diff --git "a/small/trainer_state.json" "b/small/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/small/trainer_state.json"
@@ -0,0 +1,78019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.7130711425593769,
+  "eval_steps": 500,
+  "global_step": 13000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2099778765107052e-05,
+      "loss": 6.2494,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2099557530214105e-05,
+      "loss": 5.9379,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2099336295321155e-05,
+      "loss": 5.9627,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2099115060428208e-05,
+      "loss": 5.9533,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2098893825535259e-05,
+      "loss": 6.0292,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2098672590642312e-05,
+      "loss": 6.1759,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2098451355749364e-05,
+      "loss": 5.8527,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2098230120856417e-05,
+      "loss": 5.7614,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2098008885963468e-05,
+      "loss": 5.9216,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.209778765107052e-05,
+      "loss": 5.6604,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2097566416177573e-05,
+      "loss": 5.8151,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2097345181284624e-05,
+      "loss": 5.4899,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2097123946391677e-05,
+      "loss": 6.0202,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2096902711498727e-05,
+      "loss": 5.8332,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.209668147660578e-05,
+      "loss": 5.4777,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2096460241712833e-05,
+      "loss": 5.8984,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2096239006819885e-05,
+      "loss": 5.4229,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2096017771926938e-05,
+      "loss": 5.5878,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.209579653703399e-05,
+      "loss": 5.4107,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2095575302141041e-05,
+      "loss": 5.5393,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2095354067248092e-05,
+      "loss": 5.4162,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2095132832355145e-05,
+      "loss": 5.3331,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2094911597462198e-05,
+      "loss": 5.3474,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.209469036256925e-05,
+      "loss": 5.3807,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2094469127676301e-05,
+      "loss": 5.0419,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2094247892783354e-05,
+      "loss": 5.2377,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2094026657890406e-05,
+      "loss": 5.3479,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2093805422997457e-05,
+      "loss": 5.2335,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.209358418810451e-05,
+      "loss": 5.1587,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2093362953211562e-05,
+      "loss": 5.1019,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2093141718318613e-05,
+      "loss": 5.0818,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2092920483425666e-05,
+      "loss": 5.1448,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2092699248532719e-05,
+      "loss": 5.1841,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2092478013639771e-05,
+      "loss": 5.032,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2092256778746824e-05,
+      "loss": 4.861,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2092035543853875e-05,
+      "loss": 5.1035,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2091814308960926e-05,
+      "loss": 4.737,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2091593074067978e-05,
+      "loss": 4.7012,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2091371839175031e-05,
+      "loss": 4.7071,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2091150604282083e-05,
+      "loss": 4.7569,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2090929369389136e-05,
+      "loss": 4.7605,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2090708134496187e-05,
+      "loss": 4.5505,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.209048689960324e-05,
+      "loss": 4.869,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.209026566471029e-05,
+      "loss": 4.5049,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2090044429817343e-05,
+      "loss": 4.2269,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2089823194924396e-05,
+      "loss": 4.5952,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2089601960031447e-05,
+      "loss": 4.4343,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.20893807251385e-05,
+      "loss": 4.4084,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2089159490245552e-05,
+      "loss": 4.5974,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2088938255352605e-05,
+      "loss": 4.4134,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2088717020459657e-05,
+      "loss": 4.4387,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2088495785566708e-05,
+      "loss": 4.2782,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2088274550673759e-05,
+      "loss": 4.3694,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2088053315780812e-05,
+      "loss": 4.3067,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2087832080887864e-05,
+      "loss": 4.1674,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2087610845994917e-05,
+      "loss": 4.3358,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.208738961110197e-05,
+      "loss": 4.1676,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.208716837620902e-05,
+      "loss": 4.0904,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2086947141316073e-05,
+      "loss": 3.9501,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2086725906423126e-05,
+      "loss": 4.1288,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2086504671530176e-05,
+      "loss": 4.174,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2086283436637229e-05,
+      "loss": 4.1424,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.208606220174428e-05,
+      "loss": 4.1296,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2085840966851333e-05,
+      "loss": 3.9578,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2085619731958385e-05,
+      "loss": 3.9026,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2085398497065438e-05,
+      "loss": 3.9955,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.208517726217249e-05,
+      "loss": 3.7823,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2084956027279541e-05,
+      "loss": 3.9523,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2084734792386592e-05,
+      "loss": 3.8643,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2084513557493645e-05,
+      "loss": 3.887,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2084292322600698e-05,
+      "loss": 3.7514,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.208407108770775e-05,
+      "loss": 3.5862,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2083849852814803e-05,
+      "loss": 3.7149,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2083628617921854e-05,
+      "loss": 3.6303,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2083407383028906e-05,
+      "loss": 3.6516,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2083186148135959e-05,
+      "loss": 3.759,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.208296491324301e-05,
+      "loss": 3.7953,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2082743678350062e-05,
+      "loss": 3.7143,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2082522443457115e-05,
+      "loss": 3.5918,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2082301208564166e-05,
+      "loss": 3.6857,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2082079973671219e-05,
+      "loss": 3.749,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2081858738778271e-05,
+      "loss": 3.6016,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2081637503885324e-05,
+      "loss": 3.5423,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2081416268992375e-05,
+      "loss": 3.489,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2081195034099426e-05,
+      "loss": 3.4033,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2080973799206478e-05,
+      "loss": 3.3356,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.208075256431353e-05,
+      "loss": 3.2311,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2080531329420583e-05,
+      "loss": 3.2486,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2080310094527636e-05,
+      "loss": 3.1463,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2080088859634689e-05,
+      "loss": 3.1161,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.207986762474174e-05,
+      "loss": 3.1493,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2079646389848792e-05,
+      "loss": 3.168,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2079425154955843e-05,
+      "loss": 3.0599,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2079203920062896e-05,
+      "loss": 3.087,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2078982685169948e-05,
+      "loss": 2.9003,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2078761450277e-05,
+      "loss": 2.9097,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2078540215384052e-05,
+      "loss": 2.9576,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2078318980491104e-05,
+      "loss": 3.1639,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2078097745598157e-05,
+      "loss": 2.7802,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2077876510705208e-05,
+      "loss": 3.2021,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.207765527581226e-05,
+      "loss": 2.6688,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2077434040919312e-05,
+      "loss": 2.9469,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2077212806026364e-05,
+      "loss": 3.0295,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2076991571133417e-05,
+      "loss": 2.7608,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.207677033624047e-05,
+      "loss": 2.5818,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2076549101347522e-05,
+      "loss": 2.6256,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2076327866454573e-05,
+      "loss": 2.6644,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2076106631561626e-05,
+      "loss": 2.7138,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2075885396668676e-05,
+      "loss": 2.7238,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2075664161775729e-05,
+      "loss": 3.0339,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2075442926882782e-05,
+      "loss": 2.5504,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2075221691989834e-05,
+      "loss": 2.6765,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2075000457096885e-05,
+      "loss": 2.6176,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2074779222203938e-05,
+      "loss": 3.1166,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.207455798731099e-05,
+      "loss": 2.8893,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2074336752418041e-05,
+      "loss": 2.5406,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2074115517525094e-05,
+      "loss": 2.6891,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2073894282632145e-05,
+      "loss": 2.3684,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2073673047739197e-05,
+      "loss": 2.5043,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.207345181284625e-05,
+      "loss": 2.6407,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2073230577953303e-05,
+      "loss": 2.5009,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2073009343060355e-05,
+      "loss": 2.6209,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2072788108167408e-05,
+      "loss": 2.5166,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2072566873274459e-05,
+      "loss": 2.524,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.207234563838151e-05,
+      "loss": 2.3792,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2072124403488562e-05,
+      "loss": 2.8487,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2071903168595615e-05,
+      "loss": 2.6234,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2071681933702668e-05,
+      "loss": 2.9876,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2071460698809719e-05,
+      "loss": 2.6452,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2071239463916771e-05,
+      "loss": 2.3216,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2071018229023824e-05,
+      "loss": 2.8924,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2070796994130875e-05,
+      "loss": 2.6588,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2070575759237927e-05,
+      "loss": 2.5873,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.207035452434498e-05,
+      "loss": 2.4033,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.207013328945203e-05,
+      "loss": 2.3411,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2069912054559083e-05,
+      "loss": 2.6996,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2069690819666136e-05,
+      "loss": 2.4712,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2069469584773189e-05,
+      "loss": 2.9068,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2069248349880241e-05,
+      "loss": 2.5642,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2069027114987292e-05,
+      "loss": 2.5955,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2068805880094343e-05,
+      "loss": 2.2884,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2068584645201396e-05,
+      "loss": 2.4735,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2068363410308448e-05,
+      "loss": 2.4379,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2068142175415501e-05,
+      "loss": 2.4136,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2067920940522554e-05,
+      "loss": 2.3871,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2067699705629604e-05,
+      "loss": 2.3193,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2067478470736657e-05,
+      "loss": 2.3528,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.206725723584371e-05,
+      "loss": 2.241,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.206703600095076e-05,
+      "loss": 2.3944,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2066814766057813e-05,
+      "loss": 2.3461,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2066593531164864e-05,
+      "loss": 2.1517,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2066372296271917e-05,
+      "loss": 2.622,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.206615106137897e-05,
+      "loss": 2.3813,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2065929826486022e-05,
+      "loss": 2.1511,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2065708591593075e-05,
+      "loss": 2.3742,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2065487356700125e-05,
+      "loss": 2.2326,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2065266121807176e-05,
+      "loss": 2.6104,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2065044886914229e-05,
+      "loss": 2.2979,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2064823652021282e-05,
+      "loss": 2.3334,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2064602417128334e-05,
+      "loss": 2.621,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2064381182235387e-05,
+      "loss": 2.2042,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2064159947342438e-05,
+      "loss": 2.03,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.206393871244949e-05,
+      "loss": 2.5569,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2063717477556543e-05,
+      "loss": 2.5348,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2063496242663594e-05,
+      "loss": 2.4624,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2063275007770647e-05,
+      "loss": 2.322,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2063053772877697e-05,
+      "loss": 2.3216,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.206283253798475e-05,
+      "loss": 2.2757,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2062611303091803e-05,
+      "loss": 2.3139,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2062390068198855e-05,
+      "loss": 2.4493,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2062168833305908e-05,
+      "loss": 2.3694,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2061947598412959e-05,
+      "loss": 2.501,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.206172636352001e-05,
+      "loss": 2.311,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2061505128627062e-05,
+      "loss": 2.3907,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2061283893734115e-05,
+      "loss": 2.307,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2061062658841168e-05,
+      "loss": 2.3019,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.206084142394822e-05,
+      "loss": 2.6038,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2060620189055271e-05,
+      "loss": 2.311,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2060398954162324e-05,
+      "loss": 2.5681,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2060177719269376e-05,
+      "loss": 2.1438,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2059956484376427e-05,
+      "loss": 2.4309,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.205973524948348e-05,
+      "loss": 2.0145,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2059514014590532e-05,
+      "loss": 2.2559,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2059292779697583e-05,
+      "loss": 2.0744,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2059071544804636e-05,
+      "loss": 2.3087,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2058850309911689e-05,
+      "loss": 2.246,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2058629075018741e-05,
+      "loss": 2.1397,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2058407840125792e-05,
+      "loss": 2.4957,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2058186605232845e-05,
+      "loss": 2.3072,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2057965370339896e-05,
+      "loss": 2.1327,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2057744135446948e-05,
+      "loss": 2.2465,
+      "step": 191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2057522900554001e-05,
+      "loss": 2.3005,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2057301665661053e-05,
+      "loss": 2.3181,
+      "step": 193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2057080430768106e-05,
+      "loss": 2.1322,
+      "step": 194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2056859195875157e-05,
+      "loss": 2.3062,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.205663796098221e-05,
+      "loss": 2.5299,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.205641672608926e-05,
+      "loss": 2.2019,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2056195491196313e-05,
+      "loss": 2.3158,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2055974256303366e-05,
+      "loss": 2.1957,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2055753021410417e-05,
+      "loss": 2.0758,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.205553178651747e-05,
+      "loss": 2.2198,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2055310551624522e-05,
+      "loss": 2.2026,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2055089316731575e-05,
+      "loss": 2.205,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2054868081838625e-05,
+      "loss": 2.2121,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2054646846945678e-05,
+      "loss": 2.1506,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2054425612052729e-05,
+      "loss": 2.0869,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2054204377159782e-05,
+      "loss": 1.9535,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2053983142266834e-05,
+      "loss": 2.0574,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2053761907373887e-05,
+      "loss": 2.1657,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.205354067248094e-05,
+      "loss": 2.105,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.205331943758799e-05,
+      "loss": 2.1555,
+      "step": 211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2053098202695043e-05,
+      "loss": 2.2332,
+      "step": 212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2052876967802094e-05,
+      "loss": 2.3403,
+      "step": 213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2052655732909146e-05,
+      "loss": 2.108,
+      "step": 214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2052434498016199e-05,
+      "loss": 2.0836,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2052213263123252e-05,
+      "loss": 2.1525,
+      "step": 216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2051992028230303e-05,
+      "loss": 2.0818,
+      "step": 217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2051770793337355e-05,
+      "loss": 2.3257,
+      "step": 218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2051549558444408e-05,
+      "loss": 2.1242,
+      "step": 219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2051328323551459e-05,
+      "loss": 2.3579,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2051107088658511e-05,
+      "loss": 1.9913,
+      "step": 221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2050885853765562e-05,
+      "loss": 2.1765,
+      "step": 222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2050664618872615e-05,
+      "loss": 2.1032,
+      "step": 223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2050443383979668e-05,
+      "loss": 2.0624,
+      "step": 224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.205022214908672e-05,
+      "loss": 2.0947,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2050000914193773e-05,
+      "loss": 2.2217,
+      "step": 226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2049779679300825e-05,
+      "loss": 1.9053,
+      "step": 227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2049558444407876e-05,
+      "loss": 2.2266,
+      "step": 228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2049337209514927e-05,
+      "loss": 2.4092,
+      "step": 229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.204911597462198e-05,
+      "loss": 2.0461,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2048894739729032e-05,
+      "loss": 2.3026,
+      "step": 231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2048673504836085e-05,
+      "loss": 2.1552,
+      "step": 232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2048452269943136e-05,
+      "loss": 2.2422,
+      "step": 233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2048231035050189e-05,
+      "loss": 2.233,
+      "step": 234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2048009800157241e-05,
+      "loss": 2.1816,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2047788565264294e-05,
+      "loss": 2.2464,
+      "step": 236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2047567330371345e-05,
+      "loss": 2.2997,
+      "step": 237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2047346095478397e-05,
+      "loss": 2.124,
+      "step": 238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2047124860585448e-05,
+      "loss": 2.0221,
+      "step": 239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.20469036256925e-05,
+      "loss": 1.9506,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2046682390799553e-05,
+      "loss": 2.0779,
+      "step": 241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2046461155906606e-05,
+      "loss": 2.0282,
+      "step": 242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2046239921013659e-05,
+      "loss": 2.281,
+      "step": 243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.204601868612071e-05,
+      "loss": 1.8426,
+      "step": 244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.204579745122776e-05,
+      "loss": 1.9389,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2045576216334813e-05,
+      "loss": 1.7942,
+      "step": 246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2045354981441866e-05,
+      "loss": 2.3796,
+      "step": 247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2045133746548918e-05,
+      "loss": 1.9509,
+      "step": 248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.204491251165597e-05,
+      "loss": 2.3517,
+      "step": 249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2044691276763022e-05,
+      "loss": 2.045,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2044470041870074e-05,
+      "loss": 2.278,
+      "step": 251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2044248806977127e-05,
+      "loss": 2.2445,
+      "step": 252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2044027572084178e-05,
+      "loss": 2.1946,
+      "step": 253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.204380633719123e-05,
+      "loss": 2.2496,
+      "step": 254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2043585102298282e-05,
+      "loss": 2.0041,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2043363867405334e-05,
+      "loss": 1.9711,
+      "step": 256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2043142632512387e-05,
+      "loss": 1.889,
+      "step": 257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.204292139761944e-05,
+      "loss": 1.9935,
+      "step": 258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2042700162726492e-05,
+      "loss": 2.1702,
+      "step": 259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2042478927833543e-05,
+      "loss": 2.0582,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2042257692940594e-05,
+      "loss": 2.0215,
+      "step": 261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2042036458047646e-05,
+      "loss": 1.9359,
+      "step": 262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2041815223154699e-05,
+      "loss": 1.8524,
+      "step": 263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2041593988261752e-05,
+      "loss": 1.9025,
+      "step": 264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2041372753368804e-05,
+      "loss": 2.3437,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2041151518475855e-05,
+      "loss": 2.1731,
+      "step": 266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2040930283582908e-05,
+      "loss": 2.3649,
+      "step": 267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.204070904868996e-05,
+      "loss": 1.8324,
+      "step": 268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2040487813797011e-05,
+      "loss": 1.8753,
+      "step": 269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2040266578904064e-05,
+      "loss": 2.0902,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2040045344011115e-05,
+      "loss": 2.1648,
+      "step": 271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2039824109118167e-05,
+      "loss": 1.9878,
+      "step": 272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.203960287422522e-05,
+      "loss": 1.9437,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2039381639332273e-05,
+      "loss": 2.0758,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2039160404439325e-05,
+      "loss": 2.0224,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2038939169546376e-05,
+      "loss": 2.0,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2038717934653429e-05,
+      "loss": 2.0943,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.203849669976048e-05,
+      "loss": 1.9301,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2038275464867532e-05,
+      "loss": 1.8916,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2038054229974585e-05,
+      "loss": 2.0877,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2037832995081638e-05,
+      "loss": 1.9178,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2037611760188688e-05,
+      "loss": 2.0196,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2037390525295741e-05,
