diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,11101 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.999458581483487,
+  "global_step": 3692,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 2.6714,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 2.4616,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 2.5089,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1621621621621624e-05,
+      "loss": 2.5389,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.702702702702703e-05,
+      "loss": 2.4892,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2432432432432436e-05,
+      "loss": 2.4949,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.783783783783784e-05,
+      "loss": 2.5016,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.324324324324325e-05,
+      "loss": 2.5727,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8648648648648654e-05,
+      "loss": 2.605,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.405405405405406e-05,
+      "loss": 2.5045,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9459459459459466e-05,
+      "loss": 2.5509,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.486486486486487e-05,
+      "loss": 2.6363,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.027027027027028e-05,
+      "loss": 2.3918,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.567567567567568e-05,
+      "loss": 2.5172,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.108108108108109e-05,
+      "loss": 2.6391,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.64864864864865e-05,
+      "loss": 2.3696,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.18918918918919e-05,
+      "loss": 2.4359,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.729729729729731e-05,
+      "loss": 2.5197,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999998153008212e-05,
+      "loss": 2.5001,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999983377082087e-05,
+      "loss": 2.7154,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999953825273504e-05,
+      "loss": 2.5842,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999909497669792e-05,
+      "loss": 2.4231,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99985039440195e-05,
+      "loss": 2.6799,
+      "step": 46
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999776515644638e-05,
+      "loss": 2.5285,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999687861616181e-05,
+      "loss": 2.6049,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999584432578569e-05,
+      "loss": 2.46,
+      "step": 52
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999466228837451e-05,
+      "loss": 2.5835,
+      "step": 54
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999333250742145e-05,
+      "loss": 2.6909,
+      "step": 56
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999185498685624e-05,
+      "loss": 2.4055,
+      "step": 58
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999022973104525e-05,
+      "loss": 2.5816,
+      "step": 60
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99884567447914e-05,
+      "loss": 2.5871,
+      "step": 62
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998653603333418e-05,
+      "loss": 2.701,
+      "step": 64
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998446760234966e-05,
+      "loss": 2.5853,
+      "step": 66
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998225145795046e-05,
+      "loss": 2.4068,
+      "step": 68
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.997988760668566e-05,
+      "loss": 2.5417,
+      "step": 70
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.997737605554092e-05,
+      "loss": 2.531,
+      "step": 72
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.997471681193833e-05,
+      "loss": 2.5509,
+      "step": 74
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.997190988373644e-05,
+      "loss": 2.4764,
+      "step": 76
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.996895527923023e-05,
+      "loss": 2.4522,
+      "step": 78
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.996585300715116e-05,
+      "loss": 2.332,
+      "step": 80
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.996260307666696e-05,
+      "loss": 2.518,
+      "step": 82
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.995920549738183e-05,
+      "loss": 2.5257,
+      "step": 84
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.995566027933621e-05,
+      "loss": 2.4631,
+      "step": 86
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.995196743300692e-05,
+      "loss": 2.6791,
+      "step": 88
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.994812696930698e-05,
+      "loss": 2.5043,
+      "step": 90
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.994413889958568e-05,
+      "loss": 2.5534,
+      "step": 92
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.994000323562852e-05,
+      "loss": 2.3725,
+      "step": 94
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.993571998965714e-05,
+      "loss": 2.7813,
+      "step": 96
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.993128917432934e-05,
+      "loss": 2.4455,
+      "step": 98
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.992671080273903e-05,
+      "loss": 2.4049,
+      "step": 100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.992198488841611e-05,
+      "loss": 2.4881,
+      "step": 102
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.991711144532654e-05,
+      "loss": 2.4836,
+      "step": 104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.991209048787228e-05,
+      "loss": 2.4244,
+      "step": 106
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.990692203089119e-05,
+      "loss": 2.5105,
+      "step": 108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.9901606089657e-05,
+      "loss": 2.6217,
+      "step": 110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.989614267987933e-05,
+      "loss": 2.5529,
+      "step": 112
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.989053181770356e-05,
+      "loss": 2.512,
+      "step": 114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.988477351971084e-05,
+      "loss": 2.5365,
+      "step": 116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.9878867802918e-05,
+      "loss": 2.6218,
+      "step": 118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.987281468477756e-05,
+      "loss": 2.4998,
+      "step": 120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.986661418317759e-05,
+      "loss": 2.4597,
+      "step": 122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.986026631644173e-05,
+      "loss": 2.6517,
+      "step": 124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.985377110332912e-05,
+      "loss": 2.6685,
+      "step": 126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.984712856303432e-05,
+      "loss": 2.5139,
+      "step": 128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.984033871518727e-05,
+      "loss": 2.4927,
+      "step": 130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.983340157985324e-05,
+      "loss": 2.4832,
+      "step": 132
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.982631717753275e-05,
+      "loss": 2.5724,
+      "step": 134
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.981908552916153e-05,
+      "loss": 2.5026,
+      "step": 136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.981170665611046e-05,
+      "loss": 2.4104,
+      "step": 138
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.980418058018547e-05,
+      "loss": 2.4806,
+      "step": 140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.979650732362753e-05,
+      "loss": 2.3937,
+      "step": 142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.978868690911253e-05,
+      "loss": 2.7488,
+      "step": 144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.978071935975126e-05,
+      "loss": 2.5604,
+      "step": 146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.977260469908931e-05,
+      "loss": 2.4197,
+      "step": 148
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.976434295110701e-05,
+      "loss": 2.6003,
+      "step": 150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.975593414021938e-05,
+      "loss": 2.4822,
+      "step": 152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.974737829127602e-05,
+      "loss": 2.4176,
+      "step": 154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.973867542956104e-05,
+      "loss": 2.5665,
+      "step": 156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.972982558079303e-05,
+      "loss": 2.7229,
+      "step": 158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.972082877112494e-05,
+      "loss": 2.7438,
+      "step": 160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.971168502714401e-05,
+      "loss": 2.5602,
+      "step": 162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.970239437587174e-05,
+      "loss": 2.5438,
+      "step": 164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.96929568447637e-05,
+      "loss": 2.5334,
+      "step": 166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.968337246170956e-05,
+      "loss": 2.496,
+      "step": 168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.967364125503295e-05,
+      "loss": 2.4688,
+      "step": 170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.966376325349143e-05,
+      "loss": 2.5845,
+      "step": 172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.965373848627631e-05,
+      "loss": 2.3486,
+      "step": 174
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.964356698301264e-05,
+      "loss": 2.6254,
+      "step": 176
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.963324877375912e-05,
+      "loss": 2.2205,
+      "step": 178
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.9622783889008e-05,
+      "loss": 2.4156,
+      "step": 180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.961217235968494e-05,
+      "loss": 2.6328,
+      "step": 182
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.960141421714897e-05,
+      "loss": 2.5354,
+      "step": 184
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.959050949319244e-05,
+      "loss": 2.4779,
+      "step": 186
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.957945822004083e-05,
+      "loss": 2.4148,
+      "step": 188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.956826043035268e-05,
+      "loss": 2.5947,
+      "step": 190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.95569161572196e-05,
+      "loss": 2.6662,
+      "step": 192
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.954542543416599e-05,
+      "loss": 2.4853,
+      "step": 194
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.953378829514908e-05,
+      "loss": 2.4561,
+      "step": 196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.952200477455881e-05,
+      "loss": 2.5139,
+      "step": 198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.951007490721766e-05,
+      "loss": 2.5978,
+      "step": 200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.949799872838061e-05,
+      "loss": 2.6275,
+      "step": 202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.948577627373503e-05,
+      "loss": 2.4584,
+      "step": 204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.947340757940053e-05,
+      "loss": 2.3811,
+      "step": 206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.946089268192895e-05,
+      "loss": 2.4847,
+      "step": 208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.944823161830407e-05,
+      "loss": 2.5872,
+      "step": 210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.943542442594177e-05,
+      "loss": 2.7864,
+      "step": 212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.942247114268964e-05,
+      "loss": 2.5991,
+      "step": 214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.940937180682706e-05,
+      "loss": 2.4734,
+      "step": 216
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.9396126457065e-05,
+      "loss": 2.6016,
+      "step": 218
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.938273513254597e-05,
+      "loss": 2.5428,
+      "step": 220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.936919787284378e-05,
+      "loss": 2.6381,
+      "step": 222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.935551471796358e-05,
+      "loss": 2.6325,
+      "step": 224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.934168570834165e-05,
+      "loss": 2.5066,
+      "step": 226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.932771088484527e-05,
+      "loss": 2.5039,
+      "step": 228
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.931359028877267e-05,
+      "loss": 2.7124,
+      "step": 230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.929932396185281e-05,
+      "loss": 2.4925,
+      "step": 232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.928491194624539e-05,
+      "loss": 2.579,
+      "step": 234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.927035428454055e-05,
+      "loss": 2.6093,
+      "step": 236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.925565101975893e-05,
+      "loss": 2.3589,
+      "step": 238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.924080219535141e-05,
+      "loss": 2.6058,
+      "step": 240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.922580785519901e-05,
+      "loss": 2.5016,
+      "step": 242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.921066804361284e-05,
+      "loss": 2.6041,
+      "step": 244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.919538280533382e-05,
+      "loss": 2.4222,
+      "step": 246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.917995218553271e-05,
+      "loss": 2.5596,
+      "step": 248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.916437622980985e-05,
+      "loss": 2.5427,
+      "step": 250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.91486549841951e-05,
+      "loss": 2.5865,
+      "step": 252
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.913278849514765e-05,
+      "loss": 2.4464,
+      "step": 254
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.911677680955596e-05,
+      "loss": 2.4279,
+      "step": 256
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.910061997473752e-05,
+      "loss": 2.4858,
+      "step": 258
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.908431803843881e-05,
+      "loss": 2.3309,
+      "step": 260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.906787104883506e-05,
+      "loss": 2.5427,
+      "step": 262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.905127905453023e-05,
+      "loss": 2.568,
+      "step": 264
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.90345421045567e-05,
+      "loss": 2.5368,
+      "step": 266
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.90176602483753e-05,
+      "loss": 2.562,
+      "step": 268
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.90006335358751e-05,
+      "loss": 2.43,
+      "step": 270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.898346201737317e-05,
+      "loss": 2.6544,
+      "step": 272
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.896614574361454e-05,
+      "loss": 2.5701,
+      "step": 274
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.894868476577201e-05,
+      "loss": 2.5294,
+      "step": 276
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.893107913544609e-05,
+      "loss": 2.6586,
+      "step": 278
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.891332890466463e-05,
+      "loss": 2.4904,
+      "step": 280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.88954341258829e-05,
+      "loss": 2.5686,
+      "step": 282
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.887739485198331e-05,
+      "loss": 2.5522,
+      "step": 284
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.885921113627525e-05,
+      "loss": 2.6227,
+      "step": 286
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.884088303249501e-05,
+      "loss": 2.6265,
+      "step": 288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.882241059480555e-05,
+      "loss": 2.6851,
+      "step": 290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.880379387779637e-05,
+      "loss": 2.4501,
+      "step": 292
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.878503293648332e-05,
+      "loss": 2.3563,
+      "step": 294
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.876612782630848e-05,
+      "loss": 2.3076,
+      "step": 296
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.874707860313997e-05,
+      "loss": 2.5158,
+      "step": 298
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.87278853232718e-05,
+      "loss": 2.4241,
+      "step": 300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.87085480434237e-05,
+      "loss": 2.6496,
+      "step": 302
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.868906682074093e-05,
+      "loss": 2.5175,
+      "step": 304
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.866944171279411e-05,
+      "loss": 2.4658,
+      "step": 306
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.864967277757911e-05,
+      "loss": 2.3329,
+      "step": 308
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.862976007351683e-05,
+      "loss": 2.6255,
+      "step": 310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.860970365945299e-05,
+      "loss": 2.5342,
+      "step": 312
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.858950359465805e-05,
+      "loss": 2.3472,
+      "step": 314
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.856915993882696e-05,
+      "loss": 2.4315,
+      "step": 316
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.854867275207901e-05,
+      "loss": 2.5811,
+      "step": 318
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.852804209495766e-05,
+      "loss": 2.5231,
+      "step": 320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.850726802843034e-05,
+      "loss": 2.6642,
+      "step": 322
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.84863506138883e-05,
+      "loss": 2.5686,
+      "step": 324
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.846528991314639e-05,
+      "loss": 2.7031,
+      "step": 326
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.844408598844288e-05,
+      "loss": 2.4843,
+      "step": 328
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.842273890243936e-05,
+      "loss": 2.6201,
+      "step": 330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.840124871822041e-05,
+      "loss": 2.4424,
+      "step": 332
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.837961549929356e-05,
+      "loss": 2.5032,
+      "step": 334
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.835783930958897e-05,
+      "loss": 2.6527,
+      "step": 336
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.833592021345937e-05,
+      "loss": 2.4956,
+      "step": 338
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.831385827567975e-05,
+      "loss": 2.3891,
+      "step": 340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.829165356144727e-05,
+      "loss": 2.2116,
+      "step": 342
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.826930613638098e-05,
+      "loss": 2.5029,
+      "step": 344
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.824681606652168e-05,
+      "loss": 2.5519,
+      "step": 346
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.822418341833172e-05,
+      "loss": 2.5432,
+      "step": 348
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.82014082586948e-05,
+      "loss": 2.5587,
+      "step": 350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.817849065491575e-05,
+      "loss": 2.552,
+      "step": 352
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.815543067472039e-05,
+      "loss": 2.4258,
+      "step": 354
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.813222838625521e-05,
+      "loss": 2.4393,
+      "step": 356
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.810888385808732e-05,
+      "loss": 2.5239,
+      "step": 358
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.808539715920414e-05,
+      "loss": 2.4571,
+      "step": 360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.806176835901328e-05,
+      "loss": 2.5202,
+      "step": 362
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.803799752734219e-05,
+      "loss": 2.4761,
+      "step": 364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.801408473443816e-05,
+      "loss": 2.8216,
+      "step": 366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.79900300509679e-05,
+      "loss": 2.6249,
+      "step": 368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.796583354801752e-05,
+      "loss": 2.5059,
+      "step": 370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.794149529709216e-05,
+      "loss": 2.5326,
+      "step": 372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.791701537011591e-05,
+      "loss": 2.7352,
+      "step": 374
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.789239383943152e-05,
+      "loss": 2.6958,
+      "step": 376
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.78676307778002e-05,
+      "loss": 2.4794,
+      "step": 378
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.784272625840136e-05,
+      "loss": 2.5559,
+      "step": 380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.781768035483256e-05,
+      "loss": 2.4731,
+      "step": 382
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.779249314110909e-05,
+      "loss": 2.5427,
+      "step": 384
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.776716469166384e-05,
+      "loss": 2.6534,
+      "step": 386
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.774169508134715e-05,
+      "loss": 2.5991,
+      "step": 388
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.771608438542639e-05,
+      "loss": 2.4887,
+      "step": 390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.769033267958598e-05,
+      "loss": 2.5762,
+      "step": 392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.766444003992703e-05,
+      "loss": 2.3775,
+      "step": 394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.763840654296706e-05,
+      "loss": 2.4067,
+      "step": 396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.761223226563996e-05,
+      "loss": 2.3338,
+      "step": 398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.758591728529555e-05,
+      "loss": 2.4981,
+      "step": 400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.755946167969952e-05,
+      "loss": 2.402,
+      "step": 402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.753286552703312e-05,
+      "loss": 2.7678,
+      "step": 404
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.750612890589293e-05,
+      "loss": 2.5216,
+      "step": 406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.747925189529063e-05,
+      "loss": 2.3811,
+      "step": 408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.745223457465282e-05,
+      "loss": 2.4442,
+      "step": 410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.742507702382071e-05,
+      "loss": 2.4474,
+      "step": 412
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.739777932304992e-05,
+      "loss": 2.5238,
+      "step": 414
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.737034155301024e-05,
+      "loss": 2.4573,
+      "step": 416
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.734276379478538e-05,
+      "loss": 2.5096,
+      "step": 418
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.731504612987279e-05,
+      "loss": 2.5997,
+      "step": 420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.728718864018329e-05,
+      "loss": 2.3851,
+      "step": 422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.725919140804099e-05,
+      "loss": 2.5155,
+      "step": 424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.72310545161829e-05,
+      "loss": 2.614,
+      "step": 426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.72027780477588e-05,
+      "loss": 2.5027,
+      "step": 428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.717436208633088e-05,
+      "loss": 2.4011,
+      "step": 430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.714580671587366e-05,
+      "loss": 2.5327,
+      "step": 432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.711711202077354e-05,
+      "loss": 2.3772,
+      "step": 434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.708827808582871e-05,
+      "loss": 2.5332,
+      "step": 436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.70593049962488e-05,
+      "loss": 2.4211,
+      "step": 438
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.703019283765471e-05,
+      "loss": 2.5811,
+      "step": 440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.700094169607828e-05,
+      "loss": 2.5859,
+      "step": 442
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.697155165796209e-05,
+      "loss": 2.7103,
+      "step": 444
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.694202281015918e-05,
+      "loss": 2.5435,
+      "step": 446
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.691235523993278e-05,
+      "loss": 2.4327,
+      "step": 448
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.688254903495609e-05,
+      "loss": 2.4772,
+      "step": 450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.685260428331202e-05,
+      "loss": 2.5445,
+      "step": 452
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.682252107349288e-05,
+      "loss": 2.5334,
+      "step": 454
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.679229949440015e-05,
+      "loss": 2.4343,
+      "step": 456
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.676193963534423e-05,
+      "loss": 2.4341,
+      "step": 458
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.673144158604419e-05,
+      "loss": 2.5318,
+      "step": 460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.67008054366274e-05,
+      "loss": 2.435,
+      "step": 462
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.667003127762942e-05,
+      "loss": 2.6514,
+      "step": 464
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.663911919999362e-05,
+      "loss": 2.4744,
+      "step": 466
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.660806929507095e-05,
+      "loss": 2.5498,
+      "step": 468
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.657688165461965e-05,
+      "loss": 2.6276,
+      "step": 470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.654555637080502e-05,
+      "loss": 2.6097,
+      "step": 472
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.65140935361991e-05,
+      "loss": 2.3675,
+      "step": 474
