss": 0.0797,
+      "step": 9602
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 2.4637949260042283e-05,
+      "loss": 0.0558,
+      "step": 9604
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 2.4632663847780126e-05,
+      "loss": 0.0621,
+      "step": 9606
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 2.4627378435517972e-05,
+      "loss": 0.0617,
+      "step": 9608
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 2.4622093023255815e-05,
+      "loss": 0.1126,
+      "step": 9610
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 2.4616807610993658e-05,
+      "loss": 0.0502,
+      "step": 9612
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 2.46115221987315e-05,
+      "loss": 0.0547,
+      "step": 9614
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 2.4606236786469346e-05,
+      "loss": 0.0615,
+      "step": 9616
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 2.460095137420719e-05,
+      "loss": 0.0155,
+      "step": 9618
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 2.459566596194503e-05,
+      "loss": 0.0557,
+      "step": 9620
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 2.4590380549682874e-05,
+      "loss": 0.0594,
+      "step": 9622
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 2.458509513742072e-05,
+      "loss": 0.0639,
+      "step": 9624
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 2.4579809725158563e-05,
+      "loss": 0.0344,
+      "step": 9626
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 2.457452431289641e-05,
+      "loss": 0.0442,
+      "step": 9628
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 2.4569238900634252e-05,
+      "loss": 0.1143,
+      "step": 9630
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 2.4563953488372094e-05,
+      "loss": 0.0763,
+      "step": 9632
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 2.455866807610994e-05,
+      "loss": 0.0544,
+      "step": 9634
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 2.4553382663847783e-05,
+      "loss": 0.0375,
+      "step": 9636
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 2.4548097251585626e-05,
+      "loss": 0.0968,
+      "step": 9638
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 2.454281183932347e-05,
+      "loss": 0.1139,
+      "step": 9640
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 2.453752642706131e-05,
+      "loss": 0.0868,
+      "step": 9642
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 2.4532241014799157e-05,
+      "loss": 0.0274,
+      "step": 9644
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 2.4526955602537e-05,
+      "loss": 0.1256,
+      "step": 9646
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 2.4521670190274843e-05,
+      "loss": 0.123,
+      "step": 9648
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 2.4516384778012685e-05,
+      "loss": 0.0492,
+      "step": 9650
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 2.4511099365750528e-05,
+      "loss": 0.0606,
+      "step": 9652
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 2.4505813953488374e-05,
+      "loss": 0.0915,
+      "step": 9654
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 2.4500528541226217e-05,
+      "loss": 0.0606,
+      "step": 9656
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 2.449524312896406e-05,
+      "loss": 0.0424,
+      "step": 9658
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 2.4489957716701902e-05,
+      "loss": 0.0585,
+      "step": 9660
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 2.4484672304439748e-05,
+      "loss": 0.0735,
+      "step": 9662
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 2.447938689217759e-05,
+      "loss": 0.0558,
+      "step": 9664
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 2.4474101479915433e-05,
+      "loss": 0.0444,
+      "step": 9666
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 2.4468816067653276e-05,
+      "loss": 0.0697,
+      "step": 9668
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 2.446353065539112e-05,
+      "loss": 0.1268,
+      "step": 9670
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 2.4458245243128965e-05,
+      "loss": 0.1274,
+      "step": 9672
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 2.4452959830866807e-05,
+      "loss": 0.1141,
+      "step": 9674
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 2.4447674418604654e-05,
+      "loss": 0.0624,
+      "step": 9676
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 2.4442389006342496e-05,
+      "loss": 0.054,
+      "step": 9678
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 2.4437103594080342e-05,
+      "loss": 0.0644,
+      "step": 9680
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 2.4431818181818185e-05,
+      "loss": 0.0435,
+      "step": 9682
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 2.4426532769556028e-05,
+      "loss": 0.0393,
+      "step": 9684
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 2.442124735729387e-05,
+      "loss": 0.1323,
+      "step": 9686
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 2.4415961945031713e-05,
+      "loss": 0.1019,
+      "step": 9688
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 2.441067653276956e-05,
+      "loss": 0.0803,
+      "step": 9690
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 2.44053911205074e-05,
+      "loss": 0.0582,
+      "step": 9692
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 2.4400105708245244e-05,
+      "loss": 0.0659,
+      "step": 9694
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 2.4394820295983087e-05,
+      "loss": 0.0528,
+      "step": 9696
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 2.4389534883720933e-05,
+      "loss": 0.0614,
+      "step": 9698
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 2.4384249471458776e-05,
+      "loss": 0.1287,
+      "step": 9700
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 2.437896405919662e-05,
+      "loss": 0.211,
+      "step": 9702
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 2.437367864693446e-05,
+      "loss": 0.0376,
+      "step": 9704
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 2.4368393234672304e-05,
+      "loss": 0.0271,
+      "step": 9706
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 2.436310782241015e-05,
+      "loss": 0.1002,
+      "step": 9708
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 2.4357822410147992e-05,
+      "loss": 0.1325,
+      "step": 9710
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 2.4352536997885835e-05,
+      "loss": 0.0062,
+      "step": 9712
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 2.4347251585623678e-05,
+      "loss": 0.068,
+      "step": 9714
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 2.4341966173361524e-05,
+      "loss": 0.0643,
+      "step": 9716
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 2.4336680761099367e-05,
+      "loss": 0.1679,
+      "step": 9718
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 2.433139534883721e-05,
+      "loss": 0.1077,
+      "step": 9720
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 2.4326109936575052e-05,
+      "loss": 0.0535,
+      "step": 9722
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 2.4320824524312898e-05,
+      "loss": 0.1192,
+      "step": 9724
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 2.431553911205074e-05,
+      "loss": 0.057,
+      "step": 9726
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 2.4310253699788583e-05,
+      "loss": 0.1002,
+      "step": 9728
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.430496828752643e-05,
+      "loss": 0.2332,
+      "step": 9730
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.4299682875264272e-05,
+      "loss": 0.0869,
+      "step": 9732
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.4294397463002118e-05,
+      "loss": 0.2152,
+      "step": 9734
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.428911205073996e-05,
+      "loss": 0.1526,
+      "step": 9736
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.4283826638477803e-05,
+      "loss": 0.1586,
+      "step": 9738
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 2.4278541226215646e-05,
+      "loss": 0.1074,
+      "step": 9740
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 2.427325581395349e-05,
+      "loss": 0.0455,
+      "step": 9742
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 2.4267970401691335e-05,
+      "loss": 0.0448,
+      "step": 9744
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 2.4262684989429177e-05,
+      "loss": 0.0873,
+      "step": 9746
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 2.425739957716702e-05,
+      "loss": 0.0349,
+      "step": 9748
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 2.4252114164904863e-05,
+      "loss": 0.0399,
+      "step": 9750
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 2.424682875264271e-05,
+      "loss": 0.0379,
+      "step": 9752
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 2.424154334038055e-05,
+      "loss": 0.0655,
+      "step": 9754
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 2.4236257928118394e-05,
+      "loss": 0.0885,
+      "step": 9756
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 2.4230972515856237e-05,
+      "loss": 0.0374,
+      "step": 9758
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 2.422568710359408e-05,
+      "loss": 0.0658,
+      "step": 9760
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 2.4220401691331926e-05,
+      "loss": 0.042,
+      "step": 9762
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 2.421511627906977e-05,
+      "loss": 0.0622,
+      "step": 9764
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 2.420983086680761e-05,
+      "loss": 0.0568,
+      "step": 9766
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 2.4204545454545454e-05,
+      "loss": 0.0538,
+      "step": 9768
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 2.41992600422833e-05,
+      "loss": 0.0909,
+      "step": 9770
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 2.4193974630021142e-05,
+      "loss": 0.1713,
+      "step": 9772
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 2.4188689217758985e-05,
+      "loss": 0.0973,
+      "step": 9774
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 2.4183403805496828e-05,
+      "loss": 0.0276,
+      "step": 9776
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.4178118393234674e-05,
+      "loss": 0.063,
+      "step": 9778
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.4172832980972516e-05,
+      "loss": 0.0193,
+      "step": 9780
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.4167547568710363e-05,
+      "loss": 0.0401,
+      "step": 9782
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.4162262156448205e-05,
+      "loss": 0.1097,
+      "step": 9784
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.4156976744186048e-05,
+      "loss": 0.0611,
+      "step": 9786
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 2.4151691331923894e-05,
+      "loss": 0.0231,
+      "step": 9788
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 2.4146405919661737e-05,
+      "loss": 0.145,
+      "step": 9790
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 2.414112050739958e-05,
+      "loss": 0.0839,
+      "step": 9792
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 2.4135835095137422e-05,
+      "loss": 0.0581,
+      "step": 9794
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.4130549682875265e-05,
+      "loss": 0.0452,
+      "step": 9796
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.412526427061311e-05,
+      "loss": 0.0309,
+      "step": 9798
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.4119978858350953e-05,
+      "loss": 0.1888,
+      "step": 9800
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.4114693446088796e-05,
+      "loss": 0.1226,
+      "step": 9802
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.410940803382664e-05,
+      "loss": 0.0596,
+      "step": 9804
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 2.4104122621564485e-05,
+      "loss": 0.0507,
+      "step": 9806
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 2.4098837209302327e-05,
+      "loss": 0.0564,
+      "step": 9808
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 2.409355179704017e-05,
+      "loss": 0.0319,
+      "step": 9810
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 2.4088266384778013e-05,
+      "loss": 0.1549,
+      "step": 9812
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 2.4082980972515855e-05,
+      "loss": 0.1253,
+      "step": 9814
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 2.40776955602537e-05,
+      "loss": 0.0819,
+      "step": 9816
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 2.4072410147991544e-05,
+      "loss": 0.126,
+      "step": 9818
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 2.4067124735729387e-05,
+      "loss": 0.0943,
+      "step": 9820
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 2.406183932346723e-05,
+      "loss": 0.0304,
+      "step": 9822
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 2.4056553911205076e-05,
+      "loss": 0.1265,
+      "step": 9824
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 2.4051268498942918e-05,
+      "loss": 0.0787,
+      "step": 9826
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 2.404598308668076e-05,
+      "loss": 0.104,
+      "step": 9828
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 2.4040697674418604e-05,
+      "loss": 0.0846,
+      "step": 9830
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 2.403541226215645e-05,
+      "loss": 0.0298,
+      "step": 9832
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 2.4030126849894292e-05,
+      "loss": 0.1509,
+      "step": 9834
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 2.402484143763214e-05,
+      "loss": 0.0735,
+      "step": 9836
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 2.401955602536998e-05,
+      "loss": 0.0591,
+      "step": 9838
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 2.4014270613107824e-05,
+      "loss": 0.0444,
+      "step": 9840
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 2.400898520084567e-05,
+      "loss": 0.032,
+      "step": 9842
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 2.4003699788583512e-05,
+      "loss": 0.103,
+      "step": 9844
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 2.3998414376321355e-05,
+      "loss": 0.0892,
+      "step": 9846
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 2.3993128964059198e-05,
+      "loss": 0.0488,
+      "step": 9848
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 2.398784355179704e-05,
+      "loss": 0.0486,
+      "step": 9850
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 2.3982558139534887e-05,
+      "loss": 0.0858,
+      "step": 9852
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 2.397727272727273e-05,
+      "loss": 0.0312,
+      "step": 9854
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 2.3971987315010572e-05,
+      "loss": 0.0556,
+      "step": 9856
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 2.3966701902748415e-05,
+      "loss": 0.0405,
+      "step": 9858
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 2.396141649048626e-05,
+      "loss": 0.0065,
+      "step": 9860
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 2.3956131078224103e-05,
+      "loss": 0.0873,
+      "step": 9862
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 2.3950845665961946e-05,
+      "loss": 0.116,
+      "step": 9864
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 2.394556025369979e-05,
+      "loss": 0.0445,
+      "step": 9866
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 2.394027484143763e-05,
+      "loss": 0.0315,
+      "step": 9868
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 2.3934989429175477e-05,
+      "loss": 0.0742,
+      "step": 9870
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 2.392970401691332e-05,
+      "loss": 0.0774,
+      "step": 9872
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 2.3924418604651163e-05,
+      "loss": 0.1122,
+      "step": 9874
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 2.3919133192389005e-05,
+      "loss": 0.0437,
+      "step": 9876
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 2.391384778012685e-05,
+      "loss": 0.1272,
+      "step": 9878
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 2.3908562367864694e-05,
+      "loss": 0.0312,
+      "step": 9880
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 2.3903276955602537e-05,
+      "loss": 0.1325,
+      "step": 9882
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 2.3897991543340383e-05,
+      "loss": 0.0612,
+      "step": 9884
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 2.3892706131078225e-05,
+      "loss": 0.0457,
+      "step": 9886
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 2.3887420718816068e-05,
+      "loss": 0.1169,
+      "step": 9888
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 2.3882135306553914e-05,
+      "loss": 0.031,
+      "step": 9890
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 2.3876849894291757e-05,
+      "loss": 0.0832,
+      "step": 9892
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 2.38715644820296e-05,
+      "loss": 0.069,
+      "step": 9894
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 2.3866279069767446e-05,
+      "loss": 0.0545,
+      "step": 9896
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 2.3860993657505288e-05,
+      "loss": 0.0588,
+      "step": 9898
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 2.385570824524313e-05,
+      "loss": 0.1478,
+      "step": 9900
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 2.3850422832980974e-05,
+      "loss": 0.057,
+      "step": 9902
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 2.3845137420718816e-05,
+      "loss": 0.1162,
+      "step": 9904
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 2.3839852008456662e-05,
+      "loss": 0.0453,
+      "step": 9906
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 2.3834566596194505e-05,
+      "loss": 0.0668,
+      "step": 9908
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 2.3829281183932348e-05,
+      "loss": 0.0392,
+      "step": 9910
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 2.382399577167019e-05,
+      "loss": 0.0875,
+      "step": 9912
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 2.3818710359408036e-05,
+      "loss": 0.0435,
+      "step": 9914
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 2.381342494714588e-05,
+      "loss": 0.1691,
+      "step": 9916
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 2.3808139534883722e-05,
+      "loss": 0.0769,
+      "step": 9918
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 2.3802854122621564e-05,
+      "loss": 0.0625,
+      "step": 9920
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 2.3797568710359407e-05,
+      "loss": 0.0797,
+      "step": 9922
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 2.3792283298097253e-05,
+      "loss": 0.0506,
+      "step": 9924
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 2.3786997885835096e-05,
+      "loss": 0.0871,
+      "step": 9926
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 2.378171247357294e-05,
+      "loss": 0.1237,
+      "step": 9928
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 2.377642706131078e-05,
+      "loss": 0.1047,
+      "step": 9930
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 2.3771141649048627e-05,
+      "loss": 0.07,
+      "step": 9932
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 2.376585623678647e-05,
+      "loss": 0.0599,
+      "step": 9934
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 2.3760570824524313e-05,
+      "loss": 0.0793,
+      "step": 9936
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 2.375528541226216e-05,
+      "loss": 0.1223,
+      "step": 9938
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 2.375e-05,
+      "loss": 0.0498,
+      "step": 9940
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 2.3744714587737847e-05,
+      "loss": 0.1055,
+      "step": 9942
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 2.373942917547569e-05,
+      "loss": 0.0825,
+      "step": 9944
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 2.3734143763213533e-05,
+      "loss": 0.023,
+      "step": 9946
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 2.3728858350951375e-05,
+      "loss": 0.1028,
+      "step": 9948
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 2.372357293868922e-05,
+      "loss": 0.0584,
+      "step": 9950
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 2.3718287526427064e-05,
+      "loss": 0.1151,
+      "step": 9952
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 2.3713002114164907e-05,
+      "loss": 0.0949,
+      "step": 9954
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 2.370771670190275e-05,
+      "loss": 0.1587,
+      "step": 9956
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 2.3702431289640592e-05,
+      "loss": 0.0479,
+      "step": 9958
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 2.3697145877378438e-05,
+      "loss": 0.0412,
+      "step": 9960
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 2.369186046511628e-05,
+      "loss": 0.1578,
+      "step": 9962
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 2.3686575052854124e-05,
+      "loss": 0.1135,
+      "step": 9964
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 2.3681289640591966e-05,
+      "loss": 0.1214,
+      "step": 9966
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 2.3676004228329812e-05,
+      "loss": 0.0285,
+      "step": 9968
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 2.3670718816067655e-05,
+      "loss": 0.0551,
+      "step": 9970
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 2.3665433403805498e-05,
+      "loss": 0.0443,
+      "step": 9972
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 2.366014799154334e-05,
+      "loss": 0.0498,
+      "step": 9974
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 2.3654862579281183e-05,
+      "loss": 0.283,
+      "step": 9976
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 2.364957716701903e-05,
+      "loss": 0.0721,
+      "step": 9978
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 2.364429175475687e-05,
+      "loss": 0.0452,
+      "step": 9980
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 2.3639006342494714e-05,
+      "loss": 0.1145,
+      "step": 9982
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 2.3633720930232557e-05,
+      "loss": 0.163,
+      "step": 9984
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.3628435517970403e-05,
+      "loss": 0.0282,
+      "step": 9986
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.3623150105708246e-05,
+      "loss": 0.0182,
+      "step": 9988
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.361786469344609e-05,
+      "loss": 0.0602,
+      "step": 9990
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.3612579281183934e-05,
+      "loss": 0.1005,
+      "step": 9992
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.3607293868921777e-05,
+      "loss": 0.0487,
+      "step": 9994
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 2.3602008456659623e-05,
+      "loss": 0.1124,
+      "step": 9996
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 2.3596723044397466e-05,
+      "loss": 0.0693,
+      "step": 9998
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 2.359143763213531e-05,
+      "loss": 0.05,
+      "step": 10000
+    },
+    {
+      "epoch": 10.57,
+      "eval_cer": 0.02638928469649473,
+      "eval_loss": 0.38202860951423645,
+      "eval_runtime": 128.6979,
+      "eval_samples_per_second": 6.535,
+      "eval_steps_per_second": 0.824,
+      "step": 10000
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 2.358615221987315e-05,
+      "loss": 0.0817,
+      "step": 10002
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 2.3580866807610994e-05,
+      "loss": 0.0701,
+      "step": 10004
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 2.357558139534884e-05,
+      "loss": 0.0567,
+      "step": 10006
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 2.3570295983086683e-05,
+      "loss": 0.0494,
+      "step": 10008
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 2.3565010570824525e-05,
+      "loss": 0.0508,
+      "step": 10010
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 2.3559725158562368e-05,
+      "loss": 0.1178,
+      "step": 10012
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 2.3554439746300214e-05,
+      "loss": 0.0839,
+      "step": 10014
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 2.3549154334038057e-05,
+      "loss": 0.1188,
+      "step": 10016
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 2.35438689217759e-05,
+      "loss": 0.0868,
+      "step": 10018
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 2.3538583509513742e-05,
+      "loss": 0.0215,
+      "step": 10020
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 2.3533298097251585e-05,
+      "loss": 0.0431,
+      "step": 10022
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 2.352801268498943e-05,
+      "loss": 0.0646,
+      "step": 10024
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 2.3522727272727273e-05,
+      "loss": 0.0371,
+      "step": 10026
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 2.3517441860465116e-05,
+      "loss": 0.0433,
+      "step": 10028
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 2.351215644820296e-05,
+      "loss": 0.0562,
+      "step": 10030
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 2.3506871035940805e-05,
+      "loss": 0.0157,
+      "step": 10032
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 2.3501585623678648e-05,
+      "loss": 0.0435,
+      "step": 10034
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 2.349630021141649e-05,
+      "loss": 0.0897,
+      "step": 10036
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 2.3491014799154333e-05,
+      "loss": 0.184,
+      "step": 10038
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 2.348572938689218e-05,
+      "loss": 0.2017,
+      "step": 10040
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 2.348044397463002e-05,
+      "loss": 0.1572,
+      "step": 10042
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 2.3475158562367868e-05,
+      "loss": 0.0323,
+      "step": 10044
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 2.346987315010571e-05,
+      "loss": 0.1133,
+      "step": 10046
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 2.3464587737843553e-05,
+      "loss": 0.1693,
+      "step": 10048
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 2.34593023255814e-05,
+      "loss": 0.1112,
+      "step": 10050
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 2.3454016913319242e-05,
+      "loss": 0.137,
+      "step": 10052
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 2.3448731501057084e-05,
+      "loss": 0.0497,
+      "step": 10054
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 2.3443446088794927e-05,
+      "loss": 0.1043,
+      "step": 10056
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 2.343816067653277e-05,
+      "loss": 0.0784,
+      "step": 10058
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 2.3432875264270616e-05,
+      "loss": 0.0274,
+      "step": 10060
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 2.342758985200846e-05,
+      "loss": 0.0521,
+      "step": 10062
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 2.34223044397463e-05,
+      "loss": 0.0672,
+      "step": 10064
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 2.3417019027484144e-05,
+      "loss": 0.0268,
+      "step": 10066
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 2.341173361522199e-05,
+      "loss": 0.0964,
+      "step": 10068
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 2.3406448202959833e-05,
+      "loss": 0.0146,
+      "step": 10070
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 2.3401162790697675e-05,
+      "loss": 0.052,
+      "step": 10072
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 2.3395877378435518e-05,
+      "loss": 0.0886,
+      "step": 10074
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 2.339059196617336e-05,
+      "loss": 0.0271,
+      "step": 10076
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 2.3385306553911207e-05,
+      "loss": 0.0203,
+      "step": 10078
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 2.338002114164905e-05,
+      "loss": 0.0978,
+      "step": 10080
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 2.3374735729386892e-05,
+      "loss": 0.0224,
+      "step": 10082
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 2.3369450317124735e-05,
+      "loss": 0.0761,
+      "step": 10084
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 2.336416490486258e-05,
+      "loss": 0.0888,
+      "step": 10086
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 2.3358879492600423e-05,
+      "loss": 0.0903,
+      "step": 10088
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 2.3353594080338266e-05,
+      "loss": 0.1393,
+      "step": 10090
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 2.3348308668076112e-05,
+      "loss": 0.0509,
+      "step": 10092
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 2.3343023255813955e-05,
+      "loss": 0.0558,
+      "step": 10094
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 2.3337737843551797e-05,
+      "loss": 0.0986,
+      "step": 10096
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 2.3332452431289643e-05,
+      "loss": 0.0502,
+      "step": 10098
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 2.3327167019027486e-05,
+      "loss": 0.0581,
+      "step": 10100
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 2.332188160676533e-05,
+      "loss": 0.0655,
+      "step": 10102
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 2.3316596194503175e-05,
+      "loss": 0.0805,
+      "step": 10104
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 2.3311310782241018e-05,
+      "loss": 0.0529,
+      "step": 10106
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 2.330602536997886e-05,
+      "loss": 0.1108,
+      "step": 10108
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 2.3300739957716703e-05,
+      "loss": 0.0586,
+      "step": 10110
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 2.3295454545454546e-05,
+      "loss": 0.0512,
+      "step": 10112
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 2.329016913319239e-05,
+      "loss": 0.0463,
+      "step": 10114
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 2.3284883720930234e-05,
+      "loss": 0.0569,
+      "step": 10116
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 2.3279598308668077e-05,
+      "loss": 0.0277,
+      "step": 10118
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 2.327431289640592e-05,
+      "loss": 0.0731,
+      "step": 10120
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 2.3269027484143766e-05,
+      "loss": 0.0775,
+      "step": 10122
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 2.326374207188161e-05,
+      "loss": 0.04,
+      "step": 10124
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 2.325845665961945e-05,
+      "loss": 0.084,
+      "step": 10126
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 2.3253171247357294e-05,
+      "loss": 0.056,
+      "step": 10128
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 2.3247885835095136e-05,
+      "loss": 0.0541,
+      "step": 10130
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 2.3242600422832982e-05,
+      "loss": 0.0266,
+      "step": 10132
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 2.3237315010570825e-05,
+      "loss": 0.0888,
+      "step": 10134
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 2.3232029598308668e-05,
+      "loss": 0.0318,
+      "step": 10136
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 2.322674418604651e-05,
+      "loss": 0.0515,
+      "step": 10138
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 2.3221458773784357e-05,
+      "loss": 0.0594,
+      "step": 10140
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 2.32161733615222e-05,
+      "loss": 0.0935,
+      "step": 10142
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 2.3210887949260042e-05,
+      "loss": 0.0298,
+      "step": 10144
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 2.3205602536997888e-05,
+      "loss": 0.1176,
+      "step": 10146
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 2.320031712473573e-05,
+      "loss": 0.0506,
+      "step": 10148
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 2.3195031712473577e-05,
+      "loss": 0.0595,
+      "step": 10150
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 2.318974630021142e-05,
+      "loss": 0.0413,
+      "step": 10152
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 2.3184460887949262e-05,
+      "loss": 0.102,
+      "step": 10154
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 2.3179175475687105e-05,
+      "loss": 0.0363,
+      "step": 10156
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 2.317389006342495e-05,
+      "loss": 0.0616,
+      "step": 10158
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 2.3168604651162793e-05,
+      "loss": 0.1399,
+      "step": 10160
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 2.3163319238900636e-05,
+      "loss": 0.0462,
+      "step": 10162
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 2.315803382663848e-05,
+      "loss": 0.0246,
+      "step": 10164
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 2.315274841437632e-05,
+      "loss": 0.0721,
+      "step": 10166
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 2.3147463002114167e-05,
+      "loss": 0.0664,
+      "step": 10168
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 2.314217758985201e-05,
+      "loss": 0.0476,
+      "step": 10170
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 2.3136892177589853e-05,
+      "loss": 0.0586,
+      "step": 10172
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 2.3131606765327695e-05,
+      "loss": 0.0424,
+      "step": 10174
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 2.312632135306554e-05,
+      "loss": 0.0312,
+      "step": 10176
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 2.3121035940803384e-05,
+      "loss": 0.0746,
+      "step": 10178
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 2.3115750528541227e-05,
+      "loss": 0.0858,
+      "step": 10180
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 2.311046511627907e-05,
+      "loss": 0.0282,
+      "step": 10182
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 2.3105179704016912e-05,
+      "loss": 0.0502,
+      "step": 10184
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 2.3099894291754758e-05,
+      "loss": 0.0535,
+      "step": 10186
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 2.30946088794926e-05,
+      "loss": 0.0495,
+      "step": 10188
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 2.3089323467230444e-05,
+      "loss": 0.1278,
+      "step": 10190
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 2.3084038054968286e-05,
+      "loss": 0.1004,
+      "step": 10192
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 2.3078752642706132e-05,
+      "loss": 0.0488,
+      "step": 10194
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 2.3073467230443975e-05,
+      "loss": 0.0498,
+      "step": 10196
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 2.3068181818181818e-05,
+      "loss": 0.0595,
+      "step": 10198
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 2.3062896405919664e-05,
+      "loss": 0.0699,
+      "step": 10200
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 2.3057610993657506e-05,
+      "loss": 0.0941,
+      "step": 10202
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 2.3052325581395352e-05,
+      "loss": 0.0409,
+      "step": 10204
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 2.3047040169133195e-05,
+      "loss": 0.0915,
+      "step": 10206
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 2.3041754756871038e-05,
+      "loss": 0.0957,
+      "step": 10208
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 2.303646934460888e-05,
+      "loss": 0.0636,
+      "step": 10210
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 2.3031183932346727e-05,
+      "loss": 0.0596,
+      "step": 10212
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 2.302589852008457e-05,
+      "loss": 0.05,
+      "step": 10214
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 2.3020613107822412e-05,
+      "loss": 0.0246,
+      "step": 10216
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 2.3015327695560255e-05,
+      "loss": 0.142,
+      "step": 10218
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 2.3010042283298097e-05,
+      "loss": 0.0424,
+      "step": 10220
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 2.3004756871035943e-05,
+      "loss": 0.0676,
+      "step": 10222
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 2.2999471458773786e-05,
+      "loss": 0.0551,
+      "step": 10224
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 2.299418604651163e-05,
+      "loss": 0.0686,
+      "step": 10226
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 2.298890063424947e-05,
+      "loss": 0.1185,
+      "step": 10228
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 2.2983615221987317e-05,
+      "loss": 0.0675,
+      "step": 10230
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 2.297832980972516e-05,
+      "loss": 0.0656,
+      "step": 10232
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 2.2973044397463003e-05,
+      "loss": 0.0615,
+      "step": 10234
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 2.2967758985200845e-05,
+      "loss": 0.0193,
+      "step": 10236
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 2.2962473572938688e-05,
+      "loss": 0.1593,
+      "step": 10238
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 2.2957188160676534e-05,
+      "loss": 0.1292,
+      "step": 10240
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 2.2951902748414377e-05,
+      "loss": 0.0303,
+      "step": 10242
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 2.294661733615222e-05,
+      "loss": 0.1374,
+      "step": 10244
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 2.2941331923890062e-05,
+      "loss": 0.0827,
+      "step": 10246
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 2.2936046511627908e-05,
+      "loss": 0.0772,
+      "step": 10248
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 2.293076109936575e-05,
+      "loss": 0.0461,
+      "step": 10250
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 2.2925475687103597e-05,
+      "loss": 0.0469,
+      "step": 10252
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 2.292019027484144e-05,
+      "loss": 0.0446,
+      "step": 10254
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 2.2914904862579282e-05,
+      "loss": 0.0505,
+      "step": 10256
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 2.290961945031713e-05,
+      "loss": 0.017,
+      "step": 10258
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 2.290433403805497e-05,
+      "loss": 0.0467,
+      "step": 10260
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 2.2899048625792814e-05,
+      "loss": 0.136,
+      "step": 10262
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 2.2893763213530656e-05,
+      "loss": 0.0482,
+      "step": 10264
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 2.2888477801268502e-05,
+      "loss": 0.065,
+      "step": 10266
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 2.2883192389006345e-05,
+      "loss": 0.0573,
+      "step": 10268
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 2.2877906976744188e-05,
+      "loss": 0.0658,
+      "step": 10270
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 2.287262156448203e-05,
+      "loss": 0.0434,
+      "step": 10272
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 2.2867336152219873e-05,
+      "loss": 0.1148,
+      "step": 10274
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 2.286205073995772e-05,
+      "loss": 0.0269,
+      "step": 10276
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 2.2856765327695562e-05,
+      "loss": 0.06,
+      "step": 10278
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 2.2851479915433404e-05,
+      "loss": 0.0524,
+      "step": 10280
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 2.2846194503171247e-05,
+      "loss": 0.0526,
+      "step": 10282
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 2.2840909090909093e-05,
+      "loss": 0.0903,
+      "step": 10284
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 2.2835623678646936e-05,
+      "loss": 0.0244,
+      "step": 10286
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 2.283033826638478e-05,
+      "loss": 0.0735,
+      "step": 10288
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 2.282505285412262e-05,
+      "loss": 0.0204,
+      "step": 10290
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 2.2819767441860464e-05,
+      "loss": 0.0459,
+      "step": 10292
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 2.281448202959831e-05,
+      "loss": 0.0396,
+      "step": 10294
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 2.2809196617336153e-05,
+      "loss": 0.0582,
+      "step": 10296
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 2.2803911205073995e-05,
+      "loss": 0.0827,
+      "step": 10298
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 2.2798625792811838e-05,
+      "loss": 0.2031,
+      "step": 10300
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 2.2793340380549684e-05,
+      "loss": 0.0703,
+      "step": 10302
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 2.2788054968287527e-05,
+      "loss": 0.0775,
+      "step": 10304
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 2.2782769556025373e-05,
+      "loss": 0.0718,
+      "step": 10306
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 2.2777484143763215e-05,
+      "loss": 0.1008,
+      "step": 10308
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 2.2772198731501058e-05,
+      "loss": 0.0488,
+      "step": 10310
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 2.2766913319238904e-05,
+      "loss": 0.091,
+      "step": 10312
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 2.2761627906976747e-05,
+      "loss": 0.0164,
+      "step": 10314
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 2.275634249471459e-05,
+      "loss": 0.0562,
+      "step": 10316
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 2.2751057082452432e-05,
+      "loss": 0.1001,
+      "step": 10318
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 2.2745771670190275e-05,
+      "loss": 0.0833,
+      "step": 10320
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 2.274048625792812e-05,
+      "loss": 0.0365,
+      "step": 10322
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 2.2735200845665964e-05,
+      "loss": 0.0233,
+      "step": 10324
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 2.2729915433403806e-05,
+      "loss": 0.0548,
+      "step": 10326
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 2.272463002114165e-05,
+      "loss": 0.082,
+      "step": 10328
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 2.2719344608879495e-05,
+      "loss": 0.0264,
+      "step": 10330
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 2.2714059196617338e-05,
+      "loss": 0.1164,
+      "step": 10332
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 2.270877378435518e-05,
+      "loss": 0.1032,
+      "step": 10334
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 2.2703488372093023e-05,
+      "loss": 0.0402,
+      "step": 10336
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 2.2698202959830866e-05,
+      "loss": 0.0097,
+      "step": 10338
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 2.2692917547568712e-05,
+      "loss": 0.0675,
+      "step": 10340
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 2.2687632135306554e-05,
+      "loss": 0.1228,
+      "step": 10342
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 2.2682346723044397e-05,
+      "loss": 0.1078,
+      "step": 10344
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 2.267706131078224e-05,
+      "loss": 0.0542,
+      "step": 10346
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 2.2671775898520086e-05,
+      "loss": 0.046,
+      "step": 10348
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 2.266649048625793e-05,
+      "loss": 0.0806,
+      "step": 10350
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 2.266120507399577e-05,
+      "loss": 0.0393,
+      "step": 10352
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 2.2655919661733617e-05,
+      "loss": 0.0835,
+      "step": 10354
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 2.265063424947146e-05,
+      "loss": 0.0646,
+      "step": 10356
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 2.2645348837209303e-05,
+      "loss": 0.0517,
+      "step": 10358
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 2.264006342494715e-05,
+      "loss": 0.0558,
+      "step": 10360
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 2.263477801268499e-05,
+      "loss": 0.1552,
+      "step": 10362
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 2.2629492600422834e-05,
+      "loss": 0.0535,
+      "step": 10364
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 2.262420718816068e-05,
+      "loss": 0.0364,
+      "step": 10366
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 2.2618921775898523e-05,
+      "loss": 0.0611,
+      "step": 10368
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 2.2613636363636365e-05,
+      "loss": 0.0374,
+      "step": 10370
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 2.2608350951374208e-05,
+      "loss": 0.0747,
+      "step": 10372
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 2.260306553911205e-05,
+      "loss": 0.0561,
+      "step": 10374
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 2.2597780126849897e-05,
+      "loss": 0.1045,
+      "step": 10376
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 2.259249471458774e-05,
+      "loss": 0.0336,
+      "step": 10378
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 2.2587209302325582e-05,
+      "loss": 0.1093,
+      "step": 10380
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 2.2581923890063425e-05,
+      "loss": 0.1052,
+      "step": 10382
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 2.257663847780127e-05,
+      "loss": 0.0394,
+      "step": 10384
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 2.2571353065539113e-05,
+      "loss": 0.0291,
+      "step": 10386
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 2.2566067653276956e-05,
+      "loss": 0.0769,
+      "step": 10388
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 2.25607822410148e-05,
+      "loss": 0.1155,
+      "step": 10390
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 2.255549682875264e-05,
+      "loss": 0.0829,
+      "step": 10392
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 2.2550211416490488e-05,
+      "loss": 0.0843,
+      "step": 10394
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 2.254492600422833e-05,
+      "loss": 0.0408,
+      "step": 10396
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 2.2539640591966173e-05,
+      "loss": 0.1647,
+      "step": 10398
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 2.2534355179704016e-05,
+      "loss": 0.0599,
+      "step": 10400
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 2.252906976744186e-05,
+      "loss": 0.0424,
+      "step": 10402
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 2.2523784355179704e-05,
+      "loss": 0.09,
+      "step": 10404
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 2.2518498942917547e-05,
+      "loss": 0.0305,
+      "step": 10406
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 2.2513213530655393e-05,
+      "loss": 0.0331,
+      "step": 10408
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 2.2507928118393236e-05,
+      "loss": 0.0907,
+      "step": 10410
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 2.2502642706131082e-05,
+      "loss": 0.0668,
+      "step": 10412
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 2.2497357293868924e-05,
+      "loss": 0.1693,
+      "step": 10414
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 2.2492071881606767e-05,
+      "loss": 0.0328,
+      "step": 10416
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 2.248678646934461e-05,
+      "loss": 0.0536,
+      "step": 10418
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 2.2481501057082456e-05,
+      "loss": 0.1321,
+      "step": 10420
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 2.24762156448203e-05,
+      "loss": 0.0779,
+      "step": 10422
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 2.247093023255814e-05,
+      "loss": 0.0142,
+      "step": 10424
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 2.2465644820295984e-05,
+      "loss": 0.0312,
+      "step": 10426
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 2.2460359408033827e-05,
+      "loss": 0.1316,
+      "step": 10428
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 2.2455073995771673e-05,
+      "loss": 0.0545,
+      "step": 10430
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 2.2449788583509515e-05,
+      "loss": 0.0712,
+      "step": 10432
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 2.2444503171247358e-05,
+      "loss": 0.0942,
+      "step": 10434
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 2.24392177589852e-05,
+      "loss": 0.0524,
+      "step": 10436
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 2.2433932346723047e-05,
+      "loss": 0.0469,
+      "step": 10438
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 2.242864693446089e-05,
+      "loss": 0.0925,
+      "step": 10440
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 2.2423361522198732e-05,
+      "loss": 0.045,
+      "step": 10442
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 2.2418076109936575e-05,
+      "loss": 0.0913,
+      "step": 10444
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 2.2412790697674417e-05,
+      "loss": 0.0557,
+      "step": 10446
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 2.2407505285412263e-05,
+      "loss": 0.036,
+      "step": 10448
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 2.2402219873150106e-05,
+      "loss": 0.0623,
+      "step": 10450
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 2.239693446088795e-05,
+      "loss": 0.0375,
+      "step": 10452
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 2.239164904862579e-05,
+      "loss": 0.1366,
+      "step": 10454
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 2.2386363636363637e-05,
+      "loss": 0.0478,
+      "step": 10456
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 2.238107822410148e-05,
+      "loss": 0.0697,
+      "step": 10458
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 2.2375792811839326e-05,
+      "loss": 0.0432,
+      "step": 10460
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 2.237050739957717e-05,
+      "loss": 0.0327,
+      "step": 10462
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 2.236522198731501e-05,
+      "loss": 0.0793,
+      "step": 10464
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 2.2359936575052858e-05,
+      "loss": 0.0506,
+      "step": 10466
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 2.23546511627907e-05,
+      "loss": 0.0744,
+      "step": 10468
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 2.2349365750528543e-05,
+      "loss": 0.0531,
+      "step": 10470
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 2.2344080338266386e-05,
+      "loss": 0.1243,
+      "step": 10472
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 2.233879492600423e-05,
+      "loss": 0.0289,
+      "step": 10474
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 2.2333509513742074e-05,
+      "loss": 0.0458,
+      "step": 10476
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 2.2328224101479917e-05,
+      "loss": 0.0888,
+      "step": 10478
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 2.232293868921776e-05,
+      "loss": 0.0729,
+      "step": 10480
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 2.2317653276955602e-05,
+      "loss": 0.054,
+      "step": 10482
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 2.231236786469345e-05,
+      "loss": 0.0521,
+      "step": 10484
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 2.230708245243129e-05,
+      "loss": 0.0531,
+      "step": 10486
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 2.2301797040169134e-05,
+      "loss": 0.0542,
+      "step": 10488
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 2.2296511627906976e-05,
+      "loss": 0.0279,
+      "step": 10490
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 2.2291226215644823e-05,
+      "loss": 0.0912,
+      "step": 10492
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 2.2285940803382665e-05,
+      "loss": 0.0324,
+      "step": 10494
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 2.2280655391120508e-05,
+      "loss": 0.0314,
+      "step": 10496
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 2.227536997885835e-05,
+      "loss": 0.1059,
+      "step": 10498
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 2.2270084566596193e-05,
+      "loss": 0.042,
+      "step": 10500
+    },
+    {
+      "epoch": 11.1,
+      "eval_cer": 0.04098033627814192,
+      "eval_loss": 0.8228550553321838,
+      "eval_runtime": 124.2745,
+      "eval_samples_per_second": 6.767,
+      "eval_steps_per_second": 0.853,
+      "step": 10500
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 2.226479915433404e-05,
+      "loss": 0.0726,
+      "step": 10502
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 2.2259513742071882e-05,
+      "loss": 0.0406,
+      "step": 10504
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 2.2254228329809725e-05,
+      "loss": 0.0535,
+      "step": 10506
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 2.2248942917547567e-05,
+      "loss": 0.0596,
+      "step": 10508
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 2.2243657505285413e-05,
+      "loss": 0.0626,
+      "step": 10510
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 2.2238372093023256e-05,
+      "loss": 0.0582,
+      "step": 10512
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 2.2233086680761102e-05,
+      "loss": 0.1281,
+      "step": 10514
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 2.2227801268498945e-05,
+      "loss": 0.0823,
+      "step": 10516
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 2.2222515856236787e-05,
+      "loss": 0.0629,
+      "step": 10518
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 2.2217230443974633e-05,
+      "loss": 0.0487,
+      "step": 10520
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 2.2211945031712476e-05,
+      "loss": 0.0708,
+      "step": 10522
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 2.220665961945032e-05,
+      "loss": 0.073,
+      "step": 10524
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 2.220137420718816e-05,
+      "loss": 0.1439,
+      "step": 10526
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 2.2196088794926008e-05,
+      "loss": 0.0255,
+      "step": 10528
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 2.219080338266385e-05,
+      "loss": 0.0569,
+      "step": 10530
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 2.2185517970401693e-05,
+      "loss": 0.0554,
+      "step": 10532
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 2.2180232558139536e-05,
+      "loss": 0.0647,
+      "step": 10534
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 2.2174947145877378e-05,
+      "loss": 0.0534,
+      "step": 10536
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 2.2169661733615224e-05,
+      "loss": 0.0495,
+      "step": 10538
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 2.2164376321353067e-05,
+      "loss": 0.128,
+      "step": 10540
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 2.215909090909091e-05,
+      "loss": 0.1084,
+      "step": 10542
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 2.2153805496828752e-05,
+      "loss": 0.0622,
+      "step": 10544
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 2.21485200845666e-05,
+      "loss": 0.0289,
+      "step": 10546
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 2.214323467230444e-05,
+      "loss": 0.0682,
+      "step": 10548
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 2.2137949260042284e-05,
+      "loss": 0.0404,
+      "step": 10550
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 2.2132663847780126e-05,
+      "loss": 0.0743,
+      "step": 10552
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 2.212737843551797e-05,
+      "loss": 0.0205,
+      "step": 10554
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 2.2122093023255815e-05,
+      "loss": 0.0295,
+      "step": 10556
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 2.2116807610993658e-05,
+      "loss": 0.0505,
+      "step": 10558
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 2.21115221987315e-05,
+      "loss": 0.0379,
+      "step": 10560
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 2.2106236786469346e-05,
+      "loss": 0.0194,
+      "step": 10562
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 2.210095137420719e-05,
+      "loss": 0.0343,
+      "step": 10564
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 2.2095665961945032e-05,
+      "loss": 0.077,
+      "step": 10566
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 2.2090380549682878e-05,
+      "loss": 0.0756,
+      "step": 10568
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 2.208509513742072e-05,
+      "loss": 0.049,
+      "step": 10570
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 2.2079809725158563e-05,
+      "loss": 0.0582,
+      "step": 10572
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 2.207452431289641e-05,
+      "loss": 0.0347,
+      "step": 10574
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 2.2069238900634252e-05,
+      "loss": 0.0778,
+      "step": 10576
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 2.2063953488372095e-05,
+      "loss": 0.0305,
+      "step": 10578
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 2.2058668076109937e-05,
+      "loss": 0.098,
+      "step": 10580
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 2.2053382663847783e-05,
+      "loss": 0.0289,
+      "step": 10582
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 2.2048097251585626e-05,
+      "loss": 0.0603,
+      "step": 10584
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 2.204281183932347e-05,
+      "loss": 0.0724,
+      "step": 10586
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 2.203752642706131e-05,
+      "loss": 0.0864,
+      "step": 10588
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 2.2032241014799154e-05,
+      "loss": 0.1251,
+      "step": 10590
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 2.2026955602537e-05,
+      "loss": 0.0161,
+      "step": 10592
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 2.2021670190274843e-05,
+      "loss": 0.0633,
+      "step": 10594
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 2.2016384778012685e-05,
+      "loss": 0.0709,
+      "step": 10596
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 2.2011099365750528e-05,
+      "loss": 0.0971,
+      "step": 10598
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 2.2005813953488374e-05,
+      "loss": 0.0582,
+      "step": 10600
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 2.2000528541226217e-05,
+      "loss": 0.0354,
+      "step": 10602
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 2.199524312896406e-05,
+      "loss": 0.0754,
+      "step": 10604
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 2.1989957716701902e-05,
+      "loss": 0.0231,
+      "step": 10606
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 2.1984672304439745e-05,
+      "loss": 0.0977,
+      "step": 10608
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 2.197938689217759e-05,
+      "loss": 0.0853,
+      "step": 10610
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 2.1974101479915434e-05,
+      "loss": 0.0304,
+      "step": 10612
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 2.1968816067653276e-05,
+      "loss": 0.0818,
+      "step": 10614
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 2.1963530655391122e-05,
+      "loss": 0.1034,
+      "step": 10616
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 2.1958245243128965e-05,
+      "loss": 0.0743,
+      "step": 10618
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 2.195295983086681e-05,
+      "loss": 0.0717,
+      "step": 10620
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 2.1947674418604654e-05,
+      "loss": 0.0263,
+      "step": 10622
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 2.1942389006342496e-05,
+      "loss": 0.065,
+      "step": 10624
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 2.193710359408034e-05,
+      "loss": 0.1063,
+      "step": 10626
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 2.1931818181818185e-05,
+      "loss": 0.0699,
+      "step": 10628
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 2.1926532769556028e-05,
+      "loss": 0.0324,
+      "step": 10630
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 2.192124735729387e-05,
+      "loss": 0.0283,
+      "step": 10632
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 2.1915961945031713e-05,
+      "loss": 0.0296,
+      "step": 10634
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 2.191067653276956e-05,
+      "loss": 0.0686,
+      "step": 10636
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 2.1905391120507402e-05,
+      "loss": 0.0577,
+      "step": 10638
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 2.1900105708245245e-05,
+      "loss": 0.0288,
+      "step": 10640
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 2.1894820295983087e-05,
+      "loss": 0.1001,
+      "step": 10642
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 2.188953488372093e-05,
+      "loss": 0.0518,
+      "step": 10644
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 2.1884249471458776e-05,
+      "loss": 0.0222,
+      "step": 10646
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 2.187896405919662e-05,
+      "loss": 0.0773,
+      "step": 10648