+      "loss": 1.9875,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2037169290402794e-05,
+      "loss": 2.0184,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2036948055509845e-05,
+      "loss": 1.7628,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2036726820616897e-05,
+      "loss": 1.8997,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.203650558572395e-05,
+      "loss": 2.1896,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2036284350831e-05,
+      "loss": 1.851,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2036063115938053e-05,
+      "loss": 2.0362,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2035841881045106e-05,
+      "loss": 1.8698,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2035620646152159e-05,
+      "loss": 2.0835,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.203539941125921e-05,
+      "loss": 1.9232,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2035178176366262e-05,
+      "loss": 1.7713,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2034956941473313e-05,
+      "loss": 2.3041,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2034735706580366e-05,
+      "loss": 2.2091,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2034514471687418e-05,
+      "loss": 1.905,
+      "step": 296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2034293236794471e-05,
+      "loss": 2.1838,
+      "step": 297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2034072001901524e-05,
+      "loss": 2.0544,
+      "step": 298
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2033850767008574e-05,
+      "loss": 1.943,
+      "step": 299
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2033629532115627e-05,
+      "loss": 1.7915,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2033408297222678e-05,
+      "loss": 2.1064,
+      "step": 301
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.203318706232973e-05,
+      "loss": 2.1636,
+      "step": 302
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2032965827436783e-05,
+      "loss": 2.1027,
+      "step": 303
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2032744592543834e-05,
+      "loss": 1.7855,
+      "step": 304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2032523357650887e-05,
+      "loss": 1.8238,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.203230212275794e-05,
+      "loss": 1.7987,
+      "step": 306
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2032080887864992e-05,
+      "loss": 2.0291,
+      "step": 307
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2031859652972043e-05,
+      "loss": 1.9054,
+      "step": 308
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2031638418079095e-05,
+      "loss": 1.7622,
+      "step": 309
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2031417183186146e-05,
+      "loss": 2.1124,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2031195948293199e-05,
+      "loss": 1.9258,
+      "step": 311
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2030974713400252e-05,
+      "loss": 2.1088,
+      "step": 312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2030753478507304e-05,
+      "loss": 1.8885,
+      "step": 313
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2030532243614357e-05,
+      "loss": 1.9167,
+      "step": 314
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2030311008721408e-05,
+      "loss": 1.6188,
+      "step": 315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.203008977382846e-05,
+      "loss": 1.9314,
+      "step": 316
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2029868538935511e-05,
+      "loss": 1.8337,
+      "step": 317
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2029647304042564e-05,
+      "loss": 1.9447,
+      "step": 318
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2029426069149616e-05,
+      "loss": 2.0193,
+      "step": 319
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2029204834256669e-05,
+      "loss": 1.7692,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.202898359936372e-05,
+      "loss": 1.915,
+      "step": 321
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2028762364470773e-05,
+      "loss": 1.886,
+      "step": 322
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2028541129577825e-05,
+      "loss": 1.9713,
+      "step": 323
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2028319894684878e-05,
+      "loss": 2.0555,
+      "step": 324
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2028098659791929e-05,
+      "loss": 1.9737,
+      "step": 325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.202787742489898e-05,
+      "loss": 1.8038,
+      "step": 326
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2027656190006032e-05,
+      "loss": 1.7807,
+      "step": 327
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2027434955113085e-05,
+      "loss": 1.8351,
+      "step": 328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2027213720220138e-05,
+      "loss": 2.0682,
+      "step": 329
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.202699248532719e-05,
+      "loss": 1.8613,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2026771250434243e-05,
+      "loss": 2.0307,
+      "step": 331
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2026550015541294e-05,
+      "loss": 1.6822,
+      "step": 332
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2026328780648345e-05,
+      "loss": 2.0864,
+      "step": 333
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2026107545755397e-05,
+      "loss": 2.1202,
+      "step": 334
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.202588631086245e-05,
+      "loss": 2.0163,
+      "step": 335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2025665075969502e-05,
+      "loss": 1.8257,
+      "step": 336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2025443841076553e-05,
+      "loss": 1.9025,
+      "step": 337
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2025222606183606e-05,
+      "loss": 2.0835,
+      "step": 338
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2025001371290659e-05,
+      "loss": 1.7888,
+      "step": 339
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2024780136397711e-05,
+      "loss": 1.959,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2024558901504762e-05,
+      "loss": 1.841,
+      "step": 341
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2024337666611815e-05,
+      "loss": 1.7588,
+      "step": 342
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2024116431718866e-05,
+      "loss": 1.7913,
+      "step": 343
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2023895196825918e-05,
+      "loss": 1.9008,
+      "step": 344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2023673961932971e-05,
+      "loss": 2.0296,
+      "step": 345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2023452727040023e-05,
+      "loss": 1.8622,
+      "step": 346
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2023231492147076e-05,
+      "loss": 1.8546,
+      "step": 347
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2023010257254127e-05,
+      "loss": 2.117,
+      "step": 348
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2022789022361178e-05,
+      "loss": 1.9975,
+      "step": 349
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.202256778746823e-05,
+      "loss": 1.9062,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2022346552575283e-05,
+      "loss": 2.1332,
+      "step": 351
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2022125317682336e-05,
+      "loss": 1.8583,
+      "step": 352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2021904082789387e-05,
+      "loss": 2.0382,
+      "step": 353
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.202168284789644e-05,
+      "loss": 1.973,
+      "step": 354
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2021461613003492e-05,
+      "loss": 1.8628,
+      "step": 355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2021240378110545e-05,
+      "loss": 1.8892,
+      "step": 356
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2021019143217595e-05,
+      "loss": 1.8121,
+      "step": 357
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2020797908324648e-05,
+      "loss": 1.9993,
+      "step": 358
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2020576673431699e-05,
+      "loss": 1.7818,
+      "step": 359
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2020355438538752e-05,
+      "loss": 1.8885,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2020134203645804e-05,
+      "loss": 2.0092,
+      "step": 361
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2019912968752857e-05,
+      "loss": 1.8893,
+      "step": 362
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.201969173385991e-05,
+      "loss": 1.9205,
+      "step": 363
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.201947049896696e-05,
+      "loss": 1.9489,
+      "step": 364
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2019249264074013e-05,
+      "loss": 1.9171,
+      "step": 365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2019028029181064e-05,
+      "loss": 1.8895,
+      "step": 366
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2018806794288116e-05,
+      "loss": 1.7185,
+      "step": 367
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2018585559395169e-05,
+      "loss": 1.8327,
+      "step": 368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2018364324502222e-05,
+      "loss": 2.0502,
+      "step": 369
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2018143089609273e-05,
+      "loss": 1.9539,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2017921854716325e-05,
+      "loss": 2.1759,
+      "step": 371
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2017700619823378e-05,
+      "loss": 1.6659,
+      "step": 372
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2017479384930429e-05,
+      "loss": 1.9238,
+      "step": 373
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2017258150037481e-05,
+      "loss": 1.8925,
+      "step": 374
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2017036915144532e-05,
+      "loss": 2.1421,
+      "step": 375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2016815680251585e-05,
+      "loss": 1.8726,
+      "step": 376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2016594445358637e-05,
+      "loss": 2.0468,
+      "step": 377
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.201637321046569e-05,
+      "loss": 1.8589,
+      "step": 378
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2016151975572743e-05,
+      "loss": 1.5927,
+      "step": 379
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2015930740679794e-05,
+      "loss": 1.8779,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2015709505786846e-05,
+      "loss": 1.9629,
+      "step": 381
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2015488270893897e-05,
+      "loss": 1.9329,
+      "step": 382
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.201526703600095e-05,
+      "loss": 1.8599,
+      "step": 383
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2015045801108002e-05,
+      "loss": 1.9864,
+      "step": 384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2014824566215055e-05,
+      "loss": 1.83,
+      "step": 385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2014603331322106e-05,
+      "loss": 1.7904,
+      "step": 386
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2014382096429159e-05,
+      "loss": 1.9708,
+      "step": 387
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2014160861536211e-05,
+      "loss": 2.0796,
+      "step": 388
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2013939626643262e-05,
+      "loss": 1.8932,
+      "step": 389
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2013718391750315e-05,
+      "loss": 1.931,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2013497156857367e-05,
+      "loss": 1.8302,
+      "step": 391
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2013275921964418e-05,
+      "loss": 1.773,
+      "step": 392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.201305468707147e-05,
+      "loss": 1.895,
+      "step": 393
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2012833452178523e-05,
+      "loss": 1.7915,
+      "step": 394
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2012612217285576e-05,
+      "loss": 1.8091,
+      "step": 395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2012390982392627e-05,
+      "loss": 1.9695,
+      "step": 396
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.201216974749968e-05,
+      "loss": 1.9148,
+      "step": 397
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.201194851260673e-05,
+      "loss": 1.8008,
+      "step": 398
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2011727277713783e-05,
+      "loss": 1.9085,
+      "step": 399
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2011506042820836e-05,
+      "loss": 1.9472,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2011284807927888e-05,
+      "loss": 1.8137,
+      "step": 401
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2011063573034941e-05,
+      "loss": 1.9906,
+      "step": 402
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2010842338141992e-05,
+      "loss": 1.7841,
+      "step": 403
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2010621103249044e-05,
+      "loss": 1.8728,
+      "step": 404
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2010399868356095e-05,
+      "loss": 2.0759,
+      "step": 405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2010178633463148e-05,
+      "loss": 1.8314,
+      "step": 406
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.20099573985702e-05,
+      "loss": 1.8341,
+      "step": 407
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2009736163677252e-05,
+      "loss": 1.7549,
+      "step": 408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2009514928784304e-05,
+      "loss": 2.1419,
+      "step": 409
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2009293693891357e-05,
+      "loss": 1.7279,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.200907245899841e-05,
+      "loss": 1.7952,
+      "step": 411
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2008851224105462e-05,
+      "loss": 2.0525,
+      "step": 412
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2008629989212513e-05,
+      "loss": 1.9569,
+      "step": 413
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2008408754319564e-05,
+      "loss": 1.6041,
+      "step": 414
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2008187519426616e-05,
+      "loss": 2.0107,
+      "step": 415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2007966284533669e-05,
+      "loss": 1.8068,
+      "step": 416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2007745049640722e-05,
+      "loss": 1.8754,
+      "step": 417
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2007523814747774e-05,
+      "loss": 1.7861,
+      "step": 418
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2007302579854825e-05,
+      "loss": 1.8519,
+      "step": 419
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2007081344961878e-05,
+      "loss": 1.8379,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2006860110068929e-05,
+      "loss": 1.821,
+      "step": 421
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2006638875175981e-05,
+      "loss": 1.7279,
+      "step": 422
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2006417640283034e-05,
+      "loss": 1.9146,
+      "step": 423
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2006196405390087e-05,
+      "loss": 1.9633,
+      "step": 424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2005975170497137e-05,
+      "loss": 1.9419,
+      "step": 425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.200575393560419e-05,
+      "loss": 1.8026,
+      "step": 426
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2005532700711243e-05,
+      "loss": 1.8423,
+      "step": 427
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2005311465818295e-05,
+      "loss": 1.9496,
+      "step": 428
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2005090230925346e-05,
+      "loss": 2.0772,
+      "step": 429
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2004868996032397e-05,
+      "loss": 2.0318,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.200464776113945e-05,
+      "loss": 1.9078,
+      "step": 431
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2004426526246502e-05,
+      "loss": 1.838,
+      "step": 432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2004205291353555e-05,
+      "loss": 1.983,
+      "step": 433
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2003984056460608e-05,
+      "loss": 1.805,
+      "step": 434
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.200376282156766e-05,
+      "loss": 1.9786,
+      "step": 435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2003541586674711e-05,
+      "loss": 1.7417,
+      "step": 436
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2003320351781762e-05,
+      "loss": 1.8441,
+      "step": 437
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2003099116888815e-05,
+      "loss": 1.88,
+      "step": 438
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2002877881995867e-05,
+      "loss": 1.7053,
+      "step": 439
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.200265664710292e-05,
+      "loss": 1.7682,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.200243541220997e-05,
+      "loss": 1.8722,
+      "step": 441
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2002214177317023e-05,
+      "loss": 1.975,
+      "step": 442
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2001992942424076e-05,
+      "loss": 1.8161,
+      "step": 443
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2001771707531129e-05,
+      "loss": 1.845,
+      "step": 444
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.200155047263818e-05,
+      "loss": 1.9144,
+      "step": 445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.200132923774523e-05,
+      "loss": 1.8992,
+      "step": 446
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2001108002852283e-05,
+      "loss": 1.8648,
+      "step": 447
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2000886767959336e-05,
+      "loss": 2.0998,
+      "step": 448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2000665533066388e-05,
+      "loss": 1.805,
+      "step": 449
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2000444298173441e-05,
+      "loss": 1.755,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2000223063280493e-05,
+      "loss": 1.9506,
+      "step": 451
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2000001828387544e-05,
+      "loss": 1.7503,
+      "step": 452
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1999780593494595e-05,
+      "loss": 1.8686,
+      "step": 453
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1999559358601648e-05,
+      "loss": 1.9213,
+      "step": 454
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.19993381237087e-05,
+      "loss": 1.7717,
+      "step": 455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1999116888815753e-05,
+      "loss": 1.8258,
+      "step": 456
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1998895653922804e-05,
+      "loss": 1.6498,
+      "step": 457
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1998674419029857e-05,
+      "loss": 1.8013,
+      "step": 458
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.199845318413691e-05,
+      "loss": 1.9107,
+      "step": 459
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1998231949243962e-05,
+      "loss": 1.8218,
+      "step": 460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1998010714351013e-05,
+      "loss": 1.7679,
+      "step": 461
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1997789479458065e-05,
+      "loss": 1.7131,
+      "step": 462
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1997568244565116e-05,
+      "loss": 1.7763,
+      "step": 463
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1997347009672169e-05,
+      "loss": 1.6624,
+      "step": 464
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1997125774779222e-05,
+      "loss": 1.8549,
+      "step": 465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1996904539886274e-05,
+      "loss": 2.0371,
+      "step": 466
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1996683304993327e-05,
+      "loss": 1.9718,
+      "step": 467
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1996462070100378e-05,
+      "loss": 2.0965,
+      "step": 468
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.199624083520743e-05,
+      "loss": 1.6866,
+      "step": 469
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1996019600314481e-05,
+      "loss": 1.808,
+      "step": 470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1995798365421534e-05,
+      "loss": 1.6926,
+      "step": 471
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1995577130528586e-05,
+      "loss": 1.8374,
+      "step": 472
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1995355895635639e-05,
+      "loss": 1.7353,
+      "step": 473
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.199513466074269e-05,
+      "loss": 1.872,
+      "step": 474
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1994913425849743e-05,
+      "loss": 1.8178,
+      "step": 475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1994692190956795e-05,
+      "loss": 1.8045,
+      "step": 476
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1994470956063846e-05,
+      "loss": 1.8285,
+      "step": 477
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1994249721170899e-05,
+      "loss": 1.8117,
+      "step": 478
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.199402848627795e-05,
+      "loss": 1.9279,
+      "step": 479
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1993807251385002e-05,
+      "loss": 1.6499,
+      "step": 480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1993586016492055e-05,
+      "loss": 1.8115,
+      "step": 481
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1993364781599108e-05,
+      "loss": 1.9033,
+      "step": 482
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.199314354670616e-05,
+      "loss": 1.8269,
+      "step": 483
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1992922311813211e-05,
+      "loss": 1.8419,
+      "step": 484
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1992701076920264e-05,
+      "loss": 1.762,
+      "step": 485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1992479842027315e-05,
+      "loss": 1.7094,
+      "step": 486
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1992258607134367e-05,
+      "loss": 1.5908,
+      "step": 487
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.199203737224142e-05,
+      "loss": 1.7651,
+      "step": 488
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1991816137348472e-05,
+      "loss": 1.749,
+      "step": 489
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1991594902455523e-05,
+      "loss": 1.849,
+      "step": 490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1991373667562576e-05,
+      "loss": 2.0595,
+      "step": 491
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1991152432669629e-05,
+      "loss": 1.8275,
+      "step": 492
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.199093119777668e-05,
+      "loss": 1.9901,
+      "step": 493
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1990709962883732e-05,
+      "loss": 1.6606,
+      "step": 494
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1990488727990785e-05,
+      "loss": 1.8877,
+      "step": 495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1990267493097836e-05,
+      "loss": 1.7903,
+      "step": 496
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1990046258204888e-05,
+      "loss": 1.8785,
+      "step": 497
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.198982502331194e-05,
+      "loss": 1.8374,
+      "step": 498
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1989603788418993e-05,
+      "loss": 1.8251,
+      "step": 499
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1989382553526046e-05,
+      "loss": 1.7369,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1989161318633097e-05,
+      "loss": 1.7855,
+      "step": 501
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1988940083740148e-05,
+      "loss": 1.8959,
+      "step": 502
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.19887188488472e-05,
+      "loss": 1.7022,
+      "step": 503
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1988497613954253e-05,
+      "loss": 1.6233,
+      "step": 504
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1988276379061306e-05,
+      "loss": 1.8996,
+      "step": 505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1988055144168358e-05,
+      "loss": 1.8047,
+      "step": 506
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.198783390927541e-05,
+      "loss": 1.6935,
+      "step": 507
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1987612674382462e-05,
+      "loss": 1.7898,
+      "step": 508
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1987391439489513e-05,
+      "loss": 1.8066,
+      "step": 509
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1987170204596565e-05,
+      "loss": 1.8782,
+      "step": 510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1986948969703618e-05,
+      "loss": 1.9456,
+      "step": 511
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1986727734810669e-05,
+      "loss": 1.6901,
+      "step": 512
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1986506499917722e-05,
+      "loss": 1.8026,
+      "step": 513
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1986285265024774e-05,
+      "loss": 1.7128,
+      "step": 514
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1986064030131827e-05,
+      "loss": 1.7497,
+      "step": 515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.198584279523888e-05,
+      "loss": 1.676,
+      "step": 516
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.198562156034593e-05,
+      "loss": 1.7868,
+      "step": 517
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1985400325452981e-05,
+      "loss": 1.9384,
+      "step": 518
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1985179090560034e-05,
+      "loss": 1.8414,
+      "step": 519
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1984957855667086e-05,
+      "loss": 1.8388,
+      "step": 520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1984736620774139e-05,
+      "loss": 1.695,
+      "step": 521
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1984515385881192e-05,
+      "loss": 1.9211,
+      "step": 522
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1984294150988243e-05,
+      "loss": 2.0574,
+      "step": 523
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1984072916095295e-05,
+      "loss": 1.6197,
+      "step": 524
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1983851681202346e-05,
+      "loss": 1.7814,
+      "step": 525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1983630446309399e-05,
+      "loss": 1.7549,
+      "step": 526
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1983409211416451e-05,
+      "loss": 1.8049,
+      "step": 527
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1983187976523504e-05,
+      "loss": 1.725,
+      "step": 528
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1982966741630555e-05,
+      "loss": 1.7561,
+      "step": 529
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1982745506737607e-05,
+      "loss": 1.9584,
+      "step": 530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.198252427184466e-05,
+      "loss": 1.843,
+      "step": 531
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1982303036951713e-05,
+      "loss": 1.7888,
+      "step": 532
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1982081802058764e-05,
+      "loss": 1.994,
+      "step": 533
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1981860567165815e-05,
+      "loss": 1.7437,
+      "step": 534
