diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7333 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9972316210396803,
+  "global_step": 1218,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.405405405405406e-07,
+      "loss": 1.9685,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0810810810810812e-06,
+      "loss": 2.0077,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6216216216216219e-06,
+      "loss": 1.7185,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1621621621621623e-06,
+      "loss": 1.5678,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.702702702702703e-06,
+      "loss": 1.6993,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2432432432432437e-06,
+      "loss": 1.7094,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7837837837837844e-06,
+      "loss": 1.6745,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.324324324324325e-06,
+      "loss": 1.7161,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.864864864864866e-06,
+      "loss": 1.5671,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 1.5681,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.945945945945947e-06,
+      "loss": 1.6126,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.486486486486487e-06,
+      "loss": 1.7084,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.027027027027028e-06,
+      "loss": 1.5979,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.567567567567569e-06,
+      "loss": 1.5169,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.108108108108109e-06,
+      "loss": 1.529,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.64864864864865e-06,
+      "loss": 1.5377,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.189189189189191e-06,
+      "loss": 1.6252,
+      "step": 17
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.729729729729732e-06,
+      "loss": 1.5163,
+      "step": 18
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.027027027027027e-05,
+      "loss": 1.535,
+      "step": 19
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 1.5056,
+      "step": 20
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1351351351351352e-05,
+      "loss": 1.5673,
+      "step": 21
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1891891891891894e-05,
+      "loss": 1.5766,
+      "step": 22
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2432432432432433e-05,
+      "loss": 1.5272,
+      "step": 23
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2972972972972975e-05,
+      "loss": 1.551,
+      "step": 24
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3513513513513515e-05,
+      "loss": 1.6771,
+      "step": 25
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4054054054054055e-05,
+      "loss": 1.5944,
+      "step": 26
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4594594594594596e-05,
+      "loss": 1.6029,
+      "step": 27
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5135135135135138e-05,
+      "loss": 1.5741,
+      "step": 28
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5675675675675676e-05,
+      "loss": 1.5831,
+      "step": 29
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 1.5362,
+      "step": 30
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6756756756756757e-05,
+      "loss": 1.5485,
+      "step": 31
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.72972972972973e-05,
+      "loss": 1.6209,
+      "step": 32
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.783783783783784e-05,
+      "loss": 1.505,
+      "step": 33
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8378378378378383e-05,
+      "loss": 1.5399,
+      "step": 34
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.891891891891892e-05,
+      "loss": 1.497,
+      "step": 35
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9459459459459463e-05,
+      "loss": 1.6058,
+      "step": 36
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 1.5012,
+      "step": 37
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999996461903301e-05,
+      "loss": 1.5813,
+      "step": 38
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999858476382388e-05,
+      "loss": 1.5735,
+      "step": 39
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999681572799226e-05,
+      "loss": 1.4149,
+      "step": 40
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999433909535333e-05,
+      "loss": 1.542,
+      "step": 41
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999115488343213e-05,
+      "loss": 1.5271,
+      "step": 42
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999872631147608e-05,
+      "loss": 1.4501,
+      "step": 43
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999826638168783e-05,
+      "loss": 1.4914,
+      "step": 44
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997735702233006e-05,
+      "loss": 1.4565,
+      "step": 45
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99971342768668e-05,
+      "loss": 1.4538,
+      "step": 46
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999646210984502e-05,
+      "loss": 1.552,
+      "step": 47
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999571920592405e-05,
+      "loss": 1.5556,
+      "step": 48
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994905570360817e-05,
+      "loss": 1.5317,
+      "step": 49
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999402120891276e-05,
+      "loss": 1.4997,
+      "step": 50
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99930661278378e-05,
+      "loss": 1.4592,
+      "step": 51
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992040333894273e-05,
+      "loss": 1.4862,
+      "step": 52
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990943834340893e-05,
+      "loss": 1.6162,
+      "step": 53
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989776636936705e-05,
+      "loss": 1.5054,
+      "step": 54
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988538749941024e-05,
+      "loss": 1.5535,
+      "step": 55
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987230182113374e-05,
+      "loss": 1.5145,
+      "step": 56
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998585094271344e-05,
+      "loss": 1.4675,
+      "step": 57
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998440104150098e-05,
+      "loss": 1.5726,
+      "step": 58
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998288048873578e-05,
+      "loss": 1.5489,
+      "step": 59
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981289295177566e-05,
+      "loss": 1.5856,
+      "step": 60
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979627472085927e-05,
+      "loss": 1.4966,
+      "step": 61
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997789503122025e-05,
+      "loss": 1.5644,
+      "step": 62
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9976091984839616e-05,
+      "loss": 1.49,
+      "step": 63
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974218345702733e-05,
+      "loss": 1.561,
+      "step": 64
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972274127067838e-05,
+      "loss": 1.5704,
+      "step": 65
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997025934269259e-05,
+      "loss": 1.5434,
+      "step": 66
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968174006833996e-05,
+      "loss": 1.488,
+      "step": 67
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966018134248296e-05,
+      "loss": 1.5842,
+      "step": 68
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963791740190863e-05,
+      "loss": 1.5373,
+      "step": 69
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996149484041609e-05,
+      "loss": 1.436,
+      "step": 70
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959127451177287e-05,
+      "loss": 1.5963,
+      "step": 71
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956689589226555e-05,
+      "loss": 1.5238,
+      "step": 72
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954181271814673e-05,
+      "loss": 1.4124,
+      "step": 73
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951602516690988e-05,
+      "loss": 1.4398,
+      "step": 74
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948953342103268e-05,
+      "loss": 1.4836,
+      "step": 75
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994623376679758e-05,
+      "loss": 1.5311,
+      "step": 76
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943443810018174e-05,
+      "loss": 1.4614,
+      "step": 77
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940583491507314e-05,
+      "loss": 1.5746,
+      "step": 78
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993765283150517e-05,
+      "loss": 1.4914,
+      "step": 79
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934651850749663e-05,
+      "loss": 1.5658,
+      "step": 80
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931580570476306e-05,
+      "loss": 1.5893,
+      "step": 81
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928439012418076e-05,
+      "loss": 1.5762,
+      "step": 82
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925227198805247e-05,
+      "loss": 1.4619,
+      "step": 83
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921945152365235e-05,
+      "loss": 1.6507,
+      "step": 84
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918592896322432e-05,
+      "loss": 1.5032,
+      "step": 85
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915170454398045e-05,
+      "loss": 1.5206,
+      "step": 86
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911677850809943e-05,
+      "loss": 1.5568,
+      "step": 87
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908115110272463e-05,
+      "loss": 1.5528,
+      "step": 88
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904482257996244e-05,
+      "loss": 1.5943,
+      "step": 89
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990077931968805e-05,
+      "loss": 1.5335,
+      "step": 90
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897006321550592e-05,
+      "loss": 1.5123,
+      "step": 91
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893163290282335e-05,
+      "loss": 1.442,
+      "step": 92
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889250253077306e-05,
+      "loss": 1.5288,
+      "step": 93
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885267237624923e-05,
+      "loss": 1.5926,
+      "step": 94
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988121427210976e-05,
+      "loss": 1.6138,
+      "step": 95
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98770913852114e-05,
+      "loss": 1.4996,
+      "step": 96
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872898606104175e-05,
+      "loss": 1.5081,
+      "step": 97
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868635964457007e-05,
+      "loss": 1.5397,
+      "step": 98
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986430349043317e-05,
+      "loss": 1.5946,
+      "step": 99
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859901214690094e-05,
+      "loss": 1.5384,
+      "step": 100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855429168379127e-05,
+      "loss": 1.5845,
+      "step": 101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985088738314534e-05,
+      "loss": 1.4658,
+      "step": 102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846275891127275e-05,
+      "loss": 1.5455,
+      "step": 103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841594724956746e-05,
+      "loss": 1.5214,
+      "step": 104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836843917758593e-05,
+      "loss": 1.4254,
+      "step": 105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983202350315044e-05,
+      "loss": 1.5212,
+      "step": 106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982713351524248e-05,
+      "loss": 1.5146,
+      "step": 107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982217398863721e-05,
+      "loss": 1.4277,
+      "step": 108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98171449584292e-05,
+      "loss": 1.425,
+      "step": 109
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812046460204837e-05,
+      "loss": 1.5144,
+      "step": 110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806878530042083e-05,
+      "loss": 1.6131,
+      "step": 111
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801641204510216e-05,
+      "loss": 1.5409,
+      "step": 112
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796334520669555e-05,
+      "loss": 1.4707,
+      "step": 113
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790958516071228e-05,
+      "loss": 1.537,
+      "step": 114
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978551322875688e-05,
+      "loss": 1.4324,
+      "step": 115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977999869725842e-05,
+      "loss": 1.4989,
+      "step": 116
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977441496059774e-05,
+      "loss": 1.4105,
+      "step": 117
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768762058286433e-05,
+      "loss": 1.4307,
+      "step": 118
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976304003032554e-05,
+      "loss": 1.4543,
+      "step": 119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9757248917205228e-05,
+      "loss": 1.4607,
+      "step": 120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975138875990454e-05,
+      "loss": 1.6315,
+      "step": 121
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974545959989108e-05,
+      "loss": 1.4523,
+      "step": 122
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739461479120727e-05,
+      "loss": 1.543,
+      "step": 123
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973339444003735e-05,
+      "loss": 1.4478,
+      "step": 124
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727258525572487e-05,
+      "loss": 1.4982,
+      "step": 125
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721053779145057e-05,
+      "loss": 1.487,
+      "step": 126
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714780244661044e-05,
+      "loss": 1.5652,
+      "step": 127
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708437966513196e-05,
+      "loss": 1.4799,
+      "step": 128
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702026989580694e-05,
+      "loss": 1.5135,
+      "step": 129
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969554735922885e-05,
+      "loss": 1.4226,
+      "step": 130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968899912130879e-05,
+      "loss": 1.5302,
+      "step": 131
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682382322157103e-05,
+      "loss": 1.4938,
+      "step": 132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675697008595545e-05,
+      "loss": 1.3944,
+      "step": 133
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668943227930686e-05,
+      "loss": 1.5336,
+      "step": 134
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966212102795358e-05,
+      "loss": 1.4832,
+      "step": 135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965523045693944e-05,
+      "loss": 1.4443,
+      "step": 136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964827156364728e-05,
+      "loss": 1.5658,
+      "step": 137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964124439731957e-05,
+      "loss": 1.4998,
+      "step": 138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634149007681894e-05,
+      "loss": 1.5682,
+      "step": 139
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962698544494261e-05,
+      "loss": 1.4023,
+      "step": 140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619753759792466e-05,
+      "loss": 1.4763,
+      "step": 141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961245400340427e-05,
+      "loss": 1.4863,
+      "step": 142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605086227432512e-05,
+      "loss": 1.5116,
+      "step": 143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597650484012997e-05,
+      "loss": 1.5353,
+      "step": 144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9590146825762476e-05,
+      "loss": 1.4939,
+      "step": 145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9582575305778297e-05,
+      "loss": 1.4486,
+      "step": 146
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9574935977637994e-05,
+      "loss": 1.454,
+      "step": 147
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9567228895398936e-05,
+      "loss": 1.5772,
+      "step": 148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955945411359792e-05,
+      "loss": 1.5268,
+      "step": 149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9551611687250808e-05,
+      "loss": 1.4294,
+      "step": 150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543701671852127e-05,
+      "loss": 1.4364,
+      "step": 151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535724123374674e-05,
+      "loss": 1.4264,
+      "step": 152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952767909826913e-05,
+      "loss": 1.5548,
+      "step": 153
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951956665346364e-05,
+      "loss": 1.4417,
+      "step": 154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951138684636344e-05,
+      "loss": 1.4987,
+      "step": 155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9503139734850426e-05,
+      "loss": 1.5214,
+      "step": 156
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9494825377282746e-05,
