diff --git "a/checkpoint-800/trainer_state.json" "b/checkpoint-800/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-800/trainer_state.json"
@@ -0,0 +1,4816 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.969547831436481,
+  "global_step": 800,
+  "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": 1.5039,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0810810810810812e-06,
+      "loss": 1.5785,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6216216216216219e-06,
+      "loss": 1.2981,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1621621621621623e-06,
+      "loss": 1.1631,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.702702702702703e-06,
+      "loss": 1.2493,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2432432432432437e-06,
+      "loss": 1.2413,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7837837837837844e-06,
+      "loss": 1.2368,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.324324324324325e-06,
+      "loss": 1.2152,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.864864864864866e-06,
+      "loss": 1.1487,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 1.116,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.945945945945947e-06,
+      "loss": 1.179,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.486486486486487e-06,
+      "loss": 1.2134,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.027027027027028e-06,
+      "loss": 1.1641,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.567567567567569e-06,
+      "loss": 1.0623,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.108108108108109e-06,
+      "loss": 1.0922,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.64864864864865e-06,
+      "loss": 1.0979,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.189189189189191e-06,
+      "loss": 1.1572,
+      "step": 17
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.729729729729732e-06,
+      "loss": 1.0788,
+      "step": 18
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.027027027027027e-05,
+      "loss": 1.1467,
+      "step": 19
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 1.0852,
+      "step": 20
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1351351351351352e-05,
+      "loss": 1.1176,
+      "step": 21
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1891891891891894e-05,
+      "loss": 1.1049,
+      "step": 22
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2432432432432433e-05,
+      "loss": 1.0972,
+      "step": 23
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2972972972972975e-05,
+      "loss": 1.1404,
+      "step": 24
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3513513513513515e-05,
+      "loss": 1.1416,
+      "step": 25
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4054054054054055e-05,
+      "loss": 1.1458,
+      "step": 26
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4594594594594596e-05,
+      "loss": 1.1497,
+      "step": 27
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5135135135135138e-05,
+      "loss": 1.14,
+      "step": 28
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5675675675675676e-05,
+      "loss": 1.0808,
+      "step": 29
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 1.0996,
+      "step": 30
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6756756756756757e-05,
+      "loss": 1.1064,
+      "step": 31
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.72972972972973e-05,
+      "loss": 1.1823,
+      "step": 32
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.783783783783784e-05,
+      "loss": 1.0483,
+      "step": 33
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8378378378378383e-05,
+      "loss": 1.1019,
+      "step": 34
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.891891891891892e-05,
+      "loss": 1.0325,
+      "step": 35
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9459459459459463e-05,
+      "loss": 1.1006,
+      "step": 36
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 1.0715,
+      "step": 37
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999996461903301e-05,
+      "loss": 1.1542,
+      "step": 38
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999858476382388e-05,
+      "loss": 1.1402,
+      "step": 39
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999681572799226e-05,
+      "loss": 1.0163,
+      "step": 40
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999433909535333e-05,
+      "loss": 1.0851,
+      "step": 41
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999115488343213e-05,
+      "loss": 1.1119,
+      "step": 42
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999872631147608e-05,
+      "loss": 1.0726,
+      "step": 43
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999826638168783e-05,
+      "loss": 1.0927,
+      "step": 44
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997735702233006e-05,
+      "loss": 1.0697,
+      "step": 45
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99971342768668e-05,
+      "loss": 1.0363,
+      "step": 46
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999646210984502e-05,
+      "loss": 1.1251,
+      "step": 47
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999571920592405e-05,
+      "loss": 1.127,
+      "step": 48
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994905570360817e-05,
+      "loss": 1.0994,
+      "step": 49
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999402120891276e-05,
+      "loss": 1.0688,
+      "step": 50
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99930661278378e-05,
+      "loss": 1.0544,
+      "step": 51
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992040333894273e-05,
+      "loss": 1.0908,
+      "step": 52
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990943834340893e-05,
+      "loss": 1.1394,
+      "step": 53
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989776636936705e-05,
+      "loss": 1.1065,
+      "step": 54
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988538749941024e-05,
+      "loss": 1.0752,
+      "step": 55
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987230182113374e-05,
+      "loss": 1.076,
+      "step": 56
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998585094271344e-05,
+      "loss": 1.0504,
+      "step": 57
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998440104150098e-05,
+      "loss": 1.0934,
+      "step": 58
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998288048873578e-05,
+      "loss": 1.1355,
+      "step": 59
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981289295177566e-05,
+      "loss": 1.1693,
+      "step": 60
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979627472085927e-05,
+      "loss": 1.0514,
+      "step": 61
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997789503122025e-05,
+      "loss": 1.119,
+      "step": 62
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9976091984839616e-05,
+      "loss": 1.0635,
+      "step": 63
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974218345702733e-05,
+      "loss": 1.1111,
+      "step": 64
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972274127067838e-05,
+      "loss": 1.1732,
+      "step": 65
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997025934269259e-05,
+      "loss": 1.149,
+      "step": 66
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968174006833996e-05,
+      "loss": 1.1011,
+      "step": 67
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966018134248296e-05,
+      "loss": 1.1367,
+      "step": 68
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963791740190863e-05,
+      "loss": 1.1354,
+      "step": 69
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996149484041609e-05,
+      "loss": 1.0343,
+      "step": 70
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959127451177287e-05,
+      "loss": 1.1997,
+      "step": 71
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956689589226555e-05,
+      "loss": 1.1131,
+      "step": 72
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954181271814673e-05,
+      "loss": 1.0335,
+      "step": 73
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951602516690988e-05,
+      "loss": 1.0786,
+      "step": 74
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948953342103268e-05,
+      "loss": 1.0926,
+      "step": 75
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994623376679758e-05,
+      "loss": 1.1353,
+      "step": 76
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943443810018174e-05,
+      "loss": 1.0605,
+      "step": 77
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940583491507314e-05,
+      "loss": 1.1255,
+      "step": 78
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993765283150517e-05,
+      "loss": 1.0541,
+      "step": 79
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934651850749663e-05,
+      "loss": 1.1197,
+      "step": 80
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931580570476306e-05,
+      "loss": 1.1297,
+      "step": 81
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928439012418076e-05,
+      "loss": 1.1324,
+      "step": 82
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925227198805247e-05,
+      "loss": 1.0846,
+      "step": 83
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921945152365235e-05,
+      "loss": 1.1831,
+      "step": 84
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918592896322432e-05,
+      "loss": 1.0921,
+      "step": 85
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915170454398045e-05,
+      "loss": 1.1037,
+      "step": 86
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911677850809943e-05,
+      "loss": 1.125,
+      "step": 87
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908115110272463e-05,
+      "loss": 1.1279,
+      "step": 88
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904482257996244e-05,
+      "loss": 1.1427,
+      "step": 89
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990077931968805e-05,
+      "loss": 1.1475,
+      "step": 90
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897006321550592e-05,
+      "loss": 1.136,
+      "step": 91
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893163290282335e-05,
+      "loss": 1.0469,
+      "step": 92
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889250253077306e-05,
+      "loss": 1.1028,
+      "step": 93
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885267237624923e-05,
+      "loss": 1.1775,
+      "step": 94
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988121427210976e-05,
+      "loss": 1.1744,
+      "step": 95
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98770913852114e-05,
+      "loss": 1.1049,
+      "step": 96
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872898606104175e-05,
+      "loss": 1.1085,
+      "step": 97
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868635964457007e-05,
+      "loss": 1.1153,
+      "step": 98
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986430349043317e-05,
+      "loss": 1.1637,
+      "step": 99
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859901214690094e-05,
+      "loss": 1.0971,
+      "step": 100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855429168379127e-05,
+      "loss": 1.1537,
+      "step": 101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985088738314534e-05,
+      "loss": 1.0882,
+      "step": 102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846275891127275e-05,