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.648249324378044e-05,
+      "loss": 2.469,
+      "step": 476
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.64507555869338e-05,
+      "loss": 2.5554,
+      "step": 478
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.641888065944984e-05,
+      "loss": 2.35,
+      "step": 480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.638686855552494e-05,
+      "loss": 2.495,
+      "step": 482
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.635471936976081e-05,
+      "loss": 2.4547,
+      "step": 484
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63224331971643e-05,
+      "loss": 2.3311,
+      "step": 486
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.629001013314705e-05,
+      "loss": 2.5145,
+      "step": 488
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.625745027352526e-05,
+      "loss": 2.5413,
+      "step": 490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.622475371451939e-05,
+      "loss": 2.5209,
+      "step": 492
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.619192055275386e-05,
+      "loss": 2.4376,
+      "step": 494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.615895088525677e-05,
+      "loss": 2.455,
+      "step": 496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.612584480945964e-05,
+      "loss": 2.3267,
+      "step": 498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.609260242319709e-05,
+      "loss": 2.6783,
+      "step": 500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.605922382470658e-05,
+      "loss": 2.4699,
+      "step": 502
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.602570911262805e-05,
+      "loss": 2.4731,
+      "step": 504
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.599205838600375e-05,
+      "loss": 2.4779,
+      "step": 506
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.595827174427786e-05,
+      "loss": 2.6002,
+      "step": 508
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.592434928729616e-05,
+      "loss": 2.5211,
+      "step": 510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.589029111530586e-05,
+      "loss": 2.5308,
+      "step": 512
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.585609732895517e-05,
+      "loss": 2.5043,
+      "step": 514
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.582176802929315e-05,
+      "loss": 2.4683,
+      "step": 516
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.578730331776924e-05,
+      "loss": 2.4226,
+      "step": 518
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.575270329623309e-05,
+      "loss": 2.3188,
+      "step": 520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.571796806693422e-05,
+      "loss": 2.6134,
+      "step": 522
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.568309773252171e-05,
+      "loss": 2.3866,
+      "step": 524
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.564809239604388e-05,
+      "loss": 2.6209,
+      "step": 526
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.5612952160948e-05,
+      "loss": 2.5562,
+      "step": 528
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.557767713108009e-05,
+      "loss": 2.6116,
+      "step": 530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.554226741068432e-05,
+      "loss": 2.5081,
+      "step": 532
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.550672310440311e-05,
+      "loss": 2.7225,
+      "step": 534
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.547104431727647e-05,
+      "loss": 2.6353,
+      "step": 536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.543523115474187e-05,
+      "loss": 2.5658,
+      "step": 538
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.539928372263387e-05,
+      "loss": 2.5224,
+      "step": 540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.536320212718382e-05,
+      "loss": 2.702,
+      "step": 542
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.532698647501958e-05,
+      "loss": 2.4106,
+      "step": 544
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.529063687316513e-05,
+      "loss": 2.5009,
+      "step": 546
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.525415342904034e-05,
+      "loss": 2.4723,
+      "step": 548
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.521753625046056e-05,
+      "loss": 2.3118,
+      "step": 550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.51807854456364e-05,
+      "loss": 2.6302,
+      "step": 552
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.51439011231733e-05,
+      "loss": 2.3981,
+      "step": 554
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.510688339207133e-05,
+      "loss": 2.4194,
+      "step": 556
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.506973236172478e-05,
+      "loss": 2.5114,
+      "step": 558
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.503244814192187e-05,
+      "loss": 2.4885,
+      "step": 560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.499503084284441e-05,
+      "loss": 2.4262,
+      "step": 562
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.49574805750675e-05,
+      "loss": 2.2484,
+      "step": 564
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.491979744955915e-05,
+      "loss": 2.3817,
+      "step": 566
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.488198157768005e-05,
+      "loss": 2.455,
+      "step": 568
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.484403307118312e-05,
+      "loss": 2.3778,
+      "step": 570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.480595204221329e-05,
+      "loss": 2.5499,
+      "step": 572
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.47677386033071e-05,
+      "loss": 2.5838,
+      "step": 574
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.472939286739235e-05,
+      "loss": 2.5077,
+      "step": 576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.469091494778785e-05,
+      "loss": 2.6054,
+      "step": 578
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.465230495820303e-05,
+      "loss": 2.4767,
+      "step": 580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.461356301273758e-05,
+      "loss": 2.6251,
+      "step": 582
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.45746892258812e-05,
+      "loss": 2.4725,
+      "step": 584
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.453568371251316e-05,
+      "loss": 2.5408,
+      "step": 586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.4496546587902e-05,
+      "loss": 2.5397,
+      "step": 588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.445727796770524e-05,
+      "loss": 2.666,
+      "step": 590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.441787796796896e-05,
+      "loss": 2.3805,
+      "step": 592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.43783467051275e-05,
+      "loss": 2.57,
+      "step": 594
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.43386842960031e-05,
+      "loss": 2.6776,
+      "step": 596
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.429889085780557e-05,
+      "loss": 2.447,
+      "step": 598
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.425896650813196e-05,
+      "loss": 2.6253,
+      "step": 600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.421891136496612e-05,
+      "loss": 2.3422,
+      "step": 602
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.41787255466785e-05,
+      "loss": 2.3565,
+      "step": 604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.413840917202566e-05,
+      "loss": 2.4135,
+      "step": 606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.409796236014999e-05,
+      "loss": 2.6778,
+      "step": 608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.405738523057938e-05,
+      "loss": 2.4313,
+      "step": 610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.401667790322679e-05,
+      "loss": 2.4427,
+      "step": 612
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.397584049838996e-05,
+      "loss": 2.6661,
+      "step": 614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.393487313675102e-05,
+      "loss": 2.4825,
+      "step": 616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.389377593937618e-05,
+      "loss": 2.5834,
+      "step": 618
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.38525490277153e-05,
+      "loss": 2.4413,
+      "step": 620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.38111925236016e-05,
+      "loss": 2.5265,
+      "step": 622
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.376970654925124e-05,
+      "loss": 2.5181,
+      "step": 624
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.372809122726299e-05,
+      "loss": 2.6319,
+      "step": 626
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.368634668061791e-05,
+      "loss": 2.7302,
+      "step": 628
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.364447303267889e-05,
+      "loss": 2.5624,
+      "step": 630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.360247040719039e-05,
+      "loss": 2.4739,
+      "step": 632
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.356033892827796e-05,
+      "loss": 2.3922,
+      "step": 634
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.3518078720448e-05,
+      "loss": 2.5975,
+      "step": 636
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.347568990858726e-05,
+      "loss": 2.4533,
+      "step": 638
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.343317261796262e-05,
+      "loss": 2.4955,
+      "step": 640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.339052697422057e-05,
+      "loss": 2.4098,
+      "step": 642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.334775310338694e-05,
+      "loss": 2.705,
+      "step": 644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.330485113186648e-05,
+      "loss": 2.4335,
+      "step": 646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.326182118644254e-05,
+      "loss": 2.6452,
+      "step": 648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.321866339427658e-05,
+      "loss": 2.4124,
+      "step": 650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.317537788290794e-05,
+      "loss": 2.3303,
+      "step": 652
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.313196478025337e-05,
+      "loss": 2.4199,
+      "step": 654
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.308842421460667e-05,
+      "loss": 2.4577,
+      "step": 656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.304475631463834e-05,
+      "loss": 2.5357,
+      "step": 658
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.300096120939516e-05,
+      "loss": 2.418,
+      "step": 660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.29570390282998e-05,
+      "loss": 2.4678,
+      "step": 662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.29129899011505e-05,
+      "loss": 2.3826,
+      "step": 664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.286881395812066e-05,
+      "loss": 2.4426,
+      "step": 666
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.28245113297584e-05,
+      "loss": 2.3528,
+      "step": 668
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.278008214698624e-05,
+      "loss": 2.4743,
+      "step": 670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.27355265411007e-05,
+      "loss": 2.3887,
+      "step": 672
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.26908446437719e-05,
+      "loss": 2.468,
+      "step": 674
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.264603658704318e-05,
+      "loss": 2.352,
+      "step": 676
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.260110250333066e-05,
+      "loss": 2.4429,
+      "step": 678
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.255604252542296e-05,
+      "loss": 2.3712,
+      "step": 680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.251085678648072e-05,
+      "loss": 2.5003,
+      "step": 682
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.246554542003618e-05,
+      "loss": 2.5098,
+      "step": 684
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.24201085599929e-05,
+      "loss": 2.6184,
+      "step": 686
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.237454634062525e-05,
+      "loss": 2.4172,
+      "step": 688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.23288588965781e-05,
+      "loss": 2.4234,
+      "step": 690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.228304636286633e-05,
+      "loss": 2.6617,
+      "step": 692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.223710887487453e-05,
+      "loss": 2.422,
+      "step": 694
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.219104656835654e-05,
+      "loss": 2.6878,
+      "step": 696
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.214485957943503e-05,
+      "loss": 2.6575,
+      "step": 698
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.209854804460121e-05,
+      "loss": 2.369,
+      "step": 700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.205211210071426e-05,
+      "loss": 2.5432,
+      "step": 702
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.200555188500103e-05,
+      "loss": 2.5313,
+      "step": 704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.195886753505565e-05,
+      "loss": 2.3887,
+      "step": 706
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.191205918883909e-05,
+      "loss": 2.6655,
+      "step": 708
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.18651269846787e-05,
+      "loss": 2.5833,
+      "step": 710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.181807106126792e-05,
+      "loss": 2.6638,
+      "step": 712
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.177089155766574e-05,
+      "loss": 2.4395,
+      "step": 714
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.172358861329641e-05,
+      "loss": 2.5247,
+      "step": 716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.167616236794894e-05,
+      "loss": 2.4711,
+      "step": 718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.162861296177671e-05,
+      "loss": 2.4537,
+      "step": 720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.158094053529709e-05,
+      "loss": 2.4404,
+      "step": 722
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.153314522939096e-05,
+      "loss": 2.4599,
+      "step": 724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.148522718530236e-05,
+      "loss": 2.5289,
+      "step": 726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.143718654463804e-05,
+      "loss": 2.2966,
+      "step": 728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.138902344936706e-05,
+      "loss": 2.4635,
+      "step": 730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.134073804182033e-05,
+      "loss": 2.6182,
+      "step": 732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.129233046469022e-05,
+      "loss": 2.6568,
+      "step": 734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.124380086103013e-05,
+      "loss": 2.5841,
+      "step": 736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.11951493742541e-05,
+      "loss": 2.609,
+      "step": 738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.114637614813634e-05,
+      "loss": 2.3299,
+      "step": 740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.109748132681082e-05,
+      "loss": 2.5093,
+      "step": 742
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.104846505477083e-05,
+      "loss": 2.4223,
+      "step": 744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.09993274768686e-05,
+      "loss": 2.4636,
+      "step": 746
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.095006873831479e-05,
+      "loss": 2.3136,
+      "step": 748
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.090068898467823e-05,
+      "loss": 2.557,
+      "step": 750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.085118836188521e-05,
+      "loss": 2.4634,
+      "step": 752
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.080156701621936e-05,
+      "loss": 2.5238,
+      "step": 754
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.075182509432095e-05,
+      "loss": 2.4833,
+      "step": 756
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.070196274318666e-05,
+      "loss": 2.6603,
+      "step": 758
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.0651980110169e-05,
+      "loss": 2.4763,
+      "step": 760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.060187734297599e-05,
+      "loss": 2.4662,
+      "step": 762
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.055165458967063e-05,
+      "loss": 2.4409,
+      "step": 764
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.050131199867052e-05,
+      "loss": 2.5474,
+      "step": 766
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.045084971874738e-05,
+      "loss": 2.5071,
+      "step": 768
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.040026789902665e-05,
+      "loss": 2.4774,
+      "step": 770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.034956668898706e-05,
+      "loss": 2.4119,
+      "step": 772
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.029874623846011e-05,
+      "loss": 2.4335,
+      "step": 774
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.02478066976297e-05,
+      "loss": 2.3666,
+      "step": 776
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.019674821703166e-05,
+      "loss": 2.5817,
+      "step": 778
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.014557094755331e-05,
+      "loss": 2.2798,
+      "step": 780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.009427504043305e-05,
+      "loss": 2.5065,
+      "step": 782
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.004286064725982e-05,
+      "loss": 2.4121,
+      "step": 784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.999132791997271e-05,
+      "loss": 2.5618,
+      "step": 786
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.993967701086057e-05,
+      "loss": 2.7772,
+      "step": 788
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.988790807256143e-05,
+      "loss": 2.3717,
+      "step": 790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.983602125806216e-05,
+      "loss": 2.5273,
+      "step": 792
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.978401672069797e-05,
+      "loss": 2.5246,
+      "step": 794
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.973189461415194e-05,
+      "loss": 2.5115,
+      "step": 796
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.967965509245461e-05,
+      "loss": 2.583,
+      "step": 798
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.962729830998353e-05,
+      "loss": 2.4989,
+      "step": 800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.957482442146272e-05,
+      "loss": 2.4011,
+      "step": 802
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.952223358196227e-05,
+      "loss": 2.424,
+      "step": 804
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.946952594689797e-05,
+      "loss": 2.5144,
+      "step": 806
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.941670167203067e-05,
+      "loss": 2.4956,
+      "step": 808
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.936376091346595e-05,
+      "loss": 2.5917,
+      "step": 810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.931070382765359e-05,
+      "loss": 2.3386,
+      "step": 812
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.925753057138719e-05,
+      "loss": 2.4911,
+      "step": 814
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.920424130180363e-05,
+      "loss": 2.5727,
+      "step": 816
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.915083617638262e-05,
+      "loss": 2.4148,
+      "step": 818
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.909731535294628e-05,
+      "loss": 2.4859,
+      "step": 820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.904367898965857e-05,
+      "loss": 2.4704,
+      "step": 822
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.898992724502498e-05,
+      "loss": 2.5904,
+      "step": 824
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.893606027789192e-05,
+      "loss": 2.5586,
+      "step": 826
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.888207824744629e-05,
+      "loss": 2.4723,
+      "step": 828
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.882798131321508e-05,
+      "loss": 2.5011,
+      "step": 830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.877376963506477e-05,
+      "loss": 2.6237,
+      "step": 832
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.871944337320102e-05,
+      "loss": 2.548,
+      "step": 834
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.866500268816803e-05,
+      "loss": 2.3527,
+      "step": 836
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.861044774084815e-05,
+      "loss": 2.6638,
+      "step": 838
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.855577869246142e-05,
+      "loss": 2.4873,
+      "step": 840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.850099570456509e-05,
+      "loss": 2.4461,
+      "step": 842
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.844609893905309e-05,
+      "loss": 2.4031,
+      "step": 844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.839108855815557e-05,
+      "loss": 2.5516,
+      "step": 846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.833596472443848e-05,
+      "loss": 2.4283,
+      "step": 848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.828072760080299e-05,
+      "loss": 2.2932,
+      "step": 850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.822537735048512e-05,
+      "loss": 2.3761,
+      "step": 852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.816991413705516e-05,
+      "loss": 2.4804,
+      "step": 854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.811433812441722e-05,
+      "loss": 2.6496,
+      "step": 856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.80586494768088e-05,
+      "loss": 2.4868,
+      "step": 858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.800284835880024e-05,
+      "loss": 2.679,
+      "step": 860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.79469349352942e-05,
+      "loss": 2.6229,
+      "step": 862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.78909093715253e-05,
+      "loss": 2.5438,
+      "step": 864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.783477183305949e-05,
+      "loss": 2.4863,
+      "step": 866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.777852248579367e-05,
+      "loss": 2.5205,
+      "step": 868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.772216149595513e-05,
+      "loss": 2.4179,
+      "step": 870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.766568903010113e-05,
+      "loss": 2.4653,
+      "step": 872
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.76091052551183e-05,
+      "loss": 2.3727,
+      "step": 874
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.755241033822224e-05,
+      "loss": 2.5503,
+      "step": 876
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.7495604446957e-05,
+      "loss": 2.4288,
+      "step": 878
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.743868774919458e-05,
+      "loss": 2.451,
+      "step": 880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.738166041313439e-05,
+      "loss": 2.3869,
+      "step": 882
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.732452260730286e-05,
+      "loss": 2.5419,
+      "step": 884
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.726727450055287e-05,
+      "loss": 2.3962,
+      "step": 886
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.720991626206321e-05,
+      "loss": 2.4672,
+      "step": 888
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.715244806133816e-05,
+      "loss": 2.4988,
+      "step": 890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.7094870068207e-05,
+      "loss": 2.2557,
+      "step": 892
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.703718245282337e-05,
+      "loss": 2.5007,
+      "step": 894
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.697938538566499e-05,
+      "loss": 2.4908,
+      "step": 896
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.69214790375329e-05,
+      "loss": 2.4346,
+      "step": 898
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.686346357955117e-05,
+      "loss": 2.2897,
+      "step": 900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.68053391831663e-05,
+      "loss": 2.3337,
+      "step": 902
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.674710602014671e-05,
+      "loss": 2.4618,
+      "step": 904
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.668876426258221e-05,
+      "loss": 2.5041,
+      "step": 906
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.66303140828836e-05,
+      "loss": 2.4207,
+      "step": 908
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.657175565378206e-05,
+      "loss": 2.4657,
+      "step": 910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.651308914832862e-05,
+      "loss": 2.5422,
+      "step": 912
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.645431473989376e-05,
+      "loss": 2.4069,
+      "step": 914
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.63954326021668e-05,
+      "loss": 2.6141,
+      "step": 916
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.633644290915545e-05,
+      "loss": 2.7452,
+      "step": 918
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.627734583518521e-05,
+      "loss": 2.4625,
+      "step": 920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.621814155489895e-05,
+      "loss": 2.3913,
+      "step": 922
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.615883024325636e-05,
+      "loss": 3.1472,
+      "step": 924
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.609941207553342e-05,
+      "loss": 2.4791,
+      "step": 926
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.603988722732186e-05,
+      "loss": 2.4555,
+      "step": 928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.598025587452873e-05,
+      "loss": 2.5092,
+      "step": 930
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.592051819337579e-05,
+      "loss": 2.5088,
+      "step": 932
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.586067436039899e-05,
+      "loss": 2.5663,
+      "step": 934
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.580072455244801e-05,
+      "loss": 2.5562,
+      "step": 936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.574066894668573e-05,
+      "loss": 2.4265,
+      "step": 938
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.568050772058762e-05,
+      "loss": 2.473,
+      "step": 940
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.562024105194133e-05,