+      "loss": 1.4836,
+      "step": 157
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486443832494414e-05,
+      "loss": 1.4791,
+      "step": 158
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477995159794854e-05,
+      "loss": 1.5139,
+      "step": 159
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9469479418968506e-05,
+      "loss": 1.4858,
+      "step": 160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9460896670274408e-05,
+      "loss": 1.3027,
+      "step": 161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9452246974445743e-05,
+      "loss": 1.5751,
+      "step": 162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443530392689434e-05,
+      "loss": 1.4963,
+      "step": 163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94347469866857e-05,
+      "loss": 1.5052,
+      "step": 164
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9425896818587615e-05,
+      "loss": 1.4721,
+      "step": 165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941697995102069e-05,
+      "loss": 1.5447,
+      "step": 166
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9407996447082394e-05,
+      "loss": 1.5826,
+      "step": 167
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939894637034174e-05,
+      "loss": 1.538,
+      "step": 168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9389829784838833e-05,
+      "loss": 1.4421,
+      "step": 169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938064675508438e-05,
+      "loss": 1.3979,
+      "step": 170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9371397346059286e-05,
+      "loss": 1.5234,
+      "step": 171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936208162321415e-05,
+      "loss": 1.4172,
+      "step": 172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352699652468835e-05,
+      "loss": 1.464,
+      "step": 173
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9343251500211977e-05,
+      "loss": 1.4958,
+      "step": 174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933373723330053e-05,
+      "loss": 1.4263,
+      "step": 175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9324156919059286e-05,
+      "loss": 1.5034,
+      "step": 176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.93145106252804e-05,
+      "loss": 1.5243,
+      "step": 177
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9304798420222918e-05,
+      "loss": 1.4999,
+      "step": 178
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9295020372612276e-05,
+      "loss": 1.536,
+      "step": 179
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9285176551639826e-05,
+      "loss": 1.5425,
+      "step": 180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9275267026962358e-05,
+      "loss": 1.6061,
+      "step": 181
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9265291868701584e-05,
+      "loss": 1.4502,
+      "step": 182
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9255251147443646e-05,
+      "loss": 1.567,
+      "step": 183
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924514493423864e-05,
+      "loss": 1.4235,
+      "step": 184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9234973300600074e-05,
+      "loss": 1.4931,
+      "step": 185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92247363185044e-05,
+      "loss": 1.503,
+      "step": 186
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9214434060390484e-05,
+      "loss": 1.5545,
+      "step": 187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9204066599159094e-05,
+      "loss": 1.4933,
+      "step": 188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9193634008172396e-05,
+      "loss": 1.5112,
+      "step": 189
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9183136361253417e-05,
+      "loss": 1.4536,
+      "step": 190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917257373268554e-05,
+      "loss": 1.4638,
+      "step": 191
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916194619721196e-05,
+      "loss": 1.4806,
+      "step": 192
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915125383003518e-05,
+      "loss": 1.4791,
+      "step": 193
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914049670681646e-05,
+      "loss": 1.5478,
+      "step": 194
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912967490367528e-05,
+      "loss": 1.4017,
+      "step": 195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9118788497188815e-05,
+      "loss": 1.3816,
+      "step": 196
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9107837564391376e-05,
+      "loss": 1.4484,
+      "step": 197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9096822182773887e-05,
+      "loss": 1.3673,
+      "step": 198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9085742430283322e-05,
+      "loss": 1.4965,
+      "step": 199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907459838532215e-05,
+      "loss": 1.5041,
+      "step": 200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9063390126747778e-05,
+      "loss": 1.5221,
+      "step": 201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9052117733872025e-05,
+      "loss": 1.4985,
+      "step": 202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904078128646052e-05,
+      "loss": 1.5436,
+      "step": 203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902938086473215e-05,
+      "loss": 1.5542,
+      "step": 204
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901791654935852e-05,
+      "loss": 1.4878,
+      "step": 205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9006388421463322e-05,
+      "loss": 1.4366,
+      "step": 206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899479656262183e-05,
+      "loss": 1.5619,
+      "step": 207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898314105486028e-05,
+      "loss": 1.4125,
+      "step": 208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8971421980655295e-05,
+      "loss": 1.4888,
+      "step": 209
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8959639422933316e-05,
+      "loss": 1.4248,
+      "step": 210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894779346506999e-05,
+      "loss": 1.4671,
+      "step": 211
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893588419088962e-05,
+      "loss": 1.4576,
+      "step": 212
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892391168466452e-05,
+      "loss": 1.5892,
+      "step": 213
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891187603111447e-05,
+      "loss": 1.6097,
+      "step": 214
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8899777315406073e-05,
+      "loss": 1.428,
+      "step": 215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8887615623152188e-05,
+      "loss": 1.4905,
+      "step": 216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88753910404113e-05,
+      "loss": 1.5599,
+      "step": 217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8863103653686917e-05,
+      "loss": 1.3781,
+      "step": 218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8850753549926967e-05,
+      "loss": 1.427,
+      "step": 219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8838340816523175e-05,
+      "loss": 1.5079,
+      "step": 220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8825865541310438e-05,
+      "loss": 1.5375,
+      "step": 221
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8813327812566217e-05,
+      "loss": 1.5392,
+      "step": 222
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880072771900991e-05,
+      "loss": 1.4198,
+      "step": 223
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878806534980221e-05,
+      "loss": 1.5668,
+      "step": 224
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8775340794544497e-05,
+      "loss": 1.4315,
+      "step": 225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876255414327818e-05,
+      "loss": 1.5012,
+      "step": 226
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8749705486484074e-05,
+      "loss": 1.5996,
+      "step": 227
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8736794915081765e-05,
+      "loss": 1.4609,
+      "step": 228
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8723822520428954e-05,
+      "loss": 1.4306,
+      "step": 229
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8710788394320807e-05,
+      "loss": 1.3959,
+      "step": 230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8697692628989327e-05,
+      "loss": 1.4551,
+      "step": 231
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868453531710268e-05,
+      "loss": 1.4901,
+      "step": 232
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8671316551764552e-05,
+      "loss": 1.4911,
+      "step": 233
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865803642651348e-05,
+      "loss": 1.509,
+      "step": 234
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8644695035322203e-05,
+      "loss": 1.5201,
+      "step": 235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8631292472596978e-05,
+      "loss": 1.5627,
+      "step": 236
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8617828833176935e-05,
+      "loss": 1.4702,
+      "step": 237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860430421233339e-05,
+      "loss": 1.4672,
+      "step": 238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859071870576918e-05,
+      "loss": 1.4023,
+      "step": 239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857707240961797e-05,
+      "loss": 1.4423,
+      "step": 240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8563365420443594e-05,
+      "loss": 1.4925,
+      "step": 241
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854959783523936e-05,
+      "loss": 1.4703,
+      "step": 242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853576975142736e-05,
+      "loss": 1.499,
+      "step": 243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852188126685779e-05,
+      "loss": 1.4785,
+      "step": 244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8507932479808254e-05,
+      "loss": 1.3313,
+      "step": 245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8493923488983066e-05,
+      "loss": 1.4063,
+      "step": 246
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847985439351256e-05,
+      "loss": 1.5201,
+      "step": 247
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846572529295237e-05,
+      "loss": 1.5639,
+      "step": 248
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845153628728274e-05,
+      "loss": 1.556,
+      "step": 249
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8437287476907828e-05,
+      "loss": 1.5443,
+      "step": 250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842297896265497e-05,
+      "loss": 1.394,
+      "step": 251
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8408610845773974e-05,
+      "loss": 1.5062,
+      "step": 252
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8394183227936418e-05,
+      "loss": 1.4301,
+      "step": 253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8379696211234918e-05,
+      "loss": 1.5224,
+      "step": 254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8365149898182403e-05,
+      "loss": 1.4614,
+      "step": 255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8350544391711396e-05,
+      "loss": 1.4176,
+      "step": 256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833587979517329e-05,
+      "loss": 1.4349,
+      "step": 257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8321156212337604e-05,
+      "loss": 1.438,
+      "step": 258
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830637374739126e-05,
+      "loss": 1.5304,
+      "step": 259
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829153250493783e-05,
+      "loss": 1.3957,
+      "step": 260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827663258999683e-05,
+      "loss": 1.4895,
+      "step": 261
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8261674108002925e-05,
+      "loss": 1.4291,
+      "step": 262
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824665716480524e-05,
+      "loss": 1.4376,
+      "step": 263
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823158186666656e-05,
+      "loss": 1.4583,
+      "step": 264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821644832026261e-05,
+      "loss": 1.5522,
+      "step": 265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82012566326813e-05,
+      "loss": 1.5009,
+      "step": 266
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8186006911421937e-05,
+      "loss": 1.5268,
+      "step": 267
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817069926439451e-05,
+      "loss": 1.5403,
+      "step": 268
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8155333799918883e-05,
+      "loss": 1.5086,
+      "step": 269
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8139910626724058e-05,
+      "loss": 1.4217,
+      "step": 270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8124429853947387e-05,
+      "loss": 1.5121,
+      "step": 271
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8108891591133812e-05,
+      "loss": 1.4745,
+      "step": 272
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809329594823509e-05,
+      "loss": 1.4478,
+      "step": 273
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8077643035609006e-05,
+      "loss": 1.5936,
+      "step": 274
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806193296401859e-05,
+      "loss": 1.434,
+      "step": 275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804616584463136e-05,
+      "loss": 1.5192,
+      "step": 276
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803034178901849e-05,
+      "loss": 1.4611,
+      "step": 277
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8014460909154058e-05,
+      "loss": 1.4463,
+      "step": 278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799852331741425e-05,
+      "loss": 1.5376,
+      "step": 279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7982529126576543e-05,
+      "loss": 1.5143,
+      "step": 280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7966478449818925e-05,
+      "loss": 1.5489,
+      "step": 281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7950371400719087e-05,
+      "loss": 1.5324,
+      "step": 282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7934208093253625e-05,
+      "loss": 1.4892,
+      "step": 283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7917988641797227e-05,
+      "loss": 1.5164,
+      "step": 284
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7901713161121873e-05,
+      "loss": 1.4834,
+      "step": 285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7885381766396008e-05,
+      "loss": 1.418,
+      "step": 286
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786899457318374e-05,
+      "loss": 1.4374,
+      "step": 287
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7852551697444017e-05,
+      "loss": 1.4375,
+      "step": 288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783605325552981e-05,
+      "loss": 1.4484,
+      "step": 289
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7819499364187282e-05,
+      "loss": 1.507,
+      "step": 290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780289014055497e-05,
+      "loss": 1.3947,
+      "step": 291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7786225702162955e-05,
+      "loss": 1.4528,
+      "step": 292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7769506166932026e-05,
+      "loss": 1.5722,
+      "step": 293
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7752731653172847e-05,
+      "loss": 1.4561,
+      "step": 294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7735902279585118e-05,
+      "loss": 1.4411,
+      "step": 295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7719018165256745e-05,
+      "loss": 1.4508,
+      "step": 296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7702079429662986e-05,
+      "loss": 1.5283,
+      "step": 297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7685086192665605e-05,
+      "loss": 1.2922,
+      "step": 298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7668038574512045e-05,
+      "loss": 1.4048,
+      "step": 299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7650936695834536e-05,
+      "loss": 1.4329,
+      "step": 300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763378067764929e-05,
+      "loss": 1.5082,
+      "step": 301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7616570641355602e-05,
+      "loss": 1.3541,
+      "step": 302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759930670873502e-05,
+      "loss": 1.4544,
+      "step": 303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758198900195047e-05,
+      "loss": 1.557,
+      "step": 304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7564617643545395e-05,
+      "loss": 1.4158,
+      "step": 305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7547192756442887e-05,
+      "loss": 1.4912,
+      "step": 306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7529714463944815e-05,
+      "loss": 1.4887,
+      "step": 307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751218288973096e-05,
+      "loss": 1.5082,
+      "step": 308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7494598157858127e-05,
+      "loss": 1.4374,
+      "step": 309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7476960392759284e-05,
+      "loss": 1.4352,
