diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,11102 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.40017342293518315,
+  "global_step": 1846,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99783672585535e-06,
+      "loss": 5.0039,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.995673451710698e-06,
+      "loss": 5.418,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.993510177566045e-06,
+      "loss": 5.2227,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.991346903421395e-06,
+      "loss": 5.2734,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.989183629276742e-06,
+      "loss": 4.6602,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.987020355132093e-06,
+      "loss": 5.4844,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.984857080987443e-06,
+      "loss": 5.5938,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.98269380684279e-06,
+      "loss": 5.2266,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.980530532698139e-06,
+      "loss": 4.5703,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.978367258553486e-06,
+      "loss": 5.4844,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.976203984408836e-06,
+      "loss": 5.082,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.974040710264185e-06,
+      "loss": 5.0938,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.971877436119532e-06,
+      "loss": 4.8438,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.969714161974882e-06,
+      "loss": 5.4297,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.96755088783023e-06,
+      "loss": 5.5,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.965387613685578e-06,
+      "loss": 4.7305,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.963224339540926e-06,
+      "loss": 5.0391,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.961061065396277e-06,
+      "loss": 5.5859,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.958897791251625e-06,
+      "loss": 5.4688,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.956734517106974e-06,
+      "loss": 4.4297,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.954571242962323e-06,
+      "loss": 5.6602,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.952407968817671e-06,
+      "loss": 5.3359,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.95024469467302e-06,
+      "loss": 5.2422,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.948081420528369e-06,
+      "loss": 5.1602,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.945918146383716e-06,
+      "loss": 5.3984,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.943754872239066e-06,
+      "loss": 4.9766,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.941591598094413e-06,
+      "loss": 4.9141,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.939428323949763e-06,
+      "loss": 4.543,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.93726504980511e-06,
+      "loss": 5.6094,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.935101775660459e-06,
+      "loss": 5.4531,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.93293850151581e-06,
+      "loss": 4.6562,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.930775227371158e-06,
+      "loss": 5.5781,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.928611953226507e-06,
+      "loss": 4.8359,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.926448679081855e-06,
+      "loss": 5.6445,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.924285404937204e-06,
+      "loss": 5.2656,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.92212213079255e-06,
+      "loss": 5.0742,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9199588566479e-06,
+      "loss": 5.4961,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.91779558250325e-06,
+      "loss": 4.6523,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.915632308358597e-06,
+      "loss": 4.957,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.913469034213945e-06,
+      "loss": 5.5039,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.911305760069296e-06,
+      "loss": 5.5938,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.909142485924643e-06,
+      "loss": 5.1797,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.906979211779993e-06,
+      "loss": 5.2539,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.90481593763534e-06,
+      "loss": 5.125,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.902652663490689e-06,
+      "loss": 5.0195,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.900489389346039e-06,
+      "loss": 5.4141,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.898326115201386e-06,
+      "loss": 4.625,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.896162841056735e-06,
+      "loss": 5.1094,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.893999566912083e-06,
+      "loss": 4.8086,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.891836292767432e-06,
+      "loss": 4.9297,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.88967301862278e-06,
+      "loss": 5.3945,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.88750974447813e-06,
+      "loss": 5.207,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.88534647033348e-06,
+      "loss": 4.8047,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.883183196188828e-06,
+      "loss": 5.3477,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.881019922044177e-06,
+      "loss": 4.9648,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.878856647899524e-06,
+      "loss": 5.3945,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.876693373754873e-06,
+      "loss": 4.9609,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.874530099610221e-06,
+      "loss": 5.6484,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.87236682546557e-06,
+      "loss": 5.4609,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.870203551320919e-06,
+      "loss": 5.2539,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.868040277176266e-06,
+      "loss": 4.5781,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.865877003031616e-06,
+      "loss": 5.1562,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.863713728886965e-06,
+      "loss": 4.7734,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.861550454742313e-06,
+      "loss": 4.5039,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.859387180597662e-06,
+      "loss": 5.2461,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.857223906453012e-06,
+      "loss": 5.1367,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.85506063230836e-06,
+      "loss": 5.2344,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.852897358163708e-06,
+      "loss": 4.5,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.850734084019056e-06,
+      "loss": 5.3086,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.848570809874405e-06,
+      "loss": 4.5469,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.846407535729754e-06,
+      "loss": 5.1797,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.844244261585102e-06,
+      "loss": 5.4336,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.84208098744045e-06,
+      "loss": 4.9883,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.8399177132958e-06,
+      "loss": 5.2617,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.837754439151148e-06,
+      "loss": 5.0703,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.835591165006497e-06,
+      "loss": 5.0742,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.833427890861846e-06,
+      "loss": 4.2539,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.831264616717194e-06,
+      "loss": 5.4297,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.829101342572543e-06,
+      "loss": 5.1406,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.826938068427892e-06,
+      "loss": 4.582,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.824774794283239e-06,
+      "loss": 5.4414,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.822611520138589e-06,
+      "loss": 5.0859,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.820448245993938e-06,
+      "loss": 5.8906,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.818284971849286e-06,
+      "loss": 4.0859,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.816121697704635e-06,
+      "loss": 5.5586,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.813958423559984e-06,
+      "loss": 5.5039,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.811795149415332e-06,
+      "loss": 4.8281,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.809631875270681e-06,
+      "loss": 5.2188,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.80746860112603e-06,
+      "loss": 5.1914,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.805305326981378e-06,
+      "loss": 4.9805,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.803142052836727e-06,
+      "loss": 5.1211,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.800978778692076e-06,
+      "loss": 4.9648,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.798815504547423e-06,
+      "loss": 5.0391,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.796652230402773e-06,
+      "loss": 4.543,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.79448895625812e-06,
+      "loss": 4.4766,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.792325682113469e-06,
+      "loss": 3.9922,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.790162407968819e-06,
+      "loss": 4.6719,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.787999133824168e-06,
+      "loss": 4.6914,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.785835859679516e-06,
+      "loss": 5.1484,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.783672585534867e-06,
+      "loss": 5.1523,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.781509311390212e-06,
+      "loss": 5.4727,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.779346037245562e-06,
+      "loss": 5.2539,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.777182763100911e-06,
+      "loss": 5.2852,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.77501948895626e-06,
+      "loss": 5.2969,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.772856214811608e-06,
+      "loss": 5.2852,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.770692940666955e-06,
+      "loss": 5.3398,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.768529666522304e-06,
+      "loss": 4.6367,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.766366392377652e-06,
+      "loss": 5.2539,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.764203118233003e-06,
+      "loss": 5.3594,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.762039844088351e-06,
+      "loss": 5.4883,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.7598765699437e-06,
+      "loss": 5.3594,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.757713295799047e-06,
+      "loss": 5.3789,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.755550021654396e-06,
+      "loss": 4.8281,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.753386747509746e-06,
+      "loss": 4.7305,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.751223473365093e-06,
+      "loss": 4.8516,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.749060199220442e-06,
+      "loss": 5.1328,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.746896925075792e-06,
+      "loss": 5.0039,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.744733650931139e-06,
+      "loss": 5.3477,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.742570376786488e-06,
+      "loss": 4.6758,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.740407102641836e-06,
+      "loss": 4.5234,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.738243828497187e-06,
+      "loss": 5.1172,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.736080554352535e-06,
+      "loss": 4.9609,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.733917280207882e-06,
+      "loss": 4.1367,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.731754006063233e-06,
+      "loss": 4.6328,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.729590731918581e-06,
+      "loss": 5.3164,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.727427457773928e-06,
+      "loss": 4.6406,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.725264183629277e-06,
+      "loss": 5.4844,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.723100909484626e-06,
+      "loss": 4.3984,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.720937635339974e-06,
+      "loss": 5.1484,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.718774361195323e-06,
+      "loss": 5.4648,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.716611087050672e-06,
+      "loss": 5.3359,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.71444781290602e-06,
+      "loss": 4.9219,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.712284538761369e-06,
+      "loss": 4.2539,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.71012126461672e-06,
+      "loss": 4.6484,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.707957990472066e-06,
+      "loss": 5.4648,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.705794716327417e-06,
+      "loss": 4.9961,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.703631442182765e-06,
+      "loss": 4.2422,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.701468168038112e-06,
+      "loss": 4.0039,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.69930489389346e-06,
+      "loss": 5.8984,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.69714161974881e-06,
+      "loss": 5.3984,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.694978345604158e-06,
+      "loss": 4.043,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.692815071459507e-06,
+      "loss": 4.9297,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.690651797314855e-06,
+      "loss": 4.75,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.688488523170206e-06,
+      "loss": 5.0312,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.686325249025553e-06,
+      "loss": 5.3047,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.684161974880901e-06,
+      "loss": 5.4453,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.68199870073625e-06,
+      "loss": 4.5195,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.679835426591599e-06,
+      "loss": 4.9023,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.677672152446947e-06,
+      "loss": 5.4531,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.675508878302296e-06,
+      "loss": 4.9141,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.673345604157645e-06,
+      "loss": 5.2031,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.671182330012993e-06,
+      "loss": 4.8125,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.669019055868342e-06,
+      "loss": 5.3242,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.66685578172369e-06,
+      "loss": 5.7188,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.66469250757904e-06,
+      "loss": 5.0781,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.66252923343439e-06,
+      "loss": 5.3438,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.660365959289737e-06,
+      "loss": 4.5586,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.658202685145085e-06,
+      "loss": 5.6875,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.656039411000434e-06,
+      "loss": 5.168,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.653876136855783e-06,
+      "loss": 5.3672,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.651712862711131e-06,
+      "loss": 5.2539,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.64954958856648e-06,
+      "loss": 5.1367,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.647386314421827e-06,
+      "loss": 4.5117,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.645223040277176e-06,
+      "loss": 5.3047,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.643059766132526e-06,
+      "loss": 5.5547,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.640896491987875e-06,
+      "loss": 5.3945,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.638733217843223e-06,
+      "loss": 5.3125,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.636569943698572e-06,
+      "loss": 4.2383,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.63440666955392e-06,
+      "loss": 5.1133,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.63224339540927e-06,
+      "loss": 4.6445,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.630080121264616e-06,
+      "loss": 5.5586,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.627916847119967e-06,
+      "loss": 5.0938,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.625753572975315e-06,
+      "loss": 5.3359,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.623590298830662e-06,
+      "loss": 4.9688,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.621427024686013e-06,
+      "loss": 4.3984,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.61926375054136e-06,
+      "loss": 5.5117,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.61710047639671e-06,
+      "loss": 5.3125,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.614937202252058e-06,
+      "loss": 5.4922,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.612773928107407e-06,
+      "loss": 5.2539,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.610610653962756e-06,
+      "loss": 4.8828,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.608447379818104e-06,
+      "loss": 5.0977,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.606284105673453e-06,
+      "loss": 5.1836,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.6041208315288e-06,
+      "loss": 5.3359,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.601957557384149e-06,
+      "loss": 5.1289,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.599794283239499e-06,
+      "loss": 4.4023,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.597631009094846e-06,
+      "loss": 5.3906,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.595467734950196e-06,
+      "loss": 5.4609,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.593304460805545e-06,
+      "loss": 5.4844,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.591141186660894e-06,
+      "loss": 5.293,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.588977912516242e-06,
+      "loss": 5.4492,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.58681463837159e-06,
+      "loss": 4.7344,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.58465136422694e-06,
+      "loss": 4.0977,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.582488090082288e-06,
+      "loss": 4.9531,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.580324815937635e-06,
+      "loss": 5.2656,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.578161541792984e-06,
+      "loss": 5.4102,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.575998267648333e-06,
+      "loss": 4.7539,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.573834993503681e-06,
+      "loss": 5.3789,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.57167171935903e-06,
+      "loss": 5.2109,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.569508445214379e-06,
+      "loss": 5.2734,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.567345171069729e-06,
+      "loss": 5.3984,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.565181896925078e-06,
+      "loss": 5.3281,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.563018622780425e-06,
+      "loss": 4.2578,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.560855348635773e-06,
+      "loss": 5.0703,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.558692074491122e-06,
+      "loss": 5.0703,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.55652880034647e-06,
+      "loss": 5.0898,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.55436552620182e-06,
+      "loss": 5.1172,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.55220225205717e-06,
+      "loss": 4.668,
+      "step": 207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.550038977912517e-06,
+      "loss": 5.4414,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.547875703767865e-06,
+      "loss": 5.3945,
+      "step": 209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.545712429623214e-06,
+      "loss": 5.418,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.543549155478562e-06,
+      "loss": 5.2773,
+      "step": 211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.541385881333913e-06,
+      "loss": 5.3242,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.539222607189261e-06,
+      "loss": 4.1094,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.537059333044608e-06,
+      "loss": 5.4961,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.534896058899957e-06,
+      "loss": 4.5938,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.532732784755306e-06,
+      "loss": 5.3047,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.530569510610654e-06,
+      "loss": 5.0703,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.528406236466003e-06,
+      "loss": 5.1953,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.526242962321353e-06,
+      "loss": 5.5352,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.5240796881767e-06,
+      "loss": 5.25,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.521916414032049e-06,
+      "loss": 5.3711,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.519753139887398e-06,
+      "loss": 4.8281,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.517589865742746e-06,
+      "loss": 5.375,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.515426591598097e-06,
+      "loss": 4.9102,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.513263317453444e-06,
+      "loss": 5.3906,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.511100043308792e-06,
+      "loss": 5.6406,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.508936769164143e-06,
+      "loss": 5.3281,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.50677349501949e-06,
+      "loss": 5.5312,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.504610220874838e-06,
+      "loss": 5.4414,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.502446946730187e-06,
+      "loss": 5.3984,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.500283672585536e-06,
+      "loss": 4.6875,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.498120398440884e-06,
+      "loss": 5.4922,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.495957124296233e-06,
+      "loss": 5.0273,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.493793850151582e-06,
+      "loss": 5.5547,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.49163057600693e-06,
+      "loss": 5.3398,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.489467301862279e-06,
+      "loss": 5.2656,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.487304027717628e-06,
+      "loss": 4.3828,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.485140753572976e-06,
+      "loss": 5.0898,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.482977479428325e-06,
+      "loss": 5.6172,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.480814205283672e-06,
+      "loss": 5.1094,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.478650931139022e-06,
+      "loss": 5.0859,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.47648765699437e-06,
+      "loss": 5.4609,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.47432438284972e-06,
+      "loss": 5.4805,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.472161108705068e-06,
+      "loss": 4.4102,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.469997834560417e-06,
+      "loss": 5.0352,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.467834560415766e-06,
+      "loss": 5.4688,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.465671286271116e-06,
+      "loss": 5.5078,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.463508012126463e-06,
+      "loss": 4.8438,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.461344737981811e-06,
+      "loss": 5.0234,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.45918146383716e-06,
+      "loss": 5.375,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.457018189692509e-06,
+      "loss": 5.4141,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.454854915547857e-06,
+      "loss": 4.6875,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.452691641403206e-06,
+      "loss": 5.0703,
+      "step": 253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.450528367258553e-06,
+      "loss": 4.7305,
+      "step": 254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.448365093113903e-06,
+      "loss": 5.2188,
+      "step": 255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.446201818969252e-06,
+      "loss": 5.1406,
+      "step": 256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.4440385448246e-06,
+      "loss": 5.043,
+      "step": 257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.44187527067995e-06,
+      "loss": 5.0352,
+      "step": 258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.439711996535298e-06,
+      "loss": 5.0664,
+      "step": 259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.437548722390647e-06,
+      "loss": 5.1133,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.435385448245995e-06,
+      "loss": 3.5605,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.433222174101342e-06,
+      "loss": 5.1562,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.431058899956693e-06,
+      "loss": 4.5938,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.428895625812041e-06,
+      "loss": 5.6016,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.426732351667388e-06,
+      "loss": 5.4688,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.424569077522737e-06,