+      "loss": 2.5223,
+      "step": 942
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.555986911884609e-05,
+      "loss": 2.3263,
+      "step": 944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.549939209971221e-05,
+      "loss": 2.2938,
+      "step": 946
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.543881017326057e-05,
+      "loss": 2.321,
+      "step": 948
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.537812351852201e-05,
+      "loss": 2.4323,
+      "step": 950
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.531733231483694e-05,
+      "loss": 2.365,
+      "step": 952
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.525643674185466e-05,
+      "loss": 2.4085,
+      "step": 954
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.519543697953296e-05,
+      "loss": 2.4288,
+      "step": 956
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.51343332081375e-05,
+      "loss": 2.6551,
+      "step": 958
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.50731256082413e-05,
+      "loss": 2.4887,
+      "step": 960
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.501181436072422e-05,
+      "loss": 2.6168,
+      "step": 962
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.495039964677241e-05,
+      "loss": 2.4247,
+      "step": 964
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.488888164787782e-05,
+      "loss": 2.5132,
+      "step": 966
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.482726054583761e-05,
+      "loss": 2.5011,
+      "step": 968
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.476553652275356e-05,
+      "loss": 2.4964,
+      "step": 970
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.47037097610317e-05,
+      "loss": 2.3202,
+      "step": 972
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.464178044338162e-05,
+      "loss": 2.2058,
+      "step": 974
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.4579748752816e-05,
+      "loss": 2.585,
+      "step": 976
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.451761487265003e-05,
+      "loss": 2.3743,
+      "step": 978
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.44553789865009e-05,
+      "loss": 2.2927,
+      "step": 980
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.439304127828728e-05,
+      "loss": 2.3899,
+      "step": 982
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.433060193222868e-05,
+      "loss": 2.462,
+      "step": 984
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.426806113284502e-05,
+      "loss": 2.4369,
+      "step": 986
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.420541906495599e-05,
+      "loss": 2.4967,
+      "step": 988
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.414267591368058e-05,
+      "loss": 2.6217,
+      "step": 990
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.407983186443653e-05,
+      "loss": 2.6545,
+      "step": 992
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.401688710293967e-05,
+      "loss": 2.4993,
+      "step": 994
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.395384181520351e-05,
+      "loss": 2.3227,
+      "step": 996
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.389069618753865e-05,
+      "loss": 2.325,
+      "step": 998
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.382745040655212e-05,
+      "loss": 2.6491,
+      "step": 1000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.376410465914705e-05,
+      "loss": 2.4874,
+      "step": 1002
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.370065913252188e-05,
+      "loss": 2.505,
+      "step": 1004
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.363711401417e-05,
+      "loss": 2.4867,
+      "step": 1006
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.357346949187906e-05,
+      "loss": 2.2378,
+      "step": 1008
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.350972575373047e-05,
+      "loss": 2.372,
+      "step": 1010
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.344588298809887e-05,
+      "loss": 2.3432,
+      "step": 1012
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.338194138365151e-05,
+      "loss": 2.6878,
+      "step": 1014
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.331790112934777e-05,
+      "loss": 2.4083,
+      "step": 1016
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.325376241443849e-05,
+      "loss": 2.4451,
+      "step": 1018
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.318952542846557e-05,
+      "loss": 2.3759,
+      "step": 1020
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.312519036126125e-05,
+      "loss": 2.5355,
+      "step": 1022
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.306075740294763e-05,
+      "loss": 2.4161,
+      "step": 1024
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.299622674393614e-05,
+      "loss": 2.3455,
+      "step": 1026
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.293159857492686e-05,
+      "loss": 2.469,
+      "step": 1028
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.28668730869081e-05,
+      "loss": 2.3113,
+      "step": 1030
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.280205047115572e-05,
+      "loss": 2.4072,
+      "step": 1032
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.273713091923264e-05,
+      "loss": 2.5218,
+      "step": 1034
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.267211462298822e-05,
+      "loss": 2.374,
+      "step": 1036
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.260700177455773e-05,
+      "loss": 2.452,
+      "step": 1038
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.254179256636179e-05,
+      "loss": 2.4523,
+      "step": 1040
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.247648719110572e-05,
+      "loss": 2.5231,
+      "step": 1042
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.241108584177911e-05,
+      "loss": 2.5678,
+      "step": 1044
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.234558871165512e-05,
+      "loss": 2.449,
+      "step": 1046
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.227999599428995e-05,
+      "loss": 2.3786,
+      "step": 1048
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.221430788352233e-05,
+      "loss": 2.3994,
+      "step": 1050
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.214852457347286e-05,
+      "loss": 2.5034,
+      "step": 1052
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.208264625854347e-05,
+      "loss": 2.2819,
+      "step": 1054
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.201667313341685e-05,
+      "loss": 2.4361,
+      "step": 1056
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.19506053930559e-05,
+      "loss": 2.3855,
+      "step": 1058
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.18844432327031e-05,
+      "loss": 2.4898,
+      "step": 1060
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.181818684787992e-05,
+      "loss": 2.5017,
+      "step": 1062
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.175183643438635e-05,
+      "loss": 2.402,
+      "step": 1064
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.168539218830024e-05,
+      "loss": 2.3225,
+      "step": 1066
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.16188543059767e-05,
+      "loss": 2.3171,
+      "step": 1068
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.155222298404756e-05,
+      "loss": 2.5654,
+      "step": 1070
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.148549841942082e-05,
+      "loss": 2.3448,
+      "step": 1072
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.141868080927996e-05,
+      "loss": 2.2422,
+      "step": 1074
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.135177035108352e-05,
+      "loss": 2.6608,
+      "step": 1076
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.128476724256431e-05,
+      "loss": 2.486,
+      "step": 1078
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.121767168172904e-05,
+      "loss": 2.3347,
+      "step": 1080
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.115048386685757e-05,
+      "loss": 2.4229,
+      "step": 1082
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.108320399650244e-05,
+      "loss": 2.6345,
+      "step": 1084
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.101583226948819e-05,
+      "loss": 2.5783,
+      "step": 1086
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.09483688849108e-05,
+      "loss": 2.4985,
+      "step": 1088
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.088081404213718e-05,
+      "loss": 2.3184,
+      "step": 1090
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.081316794080445e-05,
+      "loss": 2.324,
+      "step": 1092
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.074543078081946e-05,
+      "loss": 2.504,
+      "step": 1094
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.067760276235812e-05,
+      "loss": 2.3798,
+      "step": 1096
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.060968408586489e-05,
+      "loss": 2.4197,
+      "step": 1098
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.054167495205207e-05,
+      "loss": 2.4555,
+      "step": 1100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.047357556189936e-05,
+      "loss": 2.6626,
+      "step": 1102
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.040538611665314e-05,
+      "loss": 2.5664,
+      "step": 1104
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.033710681782592e-05,
+      "loss": 2.4436,
+      "step": 1106
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.026873786719573e-05,
+      "loss": 2.5044,
+      "step": 1108
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.02002794668056e-05,
+      "loss": 2.488,
+      "step": 1110
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.013173181896283e-05,
+      "loss": 2.4565,
+      "step": 1112
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.006309512623848e-05,
+      "loss": 2.5484,
+      "step": 1114
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.99943695914668e-05,
+      "loss": 2.4438,
+      "step": 1116
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.992555541774452e-05,
+      "loss": 2.4668,
+      "step": 1118
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.985665280843035e-05,
+      "loss": 2.5129,
+      "step": 1120
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.978766196714436e-05,
+      "loss": 2.2599,
+      "step": 1122
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.97185830977673e-05,
+      "loss": 2.4388,
+      "step": 1124
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.964941640444014e-05,
+      "loss": 2.5566,
+      "step": 1126
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.958016209156331e-05,
+      "loss": 2.3852,
+      "step": 1128
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.951082036379625e-05,
+      "loss": 2.3447,
+      "step": 1130
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.944139142605665e-05,
+      "loss": 2.471,
+      "step": 1132
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.937187548351996e-05,
+      "loss": 2.4846,
+      "step": 1134
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.930227274161877e-05,
+      "loss": 2.433,
+      "step": 1136
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.923258340604212e-05,
+      "loss": 2.7046,
+      "step": 1138
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.916280768273498e-05,
+      "loss": 2.2928,
+      "step": 1140
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.909294577789766e-05,
+      "loss": 2.5962,
+      "step": 1142
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.902299789798505e-05,
+      "loss": 2.4707,
+      "step": 1144
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.895296424970618e-05,
+      "loss": 2.4212,
+      "step": 1146
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.888284504002352e-05,
+      "loss": 2.5168,
+      "step": 1148
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.881264047615245e-05,
+      "loss": 2.5038,
+      "step": 1150
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.874235076556046e-05,
+      "loss": 2.2647,
+      "step": 1152
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.867197611596683e-05,
+      "loss": 2.5225,
+      "step": 1154
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.860151673534168e-05,
+      "loss": 2.3552,
+      "step": 1156
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.853097283190567e-05,
+      "loss": 2.5299,
+      "step": 1158
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.846034461412912e-05,
+      "loss": 2.476,
+      "step": 1160
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.838963229073162e-05,
+      "loss": 2.3523,
+      "step": 1162
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.831883607068125e-05,
+      "loss": 2.4746,
+      "step": 1164
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.824795616319402e-05,
+      "loss": 2.4551,
+      "step": 1166
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.817699277773325e-05,
+      "loss": 2.4863,
+      "step": 1168
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.810594612400898e-05,
+      "loss": 2.5789,
+      "step": 1170
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.803481641197733e-05,
+      "loss": 2.487,
+      "step": 1172
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.796360385183984e-05,
+      "loss": 2.5997,
+      "step": 1174
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.789230865404287e-05,
+      "loss": 2.3587,
+      "step": 1176
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.782093102927703e-05,
+      "loss": 2.7109,
+      "step": 1178
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.77494711884765e-05,
+      "loss": 2.5783,
+      "step": 1180
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.767792934281843e-05,
+      "loss": 2.4947,
+      "step": 1182
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.76063057037223e-05,
+      "loss": 2.3812,
+      "step": 1184
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.753460048284928e-05,
+      "loss": 2.3337,
+      "step": 1186
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.74628138921017e-05,
+      "loss": 2.5691,
+      "step": 1188
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.739094614362229e-05,
+      "loss": 2.4811,
+      "step": 1190
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.731899744979364e-05,
+      "loss": 2.618,
+      "step": 1192
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.724696802323755e-05,
+      "loss": 2.2892,
+      "step": 1194
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.717485807681437e-05,
+      "loss": 2.3032,
+      "step": 1196
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.710266782362247e-05,
+      "loss": 2.4592,
+      "step": 1198
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.703039747699747e-05,
+      "loss": 2.3496,
+      "step": 1200
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.695804725051172e-05,
+      "loss": 2.423,
+      "step": 1202
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.68856173579736e-05,
+      "loss": 2.4122,
+      "step": 1204
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.681310801342696e-05,
+      "loss": 2.3985,
+      "step": 1206
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.674051943115042e-05,
+      "loss": 2.2799,
+      "step": 1208
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.666785182565677e-05,
+      "loss": 2.3947,
+      "step": 1210
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.65951054116923e-05,
+      "loss": 2.3299,
+      "step": 1212
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.652228040423622e-05,
+      "loss": 2.274,
+      "step": 1214
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.644937701850002e-05,
+      "loss": 2.3697,
+      "step": 1216
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.637639546992677e-05,
+      "loss": 2.3167,
+      "step": 1218
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.630333597419054e-05,
+      "loss": 2.4688,
+      "step": 1220
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.623019874719579e-05,
+      "loss": 2.2979,
+      "step": 1222
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.61569840050766e-05,
+      "loss": 2.4614,
+      "step": 1224
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.60836919641962e-05,
+      "loss": 2.5093,
+      "step": 1226
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.60103228411462e-05,
+      "loss": 2.4832,
+      "step": 1228
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.593687685274609e-05,
+      "loss": 2.4112,
+      "step": 1230
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.586335421604238e-05,
+      "loss": 2.3033,
+      "step": 1232
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.578975514830821e-05,
+      "loss": 2.6554,
+      "step": 1234
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.571607986704252e-05,
+      "loss": 2.3495,
+      "step": 1236
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.564232858996949e-05,
+      "loss": 2.4517,
+      "step": 1238
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.556850153503787e-05,
+      "loss": 2.4985,
+      "step": 1240
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.549459892042041e-05,
+      "loss": 2.5046,
+      "step": 1242
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.542062096451305e-05,
+      "loss": 2.5004,
+      "step": 1244
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.534656788593446e-05,
+      "loss": 2.3215,
+      "step": 1246
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.527243990352529e-05,
+      "loss": 2.5481,
+      "step": 1248
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.519823723634753e-05,
+      "loss": 2.3608,
+      "step": 1250
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.51239601036839e-05,
+      "loss": 2.2113,
+      "step": 1252
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.504960872503715e-05,
+      "loss": 2.6318,
+      "step": 1254
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.497518332012946e-05,
+      "loss": 2.3967,
+      "step": 1256
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.490068410890175e-05,
+      "loss": 2.1024,
+      "step": 1258
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.48261113115131e-05,
+      "loss": 2.5322,
+      "step": 1260
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.475146514834001e-05,
+      "loss": 2.3737,
+      "step": 1262
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.46767458399758e-05,
+      "loss": 2.4803,
+      "step": 1264
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.460195360722995e-05,
+      "loss": 2.1737,
+      "step": 1266
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.452708867112745e-05,
+      "loss": 2.5601,
+      "step": 1268
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.44521512529081e-05,
+      "loss": 2.5452,
+      "step": 1270
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.437714157402598e-05,
+      "loss": 2.3953,
+      "step": 1272
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.430205985614864e-05,
+      "loss": 2.4914,
+      "step": 1274
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.422690632115654e-05,
+      "loss": 2.3997,
+      "step": 1276
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.41516811911424e-05,
+      "loss": 2.2561,
+      "step": 1278
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.407638468841047e-05,
+      "loss": 2.6531,
+      "step": 1280
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.400101703547597e-05,
+      "loss": 2.6299,
+      "step": 1282
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.392557845506432e-05,
+      "loss": 2.4573,
+      "step": 1284
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.385006917011063e-05,
+      "loss": 2.5633,
+      "step": 1286
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.377448940375887e-05,
+      "loss": 2.6371,
+      "step": 1288
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.369883937936136e-05,
+      "loss": 2.2814,
+      "step": 1290
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.362311932047797e-05,
+      "loss": 2.6985,
+      "step": 1292
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.354732945087563e-05,
+      "loss": 2.3274,
+      "step": 1294
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.34714699945275e-05,
+      "loss": 2.2417,
+      "step": 1296
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.33955411756124e-05,
+      "loss": 2.4285,
+      "step": 1298
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.331954321851418e-05,
+      "loss": 2.4677,
+      "step": 1300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.32434763478209e-05,
+      "loss": 2.4342,
+      "step": 1302
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.316734078832438e-05,
+      "loss": 2.3903,
+      "step": 1304
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.309113676501939e-05,
+      "loss": 2.4379,
+      "step": 1306
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.301486450310298e-05,
+      "loss": 2.4929,
+      "step": 1308
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.293852422797391e-05,
+      "loss": 2.4626,
+      "step": 1310
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.286211616523193e-05,
+      "loss": 2.5199,
+      "step": 1312
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.278564054067709e-05,
+      "loss": 2.3659,
+      "step": 1314
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.270909758030912e-05,
+      "loss": 2.4869,
+      "step": 1316
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.263248751032671e-05,
+      "loss": 2.5166,
+      "step": 1318
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.255581055712688e-05,
+      "loss": 2.139,
+      "step": 1320
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.247906694730437e-05,
+      "loss": 2.4807,
+      "step": 1322
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.24022569076508e-05,
+      "loss": 2.4607,
+      "step": 1324
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.232538066515414e-05,
+      "loss": 2.3367,
+      "step": 1326
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.224843844699803e-05,
+      "loss": 2.6005,
+      "step": 1328
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.217143048056108e-05,
+      "loss": 2.3467,
+      "step": 1330
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.209435699341613e-05,
+      "loss": 2.4132,
+      "step": 1332
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.201721821332973e-05,
+      "loss": 2.3049,
+      "step": 1334
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.194001436826135e-05,
+      "loss": 2.3176,
+      "step": 1336
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.18627456863627e-05,
+      "loss": 2.5401,
+      "step": 1338
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.178541239597717e-05,
+      "loss": 2.4131,
+      "step": 1340
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.170801472563903e-05,
+      "loss": 2.4554,
+      "step": 1342
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.163055290407282e-05,
+      "loss": 2.405,
+      "step": 1344
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.155302716019263e-05,
+      "loss": 2.4435,
+      "step": 1346
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.14754377231015e-05,
+      "loss": 2.4068,
+      "step": 1348
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.139778482209068e-05,
+      "loss": 2.4863,
+      "step": 1350
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.132006868663894e-05,
+      "loss": 2.3856,
+      "step": 1352
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.124228954641196e-05,
+      "loss": 2.3076,
+      "step": 1354
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.116444763126158e-05,
+      "loss": 2.334,
+      "step": 1356
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.108654317122515e-05,
+      "loss": 2.3639,
+      "step": 1358
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.100857639652489e-05,
+      "loss": 2.7099,
+      "step": 1360
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.093054753756713e-05,
+      "loss": 2.6381,
+      "step": 1362
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.085245682494168e-05,
+      "loss": 2.4935,
+      "step": 1364
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.077430448942117e-05,
+      "loss": 2.3986,
+      "step": 1366
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.069609076196029e-05,
+      "loss": 2.5647,
+      "step": 1368
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.061781587369519e-05,
+      "loss": 2.34,
+      "step": 1370
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.053948005594273e-05,
+      "loss": 2.5114,
+      "step": 1372
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.046108354019987e-05,
+      "loss": 2.4023,
+      "step": 1374
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.038262655814291e-05,
+      "loss": 2.5273,
+      "step": 1376
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.030410934162684e-05,
+      "loss": 2.3996,
+      "step": 1378
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.022553212268469e-05,
+      "loss": 2.5281,
+      "step": 1380
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.014689513352675e-05,
+      "loss": 2.2748,
+      "step": 1382
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.006819860654001e-05,
+      "loss": 2.6029,
+      "step": 1384
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.998944277428734e-05,
+      "loss": 2.3091,
+      "step": 1386
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.991062786950691e-05,
+      "loss": 2.4689,
+      "step": 1388
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.983175412511145e-05,
+      "loss": 2.4879,
+      "step": 1390
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.975282177418756e-05,
+      "loss": 2.5369,
+      "step": 1392
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.967383104999505e-05,
+      "loss": 2.4159,
+      "step": 1394
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.959478218596625e-05,
+      "loss": 2.5748,
+      "step": 1396
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.951567541570523e-05,
+      "loss": 2.5503,
+      "step": 1398
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.943651097298727e-05,
+      "loss": 2.5094,
+      "step": 1400
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.935728909175805e-05,
+      "loss": 2.3161,
+      "step": 1402