+      "loss": 1.128,
+      "step": 103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841594724956746e-05,
+      "loss": 1.14,
+      "step": 104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836843917758593e-05,
+      "loss": 1.0273,
+      "step": 105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983202350315044e-05,
+      "loss": 1.0891,
+      "step": 106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982713351524248e-05,
+      "loss": 1.106,
+      "step": 107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982217398863721e-05,
+      "loss": 1.0599,
+      "step": 108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98171449584292e-05,
+      "loss": 1.0396,
+      "step": 109
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812046460204837e-05,
+      "loss": 1.0767,
+      "step": 110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806878530042083e-05,
+      "loss": 1.1568,
+      "step": 111
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801641204510216e-05,
+      "loss": 1.1534,
+      "step": 112
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796334520669555e-05,
+      "loss": 1.0332,
+      "step": 113
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790958516071228e-05,
+      "loss": 1.1018,
+      "step": 114
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978551322875688e-05,
+      "loss": 1.0634,
+      "step": 115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977999869725842e-05,
+      "loss": 1.0766,
+      "step": 116
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977441496059774e-05,
+      "loss": 1.0456,
+      "step": 117
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768762058286433e-05,
+      "loss": 1.0047,
+      "step": 118
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976304003032554e-05,
+      "loss": 1.0798,
+      "step": 119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9757248917205228e-05,
+      "loss": 1.1055,
+      "step": 120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975138875990454e-05,
+      "loss": 1.1921,
+      "step": 121
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974545959989108e-05,
+      "loss": 1.0465,
+      "step": 122
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739461479120727e-05,
+      "loss": 1.129,
+      "step": 123
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973339444003735e-05,
+      "loss": 1.0773,
+      "step": 124
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727258525572487e-05,
+      "loss": 1.0955,
+      "step": 125
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721053779145057e-05,
+      "loss": 1.0941,
+      "step": 126
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714780244661044e-05,
+      "loss": 1.1672,
+      "step": 127
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708437966513196e-05,
+      "loss": 1.1014,
+      "step": 128
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702026989580694e-05,
+      "loss": 1.1272,
+      "step": 129
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969554735922885e-05,
+      "loss": 1.0398,
+      "step": 130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968899912130879e-05,
+      "loss": 1.106,
+      "step": 131
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682382322157103e-05,
+      "loss": 1.1033,
+      "step": 132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675697008595545e-05,
+      "loss": 1.0154,
+      "step": 133
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668943227930686e-05,
+      "loss": 1.1219,
+      "step": 134
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966212102795358e-05,
+      "loss": 1.0824,
+      "step": 135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965523045693944e-05,
+      "loss": 1.0581,
+      "step": 136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964827156364728e-05,
+      "loss": 1.1327,
+      "step": 137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964124439731957e-05,
+      "loss": 1.1144,
+      "step": 138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634149007681894e-05,
+      "loss": 1.1436,
+      "step": 139
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962698544494261e-05,
+      "loss": 1.0457,
+      "step": 140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619753759792466e-05,
+      "loss": 1.1053,
+      "step": 141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961245400340427e-05,
+      "loss": 1.0881,
+      "step": 142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605086227432512e-05,
+      "loss": 1.1058,
+      "step": 143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597650484012997e-05,
+      "loss": 1.1418,
+      "step": 144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9590146825762476e-05,
+      "loss": 1.0623,
+      "step": 145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9582575305778297e-05,
+      "loss": 1.0987,
+      "step": 146
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9574935977637994e-05,
+      "loss": 1.0764,
+      "step": 147
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9567228895398936e-05,
+      "loss": 1.153,
+      "step": 148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955945411359792e-05,
+      "loss": 1.1233,
+      "step": 149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9551611687250808e-05,
+      "loss": 1.0827,
+      "step": 150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543701671852127e-05,
+      "loss": 1.0487,
+      "step": 151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535724123374674e-05,
+      "loss": 1.0795,
+      "step": 152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952767909826913e-05,
+      "loss": 1.15,
+      "step": 153
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951956665346364e-05,
+      "loss": 1.0645,
+      "step": 154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951138684636344e-05,
+      "loss": 1.0764,
+      "step": 155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9503139734850426e-05,
+      "loss": 1.1185,
+      "step": 156
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9494825377282746e-05,
+      "loss": 1.0879,
+      "step": 157
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486443832494414e-05,
+      "loss": 1.0558,
+      "step": 158
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477995159794854e-05,
+      "loss": 1.1042,
+      "step": 159
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9469479418968506e-05,
+      "loss": 1.1029,
+      "step": 160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9460896670274408e-05,
+      "loss": 0.9836,
+      "step": 161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9452246974445743e-05,
+      "loss": 1.1325,
+      "step": 162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443530392689434e-05,
+      "loss": 1.1018,
+      "step": 163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94347469866857e-05,
+      "loss": 1.1525,
+      "step": 164
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9425896818587615e-05,
+      "loss": 1.0719,
+      "step": 165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941697995102069e-05,
+      "loss": 1.1595,
+      "step": 166
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9407996447082394e-05,
+      "loss": 1.1698,
+      "step": 167
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939894637034174e-05,
+      "loss": 1.131,
+      "step": 168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9389829784838833e-05,
+      "loss": 1.0604,
+      "step": 169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938064675508438e-05,
+      "loss": 1.0422,
+      "step": 170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9371397346059286e-05,
+      "loss": 1.1098,
+      "step": 171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936208162321415e-05,
+      "loss": 1.0682,
+      "step": 172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352699652468835e-05,
+      "loss": 1.0763,
+      "step": 173
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9343251500211977e-05,
+      "loss": 1.1038,
+      "step": 174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933373723330053e-05,
+      "loss": 1.0613,
+      "step": 175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9324156919059286e-05,
+      "loss": 1.1157,
+      "step": 176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.93145106252804e-05,
+      "loss": 1.1264,
+      "step": 177
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9304798420222918e-05,
+      "loss": 1.0787,
+      "step": 178
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9295020372612276e-05,
+      "loss": 1.1298,
+      "step": 179
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9285176551639826e-05,
+      "loss": 1.1175,
+      "step": 180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9275267026962358e-05,
+      "loss": 1.1686,
+      "step": 181
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9265291868701584e-05,
+      "loss": 1.0939,
+      "step": 182
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9255251147443646e-05,
+      "loss": 1.1674,
+      "step": 183
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924514493423864e-05,
+      "loss": 1.0747,
+      "step": 184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9234973300600074e-05,
+      "loss": 1.1085,
+      "step": 185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92247363185044e-05,
+      "loss": 1.1157,
+      "step": 186
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9214434060390484e-05,
+      "loss": 1.1827,
+      "step": 187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9204066599159094e-05,
+      "loss": 1.1126,
+      "step": 188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9193634008172396e-05,
+      "loss": 1.1158,
+      "step": 189
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9183136361253417e-05,
+      "loss": 1.0609,
+      "step": 190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917257373268554e-05,
+      "loss": 1.0617,
+      "step": 191
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916194619721196e-05,
+      "loss": 1.0724,
+      "step": 192
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915125383003518e-05,
+      "loss": 1.0784,
+      "step": 193
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914049670681646e-05,
+      "loss": 1.1141,
+      "step": 194
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912967490367528e-05,
+      "loss": 1.0305,
+      "step": 195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9118788497188815e-05,
+      "loss": 1.026,
+      "step": 196
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9107837564391376e-05,
+      "loss": 1.0663,
+      "step": 197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9096822182773887e-05,
+      "loss": 1.0368,
+      "step": 198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9085742430283322e-05,
+      "loss": 1.1052,
+      "step": 199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907459838532215e-05,
+      "loss": 1.0834,
+      "step": 200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9063390126747778e-05,
+      "loss": 1.1579,
+      "step": 201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9052117733872025e-05,
+      "loss": 1.1034,
+      "step": 202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904078128646052e-05,
+      "loss": 1.1173,