+      "loss": 5.2031,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.422405803378086e-06,
+      "loss": 5.5352,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.420242529233436e-06,
+      "loss": 5.2656,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.418079255088785e-06,
+      "loss": 4.5703,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.415915980944133e-06,
+      "loss": 5.6172,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.413752706799482e-06,
+      "loss": 5.1562,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.41158943265483e-06,
+      "loss": 4.7109,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.409426158510178e-06,
+      "loss": 5.2578,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.407262884365526e-06,
+      "loss": 5.0234,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.405099610220877e-06,
+      "loss": 5.2695,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.402936336076224e-06,
+      "loss": 4.7031,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.400773061931572e-06,
+      "loss": 5.1328,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.398609787786923e-06,
+      "loss": 4.6719,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.39644651364227e-06,
+      "loss": 5.3555,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.39428323949762e-06,
+      "loss": 5.3281,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.392119965352969e-06,
+      "loss": 4.8867,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.389956691208315e-06,
+      "loss": 4.8828,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.387793417063666e-06,
+      "loss": 3.8965,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.385630142919014e-06,
+      "loss": 4.9609,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.383466868774361e-06,
+      "loss": 5.2695,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.38130359462971e-06,
+      "loss": 4.9375,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.379140320485059e-06,
+      "loss": 4.9648,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.376977046340407e-06,
+      "loss": 4.5156,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.374813772195756e-06,
+      "loss": 4.875,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.372650498051106e-06,
+      "loss": 4.8906,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.370487223906455e-06,
+      "loss": 5.2734,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.368323949761804e-06,
+      "loss": 5.0,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.36616067561715e-06,
+      "loss": 5.0703,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.3639974014725e-06,
+      "loss": 4.5781,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.36183412732785e-06,
+      "loss": 5.0117,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.359670853183197e-06,
+      "loss": 5.1367,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.357507579038545e-06,
+      "loss": 4.5117,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.355344304893892e-06,
+      "loss": 5.4141,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.353181030749243e-06,
+      "loss": 5.1953,
+      "step": 299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.351017756604591e-06,
+      "loss": 4.6953,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.34885448245994e-06,
+      "loss": 4.8086,
+      "step": 301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.346691208315289e-06,
+      "loss": 4.3711,
+      "step": 302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.344527934170639e-06,
+      "loss": 4.8984,
+      "step": 303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.342364660025986e-06,
+      "loss": 4.1758,
+      "step": 304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.340201385881335e-06,
+      "loss": 4.3516,
+      "step": 305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.338038111736683e-06,
+      "loss": 5.1641,
+      "step": 306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.335874837592032e-06,
+      "loss": 5.4023,
+      "step": 307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.33371156344738e-06,
+      "loss": 5.1289,
+      "step": 308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.33154828930273e-06,
+      "loss": 5.2344,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.329385015158076e-06,
+      "loss": 5.1562,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.327221741013427e-06,
+      "loss": 4.8438,
+      "step": 311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.325058466868775e-06,
+      "loss": 5.1484,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.322895192724122e-06,
+      "loss": 4.7969,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.320731918579473e-06,
+      "loss": 5.4492,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.318568644434823e-06,
+      "loss": 5.0156,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.31640537029017e-06,
+      "loss": 5.4062,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.314242096145518e-06,
+      "loss": 5.2344,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.312078822000867e-06,
+      "loss": 5.1875,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.309915547856216e-06,
+      "loss": 4.8281,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.307752273711564e-06,
+      "loss": 5.5664,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.305588999566913e-06,
+      "loss": 4.9883,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.303425725422262e-06,
+      "loss": 5.3164,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.30126245127761e-06,
+      "loss": 4.4297,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.299099177132957e-06,
+      "loss": 5.207,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.296935902988308e-06,
+      "loss": 4.5156,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.294772628843656e-06,
+      "loss": 4.5781,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.292609354699005e-06,
+      "loss": 4.5547,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.290446080554354e-06,
+      "loss": 5.0234,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.288282806409702e-06,
+      "loss": 4.8555,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.28611953226505e-06,
+      "loss": 4.2188,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.2839562581204e-06,
+      "loss": 5.5,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.281792983975747e-06,
+      "loss": 4.7891,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.279629709831095e-06,
+      "loss": 4.8594,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.277466435686446e-06,
+      "loss": 5.6758,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.275303161541793e-06,
+      "loss": 5.6953,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.273139887397141e-06,
+      "loss": 5.4258,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.270976613252492e-06,
+      "loss": 4.8711,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.268813339107839e-06,
+      "loss": 5.4531,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.266650064963189e-06,
+      "loss": 5.3672,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.264486790818538e-06,
+      "loss": 4.5117,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.262323516673886e-06,
+      "loss": 4.8633,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.260160242529235e-06,
+      "loss": 5.0195,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.257996968384582e-06,
+      "loss": 5.3047,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.25583369423993e-06,
+      "loss": 5.0391,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.25367042009528e-06,
+      "loss": 5.4375,
+      "step": 345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.25150714595063e-06,
+      "loss": 5.3281,
+      "step": 346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.249343871805977e-06,
+      "loss": 5.457,
+      "step": 347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.247180597661325e-06,
+      "loss": 5.2305,
+      "step": 348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.245017323516676e-06,
+      "loss": 5.043,
+      "step": 349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.242854049372023e-06,
+      "loss": 5.125,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.240690775227373e-06,
+      "loss": 4.4883,
+      "step": 351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.238527501082722e-06,
+      "loss": 4.0625,
+      "step": 352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.23636422693807e-06,
+      "loss": 4.7227,
+      "step": 353
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.234200952793419e-06,
+      "loss": 5.7852,
+      "step": 354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.232037678648766e-06,
+      "loss": 4.8828,
+      "step": 355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.229874404504114e-06,
+      "loss": 4.9297,
+      "step": 356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.227711130359463e-06,
+      "loss": 4.9453,
+      "step": 357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.225547856214812e-06,
+      "loss": 5.4219,
+      "step": 358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.22338458207016e-06,
+      "loss": 5.2188,
+      "step": 359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.22122130792551e-06,
+      "loss": 4.9961,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.21905803378086e-06,
+      "loss": 5.3711,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.216894759636206e-06,
+      "loss": 5.4531,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.214731485491555e-06,
+      "loss": 5.1016,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.212568211346904e-06,
+      "loss": 4.5938,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.210404937202252e-06,
+      "loss": 4.3672,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.208241663057601e-06,
+      "loss": 5.3047,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.20607838891295e-06,
+      "loss": 5.2109,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.203915114768298e-06,
+      "loss": 4.8359,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.201751840623645e-06,
+      "loss": 5.1953,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.199588566478996e-06,
+      "loss": 4.7734,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.197425292334344e-06,
+      "loss": 5.2422,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.195262018189693e-06,
+      "loss": 4.793,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.193098744045043e-06,
+      "loss": 4.8281,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.190935469900392e-06,
+      "loss": 4.9688,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.188772195755739e-06,
+      "loss": 4.9141,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.186608921611088e-06,
+      "loss": 5.4844,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.184445647466436e-06,
+      "loss": 5.4258,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.182282373321785e-06,
+      "loss": 4.0469,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.180119099177134e-06,
+      "loss": 4.4609,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.177955825032482e-06,
+      "loss": 5.4297,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.17579255088783e-06,
+      "loss": 4.3906,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.17362927674318e-06,
+      "loss": 5.0938,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.171466002598528e-06,
+      "loss": 5.0312,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.169302728453877e-06,
+      "loss": 5.3281,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.167139454309226e-06,
+      "loss": 5.293,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.164976180164574e-06,
+      "loss": 5.4414,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.162812906019923e-06,
+      "loss": 5.043,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.160649631875271e-06,
+      "loss": 5.5,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.15848635773062e-06,
+      "loss": 5.0,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.156323083585969e-06,
+      "loss": 5.1719,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.154159809441317e-06,
+      "loss": 4.8633,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.151996535296666e-06,
+      "loss": 4.8008,
+      "step": 392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.149833261152013e-06,
+      "loss": 5.4297,
+      "step": 393
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.147669987007363e-06,
+      "loss": 5.125,
+      "step": 394
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.145506712862712e-06,
+      "loss": 5.0312,
+      "step": 395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.14334343871806e-06,
+      "loss": 5.1484,
+      "step": 396
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.14118016457341e-06,
+      "loss": 5.1367,
+      "step": 397
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.139016890428758e-06,
+      "loss": 5.2344,
+      "step": 398
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.136853616284107e-06,
+      "loss": 5.0625,
+      "step": 399
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.134690342139454e-06,
+      "loss": 5.2578,
+      "step": 400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.132527067994802e-06,
+      "loss": 4.9688,
+      "step": 401
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.130363793850153e-06,
+      "loss": 4.457,
+      "step": 402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.1282005197055e-06,
+      "loss": 5.0234,
+      "step": 403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.12603724556085e-06,
+      "loss": 4.3789,
+      "step": 404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.123873971416199e-06,
+      "loss": 5.0859,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.121710697271547e-06,
+      "loss": 5.1523,
+      "step": 406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.119547423126896e-06,
+      "loss": 4.7031,
+      "step": 407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.117384148982245e-06,
+      "loss": 4.9922,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.115220874837593e-06,
+      "loss": 5.1484,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.113057600692942e-06,
+      "loss": 4.6406,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.11089432654829e-06,
+      "loss": 4.8125,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.10873105240364e-06,
+      "loss": 5.1562,
+      "step": 412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.106567778258986e-06,
+      "loss": 5.2422,
+      "step": 413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.104404504114337e-06,
+      "loss": 4.2891,
+      "step": 414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.102241229969684e-06,
+      "loss": 5.2969,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.100077955825032e-06,
+      "loss": 5.875,
+      "step": 416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.097914681680383e-06,
+      "loss": 4.9297,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.095751407535731e-06,
+      "loss": 4.8125,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.09358813339108e-06,
+      "loss": 5.2227,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.091424859246427e-06,
+      "loss": 4.0859,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.089261585101775e-06,
+      "loss": 5.1094,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.087098310957126e-06,
+      "loss": 5.6484,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.084935036812473e-06,
+      "loss": 3.5898,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.082771762667823e-06,
+      "loss": 4.5625,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.080608488523172e-06,
+      "loss": 4.7188,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.078445214378519e-06,
+      "loss": 5.3164,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.076281940233867e-06,
+      "loss": 4.0508,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.074118666089216e-06,
+      "loss": 5.3164,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.071955391944566e-06,
+      "loss": 5.1836,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.069792117799915e-06,
+      "loss": 4.8164,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.067628843655262e-06,
+      "loss": 5.2891,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.06546556951061e-06,
+      "loss": 4.9492,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.06330229536596e-06,
+      "loss": 4.5781,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.061139021221308e-06,
+      "loss": 5.3203,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.058975747076657e-06,
+      "loss": 4.7188,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.056812472932005e-06,
+      "loss": 5.2148,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.054649198787354e-06,
+      "loss": 4.0117,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.052485924642703e-06,
+      "loss": 5.2617,
+      "step": 438
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.050322650498051e-06,
+      "loss": 5.7031,
+      "step": 439
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.0481593763534e-06,
+      "loss": 5.0078,
+      "step": 440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.045996102208749e-06,
+      "loss": 5.293,
+      "step": 441
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.043832828064099e-06,
+      "loss": 5.1484,
+      "step": 442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.041669553919446e-06,
+      "loss": 4.9492,
+      "step": 443
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.039506279774796e-06,
+      "loss": 5.2461,
+      "step": 444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.037343005630145e-06,
+      "loss": 5.1758,
+      "step": 445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.035179731485492e-06,
+      "loss": 5.0078,
+      "step": 446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.03301645734084e-06,
+      "loss": 4.9961,
+      "step": 447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.03085318319619e-06,
+      "loss": 5.3438,
+      "step": 448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.028689909051538e-06,
+      "loss": 4.2305,
+      "step": 449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.026526634906887e-06,
+      "loss": 5.2695,
+      "step": 450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.024363360762235e-06,
+      "loss": 5.3477,
+      "step": 451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.022200086617584e-06,
+      "loss": 5.1562,
+      "step": 452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.020036812472933e-06,
+      "loss": 5.2578,
+      "step": 453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.017873538328281e-06,
+      "loss": 5.2852,
+      "step": 454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.01571026418363e-06,
+      "loss": 5.1758,
+      "step": 455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.01354699003898e-06,
+      "loss": 4.1016,
+      "step": 456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.011383715894325e-06,
+      "loss": 4.9258,
+      "step": 457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.009220441749676e-06,
+      "loss": 4.7773,
+      "step": 458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.007057167605024e-06,
+      "loss": 5.1133,
+      "step": 459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.004893893460373e-06,
+      "loss": 4.9453,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.002730619315722e-06,
+      "loss": 5.168,
+      "step": 461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.00056734517107e-06,
+      "loss": 4.875,
+      "step": 462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.998404071026419e-06,
+      "loss": 4.6328,
+      "step": 463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.99624079688177e-06,
+      "loss": 5.1875,
+      "step": 464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.994077522737116e-06,
+      "loss": 5.3281,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.991914248592465e-06,
+      "loss": 4.8438,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.989750974447814e-06,
+      "loss": 4.8633,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.987587700303162e-06,
+      "loss": 5.3477,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.985424426158511e-06,
+      "loss": 5.6211,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.98326115201386e-06,
+      "loss": 4.7969,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.981097877869207e-06,
+      "loss": 4.3203,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.978934603724555e-06,
+      "loss": 5.3281,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.976771329579906e-06,
+      "loss": 4.8672,
+      "step": 473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.974608055435254e-06,
+      "loss": 4.9531,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.972444781290603e-06,
+      "loss": 4.7266,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.970281507145953e-06,
+      "loss": 5.0391,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.968118233001299e-06,
+      "loss": 5.0117,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.965954958856649e-06,
+      "loss": 5.0234,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.963791684711998e-06,
+      "loss": 4.3672,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.961628410567346e-06,
+      "loss": 5.5195,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.959465136422695e-06,
+      "loss": 5.207,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.957301862278044e-06,
+      "loss": 4.2617,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.95513858813339e-06,
+      "loss": 5.0742,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.95297531398874e-06,
+      "loss": 5.0859,
+      "step": 484
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.95081203984409e-06,
+      "loss": 5.1289,
+      "step": 485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.948648765699438e-06,
+      "loss": 4.2109,
+      "step": 486
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.946485491554787e-06,
+      "loss": 5.0352,
+      "step": 487
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.944322217410136e-06,
+      "loss": 5.3945,
+      "step": 488
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.942158943265484e-06,
+      "loss": 5.3711,
+      "step": 489
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.939995669120833e-06,
+      "loss": 4.7305,
+      "step": 490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.93783239497618e-06,
+      "loss": 5.4688,
+      "step": 491
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.93566912083153e-06,
+      "loss": 5.4922,
+      "step": 492
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.933505846686877e-06,
+      "loss": 5.082,
+      "step": 493
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.931342572542226e-06,
+      "loss": 4.7891,
+      "step": 494
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.929179298397576e-06,
+      "loss": 5.0156,
+      "step": 495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.927016024252923e-06,
+      "loss": 4.8633,
+      "step": 496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.924852750108273e-06,
+      "loss": 5.1562,
+      "step": 497
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.922689475963622e-06,
+      "loss": 4.793,
+      "step": 498
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.920526201818969e-06,
+      "loss": 5.0273,
+      "step": 499
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.91836292767432e-06,
+      "loss": 4.2539,
+      "step": 500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.916199653529668e-06,
+      "loss": 5.1367,
+      "step": 501
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.914036379385015e-06,
+      "loss": 4.8359,
+      "step": 502
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.911873105240364e-06,
+      "loss": 4.9609,
+      "step": 503
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.909709831095712e-06,
+      "loss": 5.1172,
+      "step": 504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.907546556951061e-06,
+      "loss": 5.3242,
+      "step": 505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.90538328280641e-06,
+      "loss": 4.75,
+      "step": 506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.90322000866176e-06,
+      "loss": 4.9688,
+      "step": 507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.901056734517109e-06,
+      "loss": 5.5469,
+      "step": 508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.898893460372457e-06,
+      "loss": 5.1875,
+      "step": 509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.896730186227806e-06,
+      "loss": 5.082,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.894566912083153e-06,
+      "loss": 4.1367,
+      "step": 511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.892403637938503e-06,
+      "loss": 5.1875,
+      "step": 512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.890240363793852e-06,
+      "loss": 5.6211,
+      "step": 513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.888077089649199e-06,
+      "loss": 5.3281,
+      "step": 514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.88591381550455e-06,