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.927801000613298e-05,
+      "loss": 2.3559,
+      "step": 1404
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.919867395039652e-05,
+      "loss": 2.4446,
+      "step": 1406
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.91192811590015e-05,
+      "loss": 2.3579,
+      "step": 1408
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.903983186656844e-05,
+      "loss": 2.3263,
+      "step": 1410
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.896032630788476e-05,
+      "loss": 2.4279,
+      "step": 1412
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.888076471790424e-05,
+      "loss": 2.3288,
+      "step": 1414
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.880114733174615e-05,
+      "loss": 2.3714,
+      "step": 1416
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.872147438469476e-05,
+      "loss": 2.5845,
+      "step": 1418
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.864174611219841e-05,
+      "loss": 2.2575,
+      "step": 1420
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.856196274986907e-05,
+      "loss": 2.7716,
+      "step": 1422
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.848212453348137e-05,
+      "loss": 2.408,
+      "step": 1424
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.840223169897217e-05,
+      "loss": 2.5191,
+      "step": 1426
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.832228448243964e-05,
+      "loss": 2.3474,
+      "step": 1428
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.824228312014274e-05,
+      "loss": 2.5852,
+      "step": 1430
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.816222784850038e-05,
+      "loss": 2.5364,
+      "step": 1432
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.80821189040908e-05,
+      "loss": 2.26,
+      "step": 1434
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.800195652365087e-05,
+      "loss": 2.4253,
+      "step": 1436
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.792174094407533e-05,
+      "loss": 2.3855,
+      "step": 1438
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.784147240241619e-05,
+      "loss": 2.2678,
+      "step": 1440
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.776115113588194e-05,
+      "loss": 2.4646,
+      "step": 1442
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.76807773818369e-05,
+      "loss": 2.6316,
+      "step": 1444
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.760035137780046e-05,
+      "loss": 2.5357,
+      "step": 1446
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.751987336144648e-05,
+      "loss": 2.5943,
+      "step": 1448
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.743934357060246e-05,
+      "loss": 2.4468,
+      "step": 1450
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.735876224324895e-05,
+      "loss": 2.3678,
+      "step": 1452
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.72781296175188e-05,
+      "loss": 2.3095,
+      "step": 1454
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.719744593169641e-05,
+      "loss": 2.4335,
+      "step": 1456
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.711671142421714e-05,
+      "loss": 2.5255,
+      "step": 1458
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.703592633366647e-05,
+      "loss": 2.3837,
+      "step": 1460
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.695509089877943e-05,
+      "loss": 2.5474,
+      "step": 1462
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.687420535843975e-05,
+      "loss": 2.5055,
+      "step": 1464
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.679326995167932e-05,
+      "loss": 2.4212,
+      "step": 1466
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.671228491767728e-05,
+      "loss": 2.4475,
+      "step": 1468
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.663125049575956e-05,
+      "loss": 2.5926,
+      "step": 1470
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.655016692539793e-05,
+      "loss": 2.2358,
+      "step": 1472
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.646903444620949e-05,
+      "loss": 2.6164,
+      "step": 1474
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.63878532979558e-05,
+      "loss": 2.5778,
+      "step": 1476
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.630662372054227e-05,
+      "loss": 2.4401,
+      "step": 1478
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.622534595401746e-05,
+      "loss": 2.4327,
+      "step": 1480
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.614402023857232e-05,
+      "loss": 2.3336,
+      "step": 1482
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.606264681453946e-05,
+      "loss": 2.4107,
+      "step": 1484
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.598122592239255e-05,
+      "loss": 2.4793,
+      "step": 1486
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.589975780274544e-05,
+      "loss": 2.6092,
+      "step": 1488
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.581824269635166e-05,
+      "loss": 2.4823,
+      "step": 1490
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.57366808441035e-05,
+      "loss": 2.3623,
+      "step": 1492
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.565507248703144e-05,
+      "loss": 2.5841,
+      "step": 1494
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.557341786630339e-05,
+      "loss": 2.3636,
+      "step": 1496
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.549171722322395e-05,
+      "loss": 2.5033,
+      "step": 1498
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.540997079923376e-05,
+      "loss": 2.5465,
+      "step": 1500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.532817883590874e-05,
+      "loss": 2.4308,
+      "step": 1502
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.524634157495935e-05,
+      "loss": 2.6063,
+      "step": 1504
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.516445925822997e-05,
+      "loss": 2.3648,
+      "step": 1506
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.508253212769808e-05,
+      "loss": 2.5649,
+      "step": 1508
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.500056042547364e-05,
+      "loss": 2.4303,
+      "step": 1510
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.491854439379827e-05,
+      "loss": 2.2518,
+      "step": 1512
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.483648427504467e-05,
+      "loss": 2.6185,
+      "step": 1514
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.475438031171574e-05,
+      "loss": 2.4631,
+      "step": 1516
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.4672232746444e-05,
+      "loss": 2.5055,
+      "step": 1518
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.459004182199082e-05,
+      "loss": 2.4789,
+      "step": 1520
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.45078077812457e-05,
+      "loss": 2.518,
+      "step": 1522
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.442553086722554e-05,
+      "loss": 2.2487,
+      "step": 1524
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.434321132307394e-05,
+      "loss": 2.4873,
+      "step": 1526
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.426084939206051e-05,
+      "loss": 2.4427,
+      "step": 1528
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.417844531758009e-05,
+      "loss": 2.5523,
+      "step": 1530
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.40959993431521e-05,
+      "loss": 2.4331,
+      "step": 1532
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.401351171241971e-05,
+      "loss": 2.2483,
+      "step": 1534
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.393098266914925e-05,
+      "loss": 2.3769,
+      "step": 1536
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.384841245722945e-05,
+      "loss": 2.4459,
+      "step": 1538
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.376580132067065e-05,
+      "loss": 2.4104,
+      "step": 1540
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.368314950360415e-05,
+      "loss": 2.3963,
+      "step": 1542
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.360045725028146e-05,
+      "loss": 2.4358,
+      "step": 1544
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.351772480507363e-05,
+      "loss": 2.3851,
+      "step": 1546
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.34349524124704e-05,
+      "loss": 2.3434,
+      "step": 1548
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.335214031707965e-05,
+      "loss": 2.3168,
+      "step": 1550
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.326928876362652e-05,
+      "loss": 2.5622,
+      "step": 1552
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.318639799695285e-05,
+      "loss": 2.4061,
+      "step": 1554
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.310346826201621e-05,
+      "loss": 2.6289,
+      "step": 1556
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.302049980388948e-05,
+      "loss": 2.4561,
+      "step": 1558
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.29374928677599e-05,
+      "loss": 2.4697,
+      "step": 1560
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.28544476989284e-05,
+      "loss": 2.4481,
+      "step": 1562
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.277136454280898e-05,
+      "loss": 2.5529,
+      "step": 1564
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.268824364492782e-05,
+      "loss": 2.4358,
+      "step": 1566
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.260508525092266e-05,
+      "loss": 2.3754,
+      "step": 1568
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.252188960654204e-05,
+      "loss": 2.5845,
+      "step": 1570
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.243865695764459e-05,
+      "loss": 2.5552,
+      "step": 1572
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.235538755019832e-05,
+      "loss": 2.4616,
+      "step": 1574
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.227208163027982e-05,
+      "loss": 2.3196,
+      "step": 1576
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.218873944407361e-05,
+      "loss": 2.4119,
+      "step": 1578
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.210536123787138e-05,
+      "loss": 2.2707,
+      "step": 1580
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.202194725807127e-05,
+      "loss": 2.7299,
+      "step": 1582
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.19384977511771e-05,
+      "loss": 2.2659,
+      "step": 1584
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.185501296379777e-05,
+      "loss": 2.5439,
+      "step": 1586
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.177149314264631e-05,
+      "loss": 2.6154,
+      "step": 1588
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.168793853453943e-05,
+      "loss": 2.5537,
+      "step": 1590
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.160434938639648e-05,
+      "loss": 2.3475,
+      "step": 1592
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.152072594523906e-05,
+      "loss": 2.3385,
+      "step": 1594
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.143706845818992e-05,
+      "loss": 2.4313,
+      "step": 1596
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.135337717247261e-05,
+      "loss": 2.3323,
+      "step": 1598
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.12696523354104e-05,
+      "loss": 2.4587,
+      "step": 1600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.118589419442584e-05,
+      "loss": 2.6458,
+      "step": 1602
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.110210299703982e-05,
+      "loss": 2.5148,
+      "step": 1604
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.101827899087094e-05,
+      "loss": 2.5324,
+      "step": 1606
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.0934422423634744e-05,
+      "loss": 2.4962,
+      "step": 1608
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.085053354314302e-05,
+      "loss": 2.4868,
+      "step": 1610
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.076661259730305e-05,
+      "loss": 2.506,
+      "step": 1612
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.068265983411685e-05,
+      "loss": 2.3774,
+      "step": 1614
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.05986755016805e-05,
+      "loss": 2.4287,
+      "step": 1616
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.051465984818332e-05,
+      "loss": 2.3267,
+      "step": 1618
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.043061312190723e-05,
+      "loss": 2.4453,
+      "step": 1620
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.034653557122598e-05,
+      "loss": 2.4022,
+      "step": 1622
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.0262427444604384e-05,
+      "loss": 2.5554,
+      "step": 1624
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.017828899059763e-05,
+      "loss": 2.4102,
+      "step": 1626
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.009412045785051e-05,
+      "loss": 2.3266,
+      "step": 1628
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.000992209509676e-05,
+      "loss": 2.2382,
+      "step": 1630
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.9925694151158184e-05,
+      "loss": 2.2763,
+      "step": 1632
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.984143687494409e-05,
+      "loss": 2.481,
+      "step": 1634
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.975715051545039e-05,
+      "loss": 2.5598,
+      "step": 1636
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.9672835321759016e-05,
+      "loss": 2.3289,
+      "step": 1638
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.958849154303704e-05,
+      "loss": 2.4317,
+      "step": 1640
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.9504119428536076e-05,
+      "loss": 2.448,
+      "step": 1642
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.9419719227591405e-05,
+      "loss": 2.2034,
+      "step": 1644
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.933529118962138e-05,
+      "loss": 2.4841,
+      "step": 1646
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.925083556412657e-05,
+      "loss": 2.5998,
+      "step": 1648
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.916635260068909e-05,
+      "loss": 2.5288,
+      "step": 1650
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.908184254897182e-05,
+      "loss": 2.5148,
+      "step": 1652
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.899730565871774e-05,
+      "loss": 2.5166,
+      "step": 1654
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.891274217974907e-05,
+      "loss": 2.4235,
+      "step": 1656
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.8828152361966685e-05,
+      "loss": 2.5575,
+      "step": 1658
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.874353645534922e-05,
+      "loss": 2.4232,
+      "step": 1660
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.865889470995248e-05,
+      "loss": 2.2509,
+      "step": 1662
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.857422737590857e-05,
+      "loss": 2.2636,
+      "step": 1664
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.8489534703425256e-05,
+      "loss": 2.4923,
+      "step": 1666
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.8404816942785134e-05,
+      "loss": 2.3899,
+      "step": 1668
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.8320074344345e-05,
+      "loss": 2.4698,
+      "step": 1670
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.8235307158535e-05,
+      "loss": 2.65,
+      "step": 1672
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.8150515635858e-05,
+      "loss": 2.4687,
+      "step": 1674
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.806570002688869e-05,
+      "loss": 2.4793,
+      "step": 1676
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.798086058227304e-05,
+      "loss": 2.2238,
+      "step": 1678
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.78959975527274e-05,
+      "loss": 2.4365,
+      "step": 1680
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.781111118903785e-05,
+      "loss": 2.4891,
+      "step": 1682
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.772620174205938e-05,
+      "loss": 2.3248,
+      "step": 1684
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.764126946271526e-05,
+      "loss": 2.6325,
+      "step": 1686
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.755631460199616e-05,
+      "loss": 2.4281,
+      "step": 1688
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.747133741095956e-05,
+      "loss": 2.4829,
+      "step": 1690
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.738633814072888e-05,
+      "loss": 2.3321,
+      "step": 1692
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.730131704249278e-05,
+      "loss": 2.5413,
+      "step": 1694
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.721627436750449e-05,
+      "loss": 2.0978,
+      "step": 1696
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.713121036708091e-05,
+      "loss": 2.5072,
+      "step": 1698
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.704612529260205e-05,
+      "loss": 2.4096,
+      "step": 1700
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.6961019395510126e-05,
+      "loss": 2.3686,
+      "step": 1702
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.6875892927308936e-05,
+      "loss": 2.475,
+      "step": 1704
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.679074613956307e-05,
+      "loss": 2.375,
+      "step": 1706
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.6705579283897116e-05,
+      "loss": 2.5238,
+      "step": 1708
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.662039261199502e-05,
+      "loss": 2.5842,
+      "step": 1710
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.6535186375599266e-05,
+      "loss": 2.5468,
+      "step": 1712
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.644996082651017e-05,
+      "loss": 2.4626,
+      "step": 1714
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.636471621658508e-05,
+      "loss": 2.5552,
+      "step": 1716
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.627945279773774e-05,
+      "loss": 2.2431,
+      "step": 1718
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.61941708219374e-05,
+      "loss": 2.36,
+      "step": 1720
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.6108870541208224e-05,
+      "loss": 2.3865,
+      "step": 1722
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.602355220762838e-05,
+      "loss": 2.5472,
+      "step": 1724
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.593821607332952e-05,
+      "loss": 2.3935,
+      "step": 1726
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.585286239049574e-05,
+      "loss": 2.5526,
+      "step": 1728
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.576749141136313e-05,
+      "loss": 2.5119,
+      "step": 1730
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.568210338821881e-05,
+      "loss": 2.3868,
+      "step": 1732
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.5596698573400306e-05,
+      "loss": 2.4324,
+      "step": 1734
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.5511277219294765e-05,
+      "loss": 2.4088,
+      "step": 1736
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.54258395783382e-05,
+      "loss": 2.447,
+      "step": 1738
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.534038590301476e-05,
+      "loss": 2.4857,
+      "step": 1740
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.5254916445855974e-05,
+      "loss": 2.3698,
+      "step": 1742
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.5169431459440014e-05,
+      "loss": 2.5048,
+      "step": 1744
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.508393119639094e-05,
+      "loss": 2.5057,
+      "step": 1746
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.499841590937795e-05,
+      "loss": 2.4211,
+      "step": 1748
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.491288585111467e-05,
+      "loss": 2.7328,
+      "step": 1750
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.4827341274358344e-05,
+      "loss": 2.5598,
+      "step": 1752
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.4741782431909136e-05,
+      "loss": 2.2472,
+      "step": 1754
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.465620957660938e-05,
+      "loss": 2.4122,
+      "step": 1756
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.457062296134279e-05,
+      "loss": 2.4685,
+      "step": 1758
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.448502283903377e-05,
+      "loss": 2.5201,
+      "step": 1760
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.439940946264662e-05,
+      "loss": 2.5483,
+      "step": 1762
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.4313783085184825e-05,
+      "loss": 2.4956,
+      "step": 1764
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.422814395969029e-05,
+      "loss": 2.5378,
+      "step": 1766
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.414249233924258e-05,
+      "loss": 2.218,
+      "step": 1768
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.40568284769582e-05,
+      "loss": 2.5364,
+      "step": 1770
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.39711526259898e-05,
+      "loss": 2.4421,
+      "step": 1772
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.388546503952551e-05,
+      "loss": 2.3293,
+      "step": 1774
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.379976597078808e-05,
+      "loss": 2.4037,
+      "step": 1776
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.371405567303428e-05,
+      "loss": 2.4786,
+      "step": 1778
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.362833439955396e-05,
+      "loss": 2.2926,
+      "step": 1780
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.354260240366947e-05,
+      "loss": 2.5112,
+      "step": 1782
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.3456859938734836e-05,
+      "loss": 2.2066,
+      "step": 1784
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.337110725813501e-05,
+      "loss": 2.4357,
+      "step": 1786
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.328534461528515e-05,
+      "loss": 2.502,
+      "step": 1788
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3199572263629824e-05,
+      "loss": 2.2781,
+      "step": 1790
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3113790456642345e-05,
+      "loss": 2.3274,
+      "step": 1792
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3027999447823905e-05,
+      "loss": 2.4531,
+      "step": 1794
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2942199490702924e-05,
+      "loss": 2.4264,
+      "step": 1796
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.285639083883428e-05,
+      "loss": 2.3976,
+      "step": 1798
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.27705737457985e-05,
+      "loss": 2.3159,
+      "step": 1800
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.268474846520112e-05,
+      "loss": 2.3113,
+      "step": 1802
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.259891525067179e-05,
+      "loss": 2.3999,
+      "step": 1804
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.251307435586368e-05,
+      "loss": 2.4202,
+      "step": 1806
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.2427226034452614e-05,
+      "loss": 2.4635,
+      "step": 1808
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.23413705401364e-05,
+      "loss": 2.588,
+      "step": 1810
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.225550812663399e-05,
+      "loss": 2.3913,
+      "step": 1812
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.216963904768485e-05,
+      "loss": 2.3559,
+      "step": 1814
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.2083763557048056e-05,
+      "loss": 2.3511,
+      "step": 1816
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.1997881908501736e-05,
+      "loss": 2.0888,
+      "step": 1818
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.191199435584211e-05,
+      "loss": 2.3658,
+      "step": 1820
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.182610115288295e-05,
+      "loss": 2.4578,
+      "step": 1822
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.174020255345464e-05,
+      "loss": 2.5353,
+      "step": 1824
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.1654298811403556e-05,
+      "loss": 2.3506,
+      "step": 1826
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.1568390180591265e-05,
+      "loss": 2.43,
+      "step": 1828
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.148247691489377e-05,
+      "loss": 2.5092,
+      "step": 1830
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.139655926820078e-05,
+      "loss": 2.4586,
+      "step": 1832
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.131063749441496e-05,
+      "loss": 2.3623,
+      "step": 1834
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.1224711847451145e-05,
+      "loss": 2.5055,
+      "step": 1836
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.113878258123563e-05,
+      "loss": 2.3857,
+      "step": 1838
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.105284994970543e-05,
+      "loss": 2.6249,
+      "step": 1840
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.096691420680745e-05,
+      "loss": 2.408,
+      "step": 1842
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.088097560649784e-05,
+      "loss": 2.5748,
+      "step": 1844
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.0795034402741185e-05,
+      "loss": 2.2292,
+      "step": 1846
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.06661182712092e-05,
+      "loss": 3.0448,
+      "step": 1848
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.0580171669978546e-05,
+      "loss": 2.4891,
+      "step": 1850
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.049422335423252e-05,
+      "loss": 2.5769,
+      "step": 1852
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.04082735779644e-05,
+      "loss": 2.3243,
+      "step": 1854
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.032232259517179e-05,
+      "loss": 2.5441,
+      "step": 1856
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.023637065985585e-05,
+      "loss": 2.4429,
+      "step": 1858
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.015041802602057e-05,
+      "loss": 2.3943,
+      "step": 1860
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0064464947672e-05,
+      "loss": 2.3169,
+      "step": 1862