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1981639332272867e-05,
+      "loss": 1.7606,
+      "step": 535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.198141809737992e-05,
+      "loss": 1.8532,
+      "step": 536
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1981196862486972e-05,
+      "loss": 1.9772,
+      "step": 537
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1980975627594025e-05,
+      "loss": 1.8858,
+      "step": 538
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1980754392701076e-05,
+      "loss": 1.9425,
+      "step": 539
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1980533157808129e-05,
+      "loss": 1.8709,
+      "step": 540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.198031192291518e-05,
+      "loss": 1.5773,
+      "step": 541
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1980090688022232e-05,
+      "loss": 1.7719,
+      "step": 542
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1979869453129285e-05,
+      "loss": 1.7198,
+      "step": 543
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1979648218236337e-05,
+      "loss": 1.6952,
+      "step": 544
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1979426983343388e-05,
+      "loss": 1.6439,
+      "step": 545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.197920574845044e-05,
+      "loss": 1.8398,
+      "step": 546
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1978984513557493e-05,
+      "loss": 1.8271,
+      "step": 547
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1978763278664546e-05,
+      "loss": 1.9786,
+      "step": 548
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1978542043771597e-05,
+      "loss": 1.7998,
+      "step": 549
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1978320808878648e-05,
+      "loss": 2.0632,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.19780995739857e-05,
+      "loss": 1.8209,
+      "step": 551
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1977878339092753e-05,
+      "loss": 1.8374,
+      "step": 552
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1977657104199806e-05,
+      "loss": 2.0257,
+      "step": 553
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1977435869306858e-05,
+      "loss": 1.9111,
+      "step": 554
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1977214634413911e-05,
+      "loss": 1.8641,
+      "step": 555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1976993399520962e-05,
+      "loss": 1.8179,
+      "step": 556
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1976772164628014e-05,
+      "loss": 1.8207,
+      "step": 557
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1976550929735065e-05,
+      "loss": 1.7408,
+      "step": 558
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1976329694842118e-05,
+      "loss": 1.799,
+      "step": 559
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.197610845994917e-05,
+      "loss": 1.7967,
+      "step": 560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1975887225056222e-05,
+      "loss": 1.8975,
+      "step": 561
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1975665990163274e-05,
+      "loss": 1.7899,
+      "step": 562
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1975444755270327e-05,
+      "loss": 1.6941,
+      "step": 563
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.197522352037738e-05,
+      "loss": 1.8821,
+      "step": 564
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.197500228548443e-05,
+      "loss": 1.8949,
+      "step": 565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1974781050591483e-05,
+      "loss": 1.9222,
+      "step": 566
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1974559815698534e-05,
+      "loss": 1.9826,
+      "step": 567
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1974338580805586e-05,
+      "loss": 1.7985,
+      "step": 568
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1974117345912639e-05,
+      "loss": 1.7111,
+      "step": 569
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1973896111019692e-05,
+      "loss": 1.7498,
+      "step": 570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1973674876126744e-05,
+      "loss": 1.6509,
+      "step": 571
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1973453641233795e-05,
+      "loss": 1.8558,
+      "step": 572
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1973232406340848e-05,
+      "loss": 1.7582,
+      "step": 573
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1973011171447899e-05,
+      "loss": 1.6918,
+      "step": 574
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1972789936554951e-05,
+      "loss": 1.6413,
+      "step": 575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1972568701662004e-05,
+      "loss": 1.822,
+      "step": 576
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1972347466769057e-05,
+      "loss": 1.9063,
+      "step": 577
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1972126231876107e-05,
+      "loss": 1.8411,
+      "step": 578
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.197190499698316e-05,
+      "loss": 1.7243,
+      "step": 579
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1971683762090213e-05,
+      "loss": 1.7729,
+      "step": 580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1971462527197264e-05,
+      "loss": 1.8975,
+      "step": 581
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1971241292304316e-05,
+      "loss": 1.8467,
+      "step": 582
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1971020057411367e-05,
+      "loss": 1.9479,
+      "step": 583
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.197079882251842e-05,
+      "loss": 1.8225,
+      "step": 584
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1970577587625472e-05,
+      "loss": 1.8203,
+      "step": 585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1970356352732525e-05,
+      "loss": 1.8715,
+      "step": 586
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1970135117839578e-05,
+      "loss": 1.7911,
+      "step": 587
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.196991388294663e-05,
+      "loss": 1.9704,
+      "step": 588
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1969692648053681e-05,
+      "loss": 1.7054,
+      "step": 589
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1969471413160732e-05,
+      "loss": 1.9552,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1969250178267785e-05,
+      "loss": 1.7802,
+      "step": 591
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1969028943374837e-05,
+      "loss": 1.6781,
+      "step": 592
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.196880770848189e-05,
+      "loss": 1.9838,
+      "step": 593
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.196858647358894e-05,
+      "loss": 1.7822,
+      "step": 594
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1968365238695993e-05,
+      "loss": 1.7684,
+      "step": 595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1968144003803046e-05,
+      "loss": 1.7029,
+      "step": 596
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1967922768910097e-05,
+      "loss": 1.8811,
+      "step": 597
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.196770153401715e-05,
+      "loss": 1.8499,
+      "step": 598
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1967480299124202e-05,
+      "loss": 1.8912,
+      "step": 599
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1967259064231253e-05,
+      "loss": 1.6229,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1967037829338306e-05,
+      "loss": 1.7749,
+      "step": 601
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1966816594445358e-05,
+      "loss": 1.7644,
+      "step": 602
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1966595359552411e-05,
+      "loss": 1.6864,
+      "step": 603
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1966374124659463e-05,
+      "loss": 1.7691,
+      "step": 604
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1966152889766514e-05,
+      "loss": 1.8626,
+      "step": 605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1965931654873565e-05,
+      "loss": 1.7435,
+      "step": 606
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1965710419980618e-05,
+      "loss": 1.8083,
+      "step": 607
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.196548918508767e-05,
+      "loss": 1.7676,
+      "step": 608
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1965267950194723e-05,
+      "loss": 1.7467,
+      "step": 609
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1965046715301776e-05,
+      "loss": 1.9057,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1964825480408827e-05,
+      "loss": 1.7589,
+      "step": 611
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.196460424551588e-05,
+      "loss": 1.6533,
+      "step": 612
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.196438301062293e-05,
+      "loss": 1.8391,
+      "step": 613
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1964161775729983e-05,
+      "loss": 1.654,
+      "step": 614
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1963940540837035e-05,
+      "loss": 1.8614,
+      "step": 615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1963719305944086e-05,
+      "loss": 1.941,
+      "step": 616
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1963498071051139e-05,
+      "loss": 1.7726,
+      "step": 617
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1963276836158192e-05,
+      "loss": 2.0001,
+      "step": 618
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1963055601265244e-05,
+      "loss": 1.7198,
+      "step": 619
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1962834366372297e-05,
+      "loss": 1.7697,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1962613131479348e-05,
+      "loss": 1.8242,
+      "step": 621
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1962391896586399e-05,
+      "loss": 1.7087,
+      "step": 622
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1962170661693451e-05,
+      "loss": 1.8136,
+      "step": 623
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1961949426800504e-05,
+      "loss": 1.773,
+      "step": 624
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1961728191907556e-05,
+      "loss": 1.8625,
+      "step": 625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1961506957014609e-05,
+      "loss": 1.6965,
+      "step": 626
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.196128572212166e-05,
+      "loss": 1.8636,
+      "step": 627
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1961064487228713e-05,
+      "loss": 1.6231,
+      "step": 628
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1960843252335764e-05,
+      "loss": 1.8446,
+      "step": 629
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1960622017442816e-05,
+      "loss": 1.8904,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1960400782549869e-05,
+      "loss": 1.7709,
+      "step": 631
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1960179547656921e-05,
+      "loss": 1.818,
+      "step": 632
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1959958312763972e-05,
+      "loss": 1.7562,
+      "step": 633
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1959737077871025e-05,
+      "loss": 1.8099,
+      "step": 634
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1959515842978078e-05,
+      "loss": 1.8727,
+      "step": 635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.195929460808513e-05,
+      "loss": 1.8275,
+      "step": 636
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1959073373192181e-05,
+      "loss": 1.7047,
+      "step": 637
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1958852138299232e-05,
+      "loss": 1.8598,
+      "step": 638
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1958630903406285e-05,
+      "loss": 2.0065,
+      "step": 639
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1958409668513337e-05,
+      "loss": 1.675,
+      "step": 640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.195818843362039e-05,
+      "loss": 1.9187,
+      "step": 641
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1957967198727442e-05,
+      "loss": 1.6532,
+      "step": 642
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1957745963834493e-05,
+      "loss": 2.066,
+      "step": 643
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1957524728941546e-05,
+      "loss": 1.8735,
+      "step": 644
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1957303494048599e-05,
+      "loss": 1.6363,
+      "step": 645
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.195708225915565e-05,
+      "loss": 1.6705,
+      "step": 646
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1956861024262702e-05,
+      "loss": 1.8533,
+      "step": 647
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1956639789369755e-05,
+      "loss": 1.5496,
+      "step": 648
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1956418554476806e-05,
+      "loss": 1.8774,
+      "step": 649
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1956197319583858e-05,
+      "loss": 1.8601,
+      "step": 650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.195597608469091e-05,
+      "loss": 1.7863,
+      "step": 651
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1955754849797963e-05,
+      "loss": 1.7216,
+      "step": 652
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1955533614905014e-05,
+      "loss": 1.7468,
+      "step": 653
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1955312380012065e-05,
+      "loss": 1.8638,
+      "step": 654
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1955091145119118e-05,
+      "loss": 1.9798,
+      "step": 655
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.195486991022617e-05,
+      "loss": 1.7882,
+      "step": 656
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1954648675333223e-05,
+      "loss": 1.8907,
+      "step": 657
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1954427440440276e-05,
+      "loss": 1.8342,
+      "step": 658
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1954206205547328e-05,
+      "loss": 1.6836,
+      "step": 659
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.195398497065438e-05,
+      "loss": 1.9493,
+      "step": 660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1953763735761432e-05,
+      "loss": 1.7747,
+      "step": 661
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1953542500868483e-05,
+      "loss": 1.9663,
+      "step": 662
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1953321265975535e-05,
+      "loss": 1.8274,
+      "step": 663
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1953100031082588e-05,
+      "loss": 1.8305,
+      "step": 664
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1952878796189639e-05,
+      "loss": 1.6622,
+      "step": 665
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1952657561296692e-05,
+      "loss": 1.8447,
+      "step": 666
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1952436326403744e-05,
+      "loss": 1.8634,
+      "step": 667
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1952215091510797e-05,
+      "loss": 1.8737,
+      "step": 668
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1951993856617848e-05,
+      "loss": 1.8033,
+      "step": 669
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.19517726217249e-05,
+      "loss": 1.8334,
+      "step": 670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1951551386831951e-05,
+      "loss": 1.7158,
+      "step": 671
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1951330151939004e-05,
+      "loss": 1.7765,
+      "step": 672
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1951108917046056e-05,
+      "loss": 1.6986,
+      "step": 673
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1950887682153109e-05,
+      "loss": 1.7305,
+      "step": 674
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1950666447260162e-05,
+      "loss": 1.8887,
+      "step": 675
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1950445212367213e-05,
+      "loss": 1.6838,
+      "step": 676
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1950223977474265e-05,
+      "loss": 1.8306,
+      "step": 677
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1950002742581316e-05,
+      "loss": 1.9733,
+      "step": 678
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1949781507688369e-05,
+      "loss": 1.7305,
+      "step": 679
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1949560272795421e-05,
+      "loss": 1.7655,
+      "step": 680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1949339037902474e-05,
+      "loss": 1.9006,
+      "step": 681
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1949117803009525e-05,
+      "loss": 1.78,
+      "step": 682
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1948896568116577e-05,
+      "loss": 1.6964,
+      "step": 683
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.194867533322363e-05,
+      "loss": 1.6342,
+      "step": 684
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1948454098330681e-05,
+      "loss": 2.0197,
+      "step": 685
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1948232863437734e-05,
+      "loss": 1.7521,
+      "step": 686
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1948011628544785e-05,
+      "loss": 1.8843,
+      "step": 687
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1947790393651837e-05,
+      "loss": 1.6181,
+      "step": 688
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.194756915875889e-05,
+      "loss": 1.8668,
+      "step": 689
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1947347923865942e-05,
+      "loss": 1.8206,
+      "step": 690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1947126688972995e-05,
+      "loss": 1.7879,
+      "step": 691
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1946905454080048e-05,
+      "loss": 1.6594,
+      "step": 692
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1946684219187099e-05,
+      "loss": 1.7116,
+      "step": 693
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.194646298429415e-05,
+      "loss": 1.9302,
+      "step": 694
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1946241749401202e-05,
+      "loss": 1.7969,
+      "step": 695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1946020514508255e-05,
+      "loss": 1.5646,
+      "step": 696
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1945799279615307e-05,
+      "loss": 1.9959,
+      "step": 697
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1945578044722358e-05,
+      "loss": 1.7048,
+      "step": 698
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.194535680982941e-05,
+      "loss": 1.9736,
+      "step": 699
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1945135574936463e-05,
+      "loss": 1.67,
+      "step": 700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1944914340043514e-05,
+      "loss": 1.8665,
+      "step": 701
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1944693105150567e-05,
+      "loss": 1.76,
+      "step": 702
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.194447187025762e-05,
+      "loss": 1.6851,
+      "step": 703
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.194425063536467e-05,
+      "loss": 1.7348,
+      "step": 704
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1944029400471723e-05,
+      "loss": 1.7796,
+      "step": 705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1943808165578776e-05,
+      "loss": 1.7398,
+      "step": 706
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1943586930685828e-05,
+      "loss": 1.8887,
+      "step": 707
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1943365695792881e-05,
+      "loss": 1.7942,
+      "step": 708
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1943144460899932e-05,
+      "loss": 1.6948,
+      "step": 709
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1942923226006983e-05,
+      "loss": 1.8689,
+      "step": 710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1942701991114035e-05,
+      "loss": 1.7016,
+      "step": 711
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1942480756221088e-05,
+      "loss": 1.7724,
+      "step": 712
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.194225952132814e-05,
+      "loss": 1.6469,
+      "step": 713
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1942038286435193e-05,
+      "loss": 1.6649,
+      "step": 714
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1941817051542244e-05,
+      "loss": 1.6566,
+      "step": 715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1941595816649297e-05,
+      "loss": 1.9367,
+      "step": 716
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1941374581756348e-05,
+      "loss": 1.9762,
+      "step": 717
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.19411533468634e-05,
+      "loss": 1.7355,
+      "step": 718
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1940932111970453e-05,
+      "loss": 1.721,
+      "step": 719
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1940710877077504e-05,
+      "loss": 1.8193,
+      "step": 720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1940489642184556e-05,
+      "loss": 1.5788,
+      "step": 721
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1940268407291609e-05,
+      "loss": 1.8623,
+      "step": 722
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1940047172398662e-05,
+      "loss": 1.8758,
+      "step": 723
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1939825937505714e-05,
+      "loss": 1.9503,
+      "step": 724
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1939604702612765e-05,
+      "loss": 1.8394,
+      "step": 725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1939383467719816e-05,
+      "loss": 1.7832,
+      "step": 726
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1939162232826869e-05,
+      "loss": 1.7016,
+      "step": 727
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1938940997933921e-05,
+      "loss": 1.7279,
+      "step": 728
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1938719763040974e-05,
+      "loss": 1.6689,
+      "step": 729
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1938498528148027e-05,
+      "loss": 1.8168,
+      "step": 730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1938277293255077e-05,
+      "loss": 1.6908,
+      "step": 731
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.193805605836213e-05,
+      "loss": 1.7178,
+      "step": 732
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1937834823469183e-05,
+      "loss": 1.6441,
+      "step": 733
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1937613588576234e-05,
+      "loss": 1.4902,
+      "step": 734
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1937392353683286e-05,
+      "loss": 1.8743,
+      "step": 735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1937171118790337e-05,
+      "loss": 1.8876,
+      "step": 736
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.193694988389739e-05,
+      "loss": 1.8034,
+      "step": 737
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1936728649004442e-05,
+      "loss": 1.7743,
+      "step": 738
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1936507414111495e-05,
+      "loss": 1.9182,
+      "step": 739
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1936286179218548e-05,
+      "loss": 1.8462,
+      "step": 740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1936064944325598e-05,
+      "loss": 1.8658,
+      "step": 741
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.193584370943265e-05,
+      "loss": 1.7159,
+      "step": 742
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1935622474539702e-05,
+      "loss": 1.629,
+      "step": 743
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1935401239646755e-05,
+      "loss": 1.776,
+      "step": 744
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1935180004753807e-05,
+      "loss": 1.7877,
+      "step": 745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.193495876986086e-05,
+      "loss": 1.5661,
+      "step": 746
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.193473753496791e-05,
+      "loss": 1.6865,
+      "step": 747
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1934516300074963e-05,
+      "loss": 1.9324,
+      "step": 748
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1934295065182016e-05,
+      "loss": 1.7774,
+      "step": 749
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1934073830289067e-05,
+      "loss": 1.7717,
+      "step": 750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.193385259539612e-05,
+      "loss": 1.7717,
+      "step": 751
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1933631360503172e-05,
+      "loss": 1.7904,
+      "step": 752
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1933410125610223e-05,
+      "loss": 1.6102,
+      "step": 753
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1933188890717276e-05,
+      "loss": 1.7395,
+      "step": 754
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1932967655824328e-05,
+      "loss": 1.7517,
+      "step": 755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1932746420931381e-05,
+      "loss": 1.7857,
+      "step": 756
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1932525186038432e-05,
+      "loss": 1.6827,
+      "step": 757
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1932303951145483e-05,
+      "loss": 1.5824,
+      "step": 758
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1932082716252535e-05,
+      "loss": 1.6816,
+      "step": 759
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1931861481359588e-05,
+      "loss": 1.6733,
+      "step": 760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.193164024646664e-05,
+      "loss": 1.7992,
+      "step": 761
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1931419011573693e-05,
+      "loss": 1.7233,
+      "step": 762
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1931197776680746e-05,
+      "loss": 1.7918,
+      "step": 763
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1930976541787797e-05,
+      "loss": 1.8969,
+      "step": 764
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.193075530689485e-05,
+      "loss": 1.7503,
+      "step": 765
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.19305340720019e-05,
+      "loss": 1.7614,
+      "step": 766
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1930312837108953e-05,
+      "loss": 1.7467,
+      "step": 767
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1930091602216005e-05,
+      "loss": 1.7097,
+      "step": 768
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1929870367323056e-05,
+      "loss": 1.7648,
+      "step": 769
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1929649132430109e-05,
+      "loss": 1.7165,
+      "step": 770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1929427897537162e-05,
+      "loss": 1.6739,
+      "step": 771
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1929206662644214e-05,
+      "loss": 1.7839,
+      "step": 772
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1928985427751265e-05,
+      "loss": 1.6038,
+      "step": 773
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1928764192858318e-05,
+      "loss": 1.7033,
+      "step": 774
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1928542957965369e-05,
+      "loss": 1.8716,
+      "step": 775
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1928321723072421e-05,
+      "loss": 1.7298,
+      "step": 776
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1928100488179474e-05,
+      "loss": 1.6073,
+      "step": 777
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1927879253286526e-05,
+      "loss": 1.9102,
+      "step": 778
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1927658018393579e-05,
+      "loss": 1.7041,
+      "step": 779
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.192743678350063e-05,
+      "loss": 1.7373,
+      "step": 780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1927215548607683e-05,