+      "loss": 5.2539,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.883750541359896e-06,
+      "loss": 4.9414,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.881587267215245e-06,
+      "loss": 5.2891,
+      "step": 517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.879423993070594e-06,
+      "loss": 4.4922,
+      "step": 518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.877260718925942e-06,
+      "loss": 4.9023,
+      "step": 519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.875097444781293e-06,
+      "loss": 5.0547,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.872934170636641e-06,
+      "loss": 5.1992,
+      "step": 521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.870770896491988e-06,
+      "loss": 4.9805,
+      "step": 522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.868607622347337e-06,
+      "loss": 5.3242,
+      "step": 523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.866444348202686e-06,
+      "loss": 4.9648,
+      "step": 524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.864281074058034e-06,
+      "loss": 5.1094,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.862117799913383e-06,
+      "loss": 4.4883,
+      "step": 526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.85995452576873e-06,
+      "loss": 4.3828,
+      "step": 527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.85779125162408e-06,
+      "loss": 5.2383,
+      "step": 528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.855627977479429e-06,
+      "loss": 4.7266,
+      "step": 529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.853464703334777e-06,
+      "loss": 4.9141,
+      "step": 530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.851301429190126e-06,
+      "loss": 5.4766,
+      "step": 531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.849138155045476e-06,
+      "loss": 5.1641,
+      "step": 532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.846974880900823e-06,
+      "loss": 5.2734,
+      "step": 533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.844811606756172e-06,
+      "loss": 5.0234,
+      "step": 534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.84264833261152e-06,
+      "loss": 4.9688,
+      "step": 535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.84048505846687e-06,
+      "loss": 5.0547,
+      "step": 536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.838321784322218e-06,
+      "loss": 5.3281,
+      "step": 537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.836158510177567e-06,
+      "loss": 4.6875,
+      "step": 538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.833995236032915e-06,
+      "loss": 4.4141,
+      "step": 539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.831831961888264e-06,
+      "loss": 5.1562,
+      "step": 540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.829668687743613e-06,
+      "loss": 4.8672,
+      "step": 541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.827505413598961e-06,
+      "loss": 4.4297,
+      "step": 542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.82534213945431e-06,
+      "loss": 5.0,
+      "step": 543
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.823178865309659e-06,
+      "loss": 5.1055,
+      "step": 544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.821015591165007e-06,
+      "loss": 5.3008,
+      "step": 545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.818852317020356e-06,
+      "loss": 5.6211,
+      "step": 546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.816689042875706e-06,
+      "loss": 5.1094,
+      "step": 547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.814525768731053e-06,
+      "loss": 4.9844,
+      "step": 548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.812362494586402e-06,
+      "loss": 5.3438,
+      "step": 549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.810199220441749e-06,
+      "loss": 5.2422,
+      "step": 550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.8080359462971e-06,
+      "loss": 4.9297,
+      "step": 551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.805872672152448e-06,
+      "loss": 4.7578,
+      "step": 552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.803709398007797e-06,
+      "loss": 4.9609,
+      "step": 553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.801546123863145e-06,
+      "loss": 4.3125,
+      "step": 554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.799382849718494e-06,
+      "loss": 4.7578,
+      "step": 555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.797219575573843e-06,
+      "loss": 5.4375,
+      "step": 556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.795056301429191e-06,
+      "loss": 5.3906,
+      "step": 557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.79289302728454e-06,
+      "loss": 5.207,
+      "step": 558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.790729753139889e-06,
+      "loss": 4.1875,
+      "step": 559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.788566478995236e-06,
+      "loss": 5.3672,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.786403204850584e-06,
+      "loss": 4.1289,
+      "step": 561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.784239930705933e-06,
+      "loss": 5.3203,
+      "step": 562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.782076656561283e-06,
+      "loss": 4.8438,
+      "step": 563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.779913382416632e-06,
+      "loss": 5.625,
+      "step": 564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.77775010827198e-06,
+      "loss": 4.8047,
+      "step": 565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.77558683412733e-06,
+      "loss": 5.1523,
+      "step": 566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.773423559982676e-06,
+      "loss": 5.0195,
+      "step": 567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.771260285838026e-06,
+      "loss": 5.1602,
+      "step": 568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.769097011693375e-06,
+      "loss": 5.4922,
+      "step": 569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.766933737548722e-06,
+      "loss": 4.7031,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.764770463404072e-06,
+      "loss": 4.4805,
+      "step": 571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.762607189259421e-06,
+      "loss": 5.5664,
+      "step": 572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.760443915114768e-06,
+      "loss": 5.3047,
+      "step": 573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.758280640970117e-06,
+      "loss": 4.9609,
+      "step": 574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.756117366825465e-06,
+      "loss": 5.1484,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.753954092680816e-06,
+      "loss": 5.1406,
+      "step": 576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.751790818536164e-06,
+      "loss": 4.7461,
+      "step": 577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.749627544391513e-06,
+      "loss": 5.25,
+      "step": 578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.747464270246862e-06,
+      "loss": 3.8379,
+      "step": 579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.745300996102209e-06,
+      "loss": 4.5977,
+      "step": 580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.743137721957559e-06,
+      "loss": 4.9336,
+      "step": 581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.740974447812906e-06,
+      "loss": 5.3828,
+      "step": 582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.738811173668256e-06,
+      "loss": 5.1016,
+      "step": 583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.736647899523603e-06,
+      "loss": 4.8984,
+      "step": 584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.734484625378952e-06,
+      "loss": 4.9766,
+      "step": 585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.7323213512343e-06,
+      "loss": 5.1875,
+      "step": 586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.73015807708965e-06,
+      "loss": 4.9805,
+      "step": 587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.727994802945e-06,
+      "loss": 4.8516,
+      "step": 588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.725831528800347e-06,
+      "loss": 5.3906,
+      "step": 589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.723668254655697e-06,
+      "loss": 5.6367,
+      "step": 590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.721504980511046e-06,
+      "loss": 5.3477,
+      "step": 591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.719341706366393e-06,
+      "loss": 5.2891,
+      "step": 592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.717178432221741e-06,
+      "loss": 5.0391,
+      "step": 593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.71501515807709e-06,
+      "loss": 5.1133,
+      "step": 594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.712851883932439e-06,
+      "loss": 4.7188,
+      "step": 595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.710688609787787e-06,
+      "loss": 4.2656,
+      "step": 596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.708525335643136e-06,
+      "loss": 4.9453,
+      "step": 597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.706362061498486e-06,
+      "loss": 5.2695,
+      "step": 598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.704198787353833e-06,
+      "loss": 4.3945,
+      "step": 599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.702035513209184e-06,
+      "loss": 5.3711,
+      "step": 600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.69987223906453e-06,
+      "loss": 4.9844,
+      "step": 601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.697708964919879e-06,
+      "loss": 5.4336,
+      "step": 602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.69554569077523e-06,
+      "loss": 5.4414,
+      "step": 603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.693382416630576e-06,
+      "loss": 5.1055,
+      "step": 604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.691219142485925e-06,
+      "loss": 4.2305,
+      "step": 605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.689055868341274e-06,
+      "loss": 5.0625,
+      "step": 606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.686892594196622e-06,
+      "loss": 5.3672,
+      "step": 607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.684729320051971e-06,
+      "loss": 4.9336,
+      "step": 608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.68256604590732e-06,
+      "loss": 4.0586,
+      "step": 609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.68040277176267e-06,
+      "loss": 5.1406,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.678239497618019e-06,
+      "loss": 4.8008,
+      "step": 611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.676076223473367e-06,
+      "loss": 5.1875,
+      "step": 612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.673912949328714e-06,
+      "loss": 5.1328,
+      "step": 613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.671749675184063e-06,
+      "loss": 4.9883,
+      "step": 614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.669586401039413e-06,
+      "loss": 5.0469,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.66742312689476e-06,
+      "loss": 5.0859,
+      "step": 616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.665259852750109e-06,
+      "loss": 5.1992,
+      "step": 617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.663096578605456e-06,
+      "loss": 5.0312,
+      "step": 618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.660933304460806e-06,
+      "loss": 4.7344,
+      "step": 619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.658770030316153e-06,
+      "loss": 5.1016,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.656606756171504e-06,
+      "loss": 4.9883,
+      "step": 621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.654443482026852e-06,
+      "loss": 5.0469,
+      "step": 622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.652280207882201e-06,
+      "loss": 5.1289,
+      "step": 623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.65011693373755e-06,
+      "loss": 5.3047,
+      "step": 624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.647953659592898e-06,
+      "loss": 4.1406,
+      "step": 625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.645790385448247e-06,
+      "loss": 4.8984,
+      "step": 626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.643627111303596e-06,
+      "loss": 5.2656,
+      "step": 627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.641463837158944e-06,
+      "loss": 4.4336,
+      "step": 628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.639300563014293e-06,
+      "loss": 5.1953,
+      "step": 629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.63713728886964e-06,
+      "loss": 4.9141,
+      "step": 630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.63497401472499e-06,
+      "loss": 4.5547,
+      "step": 631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.632810740580337e-06,
+      "loss": 4.875,
+      "step": 632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.630647466435688e-06,
+      "loss": 5.2344,
+      "step": 633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.628484192291036e-06,
+      "loss": 5.2578,
+      "step": 634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.626320918146385e-06,
+      "loss": 4.918,
+      "step": 635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.624157644001733e-06,
+      "loss": 4.293,
+      "step": 636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.621994369857082e-06,
+      "loss": 5.4219,
+      "step": 637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.619831095712429e-06,
+      "loss": 4.9102,
+      "step": 638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.61766782156778e-06,
+      "loss": 4.7266,
+      "step": 639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.615504547423128e-06,
+      "loss": 5.0938,
+      "step": 640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.613341273278477e-06,
+      "loss": 5.375,
+      "step": 641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.611177999133825e-06,
+      "loss": 5.0273,
+      "step": 642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.609014724989174e-06,
+      "loss": 5.3281,
+      "step": 643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.606851450844523e-06,
+      "loss": 4.9414,
+      "step": 644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.604688176699871e-06,
+      "loss": 4.3242,
+      "step": 645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.60252490255522e-06,
+      "loss": 5.0742,
+      "step": 646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.600361628410569e-06,
+      "loss": 4.9023,
+      "step": 647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.598198354265917e-06,
+      "loss": 5.0078,
+      "step": 648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.596035080121266e-06,
+      "loss": 5.2031,
+      "step": 649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.593871805976613e-06,
+      "loss": 5.1758,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.591708531831963e-06,
+      "loss": 4.8164,
+      "step": 651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.58954525768731e-06,
+      "loss": 5.3164,
+      "step": 652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.587381983542659e-06,
+      "loss": 5.5586,
+      "step": 653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.585218709398008e-06,
+      "loss": 4.9688,
+      "step": 654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.583055435253356e-06,
+      "loss": 5.0469,
+      "step": 655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.580892161108707e-06,
+      "loss": 5.3789,
+      "step": 656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.578728886964054e-06,
+      "loss": 5.0312,
+      "step": 657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.576565612819402e-06,
+      "loss": 5.1836,
+      "step": 658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.574402338674753e-06,
+      "loss": 5.2852,
+      "step": 659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.5722390645301e-06,
+      "loss": 4.9531,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.57007579038545e-06,
+      "loss": 5.5469,
+      "step": 661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.567912516240799e-06,
+      "loss": 5.2031,
+      "step": 662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.565749242096146e-06,
+      "loss": 5.375,
+      "step": 663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.563585967951494e-06,
+      "loss": 4.9609,
+      "step": 664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.561422693806843e-06,
+      "loss": 5.1484,
+      "step": 665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.559259419662192e-06,
+      "loss": 4.9375,
+      "step": 666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.55709614551754e-06,
+      "loss": 5.0195,
+      "step": 667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.55493287137289e-06,
+      "loss": 4.9531,
+      "step": 668
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.552769597228237e-06,
+      "loss": 5.1211,
+      "step": 669
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.550606323083586e-06,
+      "loss": 5.1406,
+      "step": 670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.548443048938937e-06,
+      "loss": 4.4844,
+      "step": 671
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.546279774794283e-06,
+      "loss": 5.2773,
+      "step": 672
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.544116500649632e-06,
+      "loss": 4.6094,
+      "step": 673
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.541953226504982e-06,
+      "loss": 4.8789,
+      "step": 674
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.53978995236033e-06,
+      "loss": 4.3008,
+      "step": 675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.537626678215678e-06,
+      "loss": 4.8633,
+      "step": 676
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.535463404071027e-06,
+      "loss": 4.6953,
+      "step": 677
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.533300129926375e-06,
+      "loss": 4.7031,
+      "step": 678
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.531136855781726e-06,
+      "loss": 4.8008,
+      "step": 679
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.528973581637074e-06,
+      "loss": 4.7031,
+      "step": 680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.526810307492423e-06,
+      "loss": 5.4062,
+      "step": 681
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.524647033347772e-06,
+      "loss": 5.1641,
+      "step": 682
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.522483759203119e-06,
+      "loss": 4.8047,
+      "step": 683
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.520320485058467e-06,
+      "loss": 4.6055,
+      "step": 684
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.518157210913816e-06,
+      "loss": 4.9531,
+      "step": 685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.515993936769165e-06,
+      "loss": 5.2266,
+      "step": 686
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.513830662624513e-06,
+      "loss": 4.0234,
+      "step": 687
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.51166738847986e-06,
+      "loss": 4.3359,
+      "step": 688
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.509504114335209e-06,
+      "loss": 5.3906,
+      "step": 689
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.50734084019056e-06,
+      "loss": 5.3594,
+      "step": 690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.505177566045908e-06,
+      "loss": 4.4336,
+      "step": 691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.503014291901257e-06,
+      "loss": 5.1797,
+      "step": 692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.500851017756607e-06,
+      "loss": 5.4922,
+      "step": 693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.498687743611952e-06,
+      "loss": 5.1562,
+      "step": 694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.496524469467303e-06,
+      "loss": 5.3555,
+      "step": 695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.494361195322651e-06,
+      "loss": 4.1992,
+      "step": 696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.492197921178e-06,
+      "loss": 5.2227,
+      "step": 697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.490034647033349e-06,
+      "loss": 4.8086,
+      "step": 698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.487871372888697e-06,
+      "loss": 4.7969,
+      "step": 699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.485708098744044e-06,
+      "loss": 5.0195,
+      "step": 700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.483544824599393e-06,
+      "loss": 4.6172,
+      "step": 701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.481381550454743e-06,
+      "loss": 4.5,
+      "step": 702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.479218276310092e-06,
+      "loss": 5.1836,
+      "step": 703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.47705500216544e-06,
+      "loss": 4.918,
+      "step": 704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.47489172802079e-06,
+      "loss": 4.5469,
+      "step": 705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.472728453876138e-06,
+      "loss": 5.043,
+      "step": 706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.470565179731486e-06,
+      "loss": 4.3359,
+      "step": 707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.468401905586835e-06,
+      "loss": 4.6719,
+      "step": 708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.466238631442182e-06,
+      "loss": 5.3828,
+      "step": 709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.464075357297532e-06,
+      "loss": 4.5977,
+      "step": 710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.461912083152881e-06,
+      "loss": 5.1562,
+      "step": 711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.45974880900823e-06,
+      "loss": 4.9453,
+      "step": 712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.457585534863578e-06,
+      "loss": 4.8711,
+      "step": 713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.455422260718927e-06,
+      "loss": 5.1055,
+      "step": 714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.453258986574276e-06,
+      "loss": 5.1797,
+      "step": 715
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.451095712429624e-06,
+      "loss": 4.9141,
+      "step": 716
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.448932438284973e-06,
+      "loss": 5.2148,
+      "step": 717
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.446769164140322e-06,
+      "loss": 5.2695,
+      "step": 718
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.444605889995669e-06,
+      "loss": 3.1699,
+      "step": 719
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.442442615851017e-06,
+      "loss": 5.2383,
+      "step": 720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.440279341706366e-06,
+      "loss": 4.582,
+      "step": 721
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.438116067561715e-06,
+      "loss": 5.3281,
+      "step": 722
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.435952793417063e-06,
+      "loss": 4.9766,
+      "step": 723
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.433789519272414e-06,
+      "loss": 3.9102,
+      "step": 724
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.431626245127762e-06,
+      "loss": 4.8281,
+      "step": 725
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.429462970983111e-06,
+      "loss": 5.0625,
+      "step": 726
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.42729969683846e-06,
+      "loss": 5.4297,
+      "step": 727
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.425136422693807e-06,
+      "loss": 5.2227,
+      "step": 728
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.422973148549157e-06,
+      "loss": 4.7227,
+      "step": 729
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.420809874404506e-06,
+      "loss": 4.6094,
+      "step": 730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.418646600259853e-06,
+      "loss": 5.2031,
+      "step": 731
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.416483326115203e-06,
+      "loss": 4.9531,
+      "step": 732
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.41432005197055e-06,
+      "loss": 4.6172,
+      "step": 733
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.412156777825899e-06,
+      "loss": 5.1406,
+      "step": 734
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.409993503681247e-06,
+      "loss": 4.3516,
+      "step": 735
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.407830229536596e-06,
+      "loss": 4.5078,
+      "step": 736
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.405666955391946e-06,
+      "loss": 4.2461,
+      "step": 737
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.403503681247295e-06,
+      "loss": 5.375,
+      "step": 738
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.401340407102644e-06,
+      "loss": 4.6484,
+      "step": 739
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.39917713295799e-06,
+      "loss": 5.0078,
+      "step": 740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.397013858813339e-06,