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.9978511678817496e-05,
+      "loss": 2.4604,
+      "step": 1864
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.989255847346499e-05,
+      "loss": 2.3745,
+      "step": 1866
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.980660558562222e-05,
+      "loss": 2.3082,
+      "step": 1868
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.972065326929598e-05,
+      "loss": 2.4983,
+      "step": 1870
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.963470177849135e-05,
+      "loss": 2.3494,
+      "step": 1872
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.954875136721104e-05,
+      "loss": 2.2882,
+      "step": 1874
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.946280228945453e-05,
+      "loss": 2.401,
+      "step": 1876
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9376854799217327e-05,
+      "loss": 2.3044,
+      "step": 1878
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.929090915049029e-05,
+      "loss": 2.51,
+      "step": 1880
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.920496559725883e-05,
+      "loss": 2.5332,
+      "step": 1882
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.911902439350217e-05,
+      "loss": 2.449,
+      "step": 1884
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9033085793192574e-05,
+      "loss": 2.4766,
+      "step": 1886
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.894715005029459e-05,
+      "loss": 2.5255,
+      "step": 1888
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8861217418764374e-05,
+      "loss": 2.4169,
+      "step": 1890
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8775288152548866e-05,
+      "loss": 2.4542,
+      "step": 1892
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.868936250558506e-05,
+      "loss": 2.3703,
+      "step": 1894
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8603440731799216e-05,
+      "loss": 2.4712,
+      "step": 1896
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.851752308510624e-05,
+      "loss": 2.3871,
+      "step": 1898
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.843160981940875e-05,
+      "loss": 2.3593,
+      "step": 1900
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.8345701188596456e-05,
+      "loss": 2.3834,
+      "step": 1902
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.825979744654536e-05,
+      "loss": 2.5184,
+      "step": 1904
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.817389884711705e-05,
+      "loss": 2.3947,
+      "step": 1906
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.8088005644157895e-05,
+      "loss": 2.3947,
+      "step": 1908
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.800211809149829e-05,
+      "loss": 2.332,
+      "step": 1910
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.791623644295195e-05,
+      "loss": 2.4736,
+      "step": 1912
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.7830360952315164e-05,
+      "loss": 2.3257,
+      "step": 1914
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.774449187336602e-05,
+      "loss": 2.4029,
+      "step": 1916
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.765862945986362e-05,
+      "loss": 2.2457,
+      "step": 1918
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.7572773965547384e-05,
+      "loss": 2.4841,
+      "step": 1920
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.7486925644136324e-05,
+      "loss": 2.2552,
+      "step": 1922
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.740108474932822e-05,
+      "loss": 2.2952,
+      "step": 1924
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.731525153479891e-05,
+      "loss": 2.6216,
+      "step": 1926
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.72294262542015e-05,
+      "loss": 2.6685,
+      "step": 1928
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.7143609161165736e-05,
+      "loss": 2.5377,
+      "step": 1930
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.705780050929708e-05,
+      "loss": 2.3924,
+      "step": 1932
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.697200055217612e-05,
+      "loss": 2.3375,
+      "step": 1934
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.688620954335766e-05,
+      "loss": 2.5853,
+      "step": 1936
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.680042773637018e-05,
+      "loss": 2.4731,
+      "step": 1938
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.671465538471486e-05,
+      "loss": 2.3143,
+      "step": 1940
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6628892741865e-05,
+      "loss": 2.3821,
+      "step": 1942
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.654314006126516e-05,
+      "loss": 2.5513,
+      "step": 1944
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.645739759633054e-05,
+      "loss": 2.3581,
+      "step": 1946
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.637166560044605e-05,
+      "loss": 2.3677,
+      "step": 1948
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.628594432696573e-05,
+      "loss": 2.2615,
+      "step": 1950
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.620023402921191e-05,
+      "loss": 2.4965,
+      "step": 1952
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.61145349604745e-05,
+      "loss": 2.3364,
+      "step": 1954
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.602884737401022e-05,
+      "loss": 2.1789,
+      "step": 1956
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.594317152304183e-05,
+      "loss": 2.229,
+      "step": 1958
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.5857507660757424e-05,
+      "loss": 2.7305,
+      "step": 1960
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.5771856040309716e-05,
+      "loss": 2.4481,
+      "step": 1962
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.568621691481519e-05,
+      "loss": 2.5205,
+      "step": 1964
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.5600590537353397e-05,
+      "loss": 2.4775,
+      "step": 1966
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.551497716096624e-05,
+      "loss": 2.5465,
+      "step": 1968
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.5429377038657214e-05,
+      "loss": 2.4793,
+      "step": 1970
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.534379042339063e-05,
+      "loss": 2.5299,
+      "step": 1972
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.5258217568090876e-05,
+      "loss": 2.4299,
+      "step": 1974
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.517265872564167e-05,
+      "loss": 2.4201,
+      "step": 1976
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.508711414888534e-05,
+      "loss": 2.3312,
+      "step": 1978
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.5001584090622065e-05,
+      "loss": 2.3622,
+      "step": 1980
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.491606880360909e-05,
+      "loss": 2.3231,
+      "step": 1982
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.483056854055999e-05,
+      "loss": 2.3689,
+      "step": 1984
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.474508355414404e-05,
+      "loss": 2.4958,
+      "step": 1986
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.465961409698525e-05,
+      "loss": 2.5048,
+      "step": 1988
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.457416042166181e-05,
+      "loss": 2.4204,
+      "step": 1990
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.448872278070523e-05,
+      "loss": 2.5445,
+      "step": 1992
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.4403301426599706e-05,
+      "loss": 2.1568,
+      "step": 1994
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.431789661178121e-05,
+      "loss": 2.2937,
+      "step": 1996
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.423250858863689e-05,
+      "loss": 2.336,
+      "step": 1998
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.4147137609504266e-05,
+      "loss": 2.6334,
+      "step": 2000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.4061783926670496e-05,
+      "loss": 2.2973,
+      "step": 2002
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.3976447792371624e-05,
+      "loss": 2.5251,
+      "step": 2004
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.38911294587918e-05,
+      "loss": 2.4591,
+      "step": 2006
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.38058291780626e-05,
+      "loss": 2.4621,
+      "step": 2008
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.372054720226227e-05,
+      "loss": 2.3918,
+      "step": 2010
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.3635283783414924e-05,
+      "loss": 2.357,
+      "step": 2012
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.3550039173489845e-05,
+      "loss": 2.3846,
+      "step": 2014
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.346481362440074e-05,
+      "loss": 2.421,
+      "step": 2016
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.337960738800498e-05,
+      "loss": 2.523,
+      "step": 2018
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.3294420716102895e-05,
+      "loss": 2.3806,
+      "step": 2020
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.320925386043696e-05,
+      "loss": 2.4476,
+      "step": 2022
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.3124107072691055e-05,
+      "loss": 2.4976,
+      "step": 2024
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.3038980604489885e-05,
+      "loss": 2.6172,
+      "step": 2026
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.2953874707397964e-05,
+      "loss": 2.4828,
+      "step": 2028
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.28687896329191e-05,
+      "loss": 2.3404,
+      "step": 2030
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.278372563249552e-05,
+      "loss": 2.4298,
+      "step": 2032
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.269868295750722e-05,
+      "loss": 2.5339,
+      "step": 2034
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.261366185927114e-05,
+      "loss": 2.2219,
+      "step": 2036
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.252866258904045e-05,
+      "loss": 2.3277,
+      "step": 2038
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.2443685398003835e-05,
+      "loss": 2.3991,
+      "step": 2040
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.235873053728475e-05,
+      "loss": 2.3344,
+      "step": 2042
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.227379825794063e-05,
+      "loss": 2.3301,
+      "step": 2044
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.218888881096217e-05,
+      "loss": 2.4981,
+      "step": 2046
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.21040024472726e-05,
+      "loss": 2.4976,
+      "step": 2048
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.201913941772696e-05,
+      "loss": 2.6412,
+      "step": 2050
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.193429997311132e-05,
+      "loss": 2.3847,
+      "step": 2052
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.184948436414203e-05,
+      "loss": 2.2447,
+      "step": 2054
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.1764692841464995e-05,
+      "loss": 2.5925,
+      "step": 2056
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.1679925655655e-05,
+      "loss": 2.3348,
+      "step": 2058
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.159518305721487e-05,
+      "loss": 2.5109,
+      "step": 2060
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.151046529657477e-05,
+      "loss": 2.5121,
+      "step": 2062
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.142577262409144e-05,
+      "loss": 2.4563,
+      "step": 2064
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.134110529004753e-05,
+      "loss": 2.4912,
+      "step": 2066
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.1256463544650783e-05,
+      "loss": 2.4457,
+      "step": 2068
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.117184763803334e-05,
+      "loss": 2.4428,
+      "step": 2070
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.108725782025092e-05,
+      "loss": 2.3274,
+      "step": 2072
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.1002694341282276e-05,
+      "loss": 2.4062,
+      "step": 2074
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.0918157451028185e-05,
+      "loss": 2.6018,
+      "step": 2076
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.083364739931092e-05,
+      "loss": 2.4493,
+      "step": 2078
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.0749164435873425e-05,
+      "loss": 2.5798,
+      "step": 2080
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.0664708810378625e-05,
+      "loss": 2.2729,
+      "step": 2082
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.05802807724086e-05,
+      "loss": 2.3844,
+      "step": 2084
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.049588057146394e-05,
+      "loss": 2.402,
+      "step": 2086
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.041150845696296e-05,
+      "loss": 2.4163,
+      "step": 2088
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.032716467824099e-05,
+      "loss": 2.428,
+      "step": 2090
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.0242849484549623e-05,
+      "loss": 2.4803,
+      "step": 2092
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.015856312505593e-05,
+      "loss": 2.2398,
+      "step": 2094
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.0074305848841814e-05,
+      "loss": 2.2521,
+      "step": 2096
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.9990077904903254e-05,
+      "loss": 2.3918,
+      "step": 2098
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.99058795421495e-05,
+      "loss": 2.519,
+      "step": 2100
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.982171100940239e-05,
+      "loss": 2.4067,
+      "step": 2102
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.973757255539562e-05,
+      "loss": 2.4408,
+      "step": 2104
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.965346442877403e-05,
+      "loss": 2.4309,
+      "step": 2106
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.9569386878092774e-05,
+      "loss": 2.2934,
+      "step": 2108
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.94853401518167e-05,
+      "loss": 2.356,
+      "step": 2110
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.94013244983195e-05,
+      "loss": 2.3314,
+      "step": 2112
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.9317340165883156e-05,
+      "loss": 2.4871,
+      "step": 2114
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.923338740269696e-05,
+      "loss": 2.4932,
+      "step": 2116
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.9149466456857e-05,
+      "loss": 2.2894,
+      "step": 2118
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.906557757636526e-05,
+      "loss": 2.4726,
+      "step": 2120
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.898172100912908e-05,
+      "loss": 2.3281,
+      "step": 2122
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8897897002960195e-05,
+      "loss": 2.6179,
+      "step": 2124
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8814105805574166e-05,
+      "loss": 2.3778,
+      "step": 2126
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.873034766458959e-05,
+      "loss": 2.4198,
+      "step": 2128
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.86466228275274e-05,
+      "loss": 2.318,
+      "step": 2130
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.856293154181009e-05,
+      "loss": 2.3894,
+      "step": 2132
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.847927405476097e-05,
+      "loss": 2.388,
+      "step": 2134
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.839565061360352e-05,
+      "loss": 2.4026,
+      "step": 2136
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.831206146546059e-05,
+      "loss": 2.3764,
+      "step": 2138
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.82285068573537e-05,
+      "loss": 2.2921,
+      "step": 2140
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.814498703620226e-05,
+      "loss": 2.5344,
+      "step": 2142
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.80615022488229e-05,
+      "loss": 2.3933,
+      "step": 2144
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.797805274192875e-05,
+      "loss": 2.5373,
+      "step": 2146
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.789463876212863e-05,
+      "loss": 2.2517,
+      "step": 2148
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.781126055592641e-05,
+      "loss": 2.5876,
+      "step": 2150
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.772791836972019e-05,
+      "loss": 2.5212,
+      "step": 2152
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.764461244980169e-05,
+      "loss": 2.4556,
+      "step": 2154
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.7561343042355415e-05,
+      "loss": 2.4632,
+      "step": 2156
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.747811039345798e-05,
+      "loss": 2.333,
+      "step": 2158
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.739491474907735e-05,
+      "loss": 2.3092,
+      "step": 2160
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.731175635507219e-05,
+      "loss": 2.4144,
+      "step": 2162
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.722863545719103e-05,
+      "loss": 2.51,
+      "step": 2164
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.7145552301071594e-05,
+      "loss": 2.449,
+      "step": 2166
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.706250713224011e-05,
+      "loss": 2.4989,
+      "step": 2168
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.697950019611054e-05,
+      "loss": 2.4551,
+      "step": 2170
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.689653173798381e-05,
+      "loss": 2.4758,
+      "step": 2172
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.681360200304718e-05,
+      "loss": 2.5718,
+      "step": 2174
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.673071123637347e-05,
+      "loss": 2.4931,
+      "step": 2176
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.664785968292036e-05,
+      "loss": 2.5029,
+      "step": 2178
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.656504758752961e-05,
+      "loss": 2.4353,
+      "step": 2180
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.64822751949264e-05,
+      "loss": 2.2827,
+      "step": 2182
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.639954274971854e-05,
+      "loss": 2.4667,
+      "step": 2184
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.631685049639586e-05,
+      "loss": 2.3599,
+      "step": 2186
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.623419867932937e-05,
+      "loss": 2.3421,
+      "step": 2188
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.6151587542770567e-05,
+      "loss": 2.4358,
+      "step": 2190
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.6069017330850754e-05,
+      "loss": 2.4472,
+      "step": 2192
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.598648828758031e-05,
+      "loss": 2.4576,
+      "step": 2194
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.590400065684792e-05,
+      "loss": 2.5057,
+      "step": 2196
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.582155468241993e-05,
+      "loss": 2.4759,
+      "step": 2198
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.573915060793949e-05,
+      "loss": 2.1946,
+      "step": 2200
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.5656788676926066e-05,
+      "loss": 2.562,
+      "step": 2202
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.557446913277448e-05,
+      "loss": 2.7023,
+      "step": 2204
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.5492192218754326e-05,
+      "loss": 2.5342,
+      "step": 2206
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.540995817800917e-05,
+      "loss": 2.4178,
+      "step": 2208
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.532776725355601e-05,
+      "loss": 2.4072,
+      "step": 2210
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.5245619688284274e-05,
+      "loss": 2.3832,
+      "step": 2212
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.516351572495535e-05,
+      "loss": 2.5958,
+      "step": 2214
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.508145560620173e-05,
+      "loss": 2.3778,
+      "step": 2216
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.499943957452637e-05,
+      "loss": 2.3753,
+      "step": 2218
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.4917467872301934e-05,
+      "loss": 2.4079,
+      "step": 2220
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.4835540741770054e-05,
+      "loss": 2.4669,
+      "step": 2222
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.4753658425040656e-05,
+      "loss": 2.4278,
+      "step": 2224
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.467182116409127e-05,
+      "loss": 2.4575,
+      "step": 2226
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.459002920076625e-05,
+      "loss": 2.5202,
+      "step": 2228
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.450828277677606e-05,
+      "loss": 2.4366,
+      "step": 2230
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.442658213369662e-05,
+      "loss": 2.4061,
+      "step": 2232
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.434492751296856e-05,
+      "loss": 2.393,
+      "step": 2234
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.426331915589651e-05,
+      "loss": 2.3859,
+      "step": 2236
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.418175730364836e-05,
+      "loss": 2.516,
+      "step": 2238
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.4100242197254564e-05,
+      "loss": 2.2877,
+      "step": 2240
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.401877407760747e-05,
+      "loss": 2.3637,
+      "step": 2242
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.393735318546054e-05,
+      "loss": 2.4131,
+      "step": 2244
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.38559797614277e-05,
+      "loss": 2.4939,
+      "step": 2246
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.377465404598253e-05,
+      "loss": 2.4648,
+      "step": 2248
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.369337627945774e-05,
+      "loss": 2.504,
+      "step": 2250
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.3612146702044226e-05,
+      "loss": 2.5889,
+      "step": 2252
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.3530965553790526e-05,
+      "loss": 2.5468,
+      "step": 2254
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.3449833074602064e-05,
+      "loss": 2.3507,
+      "step": 2256
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.336874950424046e-05,
+      "loss": 2.418,
+      "step": 2258
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.328771508232273e-05,
+      "loss": 2.424,
+      "step": 2260
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.320673004832071e-05,
+      "loss": 2.4613,
+      "step": 2262
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.312579464156025e-05,
+      "loss": 2.5197,
+      "step": 2264
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.304490910122058e-05,
+      "loss": 2.3393,
+      "step": 2266
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.2964073666333536e-05,
+      "loss": 2.2088,
+      "step": 2268
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.2883288575782875e-05,
+      "loss": 2.4951,
+      "step": 2270
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.2802554068303596e-05,
+      "loss": 2.3728,
+      "step": 2272
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.272187038248121e-05,
+      "loss": 2.2495,
+      "step": 2274
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.264123775675106e-05,
+      "loss": 2.3507,
+      "step": 2276
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.256065642939756e-05,
+      "loss": 2.3987,
+      "step": 2278
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.248012663855353e-05,
+      "loss": 2.4537,
+      "step": 2280
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.239964862219954e-05,
+      "loss": 2.4883,
+      "step": 2282
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.231922261816311e-05,
+      "loss": 2.1862,
+      "step": 2284
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.223884886411807e-05,
+      "loss": 2.4353,
+      "step": 2286
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.215852759758381e-05,
+      "loss": 2.2626,
+      "step": 2288
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.2078259055924675e-05,
+      "loss": 2.3032,
+      "step": 2290
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.199804347634915e-05,
+      "loss": 2.3917,
+      "step": 2292
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.191788109590922e-05,
+      "loss": 2.4218,
+      "step": 2294
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.183777215149962e-05,
+      "loss": 2.3915,
+      "step": 2296
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.175771687985726e-05,
+      "loss": 2.5188,
+      "step": 2298
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.167771551756036e-05,
+      "loss": 2.4356,
+      "step": 2300
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.159776830102784e-05,
+      "loss": 2.3713,
+      "step": 2302
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.1517875466518626e-05,
+      "loss": 2.488,
+      "step": 2304
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.1438037250130944e-05,
+      "loss": 2.5058,
+      "step": 2306
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.135825388780159e-05,
+      "loss": 2.5503,
+      "step": 2308
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.127852561530526e-05,
+      "loss": 2.4376,
+      "step": 2310
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.1198852668253856e-05,
+      "loss": 2.4054,
+      "step": 2312
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.111923528209577e-05,
+      "loss": 2.3189,
+      "step": 2314
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.103967369211525e-05,
+      "loss": 2.3736,
+      "step": 2316
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.096016813343158e-05,
+      "loss": 2.4304,
+      "step": 2318
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.08807188409985e-05,
+      "loss": 2.3717,
+      "step": 2320
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.080132604960349e-05,
+      "loss": 2.3996,