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 2.187367864693446e-05,
+      "loss": 0.0045,
+      "step": 10650
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 2.1868393234672304e-05,
+      "loss": 0.0173,
+      "step": 10652
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 2.1863107822410147e-05,
+      "loss": 0.0316,
+      "step": 10654
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 2.1857822410147993e-05,
+      "loss": 0.1172,
+      "step": 10656
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 2.1852536997885835e-05,
+      "loss": 0.0507,
+      "step": 10658
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 2.1847251585623678e-05,
+      "loss": 0.0951,
+      "step": 10660
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 2.184196617336152e-05,
+      "loss": 0.0199,
+      "step": 10662
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 2.1836680761099367e-05,
+      "loss": 0.0583,
+      "step": 10664
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 2.183139534883721e-05,
+      "loss": 0.041,
+      "step": 10666
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 2.1826109936575052e-05,
+      "loss": 0.0624,
+      "step": 10668
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 2.1820824524312898e-05,
+      "loss": 0.0803,
+      "step": 10670
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 2.181553911205074e-05,
+      "loss": 0.1347,
+      "step": 10672
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 2.1810253699788587e-05,
+      "loss": 0.0725,
+      "step": 10674
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 2.180496828752643e-05,
+      "loss": 0.0355,
+      "step": 10676
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 2.1799682875264272e-05,
+      "loss": 0.0703,
+      "step": 10678
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 2.1794397463002115e-05,
+      "loss": 0.0373,
+      "step": 10680
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 2.178911205073996e-05,
+      "loss": 0.0535,
+      "step": 10682
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 2.1783826638477804e-05,
+      "loss": 0.0635,
+      "step": 10684
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 2.1778541226215646e-05,
+      "loss": 0.0548,
+      "step": 10686
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 2.177325581395349e-05,
+      "loss": 0.0797,
+      "step": 10688
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 2.176797040169133e-05,
+      "loss": 0.1646,
+      "step": 10690
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 2.1762684989429178e-05,
+      "loss": 0.0078,
+      "step": 10692
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 2.175739957716702e-05,
+      "loss": 0.039,
+      "step": 10694
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 2.1752114164904863e-05,
+      "loss": 0.0683,
+      "step": 10696
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 2.1746828752642706e-05,
+      "loss": 0.0309,
+      "step": 10698
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 2.1741543340380552e-05,
+      "loss": 0.0709,
+      "step": 10700
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 2.1736257928118394e-05,
+      "loss": 0.0945,
+      "step": 10702
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 2.1730972515856237e-05,
+      "loss": 0.0391,
+      "step": 10704
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 2.172568710359408e-05,
+      "loss": 0.0402,
+      "step": 10706
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 2.1720401691331922e-05,
+      "loss": 0.1571,
+      "step": 10708
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 2.171511627906977e-05,
+      "loss": 0.0544,
+      "step": 10710
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 2.170983086680761e-05,
+      "loss": 0.0227,
+      "step": 10712
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 2.1704545454545454e-05,
+      "loss": 0.0334,
+      "step": 10714
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 2.1699260042283297e-05,
+      "loss": 0.041,
+      "step": 10716
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 2.1693974630021143e-05,
+      "loss": 0.108,
+      "step": 10718
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 2.1688689217758985e-05,
+      "loss": 0.0199,
+      "step": 10720
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 2.168340380549683e-05,
+      "loss": 0.0907,
+      "step": 10722
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.1678118393234674e-05,
+      "loss": 0.0606,
+      "step": 10724
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.1672832980972517e-05,
+      "loss": 0.0314,
+      "step": 10726
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.1667547568710363e-05,
+      "loss": 0.0559,
+      "step": 10728
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.1662262156448205e-05,
+      "loss": 0.0155,
+      "step": 10730
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.1656976744186048e-05,
+      "loss": 0.068,
+      "step": 10732
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 2.165169133192389e-05,
+      "loss": 0.037,
+      "step": 10734
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 2.1646405919661737e-05,
+      "loss": 0.0204,
+      "step": 10736
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 2.164112050739958e-05,
+      "loss": 0.0349,
+      "step": 10738
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 2.1635835095137422e-05,
+      "loss": 0.0516,
+      "step": 10740
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.1630549682875265e-05,
+      "loss": 0.0482,
+      "step": 10742
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.1625264270613107e-05,
+      "loss": 0.022,
+      "step": 10744
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.1619978858350954e-05,
+      "loss": 0.0717,
+      "step": 10746
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.1614693446088796e-05,
+      "loss": 0.0435,
+      "step": 10748
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.160940803382664e-05,
+      "loss": 0.0449,
+      "step": 10750
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 2.160412262156448e-05,
+      "loss": 0.0319,
+      "step": 10752
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 2.1598837209302328e-05,
+      "loss": 0.0255,
+      "step": 10754
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 2.159355179704017e-05,
+      "loss": 0.1095,
+      "step": 10756
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 2.1588266384778013e-05,
+      "loss": 0.0939,
+      "step": 10758
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 2.1582980972515856e-05,
+      "loss": 0.055,
+      "step": 10760
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.15776955602537e-05,
+      "loss": 0.1075,
+      "step": 10762
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.1572410147991544e-05,
+      "loss": 0.0374,
+      "step": 10764
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.1567124735729387e-05,
+      "loss": 0.0753,
+      "step": 10766
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.156183932346723e-05,
+      "loss": 0.1265,
+      "step": 10768
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.1556553911205072e-05,
+      "loss": 0.0542,
+      "step": 10770
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 2.155126849894292e-05,
+      "loss": 0.0173,
+      "step": 10772
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 2.154598308668076e-05,
+      "loss": 0.0681,
+      "step": 10774
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 2.1540697674418607e-05,
+      "loss": 0.0296,
+      "step": 10776
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 2.153541226215645e-05,
+      "loss": 0.0658,
+      "step": 10778
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 2.1530126849894293e-05,
+      "loss": 0.1054,
+      "step": 10780
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 2.152484143763214e-05,
+      "loss": 0.1244,
+      "step": 10782
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 2.151955602536998e-05,
+      "loss": 0.026,
+      "step": 10784
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 2.1514270613107824e-05,
+      "loss": 0.0251,
+      "step": 10786
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 2.1508985200845667e-05,
+      "loss": 0.0284,
+      "step": 10788
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 2.1503699788583513e-05,
+      "loss": 0.0288,
+      "step": 10790
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 2.1498414376321355e-05,
+      "loss": 0.0914,
+      "step": 10792
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 2.1493128964059198e-05,
+      "loss": 0.0433,
+      "step": 10794
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 2.148784355179704e-05,
+      "loss": 0.0501,
+      "step": 10796
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 2.1482558139534883e-05,
+      "loss": 0.0046,
+      "step": 10798
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.147727272727273e-05,
+      "loss": 0.0498,
+      "step": 10800
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.1471987315010572e-05,
+      "loss": 0.0928,
+      "step": 10802
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.1466701902748415e-05,
+      "loss": 0.0529,
+      "step": 10804
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.1461416490486257e-05,
+      "loss": 0.0392,
+      "step": 10806
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.1456131078224103e-05,
+      "loss": 0.0596,
+      "step": 10808
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.1450845665961946e-05,
+      "loss": 0.0378,
+      "step": 10810
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.144556025369979e-05,
+      "loss": 0.0409,
+      "step": 10812
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.144027484143763e-05,
+      "loss": 0.0267,
+      "step": 10814
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.1434989429175474e-05,
+      "loss": 0.0452,
+      "step": 10816
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.142970401691332e-05,
+      "loss": 0.0531,
+      "step": 10818
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.1424418604651163e-05,
+      "loss": 0.0386,
+      "step": 10820
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.1419133192389006e-05,
+      "loss": 0.063,
+      "step": 10822
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.141384778012685e-05,
+      "loss": 0.0325,
+      "step": 10824
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.1408562367864694e-05,
+      "loss": 0.0868,
+      "step": 10826
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 2.1403276955602537e-05,
+      "loss": 0.0804,
+      "step": 10828
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 2.1397991543340383e-05,
+      "loss": 0.0758,
+      "step": 10830
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 2.1392706131078226e-05,
+      "loss": 0.1376,
+      "step": 10832
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 2.138742071881607e-05,
+      "loss": 0.0455,
+      "step": 10834
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 2.1382135306553914e-05,
+      "loss": 0.0557,
+      "step": 10836
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 2.1376849894291757e-05,
+      "loss": 0.004,
+      "step": 10838
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 2.13715644820296e-05,
+      "loss": 0.0539,
+      "step": 10840
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 2.1366279069767442e-05,
+      "loss": 0.1415,
+      "step": 10842
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 2.136099365750529e-05,
+      "loss": 0.0817,
+      "step": 10844
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.135570824524313e-05,
+      "loss": 0.0372,
+      "step": 10846
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.1350422832980974e-05,
+      "loss": 0.0595,
+      "step": 10848
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.1345137420718817e-05,
+      "loss": 0.04,
+      "step": 10850
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.133985200845666e-05,
+      "loss": 0.0356,
+      "step": 10852
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.1334566596194505e-05,
+      "loss": 0.0702,
+      "step": 10854
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 2.1329281183932348e-05,
+      "loss": 0.0667,
+      "step": 10856
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 2.132399577167019e-05,
+      "loss": 0.0514,
+      "step": 10858
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 2.1318710359408033e-05,
+      "loss": 0.1231,
+      "step": 10860
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 2.131342494714588e-05,
+      "loss": 0.0405,
+      "step": 10862
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 2.1308139534883722e-05,
+      "loss": 0.0731,
+      "step": 10864
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.1302854122621565e-05,
+      "loss": 0.0312,
+      "step": 10866
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.1297568710359407e-05,
+      "loss": 0.0644,
+      "step": 10868
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.129228329809725e-05,
+      "loss": 0.0827,
+      "step": 10870
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.1286997885835096e-05,
+      "loss": 0.0253,
+      "step": 10872
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.128171247357294e-05,
+      "loss": 0.0523,
+      "step": 10874
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 2.127642706131078e-05,
+      "loss": 0.0856,
+      "step": 10876
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 2.1271141649048627e-05,
+      "loss": 0.065,
+      "step": 10878
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 2.126585623678647e-05,
+      "loss": 0.0138,
+      "step": 10880
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 2.1260570824524316e-05,
+      "loss": 0.0285,
+      "step": 10882
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.125528541226216e-05,
+      "loss": 0.0584,
+      "step": 10884
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.125e-05,
+      "loss": 0.0262,
+      "step": 10886
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.1244714587737844e-05,
+      "loss": 0.0238,
+      "step": 10888
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.123942917547569e-05,
+      "loss": 0.0802,
+      "step": 10890
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.1234143763213533e-05,
+      "loss": 0.0672,
+      "step": 10892
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.1228858350951376e-05,
+      "loss": 0.0377,
+      "step": 10894
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.1223572938689218e-05,
+      "loss": 0.028,
+      "step": 10896
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.1218287526427064e-05,
+      "loss": 0.0638,
+      "step": 10898
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.1213002114164907e-05,
+      "loss": 0.0875,
+      "step": 10900
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.120771670190275e-05,
+      "loss": 0.0416,
+      "step": 10902
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.1202431289640592e-05,
+      "loss": 0.006,
+      "step": 10904
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.1197145877378435e-05,
+      "loss": 0.0266,
+      "step": 10906
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.119186046511628e-05,
+      "loss": 0.0754,
+      "step": 10908
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.1186575052854124e-05,
+      "loss": 0.0288,
+      "step": 10910
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.1181289640591966e-05,
+      "loss": 0.0511,
+      "step": 10912
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 2.117600422832981e-05,
+      "loss": 0.0647,
+      "step": 10914
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 2.1170718816067655e-05,
+      "loss": 0.0274,
+      "step": 10916
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 2.1165433403805498e-05,
+      "loss": 0.0344,
+      "step": 10918
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 2.116014799154334e-05,
+      "loss": 0.0904,
+      "step": 10920
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 2.1154862579281183e-05,
+      "loss": 0.0463,
+      "step": 10922
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 2.1149577167019026e-05,
+      "loss": 0.139,
+      "step": 10924
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 2.1144291754756872e-05,
+      "loss": 0.0338,
+      "step": 10926
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 2.1139006342494715e-05,
+      "loss": 0.0309,
+      "step": 10928
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 2.113372093023256e-05,
+      "loss": 0.0738,
+      "step": 10930
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 2.1128435517970403e-05,
+      "loss": 0.0942,
+      "step": 10932
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 2.1123150105708246e-05,
+      "loss": 0.0721,
+      "step": 10934
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 2.1117864693446092e-05,
+      "loss": 0.1223,
+      "step": 10936
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 2.1112579281183935e-05,
+      "loss": 0.0341,
+      "step": 10938
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 2.1107293868921777e-05,
+      "loss": 0.0784,
+      "step": 10940
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 2.110200845665962e-05,
+      "loss": 0.0329,
+      "step": 10942
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 2.1096723044397466e-05,
+      "loss": 0.0208,
+      "step": 10944
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 2.109143763213531e-05,
+      "loss": 0.0544,
+      "step": 10946
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 2.108615221987315e-05,
+      "loss": 0.0241,
+      "step": 10948
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 2.1080866807610994e-05,
+      "loss": 0.0443,
+      "step": 10950
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 2.107558139534884e-05,
+      "loss": 0.0468,
+      "step": 10952
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 2.1070295983086683e-05,
+      "loss": 0.0101,
+      "step": 10954
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 2.1065010570824526e-05,
+      "loss": 0.132,
+      "step": 10956
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 2.1059725158562368e-05,
+      "loss": 0.0627,
+      "step": 10958
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 2.105443974630021e-05,
+      "loss": 0.033,
+      "step": 10960
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 2.1049154334038057e-05,
+      "loss": 0.0594,
+      "step": 10962
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 2.10438689217759e-05,
+      "loss": 0.0433,
+      "step": 10964
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 2.1038583509513742e-05,
+      "loss": 0.0288,
+      "step": 10966
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 2.1033298097251585e-05,
+      "loss": 0.0383,
+      "step": 10968
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 2.102801268498943e-05,
+      "loss": 0.118,
+      "step": 10970
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 2.1022727272727274e-05,
+      "loss": 0.0767,
+      "step": 10972
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 2.1017441860465116e-05,
+      "loss": 0.1311,
+      "step": 10974
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 2.101215644820296e-05,
+      "loss": 0.0383,
+      "step": 10976
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 2.10068710359408e-05,
+      "loss": 0.0315,
+      "step": 10978
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 2.1001585623678648e-05,
+      "loss": 0.1211,
+      "step": 10980
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 2.099630021141649e-05,
+      "loss": 0.1321,
+      "step": 10982
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 2.0991014799154336e-05,
+      "loss": 0.0749,
+      "step": 10984
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 2.098572938689218e-05,
+      "loss": 0.0581,
+      "step": 10986
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 2.0980443974630022e-05,
+      "loss": 0.0501,
+      "step": 10988
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 2.0975158562367868e-05,
+      "loss": 0.1586,
+      "step": 10990
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 2.096987315010571e-05,
+      "loss": 0.1585,
+      "step": 10992
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 2.0964587737843553e-05,
+      "loss": 0.016,
+      "step": 10994
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 2.0959302325581396e-05,
+      "loss": 0.0378,
+      "step": 10996
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 2.0954016913319242e-05,
+      "loss": 0.0257,
+      "step": 10998
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 2.0948731501057085e-05,
+      "loss": 0.0754,
+      "step": 11000
+    },
+    {
+      "epoch": 11.63,
+      "eval_cer": 0.027586206896551724,
+      "eval_loss": 0.871033787727356,
+      "eval_runtime": 128.8039,
+      "eval_samples_per_second": 6.529,
+      "eval_steps_per_second": 0.823,
+      "step": 11000
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 2.0943446088794927e-05,
+      "loss": 0.0617,
+      "step": 11002
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 2.093816067653277e-05,
+      "loss": 0.0261,
+      "step": 11004
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 2.0932875264270613e-05,
+      "loss": 0.0115,
+      "step": 11006
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 2.092758985200846e-05,
+      "loss": 0.0594,
+      "step": 11008
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 2.09223044397463e-05,
+      "loss": 0.0841,
+      "step": 11010
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 2.0917019027484144e-05,
+      "loss": 0.0745,
+      "step": 11012
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 2.0911733615221987e-05,
+      "loss": 0.0111,
+      "step": 11014
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 2.0906448202959833e-05,
+      "loss": 0.0268,
+      "step": 11016
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 2.0901162790697675e-05,
+      "loss": 0.0558,
+      "step": 11018
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 2.0895877378435518e-05,
+      "loss": 0.0958,
+      "step": 11020
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 2.089059196617336e-05,
+      "loss": 0.0142,
+      "step": 11022
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 2.0885306553911203e-05,
+      "loss": 0.013,
+      "step": 11024
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 2.088002114164905e-05,
+      "loss": 0.0168,
+      "step": 11026
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 2.0874735729386892e-05,
+      "loss": 0.0582,
+      "step": 11028
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 2.0869450317124735e-05,
+      "loss": 0.1415,
+      "step": 11030
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 2.086416490486258e-05,
+      "loss": 0.0226,
+      "step": 11032
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 2.0858879492600424e-05,
+      "loss": 0.0594,
+      "step": 11034
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 2.0853594080338266e-05,
+      "loss": 0.0904,
+      "step": 11036
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 2.0848308668076112e-05,
+      "loss": 0.0244,
+      "step": 11038
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 2.0843023255813955e-05,
+      "loss": 0.0395,
+      "step": 11040
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 2.0837737843551798e-05,
+      "loss": 0.0772,
+      "step": 11042
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 2.0832452431289644e-05,
+      "loss": 0.0739,
+      "step": 11044
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 2.0827167019027486e-05,
+      "loss": 0.0474,
+      "step": 11046
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 2.082188160676533e-05,
+      "loss": 0.0723,
+      "step": 11048
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 2.0816596194503172e-05,
+      "loss": 0.0151,
+      "step": 11050
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 2.0811310782241018e-05,
+      "loss": 0.024,
+      "step": 11052
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 2.080602536997886e-05,
+      "loss": 0.0285,
+      "step": 11054
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 2.0800739957716703e-05,
+      "loss": 0.146,
+      "step": 11056
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 2.0795454545454546e-05,
+      "loss": 0.056,
+      "step": 11058
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 2.079016913319239e-05,
+      "loss": 0.0106,
+      "step": 11060
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 2.0784883720930235e-05,
+      "loss": 0.0663,
+      "step": 11062
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 2.0779598308668077e-05,
+      "loss": 0.1001,
+      "step": 11064
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 2.077431289640592e-05,
+      "loss": 0.0979,
+      "step": 11066
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 2.0769027484143763e-05,
+      "loss": 0.0581,
+      "step": 11068
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 2.076374207188161e-05,
+      "loss": 0.0471,
+      "step": 11070
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 2.075845665961945e-05,
+      "loss": 0.0102,
+      "step": 11072
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 2.0753171247357294e-05,
+      "loss": 0.0279,
+      "step": 11074
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 2.0747885835095137e-05,
+      "loss": 0.0497,
+      "step": 11076
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 2.074260042283298e-05,
+      "loss": 0.0217,
+      "step": 11078
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 2.0737315010570825e-05,
+      "loss": 0.0711,
+      "step": 11080
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 2.0732029598308668e-05,
+      "loss": 0.0176,
+      "step": 11082
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 2.072674418604651e-05,
+      "loss": 0.116,
+      "step": 11084
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 2.0721458773784357e-05,
+      "loss": 0.0255,
+      "step": 11086
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 2.07161733615222e-05,
+      "loss": 0.0766,
+      "step": 11088
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 2.0710887949260045e-05,
+      "loss": 0.0453,
+      "step": 11090
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 2.0705602536997888e-05,
+      "loss": 0.0803,
+      "step": 11092
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 2.070031712473573e-05,
+      "loss": 0.0871,
+      "step": 11094
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 2.0695031712473573e-05,
+      "loss": 0.0988,
+      "step": 11096
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 2.068974630021142e-05,
+      "loss": 0.0257,
+      "step": 11098
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 2.0684460887949262e-05,
+      "loss": 0.1108,
+      "step": 11100
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 2.0679175475687105e-05,
+      "loss": 0.0522,
+      "step": 11102
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 2.0673890063424948e-05,
+      "loss": 0.0438,
+      "step": 11104
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 2.0668604651162794e-05,
+      "loss": 0.0334,
+      "step": 11106
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 2.0663319238900636e-05,
+      "loss": 0.0322,
+      "step": 11108
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 2.065803382663848e-05,
+      "loss": 0.0284,
+      "step": 11110
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 2.065274841437632e-05,
+      "loss": 0.0106,
+      "step": 11112
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 2.0647463002114164e-05,
+      "loss": 0.0164,
+      "step": 11114
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 2.064217758985201e-05,
+      "loss": 0.0614,
+      "step": 11116
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 2.0636892177589853e-05,
+      "loss": 0.0688,
+      "step": 11118
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 2.0631606765327696e-05,
+      "loss": 0.0633,
+      "step": 11120
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 2.062632135306554e-05,
+      "loss": 0.1126,
+      "step": 11122
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 2.0621035940803384e-05,
+      "loss": 0.004,
+      "step": 11124
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 2.0615750528541227e-05,
+      "loss": 0.0379,
+      "step": 11126
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 2.061046511627907e-05,
+      "loss": 0.085,
+      "step": 11128
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 2.0605179704016912e-05,
+      "loss": 0.0781,
+      "step": 11130
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 2.0599894291754755e-05,
+      "loss": 0.0755,
+      "step": 11132
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 2.05946088794926e-05,
+      "loss": 0.0244,
+      "step": 11134
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 2.0589323467230444e-05,
+      "loss": 0.0323,
+      "step": 11136
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 2.0584038054968287e-05,
+      "loss": 0.0872,
+      "step": 11138
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 2.0578752642706133e-05,
+      "loss": 0.0902,
+      "step": 11140
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 2.0573467230443975e-05,
+      "loss": 0.0517,
+      "step": 11142
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 2.056818181818182e-05,
+      "loss": 0.1477,
+      "step": 11144
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 2.0562896405919664e-05,
+      "loss": 0.0487,
+      "step": 11146
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 2.0557610993657507e-05,
+      "loss": 0.0588,
+      "step": 11148
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 2.055232558139535e-05,
+      "loss": 0.0118,
+      "step": 11150
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 2.0547040169133195e-05,
+      "loss": 0.0789,
+      "step": 11152
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 2.0541754756871038e-05,
+      "loss": 0.0406,
+      "step": 11154
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 2.053646934460888e-05,
+      "loss": 0.0636,
+      "step": 11156
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 2.0531183932346723e-05,
+      "loss": 0.0327,
+      "step": 11158
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 2.052589852008457e-05,
+      "loss": 0.0546,
+      "step": 11160
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 2.0520613107822412e-05,
+      "loss": 0.0355,
+      "step": 11162
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 2.0515327695560255e-05,
+      "loss": 0.053,
+      "step": 11164
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 2.0510042283298097e-05,
+      "loss": 0.0439,
+      "step": 11166
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 2.050475687103594e-05,
+      "loss": 0.0155,
+      "step": 11168
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 2.0499471458773786e-05,
+      "loss": 0.0532,
+      "step": 11170
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 2.049418604651163e-05,
+      "loss": 0.0108,
+      "step": 11172
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 2.048890063424947e-05,
+      "loss": 0.0571,
+      "step": 11174
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 2.0483615221987314e-05,
+      "loss": 0.0961,
+      "step": 11176
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 2.047832980972516e-05,
+      "loss": 0.0407,
+      "step": 11178
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 2.0473044397463003e-05,
+      "loss": 0.0304,
+      "step": 11180
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 2.0467758985200846e-05,
+      "loss": 0.0345,
+      "step": 11182
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 2.0462473572938688e-05,
+      "loss": 0.019,
+      "step": 11184
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 2.045718816067653e-05,
+      "loss": 0.071,
+      "step": 11186
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 2.0451902748414377e-05,
+      "loss": 0.2046,
+      "step": 11188
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 2.044661733615222e-05,
+      "loss": 0.0732,
+      "step": 11190
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 2.0441331923890066e-05,
+      "loss": 0.0416,
+      "step": 11192
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 2.043604651162791e-05,
+      "loss": 0.031,
+      "step": 11194
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 2.043076109936575e-05,
+      "loss": 0.0405,
+      "step": 11196
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 2.0425475687103597e-05,
+      "loss": 0.0323,
+      "step": 11198
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 2.042019027484144e-05,
+      "loss": 0.063,
+      "step": 11200
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 2.0414904862579282e-05,
+      "loss": 0.1026,
+      "step": 11202
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 2.0409619450317125e-05,
+      "loss": 0.0508,
+      "step": 11204
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 2.040433403805497e-05,
+      "loss": 0.0286,
+      "step": 11206
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 2.0399048625792814e-05,
+      "loss": 0.0337,
+      "step": 11208
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 2.0393763213530657e-05,
+      "loss": 0.061,
+      "step": 11210
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 2.03884778012685e-05,
+      "loss": 0.0663,
+      "step": 11212
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 2.0383192389006345e-05,
+      "loss": 0.091,
+      "step": 11214
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 2.0377906976744188e-05,
+      "loss": 0.0043,
+      "step": 11216
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 2.037262156448203e-05,
+      "loss": 0.0518,
+      "step": 11218
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 2.0367336152219873e-05,
+      "loss": 0.0485,
+      "step": 11220
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 2.0362050739957716e-05,
+      "loss": 0.0537,
+      "step": 11222
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 2.0356765327695562e-05,
+      "loss": 0.0262,
+      "step": 11224
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 2.0351479915433405e-05,
+      "loss": 0.1172,
+      "step": 11226
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 2.0346194503171247e-05,
+      "loss": 0.0423,
+      "step": 11228
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 2.034355179704017e-05,
+      "loss": 0.0751,
+      "step": 11230
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 2.0338266384778013e-05,
+      "loss": 0.0498,
+      "step": 11232
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 2.033298097251586e-05,
+      "loss": 0.0332,
+      "step": 11234
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 2.0327695560253702e-05,
+      "loss": 0.0439,
+      "step": 11236
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 2.0322410147991544e-05,
+      "loss": 0.0399,
+      "step": 11238
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 2.0317124735729387e-05,
+      "loss": 0.0337,
+      "step": 11240
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 2.0311839323467233e-05,
+      "loss": 0.0582,
+      "step": 11242
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 2.0306553911205076e-05,
+      "loss": 0.1019,
+      "step": 11244
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 2.030126849894292e-05,
+      "loss": 0.0754,
+      "step": 11246
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 2.029598308668076e-05,
+      "loss": 0.0471,
+      "step": 11248
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 2.0290697674418604e-05,
+      "loss": 0.0311,
+      "step": 11250
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 2.028541226215645e-05,
+      "loss": 0.0573,
+      "step": 11252
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 2.0280126849894293e-05,
+      "loss": 0.0474,
+      "step": 11254
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 2.0274841437632135e-05,
+      "loss": 0.0533,
+      "step": 11256
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 2.0269556025369978e-05,
+      "loss": 0.1002,
+      "step": 11258
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 2.0264270613107824e-05,
+      "loss": 0.045,
+      "step": 11260
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 2.0258985200845667e-05,
+      "loss": 0.0355,
+      "step": 11262
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 2.025369978858351e-05,
+      "loss": 0.0453,
+      "step": 11264
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 2.0248414376321352e-05,
+      "loss": 0.0356,
+      "step": 11266
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 2.0243128964059198e-05,
+      "loss": 0.0951,
+      "step": 11268
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 2.023784355179704e-05,
+      "loss": 0.0593,
+      "step": 11270
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 2.0232558139534883e-05,
+      "loss": 0.046,
+      "step": 11272
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 2.022727272727273e-05,
+      "loss": 0.0185,
+      "step": 11274
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 2.0221987315010572e-05,
+      "loss": 0.041,
+      "step": 11276
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 2.0216701902748418e-05,
+      "loss": 0.0857,
+      "step": 11278
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 2.021141649048626e-05,
+      "loss": 0.0111,
+      "step": 11280
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 2.0206131078224104e-05,
+      "loss": 0.0913,
+      "step": 11282
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 2.0200845665961946e-05,
+      "loss": 0.0448,
+      "step": 11284
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 2.019556025369979e-05,
+      "loss": 0.0471,
+      "step": 11286
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 2.0190274841437635e-05,
+      "loss": 0.0396,
+      "step": 11288
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 2.0184989429175478e-05,
+      "loss": 0.0769,
+      "step": 11290
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 2.017970401691332e-05,
+      "loss": 0.053,
+      "step": 11292
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 2.0174418604651163e-05,
+      "loss": 0.1192,
+      "step": 11294
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 2.016913319238901e-05,
+      "loss": 0.0102,
+      "step": 11296
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 2.0163847780126852e-05,
+      "loss": 0.0832,
+      "step": 11298
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 2.0158562367864694e-05,
+      "loss": 0.0162,
+      "step": 11300
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 2.0153276955602537e-05,
+      "loss": 0.0205,
+      "step": 11302
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 2.014799154334038e-05,
+      "loss": 0.016,
+      "step": 11304
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 2.0142706131078226e-05,
+      "loss": 0.0455,
+      "step": 11306
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 2.013742071881607e-05,
+      "loss": 0.0433,
+      "step": 11308
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 2.013213530655391e-05,
+      "loss": 0.0466,
+      "step": 11310
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 2.0126849894291754e-05,
+      "loss": 0.0151,
+      "step": 11312
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 2.01215644820296e-05,
+      "loss": 0.0353,
+      "step": 11314
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 2.0116279069767443e-05,
+      "loss": 0.1275,
+      "step": 11316
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 2.0110993657505285e-05,
+      "loss": 0.1457,
+      "step": 11318
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 2.0105708245243128e-05,
+      "loss": 0.0438,
+      "step": 11320
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 2.0100422832980974e-05,
+      "loss": 0.0343,
+      "step": 11322
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 2.0095137420718817e-05,
+      "loss": 0.1007,
+      "step": 11324
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 2.0089852008456663e-05,
+      "loss": 0.0562,
+      "step": 11326
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 2.0084566596194505e-05,
+      "loss": 0.0126,
+      "step": 11328
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 2.0079281183932348e-05,
+      "loss": 0.0698,
+      "step": 11330
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 2.007399577167019e-05,
+      "loss": 0.0589,
+      "step": 11332
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 2.0068710359408037e-05,
+      "loss": 0.1048,
+      "step": 11334
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 2.006342494714588e-05,
+      "loss": 0.0315,
+      "step": 11336
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 2.0058139534883722e-05,
+      "loss": 0.0187,
+      "step": 11338
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 2.0052854122621565e-05,
+      "loss": 0.0887,
+      "step": 11340
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 2.004756871035941e-05,
+      "loss": 0.069,
+      "step": 11342
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 2.0042283298097253e-05,
+      "loss": 0.0561,
+      "step": 11344
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 2.0036997885835096e-05,
+      "loss": 0.0311,
+      "step": 11346
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 2.003171247357294e-05,
+      "loss": 0.0373,
+      "step": 11348
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 2.002642706131078e-05,
+      "loss": 0.0202,
+      "step": 11350
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 2.0021141649048628e-05,
+      "loss": 0.0274,
+      "step": 11352
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 2.001585623678647e-05,
+      "loss": 0.0598,
+      "step": 11354
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 2.0010570824524313e-05,
+      "loss": 0.0263,
+      "step": 11356
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 2.0005285412262156e-05,
+      "loss": 0.0716,
+      "step": 11358
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0086,
+      "step": 11360
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.9994714587737844e-05,
+      "loss": 0.0221,
+      "step": 11362
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.9989429175475687e-05,
+      "loss": 0.037,
+      "step": 11364
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.998414376321353e-05,
+      "loss": 0.0976,
+      "step": 11366
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.9978858350951372e-05,
+      "loss": 0.0408,
+      "step": 11368
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.997357293868922e-05,
+      "loss": 0.025,
+      "step": 11370
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.996828752642706e-05,
+      "loss": 0.0182,
+      "step": 11372
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.9963002114164904e-05,
+      "loss": 0.0515,
+      "step": 11374
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.995771670190275e-05,
+      "loss": 0.0557,
+      "step": 11376
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.9952431289640592e-05,
+      "loss": 0.0922,
+      "step": 11378
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.994714587737844e-05,
+      "loss": 0.0551,
+      "step": 11380
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.994186046511628e-05,
+      "loss": 0.1137,
+      "step": 11382
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.9936575052854124e-05,
+      "loss": 0.0116,
+      "step": 11384
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.9931289640591967e-05,
+      "loss": 0.0234,
+      "step": 11386
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.9926004228329813e-05,
+      "loss": 0.0043,
+      "step": 11388
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.9920718816067655e-05,
+      "loss": 0.0436,
+      "step": 11390
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.9915433403805498e-05,
+      "loss": 0.0186,
+      "step": 11392
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.991014799154334e-05,
+      "loss": 0.0104,
+      "step": 11394
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.9904862579281187e-05,
+      "loss": 0.0784,
+      "step": 11396
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.989957716701903e-05,
+      "loss": 0.0521,
+      "step": 11398
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.9894291754756872e-05,
+      "loss": 0.0724,
+      "step": 11400
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.9889006342494715e-05,
+      "loss": 0.0619,
+      "step": 11402
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.9883720930232557e-05,
+      "loss": 0.0655,
+      "step": 11404
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.9878435517970403e-05,
+      "loss": 0.0641,
+      "step": 11406
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.9873150105708246e-05,
+      "loss": 0.0981,
+      "step": 11408
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.986786469344609e-05,
+      "loss": 0.0054,
+      "step": 11410
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.986257928118393e-05,
+      "loss": 0.0116,
+      "step": 11412
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.9857293868921777e-05,
+      "loss": 0.0746,
+      "step": 11414
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.985200845665962e-05,
+      "loss": 0.0258,
+      "step": 11416
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.9846723044397463e-05,
+      "loss": 0.0716,
+      "step": 11418
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.9841437632135305e-05,
+      "loss": 0.0096,
+      "step": 11420
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.9836152219873148e-05,
+      "loss": 0.0508,
+      "step": 11422
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.9830866807610994e-05,
+      "loss": 0.0379,
+      "step": 11424
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.9825581395348837e-05,
+      "loss": 0.0356,
+      "step": 11426
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.9820295983086683e-05,
+      "loss": 0.0137,
+      "step": 11428
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.9815010570824526e-05,
+      "loss": 0.0478,
+      "step": 11430
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.980972515856237e-05,
+      "loss": 0.0576,
+      "step": 11432
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.9804439746300214e-05,
+      "loss": 0.0398,
+      "step": 11434
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.9799154334038057e-05,
+      "loss": 0.0909,
+      "step": 11436
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.97938689217759e-05,
+      "loss": 0.0252,
+      "step": 11438
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.9788583509513742e-05,
+      "loss": 0.0412,
+      "step": 11440
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.978329809725159e-05,
+      "loss": 0.0248,
+      "step": 11442
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.977801268498943e-05,
+      "loss": 0.0725,
+      "step": 11444
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.9772727272727274e-05,
+      "loss": 0.03,
+      "step": 11446
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.9767441860465116e-05,
+      "loss": 0.0491,
+      "step": 11448
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.9762156448202962e-05,
+      "loss": 0.0061,
+      "step": 11450
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.9756871035940805e-05,
+      "loss": 0.0138,
+      "step": 11452
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.9751585623678648e-05,
+      "loss": 0.0164,
+      "step": 11454
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.974630021141649e-05,
+      "loss": 0.0317,
+      "step": 11456
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.9741014799154333e-05,
+      "loss": 0.0404,
+      "step": 11458
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.973572938689218e-05,
+      "loss": 0.0162,
+      "step": 11460
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.9730443974630022e-05,
+      "loss": 0.0183,
+      "step": 11462
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.9725158562367865e-05,
+      "loss": 0.0297,
+      "step": 11464
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.9719873150105707e-05,
+      "loss": 0.038,
+      "step": 11466
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.9714587737843553e-05,
+      "loss": 0.0497,
+      "step": 11468
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.9709302325581396e-05,
+      "loss": 0.0859,
+      "step": 11470
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.970401691331924e-05,
+      "loss": 0.0235,
+      "step": 11472
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.969873150105708e-05,
+      "loss": 0.0481,
+      "step": 11474
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.9693446088794927e-05,
+      "loss": 0.0667,
+      "step": 11476
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.968816067653277e-05,
+      "loss": 0.0276,
+      "step": 11478
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.9682875264270613e-05,
+      "loss": 0.0065,
+      "step": 11480
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.967758985200846e-05,
+      "loss": 0.0434,
+      "step": 11482
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.96723044397463e-05,
+      "loss": 0.0375,
+      "step": 11484
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.9667019027484148e-05,
+      "loss": 0.0296,
+      "step": 11486
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.966173361522199e-05,
+      "loss": 0.0439,
+      "step": 11488
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.9656448202959833e-05,
+      "loss": 0.0253,
+      "step": 11490
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.9651162790697676e-05,
+      "loss": 0.0521,
+      "step": 11492
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.9645877378435518e-05,
+      "loss": 0.0333,
+      "step": 11494
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.9640591966173364e-05,
+      "loss": 0.0429,
+      "step": 11496
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.9635306553911207e-05,
+      "loss": 0.1049,
+      "step": 11498
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.963002114164905e-05,
+      "loss": 0.129,
+      "step": 11500
+    },
+    {
+      "epoch": 12.16,
+      "eval_cer": 0.059048161869478484,
+      "eval_loss": 0.6092634797096252,
+      "eval_runtime": 125.9114,
+      "eval_samples_per_second": 6.679,
+      "eval_steps_per_second": 0.842,
+      "step": 11500
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.9624735729386892e-05,
+      "loss": 0.0586,
+      "step": 11502
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.961945031712474e-05,
+      "loss": 0.0584,
+      "step": 11504
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.961416490486258e-05,
+      "loss": 0.0592,
+      "step": 11506
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.9608879492600424e-05,
+      "loss": 0.0383,
+      "step": 11508
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.9603594080338266e-05,
+      "loss": 0.0369,
+      "step": 11510
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.959830866807611e-05,
+      "loss": 0.0529,
+      "step": 11512
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.9593023255813955e-05,
+      "loss": 0.0497,
+      "step": 11514
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.9587737843551798e-05,
+      "loss": 0.0366,
+      "step": 11516
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.958245243128964e-05,
+      "loss": 0.044,
+      "step": 11518
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.9577167019027483e-05,
+      "loss": 0.0366,
+      "step": 11520
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.957188160676533e-05,
+      "loss": 0.0423,
+      "step": 11522
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.9566596194503172e-05,
+      "loss": 0.0449,
+      "step": 11524
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.9561310782241014e-05,
+      "loss": 0.0332,
+      "step": 11526
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.9556025369978857e-05,
+      "loss": 0.0076,
+      "step": 11528
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.9550739957716703e-05,
+      "loss": 0.035,
+      "step": 11530
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.9545454545454546e-05,
+      "loss": 0.0676,
+      "step": 11532
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.9540169133192392e-05,
+      "loss": 0.0041,
+      "step": 11534
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.9534883720930235e-05,
+      "loss": 0.0308,
+      "step": 11536
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.9529598308668077e-05,
+      "loss": 0.0438,
+      "step": 11538
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.9524312896405923e-05,
+      "loss": 0.0225,
+      "step": 11540
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.9519027484143766e-05,
+      "loss": 0.0332,
+      "step": 11542
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.951374207188161e-05,
+      "loss": 0.0264,
+      "step": 11544
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.950845665961945e-05,
+      "loss": 0.0327,
+      "step": 11546
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.9503171247357294e-05,
+      "loss": 0.0606,
+      "step": 11548
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.949788583509514e-05,
+      "loss": 0.0591,
+      "step": 11550
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.9492600422832983e-05,
+      "loss": 0.0971,
+      "step": 11552
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.9487315010570825e-05,
+      "loss": 0.0467,
+      "step": 11554
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.9482029598308668e-05,
+      "loss": 0.0173,
+      "step": 11556
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.9476744186046514e-05,
+      "loss": 0.0635,
+      "step": 11558
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.9471458773784357e-05,
+      "loss": 0.0159,
+      "step": 11560
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.94661733615222e-05,
+      "loss": 0.0232,
+      "step": 11562
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.9460887949260042e-05,
+      "loss": 0.022,
+      "step": 11564
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.9455602536997885e-05,
+      "loss": 0.0191,
+      "step": 11566
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.945031712473573e-05,
+      "loss": 0.0209,
+      "step": 11568
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.9445031712473574e-05,
+      "loss": 0.0111,
+      "step": 11570
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.9439746300211416e-05,
+      "loss": 0.0271,
+      "step": 11572
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.943446088794926e-05,
+      "loss": 0.0219,
+      "step": 11574
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.9429175475687105e-05,
+      "loss": 0.0302,
+      "step": 11576
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.9423890063424948e-05,
+      "loss": 0.059,
+      "step": 11578
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.941860465116279e-05,
+      "loss": 0.0582,
+      "step": 11580
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.9413319238900633e-05,
+      "loss": 0.0951,
+      "step": 11582
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.940803382663848e-05,
+      "loss": 0.0089,
+      "step": 11584
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.9402748414376322e-05,
+      "loss": 0.0225,
+      "step": 11586
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.9397463002114168e-05,
+      "loss": 0.1043,
+      "step": 11588
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.939217758985201e-05,
+      "loss": 0.0561,
+      "step": 11590
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.9386892177589853e-05,
+      "loss": 0.0499,
+      "step": 11592
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.93816067653277e-05,
+      "loss": 0.0464,
+      "step": 11594
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.9376321353065542e-05,
+      "loss": 0.0498,
+      "step": 11596
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.9371035940803385e-05,
+      "loss": 0.0296,
+      "step": 11598
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.9365750528541227e-05,
+      "loss": 0.0946,
+      "step": 11600
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.936046511627907e-05,
+      "loss": 0.094,
+      "step": 11602
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.9355179704016916e-05,
+      "loss": 0.1682,
+      "step": 11604
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.934989429175476e-05,
+      "loss": 0.0512,
+      "step": 11606
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.93446088794926e-05,
+      "loss": 0.1028,
+      "step": 11608
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.9339323467230444e-05,
+      "loss": 0.1174,
+      "step": 11610
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.933403805496829e-05,
+      "loss": 0.0504,
+      "step": 11612
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.9328752642706133e-05,
+      "loss": 0.0252,
+      "step": 11614
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.9323467230443975e-05,
+      "loss": 0.0987,
+      "step": 11616
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.9318181818181818e-05,
+      "loss": 0.0166,
+      "step": 11618
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.931289640591966e-05,
+      "loss": 0.1345,
+      "step": 11620
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.9307610993657507e-05,
+      "loss": 0.018,
+      "step": 11622
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.930232558139535e-05,
+      "loss": 0.0405,
+      "step": 11624
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.9297040169133192e-05,
+      "loss": 0.0127,
+      "step": 11626
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.9291754756871035e-05,
+      "loss": 0.0211,
+      "step": 11628
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.928646934460888e-05,
+      "loss": 0.2615,
+      "step": 11630
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.9281183932346724e-05,
+      "loss": 0.0352,
+      "step": 11632
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.9275898520084566e-05,
+      "loss": 0.062,
+      "step": 11634
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.9270613107822412e-05,
+      "loss": 0.0076,
+      "step": 11636
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.9265327695560255e-05,
+      "loss": 0.1104,
+      "step": 11638
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.9260042283298098e-05,
+      "loss": 0.0247,
+      "step": 11640
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.9254756871035944e-05,
+      "loss": 0.0621,
+      "step": 11642
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.9249471458773786e-05,
+      "loss": 0.0341,
+      "step": 11644
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.924418604651163e-05,
+      "loss": 0.0243,
+      "step": 11646
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.9238900634249475e-05,
+      "loss": 0.0353,
+      "step": 11648
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.9233615221987318e-05,
+      "loss": 0.0273,
+      "step": 11650
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.922832980972516e-05,
+      "loss": 0.0557,
+      "step": 11652
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.9223044397463003e-05,
+      "loss": 0.0112,
+      "step": 11654
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.9217758985200846e-05,
+      "loss": 0.0363,
+      "step": 11656
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.9212473572938692e-05,
+      "loss": 0.0698,
+      "step": 11658
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.9207188160676534e-05,
+      "loss": 0.16,
+      "step": 11660
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.9201902748414377e-05,
+      "loss": 0.089,
+      "step": 11662
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.919661733615222e-05,
+      "loss": 0.0624,
+      "step": 11664
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.9191331923890062e-05,
+      "loss": 0.0254,
+      "step": 11666
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.918604651162791e-05,
+      "loss": 0.0436,
+      "step": 11668
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.918076109936575e-05,
+      "loss": 0.0417,
+      "step": 11670
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.9175475687103594e-05,
+      "loss": 0.032,
+      "step": 11672
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.9170190274841437e-05,
+      "loss": 0.0656,
+      "step": 11674
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.9164904862579283e-05,
+      "loss": 0.0597,
+      "step": 11676
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.9159619450317125e-05,
+      "loss": 0.0349,
+      "step": 11678
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.9154334038054968e-05,
+      "loss": 0.0432,
+      "step": 11680
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.914904862579281e-05,
+      "loss": 0.0244,
+      "step": 11682
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.9143763213530653e-05,
+      "loss": 0.0721,
+      "step": 11684
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.91384778012685e-05,
+      "loss": 0.0318,
+      "step": 11686
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.9133192389006342e-05,
+      "loss": 0.0717,
+      "step": 11688
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.9127906976744188e-05,
+      "loss": 0.0474,
+      "step": 11690
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.912262156448203e-05,
+      "loss": 0.0258,
+      "step": 11692
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.9117336152219877e-05,
+      "loss": 0.0452,
+      "step": 11694
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.911205073995772e-05,
+      "loss": 0.0167,
+      "step": 11696
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.9106765327695562e-05,
+      "loss": 0.0303,
+      "step": 11698
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.9101479915433405e-05,