+      "step": 310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7459269719242665e-05,
+      "loss": 1.3782,
+      "step": 311
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74415262624909e-05,
+      "loss": 1.5327,
+      "step": 312
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742373014806012e-05,
+      "loss": 1.5196,
+      "step": 313
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740588150187907e-05,
+      "loss": 1.5719,
+      "step": 314
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7387980450248222e-05,
+      "loss": 1.5836,
+      "step": 315
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7370027119838884e-05,
+      "loss": 1.4612,
+      "step": 316
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735202163769229e-05,
+      "loss": 1.5138,
+      "step": 317
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7333964131218714e-05,
+      "loss": 1.46,
+      "step": 318
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7315854728196568e-05,
+      "loss": 1.5212,
+      "step": 319
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729769355677149e-05,
+      "loss": 1.4436,
+      "step": 320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7279480745455433e-05,
+      "loss": 1.446,
+      "step": 321
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7261216423125782e-05,
+      "loss": 1.4478,
+      "step": 322
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724290071902441e-05,
+      "loss": 1.5164,
+      "step": 323
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7224533762756775e-05,
+      "loss": 1.5921,
+      "step": 324
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720611568429103e-05,
+      "loss": 1.3207,
+      "step": 325
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718764661395704e-05,
+      "loss": 1.4776,
+      "step": 326
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.716912668244553e-05,
+      "loss": 1.4853,
+      "step": 327
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715055602080711e-05,
+      "loss": 1.4779,
+      "step": 328
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7131934760451385e-05,
+      "loss": 1.3848,
+      "step": 329
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7113263033145985e-05,
+      "loss": 1.5338,
+      "step": 330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7094540971015663e-05,
+      "loss": 1.335,
+      "step": 331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7075768706541355e-05,
+      "loss": 1.4512,
+      "step": 332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7056946372559234e-05,
+      "loss": 1.3966,
+      "step": 333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7038074102259775e-05,
+      "loss": 1.5724,
+      "step": 334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7019152029186817e-05,
+      "loss": 1.3596,
+      "step": 335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.70001802872366e-05,
+      "loss": 1.4671,
+      "step": 336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6981159010656847e-05,
+      "loss": 1.484,
+      "step": 337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6962088334045785e-05,
+      "loss": 1.5372,
+      "step": 338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694296839235121e-05,
+      "loss": 1.4365,
+      "step": 339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692379932086953e-05,
+      "loss": 1.4995,
+      "step": 340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6904581255244802e-05,
+      "loss": 1.4586,
+      "step": 341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688531433146777e-05,
+      "loss": 1.4517,
+      "step": 342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6865998685874923e-05,
+      "loss": 1.513,
+      "step": 343
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6846634455147498e-05,
+      "loss": 1.4483,
+      "step": 344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6827221776310532e-05,
+      "loss": 1.3291,
+      "step": 345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6807760786731905e-05,
+      "loss": 1.3948,
+      "step": 346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6788251624121335e-05,
+      "loss": 1.536,
+      "step": 347
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6768694426529432e-05,
+      "loss": 1.4636,
+      "step": 348
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6749089332346714e-05,
+      "loss": 1.4605,
+      "step": 349
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672943648030261e-05,
+      "loss": 1.4914,
+      "step": 350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6709736009464504e-05,
+      "loss": 1.4608,
+      "step": 351
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668998805923675e-05,
+      "loss": 1.5315,
+      "step": 352
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6670192769359643e-05,
+      "loss": 1.4783,
+      "step": 353
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6650350279908497e-05,
+      "loss": 1.5233,
+      "step": 354
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6630460731292597e-05,
+      "loss": 1.4558,
+      "step": 355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661052426425424e-05,
+      "loss": 1.4945,
+      "step": 356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6590541019867722e-05,
+      "loss": 1.5253,
+      "step": 357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6570511139538348e-05,
+      "loss": 1.3644,
+      "step": 358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655043476500142e-05,
+      "loss": 1.3728,
+      "step": 359
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6530312038321247e-05,
+      "loss": 1.5508,
+      "step": 360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6510143101890136e-05,
+      "loss": 1.3814,
+      "step": 361
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6489928098427383e-05,
+      "loss": 1.51,
+      "step": 362
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6469667170978258e-05,
+      "loss": 1.3647,
+      "step": 363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6449360462913005e-05,
+      "loss": 1.4148,
+      "step": 364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642900811792582e-05,
+      "loss": 1.5278,
+      "step": 365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640861028003383e-05,
+      "loss": 1.4379,
+      "step": 366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6388167093576083e-05,
+      "loss": 1.4898,
+      "step": 367
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6367678703212515e-05,
+      "loss": 1.4376,
+      "step": 368
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6347145253922942e-05,
+      "loss": 1.5371,
+      "step": 369
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632656689100602e-05,
+      "loss": 1.4784,
+      "step": 370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6305943760078226e-05,
+      "loss": 1.4778,
+      "step": 371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.628527600707283e-05,
+      "loss": 1.4496,
+      "step": 372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6264563778238834e-05,
+      "loss": 1.4265,
+      "step": 373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6243807220139988e-05,
+      "loss": 1.4144,
+      "step": 374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6223006479653708e-05,
+      "loss": 1.3847,
+      "step": 375
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6202161703970057e-05,
+      "loss": 1.4024,
+      "step": 376
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6181273040590696e-05,
+      "loss": 1.4366,
+      "step": 377
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616034063732785e-05,
+      "loss": 1.3984,
+      "step": 378
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613936464230325e-05,
+      "loss": 1.4135,
+      "step": 379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6118345203947093e-05,
+      "loss": 1.5571,
+      "step": 380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6097282470996997e-05,
+      "loss": 1.4876,
+      "step": 381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6076176592496926e-05,
+      "loss": 1.3847,
+      "step": 382
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605502771779616e-05,
+      "loss": 1.4314,
+      "step": 383
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603383599654823e-05,
+      "loss": 1.4177,
+      "step": 384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601260157870985e-05,
+      "loss": 1.4774,
+      "step": 385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599132461453987e-05,
+      "loss": 1.4068,
+      "step": 386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5970005254598204e-05,
+      "loss": 1.4833,
+      "step": 387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.594864364974476e-05,
+      "loss": 1.4635,
+      "step": 388
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592723995113839e-05,
+      "loss": 1.4766,
+      "step": 389
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5905794310235808e-05,
+      "loss": 1.4363,
+      "step": 390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5884306878790512e-05,
+      "loss": 1.4122,
+      "step": 391
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586277780885172e-05,
+      "loss": 1.4629,
+      "step": 392
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58412072527633e-05,
+      "loss": 1.5554,
+      "step": 393
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5819595363162682e-05,
+      "loss": 1.4584,
+      "step": 394
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5797942292979767e-05,
+      "loss": 1.4806,
+      "step": 395
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577624819543587e-05,
+      "loss": 1.4799,
+      "step": 396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5754513224042625e-05,
+      "loss": 1.5323,
+      "step": 397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573273753260089e-05,
+      "loss": 1.5143,
+      "step": 398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571092127519967e-05,
+      "loss": 1.3724,
+      "step": 399
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568906460621502e-05,
+      "loss": 1.4376,
+      "step": 400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566716768030896e-05,
+      "loss": 1.516,
+      "step": 401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5645230652428367e-05,
+      "loss": 1.3851,
+      "step": 402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5623253677803897e-05,
+      "loss": 1.4896,
+      "step": 403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5601236911948876e-05,
+      "loss": 1.4985,
+      "step": 404
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5579180510658187e-05,
+      "loss": 1.4713,
+      "step": 405
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5557084630007206e-05,
+      "loss": 1.5704,
+      "step": 406
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5534949426350642e-05,
+      "loss": 1.1073,
+      "step": 407
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551277505632149e-05,
+      "loss": 1.0673,
+      "step": 408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549056167682987e-05,
+      "loss": 1.1634,
+      "step": 409
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546830944506196e-05,
+      "loss": 1.1615,
+      "step": 410
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544601851847885e-05,
+      "loss": 1.0975,
+      "step": 411
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542368905481545e-05,
+      "loss": 1.1318,
+      "step": 412
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5401321212079366e-05,
+      "loss": 0.9377,
+      "step": 413
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5378915148549772e-05,
+      "loss": 1.0874,
+      "step": 414
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5356471022776315e-05,
+      "loss": 1.0396,
+      "step": 415
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5333988993577958e-05,
+      "loss": 1.0781,
+      "step": 416
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5311469220041903e-05,
+      "loss": 1.0133,
+      "step": 417
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5288911861522413e-05,
+      "loss": 1.0539,
+      "step": 418
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526631707763972e-05,
+      "loss": 1.2074,
+      "step": 419
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5243685028278888e-05,
+      "loss": 1.1373,
+      "step": 420
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5221015873588672e-05,
+      "loss": 0.9648,
+      "step": 421
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5198309773980397e-05,
+      "loss": 1.0255,
+      "step": 422
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5175566890126812e-05,
+      "loss": 1.0889,
+      "step": 423
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5152787382960968e-05,
+      "loss": 1.0913,
+      "step": 424
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5129971413675055e-05,
+      "loss": 1.0329,
+      "step": 425
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.510711914371929e-05,
+      "loss": 1.0569,
+      "step": 426
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5084230734800754e-05,
+      "loss": 1.0238,
+      "step": 427
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5061306348882252e-05,
+      "loss": 1.0904,
+      "step": 428
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5038346148181178e-05,
+      "loss": 1.1452,
+      "step": 429
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5015350295168344e-05,
+      "loss": 1.0408,
+      "step": 430
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4992318952566862e-05,
+      "loss": 1.1252,
+      "step": 431
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4969252283350964e-05,
+      "loss": 0.9937,
+      "step": 432
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4946150450744859e-05,
+      "loss": 1.0751,
+      "step": 433
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4923013618221584e-05,
+      "loss": 1.1261,
+      "step": 434
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4899841949501845e-05,
+      "loss": 1.1045,
+      "step": 435
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4876635608552845e-05,
+      "loss": 1.029,
+      "step": 436
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4853394759587146e-05,
+      "loss": 1.1211,
+      "step": 437
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4830119567061484e-05,
+      "loss": 1.0406,
+      "step": 438
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4806810195675627e-05,
+      "loss": 1.0517,
+      "step": 439
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4783466810371195e-05,
+      "loss": 1.0026,
+      "step": 440
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4760089576330493e-05,
+      "loss": 1.0275,
+      "step": 441
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4736678658975357e-05,
+      "loss": 1.0359,
+      "step": 442
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.471323422396596e-05,
+      "loss": 1.164,
+      "step": 443
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4689756437199658e-05,
+      "loss": 1.1032,
+      "step": 444
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4666245464809818e-05,
+      "loss": 1.0966,
+      "step": 445
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4642701473164618e-05,
+      "loss": 1.1152,
+      "step": 446
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4619124628865904e-05,
+      "loss": 1.0731,
+      "step": 447
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.459551509874798e-05,
+      "loss": 1.0371,
+      "step": 448
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4571873049876452e-05,
+      "loss": 1.0888,
+      "step": 449
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.454819864954703e-05,
+      "loss": 1.0688,
+      "step": 450
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4524492065284344e-05,
+      "loss": 1.0156,
+      "step": 451
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4500753464840775e-05,
+      "loss": 1.1587,
+      "step": 452
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4476983016195245e-05,
+      "loss": 1.0391,
+      "step": 453
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4453180887552052e-05,
+      "loss": 1.0478,
+      "step": 454
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4429347247339656e-05,
+      "loss": 1.1598,
+      "step": 455
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4405482264209512e-05,
+      "loss": 0.9857,
+      "step": 456
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4381586107034849e-05,
+      "loss": 1.1735,
+      "step": 457
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4357658944909496e-05,
+      "loss": 1.0613,
+      "step": 458
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4333700947146686e-05,
+      "loss": 1.1052,
+      "step": 459
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4309712283277839e-05,
+      "loss": 1.1275,
+      "step": 460
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4285693123051385e-05,
+      "loss": 1.0158,
+      "step": 461
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4261643636431539e-05,
+      "loss": 1.1073,
+      "step": 462
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4237563993597133e-05,