+      "step": 2322
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.072198999386704e-05,
+      "loss": 2.5024,
+      "step": 2324
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.064271090824197e-05,
+      "loss": 2.3109,
+      "step": 2326
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.056348902701274e-05,
+      "loss": 2.376,
+      "step": 2328
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.0484324584294783e-05,
+      "loss": 2.5198,
+      "step": 2330
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.040521781403377e-05,
+      "loss": 2.2787,
+      "step": 2332
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.0326168950004964e-05,
+      "loss": 2.3356,
+      "step": 2334
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.0247178225812435e-05,
+      "loss": 2.4738,
+      "step": 2336
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.0168245874888557e-05,
+      "loss": 2.41,
+      "step": 2338
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.00893721304931e-05,
+      "loss": 2.4233,
+      "step": 2340
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0010557225712667e-05,
+      "loss": 2.4596,
+      "step": 2342
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.993180139345999e-05,
+      "loss": 2.5253,
+      "step": 2344
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.9853104866473246e-05,
+      "loss": 2.3779,
+      "step": 2346
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.977446787731532e-05,
+      "loss": 2.397,
+      "step": 2348
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9695890658373164e-05,
+      "loss": 2.4016,
+      "step": 2350
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.96173734418571e-05,
+      "loss": 2.3168,
+      "step": 2352
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9538916459800136e-05,
+      "loss": 2.2834,
+      "step": 2354
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9460519944057284e-05,
+      "loss": 2.4692,
+      "step": 2356
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9382184126304834e-05,
+      "loss": 2.1485,
+      "step": 2358
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9303909238039718e-05,
+      "loss": 2.1016,
+      "step": 2360
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9225695510578843e-05,
+      "loss": 2.1819,
+      "step": 2362
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9147543175058335e-05,
+      "loss": 2.4526,
+      "step": 2364
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9069452462432883e-05,
+      "loss": 2.5037,
+      "step": 2366
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.899142360347511e-05,
+      "loss": 2.223,
+      "step": 2368
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8913456828774854e-05,
+      "loss": 2.4296,
+      "step": 2370
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.883555236873845e-05,
+      "loss": 2.3899,
+      "step": 2372
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.875771045358805e-05,
+      "loss": 2.4758,
+      "step": 2374
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8679931313361053e-05,
+      "loss": 2.4788,
+      "step": 2376
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.860221517790933e-05,
+      "loss": 2.4869,
+      "step": 2378
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8524562276898513e-05,
+      "loss": 2.4313,
+      "step": 2380
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8446972839807384e-05,
+      "loss": 2.2432,
+      "step": 2382
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8369447095927195e-05,
+      "loss": 2.2748,
+      "step": 2384
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8291985274360983e-05,
+      "loss": 2.6419,
+      "step": 2386
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8214587604022847e-05,
+      "loss": 2.4079,
+      "step": 2388
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8137254313637306e-05,
+      "loss": 2.3603,
+      "step": 2390
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.805998563173866e-05,
+      "loss": 2.2753,
+      "step": 2392
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.798278178667028e-05,
+      "loss": 2.3884,
+      "step": 2394
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.790564300658387e-05,
+      "loss": 2.5817,
+      "step": 2396
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7828569519438942e-05,
+      "loss": 2.4844,
+      "step": 2398
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.775156155300197e-05,
+      "loss": 2.4505,
+      "step": 2400
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7674619334845876e-05,
+      "loss": 2.443,
+      "step": 2402
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7597743092349217e-05,
+      "loss": 2.3359,
+      "step": 2404
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.752093305269565e-05,
+      "loss": 2.4345,
+      "step": 2406
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.7444189442873115e-05,
+      "loss": 2.2828,
+      "step": 2408
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.7367512489673312e-05,
+      "loss": 2.5291,
+      "step": 2410
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.7290902419690895e-05,
+      "loss": 2.391,
+      "step": 2412
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.7214359459322924e-05,
+      "loss": 2.06,
+      "step": 2414
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.7137883834768073e-05,
+      "loss": 2.6396,
+      "step": 2416
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.7061475772026086e-05,
+      "loss": 2.4143,
+      "step": 2418
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.698513549689703e-05,
+      "loss": 2.4331,
+      "step": 2420
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6908863234980636e-05,
+      "loss": 2.4801,
+      "step": 2422
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.6832659211675627e-05,
+      "loss": 2.3912,
+      "step": 2424
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.67565236521791e-05,
+      "loss": 2.4359,
+      "step": 2426
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.668045678148584e-05,
+      "loss": 2.5352,
+      "step": 2428
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.6604458824387614e-05,
+      "loss": 2.4982,
+      "step": 2430
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.6528530005472518e-05,
+      "loss": 2.4898,
+      "step": 2432
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.6452670549124375e-05,
+      "loss": 2.5291,
+      "step": 2434
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.637688067952204e-05,
+      "loss": 2.3178,
+      "step": 2436
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.630116062063867e-05,
+      "loss": 2.5588,
+      "step": 2438
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.622551059624113e-05,
+      "loss": 2.3862,
+      "step": 2440
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.614993082988937e-05,
+      "loss": 2.3327,
+      "step": 2442
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.607442154493568e-05,
+      "loss": 2.3323,
+      "step": 2444
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.599898296452406e-05,
+      "loss": 2.237,
+      "step": 2446
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.592361531158952e-05,
+      "loss": 2.1117,
+      "step": 2448
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.5848318808857606e-05,
+      "loss": 2.3355,
+      "step": 2450
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.5773093678843473e-05,
+      "loss": 2.3701,
+      "step": 2452
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.5697940143851375e-05,
+      "loss": 2.4158,
+      "step": 2454
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.5622858425974018e-05,
+      "loss": 2.4807,
+      "step": 2456
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.5547848747091897e-05,
+      "loss": 2.2695,
+      "step": 2458
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.5472911328872574e-05,
+      "loss": 2.4957,
+      "step": 2460
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.5398046392770054e-05,
+      "loss": 2.3591,
+      "step": 2462
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.532325416002419e-05,
+      "loss": 2.6757,
+      "step": 2464
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.524853485166e-05,
+      "loss": 2.3271,
+      "step": 2466
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.517388868848692e-05,
+      "loss": 2.3084,
+      "step": 2468
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.5099315891098264e-05,
+      "loss": 2.3597,
+      "step": 2470
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.5024816679870556e-05,
+      "loss": 2.5597,
+      "step": 2472
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.495039127496287e-05,
+      "loss": 2.4667,
+      "step": 2474
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.4876039896316123e-05,
+      "loss": 2.2991,
+      "step": 2476
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.4801762763652474e-05,
+      "loss": 2.27,
+      "step": 2478
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.4727560096474706e-05,
+      "loss": 2.5184,
+      "step": 2480
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.4653432114065544e-05,
+      "loss": 2.5034,
+      "step": 2482
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.457937903548695e-05,
+      "loss": 2.3994,
+      "step": 2484
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.450540107957961e-05,
+      "loss": 2.229,
+      "step": 2486
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.443149846496212e-05,
+      "loss": 2.4133,
+      "step": 2488
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.4357671410030526e-05,
+      "loss": 2.5226,
+      "step": 2490
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.4283920132957482e-05,
+      "loss": 2.3836,
+      "step": 2492
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.42102448516918e-05,
+      "loss": 2.4287,
+      "step": 2494
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.413664578395761e-05,
+      "loss": 2.4322,
+      "step": 2496
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.4063123147253923e-05,
+      "loss": 2.3545,
+      "step": 2498
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.398967715885379e-05,
+      "loss": 2.3359,
+      "step": 2500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.391630803580382e-05,
+      "loss": 2.4889,
+      "step": 2502
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.3843015994923412e-05,
+      "loss": 2.3731,
+      "step": 2504
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.3769801252804213e-05,
+      "loss": 2.2901,
+      "step": 2506
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.3696664025809458e-05,
+      "loss": 2.3341,
+      "step": 2508
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.3623604530073245e-05,
+      "loss": 2.3624,
+      "step": 2510
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.3550622981499988e-05,
+      "loss": 2.4377,
+      "step": 2512
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.3477719595763774e-05,
+      "loss": 2.2931,
+      "step": 2514
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.340489458830772e-05,
+      "loss": 2.3726,
+      "step": 2516
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.3332148174343254e-05,
+      "loss": 2.2644,
+      "step": 2518
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.3259480568849586e-05,
+      "loss": 2.5434,
+      "step": 2520
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.3186891986573035e-05,
+      "loss": 2.2445,
+      "step": 2522
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.3114382642026404e-05,
+      "loss": 2.3012,
+      "step": 2524
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.3041952749488304e-05,
+      "loss": 2.298,
+      "step": 2526
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2969602523002543e-05,
+      "loss": 2.3226,
+      "step": 2528
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2897332176377528e-05,
+      "loss": 2.4809,
+      "step": 2530
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2825141923185632e-05,
+      "loss": 2.3514,
+      "step": 2532
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.275303197676248e-05,
+      "loss": 2.4344,
+      "step": 2534
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2681002550206355e-05,
+      "loss": 2.3313,
+      "step": 2536
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2609053856377714e-05,
+      "loss": 2.1924,
+      "step": 2538
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2537186107898313e-05,
+      "loss": 2.1984,
+      "step": 2540
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2465399517150722e-05,
+      "loss": 2.4612,
+      "step": 2542
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2393694296277707e-05,
+      "loss": 2.3225,
+      "step": 2544
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2322070657181583e-05,
+      "loss": 2.3635,
+      "step": 2546
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2250528811523513e-05,
+      "loss": 2.4144,
+      "step": 2548
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2179068970722978e-05,
+      "loss": 2.5847,
+      "step": 2550
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2107691345957133e-05,
+      "loss": 2.3221,
+      "step": 2552
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.203639614816017e-05,
+      "loss": 2.4227,
+      "step": 2554
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.196518358802268e-05,
+      "loss": 2.4364,
+      "step": 2556
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1894053875991017e-05,
+      "loss": 2.4288,
+      "step": 2558
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.182300722226675e-05,
+      "loss": 2.3931,
+      "step": 2560
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1752043836806002e-05,
+      "loss": 2.4772,
+      "step": 2562
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1681163929318777e-05,
+      "loss": 2.3936,
+      "step": 2564
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1610367709268387e-05,
+      "loss": 2.3759,
+      "step": 2566
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1539655385870877e-05,
+      "loss": 2.4427,
+      "step": 2568
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1469027168094347e-05,
+      "loss": 2.387,
+      "step": 2570
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1398483264658313e-05,
+      "loss": 2.2637,
+      "step": 2572
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.132802388403319e-05,
+      "loss": 2.3364,
+      "step": 2574
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.125764923443953e-05,
+      "loss": 2.3348,
+      "step": 2576
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.118735952384757e-05,
+      "loss": 2.34,
+      "step": 2578
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1117154959976482e-05,
+      "loss": 2.2867,
+      "step": 2580
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.104703575029385e-05,
+      "loss": 2.4191,
+      "step": 2582
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.097700210201497e-05,
+      "loss": 2.2275,
+      "step": 2584
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.090705422210237e-05,
+      "loss": 2.6198,
+      "step": 2586
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0837192317265016e-05,
+      "loss": 2.364,
+      "step": 2588
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0767416593957894e-05,
+      "loss": 2.2663,
+      "step": 2590
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0697727258381238e-05,
+      "loss": 2.2649,
+      "step": 2592
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0628124516480046e-05,
+      "loss": 2.5761,
+      "step": 2594
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0558608573943354e-05,
+      "loss": 2.2132,
+      "step": 2596
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0489179636203766e-05,
+      "loss": 2.3719,
+      "step": 2598
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0419837908436688e-05,
+      "loss": 2.4978,
+      "step": 2600
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0350583595559865e-05,
+      "loss": 2.2988,
+      "step": 2602
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0281416902232708e-05,
+      "loss": 2.255,
+      "step": 2604
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.021233803285567e-05,
+      "loss": 2.2799,
+      "step": 2606
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.014334719156966e-05,
+      "loss": 2.2972,
+      "step": 2608
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0074444582255485e-05,
+      "loss": 2.4158,
+      "step": 2610
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0005630408533215e-05,
+      "loss": 2.3353,
+      "step": 2612
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9936904873761536e-05,
+      "loss": 2.3829,
+      "step": 2614
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9868268181037185e-05,
+      "loss": 2.1709,
+      "step": 2616
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9799720533194404e-05,
+      "loss": 2.549,
+      "step": 2618
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9731262132804274e-05,
+      "loss": 2.5804,
+      "step": 2620
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.966289318217411e-05,
+      "loss": 2.5311,
+      "step": 2622
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.959461388334686e-05,
+      "loss": 2.3825,
+      "step": 2624
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9526424438100642e-05,
+      "loss": 2.3505,
+      "step": 2626
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9458325047947938e-05,
+      "loss": 2.3793,
+      "step": 2628
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9390315914135125e-05,
+      "loss": 2.2617,
+      "step": 2630
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9322397237641875e-05,
+      "loss": 2.5081,
+      "step": 2632
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.925456921918055e-05,
+      "loss": 2.578,
+      "step": 2634
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.918683205919557e-05,
+      "loss": 2.3566,
+      "step": 2636
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9119185957862835e-05,
+      "loss": 2.5683,
+      "step": 2638
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9051631115089196e-05,
+      "loss": 2.213,
+      "step": 2640
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8984167730511825e-05,
+      "loss": 2.6764,
+      "step": 2642
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8916796003497572e-05,
+      "loss": 2.422,
+      "step": 2644
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8849516133142432e-05,
+      "loss": 2.1258,
+      "step": 2646
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8782328318270964e-05,
+      "loss": 2.505,
+      "step": 2648
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8715232757435704e-05,
+      "loss": 2.2427,
+      "step": 2650
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.864822964891651e-05,
+      "loss": 2.3599,
+      "step": 2652
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8581319190720035e-05,
+      "loss": 2.679,
+      "step": 2654
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.851450158057918e-05,
+      "loss": 2.3708,
+      "step": 2656
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.844777701595244e-05,
+      "loss": 2.414,
+      "step": 2658
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.83811456940233e-05,
+      "loss": 2.524,
+      "step": 2660
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8314607811699762e-05,
+      "loss": 2.4521,
+      "step": 2662
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.824816356561364e-05,
+      "loss": 2.3931,
+      "step": 2664
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8181813152120092e-05,
+      "loss": 2.3704,
+      "step": 2666
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8115556767296914e-05,
+      "loss": 2.5238,
+      "step": 2668
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.804939460694411e-05,
+      "loss": 2.5418,
+      "step": 2670
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7983326866583144e-05,
+      "loss": 2.6173,
+      "step": 2672
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7917353741456545e-05,
+      "loss": 2.4272,
+      "step": 2674
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7851475426527142e-05,
+      "loss": 2.4986,
+      "step": 2676
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7785692116477682e-05,
+      "loss": 2.4365,
+      "step": 2678
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.772000400571005e-05,
+      "loss": 2.5869,
+      "step": 2680
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.76544112883449e-05,
+      "loss": 2.4987,
+      "step": 2682
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.7588914158220898e-05,
+      "loss": 2.4701,
+      "step": 2684
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.7523512808894288e-05,
+      "loss": 2.5142,
+      "step": 2686
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.7458207433638223e-05,
+      "loss": 2.518,
+      "step": 2688
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.7392998225442263e-05,
+      "loss": 2.386,
+      "step": 2690
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.732788537701179e-05,
+      "loss": 2.2214,
+      "step": 2692
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.726286908076738e-05,
+      "loss": 2.358,
+      "step": 2694
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.7197949528844286e-05,
+      "loss": 2.5727,
+      "step": 2696
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.7133126913091903e-05,
+      "loss": 2.5317,
+      "step": 2698
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.706840142507315e-05,
+      "loss": 2.2929,
+      "step": 2700
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.700377325606388e-05,
+      "loss": 2.4207,
+      "step": 2702
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.6939242597052373e-05,
+      "loss": 2.4398,
+      "step": 2704
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.6874809638738754e-05,
+      "loss": 2.3671,
+      "step": 2706
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.681047457153444e-05,
+      "loss": 2.5831,
+      "step": 2708
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.6746237585561524e-05,
+      "loss": 2.414,
+      "step": 2710
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.6682098870652236e-05,
+      "loss": 2.2996,
+      "step": 2712
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.6618058616348492e-05,
+      "loss": 2.4037,
+      "step": 2714
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.655411701190115e-05,
+      "loss": 2.592,
+      "step": 2716
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.6490274246269533e-05,
+      "loss": 2.2498,
+      "step": 2718
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.642653050812094e-05,
+      "loss": 2.2538,
+      "step": 2720
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.636288598583e-05,
+      "loss": 2.3926,
+      "step": 2722
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.629934086747813e-05,
+      "loss": 2.5224,
+      "step": 2724
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6235895340852964e-05,
+      "loss": 2.3785,
+      "step": 2726
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6172549593447877e-05,
+      "loss": 2.4254,
+      "step": 2728
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6109303812461375e-05,
+      "loss": 2.2977,
+      "step": 2730
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.60461581847965e-05,
+      "loss": 2.4296,
+      "step": 2732
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.598311289706033e-05,
+      "loss": 2.1917,
+      "step": 2734
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.592016813556347e-05,
+      "loss": 2.361,
+      "step": 2736
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5857324086319414e-05,
+      "loss": 2.3198,
+      "step": 2738
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.579458093504403e-05,
+      "loss": 2.3945,
+      "step": 2740
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5731938867155e-05,
+      "loss": 2.2314,
+      "step": 2742
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5669398067771324e-05,
+      "loss": 2.4571,
+      "step": 2744
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.560695872171273e-05,
+      "loss": 2.2473,
+      "step": 2746
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5544621013499094e-05,
+      "loss": 2.4553,
+      "step": 2748
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.548238512734998e-05,
+      "loss": 2.3213,
+      "step": 2750
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.542025124718401e-05,
+      "loss": 2.3302,
+      "step": 2752
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.535821955661839e-05,
+      "loss": 2.2468,
+      "step": 2754
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5296290238968303e-05,
+      "loss": 2.3087,
+      "step": 2756
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5234463477246452e-05,
+      "loss": 2.4679,
+      "step": 2758
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5172739454162405e-05,
+      "loss": 2.3439,
+      "step": 2760
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5111118352122183e-05,
+      "loss": 2.2882,
+      "step": 2762
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5049600353227588e-05,
+      "loss": 2.4456,
+      "step": 2764
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4988185639275798e-05,
+      "loss": 2.3367,
+      "step": 2766
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4926874391758716e-05,
+      "loss": 2.3341,
+      "step": 2768
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4865666791862521e-05,
+      "loss": 2.7886,
+      "step": 2770
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4804563020467044e-05,
+      "loss": 2.6835,
+      "step": 2772
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4743563258145353e-05,
+      "loss": 2.3864,
+      "step": 2774
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4682667685163071e-05,
+      "loss": 2.3261,
+      "step": 2776
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4621876481477987e-05,
+      "loss": 2.4467,
+      "step": 2778
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4561189826739446e-05,
+      "loss": 2.3331,
+      "step": 2780
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.45006079002878e-05,