+      "loss": 4.9648,
+      "step": 741
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.39485058466869e-06,
+      "loss": 4.4141,
+      "step": 742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.392687310524036e-06,
+      "loss": 5.0664,
+      "step": 743
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.390524036379387e-06,
+      "loss": 4.5508,
+      "step": 744
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.388360762234735e-06,
+      "loss": 4.7891,
+      "step": 745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.386197488090082e-06,
+      "loss": 4.7461,
+      "step": 746
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.384034213945431e-06,
+      "loss": 4.3867,
+      "step": 747
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.38187093980078e-06,
+      "loss": 3.7891,
+      "step": 748
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.37970766565613e-06,
+      "loss": 4.7578,
+      "step": 749
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.377544391511479e-06,
+      "loss": 5.0586,
+      "step": 750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.375381117366826e-06,
+      "loss": 5.1797,
+      "step": 751
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.373217843222174e-06,
+      "loss": 4.7578,
+      "step": 752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.371054569077523e-06,
+      "loss": 3.9746,
+      "step": 753
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.368891294932872e-06,
+      "loss": 4.4102,
+      "step": 754
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.36672802078822e-06,
+      "loss": 4.7305,
+      "step": 755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.364564746643569e-06,
+      "loss": 4.4961,
+      "step": 756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.362401472498918e-06,
+      "loss": 4.1484,
+      "step": 757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.360238198354266e-06,
+      "loss": 4.3281,
+      "step": 758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.358074924209615e-06,
+      "loss": 4.8789,
+      "step": 759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.355911650064964e-06,
+      "loss": 4.625,
+      "step": 760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.353748375920312e-06,
+      "loss": 5.1562,
+      "step": 761
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.351585101775661e-06,
+      "loss": 5.2031,
+      "step": 762
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.34942182763101e-06,
+      "loss": 5.1641,
+      "step": 763
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.34725855348636e-06,
+      "loss": 4.4844,
+      "step": 764
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.345095279341707e-06,
+      "loss": 4.9297,
+      "step": 765
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.342932005197056e-06,
+      "loss": 5.1055,
+      "step": 766
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.340768731052404e-06,
+      "loss": 4.9453,
+      "step": 767
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.338605456907753e-06,
+      "loss": 4.3242,
+      "step": 768
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.336442182763102e-06,
+      "loss": 4.9648,
+      "step": 769
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.33427890861845e-06,
+      "loss": 4.3359,
+      "step": 770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.332115634473799e-06,
+      "loss": 5.1719,
+      "step": 771
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.329952360329148e-06,
+      "loss": 4.4805,
+      "step": 772
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.327789086184496e-06,
+      "loss": 5.2227,
+      "step": 773
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.325625812039845e-06,
+      "loss": 3.8477,
+      "step": 774
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.323462537895194e-06,
+      "loss": 5.1094,
+      "step": 775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.321299263750542e-06,
+      "loss": 4.793,
+      "step": 776
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.31913598960589e-06,
+      "loss": 4.3633,
+      "step": 777
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.31697271546124e-06,
+      "loss": 4.9531,
+      "step": 778
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.314809441316586e-06,
+      "loss": 5.2695,
+      "step": 779
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.312646167171937e-06,
+      "loss": 5.4023,
+      "step": 780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.310482893027285e-06,
+      "loss": 4.7969,
+      "step": 781
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.308319618882634e-06,
+      "loss": 4.3477,
+      "step": 782
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.306156344737983e-06,
+      "loss": 5.0156,
+      "step": 783
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.30399307059333e-06,
+      "loss": 5.25,
+      "step": 784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.30182979644868e-06,
+      "loss": 4.8281,
+      "step": 785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.299666522304029e-06,
+      "loss": 4.8516,
+      "step": 786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.297503248159376e-06,
+      "loss": 4.75,
+      "step": 787
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.295339974014726e-06,
+      "loss": 4.8242,
+      "step": 788
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.293176699870075e-06,
+      "loss": 4.6133,
+      "step": 789
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.291013425725422e-06,
+      "loss": 4.4922,
+      "step": 790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.28885015158077e-06,
+      "loss": 4.9414,
+      "step": 791
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.286686877436119e-06,
+      "loss": 4.9102,
+      "step": 792
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.28452360329147e-06,
+      "loss": 5.1602,
+      "step": 793
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.282360329146818e-06,
+      "loss": 5.0938,
+      "step": 794
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.280197055002167e-06,
+      "loss": 4.7031,
+      "step": 795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.278033780857515e-06,
+      "loss": 4.9766,
+      "step": 796
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.275870506712862e-06,
+      "loss": 4.875,
+      "step": 797
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.273707232568213e-06,
+      "loss": 5.0273,
+      "step": 798
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.27154395842356e-06,
+      "loss": 4.918,
+      "step": 799
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.26938068427891e-06,
+      "loss": 4.6758,
+      "step": 800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.267217410134259e-06,
+      "loss": 5.3438,
+      "step": 801
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.265054135989606e-06,
+      "loss": 5.4688,
+      "step": 802
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.262890861844954e-06,
+      "loss": 4.9961,
+      "step": 803
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.260727587700303e-06,
+      "loss": 4.8516,
+      "step": 804
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.258564313555653e-06,
+      "loss": 4.1719,
+      "step": 805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.256401039411002e-06,
+      "loss": 5.0664,
+      "step": 806
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.25423776526635e-06,
+      "loss": 4.6523,
+      "step": 807
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.2520744911217e-06,
+      "loss": 4.8438,
+      "step": 808
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.249911216977048e-06,
+      "loss": 4.3438,
+      "step": 809
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.247747942832397e-06,
+      "loss": 5.1367,
+      "step": 810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.245584668687743e-06,
+      "loss": 5.1758,
+      "step": 811
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.243421394543092e-06,
+      "loss": 4.9297,
+      "step": 812
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.24125812039844e-06,
+      "loss": 5.3438,
+      "step": 813
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.23909484625379e-06,
+      "loss": 5.168,
+      "step": 814
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.23693157210914e-06,
+      "loss": 5.2305,
+      "step": 815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.234768297964487e-06,
+      "loss": 5.3281,
+      "step": 816
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.232605023819837e-06,
+      "loss": 4.9531,
+      "step": 817
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.230441749675184e-06,
+      "loss": 4.9219,
+      "step": 818
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.228278475530533e-06,
+      "loss": 4.5938,
+      "step": 819
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.226115201385883e-06,
+      "loss": 4.668,
+      "step": 820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.22395192724123e-06,
+      "loss": 5.1992,
+      "step": 821
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.221788653096579e-06,
+      "loss": 4.6172,
+      "step": 822
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.219625378951927e-06,
+      "loss": 4.9922,
+      "step": 823
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.217462104807276e-06,
+      "loss": 4.543,
+      "step": 824
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.215298830662625e-06,
+      "loss": 3.9961,
+      "step": 825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.213135556517973e-06,
+      "loss": 5.1953,
+      "step": 826
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.210972282373324e-06,
+      "loss": 5.6328,
+      "step": 827
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.208809008228672e-06,
+      "loss": 5.1172,
+      "step": 828
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.206645734084021e-06,
+      "loss": 4.5508,
+      "step": 829
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.204482459939368e-06,
+      "loss": 4.25,
+      "step": 830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.202319185794717e-06,
+      "loss": 4.9922,
+      "step": 831
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.200155911650067e-06,
+      "loss": 5.1953,
+      "step": 832
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.197992637505414e-06,
+      "loss": 5.2461,
+      "step": 833
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.195829363360763e-06,
+      "loss": 4.9922,
+      "step": 834
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.193666089216113e-06,
+      "loss": 4.625,
+      "step": 835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.19150281507146e-06,
+      "loss": 5.4141,
+      "step": 836
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.189339540926809e-06,
+      "loss": 3.8438,
+      "step": 837
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.187176266782157e-06,
+      "loss": 4.8828,
+      "step": 838
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.185012992637506e-06,
+      "loss": 4.9883,
+      "step": 839
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.182849718492856e-06,
+      "loss": 5.5078,
+      "step": 840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.180686444348205e-06,
+      "loss": 5.1875,
+      "step": 841
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.178523170203552e-06,
+      "loss": 5.0859,
+      "step": 842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.1763598960589e-06,
+      "loss": 3.9609,
+      "step": 843
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.17419662191425e-06,
+      "loss": 4.6055,
+      "step": 844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.172033347769598e-06,
+      "loss": 5.2891,
+      "step": 845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.169870073624947e-06,
+      "loss": 4.9336,
+      "step": 846
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.167706799480293e-06,
+      "loss": 4.8828,
+      "step": 847
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.165543525335644e-06,
+      "loss": 5.1016,
+      "step": 848
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.163380251190992e-06,
+      "loss": 5.043,
+      "step": 849
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.161216977046341e-06,
+      "loss": 5.1602,
+      "step": 850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.15905370290169e-06,
+      "loss": 4.9141,
+      "step": 851
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.156890428757038e-06,
+      "loss": 4.625,
+      "step": 852
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.154727154612387e-06,
+      "loss": 4.4375,
+      "step": 853
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.152563880467736e-06,
+      "loss": 4.7891,
+      "step": 854
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.150400606323083e-06,
+      "loss": 4.1172,
+      "step": 855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.148237332178433e-06,
+      "loss": 4.8555,
+      "step": 856
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.146074058033782e-06,
+      "loss": 4.5117,
+      "step": 857
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.14391078388913e-06,
+      "loss": 4.4492,
+      "step": 858
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.141747509744479e-06,
+      "loss": 5.2617,
+      "step": 859
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.139584235599828e-06,
+      "loss": 5.1836,
+      "step": 860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.137420961455176e-06,
+      "loss": 4.4023,
+      "step": 861
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.135257687310525e-06,
+      "loss": 4.9141,
+      "step": 862
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.133094413165874e-06,
+      "loss": 4.9531,
+      "step": 863
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.130931139021222e-06,
+      "loss": 5.1484,
+      "step": 864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.128767864876571e-06,
+      "loss": 4.9961,
+      "step": 865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.12660459073192e-06,
+      "loss": 5.0664,
+      "step": 866
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.124441316587267e-06,
+      "loss": 5.2656,
+      "step": 867
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.122278042442617e-06,
+      "loss": 4.5859,
+      "step": 868
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.120114768297966e-06,
+      "loss": 3.7559,
+      "step": 869
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.117951494153313e-06,
+      "loss": 5.2695,
+      "step": 870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.115788220008663e-06,
+      "loss": 4.8828,
+      "step": 871
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.113624945864012e-06,
+      "loss": 5.0625,
+      "step": 872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.11146167171936e-06,
+      "loss": 4.3672,
+      "step": 873
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.109298397574709e-06,
+      "loss": 4.6562,
+      "step": 874
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.107135123430058e-06,
+      "loss": 4.2578,
+      "step": 875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.104971849285406e-06,
+      "loss": 4.4023,
+      "step": 876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.102808575140755e-06,
+      "loss": 5.4258,
+      "step": 877
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.100645300996104e-06,
+      "loss": 5.2578,
+      "step": 878
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.098482026851452e-06,
+      "loss": 4.9336,
+      "step": 879
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.096318752706801e-06,
+      "loss": 4.8711,
+      "step": 880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.094155478562148e-06,
+      "loss": 5.0938,
+      "step": 881
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.091992204417496e-06,
+      "loss": 3.6211,
+      "step": 882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.089828930272847e-06,
+      "loss": 4.3125,
+      "step": 883
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.087665656128195e-06,
+      "loss": 4.8789,
+      "step": 884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.085502381983544e-06,
+      "loss": 5.3398,
+      "step": 885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.083339107838891e-06,
+      "loss": 5.3711,
+      "step": 886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.08117583369424e-06,
+      "loss": 4.3516,
+      "step": 887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.07901255954959e-06,
+      "loss": 4.9414,
+      "step": 888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.076849285404937e-06,
+      "loss": 4.6328,
+      "step": 889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.074686011260286e-06,
+      "loss": 4.5898,
+      "step": 890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.072522737115636e-06,
+      "loss": 5.1367,
+      "step": 891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.070359462970983e-06,
+      "loss": 5.1016,
+      "step": 892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.068196188826332e-06,
+      "loss": 4.6797,
+      "step": 893
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.06603291468168e-06,
+      "loss": 4.1172,
+      "step": 894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.063869640537029e-06,
+      "loss": 4.7305,
+      "step": 895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.06170636639238e-06,
+      "loss": 4.8984,
+      "step": 896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.059543092247728e-06,
+      "loss": 5.2344,
+      "step": 897
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.057379818103077e-06,
+      "loss": 5.1172,
+      "step": 898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.055216543958425e-06,
+      "loss": 5.1016,
+      "step": 899
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.053053269813772e-06,
+      "loss": 5.2734,
+      "step": 900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.050889995669121e-06,
+      "loss": 5.207,
+      "step": 901
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.04872672152447e-06,
+      "loss": 5.125,
+      "step": 902
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.04656344737982e-06,
+      "loss": 5.0039,
+      "step": 903
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.044400173235167e-06,
+      "loss": 5.2227,
+      "step": 904
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.042236899090516e-06,
+      "loss": 4.582,
+      "step": 905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.040073624945864e-06,
+      "loss": 4.832,
+      "step": 906
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.037910350801213e-06,
+      "loss": 4.8867,
+      "step": 907
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.035747076656563e-06,
+      "loss": 4.2969,
+      "step": 908
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.03358380251191e-06,
+      "loss": 5.1953,
+      "step": 909
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.031420528367259e-06,
+      "loss": 5.1328,
+      "step": 910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.02925725422261e-06,
+      "loss": 4.8359,
+      "step": 911
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.027093980077956e-06,
+      "loss": 3.75,
+      "step": 912
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.024930705933305e-06,
+      "loss": 4.7422,
+      "step": 913
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.022767431788654e-06,
+      "loss": 5.168,
+      "step": 914
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.020604157644002e-06,
+      "loss": 4.4492,
+      "step": 915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.01844088349935e-06,
+      "loss": 4.7734,
+      "step": 916
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.0162776093547e-06,
+      "loss": 4.8086,
+      "step": 917
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.01411433521005e-06,
+      "loss": 5.25,
+      "step": 918
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.011951061065397e-06,
+      "loss": 5.3516,
+      "step": 919
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.009787786920745e-06,
+      "loss": 4.4844,
+      "step": 920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.007624512776094e-06,
+      "loss": 4.582,
+      "step": 921
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.005461238631443e-06,
+      "loss": 3.7168,
+      "step": 922
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.003297964486791e-06,
+      "loss": 4.2773,
+      "step": 923
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.00113469034214e-06,
+      "loss": 3.4414,
+      "step": 924
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.998971416197489e-06,
+      "loss": 5.0312,
+      "step": 925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.996808142052836e-06,
+      "loss": 5.2812,
+      "step": 926
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.994644867908186e-06,
+      "loss": 5.0469,
+      "step": 927
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.992481593763535e-06,
+      "loss": 4.6328,
+      "step": 928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.990318319618883e-06,
+      "loss": 5.0859,
+      "step": 929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.988155045474234e-06,
+      "loss": 5.2383,
+      "step": 930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.985991771329582e-06,
+      "loss": 4.5938,
+      "step": 931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.98382849718493e-06,
+      "loss": 4.7188,
+      "step": 932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.981665223040278e-06,
+      "loss": 4.8359,
+      "step": 933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.979501948895627e-06,
+      "loss": 4.3477,
+      "step": 934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.977338674750975e-06,
+      "loss": 5.1953,
+      "step": 935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.975175400606324e-06,
+      "loss": 4.9258,
+      "step": 936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.973012126461673e-06,
+      "loss": 5.0859,
+      "step": 937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.97084885231702e-06,
+      "loss": 5.2344,
+      "step": 938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.96868557817237e-06,
+      "loss": 5.0938,
+      "step": 939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.966522304027719e-06,
+      "loss": 4.9531,
+      "step": 940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.964359029883067e-06,
+      "loss": 4.1719,
+      "step": 941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.962195755738416e-06,
+      "loss": 4.9375,
+      "step": 942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.960032481593765e-06,
+      "loss": 4.8047,
+      "step": 943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.957869207449113e-06,
+      "loss": 4.9336,
+      "step": 944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.955705933304462e-06,
+      "loss": 4.957,
+      "step": 945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.95354265915981e-06,
+      "loss": 4.4805,
+      "step": 946
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.95137938501516e-06,
+      "loss": 4.375,
+      "step": 947
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.949216110870506e-06,
+      "loss": 4.7656,
+      "step": 948
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.947052836725857e-06,
+      "loss": 4.8438,
+      "step": 949
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.944889562581204e-06,
+      "loss": 5.2188,
+      "step": 950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.942726288436552e-06,
+      "loss": 4.7266,
+      "step": 951
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.940563014291903e-06,
+      "loss": 5.4023,
+      "step": 952
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.938399740147251e-06,
+      "loss": 4.5625,
+      "step": 953
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.9362364660026e-06,
+      "loss": 4.5273,
+      "step": 954
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.934073191857948e-06,
+      "loss": 5.0742,
+      "step": 955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.931909917713297e-06,
+      "loss": 4.5625,
+      "step": 956
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.929746643568644e-06,
+      "loss": 4.3125,
+      "step": 957
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.927583369423993e-06,
+      "loss": 4.5664,
+      "step": 958
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.925420095279343e-06,
+      "loss": 5.5156,
+      "step": 959
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.92325682113469e-06,
+      "loss": 5.5859,
+      "step": 960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.92109354699004e-06,
+      "loss": 4.7344,
+      "step": 961
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.918930272845389e-06,
+      "loss": 4.8125,
+      "step": 962
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.916766998700738e-06,
+      "loss": 4.8008,
+      "step": 963
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.914603724556086e-06,
+      "loss": 5.0195,
+      "step": 964