+      "loss": 0.0828,
+      "step": 11700
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.9096194503171247e-05,
+      "loss": 0.015,
+      "step": 11702
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.9090909090909094e-05,
+      "loss": 0.036,
+      "step": 11704
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.9085623678646936e-05,
+      "loss": 0.0045,
+      "step": 11706
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.908033826638478e-05,
+      "loss": 0.0425,
+      "step": 11708
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.907505285412262e-05,
+      "loss": 0.1253,
+      "step": 11710
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.9069767441860468e-05,
+      "loss": 0.051,
+      "step": 11712
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.906448202959831e-05,
+      "loss": 0.0164,
+      "step": 11714
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.9059196617336153e-05,
+      "loss": 0.0591,
+      "step": 11716
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.9053911205073996e-05,
+      "loss": 0.039,
+      "step": 11718
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.904862579281184e-05,
+      "loss": 0.0519,
+      "step": 11720
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.9043340380549684e-05,
+      "loss": 0.0243,
+      "step": 11722
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.9038054968287527e-05,
+      "loss": 0.0222,
+      "step": 11724
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.903276955602537e-05,
+      "loss": 0.0417,
+      "step": 11726
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.9027484143763212e-05,
+      "loss": 0.0052,
+      "step": 11728
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.902219873150106e-05,
+      "loss": 0.1375,
+      "step": 11730
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.90169133192389e-05,
+      "loss": 0.0676,
+      "step": 11732
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.9011627906976744e-05,
+      "loss": 0.0273,
+      "step": 11734
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.9006342494714586e-05,
+      "loss": 0.0784,
+      "step": 11736
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.9001057082452433e-05,
+      "loss": 0.0367,
+      "step": 11738
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.8995771670190275e-05,
+      "loss": 0.0331,
+      "step": 11740
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.8990486257928118e-05,
+      "loss": 0.0026,
+      "step": 11742
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.8985200845665964e-05,
+      "loss": 0.0098,
+      "step": 11744
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.8979915433403807e-05,
+      "loss": 0.0093,
+      "step": 11746
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.8974630021141653e-05,
+      "loss": 0.0568,
+      "step": 11748
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.8969344608879495e-05,
+      "loss": 0.0363,
+      "step": 11750
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.8964059196617338e-05,
+      "loss": 0.0386,
+      "step": 11752
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.895877378435518e-05,
+      "loss": 0.0281,
+      "step": 11754
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.8953488372093023e-05,
+      "loss": 0.018,
+      "step": 11756
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.894820295983087e-05,
+      "loss": 0.0129,
+      "step": 11758
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.8942917547568712e-05,
+      "loss": 0.0137,
+      "step": 11760
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.8937632135306555e-05,
+      "loss": 0.0773,
+      "step": 11762
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.8932346723044397e-05,
+      "loss": 0.069,
+      "step": 11764
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.8927061310782243e-05,
+      "loss": 0.0699,
+      "step": 11766
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.8921775898520086e-05,
+      "loss": 0.0518,
+      "step": 11768
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.891649048625793e-05,
+      "loss": 0.0252,
+      "step": 11770
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.891120507399577e-05,
+      "loss": 0.0087,
+      "step": 11772
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.8905919661733614e-05,
+      "loss": 0.0321,
+      "step": 11774
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.890063424947146e-05,
+      "loss": 0.0232,
+      "step": 11776
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.8895348837209303e-05,
+      "loss": 0.0225,
+      "step": 11778
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.8890063424947146e-05,
+      "loss": 0.043,
+      "step": 11780
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.8884778012684988e-05,
+      "loss": 0.0319,
+      "step": 11782
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.8879492600422834e-05,
+      "loss": 0.0238,
+      "step": 11784
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.8874207188160677e-05,
+      "loss": 0.044,
+      "step": 11786
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.886892177589852e-05,
+      "loss": 0.0721,
+      "step": 11788
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.8863636363636362e-05,
+      "loss": 0.1443,
+      "step": 11790
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.885835095137421e-05,
+      "loss": 0.1707,
+      "step": 11792
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.885306553911205e-05,
+      "loss": 0.0278,
+      "step": 11794
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.8847780126849897e-05,
+      "loss": 0.0472,
+      "step": 11796
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.884249471458774e-05,
+      "loss": 0.0468,
+      "step": 11798
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.8837209302325582e-05,
+      "loss": 0.0368,
+      "step": 11800
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.883192389006343e-05,
+      "loss": 0.0713,
+      "step": 11802
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.882663847780127e-05,
+      "loss": 0.0581,
+      "step": 11804
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.8821353065539114e-05,
+      "loss": 0.1298,
+      "step": 11806
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.8816067653276956e-05,
+      "loss": 0.1243,
+      "step": 11808
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.88107822410148e-05,
+      "loss": 0.078,
+      "step": 11810
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.8805496828752645e-05,
+      "loss": 0.0132,
+      "step": 11812
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.8800211416490488e-05,
+      "loss": 0.0336,
+      "step": 11814
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.879492600422833e-05,
+      "loss": 0.0503,
+      "step": 11816
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.8789640591966173e-05,
+      "loss": 0.0343,
+      "step": 11818
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.878435517970402e-05,
+      "loss": 0.0146,
+      "step": 11820
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.8779069767441862e-05,
+      "loss": 0.0561,
+      "step": 11822
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.8773784355179705e-05,
+      "loss": 0.141,
+      "step": 11824
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.8768498942917547e-05,
+      "loss": 0.0475,
+      "step": 11826
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.876321353065539e-05,
+      "loss": 0.0365,
+      "step": 11828
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.8757928118393236e-05,
+      "loss": 0.0063,
+      "step": 11830
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.875264270613108e-05,
+      "loss": 0.0754,
+      "step": 11832
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.874735729386892e-05,
+      "loss": 0.06,
+      "step": 11834
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.8742071881606764e-05,
+      "loss": 0.1181,
+      "step": 11836
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.873678646934461e-05,
+      "loss": 0.0563,
+      "step": 11838
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.8731501057082453e-05,
+      "loss": 0.0358,
+      "step": 11840
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.8726215644820295e-05,
+      "loss": 0.0087,
+      "step": 11842
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.8720930232558138e-05,
+      "loss": 0.0112,
+      "step": 11844
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.8715644820295984e-05,
+      "loss": 0.0286,
+      "step": 11846
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.8710359408033827e-05,
+      "loss": 0.0281,
+      "step": 11848
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.8705073995771673e-05,
+      "loss": 0.0756,
+      "step": 11850
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.8699788583509516e-05,
+      "loss": 0.0139,
+      "step": 11852
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.8694503171247358e-05,
+      "loss": 0.0402,
+      "step": 11854
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.8689217758985204e-05,
+      "loss": 0.023,
+      "step": 11856
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.8683932346723047e-05,
+      "loss": 0.0187,
+      "step": 11858
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.867864693446089e-05,
+      "loss": 0.0562,
+      "step": 11860
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.8673361522198732e-05,
+      "loss": 0.0147,
+      "step": 11862
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.8668076109936575e-05,
+      "loss": 0.0217,
+      "step": 11864
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.866279069767442e-05,
+      "loss": 0.0082,
+      "step": 11866
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.8657505285412264e-05,
+      "loss": 0.039,
+      "step": 11868
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.8652219873150106e-05,
+      "loss": 0.0157,
+      "step": 11870
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.864693446088795e-05,
+      "loss": 0.0795,
+      "step": 11872
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.8641649048625795e-05,
+      "loss": 0.0533,
+      "step": 11874
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.8636363636363638e-05,
+      "loss": 0.0508,
+      "step": 11876
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.863107822410148e-05,
+      "loss": 0.0334,
+      "step": 11878
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.8625792811839323e-05,
+      "loss": 0.022,
+      "step": 11880
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.8620507399577166e-05,
+      "loss": 0.0166,
+      "step": 11882
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.8615221987315012e-05,
+      "loss": 0.0703,
+      "step": 11884
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.8609936575052855e-05,
+      "loss": 0.0374,
+      "step": 11886
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.8604651162790697e-05,
+      "loss": 0.0318,
+      "step": 11888
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.859936575052854e-05,
+      "loss": 0.1932,
+      "step": 11890
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.8594080338266386e-05,
+      "loss": 0.0277,
+      "step": 11892
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.858879492600423e-05,
+      "loss": 0.0648,
+      "step": 11894
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.858350951374207e-05,
+      "loss": 0.0769,
+      "step": 11896
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.8578224101479917e-05,
+      "loss": 0.0774,
+      "step": 11898
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.857293868921776e-05,
+      "loss": 0.0401,
+      "step": 11900
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.8567653276955606e-05,
+      "loss": 0.0181,
+      "step": 11902
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.856236786469345e-05,
+      "loss": 0.0538,
+      "step": 11904
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.855708245243129e-05,
+      "loss": 0.0022,
+      "step": 11906
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.8551797040169134e-05,
+      "loss": 0.0727,
+      "step": 11908
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.854651162790698e-05,
+      "loss": 0.003,
+      "step": 11910
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.8541226215644823e-05,
+      "loss": 0.0089,
+      "step": 11912
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.8535940803382666e-05,
+      "loss": 0.0329,
+      "step": 11914
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.8530655391120508e-05,
+      "loss": 0.0231,
+      "step": 11916
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.852536997885835e-05,
+      "loss": 0.0387,
+      "step": 11918
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.8520084566596197e-05,
+      "loss": 0.061,
+      "step": 11920
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.851479915433404e-05,
+      "loss": 0.0585,
+      "step": 11922
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.8509513742071882e-05,
+      "loss": 0.0702,
+      "step": 11924
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.8504228329809725e-05,
+      "loss": 0.0598,
+      "step": 11926
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.849894291754757e-05,
+      "loss": 0.024,
+      "step": 11928
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.8493657505285414e-05,
+      "loss": 0.0682,
+      "step": 11930
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.8488372093023256e-05,
+      "loss": 0.0347,
+      "step": 11932
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.84830866807611e-05,
+      "loss": 0.0395,
+      "step": 11934
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.847780126849894e-05,
+      "loss": 0.0333,
+      "step": 11936
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.8472515856236788e-05,
+      "loss": 0.0432,
+      "step": 11938
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.846723044397463e-05,
+      "loss": 0.0321,
+      "step": 11940
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.8461945031712473e-05,
+      "loss": 0.0308,
+      "step": 11942
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.8456659619450316e-05,
+      "loss": 0.0345,
+      "step": 11944
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.8451374207188162e-05,
+      "loss": 0.0509,
+      "step": 11946
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.8446088794926004e-05,
+      "loss": 0.0276,
+      "step": 11948
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.8440803382663847e-05,
+      "loss": 0.0526,
+      "step": 11950
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.8435517970401693e-05,
+      "loss": 0.0226,
+      "step": 11952
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.8430232558139536e-05,
+      "loss": 0.0062,
+      "step": 11954
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.8424947145877382e-05,
+      "loss": 0.0682,
+      "step": 11956
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.8419661733615225e-05,
+      "loss": 0.0851,
+      "step": 11958
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.8414376321353067e-05,
+      "loss": 0.0219,
+      "step": 11960
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.840909090909091e-05,
+      "loss": 0.0094,
+      "step": 11962
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.8403805496828756e-05,
+      "loss": 0.0361,
+      "step": 11964
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.83985200845666e-05,
+      "loss": 0.1176,
+      "step": 11966
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.839323467230444e-05,
+      "loss": 0.0238,
+      "step": 11968
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.8387949260042284e-05,
+      "loss": 0.0309,
+      "step": 11970
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.8382663847780127e-05,
+      "loss": 0.0262,
+      "step": 11972
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.8377378435517973e-05,
+      "loss": 0.0233,
+      "step": 11974
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.8372093023255815e-05,
+      "loss": 0.0249,
+      "step": 11976
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.8366807610993658e-05,
+      "loss": 0.1753,
+      "step": 11978
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.83615221987315e-05,
+      "loss": 0.0459,
+      "step": 11980
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.8356236786469347e-05,
+      "loss": 0.0319,
+      "step": 11982
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.835095137420719e-05,
+      "loss": 0.088,
+      "step": 11984
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.8345665961945032e-05,
+      "loss": 0.0274,
+      "step": 11986
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.8340380549682875e-05,
+      "loss": 0.0498,
+      "step": 11988
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.8335095137420718e-05,
+      "loss": 0.0145,
+      "step": 11990
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.8329809725158564e-05,
+      "loss": 0.0204,
+      "step": 11992
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.8324524312896406e-05,
+      "loss": 0.0554,
+      "step": 11994
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.831923890063425e-05,
+      "loss": 0.0453,
+      "step": 11996
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.831395348837209e-05,
+      "loss": 0.0281,
+      "step": 11998
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.8308668076109938e-05,
+      "loss": 0.0519,
+      "step": 12000
+    },
+    {
+      "epoch": 12.68,
+      "eval_cer": 0.03493872898261613,
+      "eval_loss": 0.7152830958366394,
+      "eval_runtime": 126.707,
+      "eval_samples_per_second": 6.637,
+      "eval_steps_per_second": 0.837,
+      "step": 12000
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.830338266384778e-05,
+      "loss": 0.0397,
+      "step": 12002
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.8298097251585626e-05,
+      "loss": 0.0566,
+      "step": 12004
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.829281183932347e-05,
+      "loss": 0.0203,
+      "step": 12006
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.8287526427061312e-05,
+      "loss": 0.0205,
+      "step": 12008
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.8282241014799158e-05,
+      "loss": 0.0227,
+      "step": 12010
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.8276955602537e-05,
+      "loss": 0.0056,
+      "step": 12012
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.8271670190274843e-05,
+      "loss": 0.0421,
+      "step": 12014
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.8266384778012686e-05,
+      "loss": 0.0131,
+      "step": 12016
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.826109936575053e-05,
+      "loss": 0.0164,
+      "step": 12018
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.8255813953488375e-05,
+      "loss": 0.0469,
+      "step": 12020
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.8250528541226217e-05,
+      "loss": 0.0405,
+      "step": 12022
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.824524312896406e-05,
+      "loss": 0.0053,
+      "step": 12024
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.8239957716701903e-05,
+      "loss": 0.0114,
+      "step": 12026
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.823467230443975e-05,
+      "loss": 0.0044,
+      "step": 12028
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.822938689217759e-05,
+      "loss": 0.0255,
+      "step": 12030
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.8224101479915434e-05,
+      "loss": 0.0303,
+      "step": 12032
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.8218816067653277e-05,
+      "loss": 0.0222,
+      "step": 12034
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.821353065539112e-05,
+      "loss": 0.0577,
+      "step": 12036
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.8208245243128965e-05,
+      "loss": 0.0462,
+      "step": 12038
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.8202959830866808e-05,
+      "loss": 0.0222,
+      "step": 12040
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.819767441860465e-05,
+      "loss": 0.0696,
+      "step": 12042
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.8192389006342493e-05,
+      "loss": 0.0139,
+      "step": 12044
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.818710359408034e-05,
+      "loss": 0.0103,
+      "step": 12046
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 0.0333,
+      "step": 12048
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.8176532769556025e-05,
+      "loss": 0.0181,
+      "step": 12050
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.8171247357293867e-05,
+      "loss": 0.0193,
+      "step": 12052
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.8165961945031713e-05,
+      "loss": 0.051,
+      "step": 12054
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.8160676532769556e-05,
+      "loss": 0.0071,
+      "step": 12056
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.8155391120507402e-05,
+      "loss": 0.0019,
+      "step": 12058
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.8150105708245245e-05,
+      "loss": 0.0435,
+      "step": 12060
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.8144820295983088e-05,
+      "loss": 0.0204,
+      "step": 12062
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.8139534883720934e-05,
+      "loss": 0.0172,
+      "step": 12064
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.8134249471458776e-05,
+      "loss": 0.0118,
+      "step": 12066
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.812896405919662e-05,
+      "loss": 0.0241,
+      "step": 12068
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.812367864693446e-05,
+      "loss": 0.0647,
+      "step": 12070
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.8118393234672304e-05,
+      "loss": 0.0233,
+      "step": 12072
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.811310782241015e-05,
+      "loss": 0.0086,
+      "step": 12074
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.8107822410147993e-05,
+      "loss": 0.0489,
+      "step": 12076
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.8102536997885836e-05,
+      "loss": 0.0581,
+      "step": 12078
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.809725158562368e-05,
+      "loss": 0.0231,
+      "step": 12080
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.8091966173361524e-05,
+      "loss": 0.041,
+      "step": 12082
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.8086680761099367e-05,
+      "loss": 0.034,
+      "step": 12084
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.808139534883721e-05,
+      "loss": 0.0653,
+      "step": 12086
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.8076109936575052e-05,
+      "loss": 0.0455,
+      "step": 12088
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.8070824524312895e-05,
+      "loss": 0.0227,
+      "step": 12090
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.806553911205074e-05,
+      "loss": 0.1526,
+      "step": 12092
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.8060253699788584e-05,
+      "loss": 0.0399,
+      "step": 12094
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.8054968287526427e-05,
+      "loss": 0.0096,
+      "step": 12096
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.804968287526427e-05,
+      "loss": 0.081,
+      "step": 12098
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.8044397463002115e-05,
+      "loss": 0.0674,
+      "step": 12100
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.8039112050739958e-05,
+      "loss": 0.0201,
+      "step": 12102
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.80338266384778e-05,
+      "loss": 0.0162,
+      "step": 12104
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.8028541226215647e-05,
+      "loss": 0.1635,
+      "step": 12106
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.802325581395349e-05,
+      "loss": 0.0519,
+      "step": 12108
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.8017970401691332e-05,
+      "loss": 0.0668,
+      "step": 12110
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.8012684989429178e-05,
+      "loss": 0.0052,
+      "step": 12112
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.800739957716702e-05,
+      "loss": 0.0597,
+      "step": 12114
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.8002114164904863e-05,
+      "loss": 0.0481,
+      "step": 12116
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.799682875264271e-05,
+      "loss": 0.0407,
+      "step": 12118
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.7991543340380552e-05,
+      "loss": 0.0751,
+      "step": 12120
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.7986257928118395e-05,
+      "loss": 0.0375,
+      "step": 12122
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.7980972515856237e-05,
+      "loss": 0.0605,
+      "step": 12124
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.797568710359408e-05,
+      "loss": 0.0272,
+      "step": 12126
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.7970401691331926e-05,
+      "loss": 0.0457,
+      "step": 12128
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.796511627906977e-05,
+      "loss": 0.0282,
+      "step": 12130
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.795983086680761e-05,
+      "loss": 0.0617,
+      "step": 12132
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.7954545454545454e-05,
+      "loss": 0.0221,
+      "step": 12134
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.79492600422833e-05,
+      "loss": 0.0818,
+      "step": 12136
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.7943974630021143e-05,
+      "loss": 0.0797,
+      "step": 12138
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.7938689217758986e-05,
+      "loss": 0.0649,
+      "step": 12140
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.7933403805496828e-05,
+      "loss": 0.0281,
+      "step": 12142
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.792811839323467e-05,
+      "loss": 0.0125,
+      "step": 12144
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.7922832980972517e-05,
+      "loss": 0.0172,
+      "step": 12146
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.791754756871036e-05,
+      "loss": 0.023,
+      "step": 12148
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.7912262156448202e-05,
+      "loss": 0.0441,
+      "step": 12150
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.7906976744186045e-05,
+      "loss": 0.0704,
+      "step": 12152
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.790169133192389e-05,
+      "loss": 0.1245,
+      "step": 12154
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.7896405919661734e-05,
+      "loss": 0.0597,
+      "step": 12156
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.7891120507399576e-05,
+      "loss": 0.07,
+      "step": 12158
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.7885835095137422e-05,
+      "loss": 0.069,
+      "step": 12160
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.7880549682875265e-05,
+      "loss": 0.0785,
+      "step": 12162
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.787526427061311e-05,
+      "loss": 0.0217,
+      "step": 12164
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.7869978858350954e-05,
+      "loss": 0.0645,
+      "step": 12166
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.7864693446088797e-05,
+      "loss": 0.0546,
+      "step": 12168
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.785940803382664e-05,
+      "loss": 0.0539,
+      "step": 12170
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.7854122621564485e-05,
+      "loss": 0.1025,
+      "step": 12172
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.7848837209302328e-05,
+      "loss": 0.055,
+      "step": 12174
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.784355179704017e-05,
+      "loss": 0.0331,
+      "step": 12176
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.7838266384778013e-05,
+      "loss": 0.0556,
+      "step": 12178
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.7832980972515856e-05,
+      "loss": 0.0582,
+      "step": 12180
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.7827695560253702e-05,
+      "loss": 0.072,
+      "step": 12182
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.7822410147991545e-05,
+      "loss": 0.0276,
+      "step": 12184
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.7817124735729387e-05,
+      "loss": 0.0059,
+      "step": 12186
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.781183932346723e-05,
+      "loss": 0.0242,
+      "step": 12188
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.7806553911205076e-05,
+      "loss": 0.0488,
+      "step": 12190
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.780126849894292e-05,
+      "loss": 0.0418,
+      "step": 12192
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.779598308668076e-05,
+      "loss": 0.0164,
+      "step": 12194
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.7790697674418604e-05,
+      "loss": 0.0877,
+      "step": 12196
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.7785412262156447e-05,
+      "loss": 0.1477,
+      "step": 12198
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.7780126849894293e-05,
+      "loss": 0.0141,
+      "step": 12200
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.7774841437632136e-05,
+      "loss": 0.0179,
+      "step": 12202
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.7769556025369978e-05,
+      "loss": 0.0325,
+      "step": 12204
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.776427061310782e-05,
+      "loss": 0.0368,
+      "step": 12206
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.7758985200845667e-05,
+      "loss": 0.0178,
+      "step": 12208
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.775369978858351e-05,
+      "loss": 0.038,
+      "step": 12210
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.7748414376321352e-05,
+      "loss": 0.011,
+      "step": 12212
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.77431289640592e-05,
+      "loss": 0.0098,
+      "step": 12214
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.773784355179704e-05,
+      "loss": 0.0482,
+      "step": 12216
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.7732558139534887e-05,
+      "loss": 0.0697,
+      "step": 12218
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.772727272727273e-05,
+      "loss": 0.0124,
+      "step": 12220
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.7721987315010572e-05,
+      "loss": 0.1065,
+      "step": 12222
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.7716701902748415e-05,
+      "loss": 0.1214,
+      "step": 12224
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.771141649048626e-05,
+      "loss": 0.0112,
+      "step": 12226
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.7706131078224104e-05,
+      "loss": 0.0301,
+      "step": 12228
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.7700845665961946e-05,
+      "loss": 0.1167,
+      "step": 12230
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.769556025369979e-05,
+      "loss": 0.0479,
+      "step": 12232
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.7690274841437632e-05,
+      "loss": 0.0196,
+      "step": 12234
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.7684989429175478e-05,
+      "loss": 0.0172,
+      "step": 12236
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.767970401691332e-05,
+      "loss": 0.0264,
+      "step": 12238
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.7674418604651163e-05,
+      "loss": 0.121,
+      "step": 12240
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.7669133192389006e-05,
+      "loss": 0.0175,
+      "step": 12242
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.7663847780126852e-05,
+      "loss": 0.0459,
+      "step": 12244
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.7658562367864695e-05,
+      "loss": 0.0183,
+      "step": 12246
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.7653276955602537e-05,
+      "loss": 0.0601,
+      "step": 12248
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.764799154334038e-05,
+      "loss": 0.0104,
+      "step": 12250
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.7642706131078223e-05,
+      "loss": 0.0101,
+      "step": 12252
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.763742071881607e-05,
+      "loss": 0.0273,
+      "step": 12254
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.763213530655391e-05,
+      "loss": 0.0343,
+      "step": 12256
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.7626849894291754e-05,
+      "loss": 0.0275,
+      "step": 12258
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.7621564482029597e-05,
+      "loss": 0.0227,
+      "step": 12260
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.7616279069767443e-05,
+      "loss": 0.0143,
+      "step": 12262
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.7610993657505285e-05,
+      "loss": 0.0261,
+      "step": 12264
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.760570824524313e-05,
+      "loss": 0.0837,
+      "step": 12266
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.7600422832980974e-05,
+      "loss": 0.052,
+      "step": 12268
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.7595137420718817e-05,
+      "loss": 0.0346,
+      "step": 12270
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.7589852008456663e-05,
+      "loss": 0.0276,
+      "step": 12272
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.7584566596194506e-05,
+      "loss": 0.0328,
+      "step": 12274
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.7579281183932348e-05,
+      "loss": 0.0346,
+      "step": 12276
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.757399577167019e-05,
+      "loss": 0.0911,
+      "step": 12278
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.7568710359408037e-05,
+      "loss": 0.0286,
+      "step": 12280
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.756342494714588e-05,
+      "loss": 0.0165,
+      "step": 12282
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.7558139534883722e-05,
+      "loss": 0.0112,
+      "step": 12284
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.7552854122621565e-05,
+      "loss": 0.0442,
+      "step": 12286
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.7547568710359408e-05,
+      "loss": 0.0113,
+      "step": 12288
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.7542283298097254e-05,
+      "loss": 0.0397,
+      "step": 12290
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.7536997885835096e-05,
+      "loss": 0.0113,
+      "step": 12292
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.753171247357294e-05,
+      "loss": 0.0359,
+      "step": 12294
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.7526427061310782e-05,
+      "loss": 0.0269,
+      "step": 12296
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.7521141649048628e-05,
+      "loss": 0.0205,
+      "step": 12298
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.751585623678647e-05,
+      "loss": 0.006,
+      "step": 12300
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.7510570824524313e-05,
+      "loss": 0.0539,
+      "step": 12302
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.7505285412262156e-05,
+      "loss": 0.0533,
+      "step": 12304
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.75e-05,
+      "loss": 0.067,
+      "step": 12306
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.7494714587737845e-05,
+      "loss": 0.0075,
+      "step": 12308
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.7489429175475687e-05,
+      "loss": 0.0233,
+      "step": 12310
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.748414376321353e-05,
+      "loss": 0.0528,
+      "step": 12312
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.7478858350951376e-05,
+      "loss": 0.0479,
+      "step": 12314
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.747357293868922e-05,
+      "loss": 0.1347,
+      "step": 12316
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.746828752642706e-05,
+      "loss": 0.0294,
+      "step": 12318
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.7463002114164907e-05,
+      "loss": 0.0273,
+      "step": 12320
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.745771670190275e-05,
+      "loss": 0.0432,
+      "step": 12322
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.7452431289640593e-05,
+      "loss": 0.106,
+      "step": 12324
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.744714587737844e-05,
+      "loss": 0.0854,
+      "step": 12326
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.744186046511628e-05,
+      "loss": 0.0735,
+      "step": 12328
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.7436575052854124e-05,
+      "loss": 0.0072,
+      "step": 12330
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.7431289640591967e-05,
+      "loss": 0.0377,
+      "step": 12332
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.7426004228329813e-05,
+      "loss": 0.0226,
+      "step": 12334
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.7420718816067655e-05,
+      "loss": 0.0568,
+      "step": 12336
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.7415433403805498e-05,
+      "loss": 0.0474,
+      "step": 12338
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.741014799154334e-05,
+      "loss": 0.0052,
+      "step": 12340
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.7404862579281183e-05,
+      "loss": 0.0388,
+      "step": 12342
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.739957716701903e-05,
+      "loss": 0.0411,
+      "step": 12344
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.7394291754756872e-05,
+      "loss": 0.0087,
+      "step": 12346
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.7389006342494715e-05,
+      "loss": 0.0273,
+      "step": 12348
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.7383720930232558e-05,
+      "loss": 0.0399,
+      "step": 12350
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.73784355179704e-05,
+      "loss": 0.003,
+      "step": 12352
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.7373150105708246e-05,
+      "loss": 0.0108,
+      "step": 12354
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.736786469344609e-05,
+      "loss": 0.0944,
+      "step": 12356
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.736257928118393e-05,
+      "loss": 0.088,
+      "step": 12358
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.7357293868921774e-05,
+      "loss": 0.0058,
+      "step": 12360
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.735200845665962e-05,
+      "loss": 0.0372,
+      "step": 12362
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.7346723044397463e-05,
+      "loss": 0.0398,
+      "step": 12364
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.7341437632135306e-05,
+      "loss": 0.0178,
+      "step": 12366
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.7336152219873152e-05,
+      "loss": 0.0135,
+      "step": 12368
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.7330866807610994e-05,
+      "loss": 0.0077,
+      "step": 12370
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.732558139534884e-05,
+      "loss": 0.016,
+      "step": 12372
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.7320295983086683e-05,
+      "loss": 0.1078,
+      "step": 12374
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.7315010570824526e-05,
+      "loss": 0.0113,
+      "step": 12376
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.730972515856237e-05,
+      "loss": 0.032,
+      "step": 12378
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.7304439746300215e-05,
+      "loss": 0.0365,
+      "step": 12380
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.7299154334038057e-05,
+      "loss": 0.0738,
+      "step": 12382
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.72938689217759e-05,
+      "loss": 0.0072,
+      "step": 12384
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.7288583509513743e-05,
+      "loss": 0.0046,
+      "step": 12386
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.7283298097251585e-05,
+      "loss": 0.0382,
+      "step": 12388
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.727801268498943e-05,
+      "loss": 0.0071,
+      "step": 12390
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.7272727272727274e-05,
+      "loss": 0.005,
+      "step": 12392
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.7267441860465117e-05,
+      "loss": 0.1507,
+      "step": 12394
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.726215644820296e-05,
+      "loss": 0.0189,
+      "step": 12396
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.7256871035940805e-05,
+      "loss": 0.0148,
+      "step": 12398
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.7251585623678648e-05,
+      "loss": 0.0264,
+      "step": 12400
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.724630021141649e-05,
+      "loss": 0.0133,
+      "step": 12402
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.7241014799154333e-05,
+      "loss": 0.0567,
+      "step": 12404
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.7235729386892176e-05,
+      "loss": 0.0368,
+      "step": 12406
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.7230443974630022e-05,
+      "loss": 0.0566,
+      "step": 12408
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.7225158562367865e-05,
+      "loss": 0.0335,
+      "step": 12410
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.7219873150105707e-05,
+      "loss": 0.0466,
+      "step": 12412
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.721458773784355e-05,
+      "loss": 0.0176,
+      "step": 12414
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.7209302325581396e-05,
+      "loss": 0.0085,
+      "step": 12416
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.720401691331924e-05,
+      "loss": 0.0475,
+      "step": 12418
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.719873150105708e-05,
+      "loss": 0.0202,
+      "step": 12420
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.7193446088794928e-05,
+      "loss": 0.0277,
+      "step": 12422
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.718816067653277e-05,
+      "loss": 0.0702,
+      "step": 12424
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.7182875264270616e-05,
+      "loss": 0.0043,
+      "step": 12426
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.717758985200846e-05,
+      "loss": 0.0044,
+      "step": 12428
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.71723044397463e-05,
+      "loss": 0.0139,
+      "step": 12430
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.7167019027484144e-05,
+      "loss": 0.0196,
+      "step": 12432
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.716173361522199e-05,
+      "loss": 0.0219,
+      "step": 12434
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.7156448202959833e-05,
+      "loss": 0.007,
+      "step": 12436
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.7151162790697676e-05,
+      "loss": 0.063,
+      "step": 12438
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.714587737843552e-05,
+      "loss": 0.0267,
+      "step": 12440
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.714059196617336e-05,
+      "loss": 0.0312,
+      "step": 12442
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.7135306553911207e-05,
+      "loss": 0.0269,
+      "step": 12444
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.713002114164905e-05,
+      "loss": 0.0248,
+      "step": 12446
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.7124735729386892e-05,
+      "loss": 0.0452,
+      "step": 12448
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.7119450317124735e-05,
+      "loss": 0.0167,
+      "step": 12450
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.711416490486258e-05,
+      "loss": 0.0416,
+      "step": 12452
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.7108879492600424e-05,
+      "loss": 0.017,
+      "step": 12454
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.7103594080338267e-05,
+      "loss": 0.0101,
+      "step": 12456
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.709830866807611e-05,
+      "loss": 0.0318,
+      "step": 12458
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.7093023255813952e-05,
+      "loss": 0.0352,
+      "step": 12460
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.7087737843551798e-05,
+      "loss": 0.0109,
+      "step": 12462
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.708245243128964e-05,
+      "loss": 0.0253,
+      "step": 12464
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.7077167019027483e-05,
+      "loss": 0.0144,
+      "step": 12466
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.7071881606765326e-05,
+      "loss": 0.0045,
+      "step": 12468
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.7066596194503172e-05,
+      "loss": 0.0683,
+      "step": 12470
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.7061310782241015e-05,
+      "loss": 0.0128,
+      "step": 12472
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.705602536997886e-05,
+      "loss": 0.0172,
+      "step": 12474
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.7050739957716703e-05,
+      "loss": 0.0041,
+      "step": 12476
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.7045454545454546e-05,
+      "loss": 0.044,
+      "step": 12478
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.7040169133192392e-05,
+      "loss": 0.0309,
+      "step": 12480
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.7034883720930235e-05,
+      "loss": 0.0039,
+      "step": 12482
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.7029598308668078e-05,
+      "loss": 0.0094,
+      "step": 12484
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.702431289640592e-05,
+      "loss": 0.0042,
+      "step": 12486
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.7019027484143766e-05,
+      "loss": 0.0219,
+      "step": 12488
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.701374207188161e-05,
+      "loss": 0.0145,
+      "step": 12490
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.700845665961945e-05,
+      "loss": 0.0336,
+      "step": 12492
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.7003171247357294e-05,
+      "loss": 0.0401,
+      "step": 12494
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.6997885835095137e-05,
+      "loss": 0.0391,
+      "step": 12496
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.6992600422832983e-05,
+      "loss": 0.038,
+      "step": 12498
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.6987315010570826e-05,
+      "loss": 0.0196,
+      "step": 12500
+    },
+    {
+      "epoch": 13.21,
+      "eval_cer": 0.06223995440296381,
+      "eval_loss": 0.755074143409729,
+      "eval_runtime": 124.4091,
+      "eval_samples_per_second": 6.76,
+      "eval_steps_per_second": 0.852,
+      "step": 12500
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.698202959830867e-05,
+      "loss": 0.007,
+      "step": 12502
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.697674418604651e-05,
+      "loss": 0.025,
+      "step": 12504
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.6971458773784357e-05,
+      "loss": 0.029,
+      "step": 12506
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.69661733615222e-05,
+      "loss": 0.0127,
+      "step": 12508
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.6960887949260042e-05,
+      "loss": 0.0133,
+      "step": 12510
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.6955602536997885e-05,
+      "loss": 0.0413,
+      "step": 12512
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.6950317124735728e-05,
+      "loss": 0.0103,
+      "step": 12514
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.6945031712473574e-05,
+      "loss": 0.0451,
+      "step": 12516
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.6939746300211416e-05,
+      "loss": 0.0842,
+      "step": 12518
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.693446088794926e-05,
+      "loss": 0.0592,
+      "step": 12520
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.6929175475687102e-05,
+      "loss": 0.0405,
+      "step": 12522
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.6923890063424948e-05,
+      "loss": 0.0995,
+      "step": 12524
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.691860465116279e-05,
+      "loss": 0.0079,
+      "step": 12526
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.6913319238900637e-05,
+      "loss": 0.0506,
+      "step": 12528
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.690803382663848e-05,
+      "loss": 0.0021,
+      "step": 12530
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.6902748414376322e-05,
+      "loss": 0.0799,
+      "step": 12532
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.6897463002114168e-05,
+      "loss": 0.0053,
+      "step": 12534
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.689217758985201e-05,
+      "loss": 0.0199,
+      "step": 12536
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.6886892177589853e-05,
+      "loss": 0.0066,
+      "step": 12538
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.6881606765327696e-05,
+      "loss": 0.015,
+      "step": 12540
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.6876321353065542e-05,
+      "loss": 0.0059,
+      "step": 12542
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.6871035940803385e-05,
+      "loss": 0.0768,
+      "step": 12544
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.6865750528541227e-05,
+      "loss": 0.0812,
+      "step": 12546
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.686046511627907e-05,
+      "loss": 0.0314,
+      "step": 12548
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.6855179704016913e-05,
+      "loss": 0.0307,
+      "step": 12550
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.684989429175476e-05,
+      "loss": 0.0369,
+      "step": 12552
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.68446088794926e-05,
+      "loss": 0.0515,
+      "step": 12554
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.6839323467230444e-05,
+      "loss": 0.0093,
+      "step": 12556
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.6834038054968287e-05,
+      "loss": 0.0179,
+      "step": 12558
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.6828752642706133e-05,
+      "loss": 0.0736,
+      "step": 12560
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.6823467230443976e-05,
+      "loss": 0.0317,
+      "step": 12562
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.6818181818181818e-05,
+      "loss": 0.0386,
+      "step": 12564
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.681289640591966e-05,
+      "loss": 0.0064,
+      "step": 12566
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.6807610993657504e-05,
+      "loss": 0.0181,
+      "step": 12568
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.680232558139535e-05,
+      "loss": 0.0671,
+      "step": 12570
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.6797040169133192e-05,
+      "loss": 0.0633,
+      "step": 12572
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.6791754756871035e-05,
+      "loss": 0.0319,
+      "step": 12574
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.678646934460888e-05,
+      "loss": 0.0428,
+      "step": 12576
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.6781183932346724e-05,
+      "loss": 0.0128,
+      "step": 12578
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.6775898520084566e-05,
+      "loss": 0.0144,
+      "step": 12580
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.6770613107822412e-05,
+      "loss": 0.0642,
+      "step": 12582
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.6765327695560255e-05,
+      "loss": 0.0288,
+      "step": 12584
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.6760042283298098e-05,
+      "loss": 0.0676,
+      "step": 12586
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.6754756871035944e-05,
+      "loss": 0.0565,
+      "step": 12588
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.6749471458773787e-05,
+      "loss": 0.0209,
+      "step": 12590
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.674418604651163e-05,
+      "loss": 0.0521,
+      "step": 12592
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.6738900634249472e-05,
+      "loss": 0.0486,
+      "step": 12594
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.6733615221987318e-05,
+      "loss": 0.0414,
+      "step": 12596
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.672832980972516e-05,
+      "loss": 0.1086,
+      "step": 12598
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.6723044397463003e-05,
+      "loss": 0.0127,
+      "step": 12600
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.6717758985200846e-05,
+      "loss": 0.0137,
+      "step": 12602
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.671247357293869e-05,
+      "loss": 0.0251,
+      "step": 12604
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.6707188160676535e-05,
+      "loss": 0.0726,
+      "step": 12606
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.6701902748414377e-05,
+      "loss": 0.0432,
+      "step": 12608
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.669661733615222e-05,
+      "loss": 0.0551,
+      "step": 12610
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.6691331923890063e-05,
+      "loss": 0.022,
+      "step": 12612
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.668604651162791e-05,
+      "loss": 0.0516,
+      "step": 12614
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.668076109936575e-05,
+      "loss": 0.0044,
+      "step": 12616
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.6675475687103594e-05,
+      "loss": 0.0202,
+      "step": 12618
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.6670190274841437e-05,
+      "loss": 0.0538,
+      "step": 12620
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.666490486257928e-05,
+      "loss": 0.0165,
+      "step": 12622
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.6659619450317125e-05,
+      "loss": 0.0322,
+      "step": 12624
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.6654334038054968e-05,
+      "loss": 0.043,
+      "step": 12626
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.664904862579281e-05,
+      "loss": 0.0144,
+      "step": 12628
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.6643763213530657e-05,
+      "loss": 0.0054,
+      "step": 12630
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.66384778012685e-05,
+      "loss": 0.0574,
+      "step": 12632
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.6633192389006346e-05,
+      "loss": 0.0266,
+      "step": 12634
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.6627906976744188e-05,
+      "loss": 0.0316,
+      "step": 12636
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.662262156448203e-05,
+      "loss": 0.0876,
+      "step": 12638
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.6617336152219874e-05,
+      "loss": 0.0855,
+      "step": 12640
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.661205073995772e-05,
+      "loss": 0.1275,
+      "step": 12642
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.6606765327695562e-05,
+      "loss": 0.0203,
+      "step": 12644
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.6601479915433405e-05,
+      "loss": 0.0323,
+      "step": 12646
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.6596194503171248e-05,
+      "loss": 0.0605,
+      "step": 12648
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.6590909090909094e-05,
+      "loss": 0.0931,
+      "step": 12650
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.6585623678646936e-05,
+      "loss": 0.0625,
+      "step": 12652
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.658033826638478e-05,
+      "loss": 0.1087,
+      "step": 12654
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.6575052854122622e-05,
+      "loss": 0.0316,
+      "step": 12656
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.6569767441860464e-05,
+      "loss": 0.0726,
+      "step": 12658
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.656448202959831e-05,
+      "loss": 0.0508,
+      "step": 12660
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.6559196617336153e-05,
+      "loss": 0.0491,
+      "step": 12662
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.6553911205073996e-05,
+      "loss": 0.0286,
+      "step": 12664
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.654862579281184e-05,
+      "loss": 0.0067,
+      "step": 12666
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.6543340380549685e-05,
+      "loss": 0.0397,
+      "step": 12668
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.6538054968287527e-05,
+      "loss": 0.0147,
+      "step": 12670
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.653276955602537e-05,
+      "loss": 0.0627,
+      "step": 12672
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.6527484143763213e-05,
+      "loss": 0.0232,
+      "step": 12674
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.6522198731501055e-05,
+      "loss": 0.0828,
+      "step": 12676
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.65169133192389e-05,
+      "loss": 0.0694,
+      "step": 12678
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.6511627906976744e-05,
+      "loss": 0.0384,
+      "step": 12680
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.650634249471459e-05,
+      "loss": 0.0524,
+      "step": 12682
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.6501057082452433e-05,
+      "loss": 0.0372,
+      "step": 12684
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.6495771670190275e-05,
+      "loss": 0.0427,
+      "step": 12686
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.649048625792812e-05,
+      "loss": 0.0364,
+      "step": 12688
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.6485200845665964e-05,
+      "loss": 0.0206,
+      "step": 12690
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.6479915433403807e-05,
+      "loss": 0.0788,
+      "step": 12692
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.647463002114165e-05,
+      "loss": 0.0328,
+      "step": 12694
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.6469344608879496e-05,
+      "loss": 0.0387,
+      "step": 12696
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.6464059196617338e-05,
+      "loss": 0.0256,
+      "step": 12698
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.645877378435518e-05,
+      "loss": 0.0057,
+      "step": 12700
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.6453488372093024e-05,
+      "loss": 0.0207,
+      "step": 12702
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.6448202959830866e-05,
+      "loss": 0.026,
+      "step": 12704
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.6442917547568712e-05,
+      "loss": 0.0282,
+      "step": 12706
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.6437632135306555e-05,
+      "loss": 0.0158,
+      "step": 12708
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.6432346723044398e-05,
+      "loss": 0.0053,
+      "step": 12710
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.642706131078224e-05,
+      "loss": 0.0359,
+      "step": 12712
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.6421775898520086e-05,
+      "loss": 0.0409,
+      "step": 12714
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.641649048625793e-05,
+      "loss": 0.0491,
+      "step": 12716
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.641120507399577e-05,
+      "loss": 0.0802,
+      "step": 12718
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.6405919661733614e-05,
+      "loss": 0.014,
+      "step": 12720
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.6400634249471457e-05,
+      "loss": 0.0092,
+      "step": 12722
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.6395348837209303e-05,
+      "loss": 0.017,
+      "step": 12724
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.6390063424947146e-05,
+      "loss": 0.0035,
+      "step": 12726
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.638477801268499e-05,
+      "loss": 0.0399,
+      "step": 12728
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.637949260042283e-05,
+      "loss": 0.047,
+      "step": 12730
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.6374207188160677e-05,
+      "loss": 0.0686,
+      "step": 12732
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.636892177589852e-05,
+      "loss": 0.0192,
+      "step": 12734
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.6363636363636366e-05,
+      "loss": 0.0117,
+      "step": 12736
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.635835095137421e-05,
+      "loss": 0.0389,
+      "step": 12738
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.635306553911205e-05,
+      "loss": 0.013,
+      "step": 12740
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.6347780126849897e-05,
+      "loss": 0.0425,
+      "step": 12742
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.634249471458774e-05,
+      "loss": 0.0402,
+      "step": 12744
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.6337209302325583e-05,
+      "loss": 0.0108,
+      "step": 12746
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.6331923890063425e-05,
+      "loss": 0.0307,
+      "step": 12748
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.632663847780127e-05,
+      "loss": 0.0325,
+      "step": 12750
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.6321353065539114e-05,
+      "loss": 0.0307,
+      "step": 12752
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.6316067653276957e-05,
+      "loss": 0.0683,
+      "step": 12754
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.63107822410148e-05,
+      "loss": 0.1128,
+      "step": 12756
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.6305496828752642e-05,
+      "loss": 0.0099,
+      "step": 12758
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.6300211416490488e-05,
+      "loss": 0.0266,
+      "step": 12760