+      "loss": 1.0065,
+      "step": 463
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4213454364940362e-05,
+      "loss": 1.0783,
+      "step": 464
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4189314921065629e-05,
+      "loss": 1.1141,
+      "step": 465
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4165145832788305e-05,
+      "loss": 1.0855,
+      "step": 466
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4140947271133536e-05,
+      "loss": 1.0268,
+      "step": 467
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4116719407335022e-05,
+      "loss": 1.1203,
+      "step": 468
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4092462412833811e-05,
+      "loss": 1.0887,
+      "step": 469
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.406817645927709e-05,
+      "loss": 0.9545,
+      "step": 470
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4043861718516964e-05,
+      "loss": 1.0331,
+      "step": 471
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4019518362609239e-05,
+      "loss": 1.0864,
+      "step": 472
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.399514656381221e-05,
+      "loss": 1.0886,
+      "step": 473
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3970746494585439e-05,
+      "loss": 1.1292,
+      "step": 474
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3946318327588534e-05,
+      "loss": 1.0675,
+      "step": 475
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3921862235679929e-05,
+      "loss": 1.1474,
+      "step": 476
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.389737839191566e-05,
+      "loss": 1.039,
+      "step": 477
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3872866969548143e-05,
+      "loss": 1.073,
+      "step": 478
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.384832814202494e-05,
+      "loss": 1.1149,
+      "step": 479
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3823762082987544e-05,
+      "loss": 1.1452,
+      "step": 480
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3799168966270139e-05,
+      "loss": 1.0943,
+      "step": 481
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3774548965898371e-05,
+      "loss": 1.0701,
+      "step": 482
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3749902256088125e-05,
+      "loss": 1.0005,
+      "step": 483
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3725229011244294e-05,
+      "loss": 1.1638,
+      "step": 484
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3700529405959517e-05,
+      "loss": 1.0659,
+      "step": 485
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3675803615012993e-05,
+      "loss": 1.0407,
+      "step": 486
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3651051813369188e-05,
+      "loss": 1.0907,
+      "step": 487
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3626274176176645e-05,
+      "loss": 1.1026,
+      "step": 488
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3601470878766714e-05,
+      "loss": 1.1216,
+      "step": 489
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3576642096652322e-05,
+      "loss": 1.0523,
+      "step": 490
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3551788005526738e-05,
+      "loss": 1.0775,
+      "step": 491
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3526908781262314e-05,
+      "loss": 1.1233,
+      "step": 492
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3502004599909255e-05,
+      "loss": 1.1803,
+      "step": 493
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3477075637694362e-05,
+      "loss": 1.1263,
+      "step": 494
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3452122071019797e-05,
+      "loss": 1.0823,
+      "step": 495
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3427144076461818e-05,
+      "loss": 1.1011,
+      "step": 496
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3402141830769551e-05,
+      "loss": 0.9963,
+      "step": 497
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3377115510863716e-05,
+      "loss": 1.0976,
+      "step": 498
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3352065293835399e-05,
+      "loss": 1.0262,
+      "step": 499
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3326991356944776e-05,
+      "loss": 1.0757,
+      "step": 500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3301893877619874e-05,
+      "loss": 1.0232,
+      "step": 501
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3276773033455312e-05,
+      "loss": 1.0653,
+      "step": 502
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3251629002211042e-05,
+      "loss": 1.0742,
+      "step": 503
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.322646196181109e-05,
+      "loss": 1.103,
+      "step": 504
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3201272090342303e-05,
+      "loss": 1.1171,
+      "step": 505
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3176059566053083e-05,
+      "loss": 1.0204,
+      "step": 506
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3150824567352128e-05,
+      "loss": 1.0055,
+      "step": 507
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3125567272807167e-05,
+      "loss": 1.0891,
+      "step": 508
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3100287861143703e-05,
+      "loss": 1.0075,
+      "step": 509
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3074986511243741e-05,
+      "loss": 1.0382,
+      "step": 510
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3049663402144528e-05,
+      "loss": 1.0055,
+      "step": 511
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.302431871303728e-05,
+      "loss": 1.0861,
+      "step": 512
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2998952623265917e-05,
+      "loss": 1.0948,
+      "step": 513
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2973565312325798e-05,
+      "loss": 1.0362,
+      "step": 514
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2948156959862446e-05,
+      "loss": 1.1207,
+      "step": 515
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2922727745670276e-05,
+      "loss": 1.0854,
+      "step": 516
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2897277849691326e-05,
+      "loss": 1.0418,
+      "step": 517
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2871807452013977e-05,
+      "loss": 1.0269,
+      "step": 518
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.28463167328717e-05,
+      "loss": 1.0621,
+      "step": 519
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2820805872641745e-05,
+      "loss": 1.0765,
+      "step": 520
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2795275051843893e-05,
+      "loss": 1.0733,
+      "step": 521
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.276972445113917e-05,
+      "loss": 0.9789,
+      "step": 522
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2744154251328573e-05,
+      "loss": 1.0535,
+      "step": 523
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2718564633351773e-05,
+      "loss": 1.0474,
+      "step": 524
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2692955778285865e-05,
+      "loss": 1.0164,
+      "step": 525
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.266732786734405e-05,
+      "loss": 1.0503,
+      "step": 526
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2641681081874394e-05,
+      "loss": 1.0523,
+      "step": 527
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2616015603358497e-05,
+      "loss": 1.0932,
+      "step": 528
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2590331613410261e-05,
+      "loss": 0.9713,
+      "step": 529
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2564629293774561e-05,
+      "loss": 1.0768,
+      "step": 530
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.253890882632598e-05,
+      "loss": 1.1055,
+      "step": 531
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2513170393067527e-05,
+      "loss": 1.075,
+      "step": 532
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2487414176129322e-05,
+      "loss": 1.1094,
+      "step": 533
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246164035776735e-05,
+      "loss": 1.1134,
+      "step": 534
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2435849120362123e-05,
+      "loss": 0.9916,
+      "step": 535
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2410040646417431e-05,
+      "loss": 1.0974,
+      "step": 536
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2384215118559027e-05,
+      "loss": 0.9745,
+      "step": 537
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.235837271953334e-05,
+      "loss": 1.0322,
+      "step": 538
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2332513632206183e-05,
+      "loss": 1.1612,
+      "step": 539
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2306638039561455e-05,
+      "loss": 1.0081,
+      "step": 540
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2280746124699864e-05,
+      "loss": 1.1934,
+      "step": 541
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2254838070837596e-05,
+      "loss": 0.9869,
+      "step": 542
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2228914061305059e-05,
+      "loss": 1.0878,
+      "step": 543
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2202974279545554e-05,
+      "loss": 1.0396,
+      "step": 544
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2177018909113994e-05,
+      "loss": 0.9445,
+      "step": 545
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.21510481336756e-05,
+      "loss": 1.0257,
+      "step": 546
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2125062137004602e-05,
+      "loss": 1.1219,
+      "step": 547
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2099061102982939e-05,
+      "loss": 0.9798,
+      "step": 548
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2073045215598953e-05,
+      "loss": 1.0811,
+      "step": 549
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.20470146589461e-05,
+      "loss": 1.1189,
+      "step": 550
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2020969617221627e-05,
+      "loss": 1.0373,
+      "step": 551
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.199491027472529e-05,
+      "loss": 1.0027,
+      "step": 552
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1968836815858038e-05,
+      "loss": 1.099,
+      "step": 553
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1942749425120704e-05,
+      "loss": 1.0822,
+      "step": 554
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1916648287112714e-05,
+      "loss": 1.0227,
+      "step": 555
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1890533586530766e-05,
+      "loss": 1.0001,
+      "step": 556
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1864405508167532e-05,
+      "loss": 1.116,
+      "step": 557
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1838264236910348e-05,
+      "loss": 1.0277,
+      "step": 558
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1812109957739907e-05,
+      "loss": 1.0403,
+      "step": 559
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1785942855728945e-05,
+      "loss": 1.0276,
+      "step": 560
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1759763116040936e-05,
+      "loss": 0.9857,
+      "step": 561
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1733570923928785e-05,
+      "loss": 1.095,
+      "step": 562
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1707366464733501e-05,
+      "loss": 1.1235,
+      "step": 563
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1681149923882913e-05,
+      "loss": 0.9838,
+      "step": 564
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1654921486890327e-05,
+      "loss": 1.0423,
+      "step": 565
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1628681339353244e-05,
+      "loss": 0.955,
+      "step": 566
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1602429666952015e-05,
+      "loss": 1.0012,
+      "step": 567
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1576166655448558e-05,
+      "loss": 1.0756,
+      "step": 568
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1549892490685018e-05,
+      "loss": 1.0529,
+      "step": 569
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1523607358582462e-05,
+      "loss": 1.1125,
+      "step": 570
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.149731144513958e-05,
+      "loss": 1.0864,
+      "step": 571
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1471004936431327e-05,
+      "loss": 1.0284,
+      "step": 572
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.144468801860766e-05,
+      "loss": 1.0784,
+      "step": 573
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1418360877892165e-05,
+      "loss": 1.1312,
+      "step": 574
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1392023700580796e-05,
+      "loss": 1.0445,
+      "step": 575
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1365676673040502e-05,
+      "loss": 1.0802,
+      "step": 576
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.133931998170795e-05,
+      "loss": 1.1193,
+      "step": 577
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1312953813088183e-05,
+      "loss": 1.0525,
+      "step": 578
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1286578353753313e-05,
+      "loss": 1.0297,
+      "step": 579
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1260193790341186e-05,
+      "loss": 0.9857,
+      "step": 580
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1233800309554083e-05,
+      "loss": 1.0877,
+      "step": 581
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1207398098157371e-05,
+      "loss": 1.0841,
+      "step": 582
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1180987342978209e-05,
+      "loss": 1.0807,
+      "step": 583
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1154568230904204e-05,
+      "loss": 0.9939,
+      "step": 584
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1128140948882107e-05,
+      "loss": 1.0699,
+      "step": 585
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1101705683916473e-05,
+      "loss": 1.1226,
+      "step": 586
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1075262623068352e-05,
+      "loss": 1.0694,
+      "step": 587
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1048811953453955e-05,
+      "loss": 1.1485,
+      "step": 588
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1022353862243338e-05,
+      "loss": 1.1421,
+      "step": 589
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0995888536659067e-05,
+      "loss": 0.981,
+      "step": 590
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.096941616397491e-05,
+      "loss": 1.099,
+      "step": 591
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0942936931514492e-05,
+      "loss": 0.9755,
+      "step": 592
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0916451026649981e-05,
+      "loss": 1.1162,
+      "step": 593
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.088995863680077e-05,
+      "loss": 0.9486,
+      "step": 594
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0863459949432122e-05,
+      "loss": 1.0728,
+      "step": 595
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0836955152053883e-05,
+      "loss": 1.022,
+      "step": 596
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.081044443221912e-05,
+      "loss": 1.0214,
+      "step": 597
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0783927977522819e-05,
+      "loss": 1.0447,
+      "step": 598
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0757405975600534e-05,
+      "loss": 1.1324,
+      "step": 599
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0730878614127087e-05,
+      "loss": 1.0334,
+      "step": 600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0704346080815218e-05,
+      "loss": 0.9593,
+      "step": 601
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0677808563414256e-05,
+      "loss": 1.0855,
+      "step": 602
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0651266249708816e-05,
+      "loss": 0.9969,
+      "step": 603
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0624719327517434e-05,
+      "loss": 1.0853,
+      "step": 604
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0598167984691276e-05,
+      "loss": 1.0692,
+      "step": 605
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057161240911277e-05,
+      "loss": 1.1015,
+      "step": 606
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0545052788694312e-05,
+      "loss": 1.0251,
+      "step": 607
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0518489311376905e-05,
+      "loss": 1.0395,
+      "step": 608
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0491922165128853e-05,
+      "loss": 1.1046,
+      "step": 609
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0465351537944429e-05,
+      "loss": 1.0587,
+      "step": 610
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.043877761784252e-05,
+      "loss": 1.0852,
+      "step": 611
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0412200592865331e-05,
+      "loss": 1.0562,
+      "step": 612
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0385620651077024e-05,
+      "loss": 1.1277,
+      "step": 613
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0359037980562416e-05,
+      "loss": 1.1861,
+      "step": 614
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0332452769425619e-05,