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.912440450411435e-06,
+      "loss": 5.293,
+      "step": 965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.910277176266784e-06,
+      "loss": 4.8438,
+      "step": 966
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.908113902122132e-06,
+      "loss": 5.1133,
+      "step": 967
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.905950627977481e-06,
+      "loss": 4.7266,
+      "step": 968
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.90378735383283e-06,
+      "loss": 4.2891,
+      "step": 969
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.901624079688177e-06,
+      "loss": 5.3906,
+      "step": 970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.899460805543527e-06,
+      "loss": 5.3555,
+      "step": 971
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.897297531398874e-06,
+      "loss": 5.1992,
+      "step": 972
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.895134257254223e-06,
+      "loss": 5.1133,
+      "step": 973
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.892970983109571e-06,
+      "loss": 5.0547,
+      "step": 974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.890807708964922e-06,
+      "loss": 4.293,
+      "step": 975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.88864443482027e-06,
+      "loss": 4.7422,
+      "step": 976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.886481160675617e-06,
+      "loss": 5.1953,
+      "step": 977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.884317886530966e-06,
+      "loss": 5.0156,
+      "step": 978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.882154612386315e-06,
+      "loss": 4.8633,
+      "step": 979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.879991338241663e-06,
+      "loss": 4.4805,
+      "step": 980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.877828064097014e-06,
+      "loss": 4.8633,
+      "step": 981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.87566478995236e-06,
+      "loss": 5.1914,
+      "step": 982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.87350151580771e-06,
+      "loss": 4.7188,
+      "step": 983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.871338241663058e-06,
+      "loss": 4.625,
+      "step": 984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.869174967518407e-06,
+      "loss": 4.7656,
+      "step": 985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.867011693373755e-06,
+      "loss": 4.9883,
+      "step": 986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.864848419229106e-06,
+      "loss": 4.0078,
+      "step": 987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.862685145084452e-06,
+      "loss": 5.0781,
+      "step": 988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.860521870939801e-06,
+      "loss": 5.1797,
+      "step": 989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.85835859679515e-06,
+      "loss": 4.4805,
+      "step": 990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.856195322650498e-06,
+      "loss": 5.1992,
+      "step": 991
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.854032048505847e-06,
+      "loss": 5.1641,
+      "step": 992
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.851868774361196e-06,
+      "loss": 5.1367,
+      "step": 993
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.849705500216544e-06,
+      "loss": 4.9609,
+      "step": 994
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.847542226071893e-06,
+      "loss": 5.2188,
+      "step": 995
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.845378951927242e-06,
+      "loss": 5.1016,
+      "step": 996
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.84321567778259e-06,
+      "loss": 4.8047,
+      "step": 997
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.841052403637939e-06,
+      "loss": 4.6055,
+      "step": 998
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.83888912949329e-06,
+      "loss": 5.1797,
+      "step": 999
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.836725855348636e-06,
+      "loss": 5.3086,
+      "step": 1000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.834562581203987e-06,
+      "loss": 5.1602,
+      "step": 1001
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.832399307059335e-06,
+      "loss": 5.4766,
+      "step": 1002
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.830236032914682e-06,
+      "loss": 4.918,
+      "step": 1003
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.828072758770031e-06,
+      "loss": 5.0625,
+      "step": 1004
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.82590948462538e-06,
+      "loss": 5.3203,
+      "step": 1005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.823746210480728e-06,
+      "loss": 5.1875,
+      "step": 1006
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.821582936336077e-06,
+      "loss": 4.2109,
+      "step": 1007
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.819419662191424e-06,
+      "loss": 5.1602,
+      "step": 1008
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.817256388046773e-06,
+      "loss": 4.4531,
+      "step": 1009
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.815093113902123e-06,
+      "loss": 5.0625,
+      "step": 1010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.812929839757472e-06,
+      "loss": 4.6367,
+      "step": 1011
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.81076656561282e-06,
+      "loss": 3.7793,
+      "step": 1012
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.808603291468169e-06,
+      "loss": 4.6797,
+      "step": 1013
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.806440017323516e-06,
+      "loss": 5.2734,
+      "step": 1014
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.804276743178866e-06,
+      "loss": 4.9414,
+      "step": 1015
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.802113469034213e-06,
+      "loss": 4.7227,
+      "step": 1016
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.799950194889564e-06,
+      "loss": 4.7852,
+      "step": 1017
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.797786920744912e-06,
+      "loss": 4.8555,
+      "step": 1018
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.79562364660026e-06,
+      "loss": 4.9023,
+      "step": 1019
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.793460372455608e-06,
+      "loss": 5.1523,
+      "step": 1020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.791297098310956e-06,
+      "loss": 4.1836,
+      "step": 1021
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.789133824166307e-06,
+      "loss": 4.3008,
+      "step": 1022
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.786970550021656e-06,
+      "loss": 4.7422,
+      "step": 1023
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.784807275877004e-06,
+      "loss": 4.9219,
+      "step": 1024
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.782644001732353e-06,
+      "loss": 4.8242,
+      "step": 1025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.780480727587701e-06,
+      "loss": 4.3945,
+      "step": 1026
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.77831745344305e-06,
+      "loss": 4.918,
+      "step": 1027
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.776154179298397e-06,
+      "loss": 4.8867,
+      "step": 1028
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.773990905153746e-06,
+      "loss": 5.4648,
+      "step": 1029
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.771827631009096e-06,
+      "loss": 4.9258,
+      "step": 1030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.769664356864443e-06,
+      "loss": 4.8789,
+      "step": 1031
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.767501082719793e-06,
+      "loss": 4.6562,
+      "step": 1032
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.765337808575144e-06,
+      "loss": 5.2031,
+      "step": 1033
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.763174534430489e-06,
+      "loss": 4.9922,
+      "step": 1034
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.76101126028584e-06,
+      "loss": 4.8555,
+      "step": 1035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.758847986141188e-06,
+      "loss": 3.0059,
+      "step": 1036
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.756684711996537e-06,
+      "loss": 4.8945,
+      "step": 1037
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.754521437851885e-06,
+      "loss": 5.0352,
+      "step": 1038
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.752358163707232e-06,
+      "loss": 5.2266,
+      "step": 1039
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.750194889562581e-06,
+      "loss": 4.0508,
+      "step": 1040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.74803161541793e-06,
+      "loss": 5.4062,
+      "step": 1041
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.745868341273278e-06,
+      "loss": 5.1797,
+      "step": 1042
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.743705067128627e-06,
+      "loss": 4.7305,
+      "step": 1043
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.741541792983976e-06,
+      "loss": 5.0352,
+      "step": 1044
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.739378518839326e-06,
+      "loss": 4.5898,
+      "step": 1045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.737215244694675e-06,
+      "loss": 3.877,
+      "step": 1046
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.735051970550022e-06,
+      "loss": 5.4141,
+      "step": 1047
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.73288869640537e-06,
+      "loss": 4.9062,
+      "step": 1048
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.73072542226072e-06,
+      "loss": 5.1133,
+      "step": 1049
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.728562148116068e-06,
+      "loss": 4.6172,
+      "step": 1050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.726398873971416e-06,
+      "loss": 4.8555,
+      "step": 1051
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.724235599826767e-06,
+      "loss": 4.332,
+      "step": 1052
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.722072325682114e-06,
+      "loss": 4.9766,
+      "step": 1053
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.719909051537462e-06,
+      "loss": 5.1602,
+      "step": 1054
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.717745777392811e-06,
+      "loss": 4.4492,
+      "step": 1055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.71558250324816e-06,
+      "loss": 5.0352,
+      "step": 1056
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.71341922910351e-06,
+      "loss": 4.2305,
+      "step": 1057
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.711255954958859e-06,
+      "loss": 4.6562,
+      "step": 1058
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.709092680814205e-06,
+      "loss": 3.3281,
+      "step": 1059
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.706929406669554e-06,
+      "loss": 5.0078,
+      "step": 1060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.704766132524903e-06,
+      "loss": 4.6602,
+      "step": 1061
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.702602858380251e-06,
+      "loss": 4.6562,
+      "step": 1062
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.7004395842356e-06,
+      "loss": 4.7109,
+      "step": 1063
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.69827631009095e-06,
+      "loss": 4.9531,
+      "step": 1064
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.696113035946296e-06,
+      "loss": 4.9414,
+      "step": 1065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.693949761801646e-06,
+      "loss": 4.3398,
+      "step": 1066
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.691786487656995e-06,
+      "loss": 5.1094,
+      "step": 1067
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.689623213512343e-06,
+      "loss": 5.0,
+      "step": 1068
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.687459939367694e-06,
+      "loss": 4.9531,
+      "step": 1069
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.685296665223042e-06,
+      "loss": 4.9102,
+      "step": 1070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.68313339107839e-06,
+      "loss": 4.8164,
+      "step": 1071
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.68097011693374e-06,
+      "loss": 5.0547,
+      "step": 1072
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.678806842789087e-06,
+      "loss": 5.1875,
+      "step": 1073
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.676643568644435e-06,
+      "loss": 4.3359,
+      "step": 1074
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.674480294499784e-06,
+      "loss": 3.9531,
+      "step": 1075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.672317020355133e-06,
+      "loss": 5.2852,
+      "step": 1076
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.670153746210481e-06,
+      "loss": 4.8867,
+      "step": 1077
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.66799047206583e-06,
+      "loss": 5.2773,
+      "step": 1078
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.665827197921179e-06,
+      "loss": 4.4141,
+      "step": 1079
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.663663923776527e-06,
+      "loss": 5.0273,
+      "step": 1080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.661500649631876e-06,
+      "loss": 5.4062,
+      "step": 1081
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.659337375487225e-06,
+      "loss": 4.7539,
+      "step": 1082
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.657174101342573e-06,
+      "loss": 5.0781,
+      "step": 1083
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.65501082719792e-06,
+      "loss": 5.0078,
+      "step": 1084
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.65284755305327e-06,
+      "loss": 4.1562,
+      "step": 1085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.65068427890862e-06,
+      "loss": 4.6016,
+      "step": 1086
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.648521004763966e-06,
+      "loss": 3.9824,
+      "step": 1087
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.646357730619317e-06,
+      "loss": 4.6289,
+      "step": 1088
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.644194456474665e-06,
+      "loss": 5.1562,
+      "step": 1089
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.642031182330014e-06,
+      "loss": 4.8984,
+      "step": 1090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.639867908185363e-06,
+      "loss": 4.375,
+      "step": 1091
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.637704634040711e-06,
+      "loss": 5.5625,
+      "step": 1092
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.63554135989606e-06,
+      "loss": 3.5898,
+      "step": 1093
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.633378085751408e-06,
+      "loss": 5.168,
+      "step": 1094
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.631214811606757e-06,
+      "loss": 5.3047,
+      "step": 1095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.629051537462106e-06,
+      "loss": 5.1328,
+      "step": 1096
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.6268882633174545e-06,
+      "loss": 5.0781,
+      "step": 1097
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.624724989172802e-06,
+      "loss": 4.1992,
+      "step": 1098
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.622561715028151e-06,
+      "loss": 5.1992,
+      "step": 1099
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.6203984408835e-06,
+      "loss": 5.0312,
+      "step": 1100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.618235166738849e-06,
+      "loss": 4.8359,
+      "step": 1101
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.616071892594198e-06,
+      "loss": 4.3867,
+      "step": 1102
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.613908618449547e-06,
+      "loss": 3.7598,
+      "step": 1103
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.611745344304894e-06,
+      "loss": 5.2422,
+      "step": 1104
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.609582070160243e-06,
+      "loss": 4.3438,
+      "step": 1105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.607418796015592e-06,
+      "loss": 4.9727,
+      "step": 1106
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.60525552187094e-06,
+      "loss": 4.9688,
+      "step": 1107
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.603092247726289e-06,
+      "loss": 5.1914,
+      "step": 1108
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.600928973581636e-06,
+      "loss": 3.9668,
+      "step": 1109
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.598765699436985e-06,
+      "loss": 4.5977,
+      "step": 1110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.596602425292334e-06,
+      "loss": 4.5039,
+      "step": 1111
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.5944391511476835e-06,
+      "loss": 5.3281,
+      "step": 1112
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.592275877003032e-06,
+      "loss": 5.043,
+      "step": 1113
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.590112602858382e-06,
+      "loss": 4.9961,
+      "step": 1114
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.5879493287137295e-06,
+      "loss": 4.8555,
+      "step": 1115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.585786054569079e-06,
+      "loss": 4.7383,
+      "step": 1116
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.583622780424427e-06,
+      "loss": 5.0234,
+      "step": 1117
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.581459506279775e-06,
+      "loss": 4.8438,
+      "step": 1118
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.579296232135124e-06,
+      "loss": 4.7305,
+      "step": 1119
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.577132957990473e-06,
+      "loss": 5.1484,
+      "step": 1120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.5749696838458206e-06,
+      "loss": 5.0625,
+      "step": 1121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.57280640970117e-06,
+      "loss": 4.3359,
+      "step": 1122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.570643135556518e-06,
+      "loss": 4.0039,
+      "step": 1123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.568479861411867e-06,
+      "loss": 4.1992,
+      "step": 1124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.566316587267216e-06,
+      "loss": 4.9531,
+      "step": 1125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.5641533131225655e-06,
+      "loss": 4.9766,
+      "step": 1126
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.561990038977913e-06,
+      "loss": 4.5078,
+      "step": 1127
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.559826764833262e-06,
+      "loss": 3.8027,
+      "step": 1128
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.5576634906886115e-06,
+      "loss": 4.0977,
+      "step": 1129
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.5555002165439585e-06,
+      "loss": 4.375,
+      "step": 1130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.553336942399307e-06,
+      "loss": 3.9414,
+      "step": 1131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.551173668254657e-06,
+      "loss": 4.7852,
+      "step": 1132
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.5490103941100045e-06,
+      "loss": 4.4414,
+      "step": 1133
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.546847119965354e-06,
+      "loss": 4.9766,
+      "step": 1134
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.544683845820703e-06,
+      "loss": 4.8633,
+      "step": 1135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.542520571676052e-06,
+      "loss": 4.75,
+      "step": 1136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.5403572975314e-06,
+      "loss": 5.3125,
+      "step": 1137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.538194023386748e-06,
+      "loss": 4.4922,
+      "step": 1138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.536030749242097e-06,
+      "loss": 4.8359,
+      "step": 1139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.533867475097446e-06,
+      "loss": 4.7188,
+      "step": 1140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.531704200952794e-06,
+      "loss": 4.4453,
+      "step": 1141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.529540926808142e-06,
+      "loss": 5.1289,
+      "step": 1142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.52737765266349e-06,
+      "loss": 5.1562,
+      "step": 1143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.525214378518839e-06,
+      "loss": 4.7461,
+      "step": 1144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.523051104374188e-06,
+      "loss": 4.8594,
+      "step": 1145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.520887830229537e-06,
+      "loss": 5.1406,
+      "step": 1146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.5187245560848865e-06,
+      "loss": 4.4375,
+      "step": 1147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.516561281940236e-06,
+      "loss": 4.793,
+      "step": 1148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.514398007795582e-06,
+      "loss": 3.9336,
+      "step": 1149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.512234733650932e-06,
+      "loss": 4.4219,
+      "step": 1150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.51007145950628e-06,
+      "loss": 4.5078,
+      "step": 1151
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.507908185361629e-06,
+      "loss": 5.1953,
+      "step": 1152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.505744911216978e-06,
+      "loss": 5.2578,
+      "step": 1153
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.503581637072327e-06,
+      "loss": 4.7891,
+      "step": 1154
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.501418362927674e-06,
+      "loss": 5.0547,
+      "step": 1155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.499255088783023e-06,
+      "loss": 5.0195,
+      "step": 1156
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.497091814638372e-06,
+      "loss": 4.9727,
+      "step": 1157
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.494928540493721e-06,
+      "loss": 4.7188,
+      "step": 1158
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.49276526634907e-06,
+      "loss": 5.3398,
+      "step": 1159
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.490601992204419e-06,
+      "loss": 5.0312,
+      "step": 1160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.488438718059768e-06,
+      "loss": 4.4453,
+      "step": 1161
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.4862754439151155e-06,
+      "loss": 5.0859,
+      "step": 1162
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.484112169770464e-06,
+      "loss": 5.0352,
+      "step": 1163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.481948895625812e-06,
+      "loss": 4.2539,
+      "step": 1164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.4797856214811615e-06,
+      "loss": 4.1562,
+      "step": 1165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.477622347336511e-06,
+      "loss": 5.1641,
+      "step": 1166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.475459073191859e-06,
+      "loss": 4.7344,
+      "step": 1167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.473295799047207e-06,
+      "loss": 4.2695,
+      "step": 1168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.471132524902555e-06,
+      "loss": 4.6406,
+      "step": 1169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.468969250757905e-06,
+      "loss": 5.1172,
+      "step": 1170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.466805976613253e-06,
+      "loss": 5.3516,
+      "step": 1171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.464642702468603e-06,
+      "loss": 3.6094,
+      "step": 1172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.462479428323951e-06,
+      "loss": 4.5312,
+      "step": 1173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.460316154179298e-06,
+      "loss": 4.5117,
+      "step": 1174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.458152880034647e-06,
+      "loss": 5.0742,
+      "step": 1175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.455989605889996e-06,
+      "loss": 5.0391,
+      "step": 1176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.453826331745344e-06,
+      "loss": 4.6992,
+      "step": 1177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.451663057600693e-06,
+      "loss": 4.8047,
+      "step": 1178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.449499783456043e-06,
+      "loss": 5.0391,
+      "step": 1179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.447336509311391e-06,
+      "loss": 5.1016,
+      "step": 1180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.445173235166739e-06,
+      "loss": 5.1094,
+      "step": 1181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.443009961022089e-06,
+      "loss": 3.3613,
+      "step": 1182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.4408466868774365e-06,
+      "loss": 4.8008,
+      "step": 1183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.438683412732786e-06,
+      "loss": 5.4219,
+      "step": 1184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.436520138588135e-06,
+      "loss": 5.1641,
+      "step": 1185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.4343568644434825e-06,
+      "loss": 5.0625,