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.629492600422833e-05,
+      "loss": 0.0508,
+      "step": 12762
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.6289640591966173e-05,
+      "loss": 0.0086,
+      "step": 12764
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.6284355179704016e-05,
+      "loss": 0.0065,
+      "step": 12766
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1.6279069767441862e-05,
+      "loss": 0.0293,
+      "step": 12768
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1.6273784355179705e-05,
+      "loss": 0.0284,
+      "step": 12770
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1.6268498942917548e-05,
+      "loss": 0.049,
+      "step": 12772
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1.626321353065539e-05,
+      "loss": 0.0055,
+      "step": 12774
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.6257928118393233e-05,
+      "loss": 0.0229,
+      "step": 12776
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.625264270613108e-05,
+      "loss": 0.025,
+      "step": 12778
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.624735729386892e-05,
+      "loss": 0.0119,
+      "step": 12780
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.6242071881606764e-05,
+      "loss": 0.0036,
+      "step": 12782
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.623678646934461e-05,
+      "loss": 0.0205,
+      "step": 12784
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 1.6231501057082453e-05,
+      "loss": 0.0826,
+      "step": 12786
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 1.6226215644820296e-05,
+      "loss": 0.0318,
+      "step": 12788
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 1.6220930232558142e-05,
+      "loss": 0.06,
+      "step": 12790
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 1.6215644820295984e-05,
+      "loss": 0.0567,
+      "step": 12792
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 1.6210359408033827e-05,
+      "loss": 0.0042,
+      "step": 12794
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.6205073995771673e-05,
+      "loss": 0.0224,
+      "step": 12796
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.6199788583509516e-05,
+      "loss": 0.0903,
+      "step": 12798
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.619450317124736e-05,
+      "loss": 0.0358,
+      "step": 12800
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.61892177589852e-05,
+      "loss": 0.0387,
+      "step": 12802
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.6183932346723047e-05,
+      "loss": 0.012,
+      "step": 12804
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.617864693446089e-05,
+      "loss": 0.0068,
+      "step": 12806
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.6173361522198733e-05,
+      "loss": 0.0562,
+      "step": 12808
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.6168076109936575e-05,
+      "loss": 0.0161,
+      "step": 12810
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.6162790697674418e-05,
+      "loss": 0.1477,
+      "step": 12812
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 1.6157505285412264e-05,
+      "loss": 0.0068,
+      "step": 12814
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 1.6152219873150107e-05,
+      "loss": 0.0012,
+      "step": 12816
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 1.614693446088795e-05,
+      "loss": 0.0062,
+      "step": 12818
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 1.6141649048625792e-05,
+      "loss": 0.0027,
+      "step": 12820
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 1.6136363636363638e-05,
+      "loss": 0.0334,
+      "step": 12822
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.613107822410148e-05,
+      "loss": 0.0741,
+      "step": 12824
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.6125792811839323e-05,
+      "loss": 0.0322,
+      "step": 12826
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.6120507399577166e-05,
+      "loss": 0.0601,
+      "step": 12828
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.611522198731501e-05,
+      "loss": 0.0885,
+      "step": 12830
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.6109936575052855e-05,
+      "loss": 0.0439,
+      "step": 12832
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 1.6104651162790697e-05,
+      "loss": 0.0516,
+      "step": 12834
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 1.609936575052854e-05,
+      "loss": 0.0019,
+      "step": 12836
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 1.6094080338266386e-05,
+      "loss": 0.0077,
+      "step": 12838
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 1.608879492600423e-05,
+      "loss": 0.0339,
+      "step": 12840
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.6083509513742075e-05,
+      "loss": 0.095,
+      "step": 12842
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.6078224101479918e-05,
+      "loss": 0.0199,
+      "step": 12844
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.607293868921776e-05,
+      "loss": 0.0221,
+      "step": 12846
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.6067653276955603e-05,
+      "loss": 0.0254,
+      "step": 12848
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.606236786469345e-05,
+      "loss": 0.0099,
+      "step": 12850
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.605708245243129e-05,
+      "loss": 0.0319,
+      "step": 12852
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.6051797040169134e-05,
+      "loss": 0.0054,
+      "step": 12854
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.6046511627906977e-05,
+      "loss": 0.0582,
+      "step": 12856
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.6041226215644823e-05,
+      "loss": 0.0241,
+      "step": 12858
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.6035940803382666e-05,
+      "loss": 0.0605,
+      "step": 12860
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.603065539112051e-05,
+      "loss": 0.0041,
+      "step": 12862
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.602536997885835e-05,
+      "loss": 0.0449,
+      "step": 12864
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.6020084566596194e-05,
+      "loss": 0.037,
+      "step": 12866
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.601479915433404e-05,
+      "loss": 0.0217,
+      "step": 12868
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.6009513742071882e-05,
+      "loss": 0.0166,
+      "step": 12870
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.6004228329809725e-05,
+      "loss": 0.0344,
+      "step": 12872
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.5998942917547568e-05,
+      "loss": 0.0757,
+      "step": 12874
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.5993657505285414e-05,
+      "loss": 0.0141,
+      "step": 12876
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.5988372093023257e-05,
+      "loss": 0.0195,
+      "step": 12878
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.59830866807611e-05,
+      "loss": 0.0059,
+      "step": 12880
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.5977801268498942e-05,
+      "loss": 0.0134,
+      "step": 12882
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.5972515856236785e-05,
+      "loss": 0.0369,
+      "step": 12884
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.596723044397463e-05,
+      "loss": 0.0202,
+      "step": 12886
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.5961945031712473e-05,
+      "loss": 0.0057,
+      "step": 12888
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.5956659619450316e-05,
+      "loss": 0.0126,
+      "step": 12890
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.5951374207188162e-05,
+      "loss": 0.0295,
+      "step": 12892
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.5946088794926005e-05,
+      "loss": 0.0557,
+      "step": 12894
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.594080338266385e-05,
+      "loss": 0.0056,
+      "step": 12896
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.5935517970401693e-05,
+      "loss": 0.0762,
+      "step": 12898
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.5930232558139536e-05,
+      "loss": 0.1435,
+      "step": 12900
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.592494714587738e-05,
+      "loss": 0.0182,
+      "step": 12902
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.5919661733615225e-05,
+      "loss": 0.0355,
+      "step": 12904
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.5914376321353067e-05,
+      "loss": 0.0059,
+      "step": 12906
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.590909090909091e-05,
+      "loss": 0.0151,
+      "step": 12908
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.5903805496828753e-05,
+      "loss": 0.0204,
+      "step": 12910
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.58985200845666e-05,
+      "loss": 0.006,
+      "step": 12912
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.589323467230444e-05,
+      "loss": 0.0701,
+      "step": 12914
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.5887949260042284e-05,
+      "loss": 0.0503,
+      "step": 12916
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.5882663847780127e-05,
+      "loss": 0.0082,
+      "step": 12918
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.587737843551797e-05,
+      "loss": 0.0953,
+      "step": 12920
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.5872093023255816e-05,
+      "loss": 0.0477,
+      "step": 12922
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.586680761099366e-05,
+      "loss": 0.0165,
+      "step": 12924
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.58615221987315e-05,
+      "loss": 0.0203,
+      "step": 12926
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.5856236786469344e-05,
+      "loss": 0.0265,
+      "step": 12928
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.585095137420719e-05,
+      "loss": 0.0096,
+      "step": 12930
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.5845665961945032e-05,
+      "loss": 0.001,
+      "step": 12932
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.5840380549682875e-05,
+      "loss": 0.0594,
+      "step": 12934
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.5835095137420718e-05,
+      "loss": 0.0072,
+      "step": 12936
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.582980972515856e-05,
+      "loss": 0.0511,
+      "step": 12938
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.5824524312896406e-05,
+      "loss": 0.0159,
+      "step": 12940
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.581923890063425e-05,
+      "loss": 0.0259,
+      "step": 12942
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.5813953488372095e-05,
+      "loss": 0.1291,
+      "step": 12944
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.5808668076109938e-05,
+      "loss": 0.043,
+      "step": 12946
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.580338266384778e-05,
+      "loss": 0.0402,
+      "step": 12948
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.5798097251585627e-05,
+      "loss": 0.0065,
+      "step": 12950
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.579281183932347e-05,
+      "loss": 0.0221,
+      "step": 12952
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.5787526427061312e-05,
+      "loss": 0.0411,
+      "step": 12954
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.5782241014799155e-05,
+      "loss": 0.0816,
+      "step": 12956
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.5776955602537e-05,
+      "loss": 0.0357,
+      "step": 12958
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.5771670190274843e-05,
+      "loss": 0.0548,
+      "step": 12960
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.5766384778012686e-05,
+      "loss": 0.0553,
+      "step": 12962
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.576109936575053e-05,
+      "loss": 0.0188,
+      "step": 12964
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.5755813953488375e-05,
+      "loss": 0.066,
+      "step": 12966
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.5750528541226217e-05,
+      "loss": 0.0554,
+      "step": 12968
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.574524312896406e-05,
+      "loss": 0.0076,
+      "step": 12970
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.5739957716701903e-05,
+      "loss": 0.0499,
+      "step": 12972
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.5734672304439745e-05,
+      "loss": 0.0239,
+      "step": 12974
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 1.572938689217759e-05,
+      "loss": 0.009,
+      "step": 12976
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 1.5724101479915434e-05,
+      "loss": 0.0518,
+      "step": 12978
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 1.5718816067653277e-05,
+      "loss": 0.0334,
+      "step": 12980
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 1.571353065539112e-05,
+      "loss": 0.0415,
+      "step": 12982
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.5708245243128966e-05,
+      "loss": 0.0461,
+      "step": 12984
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.5702959830866808e-05,
+      "loss": 0.0394,
+      "step": 12986
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.569767441860465e-05,
+      "loss": 0.0133,
+      "step": 12988
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.5692389006342494e-05,
+      "loss": 0.0237,
+      "step": 12990
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.5687103594080336e-05,
+      "loss": 0.0313,
+      "step": 12992
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.5681818181818182e-05,
+      "loss": 0.0376,
+      "step": 12994
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.5676532769556025e-05,
+      "loss": 0.0207,
+      "step": 12996
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.567124735729387e-05,
+      "loss": 0.0201,
+      "step": 12998
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.5665961945031714e-05,
+      "loss": 0.0495,
+      "step": 13000
+    },
+    {
+      "epoch": 13.74,
+      "eval_cer": 0.04098033627814192,
+      "eval_loss": 0.49177786707878113,
+      "eval_runtime": 126.4592,
+      "eval_samples_per_second": 6.65,
+      "eval_steps_per_second": 0.838,
+      "step": 13000
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.566067653276956e-05,
+      "loss": 0.0182,
+      "step": 13002
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 1.5655391120507402e-05,
+      "loss": 0.0455,
+      "step": 13004
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 1.5650105708245245e-05,
+      "loss": 0.0239,
+      "step": 13006
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 1.5644820295983088e-05,
+      "loss": 0.0461,
+      "step": 13008
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 1.563953488372093e-05,
+      "loss": 0.0214,
+      "step": 13010
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 1.5634249471458776e-05,
+      "loss": 0.0747,
+      "step": 13012
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.562896405919662e-05,
+      "loss": 0.0303,
+      "step": 13014
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.5623678646934462e-05,
+      "loss": 0.0234,
+      "step": 13016
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.5618393234672305e-05,
+      "loss": 0.0324,
+      "step": 13018
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.5613107822410147e-05,
+      "loss": 0.0018,
+      "step": 13020
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.5607822410147993e-05,
+      "loss": 0.0218,
+      "step": 13022
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.5602536997885836e-05,
+      "loss": 0.0367,
+      "step": 13024
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.559725158562368e-05,
+      "loss": 0.0582,
+      "step": 13026
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.559196617336152e-05,
+      "loss": 0.0355,
+      "step": 13028
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.5586680761099367e-05,
+      "loss": 0.0429,
+      "step": 13030
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 1.558139534883721e-05,
+      "loss": 0.0299,
+      "step": 13032
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 1.5576109936575053e-05,
+      "loss": 0.0404,
+      "step": 13034
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 1.5570824524312895e-05,
+      "loss": 0.0406,
+      "step": 13036
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 1.5565539112050738e-05,
+      "loss": 0.0185,
+      "step": 13038
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 1.5560253699788584e-05,
+      "loss": 0.0204,
+      "step": 13040
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.5554968287526427e-05,
+      "loss": 0.0058,
+      "step": 13042
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.554968287526427e-05,
+      "loss": 0.0612,
+      "step": 13044
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.5544397463002115e-05,
+      "loss": 0.0198,
+      "step": 13046
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.5539112050739958e-05,
+      "loss": 0.0196,
+      "step": 13048
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.5533826638477804e-05,
+      "loss": 0.0012,
+      "step": 13050
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 1.5528541226215647e-05,
+      "loss": 0.029,
+      "step": 13052
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 1.552325581395349e-05,
+      "loss": 0.0228,
+      "step": 13054
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 1.5517970401691332e-05,
+      "loss": 0.0367,
+      "step": 13056
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 1.5512684989429178e-05,
+      "loss": 0.0608,
+      "step": 13058
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.550739957716702e-05,
+      "loss": 0.0569,
+      "step": 13060
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.5502114164904864e-05,
+      "loss": 0.0054,
+      "step": 13062
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.5496828752642706e-05,
+      "loss": 0.0158,
+      "step": 13064
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.5491543340380552e-05,
+      "loss": 0.0341,
+      "step": 13066
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.5486257928118395e-05,
+      "loss": 0.0202,
+      "step": 13068
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 1.5480972515856238e-05,
+      "loss": 0.0623,
+      "step": 13070
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 1.547568710359408e-05,
+      "loss": 0.0062,
+      "step": 13072
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 1.5470401691331923e-05,
+      "loss": 0.0139,
+      "step": 13074
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 1.546511627906977e-05,
+      "loss": 0.0107,
+      "step": 13076
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 1.5459830866807612e-05,
+      "loss": 0.0138,
+      "step": 13078
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.5454545454545454e-05,
+      "loss": 0.0398,
+      "step": 13080
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.5449260042283297e-05,
+      "loss": 0.0256,
+      "step": 13082
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.5443974630021143e-05,
+      "loss": 0.0062,
+      "step": 13084
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.5438689217758986e-05,
+      "loss": 0.0163,
+      "step": 13086
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 1.543340380549683e-05,
+      "loss": 0.0525,
+      "step": 13088
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 1.542811839323467e-05,
+      "loss": 0.1588,
+      "step": 13090
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 1.5422832980972514e-05,
+      "loss": 0.0386,
+      "step": 13092
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 1.541754756871036e-05,
+      "loss": 0.0213,
+      "step": 13094
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 1.5412262156448203e-05,
+      "loss": 0.0187,
+      "step": 13096
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.5406976744186045e-05,
+      "loss": 0.0402,
+      "step": 13098
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.540169133192389e-05,
+      "loss": 0.0738,
+      "step": 13100
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.5396405919661734e-05,
+      "loss": 0.0397,
+      "step": 13102
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.539112050739958e-05,
+      "loss": 0.0111,
+      "step": 13104
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.5385835095137423e-05,
+      "loss": 0.0102,
+      "step": 13106
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.5380549682875265e-05,
+      "loss": 0.089,
+      "step": 13108
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.5375264270613108e-05,
+      "loss": 0.0461,
+      "step": 13110
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.5369978858350954e-05,
+      "loss": 0.0456,
+      "step": 13112
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.5364693446088797e-05,
+      "loss": 0.0304,
+      "step": 13114
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.535940803382664e-05,
+      "loss": 0.0566,
+      "step": 13116
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.5354122621564482e-05,
+      "loss": 0.0024,
+      "step": 13118
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.5348837209302328e-05,
+      "loss": 0.0571,
+      "step": 13120
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.534355179704017e-05,
+      "loss": 0.0012,
+      "step": 13122
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.5338266384778014e-05,
+      "loss": 0.0618,
+      "step": 13124
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 1.5332980972515856e-05,
+      "loss": 0.035,
+      "step": 13126
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 1.53276955602537e-05,
+      "loss": 0.0343,
+      "step": 13128
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 1.5322410147991545e-05,
+      "loss": 0.0398,
+      "step": 13130
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 1.5317124735729388e-05,
+      "loss": 0.1157,
+      "step": 13132
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 1.531183932346723e-05,
+      "loss": 0.0126,
+      "step": 13134
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 1.5306553911205073e-05,
+      "loss": 0.0197,
+      "step": 13136
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 1.530126849894292e-05,
+      "loss": 0.0207,
+      "step": 13138
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 1.529598308668076e-05,
+      "loss": 0.0249,
+      "step": 13140
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 1.5290697674418604e-05,
+      "loss": 0.0116,
+      "step": 13142
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 1.5285412262156447e-05,
+      "loss": 0.0212,
+      "step": 13144
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.528012684989429e-05,
+      "loss": 0.0302,
+      "step": 13146
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.5274841437632136e-05,
+      "loss": 0.0309,
+      "step": 13148
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.526955602536998e-05,
+      "loss": 0.0287,
+      "step": 13150
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.5264270613107824e-05,
+      "loss": 0.0315,
+      "step": 13152
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.5258985200845667e-05,
+      "loss": 0.0123,
+      "step": 13154
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 1.5253699788583512e-05,
+      "loss": 0.0107,
+      "step": 13156
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 1.5248414376321354e-05,
+      "loss": 0.0162,
+      "step": 13158
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 1.5243128964059197e-05,
+      "loss": 0.0362,
+      "step": 13160
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 1.523784355179704e-05,
+      "loss": 0.0339,
+      "step": 13162
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.5232558139534884e-05,
+      "loss": 0.0635,
+      "step": 13164
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.5227272727272728e-05,
+      "loss": 0.024,
+      "step": 13166
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.5221987315010573e-05,
+      "loss": 0.023,
+      "step": 13168
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.5216701902748415e-05,
+      "loss": 0.0562,
+      "step": 13170
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.5211416490486258e-05,
+      "loss": 0.0143,
+      "step": 13172
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 1.5206131078224104e-05,
+      "loss": 0.0143,
+      "step": 13174
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 1.5200845665961947e-05,
+      "loss": 0.0185,
+      "step": 13176
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 1.519556025369979e-05,
+      "loss": 0.0204,
+      "step": 13178
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 1.5190274841437632e-05,
+      "loss": 0.0322,
+      "step": 13180
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 1.5184989429175475e-05,
+      "loss": 0.047,
+      "step": 13182
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.517970401691332e-05,
+      "loss": 0.0296,
+      "step": 13184
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.5174418604651163e-05,
+      "loss": 0.0115,
+      "step": 13186
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.5169133192389006e-05,
+      "loss": 0.0086,
+      "step": 13188
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.516384778012685e-05,
+      "loss": 0.0098,
+      "step": 13190
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 1.5158562367864695e-05,
+      "loss": 0.0352,
+      "step": 13192
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 1.5153276955602538e-05,
+      "loss": 0.003,
+      "step": 13194
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 1.5147991543340382e-05,
+      "loss": 0.0179,
+      "step": 13196
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 1.5142706131078225e-05,
+      "loss": 0.0103,
+      "step": 13198
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 1.5137420718816067e-05,
+      "loss": 0.0212,
+      "step": 13200
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.5132135306553913e-05,
+      "loss": 0.0327,
+      "step": 13202
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.5126849894291756e-05,
+      "loss": 0.0174,
+      "step": 13204
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.5121564482029599e-05,
+      "loss": 0.0536,
+      "step": 13206
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.5116279069767441e-05,
+      "loss": 0.0491,
+      "step": 13208
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.5110993657505287e-05,
+      "loss": 0.0246,
+      "step": 13210
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.510570824524313e-05,
+      "loss": 0.0733,
+      "step": 13212
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.5100422832980973e-05,
+      "loss": 0.005,
+      "step": 13214
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.5095137420718815e-05,
+      "loss": 0.0153,
+      "step": 13216
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.508985200845666e-05,
+      "loss": 0.0374,
+      "step": 13218
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.5084566596194504e-05,
+      "loss": 0.035,
+      "step": 13220
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 1.5079281183932348e-05,
+      "loss": 0.0276,
+      "step": 13222
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 1.5073995771670191e-05,
+      "loss": 0.0337,
+      "step": 13224
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 1.5068710359408034e-05,
+      "loss": 0.0762,
+      "step": 13226
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 1.506342494714588e-05,
+      "loss": 0.0072,
+      "step": 13228
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.5058139534883723e-05,
+      "loss": 0.0254,
+      "step": 13230
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.5052854122621565e-05,
+      "loss": 0.0068,
+      "step": 13232
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.5047568710359408e-05,
+      "loss": 0.018,
+      "step": 13234
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.504228329809725e-05,
+      "loss": 0.0269,
+      "step": 13236
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.5036997885835097e-05,
+      "loss": 0.054,
+      "step": 13238
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.503171247357294e-05,
+      "loss": 0.0689,
+      "step": 13240
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.5026427061310782e-05,
+      "loss": 0.0259,
+      "step": 13242
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.5021141649048626e-05,
+      "loss": 0.0249,
+      "step": 13244
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.501585623678647e-05,
+      "loss": 0.0192,
+      "step": 13246
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.5010570824524315e-05,
+      "loss": 0.0105,
+      "step": 13248
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.5005285412262158e-05,
+      "loss": 0.0108,
+      "step": 13250
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.5e-05,
+      "loss": 0.0063,
+      "step": 13252
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.4994714587737843e-05,
+      "loss": 0.0109,
+      "step": 13254
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.4989429175475689e-05,
+      "loss": 0.0026,
+      "step": 13256
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.4984143763213532e-05,
+      "loss": 0.0211,
+      "step": 13258
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 1.4978858350951374e-05,
+      "loss": 0.0473,
+      "step": 13260
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 1.4973572938689217e-05,
+      "loss": 0.0033,
+      "step": 13262
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 1.4968287526427063e-05,
+      "loss": 0.0232,
+      "step": 13264
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 1.4963002114164906e-05,
+      "loss": 0.0317,
+      "step": 13266
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.4957716701902749e-05,
+      "loss": 0.0207,
+      "step": 13268
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.4952431289640593e-05,
+      "loss": 0.0295,
+      "step": 13270
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.4947145877378436e-05,
+      "loss": 0.0027,
+      "step": 13272
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.4941860465116282e-05,
+      "loss": 0.0149,
+      "step": 13274
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.4936575052854124e-05,
+      "loss": 0.0253,
+      "step": 13276
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.4931289640591967e-05,
+      "loss": 0.0139,
+      "step": 13278
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.492600422832981e-05,
+      "loss": 0.0864,
+      "step": 13280
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.4920718816067656e-05,
+      "loss": 0.0691,
+      "step": 13282
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.4915433403805498e-05,
+      "loss": 0.015,
+      "step": 13284
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.4910147991543341e-05,
+      "loss": 0.0429,
+      "step": 13286
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.4904862579281184e-05,
+      "loss": 0.0207,
+      "step": 13288
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.4899577167019026e-05,
+      "loss": 0.0256,
+      "step": 13290
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.4894291754756872e-05,
+      "loss": 0.0107,
+      "step": 13292
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.4889006342494715e-05,
+      "loss": 0.0192,
+      "step": 13294
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.488372093023256e-05,
+      "loss": 0.0007,
+      "step": 13296
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.4878435517970402e-05,
+      "loss": 0.087,
+      "step": 13298
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.4873150105708247e-05,
+      "loss": 0.003,
+      "step": 13300
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.4867864693446091e-05,
+      "loss": 0.0655,
+      "step": 13302
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.4862579281183934e-05,
+      "loss": 0.0009,
+      "step": 13304
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.4857293868921776e-05,
+      "loss": 0.0076,
+      "step": 13306
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.4852008456659619e-05,
+      "loss": 0.0231,
+      "step": 13308
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.4846723044397465e-05,
+      "loss": 0.0266,
+      "step": 13310
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.4841437632135308e-05,
+      "loss": 0.0252,
+      "step": 13312
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.483615221987315e-05,
+      "loss": 0.0915,
+      "step": 13314
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.4830866807610993e-05,
+      "loss": 0.0135,
+      "step": 13316
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.4825581395348839e-05,
+      "loss": 0.019,
+      "step": 13318
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.4820295983086682e-05,
+      "loss": 0.0622,
+      "step": 13320
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.4815010570824524e-05,
+      "loss": 0.0025,
+      "step": 13322
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.4809725158562369e-05,
+      "loss": 0.0052,
+      "step": 13324
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.4804439746300211e-05,
+      "loss": 0.0065,
+      "step": 13326
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.4799154334038057e-05,
+      "loss": 0.063,
+      "step": 13328
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.47938689217759e-05,
+      "loss": 0.0188,
+      "step": 13330
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.4788583509513743e-05,
+      "loss": 0.0182,
+      "step": 13332
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.4783298097251585e-05,
+      "loss": 0.021,
+      "step": 13334
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.4778012684989432e-05,
+      "loss": 0.0099,
+      "step": 13336
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.4772727272727274e-05,
+      "loss": 0.0057,
+      "step": 13338
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.4767441860465117e-05,
+      "loss": 0.0193,
+      "step": 13340
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.476215644820296e-05,
+      "loss": 0.0487,
+      "step": 13342
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.4756871035940802e-05,
+      "loss": 0.0194,
+      "step": 13344
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.4751585623678648e-05,
+      "loss": 0.0289,
+      "step": 13346
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.4746300211416491e-05,
+      "loss": 0.0224,
+      "step": 13348
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.4741014799154335e-05,
+      "loss": 0.0137,
+      "step": 13350
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.4735729386892178e-05,
+      "loss": 0.0165,
+      "step": 13352
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.473044397463002e-05,
+      "loss": 0.0291,
+      "step": 13354
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.4725158562367867e-05,
+      "loss": 0.0642,
+      "step": 13356
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.471987315010571e-05,
+      "loss": 0.0494,
+      "step": 13358
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.4714587737843552e-05,
+      "loss": 0.0077,
+      "step": 13360
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.4709302325581395e-05,
+      "loss": 0.0222,
+      "step": 13362
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.470401691331924e-05,
+      "loss": 0.0185,
+      "step": 13364
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.4698731501057083e-05,
+      "loss": 0.0556,
+      "step": 13366
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.4693446088794926e-05,
+      "loss": 0.0324,
+      "step": 13368
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.4688160676532769e-05,
+      "loss": 0.0105,
+      "step": 13370
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.4682875264270613e-05,
+      "loss": 0.0087,
+      "step": 13372
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.4677589852008458e-05,
+      "loss": 0.0375,
+      "step": 13374
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.4672304439746302e-05,
+      "loss": 0.0014,
+      "step": 13376
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.4667019027484145e-05,
+      "loss": 0.019,
+      "step": 13378
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.4661733615221987e-05,
+      "loss": 0.0439,
+      "step": 13380
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 1.4656448202959833e-05,
+      "loss": 0.017,
+      "step": 13382
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 1.4651162790697676e-05,
+      "loss": 0.0047,
+      "step": 13384
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 1.4645877378435519e-05,
+      "loss": 0.0519,
+      "step": 13386
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 1.4640591966173361e-05,
+      "loss": 0.0107,
+      "step": 13388
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 1.4635306553911204e-05,
+      "loss": 0.0064,
+      "step": 13390
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.463002114164905e-05,
+      "loss": 0.0832,
+      "step": 13392
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.4624735729386893e-05,
+      "loss": 0.0079,
+      "step": 13394
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.4619450317124735e-05,
+      "loss": 0.0031,
+      "step": 13396
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.461416490486258e-05,
+      "loss": 0.0436,
+      "step": 13398
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.4608879492600424e-05,
+      "loss": 0.0132,
+      "step": 13400
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.4603594080338267e-05,
+      "loss": 0.0417,
+      "step": 13402
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.4598308668076111e-05,
+      "loss": 0.0505,
+      "step": 13404
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.4593023255813954e-05,
+      "loss": 0.0369,
+      "step": 13406
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.4587737843551796e-05,
+      "loss": 0.072,
+      "step": 13408
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 1.4582452431289643e-05,
+      "loss": 0.0264,
+      "step": 13410
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 1.4577167019027485e-05,
+      "loss": 0.0184,
+      "step": 13412
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 1.4571881606765328e-05,
+      "loss": 0.0311,
+      "step": 13414
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 1.456659619450317e-05,
+      "loss": 0.0128,
+      "step": 13416
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 1.4561310782241017e-05,
+      "loss": 0.0044,
+      "step": 13418
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.455602536997886e-05,
+      "loss": 0.007,
+      "step": 13420
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.4550739957716702e-05,
+      "loss": 0.0169,
+      "step": 13422
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.4545454545454545e-05,
+      "loss": 0.0117,
+      "step": 13424
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.4540169133192389e-05,
+      "loss": 0.0298,
+      "step": 13426
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.4534883720930233e-05,
+      "loss": 0.0403,
+      "step": 13428
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 1.4529598308668078e-05,
+      "loss": 0.0188,
+      "step": 13430
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 1.452431289640592e-05,
+      "loss": 0.0029,
+      "step": 13432
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 1.4519027484143763e-05,
+      "loss": 0.0201,
+      "step": 13434
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 1.4513742071881609e-05,
+      "loss": 0.0215,
+      "step": 13436
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.4508456659619452e-05,
+      "loss": 0.0227,
+      "step": 13438
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.4503171247357294e-05,
+      "loss": 0.0074,
+      "step": 13440
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.4497885835095137e-05,
+      "loss": 0.0111,
+      "step": 13442
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.449260042283298e-05,
+      "loss": 0.0031,
+      "step": 13444
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.4487315010570826e-05,
+      "loss": 0.0568,
+      "step": 13446
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.4482029598308669e-05,
+      "loss": 0.0475,
+      "step": 13448
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.4476744186046511e-05,
+      "loss": 0.0066,
+      "step": 13450
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.4471458773784356e-05,
+      "loss": 0.0605,
+      "step": 13452
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.44661733615222e-05,
+      "loss": 0.0407,
+      "step": 13454
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.4460887949260044e-05,
+      "loss": 0.0279,
+      "step": 13456
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.4455602536997887e-05,
+      "loss": 0.009,
+      "step": 13458
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.445031712473573e-05,
+      "loss": 0.0206,
+      "step": 13460
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.4445031712473572e-05,
+      "loss": 0.1183,
+      "step": 13462
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.4439746300211418e-05,
+      "loss": 0.039,
+      "step": 13464
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.4434460887949261e-05,
+      "loss": 0.0591,
+      "step": 13466
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.4429175475687104e-05,
+      "loss": 0.0715,
+      "step": 13468
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.4423890063424946e-05,
+      "loss": 0.024,
+      "step": 13470
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.4418604651162792e-05,
+      "loss": 0.0236,
+      "step": 13472
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.4413319238900635e-05,
+      "loss": 0.0232,
+      "step": 13474
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 1.4408033826638478e-05,
+      "loss": 0.0394,
+      "step": 13476
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 1.4402748414376322e-05,
+      "loss": 0.0025,
+      "step": 13478
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 1.4397463002114165e-05,
+      "loss": 0.0709,
+      "step": 13480
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 1.439217758985201e-05,
+      "loss": 0.0878,
+      "step": 13482
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 1.4386892177589854e-05,
+      "loss": 0.008,
+      "step": 13484
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.4381606765327696e-05,
+      "loss": 0.02,
+      "step": 13486
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.4376321353065539e-05,
+      "loss": 0.0175,
+      "step": 13488
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.4371035940803385e-05,
+      "loss": 0.018,
+      "step": 13490
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.4365750528541228e-05,
+      "loss": 0.0914,
+      "step": 13492
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.436046511627907e-05,
+      "loss": 0.0175,
+      "step": 13494
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.4355179704016913e-05,
+      "loss": 0.0415,
+      "step": 13496
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.4349894291754756e-05,
+      "loss": 0.0276,
+      "step": 13498
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.4344608879492602e-05,
+      "loss": 0.0012,
+      "step": 13500
+    },
+    {
+      "epoch": 14.27,
+      "eval_cer": 0.09563978341407808,
+      "eval_loss": 1.003995656967163,
+      "eval_runtime": 120.9517,
+      "eval_samples_per_second": 6.953,
+      "eval_steps_per_second": 0.876,
+      "step": 13500
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.4339323467230444e-05,
+      "loss": 0.1183,
+      "step": 13502
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.4334038054968287e-05,
+      "loss": 0.062,
+      "step": 13504
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.4328752642706131e-05,
+      "loss": 0.0552,
+      "step": 13506
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.4323467230443976e-05,
+      "loss": 0.0046,
+      "step": 13508
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.431818181818182e-05,
+      "loss": 0.0243,
+      "step": 13510
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.4312896405919663e-05,
+      "loss": 0.1092,
+      "step": 13512
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.4307610993657506e-05,
+      "loss": 0.1191,
+      "step": 13514
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.4302325581395348e-05,
+      "loss": 0.0201,
+      "step": 13516
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.4297040169133194e-05,
+      "loss": 0.0286,
+      "step": 13518
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.4291754756871037e-05,
+      "loss": 0.0275,
+      "step": 13520
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.428646934460888e-05,
+      "loss": 0.0763,
+      "step": 13522
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.4281183932346722e-05,
+      "loss": 0.1111,
+      "step": 13524
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.4275898520084568e-05,
+      "loss": 0.0027,
+      "step": 13526
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.4270613107822411e-05,
+      "loss": 0.0311,
+      "step": 13528
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.4265327695560254e-05,
+      "loss": 0.0032,
+      "step": 13530
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.4260042283298098e-05,
+      "loss": 0.0262,
+      "step": 13532
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 1.425475687103594e-05,
+      "loss": 0.0174,
+      "step": 13534
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 1.4249471458773787e-05,
+      "loss": 0.0109,
+      "step": 13536
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 1.424418604651163e-05,
+      "loss": 0.0691,
+      "step": 13538
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 1.4238900634249472e-05,
+      "loss": 0.1051,
+      "step": 13540
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.4233615221987315e-05,
+      "loss": 0.1075,
+      "step": 13542
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.422832980972516e-05,
+      "loss": 0.0403,
+      "step": 13544
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.4223044397463003e-05,
+      "loss": 0.0249,
+      "step": 13546
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.4217758985200846e-05,
+      "loss": 0.0125,
+      "step": 13548
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.4212473572938689e-05,
+      "loss": 0.0637,
+      "step": 13550
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 1.4207188160676532e-05,
+      "loss": 0.0158,
+      "step": 13552
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 1.4201902748414378e-05,
+      "loss": 0.0537,
+      "step": 13554
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 1.419661733615222e-05,
+      "loss": 0.027,
+      "step": 13556
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 1.4191331923890065e-05,
+      "loss": 0.0184,
+      "step": 13558
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 1.4186046511627907e-05,
+      "loss": 0.0254,
+      "step": 13560
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.4180761099365752e-05,
+      "loss": 0.0231,
+      "step": 13562
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.4175475687103596e-05,
+      "loss": 0.0292,
+      "step": 13564
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.4170190274841439e-05,
+      "loss": 0.0091,
+      "step": 13566
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.4164904862579281e-05,
+      "loss": 0.0139,
+      "step": 13568
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.4159619450317124e-05,
+      "loss": 0.0491,
+      "step": 13570
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 1.415433403805497e-05,
+      "loss": 0.0261,
+      "step": 13572
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 1.4149048625792813e-05,
+      "loss": 0.0122,
+      "step": 13574
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 1.4143763213530655e-05,
+      "loss": 0.0122,
+      "step": 13576
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 1.4138477801268498e-05,
+      "loss": 0.049,
+      "step": 13578
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 1.4133192389006344e-05,
+      "loss": 0.0109,
+      "step": 13580
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 1.4127906976744187e-05,
+      "loss": 0.1036,
+      "step": 13582
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 1.412262156448203e-05,
+      "loss": 0.0188,
+      "step": 13584
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 1.4117336152219874e-05,
+      "loss": 0.0124,
+      "step": 13586
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 1.4112050739957717e-05,
+      "loss": 0.0425,
+      "step": 13588
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.4106765327695563e-05,
+      "loss": 0.0123,
+      "step": 13590
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.4101479915433405e-05,
+      "loss": 0.0958,
+      "step": 13592
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.4096194503171248e-05,
+      "loss": 0.0522,
+      "step": 13594
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.409090909090909e-05,
+      "loss": 0.0211,
+      "step": 13596
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.4085623678646937e-05,
+      "loss": 0.0055,
+      "step": 13598
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 1.408033826638478e-05,
+      "loss": 0.0014,
+      "step": 13600
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 1.4075052854122622e-05,
+      "loss": 0.0414,
+      "step": 13602
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 1.4069767441860465e-05,
+      "loss": 0.0068,
+      "step": 13604
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 1.4064482029598307e-05,
+      "loss": 0.0163,
+      "step": 13606
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 1.4059196617336153e-05,
+      "loss": 0.0078,
+      "step": 13608
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 1.4053911205073996e-05,
+      "loss": 0.0214,
+      "step": 13610
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 1.404862579281184e-05,
+      "loss": 0.0221,
+      "step": 13612
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 1.4043340380549683e-05,
+      "loss": 0.0048,
+      "step": 13614
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 1.403805496828753e-05,
+      "loss": 0.0113,
+      "step": 13616
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 1.4032769556025372e-05,
+      "loss": 0.0915,
+      "step": 13618
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 1.4027484143763215e-05,
+      "loss": 0.0354,
+      "step": 13620
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 1.4022198731501057e-05,
+      "loss": 0.0057,
+      "step": 13622
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 1.40169133192389e-05,
+      "loss": 0.003,
+      "step": 13624
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 1.4011627906976746e-05,
+      "loss": 0.0131,
+      "step": 13626
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.4006342494714589e-05,
+      "loss": 0.0242,
+      "step": 13628
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.4001057082452431e-05,
+      "loss": 0.018,
+      "step": 13630
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.3995771670190274e-05,
+      "loss": 0.0172,
+      "step": 13632
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.399048625792812e-05,
+      "loss": 0.0313,
+      "step": 13634
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.3985200845665963e-05,
+      "loss": 0.0219,
+      "step": 13636
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 1.3979915433403807e-05,
+      "loss": 0.0044,
+      "step": 13638
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 1.397463002114165e-05,
+      "loss": 0.0299,
+      "step": 13640
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 1.3969344608879492e-05,
+      "loss": 0.0146,
+      "step": 13642
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 1.3964059196617338e-05,
+      "loss": 0.0114,
+      "step": 13644
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 1.3958773784355181e-05,
+      "loss": 0.1177,
+      "step": 13646
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.3953488372093024e-05,
+      "loss": 0.0769,
+      "step": 13648
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.3948202959830866e-05,
+      "loss": 0.0479,
+      "step": 13650
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.3942917547568712e-05,
+      "loss": 0.0186,
+      "step": 13652
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.3937632135306555e-05,
+      "loss": 0.0156,
+      "step": 13654
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.3932346723044398e-05,
+      "loss": 0.0296,
+      "step": 13656
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.392706131078224e-05,
+      "loss": 0.0376,
+      "step": 13658
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.3921775898520085e-05,
+      "loss": 0.0162,
+      "step": 13660
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.391649048625793e-05,
+      "loss": 0.02,
+      "step": 13662
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.3911205073995774e-05,
+      "loss": 0.0107,
+      "step": 13664
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.3905919661733616e-05,
+      "loss": 0.0131,
+      "step": 13666
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.3900634249471459e-05,
+      "loss": 0.0253,
+      "step": 13668
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.3895348837209305e-05,
+      "loss": 0.0241,
+      "step": 13670
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.3890063424947148e-05,
+      "loss": 0.0411,
+      "step": 13672
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.388477801268499e-05,
+      "loss": 0.0222,
+      "step": 13674
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.3879492600422833e-05,
+      "loss": 0.0701,
+      "step": 13676
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.3874207188160676e-05,
+      "loss": 0.0616,
+      "step": 13678
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.3868921775898522e-05,
+      "loss": 0.0431,
+      "step": 13680
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.3863636363636364e-05,
+      "loss": 0.0509,
+      "step": 13682
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.3858350951374207e-05,
+      "loss": 0.0369,
+      "step": 13684
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.3853065539112051e-05,
+      "loss": 0.0113,
+      "step": 13686
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.3847780126849896e-05,
+      "loss": 0.051,
+      "step": 13688
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.3842494714587738e-05,
+      "loss": 0.0501,
+      "step": 13690
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.3837209302325583e-05,
+      "loss": 0.0243,
+      "step": 13692
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.3831923890063426e-05,
+      "loss": 0.0398,
+      "step": 13694
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.3826638477801268e-05,
+      "loss": 0.0111,
+      "step": 13696
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.3821353065539114e-05,
+      "loss": 0.0178,
+      "step": 13698
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.3816067653276957e-05,
+      "loss": 0.0069,
+      "step": 13700
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.38107822410148e-05,
+      "loss": 0.0212,
+      "step": 13702
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.3805496828752642e-05,
+      "loss": 0.0212,
+      "step": 13704
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.3800211416490485e-05,
+      "loss": 0.0104,
+      "step": 13706
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.3794926004228331e-05,
+      "loss": 0.0028,
+      "step": 13708
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.3789640591966174e-05,
+      "loss": 0.0084,
+      "step": 13710
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.3784355179704016e-05,
+      "loss": 0.0075,
+      "step": 13712
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.377906976744186e-05,
+      "loss": 0.0149,
+      "step": 13714
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.3773784355179705e-05,
+      "loss": 0.0051,
+      "step": 13716
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.376849894291755e-05,
+      "loss": 0.0321,
+      "step": 13718
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.3763213530655392e-05,
+      "loss": 0.0161,
+      "step": 13720
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.3757928118393235e-05,
+      "loss": 0.0121,
+      "step": 13722
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.3752642706131077e-05,
+      "loss": 0.0102,
+      "step": 13724
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.3747357293868924e-05,
+      "loss": 0.005,
+      "step": 13726
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.3742071881606766e-05,
+      "loss": 0.0073,
+      "step": 13728
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.3736786469344609e-05,
+      "loss": 0.0379,
+      "step": 13730
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 1.3731501057082452e-05,
+      "loss": 0.0563,
+      "step": 13732
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 1.3726215644820298e-05,
+      "loss": 0.0341,
+      "step": 13734
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 1.372093023255814e-05,
+      "loss": 0.0172,
+      "step": 13736
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 1.3715644820295983e-05,
+      "loss": 0.0016,
+      "step": 13738
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 1.3710359408033827e-05,
+      "loss": 0.0094,
+      "step": 13740
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.370507399577167e-05,
+      "loss": 0.0734,
+      "step": 13742
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.3699788583509516e-05,
+      "loss": 0.0563,
+      "step": 13744
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.3694503171247359e-05,
+      "loss": 0.0231,
+      "step": 13746
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.3689217758985201e-05,
+      "loss": 0.0068,
+      "step": 13748
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.3683932346723044e-05,
+      "loss": 0.0094,
+      "step": 13750
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.367864693446089e-05,
+      "loss": 0.0178,
+      "step": 13752
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.3673361522198733e-05,
+      "loss": 0.0592,
+      "step": 13754
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.3668076109936575e-05,
+      "loss": 0.0183,
+      "step": 13756
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.3662790697674418e-05,
+      "loss": 0.0065,
+      "step": 13758
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.365750528541226e-05,
+      "loss": 0.0021,
+      "step": 13760
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.3652219873150107e-05,
+      "loss": 0.0198,
+      "step": 13762
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.364693446088795e-05,
+      "loss": 0.0197,
+      "step": 13764
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.3641649048625794e-05,
+      "loss": 0.0105,
+      "step": 13766
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 0.0388,
+      "step": 13768
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.3631078224101481e-05,
+      "loss": 0.0011,
+      "step": 13770
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.3625792811839325e-05,
+      "loss": 0.0006,
+      "step": 13772
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.3620507399577168e-05,
+      "loss": 0.0233,
+      "step": 13774
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.361522198731501e-05,
+      "loss": 0.0077,
+      "step": 13776
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.3609936575052853e-05,
+      "loss": 0.003,
+      "step": 13778
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.36046511627907e-05,
+      "loss": 0.0128,
+      "step": 13780
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.3599365750528542e-05,
+      "loss": 0.0146,
+      "step": 13782
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.3594080338266385e-05,
+      "loss": 0.0699,
+      "step": 13784
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.3588794926004227e-05,
+      "loss": 0.0145,
+      "step": 13786
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 1.3583509513742073e-05,
+      "loss": 0.0151,
+      "step": 13788
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 1.3578224101479916e-05,
+      "loss": 0.039,
+      "step": 13790
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 1.3572938689217759e-05,
+      "loss": 0.0277,
+      "step": 13792
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 1.3567653276955603e-05,
+      "loss": 0.0039,
+      "step": 13794
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 1.3562367864693446e-05,
+      "loss": 0.0024,
+      "step": 13796
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.3557082452431292e-05,
+      "loss": 0.0033,
+      "step": 13798
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.3551797040169135e-05,
+      "loss": 0.0173,
+      "step": 13800
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.3546511627906977e-05,
+      "loss": 0.0648,
+      "step": 13802
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.354122621564482e-05,
+      "loss": 0.0049,
+      "step": 13804
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.3535940803382666e-05,
+      "loss": 0.0268,
+      "step": 13806
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.3530655391120509e-05,
+      "loss": 0.0133,
+      "step": 13808
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.3525369978858351e-05,
+      "loss": 0.0039,
+      "step": 13810
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.3520084566596194e-05,
+      "loss": 0.0697,
+      "step": 13812
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.3514799154334037e-05,
+      "loss": 0.0069,
+      "step": 13814
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.3509513742071883e-05,
+      "loss": 0.0679,