+      "loss": 2.2346,
+      "step": 2782
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4440130881153917e-05,
+      "loss": 2.3942,
+      "step": 2784
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.437975894805867e-05,
+      "loss": 2.441,
+      "step": 2786
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4319492279412388e-05,
+      "loss": 2.3998,
+      "step": 2788
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.425933105331429e-05,
+      "loss": 2.518,
+      "step": 2790
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.419927544755199e-05,
+      "loss": 2.3147,
+      "step": 2792
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4139325639601015e-05,
+      "loss": 2.2925,
+      "step": 2794
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4079481806624217e-05,
+      "loss": 2.5182,
+      "step": 2796
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4019744125471274e-05,
+      "loss": 2.4969,
+      "step": 2798
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3960112772678125e-05,
+      "loss": 2.4316,
+      "step": 2800
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3900587924466585e-05,
+      "loss": 2.3239,
+      "step": 2802
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3841169756743649e-05,
+      "loss": 2.6349,
+      "step": 2804
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.378185844510107e-05,
+      "loss": 2.3982,
+      "step": 2806
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3722654164814796e-05,
+      "loss": 2.4663,
+      "step": 2808
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.366355709084456e-05,
+      "loss": 2.4762,
+      "step": 2810
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3604567397833201e-05,
+      "loss": 2.4103,
+      "step": 2812
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.354568526010624e-05,
+      "loss": 2.4714,
+      "step": 2814
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3486910851671374e-05,
+      "loss": 2.1582,
+      "step": 2816
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.342824434621795e-05,
+      "loss": 2.3474,
+      "step": 2818
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3369685917116408e-05,
+      "loss": 2.3022,
+      "step": 2820
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3311235737417793e-05,
+      "loss": 2.2013,
+      "step": 2822
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3252893979853304e-05,
+      "loss": 2.5426,
+      "step": 2824
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.319466081683371e-05,
+      "loss": 2.3739,
+      "step": 2826
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3136536420448841e-05,
+      "loss": 2.3773,
+      "step": 2828
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.307852096246711e-05,
+      "loss": 2.4481,
+      "step": 2830
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.302061461433502e-05,
+      "loss": 2.5957,
+      "step": 2832
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2962817547176625e-05,
+      "loss": 2.5113,
+      "step": 2834
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2905129931793009e-05,
+      "loss": 2.3745,
+      "step": 2836
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2847551938661839e-05,
+      "loss": 2.3667,
+      "step": 2838
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2790083737936798e-05,
+      "loss": 2.3051,
+      "step": 2840
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2732725499447146e-05,
+      "loss": 2.2803,
+      "step": 2842
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2675477392697139e-05,
+      "loss": 2.3317,
+      "step": 2844
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2618339586865625e-05,
+      "loss": 2.5069,
+      "step": 2846
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2561312250805435e-05,
+      "loss": 2.403,
+      "step": 2848
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2504395553043008e-05,
+      "loss": 2.3479,
+      "step": 2850
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2447589661777759e-05,
+      "loss": 2.2771,
+      "step": 2852
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.239089474488171e-05,
+      "loss": 2.3368,
+      "step": 2854
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2334310969898871e-05,
+      "loss": 2.3194,
+      "step": 2856
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.227783850404487e-05,
+      "loss": 2.3038,
+      "step": 2858
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2221477514206337e-05,
+      "loss": 2.5084,
+      "step": 2860
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.216522816694053e-05,
+      "loss": 2.2501,
+      "step": 2862
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2109090628474718e-05,
+      "loss": 2.2798,
+      "step": 2864
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2053065064705805e-05,
+      "loss": 2.2456,
+      "step": 2866
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.1997151641199772e-05,
+      "loss": 2.4106,
+      "step": 2868
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.1941350523191208e-05,
+      "loss": 2.4705,
+      "step": 2870
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.1885661875582783e-05,
+      "loss": 2.5891,
+      "step": 2872
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.183008586294485e-05,
+      "loss": 2.2367,
+      "step": 2874
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.1774622649514889e-05,
+      "loss": 2.5675,
+      "step": 2876
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.1719272399197023e-05,
+      "loss": 2.4596,
+      "step": 2878
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.166403527556153e-05,
+      "loss": 2.2995,
+      "step": 2880
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.1608911441844429e-05,
+      "loss": 2.2225,
+      "step": 2882
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.155390106094692e-05,
+      "loss": 2.2498,
+      "step": 2884
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.1499004295434918e-05,
+      "loss": 2.3428,
+      "step": 2886
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.1444221307538571e-05,
+      "loss": 2.3654,
+      "step": 2888
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.1389552259151864e-05,
+      "loss": 2.3089,
+      "step": 2890
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.1334997311832002e-05,
+      "loss": 2.3778,
+      "step": 2892
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1280556626799005e-05,
+      "loss": 2.3831,
+      "step": 2894
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1226230364935226e-05,
+      "loss": 2.4711,
+      "step": 2896
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1172018686784935e-05,
+      "loss": 2.5057,
+      "step": 2898
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1117921752553723e-05,
+      "loss": 2.3913,
+      "step": 2900
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.106393972210809e-05,
+      "loss": 2.5023,
+      "step": 2902
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1010072754975014e-05,
+      "loss": 2.3522,
+      "step": 2904
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.095632101034143e-05,
+      "loss": 2.6258,
+      "step": 2906
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.0902684647053735e-05,
+      "loss": 2.3644,
+      "step": 2908
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.0849163823617375e-05,
+      "loss": 2.4708,
+      "step": 2910
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.0795758698196368e-05,
+      "loss": 2.2643,
+      "step": 2912
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.0742469428612816e-05,
+      "loss": 2.4429,
+      "step": 2914
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.0689296172346431e-05,
+      "loss": 2.2625,
+      "step": 2916
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.0636239086534072e-05,
+      "loss": 2.3484,
+      "step": 2918
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.0583298327969338e-05,
+      "loss": 2.4041,
+      "step": 2920
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.0530474053102034e-05,
+      "loss": 2.1622,
+      "step": 2922
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.047776641803772e-05,
+      "loss": 2.417,
+      "step": 2924
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.0425175578537299e-05,
+      "loss": 2.3336,
+      "step": 2926
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.0372701690016474e-05,
+      "loss": 2.4013,
+      "step": 2928
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.0320344907545388e-05,
+      "loss": 2.2072,
+      "step": 2930
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.0268105385848064e-05,
+      "loss": 2.5827,
+      "step": 2932
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.0215983279302049e-05,
+      "loss": 2.5077,
+      "step": 2934
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.0163978741937847e-05,
+      "loss": 2.512,
+      "step": 2936
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.0112091927438583e-05,
+      "loss": 2.457,
+      "step": 2938
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.0060322989139442e-05,
+      "loss": 2.5785,
+      "step": 2940
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.0008672080027298e-05,
+      "loss": 2.4435,
+      "step": 2942
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.957139352740191e-06,
+      "loss": 2.2668,
+      "step": 2944
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.90572495956696e-06,
+      "loss": 2.4811,
+      "step": 2946
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.854429052446684e-06,
+      "loss": 2.4208,
+      "step": 2948
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.803251782968358e-06,
+      "loss": 2.4254,
+      "step": 2950
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.752193302370315e-06,
+      "loss": 2.3277,
+      "step": 2952
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.701253761539897e-06,
+      "loss": 2.2717,
+      "step": 2954
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.650433311012946e-06,
+      "loss": 2.4127,
+      "step": 2956
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.599732100973357e-06,
+      "loss": 2.6305,
+      "step": 2958
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.549150281252633e-06,
+      "loss": 2.3212,
+      "step": 2960
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.498688001329486e-06,
+      "loss": 2.3556,
+      "step": 2962
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.448345410329379e-06,
+      "loss": 2.3342,
+      "step": 2964
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.398122657024022e-06,
+      "loss": 2.454,
+      "step": 2966
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.348019889831006e-06,
+      "loss": 2.6068,
+      "step": 2968
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.298037256813347e-06,
+      "loss": 2.6167,
+      "step": 2970
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.248174905679058e-06,
+      "loss": 2.3684,
+      "step": 2972
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.198432983780658e-06,
+      "loss": 2.5119,
+      "step": 2974
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.14881163811479e-06,
+      "loss": 2.3928,
+      "step": 2976
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.099311015321782e-06,
+      "loss": 2.3656,
+      "step": 2978
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.049931261685207e-06,
+      "loss": 2.3461,
+      "step": 2980
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.000672523131431e-06,
+      "loss": 2.3134,
+      "step": 2982
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.951534945229172e-06,
+      "loss": 2.3456,
+      "step": 2984
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.902518673189192e-06,
+      "loss": 2.3656,
+      "step": 2986
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.853623851863663e-06,
+      "loss": 2.3751,
+      "step": 2988
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.804850625745897e-06,
+      "loss": 2.5872,
+      "step": 2990
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.756199138969866e-06,
+      "loss": 2.2217,
+      "step": 2992
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.707669535309793e-06,
+      "loss": 2.0714,
+      "step": 2994
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.659261958179688e-06,
+      "loss": 2.5951,
+      "step": 2996
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.610976550632943e-06,
+      "loss": 2.3067,
+      "step": 2998
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.562813455361957e-06,
+      "loss": 2.3471,
+      "step": 3000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.514772814697653e-06,
+      "loss": 2.4585,
+      "step": 3002
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.466854770609062e-06,
+      "loss": 2.199,
+      "step": 3004
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.419059464702927e-06,
+      "loss": 2.2591,
+      "step": 3006
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.371387038223289e-06,
+      "loss": 2.3367,
+      "step": 3008
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.323837632051062e-06,
+      "loss": 2.5848,
+      "step": 3010
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.27641138670358e-06,
+      "loss": 2.2525,
+      "step": 3012
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.229108442334255e-06,
+      "loss": 2.7048,
+      "step": 3014
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.18192893873208e-06,
+      "loss": 2.4397,
+      "step": 3016
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.134873015321303e-06,
+      "loss": 2.3919,
+      "step": 3018
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.087940811160916e-06,
+      "loss": 2.3169,
+      "step": 3020
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.041132464944351e-06,
+      "loss": 2.4048,
+      "step": 3022
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 7.994448114998975e-06,
+      "loss": 2.5458,
+      "step": 3024
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 7.947887899285761e-06,
+      "loss": 2.2902,
+      "step": 3026
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 7.901451955398792e-06,
+      "loss": 2.4315,
+      "step": 3028
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 7.855140420564965e-06,
+      "loss": 2.5107,
+      "step": 3030
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 7.808953431643467e-06,
+      "loss": 2.3578,
+      "step": 3032
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.762891125125476e-06,
+      "loss": 2.3267,
+      "step": 3034
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.716953637133677e-06,
+      "loss": 2.3038,
+      "step": 3036
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.671141103421919e-06,
+      "loss": 2.2405,
+      "step": 3038
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.625453659374754e-06,
+      "loss": 2.2669,
+      "step": 3040
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.579891440007103e-06,
+      "loss": 2.4222,
+      "step": 3042
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.534454579963829e-06,
+      "loss": 2.4086,
+      "step": 3044
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.489143213519301e-06,
+      "loss": 2.3461,
+      "step": 3046
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.44395747457704e-06,
+      "loss": 2.224,
+      "step": 3048
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.398897496669338e-06,
+      "loss": 2.316,
+      "step": 3050
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.353963412956838e-06,
+      "loss": 2.4673,
+      "step": 3052
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.309155356228109e-06,
+      "loss": 2.3921,
+      "step": 3054
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.264473458899301e-06,
+      "loss": 2.3709,
+      "step": 3056
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.219917853013764e-06,
+      "loss": 2.5216,
+      "step": 3058
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.175488670241609e-06,
+      "loss": 2.4435,
+      "step": 3060
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.131186041879357e-06,
+      "loss": 2.4123,
+      "step": 3062
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.0870100988495004e-06,
+      "loss": 2.2985,
+      "step": 3064
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.0429609717002076e-06,
+      "loss": 2.4648,
+      "step": 3066
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.999038790604856e-06,
+      "loss": 2.4027,
+      "step": 3068
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.955243685361673e-06,
+      "loss": 2.5828,
+      "step": 3070
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.911575785393326e-06,
+      "loss": 2.2331,
+      "step": 3072
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.868035219746638e-06,
+      "loss": 2.3046,
+      "step": 3074
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.824622117092078e-06,
+      "loss": 2.3877,
+      "step": 3076
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.781336605723432e-06,
+      "loss": 2.307,
+      "step": 3078
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.738178813557472e-06,
+      "loss": 2.4418,
+      "step": 3080
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.695148868133516e-06,
+      "loss": 2.3749,
+      "step": 3082
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.652246896613068e-06,
+      "loss": 2.4227,
+      "step": 3084
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.609473025779434e-06,
+      "loss": 2.5151,
+      "step": 3086
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.566827382037383e-06,
+      "loss": 2.4882,
+      "step": 3088
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.524310091412739e-06,
+      "loss": 2.3111,
+      "step": 3090
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.481921279552023e-06,
+      "loss": 2.3321,
+      "step": 3092
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.439661071722048e-06,
+      "loss": 2.2051,
+      "step": 3094
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.397529592809614e-06,
+      "loss": 2.3448,
+      "step": 3096
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.355526967321112e-06,
+      "loss": 2.4095,
+      "step": 3098
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.313653319382107e-06,
+      "loss": 2.1535,
+      "step": 3100
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.271908772737017e-06,
+      "loss": 2.3662,
+      "step": 3102
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.2302934507487755e-06,
+      "loss": 2.1468,
+      "step": 3104
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.188807476398412e-06,
+      "loss": 2.4795,
+      "step": 3106
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.147450972284696e-06,
+      "loss": 2.353,
+      "step": 3108
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.106224060623822e-06,
+      "loss": 2.3579,
+      "step": 3110
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.065126863248976e-06,
+      "loss": 2.3739,
+      "step": 3112
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.0241595016100545e-06,
+      "loss": 2.3866,
+      "step": 3114
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.98332209677322e-06,
+      "loss": 2.2253,
+      "step": 3116
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.942614769420629e-06,
+      "loss": 2.3605,
+      "step": 3118
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.902037639850011e-06,
+      "loss": 2.3848,
+      "step": 3120
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.86159082797435e-06,
+      "loss": 2.1943,
+      "step": 3122
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.8212744533215016e-06,
+      "loss": 2.3193,
+      "step": 3124
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.781088635033882e-06,
+      "loss": 2.5142,
+      "step": 3126
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.741033491868047e-06,
+      "loss": 2.5701,
+      "step": 3128
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.701109142194422e-06,
+      "loss": 2.5069,
+      "step": 3130
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.6613157039969055e-06,
+      "loss": 2.4168,
+      "step": 3132
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.621653294872514e-06,
+      "loss": 2.4338,
+      "step": 3134
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.582122032031051e-06,
+      "loss": 2.4563,
+      "step": 3136
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.542722032294761e-06,
+      "loss": 2.2138,
+      "step": 3138
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.503453412098003e-06,
+      "loss": 2.6032,
+      "step": 3140
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.464316287486859e-06,
+      "loss": 2.3332,
+      "step": 3142
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.425310774118802e-06,
+      "loss": 2.3154,
+      "step": 3144
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.386436987262416e-06,
+      "loss": 2.6818,
+      "step": 3146
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.347695041796985e-06,
+      "loss": 2.2799,
+      "step": 3148
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.309085052212165e-06,
+      "loss": 2.5646,
+      "step": 3150
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.270607132607663e-06,
+      "loss": 2.3395,
+      "step": 3152
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.232261396692911e-06,
+      "loss": 2.4606,
+      "step": 3154
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.194047957786713e-06,
+      "loss": 2.3552,
+      "step": 3156
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.155966928816885e-06,
+      "loss": 2.5682,
+      "step": 3158
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.118018422319948e-06,
+      "loss": 2.4571,
+      "step": 3160
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.080202550440849e-06,
+      "loss": 2.24,
+      "step": 3162
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.042519424932513e-06,
+      "loss": 2.5308,
+      "step": 3164
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.0049691571555925e-06,
+      "loss": 2.4177,
+      "step": 3166
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.967551858078129e-06,
+      "loss": 2.6177,
+      "step": 3168
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.930267638275221e-06,
+      "loss": 2.505,
+      "step": 3170
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.893116607928677e-06,
+      "loss": 2.5166,
+      "step": 3172
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.856098876826709e-06,
+      "loss": 2.0793,
+      "step": 3174
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.819214554363616e-06,
+      "loss": 2.4421,
+      "step": 3176
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.782463749539446e-06,
+      "loss": 2.3317,
+      "step": 3178
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.745846570959672e-06,
+      "loss": 2.4747,
+      "step": 3180
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.70936312683487e-06,
+      "loss": 2.2323,
+      "step": 3182
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.673013524980424e-06,
+      "loss": 2.3297,
+      "step": 3184
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.63679787281619e-06,
+      "loss": 2.5994,
+      "step": 3186
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6007162773661515e-06,
+      "loss": 2.2933,
+      "step": 3188
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.564768845258139e-06,
+      "loss": 2.4649,
+      "step": 3190
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.528955682723529e-06,
+      "loss": 2.3754,
+      "step": 3192
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.4932768955968876e-06,
+      "loss": 2.6034,
+      "step": 3194
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.4577325893156715e-06,
+      "loss": 2.5477,
+      "step": 3196
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.422322868919937e-06,
+      "loss": 2.3983,
+      "step": 3198
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.3870478390519884e-06,
+      "loss": 2.3261,
+      "step": 3200
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.3519076039561345e-06,
+      "loss": 2.4168,
+      "step": 3202
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.316902267478296e-06,
+      "loss": 2.4235,
+      "step": 3204
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.2820319330657835e-06,
+      "loss": 2.2992,
+      "step": 3206
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.2472967037669066e-06,
+      "loss": 2.4394,
+      "step": 3208
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.2126966822307715e-06,
+      "loss": 2.374,
+      "step": 3210
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.178231970706858e-06,
+      "loss": 2.4277,
+      "step": 3212
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.1439026710448355e-06,
+      "loss": 2.4958,
+      "step": 3214
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.109708884694158e-06,
+      "loss": 2.3339,
+      "step": 3216
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.075650712703849e-06,
+      "loss": 2.3244,
+      "step": 3218
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.041728255722154e-06,
+      "loss": 2.4202,
+      "step": 3220
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.0079416139962525e-06,
+      "loss": 2.4348,
+      "step": 3222
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.974290887371951e-06,
+      "loss": 2.5305,
+      "step": 3224
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.940776175293431e-06,
+      "loss": 2.3909,
+      "step": 3226
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.9073975768029124e-06,
+      "loss": 2.4669,
+      "step": 3228
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.8741551905403735e-06,
+      "loss": 2.5117,
+      "step": 3230
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.8410491147432395e-06,
+      "loss": 2.3205,
+      "step": 3232
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.808079447246149e-06,
+      "loss": 2.5002,
+      "step": 3234
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.7752462854806213e-06,
+      "loss": 2.5181,
+      "step": 3236
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.7425497264747534e-06,
+      "loss": 2.5206,
+      "step": 3238
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.7099898668529642e-06,
+      "loss": 2.4898,
+      "step": 3240