+      "step": 1186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.432193590298831e-06,
+      "loss": 4.4688,
+      "step": 1187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.43003031615418e-06,
+      "loss": 4.9805,
+      "step": 1188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.427867042009528e-06,
+      "loss": 5.2109,
+      "step": 1189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.425703767864877e-06,
+      "loss": 4.1719,
+      "step": 1190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.423540493720226e-06,
+      "loss": 4.2109,
+      "step": 1191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.421377219575575e-06,
+      "loss": 5.2383,
+      "step": 1192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.419213945430924e-06,
+      "loss": 5.0977,
+      "step": 1193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.417050671286273e-06,
+      "loss": 4.1289,
+      "step": 1194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.41488739714162e-06,
+      "loss": 4.5469,
+      "step": 1195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.412724122996969e-06,
+      "loss": 3.9688,
+      "step": 1196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.4105608488523185e-06,
+      "loss": 5.2109,
+      "step": 1197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.408397574707666e-06,
+      "loss": 4.9258,
+      "step": 1198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.406234300563016e-06,
+      "loss": 5.0234,
+      "step": 1199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.4040710264183645e-06,
+      "loss": 5.2734,
+      "step": 1200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.4019077522737115e-06,
+      "loss": 4.3906,
+      "step": 1201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.399744478129061e-06,
+      "loss": 4.7852,
+      "step": 1202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.39758120398441e-06,
+      "loss": 4.125,
+      "step": 1203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.395417929839759e-06,
+      "loss": 4.9453,
+      "step": 1204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.393254655695107e-06,
+      "loss": 5.1133,
+      "step": 1205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.391091381550457e-06,
+      "loss": 5.125,
+      "step": 1206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.388928107405804e-06,
+      "loss": 4.7852,
+      "step": 1207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.386764833261152e-06,
+      "loss": 5.0664,
+      "step": 1208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.384601559116501e-06,
+      "loss": 4.5977,
+      "step": 1209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.38243828497185e-06,
+      "loss": 5.0312,
+      "step": 1210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.380275010827198e-06,
+      "loss": 4.9258,
+      "step": 1211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.378111736682546e-06,
+      "loss": 4.8711,
+      "step": 1212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.375948462537895e-06,
+      "loss": 4.7852,
+      "step": 1213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.373785188393244e-06,
+      "loss": 5.0156,
+      "step": 1214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.3716219142485935e-06,
+      "loss": 5.2852,
+      "step": 1215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.369458640103942e-06,
+      "loss": 4.7539,
+      "step": 1216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.367295365959291e-06,
+      "loss": 4.8633,
+      "step": 1217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.3651320918146395e-06,
+      "loss": 5.3203,
+      "step": 1218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.362968817669989e-06,
+      "loss": 5.4336,
+      "step": 1219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.360805543525336e-06,
+      "loss": 4.4883,
+      "step": 1220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.358642269380685e-06,
+      "loss": 4.9141,
+      "step": 1221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.356478995236034e-06,
+      "loss": 5.4297,
+      "step": 1222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.354315721091382e-06,
+      "loss": 4.5703,
+      "step": 1223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.352152446946731e-06,
+      "loss": 5.2812,
+      "step": 1224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.34998917280208e-06,
+      "loss": 4.4648,
+      "step": 1225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.347825898657428e-06,
+      "loss": 4.8945,
+      "step": 1226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.345662624512777e-06,
+      "loss": 4.9414,
+      "step": 1227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.343499350368126e-06,
+      "loss": 5.2188,
+      "step": 1228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.341336076223474e-06,
+      "loss": 4.7852,
+      "step": 1229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.339172802078823e-06,
+      "loss": 5.0117,
+      "step": 1230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.337009527934172e-06,
+      "loss": 4.4961,
+      "step": 1231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.334846253789519e-06,
+      "loss": 4.6367,
+      "step": 1232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.332682979644868e-06,
+      "loss": 4.7383,
+      "step": 1233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.330519705500217e-06,
+      "loss": 5.2344,
+      "step": 1234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.328356431355566e-06,
+      "loss": 4.5859,
+      "step": 1235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.3261931572109145e-06,
+      "loss": 5.0117,
+      "step": 1236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.324029883066264e-06,
+      "loss": 4.7695,
+      "step": 1237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.321866608921613e-06,
+      "loss": 5.2617,
+      "step": 1238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.31970333477696e-06,
+      "loss": 4.5039,
+      "step": 1239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.317540060632309e-06,
+      "loss": 3.9297,
+      "step": 1240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.315376786487658e-06,
+      "loss": 4.8203,
+      "step": 1241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.313213512343006e-06,
+      "loss": 4.4688,
+      "step": 1242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.311050238198355e-06,
+      "loss": 4.9219,
+      "step": 1243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.308886964053704e-06,
+      "loss": 4.2188,
+      "step": 1244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.306723689909051e-06,
+      "loss": 3.9902,
+      "step": 1245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.3045604157644e-06,
+      "loss": 4.6641,
+      "step": 1246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.302397141619749e-06,
+      "loss": 4.9375,
+      "step": 1247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.300233867475098e-06,
+      "loss": 4.6406,
+      "step": 1248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.298070593330447e-06,
+      "loss": 5.1289,
+      "step": 1249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.2959073191857965e-06,
+      "loss": 5.1172,
+      "step": 1250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.2937440450411435e-06,
+      "loss": 4.6094,
+      "step": 1251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.291580770896492e-06,
+      "loss": 4.5195,
+      "step": 1252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.289417496751842e-06,
+      "loss": 5.1133,
+      "step": 1253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.2872542226071895e-06,
+      "loss": 4.9961,
+      "step": 1254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.285090948462539e-06,
+      "loss": 5.3867,
+      "step": 1255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.282927674317888e-06,
+      "loss": 5.25,
+      "step": 1256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.280764400173235e-06,
+      "loss": 5.2422,
+      "step": 1257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.278601126028584e-06,
+      "loss": 5.0117,
+      "step": 1258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.276437851883933e-06,
+      "loss": 5.1523,
+      "step": 1259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.274274577739282e-06,
+      "loss": 4.2578,
+      "step": 1260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.272111303594631e-06,
+      "loss": 4.4297,
+      "step": 1261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.2699480294499804e-06,
+      "loss": 5.0156,
+      "step": 1262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.267784755305328e-06,
+      "loss": 3.9902,
+      "step": 1263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.265621481160678e-06,
+      "loss": 5.2031,
+      "step": 1264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.2634582070160256e-06,
+      "loss": 5.1562,
+      "step": 1265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.261294932871373e-06,
+      "loss": 4.7812,
+      "step": 1266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.259131658726721e-06,
+      "loss": 4.9219,
+      "step": 1267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.2569683845820715e-06,
+      "loss": 4.7383,
+      "step": 1268
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.254805110437419e-06,
+      "loss": 5.0469,
+      "step": 1269
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.252641836292766e-06,
+      "loss": 5.043,
+      "step": 1270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.250478562148117e-06,
+      "loss": 5.1875,
+      "step": 1271
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.248315288003465e-06,
+      "loss": 5.0625,
+      "step": 1272
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.246152013858814e-06,
+      "loss": 5.1055,
+      "step": 1273
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.243988739714163e-06,
+      "loss": 4.5195,
+      "step": 1274
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.241825465569512e-06,
+      "loss": 4.4414,
+      "step": 1275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.239662191424859e-06,
+      "loss": 4.5664,
+      "step": 1276
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.237498917280208e-06,
+      "loss": 4.7305,
+      "step": 1277
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.235335643135557e-06,
+      "loss": 5.2891,
+      "step": 1278
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.233172368990905e-06,
+      "loss": 4.625,
+      "step": 1279
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.231009094846254e-06,
+      "loss": 4.25,
+      "step": 1280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.228845820701603e-06,
+      "loss": 4.4141,
+      "step": 1281
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.226682546556953e-06,
+      "loss": 4.0586,
+      "step": 1282
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.224519272412301e-06,
+      "loss": 4.9219,
+      "step": 1283
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.222355998267649e-06,
+      "loss": 4.8984,
+      "step": 1284
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.220192724122997e-06,
+      "loss": 5.1602,
+      "step": 1285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.2180294499783465e-06,
+      "loss": 5.3281,
+      "step": 1286
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.215866175833696e-06,
+      "loss": 4.5078,
+      "step": 1287
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.213702901689044e-06,
+      "loss": 4.2305,
+      "step": 1288
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.2115396275443925e-06,
+      "loss": 4.7734,
+      "step": 1289
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.209376353399741e-06,
+      "loss": 5.3242,
+      "step": 1290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.207213079255089e-06,
+      "loss": 4.3711,
+      "step": 1291
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.205049805110438e-06,
+      "loss": 4.9961,
+      "step": 1292
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.202886530965787e-06,
+      "loss": 4.9688,
+      "step": 1293
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.200723256821136e-06,
+      "loss": 5.0703,
+      "step": 1294
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.198559982676485e-06,
+      "loss": 4.0078,
+      "step": 1295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.196396708531834e-06,
+      "loss": 4.8438,
+      "step": 1296
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.194233434387181e-06,
+      "loss": 5.2031,
+      "step": 1297
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.192070160242529e-06,
+      "loss": 5.2656,
+      "step": 1298
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.189906886097879e-06,
+      "loss": 4.9414,
+      "step": 1299
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.187743611953228e-06,
+      "loss": 4.6914,
+      "step": 1300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.185580337808574e-06,
+      "loss": 5.0859,
+      "step": 1301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.183417063663926e-06,
+      "loss": 4.8945,
+      "step": 1302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.181253789519273e-06,
+      "loss": 4.5781,
+      "step": 1303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.179090515374621e-06,
+      "loss": 5.0625,
+      "step": 1304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.176927241229969e-06,
+      "loss": 4.6953,
+      "step": 1305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.17476396708532e-06,
+      "loss": 4.6953,
+      "step": 1306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.1726006929406675e-06,
+      "loss": 4.8477,
+      "step": 1307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.170437418796017e-06,
+      "loss": 4.7695,
+      "step": 1308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.168274144651365e-06,
+      "loss": 4.668,
+      "step": 1309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.166110870506713e-06,
+      "loss": 4.6523,
+      "step": 1310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.163947596362062e-06,
+      "loss": 4.8477,
+      "step": 1311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.161784322217411e-06,
+      "loss": 5.0664,
+      "step": 1312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.1596210480727594e-06,
+      "loss": 5.1016,
+      "step": 1313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.157457773928108e-06,
+      "loss": 4.9766,
+      "step": 1314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.155294499783456e-06,
+      "loss": 4.2266,
+      "step": 1315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.1531312256388046e-06,
+      "loss": 4.9805,
+      "step": 1316
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.150967951494154e-06,
+      "loss": 3.8438,
+      "step": 1317
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.1488046773495036e-06,
+      "loss": 4.8398,
+      "step": 1318
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.146641403204851e-06,
+      "loss": 5.1211,
+      "step": 1319
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.144478129060201e-06,
+      "loss": 3.9531,
+      "step": 1320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.1423148549155495e-06,
+      "loss": 4.9297,
+      "step": 1321
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.1401515807708965e-06,
+      "loss": 4.9297,
+      "step": 1322
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.137988306626246e-06,
+      "loss": 5.0,
+      "step": 1323
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.135825032481595e-06,
+      "loss": 4.6328,
+      "step": 1324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.1336617583369425e-06,
+      "loss": 4.1133,
+      "step": 1325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.131498484192292e-06,
+      "loss": 3.877,
+      "step": 1326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.129335210047641e-06,
+      "loss": 4.8711,
+      "step": 1327
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.127171935902988e-06,
+      "loss": 4.5,
+      "step": 1328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.125008661758338e-06,
+      "loss": 5.2266,
+      "step": 1329
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.1228453876136875e-06,
+      "loss": 4.8711,
+      "step": 1330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.120682113469035e-06,
+      "loss": 4.9492,
+      "step": 1331
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.118518839324383e-06,
+      "loss": 4.8828,
+      "step": 1332
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.1163555651797334e-06,
+      "loss": 4.0859,
+      "step": 1333
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.114192291035081e-06,
+      "loss": 4.7812,
+      "step": 1334
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.112029016890428e-06,
+      "loss": 4.3594,
+      "step": 1335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.1098657427457786e-06,
+      "loss": 4.7344,
+      "step": 1336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.107702468601126e-06,
+      "loss": 4.7461,
+      "step": 1337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.105539194456474e-06,
+      "loss": 5.1289,
+      "step": 1338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.103375920311824e-06,
+      "loss": 5.3438,
+      "step": 1339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.1012126461671715e-06,
+      "loss": 4.7734,
+      "step": 1340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.099049372022521e-06,
+      "loss": 5.1758,
+      "step": 1341
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.09688609787787e-06,
+      "loss": 3.8125,
+      "step": 1342
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.094722823733219e-06,
+      "loss": 5.1992,
+      "step": 1343
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.092559549588567e-06,
+      "loss": 5.1133,
+      "step": 1344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.090396275443916e-06,
+      "loss": 4.0703,
+      "step": 1345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.088233001299265e-06,
+      "loss": 5.0234,
+      "step": 1346
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.086069727154612e-06,
+      "loss": 4.1562,
+      "step": 1347
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.083906453009961e-06,
+      "loss": 4.5938,
+      "step": 1348
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.08174317886531e-06,
+      "loss": 5.125,
+      "step": 1349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.079579904720658e-06,
+      "loss": 4.6016,
+      "step": 1350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.0774166305760076e-06,
+      "loss": 5.3125,
+      "step": 1351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.075253356431357e-06,
+      "loss": 3.9082,
+      "step": 1352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.073090082286706e-06,
+      "loss": 5.0234,
+      "step": 1353
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.0709268081420535e-06,
+      "loss": 4.7656,
+      "step": 1354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.068763533997402e-06,
+      "loss": 4.7031,
+      "step": 1355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.066600259852751e-06,
+      "loss": 4.6367,
+      "step": 1356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.0644369857080995e-06,
+      "loss": 4.9023,
+      "step": 1357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.062273711563449e-06,
+      "loss": 4.9961,
+      "step": 1358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.060110437418797e-06,
+      "loss": 5.1328,
+      "step": 1359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.057947163274145e-06,
+      "loss": 4.5156,
+      "step": 1360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.055783889129494e-06,
+      "loss": 4.8828,
+      "step": 1361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.053620614984842e-06,
+      "loss": 4.5352,
+      "step": 1362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.051457340840191e-06,
+      "loss": 4.4531,
+      "step": 1363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.049294066695541e-06,
+      "loss": 5.4258,
+      "step": 1364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.047130792550889e-06,
+      "loss": 5.0977,
+      "step": 1365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.044967518406236e-06,
+      "loss": 4.8633,
+      "step": 1366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.042804244261588e-06,
+      "loss": 5.0312,
+      "step": 1367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.040640970116934e-06,
+      "loss": 5.0078,
+      "step": 1368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.0384776959722826e-06,
+      "loss": 4.7109,
+      "step": 1369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.036314421827631e-06,
+      "loss": 4.8438,
+      "step": 1370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.034151147682981e-06,
+      "loss": 4.8438,
+      "step": 1371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.031987873538328e-06,
+      "loss": 4.6328,
+      "step": 1372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.029824599393676e-06,
+      "loss": 4.8672,
+      "step": 1373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.027661325249026e-06,
+      "loss": 3.9766,
+      "step": 1374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.0254980511043745e-06,
+      "loss": 4.9102,
+      "step": 1375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.023334776959724e-06,
+      "loss": 4.7773,
+      "step": 1376
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.021171502815073e-06,
+      "loss": 4.8477,
+      "step": 1377
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.0190082286704205e-06,
+      "loss": 5.0781,
+      "step": 1378
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.016844954525769e-06,
+      "loss": 4.8672,
+      "step": 1379
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.014681680381118e-06,
+      "loss": 4.8477,
+      "step": 1380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.012518406236466e-06,
+      "loss": 4.9141,
+      "step": 1381
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.010355132091815e-06,
+      "loss": 4.8438,
+      "step": 1382
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.008191857947164e-06,
+      "loss": 5.1406,
+      "step": 1383
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.006028583802512e-06,
+      "loss": 5.0352,
+      "step": 1384
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.00386530965786e-06,
+      "loss": 5.0352,
+      "step": 1385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.001702035513209e-06,
+      "loss": 5.3516,
+      "step": 1386
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.999538761368558e-06,
+      "loss": 4.9375,
+      "step": 1387
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.997375487223907e-06,
+      "loss": 4.8086,
+      "step": 1388
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.9952122130792566e-06,
+      "loss": 4.7266,
+      "step": 1389
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.993048938934604e-06,
+      "loss": 4.6797,
+      "step": 1390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.990885664789954e-06,
+      "loss": 5.0703,
+      "step": 1391
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.9887223906453025e-06,
+      "loss": 3.6777,
+      "step": 1392
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.9865591165006495e-06,
+      "loss": 5.0781,
+      "step": 1393
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.984395842355999e-06,
+      "loss": 4.7812,
+      "step": 1394
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.982232568211348e-06,
+      "loss": 5.1445,
+      "step": 1395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.9800692940666955e-06,
+      "loss": 4.043,
+      "step": 1396
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.977906019922045e-06,
+      "loss": 5.3555,
+      "step": 1397
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.9757427457773945e-06,
+      "loss": 5.1094,
+      "step": 1398
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.973579471632742e-06,
+      "loss": 4.7617,
+      "step": 1399
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.97141619748809e-06,
+      "loss": 5.1484,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.9692529233434404e-06,
+      "loss": 4.7734,
+      "step": 1401
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.967089649198788e-06,
+      "loss": 4.6875,
+      "step": 1402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.964926375054136e-06,
+      "loss": 4.3047,
+      "step": 1403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.962763100909484e-06,
+      "loss": 5.1172,
+      "step": 1404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.960599826764833e-06,
+      "loss": 4.2734,
+      "step": 1405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.958436552620181e-06,
+      "loss": 4.5039,
+      "step": 1406
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.956273278475531e-06,
+      "loss": 4.8828,
+      "step": 1407
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.954110004330879e-06,