+      "step": 203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902938086473215e-05,
+      "loss": 1.1546,
+      "step": 204
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901791654935852e-05,
+      "loss": 1.1274,
+      "step": 205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9006388421463322e-05,
+      "loss": 1.0896,
+      "step": 206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899479656262183e-05,
+      "loss": 1.1937,
+      "step": 207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898314105486028e-05,
+      "loss": 1.0613,
+      "step": 208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8971421980655295e-05,
+      "loss": 1.0896,
+      "step": 209
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8959639422933316e-05,
+      "loss": 1.0578,
+      "step": 210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894779346506999e-05,
+      "loss": 1.0935,
+      "step": 211
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893588419088962e-05,
+      "loss": 1.0655,
+      "step": 212
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892391168466452e-05,
+      "loss": 1.1562,
+      "step": 213
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891187603111447e-05,
+      "loss": 1.1434,
+      "step": 214
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8899777315406073e-05,
+      "loss": 1.0462,
+      "step": 215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8887615623152188e-05,
+      "loss": 1.1151,
+      "step": 216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88753910404113e-05,
+      "loss": 1.1457,
+      "step": 217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8863103653686917e-05,
+      "loss": 1.0401,
+      "step": 218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8850753549926967e-05,
+      "loss": 1.074,
+      "step": 219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8838340816523175e-05,
+      "loss": 1.14,
+      "step": 220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8825865541310438e-05,
+      "loss": 1.1483,
+      "step": 221
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8813327812566217e-05,
+      "loss": 1.156,
+      "step": 222
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880072771900991e-05,
+      "loss": 1.0334,
+      "step": 223
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878806534980221e-05,
+      "loss": 1.1854,
+      "step": 224
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8775340794544497e-05,
+      "loss": 1.0593,
+      "step": 225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876255414327818e-05,
+      "loss": 1.0863,
+      "step": 226
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8749705486484074e-05,
+      "loss": 1.2038,
+      "step": 227
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8736794915081765e-05,
+      "loss": 1.0991,
+      "step": 228
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8723822520428954e-05,
+      "loss": 1.0572,
+      "step": 229
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8710788394320807e-05,
+      "loss": 1.0798,
+      "step": 230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8697692628989327e-05,
+      "loss": 1.082,
+      "step": 231
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868453531710268e-05,
+      "loss": 1.1121,
+      "step": 232
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8671316551764552e-05,
+      "loss": 1.089,
+      "step": 233
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865803642651348e-05,
+      "loss": 1.1132,
+      "step": 234
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8644695035322203e-05,
+      "loss": 1.1323,
+      "step": 235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8631292472596978e-05,
+      "loss": 1.1352,
+      "step": 236
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8617828833176935e-05,
+      "loss": 1.0801,
+      "step": 237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860430421233339e-05,
+      "loss": 1.1049,
+      "step": 238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859071870576918e-05,
+      "loss": 1.0518,
+      "step": 239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857707240961797e-05,
+      "loss": 1.0691,
+      "step": 240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8563365420443594e-05,
+      "loss": 1.0941,
+      "step": 241
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854959783523936e-05,
+      "loss": 1.1016,
+      "step": 242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853576975142736e-05,
+      "loss": 1.0896,
+      "step": 243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852188126685779e-05,
+      "loss": 1.0822,
+      "step": 244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8507932479808254e-05,
+      "loss": 0.9657,
+      "step": 245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8493923488983066e-05,
+      "loss": 1.0548,
+      "step": 246
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847985439351256e-05,
+      "loss": 1.0908,
+      "step": 247
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846572529295237e-05,
+      "loss": 1.1709,
+      "step": 248
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845153628728274e-05,
+      "loss": 1.1686,
+      "step": 249
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8437287476907828e-05,
+      "loss": 1.1704,
+      "step": 250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842297896265497e-05,
+      "loss": 1.054,
+      "step": 251
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8408610845773974e-05,
+      "loss": 1.1077,
+      "step": 252
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8394183227936418e-05,
+      "loss": 1.0797,
+      "step": 253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8379696211234918e-05,
+      "loss": 1.1079,
+      "step": 254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8365149898182403e-05,
+      "loss": 1.11,
+      "step": 255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8350544391711396e-05,
+      "loss": 1.0777,
+      "step": 256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833587979517329e-05,
+      "loss": 1.0761,
+      "step": 257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8321156212337604e-05,
+      "loss": 1.0941,
+      "step": 258
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830637374739126e-05,
+      "loss": 1.128,
+      "step": 259
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829153250493783e-05,
+      "loss": 1.0112,
+      "step": 260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827663258999683e-05,
+      "loss": 1.0975,
+      "step": 261
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8261674108002925e-05,
+      "loss": 1.0442,
+      "step": 262
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824665716480524e-05,
+      "loss": 1.0653,
+      "step": 263
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823158186666656e-05,
+      "loss": 1.0561,
+      "step": 264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821644832026261e-05,
+      "loss": 1.1615,
+      "step": 265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82012566326813e-05,
+      "loss": 1.1045,
+      "step": 266
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8186006911421937e-05,
+      "loss": 1.1486,
+      "step": 267
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817069926439451e-05,
+      "loss": 1.1622,
+      "step": 268
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8155333799918883e-05,
+      "loss": 1.0986,
+      "step": 269
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8139910626724058e-05,
+      "loss": 1.0416,
+      "step": 270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8124429853947387e-05,
+      "loss": 1.1328,
+      "step": 271
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8108891591133812e-05,
+      "loss": 1.0921,
+      "step": 272
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809329594823509e-05,
+      "loss": 1.0795,
+      "step": 273
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8077643035609006e-05,
+      "loss": 1.1597,
+      "step": 274
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806193296401859e-05,
+      "loss": 1.0555,
+      "step": 275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804616584463136e-05,
+      "loss": 1.1308,
+      "step": 276
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803034178901849e-05,
+      "loss": 1.0728,
+      "step": 277
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8014460909154058e-05,
+      "loss": 1.0919,
+      "step": 278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799852331741425e-05,
+      "loss": 1.1573,
+      "step": 279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7982529126576543e-05,
+      "loss": 1.0903,
+      "step": 280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7966478449818925e-05,
+      "loss": 1.1511,
+      "step": 281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7950371400719087e-05,
+      "loss": 1.1447,
+      "step": 282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7934208093253625e-05,
+      "loss": 1.1149,
+      "step": 283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7917988641797227e-05,
+      "loss": 1.1297,
+      "step": 284
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7901713161121873e-05,
+      "loss": 1.0903,
+      "step": 285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7885381766396008e-05,
+      "loss": 1.0677,
+      "step": 286
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786899457318374e-05,
+      "loss": 1.0687,
+      "step": 287
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7852551697444017e-05,
+      "loss": 1.0536,
+      "step": 288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783605325552981e-05,
+      "loss": 1.0364,
+      "step": 289
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7819499364187282e-05,
+      "loss": 1.1249,
+      "step": 290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780289014055497e-05,
+      "loss": 1.0534,
+      "step": 291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7786225702162955e-05,
+      "loss": 1.0512,
+      "step": 292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7769506166932026e-05,
+      "loss": 1.1828,
+      "step": 293
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7752731653172847e-05,
+      "loss": 1.064,
+      "step": 294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7735902279585118e-05,
+      "loss": 1.0973,
+      "step": 295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7719018165256745e-05,
+      "loss": 1.0576,
+      "step": 296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7702079429662986e-05,
+      "loss": 1.1355,
+      "step": 297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7685086192665605e-05,
+      "loss": 0.9462,
+      "step": 298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7668038574512045e-05,
+      "loss": 1.0778,
+      "step": 299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7650936695834536e-05,
+      "loss": 1.1013,
+      "step": 300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763378067764929e-05,
+      "loss": 1.0889,
+      "step": 301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7616570641355602e-05,
+      "loss": 1.013,
+      "step": 302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759930670873502e-05,
+      "loss": 1.0847,
+      "step": 303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758198900195047e-05,