+      "loss": 1.0468,
+      "step": 615
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0305865205788728e-05,
+      "loss": 1.1171,
+      "step": 616
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0279275477790487e-05,
+      "loss": 1.0895,
+      "step": 617
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0252683773584953e-05,
+      "loss": 1.1341,
+      "step": 618
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0226090281340168e-05,
+      "loss": 0.9769,
+      "step": 619
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0199495189236828e-05,
+      "loss": 0.9259,
+      "step": 620
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0172898685466947e-05,
+      "loss": 1.1301,
+      "step": 621
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0146300958232528e-05,
+      "loss": 1.0171,
+      "step": 622
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0119702195744236e-05,
+      "loss": 1.062,
+      "step": 623
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0093102586220056e-05,
+      "loss": 0.9874,
+      "step": 624
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0066502317883969e-05,
+      "loss": 1.0152,
+      "step": 625
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0039901578964619e-05,
+      "loss": 1.0578,
+      "step": 626
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0013300557693981e-05,
+      "loss": 1.0976,
+      "step": 627
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.986699442306025e-06,
+      "loss": 1.0745,
+      "step": 628
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.960098421035383e-06,
+      "loss": 1.0151,
+      "step": 629
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.933497682116035e-06,
+      "loss": 1.0811,
+      "step": 630
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.906897413779949e-06,
+      "loss": 1.1467,
+      "step": 631
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88029780425577e-06,
+      "loss": 1.1107,
+      "step": 632
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.853699041767473e-06,
+      "loss": 1.0815,
+      "step": 633
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.827101314533056e-06,
+      "loss": 1.0757,
+      "step": 634
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.800504810763176e-06,
+      "loss": 1.0236,
+      "step": 635
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.773909718659831e-06,
+      "loss": 1.0345,
+      "step": 636
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.747316226415052e-06,
+      "loss": 1.0998,
+      "step": 637
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.720724522209518e-06,
+      "loss": 1.0581,
+      "step": 638
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.694134794211277e-06,
+      "loss": 1.0571,
+      "step": 639
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.667547230574386e-06,
+      "loss": 1.0675,
+      "step": 640
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.64096201943759e-06,
+      "loss": 1.0064,
+      "step": 641
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.61437934892298e-06,
+      "loss": 1.0512,
+      "step": 642
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.587799407134672e-06,
+      "loss": 1.0125,
+      "step": 643
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.56122238215748e-06,
+      "loss": 1.1364,
+      "step": 644
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.534648462055576e-06,
+      "loss": 0.9677,
+      "step": 645
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.50807783487115e-06,
+      "loss": 0.9366,
+      "step": 646
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.481510688623098e-06,
+      "loss": 1.099,
+      "step": 647
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.454947211305691e-06,
+      "loss": 1.1557,
+      "step": 648
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.42838759088723e-06,
+      "loss": 1.1092,
+      "step": 649
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.401832015308728e-06,
+      "loss": 1.0027,
+      "step": 650
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.375280672482567e-06,
+      "loss": 0.9507,
+      "step": 651
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.348733750291186e-06,
+      "loss": 1.0384,
+      "step": 652
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.322191436585745e-06,
+      "loss": 1.1106,
+      "step": 653
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.295653919184787e-06,
+      "loss": 0.9639,
+      "step": 654
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.269121385872915e-06,
+      "loss": 1.1055,
+      "step": 655
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.242594024399467e-06,
+      "loss": 1.0294,
+      "step": 656
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.216072022477183e-06,
+      "loss": 1.0803,
+      "step": 657
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.189555567780882e-06,
+      "loss": 1.0642,
+      "step": 658
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.16304484794612e-06,
+      "loss": 1.0686,
+      "step": 659
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.13654005056788e-06,
+      "loss": 1.0217,
+      "step": 660
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.110041363199233e-06,
+      "loss": 1.0821,
+      "step": 661
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.083548973350019e-06,
+      "loss": 1.01,
+      "step": 662
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.057063068485513e-06,
+      "loss": 1.0153,
+      "step": 663
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.030583836025093e-06,
+      "loss": 1.0438,
+      "step": 664
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.004111463340935e-06,
+      "loss": 1.008,
+      "step": 665
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.977646137756662e-06,
+      "loss": 0.9625,
+      "step": 666
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.951188046546048e-06,
+      "loss": 0.9189,
+      "step": 667
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.924737376931651e-06,
+      "loss": 1.0021,
+      "step": 668
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.898294316083529e-06,
+      "loss": 1.0809,
+      "step": 669
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.871859051117896e-06,
+      "loss": 1.117,
+      "step": 670
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.8454317690958e-06,
+      "loss": 1.1136,
+      "step": 671
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.819012657021794e-06,
+      "loss": 1.0478,
+      "step": 672
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.79260190184263e-06,
+      "loss": 1.035,
+      "step": 673
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.76619969044592e-06,
+      "loss": 1.086,
+      "step": 674
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.739806209658812e-06,
+      "loss": 1.135,
+      "step": 675
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.713421646246692e-06,
+      "loss": 0.9785,
+      "step": 676
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.687046186911819e-06,
+      "loss": 0.9707,
+      "step": 677
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.660680018292053e-06,
+      "loss": 1.0241,
+      "step": 678
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.634323326959501e-06,
+      "loss": 1.0647,
+      "step": 679
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.60797629941921e-06,
+      "loss": 1.1372,
+      "step": 680
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.581639122107837e-06,
+      "loss": 1.0297,
+      "step": 681
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.555311981392342e-06,
+      "loss": 0.9526,
+      "step": 682
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.528995063568673e-06,
+      "loss": 0.9586,
+      "step": 683
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.502688554860426e-06,
+      "loss": 1.092,
+      "step": 684
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.47639264141754e-06,
+      "loss": 1.0226,
+      "step": 685
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.450107509314983e-06,
+      "loss": 1.085,
+      "step": 686
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.423833344551443e-06,
+      "loss": 1.1103,
+      "step": 687
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.397570333047985e-06,
+      "loss": 0.986,
+      "step": 688
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.37131866064676e-06,
+      "loss": 1.0804,
+      "step": 689
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.345078513109677e-06,
+      "loss": 1.0916,
+      "step": 690
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.31885007611709e-06,
+      "loss": 1.0922,
+      "step": 691
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.2926335352665e-06,
+      "loss": 1.0426,
+      "step": 692
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.266429076071221e-06,
+      "loss": 1.0726,
+      "step": 693
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.240236883959067e-06,
+      "loss": 1.0394,
+      "step": 694
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.214057144271058e-06,
+      "loss": 1.0831,
+      "step": 695
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.187890042260094e-06,
+      "loss": 1.0468,
+      "step": 696
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.161735763089654e-06,
+      "loss": 1.0715,
+      "step": 697
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.13559449183247e-06,
+      "loss": 1.0629,
+      "step": 698
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.109466413469238e-06,
+      "loss": 1.0247,
+      "step": 699
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.083351712887288e-06,
+      "loss": 1.0005,
+      "step": 700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.057250574879296e-06,
+      "loss": 1.2048,
+      "step": 701
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.031163184141965e-06,
+      "loss": 1.0668,
+      "step": 702
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.005089725274711e-06,
+      "loss": 1.1425,
+      "step": 703
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.979030382778376e-06,
+      "loss": 0.9857,
+      "step": 704
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.952985341053902e-06,
+      "loss": 1.0851,
+      "step": 705
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92695478440105e-06,
+      "loss": 1.068,
+      "step": 706
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.900938897017064e-06,
+      "loss": 0.962,
+      "step": 707
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.874937862995401e-06,
+      "loss": 1.0436,
+      "step": 708
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.848951866324402e-06,
+      "loss": 1.0016,
+      "step": 709
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.822981090886011e-06,
+      "loss": 1.0617,
+      "step": 710
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.79702572045445e-06,
+      "loss": 1.0548,
+      "step": 711
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.771085938694943e-06,
+      "loss": 1.0146,
+      "step": 712
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.745161929162405e-06,
+      "loss": 1.124,
+      "step": 713
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.719253875300138e-06,
+      "loss": 1.0498,
+      "step": 714
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.693361960438548e-06,
+      "loss": 1.1152,
+      "step": 715
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.667486367793822e-06,
+      "loss": 1.0943,
+      "step": 716
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.641627280466663e-06,
+      "loss": 1.0807,
+      "step": 717
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.615784881440975e-06,
+      "loss": 1.0324,
+      "step": 718
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.589959353582574e-06,
+      "loss": 1.0629,
+      "step": 719
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.564150879637882e-06,
+      "loss": 1.0459,
+      "step": 720
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.538359642232654e-06,
+      "loss": 1.0255,
+      "step": 721
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5125858238706785e-06,
+      "loss": 1.0333,
+      "step": 722
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.486829606932478e-06,
+      "loss": 1.0706,
+      "step": 723
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.461091173674022e-06,
+      "loss": 1.0349,
+      "step": 724
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.43537070622544e-06,
+      "loss": 1.0736,
+      "step": 725
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.40966838658974e-06,
+      "loss": 0.9974,
+      "step": 726
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.383984396641506e-06,
+      "loss": 1.0728,
+      "step": 727
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.358318918125613e-06,
+      "loss": 1.0178,
+      "step": 728
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.332672132655953e-06,
+      "loss": 1.0454,
+      "step": 729
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.307044221714139e-06,
+      "loss": 1.0209,
+      "step": 730
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2814353666482276e-06,
+      "loss": 1.1334,
+      "step": 731
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2558457486714316e-06,
+      "loss": 0.9714,
+      "step": 732
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.230275548860833e-06,
+      "loss": 1.0516,
+      "step": 733
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2047249481561125e-06,
+      "loss": 1.1283,
+      "step": 734
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.179194127358258e-06,
+      "loss": 1.0491,
+      "step": 735
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.153683267128304e-06,
+      "loss": 1.0826,
+      "step": 736
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.128192547986023e-06,
+      "loss": 1.1392,
+      "step": 737
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.102722150308678e-06,
+      "loss": 1.0416,
+      "step": 738
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.077272254329726e-06,
+      "loss": 1.0953,
+      "step": 739
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.051843040137558e-06,
+      "loss": 1.0443,
+      "step": 740
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.026434687674204e-06,
+      "loss": 1.0803,
+      "step": 741
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.001047376734087e-06,
+      "loss": 1.0346,
+      "step": 742
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.975681286962724e-06,
+      "loss": 1.0129,
+      "step": 743
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9503365978554735e-06,
+      "loss": 1.0135,
+      "step": 744
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.925013488756264e-06,
+      "loss": 1.0746,
+      "step": 745
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8997121388563e-06,
+      "loss": 1.0111,
+      "step": 746
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.874432727192837e-06,
+      "loss": 0.9851,
+      "step": 747
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.849175432647875e-06,
+      "loss": 0.9814,
+      "step": 748
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.823940433946921e-06,
+      "loss": 0.9906,
+      "step": 749
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.798727909657698e-06,
+      "loss": 1.0407,
+      "step": 750
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.773538038188912e-06,
+      "loss": 1.051,
+      "step": 751
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.74837099778896e-06,
+      "loss": 1.1052,
+      "step": 752
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.723226966544691e-06,
+      "loss": 0.9945,
+      "step": 753
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.69810612238013e-06,
+      "loss": 1.0537,
+      "step": 754
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.673008643055228e-06,
+      "loss": 1.015,
+      "step": 755
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6479347061646046e-06,
+      "loss": 1.1698,
+      "step": 756
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.622884489136286e-06,
+      "loss": 1.058,
+      "step": 757
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.597858169230454e-06,
+      "loss": 1.0748,
+      "step": 758
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.572855923538186e-06,
+      "loss": 1.0525,
+      "step": 759
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.547877928980206e-06,
+      "loss": 1.0324,
+      "step": 760
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.522924362305639e-06,
+      "loss": 1.0841,
+      "step": 761
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.497995400090748e-06,
+      "loss": 1.0314,
+      "step": 762
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4730912187376895e-06,
+      "loss": 0.93,
+      "step": 763
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.448211994473263e-06,
+      "loss": 1.0023,
+      "step": 764
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.42335790334768e-06,
+      "loss": 1.1017,
+      "step": 765
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.398529121233291e-06,
+      "loss": 0.9936,
+      "step": 766
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.373725823823359e-06,
+      "loss": 1.0008,