+      "loss": 3.8809,
+      "step": 1408
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.951946730186229e-06,
+      "loss": 4.7695,
+      "step": 1409
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.9497834560415775e-06,
+      "loss": 4.5273,
+      "step": 1410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.947620181896927e-06,
+      "loss": 4.8828,
+      "step": 1411
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.945456907752274e-06,
+      "loss": 4.9023,
+      "step": 1412
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.943293633607623e-06,
+      "loss": 4.3477,
+      "step": 1413
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.941130359462972e-06,
+      "loss": 4.8438,
+      "step": 1414
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.93896708531832e-06,
+      "loss": 4.6406,
+      "step": 1415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.9368038111736695e-06,
+      "loss": 4.3164,
+      "step": 1416
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.934640537029018e-06,
+      "loss": 4.7695,
+      "step": 1417
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.932477262884365e-06,
+      "loss": 4.9375,
+      "step": 1418
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.930313988739715e-06,
+      "loss": 4.8789,
+      "step": 1419
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.928150714595063e-06,
+      "loss": 4.9844,
+      "step": 1420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.925987440450413e-06,
+      "loss": 4.9141,
+      "step": 1421
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.923824166305761e-06,
+      "loss": 4.6797,
+      "step": 1422
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.921660892161111e-06,
+      "loss": 3.8164,
+      "step": 1423
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.919497618016458e-06,
+      "loss": 3.6621,
+      "step": 1424
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.9173343438718065e-06,
+      "loss": 4.625,
+      "step": 1425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.915171069727156e-06,
+      "loss": 3.6191,
+      "step": 1426
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.913007795582504e-06,
+      "loss": 4.0,
+      "step": 1427
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.910844521437852e-06,
+      "loss": 5.3438,
+      "step": 1428
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.908681247293202e-06,
+      "loss": 4.5898,
+      "step": 1429
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.906517973148549e-06,
+      "loss": 4.5781,
+      "step": 1430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.904354699003898e-06,
+      "loss": 5.0547,
+      "step": 1431
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.902191424859247e-06,
+      "loss": 5.293,
+      "step": 1432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.900028150714596e-06,
+      "loss": 3.9355,
+      "step": 1433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.8978648765699445e-06,
+      "loss": 4.8281,
+      "step": 1434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.895701602425293e-06,
+      "loss": 5.125,
+      "step": 1435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.893538328280643e-06,
+      "loss": 4.7031,
+      "step": 1436
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.89137505413599e-06,
+      "loss": 5.1406,
+      "step": 1437
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.889211779991338e-06,
+      "loss": 4.9141,
+      "step": 1438
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.887048505846688e-06,
+      "loss": 3.6426,
+      "step": 1439
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.8848852317020356e-06,
+      "loss": 3.8613,
+      "step": 1440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.882721957557384e-06,
+      "loss": 4.9766,
+      "step": 1441
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.880558683412734e-06,
+      "loss": 4.8203,
+      "step": 1442
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.8783954092680815e-06,
+      "loss": 5.1328,
+      "step": 1443
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.876232135123431e-06,
+      "loss": 4.6289,
+      "step": 1444
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.87406886097878e-06,
+      "loss": 4.9023,
+      "step": 1445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.8719055868341275e-06,
+      "loss": 4.7109,
+      "step": 1446
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.869742312689477e-06,
+      "loss": 4.9609,
+      "step": 1447
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.867579038544826e-06,
+      "loss": 4.9023,
+      "step": 1448
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.865415764400173e-06,
+      "loss": 5.0234,
+      "step": 1449
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.863252490255522e-06,
+      "loss": 5.0078,
+      "step": 1450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.861089216110871e-06,
+      "loss": 5.0547,
+      "step": 1451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.8589259419662194e-06,
+      "loss": 4.75,
+      "step": 1452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.856762667821568e-06,
+      "loss": 4.2148,
+      "step": 1453
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.854599393676918e-06,
+      "loss": 4.8242,
+      "step": 1454
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.852436119532266e-06,
+      "loss": 4.9453,
+      "step": 1455
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.850272845387616e-06,
+      "loss": 5.0195,
+      "step": 1456
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.848109571242964e-06,
+      "loss": 5.1719,
+      "step": 1457
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.845946297098311e-06,
+      "loss": 4.9219,
+      "step": 1458
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.843783022953661e-06,
+      "loss": 4.9531,
+      "step": 1459
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.8416197488090096e-06,
+      "loss": 4.5586,
+      "step": 1460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.839456474664357e-06,
+      "loss": 4.4141,
+      "step": 1461
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.837293200519704e-06,
+      "loss": 3.8984,
+      "step": 1462
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.835129926375055e-06,
+      "loss": 4.9727,
+      "step": 1463
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.8329666522304025e-06,
+      "loss": 4.1406,
+      "step": 1464
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.830803378085752e-06,
+      "loss": 4.3945,
+      "step": 1465
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.828640103941101e-06,
+      "loss": 4.0664,
+      "step": 1466
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.82647682979645e-06,
+      "loss": 5.1289,
+      "step": 1467
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.824313555651797e-06,
+      "loss": 4.9492,
+      "step": 1468
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.822150281507146e-06,
+      "loss": 4.5781,
+      "step": 1469
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.819987007362495e-06,
+      "loss": 4.9375,
+      "step": 1470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.817823733217843e-06,
+      "loss": 4.0703,
+      "step": 1471
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.815660459073193e-06,
+      "loss": 4.75,
+      "step": 1472
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.813497184928541e-06,
+      "loss": 4.8789,
+      "step": 1473
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.811333910783888e-06,
+      "loss": 5.0156,
+      "step": 1474
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.809170636639238e-06,
+      "loss": 5.0352,
+      "step": 1475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.807007362494586e-06,
+      "loss": 4.4258,
+      "step": 1476
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.804844088349936e-06,
+      "loss": 5.4297,
+      "step": 1477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.8026808142052845e-06,
+      "loss": 5.1641,
+      "step": 1478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.800517540060634e-06,
+      "loss": 4.6641,
+      "step": 1479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.798354265915982e-06,
+      "loss": 5.1133,
+      "step": 1480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.7961909917713305e-06,
+      "loss": 3.8691,
+      "step": 1481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.794027717626679e-06,
+      "loss": 4.6055,
+      "step": 1482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.791864443482027e-06,
+      "loss": 5.3281,
+      "step": 1483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.789701169337376e-06,
+      "loss": 3.9395,
+      "step": 1484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.787537895192725e-06,
+      "loss": 5.1172,
+      "step": 1485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.785374621048073e-06,
+      "loss": 4.6914,
+      "step": 1486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.7832113469034225e-06,
+      "loss": 5.3516,
+      "step": 1487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.78104807275877e-06,
+      "loss": 5.0156,
+      "step": 1488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.778884798614119e-06,
+      "loss": 5.0469,
+      "step": 1489
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.7767215244694684e-06,
+      "loss": 4.6875,
+      "step": 1490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.774558250324817e-06,
+      "loss": 4.7109,
+      "step": 1491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.772394976180166e-06,
+      "loss": 4.9414,
+      "step": 1492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.770231702035513e-06,
+      "loss": 4.6016,
+      "step": 1493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.768068427890864e-06,
+      "loss": 3.9297,
+      "step": 1494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.765905153746211e-06,
+      "loss": 3.7754,
+      "step": 1495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.763741879601559e-06,
+      "loss": 4.1055,
+      "step": 1496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.761578605456909e-06,
+      "loss": 3.3457,
+      "step": 1497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.759415331312257e-06,
+      "loss": 4.7695,
+      "step": 1498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.7572520571676055e-06,
+      "loss": 5.1016,
+      "step": 1499
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.755088783022955e-06,
+      "loss": 4.9688,
+      "step": 1500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.752925508878303e-06,
+      "loss": 5.2656,
+      "step": 1501
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.750762234733651e-06,
+      "loss": 4.2734,
+      "step": 1502
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.748598960589e-06,
+      "loss": 4.7539,
+      "step": 1503
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.746435686444349e-06,
+      "loss": 4.5742,
+      "step": 1504
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.7442724122996975e-06,
+      "loss": 5.2305,
+      "step": 1505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.742109138155046e-06,
+      "loss": 4.9727,
+      "step": 1506
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.739945864010394e-06,
+      "loss": 5.1406,
+      "step": 1507
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.7377825898657426e-06,
+      "loss": 4.8164,
+      "step": 1508
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.735619315721091e-06,
+      "loss": 4.8125,
+      "step": 1509
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.733456041576441e-06,
+      "loss": 4.625,
+      "step": 1510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.731292767431789e-06,
+      "loss": 5.0156,
+      "step": 1511
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.729129493287139e-06,
+      "loss": 4.9648,
+      "step": 1512
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.7269662191424876e-06,
+      "loss": 5.1328,
+      "step": 1513
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.7248029449978345e-06,
+      "loss": 4.9883,
+      "step": 1514
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.722639670853184e-06,
+      "loss": 5.0742,
+      "step": 1515
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.720476396708533e-06,
+      "loss": 5.0078,
+      "step": 1516
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.7183131225638805e-06,
+      "loss": 5.1016,
+      "step": 1517
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.71614984841923e-06,
+      "loss": 4.5703,
+      "step": 1518
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.7139865742745795e-06,
+      "loss": 4.5,
+      "step": 1519
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.711823300129926e-06,
+      "loss": 4.9023,
+      "step": 1520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.709660025985275e-06,
+      "loss": 4.8828,
+      "step": 1521
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.707496751840625e-06,
+      "loss": 4.1055,
+      "step": 1522
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.705333477695973e-06,
+      "loss": 3.7773,
+      "step": 1523
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.703170203551323e-06,
+      "loss": 4.7734,
+      "step": 1524
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.7010069294066714e-06,
+      "loss": 4.668,
+      "step": 1525
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.698843655262019e-06,
+      "loss": 4.9609,
+      "step": 1526
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.696680381117366e-06,
+      "loss": 4.793,
+      "step": 1527
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.6945171069727166e-06,
+      "loss": 4.9492,
+      "step": 1528
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.692353832828064e-06,
+      "loss": 4.582,
+      "step": 1529
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.690190558683412e-06,
+      "loss": 4.3945,
+      "step": 1530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.6880272845387625e-06,
+      "loss": 4.9688,
+      "step": 1531
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.6858640103941095e-06,
+      "loss": 4.0977,
+      "step": 1532
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.683700736249459e-06,
+      "loss": 4.8281,
+      "step": 1533
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.681537462104808e-06,
+      "loss": 5.0781,
+      "step": 1534
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.679374187960157e-06,
+      "loss": 4.8008,
+      "step": 1535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.677210913815505e-06,
+      "loss": 4.832,
+      "step": 1536
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.6750476396708545e-06,
+      "loss": 4.9141,
+      "step": 1537
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.672884365526203e-06,
+      "loss": 4.7734,
+      "step": 1538
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.67072109138155e-06,
+      "loss": 4.8945,
+      "step": 1539
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.6685578172369e-06,
+      "loss": 4.0391,
+      "step": 1540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.666394543092248e-06,
+      "loss": 4.3867,
+      "step": 1541
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.664231268947596e-06,
+      "loss": 4.9883,
+      "step": 1542
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.662067994802946e-06,
+      "loss": 4.6562,
+      "step": 1543
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.659904720658294e-06,
+      "loss": 4.4648,
+      "step": 1544
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.657741446513644e-06,
+      "loss": 3.8789,
+      "step": 1545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.6555781723689916e-06,
+      "loss": 4.6992,
+      "step": 1546
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.653414898224341e-06,
+      "loss": 4.3281,
+      "step": 1547
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.651251624079689e-06,
+      "loss": 5.2227,
+      "step": 1548
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.6490883499350375e-06,
+      "loss": 3.6582,
+      "step": 1549
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.646925075790387e-06,
+      "loss": 5.4375,
+      "step": 1550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.644761801645735e-06,
+      "loss": 4.6797,
+      "step": 1551
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.642598527501083e-06,
+      "loss": 4.9102,
+      "step": 1552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.640435253356432e-06,
+      "loss": 4.8594,
+      "step": 1553
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.63827197921178e-06,
+      "loss": 4.1914,
+      "step": 1554
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.6361087050671295e-06,
+      "loss": 4.5547,
+      "step": 1555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.633945430922478e-06,
+      "loss": 3.8457,
+      "step": 1556
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.631782156777828e-06,
+      "loss": 4.4805,
+      "step": 1557
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.629618882633175e-06,
+      "loss": 4.5703,
+      "step": 1558
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.627455608488526e-06,
+      "loss": 4.9375,
+      "step": 1559
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.625292334343873e-06,
+      "loss": 4.4141,
+      "step": 1560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.623129060199221e-06,
+      "loss": 4.4258,
+      "step": 1561
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.620965786054571e-06,
+      "loss": 4.8984,
+      "step": 1562
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.618802511909919e-06,
+      "loss": 4.4688,
+      "step": 1563
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.616639237765266e-06,
+      "loss": 5.1406,
+      "step": 1564
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.614475963620617e-06,
+      "loss": 4.7656,
+      "step": 1565
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.612312689475964e-06,
+      "loss": 5.2188,
+      "step": 1566
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.6101494153313125e-06,
+      "loss": 4.9766,
+      "step": 1567
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.607986141186662e-06,
+      "loss": 5.2578,
+      "step": 1568
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.605822867042011e-06,
+      "loss": 4.5977,
+      "step": 1569
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.603659592897359e-06,
+      "loss": 4.7852,
+      "step": 1570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.601496318752707e-06,
+      "loss": 5.0469,
+      "step": 1571
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.599333044608056e-06,
+      "loss": 5.2461,
+      "step": 1572
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.5971697704634045e-06,
+      "loss": 4.8711,
+      "step": 1573
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.595006496318753e-06,
+      "loss": 4.6016,
+      "step": 1574
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.592843222174103e-06,
+      "loss": 4.9688,
+      "step": 1575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.5906799480294504e-06,
+      "loss": 4.8555,
+      "step": 1576
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.588516673884798e-06,
+      "loss": 4.7539,
+      "step": 1577
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.586353399740148e-06,
+      "loss": 4.8164,
+      "step": 1578
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.584190125595496e-06,
+      "loss": 4.7695,
+      "step": 1579
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.582026851450846e-06,
+      "loss": 4.7539,
+      "step": 1580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.579863577306195e-06,
+      "loss": 4.7734,
+      "step": 1581
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.577700303161542e-06,
+      "loss": 4.4609,
+      "step": 1582
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.575537029016892e-06,
+      "loss": 4.7891,
+      "step": 1583
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.5733737548722406e-06,
+      "loss": 5.2109,
+      "step": 1584
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.5712104807275875e-06,
+      "loss": 4.9102,
+      "step": 1585
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.569047206582937e-06,
+      "loss": 4.9688,
+      "step": 1586
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.566883932438286e-06,
+      "loss": 5.2031,
+      "step": 1587
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.564720658293634e-06,
+      "loss": 4.7383,
+      "step": 1588
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.562557384148983e-06,
+      "loss": 4.4922,
+      "step": 1589
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.5603941100043325e-06,
+      "loss": 4.6875,
+      "step": 1590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.55823083585968e-06,
+      "loss": 4.5469,
+      "step": 1591
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.556067561715028e-06,
+      "loss": 4.6211,
+      "step": 1592
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.5539042875703785e-06,
+      "loss": 5.168,
+      "step": 1593
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.551741013425726e-06,
+      "loss": 4.6523,
+      "step": 1594
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.549577739281074e-06,
+      "loss": 4.9062,
+      "step": 1595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.5474144651364244e-06,
+      "loss": 4.9648,
+      "step": 1596
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.545251190991771e-06,
+      "loss": 4.8672,
+      "step": 1597
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.543087916847119e-06,
+      "loss": 3.5586,
+      "step": 1598
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.540924642702469e-06,
+      "loss": 4.9961,
+      "step": 1599
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.538761368557817e-06,
+      "loss": 4.5,
+      "step": 1600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.536598094413167e-06,
+      "loss": 5.0781,
+      "step": 1601
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.5344348202685155e-06,
+      "loss": 4.7031,
+      "step": 1602
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.532271546123865e-06,
+      "loss": 4.7461,
+      "step": 1603
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.530108271979212e-06,
+      "loss": 4.7695,
+      "step": 1604
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.527944997834561e-06,
+      "loss": 5.0469,
+      "step": 1605
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.52578172368991e-06,
+      "loss": 4.8945,
+      "step": 1606
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.523618449545258e-06,
+      "loss": 5.0117,
+      "step": 1607
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.5214551754006075e-06,
+      "loss": 5.0938,
+      "step": 1608
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.519291901255956e-06,
+      "loss": 4.6016,
+      "step": 1609
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.517128627111303e-06,
+      "loss": 4.9258,
+      "step": 1610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.514965352966653e-06,
+      "loss": 4.5938,
+      "step": 1611
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.512802078822001e-06,
+      "loss": 5.0586,
+      "step": 1612
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.510638804677351e-06,
+      "loss": 4.5,
+      "step": 1613
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.5084755305326994e-06,
+      "loss": 4.9297,
+      "step": 1614
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.506312256388049e-06,
+      "loss": 5.0625,
+      "step": 1615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.504148982243396e-06,
+      "loss": 4.8633,
+      "step": 1616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.5019857080987446e-06,
+      "loss": 4.6914,
+      "step": 1617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.499822433954094e-06,
+      "loss": 4.9688,
+      "step": 1618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.497659159809442e-06,
+      "loss": 5.0898,
+      "step": 1619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.4954958856647905e-06,
+      "loss": 5.1953,
+      "step": 1620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.49333261152014e-06,
+      "loss": 4.7578,
+      "step": 1621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.491169337375487e-06,
+      "loss": 4.5039,
+      "step": 1622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.489006063230835e-06,
+      "loss": 4.8789,
+      "step": 1623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.486842789086185e-06,
+      "loss": 4.0352,
+      "step": 1624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.484679514941535e-06,
+      "loss": 3.3906,
+      "step": 1625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.4825162407968825e-06,
+      "loss": 4.7734,
+      "step": 1626
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.480352966652233e-06,
+      "loss": 4.5117,
+      "step": 1627
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.478189692507581e-06,
+      "loss": 4.4219,
+      "step": 1628
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.476026418362928e-06,