+      "loss": 1.7944,
+      "step": 781
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1926994313714734e-05,
+      "loss": 1.6113,
+      "step": 782
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1926773078821786e-05,
+      "loss": 1.7746,
+      "step": 783
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1926551843928839e-05,
+      "loss": 1.9446,
+      "step": 784
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1926330609035891e-05,
+      "loss": 1.7319,
+      "step": 785
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1926109374142942e-05,
+      "loss": 1.6554,
+      "step": 786
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1925888139249995e-05,
+      "loss": 1.7357,
+      "step": 787
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1925666904357047e-05,
+      "loss": 1.7777,
+      "step": 788
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1925445669464098e-05,
+      "loss": 1.6847,
+      "step": 789
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1925224434571151e-05,
+      "loss": 1.8348,
+      "step": 790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1925003199678202e-05,
+      "loss": 1.6002,
+      "step": 791
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1924781964785255e-05,
+      "loss": 1.7908,
+      "step": 792
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1924560729892307e-05,
+      "loss": 1.8806,
+      "step": 793
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.192433949499936e-05,
+      "loss": 1.6721,
+      "step": 794
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1924118260106412e-05,
+      "loss": 1.6242,
+      "step": 795
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1923897025213465e-05,
+      "loss": 1.642,
+      "step": 796
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1923675790320516e-05,
+      "loss": 1.8211,
+      "step": 797
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1923454555427567e-05,
+      "loss": 1.8474,
+      "step": 798
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.192323332053462e-05,
+      "loss": 1.7889,
+      "step": 799
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1923012085641672e-05,
+      "loss": 1.6975,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1922790850748725e-05,
+      "loss": 1.6823,
+      "step": 801
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1922569615855776e-05,
+      "loss": 1.516,
+      "step": 802
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1922348380962828e-05,
+      "loss": 1.6646,
+      "step": 803
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.192212714606988e-05,
+      "loss": 1.8115,
+      "step": 804
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1921905911176932e-05,
+      "loss": 1.7377,
+      "step": 805
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1921684676283984e-05,
+      "loss": 1.6636,
+      "step": 806
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1921463441391037e-05,
+      "loss": 1.8727,
+      "step": 807
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1921242206498088e-05,
+      "loss": 1.9068,
+      "step": 808
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.192102097160514e-05,
+      "loss": 1.6938,
+      "step": 809
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1920799736712193e-05,
+      "loss": 1.7536,
+      "step": 810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1920578501819246e-05,
+      "loss": 1.6906,
+      "step": 811
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1920357266926298e-05,
+      "loss": 1.713,
+      "step": 812
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.192013603203335e-05,
+      "loss": 1.689,
+      "step": 813
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.19199147971404e-05,
+      "loss": 1.7188,
+      "step": 814
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1919693562247453e-05,
+      "loss": 1.6583,
+      "step": 815
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1919472327354505e-05,
+      "loss": 1.6888,
+      "step": 816
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1919251092461558e-05,
+      "loss": 1.5497,
+      "step": 817
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.191902985756861e-05,
+      "loss": 1.4545,
+      "step": 818
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1918808622675662e-05,
+      "loss": 1.646,
+      "step": 819
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1918587387782714e-05,
+      "loss": 1.7241,
+      "step": 820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1918366152889767e-05,
+      "loss": 1.7051,
+      "step": 821
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1918144917996818e-05,
+      "loss": 1.9028,
+      "step": 822
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.191792368310387e-05,
+      "loss": 1.6897,
+      "step": 823
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1917702448210921e-05,
+      "loss": 1.7959,
+      "step": 824
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1917481213317974e-05,
+      "loss": 1.7387,
+      "step": 825
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1917259978425026e-05,
+      "loss": 1.6997,
+      "step": 826
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1917038743532079e-05,
+      "loss": 1.71,
+      "step": 827
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1916817508639132e-05,
+      "loss": 1.6626,
+      "step": 828
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1916596273746183e-05,
+      "loss": 1.7942,
+      "step": 829
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1916375038853233e-05,
+      "loss": 1.839,
+      "step": 830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1916153803960286e-05,
+      "loss": 1.7805,
+      "step": 831
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1915932569067339e-05,
+      "loss": 1.4959,
+      "step": 832
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1915711334174391e-05,
+      "loss": 1.7486,
+      "step": 833
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1915490099281444e-05,
+      "loss": 1.7959,
+      "step": 834
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1915268864388495e-05,
+      "loss": 1.7179,
+      "step": 835
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1915047629495547e-05,
+      "loss": 1.6691,
+      "step": 836
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.19148263946026e-05,
+      "loss": 1.6311,
+      "step": 837
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1914605159709651e-05,
+      "loss": 1.6578,
+      "step": 838
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1914383924816704e-05,
+      "loss": 1.8024,
+      "step": 839
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1914162689923755e-05,
+      "loss": 1.6295,
+      "step": 840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1913941455030807e-05,
+      "loss": 1.8069,
+      "step": 841
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.191372022013786e-05,
+      "loss": 1.9045,
+      "step": 842
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1913498985244912e-05,
+      "loss": 1.7291,
+      "step": 843
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1913277750351965e-05,
+      "loss": 1.7648,
+      "step": 844
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1913056515459016e-05,
+      "loss": 1.94,
+      "step": 845
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1912835280566067e-05,
+      "loss": 1.7404,
+      "step": 846
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.191261404567312e-05,
+      "loss": 1.7535,
+      "step": 847
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1912392810780172e-05,
+      "loss": 1.7499,
+      "step": 848
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1912171575887225e-05,
+      "loss": 1.6599,
+      "step": 849
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1911950340994277e-05,
+      "loss": 1.7683,
+      "step": 850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1911729106101328e-05,
+      "loss": 1.7076,
+      "step": 851
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.191150787120838e-05,
+      "loss": 1.7784,
+      "step": 852
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1911286636315433e-05,
+      "loss": 1.6835,
+      "step": 853
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1911065401422484e-05,
+      "loss": 1.7836,
+      "step": 854
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1910844166529537e-05,
+      "loss": 1.8203,
+      "step": 855
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.191062293163659e-05,
+      "loss": 1.6515,
+      "step": 856
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.191040169674364e-05,
+      "loss": 1.6413,
+      "step": 857
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1910180461850693e-05,
+      "loss": 1.6568,
+      "step": 858
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1909959226957746e-05,
+      "loss": 1.6478,
+      "step": 859
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1909737992064798e-05,
+      "loss": 1.5361,
+      "step": 860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.190951675717185e-05,
+      "loss": 1.733,
+      "step": 861
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.19092955222789e-05,
+      "loss": 1.6886,
+      "step": 862
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1909074287385953e-05,
+      "loss": 1.8457,
+      "step": 863
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1908853052493005e-05,
+      "loss": 1.6941,
+      "step": 864
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1908631817600058e-05,
+      "loss": 1.6631,
+      "step": 865
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.190841058270711e-05,
+      "loss": 1.7079,
+      "step": 866
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1908189347814163e-05,
+      "loss": 1.8035,
+      "step": 867
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1907968112921214e-05,
+      "loss": 1.6662,
+      "step": 868
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1907746878028267e-05,
+      "loss": 1.6819,
+      "step": 869
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1907525643135318e-05,
+      "loss": 1.78,
+      "step": 870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.190730440824237e-05,
+      "loss": 1.6904,
+      "step": 871
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1907083173349423e-05,
+      "loss": 1.7814,
+      "step": 872
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1906861938456474e-05,
+      "loss": 1.7996,
+      "step": 873
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1906640703563526e-05,
+      "loss": 1.5813,
+      "step": 874
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1906419468670579e-05,
+      "loss": 1.6793,
+      "step": 875
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1906198233777632e-05,
+      "loss": 1.7956,
+      "step": 876
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1905976998884683e-05,
+      "loss": 1.5759,
+      "step": 877
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1905755763991735e-05,
+      "loss": 1.8151,
+      "step": 878
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1905534529098786e-05,
+      "loss": 1.7258,
+      "step": 879
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1905313294205839e-05,
+      "loss": 1.6973,
+      "step": 880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1905092059312891e-05,
+      "loss": 1.7906,
+      "step": 881
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1904870824419944e-05,
+      "loss": 1.6517,
+      "step": 882
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1904649589526996e-05,
+      "loss": 1.9065,
+      "step": 883
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1904428354634047e-05,
+      "loss": 1.7012,
+      "step": 884
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.19042071197411e-05,
+      "loss": 1.7053,
+      "step": 885
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1903985884848151e-05,
+      "loss": 1.7651,
+      "step": 886
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1903764649955204e-05,
+      "loss": 1.9063,
+      "step": 887
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1903543415062256e-05,
+      "loss": 1.8646,
+      "step": 888
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1903322180169309e-05,
+      "loss": 1.5881,
+      "step": 889
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.190310094527636e-05,
+      "loss": 1.9981,
+      "step": 890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1902879710383412e-05,
+      "loss": 1.8703,
+      "step": 891
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1902658475490465e-05,
+      "loss": 1.7378,
+      "step": 892
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1902437240597516e-05,
+      "loss": 1.5968,
+      "step": 893
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1902216005704568e-05,
+      "loss": 1.7235,
+      "step": 894
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.190199477081162e-05,
+      "loss": 1.6335,
+      "step": 895
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1901773535918672e-05,
+      "loss": 1.5456,
+      "step": 896
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1901552301025725e-05,
+      "loss": 1.7009,
+      "step": 897
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1901331066132777e-05,
+      "loss": 1.8051,
+      "step": 898
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.190110983123983e-05,
+      "loss": 1.6499,
+      "step": 899
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1900888596346882e-05,
+      "loss": 1.5866,
+      "step": 900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1900667361453933e-05,
+      "loss": 1.7845,
+      "step": 901
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1900446126560984e-05,
+      "loss": 1.7024,
+      "step": 902
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1900224891668037e-05,
+      "loss": 1.7852,
+      "step": 903
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.190000365677509e-05,
+      "loss": 1.6776,
+      "step": 904
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1899782421882142e-05,
+      "loss": 1.6227,
+      "step": 905
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1899561186989193e-05,
+      "loss": 1.7685,
+      "step": 906
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1899339952096246e-05,
+      "loss": 1.6078,
+      "step": 907
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1899118717203298e-05,
+      "loss": 1.7715,
+      "step": 908
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.189889748231035e-05,
+      "loss": 1.7254,
+      "step": 909
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1898676247417402e-05,
+      "loss": 1.7372,
+      "step": 910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1898455012524454e-05,
+      "loss": 1.6885,
+      "step": 911
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1898233777631505e-05,
+      "loss": 1.8769,
+      "step": 912
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1898012542738558e-05,
+      "loss": 1.6156,
+      "step": 913
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.189779130784561e-05,
+      "loss": 1.7405,
+      "step": 914
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1897570072952663e-05,
+      "loss": 1.7053,
+      "step": 915
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1897348838059716e-05,
+      "loss": 1.6518,
+      "step": 916
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1897127603166767e-05,
+      "loss": 1.7365,
+      "step": 917
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1896906368273818e-05,
+      "loss": 1.704,
+      "step": 918
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.189668513338087e-05,
+      "loss": 1.7273,
+      "step": 919
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1896463898487923e-05,
+      "loss": 1.5147,
+      "step": 920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1896242663594975e-05,
+      "loss": 1.7125,
+      "step": 921
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1896021428702028e-05,
+      "loss": 1.7563,
+      "step": 922
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1895800193809079e-05,
+      "loss": 1.7711,
+      "step": 923
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1895578958916132e-05,
+      "loss": 1.5452,
+      "step": 924
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1895357724023184e-05,
+      "loss": 1.6344,
+      "step": 925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1895136489130235e-05,
+      "loss": 1.7968,
+      "step": 926
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1894915254237288e-05,
+      "loss": 1.6071,
+      "step": 927
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1894694019344339e-05,
+      "loss": 1.7251,
+      "step": 928
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1894472784451391e-05,
+      "loss": 1.764,
+      "step": 929
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1894251549558444e-05,
+      "loss": 1.7152,
+      "step": 930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1894030314665496e-05,
+      "loss": 1.7365,
+      "step": 931
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1893809079772549e-05,
+      "loss": 1.7158,
+      "step": 932
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.18935878448796e-05,
+      "loss": 1.7477,
+      "step": 933
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1893366609986651e-05,
+      "loss": 1.7816,
+      "step": 934
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1893145375093704e-05,
+      "loss": 1.7759,
+      "step": 935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1892924140200756e-05,
+      "loss": 1.688,
+      "step": 936
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1892702905307809e-05,
+      "loss": 1.7261,
+      "step": 937
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1892481670414861e-05,
+      "loss": 1.7297,
+      "step": 938
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1892260435521912e-05,
+      "loss": 1.7129,
+      "step": 939
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1892039200628965e-05,
+      "loss": 1.7553,
+      "step": 940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1891817965736017e-05,
+      "loss": 1.4814,
+      "step": 941
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1891596730843068e-05,
+      "loss": 1.7398,
+      "step": 942
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1891375495950121e-05,
+      "loss": 1.5658,
+      "step": 943
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1891154261057172e-05,
+      "loss": 1.8237,
+      "step": 944
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1890933026164225e-05,
+      "loss": 1.7541,
+      "step": 945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1890711791271277e-05,
+      "loss": 1.5975,
+      "step": 946
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.189049055637833e-05,
+      "loss": 1.5441,
+      "step": 947
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1890269321485382e-05,
+      "loss": 1.6359,
+      "step": 948
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1890048086592433e-05,
+      "loss": 1.549,
+      "step": 949
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1889826851699484e-05,
+      "loss": 1.7403,
+      "step": 950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1889605616806537e-05,
+      "loss": 1.6739,
+      "step": 951
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.188938438191359e-05,
+      "loss": 1.5686,
+      "step": 952
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1889163147020642e-05,
+      "loss": 1.5181,
+      "step": 953
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1888941912127695e-05,
+      "loss": 1.613,
+      "step": 954
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1888720677234746e-05,
+      "loss": 1.5635,
+      "step": 955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1888499442341798e-05,
+      "loss": 1.6362,
+      "step": 956
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.188827820744885e-05,
+      "loss": 1.6551,
+      "step": 957
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1888056972555902e-05,
+      "loss": 1.8393,
+      "step": 958
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1887835737662954e-05,
+      "loss": 1.6344,
+      "step": 959
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1887614502770007e-05,
+      "loss": 1.5098,
+      "step": 960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1887393267877058e-05,
+      "loss": 1.792,
+      "step": 961
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.188717203298411e-05,
+      "loss": 1.8036,
+      "step": 962
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1886950798091163e-05,
+      "loss": 1.7222,
+      "step": 963
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1886729563198216e-05,
+      "loss": 1.634,
+      "step": 964
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1886508328305267e-05,
+      "loss": 1.8661,
+      "step": 965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.188628709341232e-05,
+      "loss": 1.79,
+      "step": 966
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.188606585851937e-05,
+      "loss": 1.822,
+      "step": 967
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1885844623626423e-05,
+      "loss": 1.716,
+      "step": 968
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1885623388733475e-05,
+      "loss": 1.6704,
+      "step": 969
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1885402153840528e-05,
+      "loss": 1.675,
+      "step": 970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.188518091894758e-05,
+      "loss": 1.6141,
+      "step": 971
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1884959684054632e-05,
+      "loss": 1.5948,
+      "step": 972
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1884738449161684e-05,
+      "loss": 1.7245,
+      "step": 973
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1884517214268735e-05,
+      "loss": 1.6879,
+      "step": 974
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1884295979375788e-05,
+      "loss": 1.6657,
+      "step": 975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.188407474448284e-05,
+      "loss": 1.55,
+      "step": 976
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1883853509589891e-05,
+      "loss": 1.7328,
+      "step": 977
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1883632274696944e-05,
+      "loss": 1.5977,
+      "step": 978
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1883411039803996e-05,
+      "loss": 1.6002,
+      "step": 979
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1883189804911049e-05,
+      "loss": 1.5748,
+      "step": 980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.18829685700181e-05,
+      "loss": 1.6865,
+      "step": 981
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1882747335125153e-05,
+      "loss": 1.7054,
+      "step": 982
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1882526100232203e-05,
+      "loss": 1.6075,
+      "step": 983
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1882304865339256e-05,
+      "loss": 1.5778,
+      "step": 984
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1882083630446309e-05,
+      "loss": 1.6772,
+      "step": 985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1881862395553361e-05,
+      "loss": 1.7636,
+      "step": 986
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1881641160660414e-05,
+      "loss": 1.6753,
+      "step": 987
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1881419925767465e-05,
+      "loss": 1.6938,
+      "step": 988
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1881198690874517e-05,
+      "loss": 1.6548,
+      "step": 989
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1880977455981568e-05,
+      "loss": 1.7519,
+      "step": 990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1880756221088621e-05,
+      "loss": 1.6785,
+      "step": 991
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1880534986195674e-05,
+      "loss": 1.6723,
+      "step": 992
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1880313751302726e-05,
+      "loss": 1.8269,
+      "step": 993
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1880092516409777e-05,
+      "loss": 1.6141,
+      "step": 994
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.187987128151683e-05,
+      "loss": 1.6374,
+      "step": 995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1879650046623882e-05,
+      "loss": 1.7318,
+      "step": 996
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1879428811730935e-05,
+      "loss": 1.7491,
+      "step": 997
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1879207576837986e-05,
+      "loss": 1.7749,
+      "step": 998
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1878986341945037e-05,
+      "loss": 1.7846,
+      "step": 999
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.187876510705209e-05,
+      "loss": 1.7536,
+      "step": 1000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1878543872159142e-05,
+      "loss": 1.5872,
+      "step": 1001
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1878322637266195e-05,
+      "loss": 1.7751,
+      "step": 1002
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1878101402373247e-05,
+      "loss": 1.5269,
+      "step": 1003
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.18778801674803e-05,
+      "loss": 1.8196,
+      "step": 1004
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.187765893258735e-05,
+      "loss": 1.7619,
+      "step": 1005
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1877437697694402e-05,
+      "loss": 1.7308,
+      "step": 1006
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1877216462801454e-05,
+      "loss": 1.6676,
+      "step": 1007
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1876995227908507e-05,
+      "loss": 1.6622,
+      "step": 1008
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.187677399301556e-05,
+      "loss": 1.7357,
+      "step": 1009
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.187655275812261e-05,
+      "loss": 1.5753,
+      "step": 1010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1876331523229663e-05,
+      "loss": 1.5509,
+      "step": 1011
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1876110288336716e-05,
+      "loss": 1.608,
+      "step": 1012
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1875889053443768e-05,
+      "loss": 1.7923,
+      "step": 1013
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.187566781855082e-05,
+      "loss": 1.6366,
+      "step": 1014
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1875446583657872e-05,
+      "loss": 1.6283,
+      "step": 1015
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1875225348764923e-05,
+      "loss": 1.6851,
+      "step": 1016
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1875004113871975e-05,
+      "loss": 1.7311,
+      "step": 1017
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1874782878979028e-05,
+      "loss": 1.6823,
+      "step": 1018
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.187456164408608e-05,
+      "loss": 1.6456,
+      "step": 1019
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1874340409193133e-05,
+      "loss": 1.6188,
+      "step": 1020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1874119174300184e-05,
+      "loss": 1.8024,
+      "step": 1021
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1873897939407235e-05,
+      "loss": 1.5292,
+      "step": 1022
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1873676704514288e-05,
+      "loss": 1.5739,
+      "step": 1023
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.187345546962134e-05,
+      "loss": 1.7221,
+      "step": 1024
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1873234234728393e-05,
+      "loss": 1.7136,
+      "step": 1025
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1873012999835445e-05,
+      "loss": 1.6437,
+      "step": 1026