+      "step": 767
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.348948186630815e-06,
+      "loss": 0.9642,
+      "step": 768
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.324196384987009e-06,
+      "loss": 1.0331,
+      "step": 769
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.2994705940404825e-06,
+      "loss": 1.1336,
+      "step": 770
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.274770988755712e-06,
+      "loss": 1.0058,
+      "step": 771
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.250097743911877e-06,
+      "loss": 1.0476,
+      "step": 772
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.225451034101631e-06,
+      "loss": 0.9599,
+      "step": 773
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.200831033729864e-06,
+      "loss": 1.0348,
+      "step": 774
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.176237917012459e-06,
+      "loss": 1.0001,
+      "step": 775
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.151671857975061e-06,
+      "loss": 1.0962,
+      "step": 776
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.12713303045186e-06,
+      "loss": 1.0898,
+      "step": 777
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.10262160808434e-06,
+      "loss": 1.0083,
+      "step": 778
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0781377643200765e-06,
+      "loss": 1.0719,
+      "step": 779
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.053681672411471e-06,
+      "loss": 0.9877,
+      "step": 780
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.029253505414565e-06,
+      "loss": 1.1058,
+      "step": 781
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.004853436187794e-06,
+      "loss": 1.0424,
+      "step": 782
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9804816373907625e-06,
+      "loss": 1.0619,
+      "step": 783
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.956138281483039e-06,
+      "loss": 1.1046,
+      "step": 784
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.931823540722912e-06,
+      "loss": 1.0912,
+      "step": 785
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.907537587166191e-06,
+      "loss": 1.0446,
+      "step": 786
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.883280592664979e-06,
+      "loss": 1.0525,
+      "step": 787
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.859052728866468e-06,
+      "loss": 0.9887,
+      "step": 788
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.834854167211699e-06,
+      "loss": 1.0369,
+      "step": 789
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.810685078934375e-06,
+      "loss": 1.0696,
+      "step": 790
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.78654563505964e-06,
+      "loss": 1.0302,
+      "step": 791
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.762436006402874e-06,
+      "loss": 1.059,
+      "step": 792
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.738356363568463e-06,
+      "loss": 1.1394,
+      "step": 793
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.714306876948621e-06,
+      "loss": 1.0213,
+      "step": 794
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.69028771672216e-06,
+      "loss": 1.0083,
+      "step": 795
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.666299052853314e-06,
+      "loss": 1.0249,
+      "step": 796
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.642341055090508e-06,
+      "loss": 1.1016,
+      "step": 797
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.618413892965158e-06,
+      "loss": 1.0996,
+      "step": 798
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5945177357904935e-06,
+      "loss": 1.0578,
+      "step": 799
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.570652752660343e-06,
+      "loss": 1.0476,
+      "step": 800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.546819112447952e-06,
+      "loss": 1.0377,
+      "step": 801
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.523016983804759e-06,
+      "loss": 1.0016,
+      "step": 802
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.499246535159231e-06,
+      "loss": 1.0697,
+      "step": 803
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.47550793471566e-06,
+      "loss": 1.1112,
+      "step": 804
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.451801350452975e-06,
+      "loss": 1.0682,
+      "step": 805
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.428126950123551e-06,
+      "loss": 1.1169,
+      "step": 806
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.404484901252023e-06,
+      "loss": 0.9758,
+      "step": 807
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3808753711341e-06,
+      "loss": 1.0212,
+      "step": 808
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.357298526835381e-06,
+      "loss": 1.0372,
+      "step": 809
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.333754535190186e-06,
+      "loss": 1.1271,
+      "step": 810
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3102435628003435e-06,
+      "loss": 1.063,
+      "step": 811
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.286765776034044e-06,
+      "loss": 1.1389,
+      "step": 812
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.263321341024646e-06,
+      "loss": 0.9841,
+      "step": 813
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.239910423669509e-06,
+      "loss": 0.7626,
+      "step": 814
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.216533189628808e-06,
+      "loss": 0.7865,
+      "step": 815
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.193189804324376e-06,
+      "loss": 0.7634,
+      "step": 816
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.169880432938519e-06,
+      "loss": 0.8215,
+      "step": 817
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.146605240412859e-06,
+      "loss": 0.7284,
+      "step": 818
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.123364391447156e-06,
+      "loss": 0.7751,
+      "step": 819
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.100158050498159e-06,
+      "loss": 0.8276,
+      "step": 820
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.076986381778417e-06,
+      "loss": 0.8012,
+      "step": 821
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.053849549255143e-06,
+      "loss": 0.755,
+      "step": 822
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.03074771664904e-06,
+      "loss": 0.7815,
+      "step": 823
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0076810474331395e-06,
+      "loss": 0.7611,
+      "step": 824
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.984649704831658e-06,
+      "loss": 0.8147,
+      "step": 825
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.961653851818827e-06,
+      "loss": 0.8307,
+      "step": 826
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.938693651117751e-06,
+      "loss": 0.7899,
+      "step": 827
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9157692651992495e-06,
+      "loss": 0.819,
+      "step": 828
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.892880856280713e-06,
+      "loss": 0.7603,
+      "step": 829
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.870028586324947e-06,
+      "loss": 0.8007,
+      "step": 830
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.847212617039037e-06,
+      "loss": 0.8197,
+      "step": 831
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.82443310987319e-06,
+      "loss": 0.8041,
+      "step": 832
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.801690226019606e-06,
+      "loss": 0.7963,
+      "step": 833
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.77898412641133e-06,
+      "loss": 0.7348,
+      "step": 834
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.756314971721115e-06,
+      "loss": 0.6906,
+      "step": 835
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.733682922360282e-06,
+      "loss": 0.7446,
+      "step": 836
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.71108813847759e-06,
+      "loss": 0.8379,
+      "step": 837
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.688530779958099e-06,
+      "loss": 0.7473,
+      "step": 838
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.666011006422041e-06,
+      "loss": 0.7785,
+      "step": 839
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.643528977223689e-06,
+      "loss": 0.8199,
+      "step": 840
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.621084851450229e-06,
+      "loss": 0.7547,
+      "step": 841
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5986787879206375e-06,
+      "loss": 0.7269,
+      "step": 842
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5763109451845515e-06,
+      "loss": 0.7745,
+      "step": 843
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.553981481521156e-06,
+      "loss": 0.7706,
+      "step": 844
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.531690554938043e-06,
+      "loss": 0.7923,
+      "step": 845
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.509438323170131e-06,
+      "loss": 0.7361,
+      "step": 846
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.487224943678513e-06,
+      "loss": 0.7218,
+      "step": 847
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.465050573649359e-06,
+      "loss": 0.6608,
+      "step": 848
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.442915369992802e-06,
+      "loss": 0.8203,
+      "step": 849
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4208194893418125e-06,
+      "loss": 0.7754,
+      "step": 850
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.398763088051127e-06,
+      "loss": 0.6917,
+      "step": 851
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3767463221961034e-06,
+      "loss": 0.7366,
+      "step": 852
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.354769347571638e-06,
+      "loss": 0.7909,
+      "step": 853
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.332832319691044e-06,
+      "loss": 0.7557,
+      "step": 854
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3109353937849815e-06,
+      "loss": 0.7832,
+      "step": 855
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.289078724800331e-06,
+      "loss": 0.7443,
+      "step": 856
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.267262467399114e-06,
+      "loss": 0.6829,
+      "step": 857
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.24548677595738e-06,
+      "loss": 0.7148,
+      "step": 858
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.22375180456413e-06,
+      "loss": 0.6658,
+      "step": 859
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.202057707020235e-06,
+      "loss": 0.6883,
+      "step": 860
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.180404636837321e-06,
+      "loss": 0.6841,
+      "step": 861
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.158792747236702e-06,
+      "loss": 0.7621,
+      "step": 862
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.137222191148282e-06,
+      "loss": 0.7775,
+      "step": 863
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.11569312120949e-06,
+      "loss": 0.7944,
+      "step": 864
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0942056897641934e-06,
+      "loss": 0.8151,
+      "step": 865
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.072760048861614e-06,
+      "loss": 0.6909,
+      "step": 866
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.051356350255246e-06,
+      "loss": 0.7683,
+      "step": 867
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0299947454018e-06,
+      "loss": 0.7324,
+      "step": 868
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.008675385460131e-06,
+      "loss": 0.8605,
+      "step": 869
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.987398421290155e-06,
+      "loss": 0.6989,
+      "step": 870
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.966164003451775e-06,
+      "loss": 0.7251,
+      "step": 871
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.944972282203844e-06,
+      "loss": 0.805,
+      "step": 872
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.923823407503076e-06,
+      "loss": 0.7412,
+      "step": 873
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.902717529003005e-06,
+      "loss": 0.8276,
+      "step": 874
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.88165479605291e-06,
+      "loss": 0.744,
+      "step": 875
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.860635357696756e-06,
+      "loss": 0.7458,
+      "step": 876
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.839659362672156e-06,
+      "loss": 0.7778,
+      "step": 877
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.818726959409305e-06,
+      "loss": 0.7298,
+      "step": 878
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7978382960299476e-06,
+      "loss": 0.8165,
+      "step": 879
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.776993520346295e-06,
+      "loss": 0.8294,
+      "step": 880
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.756192779860014e-06,
+      "loss": 0.715,
+      "step": 881
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7354362217611652e-06,
+      "loss": 0.7522,
+      "step": 882
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.714723992927177e-06,
+      "loss": 0.7537,
+      "step": 883
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.694056239921776e-06,
+      "loss": 0.8263,
+      "step": 884
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6734331089939835e-06,
+      "loss": 0.8068,
+      "step": 885
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6528547460770636e-06,
+      "loss": 0.7623,
+      "step": 886
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6323212967874866e-06,
+      "loss": 0.6378,
+      "step": 887
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6118329064239222e-06,
+      "loss": 0.7163,
+      "step": 888
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5913897199661716e-06,
+      "loss": 0.715,
+      "step": 889
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5709918820741816e-06,
+      "loss": 0.7586,
+      "step": 890
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5506395370869963e-06,
+      "loss": 0.7395,
+      "step": 891
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5303328290217453e-06,
+      "loss": 0.7425,
+      "step": 892
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5100719015726228e-06,
+      "loss": 0.7336,
+      "step": 893
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4898568981098678e-06,
+      "loss": 0.7556,
+      "step": 894
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.469687961678757e-06,
+      "loss": 0.8236,
+      "step": 895
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4495652349985844e-06,
+      "loss": 0.8095,
+      "step": 896
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.429488860461655e-06,
+      "loss": 0.7911,
+      "step": 897
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4094589801322773e-06,
+      "loss": 0.7544,
+      "step": 898
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.389475735745761e-06,
+      "loss": 0.7854,
+      "step": 899
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3695392687074045e-06,
+      "loss": 0.6854,
+      "step": 900
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3496497200915067e-06,
+      "loss": 0.7875,
+      "step": 901
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3298072306403595e-06,
+      "loss": 0.8737,
+      "step": 902
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3100119407632556e-06,
+      "loss": 0.7294,
+      "step": 903
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2902639905354948e-06,
+      "loss": 0.742,
+      "step": 904
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2705635196973927e-06,
+      "loss": 0.7496,
+      "step": 905
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2509106676532897e-06,
+      "loss": 0.8544,
+      "step": 906
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.231305573470569e-06,
+      "loss": 0.7902,
+      "step": 907
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2117483758786683e-06,
+      "loss": 0.7222,
+      "step": 908
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.192239213268099e-06,
+      "loss": 0.7472,
+      "step": 909
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.17277822368947e-06,
+      "loss": 0.7353,
+      "step": 910
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1533655448525057e-06,
+      "loss": 0.824,
+      "step": 911
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.134001314125079e-06,
+      "loss": 0.7616,
+      "step": 912
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.114685668532229e-06,
+      "loss": 0.694,
+      "step": 913
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0954187447551996e-06,
+      "loss": 0.6772,
+      "step": 914
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.076200679130471e-06,
+      "loss": 0.7149,
+      "step": 915
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0570316076487918e-06,
+      "loss": 0.6822,
+      "step": 916
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0379116659542186e-06,
+      "loss": 0.7332,
+      "step": 917
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0188409893431556e-06,
+      "loss": 0.7217,
+      "step": 918
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.999819712763402e-06,
+      "loss": 0.7803,