+      "step": 13816
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.3504228329809725e-05,
+      "loss": 0.0769,
+      "step": 13818
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.349894291754757e-05,
+      "loss": 0.0052,
+      "step": 13820
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.3493657505285412e-05,
+      "loss": 0.0194,
+      "step": 13822
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.3488372093023258e-05,
+      "loss": 0.004,
+      "step": 13824
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.3483086680761101e-05,
+      "loss": 0.0115,
+      "step": 13826
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.3477801268498944e-05,
+      "loss": 0.0701,
+      "step": 13828
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.3472515856236786e-05,
+      "loss": 0.0325,
+      "step": 13830
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.3467230443974629e-05,
+      "loss": 0.0148,
+      "step": 13832
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.3461945031712475e-05,
+      "loss": 0.0282,
+      "step": 13834
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.3456659619450318e-05,
+      "loss": 0.0206,
+      "step": 13836
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.345137420718816e-05,
+      "loss": 0.0577,
+      "step": 13838
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.3446088794926003e-05,
+      "loss": 0.0582,
+      "step": 13840
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.344080338266385e-05,
+      "loss": 0.0143,
+      "step": 13842
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.3435517970401692e-05,
+      "loss": 0.0237,
+      "step": 13844
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.3430232558139536e-05,
+      "loss": 0.0128,
+      "step": 13846
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.3424947145877379e-05,
+      "loss": 0.0338,
+      "step": 13848
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.3419661733615222e-05,
+      "loss": 0.0459,
+      "step": 13850
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.3414376321353068e-05,
+      "loss": 0.0092,
+      "step": 13852
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.340909090909091e-05,
+      "loss": 0.069,
+      "step": 13854
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 1.3403805496828753e-05,
+      "loss": 0.0052,
+      "step": 13856
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 1.3398520084566596e-05,
+      "loss": 0.0128,
+      "step": 13858
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 1.3393234672304442e-05,
+      "loss": 0.0054,
+      "step": 13860
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 1.3387949260042284e-05,
+      "loss": 0.0318,
+      "step": 13862
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.3382663847780127e-05,
+      "loss": 0.0147,
+      "step": 13864
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.337737843551797e-05,
+      "loss": 0.0183,
+      "step": 13866
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.3372093023255814e-05,
+      "loss": 0.0115,
+      "step": 13868
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.3366807610993659e-05,
+      "loss": 0.0046,
+      "step": 13870
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.3361522198731501e-05,
+      "loss": 0.0153,
+      "step": 13872
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.3356236786469346e-05,
+      "loss": 0.0257,
+      "step": 13874
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.3350951374207188e-05,
+      "loss": 0.0091,
+      "step": 13876
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.3345665961945034e-05,
+      "loss": 0.0428,
+      "step": 13878
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.3340380549682877e-05,
+      "loss": 0.006,
+      "step": 13880
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.333509513742072e-05,
+      "loss": 0.0107,
+      "step": 13882
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.3329809725158562e-05,
+      "loss": 0.0074,
+      "step": 13884
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.3324524312896405e-05,
+      "loss": 0.0027,
+      "step": 13886
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.3319238900634251e-05,
+      "loss": 0.0179,
+      "step": 13888
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.3313953488372094e-05,
+      "loss": 0.0392,
+      "step": 13890
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.3308668076109936e-05,
+      "loss": 0.0578,
+      "step": 13892
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.3303382663847779e-05,
+      "loss": 0.0031,
+      "step": 13894
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.3298097251585625e-05,
+      "loss": 0.0141,
+      "step": 13896
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.3292811839323468e-05,
+      "loss": 0.023,
+      "step": 13898
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.3287526427061312e-05,
+      "loss": 0.0334,
+      "step": 13900
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.3282241014799155e-05,
+      "loss": 0.0172,
+      "step": 13902
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.3276955602536997e-05,
+      "loss": 0.0193,
+      "step": 13904
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.3271670190274844e-05,
+      "loss": 0.012,
+      "step": 13906
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.3266384778012686e-05,
+      "loss": 0.0029,
+      "step": 13908
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.3261099365750529e-05,
+      "loss": 0.0846,
+      "step": 13910
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.3255813953488372e-05,
+      "loss": 0.0243,
+      "step": 13912
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.3250528541226218e-05,
+      "loss": 0.0498,
+      "step": 13914
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.324524312896406e-05,
+      "loss": 0.0711,
+      "step": 13916
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.3239957716701903e-05,
+      "loss": 0.0096,
+      "step": 13918
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.3234672304439746e-05,
+      "loss": 0.01,
+      "step": 13920
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.322938689217759e-05,
+      "loss": 0.0635,
+      "step": 13922
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.3224101479915434e-05,
+      "loss": 0.004,
+      "step": 13924
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.3218816067653279e-05,
+      "loss": 0.0302,
+      "step": 13926
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.3213530655391121e-05,
+      "loss": 0.0196,
+      "step": 13928
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.3208245243128964e-05,
+      "loss": 0.0089,
+      "step": 13930
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.320295983086681e-05,
+      "loss": 0.0592,
+      "step": 13932
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.3197674418604653e-05,
+      "loss": 0.0242,
+      "step": 13934
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.3192389006342495e-05,
+      "loss": 0.0562,
+      "step": 13936
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.3187103594080338e-05,
+      "loss": 0.1009,
+      "step": 13938
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.318181818181818e-05,
+      "loss": 0.0292,
+      "step": 13940
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.3176532769556027e-05,
+      "loss": 0.0123,
+      "step": 13942
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.317124735729387e-05,
+      "loss": 0.002,
+      "step": 13944
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.3165961945031712e-05,
+      "loss": 0.0556,
+      "step": 13946
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.3160676532769557e-05,
+      "loss": 0.0065,
+      "step": 13948
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.3155391120507401e-05,
+      "loss": 0.0098,
+      "step": 13950
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.3150105708245244e-05,
+      "loss": 0.0191,
+      "step": 13952
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.3144820295983088e-05,
+      "loss": 0.0013,
+      "step": 13954
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.313953488372093e-05,
+      "loss": 0.0402,
+      "step": 13956
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.3134249471458773e-05,
+      "loss": 0.0177,
+      "step": 13958
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.312896405919662e-05,
+      "loss": 0.0181,
+      "step": 13960
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.3123678646934462e-05,
+      "loss": 0.0154,
+      "step": 13962
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.3118393234672305e-05,
+      "loss": 0.0194,
+      "step": 13964
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.3113107822410147e-05,
+      "loss": 0.0338,
+      "step": 13966
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.3107822410147993e-05,
+      "loss": 0.0014,
+      "step": 13968
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.3102536997885836e-05,
+      "loss": 0.0198,
+      "step": 13970
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.3097251585623679e-05,
+      "loss": 0.0695,
+      "step": 13972
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.3091966173361521e-05,
+      "loss": 0.0272,
+      "step": 13974
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.3086680761099366e-05,
+      "loss": 0.0626,
+      "step": 13976
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.308139534883721e-05,
+      "loss": 0.011,
+      "step": 13978
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.3076109936575055e-05,
+      "loss": 0.017,
+      "step": 13980
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.3070824524312897e-05,
+      "loss": 0.0119,
+      "step": 13982
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.306553911205074e-05,
+      "loss": 0.02,
+      "step": 13984
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.3060253699788586e-05,
+      "loss": 0.0353,
+      "step": 13986
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.3054968287526429e-05,
+      "loss": 0.039,
+      "step": 13988
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.3049682875264271e-05,
+      "loss": 0.0174,
+      "step": 13990
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.3044397463002114e-05,
+      "loss": 0.0433,
+      "step": 13992
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.3039112050739957e-05,
+      "loss": 0.0249,
+      "step": 13994
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.3033826638477803e-05,
+      "loss": 0.0049,
+      "step": 13996
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 1.3028541226215645e-05,
+      "loss": 0.0084,
+      "step": 13998
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 1.3023255813953488e-05,
+      "loss": 0.043,
+      "step": 14000
+    },
+    {
+      "epoch": 14.8,
+      "eval_cer": 0.11011684240524366,
+      "eval_loss": 0.7147426009178162,
+      "eval_runtime": 118.439,
+      "eval_samples_per_second": 7.101,
+      "eval_steps_per_second": 0.895,
+      "step": 14000
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 1.3017970401691332e-05,
+      "loss": 0.0128,
+      "step": 14002
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 1.3012684989429177e-05,
+      "loss": 0.0234,
+      "step": 14004
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 1.3007399577167021e-05,
+      "loss": 0.0215,
+      "step": 14006
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 1.3002114164904864e-05,
+      "loss": 0.0191,
+      "step": 14008
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 1.2996828752642706e-05,
+      "loss": 0.0338,
+      "step": 14010
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 1.299154334038055e-05,
+      "loss": 0.0133,
+      "step": 14012
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 1.2986257928118395e-05,
+      "loss": 0.0235,
+      "step": 14014
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 1.2980972515856238e-05,
+      "loss": 0.0398,
+      "step": 14016
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 1.297568710359408e-05,
+      "loss": 0.0036,
+      "step": 14018
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 1.2970401691331923e-05,
+      "loss": 0.0039,
+      "step": 14020
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 1.296511627906977e-05,
+      "loss": 0.031,
+      "step": 14022
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 1.2959830866807612e-05,
+      "loss": 0.0258,
+      "step": 14024
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 1.2954545454545455e-05,
+      "loss": 0.0334,
+      "step": 14026
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 1.2949260042283299e-05,
+      "loss": 0.025,
+      "step": 14028
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 1.2943974630021142e-05,
+      "loss": 0.0311,
+      "step": 14030
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 1.2938689217758988e-05,
+      "loss": 0.008,
+      "step": 14032
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 1.293340380549683e-05,
+      "loss": 0.0442,
+      "step": 14034
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 1.2928118393234673e-05,
+      "loss": 0.0072,
+      "step": 14036
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 1.2922832980972516e-05,
+      "loss": 0.0121,
+      "step": 14038
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 1.2917547568710358e-05,
+      "loss": 0.0383,
+      "step": 14040
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 1.2912262156448204e-05,
+      "loss": 0.0156,
+      "step": 14042
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 1.2906976744186047e-05,
+      "loss": 0.0085,
+      "step": 14044
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 1.290169133192389e-05,
+      "loss": 0.0195,
+      "step": 14046
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 1.2896405919661732e-05,
+      "loss": 0.027,
+      "step": 14048
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 1.2891120507399579e-05,
+      "loss": 0.0062,
+      "step": 14050
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 1.2885835095137421e-05,
+      "loss": 0.0284,
+      "step": 14052
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.2880549682875266e-05,
+      "loss": 0.0132,
+      "step": 14054
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.2875264270613108e-05,
+      "loss": 0.0049,
+      "step": 14056
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.2869978858350951e-05,
+      "loss": 0.1056,
+      "step": 14058
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.2864693446088797e-05,
+      "loss": 0.0352,
+      "step": 14060
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.285940803382664e-05,
+      "loss": 0.0039,
+      "step": 14062
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 1.2854122621564482e-05,
+      "loss": 0.0626,
+      "step": 14064
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 1.2848837209302325e-05,
+      "loss": 0.0113,
+      "step": 14066
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 1.2843551797040171e-05,
+      "loss": 0.0759,
+      "step": 14068
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 1.2838266384778014e-05,
+      "loss": 0.045,
+      "step": 14070
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 1.2832980972515856e-05,
+      "loss": 0.0482,
+      "step": 14072
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 1.2827695560253699e-05,
+      "loss": 0.0117,
+      "step": 14074
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 1.2822410147991543e-05,
+      "loss": 0.0192,
+      "step": 14076
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 1.2817124735729388e-05,
+      "loss": 0.0201,
+      "step": 14078
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 1.281183932346723e-05,
+      "loss": 0.0157,
+      "step": 14080
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 1.2806553911205075e-05,
+      "loss": 0.0105,
+      "step": 14082
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 1.2801268498942918e-05,
+      "loss": 0.0715,
+      "step": 14084
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 1.2795983086680764e-05,
+      "loss": 0.0033,
+      "step": 14086
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 1.2790697674418606e-05,
+      "loss": 0.0399,
+      "step": 14088
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 1.2785412262156449e-05,
+      "loss": 0.0123,
+      "step": 14090
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 1.2780126849894292e-05,
+      "loss": 0.0195,
+      "step": 14092
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 1.2774841437632134e-05,
+      "loss": 0.0414,
+      "step": 14094
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 1.276955602536998e-05,
+      "loss": 0.0498,
+      "step": 14096
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 1.2764270613107823e-05,
+      "loss": 0.0189,
+      "step": 14098
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 1.2758985200845666e-05,
+      "loss": 0.0561,
+      "step": 14100
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 1.2753699788583508e-05,
+      "loss": 0.0084,
+      "step": 14102
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 1.2748414376321354e-05,
+      "loss": 0.0024,
+      "step": 14104
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 1.2743128964059197e-05,
+      "loss": 0.0021,
+      "step": 14106
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 1.2737843551797041e-05,
+      "loss": 0.0055,
+      "step": 14108
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 1.2732558139534884e-05,
+      "loss": 0.0232,
+      "step": 14110
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 1.2727272727272727e-05,
+      "loss": 0.0115,
+      "step": 14112
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 1.2721987315010573e-05,
+      "loss": 0.0262,
+      "step": 14114
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 1.2716701902748416e-05,
+      "loss": 0.0173,
+      "step": 14116
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 1.2711416490486258e-05,
+      "loss": 0.0069,
+      "step": 14118
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 1.27061310782241e-05,
+      "loss": 0.0197,
+      "step": 14120
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 1.2700845665961947e-05,
+      "loss": 0.0512,
+      "step": 14122
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 1.269556025369979e-05,
+      "loss": 0.0183,
+      "step": 14124
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 1.2690274841437632e-05,
+      "loss": 0.0081,
+      "step": 14126
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 1.2684989429175475e-05,
+      "loss": 0.0204,
+      "step": 14128
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 1.267970401691332e-05,
+      "loss": 0.0542,
+      "step": 14130
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 1.2674418604651164e-05,
+      "loss": 0.0014,
+      "step": 14132
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 1.2669133192389008e-05,
+      "loss": 0.0283,
+      "step": 14134
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 1.266384778012685e-05,
+      "loss": 0.0025,
+      "step": 14136
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.2658562367864693e-05,
+      "loss": 0.0045,
+      "step": 14138
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.265327695560254e-05,
+      "loss": 0.0321,
+      "step": 14140
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.2647991543340382e-05,
+      "loss": 0.0172,
+      "step": 14142
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.2642706131078225e-05,
+      "loss": 0.0381,
+      "step": 14144
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.2637420718816067e-05,
+      "loss": 0.0101,
+      "step": 14146
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.263213530655391e-05,
+      "loss": 0.0131,
+      "step": 14148
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.2626849894291756e-05,
+      "loss": 0.0297,
+      "step": 14150
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.2621564482029599e-05,
+      "loss": 0.0712,
+      "step": 14152
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.2616279069767442e-05,
+      "loss": 0.0195,
+      "step": 14154
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.2610993657505286e-05,
+      "loss": 0.0028,
+      "step": 14156
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.260570824524313e-05,
+      "loss": 0.0212,
+      "step": 14158
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.2600422832980973e-05,
+      "loss": 0.0071,
+      "step": 14160
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.2595137420718817e-05,
+      "loss": 0.0148,
+      "step": 14162
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.258985200845666e-05,
+      "loss": 0.0342,
+      "step": 14164
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.2584566596194503e-05,
+      "loss": 0.0767,
+      "step": 14166
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 1.2579281183932349e-05,
+      "loss": 0.021,
+      "step": 14168
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 1.2573995771670191e-05,
+      "loss": 0.0338,
+      "step": 14170
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 1.2568710359408034e-05,
+      "loss": 0.0312,
+      "step": 14172
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 1.2563424947145877e-05,
+      "loss": 0.0186,
+      "step": 14174
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 1.2558139534883723e-05,
+      "loss": 0.0552,
+      "step": 14176
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 1.2552854122621565e-05,
+      "loss": 0.0098,
+      "step": 14178
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 1.2547568710359408e-05,
+      "loss": 0.0243,
+      "step": 14180
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 1.254228329809725e-05,
+      "loss": 0.0223,
+      "step": 14182
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 1.2536997885835095e-05,
+      "loss": 0.0292,
+      "step": 14184
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 1.253171247357294e-05,
+      "loss": 0.0551,
+      "step": 14186
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 1.2526427061310784e-05,
+      "loss": 0.0148,
+      "step": 14188
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 1.2521141649048627e-05,
+      "loss": 0.0025,
+      "step": 14190
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 1.251585623678647e-05,
+      "loss": 0.0089,
+      "step": 14192
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 1.2510570824524315e-05,
+      "loss": 0.0835,
+      "step": 14194
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 1.2505285412262158e-05,
+      "loss": 0.0132,
+      "step": 14196
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0119,
+      "step": 14198
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 1.2494714587737843e-05,
+      "loss": 0.0015,
+      "step": 14200
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 1.2489429175475688e-05,
+      "loss": 0.0044,
+      "step": 14202
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 1.248414376321353e-05,
+      "loss": 0.0366,
+      "step": 14204
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 1.2478858350951375e-05,
+      "loss": 0.1134,
+      "step": 14206
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 1.2473572938689217e-05,
+      "loss": 0.0088,
+      "step": 14208
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 1.2468287526427062e-05,
+      "loss": 0.0351,
+      "step": 14210
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 1.2463002114164906e-05,
+      "loss": 0.0116,
+      "step": 14212
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 1.245771670190275e-05,
+      "loss": 0.0377,
+      "step": 14214
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 1.2452431289640593e-05,
+      "loss": 0.0145,
+      "step": 14216
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 1.2447145877378436e-05,
+      "loss": 0.0409,
+      "step": 14218
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 1.244186046511628e-05,
+      "loss": 0.0149,
+      "step": 14220
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 1.2436575052854123e-05,
+      "loss": 0.0156,
+      "step": 14222
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 1.2431289640591967e-05,
+      "loss": 0.011,
+      "step": 14224
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 1.242600422832981e-05,
+      "loss": 0.0027,
+      "step": 14226
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 1.2420718816067654e-05,
+      "loss": 0.0122,
+      "step": 14228
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 1.2415433403805497e-05,
+      "loss": 0.0271,
+      "step": 14230
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 1.2410147991543341e-05,
+      "loss": 0.0151,
+      "step": 14232
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 1.2404862579281184e-05,
+      "loss": 0.0016,
+      "step": 14234
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 1.2399577167019028e-05,
+      "loss": 0.0097,
+      "step": 14236
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 1.2394291754756873e-05,
+      "loss": 0.0016,
+      "step": 14238
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 1.2389006342494715e-05,
+      "loss": 0.0188,
+      "step": 14240
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 1.238372093023256e-05,
+      "loss": 0.0048,
+      "step": 14242
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 1.2378435517970402e-05,
+      "loss": 0.0289,
+      "step": 14244
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 1.2373150105708247e-05,
+      "loss": 0.0386,
+      "step": 14246
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 1.236786469344609e-05,
+      "loss": 0.0394,
+      "step": 14248
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 1.2362579281183934e-05,
+      "loss": 0.0077,
+      "step": 14250
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 1.2357293868921776e-05,
+      "loss": 0.007,
+      "step": 14252
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 1.2352008456659619e-05,
+      "loss": 0.0025,
+      "step": 14254
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 1.2346723044397463e-05,
+      "loss": 0.018,
+      "step": 14256
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 1.2341437632135306e-05,
+      "loss": 0.0049,
+      "step": 14258
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 1.233615221987315e-05,
+      "loss": 0.0545,
+      "step": 14260
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 1.2330866807610993e-05,
+      "loss": 0.0179,
+      "step": 14262
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 1.2325581395348838e-05,
+      "loss": 0.0084,
+      "step": 14264
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 1.2320295983086682e-05,
+      "loss": 0.0726,
+      "step": 14266
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 1.2315010570824526e-05,
+      "loss": 0.0107,
+      "step": 14268
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 1.2309725158562369e-05,
+      "loss": 0.0246,
+      "step": 14270
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 1.2304439746300212e-05,
+      "loss": 0.0071,
+      "step": 14272
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 1.2299154334038056e-05,
+      "loss": 0.0165,
+      "step": 14274
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 1.2293868921775899e-05,
+      "loss": 0.0012,
+      "step": 14276
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 1.2288583509513743e-05,
+      "loss": 0.0039,
+      "step": 14278
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 1.2283298097251586e-05,
+      "loss": 0.0154,
+      "step": 14280
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 1.227801268498943e-05,
+      "loss": 0.009,
+      "step": 14282
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 1.2272727272727273e-05,
+      "loss": 0.0496,
+      "step": 14284
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 1.2267441860465115e-05,
+      "loss": 0.0129,
+      "step": 14286
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 1.226215644820296e-05,
+      "loss": 0.0179,
+      "step": 14288
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 1.2256871035940804e-05,
+      "loss": 0.0161,
+      "step": 14290
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 1.2251585623678648e-05,
+      "loss": 0.0031,
+      "step": 14292
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 1.2246300211416491e-05,
+      "loss": 0.0035,
+      "step": 14294
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 1.2241014799154336e-05,
+      "loss": 0.0109,
+      "step": 14296
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 1.2235729386892178e-05,
+      "loss": 0.0152,
+      "step": 14298
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 1.2230443974630023e-05,
+      "loss": 0.0054,
+      "step": 14300
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 1.2225158562367865e-05,
+      "loss": 0.0019,
+      "step": 14302
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 1.2219873150105708e-05,
+      "loss": 0.0168,
+      "step": 14304
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 1.2214587737843552e-05,
+      "loss": 0.01,
+      "step": 14306
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 1.2209302325581395e-05,
+      "loss": 0.0634,
+      "step": 14308
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 1.220401691331924e-05,
+      "loss": 0.0297,
+      "step": 14310
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 1.2198731501057082e-05,
+      "loss": 0.0023,
+      "step": 14312
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 1.2193446088794926e-05,
+      "loss": 0.0024,
+      "step": 14314
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 1.218816067653277e-05,
+      "loss": 0.0011,
+      "step": 14316
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 1.2182875264270615e-05,
+      "loss": 0.019,
+      "step": 14318
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 1.2177589852008458e-05,
+      "loss": 0.014,
+      "step": 14320
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 1.21723044397463e-05,
+      "loss": 0.0491,
+      "step": 14322
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 1.2167019027484145e-05,
+      "loss": 0.0111,
+      "step": 14324
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 1.2161733615221987e-05,
+      "loss": 0.1101,
+      "step": 14326
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 1.2156448202959832e-05,
+      "loss": 0.0013,
+      "step": 14328
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 1.2151162790697674e-05,
+      "loss": 0.0017,
+      "step": 14330
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 1.2145877378435519e-05,
+      "loss": 0.0016,
+      "step": 14332
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 1.2140591966173362e-05,
+      "loss": 0.0131,
+      "step": 14334
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 1.2135306553911206e-05,
+      "loss": 0.0167,
+      "step": 14336
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 1.2130021141649049e-05,
+      "loss": 0.0574,
+      "step": 14338
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 1.2124735729386893e-05,
+      "loss": 0.0031,
+      "step": 14340
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 1.2119450317124736e-05,
+      "loss": 0.0017,
+      "step": 14342
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 1.211416490486258e-05,
+      "loss": 0.0011,
+      "step": 14344
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 1.2108879492600424e-05,
+      "loss": 0.0199,
+      "step": 14346
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 1.2103594080338267e-05,
+      "loss": 0.0643,
+      "step": 14348
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 1.2098308668076111e-05,
+      "loss": 0.0007,
+      "step": 14350
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 1.2093023255813954e-05,
+      "loss": 0.0077,
+      "step": 14352
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 1.2087737843551798e-05,
+      "loss": 0.0107,
+      "step": 14354
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 1.2082452431289641e-05,
+      "loss": 0.0029,
+      "step": 14356
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 1.2077167019027484e-05,
+      "loss": 0.0179,
+      "step": 14358
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 1.2071881606765328e-05,
+      "loss": 0.0099,
+      "step": 14360
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 1.206659619450317e-05,
+      "loss": 0.0019,
+      "step": 14362
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 1.2061310782241015e-05,
+      "loss": 0.0512,
+      "step": 14364
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 1.2056025369978858e-05,
+      "loss": 0.0281,
+      "step": 14366
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 1.2050739957716702e-05,
+      "loss": 0.0039,
+      "step": 14368
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 1.2045454545454547e-05,
+      "loss": 0.0081,
+      "step": 14370
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 1.2040169133192391e-05,
+      "loss": 0.038,
+      "step": 14372
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 1.2034883720930234e-05,
+      "loss": 0.0053,
+      "step": 14374
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 1.2029598308668076e-05,
+      "loss": 0.0346,
+      "step": 14376
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 1.202431289640592e-05,
+      "loss": 0.0205,
+      "step": 14378
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 1.2019027484143763e-05,
+      "loss": 0.0112,
+      "step": 14380
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 1.2013742071881608e-05,
+      "loss": 0.0561,
+      "step": 14382
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 1.200845665961945e-05,
+      "loss": 0.0052,
+      "step": 14384
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 1.2003171247357295e-05,
+      "loss": 0.0107,
+      "step": 14386
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 1.1997885835095137e-05,
+      "loss": 0.0043,
+      "step": 14388
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 1.1992600422832982e-05,
+      "loss": 0.0613,
+      "step": 14390
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 1.1987315010570824e-05,
+      "loss": 0.0243,
+      "step": 14392
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 1.1982029598308669e-05,
+      "loss": 0.0275,
+      "step": 14394
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 1.1976744186046513e-05,
+      "loss": 0.0478,
+      "step": 14396
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 1.1971458773784356e-05,
+      "loss": 0.0442,
+      "step": 14398
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 1.19661733615222e-05,
+      "loss": 0.0173,
+      "step": 14400
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 1.1960887949260043e-05,
+      "loss": 0.082,
+      "step": 14402
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 1.1955602536997887e-05,
+      "loss": 0.0412,
+      "step": 14404
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 1.195031712473573e-05,
+      "loss": 0.0079,
+      "step": 14406
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 1.1945031712473574e-05,
+      "loss": 0.0347,
+      "step": 14408
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 1.1939746300211417e-05,
+      "loss": 0.0082,
+      "step": 14410
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 1.193446088794926e-05,
+      "loss": 0.0088,
+      "step": 14412
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 1.1929175475687104e-05,
+      "loss": 0.0102,
+      "step": 14414
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 1.1923890063424947e-05,
+      "loss": 0.0421,
+      "step": 14416
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 1.1918604651162791e-05,
+      "loss": 0.0322,
+      "step": 14418
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 1.1913319238900635e-05,
+      "loss": 0.0186,
+      "step": 14420
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 1.190803382663848e-05,
+      "loss": 0.0491,
+      "step": 14422
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 1.1902748414376322e-05,
+      "loss": 0.0272,
+      "step": 14424
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 1.1897463002114167e-05,
+      "loss": 0.0482,
+      "step": 14426
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 1.189217758985201e-05,
+      "loss": 0.0494,
+      "step": 14428
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 1.1886892177589852e-05,
+      "loss": 0.0158,
+      "step": 14430
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 1.1881606765327696e-05,
+      "loss": 0.0274,
+      "step": 14432
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 1.1876321353065539e-05,
+      "loss": 0.0167,
+      "step": 14434
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 1.1871035940803384e-05,
+      "loss": 0.1149,
+      "step": 14436
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 1.1865750528541226e-05,
+      "loss": 0.0095,
+      "step": 14438
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 1.186046511627907e-05,
+      "loss": 0.0087,
+      "step": 14440
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 1.1855179704016913e-05,
+      "loss": 0.0084,
+      "step": 14442
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 1.1849894291754758e-05,
+      "loss": 0.0193,
+      "step": 14444
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 1.18446088794926e-05,
+      "loss": 0.0106,
+      "step": 14446
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 1.1839323467230445e-05,
+      "loss": 0.0039,
+      "step": 14448
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 1.1834038054968289e-05,
+      "loss": 0.0293,
+      "step": 14450
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 1.1828752642706132e-05,
+      "loss": 0.0379,
+      "step": 14452
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 1.1823467230443976e-05,
+      "loss": 0.0193,
+      "step": 14454
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 1.1818181818181819e-05,
+      "loss": 0.0078,
+      "step": 14456
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 1.1812896405919663e-05,
+      "loss": 0.0086,
+      "step": 14458
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 1.1807610993657506e-05,
+      "loss": 0.0157,
+      "step": 14460
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 1.1802325581395348e-05,
+      "loss": 0.0359,
+      "step": 14462
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 1.1797040169133193e-05,
+      "loss": 0.0405,
+      "step": 14464
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 1.1791754756871035e-05,
+      "loss": 0.0195,
+      "step": 14466
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 1.178646934460888e-05,
+      "loss": 0.0071,
+      "step": 14468
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 1.1781183932346722e-05,
+      "loss": 0.0268,
+      "step": 14470
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 1.1775898520084567e-05,
+      "loss": 0.0203,
+      "step": 14472
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 1.1770613107822411e-05,
+      "loss": 0.0033,
+      "step": 14474
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 1.1765327695560256e-05,
+      "loss": 0.0172,
+      "step": 14476
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 1.1760042283298098e-05,
+      "loss": 0.007,
+      "step": 14478
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 1.1754756871035941e-05,
+      "loss": 0.0289,
+      "step": 14480
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 1.1749471458773785e-05,
+      "loss": 0.0525,
+      "step": 14482
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 1.1744186046511628e-05,
+      "loss": 0.0071,
+      "step": 14484
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 1.1738900634249472e-05,
+      "loss": 0.0152,
+      "step": 14486
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 1.1733615221987315e-05,
+      "loss": 0.0126,
+      "step": 14488
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 1.172832980972516e-05,
+      "loss": 0.0154,
+      "step": 14490
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 1.1723044397463002e-05,
+      "loss": 0.0023,
+      "step": 14492
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 1.1717758985200846e-05,
+      "loss": 0.0349,
+      "step": 14494
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 1.1712473572938689e-05,
+      "loss": 0.0262,
+      "step": 14496
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 1.1707188160676533e-05,
+      "loss": 0.031,
+      "step": 14498
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 1.1701902748414378e-05,
+      "loss": 0.0121,
+      "step": 14500
+    },
+    {
+      "epoch": 15.33,
+      "eval_cer": 0.04895981761185523,
+      "eval_loss": 0.6679000854492188,
+      "eval_runtime": 125.3329,
+      "eval_samples_per_second": 6.71,
+      "eval_steps_per_second": 0.846,
+      "step": 14500
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 1.169661733615222e-05,
+      "loss": 0.0094,
+      "step": 14502
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 1.1691331923890065e-05,
+      "loss": 0.0073,
+      "step": 14504
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 1.1686046511627907e-05,
+      "loss": 0.0042,
+      "step": 14506
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 1.1680761099365752e-05,
+      "loss": 0.0144,
+      "step": 14508
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 1.1675475687103595e-05,
+      "loss": 0.0017,
+      "step": 14510
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 1.1670190274841439e-05,
+      "loss": 0.0413,
+      "step": 14512
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 1.1664904862579282e-05,
+      "loss": 0.0259,
+      "step": 14514
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 1.1659619450317124e-05,
+      "loss": 0.0061,
+      "step": 14516
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 1.1654334038054969e-05,
+      "loss": 0.0018,
+      "step": 14518
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 1.1649048625792811e-05,
+      "loss": 0.0007,
+      "step": 14520
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 1.1643763213530656e-05,
+      "loss": 0.0474,
+      "step": 14522
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 1.16384778012685e-05,
+      "loss": 0.0409,
+      "step": 14524
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 1.1633192389006343e-05,
+      "loss": 0.0263,
+      "step": 14526
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 1.1627906976744187e-05,
+      "loss": 0.0323,
+      "step": 14528
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 1.1622621564482031e-05,
+      "loss": 0.0008,
+      "step": 14530
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 1.1617336152219874e-05,
+      "loss": 0.0061,
+      "step": 14532
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 1.1612050739957717e-05,
+      "loss": 0.0196,
+      "step": 14534
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 1.1606765327695561e-05,
+      "loss": 0.0093,
+      "step": 14536
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 1.1601479915433404e-05,
+      "loss": 0.0166,
+      "step": 14538
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 1.1596194503171248e-05,
+      "loss": 0.0296,
+      "step": 14540
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 1.159090909090909e-05,
+      "loss": 0.0298,
+      "step": 14542
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 1.1585623678646935e-05,
+      "loss": 0.0354,
+      "step": 14544
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 1.1580338266384778e-05,
+      "loss": 0.0048,
+      "step": 14546
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 1.1575052854122622e-05,
+      "loss": 0.0014,
+      "step": 14548
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 1.1569767441860465e-05,
+      "loss": 0.0582,
+      "step": 14550
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 1.156448202959831e-05,
+      "loss": 0.0271,
+      "step": 14552
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 1.1559196617336154e-05,
+      "loss": 0.0263,
+      "step": 14554
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 1.1553911205073996e-05,
+      "loss": 0.029,
+      "step": 14556
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 1.154862579281184e-05,
+      "loss": 0.0028,
+      "step": 14558
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 1.1543340380549683e-05,
+      "loss": 0.0368,
+      "step": 14560
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 1.1538054968287528e-05,
+      "loss": 0.0295,
+      "step": 14562
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 1.153276955602537e-05,
+      "loss": 0.0139,
+      "step": 14564
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 1.1527484143763215e-05,
+      "loss": 0.0055,
+      "step": 14566
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 1.1522198731501057e-05,
+      "loss": 0.0277,
+      "step": 14568
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 1.15169133192389e-05,
+      "loss": 0.0919,
+      "step": 14570
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 1.1511627906976744e-05,
+      "loss": 0.033,
+      "step": 14572
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 1.1506342494714587e-05,
+      "loss": 0.0078,
+      "step": 14574
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 1.1501057082452431e-05,
+      "loss": 0.0171,
+      "step": 14576
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 1.1495771670190276e-05,
+      "loss": 0.0377,
+      "step": 14578
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 1.149048625792812e-05,
+      "loss": 0.0205,
+      "step": 14580
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 1.1485200845665963e-05,
+      "loss": 0.0259,
+      "step": 14582
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 1.1479915433403807e-05,
+      "loss": 0.0213,
+      "step": 14584
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 1.147463002114165e-05,
+      "loss": 0.0033,
+      "step": 14586
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 1.1469344608879493e-05,
+      "loss": 0.0379,
+      "step": 14588
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 1.1464059196617337e-05,
+      "loss": 0.0388,
+      "step": 14590
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 1.145877378435518e-05,
+      "loss": 0.0007,
+      "step": 14592
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 1.1453488372093024e-05,
+      "loss": 0.0107,
+      "step": 14594
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 1.1448202959830867e-05,
+      "loss": 0.018,
+      "step": 14596
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 1.1442917547568711e-05,
+      "loss": 0.0355,
+      "step": 14598
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 1.1437632135306554e-05,
+      "loss": 0.0101,
+      "step": 14600
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 1.1432346723044398e-05,
+      "loss": 0.0064,
+      "step": 14602
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 1.1427061310782242e-05,
+      "loss": 0.0077,
+      "step": 14604
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 1.1421775898520085e-05,
+      "loss": 0.003,
+      "step": 14606
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 1.141649048625793e-05,
+      "loss": 0.0049,
+      "step": 14608
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 1.1411205073995772e-05,
+      "loss": 0.0184,
+      "step": 14610
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 1.1405919661733616e-05,
+      "loss": 0.0231,
+      "step": 14612
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 1.140063424947146e-05,
+      "loss": 0.0215,
+      "step": 14614
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 1.1395348837209304e-05,
+      "loss": 0.0232,
+      "step": 14616
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 1.1390063424947146e-05,
+      "loss": 0.0191,
+      "step": 14618
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 1.138477801268499e-05,
+      "loss": 0.0127,
+      "step": 14620
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 1.1379492600422833e-05,
+      "loss": 0.031,
+      "step": 14622
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 1.1374207188160676e-05,
+      "loss": 0.0854,
+      "step": 14624
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 1.136892177589852e-05,
+      "loss": 0.0194,
+      "step": 14626
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 1.1363636363636365e-05,
+      "loss": 0.0172,
+      "step": 14628
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 1.1358350951374207e-05,
+      "loss": 0.0028,
+      "step": 14630
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 1.1353065539112052e-05,
+      "loss": 0.0134,
+      "step": 14632
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 1.1347780126849896e-05,
+      "loss": 0.0056,
+      "step": 14634
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 1.1342494714587739e-05,
+      "loss": 0.0076,
+      "step": 14636
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 1.1337209302325581e-05,
+      "loss": 0.0103,
+      "step": 14638
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 1.1331923890063426e-05,
+      "loss": 0.0054,
+      "step": 14640
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 1.1326638477801268e-05,
+      "loss": 0.0255,
+      "step": 14642
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 1.1321353065539113e-05,
+      "loss": 0.0609,
+      "step": 14644
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 1.1316067653276955e-05,
+      "loss": 0.0092,
+      "step": 14646
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 1.13107822410148e-05,
+      "loss": 0.013,
+      "step": 14648
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 1.1305496828752642e-05,
+      "loss": 0.024,
+      "step": 14650
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 1.1300211416490487e-05,
+      "loss": 0.0104,
+      "step": 14652
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 1.129492600422833e-05,
+      "loss": 0.0106,
+      "step": 14654
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 1.1289640591966174e-05,
+      "loss": 0.0158,
+      "step": 14656
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 1.1284355179704018e-05,
+      "loss": 0.0783,
+      "step": 14658
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 1.1279069767441861e-05,
+      "loss": 0.0025,
+      "step": 14660
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 1.1273784355179705e-05,
+      "loss": 0.055,
+      "step": 14662
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 1.1268498942917548e-05,
+      "loss": 0.0038,
+      "step": 14664
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 1.1263213530655392e-05,
+      "loss": 0.0024,
+      "step": 14666
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 1.1257928118393235e-05,
+      "loss": 0.0337,
+      "step": 14668
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 1.125264270613108e-05,
+      "loss": 0.0129,
+      "step": 14670
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 1.1247357293868922e-05,
+      "loss": 0.0065,
+      "step": 14672
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 1.1242071881606765e-05,
+      "loss": 0.0291,
+      "step": 14674
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 1.1236786469344609e-05,
+      "loss": 0.0127,
+      "step": 14676
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 1.1231501057082452e-05,
+      "loss": 0.0149,
+      "step": 14678
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 1.1226215644820296e-05,
+      "loss": 0.0389,
+      "step": 14680
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 1.122093023255814e-05,
+      "loss": 0.0657,
+      "step": 14682
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 1.1215644820295985e-05,
+      "loss": 0.0156,
+      "step": 14684
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 1.1210359408033828e-05,
+      "loss": 0.0137,
+      "step": 14686
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 1.1205073995771672e-05,
+      "loss": 0.0564,
+      "step": 14688
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 1.1199788583509515e-05,
+      "loss": 0.0165,
+      "step": 14690
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 1.1194503171247357e-05,
+      "loss": 0.0152,
+      "step": 14692
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 1.1189217758985202e-05,
+      "loss": 0.0335,
+      "step": 14694
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 1.1183932346723044e-05,
+      "loss": 0.0033,
+      "step": 14696
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 1.1178646934460889e-05,
+      "loss": 0.0171,
+      "step": 14698
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 1.1173361522198731e-05,
+      "loss": 0.0164,
+      "step": 14700
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 1.1168076109936576e-05,
+      "loss": 0.0208,
+      "step": 14702
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 1.1162790697674418e-05,
+      "loss": 0.0062,
+      "step": 14704
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 1.1157505285412263e-05,
+      "loss": 0.0031,
+      "step": 14706
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 1.1152219873150107e-05,
+      "loss": 0.0039,
+      "step": 14708
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 1.114693446088795e-05,
+      "loss": 0.04,
+      "step": 14710
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 1.1141649048625794e-05,
+      "loss": 0.0146,
+      "step": 14712
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 1.1136363636363637e-05,
+      "loss": 0.0118,
+      "step": 14714
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 1.1131078224101481e-05,
+      "loss": 0.001,
+      "step": 14716
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 1.1125792811839324e-05,
+      "loss": 0.0444,
+      "step": 14718
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 1.1120507399577168e-05,
+      "loss": 0.0169,
+      "step": 14720
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 1.1115221987315011e-05,
+      "loss": 0.0395,
+      "step": 14722
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 1.1109936575052855e-05,
+      "loss": 0.0146,
+      "step": 14724
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 1.1104651162790698e-05,
+      "loss": 0.0048,
+      "step": 14726
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 1.109936575052854e-05,
+      "loss": 0.0195,
+      "step": 14728
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 1.1094080338266385e-05,
+      "loss": 0.0036,
+      "step": 14730
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 1.1088794926004228e-05,
+      "loss": 0.0068,
+      "step": 14732
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 1.1083509513742072e-05,
+      "loss": 0.0082,
+      "step": 14734
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 1.1078224101479916e-05,
+      "loss": 0.0098,
+      "step": 14736
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 1.107293868921776e-05,
+      "loss": 0.004,
+      "step": 14738
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 1.1067653276955603e-05,
+      "loss": 0.0357,
+      "step": 14740
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 1.1062367864693448e-05,
+      "loss": 0.0158,
+      "step": 14742
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 1.105708245243129e-05,
+      "loss": 0.0059,
+      "step": 14744
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 1.1051797040169133e-05,
+      "loss": 0.0125,
+      "step": 14746
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 1.1046511627906977e-05,
+      "loss": 0.0443,
+      "step": 14748
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 1.104122621564482e-05,
+      "loss": 0.0017,
+      "step": 14750
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 1.1035940803382664e-05,
+      "loss": 0.062,
+      "step": 14752
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 1.1030655391120507e-05,
+      "loss": 0.006,
+      "step": 14754
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 1.1025369978858352e-05,
+      "loss": 0.0143,
+      "step": 14756
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 1.1020084566596194e-05,
+      "loss": 0.0055,
+      "step": 14758
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 1.1014799154334039e-05,
+      "loss": 0.0177,
+      "step": 14760
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 1.1009513742071883e-05,
+      "loss": 0.0077,
+      "step": 14762
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 1.1004228329809726e-05,
+      "loss": 0.0037,
+      "step": 14764
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 1.099894291754757e-05,
+      "loss": 0.0334,
+      "step": 14766
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 1.0993657505285413e-05,
+      "loss": 0.0067,
+      "step": 14768
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 1.0988372093023257e-05,
+      "loss": 0.0158,
+      "step": 14770
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 1.09830866807611e-05,
+      "loss": 0.0067,
+      "step": 14772
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 1.0977801268498944e-05,
+      "loss": 0.0298,
+      "step": 14774
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 1.0972515856236787e-05,
+      "loss": 0.0008,
+      "step": 14776
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 1.0967230443974631e-05,
+      "loss": 0.0064,
+      "step": 14778
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 1.0961945031712474e-05,
+      "loss": 0.0055,
+      "step": 14780
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 1.0956659619450316e-05,
+      "loss": 0.0701,
+      "step": 14782
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 1.095137420718816e-05,
+      "loss": 0.0278,
+      "step": 14784
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 1.0946088794926005e-05,
+      "loss": 0.0189,
+      "step": 14786
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 1.094080338266385e-05,
+      "loss": 0.0136,
+      "step": 14788
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 1.0935517970401692e-05,
+      "loss": 0.0049,
+      "step": 14790
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 1.0930232558139537e-05,
+      "loss": 0.0078,
+      "step": 14792
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 1.092494714587738e-05,
+      "loss": 0.0471,
+      "step": 14794
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 1.0919661733615222e-05,
+      "loss": 0.0093,
+      "step": 14796
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 1.0914376321353066e-05,
+      "loss": 0.0332,
+      "step": 14798
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 0.046,
+      "step": 14800
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 1.0903805496828753e-05,
+      "loss": 0.0186,
+      "step": 14802
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 1.0898520084566596e-05,
+      "loss": 0.0015,
+      "step": 14804
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 1.089323467230444e-05,
+      "loss": 0.0335,
+      "step": 14806
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 1.0887949260042283e-05,
+      "loss": 0.0217,
+      "step": 14808
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 1.0882663847780127e-05,
+      "loss": 0.0014,
+      "step": 14810
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 1.0877378435517972e-05,
+      "loss": 0.0465,
+      "step": 14812
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 1.0872093023255814e-05,
+      "loss": 0.0107,
+      "step": 14814
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 1.0866807610993659e-05,
+      "loss": 0.0903,
+      "step": 14816
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 1.0861522198731501e-05,
+      "loss": 0.0038,
+      "step": 14818
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 1.0856236786469346e-05,
+      "loss": 0.0018,
+      "step": 14820
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 1.0850951374207188e-05,
+      "loss": 0.0343,
+      "step": 14822
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 1.0845665961945033e-05,
+      "loss": 0.0074,
+      "step": 14824
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 1.0840380549682875e-05,
+      "loss": 0.0257,
+      "step": 14826
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 1.083509513742072e-05,
+      "loss": 0.0017,
+      "step": 14828
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 1.0829809725158563e-05,
+      "loss": 0.0189,
+      "step": 14830
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 1.0824524312896405e-05,
+      "loss": 0.0674,
+      "step": 14832
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 1.081923890063425e-05,
+      "loss": 0.012,
+      "step": 14834
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 1.0813953488372092e-05,
+      "loss": 0.034,
+      "step": 14836
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 1.0808668076109937e-05,
+      "loss": 0.0246,
+      "step": 14838
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 1.0803382663847781e-05,
+      "loss": 0.0021,
+      "step": 14840
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 1.0798097251585625e-05,
+      "loss": 0.0245,
+      "step": 14842
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 1.0792811839323468e-05,
+      "loss": 0.0014,
+      "step": 14844
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 1.0787526427061312e-05,
+      "loss": 0.0155,
+      "step": 14846
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 1.0782241014799155e-05,
+      "loss": 0.0208,
+      "step": 14848
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 1.0776955602536998e-05,
+      "loss": 0.0056,
+      "step": 14850
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 1.0771670190274842e-05,
+      "loss": 0.0019,
+      "step": 14852
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 1.0766384778012685e-05,
+      "loss": 0.031,
+      "step": 14854
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 1.0761099365750529e-05,
+      "loss": 0.0575,
+      "step": 14856
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 1.0755813953488372e-05,
+      "loss": 0.0129,
+      "step": 14858
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 1.0750528541226216e-05,
+      "loss": 0.0061,
+      "step": 14860
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 1.0745243128964059e-05,
+      "loss": 0.0038,