+      "loss": 1.1345,
+      "step": 304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7564617643545395e-05,
+      "loss": 1.0481,
+      "step": 305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7547192756442887e-05,
+      "loss": 1.0952,
+      "step": 306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7529714463944815e-05,
+      "loss": 1.0869,
+      "step": 307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751218288973096e-05,
+      "loss": 1.1197,
+      "step": 308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7494598157858127e-05,
+      "loss": 1.1011,
+      "step": 309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7476960392759284e-05,
+      "loss": 1.079,
+      "step": 310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7459269719242665e-05,
+      "loss": 1.0569,
+      "step": 311
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74415262624909e-05,
+      "loss": 1.1322,
+      "step": 312
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742373014806012e-05,
+      "loss": 1.121,
+      "step": 313
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740588150187907e-05,
+      "loss": 1.1544,
+      "step": 314
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7387980450248222e-05,
+      "loss": 1.1591,
+      "step": 315
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7370027119838884e-05,
+      "loss": 1.0956,
+      "step": 316
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735202163769229e-05,
+      "loss": 1.1096,
+      "step": 317
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7333964131218714e-05,
+      "loss": 1.0962,
+      "step": 318
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7315854728196568e-05,
+      "loss": 1.1088,
+      "step": 319
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729769355677149e-05,
+      "loss": 1.0693,
+      "step": 320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7279480745455433e-05,
+      "loss": 1.0909,
+      "step": 321
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7261216423125782e-05,
+      "loss": 1.0636,
+      "step": 322
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724290071902441e-05,
+      "loss": 1.131,
+      "step": 323
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7224533762756775e-05,
+      "loss": 1.1824,
+      "step": 324
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720611568429103e-05,
+      "loss": 1.002,
+      "step": 325
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718764661395704e-05,
+      "loss": 1.0908,
+      "step": 326
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.716912668244553e-05,
+      "loss": 1.1213,
+      "step": 327
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715055602080711e-05,
+      "loss": 1.1056,
+      "step": 328
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7131934760451385e-05,
+      "loss": 1.073,
+      "step": 329
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7113263033145985e-05,
+      "loss": 1.1817,
+      "step": 330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7094540971015663e-05,
+      "loss": 1.0157,
+      "step": 331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7075768706541355e-05,
+      "loss": 1.0835,
+      "step": 332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7056946372559234e-05,
+      "loss": 0.9943,
+      "step": 333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7038074102259775e-05,
+      "loss": 1.1713,
+      "step": 334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7019152029186817e-05,
+      "loss": 1.0421,
+      "step": 335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.70001802872366e-05,
+      "loss": 1.1,
+      "step": 336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6981159010656847e-05,
+      "loss": 1.0711,
+      "step": 337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6962088334045785e-05,
+      "loss": 1.1572,
+      "step": 338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694296839235121e-05,
+      "loss": 1.0448,
+      "step": 339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692379932086953e-05,
+      "loss": 1.0735,
+      "step": 340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6904581255244802e-05,
+      "loss": 1.0977,
+      "step": 341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688531433146777e-05,
+      "loss": 1.0777,
+      "step": 342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6865998685874923e-05,
+      "loss": 1.077,
+      "step": 343
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6846634455147498e-05,
+      "loss": 1.0797,
+      "step": 344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6827221776310532e-05,
+      "loss": 1.0048,
+      "step": 345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6807760786731905e-05,
+      "loss": 1.0392,
+      "step": 346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6788251624121335e-05,
+      "loss": 1.1693,
+      "step": 347
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6768694426529432e-05,
+      "loss": 1.1036,
+      "step": 348
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6749089332346714e-05,
+      "loss": 1.0874,
+      "step": 349
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672943648030261e-05,
+      "loss": 1.1224,
+      "step": 350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6709736009464504e-05,
+      "loss": 1.0791,
+      "step": 351
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668998805923675e-05,
+      "loss": 1.0822,
+      "step": 352
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6670192769359643e-05,
+      "loss": 1.0818,
+      "step": 353
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6650350279908497e-05,
+      "loss": 1.1335,
+      "step": 354
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6630460731292597e-05,
+      "loss": 1.0699,
+      "step": 355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661052426425424e-05,
+      "loss": 1.1318,
+      "step": 356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6590541019867722e-05,
+      "loss": 1.1362,
+      "step": 357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6570511139538348e-05,
+      "loss": 1.0223,
+      "step": 358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655043476500142e-05,
+      "loss": 1.0471,
+      "step": 359
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6530312038321247e-05,
+      "loss": 1.1665,
+      "step": 360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6510143101890136e-05,
+      "loss": 1.0409,
+      "step": 361
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6489928098427383e-05,
+      "loss": 1.1231,
+      "step": 362
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6469667170978258e-05,
+      "loss": 1.0062,
+      "step": 363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6449360462913005e-05,
+      "loss": 1.0764,
+      "step": 364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642900811792582e-05,
+      "loss": 1.1421,
+      "step": 365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640861028003383e-05,
+      "loss": 1.0809,
+      "step": 366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6388167093576083e-05,
+      "loss": 1.1295,
+      "step": 367
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6367678703212515e-05,
+      "loss": 1.0813,
+      "step": 368
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6347145253922942e-05,
+      "loss": 1.1542,
+      "step": 369
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632656689100602e-05,
+      "loss": 1.0917,
+      "step": 370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6305943760078226e-05,
+      "loss": 1.1101,
+      "step": 371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.628527600707283e-05,
+      "loss": 1.0929,
+      "step": 372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6264563778238834e-05,
+      "loss": 1.074,
+      "step": 373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6243807220139988e-05,
+      "loss": 1.0869,
+      "step": 374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6223006479653708e-05,
+      "loss": 1.0546,
+      "step": 375
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6202161703970057e-05,
+      "loss": 1.0218,
+      "step": 376
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6181273040590696e-05,
+      "loss": 1.054,
+      "step": 377
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616034063732785e-05,
+      "loss": 0.9993,
+      "step": 378
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613936464230325e-05,
+      "loss": 1.0481,
+      "step": 379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6118345203947093e-05,
+      "loss": 1.1722,
+      "step": 380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6097282470996997e-05,
+      "loss": 1.1137,
+      "step": 381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6076176592496926e-05,
+      "loss": 1.0729,
+      "step": 382
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605502771779616e-05,
+      "loss": 1.0825,
+      "step": 383
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603383599654823e-05,
+      "loss": 1.0767,
+      "step": 384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601260157870985e-05,
+      "loss": 1.1263,
+      "step": 385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599132461453987e-05,
+      "loss": 1.0745,
+      "step": 386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5970005254598204e-05,
+      "loss": 1.1024,
+      "step": 387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.594864364974476e-05,
+      "loss": 1.1268,
+      "step": 388
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592723995113839e-05,
+      "loss": 1.0998,
+      "step": 389
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5905794310235808e-05,
+      "loss": 1.0569,
+      "step": 390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5884306878790512e-05,
+      "loss": 1.0528,
+      "step": 391
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586277780885172e-05,
+      "loss": 1.0959,
+      "step": 392
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58412072527633e-05,
+      "loss": 1.167,
+      "step": 393
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5819595363162682e-05,
+      "loss": 1.11,
+      "step": 394
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5797942292979767e-05,
+      "loss": 1.1125,
+      "step": 395
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577624819543587e-05,
+      "loss": 1.1226,
+      "step": 396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5754513224042625e-05,
+      "loss": 1.134,
+      "step": 397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573273753260089e-05,
+      "loss": 1.1455,
+      "step": 398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571092127519967e-05,
+      "loss": 1.0762,
+      "step": 399
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568906460621502e-05,
+      "loss": 1.0632,
+      "step": 400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566716768030896e-05,
+      "loss": 1.1466,
+      "step": 401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5645230652428367e-05,
+      "loss": 1.035,
+      "step": 402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5623253677803897e-05,
+      "loss": 1.0866,
+      "step": 403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5601236911948876e-05,