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1872791764942496e-05,
+      "loss": 1.8151,
+      "step": 1027
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1872570530049549e-05,
+      "loss": 1.6109,
+      "step": 1028
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1872349295156602e-05,
+      "loss": 1.6463,
+      "step": 1029
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1872128060263653e-05,
+      "loss": 1.7979,
+      "step": 1030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1871906825370705e-05,
+      "loss": 1.7308,
+      "step": 1031
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1871685590477756e-05,
+      "loss": 1.6475,
+      "step": 1032
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1871464355584809e-05,
+      "loss": 1.7582,
+      "step": 1033
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1871243120691861e-05,
+      "loss": 1.7061,
+      "step": 1034
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1871021885798914e-05,
+      "loss": 1.6944,
+      "step": 1035
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1870800650905966e-05,
+      "loss": 1.7601,
+      "step": 1036
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1870579416013017e-05,
+      "loss": 1.7164,
+      "step": 1037
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1870358181120068e-05,
+      "loss": 1.7463,
+      "step": 1038
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1870136946227121e-05,
+      "loss": 1.4832,
+      "step": 1039
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1869915711334174e-05,
+      "loss": 1.8391,
+      "step": 1040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1869694476441226e-05,
+      "loss": 1.8122,
+      "step": 1041
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1869473241548279e-05,
+      "loss": 1.7825,
+      "step": 1042
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.186925200665533e-05,
+      "loss": 1.7604,
+      "step": 1043
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1869030771762382e-05,
+      "loss": 1.7952,
+      "step": 1044
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1868809536869435e-05,
+      "loss": 1.6598,
+      "step": 1045
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1868588301976486e-05,
+      "loss": 1.7321,
+      "step": 1046
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1868367067083538e-05,
+      "loss": 1.5773,
+      "step": 1047
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.186814583219059e-05,
+      "loss": 1.7333,
+      "step": 1048
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1867924597297642e-05,
+      "loss": 1.6652,
+      "step": 1049
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1867703362404695e-05,
+      "loss": 1.7759,
+      "step": 1050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1867482127511747e-05,
+      "loss": 1.7027,
+      "step": 1051
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.18672608926188e-05,
+      "loss": 1.5497,
+      "step": 1052
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.186703965772585e-05,
+      "loss": 1.604,
+      "step": 1053
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1866818422832903e-05,
+      "loss": 1.6149,
+      "step": 1054
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1866597187939954e-05,
+      "loss": 1.5908,
+      "step": 1055
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1866375953047007e-05,
+      "loss": 1.6899,
+      "step": 1056
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.186615471815406e-05,
+      "loss": 1.6732,
+      "step": 1057
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1865933483261112e-05,
+      "loss": 1.583,
+      "step": 1058
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1865712248368163e-05,
+      "loss": 1.6589,
+      "step": 1059
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1865491013475216e-05,
+      "loss": 1.6791,
+      "step": 1060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1865269778582268e-05,
+      "loss": 1.6584,
+      "step": 1061
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1865048543689319e-05,
+      "loss": 1.617,
+      "step": 1062
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1864827308796372e-05,
+      "loss": 1.7217,
+      "step": 1063
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1864606073903424e-05,
+      "loss": 1.5971,
+      "step": 1064
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1864384839010475e-05,
+      "loss": 1.7805,
+      "step": 1065
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1864163604117528e-05,
+      "loss": 1.7462,
+      "step": 1066
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.186394236922458e-05,
+      "loss": 1.6112,
+      "step": 1067
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1863721134331633e-05,
+      "loss": 1.7442,
+      "step": 1068
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1863499899438684e-05,
+      "loss": 1.7133,
+      "step": 1069
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1863278664545737e-05,
+      "loss": 1.6261,
+      "step": 1070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1863057429652788e-05,
+      "loss": 1.5674,
+      "step": 1071
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.186283619475984e-05,
+      "loss": 1.7145,
+      "step": 1072
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1862614959866893e-05,
+      "loss": 1.7811,
+      "step": 1073
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1862393724973945e-05,
+      "loss": 1.4586,
+      "step": 1074
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1862172490080998e-05,
+      "loss": 1.6075,
+      "step": 1075
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1861951255188049e-05,
+      "loss": 1.5616,
+      "step": 1076
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1861730020295102e-05,
+      "loss": 1.6931,
+      "step": 1077
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1861508785402152e-05,
+      "loss": 1.8085,
+      "step": 1078
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1861287550509205e-05,
+      "loss": 1.5359,
+      "step": 1079
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1861066315616258e-05,
+      "loss": 1.5455,
+      "step": 1080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1860845080723309e-05,
+      "loss": 1.762,
+      "step": 1081
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1860623845830361e-05,
+      "loss": 1.6682,
+      "step": 1082
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1860402610937414e-05,
+      "loss": 1.5727,
+      "step": 1083
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1860181376044466e-05,
+      "loss": 1.6908,
+      "step": 1084
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1859960141151519e-05,
+      "loss": 1.6305,
+      "step": 1085
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.185973890625857e-05,
+      "loss": 1.7485,
+      "step": 1086
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1859517671365621e-05,
+      "loss": 1.7587,
+      "step": 1087
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1859296436472673e-05,
+      "loss": 1.5869,
+      "step": 1088
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1859075201579726e-05,
+      "loss": 1.6523,
+      "step": 1089
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1858853966686779e-05,
+      "loss": 1.7167,
+      "step": 1090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1858632731793831e-05,
+      "loss": 1.7584,
+      "step": 1091
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1858411496900882e-05,
+      "loss": 1.7154,
+      "step": 1092
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1858190262007935e-05,
+      "loss": 1.7422,
+      "step": 1093
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1857969027114986e-05,
+      "loss": 1.6419,
+      "step": 1094
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1857747792222038e-05,
+      "loss": 1.7074,
+      "step": 1095
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1857526557329091e-05,
+      "loss": 1.5565,
+      "step": 1096
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1857305322436144e-05,
+      "loss": 1.579,
+      "step": 1097
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1857084087543195e-05,
+      "loss": 1.6454,
+      "step": 1098
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1856862852650247e-05,
+      "loss": 1.7104,
+      "step": 1099
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.18566416177573e-05,
+      "loss": 1.6267,
+      "step": 1100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1856420382864352e-05,
+      "loss": 1.7952,
+      "step": 1101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1856199147971403e-05,
+      "loss": 1.766,
+      "step": 1102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1855977913078454e-05,
+      "loss": 1.6254,
+      "step": 1103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1855756678185507e-05,
+      "loss": 1.61,
+      "step": 1104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.185553544329256e-05,
+      "loss": 1.5639,
+      "step": 1105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1855314208399612e-05,
+      "loss": 1.4915,
+      "step": 1106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1855092973506665e-05,
+      "loss": 1.5156,
+      "step": 1107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1854871738613717e-05,
+      "loss": 1.6214,
+      "step": 1108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1854650503720768e-05,
+      "loss": 1.6903,
+      "step": 1109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1854429268827819e-05,
+      "loss": 1.5621,
+      "step": 1110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1854208033934872e-05,
+      "loss": 1.7164,
+      "step": 1111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1853986799041924e-05,
+      "loss": 1.6115,
+      "step": 1112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1853765564148977e-05,
+      "loss": 1.6458,
+      "step": 1113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1853544329256028e-05,
+      "loss": 1.6333,
+      "step": 1114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.185332309436308e-05,
+      "loss": 1.5739,
+      "step": 1115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1853101859470133e-05,
+      "loss": 1.6178,
+      "step": 1116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1852880624577186e-05,
+      "loss": 1.6513,
+      "step": 1117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1852659389684237e-05,
+      "loss": 1.5956,
+      "step": 1118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.185243815479129e-05,
+      "loss": 1.6297,
+      "step": 1119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.185221691989834e-05,
+      "loss": 1.6358,
+      "step": 1120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1851995685005393e-05,
+      "loss": 1.5679,
+      "step": 1121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1851774450112445e-05,
+      "loss": 1.5686,
+      "step": 1122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1851553215219498e-05,
+      "loss": 1.7057,
+      "step": 1123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.185133198032655e-05,
+      "loss": 1.7268,
+      "step": 1124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1851110745433601e-05,
+      "loss": 1.8466,
+      "step": 1125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1850889510540652e-05,
+      "loss": 1.5541,
+      "step": 1126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1850668275647705e-05,
+      "loss": 1.6451,
+      "step": 1127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1850447040754758e-05,
+      "loss": 1.5237,
+      "step": 1128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.185022580586181e-05,
+      "loss": 1.805,
+      "step": 1129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1850004570968861e-05,
+      "loss": 1.676,
+      "step": 1130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1849783336075914e-05,
+      "loss": 1.6236,
+      "step": 1131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1849562101182966e-05,
+      "loss": 1.8026,
+      "step": 1132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1849340866290019e-05,
+      "loss": 1.6943,
+      "step": 1133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.184911963139707e-05,
+      "loss": 1.8,
+      "step": 1134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1848898396504123e-05,
+      "loss": 1.5439,
+      "step": 1135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1848677161611173e-05,
+      "loss": 1.5708,
+      "step": 1136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1848455926718226e-05,
+      "loss": 1.7215,
+      "step": 1137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1848234691825279e-05,
+      "loss": 1.5964,
+      "step": 1138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1848013456932331e-05,
+      "loss": 1.7039,
+      "step": 1139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1847792222039384e-05,
+      "loss": 1.518,
+      "step": 1140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1847570987146435e-05,
+      "loss": 1.6829,
+      "step": 1141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1847349752253487e-05,
+      "loss": 1.7711,
+      "step": 1142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1847128517360538e-05,
+      "loss": 1.7974,
+      "step": 1143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1846907282467591e-05,
+      "loss": 1.7143,
+      "step": 1144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1846686047574644e-05,
+      "loss": 1.537,
+      "step": 1145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1846464812681696e-05,
+      "loss": 1.7628,
+      "step": 1146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1846243577788747e-05,
+      "loss": 1.6627,
+      "step": 1147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.18460223428958e-05,
+      "loss": 1.716,
+      "step": 1148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1845801108002852e-05,
+      "loss": 1.6868,
+      "step": 1149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1845579873109903e-05,
+      "loss": 1.6424,
+      "step": 1150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1845358638216956e-05,
+      "loss": 1.7144,
+      "step": 1151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1845137403324007e-05,
+      "loss": 1.609,
+      "step": 1152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.184491616843106e-05,
+      "loss": 1.4672,
+      "step": 1153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1844694933538112e-05,
+      "loss": 1.7574,
+      "step": 1154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1844473698645165e-05,
+      "loss": 1.5722,
+      "step": 1155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1844252463752217e-05,
+      "loss": 1.5956,
+      "step": 1156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1844031228859268e-05,
+      "loss": 1.6334,
+      "step": 1157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.184380999396632e-05,
+      "loss": 1.7528,
+      "step": 1158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1843588759073372e-05,
+      "loss": 1.555,
+      "step": 1159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1843367524180424e-05,
+      "loss": 1.6905,
+      "step": 1160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1843146289287477e-05,
+      "loss": 1.7256,
+      "step": 1161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.184292505439453e-05,
+      "loss": 1.6957,
+      "step": 1162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.184270381950158e-05,
+      "loss": 1.654,
+      "step": 1163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1842482584608633e-05,
+      "loss": 1.7574,
+      "step": 1164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1842261349715686e-05,
+      "loss": 1.7662,
+      "step": 1165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1842040114822737e-05,
+      "loss": 1.6199,
+      "step": 1166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.184181887992979e-05,
+      "loss": 1.6119,
+      "step": 1167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1841597645036842e-05,
+      "loss": 1.5872,
+      "step": 1168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1841376410143893e-05,
+      "loss": 1.6324,
+      "step": 1169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1841155175250945e-05,
+      "loss": 1.6642,
+      "step": 1170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1840933940357998e-05,
+      "loss": 1.6291,
+      "step": 1171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.184071270546505e-05,
+      "loss": 1.651,
+      "step": 1172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1840491470572103e-05,
+      "loss": 1.701,
+      "step": 1173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1840270235679154e-05,
+      "loss": 1.6465,
+      "step": 1174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1840049000786205e-05,
+      "loss": 1.618,
+      "step": 1175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1839827765893258e-05,
+      "loss": 1.7984,
+      "step": 1176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.183960653100031e-05,
+      "loss": 1.5037,
+      "step": 1177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1839385296107363e-05,
+      "loss": 1.4829,
+      "step": 1178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1839164061214415e-05,
+      "loss": 1.6836,
+      "step": 1179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1838942826321466e-05,
+      "loss": 1.6871,
+      "step": 1180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1838721591428519e-05,
+      "loss": 1.6461,
+      "step": 1181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.183850035653557e-05,
+      "loss": 1.4906,
+      "step": 1182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1838279121642622e-05,
+      "loss": 1.391,
+      "step": 1183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1838057886749675e-05,
+      "loss": 1.5128,
+      "step": 1184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1837836651856726e-05,
+      "loss": 1.7679,
+      "step": 1185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1837615416963779e-05,
+      "loss": 1.5755,
+      "step": 1186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1837394182070831e-05,
+      "loss": 1.5827,
+      "step": 1187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1837172947177884e-05,
+      "loss": 1.5613,
+      "step": 1188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1836951712284936e-05,
+      "loss": 1.5395,
+      "step": 1189
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1836730477391987e-05,
+      "loss": 1.7391,
+      "step": 1190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1836509242499038e-05,
+      "loss": 1.6325,
+      "step": 1191
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1836288007606091e-05,
+      "loss": 1.5913,
+      "step": 1192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1836066772713144e-05,
+      "loss": 1.6488,
+      "step": 1193
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1835845537820196e-05,
+      "loss": 1.6314,
+      "step": 1194
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1835624302927249e-05,
+      "loss": 1.5856,
+      "step": 1195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.18354030680343e-05,
+      "loss": 1.5773,
+      "step": 1196
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1835181833141352e-05,
+      "loss": 1.7041,
+      "step": 1197
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1834960598248403e-05,
+      "loss": 1.7682,
+      "step": 1198
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1834739363355456e-05,
+      "loss": 1.7353,
+      "step": 1199
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1834518128462508e-05,
+      "loss": 1.6953,
+      "step": 1200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1834296893569561e-05,
+      "loss": 1.6736,
+      "step": 1201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1834075658676612e-05,
+      "loss": 1.6792,
+      "step": 1202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1833854423783665e-05,
+      "loss": 1.6622,
+      "step": 1203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1833633188890717e-05,
+      "loss": 1.7895,
+      "step": 1204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.183341195399777e-05,
+      "loss": 1.6339,
+      "step": 1205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.183319071910482e-05,
+      "loss": 1.7139,
+      "step": 1206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1832969484211872e-05,
+      "loss": 1.6783,
+      "step": 1207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1832748249318924e-05,
+      "loss": 1.6087,
+      "step": 1208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1832527014425977e-05,
+      "loss": 1.5437,
+      "step": 1209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.183230577953303e-05,
+      "loss": 1.6679,
+      "step": 1210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1832084544640082e-05,
+      "loss": 1.7778,
+      "step": 1211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1831863309747135e-05,
+      "loss": 1.6982,
+      "step": 1212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1831642074854186e-05,
+      "loss": 1.635,
+      "step": 1213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1831420839961237e-05,
+      "loss": 1.7338,
+      "step": 1214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1831199605068289e-05,
+      "loss": 1.6594,
+      "step": 1215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1830978370175342e-05,
+      "loss": 1.6314,
+      "step": 1216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1830757135282394e-05,
+      "loss": 1.5929,
+      "step": 1217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1830535900389445e-05,
+      "loss": 1.5874,
+      "step": 1218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1830314665496498e-05,
+      "loss": 1.5548,
+      "step": 1219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.183009343060355e-05,
+      "loss": 1.6023,
+      "step": 1220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1829872195710603e-05,
+      "loss": 1.5365,
+      "step": 1221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1829650960817654e-05,
+      "loss": 1.6214,
+      "step": 1222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1829429725924707e-05,
+      "loss": 1.7201,
+      "step": 1223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1829208491031758e-05,
+      "loss": 1.7256,
+      "step": 1224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.182898725613881e-05,
+      "loss": 1.5589,
+      "step": 1225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1828766021245863e-05,
+      "loss": 1.742,
+      "step": 1226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1828544786352915e-05,
+      "loss": 1.9865,
+      "step": 1227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1828323551459968e-05,
+      "loss": 1.6051,
+      "step": 1228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1828102316567019e-05,
+      "loss": 1.6851,
+      "step": 1229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1827881081674072e-05,
+      "loss": 1.719,
+      "step": 1230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1827659846781122e-05,
+      "loss": 1.7117,
+      "step": 1231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1827438611888175e-05,
+      "loss": 1.6233,
+      "step": 1232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1827217376995228e-05,
+      "loss": 1.5353,
+      "step": 1233
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1826996142102279e-05,
+      "loss": 1.6418,
+      "step": 1234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1826774907209331e-05,
+      "loss": 1.5701,
+      "step": 1235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1826553672316384e-05,
+      "loss": 1.5292,
+      "step": 1236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1826332437423436e-05,
+      "loss": 1.6186,
+      "step": 1237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1826111202530487e-05,
+      "loss": 1.6818,
+      "step": 1238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.182588996763754e-05,
+      "loss": 1.5842,
+      "step": 1239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1825668732744591e-05,
+      "loss": 1.7866,
+      "step": 1240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1825447497851643e-05,
+      "loss": 1.7584,
+      "step": 1241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1825226262958696e-05,
+      "loss": 1.5497,
+      "step": 1242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1825005028065749e-05,
+      "loss": 1.5722,
+      "step": 1243
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1824783793172801e-05,
+      "loss": 1.7425,
+      "step": 1244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1824562558279852e-05,
+      "loss": 1.5323,
+      "step": 1245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1824341323386905e-05,
+      "loss": 1.7142,
+      "step": 1246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1824120088493956e-05,
+      "loss": 1.658,
+      "step": 1247
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1823898853601008e-05,
+      "loss": 1.7113,
+      "step": 1248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1823677618708061e-05,
+      "loss": 1.6907,
+      "step": 1249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1823456383815114e-05,
+      "loss": 1.7172,
+      "step": 1250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1823235148922165e-05,
+      "loss": 1.8054,
+      "step": 1251
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1823013914029217e-05,
+      "loss": 1.6393,
+      "step": 1252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.182279267913627e-05,
+      "loss": 1.7682,
+      "step": 1253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.182257144424332e-05,
+      "loss": 1.6302,
+      "step": 1254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1822350209350373e-05,
+      "loss": 1.6136,
+      "step": 1255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1822128974457424e-05,
+      "loss": 1.7082,
+      "step": 1256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1821907739564477e-05,
+      "loss": 1.8075,
+      "step": 1257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.182168650467153e-05,
+      "loss": 1.4905,
+      "step": 1258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1821465269778582e-05,
+      "loss": 1.7277,
+      "step": 1259
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1821244034885635e-05,
+      "loss": 1.5435,
+      "step": 1260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1821022799992687e-05,
+      "loss": 1.6031,
+      "step": 1261
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1820801565099738e-05,
+      "loss": 1.5602,
+      "step": 1262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1820580330206789e-05,
+      "loss": 1.6445,
+      "step": 1263
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1820359095313842e-05,
+      "loss": 1.4925,
+      "step": 1264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1820137860420894e-05,
+      "loss": 1.6198,
+      "step": 1265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1819916625527947e-05,
+      "loss": 1.7822,
+      "step": 1266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1819695390634998e-05,
+      "loss": 1.583,
+      "step": 1267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.181947415574205e-05,
+      "loss": 1.7114,
+      "step": 1268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1819252920849103e-05,
+      "loss": 1.5826,
+      "step": 1269
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1819031685956154e-05,
+      "loss": 1.7028,
+      "step": 1270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1818810451063207e-05,
+      "loss": 1.7187,
+      "step": 1271