+      "step": 919
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9808479708131864e-06,
+      "loss": 0.7129,
+      "step": 920
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9619258977402253e-06,
+      "loss": 0.7679,
+      "step": 921
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.943053627440771e-06,
+      "loss": 0.7511,
+      "step": 922
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.924231293458647e-06,
+      "loss": 0.7012,
+      "step": 923
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.90545902898434e-06,
+      "loss": 0.7584,
+      "step": 924
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.886736966854019e-06,
+      "loss": 0.7744,
+      "step": 925
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8680652395486198e-06,
+      "loss": 0.797,
+      "step": 926
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.849443979192892e-06,
+      "loss": 0.7356,
+      "step": 927
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8308733175544724e-06,
+      "loss": 0.7267,
+      "step": 928
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.812353386042962e-06,
+      "loss": 0.7347,
+      "step": 929
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7938843157089734e-06,
+      "loss": 0.677,
+      "step": 930
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.775466237243226e-06,
+      "loss": 0.7115,
+      "step": 931
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7570992809755937e-06,
+      "loss": 0.8489,
+      "step": 932
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.73878357687422e-06,
+      "loss": 0.7027,
+      "step": 933
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.720519254544568e-06,
+      "loss": 0.7935,
+      "step": 934
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.702306443228516e-06,
+      "loss": 0.748,
+      "step": 935
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6841452718034343e-06,
+      "loss": 0.8216,
+      "step": 936
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.666035868781285e-06,
+      "loss": 0.8043,
+      "step": 937
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6479783623077105e-06,
+      "loss": 0.7297,
+      "step": 938
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.62997288016112e-06,
+      "loss": 0.676,
+      "step": 939
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6120195497517818e-06,
+      "loss": 0.7872,
+      "step": 940
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5941184981209354e-06,
+      "loss": 0.7718,
+      "step": 941
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5762698519398832e-06,
+      "loss": 0.7205,
+      "step": 942
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5584737375091016e-06,
+      "loss": 0.6992,
+      "step": 943
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5407302807573387e-06,
+      "loss": 0.7619,
+      "step": 944
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5230396072407204e-06,
+      "loss": 0.7293,
+      "step": 945
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5054018421418737e-06,
+      "loss": 0.7979,
+      "step": 946
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.487817110269042e-06,
+      "loss": 0.725,
+      "step": 947
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.470285536055188e-06,
+      "loss": 0.7741,
+      "step": 948
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4528072435571158e-06,
+      "loss": 0.7211,
+      "step": 949
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4353823564546064e-06,
+      "loss": 0.693,
+      "step": 950
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4180109980495293e-06,
+      "loss": 0.8077,
+      "step": 951
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4006932912649816e-06,
+      "loss": 0.7579,
+      "step": 952
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3834293586444e-06,
+      "loss": 0.7526,
+      "step": 953
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3662193223507135e-06,
+      "loss": 0.699,
+      "step": 954
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.349063304165462e-06,
+      "loss": 0.7606,
+      "step": 955
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.331961425487956e-06,
+      "loss": 0.853,
+      "step": 956
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3149138073343958e-06,
+      "loss": 0.773,
+      "step": 957
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.297920570337019e-06,
+      "loss": 0.7557,
+      "step": 958
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2809818347432598e-06,
+      "loss": 0.7767,
+      "step": 959
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2640977204148838e-06,
+      "loss": 0.8156,
+      "step": 960
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2472683468271584e-06,
+      "loss": 0.7607,
+      "step": 961
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.230493833067977e-06,
+      "loss": 0.8509,
+      "step": 962
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.213774297837047e-06,
+      "loss": 0.7933,
+      "step": 963
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1971098594450315e-06,
+      "loss": 0.7459,
+      "step": 964
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1805006358127213e-06,
+      "loss": 0.7633,
+      "step": 965
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1639467444701934e-06,
+      "loss": 0.76,
+      "step": 966
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1474483025559857e-06,
+      "loss": 0.8054,
+      "step": 967
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1310054268162628e-06,
+      "loss": 0.7315,
+      "step": 968
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.114618233603992e-06,
+      "loss": 0.8457,
+      "step": 969
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0982868388781286e-06,
+      "loss": 0.7856,
+      "step": 970
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0820113582027734e-06,
+      "loss": 0.7577,
+      "step": 971
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0657919067463773e-06,
+      "loss": 0.8005,
+      "step": 972
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0496285992809163e-06,
+      "loss": 0.7323,
+      "step": 973
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.033521550181078e-06,
+      "loss": 0.737,
+      "step": 974
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0174708734234596e-06,
+      "loss": 0.7039,
+      "step": 975
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0014766825857514e-06,
+      "loss": 0.8001,
+      "step": 976
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.985539090845943e-06,
+      "loss": 0.8213,
+      "step": 977
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9696582109815145e-06,
+      "loss": 0.7024,
+      "step": 978
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9538341553686446e-06,
+      "loss": 0.7473,
+      "step": 979
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.93806703598141e-06,
+      "loss": 0.8634,
+      "step": 980
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9223569643909978e-06,
+      "loss": 0.7337,
+      "step": 981
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9067040517649115e-06,
+      "loss": 0.675,
+      "step": 982
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8911084088661903e-06,
+      "loss": 0.7118,
+      "step": 983
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8755701460526166e-06,
+      "loss": 0.786,
+      "step": 984
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.860089373275945e-06,
+      "loss": 0.8187,
+      "step": 985
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8446662000811177e-06,
+      "loss": 0.7859,
+      "step": 986
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8293007356054903e-06,
+      "loss": 0.8582,
+      "step": 987
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8139930885780621e-06,
+      "loss": 0.7031,
+      "step": 988
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7987433673187026e-06,
+      "loss": 0.8058,
+      "step": 989
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7835516797373908e-06,
+      "loss": 0.7231,
+      "step": 990
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7684181333334437e-06,
+      "loss": 0.7504,
+      "step": 991
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7533428351947634e-06,
+      "loss": 0.7031,
+      "step": 992
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7383258919970746e-06,
+      "loss": 0.7784,
+      "step": 993
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7233674100031728e-06,
+      "loss": 0.7856,
+      "step": 994
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7084674950621694e-06,
+      "loss": 0.7928,
+      "step": 995
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6936262526087432e-06,
+      "loss": 0.8125,
+      "step": 996
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6788437876623963e-06,
+      "loss": 0.7592,
+      "step": 997
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6641202048267102e-06,
+      "loss": 0.755,
+      "step": 998
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6494556082886038e-06,
+      "loss": 0.7177,
+      "step": 999
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.634850101817601e-06,
+      "loss": 0.7806,
+      "step": 1000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6203037887650842e-06,
+      "loss": 0.7891,
+      "step": 1001
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6058167720635832e-06,
+      "loss": 0.7573,
+      "step": 1002
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5913891542260284e-06,
+      "loss": 0.7019,
+      "step": 1003
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5770210373450356e-06,
+      "loss": 0.8272,
+      "step": 1004
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5627125230921725e-06,
+      "loss": 0.781,
+      "step": 1005
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5484637127172609e-06,
+      "loss": 0.792,
+      "step": 1006
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5342747070476339e-06,
+      "loss": 0.8272,
+      "step": 1007
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.520145606487442e-06,
+      "loss": 0.7784,
+      "step": 1008
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.506076511016935e-06,
+      "loss": 0.7424,
+      "step": 1009
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4920675201917467e-06,
+      "loss": 0.7235,
+      "step": 1010
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4781187331422109e-06,
+      "loss": 0.7974,
+      "step": 1011
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4642302485726423e-06,
+      "loss": 0.7787,
+      "step": 1012
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4504021647606448e-06,
+      "loss": 0.7455,
+      "step": 1013
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4366345795564084e-06,
+      "loss": 0.7663,
+      "step": 1014
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4229275903820306e-06,
+      "loss": 0.7163,
+      "step": 1015
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.409281294230821e-06,
+      "loss": 0.7627,
+      "step": 1016
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.395695787666611e-06,
+      "loss": 0.7525,
+      "step": 1017
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3821711668230675e-06,
+      "loss": 0.7615,
+      "step": 1018
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3687075274030238e-06,
+      "loss": 0.7913,
+      "step": 1019
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3553049646777993e-06,
+      "loss": 0.7802,
+      "step": 1020
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.34196357348652e-06,
+      "loss": 0.7872,
+      "step": 1021
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3286834482354506e-06,
+      "loss": 0.7357,
+      "step": 1022
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3154646828973217e-06,
+      "loss": 0.6946,
+      "step": 1023
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3023073710106726e-06,
+      "loss": 0.7104,
+      "step": 1024
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2892116056791927e-06,
+      "loss": 0.7292,
+      "step": 1025
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2761774795710502e-06,
+      "loss": 0.7176,
+      "step": 1026
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2632050849182365e-06,
+      "loss": 0.6822,
+      "step": 1027
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2502945135159272e-06,
+      "loss": 0.7543,
+      "step": 1028
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2374458567218217e-06,
+      "loss": 0.7733,
+      "step": 1029
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2246592054555062e-06,
+      "loss": 0.8225,
+      "step": 1030
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2119346501977914e-06,
+      "loss": 0.8064,
+      "step": 1031
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1992722809900925e-06,
+      "loss": 0.7189,
+      "step": 1032
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1866721874337827e-06,
+      "loss": 0.7857,
+      "step": 1033
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1741344586895642e-06,
+      "loss": 0.7682,
+      "step": 1034
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1616591834768299e-06,
+      "loss": 0.6905,
+      "step": 1035
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.149246450073036e-06,
+      "loss": 0.7895,
+      "step": 1036
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1368963463130866e-06,
+      "loss": 0.7626,
+      "step": 1037
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1246089595887023e-06,
+      "loss": 0.715,
+      "step": 1038
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1123843768478148e-06,
+      "loss": 0.7468,
+      "step": 1039
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.100222684593929e-06,
+      "loss": 0.6507,
+      "step": 1040
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.088123968885534e-06,
+      "loss": 0.7542,
+      "step": 1041
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0760883153354818e-06,
+      "loss": 0.8011,
+      "step": 1042
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0641158091103832e-06,
+      "loss": 0.7302,
+      "step": 1043
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0522065349300103e-06,
+      "loss": 0.7747,
+      "step": 1044
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.040360577066688e-06,
+      "loss": 0.7688,
+      "step": 1045
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.028578019344706e-06,
+      "loss": 0.7384,
+      "step": 1046
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0168589451397204e-06,
+      "loss": 0.7658,
+      "step": 1047
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0052034373781716e-06,
+      "loss": 0.7751,
+      "step": 1048
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.936115785366817e-07,
+      "loss": 0.8098,
+      "step": 1049
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.820834506414866e-07,
+      "loss": 0.7774,
+      "step": 1050
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.706191352678495e-07,
+      "loss": 0.7921,
+      "step": 1051
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.592187135394826e-07,
+      "loss": 0.7449,
+      "step": 1052
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.478822661279763e-07,
+      "loss": 0.7424,
+      "step": 1053
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.366098732522233e-07,
+      "loss": 0.8077,
+      "step": 1054
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.254016146778555e-07,
+      "loss": 0.8037,
+      "step": 1055
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.1425756971668e-07,
+      "loss": 0.741,
+      "step": 1056
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.03177817226113e-07,
+      "loss": 0.764,
+      "step": 1057
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.921624356086256e-07,
+      "loss": 0.7839,
+      "step": 1058
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.81211502811189e-07,
+      "loss": 0.7792,
+      "step": 1059
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.703250963247223e-07,
+      "loss": 0.8151,
+      "step": 1060
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.595032931835423e-07,
+      "loss": 0.7714,
+      "step": 1061
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.487461699648203e-07,
+      "loss": 0.8234,
+      "step": 1062
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.380538027880425e-07,
+      "loss": 0.7568,
+      "step": 1063
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.274262673144651e-07,
+      "loss": 0.7692,
+      "step": 1064
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.168636387465856e-07,
+      "loss": 0.784,
+      "step": 1065
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.063659918276056e-07,
+      "loss": 0.7713,
+      "step": 1066
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.95933400840907e-07,
+      "loss": 0.7208,
+      "step": 1067
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.855659396095183e-07,
+      "loss": 0.6972,
+      "step": 1068
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.752636814956027e-07,
+      "loss": 0.7664,
+      "step": 1069
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.6502669939993e-07,
+      "loss": 0.7246,
+      "step": 1070
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.548550657613651e-07,