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.677566802835708e-06,
+      "loss": 2.4225,
+      "step": 3242
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.6452806302392007e-06,
+      "loss": 2.3201,
+      "step": 3244
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.6131314444750765e-06,
+      "loss": 2.4289,
+      "step": 3246
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.58111934055016e-06,
+      "loss": 2.2184,
+      "step": 3248
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.5492444130662108e-06,
+      "loss": 2.5492,
+      "step": 3250
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.517506756219563e-06,
+      "loss": 2.5086,
+      "step": 3252
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.4859064638009033e-06,
+      "loss": 2.4952,
+      "step": 3254
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.4544436291949867e-06,
+      "loss": 2.469,
+      "step": 3256
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.4231183453803604e-06,
+      "loss": 2.3824,
+      "step": 3258
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.391930704929064e-06,
+      "loss": 2.4994,
+      "step": 3260
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.360880800006383e-06,
+      "loss": 2.545,
+      "step": 3262
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.3299687223705745e-06,
+      "loss": 2.3291,
+      "step": 3264
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.299194563372604e-06,
+      "loss": 2.5543,
+      "step": 3266
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2685584139558243e-06,
+      "loss": 2.3818,
+      "step": 3268
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.238060364655765e-06,
+      "loss": 2.6038,
+      "step": 3270
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2077005055998533e-06,
+      "loss": 2.4691,
+      "step": 3272
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.177478926507127e-06,
+      "loss": 2.4399,
+      "step": 3274
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.1473957166879897e-06,
+      "loss": 2.5692,
+      "step": 3276
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.117450965043911e-06,
+      "loss": 2.4183,
+      "step": 3278
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.087644760067232e-06,
+      "loss": 2.4085,
+      "step": 3280
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.0579771898408326e-06,
+      "loss": 2.2894,
+      "step": 3282
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.0284483420379097e-06,
+      "loss": 2.2705,
+      "step": 3284
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9990583039217203e-06,
+      "loss": 2.2714,
+      "step": 3286
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9698071623452895e-06,
+      "loss": 2.366,
+      "step": 3288
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.940695003751198e-06,
+      "loss": 2.3525,
+      "step": 3290
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9117219141712947e-06,
+      "loss": 2.3377,
+      "step": 3292
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.8828879792264675e-06,
+      "loss": 2.1998,
+      "step": 3294
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.854193284126344e-06,
+      "loss": 2.3437,
+      "step": 3296
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.825637913669121e-06,
+      "loss": 2.2963,
+      "step": 3298
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.797221952241219e-06,
+      "loss": 2.3955,
+      "step": 3300
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.7689454838171147e-06,
+      "loss": 2.2326,
+      "step": 3302
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.7408085919590264e-06,
+      "loss": 2.3897,
+      "step": 3304
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.7128113598167137e-06,
+      "loss": 2.4245,
+      "step": 3306
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.684953870127227e-06,
+      "loss": 2.488,
+      "step": 3308
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.657236205214625e-06,
+      "loss": 2.3614,
+      "step": 3310
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.6296584469897743e-06,
+      "loss": 2.1686,
+      "step": 3312
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.6022206769500845e-06,
+      "loss": 2.6152,
+      "step": 3314
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.574922976179295e-06,
+      "loss": 2.3362,
+      "step": 3316
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.547765425347187e-06,
+      "loss": 2.382,
+      "step": 3318
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.520748104709375e-06,
+      "loss": 2.4045,
+      "step": 3320
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.493871094107081e-06,
+      "loss": 2.2771,
+      "step": 3322
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.467134472966892e-06,
+      "loss": 2.3296,
+      "step": 3324
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.4405383203004894e-06,
+      "loss": 2.3129,
+      "step": 3326
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.414082714704463e-06,
+      "loss": 2.2268,
+      "step": 3328
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.3877677343600524e-06,
+      "loss": 2.476,
+      "step": 3330
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.36159345703294e-06,
+      "loss": 2.5804,
+      "step": 3332
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.3355599600729915e-06,
+      "loss": 2.4219,
+      "step": 3334
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.3096673204140108e-06,
+      "loss": 2.4168,
+      "step": 3336
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.2839156145736174e-06,
+      "loss": 2.3116,
+      "step": 3338
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.2583049186528704e-06,
+      "loss": 2.3238,
+      "step": 3340
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.2328353083361562e-06,
+      "loss": 2.4897,
+      "step": 3342
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.207506858890912e-06,
+      "loss": 2.307,
+      "step": 3344
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.182319645167441e-06,
+      "loss": 2.3267,
+      "step": 3346
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.1572737415986422e-06,
+      "loss": 2.424,
+      "step": 3348
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.1323692221998257e-06,
+      "loss": 2.4612,
+      "step": 3350
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.1076061605684818e-06,
+      "loss": 2.4219,
+      "step": 3352
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.0829846298840884e-06,
+      "loss": 2.4251,
+      "step": 3354
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.058504702907843e-06,
+      "loss": 2.5063,
+      "step": 3356
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.0341664519824887e-06,
+      "loss": 2.5947,
+      "step": 3358
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.009969949032098e-06,
+      "loss": 2.4399,
+      "step": 3360
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.9859152655618498e-06,
+      "loss": 2.405,
+      "step": 3362
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.962002472657809e-06,
+      "loss": 2.3689,
+      "step": 3364
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.9382316409867264e-06,
+      "loss": 2.451,
+      "step": 3366
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.9146028407958484e-06,
+      "loss": 2.3364,
+      "step": 3368
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.8911161419126854e-06,
+      "loss": 2.3012,
+      "step": 3370
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.8677716137447954e-06,
+      "loss": 2.3833,
+      "step": 3372
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.844569325279627e-06,
+      "loss": 2.5041,
+      "step": 3374
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.8215093450842435e-06,
+      "loss": 2.2862,
+      "step": 3376
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7985917413052055e-06,
+      "loss": 2.3316,
+      "step": 3378
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7758165816682826e-06,
+      "loss": 2.2273,
+      "step": 3380
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7531839334783306e-06,
+      "loss": 2.2901,
+      "step": 3382
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7306938636190262e-06,
+      "loss": 2.5521,
+      "step": 3384
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7083464385527325e-06,
+      "loss": 2.3896,
+      "step": 3386
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.686141724320245e-06,
+      "loss": 2.2818,
+      "step": 3388
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.6640797865406288e-06,
+      "loss": 2.2733,
+      "step": 3390
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.6421606904110264e-06,
+      "loss": 2.4238,
+      "step": 3392
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.6203845007064455e-06,
+      "loss": 2.6201,
+      "step": 3394
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.5987512817795924e-06,
+      "loss": 2.422,
+      "step": 3396
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.5772610975606561e-06,
+      "loss": 2.2933,
+      "step": 3398
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.5559140115571246e-06,
+      "loss": 2.4014,
+      "step": 3400
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.5347100868536246e-06,
+      "loss": 2.3193,
+      "step": 3402
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.5136493861117097e-06,
+      "loss": 2.4959,
+      "step": 3404
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.4927319715696607e-06,
+      "loss": 2.3566,
+      "step": 3406
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.4719579050423427e-06,
+      "loss": 2.3291,
+      "step": 3408
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.4513272479209917e-06,
+      "loss": 2.3138,
+      "step": 3410
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.43084006117305e-06,
+      "loss": 2.2497,
+      "step": 3412
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.41049640534196e-06,
+      "loss": 2.2461,
+      "step": 3414
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3902963405470148e-06,
+      "loss": 2.3886,
+      "step": 3416
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.37023992648318e-06,
+      "loss": 2.2535,
+      "step": 3418
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3503272224208884e-06,
+      "loss": 2.3367,
+      "step": 3420
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.3305582872058963e-06,
+      "loss": 2.4806,
+      "step": 3422
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.3109331792590773e-06,
+      "loss": 2.4335,
+      "step": 3424
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2914519565763062e-06,
+      "loss": 2.4195,
+      "step": 3426
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2721146767282033e-06,
+      "loss": 2.4332,
+      "step": 3428
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2529213968600406e-06,
+      "loss": 2.2733,
+      "step": 3430
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.233872173691536e-06,
+      "loss": 2.3522,
+      "step": 3432
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2149670635166976e-06,
+      "loss": 2.5166,
+      "step": 3434
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.196206122203647e-06,
+      "loss": 2.4279,
+      "step": 3436
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.1775894051944514e-06,
+      "loss": 2.2575,
+      "step": 3438
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.1591169675049863e-06,
+      "loss": 2.514,
+      "step": 3440
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.140788863724751e-06,
+      "loss": 2.4809,
+      "step": 3442
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.1226051480167032e-06,
+      "loss": 2.444,
+      "step": 3444
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.1045658741171028e-06,
+      "loss": 2.3813,
+      "step": 3446
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.0866710953353731e-06,
+      "loss": 2.3445,
+      "step": 3448
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.068920864553924e-06,
+      "loss": 2.4029,
+      "step": 3450
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.0513152342279842e-06,
+      "loss": 2.4019,
+      "step": 3452
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.0338542563854748e-06,
+      "loss": 2.3407,
+      "step": 3454
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.0165379826268417e-06,
+      "loss": 2.414,
+      "step": 3456
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 9.993664641249012e-07,
+      "loss": 2.4748,
+      "step": 3458
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 9.823397516246834e-07,
+      "loss": 2.2681,
+      "step": 3460
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 9.654578954433059e-07,
+      "loss": 2.3164,
+      "step": 3462
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 9.487209454697887e-07,
+      "loss": 2.382,
+      "step": 3464
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 9.321289511649456e-07,
+      "loss": 2.3299,
+      "step": 3466
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 9.156819615612044e-07,
+      "loss": 2.3526,
+      "step": 3468
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.993800252624862e-07,
+      "loss": 2.4167,
+      "step": 3470
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.832231904440491e-07,
+      "loss": 2.5703,
+      "step": 3472
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.672115048523554e-07,
+      "loss": 2.5794,
+      "step": 3474
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.513450158049108e-07,
+      "loss": 2.3276,
+      "step": 3476
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.356237701901582e-07,
+      "loss": 2.3394,
+      "step": 3478
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.200478144672952e-07,
+      "loss": 2.3505,
+      "step": 3480
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.046171946661796e-07,
+      "loss": 2.4532,
+      "step": 3482
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.893319563871682e-07,
+      "loss": 2.513,
+      "step": 3484
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.741921448009837e-07,
+      "loss": 2.4655,
+      "step": 3486
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.591978046485926e-07,
+      "loss": 2.605,
+      "step": 3488
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.443489802410663e-07,
+      "loss": 2.4451,
+      "step": 3490
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.296457154594482e-07,
+      "loss": 2.5196,
+      "step": 3492
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.150880537546201e-07,
+      "loss": 2.2368,
+      "step": 3494
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.006760381471856e-07,
+      "loss": 2.4034,
+      "step": 3496
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.86409711227337e-07,
+      "loss": 2.5032,
+      "step": 3498
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.722891151547284e-07,
+      "loss": 2.3998,
+      "step": 3500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.583142916583574e-07,
+      "loss": 2.4205,
+      "step": 3502
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.444852820364222e-07,
+      "loss": 2.3034,
+      "step": 3504
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.30802127156227e-07,
+      "loss": 2.3639,
+      "step": 3506
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.172648674540426e-07,
+      "loss": 2.3491,
+      "step": 3508
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.038735429349962e-07,
+      "loss": 2.4925,
+      "step": 3510
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 5.90628193172943e-07,
+      "loss": 2.2371,
+      "step": 3512
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 5.775288573103666e-07,
+      "loss": 2.3831,
+      "step": 3514
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 5.645755740582404e-07,
+      "loss": 2.6221,
+      "step": 3516
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 5.517683816959219e-07,
+      "loss": 2.4171,
+      "step": 3518
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 5.391073180710638e-07,
+      "loss": 2.3518,
+      "step": 3520
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 5.265924205994644e-07,
+      "loss": 2.4263,
+      "step": 3522
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 5.14223726264973e-07,
+      "loss": 2.3707,
+      "step": 3524
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 5.020012716193901e-07,
+      "loss": 2.2659,
+      "step": 3526
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.899250927823396e-07,
+      "loss": 2.3573,
+      "step": 3528
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.779952254411913e-07,
+      "loss": 2.2359,
+      "step": 3530
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.662117048509218e-07,
+      "loss": 2.3461,
+      "step": 3532
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.545745658340206e-07,
+      "loss": 2.4581,
+      "step": 3534
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.4308384278041183e-07,
+      "loss": 2.4515,
+      "step": 3536
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.317395696473214e-07,
+      "loss": 2.4953,
+      "step": 3538
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.2054177995919374e-07,
+      "loss": 2.5276,
+      "step": 3540
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.094905068075694e-07,
+      "loss": 2.323,
+      "step": 3542
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.985857828510353e-07,
+      "loss": 2.4943,
+      "step": 3544
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.878276403150749e-07,
+      "loss": 2.4179,
+      "step": 3546
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.7721611099200693e-07,
+      "loss": 2.3685,
+      "step": 3548
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.6675122624087454e-07,
+      "loss": 2.4998,
+      "step": 3550
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.5643301698736196e-07,
+      "loss": 2.484,
+      "step": 3552
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.462615137237002e-07,
+      "loss": 2.3272,
+      "step": 3554
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.3623674650857806e-07,
+      "loss": 2.3971,
+      "step": 3556
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.2635874496705356e-07,
+      "loss": 2.544,
+      "step": 3558
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.1662753829045375e-07,
+      "loss": 2.6006,
+      "step": 3560
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0704315523631953e-07,
+      "loss": 2.3817,
+      "step": 3562
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.976056241282721e-07,
+      "loss": 2.4897,
+      "step": 3564
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8831497285599085e-07,
+      "loss": 2.3797,
+      "step": 3566
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.7917122887506364e-07,
+      "loss": 2.4418,
+      "step": 3568
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.701744192069755e-07,
+      "loss": 2.4906,
+      "step": 3570
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.613245704389644e-07,
+      "loss": 2.4531,
+      "step": 3572
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5262170872398796e-07,
+      "loss": 2.3002,
+      "step": 3574
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.440658597806178e-07,
+      "loss": 2.5243,
+      "step": 3576
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.3565704889298434e-07,
+      "loss": 2.4181,
+      "step": 3578
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.2739530091069328e-07,
+      "loss": 2.2211,
+      "step": 3580
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.1928064024874796e-07,
+      "loss": 2.0089,
+      "step": 3582
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.113130908874772e-07,
+      "loss": 2.269,
+      "step": 3584
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.0349267637247982e-07,
+      "loss": 2.4197,
+      "step": 3586
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.9581941981453579e-07,
+      "loss": 2.319,
+      "step": 3588
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.8829334388955067e-07,
+      "loss": 2.2629,
+      "step": 3590
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.80914470838478e-07,
+      "loss": 2.5493,
+      "step": 3592
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7368282246726376e-07,
+      "loss": 2.3166,
+      "step": 3594
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.6659842014677406e-07,
+      "loss": 2.5148,
+      "step": 3596
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.596612848127399e-07,
+      "loss": 2.339,
+      "step": 3598
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.5287143696568473e-07,
+      "loss": 2.308,
+      "step": 3600
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.462288966708858e-07,
+      "loss": 2.5463,
+      "step": 3602
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.397336835582741e-07,
+      "loss": 2.6243,
+      "step": 3604
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.333858168224178e-07,
+      "loss": 2.4714,
+      "step": 3606
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2718531522244447e-07,
+      "loss": 2.2877,
+      "step": 3608
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.211321970820023e-07,
+      "loss": 2.3737,
+      "step": 3610
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1522648028917116e-07,
+      "loss": 2.6491,
+      "step": 3612
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0946818229644607e-07,
+      "loss": 2.1288,
+      "step": 3614
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0385732012067607e-07,
+      "loss": 2.4494,
+      "step": 3616
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 9.839391034300316e-08,
+      "loss": 2.5278,
+      "step": 3618
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 9.307796910881794e-08,
+      "loss": 2.2621,
+      "step": 3620
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.790951212771514e-08,
+      "loss": 2.2438,
+      "step": 3622
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.28885546734548e-08,
+      "loss": 2.5771,
+      "step": 3624
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.801511158390118e-08,
+      "loss": 2.1801,
+      "step": 3626
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.328919726097838e-08,
+      "loss": 2.3484,
+      "step": 3628
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.871082567065367e-08,
+      "loss": 2.5062,
+      "step": 3630
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.42800103428598e-08,
+      "loss": 2.2535,
+      "step": 3632
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.999676437148938e-08,
+      "loss": 2.4677,
+      "step": 3634
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.5861100414322796e-08,
+      "loss": 2.2292,
+      "step": 3636
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.1873030693028177e-08,
+      "loss": 2.3609,
+      "step": 3638
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.8032566993089225e-08,
+      "loss": 2.4992,
+      "step": 3640
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.4339720663788555e-08,
+      "loss": 2.5409,
+      "step": 3642
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.079450261817997e-08,
+      "loss": 2.4727,
+      "step": 3644
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.739692333304401e-08,
+      "loss": 2.2859,
+      "step": 3646
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.4146992848854695e-08,
+      "loss": 2.3062,
+      "step": 3648
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.104472076976839e-08,
+      "loss": 2.3304,
+      "step": 3650
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.809011626357383e-08,
+      "loss": 2.4489,
+      "step": 3652
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.528318806168106e-08,
+      "loss": 2.3078,
+      "step": 3654
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2623944459082557e-08,
+      "loss": 2.3173,
+      "step": 3656
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.0112393314336565e-08,
+      "loss": 2.3973,
+      "step": 3658
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.7748542049550453e-08,
+      "loss": 2.4975,
+      "step": 3660
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.553239765034187e-08,
+      "loss": 2.4494,
+      "step": 3662
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.346396666582761e-08,
+      "loss": 2.4442,
+      "step": 3664
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.1543255208612546e-08,
+      "loss": 2.4622,
+      "step": 3666
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 9.770268954756301e-09,
+      "loss": 2.3676,
+      "step": 3668
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.145013143756597e-09,
+      "loss": 2.5927,
+      "step": 3670
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.6674925785548125e-09,
+      "loss": 2.4922,
+      "step": 3672
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.337711625497121e-09,
+      "loss": 2.1202,
+      "step": 3674
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.155674214328942e-09,
+      "loss": 2.5643,
+      "step": 3676
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.1213838382004867e-09,
+      "loss": 2.2801,
+      "step": 3678
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.234843553627908e-09,
+      "loss": 2.4424,
+      "step": 3680
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.496055980498845e-09,
+      "loss": 2.6128,
+      "step": 3682
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 9.050233020779786e-10,
+      "loss": 2.3174,
+      "step": 3684
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.6174726496817087e-10,
+      "loss": 2.4364,
+      "step": 3686
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.6622917913267088e-10,
+      "loss": 2.5562,
+      "step": 3688
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.8469917889563094e-11,
+      "loss": 2.2758,
+      "step": 3690
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.8469917889563094e-11,
+      "loss": 2.4287,
+      "step": 3692
+    },
+    {
+      "epoch": 4.0,
+      "step": 3692,
+      "total_flos": 8.7881966778581e+16,
+      "train_loss": 2.440685138283933,
+      "train_runtime": 22362.2858,
+      "train_samples_per_second": 10.572,
+      "train_steps_per_second": 0.165
+    }
+  ],
+  "max_steps": 3692,
+  "num_train_epochs": 4,
+  "total_flos": 8.7881966778581e+16,
+  "trial_name": null,
+  "trial_params": null
+}