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.181858921617026e-05,
+      "loss": 1.6329,
+      "step": 1272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.181836798127731e-05,
+      "loss": 1.6551,
+      "step": 1273
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1818146746384363e-05,
+      "loss": 1.6445,
+      "step": 1274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1817925511491415e-05,
+      "loss": 1.6331,
+      "step": 1275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1817704276598468e-05,
+      "loss": 1.8659,
+      "step": 1276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.181748304170552e-05,
+      "loss": 1.7283,
+      "step": 1277
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1817261806812571e-05,
+      "loss": 1.6891,
+      "step": 1278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1817040571919622e-05,
+      "loss": 1.7792,
+      "step": 1279
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1816819337026675e-05,
+      "loss": 1.7335,
+      "step": 1280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1816598102133728e-05,
+      "loss": 1.5727,
+      "step": 1281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.181637686724078e-05,
+      "loss": 1.7229,
+      "step": 1282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1816155632347833e-05,
+      "loss": 1.5498,
+      "step": 1283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1815934397454884e-05,
+      "loss": 1.6494,
+      "step": 1284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1815713162561936e-05,
+      "loss": 1.7601,
+      "step": 1285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1815491927668987e-05,
+      "loss": 1.6202,
+      "step": 1286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.181527069277604e-05,
+      "loss": 1.5318,
+      "step": 1287
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1815049457883093e-05,
+      "loss": 1.6421,
+      "step": 1288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1814828222990143e-05,
+      "loss": 1.6649,
+      "step": 1289
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1814606988097196e-05,
+      "loss": 1.5163,
+      "step": 1290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1814385753204249e-05,
+      "loss": 1.6395,
+      "step": 1291
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1814164518311301e-05,
+      "loss": 1.7332,
+      "step": 1292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1813943283418354e-05,
+      "loss": 1.6869,
+      "step": 1293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1813722048525405e-05,
+      "loss": 1.6626,
+      "step": 1294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1813500813632456e-05,
+      "loss": 1.8571,
+      "step": 1295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1813279578739508e-05,
+      "loss": 1.718,
+      "step": 1296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1813058343846561e-05,
+      "loss": 1.6822,
+      "step": 1297
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1812837108953614e-05,
+      "loss": 1.6957,
+      "step": 1298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1812615874060666e-05,
+      "loss": 1.7654,
+      "step": 1299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1812394639167717e-05,
+      "loss": 1.5865,
+      "step": 1300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.181217340427477e-05,
+      "loss": 1.708,
+      "step": 1301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.181195216938182e-05,
+      "loss": 1.6381,
+      "step": 1302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1811730934488873e-05,
+      "loss": 1.7111,
+      "step": 1303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1811509699595926e-05,
+      "loss": 1.5964,
+      "step": 1304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1811288464702978e-05,
+      "loss": 1.7538,
+      "step": 1305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.181106722981003e-05,
+      "loss": 1.5508,
+      "step": 1306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1810845994917082e-05,
+      "loss": 1.7011,
+      "step": 1307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1810624760024135e-05,
+      "loss": 1.8868,
+      "step": 1308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1810403525131187e-05,
+      "loss": 1.606,
+      "step": 1309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1810182290238238e-05,
+      "loss": 1.6408,
+      "step": 1310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1809961055345289e-05,
+      "loss": 1.6483,
+      "step": 1311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1809739820452342e-05,
+      "loss": 1.8826,
+      "step": 1312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1809518585559394e-05,
+      "loss": 1.9212,
+      "step": 1313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1809297350666447e-05,
+      "loss": 1.5356,
+      "step": 1314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.18090761157735e-05,
+      "loss": 1.6512,
+      "step": 1315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1808854880880552e-05,
+      "loss": 1.4958,
+      "step": 1316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1808633645987603e-05,
+      "loss": 1.636,
+      "step": 1317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1808412411094656e-05,
+      "loss": 1.6785,
+      "step": 1318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1808191176201707e-05,
+      "loss": 1.6151,
+      "step": 1319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.180796994130876e-05,
+      "loss": 1.6092,
+      "step": 1320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1807748706415812e-05,
+      "loss": 1.4687,
+      "step": 1321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1807527471522863e-05,
+      "loss": 1.8317,
+      "step": 1322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1807306236629915e-05,
+      "loss": 1.7056,
+      "step": 1323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1807085001736968e-05,
+      "loss": 1.5988,
+      "step": 1324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.180686376684402e-05,
+      "loss": 1.6696,
+      "step": 1325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1806642531951071e-05,
+      "loss": 1.6414,
+      "step": 1326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1806421297058122e-05,
+      "loss": 1.5774,
+      "step": 1327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1806200062165175e-05,
+      "loss": 1.7063,
+      "step": 1328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1805978827272228e-05,
+      "loss": 1.5609,
+      "step": 1329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.180575759237928e-05,
+      "loss": 1.6217,
+      "step": 1330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1805536357486333e-05,
+      "loss": 1.6941,
+      "step": 1331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1805315122593385e-05,
+      "loss": 1.5928,
+      "step": 1332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1805093887700436e-05,
+      "loss": 1.6091,
+      "step": 1333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1804872652807489e-05,
+      "loss": 1.6355,
+      "step": 1334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.180465141791454e-05,
+      "loss": 1.5323,
+      "step": 1335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1804430183021592e-05,
+      "loss": 1.7508,
+      "step": 1336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1804208948128645e-05,
+      "loss": 1.6898,
+      "step": 1337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1803987713235696e-05,
+      "loss": 1.6833,
+      "step": 1338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1803766478342749e-05,
+      "loss": 1.4371,
+      "step": 1339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1803545243449801e-05,
+      "loss": 1.6299,
+      "step": 1340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1803324008556854e-05,
+      "loss": 1.5521,
+      "step": 1341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1803102773663905e-05,
+      "loss": 1.4443,
+      "step": 1342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1802881538770957e-05,
+      "loss": 1.7104,
+      "step": 1343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1802660303878008e-05,
+      "loss": 1.7023,
+      "step": 1344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1802439068985061e-05,
+      "loss": 1.6977,
+      "step": 1345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1802217834092114e-05,
+      "loss": 1.5937,
+      "step": 1346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1801996599199166e-05,
+      "loss": 1.5314,
+      "step": 1347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1801775364306219e-05,
+      "loss": 1.6734,
+      "step": 1348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.180155412941327e-05,
+      "loss": 1.7195,
+      "step": 1349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1801332894520322e-05,
+      "loss": 1.7319,
+      "step": 1350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1801111659627373e-05,
+      "loss": 1.5751,
+      "step": 1351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1800890424734426e-05,
+      "loss": 1.4554,
+      "step": 1352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1800669189841478e-05,
+      "loss": 1.6583,
+      "step": 1353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1800447954948531e-05,
+      "loss": 1.5189,
+      "step": 1354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1800226720055582e-05,
+      "loss": 1.5207,
+      "step": 1355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1800005485162635e-05,
+      "loss": 1.6498,
+      "step": 1356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1799784250269687e-05,
+      "loss": 1.7185,
+      "step": 1357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1799563015376738e-05,
+      "loss": 1.5883,
+      "step": 1358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.179934178048379e-05,
+      "loss": 1.5603,
+      "step": 1359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1799120545590842e-05,
+      "loss": 1.554,
+      "step": 1360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1798899310697894e-05,
+      "loss": 1.6502,
+      "step": 1361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1798678075804947e-05,
+      "loss": 1.621,
+      "step": 1362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1798456840912e-05,
+      "loss": 1.7389,
+      "step": 1363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1798235606019052e-05,
+      "loss": 1.5391,
+      "step": 1364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1798014371126105e-05,
+      "loss": 1.6387,
+      "step": 1365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1797793136233156e-05,
+      "loss": 1.4638,
+      "step": 1366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1797571901340207e-05,
+      "loss": 1.6377,
+      "step": 1367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1797350666447259e-05,
+      "loss": 1.5907,
+      "step": 1368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1797129431554312e-05,
+      "loss": 1.6996,
+      "step": 1369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1796908196661364e-05,
+      "loss": 1.5552,
+      "step": 1370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1796686961768415e-05,
+      "loss": 1.7426,
+      "step": 1371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1796465726875468e-05,
+      "loss": 1.6488,
+      "step": 1372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.179624449198252e-05,
+      "loss": 1.5387,
+      "step": 1373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1796023257089571e-05,
+      "loss": 1.5427,
+      "step": 1374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1795802022196624e-05,
+      "loss": 1.7587,
+      "step": 1375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1795580787303677e-05,
+      "loss": 1.6083,
+      "step": 1376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1795359552410728e-05,
+      "loss": 1.5159,
+      "step": 1377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.179513831751778e-05,
+      "loss": 1.6505,
+      "step": 1378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1794917082624833e-05,
+      "loss": 1.6121,
+      "step": 1379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1794695847731885e-05,
+      "loss": 1.5288,
+      "step": 1380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1794474612838938e-05,
+      "loss": 1.4789,
+      "step": 1381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1794253377945989e-05,
+      "loss": 1.5801,
+      "step": 1382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.179403214305304e-05,
+      "loss": 1.454,
+      "step": 1383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1793810908160092e-05,
+      "loss": 1.5295,
+      "step": 1384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1793589673267145e-05,
+      "loss": 1.5136,
+      "step": 1385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1793368438374198e-05,
+      "loss": 1.7532,
+      "step": 1386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.179314720348125e-05,
+      "loss": 1.6946,
+      "step": 1387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1792925968588301e-05,
+      "loss": 1.6702,
+      "step": 1388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1792704733695354e-05,
+      "loss": 1.6935,
+      "step": 1389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1792483498802405e-05,
+      "loss": 1.643,
+      "step": 1390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1792262263909457e-05,
+      "loss": 1.5519,
+      "step": 1391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.179204102901651e-05,
+      "loss": 1.7524,
+      "step": 1392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1791819794123561e-05,
+      "loss": 1.6707,
+      "step": 1393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1791598559230613e-05,
+      "loss": 1.5226,
+      "step": 1394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1791377324337666e-05,
+      "loss": 1.5325,
+      "step": 1395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1791156089444719e-05,
+      "loss": 1.6489,
+      "step": 1396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1790934854551771e-05,
+      "loss": 1.6657,
+      "step": 1397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1790713619658822e-05,
+      "loss": 1.703,
+      "step": 1398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1790492384765873e-05,
+      "loss": 1.4468,
+      "step": 1399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1790271149872926e-05,
+      "loss": 1.6636,
+      "step": 1400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1790049914979978e-05,
+      "loss": 1.6166,
+      "step": 1401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1789828680087031e-05,
+      "loss": 1.6738,
+      "step": 1402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1789607445194084e-05,
+      "loss": 1.5954,
+      "step": 1403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1789386210301135e-05,
+      "loss": 1.6988,
+      "step": 1404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1789164975408187e-05,
+      "loss": 1.7975,
+      "step": 1405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.178894374051524e-05,
+      "loss": 1.6901,
+      "step": 1406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.178872250562229e-05,
+      "loss": 1.6759,
+      "step": 1407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1788501270729343e-05,
+      "loss": 1.6489,
+      "step": 1408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1788280035836396e-05,
+      "loss": 1.665,
+      "step": 1409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1788058800943447e-05,
+      "loss": 1.7081,
+      "step": 1410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.17878375660505e-05,
+      "loss": 1.5298,
+      "step": 1411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1787616331157552e-05,
+      "loss": 1.6934,
+      "step": 1412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1787395096264605e-05,
+      "loss": 1.4982,
+      "step": 1413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1787173861371656e-05,
+      "loss": 1.5499,
+      "step": 1414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1786952626478706e-05,
+      "loss": 1.4678,
+      "step": 1415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1786731391585759e-05,
+      "loss": 1.5627,
+      "step": 1416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1786510156692812e-05,
+      "loss": 1.5495,
+      "step": 1417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1786288921799864e-05,
+      "loss": 1.7854,
+      "step": 1418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1786067686906917e-05,
+      "loss": 1.7949,
+      "step": 1419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1785846452013968e-05,
+      "loss": 1.4805,
+      "step": 1420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.178562521712102e-05,
+      "loss": 1.6518,
+      "step": 1421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1785403982228073e-05,
+      "loss": 1.4951,
+      "step": 1422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1785182747335124e-05,
+      "loss": 1.7418,
+      "step": 1423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1784961512442177e-05,
+      "loss": 1.5125,
+      "step": 1424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.178474027754923e-05,
+      "loss": 1.5836,
+      "step": 1425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.178451904265628e-05,
+      "loss": 1.6012,
+      "step": 1426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1784297807763333e-05,
+      "loss": 1.672,
+      "step": 1427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1784076572870385e-05,
+      "loss": 1.5602,
+      "step": 1428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1783855337977438e-05,
+      "loss": 1.575,
+      "step": 1429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1783634103084489e-05,
+      "loss": 1.5079,
+      "step": 1430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.178341286819154e-05,
+      "loss": 1.467,
+      "step": 1431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1783191633298592e-05,
+      "loss": 1.7394,
+      "step": 1432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1782970398405645e-05,
+      "loss": 1.5255,
+      "step": 1433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1782749163512698e-05,
+      "loss": 1.6437,
+      "step": 1434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.178252792861975e-05,
+      "loss": 1.5259,
+      "step": 1435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1782306693726803e-05,
+      "loss": 1.4587,
+      "step": 1436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1782085458833854e-05,
+      "loss": 1.7898,
+      "step": 1437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1781864223940906e-05,
+      "loss": 1.4829,
+      "step": 1438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1781642989047957e-05,
+      "loss": 1.7068,
+      "step": 1439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.178142175415501e-05,
+      "loss": 1.4418,
+      "step": 1440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1781200519262063e-05,
+      "loss": 1.6243,
+      "step": 1441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1780979284369113e-05,
+      "loss": 1.5618,
+      "step": 1442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1780758049476166e-05,
+      "loss": 1.4209,
+      "step": 1443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1780536814583219e-05,
+      "loss": 1.5814,
+      "step": 1444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1780315579690271e-05,
+      "loss": 1.5924,
+      "step": 1445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1780094344797322e-05,
+      "loss": 1.4448,
+      "step": 1446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1779873109904375e-05,
+      "loss": 1.5867,
+      "step": 1447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1779651875011426e-05,
+      "loss": 1.6699,
+      "step": 1448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1779430640118478e-05,
+      "loss": 1.4837,
+      "step": 1449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1779209405225531e-05,
+      "loss": 1.6743,
+      "step": 1450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1778988170332584e-05,
+      "loss": 1.6765,
+      "step": 1451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1778766935439636e-05,
+      "loss": 1.5157,
+      "step": 1452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1778545700546687e-05,
+      "loss": 1.6526,
+      "step": 1453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.177832446565374e-05,
+      "loss": 1.5945,
+      "step": 1454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.177810323076079e-05,
+      "loss": 1.7335,
+      "step": 1455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1777881995867843e-05,
+      "loss": 1.5744,
+      "step": 1456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1777660760974896e-05,
+      "loss": 1.5882,
+      "step": 1457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1777439526081948e-05,
+      "loss": 1.5288,
+      "step": 1458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1777218291189e-05,
+      "loss": 1.7001,
+      "step": 1459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1776997056296052e-05,
+      "loss": 1.5451,
+      "step": 1460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1776775821403105e-05,
+      "loss": 1.5824,
+      "step": 1461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1776554586510155e-05,
+      "loss": 1.449,
+      "step": 1462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1776333351617208e-05,
+      "loss": 1.546,
+      "step": 1463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1776112116724259e-05,
+      "loss": 1.6733,
+      "step": 1464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1775890881831312e-05,
+      "loss": 1.6021,
+      "step": 1465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1775669646938364e-05,
+      "loss": 1.5848,
+      "step": 1466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1775448412045417e-05,
+      "loss": 1.5594,
+      "step": 1467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.177522717715247e-05,
+      "loss": 1.7817,
+      "step": 1468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1775005942259522e-05,
+      "loss": 1.5475,
+      "step": 1469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1774784707366573e-05,
+      "loss": 1.5482,
+      "step": 1470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1774563472473624e-05,
+      "loss": 1.7375,
+      "step": 1471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1774342237580677e-05,
+      "loss": 1.6503,
+      "step": 1472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1774121002687729e-05,
+      "loss": 1.4982,
+      "step": 1473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1773899767794782e-05,
+      "loss": 1.5749,
+      "step": 1474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1773678532901833e-05,
+      "loss": 1.3652,
+      "step": 1475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1773457298008885e-05,
+      "loss": 1.6257,
+      "step": 1476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1773236063115938e-05,
+      "loss": 1.5706,
+      "step": 1477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1773014828222989e-05,
+      "loss": 1.6279,
+      "step": 1478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1772793593330041e-05,
+      "loss": 1.5604,
+      "step": 1479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1772572358437094e-05,
+      "loss": 1.5736,
+      "step": 1480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1772351123544145e-05,
+      "loss": 1.6056,
+      "step": 1481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1772129888651198e-05,
+      "loss": 1.6218,
+      "step": 1482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.177190865375825e-05,
+      "loss": 1.7506,
+      "step": 1483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1771687418865303e-05,
+      "loss": 1.5407,
+      "step": 1484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1771466183972355e-05,
+      "loss": 1.7831,
+      "step": 1485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1771244949079406e-05,
+      "loss": 1.6516,
+      "step": 1486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1771023714186457e-05,
+      "loss": 1.5379,
+      "step": 1487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.177080247929351e-05,
+      "loss": 1.6471,
+      "step": 1488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1770581244400562e-05,
+      "loss": 1.4914,
+      "step": 1489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1770360009507615e-05,
+      "loss": 1.4819,
+      "step": 1490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1770138774614668e-05,
+      "loss": 1.5736,
+      "step": 1491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1769917539721719e-05,
+      "loss": 1.6808,
+      "step": 1492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1769696304828771e-05,
+      "loss": 1.4102,
+      "step": 1493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1769475069935824e-05,
+      "loss": 1.5992,
+      "step": 1494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1769253835042875e-05,
+      "loss": 1.6699,
+      "step": 1495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1769032600149927e-05,
+      "loss": 1.5699,
+      "step": 1496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1768811365256978e-05,
+      "loss": 1.6637,
+      "step": 1497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1768590130364031e-05,
+      "loss": 1.63,
+      "step": 1498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1768368895471083e-05,
+      "loss": 1.5407,
+      "step": 1499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1768147660578136e-05,
+      "loss": 1.6203,
+      "step": 1500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1767926425685189e-05,
+      "loss": 1.6577,
+      "step": 1501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.176770519079224e-05,
+      "loss": 1.5335,
+      "step": 1502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.176748395589929e-05,
+      "loss": 1.5695,
+      "step": 1503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1767262721006343e-05,
+      "loss": 1.4879,
+      "step": 1504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1767041486113396e-05,
+      "loss": 1.6075,
+      "step": 1505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1766820251220448e-05,
+      "loss": 1.6935,
+      "step": 1506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1766599016327501e-05,
+      "loss": 1.5908,
+      "step": 1507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1766377781434552e-05,
+      "loss": 1.4472,
+      "step": 1508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1766156546541605e-05,
+      "loss": 1.6888,
+      "step": 1509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1765935311648657e-05,
+      "loss": 1.7296,
+      "step": 1510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1765714076755708e-05,
+      "loss": 1.6869,
+      "step": 1511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.176549284186276e-05,
+      "loss": 1.5612,
+      "step": 1512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1765271606969813e-05,
+      "loss": 1.5059,
+      "step": 1513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1765050372076864e-05,
+      "loss": 1.5223,
+      "step": 1514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1764829137183917e-05,
+      "loss": 1.58,
+      "step": 1515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.176460790229097e-05,
+      "loss": 1.6866,
+      "step": 1516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1764386667398022e-05,