+      "loss": 1.1167,
+      "step": 404
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5579180510658187e-05,
+      "loss": 1.1197,
+      "step": 405
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5557084630007206e-05,
+      "loss": 1.1931,
+      "step": 406
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5534949426350642e-05,
+      "loss": 1.1023,
+      "step": 407
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551277505632149e-05,
+      "loss": 0.6942,
+      "step": 408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549056167682987e-05,
+      "loss": 0.8178,
+      "step": 409
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546830944506196e-05,
+      "loss": 0.7676,
+      "step": 410
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544601851847885e-05,
+      "loss": 0.7094,
+      "step": 411
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542368905481545e-05,
+      "loss": 0.7146,
+      "step": 412
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5401321212079366e-05,
+      "loss": 0.7071,
+      "step": 413
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5378915148549772e-05,
+      "loss": 0.7295,
+      "step": 414
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5356471022776315e-05,
+      "loss": 0.6508,
+      "step": 415
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5333988993577958e-05,
+      "loss": 0.6957,
+      "step": 416
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5311469220041903e-05,
+      "loss": 0.7019,
+      "step": 417
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5288911861522413e-05,
+      "loss": 0.7552,
+      "step": 418
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526631707763972e-05,
+      "loss": 0.8051,
+      "step": 419
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5243685028278888e-05,
+      "loss": 0.6554,
+      "step": 420
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5221015873588672e-05,
+      "loss": 0.6971,
+      "step": 421
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5198309773980397e-05,
+      "loss": 0.6483,
+      "step": 422
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5175566890126812e-05,
+      "loss": 0.7892,
+      "step": 423
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5152787382960968e-05,
+      "loss": 0.6725,
+      "step": 424
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5129971413675055e-05,
+      "loss": 0.6964,
+      "step": 425
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.510711914371929e-05,
+      "loss": 0.7293,
+      "step": 426
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5084230734800754e-05,
+      "loss": 0.6611,
+      "step": 427
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5061306348882252e-05,
+      "loss": 0.7748,
+      "step": 428
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5038346148181178e-05,
+      "loss": 0.7615,
+      "step": 429
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5015350295168344e-05,
+      "loss": 0.7051,
+      "step": 430
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4992318952566862e-05,
+      "loss": 0.7459,
+      "step": 431
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4969252283350964e-05,
+      "loss": 0.6851,
+      "step": 432
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4946150450744859e-05,
+      "loss": 0.7352,
+      "step": 433
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4923013618221584e-05,
+      "loss": 0.7288,
+      "step": 434
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4899841949501845e-05,
+      "loss": 0.7334,
+      "step": 435
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4876635608552845e-05,
+      "loss": 0.689,
+      "step": 436
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4853394759587146e-05,
+      "loss": 0.7024,
+      "step": 437
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4830119567061484e-05,
+      "loss": 0.7097,
+      "step": 438
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4806810195675627e-05,
+      "loss": 0.658,
+      "step": 439
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4783466810371195e-05,
+      "loss": 0.7137,
+      "step": 440
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4760089576330493e-05,
+      "loss": 0.6683,
+      "step": 441
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4736678658975357e-05,
+      "loss": 0.7584,
+      "step": 442
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.471323422396596e-05,
+      "loss": 0.7649,
+      "step": 443
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4689756437199658e-05,
+      "loss": 0.7315,
+      "step": 444
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4666245464809818e-05,
+      "loss": 0.7138,
+      "step": 445
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4642701473164618e-05,
+      "loss": 0.6707,
+      "step": 446
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4619124628865904e-05,
+      "loss": 0.7463,
+      "step": 447
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.459551509874798e-05,
+      "loss": 0.6355,
+      "step": 448
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4571873049876452e-05,
+      "loss": 0.7555,
+      "step": 449
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.454819864954703e-05,
+      "loss": 0.6682,
+      "step": 450
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4524492065284344e-05,
+      "loss": 0.6952,
+      "step": 451
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4500753464840775e-05,
+      "loss": 0.7286,
+      "step": 452
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4476983016195245e-05,
+      "loss": 0.7176,
+      "step": 453
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4453180887552052e-05,
+      "loss": 0.7459,
+      "step": 454
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4429347247339656e-05,
+      "loss": 0.7346,
+      "step": 455
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4405482264209512e-05,
+      "loss": 0.7064,
+      "step": 456
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4381586107034849e-05,
+      "loss": 0.7712,
+      "step": 457
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4357658944909496e-05,
+      "loss": 0.7302,
+      "step": 458
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4333700947146686e-05,
+      "loss": 0.7117,
+      "step": 459
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4309712283277839e-05,
+      "loss": 0.6974,
+      "step": 460
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4285693123051385e-05,
+      "loss": 0.7067,
+      "step": 461
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4261643636431539e-05,
+      "loss": 0.7232,
+      "step": 462
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4237563993597133e-05,
+      "loss": 0.7155,
+      "step": 463
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4213454364940362e-05,
+      "loss": 0.7625,
+      "step": 464
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4189314921065629e-05,
+      "loss": 0.7497,
+      "step": 465
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4165145832788305e-05,
+      "loss": 0.7078,
+      "step": 466
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4140947271133536e-05,
+      "loss": 0.6694,
+      "step": 467
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4116719407335022e-05,
+      "loss": 0.7768,
+      "step": 468
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4092462412833811e-05,
+      "loss": 0.6864,
+      "step": 469
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.406817645927709e-05,
+      "loss": 0.6759,
+      "step": 470
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4043861718516964e-05,
+      "loss": 0.7199,
+      "step": 471
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4019518362609239e-05,
+      "loss": 0.7102,
+      "step": 472
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.399514656381221e-05,
+      "loss": 0.7403,
+      "step": 473
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3970746494585439e-05,
+      "loss": 0.7286,
+      "step": 474
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3946318327588534e-05,
+      "loss": 0.7331,
+      "step": 475
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3921862235679929e-05,
+      "loss": 0.7195,
+      "step": 476
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.389737839191566e-05,
+      "loss": 0.6506,
+      "step": 477
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3872866969548143e-05,
+      "loss": 0.6913,
+      "step": 478
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.384832814202494e-05,
+      "loss": 0.7337,
+      "step": 479
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3823762082987544e-05,
+      "loss": 0.7577,
+      "step": 480
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3799168966270139e-05,
+      "loss": 0.6808,
+      "step": 481
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3774548965898371e-05,
+      "loss": 0.7027,
+      "step": 482
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3749902256088125e-05,
+      "loss": 0.7166,
+      "step": 483
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3725229011244294e-05,
+      "loss": 0.8069,
+      "step": 484
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3700529405959517e-05,
+      "loss": 0.6814,
+      "step": 485
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3675803615012993e-05,
+      "loss": 0.7018,
+      "step": 486
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3651051813369188e-05,
+      "loss": 0.742,
+      "step": 487
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3626274176176645e-05,
+      "loss": 0.6845,
+      "step": 488
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3601470878766714e-05,
+      "loss": 0.7453,
+      "step": 489
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3576642096652322e-05,
+      "loss": 0.7421,
+      "step": 490
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3551788005526738e-05,
+      "loss": 0.7583,
+      "step": 491
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3526908781262314e-05,
+      "loss": 0.7492,
+      "step": 492
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3502004599909255e-05,
+      "loss": 0.7243,
+      "step": 493
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3477075637694362e-05,
+      "loss": 0.7576,
+      "step": 494
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3452122071019797e-05,
+      "loss": 0.7244,
+      "step": 495
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3427144076461818e-05,
+      "loss": 0.6517,
+      "step": 496
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3402141830769551e-05,
+      "loss": 0.6985,
+      "step": 497
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3377115510863716e-05,
+      "loss": 0.7531,
+      "step": 498
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3352065293835399e-05,
+      "loss": 0.6773,
+      "step": 499
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3326991356944776e-05,
+      "loss": 0.7243,
+      "step": 500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3301893877619874e-05,
+      "loss": 0.6769,
+      "step": 501
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3276773033455312e-05,
+      "loss": 0.6559,
+      "step": 502
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3251629002211042e-05,
+      "loss": 0.8066,