+      "step": 14862
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 1.0739957716701903e-05,
+      "loss": 0.0207,
+      "step": 14864
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 1.0734672304439748e-05,
+      "loss": 0.0242,
+      "step": 14866
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 1.072938689217759e-05,
+      "loss": 0.0162,
+      "step": 14868
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 1.0724101479915435e-05,
+      "loss": 0.0286,
+      "step": 14870
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 1.0718816067653277e-05,
+      "loss": 0.0316,
+      "step": 14872
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 1.0713530655391122e-05,
+      "loss": 0.0073,
+      "step": 14874
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 1.0708245243128964e-05,
+      "loss": 0.0109,
+      "step": 14876
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 1.0702959830866809e-05,
+      "loss": 0.0047,
+      "step": 14878
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 1.0697674418604651e-05,
+      "loss": 0.0354,
+      "step": 14880
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 1.0692389006342496e-05,
+      "loss": 0.0226,
+      "step": 14882
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 1.0687103594080338e-05,
+      "loss": 0.017,
+      "step": 14884
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 1.0681818181818181e-05,
+      "loss": 0.003,
+      "step": 14886
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 1.0676532769556025e-05,
+      "loss": 0.0335,
+      "step": 14888
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 1.067124735729387e-05,
+      "loss": 0.0027,
+      "step": 14890
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 1.0665961945031714e-05,
+      "loss": 0.0122,
+      "step": 14892
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 1.0660676532769557e-05,
+      "loss": 0.0095,
+      "step": 14894
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 1.0655391120507401e-05,
+      "loss": 0.0042,
+      "step": 14896
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 1.0652748414376323e-05,
+      "loss": 0.1013,
+      "step": 14898
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 1.0647463002114165e-05,
+      "loss": 0.0157,
+      "step": 14900
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 1.064217758985201e-05,
+      "loss": 0.0084,
+      "step": 14902
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 1.0636892177589852e-05,
+      "loss": 0.0251,
+      "step": 14904
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 1.0631606765327697e-05,
+      "loss": 0.0067,
+      "step": 14906
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 1.062632135306554e-05,
+      "loss": 0.0134,
+      "step": 14908
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 1.0621035940803384e-05,
+      "loss": 0.0021,
+      "step": 14910
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 1.0615750528541226e-05,
+      "loss": 0.0048,
+      "step": 14912
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 1.0610465116279069e-05,
+      "loss": 0.0014,
+      "step": 14914
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 1.0605179704016913e-05,
+      "loss": 0.0111,
+      "step": 14916
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 1.0599894291754758e-05,
+      "loss": 0.0058,
+      "step": 14918
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 1.0594608879492602e-05,
+      "loss": 0.0129,
+      "step": 14920
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 1.0589323467230445e-05,
+      "loss": 0.0626,
+      "step": 14922
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 1.0584038054968289e-05,
+      "loss": 0.0111,
+      "step": 14924
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 1.0578752642706132e-05,
+      "loss": 0.0184,
+      "step": 14926
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 1.0573467230443976e-05,
+      "loss": 0.0208,
+      "step": 14928
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 1.0568181818181819e-05,
+      "loss": 0.0076,
+      "step": 14930
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 1.0562896405919661e-05,
+      "loss": 0.0395,
+      "step": 14932
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 1.0557610993657506e-05,
+      "loss": 0.0079,
+      "step": 14934
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 1.0552325581395349e-05,
+      "loss": 0.0393,
+      "step": 14936
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 1.0547040169133193e-05,
+      "loss": 0.0174,
+      "step": 14938
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 1.0541754756871036e-05,
+      "loss": 0.0146,
+      "step": 14940
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 1.053646934460888e-05,
+      "loss": 0.0084,
+      "step": 14942
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 1.0531183932346724e-05,
+      "loss": 0.0514,
+      "step": 14944
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 1.0525898520084567e-05,
+      "loss": 0.0202,
+      "step": 14946
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 1.0520613107822411e-05,
+      "loss": 0.011,
+      "step": 14948
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 1.0515327695560254e-05,
+      "loss": 0.0199,
+      "step": 14950
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 1.0510042283298098e-05,
+      "loss": 0.0292,
+      "step": 14952
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 1.0504756871035941e-05,
+      "loss": 0.0477,
+      "step": 14954
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 1.0499471458773785e-05,
+      "loss": 0.0244,
+      "step": 14956
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 1.0494186046511628e-05,
+      "loss": 0.0155,
+      "step": 14958
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 1.0488900634249472e-05,
+      "loss": 0.029,
+      "step": 14960
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 1.0483615221987315e-05,
+      "loss": 0.0276,
+      "step": 14962
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 1.0478329809725158e-05,
+      "loss": 0.0275,
+      "step": 14964
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 1.0473044397463002e-05,
+      "loss": 0.0165,
+      "step": 14966
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 1.0467758985200845e-05,
+      "loss": 0.0428,
+      "step": 14968
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 1.046247357293869e-05,
+      "loss": 0.0454,
+      "step": 14970
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 1.0457188160676534e-05,
+      "loss": 0.1425,
+      "step": 14972
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 1.0451902748414378e-05,
+      "loss": 0.0476,
+      "step": 14974
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 1.044661733615222e-05,
+      "loss": 0.0332,
+      "step": 14976
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 1.0441331923890065e-05,
+      "loss": 0.0403,
+      "step": 14978
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 1.0436046511627908e-05,
+      "loss": 0.0198,
+      "step": 14980
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 1.043076109936575e-05,
+      "loss": 0.0125,
+      "step": 14982
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 1.0425475687103595e-05,
+      "loss": 0.0084,
+      "step": 14984
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 1.0420190274841437e-05,
+      "loss": 0.0013,
+      "step": 14986
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 1.0414904862579282e-05,
+      "loss": 0.0103,
+      "step": 14988
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 1.0409619450317124e-05,
+      "loss": 0.0311,
+      "step": 14990
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 1.0404334038054969e-05,
+      "loss": 0.0204,
+      "step": 14992
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 1.0399048625792811e-05,
+      "loss": 0.0037,
+      "step": 14994
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 1.0393763213530656e-05,
+      "loss": 0.0013,
+      "step": 14996
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 1.03884778012685e-05,
+      "loss": 0.0351,
+      "step": 14998
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 1.0383192389006343e-05,
+      "loss": 0.0054,
+      "step": 15000
+    },
+    {
+      "epoch": 15.86,
+      "eval_cer": 0.0335138216015959,
+      "eval_loss": 0.8025035262107849,
+      "eval_runtime": 125.5761,
+      "eval_samples_per_second": 6.697,
+      "eval_steps_per_second": 0.844,
+      "step": 15000
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 1.0377906976744187e-05,
+      "loss": 0.0153,
+      "step": 15002
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 1.037262156448203e-05,
+      "loss": 0.0195,
+      "step": 15004
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 1.0367336152219874e-05,
+      "loss": 0.0055,
+      "step": 15006
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 1.0362050739957717e-05,
+      "loss": 0.0015,
+      "step": 15008
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 1.0356765327695561e-05,
+      "loss": 0.0036,
+      "step": 15010
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 1.0351479915433404e-05,
+      "loss": 0.0304,
+      "step": 15012
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 1.0346194503171248e-05,
+      "loss": 0.021,
+      "step": 15014
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 1.0340909090909091e-05,
+      "loss": 0.0065,
+      "step": 15016
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 1.0335623678646934e-05,
+      "loss": 0.006,
+      "step": 15018
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 1.0330338266384778e-05,
+      "loss": 0.0214,
+      "step": 15020
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 1.0325052854122622e-05,
+      "loss": 0.0925,
+      "step": 15022
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 1.0319767441860467e-05,
+      "loss": 0.0072,
+      "step": 15024
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 1.031448202959831e-05,
+      "loss": 0.0101,
+      "step": 15026
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 1.0309196617336154e-05,
+      "loss": 0.0601,
+      "step": 15028
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 1.0303911205073996e-05,
+      "loss": 0.0107,
+      "step": 15030
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 1.029862579281184e-05,
+      "loss": 0.0073,
+      "step": 15032
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 1.0293340380549683e-05,
+      "loss": 0.0511,
+      "step": 15034
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 1.0288054968287526e-05,
+      "loss": 0.0102,
+      "step": 15036
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 1.028276955602537e-05,
+      "loss": 0.0081,
+      "step": 15038
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 1.0277484143763213e-05,
+      "loss": 0.0111,
+      "step": 15040
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 1.0272198731501058e-05,
+      "loss": 0.0022,
+      "step": 15042
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 1.02669133192389e-05,
+      "loss": 0.0098,
+      "step": 15044
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 1.0261627906976745e-05,
+      "loss": 0.014,
+      "step": 15046
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 1.0256342494714589e-05,
+      "loss": 0.0183,
+      "step": 15048
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 1.0251057082452432e-05,
+      "loss": 0.004,
+      "step": 15050
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 1.0245771670190276e-05,
+      "loss": 0.0045,
+      "step": 15052
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 1.0240486257928119e-05,
+      "loss": 0.0175,
+      "step": 15054
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 1.0235200845665963e-05,
+      "loss": 0.0855,
+      "step": 15056
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 1.0229915433403806e-05,
+      "loss": 0.0145,
+      "step": 15058
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 1.022463002114165e-05,
+      "loss": 0.0268,
+      "step": 15060
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 1.0219344608879493e-05,
+      "loss": 0.0131,
+      "step": 15062
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 1.0214059196617337e-05,
+      "loss": 0.0231,
+      "step": 15064
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 1.020877378435518e-05,
+      "loss": 0.0278,
+      "step": 15066
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 1.0203488372093024e-05,
+      "loss": 0.0021,
+      "step": 15068
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 1.0198202959830867e-05,
+      "loss": 0.007,
+      "step": 15070
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 1.019291754756871e-05,
+      "loss": 0.0337,
+      "step": 15072
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 1.0187632135306554e-05,
+      "loss": 0.0458,
+      "step": 15074
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 1.0182346723044398e-05,
+      "loss": 0.0022,
+      "step": 15076
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 1.0177061310782243e-05,
+      "loss": 0.0135,
+      "step": 15078
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 1.0171775898520085e-05,
+      "loss": 0.0165,
+      "step": 15080
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 1.016649048625793e-05,
+      "loss": 0.0712,
+      "step": 15082
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 1.0161205073995772e-05,
+      "loss": 0.024,
+      "step": 15084
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 1.0155919661733617e-05,
+      "loss": 0.0021,
+      "step": 15086
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 1.015063424947146e-05,
+      "loss": 0.0026,
+      "step": 15088
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 1.0145348837209302e-05,
+      "loss": 0.0148,
+      "step": 15090
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 1.0140063424947146e-05,
+      "loss": 0.0023,
+      "step": 15092
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 1.0134778012684989e-05,
+      "loss": 0.0029,
+      "step": 15094
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 1.0129492600422833e-05,
+      "loss": 0.0083,
+      "step": 15096
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 1.0124207188160676e-05,
+      "loss": 0.0047,
+      "step": 15098
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 1.011892177589852e-05,
+      "loss": 0.0348,
+      "step": 15100
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 1.0113636363636365e-05,
+      "loss": 0.0033,
+      "step": 15102
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 1.0108350951374209e-05,
+      "loss": 0.0005,
+      "step": 15104
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 1.0103065539112052e-05,
+      "loss": 0.0072,
+      "step": 15106
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 1.0097780126849894e-05,
+      "loss": 0.0082,
+      "step": 15108
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 1.0092494714587739e-05,
+      "loss": 0.0054,
+      "step": 15110
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 1.0087209302325581e-05,
+      "loss": 0.0637,
+      "step": 15112
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 1.0081923890063426e-05,
+      "loss": 0.0627,
+      "step": 15114
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 1.0076638477801269e-05,
+      "loss": 0.0063,
+      "step": 15116
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 1.0071353065539113e-05,
+      "loss": 0.0118,
+      "step": 15118
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 1.0066067653276956e-05,
+      "loss": 0.0127,
+      "step": 15120
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 1.00607822410148e-05,
+      "loss": 0.0117,
+      "step": 15122
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 1.0055496828752643e-05,
+      "loss": 0.0017,
+      "step": 15124
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 1.0050211416490487e-05,
+      "loss": 0.0018,
+      "step": 15126
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 1.0044926004228331e-05,
+      "loss": 0.0018,
+      "step": 15128
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 1.0039640591966174e-05,
+      "loss": 0.0301,
+      "step": 15130
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 1.0034355179704018e-05,
+      "loss": 0.074,
+      "step": 15132
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 1.0029069767441861e-05,
+      "loss": 0.0381,
+      "step": 15134
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 1.0023784355179705e-05,
+      "loss": 0.0129,
+      "step": 15136
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 1.0018498942917548e-05,
+      "loss": 0.001,
+      "step": 15138
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 1.001321353065539e-05,
+      "loss": 0.0034,
+      "step": 15140
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 1.0007928118393235e-05,
+      "loss": 0.014,
+      "step": 15142
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 1.0002642706131078e-05,
+      "loss": 0.0118,
+      "step": 15144
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 9.997357293868922e-06,
+      "loss": 0.0318,
+      "step": 15146
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 9.992071881606765e-06,
+      "loss": 0.0006,
+      "step": 15148
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 9.98678646934461e-06,
+      "loss": 0.0669,
+      "step": 15150
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 9.981501057082452e-06,
+      "loss": 0.0426,
+      "step": 15152
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 9.976215644820296e-06,
+      "loss": 0.0088,
+      "step": 15154
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 9.97093023255814e-06,
+      "loss": 0.0141,
+      "step": 15156
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 9.965644820295983e-06,
+      "loss": 0.0079,
+      "step": 15158
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 9.960359408033828e-06,
+      "loss": 0.0047,
+      "step": 15160
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 9.95507399577167e-06,
+      "loss": 0.0025,
+      "step": 15162
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 9.949788583509515e-06,
+      "loss": 0.0007,
+      "step": 15164
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 9.944503171247357e-06,
+      "loss": 0.0009,
+      "step": 15166
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 9.939217758985202e-06,
+      "loss": 0.0158,
+      "step": 15168
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 9.933932346723044e-06,
+      "loss": 0.023,
+      "step": 15170
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 9.928646934460889e-06,
+      "loss": 0.0184,
+      "step": 15172
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 9.923361522198731e-06,
+      "loss": 0.0449,
+      "step": 15174
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 9.918076109936574e-06,
+      "loss": 0.0215,
+      "step": 15176
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 9.912790697674418e-06,
+      "loss": 0.0148,
+      "step": 15178
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 9.907505285412263e-06,
+      "loss": 0.0143,
+      "step": 15180
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 9.902219873150107e-06,
+      "loss": 0.0121,
+      "step": 15182
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 9.89693446088795e-06,
+      "loss": 0.0063,
+      "step": 15184
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 9.891649048625794e-06,
+      "loss": 0.0178,
+      "step": 15186
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 9.886363636363637e-06,
+      "loss": 0.0131,
+      "step": 15188
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 9.881078224101481e-06,
+      "loss": 0.0121,
+      "step": 15190
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 9.875792811839324e-06,
+      "loss": 0.0007,
+      "step": 15192
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 9.870507399577167e-06,
+      "loss": 0.0022,
+      "step": 15194
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 9.865221987315011e-06,
+      "loss": 0.0076,
+      "step": 15196
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 9.859936575052854e-06,
+      "loss": 0.0417,
+      "step": 15198
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 9.854651162790698e-06,
+      "loss": 0.0325,
+      "step": 15200
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 9.84936575052854e-06,
+      "loss": 0.022,
+      "step": 15202
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 9.844080338266385e-06,
+      "loss": 0.0188,
+      "step": 15204
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 9.83879492600423e-06,
+      "loss": 0.0085,
+      "step": 15206
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 9.833509513742074e-06,
+      "loss": 0.003,
+      "step": 15208
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 9.828224101479916e-06,
+      "loss": 0.0306,
+      "step": 15210
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 9.822938689217759e-06,
+      "loss": 0.0018,
+      "step": 15212
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 9.817653276955603e-06,
+      "loss": 0.0404,
+      "step": 15214
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 9.812367864693446e-06,
+      "loss": 0.0178,
+      "step": 15216
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 9.80708245243129e-06,
+      "loss": 0.0063,
+      "step": 15218
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 9.801797040169133e-06,
+      "loss": 0.0222,
+      "step": 15220
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 9.796511627906978e-06,
+      "loss": 0.0006,
+      "step": 15222
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 9.79122621564482e-06,
+      "loss": 0.0088,
+      "step": 15224
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 9.785940803382665e-06,
+      "loss": 0.014,
+      "step": 15226
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 9.780655391120507e-06,
+      "loss": 0.001,
+      "step": 15228
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 9.775369978858352e-06,
+      "loss": 0.0105,
+      "step": 15230
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 9.770084566596196e-06,
+      "loss": 0.0037,
+      "step": 15232
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 9.764799154334039e-06,
+      "loss": 0.0118,
+      "step": 15234
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 9.759513742071883e-06,
+      "loss": 0.0172,
+      "step": 15236
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 9.754228329809726e-06,
+      "loss": 0.0048,
+      "step": 15238
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 9.74894291754757e-06,
+      "loss": 0.0003,
+      "step": 15240
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 9.743657505285413e-06,
+      "loss": 0.0022,
+      "step": 15242
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 9.738372093023257e-06,
+      "loss": 0.0045,
+      "step": 15244
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 9.7330866807611e-06,
+      "loss": 0.0046,
+      "step": 15246
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 9.727801268498942e-06,
+      "loss": 0.005,
+      "step": 15248
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 9.722515856236787e-06,
+      "loss": 0.0402,
+      "step": 15250
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 9.71723044397463e-06,
+      "loss": 0.0045,
+      "step": 15252
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 9.711945031712474e-06,
+      "loss": 0.0552,
+      "step": 15254
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 9.706659619450317e-06,
+      "loss": 0.0035,
+      "step": 15256
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 9.701374207188161e-06,
+      "loss": 0.0225,
+      "step": 15258
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 9.696088794926005e-06,
+      "loss": 0.069,
+      "step": 15260
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 9.69080338266385e-06,
+      "loss": 0.0067,
+      "step": 15262
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 9.685517970401692e-06,
+      "loss": 0.0032,
+      "step": 15264
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 9.680232558139535e-06,
+      "loss": 0.0054,
+      "step": 15266
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 9.67494714587738e-06,
+      "loss": 0.0207,
+      "step": 15268
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 9.669661733615222e-06,
+      "loss": 0.0059,
+      "step": 15270
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 9.664376321353066e-06,
+      "loss": 0.0024,
+      "step": 15272
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 9.659090909090909e-06,
+      "loss": 0.0056,
+      "step": 15274
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 9.653805496828753e-06,
+      "loss": 0.0047,
+      "step": 15276
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 9.648520084566596e-06,
+      "loss": 0.0111,
+      "step": 15278
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 9.64323467230444e-06,
+      "loss": 0.0028,
+      "step": 15280
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 9.637949260042283e-06,
+      "loss": 0.001,
+      "step": 15282
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 9.632663847780127e-06,
+      "loss": 0.0099,
+      "step": 15284
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 9.627378435517972e-06,
+      "loss": 0.0105,
+      "step": 15286
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 9.622093023255814e-06,
+      "loss": 0.0016,
+      "step": 15288
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 9.616807610993659e-06,
+      "loss": 0.0077,
+      "step": 15290
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 9.611522198731502e-06,
+      "loss": 0.0269,
+      "step": 15292
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 9.606236786469346e-06,
+      "loss": 0.0276,
+      "step": 15294
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 9.600951374207189e-06,
+      "loss": 0.0022,
+      "step": 15296
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 9.595665961945031e-06,
+      "loss": 0.0045,
+      "step": 15298
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 9.590380549682876e-06,
+      "loss": 0.0041,
+      "step": 15300
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 9.585095137420718e-06,
+      "loss": 0.0145,
+      "step": 15302
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 9.579809725158563e-06,
+      "loss": 0.0014,
+      "step": 15304
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 9.574524312896405e-06,
+      "loss": 0.0012,
+      "step": 15306
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 9.56923890063425e-06,
+      "loss": 0.0182,
+      "step": 15308
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 9.563953488372094e-06,
+      "loss": 0.0387,
+      "step": 15310
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 9.558668076109938e-06,
+      "loss": 0.0252,
+      "step": 15312
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 9.553382663847781e-06,
+      "loss": 0.0138,
+      "step": 15314
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 9.548097251585624e-06,
+      "loss": 0.0383,
+      "step": 15316
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 9.542811839323468e-06,
+      "loss": 0.0089,
+      "step": 15318
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 9.53752642706131e-06,
+      "loss": 0.0074,
+      "step": 15320
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 9.532241014799155e-06,
+      "loss": 0.0173,
+      "step": 15322
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 9.526955602536998e-06,
+      "loss": 0.0107,
+      "step": 15324
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 9.521670190274842e-06,
+      "loss": 0.0162,
+      "step": 15326
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 9.516384778012685e-06,
+      "loss": 0.0182,
+      "step": 15328
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 9.51109936575053e-06,
+      "loss": 0.0373,
+      "step": 15330
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 9.505813953488372e-06,
+      "loss": 0.0092,
+      "step": 15332
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 9.500528541226216e-06,
+      "loss": 0.0283,
+      "step": 15334
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 9.495243128964059e-06,
+      "loss": 0.0025,
+      "step": 15336
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 9.489957716701903e-06,
+      "loss": 0.0232,
+      "step": 15338
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 9.484672304439748e-06,
+      "loss": 0.0578,
+      "step": 15340
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 9.47938689217759e-06,
+      "loss": 0.0006,
+      "step": 15342
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 9.474101479915435e-06,
+      "loss": 0.0544,
+      "step": 15344
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 9.468816067653277e-06,
+      "loss": 0.0355,
+      "step": 15346
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 9.463530655391122e-06,
+      "loss": 0.0202,
+      "step": 15348
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 9.458245243128964e-06,
+      "loss": 0.0019,
+      "step": 15350
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 9.452959830866807e-06,
+      "loss": 0.0169,
+      "step": 15352
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 9.447674418604651e-06,
+      "loss": 0.0172,
+      "step": 15354
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 9.442389006342494e-06,
+      "loss": 0.0081,
+      "step": 15356
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 9.437103594080338e-06,
+      "loss": 0.0047,
+      "step": 15358
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 9.431818181818181e-06,
+      "loss": 0.0017,
+      "step": 15360
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 9.426532769556026e-06,
+      "loss": 0.0409,
+      "step": 15362
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 9.42124735729387e-06,
+      "loss": 0.004,
+      "step": 15364
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 9.415961945031714e-06,
+      "loss": 0.0238,
+      "step": 15366
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 9.410676532769557e-06,
+      "loss": 0.0093,
+      "step": 15368
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 9.4053911205074e-06,
+      "loss": 0.0115,
+      "step": 15370
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 9.400105708245244e-06,
+      "loss": 0.0043,
+      "step": 15372
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 9.394820295983087e-06,
+      "loss": 0.0007,
+      "step": 15374
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 9.389534883720931e-06,
+      "loss": 0.0024,
+      "step": 15376
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 9.384249471458774e-06,
+      "loss": 0.0156,
+      "step": 15378
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 9.378964059196618e-06,
+      "loss": 0.019,
+      "step": 15380
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 9.37367864693446e-06,
+      "loss": 0.0122,
+      "step": 15382
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 9.368393234672305e-06,
+      "loss": 0.014,
+      "step": 15384
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 9.363107822410148e-06,
+      "loss": 0.0191,
+      "step": 15386
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 9.357822410147992e-06,
+      "loss": 0.0772,
+      "step": 15388
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 9.352536997885836e-06,
+      "loss": 0.0306,
+      "step": 15390
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 9.347251585623679e-06,
+      "loss": 0.0118,
+      "step": 15392
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 9.341966173361523e-06,
+      "loss": 0.0154,
+      "step": 15394
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 9.336680761099366e-06,
+      "loss": 0.0113,
+      "step": 15396
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 9.33139534883721e-06,
+      "loss": 0.0242,
+      "step": 15398
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 9.326109936575053e-06,
+      "loss": 0.0293,
+      "step": 15400
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 9.320824524312898e-06,
+      "loss": 0.0092,
+      "step": 15402
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 9.31553911205074e-06,
+      "loss": 0.0074,
+      "step": 15404
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 9.310253699788583e-06,
+      "loss": 0.0031,
+      "step": 15406
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 9.304968287526427e-06,
+      "loss": 0.0091,
+      "step": 15408
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 9.29968287526427e-06,
+      "loss": 0.0013,
+      "step": 15410
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 9.294397463002114e-06,
+      "loss": 0.0035,
+      "step": 15412
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 9.289112050739959e-06,
+      "loss": 0.0462,
+      "step": 15414
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 9.283826638477803e-06,
+      "loss": 0.0069,
+      "step": 15416
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 9.278541226215646e-06,
+      "loss": 0.0014,
+      "step": 15418
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 9.27325581395349e-06,
+      "loss": 0.0758,
+      "step": 15420
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 9.267970401691333e-06,
+      "loss": 0.0031,
+      "step": 15422
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 9.262684989429175e-06,
+      "loss": 0.0355,
+      "step": 15424
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 9.25739957716702e-06,
+      "loss": 0.0242,
+      "step": 15426
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 9.252114164904862e-06,
+      "loss": 0.002,
+      "step": 15428
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 9.246828752642707e-06,
+      "loss": 0.0156,
+      "step": 15430
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 9.24154334038055e-06,
+      "loss": 0.0538,
+      "step": 15432
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 9.236257928118394e-06,
+      "loss": 0.0066,
+      "step": 15434
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 9.230972515856237e-06,
+      "loss": 0.0657,
+      "step": 15436
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 9.225687103594081e-06,
+      "loss": 0.0134,
+      "step": 15438
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 9.220401691331924e-06,
+      "loss": 0.0129,
+      "step": 15440
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 9.215116279069768e-06,
+      "loss": 0.0293,
+      "step": 15442
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 9.209830866807612e-06,
+      "loss": 0.0413,
+      "step": 15444
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 9.204545454545455e-06,
+      "loss": 0.0044,
+      "step": 15446
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 9.1992600422833e-06,
+      "loss": 0.0584,
+      "step": 15448
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 9.193974630021142e-06,
+      "loss": 0.1936,
+      "step": 15450
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 9.188689217758986e-06,
+      "loss": 0.0127,
+      "step": 15452
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 9.183403805496829e-06,
+      "loss": 0.0157,
+      "step": 15454
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 9.178118393234673e-06,
+      "loss": 0.0288,
+      "step": 15456
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 9.172832980972516e-06,
+      "loss": 0.0008,
+      "step": 15458
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 9.167547568710359e-06,
+      "loss": 0.0394,
+      "step": 15460
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 9.162262156448203e-06,
+      "loss": 0.0032,
+      "step": 15462
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 9.156976744186046e-06,
+      "loss": 0.035,
+      "step": 15464
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 9.15169133192389e-06,
+      "loss": 0.0011,
+      "step": 15466
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 9.146405919661735e-06,
+      "loss": 0.0099,
+      "step": 15468
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 9.141120507399579e-06,
+      "loss": 0.0036,
+      "step": 15470
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 9.135835095137422e-06,
+      "loss": 0.0008,
+      "step": 15472
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 9.130549682875264e-06,
+      "loss": 0.0411,
+      "step": 15474
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 9.125264270613109e-06,
+      "loss": 0.0199,
+      "step": 15476
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 9.119978858350951e-06,
+      "loss": 0.0192,
+      "step": 15478
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 9.114693446088796e-06,
+      "loss": 0.0034,
+      "step": 15480
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 9.109408033826638e-06,
+      "loss": 0.064,
+      "step": 15482
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 9.104122621564483e-06,
+      "loss": 0.0287,
+      "step": 15484
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 9.098837209302325e-06,
+      "loss": 0.0124,
+      "step": 15486
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 9.09355179704017e-06,
+      "loss": 0.035,
+      "step": 15488
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 9.088266384778012e-06,
+      "loss": 0.0143,
+      "step": 15490
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 9.082980972515857e-06,
+      "loss": 0.0335,
+      "step": 15492
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 9.077695560253701e-06,
+      "loss": 0.0384,
+      "step": 15494
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 9.072410147991544e-06,
+      "loss": 0.039,
+      "step": 15496
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 9.067124735729388e-06,
+      "loss": 0.0183,
+      "step": 15498
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 9.06183932346723e-06,
+      "loss": 0.0215,
+      "step": 15500
+    },
+    {
+      "epoch": 16.38,
+      "eval_cer": 0.04058136221145626,
+      "eval_loss": 0.7027328014373779,
+      "eval_runtime": 129.2071,
+      "eval_samples_per_second": 6.509,
+      "eval_steps_per_second": 0.82,
+      "step": 15500
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 9.056553911205075e-06,
+      "loss": 0.0107,
+      "step": 15502
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 9.051268498942918e-06,
+      "loss": 0.0201,
+      "step": 15504
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 9.045983086680762e-06,
+      "loss": 0.0449,
+      "step": 15506
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 9.040697674418605e-06,
+      "loss": 0.0063,
+      "step": 15508
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 9.035412262156448e-06,
+      "loss": 0.0025,
+      "step": 15510
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 9.030126849894292e-06,
+      "loss": 0.0022,
+      "step": 15512
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 9.024841437632135e-06,
+      "loss": 0.0365,
+      "step": 15514
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 9.019556025369979e-06,
+      "loss": 0.0839,
+      "step": 15516
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 9.014270613107823e-06,
+      "loss": 0.0562,
+      "step": 15518
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 9.008985200845666e-06,
+      "loss": 0.0425,
+      "step": 15520
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 9.00369978858351e-06,
+      "loss": 0.0228,
+      "step": 15522
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 8.998414376321355e-06,
+      "loss": 0.0055,
+      "step": 15524
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 8.993128964059197e-06,
+      "loss": 0.01,
+      "step": 15526
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 8.98784355179704e-06,
+      "loss": 0.0198,
+      "step": 15528
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 8.982558139534884e-06,
+      "loss": 0.0038,
+      "step": 15530
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 8.977272727272727e-06,
+      "loss": 0.0086,
+      "step": 15532
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 8.971987315010571e-06,
+      "loss": 0.0288,
+      "step": 15534
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 8.966701902748414e-06,
+      "loss": 0.0007,
+      "step": 15536
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 8.961416490486259e-06,
+      "loss": 0.0206,
+      "step": 15538
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 8.956131078224101e-06,
+      "loss": 0.0066,
+      "step": 15540
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 8.950845665961946e-06,
+      "loss": 0.0089,
+      "step": 15542
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 8.945560253699788e-06,
+      "loss": 0.0103,
+      "step": 15544
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 8.940274841437633e-06,
+      "loss": 0.0196,
+      "step": 15546
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 8.934989429175477e-06,
+      "loss": 0.02,
+      "step": 15548
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 8.92970401691332e-06,
+      "loss": 0.0585,
+      "step": 15550
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 8.924418604651164e-06,
+      "loss": 0.0204,
+      "step": 15552
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 8.919133192389007e-06,
+      "loss": 0.0195,
+      "step": 15554
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 8.913847780126851e-06,
+      "loss": 0.0055,
+      "step": 15556
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 8.908562367864694e-06,
+      "loss": 0.037,
+      "step": 15558
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 8.903276955602538e-06,
+      "loss": 0.005,
+      "step": 15560
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 8.89799154334038e-06,
+      "loss": 0.0041,
+      "step": 15562
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 8.892706131078223e-06,
+      "loss": 0.0289,
+      "step": 15564
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 8.887420718816068e-06,
+      "loss": 0.0018,
+      "step": 15566
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 8.88213530655391e-06,
+      "loss": 0.0025,
+      "step": 15568
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 8.876849894291755e-06,
+      "loss": 0.0317,
+      "step": 15570
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 8.8715644820296e-06,
+      "loss": 0.0246,
+      "step": 15572
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 8.866279069767444e-06,
+      "loss": 0.0013,
+      "step": 15574
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 8.860993657505286e-06,
+      "loss": 0.0173,
+      "step": 15576
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 8.85570824524313e-06,
+      "loss": 0.018,
+      "step": 15578
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 8.850422832980973e-06,
+      "loss": 0.0055,
+      "step": 15580
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 8.845137420718816e-06,
+      "loss": 0.0109,
+      "step": 15582
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 8.83985200845666e-06,
+      "loss": 0.0013,
+      "step": 15584
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 8.834566596194503e-06,
+      "loss": 0.0017,
+      "step": 15586
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 8.829281183932347e-06,
+      "loss": 0.0039,
+      "step": 15588
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 8.82399577167019e-06,
+      "loss": 0.002,
+      "step": 15590
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 8.818710359408034e-06,
+      "loss": 0.0175,
+      "step": 15592
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 8.813424947145877e-06,
+      "loss": 0.0164,
+      "step": 15594
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 8.808139534883721e-06,
+      "loss": 0.0724,
+      "step": 15596
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 8.802854122621566e-06,
+      "loss": 0.0191,
+      "step": 15598
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 8.797568710359408e-06,
+      "loss": 0.053,
+      "step": 15600
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 8.792283298097253e-06,
+      "loss": 0.0191,
+      "step": 15602
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 8.786997885835095e-06,
+      "loss": 0.0064,
+      "step": 15604
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 8.78171247357294e-06,
+      "loss": 0.0036,
+      "step": 15606
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 8.776427061310782e-06,
+      "loss": 0.0846,
+      "step": 15608
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 8.771141649048627e-06,
+      "loss": 0.0507,
+      "step": 15610
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 8.76585623678647e-06,
+      "loss": 0.0054,
+      "step": 15612
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 8.760570824524314e-06,
+      "loss": 0.0069,
+      "step": 15614
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 8.755285412262157e-06,
+      "loss": 0.0042,
+      "step": 15616
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 8.75e-06,
+      "loss": 0.0057,
+      "step": 15618
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 8.744714587737844e-06,
+      "loss": 0.0052,
+      "step": 15620
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 8.739429175475688e-06,
+      "loss": 0.0015,
+      "step": 15622
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 8.73414376321353e-06,
+      "loss": 0.0027,
+      "step": 15624
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 8.728858350951375e-06,
+      "loss": 0.0023,
+      "step": 15626
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 8.72357293868922e-06,
+      "loss": 0.0057,
+      "step": 15628
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 8.718287526427062e-06,
+      "loss": 0.0022,
+      "step": 15630
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 8.713002114164906e-06,
+      "loss": 0.0113,
+      "step": 15632
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 8.707716701902749e-06,
+      "loss": 0.0017,
+      "step": 15634
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 8.702431289640592e-06,
+      "loss": 0.0328,
+      "step": 15636
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 8.697145877378436e-06,
+      "loss": 0.0094,
+      "step": 15638
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 8.691860465116279e-06,
+      "loss": 0.0201,
+      "step": 15640
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 8.686575052854123e-06,
+      "loss": 0.0059,
+      "step": 15642
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 8.681289640591966e-06,
+      "loss": 0.0061,
+      "step": 15644
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 8.67600422832981e-06,
+      "loss": 0.0123,
+      "step": 15646
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 8.670718816067653e-06,
+      "loss": 0.038,
+      "step": 15648
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 8.665433403805497e-06,
+      "loss": 0.0219,
+      "step": 15650
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 8.660147991543342e-06,
+      "loss": 0.0099,
+      "step": 15652
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 8.654862579281184e-06,
+      "loss": 0.019,
+      "step": 15654
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 8.649577167019029e-06,
+      "loss": 0.001,
+      "step": 15656
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 8.644291754756871e-06,
+      "loss": 0.0327,
+      "step": 15658
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 8.639006342494716e-06,
+      "loss": 0.0166,
+      "step": 15660
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 8.633720930232558e-06,
+      "loss": 0.0084,
+      "step": 15662
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 8.628435517970403e-06,
+      "loss": 0.0051,
+      "step": 15664
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 8.623150105708245e-06,
+      "loss": 0.0303,
+      "step": 15666
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 8.617864693446088e-06,
+      "loss": 0.0372,
+      "step": 15668
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 8.612579281183932e-06,
+      "loss": 0.0384,
+      "step": 15670
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 8.607293868921775e-06,
+      "loss": 0.0705,
+      "step": 15672
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 8.60200845665962e-06,
+      "loss": 0.012,
+      "step": 15674
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 8.596723044397464e-06,
+      "loss": 0.0019,
+      "step": 15676
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 8.591437632135308e-06,
+      "loss": 0.0009,
+      "step": 15678
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 8.58615221987315e-06,
+      "loss": 0.0035,
+      "step": 15680
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 8.580866807610995e-06,
+      "loss": 0.0016,
+      "step": 15682
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 8.575581395348838e-06,
+      "loss": 0.0076,
+      "step": 15684
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 8.57029598308668e-06,
+      "loss": 0.0049,
+      "step": 15686
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 8.565010570824525e-06,
+      "loss": 0.0144,
+      "step": 15688
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 8.559725158562368e-06,
+      "loss": 0.0011,
+      "step": 15690
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 8.554439746300212e-06,
+      "loss": 0.0017,
+      "step": 15692
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 8.549154334038055e-06,
+      "loss": 0.0122,
+      "step": 15694
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 8.543868921775899e-06,
+      "loss": 0.0038,
+      "step": 15696
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 8.538583509513742e-06,
+      "loss": 0.0005,
+      "step": 15698
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 8.533298097251586e-06,
+      "loss": 0.0341,
+      "step": 15700
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 8.52801268498943e-06,
+      "loss": 0.0007,
+      "step": 15702
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 8.522727272727273e-06,
+      "loss": 0.0677,
+      "step": 15704
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 8.517441860465117e-06,
+      "loss": 0.0053,
+      "step": 15706
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 8.51215644820296e-06,
+      "loss": 0.0009,
+      "step": 15708
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 8.506871035940804e-06,
+      "loss": 0.0008,
+      "step": 15710
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 8.501585623678647e-06,
+      "loss": 0.0008,
+      "step": 15712
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 8.496300211416491e-06,
+      "loss": 0.0045,
+      "step": 15714
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 8.491014799154334e-06,
+      "loss": 0.018,
+      "step": 15716
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 8.485729386892179e-06,
+      "loss": 0.0387,
+      "step": 15718
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 8.480443974630021e-06,
+      "loss": 0.0098,
+      "step": 15720
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 8.475158562367864e-06,
+      "loss": 0.0036,
+      "step": 15722
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 8.469873150105708e-06,
+      "loss": 0.0045,
+      "step": 15724
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 8.464587737843551e-06,
+      "loss": 0.023,
+      "step": 15726
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 8.459302325581395e-06,
+      "loss": 0.0013,
+      "step": 15728
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 8.45401691331924e-06,
+      "loss": 0.0501,
+      "step": 15730
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 8.448731501057084e-06,
+      "loss": 0.0244,
+      "step": 15732
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 8.443446088794927e-06,
+      "loss": 0.003,
+      "step": 15734
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 8.438160676532771e-06,
+      "loss": 0.0093,
+      "step": 15736
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 8.432875264270614e-06,
+      "loss": 0.0012,
+      "step": 15738
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 8.427589852008456e-06,
+      "loss": 0.0012,
+      "step": 15740
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 8.4223044397463e-06,
+      "loss": 0.0007,
+      "step": 15742
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 8.417019027484143e-06,
+      "loss": 0.003,
+      "step": 15744
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 8.411733615221988e-06,
+      "loss": 0.0197,
+      "step": 15746
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 8.40644820295983e-06,
+      "loss": 0.0114,
+      "step": 15748
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 8.401162790697675e-06,
+      "loss": 0.0009,
+      "step": 15750
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 8.395877378435517e-06,
+      "loss": 0.0244,
+      "step": 15752
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 8.390591966173362e-06,
+      "loss": 0.0402,
+      "step": 15754
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 8.385306553911206e-06,
+      "loss": 0.0028,
+      "step": 15756
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 8.380021141649049e-06,
+      "loss": 0.0157,
+      "step": 15758
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 8.374735729386893e-06,
+      "loss": 0.0052,
+      "step": 15760
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 8.369450317124736e-06,
+      "loss": 0.0008,
+      "step": 15762
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 8.36416490486258e-06,
+      "loss": 0.0569,
+      "step": 15764
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 8.358879492600423e-06,
+      "loss": 0.0022,
+      "step": 15766
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 8.353594080338267e-06,
+      "loss": 0.0005,
+      "step": 15768
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 8.34830866807611e-06,
+      "loss": 0.0443,
+      "step": 15770
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 8.343023255813954e-06,
+      "loss": 0.0037,
+      "step": 15772
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 8.337737843551797e-06,
+      "loss": 0.0637,
+      "step": 15774
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 8.33245243128964e-06,
+      "loss": 0.0059,
+      "step": 15776
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 8.327167019027484e-06,
+      "loss": 0.0065,
+      "step": 15778
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 8.321881606765328e-06,
+      "loss": 0.0043,
+      "step": 15780
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 8.316596194503173e-06,
+      "loss": 0.0005,
+      "step": 15782
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 8.311310782241015e-06,
+      "loss": 0.0015,
+      "step": 15784
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 8.30602536997886e-06,
+      "loss": 0.0041,
+      "step": 15786
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 8.300739957716703e-06,
+      "loss": 0.0269,
+      "step": 15788
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 8.295454545454547e-06,
+      "loss": 0.0046,
+      "step": 15790
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 8.29016913319239e-06,
+      "loss": 0.029,
+      "step": 15792
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 8.284883720930232e-06,
+      "loss": 0.0044,
+      "step": 15794
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 8.279598308668077e-06,
+      "loss": 0.0115,
+      "step": 15796
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 8.27431289640592e-06,
+      "loss": 0.0472,
+      "step": 15798
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 8.269027484143764e-06,
+      "loss": 0.0076,
+      "step": 15800
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 8.263742071881606e-06,
+      "loss": 0.0065,
+      "step": 15802
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 8.25845665961945e-06,
+      "loss": 0.0278,
+      "step": 15804
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 8.253171247357295e-06,
+      "loss": 0.0006,
+      "step": 15806
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 8.247885835095138e-06,
+      "loss": 0.0354,
+      "step": 15808
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 8.242600422832982e-06,
+      "loss": 0.0012,
+      "step": 15810
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 8.237315010570825e-06,
+      "loss": 0.0095,
+      "step": 15812
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 8.232029598308669e-06,
+      "loss": 0.0546,
+      "step": 15814
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 8.226744186046512e-06,
+      "loss": 0.0153,
+      "step": 15816
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 8.221458773784356e-06,
+      "loss": 0.0081,
+      "step": 15818
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 8.216173361522199e-06,
+      "loss": 0.0276,
+      "step": 15820
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 8.210887949260043e-06,
+      "loss": 0.0224,
+      "step": 15822
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 8.205602536997886e-06,
+      "loss": 0.0207,
+      "step": 15824
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 8.200317124735729e-06,
+      "loss": 0.0153,
+      "step": 15826
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 8.195031712473573e-06,
+      "loss": 0.0632,
+      "step": 15828
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 8.189746300211416e-06,
+      "loss": 0.0229,
+      "step": 15830
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 8.18446088794926e-06,
+      "loss": 0.004,
+      "step": 15832
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 8.179175475687104e-06,
+      "loss": 0.0379,
+      "step": 15834
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 8.173890063424949e-06,
+      "loss": 0.0025,
+      "step": 15836
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 8.168604651162791e-06,
+      "loss": 0.0197,
+      "step": 15838
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 8.163319238900636e-06,
+      "loss": 0.0026,
+      "step": 15840
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 8.158033826638478e-06,
+      "loss": 0.0005,
+      "step": 15842
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 8.152748414376321e-06,
+      "loss": 0.0158,
+      "step": 15844
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 8.147463002114165e-06,
+      "loss": 0.0148,
+      "step": 15846
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 8.142177589852008e-06,
+      "loss": 0.025,
+      "step": 15848
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 8.136892177589852e-06,
+      "loss": 0.0016,
+      "step": 15850
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 8.131606765327695e-06,
+      "loss": 0.0008,
+      "step": 15852
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 8.12632135306554e-06,
+      "loss": 0.0041,
+      "step": 15854
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 8.121035940803382e-06,
+      "loss": 0.0118,
+      "step": 15856
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 8.115750528541227e-06,
+      "loss": 0.0253,
+      "step": 15858
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 8.110465116279071e-06,
+      "loss": 0.0133,
+      "step": 15860
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 8.105179704016914e-06,
+      "loss": 0.0063,
+      "step": 15862
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 8.099894291754758e-06,
+      "loss": 0.0136,
+      "step": 15864
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 8.0946088794926e-06,
+      "loss": 0.003,
+      "step": 15866
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 8.089323467230445e-06,
+      "loss": 0.0086,
+      "step": 15868
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 8.084038054968288e-06,
+      "loss": 0.003,
+      "step": 15870
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 8.078752642706132e-06,
+      "loss": 0.0041,
+      "step": 15872
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 8.073467230443975e-06,
+      "loss": 0.0026,
+      "step": 15874
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 8.068181818181819e-06,
+      "loss": 0.0174,
+      "step": 15876
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 8.062896405919662e-06,
+      "loss": 0.0183,
+      "step": 15878
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 8.057610993657504e-06,
+      "loss": 0.005,
+      "step": 15880
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 8.052325581395349e-06,
+      "loss": 0.0087,
+      "step": 15882
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 8.047040169133193e-06,
+      "loss": 0.0174,
+      "step": 15884
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 8.041754756871037e-06,
+      "loss": 0.0063,
+      "step": 15886
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 8.03646934460888e-06,
+      "loss": 0.0044,
+      "step": 15888
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 8.031183932346724e-06,
+      "loss": 0.0145,
+      "step": 15890
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 8.025898520084567e-06,
+      "loss": 0.0314,
+      "step": 15892
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 8.020613107822412e-06,
+      "loss": 0.0004,
+      "step": 15894
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 8.015327695560254e-06,
+      "loss": 0.0327,
+      "step": 15896
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 8.010042283298097e-06,
+      "loss": 0.0079,
+      "step": 15898
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 8.004756871035941e-06,
+      "loss": 0.0295,
+      "step": 15900
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 7.999471458773784e-06,
+      "loss": 0.0153,
+      "step": 15902
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 7.994186046511628e-06,
+      "loss": 0.0069,
+      "step": 15904
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 7.988900634249471e-06,
+      "loss": 0.001,
+      "step": 15906
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 7.983615221987315e-06,
+      "loss": 0.0023,
+      "step": 15908
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 7.978329809725158e-06,