+      "loss": 1.5048,
+      "step": 1517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1764165432505073e-05,
+      "loss": 1.5228,
+      "step": 1518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1763944197612124e-05,
+      "loss": 1.6901,
+      "step": 1519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1763722962719176e-05,
+      "loss": 1.6555,
+      "step": 1520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1763501727826229e-05,
+      "loss": 1.519,
+      "step": 1521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1763280492933282e-05,
+      "loss": 1.6442,
+      "step": 1522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1763059258040334e-05,
+      "loss": 1.7013,
+      "step": 1523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1762838023147385e-05,
+      "loss": 1.5486,
+      "step": 1524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1762616788254438e-05,
+      "loss": 1.7207,
+      "step": 1525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.176239555336149e-05,
+      "loss": 1.6802,
+      "step": 1526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1762174318468541e-05,
+      "loss": 1.5108,
+      "step": 1527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1761953083575594e-05,
+      "loss": 1.5236,
+      "step": 1528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1761731848682647e-05,
+      "loss": 1.5692,
+      "step": 1529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1761510613789698e-05,
+      "loss": 1.5363,
+      "step": 1530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.176128937889675e-05,
+      "loss": 1.6096,
+      "step": 1531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1761068144003803e-05,
+      "loss": 1.4741,
+      "step": 1532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1760846909110855e-05,
+      "loss": 1.759,
+      "step": 1533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1760625674217906e-05,
+      "loss": 1.6687,
+      "step": 1534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1760404439324957e-05,
+      "loss": 1.5965,
+      "step": 1535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.176018320443201e-05,
+      "loss": 1.643,
+      "step": 1536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1759961969539062e-05,
+      "loss": 1.4844,
+      "step": 1537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1759740734646115e-05,
+      "loss": 1.5213,
+      "step": 1538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1759519499753168e-05,
+      "loss": 1.4495,
+      "step": 1539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.175929826486022e-05,
+      "loss": 1.414,
+      "step": 1540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1759077029967271e-05,
+      "loss": 1.4967,
+      "step": 1541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1758855795074324e-05,
+      "loss": 1.746,
+      "step": 1542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1758634560181375e-05,
+      "loss": 1.5961,
+      "step": 1543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1758413325288427e-05,
+      "loss": 1.6014,
+      "step": 1544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.175819209039548e-05,
+      "loss": 1.4213,
+      "step": 1545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.175797085550253e-05,
+      "loss": 1.5254,
+      "step": 1546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1757749620609583e-05,
+      "loss": 1.5779,
+      "step": 1547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1757528385716636e-05,
+      "loss": 1.6144,
+      "step": 1548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1757307150823689e-05,
+      "loss": 1.8343,
+      "step": 1549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.175708591593074e-05,
+      "loss": 1.5455,
+      "step": 1550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1756864681037792e-05,
+      "loss": 1.646,
+      "step": 1551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1756643446144843e-05,
+      "loss": 1.5608,
+      "step": 1552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1756422211251896e-05,
+      "loss": 1.4762,
+      "step": 1553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1756200976358948e-05,
+      "loss": 1.4089,
+      "step": 1554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1755979741466001e-05,
+      "loss": 1.57,
+      "step": 1555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1755758506573054e-05,
+      "loss": 1.5989,
+      "step": 1556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1755537271680104e-05,
+      "loss": 1.5941,
+      "step": 1557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1755316036787157e-05,
+      "loss": 1.6907,
+      "step": 1558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1755094801894208e-05,
+      "loss": 1.6689,
+      "step": 1559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.175487356700126e-05,
+      "loss": 1.6469,
+      "step": 1560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1754652332108313e-05,
+      "loss": 1.6278,
+      "step": 1561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1754431097215366e-05,
+      "loss": 1.552,
+      "step": 1562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1754209862322417e-05,
+      "loss": 1.5374,
+      "step": 1563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.175398862742947e-05,
+      "loss": 1.5848,
+      "step": 1564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1753767392536522e-05,
+      "loss": 1.5781,
+      "step": 1565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1753546157643573e-05,
+      "loss": 1.5852,
+      "step": 1566
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1753324922750626e-05,
+      "loss": 1.6538,
+      "step": 1567
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1753103687857676e-05,
+      "loss": 1.6603,
+      "step": 1568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1752882452964729e-05,
+      "loss": 1.7842,
+      "step": 1569
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1752661218071782e-05,
+      "loss": 1.6752,
+      "step": 1570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1752439983178834e-05,
+      "loss": 1.5971,
+      "step": 1571
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1752218748285887e-05,
+      "loss": 1.6098,
+      "step": 1572
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.175199751339294e-05,
+      "loss": 1.4572,
+      "step": 1573
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.175177627849999e-05,
+      "loss": 1.5894,
+      "step": 1574
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1751555043607041e-05,
+      "loss": 1.5956,
+      "step": 1575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1751333808714094e-05,
+      "loss": 1.4369,
+      "step": 1576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1751112573821147e-05,
+      "loss": 1.607,
+      "step": 1577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.17508913389282e-05,
+      "loss": 1.5657,
+      "step": 1578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.175067010403525e-05,
+      "loss": 1.682,
+      "step": 1579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1750448869142303e-05,
+      "loss": 1.497,
+      "step": 1580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1750227634249355e-05,
+      "loss": 1.5611,
+      "step": 1581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1750006399356408e-05,
+      "loss": 1.5603,
+      "step": 1582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1749785164463459e-05,
+      "loss": 1.7031,
+      "step": 1583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1749563929570511e-05,
+      "loss": 1.4841,
+      "step": 1584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1749342694677562e-05,
+      "loss": 1.5244,
+      "step": 1585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1749121459784615e-05,
+      "loss": 1.6458,
+      "step": 1586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1748900224891668e-05,
+      "loss": 1.5878,
+      "step": 1587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.174867898999872e-05,
+      "loss": 1.6708,
+      "step": 1588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1748457755105773e-05,
+      "loss": 1.5336,
+      "step": 1589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1748236520212824e-05,
+      "loss": 1.4699,
+      "step": 1590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1748015285319875e-05,
+      "loss": 1.3949,
+      "step": 1591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1747794050426927e-05,
+      "loss": 1.6152,
+      "step": 1592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.174757281553398e-05,
+      "loss": 1.5147,
+      "step": 1593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1747351580641032e-05,
+      "loss": 1.5941,
+      "step": 1594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1747130345748085e-05,
+      "loss": 1.6519,
+      "step": 1595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1746909110855136e-05,
+      "loss": 1.4882,
+      "step": 1596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1746687875962189e-05,
+      "loss": 1.4357,
+      "step": 1597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1746466641069241e-05,
+      "loss": 1.4978,
+      "step": 1598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1746245406176292e-05,
+      "loss": 1.4986,
+      "step": 1599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1746024171283345e-05,
+      "loss": 1.6341,
+      "step": 1600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1745802936390396e-05,
+      "loss": 1.5816,
+      "step": 1601
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1745581701497448e-05,
+      "loss": 1.5369,
+      "step": 1602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1745360466604501e-05,
+      "loss": 1.5071,
+      "step": 1603
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1745139231711554e-05,
+      "loss": 1.4664,
+      "step": 1604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1744917996818606e-05,
+      "loss": 1.6438,
+      "step": 1605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1744696761925657e-05,
+      "loss": 1.7948,
+      "step": 1606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1744475527032708e-05,
+      "loss": 1.4887,
+      "step": 1607
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.174425429213976e-05,
+      "loss": 1.6469,
+      "step": 1608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1744033057246813e-05,
+      "loss": 1.5442,
+      "step": 1609
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1743811822353866e-05,
+      "loss": 1.5075,
+      "step": 1610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1743590587460918e-05,
+      "loss": 1.721,
+      "step": 1611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.174336935256797e-05,
+      "loss": 1.657,
+      "step": 1612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1743148117675022e-05,
+      "loss": 1.5381,
+      "step": 1613
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1742926882782075e-05,
+      "loss": 1.6077,
+      "step": 1614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1742705647889125e-05,
+      "loss": 1.615,
+      "step": 1615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1742484412996178e-05,
+      "loss": 1.5462,
+      "step": 1616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1742263178103229e-05,
+      "loss": 1.5417,
+      "step": 1617
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1742041943210282e-05,
+      "loss": 1.5837,
+      "step": 1618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1741820708317334e-05,
+      "loss": 1.4792,
+      "step": 1619
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1741599473424387e-05,
+      "loss": 1.6216,
+      "step": 1620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.174137823853144e-05,
+      "loss": 1.6304,
+      "step": 1621
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.174115700363849e-05,
+      "loss": 1.5972,
+      "step": 1622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1740935768745541e-05,
+      "loss": 1.7007,
+      "step": 1623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1740714533852594e-05,
+      "loss": 1.5921,
+      "step": 1624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1740493298959647e-05,
+      "loss": 1.562,
+      "step": 1625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1740272064066699e-05,
+      "loss": 1.5915,
+      "step": 1626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1740050829173752e-05,
+      "loss": 1.6324,
+      "step": 1627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1739829594280803e-05,
+      "loss": 1.6405,
+      "step": 1628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1739608359387855e-05,
+      "loss": 1.5905,
+      "step": 1629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1739387124494908e-05,
+      "loss": 1.7515,
+      "step": 1630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1739165889601959e-05,
+      "loss": 1.7423,
+      "step": 1631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1738944654709011e-05,
+      "loss": 1.5532,
+      "step": 1632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1738723419816064e-05,
+      "loss": 1.5484,
+      "step": 1633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1738502184923115e-05,
+      "loss": 1.7151,
+      "step": 1634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1738280950030168e-05,
+      "loss": 1.5969,
+      "step": 1635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.173805971513722e-05,
+      "loss": 1.4976,
+      "step": 1636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1737838480244273e-05,
+      "loss": 1.7311,
+      "step": 1637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1737617245351324e-05,
+      "loss": 1.5921,
+      "step": 1638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1737396010458376e-05,
+      "loss": 1.5095,
+      "step": 1639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1737174775565427e-05,
+      "loss": 1.6508,
+      "step": 1640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.173695354067248e-05,
+      "loss": 1.734,
+      "step": 1641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1736732305779532e-05,
+      "loss": 1.706,
+      "step": 1642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1736511070886585e-05,
+      "loss": 1.5204,
+      "step": 1643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1736289835993638e-05,
+      "loss": 1.4846,
+      "step": 1644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1736068601100689e-05,
+      "loss": 1.6742,
+      "step": 1645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1735847366207741e-05,
+      "loss": 1.4508,
+      "step": 1646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1735626131314792e-05,
+      "loss": 1.5351,
+      "step": 1647
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1735404896421845e-05,
+      "loss": 1.676,
+      "step": 1648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1735183661528897e-05,
+      "loss": 1.6552,
+      "step": 1649
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1734962426635948e-05,
+      "loss": 1.7033,
+      "step": 1650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1734741191743001e-05,
+      "loss": 1.7263,
+      "step": 1651
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1734519956850053e-05,
+      "loss": 1.4926,
+      "step": 1652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1734298721957106e-05,
+      "loss": 1.637,
+      "step": 1653
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1734077487064157e-05,
+      "loss": 1.5177,
+      "step": 1654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.173385625217121e-05,
+      "loss": 1.4243,
+      "step": 1655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.173363501727826e-05,
+      "loss": 1.5574,
+      "step": 1656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1733413782385313e-05,
+      "loss": 1.5602,
+      "step": 1657
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1733192547492366e-05,
+      "loss": 1.7104,
+      "step": 1658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1732971312599418e-05,
+      "loss": 1.542,
+      "step": 1659
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1732750077706471e-05,
+      "loss": 1.5405,
+      "step": 1660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1732528842813522e-05,
+      "loss": 1.6992,
+      "step": 1661
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1732307607920575e-05,
+      "loss": 1.59,
+      "step": 1662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1732086373027625e-05,
+      "loss": 1.4983,
+      "step": 1663
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1731865138134678e-05,
+      "loss": 1.4796,
+      "step": 1664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.173164390324173e-05,
+      "loss": 1.5843,
+      "step": 1665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1731422668348783e-05,
+      "loss": 1.6596,
+      "step": 1666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1731201433455834e-05,
+      "loss": 1.6901,
+      "step": 1667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1730980198562887e-05,
+      "loss": 1.7009,
+      "step": 1668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.173075896366994e-05,
+      "loss": 1.5467,
+      "step": 1669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1730537728776992e-05,
+      "loss": 1.5732,
+      "step": 1670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1730316493884043e-05,
+      "loss": 1.8004,
+      "step": 1671
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1730095258991094e-05,
+      "loss": 1.6595,
+      "step": 1672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1729874024098146e-05,
+      "loss": 1.6559,
+      "step": 1673
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1729652789205199e-05,
+      "loss": 1.7054,
+      "step": 1674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1729431554312252e-05,
+      "loss": 1.5858,
+      "step": 1675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1729210319419304e-05,
+      "loss": 1.6138,
+      "step": 1676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1728989084526357e-05,
+      "loss": 1.3627,
+      "step": 1677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1728767849633408e-05,
+      "loss": 1.5774,
+      "step": 1678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1728546614740459e-05,
+      "loss": 1.7358,
+      "step": 1679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1728325379847511e-05,
+      "loss": 1.3811,
+      "step": 1680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1728104144954564e-05,
+      "loss": 1.6503,
+      "step": 1681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1727882910061617e-05,
+      "loss": 1.5281,
+      "step": 1682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1727661675168668e-05,
+      "loss": 1.6557,
+      "step": 1683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.172744044027572e-05,
+      "loss": 1.5885,
+      "step": 1684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1727219205382773e-05,
+      "loss": 1.6818,
+      "step": 1685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1726997970489825e-05,
+      "loss": 1.5469,
+      "step": 1686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1726776735596876e-05,
+      "loss": 1.6252,
+      "step": 1687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1726555500703929e-05,
+      "loss": 1.735,
+      "step": 1688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.172633426581098e-05,
+      "loss": 1.5255,
+      "step": 1689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1726113030918032e-05,
+      "loss": 1.4599,
+      "step": 1690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1725891796025085e-05,
+      "loss": 1.505,
+      "step": 1691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1725670561132138e-05,
+      "loss": 1.6714,
+      "step": 1692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.172544932623919e-05,
+      "loss": 1.6783,
+      "step": 1693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1725228091346241e-05,
+      "loss": 1.6677,
+      "step": 1694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1725006856453292e-05,
+      "loss": 1.3716,
+      "step": 1695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1724785621560345e-05,
+      "loss": 1.6282,
+      "step": 1696
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1724564386667397e-05,
+      "loss": 1.4413,
+      "step": 1697
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.172434315177445e-05,
+      "loss": 1.517,
+      "step": 1698
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1724121916881503e-05,
+      "loss": 1.6214,
+      "step": 1699
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1723900681988553e-05,
+      "loss": 1.3983,
+      "step": 1700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1723679447095606e-05,
+      "loss": 1.6486,
+      "step": 1701
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1723458212202659e-05,
+      "loss": 1.537,
+      "step": 1702
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.172323697730971e-05,
+      "loss": 1.656,
+      "step": 1703
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1723015742416762e-05,
+      "loss": 1.6551,
+      "step": 1704
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1722794507523813e-05,
+      "loss": 1.5805,
+      "step": 1705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1722573272630866e-05,
+      "loss": 1.6926,
+      "step": 1706
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1722352037737918e-05,
+      "loss": 1.5444,
+      "step": 1707
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1722130802844971e-05,
+      "loss": 1.667,
+      "step": 1708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1721909567952024e-05,
+      "loss": 1.5079,
+      "step": 1709
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1721688333059074e-05,
+      "loss": 1.6204,
+      "step": 1710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1721467098166125e-05,
+      "loss": 1.4848,
+      "step": 1711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1721245863273178e-05,
+      "loss": 1.5677,
+      "step": 1712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.172102462838023e-05,
+      "loss": 1.564,
+      "step": 1713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1720803393487283e-05,
+      "loss": 1.5732,
+      "step": 1714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1720582158594336e-05,
+      "loss": 1.6079,
+      "step": 1715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1720360923701387e-05,
+      "loss": 1.5651,
+      "step": 1716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.172013968880844e-05,
+      "loss": 1.6486,
+      "step": 1717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1719918453915492e-05,
+      "loss": 1.7312,
+      "step": 1718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1719697219022543e-05,
+      "loss": 1.4847,
+      "step": 1719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1719475984129596e-05,
+      "loss": 1.4992,
+      "step": 1720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1719254749236646e-05,
+      "loss": 1.4734,
+      "step": 1721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1719033514343699e-05,
+      "loss": 1.6969,
+      "step": 1722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1718812279450752e-05,
+      "loss": 1.6648,
+      "step": 1723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1718591044557804e-05,
+      "loss": 1.5811,
+      "step": 1724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1718369809664857e-05,
+      "loss": 1.4501,
+      "step": 1725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1718148574771908e-05,
+      "loss": 1.5468,
+      "step": 1726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.171792733987896e-05,
+      "loss": 1.5114,
+      "step": 1727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1717706104986011e-05,
+      "loss": 1.6905,
+      "step": 1728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1717484870093064e-05,
+      "loss": 1.6215,
+      "step": 1729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1717263635200117e-05,
+      "loss": 1.5452,
+      "step": 1730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.171704240030717e-05,
+      "loss": 1.4444,
+      "step": 1731
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.171682116541422e-05,
+      "loss": 1.6129,
+      "step": 1732
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1716599930521273e-05,
+      "loss": 1.5285,
+      "step": 1733
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1716378695628325e-05,
+      "loss": 1.5944,
+      "step": 1734
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1716157460735376e-05,
+      "loss": 1.5984,
+      "step": 1735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1715936225842429e-05,
+      "loss": 1.5303,
+      "step": 1736
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1715714990949481e-05,
+      "loss": 1.574,
+      "step": 1737
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1715493756056532e-05,
+      "loss": 1.4899,
+      "step": 1738
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1715272521163585e-05,
+      "loss": 1.5086,
+      "step": 1739
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1715051286270638e-05,
+      "loss": 1.6519,
+      "step": 1740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.171483005137769e-05,
+      "loss": 1.4064,
+      "step": 1741
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1714608816484741e-05,
+      "loss": 1.5285,
+      "step": 1742
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1714387581591794e-05,
+      "loss": 1.5789,
+      "step": 1743
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1714166346698845e-05,
+      "loss": 1.5403,
+      "step": 1744
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1713945111805897e-05,
+      "loss": 1.5531,
+      "step": 1745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.171372387691295e-05,
+      "loss": 1.3868,
+      "step": 1746
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1713502642020002e-05,
+      "loss": 1.5714,
+      "step": 1747
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1713281407127055e-05,
+      "loss": 1.5156,
+      "step": 1748
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1713060172234106e-05,
+      "loss": 1.4839,
+      "step": 1749
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1712838937341159e-05,
+      "loss": 1.5387,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.171261770244821e-05,
+      "loss": 1.5959,
+      "step": 1751
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1712396467555262e-05,
+      "loss": 1.6518,
+      "step": 1752
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1712175232662315e-05,
+      "loss": 1.5839,
+      "step": 1753
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1711953997769366e-05,
+      "loss": 1.5651,
+      "step": 1754
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1711732762876418e-05,
+      "loss": 1.4895,
+      "step": 1755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1711511527983471e-05,
+      "loss": 1.5451,
+      "step": 1756
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1711290293090524e-05,
+      "loss": 1.5946,
+      "step": 1757
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1711069058197576e-05,
+      "loss": 1.6066,
+      "step": 1758
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1710847823304627e-05,
+      "loss": 1.5025,
+      "step": 1759
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1710626588411678e-05,
+      "loss": 1.5586,
+      "step": 1760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.171040535351873e-05,
+      "loss": 1.5962,
+      "step": 1761
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1710184118625783e-05,