+      "step": 503
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.322646196181109e-05,
+      "loss": 0.7142,
+      "step": 504
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3201272090342303e-05,
+      "loss": 0.7048,
+      "step": 505
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3176059566053083e-05,
+      "loss": 0.7143,
+      "step": 506
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3150824567352128e-05,
+      "loss": 0.6842,
+      "step": 507
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3125567272807167e-05,
+      "loss": 0.7343,
+      "step": 508
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3100287861143703e-05,
+      "loss": 0.6544,
+      "step": 509
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3074986511243741e-05,
+      "loss": 0.6697,
+      "step": 510
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3049663402144528e-05,
+      "loss": 0.6752,
+      "step": 511
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.302431871303728e-05,
+      "loss": 0.7709,
+      "step": 512
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2998952623265917e-05,
+      "loss": 0.7113,
+      "step": 513
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2973565312325798e-05,
+      "loss": 0.7121,
+      "step": 514
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2948156959862446e-05,
+      "loss": 0.7528,
+      "step": 515
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2922727745670276e-05,
+      "loss": 0.7406,
+      "step": 516
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2897277849691326e-05,
+      "loss": 0.7273,
+      "step": 517
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2871807452013977e-05,
+      "loss": 0.6828,
+      "step": 518
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.28463167328717e-05,
+      "loss": 0.7069,
+      "step": 519
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2820805872641745e-05,
+      "loss": 0.7329,
+      "step": 520
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2795275051843893e-05,
+      "loss": 0.6676,
+      "step": 521
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.276972445113917e-05,
+      "loss": 0.7041,
+      "step": 522
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2744154251328573e-05,
+      "loss": 0.6867,
+      "step": 523
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2718564633351773e-05,
+      "loss": 0.7061,
+      "step": 524
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2692955778285865e-05,
+      "loss": 0.6561,
+      "step": 525
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.266732786734405e-05,
+      "loss": 0.7576,
+      "step": 526
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2641681081874394e-05,
+      "loss": 0.6961,
+      "step": 527
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2616015603358497e-05,
+      "loss": 0.7016,
+      "step": 528
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2590331613410261e-05,
+      "loss": 0.6771,
+      "step": 529
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2564629293774561e-05,
+      "loss": 0.7011,
+      "step": 530
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.253890882632598e-05,
+      "loss": 0.734,
+      "step": 531
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2513170393067527e-05,
+      "loss": 0.7822,
+      "step": 532
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2487414176129322e-05,
+      "loss": 0.7276,
+      "step": 533
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246164035776735e-05,
+      "loss": 0.7597,
+      "step": 534
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2435849120362123e-05,
+      "loss": 0.6397,
+      "step": 535
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2410040646417431e-05,
+      "loss": 0.7549,
+      "step": 536
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2384215118559027e-05,
+      "loss": 0.6502,
+      "step": 537
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.235837271953334e-05,
+      "loss": 0.7279,
+      "step": 538
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2332513632206183e-05,
+      "loss": 0.7003,
+      "step": 539
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2306638039561455e-05,
+      "loss": 0.711,
+      "step": 540
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2280746124699864e-05,
+      "loss": 0.6959,
+      "step": 541
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2254838070837596e-05,
+      "loss": 0.6984,
+      "step": 542
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2228914061305059e-05,
+      "loss": 0.6968,
+      "step": 543
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2202974279545554e-05,
+      "loss": 0.664,
+      "step": 544
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2177018909113994e-05,
+      "loss": 0.65,
+      "step": 545
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.21510481336756e-05,
+      "loss": 0.7142,
+      "step": 546
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2125062137004602e-05,
+      "loss": 0.7178,
+      "step": 547
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2099061102982939e-05,
+      "loss": 0.7338,
+      "step": 548
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2073045215598953e-05,
+      "loss": 0.6878,
+      "step": 549
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.20470146589461e-05,
+      "loss": 0.7476,
+      "step": 550
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2020969617221627e-05,
+      "loss": 0.6574,
+      "step": 551
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.199491027472529e-05,
+      "loss": 0.716,
+      "step": 552
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1968836815858038e-05,
+      "loss": 0.7099,
+      "step": 553
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1942749425120704e-05,
+      "loss": 0.7441,
+      "step": 554
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1916648287112714e-05,
+      "loss": 0.6379,
+      "step": 555
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1890533586530766e-05,
+      "loss": 0.6636,
+      "step": 556
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1864405508167532e-05,
+      "loss": 0.7175,
+      "step": 557
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1838264236910348e-05,
+      "loss": 0.6754,
+      "step": 558
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1812109957739907e-05,
+      "loss": 0.6965,
+      "step": 559
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1785942855728945e-05,
+      "loss": 0.7278,
+      "step": 560
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1759763116040936e-05,
+      "loss": 0.6989,
+      "step": 561
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1733570923928785e-05,
+      "loss": 0.7505,
+      "step": 562
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1707366464733501e-05,
+      "loss": 0.7006,
+      "step": 563
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1681149923882913e-05,
+      "loss": 0.6857,
+      "step": 564
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1654921486890327e-05,
+      "loss": 0.679,
+      "step": 565
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1628681339353244e-05,
+      "loss": 0.6216,
+      "step": 566
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1602429666952015e-05,
+      "loss": 0.7072,
+      "step": 567
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1576166655448558e-05,
+      "loss": 0.7212,
+      "step": 568
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1549892490685018e-05,
+      "loss": 0.6965,
+      "step": 569
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1523607358582462e-05,
+      "loss": 0.7068,
+      "step": 570
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.149731144513958e-05,
+      "loss": 0.7301,
+      "step": 571
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1471004936431327e-05,
+      "loss": 0.6699,
+      "step": 572
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.144468801860766e-05,
+      "loss": 0.7794,
+      "step": 573
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1418360877892165e-05,
+      "loss": 0.7181,
+      "step": 574
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1392023700580796e-05,
+      "loss": 0.6989,
+      "step": 575
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1365676673040502e-05,
+      "loss": 0.7535,
+      "step": 576
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.133931998170795e-05,
+      "loss": 0.737,
+      "step": 577
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1312953813088183e-05,
+      "loss": 0.7381,
+      "step": 578
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1286578353753313e-05,
+      "loss": 0.6927,
+      "step": 579
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1260193790341186e-05,
+      "loss": 0.6992,
+      "step": 580
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1233800309554083e-05,
+      "loss": 0.7445,
+      "step": 581
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1207398098157371e-05,
+      "loss": 0.7359,
+      "step": 582
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1180987342978209e-05,
+      "loss": 0.6505,
+      "step": 583
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1154568230904204e-05,
+      "loss": 0.6764,
+      "step": 584
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1128140948882107e-05,
+      "loss": 0.7704,
+      "step": 585
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1101705683916473e-05,
+      "loss": 0.7286,
+      "step": 586
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1075262623068352e-05,
+      "loss": 0.6551,
+      "step": 587
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1048811953453955e-05,
+      "loss": 0.7762,
+      "step": 588
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1022353862243338e-05,
+      "loss": 0.7439,
+      "step": 589
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0995888536659067e-05,
+      "loss": 0.7173,
+      "step": 590
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.096941616397491e-05,
+      "loss": 0.684,
+      "step": 591
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0942936931514492e-05,
+      "loss": 0.6892,
+      "step": 592
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0916451026649981e-05,
+      "loss": 0.7217,
+      "step": 593
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.088995863680077e-05,
+      "loss": 0.6731,
+      "step": 594
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0863459949432122e-05,
+      "loss": 0.7102,
+      "step": 595
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0836955152053883e-05,
+      "loss": 0.7056,
+      "step": 596
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.081044443221912e-05,
+      "loss": 0.6561,
+      "step": 597
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0783927977522819e-05,
+      "loss": 0.7552,
+      "step": 598
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0757405975600534e-05,
+      "loss": 0.7665,
+      "step": 599
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0730878614127087e-05,
+      "loss": 0.6167,
+      "step": 600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0704346080815218e-05,
+      "loss": 0.6866,
+      "step": 601
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0677808563414256e-05,
+      "loss": 0.7526,
+      "step": 602