+      "loss": 4.8828,
+      "step": 1629
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.473863144218278e-06,
+      "loss": 4.3359,
+      "step": 1630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.471699870073626e-06,
+      "loss": 4.9727,
+      "step": 1631
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.4695365959289736e-06,
+      "loss": 4.7969,
+      "step": 1632
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.467373321784322e-06,
+      "loss": 4.7617,
+      "step": 1633
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.465210047639672e-06,
+      "loss": 5.1211,
+      "step": 1634
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.463046773495019e-06,
+      "loss": 4.3555,
+      "step": 1635
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.460883499350369e-06,
+      "loss": 3.75,
+      "step": 1636
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.458720225205718e-06,
+      "loss": 4.7266,
+      "step": 1637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.456556951061066e-06,
+      "loss": 4.7266,
+      "step": 1638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.454393676916415e-06,
+      "loss": 4.5938,
+      "step": 1639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.4522304027717645e-06,
+      "loss": 4.2266,
+      "step": 1640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.450067128627111e-06,
+      "loss": 5.1836,
+      "step": 1641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.44790385448246e-06,
+      "loss": 4.3711,
+      "step": 1642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.44574058033781e-06,
+      "loss": 5.0039,
+      "step": 1643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.4435773061931575e-06,
+      "loss": 4.8672,
+      "step": 1644
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.441414032048506e-06,
+      "loss": 4.8242,
+      "step": 1645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.439250757903856e-06,
+      "loss": 4.5977,
+      "step": 1646
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.437087483759203e-06,
+      "loss": 4.6758,
+      "step": 1647
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.434924209614554e-06,
+      "loss": 4.2969,
+      "step": 1648
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.432760935469902e-06,
+      "loss": 5.0859,
+      "step": 1649
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.430597661325249e-06,
+      "loss": 4.2617,
+      "step": 1650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.428434387180599e-06,
+      "loss": 4.9805,
+      "step": 1651
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.4262711130359476e-06,
+      "loss": 4.9531,
+      "step": 1652
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.424107838891296e-06,
+      "loss": 4.2031,
+      "step": 1653
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.421944564746645e-06,
+      "loss": 5.1523,
+      "step": 1654
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.419781290601993e-06,
+      "loss": 4.9609,
+      "step": 1655
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.417618016457341e-06,
+      "loss": 4.6172,
+      "step": 1656
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.415454742312689e-06,
+      "loss": 5.0078,
+      "step": 1657
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.4132914681680395e-06,
+      "loss": 4.793,
+      "step": 1658
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.411128194023387e-06,
+      "loss": 4.3047,
+      "step": 1659
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.408964919878735e-06,
+      "loss": 4.6484,
+      "step": 1660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.406801645734086e-06,
+      "loss": 5.0039,
+      "step": 1661
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.404638371589433e-06,
+      "loss": 3.3848,
+      "step": 1662
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.402475097444781e-06,
+      "loss": 5.1055,
+      "step": 1663
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.400311823300131e-06,
+      "loss": 4.8477,
+      "step": 1664
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.398148549155479e-06,
+      "loss": 5.0898,
+      "step": 1665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.395985275010826e-06,
+      "loss": 4.6133,
+      "step": 1666
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.393822000866176e-06,
+      "loss": 5.2891,
+      "step": 1667
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.391658726721524e-06,
+      "loss": 5.0938,
+      "step": 1668
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.389495452576873e-06,
+      "loss": 4.793,
+      "step": 1669
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.387332178432222e-06,
+      "loss": 4.5273,
+      "step": 1670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.385168904287572e-06,
+      "loss": 4.8516,
+      "step": 1671
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.38300563014292e-06,
+      "loss": 5.0234,
+      "step": 1672
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.380842355998269e-06,
+      "loss": 4.6562,
+      "step": 1673
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.378679081853617e-06,
+      "loss": 4.8125,
+      "step": 1674
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.376515807708965e-06,
+      "loss": 4.8477,
+      "step": 1675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.3743525335643145e-06,
+      "loss": 4.6562,
+      "step": 1676
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.372189259419663e-06,
+      "loss": 4.2422,
+      "step": 1677
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.370025985275011e-06,
+      "loss": 4.8398,
+      "step": 1678
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.3678627111303605e-06,
+      "loss": 4.8164,
+      "step": 1679
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.365699436985708e-06,
+      "loss": 5.0352,
+      "step": 1680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.363536162841056e-06,
+      "loss": 4.9922,
+      "step": 1681
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.361372888696406e-06,
+      "loss": 3.6035,
+      "step": 1682
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.359209614551756e-06,
+      "loss": 4.9453,
+      "step": 1683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.357046340407104e-06,
+      "loss": 5.0664,
+      "step": 1684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.3548830662624524e-06,
+      "loss": 3.668,
+      "step": 1685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.352719792117802e-06,
+      "loss": 3.834,
+      "step": 1686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.350556517973149e-06,
+      "loss": 5.0781,
+      "step": 1687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.348393243828497e-06,
+      "loss": 5.125,
+      "step": 1688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.346229969683847e-06,
+      "loss": 4.3086,
+      "step": 1689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.344066695539195e-06,
+      "loss": 4.9883,
+      "step": 1690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.341903421394542e-06,
+      "loss": 4.6484,
+      "step": 1691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.339740147249893e-06,
+      "loss": 4.8672,
+      "step": 1692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.33757687310524e-06,
+      "loss": 5.0977,
+      "step": 1693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.3354135989605895e-06,
+      "loss": 3.834,
+      "step": 1694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.333250324815939e-06,
+      "loss": 4.1992,
+      "step": 1695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.331087050671288e-06,
+      "loss": 4.3633,
+      "step": 1696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.3289237765266355e-06,
+      "loss": 4.6289,
+      "step": 1697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.326760502381984e-06,
+      "loss": 5.1016,
+      "step": 1698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.324597228237333e-06,
+      "loss": 5.0703,
+      "step": 1699
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.322433954092681e-06,
+      "loss": 4.8945,
+      "step": 1700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.320270679948029e-06,
+      "loss": 4.2773,
+      "step": 1701
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.318107405803379e-06,
+      "loss": 4.8711,
+      "step": 1702
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.3159441316587266e-06,
+      "loss": 5.0625,
+      "step": 1703
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.313780857514076e-06,
+      "loss": 5.0625,
+      "step": 1704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.311617583369424e-06,
+      "loss": 4.1523,
+      "step": 1705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.3094543092247725e-06,
+      "loss": 4.9023,
+      "step": 1706
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.307291035080122e-06,
+      "loss": 4.9141,
+      "step": 1707
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.305127760935471e-06,
+      "loss": 3.6621,
+      "step": 1708
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.302964486790819e-06,
+      "loss": 4.9844,
+      "step": 1709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.300801212646168e-06,
+      "loss": 5.2656,
+      "step": 1710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.2986379385015175e-06,
+      "loss": 3.7227,
+      "step": 1711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.2964746643568645e-06,
+      "loss": 5.0586,
+      "step": 1712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.294311390212213e-06,
+      "loss": 4.8125,
+      "step": 1713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.292148116067563e-06,
+      "loss": 3.5547,
+      "step": 1714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.2899848419229105e-06,
+      "loss": 3.4336,
+      "step": 1715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.287821567778259e-06,
+      "loss": 3.832,
+      "step": 1716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.285658293633609e-06,
+      "loss": 5.0078,
+      "step": 1717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.283495019488957e-06,
+      "loss": 4.6484,
+      "step": 1718
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.281331745344306e-06,
+      "loss": 3.8574,
+      "step": 1719
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.279168471199655e-06,
+      "loss": 4.8789,
+      "step": 1720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.277005197055002e-06,
+      "loss": 4.6094,
+      "step": 1721
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.274841922910351e-06,
+      "loss": 4.4141,
+      "step": 1722
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.2726786487657006e-06,
+      "loss": 4.1836,
+      "step": 1723
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.2705153746210475e-06,
+      "loss": 4.1953,
+      "step": 1724
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.268352100476396e-06,
+      "loss": 5.125,
+      "step": 1725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.266188826331746e-06,
+      "loss": 4.6719,
+      "step": 1726
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.264025552187094e-06,
+      "loss": 5.0,
+      "step": 1727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.261862278042443e-06,
+      "loss": 5.3047,
+      "step": 1728
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.2596990038977925e-06,
+      "loss": 3.5977,
+      "step": 1729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.257535729753141e-06,
+      "loss": 5.1992,
+      "step": 1730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.255372455608488e-06,
+      "loss": 3.793,
+      "step": 1731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.253209181463838e-06,
+      "loss": 4.8594,
+      "step": 1732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.251045907319186e-06,
+      "loss": 4.7695,
+      "step": 1733
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.248882633174534e-06,
+      "loss": 4.8867,
+      "step": 1734
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.246719359029884e-06,
+      "loss": 4.6992,
+      "step": 1735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.244556084885232e-06,
+      "loss": 4.9531,
+      "step": 1736
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.242392810740579e-06,
+      "loss": 4.6328,
+      "step": 1737
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.240229536595929e-06,
+      "loss": 4.5547,
+      "step": 1738
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.238066262451277e-06,
+      "loss": 3.9648,
+      "step": 1739
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.235902988306627e-06,
+      "loss": 4.7734,
+      "step": 1740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.233739714161976e-06,
+      "loss": 4.4922,
+      "step": 1741
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.231576440017325e-06,
+      "loss": 4.5781,
+      "step": 1742
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.229413165872673e-06,
+      "loss": 4.6719,
+      "step": 1743
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.227249891728021e-06,
+      "loss": 3.9961,
+      "step": 1744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.22508661758337e-06,
+      "loss": 4.9414,
+      "step": 1745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.222923343438718e-06,
+      "loss": 4.6641,
+      "step": 1746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.2207600692940675e-06,
+      "loss": 4.6953,
+      "step": 1747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.218596795149416e-06,
+      "loss": 4.0312,
+      "step": 1748
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.216433521004764e-06,
+      "loss": 4.8516,
+      "step": 1749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.214270246860113e-06,
+      "loss": 4.3086,
+      "step": 1750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.212106972715461e-06,
+      "loss": 4.4609,
+      "step": 1751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.209943698570811e-06,
+      "loss": 4.2266,
+      "step": 1752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.207780424426159e-06,
+      "loss": 4.7617,
+      "step": 1753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.205617150281509e-06,
+      "loss": 4.625,
+      "step": 1754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.203453876136857e-06,
+      "loss": 4.9688,
+      "step": 1755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.201290601992204e-06,
+      "loss": 4.7305,
+      "step": 1756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.199127327847555e-06,
+      "loss": 4.7148,
+      "step": 1757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.196964053702902e-06,
+      "loss": 4.5508,
+      "step": 1758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.19480077955825e-06,
+      "loss": 4.9609,
+      "step": 1759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.1926375054136e-06,
+      "loss": 3.793,
+      "step": 1760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.190474231268948e-06,
+      "loss": 5.1367,
+      "step": 1761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.188310957124297e-06,
+      "loss": 4.4023,
+      "step": 1762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.186147682979645e-06,
+      "loss": 3.9746,
+      "step": 1763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.183984408834995e-06,
+      "loss": 5.0547,
+      "step": 1764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.1818211346903425e-06,
+      "loss": 5.1992,
+      "step": 1765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.179657860545691e-06,
+      "loss": 5.0742,
+      "step": 1766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.177494586401041e-06,
+      "loss": 5.1289,
+      "step": 1767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.1753313122563885e-06,
+      "loss": 4.9375,
+      "step": 1768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.173168038111736e-06,
+      "loss": 4.1445,
+      "step": 1769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.171004763967086e-06,
+      "loss": 4.7188,
+      "step": 1770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.168841489822434e-06,
+      "loss": 4.6484,
+      "step": 1771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.166678215677783e-06,
+      "loss": 4.6172,
+      "step": 1772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.164514941533132e-06,
+      "loss": 4.3555,
+      "step": 1773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.162351667388481e-06,
+      "loss": 5.0508,
+      "step": 1774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.16018839324383e-06,
+      "loss": 4.9375,
+      "step": 1775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.158025119099179e-06,
+      "loss": 4.7422,
+      "step": 1776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.155861844954526e-06,
+      "loss": 4.125,
+      "step": 1777
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.153698570809875e-06,
+      "loss": 4.6133,
+      "step": 1778
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.1515352966652245e-06,
+      "loss": 4.7031,
+      "step": 1779
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.149372022520572e-06,
+      "loss": 4.4766,
+      "step": 1780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.147208748375921e-06,
+      "loss": 5.0625,
+      "step": 1781
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.1450454742312705e-06,
+      "loss": 4.4688,
+      "step": 1782
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.1428822000866175e-06,
+      "loss": 5.1133,
+      "step": 1783
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.140718925941966e-06,
+      "loss": 4.7891,
+      "step": 1784
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.138555651797316e-06,
+      "loss": 3.998,
+      "step": 1785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.136392377652664e-06,
+      "loss": 4.8828,
+      "step": 1786
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.134229103508013e-06,
+      "loss": 4.7305,
+      "step": 1787
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.1320658293633625e-06,
+      "loss": 4.6797,
+      "step": 1788
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.1299025552187094e-06,
+      "loss": 4.9336,
+      "step": 1789
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.127739281074058e-06,
+      "loss": 5.0508,
+      "step": 1790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.125576006929408e-06,
+      "loss": 5.1562,
+      "step": 1791
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.123412732784756e-06,
+      "loss": 4.957,
+      "step": 1792
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.121249458640104e-06,
+      "loss": 4.6797,
+      "step": 1793
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.119086184495452e-06,
+      "loss": 5.0156,
+      "step": 1794
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.116922910350801e-06,
+      "loss": 4.2812,
+      "step": 1795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.11475963620615e-06,
+      "loss": 4.8398,
+      "step": 1796
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.112596362061498e-06,
+      "loss": 4.6367,
+      "step": 1797
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.110433087916847e-06,
+      "loss": 4.8008,
+      "step": 1798
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.108269813772196e-06,
+      "loss": 5.0586,
+      "step": 1799
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.1061065396275455e-06,
+      "loss": 4.9531,
+      "step": 1800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.103943265482894e-06,
+      "loss": 4.957,
+      "step": 1801
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.101779991338241e-06,
+      "loss": 5.0312,
+      "step": 1802
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.099616717193591e-06,
+      "loss": 4.6562,
+      "step": 1803
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.097453443048939e-06,
+      "loss": 5.1602,
+      "step": 1804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.095290168904288e-06,
+      "loss": 4.0,
+      "step": 1805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.093126894759637e-06,
+      "loss": 4.2734,
+      "step": 1806
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.090963620614986e-06,
+      "loss": 4.8086,
+      "step": 1807
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.088800346470334e-06,
+      "loss": 3.9883,
+      "step": 1808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.086637072325683e-06,
+      "loss": 4.7852,
+      "step": 1809
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.084473798181032e-06,
+      "loss": 4.3867,
+      "step": 1810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.08231052403638e-06,
+      "loss": 4.6445,
+      "step": 1811
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.080147249891729e-06,
+      "loss": 4.7539,
+      "step": 1812
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.077983975747078e-06,
+      "loss": 4.6094,
+      "step": 1813
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.075820701602425e-06,
+      "loss": 5.1992,
+      "step": 1814
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.0736574274577745e-06,
+      "loss": 4.7461,
+      "step": 1815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.071494153313123e-06,
+      "loss": 4.957,
+      "step": 1816
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.069330879168473e-06,
+      "loss": 4.9375,
+      "step": 1817
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.06716760502382e-06,
+      "loss": 4.0117,
+      "step": 1818
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.065004330879169e-06,
+      "loss": 4.2227,
+      "step": 1819
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.062841056734518e-06,
+      "loss": 4.8555,
+      "step": 1820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.060677782589866e-06,
+      "loss": 4.8164,
+      "step": 1821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.058514508445216e-06,
+      "loss": 4.7305,
+      "step": 1822
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.056351234300564e-06,
+      "loss": 4.9297,
+      "step": 1823
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.054187960155912e-06,
+      "loss": 4.9141,
+      "step": 1824
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.052024686011262e-06,
+      "loss": 4.7383,
+      "step": 1825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.04986141186661e-06,
+      "loss": 4.418,
+      "step": 1826
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.0476981377219576e-06,
+      "loss": 4.9766,
+      "step": 1827
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.045534863577307e-06,
+      "loss": 4.3633,
+      "step": 1828
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.043371589432656e-06,
+      "loss": 4.6836,
+      "step": 1829
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.041208315288003e-06,
+      "loss": 4.875,
+      "step": 1830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.039045041143352e-06,
+      "loss": 4.7109,
+      "step": 1831
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.036881766998701e-06,
+      "loss": 4.7695,
+      "step": 1832
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.0347184928540495e-06,
+      "loss": 5.0664,
+      "step": 1833
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.032555218709398e-06,
+      "loss": 4.8203,
+      "step": 1834
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.030391944564748e-06,
+      "loss": 4.7773,
+      "step": 1835
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.0282286704200955e-06,
+      "loss": 4.9844,
+      "step": 1836
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.026065396275444e-06,
+      "loss": 4.9531,
+      "step": 1837
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.023902122130794e-06,
+      "loss": 4.6328,
+      "step": 1838
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.0217388479861415e-06,
+      "loss": 4.5625,
+      "step": 1839
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.019575573841491e-06,
+      "loss": 4.4492,
+      "step": 1840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.01741229969684e-06,
+      "loss": 5.1328,
+      "step": 1841
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.0152490255521874e-06,
+      "loss": 3.9473,
+      "step": 1842
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.013085751407536e-06,
+      "loss": 5.0078,
+      "step": 1843
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.010922477262885e-06,
+      "loss": 4.3164,
+      "step": 1844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.008759203118234e-06,
+      "loss": 4.4648,
+      "step": 1845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.006595928973583e-06,
+      "loss": 3.9688,
+      "step": 1846
+    },
+    {
+      "epoch": 0.4,
+      "step": 1846,
+      "total_flos": 2.5288774310795674e+20,
+      "train_loss": 4.866100902288732,
+      "train_runtime": 1694.857,
+      "train_samples_per_second": 557.659,
+      "train_steps_per_second": 1.089
+    }
+  ],
+  "max_steps": 1846,
+  "num_train_epochs": 1,
+  "start_time": 1702321654.4430044,
+  "total_flos": 2.5288774310795674e+20,
+  "trial_name": null,
+  "trial_params": null
+}