+      "loss": 0.0239,
+      "step": 15910
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 7.973044397463002e-06,
+      "loss": 0.0301,
+      "step": 15912
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 7.967758985200847e-06,
+      "loss": 0.0061,
+      "step": 15914
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 7.96247357293869e-06,
+      "loss": 0.0106,
+      "step": 15916
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 7.957188160676534e-06,
+      "loss": 0.0008,
+      "step": 15918
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 7.951902748414376e-06,
+      "loss": 0.0302,
+      "step": 15920
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 7.94661733615222e-06,
+      "loss": 0.014,
+      "step": 15922
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 7.941331923890063e-06,
+      "loss": 0.007,
+      "step": 15924
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 7.936046511627908e-06,
+      "loss": 0.0229,
+      "step": 15926
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 7.93076109936575e-06,
+      "loss": 0.0071,
+      "step": 15928
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 7.925475687103595e-06,
+      "loss": 0.0022,
+      "step": 15930
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 7.920190274841438e-06,
+      "loss": 0.0043,
+      "step": 15932
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 7.91490486257928e-06,
+      "loss": 0.023,
+      "step": 15934
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 7.909619450317125e-06,
+      "loss": 0.0016,
+      "step": 15936
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 7.904334038054969e-06,
+      "loss": 0.0009,
+      "step": 15938
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 7.899048625792813e-06,
+      "loss": 0.0116,
+      "step": 15940
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 7.893763213530656e-06,
+      "loss": 0.0317,
+      "step": 15942
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 7.8884778012685e-06,
+      "loss": 0.0041,
+      "step": 15944
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 7.883192389006343e-06,
+      "loss": 0.0323,
+      "step": 15946
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 7.877906976744187e-06,
+      "loss": 0.0025,
+      "step": 15948
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 7.87262156448203e-06,
+      "loss": 0.0025,
+      "step": 15950
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 7.867336152219873e-06,
+      "loss": 0.0026,
+      "step": 15952
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 7.862050739957717e-06,
+      "loss": 0.0046,
+      "step": 15954
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 7.85676532769556e-06,
+      "loss": 0.0028,
+      "step": 15956
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 7.851479915433404e-06,
+      "loss": 0.0106,
+      "step": 15958
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 7.846194503171247e-06,
+      "loss": 0.0023,
+      "step": 15960
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 7.840909090909091e-06,
+      "loss": 0.0007,
+      "step": 15962
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 7.835623678646936e-06,
+      "loss": 0.0119,
+      "step": 15964
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 7.83033826638478e-06,
+      "loss": 0.0067,
+      "step": 15966
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 7.825052854122623e-06,
+      "loss": 0.0019,
+      "step": 15968
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 7.819767441860465e-06,
+      "loss": 0.0025,
+      "step": 15970
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 7.81448202959831e-06,
+      "loss": 0.007,
+      "step": 15972
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 7.809196617336152e-06,
+      "loss": 0.0025,
+      "step": 15974
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 7.803911205073997e-06,
+      "loss": 0.0012,
+      "step": 15976
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 7.79862579281184e-06,
+      "loss": 0.0012,
+      "step": 15978
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 7.793340380549684e-06,
+      "loss": 0.005,
+      "step": 15980
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 7.788054968287526e-06,
+      "loss": 0.0006,
+      "step": 15982
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 7.782769556025369e-06,
+      "loss": 0.0004,
+      "step": 15984
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 7.777484143763213e-06,
+      "loss": 0.002,
+      "step": 15986
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 7.772198731501058e-06,
+      "loss": 0.0265,
+      "step": 15988
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 7.766913319238902e-06,
+      "loss": 0.0074,
+      "step": 15990
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 7.761627906976745e-06,
+      "loss": 0.0025,
+      "step": 15992
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 7.756342494714589e-06,
+      "loss": 0.0256,
+      "step": 15994
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 7.751057082452432e-06,
+      "loss": 0.0007,
+      "step": 15996
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 7.745771670190276e-06,
+      "loss": 0.0028,
+      "step": 15998
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 7.740486257928119e-06,
+      "loss": 0.0029,
+      "step": 16000
+    },
+    {
+      "epoch": 16.91,
+      "eval_cer": 0.06184098033627814,
+      "eval_loss": 0.898295521736145,
+      "eval_runtime": 124.9218,
+      "eval_samples_per_second": 6.732,
+      "eval_steps_per_second": 0.849,
+      "step": 16000
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 7.735200845665962e-06,
+      "loss": 0.0036,
+      "step": 16002
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 7.729915433403806e-06,
+      "loss": 0.0021,
+      "step": 16004
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 7.724630021141649e-06,
+      "loss": 0.0006,
+      "step": 16006
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 7.719344608879493e-06,
+      "loss": 0.0126,
+      "step": 16008
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 7.714059196617336e-06,
+      "loss": 0.0391,
+      "step": 16010
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 7.70877378435518e-06,
+      "loss": 0.0153,
+      "step": 16012
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 7.703488372093023e-06,
+      "loss": 0.0036,
+      "step": 16014
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 7.698202959830867e-06,
+      "loss": 0.0109,
+      "step": 16016
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 7.692917547568711e-06,
+      "loss": 0.0096,
+      "step": 16018
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 7.687632135306554e-06,
+      "loss": 0.0146,
+      "step": 16020
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 7.682346723044398e-06,
+      "loss": 0.0411,
+      "step": 16022
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 7.677061310782241e-06,
+      "loss": 0.0046,
+      "step": 16024
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 7.671775898520085e-06,
+      "loss": 0.0061,
+      "step": 16026
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 7.666490486257928e-06,
+      "loss": 0.0031,
+      "step": 16028
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 7.661205073995772e-06,
+      "loss": 0.0096,
+      "step": 16030
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 7.655919661733615e-06,
+      "loss": 0.0065,
+      "step": 16032
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 7.65063424947146e-06,
+      "loss": 0.0007,
+      "step": 16034
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 7.645348837209302e-06,
+      "loss": 0.0067,
+      "step": 16036
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 7.640063424947145e-06,
+      "loss": 0.0016,
+      "step": 16038
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 7.63477801268499e-06,
+      "loss": 0.0039,
+      "step": 16040
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 7.629492600422834e-06,
+      "loss": 0.017,
+      "step": 16042
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 7.624207188160677e-06,
+      "loss": 0.004,
+      "step": 16044
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 7.61892177589852e-06,
+      "loss": 0.0162,
+      "step": 16046
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 7.613636363636364e-06,
+      "loss": 0.0003,
+      "step": 16048
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 7.608350951374208e-06,
+      "loss": 0.0022,
+      "step": 16050
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 7.603065539112052e-06,
+      "loss": 0.0284,
+      "step": 16052
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 7.597780126849895e-06,
+      "loss": 0.0081,
+      "step": 16054
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 7.592494714587737e-06,
+      "loss": 0.0081,
+      "step": 16056
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 7.587209302325582e-06,
+      "loss": 0.0119,
+      "step": 16058
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 7.581923890063425e-06,
+      "loss": 0.0114,
+      "step": 16060
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 7.576638477801269e-06,
+      "loss": 0.0038,
+      "step": 16062
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 7.571353065539112e-06,
+      "loss": 0.0042,
+      "step": 16064
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 7.566067653276957e-06,
+      "loss": 0.0329,
+      "step": 16066
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 7.560782241014799e-06,
+      "loss": 0.0011,
+      "step": 16068
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 7.555496828752644e-06,
+      "loss": 0.0017,
+      "step": 16070
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 7.550211416490486e-06,
+      "loss": 0.0037,
+      "step": 16072
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 7.54492600422833e-06,
+      "loss": 0.0032,
+      "step": 16074
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 7.539640591966174e-06,
+      "loss": 0.0028,
+      "step": 16076
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 7.534355179704017e-06,
+      "loss": 0.0038,
+      "step": 16078
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 7.529069767441861e-06,
+      "loss": 0.026,
+      "step": 16080
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 7.523784355179704e-06,
+      "loss": 0.0007,
+      "step": 16082
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 7.518498942917548e-06,
+      "loss": 0.001,
+      "step": 16084
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 7.513213530655391e-06,
+      "loss": 0.021,
+      "step": 16086
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 7.507928118393235e-06,
+      "loss": 0.0175,
+      "step": 16088
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 7.502642706131079e-06,
+      "loss": 0.0011,
+      "step": 16090
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 7.4973572938689215e-06,
+      "loss": 0.0009,
+      "step": 16092
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 7.492071881606766e-06,
+      "loss": 0.0044,
+      "step": 16094
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 7.4867864693446086e-06,
+      "loss": 0.0152,
+      "step": 16096
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 7.481501057082453e-06,
+      "loss": 0.0012,
+      "step": 16098
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 7.4762156448202964e-06,
+      "loss": 0.0022,
+      "step": 16100
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 7.470930232558141e-06,
+      "loss": 0.0017,
+      "step": 16102
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 7.4656448202959835e-06,
+      "loss": 0.0019,
+      "step": 16104
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 7.460359408033828e-06,
+      "loss": 0.0055,
+      "step": 16106
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 7.4550739957716705e-06,
+      "loss": 0.0068,
+      "step": 16108
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 7.449788583509513e-06,
+      "loss": 0.0033,
+      "step": 16110
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 7.4445031712473576e-06,
+      "loss": 0.0175,
+      "step": 16112
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 7.439217758985201e-06,
+      "loss": 0.0005,
+      "step": 16114
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 7.4339323467230454e-06,
+      "loss": 0.0096,
+      "step": 16116
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 7.428646934460888e-06,
+      "loss": 0.004,
+      "step": 16118
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 7.4233615221987325e-06,
+      "loss": 0.0256,
+      "step": 16120
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 7.418076109936575e-06,
+      "loss": 0.013,
+      "step": 16122
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 7.4127906976744195e-06,
+      "loss": 0.0216,
+      "step": 16124
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 7.407505285412262e-06,
+      "loss": 0.0032,
+      "step": 16126
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 7.402219873150106e-06,
+      "loss": 0.0082,
+      "step": 16128
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 7.39693446088795e-06,
+      "loss": 0.0007,
+      "step": 16130
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 7.391649048625793e-06,
+      "loss": 0.0019,
+      "step": 16132
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 7.386363636363637e-06,
+      "loss": 0.0007,
+      "step": 16134
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 7.38107822410148e-06,
+      "loss": 0.0008,
+      "step": 16136
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 7.375792811839324e-06,
+      "loss": 0.0093,
+      "step": 16138
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 7.370507399577168e-06,
+      "loss": 0.0072,
+      "step": 16140
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 7.36522198731501e-06,
+      "loss": 0.0097,
+      "step": 16142
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 7.359936575052855e-06,
+      "loss": 0.0028,
+      "step": 16144
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 7.354651162790697e-06,
+      "loss": 0.0212,
+      "step": 16146
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 7.349365750528542e-06,
+      "loss": 0.0076,
+      "step": 16148
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 7.344080338266384e-06,
+      "loss": 0.0006,
+      "step": 16150
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 7.338794926004229e-06,
+      "loss": 0.0008,
+      "step": 16152
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 7.333509513742072e-06,
+      "loss": 0.0017,
+      "step": 16154
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 7.328224101479917e-06,
+      "loss": 0.0008,
+      "step": 16156
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 7.322938689217759e-06,
+      "loss": 0.005,
+      "step": 16158
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 7.317653276955602e-06,
+      "loss": 0.0022,
+      "step": 16160
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 7.312367864693446e-06,
+      "loss": 0.001,
+      "step": 16162
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 7.30708245243129e-06,
+      "loss": 0.0022,
+      "step": 16164
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 7.301797040169133e-06,
+      "loss": 0.0132,
+      "step": 16166
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 7.296511627906977e-06,
+      "loss": 0.0175,
+      "step": 16168
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 7.291226215644821e-06,
+      "loss": 0.0066,
+      "step": 16170
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 7.285940803382664e-06,
+      "loss": 0.0012,
+      "step": 16172
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 7.280655391120508e-06,
+      "loss": 0.0157,
+      "step": 16174
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 7.275369978858351e-06,
+      "loss": 0.0004,
+      "step": 16176
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 7.2700845665961945e-06,
+      "loss": 0.0009,
+      "step": 16178
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 7.264799154334039e-06,
+      "loss": 0.0012,
+      "step": 16180
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 7.2595137420718815e-06,
+      "loss": 0.0054,
+      "step": 16182
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 7.254228329809726e-06,
+      "loss": 0.0018,
+      "step": 16184
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 7.248942917547569e-06,
+      "loss": 0.0034,
+      "step": 16186
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 7.243657505285413e-06,
+      "loss": 0.0015,
+      "step": 16188
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 7.238372093023256e-06,
+      "loss": 0.0046,
+      "step": 16190
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 7.2330866807611e-06,
+      "loss": 0.0049,
+      "step": 16192
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 7.2278012684989435e-06,
+      "loss": 0.0222,
+      "step": 16194
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 7.222515856236786e-06,
+      "loss": 0.0074,
+      "step": 16196
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 7.2172304439746305e-06,
+      "loss": 0.0017,
+      "step": 16198
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 7.211945031712473e-06,
+      "loss": 0.0007,
+      "step": 16200
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 7.2066596194503176e-06,
+      "loss": 0.0201,
+      "step": 16202
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 7.201374207188161e-06,
+      "loss": 0.0021,
+      "step": 16204
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 7.196088794926005e-06,
+      "loss": 0.0318,
+      "step": 16206
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 7.190803382663848e-06,
+      "loss": 0.0256,
+      "step": 16208
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 7.1855179704016925e-06,
+      "loss": 0.0006,
+      "step": 16210
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 7.180232558139535e-06,
+      "loss": 0.0008,
+      "step": 16212
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 7.174947145877378e-06,
+      "loss": 0.0055,
+      "step": 16214
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 7.169661733615222e-06,
+      "loss": 0.0084,
+      "step": 16216
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 7.164376321353066e-06,
+      "loss": 0.0011,
+      "step": 16218
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 7.15909090909091e-06,
+      "loss": 0.0011,
+      "step": 16220
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 7.153805496828753e-06,
+      "loss": 0.0394,
+      "step": 16222
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 7.148520084566597e-06,
+      "loss": 0.0011,
+      "step": 16224
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 7.14323467230444e-06,
+      "loss": 0.0027,
+      "step": 16226
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 7.137949260042284e-06,
+      "loss": 0.0044,
+      "step": 16228
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 7.132663847780127e-06,
+      "loss": 0.0014,
+      "step": 16230
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 7.12737843551797e-06,
+      "loss": 0.0003,
+      "step": 16232
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 7.122093023255815e-06,
+      "loss": 0.0099,
+      "step": 16234
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 7.116807610993657e-06,
+      "loss": 0.0033,
+      "step": 16236
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 7.111522198731502e-06,
+      "loss": 0.0008,
+      "step": 16238
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 7.106236786469344e-06,
+      "loss": 0.0005,
+      "step": 16240
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 7.100951374207189e-06,
+      "loss": 0.0012,
+      "step": 16242
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 7.095665961945032e-06,
+      "loss": 0.0004,
+      "step": 16244
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 7.090380549682876e-06,
+      "loss": 0.0134,
+      "step": 16246
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 7.085095137420719e-06,
+      "loss": 0.0119,
+      "step": 16248
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 7.079809725158562e-06,
+      "loss": 0.0072,
+      "step": 16250
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 7.074524312896406e-06,
+      "loss": 0.0175,
+      "step": 16252
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 7.069238900634249e-06,
+      "loss": 0.0004,
+      "step": 16254
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 7.063953488372093e-06,
+      "loss": 0.0064,
+      "step": 16256
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 7.058668076109937e-06,
+      "loss": 0.0226,
+      "step": 16258
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 7.053382663847781e-06,
+      "loss": 0.0171,
+      "step": 16260
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 7.048097251585624e-06,
+      "loss": 0.0052,
+      "step": 16262
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 7.042811839323468e-06,
+      "loss": 0.0022,
+      "step": 16264
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 7.037526427061311e-06,
+      "loss": 0.0127,
+      "step": 16266
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 7.032241014799154e-06,
+      "loss": 0.0005,
+      "step": 16268
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 7.026955602536998e-06,
+      "loss": 0.0222,
+      "step": 16270
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 7.0216701902748416e-06,
+      "loss": 0.0011,
+      "step": 16272
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 7.016384778012686e-06,
+      "loss": 0.0004,
+      "step": 16274
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 7.011099365750529e-06,
+      "loss": 0.0028,
+      "step": 16276
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 7.005813953488373e-06,
+      "loss": 0.0113,
+      "step": 16278
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 7.000528541226216e-06,
+      "loss": 0.0286,
+      "step": 16280
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 6.99524312896406e-06,
+      "loss": 0.0013,
+      "step": 16282
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 6.9899577167019035e-06,
+      "loss": 0.0129,
+      "step": 16284
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 6.984672304439746e-06,
+      "loss": 0.0817,
+      "step": 16286
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 6.9793868921775905e-06,
+      "loss": 0.0009,
+      "step": 16288
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 6.974101479915433e-06,
+      "loss": 0.0052,
+      "step": 16290
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 6.968816067653278e-06,
+      "loss": 0.0012,
+      "step": 16292
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 6.96353065539112e-06,
+      "loss": 0.014,
+      "step": 16294
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 6.958245243128965e-06,
+      "loss": 0.0027,
+      "step": 16296
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 6.952959830866808e-06,
+      "loss": 0.0091,
+      "step": 16298
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 6.9476744186046525e-06,
+      "loss": 0.0008,
+      "step": 16300
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 6.942389006342495e-06,
+      "loss": 0.0025,
+      "step": 16302
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 6.937103594080338e-06,
+      "loss": 0.0054,
+      "step": 16304
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 6.931818181818182e-06,
+      "loss": 0.0019,
+      "step": 16306
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 6.926532769556026e-06,
+      "loss": 0.0156,
+      "step": 16308
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 6.921247357293869e-06,
+      "loss": 0.0007,
+      "step": 16310
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 6.915961945031713e-06,
+      "loss": 0.0075,
+      "step": 16312
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 6.910676532769557e-06,
+      "loss": 0.0142,
+      "step": 16314
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 6.9053911205074e-06,
+      "loss": 0.0168,
+      "step": 16316
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 6.9001057082452425e-06,
+      "loss": 0.0031,
+      "step": 16318
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 6.894820295983087e-06,
+      "loss": 0.002,
+      "step": 16320
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 6.88953488372093e-06,
+      "loss": 0.0541,
+      "step": 16322
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 6.884249471458775e-06,
+      "loss": 0.0205,
+      "step": 16324
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 6.878964059196617e-06,
+      "loss": 0.0156,
+      "step": 16326
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 6.873678646934462e-06,
+      "loss": 0.011,
+      "step": 16328
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 6.8683932346723044e-06,
+      "loss": 0.0022,
+      "step": 16330
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 6.863107822410149e-06,
+      "loss": 0.0043,
+      "step": 16332
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 6.8578224101479915e-06,
+      "loss": 0.0053,
+      "step": 16334
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 6.852536997885835e-06,
+      "loss": 0.02,
+      "step": 16336
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 6.847251585623679e-06,
+      "loss": 0.0137,
+      "step": 16338
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 6.841966173361522e-06,
+      "loss": 0.0012,
+      "step": 16340
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 6.836680761099366e-06,
+      "loss": 0.0009,
+      "step": 16342
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 6.831395348837209e-06,
+      "loss": 0.0044,
+      "step": 16344
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 6.8261099365750534e-06,
+      "loss": 0.0085,
+      "step": 16346
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 6.820824524312897e-06,
+      "loss": 0.0034,
+      "step": 16348
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 6.8155391120507405e-06,
+      "loss": 0.0038,
+      "step": 16350
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 6.810253699788584e-06,
+      "loss": 0.0003,
+      "step": 16352
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 6.804968287526427e-06,
+      "loss": 0.007,
+      "step": 16354
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 6.799682875264271e-06,
+      "loss": 0.0029,
+      "step": 16356
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 6.794397463002114e-06,
+      "loss": 0.0007,
+      "step": 16358
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 6.789112050739958e-06,
+      "loss": 0.0047,
+      "step": 16360
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 6.7838266384778016e-06,
+      "loss": 0.015,
+      "step": 16362
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 6.778541226215646e-06,
+      "loss": 0.0027,
+      "step": 16364
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 6.773255813953489e-06,
+      "loss": 0.0054,
+      "step": 16366
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 6.767970401691333e-06,
+      "loss": 0.0051,
+      "step": 16368
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 6.762684989429176e-06,
+      "loss": 0.0084,
+      "step": 16370
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 6.757399577167018e-06,
+      "loss": 0.0005,
+      "step": 16372
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 6.752114164904863e-06,
+      "loss": 0.016,
+      "step": 16374
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 6.746828752642706e-06,
+      "loss": 0.0017,
+      "step": 16376
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 6.7415433403805506e-06,
+      "loss": 0.0436,
+      "step": 16378
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 6.736257928118393e-06,
+      "loss": 0.0104,
+      "step": 16380
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 6.730972515856238e-06,
+      "loss": 0.001,
+      "step": 16382
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 6.72568710359408e-06,
+      "loss": 0.0574,
+      "step": 16384
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 6.720401691331925e-06,
+      "loss": 0.0078,
+      "step": 16386
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 6.715116279069768e-06,
+      "loss": 0.0009,
+      "step": 16388
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 6.709830866807611e-06,
+      "loss": 0.0058,
+      "step": 16390
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 6.704545454545455e-06,
+      "loss": 0.0005,
+      "step": 16392
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 6.699260042283298e-06,
+      "loss": 0.0012,
+      "step": 16394
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 6.693974630021142e-06,
+      "loss": 0.0379,
+      "step": 16396
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 6.688689217758985e-06,
+      "loss": 0.0008,
+      "step": 16398
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 6.683403805496829e-06,
+      "loss": 0.0317,
+      "step": 16400
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 6.678118393234673e-06,
+      "loss": 0.0498,
+      "step": 16402
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 6.672832980972517e-06,
+      "loss": 0.0172,
+      "step": 16404
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 6.66754756871036e-06,
+      "loss": 0.0015,
+      "step": 16406
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 6.6622621564482025e-06,
+      "loss": 0.0068,
+      "step": 16408
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 6.656976744186047e-06,
+      "loss": 0.0037,
+      "step": 16410
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 6.6516913319238895e-06,
+      "loss": 0.0012,
+      "step": 16412
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 6.646405919661734e-06,
+      "loss": 0.0078,
+      "step": 16414
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 6.641120507399577e-06,
+      "loss": 0.0044,
+      "step": 16416
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 6.635835095137422e-06,
+      "loss": 0.0033,
+      "step": 16418
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 6.6305496828752644e-06,
+      "loss": 0.0222,
+      "step": 16420
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 6.625264270613109e-06,
+      "loss": 0.0202,
+      "step": 16422
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 6.6199788583509515e-06,
+      "loss": 0.0009,
+      "step": 16424
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 6.614693446088795e-06,
+      "loss": 0.0109,
+      "step": 16426
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 6.609408033826639e-06,
+      "loss": 0.0003,
+      "step": 16428
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 6.604122621564482e-06,
+      "loss": 0.0081,
+      "step": 16430
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 6.598837209302326e-06,
+      "loss": 0.0067,
+      "step": 16432
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 6.593551797040169e-06,
+      "loss": 0.0026,
+      "step": 16434
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 6.5882663847780134e-06,
+      "loss": 0.0021,
+      "step": 16436
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 6.582980972515856e-06,
+      "loss": 0.0019,
+      "step": 16438
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 6.5776955602537005e-06,
+      "loss": 0.0143,
+      "step": 16440
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 6.572410147991544e-06,
+      "loss": 0.0131,
+      "step": 16442
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 6.567124735729387e-06,
+      "loss": 0.0088,
+      "step": 16444
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 6.561839323467231e-06,
+      "loss": 0.0022,
+      "step": 16446
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 6.556553911205074e-06,
+      "loss": 0.0029,
+      "step": 16448
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 6.551268498942918e-06,
+      "loss": 0.0072,
+      "step": 16450
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 6.545983086680761e-06,
+      "loss": 0.0004,
+      "step": 16452
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 6.540697674418605e-06,
+      "loss": 0.0018,
+      "step": 16454
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 6.535412262156449e-06,
+      "loss": 0.0021,
+      "step": 16456
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 6.530126849894293e-06,
+      "loss": 0.0125,
+      "step": 16458
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 6.524841437632136e-06,
+      "loss": 0.0024,
+      "step": 16460
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 6.519556025369978e-06,
+      "loss": 0.0167,
+      "step": 16462
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 6.514270613107823e-06,
+      "loss": 0.001,
+      "step": 16464
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 6.508985200845666e-06,
+      "loss": 0.0194,
+      "step": 16466
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 6.503699788583511e-06,
+      "loss": 0.0352,
+      "step": 16468
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 6.498414376321353e-06,
+      "loss": 0.0035,
+      "step": 16470
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 6.493128964059198e-06,
+      "loss": 0.007,
+      "step": 16472
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 6.48784355179704e-06,
+      "loss": 0.0011,
+      "step": 16474
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 6.482558139534885e-06,
+      "loss": 0.0042,
+      "step": 16476
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 6.477272727272727e-06,
+      "loss": 0.004,
+      "step": 16478
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 6.471987315010571e-06,
+      "loss": 0.0087,
+      "step": 16480
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 6.466701902748415e-06,
+      "loss": 0.0063,
+      "step": 16482
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 6.461416490486258e-06,
+      "loss": 0.0038,
+      "step": 16484
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 6.456131078224102e-06,
+      "loss": 0.0028,
+      "step": 16486
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 6.450845665961945e-06,
+      "loss": 0.0137,
+      "step": 16488
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 6.445560253699789e-06,
+      "loss": 0.0012,
+      "step": 16490
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 6.440274841437633e-06,
+      "loss": 0.0024,
+      "step": 16492
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 6.4349894291754755e-06,
+      "loss": 0.028,
+      "step": 16494
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 6.42970401691332e-06,
+      "loss": 0.0153,
+      "step": 16496
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 6.4244186046511625e-06,
+      "loss": 0.0074,
+      "step": 16498
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 6.419133192389007e-06,
+      "loss": 0.0004,
+      "step": 16500
+    },
+    {
+      "epoch": 17.44,
+      "eval_cer": 0.04713593616414933,
+      "eval_loss": 0.8180399537086487,
+      "eval_runtime": 126.9565,
+      "eval_samples_per_second": 6.624,
+      "eval_steps_per_second": 0.835,
+      "step": 16500
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 6.4138477801268495e-06,
+      "loss": 0.0022,
+      "step": 16502
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 6.408562367864694e-06,
+      "loss": 0.011,
+      "step": 16504
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 6.4032769556025374e-06,
+      "loss": 0.0024,
+      "step": 16506
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 6.397991543340382e-06,
+      "loss": 0.0023,
+      "step": 16508
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 6.3927061310782245e-06,
+      "loss": 0.0145,
+      "step": 16510
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 6.387420718816067e-06,
+      "loss": 0.0111,
+      "step": 16512
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 6.3821353065539115e-06,
+      "loss": 0.0011,
+      "step": 16514
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 6.376849894291754e-06,
+      "loss": 0.0246,
+      "step": 16516
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 6.3715644820295985e-06,
+      "loss": 0.0247,
+      "step": 16518
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 6.366279069767442e-06,
+      "loss": 0.0227,
+      "step": 16520
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 6.360993657505286e-06,
+      "loss": 0.0443,
+      "step": 16522
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 6.355708245243129e-06,
+      "loss": 0.0018,
+      "step": 16524
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 6.3504228329809735e-06,
+      "loss": 0.0099,
+      "step": 16526
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 6.345137420718816e-06,
+      "loss": 0.0018,
+      "step": 16528
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 6.33985200845666e-06,
+      "loss": 0.1091,
+      "step": 16530
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 6.334566596194504e-06,
+      "loss": 0.0205,
+      "step": 16532
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 6.329281183932347e-06,
+      "loss": 0.0005,
+      "step": 16534
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 6.323995771670191e-06,
+      "loss": 0.0016,
+      "step": 16536
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 6.318710359408034e-06,
+      "loss": 0.0003,
+      "step": 16538
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 6.313424947145878e-06,
+      "loss": 0.0024,
+      "step": 16540
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 6.308139534883721e-06,
+      "loss": 0.0043,
+      "step": 16542
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 6.302854122621565e-06,
+      "loss": 0.0262,
+      "step": 16544
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 6.297568710359409e-06,
+      "loss": 0.0064,
+      "step": 16546
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 6.292283298097251e-06,
+      "loss": 0.0018,
+      "step": 16548
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 6.286997885835096e-06,
+      "loss": 0.0133,
+      "step": 16550
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 6.281712473572938e-06,
+      "loss": 0.0052,
+      "step": 16552
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 6.276427061310783e-06,
+      "loss": 0.0006,
+      "step": 16554
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 6.271141649048625e-06,
+      "loss": 0.0827,
+      "step": 16556
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 6.26585623678647e-06,
+      "loss": 0.0039,
+      "step": 16558
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 6.260570824524313e-06,
+      "loss": 0.0007,
+      "step": 16560
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 6.255285412262158e-06,
+      "loss": 0.0035,
+      "step": 16562
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0046,
+      "step": 16564
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 6.244714587737844e-06,
+      "loss": 0.008,
+      "step": 16566
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 6.239429175475687e-06,
+      "loss": 0.0105,
+      "step": 16568
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 6.234143763213531e-06,
+      "loss": 0.0166,
+      "step": 16570
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 6.228858350951375e-06,
+      "loss": 0.0054,
+      "step": 16572
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 6.223572938689218e-06,
+      "loss": 0.0008,
+      "step": 16574
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 6.218287526427061e-06,
+      "loss": 0.0012,
+      "step": 16576
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 6.213002114164905e-06,
+      "loss": 0.0022,
+      "step": 16578
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 6.2077167019027484e-06,
+      "loss": 0.0016,
+      "step": 16580
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 6.202431289640592e-06,
+      "loss": 0.0014,
+      "step": 16582
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 6.197145877378436e-06,
+      "loss": 0.0407,
+      "step": 16584
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 6.19186046511628e-06,
+      "loss": 0.0259,
+      "step": 16586
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 6.186575052854123e-06,
+      "loss": 0.0132,
+      "step": 16588
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 6.181289640591967e-06,
+      "loss": 0.0078,
+      "step": 16590
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 6.1760042283298096e-06,
+      "loss": 0.0276,
+      "step": 16592
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 6.170718816067653e-06,
+      "loss": 0.0355,
+      "step": 16594
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 6.165433403805497e-06,
+      "loss": 0.0024,
+      "step": 16596
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 6.160147991543341e-06,
+      "loss": 0.0605,
+      "step": 16598
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 6.1548625792811845e-06,
+      "loss": 0.0178,
+      "step": 16600
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 6.149577167019028e-06,
+      "loss": 0.0032,
+      "step": 16602
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 6.1442917547568715e-06,
+      "loss": 0.009,
+      "step": 16604
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 6.139006342494715e-06,
+      "loss": 0.0005,
+      "step": 16606
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 6.133720930232558e-06,
+      "loss": 0.0131,
+      "step": 16608
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 6.128435517970402e-06,
+      "loss": 0.0034,
+      "step": 16610
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 6.123150105708246e-06,
+      "loss": 0.0038,
+      "step": 16612
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 6.117864693446089e-06,
+      "loss": 0.0088,
+      "step": 16614
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 6.112579281183933e-06,
+      "loss": 0.001,
+      "step": 16616
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 6.107293868921776e-06,
+      "loss": 0.0009,
+      "step": 16618
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 6.10200845665962e-06,
+      "loss": 0.0175,
+      "step": 16620
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 6.096723044397463e-06,
+      "loss": 0.0262,
+      "step": 16622
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 6.0914376321353075e-06,
+      "loss": 0.0154,
+      "step": 16624
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 6.08615221987315e-06,
+      "loss": 0.018,
+      "step": 16626
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 6.080866807610994e-06,
+      "loss": 0.0041,
+      "step": 16628
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 6.075581395348837e-06,
+      "loss": 0.0035,
+      "step": 16630
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 6.070295983086681e-06,
+      "loss": 0.0379,
+      "step": 16632
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 6.065010570824524e-06,
+      "loss": 0.0023,
+      "step": 16634
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 6.059725158562368e-06,
+      "loss": 0.0011,
+      "step": 16636
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 6.054439746300212e-06,
+      "loss": 0.0198,
+      "step": 16638
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 6.049154334038056e-06,
+      "loss": 0.0122,
+      "step": 16640
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 6.043868921775899e-06,
+      "loss": 0.0144,
+      "step": 16642
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 6.038583509513742e-06,
+      "loss": 0.0008,
+      "step": 16644
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 6.033298097251585e-06,
+      "loss": 0.0019,
+      "step": 16646
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 6.028012684989429e-06,
+      "loss": 0.0022,
+      "step": 16648
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 6.022727272727273e-06,
+      "loss": 0.0007,
+      "step": 16650
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 6.017441860465117e-06,
+      "loss": 0.0312,
+      "step": 16652
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 6.01215644820296e-06,
+      "loss": 0.0004,
+      "step": 16654
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 6.006871035940804e-06,
+      "loss": 0.0004,
+      "step": 16656
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 6.001585623678647e-06,
+      "loss": 0.0015,
+      "step": 16658
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 5.996300211416491e-06,
+      "loss": 0.0029,
+      "step": 16660
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 5.991014799154334e-06,
+      "loss": 0.0024,
+      "step": 16662
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 5.985729386892178e-06,
+      "loss": 0.022,
+      "step": 16664
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 5.9804439746300214e-06,
+      "loss": 0.0022,
+      "step": 16666
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 5.975158562367865e-06,
+      "loss": 0.0008,
+      "step": 16668
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 5.9698731501057085e-06,
+      "loss": 0.0008,
+      "step": 16670
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 5.964587737843552e-06,
+      "loss": 0.0016,
+      "step": 16672
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 5.9593023255813955e-06,
+      "loss": 0.0059,
+      "step": 16674
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 5.95401691331924e-06,
+      "loss": 0.0033,
+      "step": 16676
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 5.948731501057083e-06,
+      "loss": 0.0013,
+      "step": 16678
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 5.943446088794926e-06,
+      "loss": 0.0049,
+      "step": 16680
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 5.9381606765327696e-06,
+      "loss": 0.0126,
+      "step": 16682
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 5.932875264270613e-06,
+      "loss": 0.0058,
+      "step": 16684
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 5.927589852008457e-06,
+      "loss": 0.0008,
+      "step": 16686
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 5.9223044397463e-06,
+      "loss": 0.0017,
+      "step": 16688
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 5.9170190274841445e-06,
+      "loss": 0.003,
+      "step": 16690
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 5.911733615221988e-06,
+      "loss": 0.0069,
+      "step": 16692
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 5.9064482029598315e-06,
+      "loss": 0.0078,
+      "step": 16694
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 5.901162790697674e-06,
+      "loss": 0.0326,
+      "step": 16696
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 5.895877378435518e-06,
+      "loss": 0.0114,
+      "step": 16698
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 5.890591966173361e-06,
+      "loss": 0.0008,
+      "step": 16700
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 5.885306553911206e-06,
+      "loss": 0.0004,
+      "step": 16702
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 5.880021141649049e-06,
+      "loss": 0.0994,
+      "step": 16704
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 5.874735729386893e-06,
+      "loss": 0.0003,
+      "step": 16706
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 5.869450317124736e-06,
+      "loss": 0.016,
+      "step": 16708
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 5.86416490486258e-06,
+      "loss": 0.0019,
+      "step": 16710
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 5.858879492600423e-06,
+      "loss": 0.0014,
+      "step": 16712
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 5.853594080338267e-06,
+      "loss": 0.0013,
+      "step": 16714
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 5.84830866807611e-06,
+      "loss": 0.0011,
+      "step": 16716
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 5.843023255813954e-06,
+      "loss": 0.0334,
+      "step": 16718
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 5.837737843551797e-06,
+      "loss": 0.0304,
+      "step": 16720
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 5.832452431289641e-06,
+      "loss": 0.0167,
+      "step": 16722
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 5.827167019027484e-06,
+      "loss": 0.0008,
+      "step": 16724
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 5.821881606765328e-06,
+      "loss": 0.0021,
+      "step": 16726
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 5.816596194503171e-06,
+      "loss": 0.0056,
+      "step": 16728
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 5.811310782241016e-06,
+      "loss": 0.0006,
+      "step": 16730
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 5.806025369978858e-06,
+      "loss": 0.0135,
+      "step": 16732
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 5.800739957716702e-06,
+      "loss": 0.0047,
+      "step": 16734
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 5.795454545454545e-06,
+      "loss": 0.0223,
+      "step": 16736
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 5.790169133192389e-06,
+      "loss": 0.0008,
+      "step": 16738
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 5.7848837209302324e-06,
+      "loss": 0.0048,
+      "step": 16740
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 5.779598308668077e-06,
+      "loss": 0.0023,
+      "step": 16742
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 5.77431289640592e-06,
+      "loss": 0.0018,
+      "step": 16744
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 5.769027484143764e-06,
+      "loss": 0.0167,
+      "step": 16746
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 5.763742071881607e-06,
+      "loss": 0.0047,
+      "step": 16748
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 5.75845665961945e-06,
+      "loss": 0.0176,
+      "step": 16750
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 5.7531712473572936e-06,
+      "loss": 0.0423,
+      "step": 16752
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 5.747885835095138e-06,
+      "loss": 0.0035,
+      "step": 16754
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 5.7426004228329814e-06,
+      "loss": 0.0093,
+      "step": 16756
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 5.737315010570825e-06,
+      "loss": 0.0033,
+      "step": 16758
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 5.7320295983086685e-06,
+      "loss": 0.0033,
+      "step": 16760
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 5.726744186046512e-06,
+      "loss": 0.0085,
+      "step": 16762
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 5.7214587737843555e-06,
+      "loss": 0.0017,
+      "step": 16764
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 5.716173361522199e-06,
+      "loss": 0.0031,
+      "step": 16766
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 5.7108879492600425e-06,
+      "loss": 0.0007,
+      "step": 16768
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 5.705602536997886e-06,
+      "loss": 0.0165,
+      "step": 16770
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 5.70031712473573e-06,
+      "loss": 0.0498,
+      "step": 16772
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 5.695031712473573e-06,
+      "loss": 0.0039,
+      "step": 16774
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 5.689746300211417e-06,
+      "loss": 0.0053,
+      "step": 16776
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 5.68446088794926e-06,
+      "loss": 0.0161,
+      "step": 16778
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 5.679175475687104e-06,
+      "loss": 0.0134,
+      "step": 16780
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 5.673890063424948e-06,
+      "loss": 0.0224,
+      "step": 16782
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 5.668604651162791e-06,
+      "loss": 0.0039,
+      "step": 16784
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 5.663319238900634e-06,
+      "loss": 0.0132,
+      "step": 16786
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 5.658033826638478e-06,
+      "loss": 0.0004,
+      "step": 16788
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 5.652748414376321e-06,
+      "loss": 0.0033,
+      "step": 16790
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 5.647463002114165e-06,
+      "loss": 0.0037,
+      "step": 16792
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 5.642177589852009e-06,
+      "loss": 0.0128,
+      "step": 16794
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 5.636892177589853e-06,
+      "loss": 0.009,
+      "step": 16796
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 5.631606765327696e-06,
+      "loss": 0.005,
+      "step": 16798
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 5.62632135306554e-06,
+      "loss": 0.0005,
+      "step": 16800
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 5.621035940803382e-06,
+      "loss": 0.0268,
+      "step": 16802
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 5.615750528541226e-06,
+      "loss": 0.0753,
+      "step": 16804
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 5.61046511627907e-06,
+      "loss": 0.0016,
+      "step": 16806
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 5.605179704016914e-06,
+      "loss": 0.009,
+      "step": 16808
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 5.599894291754757e-06,
+      "loss": 0.0004,
+      "step": 16810
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 5.594608879492601e-06,
+      "loss": 0.0033,
+      "step": 16812
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 5.589323467230444e-06,
+      "loss": 0.0091,
+      "step": 16814
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 5.584038054968288e-06,
+      "loss": 0.0008,
+      "step": 16816
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 5.578752642706131e-06,
+      "loss": 0.0018,
+      "step": 16818
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 5.573467230443975e-06,
+      "loss": 0.0183,
+      "step": 16820
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 5.568181818181818e-06,
+      "loss": 0.0005,
+      "step": 16822
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 5.562896405919662e-06,
+      "loss": 0.0205,
+      "step": 16824
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 5.5576109936575054e-06,
+      "loss": 0.0003,
+      "step": 16826
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 5.552325581395349e-06,
+      "loss": 0.0037,
+      "step": 16828
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 5.5470401691331925e-06,
+      "loss": 0.0097,
+      "step": 16830
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 5.541754756871036e-06,
+      "loss": 0.0008,
+      "step": 16832
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 5.53646934460888e-06,
+      "loss": 0.0003,
+      "step": 16834
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 5.531183932346724e-06,
+      "loss": 0.009,
+      "step": 16836
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 5.5258985200845665e-06,
+      "loss": 0.0004,
+      "step": 16838
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 5.52061310782241e-06,
+      "loss": 0.0005,
+      "step": 16840
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 5.5153276955602536e-06,
+      "loss": 0.0223,
+      "step": 16842
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 5.510042283298097e-06,
+      "loss": 0.0009,
+      "step": 16844
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 5.5047568710359415e-06,
+      "loss": 0.003,
+      "step": 16846
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 5.499471458773785e-06,
+      "loss": 0.0013,
+      "step": 16848
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 5.4941860465116285e-06,
+      "loss": 0.0194,
+      "step": 16850
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 5.488900634249472e-06,
+      "loss": 0.0211,
+      "step": 16852
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 5.4836152219873155e-06,
+      "loss": 0.0007,
+      "step": 16854
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 5.478329809725158e-06,
+      "loss": 0.0041,
+      "step": 16856
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 5.4730443974630026e-06,
+      "loss": 0.0142,
+      "step": 16858
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 5.467758985200846e-06,
+      "loss": 0.0151,
+      "step": 16860
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 5.46247357293869e-06,
+      "loss": 0.0087,
+      "step": 16862
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 5.457188160676533e-06,
+      "loss": 0.0009,
+      "step": 16864
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 5.451902748414377e-06,
+      "loss": 0.0349,
+      "step": 16866
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 5.44661733615222e-06,
+      "loss": 0.0041,
+      "step": 16868
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 5.441331923890064e-06,
+      "loss": 0.0016,
+      "step": 16870
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 5.436046511627907e-06,
+      "loss": 0.0013,
+      "step": 16872
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 5.430761099365751e-06,
+      "loss": 0.0016,
+      "step": 16874
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 5.425475687103594e-06,
+      "loss": 0.0088,
+      "step": 16876
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 5.420190274841438e-06,
+      "loss": 0.0036,
+      "step": 16878
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 5.414904862579281e-06,
+      "loss": 0.003,
+      "step": 16880
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 5.409619450317125e-06,
+      "loss": 0.0217,
+      "step": 16882
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 5.404334038054968e-06,
+      "loss": 0.0257,
+      "step": 16884
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 5.399048625792813e-06,
+      "loss": 0.0067,
+      "step": 16886
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 5.393763213530656e-06,
+      "loss": 0.0016,
+      "step": 16888
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 5.388477801268499e-06,
+      "loss": 0.0016,
+      "step": 16890
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 5.383192389006342e-06,
+      "loss": 0.001,
+      "step": 16892
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 5.377906976744186e-06,
+      "loss": 0.0179,
+      "step": 16894
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 5.372621564482029e-06,
+      "loss": 0.0012,
+      "step": 16896
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 5.367336152219874e-06,
+      "loss": 0.0012,
+      "step": 16898
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 5.362050739957717e-06,
+      "loss": 0.0142,
+      "step": 16900
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 5.356765327695561e-06,
+      "loss": 0.0005,
+      "step": 16902
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 5.351479915433404e-06,
+      "loss": 0.0021,
+      "step": 16904
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 5.346194503171248e-06,
+      "loss": 0.0076,
+      "step": 16906
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 5.3409090909090905e-06,
+      "loss": 0.0066,
+      "step": 16908
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 5.335623678646935e-06,
+      "loss": 0.0152,
+      "step": 16910
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 5.330338266384778e-06,
+      "loss": 0.0005,
+      "step": 16912
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 5.325052854122622e-06,
+      "loss": 0.017,
+      "step": 16914
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 5.3197674418604654e-06,
+      "loss": 0.0011,
+      "step": 16916
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 5.314482029598309e-06,
+      "loss": 0.0013,
+      "step": 16918
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 5.3091966173361525e-06,
+      "loss": 0.0104,
+      "step": 16920
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 5.303911205073996e-06,
+      "loss": 0.0287,
+      "step": 16922
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 5.2986257928118395e-06,
+      "loss": 0.0144,
+      "step": 16924
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 5.293340380549683e-06,
+      "loss": 0.0144,
+      "step": 16926
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 5.2880549682875265e-06,
+      "loss": 0.0034,
+      "step": 16928
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 5.28276955602537e-06,
+      "loss": 0.0273,
+      "step": 16930
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 5.277484143763214e-06,
+      "loss": 0.0043,
+      "step": 16932
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 5.272198731501057e-06,
+      "loss": 0.0009,
+      "step": 16934
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 5.266913319238901e-06,
+      "loss": 0.0027,
+      "step": 16936
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 5.261627906976745e-06,
+      "loss": 0.0109,
+      "step": 16938
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 5.2563424947145885e-06,
+      "loss": 0.0119,
+      "step": 16940
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 5.251057082452432e-06,
+      "loss": 0.0007,
+      "step": 16942
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 5.245771670190275e-06,
+      "loss": 0.0151,
+      "step": 16944
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 5.240486257928118e-06,
+      "loss": 0.0134,
+      "step": 16946
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 5.235200845665962e-06,
+      "loss": 0.0042,
+      "step": 16948
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 5.229915433403806e-06,
+      "loss": 0.0037,
+      "step": 16950
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 5.22463002114165e-06,
+      "loss": 0.002,
+      "step": 16952
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 5.219344608879493e-06,
+      "loss": 0.0198,
+      "step": 16954
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 5.214059196617337e-06,
+      "loss": 0.0402,
+      "step": 16956
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 5.20877378435518e-06,
+      "loss": 0.0003,