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0651266249708816e-05,
+      "loss": 0.6667,
+      "step": 603
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0624719327517434e-05,
+      "loss": 0.6968,
+      "step": 604
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0598167984691276e-05,
+      "loss": 0.7331,
+      "step": 605
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057161240911277e-05,
+      "loss": 0.7514,
+      "step": 606
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0545052788694312e-05,
+      "loss": 0.6374,
+      "step": 607
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0518489311376905e-05,
+      "loss": 0.7421,
+      "step": 608
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0491922165128853e-05,
+      "loss": 0.703,
+      "step": 609
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0465351537944429e-05,
+      "loss": 0.7217,
+      "step": 610
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.043877761784252e-05,
+      "loss": 0.7298,
+      "step": 611
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0412200592865331e-05,
+      "loss": 0.7088,
+      "step": 612
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0385620651077024e-05,
+      "loss": 0.7955,
+      "step": 613
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0359037980562416e-05,
+      "loss": 0.7193,
+      "step": 614
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0332452769425619e-05,
+      "loss": 0.71,
+      "step": 615
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0305865205788728e-05,
+      "loss": 0.7547,
+      "step": 616
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0279275477790487e-05,
+      "loss": 0.7515,
+      "step": 617
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0252683773584953e-05,
+      "loss": 0.7191,
+      "step": 618
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0226090281340168e-05,
+      "loss": 0.6084,
+      "step": 619
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0199495189236828e-05,
+      "loss": 0.678,
+      "step": 620
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0172898685466947e-05,
+      "loss": 0.7439,
+      "step": 621
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0146300958232528e-05,
+      "loss": 0.6651,
+      "step": 622
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0119702195744236e-05,
+      "loss": 0.6783,
+      "step": 623
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0093102586220056e-05,
+      "loss": 0.6324,
+      "step": 624
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0066502317883969e-05,
+      "loss": 0.7066,
+      "step": 625
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0039901578964619e-05,
+      "loss": 0.6981,
+      "step": 626
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0013300557693981e-05,
+      "loss": 0.7255,
+      "step": 627
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.986699442306025e-06,
+      "loss": 0.7038,
+      "step": 628
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.960098421035383e-06,
+      "loss": 0.7122,
+      "step": 629
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.933497682116035e-06,
+      "loss": 0.6878,
+      "step": 630
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.906897413779949e-06,
+      "loss": 0.7578,
+      "step": 631
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88029780425577e-06,
+      "loss": 0.7797,
+      "step": 632
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.853699041767473e-06,
+      "loss": 0.6868,
+      "step": 633
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.827101314533056e-06,
+      "loss": 0.7074,
+      "step": 634
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.800504810763176e-06,
+      "loss": 0.6887,
+      "step": 635
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.773909718659831e-06,
+      "loss": 0.7079,
+      "step": 636
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.747316226415052e-06,
+      "loss": 0.7155,
+      "step": 637
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.720724522209518e-06,
+      "loss": 0.6791,
+      "step": 638
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.694134794211277e-06,
+      "loss": 0.7382,
+      "step": 639
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.667547230574386e-06,
+      "loss": 0.666,
+      "step": 640
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.64096201943759e-06,
+      "loss": 0.7314,
+      "step": 641
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.61437934892298e-06,
+      "loss": 0.674,
+      "step": 642
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.587799407134672e-06,
+      "loss": 0.6704,
+      "step": 643
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.56122238215748e-06,
+      "loss": 0.7491,
+      "step": 644
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.534648462055576e-06,
+      "loss": 0.6328,
+      "step": 645
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.50807783487115e-06,
+      "loss": 0.6368,
+      "step": 646
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.481510688623098e-06,
+      "loss": 0.7357,
+      "step": 647
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.454947211305691e-06,
+      "loss": 0.7597,
+      "step": 648
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.42838759088723e-06,
+      "loss": 0.738,
+      "step": 649
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.401832015308728e-06,
+      "loss": 0.6966,
+      "step": 650
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.375280672482567e-06,
+      "loss": 0.6359,
+      "step": 651
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.348733750291186e-06,
+      "loss": 0.6664,
+      "step": 652
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.322191436585745e-06,
+      "loss": 0.7903,
+      "step": 653
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.295653919184787e-06,
+      "loss": 0.6504,
+      "step": 654
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.269121385872915e-06,
+      "loss": 0.6986,
+      "step": 655
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.242594024399467e-06,
+      "loss": 0.7183,
+      "step": 656
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.216072022477183e-06,
+      "loss": 0.6719,
+      "step": 657
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.189555567780882e-06,
+      "loss": 0.7101,
+      "step": 658
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.16304484794612e-06,
+      "loss": 0.6959,
+      "step": 659
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.13654005056788e-06,
+      "loss": 0.7,
+      "step": 660
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.110041363199233e-06,
+      "loss": 0.7445,
+      "step": 661
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.083548973350019e-06,
+      "loss": 0.6368,
+      "step": 662
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.057063068485513e-06,
+      "loss": 0.7267,
+      "step": 663
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.030583836025093e-06,
+      "loss": 0.7428,
+      "step": 664
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.004111463340935e-06,
+      "loss": 0.6558,
+      "step": 665
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.977646137756662e-06,
+      "loss": 0.5894,
+      "step": 666
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.951188046546048e-06,
+      "loss": 0.6661,
+      "step": 667
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.924737376931651e-06,
+      "loss": 0.6907,
+      "step": 668
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.898294316083529e-06,
+      "loss": 0.6964,
+      "step": 669
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.871859051117896e-06,
+      "loss": 0.8234,
+      "step": 670
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.8454317690958e-06,
+      "loss": 0.7296,
+      "step": 671
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.819012657021794e-06,
+      "loss": 0.6616,
+      "step": 672
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.79260190184263e-06,
+      "loss": 0.644,
+      "step": 673
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.76619969044592e-06,
+      "loss": 0.7428,
+      "step": 674
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.739806209658812e-06,
+      "loss": 0.7288,
+      "step": 675
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.713421646246692e-06,
+      "loss": 0.6562,
+      "step": 676
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.687046186911819e-06,
+      "loss": 0.6575,
+      "step": 677
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.660680018292053e-06,
+      "loss": 0.6852,
+      "step": 678
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.634323326959501e-06,
+      "loss": 0.7372,
+      "step": 679
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.60797629941921e-06,
+      "loss": 0.7377,
+      "step": 680
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.581639122107837e-06,
+      "loss": 0.6627,
+      "step": 681
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.555311981392342e-06,
+      "loss": 0.5894,
+      "step": 682
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.528995063568673e-06,
+      "loss": 0.6998,
+      "step": 683
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.502688554860426e-06,
+      "loss": 0.7196,
+      "step": 684
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.47639264141754e-06,
+      "loss": 0.6533,
+      "step": 685
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.450107509314983e-06,
+      "loss": 0.6784,
+      "step": 686
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.423833344551443e-06,
+      "loss": 0.7089,
+      "step": 687
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.397570333047985e-06,
+      "loss": 0.7154,
+      "step": 688
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.37131866064676e-06,
+      "loss": 0.736,
+      "step": 689
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.345078513109677e-06,
+      "loss": 0.7355,
+      "step": 690
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.31885007611709e-06,
+      "loss": 0.7147,
+      "step": 691
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.2926335352665e-06,
+      "loss": 0.7275,
+      "step": 692
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.266429076071221e-06,
+      "loss": 0.6845,
+      "step": 693
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.240236883959067e-06,
+      "loss": 0.7072,
+      "step": 694
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.214057144271058e-06,
+      "loss": 0.6696,
+      "step": 695
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.187890042260094e-06,
+      "loss": 0.758,
+      "step": 696
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.161735763089654e-06,
+      "loss": 0.7281,
+      "step": 697
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.13559449183247e-06,
+      "loss": 0.6864,
+      "step": 698
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.109466413469238e-06,
+      "loss": 0.6946,
+      "step": 699
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.083351712887288e-06,
+      "loss": 0.6725,
+      "step": 700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.057250574879296e-06,
+      "loss": 0.7464,
+      "step": 701
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.031163184141965e-06,
+      "loss": 0.7287,