+      "loss": 0.7973,
+      "step": 1071
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.447488525563551e-07,
+      "loss": 0.8285,
+      "step": 1072
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.347081312984194e-07,
+      "loss": 0.7857,
+      "step": 1073
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.247329730376429e-07,
+      "loss": 0.7565,
+      "step": 1074
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.148234483601746e-07,
+      "loss": 0.6821,
+      "step": 1075
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.049796273877297e-07,
+      "loss": 0.8031,
+      "step": 1076
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.952015797770862e-07,
+      "loss": 0.7517,
+      "step": 1077
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.854893747196034e-07,
+      "loss": 0.7194,
+      "step": 1078
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.758430809407169e-07,
+      "loss": 0.7053,
+      "step": 1079
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.662627666994725e-07,
+      "loss": 0.7842,
+      "step": 1080
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.567484997880247e-07,
+      "loss": 0.7345,
+      "step": 1081
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.47300347531169e-07,
+      "loss": 0.73,
+      "step": 1082
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.37918376785851e-07,
+      "loss": 0.8712,
+      "step": 1083
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.286026539407164e-07,
+      "loss": 0.8146,
+      "step": 1084
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.193532449156203e-07,
+      "loss": 0.7262,
+      "step": 1085
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.101702151611688e-07,
+      "loss": 0.7452,
+      "step": 1086
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.010536296582592e-07,
+      "loss": 0.7688,
+      "step": 1087
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.920035529176082e-07,
+      "loss": 0.7653,
+      "step": 1088
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.830200489793136e-07,
+      "loss": 0.7568,
+      "step": 1089
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.741031814123843e-07,
+      "loss": 0.711,
+      "step": 1090
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.652530133143042e-07,
+      "loss": 0.7736,
+      "step": 1091
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.564696073105669e-07,
+      "loss": 0.7911,
+      "step": 1092
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.477530255542573e-07,
+      "loss": 0.7315,
+      "step": 1093
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.391033297255932e-07,
+      "loss": 0.7599,
+      "step": 1094
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.305205810314951e-07,
+      "loss": 0.7397,
+      "step": 1095
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.220048402051503e-07,
+      "loss": 0.754,
+      "step": 1096
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.135561675055889e-07,
+      "loss": 0.8184,
+      "step": 1097
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.051746227172538e-07,
+      "loss": 0.7996,
+      "step": 1098
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.96860265149578e-07,
+      "loss": 0.715,
+      "step": 1099
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.886131536365623e-07,
+      "loss": 0.8323,
+      "step": 1100
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.804333465363609e-07,
+      "loss": 0.6795,
+      "step": 1101
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.723209017308727e-07,
+      "loss": 0.7394,
+      "step": 1102
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.6427587662532636e-07,
+      "loss": 0.8024,
+      "step": 1103
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.562983281478761e-07,
+      "loss": 0.7783,
+      "step": 1104
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4838831274919505e-07,
+      "loss": 0.726,
+      "step": 1105
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4054588640208285e-07,
+      "loss": 0.8476,
+      "step": 1106
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.327711046010663e-07,
+      "loss": 0.7458,
+      "step": 1107
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2506402236200616e-07,
+      "loss": 0.8423,
+      "step": 1108
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1742469422170417e-07,
+      "loss": 0.7711,
+      "step": 1109
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0985317423752557e-07,
+      "loss": 0.857,
+      "step": 1110
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0234951598700725e-07,
+      "loss": 0.7365,
+      "step": 1111
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.94913772567489e-07,
+      "loss": 0.6751,
+      "step": 1112
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.875459965957307e-07,
+      "loss": 0.7548,
+      "step": 1113
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.802462402075358e-07,
+      "loss": 0.8037,
+      "step": 1114
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7301455505739494e-07,
+      "loss": 0.7338,
+      "step": 1115
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6585099231810863e-07,
+      "loss": 0.754,
+      "step": 1116
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.587556026804362e-07,
+      "loss": 0.8021,
+      "step": 1117
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5172843635272403e-07,
+      "loss": 0.738,
+      "step": 1118
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4476954306056023e-07,
+      "loss": 0.7903,
+      "step": 1119
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.378789720464193e-07,
+      "loss": 0.7734,
+      "step": 1120
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.31056772069317e-07,
+      "loss": 0.8347,
+      "step": 1121
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2430299140445597e-07,
+      "loss": 0.725,
+      "step": 1122
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.176176778428974e-07,
+      "loss": 0.6322,
+      "step": 1123
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.110008786912122e-07,
+      "loss": 0.7819,
+      "step": 1124
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.044526407711501e-07,
+      "loss": 0.7837,
+      "step": 1125
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.97973010419309e-07,
+      "loss": 0.76,
+      "step": 1126
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.915620334868074e-07,
+      "loss": 0.7325,
+      "step": 1127
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.852197553389568e-07,
+      "loss": 0.7902,
+      "step": 1128
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.789462208549454e-07,
+      "loss": 0.7256,
+      "step": 1129
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.727414744275147e-07,
+      "loss": 0.728,
+      "step": 1130
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.666055599626527e-07,
+      "loss": 0.7852,
+      "step": 1131
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6053852087927436e-07,
+      "loss": 0.7888,
+      "step": 1132
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5454040010892354e-07,
+      "loss": 0.7621,
+      "step": 1133
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.486112400954621e-07,
+      "loss": 0.8157,
+      "step": 1134
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.427510827947721e-07,
+      "loss": 0.7354,
+      "step": 1135
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3695996967446178e-07,
+      "loss": 0.8115,
+      "step": 1136
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3123794171356683e-07,
+      "loss": 0.7528,
+      "step": 1137
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2558503940226296e-07,
+      "loss": 0.7649,
+      "step": 1138
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2000130274158039e-07,
+      "loss": 0.7848,
+      "step": 1139
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.144867712431198e-07,
+      "loss": 0.7815,
+      "step": 1140
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0904148392877354e-07,
+      "loss": 0.7311,
+      "step": 1141
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0366547933044712e-07,
+      "loss": 0.7462,
+      "step": 1142
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.983587954897881e-07,
+      "loss": 0.7176,
+      "step": 1143
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.931214699579176e-07,
+      "loss": 0.6902,
+      "step": 1144
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8795353979516596e-07,
+      "loss": 0.7834,
+      "step": 1145
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8285504157080414e-07,
+      "loss": 0.7117,
+      "step": 1146
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7782601136279277e-07,
+      "loss": 0.7548,
+      "step": 1147
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7286648475752122e-07,
+      "loss": 0.7306,
+      "step": 1148
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6797649684956118e-07,
+      "loss": 0.7424,
+      "step": 1149
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6315608224141023e-07,
+      "loss": 0.8624,
+      "step": 1150
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5840527504325632e-07,
+      "loss": 0.7367,
+      "step": 1151
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5372410887272814e-07,
+      "loss": 0.7027,
+      "step": 1152
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4911261685466416e-07,
+      "loss": 0.7757,
+      "step": 1153
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4457083162087383e-07,
+      "loss": 0.7566,
+      "step": 1154
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4009878530990784e-07,
+      "loss": 0.7135,
+      "step": 1155
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3569650956682944e-07,
+      "loss": 0.7213,
+      "step": 1156
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.313640355429946e-07,
+      "loss": 0.6434,
+      "step": 1157
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2710139389582654e-07,
+      "loss": 0.7905,
+      "step": 1158
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2290861478860272e-07,
+      "loss": 0.847,
+      "step": 1159
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1878572789023935e-07,
+      "loss": 0.7078,
+      "step": 1160
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1473276237507935e-07,
+      "loss": 0.6616,
+      "step": 1161
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1074974692269258e-07,
+      "loss": 0.8124,
+      "step": 1162
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.068367097176659e-07,
+      "loss": 0.7143,
+      "step": 1163
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.02993678449409e-07,
+      "loss": 0.6879,
+      "step": 1164
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.92206803119511e-08,
+      "loss": 0.7451,
+      "step": 1165
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.551774200375896e-08,
+      "loss": 0.7455,
+      "step": 1166
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.188488972753928e-08,
+      "loss": 0.8372,
+      "step": 1167
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.832214919005877e-08,
+      "loss": 0.8751,
+      "step": 1168
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.482954560195655e-08,
+      "loss": 0.7331,
+      "step": 1169
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.14071036775721e-08,
+      "loss": 0.7416,
+      "step": 1170
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.805484763476756e-08,
+      "loss": 0.7256,
+      "step": 1171
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.477280119475239e-08,
+      "loss": 0.8067,
+      "step": 1172
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.156098758192453e-08,
+      "loss": 0.7268,
+      "step": 1173
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.841942952369618e-08,
+      "loss": 0.7826,
+      "step": 1174
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.53481492503405e-08,
+      "loss": 0.7759,
+      "step": 1175
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.23471684948318e-08,
+      "loss": 0.716,
+      "step": 1176
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.9416508492688986e-08,
+      "loss": 0.7764,
+      "step": 1177
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.655618998182899e-08,
+      "loss": 0.7926,
+      "step": 1178
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.376623320241914e-08,
+      "loss": 0.8057,
+      "step": 1179
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.10466578967328e-08,
+      "loss": 0.721,
+      "step": 1180
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8397483309011726e-08,
+      "loss": 0.8369,
+      "step": 1181
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.581872818532729e-08,
+      "loss": 0.8173,
+      "step": 1182
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.331041077344944e-08,
+      "loss": 0.712,
+      "step": 1183
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.087254882271574e-08,
+      "loss": 0.7406,
+      "step": 1184
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8505159583911434e-08,
+      "loss": 0.7365,
+      "step": 1185
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6208259809139554e-08,
+      "loss": 0.6632,
+      "step": 1186
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3981865751705477e-08,
+      "loss": 0.7471,
+      "step": 1187
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.182599316600699e-08,
+      "loss": 0.7481,
+      "step": 1188
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.974065730741327e-08,
+      "loss": 0.745,
+      "step": 1189
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.772587293216611e-08,
+      "loss": 0.7976,
+      "step": 1190
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5781654297267756e-08,
+      "loss": 0.7092,
+      "step": 1191
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.3908015160385433e-08,
+      "loss": 0.7884,
+      "step": 1192
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2104968779752546e-08,
+      "loss": 0.7642,
+      "step": 1193
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.037252791407318e-08,
+      "loss": 0.7772,
+      "step": 1194
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8710704822435534e-08,
+      "loss": 0.772,
+      "step": 1195
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7119511264220844e-08,
+      "loss": 0.7676,
+      "step": 1196
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5598958499021265e-08,
+      "loss": 0.7742,
+      "step": 1197
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4149057286562128e-08,
+      "loss": 0.8167,
+      "step": 1198
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2769817886626456e-08,
+      "loss": 0.7761,
+      "step": 1199
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1461250058977248e-08,
+      "loss": 0.7587,
+      "step": 1200
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.02233630632953e-08,
+      "loss": 0.7846,
+      "step": 1201
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.056165659107053e-09,
+      "loss": 0.6886,
+      "step": 1202
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.959666105727959e-09,
+      "loss": 0.7579,
+      "step": 1203
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.933872162199207e-09,
+      "loss": 0.7947,
+      "step": 1204
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.978791087239977e-09,
+      "loss": 0.757,
+      "step": 1205
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.0944296391863825e-09,
+      "loss": 0.8102,
+      "step": 1206
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.2807940759515e-09,
+      "loss": 0.8435,
+      "step": 1207
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.53789015497763e-09,
+      "loss": 0.8808,
+      "step": 1208
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8657231331985503e-09,
+      "loss": 0.6947,
+      "step": 1209
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.264297766995105e-09,
+      "loss": 0.7229,
+      "step": 1210
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7336183121730022e-09,
+      "loss": 0.8263,
+      "step": 1211
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.273688523919514e-09,
+      "loss": 0.7912,
+      "step": 1212
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.845116567879342e-10,
+      "loss": 0.811,
+      "step": 1213
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.660904646698217e-10,
+      "loss": 0.7072,
+      "step": 1214
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1842720077390754e-10,
+      "loss": 0.756,
+      "step": 1215
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4152361761388122e-10,
+      "loss": 0.7184,
+      "step": 1216
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5380966993958655e-11,
+      "loss": 0.7508,
+      "step": 1217
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.738,
+      "step": 1218
+    },
+    {
+      "epoch": 3.0,
+      "step": 1218,
+      "total_flos": 2.7901147998899405e+17,
+      "train_loss": 1.106034378749005,
+      "train_runtime": 3656.3925,
+      "train_samples_per_second": 42.667,
+      "train_steps_per_second": 0.333
+    }
+  ],
+  "max_steps": 1218,
+  "num_train_epochs": 3,
+  "total_flos": 2.7901147998899405e+17,
+  "trial_name": null,
+  "trial_params": null
+}