+      "step": 16958
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 5.203488372093023e-06,
+      "loss": 0.0013,
+      "step": 16960
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 5.198202959830867e-06,
+      "loss": 0.0019,
+      "step": 16962
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 5.192917547568711e-06,
+      "loss": 0.0023,
+      "step": 16964
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 5.187632135306554e-06,
+      "loss": 0.0011,
+      "step": 16966
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 5.182346723044398e-06,
+      "loss": 0.0008,
+      "step": 16968
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 5.177061310782241e-06,
+      "loss": 0.0268,
+      "step": 16970
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 5.171775898520085e-06,
+      "loss": 0.0004,
+      "step": 16972
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 5.166490486257928e-06,
+      "loss": 0.0063,
+      "step": 16974
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 5.161205073995772e-06,
+      "loss": 0.0004,
+      "step": 16976
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 5.155919661733615e-06,
+      "loss": 0.0024,
+      "step": 16978
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 5.150634249471459e-06,
+      "loss": 0.0006,
+      "step": 16980
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 5.145348837209302e-06,
+      "loss": 0.0141,
+      "step": 16982
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 5.140063424947146e-06,
+      "loss": 0.0014,
+      "step": 16984
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 5.134778012684989e-06,
+      "loss": 0.0016,
+      "step": 16986
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 5.129492600422833e-06,
+      "loss": 0.0181,
+      "step": 16988
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 5.124207188160677e-06,
+      "loss": 0.0015,
+      "step": 16990
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 5.118921775898521e-06,
+      "loss": 0.0016,
+      "step": 16992
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 5.113636363636364e-06,
+      "loss": 0.0005,
+      "step": 16994
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 5.108350951374207e-06,
+      "loss": 0.0269,
+      "step": 16996
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 5.1030655391120505e-06,
+      "loss": 0.0113,
+      "step": 16998
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 5.097780126849894e-06,
+      "loss": 0.0008,
+      "step": 17000
+    },
+    {
+      "epoch": 17.97,
+      "eval_cer": 0.057395269307495014,
+      "eval_loss": 0.671505331993103,
+      "eval_runtime": 126.1941,
+      "eval_samples_per_second": 6.664,
+      "eval_steps_per_second": 0.84,
+      "step": 17000
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 5.092494714587738e-06,
+      "loss": 0.0013,
+      "step": 17002
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 5.087209302325582e-06,
+      "loss": 0.0057,
+      "step": 17004
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 5.0819238900634255e-06,
+      "loss": 0.0048,
+      "step": 17006
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 5.076638477801269e-06,
+      "loss": 0.0005,
+      "step": 17008
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 5.0713530655391125e-06,
+      "loss": 0.0004,
+      "step": 17010
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 5.066067653276956e-06,
+      "loss": 0.0017,
+      "step": 17012
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 5.0607822410147995e-06,
+      "loss": 0.0067,
+      "step": 17014
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 5.055496828752643e-06,
+      "loss": 0.0015,
+      "step": 17016
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 5.0502114164904866e-06,
+      "loss": 0.0008,
+      "step": 17018
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 5.04492600422833e-06,
+      "loss": 0.0005,
+      "step": 17020
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 5.039640591966174e-06,
+      "loss": 0.0115,
+      "step": 17022
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 5.034355179704017e-06,
+      "loss": 0.0005,
+      "step": 17024
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 5.029069767441861e-06,
+      "loss": 0.0053,
+      "step": 17026
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 5.023784355179704e-06,
+      "loss": 0.011,
+      "step": 17028
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 5.018498942917548e-06,
+      "loss": 0.0015,
+      "step": 17030
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 5.013213530655391e-06,
+      "loss": 0.0004,
+      "step": 17032
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 5.007928118393235e-06,
+      "loss": 0.0022,
+      "step": 17034
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 5.002642706131078e-06,
+      "loss": 0.0011,
+      "step": 17036
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 4.997357293868922e-06,
+      "loss": 0.001,
+      "step": 17038
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 4.992071881606765e-06,
+      "loss": 0.025,
+      "step": 17040
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 4.98678646934461e-06,
+      "loss": 0.011,
+      "step": 17042
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 4.981501057082453e-06,
+      "loss": 0.0004,
+      "step": 17044
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 4.976215644820297e-06,
+      "loss": 0.0018,
+      "step": 17046
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 4.970930232558139e-06,
+      "loss": 0.0041,
+      "step": 17048
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 4.965644820295983e-06,
+      "loss": 0.0008,
+      "step": 17050
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 4.960359408033826e-06,
+      "loss": 0.0003,
+      "step": 17052
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 4.955073995771671e-06,
+      "loss": 0.0204,
+      "step": 17054
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 4.949788583509514e-06,
+      "loss": 0.0025,
+      "step": 17056
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 4.944503171247358e-06,
+      "loss": 0.025,
+      "step": 17058
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 4.939217758985201e-06,
+      "loss": 0.0006,
+      "step": 17060
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 4.933932346723045e-06,
+      "loss": 0.0175,
+      "step": 17062
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 4.928646934460888e-06,
+      "loss": 0.0113,
+      "step": 17064
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 4.923361522198732e-06,
+      "loss": 0.0055,
+      "step": 17066
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 4.918076109936575e-06,
+      "loss": 0.0003,
+      "step": 17068
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 4.912790697674419e-06,
+      "loss": 0.0046,
+      "step": 17070
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 4.907505285412262e-06,
+      "loss": 0.0007,
+      "step": 17072
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 4.902219873150106e-06,
+      "loss": 0.0011,
+      "step": 17074
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 4.8969344608879494e-06,
+      "loss": 0.0246,
+      "step": 17076
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 4.891649048625793e-06,
+      "loss": 0.0042,
+      "step": 17078
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 4.8863636363636365e-06,
+      "loss": 0.0019,
+      "step": 17080
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 4.881078224101481e-06,
+      "loss": 0.0004,
+      "step": 17082
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 4.8757928118393235e-06,
+      "loss": 0.028,
+      "step": 17084
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 4.870507399577167e-06,
+      "loss": 0.0627,
+      "step": 17086
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 4.8652219873150105e-06,
+      "loss": 0.0159,
+      "step": 17088
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 4.859936575052854e-06,
+      "loss": 0.0006,
+      "step": 17090
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 4.854651162790698e-06,
+      "loss": 0.0037,
+      "step": 17092
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 4.849365750528542e-06,
+      "loss": 0.0007,
+      "step": 17094
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 4.8440803382663855e-06,
+      "loss": 0.0038,
+      "step": 17096
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 4.838794926004229e-06,
+      "loss": 0.0006,
+      "step": 17098
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 4.8335095137420725e-06,
+      "loss": 0.0147,
+      "step": 17100
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 4.828224101479915e-06,
+      "loss": 0.0053,
+      "step": 17102
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 4.822938689217759e-06,
+      "loss": 0.0038,
+      "step": 17104
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 4.817653276955603e-06,
+      "loss": 0.0136,
+      "step": 17106
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 4.812367864693447e-06,
+      "loss": 0.0513,
+      "step": 17108
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 4.80708245243129e-06,
+      "loss": 0.0018,
+      "step": 17110
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 4.801797040169134e-06,
+      "loss": 0.0075,
+      "step": 17112
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 4.796511627906977e-06,
+      "loss": 0.0059,
+      "step": 17114
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 4.791226215644821e-06,
+      "loss": 0.0047,
+      "step": 17116
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 4.785940803382663e-06,
+      "loss": 0.0012,
+      "step": 17118
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 4.780655391120508e-06,
+      "loss": 0.0162,
+      "step": 17120
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 4.775369978858351e-06,
+      "loss": 0.0011,
+      "step": 17122
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 4.770084566596195e-06,
+      "loss": 0.0051,
+      "step": 17124
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 4.764799154334038e-06,
+      "loss": 0.0012,
+      "step": 17126
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 4.759513742071882e-06,
+      "loss": 0.0008,
+      "step": 17128
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 4.754228329809725e-06,
+      "loss": 0.0004,
+      "step": 17130
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 4.748942917547569e-06,
+      "loss": 0.0211,
+      "step": 17132
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 4.743657505285413e-06,
+      "loss": 0.0265,
+      "step": 17134
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 4.738372093023256e-06,
+      "loss": 0.0032,
+      "step": 17136
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 4.733086680761099e-06,
+      "loss": 0.0003,
+      "step": 17138
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 4.727801268498943e-06,
+      "loss": 0.01,
+      "step": 17140
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 4.722515856236786e-06,
+      "loss": 0.0005,
+      "step": 17142
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 4.71723044397463e-06,
+      "loss": 0.0007,
+      "step": 17144
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 4.711945031712474e-06,
+      "loss": 0.0078,
+      "step": 17146
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 4.706659619450318e-06,
+      "loss": 0.0009,
+      "step": 17148
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 4.701374207188161e-06,
+      "loss": 0.0004,
+      "step": 17150
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 4.696088794926005e-06,
+      "loss": 0.0005,
+      "step": 17152
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 4.6908033826638475e-06,
+      "loss": 0.0021,
+      "step": 17154
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 4.685517970401691e-06,
+      "loss": 0.002,
+      "step": 17156
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 4.6802325581395345e-06,
+      "loss": 0.0031,
+      "step": 17158
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 4.674947145877379e-06,
+      "loss": 0.0044,
+      "step": 17160
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 4.669661733615222e-06,
+      "loss": 0.0011,
+      "step": 17162
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 4.664376321353066e-06,
+      "loss": 0.004,
+      "step": 17164
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 4.6590909090909095e-06,
+      "loss": 0.0062,
+      "step": 17166
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 4.653805496828753e-06,
+      "loss": 0.0312,
+      "step": 17168
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 4.6485200845665965e-06,
+      "loss": 0.0007,
+      "step": 17170
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 4.64323467230444e-06,
+      "loss": 0.0013,
+      "step": 17172
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 4.6379492600422835e-06,
+      "loss": 0.0014,
+      "step": 17174
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 4.632663847780127e-06,
+      "loss": 0.0465,
+      "step": 17176
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 4.6273784355179706e-06,
+      "loss": 0.0061,
+      "step": 17178
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 4.622093023255814e-06,
+      "loss": 0.0014,
+      "step": 17180
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 4.616807610993658e-06,
+      "loss": 0.0049,
+      "step": 17182
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 4.611522198731501e-06,
+      "loss": 0.0227,
+      "step": 17184
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 4.6062367864693455e-06,
+      "loss": 0.0047,
+      "step": 17186
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 4.600951374207189e-06,
+      "loss": 0.0009,
+      "step": 17188
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 4.595665961945032e-06,
+      "loss": 0.0007,
+      "step": 17190
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 4.590380549682875e-06,
+      "loss": 0.0038,
+      "step": 17192
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 4.585095137420719e-06,
+      "loss": 0.0075,
+      "step": 17194
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 4.579809725158562e-06,
+      "loss": 0.001,
+      "step": 17196
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 4.574524312896407e-06,
+      "loss": 0.0007,
+      "step": 17198
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 4.56923890063425e-06,
+      "loss": 0.0105,
+      "step": 17200
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 4.563953488372094e-06,
+      "loss": 0.0005,
+      "step": 17202
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 4.558668076109937e-06,
+      "loss": 0.0021,
+      "step": 17204
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 4.55338266384778e-06,
+      "loss": 0.001,
+      "step": 17206
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 4.548097251585623e-06,
+      "loss": 0.0008,
+      "step": 17208
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 4.542811839323467e-06,
+      "loss": 0.0007,
+      "step": 17210
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 4.537526427061311e-06,
+      "loss": 0.0013,
+      "step": 17212
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 4.532241014799155e-06,
+      "loss": 0.0004,
+      "step": 17214
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 4.526955602536998e-06,
+      "loss": 0.004,
+      "step": 17216
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 4.521670190274842e-06,
+      "loss": 0.0193,
+      "step": 17218
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 4.516384778012685e-06,
+      "loss": 0.002,
+      "step": 17220
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 4.511099365750529e-06,
+      "loss": 0.0038,
+      "step": 17222
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 4.505813953488372e-06,
+      "loss": 0.0089,
+      "step": 17224
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 4.500528541226216e-06,
+      "loss": 0.0007,
+      "step": 17226
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 4.495243128964059e-06,
+      "loss": 0.0006,
+      "step": 17228
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 4.489957716701903e-06,
+      "loss": 0.0071,
+      "step": 17230
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 4.484672304439746e-06,
+      "loss": 0.0081,
+      "step": 17232
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 4.47938689217759e-06,
+      "loss": 0.0114,
+      "step": 17234
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 4.4741014799154334e-06,
+      "loss": 0.0005,
+      "step": 17236
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 4.468816067653278e-06,
+      "loss": 0.0018,
+      "step": 17238
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 4.463530655391121e-06,
+      "loss": 0.0116,
+      "step": 17240
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 4.458245243128964e-06,
+      "loss": 0.001,
+      "step": 17242
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 4.4529598308668075e-06,
+      "loss": 0.0005,
+      "step": 17244
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 4.447674418604651e-06,
+      "loss": 0.0117,
+      "step": 17246
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 4.4423890063424945e-06,
+      "loss": 0.0073,
+      "step": 17248
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 4.437103594080338e-06,
+      "loss": 0.0003,
+      "step": 17250
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 4.4318181818181824e-06,
+      "loss": 0.0015,
+      "step": 17252
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 4.426532769556026e-06,
+      "loss": 0.0032,
+      "step": 17254
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 4.4212473572938695e-06,
+      "loss": 0.0025,
+      "step": 17256
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 4.415961945031713e-06,
+      "loss": 0.0003,
+      "step": 17258
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 4.410676532769556e-06,
+      "loss": 0.0003,
+      "step": 17260
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 4.405391120507399e-06,
+      "loss": 0.0066,
+      "step": 17262
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 4.4001057082452435e-06,
+      "loss": 0.0071,
+      "step": 17264
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 4.394820295983087e-06,
+      "loss": 0.0018,
+      "step": 17266
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 4.389534883720931e-06,
+      "loss": 0.0005,
+      "step": 17268
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 4.384249471458774e-06,
+      "loss": 0.0029,
+      "step": 17270
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 4.378964059196618e-06,
+      "loss": 0.0448,
+      "step": 17272
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 4.373678646934461e-06,
+      "loss": 0.0132,
+      "step": 17274
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 4.368393234672305e-06,
+      "loss": 0.0203,
+      "step": 17276
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 4.363107822410148e-06,
+      "loss": 0.0503,
+      "step": 17278
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 4.357822410147992e-06,
+      "loss": 0.0411,
+      "step": 17280
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 4.352536997885835e-06,
+      "loss": 0.0203,
+      "step": 17282
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 4.347251585623679e-06,
+      "loss": 0.0053,
+      "step": 17284
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 4.341966173361522e-06,
+      "loss": 0.0148,
+      "step": 17286
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 4.336680761099366e-06,
+      "loss": 0.016,
+      "step": 17288
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 4.33139534883721e-06,
+      "loss": 0.0002,
+      "step": 17290
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 4.326109936575054e-06,
+      "loss": 0.0004,
+      "step": 17292
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 4.320824524312896e-06,
+      "loss": 0.0252,
+      "step": 17294
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 4.31553911205074e-06,
+      "loss": 0.0075,
+      "step": 17296
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 4.310253699788583e-06,
+      "loss": 0.0197,
+      "step": 17298
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 4.304968287526427e-06,
+      "loss": 0.0009,
+      "step": 17300
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 4.29968287526427e-06,
+      "loss": 0.0005,
+      "step": 17302
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 4.294397463002115e-06,
+      "loss": 0.001,
+      "step": 17304
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 4.289112050739958e-06,
+      "loss": 0.0006,
+      "step": 17306
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 4.283826638477802e-06,
+      "loss": 0.0085,
+      "step": 17308
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 4.278541226215645e-06,
+      "loss": 0.0671,
+      "step": 17310
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 4.273255813953488e-06,
+      "loss": 0.0034,
+      "step": 17312
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 4.2679704016913315e-06,
+      "loss": 0.0012,
+      "step": 17314
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 4.262684989429176e-06,
+      "loss": 0.0012,
+      "step": 17316
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 4.257399577167019e-06,
+      "loss": 0.0005,
+      "step": 17318
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 4.252114164904863e-06,
+      "loss": 0.0085,
+      "step": 17320
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 4.246828752642706e-06,
+      "loss": 0.0024,
+      "step": 17322
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 4.24154334038055e-06,
+      "loss": 0.0004,
+      "step": 17324
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 4.2362579281183935e-06,
+      "loss": 0.0435,
+      "step": 17326
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 4.230972515856237e-06,
+      "loss": 0.0032,
+      "step": 17328
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 4.2256871035940805e-06,
+      "loss": 0.0005,
+      "step": 17330
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 4.220401691331924e-06,
+      "loss": 0.0004,
+      "step": 17332
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 4.2151162790697675e-06,
+      "loss": 0.0007,
+      "step": 17334
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 4.209830866807611e-06,
+      "loss": 0.0007,
+      "step": 17336
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 4.2045454545454546e-06,
+      "loss": 0.0017,
+      "step": 17338
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 4.199260042283298e-06,
+      "loss": 0.0008,
+      "step": 17340
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 4.193974630021142e-06,
+      "loss": 0.0004,
+      "step": 17342
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 4.188689217758986e-06,
+      "loss": 0.0133,
+      "step": 17344
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 4.1834038054968295e-06,
+      "loss": 0.0081,
+      "step": 17346
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 4.178118393234672e-06,
+      "loss": 0.0019,
+      "step": 17348
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 4.172832980972516e-06,
+      "loss": 0.0026,
+      "step": 17350
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 4.167547568710359e-06,
+      "loss": 0.0139,
+      "step": 17352
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 4.162262156448203e-06,
+      "loss": 0.0442,
+      "step": 17354
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 4.156976744186047e-06,
+      "loss": 0.0009,
+      "step": 17356
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 4.151691331923891e-06,
+      "loss": 0.0009,
+      "step": 17358
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 4.146405919661734e-06,
+      "loss": 0.0005,
+      "step": 17360
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 4.141120507399578e-06,
+      "loss": 0.0018,
+      "step": 17362
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 4.135835095137421e-06,
+      "loss": 0.0144,
+      "step": 17364
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 4.130549682875264e-06,
+      "loss": 0.0023,
+      "step": 17366
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 4.125264270613108e-06,
+      "loss": 0.0035,
+      "step": 17368
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 4.119978858350952e-06,
+      "loss": 0.0004,
+      "step": 17370
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 4.114693446088795e-06,
+      "loss": 0.0549,
+      "step": 17372
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 4.109408033826639e-06,
+      "loss": 0.0007,
+      "step": 17374
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 4.104122621564482e-06,
+      "loss": 0.006,
+      "step": 17376
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 4.098837209302326e-06,
+      "loss": 0.0004,
+      "step": 17378
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 4.093551797040169e-06,
+      "loss": 0.003,
+      "step": 17380
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 4.088266384778013e-06,
+      "loss": 0.0023,
+      "step": 17382
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 4.082980972515856e-06,
+      "loss": 0.0006,
+      "step": 17384
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 4.0776955602537e-06,
+      "loss": 0.0006,
+      "step": 17386
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 4.072410147991543e-06,
+      "loss": 0.0074,
+      "step": 17388
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 4.067124735729387e-06,
+      "loss": 0.0004,
+      "step": 17390
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 4.06183932346723e-06,
+      "loss": 0.0004,
+      "step": 17392
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 4.056553911205074e-06,
+      "loss": 0.0005,
+      "step": 17394
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 4.051268498942918e-06,
+      "loss": 0.0007,
+      "step": 17396
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 4.045983086680762e-06,
+      "loss": 0.0033,
+      "step": 17398
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 4.0406976744186045e-06,
+      "loss": 0.0014,
+      "step": 17400
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 4.035412262156448e-06,
+      "loss": 0.0006,
+      "step": 17402
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 4.0301268498942915e-06,
+      "loss": 0.0051,
+      "step": 17404
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 4.024841437632135e-06,
+      "loss": 0.0025,
+      "step": 17406
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 4.019556025369979e-06,
+      "loss": 0.0002,
+      "step": 17408
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 4.014270613107823e-06,
+      "loss": 0.0013,
+      "step": 17410
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 4.0089852008456664e-06,
+      "loss": 0.0169,
+      "step": 17412
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 4.00369978858351e-06,
+      "loss": 0.0014,
+      "step": 17414
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 3.9984143763213535e-06,
+      "loss": 0.0007,
+      "step": 17416
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 3.993128964059196e-06,
+      "loss": 0.0175,
+      "step": 17418
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 3.9878435517970405e-06,
+      "loss": 0.0028,
+      "step": 17420
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 3.982558139534884e-06,
+      "loss": 0.0155,
+      "step": 17422
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 3.9772727272727275e-06,
+      "loss": 0.0009,
+      "step": 17424
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 3.971987315010571e-06,
+      "loss": 0.0016,
+      "step": 17426
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 3.966701902748415e-06,
+      "loss": 0.0008,
+      "step": 17428
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 3.961416490486258e-06,
+      "loss": 0.0006,
+      "step": 17430
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 3.956131078224102e-06,
+      "loss": 0.0017,
+      "step": 17432
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 3.950845665961945e-06,
+      "loss": 0.0009,
+      "step": 17434
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 3.945560253699789e-06,
+      "loss": 0.0063,
+      "step": 17436
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 3.940274841437632e-06,
+      "loss": 0.0134,
+      "step": 17438
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 3.934989429175476e-06,
+      "loss": 0.0009,
+      "step": 17440
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 3.929704016913319e-06,
+      "loss": 0.0045,
+      "step": 17442
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 3.924418604651163e-06,
+      "loss": 0.0083,
+      "step": 17444
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 3.919133192389006e-06,
+      "loss": 0.0052,
+      "step": 17446
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 3.913847780126851e-06,
+      "loss": 0.0009,
+      "step": 17448
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 3.908562367864694e-06,
+      "loss": 0.0041,
+      "step": 17450
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 3.903276955602537e-06,
+      "loss": 0.0017,
+      "step": 17452
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 3.89799154334038e-06,
+      "loss": 0.0007,
+      "step": 17454
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 3.892706131078224e-06,
+      "loss": 0.0005,
+      "step": 17456
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 3.887420718816067e-06,
+      "loss": 0.0188,
+      "step": 17458
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 3.882135306553912e-06,
+      "loss": 0.0023,
+      "step": 17460
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 3.876849894291755e-06,
+      "loss": 0.0008,
+      "step": 17462
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 3.871564482029599e-06,
+      "loss": 0.0056,
+      "step": 17464
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 3.866279069767442e-06,
+      "loss": 0.0029,
+      "step": 17466
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 3.860993657505286e-06,
+      "loss": 0.0027,
+      "step": 17468
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 3.8557082452431285e-06,
+      "loss": 0.0302,
+      "step": 17470
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 3.850422832980973e-06,
+      "loss": 0.0005,
+      "step": 17472
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 3.845137420718816e-06,
+      "loss": 0.0003,
+      "step": 17474
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 3.83985200845666e-06,
+      "loss": 0.0004,
+      "step": 17476
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 3.834566596194503e-06,
+      "loss": 0.0151,
+      "step": 17478
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 3.829281183932347e-06,
+      "loss": 0.0004,
+      "step": 17480
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 3.82399577167019e-06,
+      "loss": 0.0031,
+      "step": 17482
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 3.818710359408034e-06,
+      "loss": 0.0012,
+      "step": 17484
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 3.813424947145878e-06,
+      "loss": 0.0023,
+      "step": 17486
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 3.808139534883721e-06,
+      "loss": 0.0004,
+      "step": 17488
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 3.8028541226215645e-06,
+      "loss": 0.0143,
+      "step": 17490
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 3.797568710359408e-06,
+      "loss": 0.0003,
+      "step": 17492
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 3.7922832980972515e-06,
+      "loss": 0.0016,
+      "step": 17494
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 3.7869978858350955e-06,
+      "loss": 0.0011,
+      "step": 17496
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 3.781712473572939e-06,
+      "loss": 0.074,
+      "step": 17498
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 3.7764270613107825e-06,
+      "loss": 0.0004,
+      "step": 17500
+    },
+    {
+      "epoch": 18.5,
+      "eval_cer": 0.06377885437446566,
+      "eval_loss": 0.9183065891265869,
+      "eval_runtime": 124.5586,
+      "eval_samples_per_second": 6.752,
+      "eval_steps_per_second": 0.851,
+      "step": 17500
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 3.771141649048626e-06,
+      "loss": 0.0004,
+      "step": 17502
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 3.76585623678647e-06,
+      "loss": 0.001,
+      "step": 17504
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 3.7605708245243126e-06,
+      "loss": 0.0004,
+      "step": 17506
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 3.7552854122621566e-06,
+      "loss": 0.0011,
+      "step": 17508
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 3.75e-06,
+      "loss": 0.0004,
+      "step": 17510
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 3.7447145877378436e-06,
+      "loss": 0.0005,
+      "step": 17512
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 3.739429175475687e-06,
+      "loss": 0.0184,
+      "step": 17514
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 3.734143763213531e-06,
+      "loss": 0.0004,
+      "step": 17516
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 3.7288583509513746e-06,
+      "loss": 0.0003,
+      "step": 17518
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 3.723572938689218e-06,
+      "loss": 0.0005,
+      "step": 17520
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 3.7182875264270616e-06,
+      "loss": 0.0016,
+      "step": 17522
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 3.7130021141649047e-06,
+      "loss": 0.0025,
+      "step": 17524
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 3.7077167019027482e-06,
+      "loss": 0.0025,
+      "step": 17526
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 3.702431289640592e-06,
+      "loss": 0.0006,
+      "step": 17528
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 3.6971458773784357e-06,
+      "loss": 0.0008,
+      "step": 17530
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 3.6918604651162792e-06,
+      "loss": 0.0327,
+      "step": 17532
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 3.6865750528541227e-06,
+      "loss": 0.0004,
+      "step": 17534
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 3.6812896405919667e-06,
+      "loss": 0.0006,
+      "step": 17536
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 3.67600422832981e-06,
+      "loss": 0.0035,
+      "step": 17538
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 3.6707188160676533e-06,
+      "loss": 0.001,
+      "step": 17540
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 3.665433403805497e-06,
+      "loss": 0.0108,
+      "step": 17542
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 3.6601479915433403e-06,
+      "loss": 0.0286,
+      "step": 17544
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 3.654862579281184e-06,
+      "loss": 0.0024,
+      "step": 17546
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 3.649577167019028e-06,
+      "loss": 0.0091,
+      "step": 17548
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 3.6442917547568713e-06,
+      "loss": 0.0008,
+      "step": 17550
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 3.639006342494715e-06,
+      "loss": 0.0015,
+      "step": 17552
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 3.6337209302325583e-06,
+      "loss": 0.0009,
+      "step": 17554
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 3.6284355179704023e-06,
+      "loss": 0.001,
+      "step": 17556
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 3.623150105708245e-06,
+      "loss": 0.0136,
+      "step": 17558
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 3.617864693446089e-06,
+      "loss": 0.0007,
+      "step": 17560
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 3.6125792811839324e-06,
+      "loss": 0.0004,
+      "step": 17562
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 3.607293868921776e-06,
+      "loss": 0.0346,
+      "step": 17564
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 3.6020084566596195e-06,
+      "loss": 0.0172,
+      "step": 17566
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 3.5967230443974634e-06,
+      "loss": 0.0003,
+      "step": 17568
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 3.591437632135307e-06,
+      "loss": 0.0003,
+      "step": 17570
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 3.5861522198731504e-06,
+      "loss": 0.0004,
+      "step": 17572
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 3.580866807610994e-06,
+      "loss": 0.0006,
+      "step": 17574
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 3.575581395348837e-06,
+      "loss": 0.0066,
+      "step": 17576
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 3.5702959830866806e-06,
+      "loss": 0.001,
+      "step": 17578
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 3.5650105708245245e-06,
+      "loss": 0.0041,
+      "step": 17580
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 3.559725158562368e-06,
+      "loss": 0.0005,
+      "step": 17582
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 3.5544397463002115e-06,
+      "loss": 0.0005,
+      "step": 17584
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 3.549154334038055e-06,
+      "loss": 0.0006,
+      "step": 17586
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 3.543868921775899e-06,
+      "loss": 0.0005,
+      "step": 17588
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 3.5385835095137425e-06,
+      "loss": 0.0015,
+      "step": 17590
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 3.533298097251586e-06,
+      "loss": 0.0029,
+      "step": 17592
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 3.528012684989429e-06,
+      "loss": 0.0004,
+      "step": 17594
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 3.5227272727272726e-06,
+      "loss": 0.002,
+      "step": 17596
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 3.517441860465116e-06,
+      "loss": 0.0069,
+      "step": 17598
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 3.51215644820296e-06,
+      "loss": 0.0047,
+      "step": 17600
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 3.5068710359408036e-06,
+      "loss": 0.0003,
+      "step": 17602
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 3.501585623678647e-06,
+      "loss": 0.0008,
+      "step": 17604
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 3.4963002114164907e-06,
+      "loss": 0.0004,
+      "step": 17606
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 3.4910147991543346e-06,
+      "loss": 0.0092,
+      "step": 17608
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 3.485729386892178e-06,
+      "loss": 0.0017,
+      "step": 17610
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 3.4804439746300212e-06,
+      "loss": 0.0029,
+      "step": 17612
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 3.4751585623678647e-06,
+      "loss": 0.001,
+      "step": 17614
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 3.4698731501057083e-06,
+      "loss": 0.0018,
+      "step": 17616
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 3.4645877378435518e-06,
+      "loss": 0.0067,
+      "step": 17618
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 3.4593023255813957e-06,
+      "loss": 0.0005,
+      "step": 17620
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 3.4540169133192392e-06,
+      "loss": 0.0138,
+      "step": 17622
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 3.4487315010570828e-06,
+      "loss": 0.0013,
+      "step": 17624
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 3.4434460887949263e-06,
+      "loss": 0.0004,
+      "step": 17626
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 3.4381606765327694e-06,
+      "loss": 0.0006,
+      "step": 17628
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 3.432875264270613e-06,
+      "loss": 0.0005,
+      "step": 17630
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 3.427589852008457e-06,
+      "loss": 0.0012,
+      "step": 17632
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 3.4223044397463003e-06,
+      "loss": 0.0254,
+      "step": 17634
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 3.417019027484144e-06,
+      "loss": 0.0005,
+      "step": 17636
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 3.4117336152219874e-06,
+      "loss": 0.0011,
+      "step": 17638
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 3.4064482029598313e-06,
+      "loss": 0.0041,
+      "step": 17640
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 3.401162790697675e-06,
+      "loss": 0.0004,
+      "step": 17642
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 3.3958773784355184e-06,
+      "loss": 0.0005,
+      "step": 17644
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 3.3905919661733615e-06,
+      "loss": 0.0051,
+      "step": 17646
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 3.385306553911205e-06,
+      "loss": 0.0032,
+      "step": 17648
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 3.3800211416490485e-06,
+      "loss": 0.0003,
+      "step": 17650
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 3.3747357293868924e-06,
+      "loss": 0.004,
+      "step": 17652
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 3.369450317124736e-06,
+      "loss": 0.007,
+      "step": 17654
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 3.3641649048625795e-06,
+      "loss": 0.0007,
+      "step": 17656
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 3.358879492600423e-06,
+      "loss": 0.0014,
+      "step": 17658
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 3.353594080338267e-06,
+      "loss": 0.0004,
+      "step": 17660
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 3.3483086680761104e-06,
+      "loss": 0.002,
+      "step": 17662
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 3.3430232558139535e-06,
+      "loss": 0.0007,
+      "step": 17664
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 3.337737843551797e-06,
+      "loss": 0.0006,
+      "step": 17666
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 3.3324524312896406e-06,
+      "loss": 0.0192,
+      "step": 17668
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 3.327167019027484e-06,
+      "loss": 0.0065,
+      "step": 17670
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 3.321881606765328e-06,
+      "loss": 0.0002,
+      "step": 17672
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 3.3165961945031716e-06,
+      "loss": 0.036,
+      "step": 17674
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 3.311310782241015e-06,
+      "loss": 0.0006,
+      "step": 17676
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 3.3060253699788586e-06,
+      "loss": 0.0006,
+      "step": 17678
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 3.3007399577167025e-06,
+      "loss": 0.0006,
+      "step": 17680
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 3.295454545454545e-06,
+      "loss": 0.0004,
+      "step": 17682
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 3.290169133192389e-06,
+      "loss": 0.0024,
+      "step": 17684
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 3.2848837209302327e-06,
+      "loss": 0.0004,
+      "step": 17686
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 3.279598308668076e-06,
+      "loss": 0.0011,
+      "step": 17688
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 3.2743128964059197e-06,
+      "loss": 0.0004,
+      "step": 17690
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 3.2690274841437636e-06,
+      "loss": 0.0007,
+      "step": 17692
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 3.263742071881607e-06,
+      "loss": 0.0026,
+      "step": 17694
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 3.2584566596194507e-06,
+      "loss": 0.0361,
+      "step": 17696
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 3.253171247357294e-06,
+      "loss": 0.0045,
+      "step": 17698
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 3.2478858350951373e-06,
+      "loss": 0.0287,
+      "step": 17700
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 3.242600422832981e-06,
+      "loss": 0.0095,
+      "step": 17702
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 3.2373150105708248e-06,
+      "loss": 0.0004,
+      "step": 17704
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 3.2320295983086683e-06,
+      "loss": 0.0014,
+      "step": 17706
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 3.226744186046512e-06,
+      "loss": 0.0005,
+      "step": 17708
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 3.2214587737843553e-06,
+      "loss": 0.002,
+      "step": 17710
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 3.2161733615221992e-06,
+      "loss": 0.0339,
+      "step": 17712
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 3.2108879492600428e-06,
+      "loss": 0.0011,
+      "step": 17714
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 3.205602536997886e-06,
+      "loss": 0.0011,
+      "step": 17716
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 3.2003171247357294e-06,
+      "loss": 0.0005,
+      "step": 17718
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 3.195031712473573e-06,
+      "loss": 0.0005,
+      "step": 17720
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 3.1897463002114164e-06,
+      "loss": 0.0006,
+      "step": 17722
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 3.1844608879492604e-06,
+      "loss": 0.0003,
+      "step": 17724
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 3.179175475687104e-06,
+      "loss": 0.0003,
+      "step": 17726
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 3.1738900634249474e-06,
+      "loss": 0.0022,
+      "step": 17728
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 3.168604651162791e-06,
+      "loss": 0.0022,
+      "step": 17730
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 3.163319238900635e-06,
+      "loss": 0.0323,
+      "step": 17732
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 3.1580338266384775e-06,
+      "loss": 0.0049,
+      "step": 17734
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 3.1527484143763215e-06,
+      "loss": 0.0007,
+      "step": 17736
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 3.147463002114165e-06,
+      "loss": 0.0311,
+      "step": 17738
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 3.1421775898520085e-06,
+      "loss": 0.0039,
+      "step": 17740
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 3.136892177589852e-06,
+      "loss": 0.0193,
+      "step": 17742
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 3.131606765327696e-06,
+      "loss": 0.0006,
+      "step": 17744
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 3.1263213530655395e-06,
+      "loss": 0.0074,
+      "step": 17746
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 3.1210359408033826e-06,
+      "loss": 0.0008,
+      "step": 17748
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 3.1157505285412265e-06,
+      "loss": 0.0019,
+      "step": 17750
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 3.11046511627907e-06,
+      "loss": 0.0037,
+      "step": 17752
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 3.1051797040169136e-06,
+      "loss": 0.0207,
+      "step": 17754
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 3.099894291754757e-06,
+      "loss": 0.0131,
+      "step": 17756
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 3.0946088794926006e-06,
+      "loss": 0.0034,
+      "step": 17758
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 3.089323467230444e-06,
+      "loss": 0.0005,
+      "step": 17760
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 3.0840380549682876e-06,
+      "loss": 0.0027,
+      "step": 17762
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 3.0787526427061316e-06,
+      "loss": 0.0404,
+      "step": 17764
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 3.0734672304439747e-06,
+      "loss": 0.0173,
+      "step": 17766
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 3.068181818181818e-06,
+      "loss": 0.0083,
+      "step": 17768
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 3.062896405919662e-06,
+      "loss": 0.0021,
+      "step": 17770
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 3.0576109936575056e-06,
+      "loss": 0.0009,
+      "step": 17772
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 3.0523255813953487e-06,
+      "loss": 0.0009,
+      "step": 17774
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 3.0470401691331927e-06,
+      "loss": 0.0037,
+      "step": 17776
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 3.041754756871036e-06,
+      "loss": 0.001,
+      "step": 17778
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 3.0364693446088797e-06,
+      "loss": 0.0078,
+      "step": 17780
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 3.0311839323467232e-06,
+      "loss": 0.0026,
+      "step": 17782
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 3.0258985200845668e-06,
+      "loss": 0.0069,
+      "step": 17784
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 3.0206131078224103e-06,
+      "loss": 0.0006,
+      "step": 17786
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 3.015327695560254e-06,
+      "loss": 0.0007,
+      "step": 17788
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 3.0100422832980977e-06,
+      "loss": 0.0286,
+      "step": 17790
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 3.004756871035941e-06,
+      "loss": 0.0008,
+      "step": 17792
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.9994714587737843e-06,
+      "loss": 0.0005,
+      "step": 17794
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.9941860465116283e-06,
+      "loss": 0.0003,
+      "step": 17796
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.988900634249472e-06,
+      "loss": 0.0008,
+      "step": 17798
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.983615221987315e-06,
+      "loss": 0.0005,
+      "step": 17800
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.978329809725159e-06,
+      "loss": 0.0007,
+      "step": 17802
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.9730443974630024e-06,
+      "loss": 0.0058,
+      "step": 17804
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.967758985200846e-06,
+      "loss": 0.0025,
+      "step": 17806
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.9624735729386894e-06,
+      "loss": 0.0277,
+      "step": 17808
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.957188160676533e-06,
+      "loss": 0.002,
+      "step": 17810
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.9519027484143764e-06,
+      "loss": 0.0102,
+      "step": 17812
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.94661733615222e-06,
+      "loss": 0.0009,
+      "step": 17814
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.941331923890064e-06,
+      "loss": 0.0006,
+      "step": 17816
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.936046511627907e-06,
+      "loss": 0.0009,
+      "step": 17818
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.9307610993657505e-06,
+      "loss": 0.0026,
+      "step": 17820
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.9254756871035944e-06,
+      "loss": 0.0004,
+      "step": 17822
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.920190274841438e-06,
+      "loss": 0.001,
+      "step": 17824
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.914904862579281e-06,
+      "loss": 0.0005,
+      "step": 17826
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.909619450317125e-06,
+      "loss": 0.0021,
+      "step": 17828
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.9043340380549685e-06,
+      "loss": 0.0007,
+      "step": 17830
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.899048625792812e-06,
+      "loss": 0.0008,
+      "step": 17832
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.8937632135306556e-06,
+      "loss": 0.0005,
+      "step": 17834
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.888477801268499e-06,
+      "loss": 0.0023,
+      "step": 17836
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.8831923890063426e-06,
+      "loss": 0.0118,
+      "step": 17838
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.877906976744186e-06,
+      "loss": 0.0006,
+      "step": 17840
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.87262156448203e-06,
+      "loss": 0.0119,
+      "step": 17842
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.867336152219873e-06,
+      "loss": 0.0009,
+      "step": 17844
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.8620507399577167e-06,
+      "loss": 0.0013,
+      "step": 17846
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.8567653276955606e-06,
+      "loss": 0.0023,
+      "step": 17848
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.851479915433404e-06,
+      "loss": 0.0005,
+      "step": 17850
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.8461945031712476e-06,
+      "loss": 0.0621,
+      "step": 17852
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.840909090909091e-06,
+      "loss": 0.0032,
+      "step": 17854
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.8356236786469347e-06,
+      "loss": 0.005,
+      "step": 17856
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.830338266384778e-06,
+      "loss": 0.0021,
+      "step": 17858
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.8250528541226217e-06,
+      "loss": 0.0103,
+      "step": 17860
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.8197674418604652e-06,
+      "loss": 0.0012,
+      "step": 17862
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.8144820295983088e-06,
+      "loss": 0.0004,
+      "step": 17864
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.8091966173361523e-06,
+      "loss": 0.0003,
+      "step": 17866
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.8039112050739962e-06,
+      "loss": 0.0003,
+      "step": 17868
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.7986257928118393e-06,
+      "loss": 0.0013,
+      "step": 17870
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.793340380549683e-06,
+      "loss": 0.0006,
+      "step": 17872
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.7880549682875268e-06,
+      "loss": 0.0406,
+      "step": 17874
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.7827695560253703e-06,
+      "loss": 0.0008,
+      "step": 17876
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.777484143763214e-06,
+      "loss": 0.0009,
+      "step": 17878
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.772198731501057e-06,
+      "loss": 0.0007,
+      "step": 17880
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.766913319238901e-06,
+      "loss": 0.0006,
+      "step": 17882
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.7616279069767444e-06,
+      "loss": 0.0003,
+      "step": 17884
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.756342494714588e-06,
+      "loss": 0.002,
+      "step": 17886
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.7510570824524314e-06,
+      "loss": 0.007,
+      "step": 17888
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.745771670190275e-06,
+      "loss": 0.006,
+      "step": 17890
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.7404862579281184e-06,
+      "loss": 0.0042,
+      "step": 17892
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.7352008456659624e-06,
+      "loss": 0.0003,
+      "step": 17894
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.7299154334038055e-06,
+      "loss": 0.0058,
+      "step": 17896
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.724630021141649e-06,
+      "loss": 0.0012,
+      "step": 17898
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.719344608879493e-06,
+      "loss": 0.0025,
+      "step": 17900
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.7140591966173364e-06,
+      "loss": 0.0011,
+      "step": 17902
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.70877378435518e-06,
+      "loss": 0.0003,
+      "step": 17904
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.703488372093023e-06,
+      "loss": 0.0016,
+      "step": 17906
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.698202959830867e-06,
+      "loss": 0.0005,
+      "step": 17908
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.6929175475687105e-06,
+      "loss": 0.0006,
+      "step": 17910
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.687632135306554e-06,
+      "loss": 0.0004,
+      "step": 17912
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.6823467230443976e-06,
+      "loss": 0.0141,
+      "step": 17914
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.677061310782241e-06,
+      "loss": 0.0009,
+      "step": 17916
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.6717758985200846e-06,
+      "loss": 0.0004,
+      "step": 17918
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.6664904862579285e-06,
+      "loss": 0.0007,
+      "step": 17920
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.661205073995772e-06,
+      "loss": 0.0022,
+      "step": 17922
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.655919661733615e-06,
+      "loss": 0.0007,
+      "step": 17924
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.6506342494714587e-06,
+      "loss": 0.0018,
+      "step": 17926
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.6453488372093026e-06,
+      "loss": 0.0003,
+      "step": 17928
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.640063424947146e-06,
+      "loss": 0.0008,
+      "step": 17930
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.6347780126849892e-06,
+      "loss": 0.0012,
+      "step": 17932
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.629492600422833e-06,
+      "loss": 0.0013,
+      "step": 17934
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.6242071881606767e-06,
+      "loss": 0.0003,
+      "step": 17936
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.61892177589852e-06,
+      "loss": 0.0012,
+      "step": 17938
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.6136363636363637e-06,
+      "loss": 0.0004,
+      "step": 17940
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.6083509513742072e-06,
+      "loss": 0.0146,
+      "step": 17942
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.6030655391120508e-06,
+      "loss": 0.0003,
+      "step": 17944
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.5977801268498947e-06,
+      "loss": 0.0022,
+      "step": 17946
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.5924947145877382e-06,
+      "loss": 0.0004,
+      "step": 17948
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.5872093023255813e-06,
+      "loss": 0.0008,
+      "step": 17950
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.581923890063425e-06,
+      "loss": 0.0023,
+      "step": 17952
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.5766384778012688e-06,
+      "loss": 0.0045,
+      "step": 17954
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.5713530655391123e-06,
+      "loss": 0.0004,
+      "step": 17956
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.5660676532769554e-06,
+      "loss": 0.0008,
+      "step": 17958
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.5607822410147993e-06,
+      "loss": 0.0017,
+      "step": 17960
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.555496828752643e-06,
+      "loss": 0.0004,
+      "step": 17962
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.5502114164904864e-06,
+      "loss": 0.0005,
+      "step": 17964
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.5449260042283303e-06,
+      "loss": 0.001,
+      "step": 17966
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.5396405919661734e-06,
+      "loss": 0.001,
+      "step": 17968
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.534355179704017e-06,
+      "loss": 0.0009,
+      "step": 17970
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.5290697674418604e-06,
+      "loss": 0.0006,
+      "step": 17972
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.5237843551797044e-06,
+      "loss": 0.0012,
+      "step": 17974
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.5184989429175475e-06,
+      "loss": 0.0093,
+      "step": 17976
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.513213530655391e-06,
+      "loss": 0.001,
+      "step": 17978
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.507928118393235e-06,
+      "loss": 0.0007,
+      "step": 17980
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.5026427061310784e-06,
+      "loss": 0.0007,
+      "step": 17982
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.4973572938689215e-06,
+      "loss": 0.0007,
+      "step": 17984
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.4920718816067655e-06,
+      "loss": 0.0109,
+      "step": 17986
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.486786469344609e-06,
+      "loss": 0.0338,
+      "step": 17988
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.4815010570824525e-06,
+      "loss": 0.0005,
+      "step": 17990
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.4762156448202965e-06,
+      "loss": 0.0145,
+      "step": 17992
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.4709302325581396e-06,
+      "loss": 0.0005,
+      "step": 17994
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.465644820295983e-06,
+      "loss": 0.011,
+      "step": 17996
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.4603594080338266e-06,
+      "loss": 0.003,
+      "step": 17998
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.4550739957716705e-06,
+      "loss": 0.0014,
+      "step": 18000
+    },
+    {
+      "epoch": 19.03,
+      "eval_cer": 0.05528640638358507,
+      "eval_loss": 0.782919704914093,
+      "eval_runtime": 125.7137,
+      "eval_samples_per_second": 6.69,
+      "eval_steps_per_second": 0.843,
+      "step": 18000
+    }
+  ],
+  "max_steps": 18920,
+  "num_train_epochs": 20,
+  "total_flos": 2.130711393070083e+20,
+  "trial_name": null,
+  "trial_params": null
+}