+      "step": 702
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.005089725274711e-06,
+      "loss": 0.7348,
+      "step": 703
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.979030382778376e-06,
+      "loss": 0.6235,
+      "step": 704
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.952985341053902e-06,
+      "loss": 0.7276,
+      "step": 705
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92695478440105e-06,
+      "loss": 0.7636,
+      "step": 706
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.900938897017064e-06,
+      "loss": 0.6621,
+      "step": 707
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.874937862995401e-06,
+      "loss": 0.6502,
+      "step": 708
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.848951866324402e-06,
+      "loss": 0.7533,
+      "step": 709
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.822981090886011e-06,
+      "loss": 0.6577,
+      "step": 710
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.79702572045445e-06,
+      "loss": 0.7331,
+      "step": 711
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.771085938694943e-06,
+      "loss": 0.7218,
+      "step": 712
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.745161929162405e-06,
+      "loss": 0.7452,
+      "step": 713
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.719253875300138e-06,
+      "loss": 0.7169,
+      "step": 714
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.693361960438548e-06,
+      "loss": 0.7362,
+      "step": 715
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.667486367793822e-06,
+      "loss": 0.7166,
+      "step": 716
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.641627280466663e-06,
+      "loss": 0.7075,
+      "step": 717
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.615784881440975e-06,
+      "loss": 0.7093,
+      "step": 718
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.589959353582574e-06,
+      "loss": 0.7078,
+      "step": 719
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.564150879637882e-06,
+      "loss": 0.7476,
+      "step": 720
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.538359642232654e-06,
+      "loss": 0.6714,
+      "step": 721
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5125858238706785e-06,
+      "loss": 0.7372,
+      "step": 722
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.486829606932478e-06,
+      "loss": 0.6916,
+      "step": 723
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.461091173674022e-06,
+      "loss": 0.7234,
+      "step": 724
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.43537070622544e-06,
+      "loss": 0.7135,
+      "step": 725
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.40966838658974e-06,
+      "loss": 0.7093,
+      "step": 726
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.383984396641506e-06,
+      "loss": 0.6133,
+      "step": 727
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.358318918125613e-06,
+      "loss": 0.6692,
+      "step": 728
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.332672132655953e-06,
+      "loss": 0.6701,
+      "step": 729
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.307044221714139e-06,
+      "loss": 0.7934,
+      "step": 730
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2814353666482276e-06,
+      "loss": 0.7,
+      "step": 731
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2558457486714316e-06,
+      "loss": 0.6637,
+      "step": 732
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.230275548860833e-06,
+      "loss": 0.7198,
+      "step": 733
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2047249481561125e-06,
+      "loss": 0.6981,
+      "step": 734
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.179194127358258e-06,
+      "loss": 0.789,
+      "step": 735
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.153683267128304e-06,
+      "loss": 0.6912,
+      "step": 736
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.128192547986023e-06,
+      "loss": 0.748,
+      "step": 737
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.102722150308678e-06,
+      "loss": 0.7162,
+      "step": 738
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.077272254329726e-06,
+      "loss": 0.7006,
+      "step": 739
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.051843040137558e-06,
+      "loss": 0.7111,
+      "step": 740
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.026434687674204e-06,
+      "loss": 0.7279,
+      "step": 741
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.001047376734087e-06,
+      "loss": 0.6927,
+      "step": 742
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.975681286962724e-06,
+      "loss": 0.6397,
+      "step": 743
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9503365978554735e-06,
+      "loss": 0.6876,
+      "step": 744
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.925013488756264e-06,
+      "loss": 0.6496,
+      "step": 745
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8997121388563e-06,
+      "loss": 0.6827,
+      "step": 746
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.874432727192837e-06,
+      "loss": 0.6476,
+      "step": 747
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.849175432647875e-06,
+      "loss": 0.7119,
+      "step": 748
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.823940433946921e-06,
+      "loss": 0.7134,
+      "step": 749
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.798727909657698e-06,
+      "loss": 0.6754,
+      "step": 750
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.773538038188912e-06,
+      "loss": 0.7139,
+      "step": 751
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.74837099778896e-06,
+      "loss": 0.688,
+      "step": 752
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.723226966544691e-06,
+      "loss": 0.6855,
+      "step": 753
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.69810612238013e-06,
+      "loss": 0.693,
+      "step": 754
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.673008643055228e-06,
+      "loss": 0.6963,
+      "step": 755
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6479347061646046e-06,
+      "loss": 0.8004,
+      "step": 756
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.622884489136286e-06,
+      "loss": 0.7051,
+      "step": 757
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.597858169230454e-06,
+      "loss": 0.6708,
+      "step": 758
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.572855923538186e-06,
+      "loss": 0.7381,
+      "step": 759
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.547877928980206e-06,
+      "loss": 0.659,
+      "step": 760
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.522924362305639e-06,
+      "loss": 0.7255,
+      "step": 761
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.497995400090748e-06,
+      "loss": 0.6643,
+      "step": 762
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4730912187376895e-06,
+      "loss": 0.6451,
+      "step": 763
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.448211994473263e-06,
+      "loss": 0.6904,
+      "step": 764
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.42335790334768e-06,
+      "loss": 0.7238,
+      "step": 765
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.398529121233291e-06,
+      "loss": 0.6466,
+      "step": 766
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.373725823823359e-06,
+      "loss": 0.6502,
+      "step": 767
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.348948186630815e-06,
+      "loss": 0.7152,
+      "step": 768
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.324196384987009e-06,
+      "loss": 0.693,
+      "step": 769
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2994705940404825e-06,
+      "loss": 0.7111,
+      "step": 770
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.274770988755712e-06,
+      "loss": 0.7255,
+      "step": 771
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.250097743911877e-06,
+      "loss": 0.6682,
+      "step": 772
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.225451034101631e-06,
+      "loss": 0.6428,
+      "step": 773
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.200831033729864e-06,
+      "loss": 0.7113,
+      "step": 774
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.176237917012459e-06,
+      "loss": 0.6725,
+      "step": 775
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.151671857975061e-06,
+      "loss": 0.7684,
+      "step": 776
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.12713303045186e-06,
+      "loss": 0.6573,
+      "step": 777
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.10262160808434e-06,
+      "loss": 0.7252,
+      "step": 778
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0781377643200765e-06,
+      "loss": 0.6943,
+      "step": 779
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.053681672411471e-06,
+      "loss": 0.7177,
+      "step": 780
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.029253505414565e-06,
+      "loss": 0.7555,
+      "step": 781
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.004853436187794e-06,
+      "loss": 0.7183,
+      "step": 782
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9804816373907625e-06,
+      "loss": 0.7157,
+      "step": 783
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.956138281483039e-06,
+      "loss": 0.6987,
+      "step": 784
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.931823540722912e-06,
+      "loss": 0.7014,
+      "step": 785
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.907537587166191e-06,
+      "loss": 0.722,
+      "step": 786
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.883280592664979e-06,
+      "loss": 0.6999,
+      "step": 787
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.859052728866468e-06,
+      "loss": 0.6672,
+      "step": 788
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.834854167211699e-06,
+      "loss": 0.7012,
+      "step": 789
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.810685078934375e-06,
+      "loss": 0.7097,
+      "step": 790
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.78654563505964e-06,
+      "loss": 0.7265,
+      "step": 791
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.762436006402874e-06,
+      "loss": 0.7105,
+      "step": 792
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.738356363568463e-06,
+      "loss": 0.7299,
+      "step": 793
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.714306876948621e-06,
+      "loss": 0.6605,
+      "step": 794
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.69028771672216e-06,
+      "loss": 0.6745,
+      "step": 795
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.666299052853314e-06,
+      "loss": 0.7132,
+      "step": 796
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.642341055090508e-06,
+      "loss": 0.7198,
+      "step": 797
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.618413892965158e-06,
+      "loss": 0.7524,
+      "step": 798
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5945177357904935e-06,
+      "loss": 0.6779,
+      "step": 799
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.570652752660343e-06,
+      "loss": 0.6845,
+      "step": 800
+    }
+  ],
+  "max_steps": 1218,
+  "num_train_epochs": 3,
+  "total_flos": 1.9341035834441728e+17,
+  "trial_name": null,
+  "trial_params": null
+}