+      "loss": 1.5422,
+      "step": 1762
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1709962883732836e-05,
+      "loss": 1.4753,
+      "step": 1763
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1709741648839888e-05,
+      "loss": 1.429,
+      "step": 1764
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.170952041394694e-05,
+      "loss": 1.6213,
+      "step": 1765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1709299179053992e-05,
+      "loss": 1.6497,
+      "step": 1766
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1709077944161043e-05,
+      "loss": 1.5193,
+      "step": 1767
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1708856709268095e-05,
+      "loss": 1.6922,
+      "step": 1768
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1708635474375148e-05,
+      "loss": 1.5687,
+      "step": 1769
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.17084142394822e-05,
+      "loss": 1.4303,
+      "step": 1770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1708193004589252e-05,
+      "loss": 1.4728,
+      "step": 1771
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1707971769696304e-05,
+      "loss": 1.633,
+      "step": 1772
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1707750534803357e-05,
+      "loss": 1.4296,
+      "step": 1773
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.170752929991041e-05,
+      "loss": 1.6292,
+      "step": 1774
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.170730806501746e-05,
+      "loss": 1.565,
+      "step": 1775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1707086830124511e-05,
+      "loss": 1.5424,
+      "step": 1776
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1706865595231564e-05,
+      "loss": 1.582,
+      "step": 1777
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1706644360338617e-05,
+      "loss": 1.5917,
+      "step": 1778
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1706423125445669e-05,
+      "loss": 1.5431,
+      "step": 1779
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1706201890552722e-05,
+      "loss": 1.7374,
+      "step": 1780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1705980655659774e-05,
+      "loss": 1.5329,
+      "step": 1781
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1705759420766825e-05,
+      "loss": 1.6309,
+      "step": 1782
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1705538185873876e-05,
+      "loss": 1.6444,
+      "step": 1783
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1705316950980929e-05,
+      "loss": 1.6196,
+      "step": 1784
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1705095716087981e-05,
+      "loss": 1.6344,
+      "step": 1785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1704874481195034e-05,
+      "loss": 1.529,
+      "step": 1786
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1704653246302085e-05,
+      "loss": 1.6389,
+      "step": 1787
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1704432011409138e-05,
+      "loss": 1.4226,
+      "step": 1788
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.170421077651619e-05,
+      "loss": 1.4682,
+      "step": 1789
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1703989541623243e-05,
+      "loss": 1.4951,
+      "step": 1790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1703768306730294e-05,
+      "loss": 1.6269,
+      "step": 1791
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1703547071837346e-05,
+      "loss": 1.5621,
+      "step": 1792
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1703325836944397e-05,
+      "loss": 1.5208,
+      "step": 1793
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.170310460205145e-05,
+      "loss": 1.4652,
+      "step": 1794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1702883367158502e-05,
+      "loss": 1.6925,
+      "step": 1795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1702662132265555e-05,
+      "loss": 1.403,
+      "step": 1796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1702440897372608e-05,
+      "loss": 1.659,
+      "step": 1797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1702219662479659e-05,
+      "loss": 1.6238,
+      "step": 1798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.170199842758671e-05,
+      "loss": 1.5452,
+      "step": 1799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1701777192693762e-05,
+      "loss": 1.5281,
+      "step": 1800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1701555957800815e-05,
+      "loss": 1.6354,
+      "step": 1801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1701334722907867e-05,
+      "loss": 1.657,
+      "step": 1802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.170111348801492e-05,
+      "loss": 1.5936,
+      "step": 1803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1700892253121971e-05,
+      "loss": 1.6513,
+      "step": 1804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1700671018229023e-05,
+      "loss": 1.573,
+      "step": 1805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1700449783336076e-05,
+      "loss": 1.5189,
+      "step": 1806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1700228548443127e-05,
+      "loss": 1.7,
+      "step": 1807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.170000731355018e-05,
+      "loss": 1.6374,
+      "step": 1808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.169978607865723e-05,
+      "loss": 1.5054,
+      "step": 1809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1699564843764283e-05,
+      "loss": 1.6181,
+      "step": 1810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1699343608871336e-05,
+      "loss": 1.7819,
+      "step": 1811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1699122373978388e-05,
+      "loss": 1.4761,
+      "step": 1812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1698901139085441e-05,
+      "loss": 1.5813,
+      "step": 1813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1698679904192492e-05,
+      "loss": 1.517,
+      "step": 1814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1698458669299545e-05,
+      "loss": 1.4587,
+      "step": 1815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1698237434406595e-05,
+      "loss": 1.6733,
+      "step": 1816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1698016199513648e-05,
+      "loss": 1.6492,
+      "step": 1817
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.16977949646207e-05,
+      "loss": 1.6898,
+      "step": 1818
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1697573729727753e-05,
+      "loss": 1.6445,
+      "step": 1819
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1697352494834804e-05,
+      "loss": 1.7004,
+      "step": 1820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1697131259941857e-05,
+      "loss": 1.4207,
+      "step": 1821
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.169691002504891e-05,
+      "loss": 1.753,
+      "step": 1822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.169668879015596e-05,
+      "loss": 1.394,
+      "step": 1823
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1696467555263013e-05,
+      "loss": 1.6601,
+      "step": 1824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1696246320370064e-05,
+      "loss": 1.4912,
+      "step": 1825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1696025085477116e-05,
+      "loss": 1.7265,
+      "step": 1826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1695803850584169e-05,
+      "loss": 1.5084,
+      "step": 1827
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1695582615691222e-05,
+      "loss": 1.3727,
+      "step": 1828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1695361380798274e-05,
+      "loss": 1.5414,
+      "step": 1829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1695140145905325e-05,
+      "loss": 1.6172,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1694918911012378e-05,
+      "loss": 1.53,
+      "step": 1831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1694697676119429e-05,
+      "loss": 1.5113,
+      "step": 1832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1694476441226481e-05,
+      "loss": 1.5177,
+      "step": 1833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1694255206333534e-05,
+      "loss": 1.7187,
+      "step": 1834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1694033971440587e-05,
+      "loss": 1.4471,
+      "step": 1835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1693812736547637e-05,
+      "loss": 1.4413,
+      "step": 1836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.169359150165469e-05,
+      "loss": 1.5283,
+      "step": 1837
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1693370266761743e-05,
+      "loss": 1.4877,
+      "step": 1838
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1693149031868794e-05,
+      "loss": 1.7305,
+      "step": 1839
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1692927796975846e-05,
+      "loss": 1.4804,
+      "step": 1840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1692706562082899e-05,
+      "loss": 1.5755,
+      "step": 1841
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.169248532718995e-05,
+      "loss": 1.5433,
+      "step": 1842
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1692264092297002e-05,
+      "loss": 1.5097,
+      "step": 1843
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1692042857404055e-05,
+      "loss": 1.3689,
+      "step": 1844
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1691821622511108e-05,
+      "loss": 1.5404,
+      "step": 1845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.169160038761816e-05,
+      "loss": 1.5894,
+      "step": 1846
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1691379152725211e-05,
+      "loss": 1.5368,
+      "step": 1847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1691157917832262e-05,
+      "loss": 1.512,
+      "step": 1848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1690936682939315e-05,
+      "loss": 1.4164,
+      "step": 1849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1690715448046367e-05,
+      "loss": 1.5509,
+      "step": 1850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.169049421315342e-05,
+      "loss": 1.5922,
+      "step": 1851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1690272978260473e-05,
+      "loss": 1.4786,
+      "step": 1852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1690051743367523e-05,
+      "loss": 1.5929,
+      "step": 1853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1689830508474576e-05,
+      "loss": 1.398,
+      "step": 1854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1689609273581627e-05,
+      "loss": 1.5883,
+      "step": 1855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.168938803868868e-05,
+      "loss": 1.7345,
+      "step": 1856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1689166803795732e-05,
+      "loss": 1.6115,
+      "step": 1857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1688945568902783e-05,
+      "loss": 1.4538,
+      "step": 1858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1688724334009836e-05,
+      "loss": 1.5117,
+      "step": 1859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1688503099116888e-05,
+      "loss": 1.6529,
+      "step": 1860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1688281864223941e-05,
+      "loss": 1.5869,
+      "step": 1861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1688060629330994e-05,
+      "loss": 1.6112,
+      "step": 1862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1687839394438044e-05,
+      "loss": 1.5031,
+      "step": 1863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1687618159545095e-05,
+      "loss": 1.5129,
+      "step": 1864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1687396924652148e-05,
+      "loss": 1.4459,
+      "step": 1865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.16871756897592e-05,
+      "loss": 1.57,
+      "step": 1866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1686954454866253e-05,
+      "loss": 1.554,
+      "step": 1867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1686733219973306e-05,
+      "loss": 1.5067,
+      "step": 1868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1686511985080357e-05,
+      "loss": 1.4829,
+      "step": 1869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.168629075018741e-05,
+      "loss": 1.5151,
+      "step": 1870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.168606951529446e-05,
+      "loss": 1.4736,
+      "step": 1871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1685848280401513e-05,
+      "loss": 1.5098,
+      "step": 1872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1685627045508566e-05,
+      "loss": 1.5174,
+      "step": 1873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1685405810615618e-05,
+      "loss": 1.5735,
+      "step": 1874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1685184575722669e-05,
+      "loss": 1.4715,
+      "step": 1875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1684963340829722e-05,
+      "loss": 1.5833,
+      "step": 1876
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1684742105936774e-05,
+      "loss": 1.4796,
+      "step": 1877
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1684520871043827e-05,
+      "loss": 1.7061,
+      "step": 1878
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1684299636150878e-05,
+      "loss": 1.5876,
+      "step": 1879
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1684078401257929e-05,
+      "loss": 1.5328,
+      "step": 1880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1683857166364981e-05,
+      "loss": 1.5373,
+      "step": 1881
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1683635931472034e-05,
+      "loss": 1.5252,
+      "step": 1882
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1683414696579087e-05,
+      "loss": 1.505,
+      "step": 1883
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1683193461686139e-05,
+      "loss": 1.43,
+      "step": 1884
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1682972226793192e-05,
+      "loss": 1.6202,
+      "step": 1885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1682750991900243e-05,
+      "loss": 1.4738,
+      "step": 1886
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1682529757007294e-05,
+      "loss": 1.4249,
+      "step": 1887
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1682308522114346e-05,
+      "loss": 1.3848,
+      "step": 1888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1682087287221399e-05,
+      "loss": 1.4939,
+      "step": 1889
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1681866052328451e-05,
+      "loss": 1.6671,
+      "step": 1890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1681644817435502e-05,
+      "loss": 1.5181,
+      "step": 1891
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1681423582542555e-05,
+      "loss": 1.6676,
+      "step": 1892
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1681202347649608e-05,
+      "loss": 1.5429,
+      "step": 1893
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.168098111275666e-05,
+      "loss": 1.6409,
+      "step": 1894
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1680759877863711e-05,
+      "loss": 1.4884,
+      "step": 1895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1680538642970764e-05,
+      "loss": 1.4422,
+      "step": 1896
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1680317408077815e-05,
+      "loss": 1.5703,
+      "step": 1897
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1680096173184867e-05,
+      "loss": 1.4877,
+      "step": 1898
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.167987493829192e-05,
+      "loss": 1.4729,
+      "step": 1899
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1679653703398972e-05,
+      "loss": 1.5844,
+      "step": 1900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1679432468506025e-05,
+      "loss": 1.3139,
+      "step": 1901
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1679211233613076e-05,
+      "loss": 1.5421,
+      "step": 1902
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1678989998720129e-05,
+      "loss": 1.7077,
+      "step": 1903
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.167876876382718e-05,
+      "loss": 1.3629,
+      "step": 1904
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1678547528934232e-05,
+      "loss": 1.6707,
+      "step": 1905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1678326294041285e-05,
+      "loss": 1.4556,
+      "step": 1906
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1678105059148337e-05,
+      "loss": 1.4428,
+      "step": 1907
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1677883824255388e-05,
+      "loss": 1.679,
+      "step": 1908
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1677662589362441e-05,
+      "loss": 1.4973,
+      "step": 1909
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1677441354469494e-05,
+      "loss": 1.5226,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1677220119576544e-05,
+      "loss": 1.4116,
+      "step": 1911
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1676998884683597e-05,
+      "loss": 1.6343,
+      "step": 1912
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1676777649790648e-05,
+      "loss": 1.5717,
+      "step": 1913
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.16765564148977e-05,
+      "loss": 1.468,
+      "step": 1914
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1676335180004753e-05,
+      "loss": 1.5117,
+      "step": 1915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1676113945111806e-05,
+      "loss": 1.4348,
+      "step": 1916
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1675892710218858e-05,
+      "loss": 1.4685,
+      "step": 1917
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.167567147532591e-05,
+      "loss": 1.5267,
+      "step": 1918
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1675450240432962e-05,
+      "loss": 1.4735,
+      "step": 1919
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1675229005540013e-05,
+      "loss": 1.4911,
+      "step": 1920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1675007770647065e-05,
+      "loss": 1.5358,
+      "step": 1921
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1674786535754118e-05,
+      "loss": 1.5344,
+      "step": 1922
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.167456530086117e-05,
+      "loss": 1.4857,
+      "step": 1923
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1674344065968222e-05,
+      "loss": 1.5738,
+      "step": 1924
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1674122831075274e-05,
+      "loss": 1.6216,
+      "step": 1925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1673901596182327e-05,
+      "loss": 1.4817,
+      "step": 1926
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1673680361289378e-05,
+      "loss": 1.4154,
+      "step": 1927
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.167345912639643e-05,
+      "loss": 1.6349,
+      "step": 1928
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1673237891503481e-05,
+      "loss": 1.5871,
+      "step": 1929
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1673016656610534e-05,
+      "loss": 1.4483,
+      "step": 1930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1672795421717586e-05,
+      "loss": 1.4254,
+      "step": 1931
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1672574186824639e-05,
+      "loss": 1.6085,
+      "step": 1932
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1672352951931692e-05,
+      "loss": 1.4651,
+      "step": 1933
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1672131717038744e-05,
+      "loss": 1.4236,
+      "step": 1934
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1671910482145795e-05,
+      "loss": 1.4454,
+      "step": 1935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1671689247252846e-05,
+      "loss": 1.4643,
+      "step": 1936
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1671468012359899e-05,
+      "loss": 1.6012,
+      "step": 1937
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1671246777466951e-05,
+      "loss": 1.5521,
+      "step": 1938
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1671025542574004e-05,
+      "loss": 1.6214,
+      "step": 1939
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1670804307681055e-05,
+      "loss": 1.4333,
+      "step": 1940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1670583072788108e-05,
+      "loss": 1.5065,
+      "step": 1941
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.167036183789516e-05,
+      "loss": 1.4154,
+      "step": 1942
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1670140603002211e-05,
+      "loss": 1.4173,
+      "step": 1943
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1669919368109264e-05,
+      "loss": 1.5965,
+      "step": 1944
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1669698133216316e-05,
+      "loss": 1.4688,
+      "step": 1945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1669476898323367e-05,
+      "loss": 1.5485,
+      "step": 1946
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.166925566343042e-05,
+      "loss": 1.4462,
+      "step": 1947
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1669034428537472e-05,
+      "loss": 1.5232,
+      "step": 1948
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1668813193644525e-05,
+      "loss": 1.5645,
+      "step": 1949
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1668591958751578e-05,
+      "loss": 1.4831,
+      "step": 1950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1668370723858629e-05,
+      "loss": 1.5504,
+      "step": 1951
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.166814948896568e-05,
+      "loss": 1.5897,
+      "step": 1952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1667928254072732e-05,
+      "loss": 1.3548,
+      "step": 1953
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1667707019179785e-05,
+      "loss": 1.572,
+      "step": 1954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1667485784286837e-05,
+      "loss": 1.3793,
+      "step": 1955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.166726454939389e-05,
+      "loss": 1.6614,
+      "step": 1956
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1667043314500941e-05,
+      "loss": 1.3716,
+      "step": 1957
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1666822079607993e-05,
+      "loss": 1.5366,
+      "step": 1958
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1666600844715044e-05,
+      "loss": 1.5852,
+      "step": 1959
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1666379609822097e-05,
+      "loss": 1.3857,
+      "step": 1960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.166615837492915e-05,
+      "loss": 1.6769,
+      "step": 1961
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.16659371400362e-05,
+      "loss": 1.7326,
+      "step": 1962
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1665715905143253e-05,
+      "loss": 1.4359,
+      "step": 1963
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1665494670250306e-05,
+      "loss": 1.5722,
+      "step": 1964
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1665273435357358e-05,
+      "loss": 1.7678,
+      "step": 1965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1665052200464411e-05,
+      "loss": 1.5301,
+      "step": 1966
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1664830965571462e-05,
+      "loss": 1.434,
+      "step": 1967
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1664609730678513e-05,
+      "loss": 1.3622,
+      "step": 1968
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1664388495785565e-05,
+      "loss": 1.464,
+      "step": 1969
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1664167260892618e-05,
+      "loss": 1.5189,
+      "step": 1970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.166394602599967e-05,
+      "loss": 1.5643,
+      "step": 1971
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1663724791106723e-05,
+      "loss": 1.5845,
+      "step": 1972
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1663503556213774e-05,
+      "loss": 1.7116,
+      "step": 1973
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1663282321320827e-05,
+      "loss": 1.5838,
+      "step": 1974
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1663061086427878e-05,
+      "loss": 1.5275,
+      "step": 1975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.166283985153493e-05,
+      "loss": 1.5436,
+      "step": 1976
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1662618616641983e-05,
+      "loss": 1.4945,
+      "step": 1977
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1662397381749036e-05,
+      "loss": 1.5542,
+      "step": 1978
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1662176146856086e-05,
+      "loss": 1.4362,
+      "step": 1979
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1661954911963139e-05,
+      "loss": 1.512,
+      "step": 1980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1661733677070192e-05,
+      "loss": 1.5363,
+      "step": 1981
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1661512442177244e-05,
+      "loss": 1.5419,
+      "step": 1982
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1661291207284295e-05,
+      "loss": 1.5021,
+      "step": 1983
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1661069972391346e-05,
+      "loss": 1.4318,
+      "step": 1984
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1660848737498399e-05,
+      "loss": 1.4557,
+      "step": 1985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1660627502605451e-05,
+      "loss": 1.6528,
+      "step": 1986
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1660406267712504e-05,
+      "loss": 1.4556,
+      "step": 1987
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1660185032819557e-05,
+      "loss": 1.5818,
+      "step": 1988
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.165996379792661e-05,
+      "loss": 1.5254,
+      "step": 1989
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.165974256303366e-05,
+      "loss": 1.5501,
+      "step": 1990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1659521328140713e-05,
+      "loss": 1.4757,
+      "step": 1991
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1659300093247764e-05,
+      "loss": 1.4778,
+      "step": 1992
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1659078858354816e-05,
+      "loss": 1.4903,
+      "step": 1993
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1658857623461869e-05,
+      "loss": 1.5557,
+      "step": 1994
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.165863638856892e-05,
+      "loss": 1.5076,
+      "step": 1995
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1658415153675972e-05,
+      "loss": 1.5895,
+      "step": 1996
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1658193918783025e-05,
+      "loss": 1.4707,
+      "step": 1997
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1657972683890078e-05,
+      "loss": 1.6641,
+      "step": 1998
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1657751448997129e-05,
+      "loss": 1.4778,
+      "step": 1999
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1657530214104181e-05,
+      "loss": 1.6383,
+      "step": 2000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1657308979211232e-05,
+      "loss": 1.6923,
+      "step": 2001
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1657087744318285e-05,
+      "loss": 1.4163,
+      "step": 2002
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1656866509425337e-05,
+      "loss": 1.5441,
+      "step": 2003
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.165664527453239e-05,
+      "loss": 1.5172,
+      "step": 2004
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1656424039639442e-05,
+      "loss": 1.5903,
+      "step": 2005
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1656202804746493e-05,
+      "loss": 1.6002,
+      "step": 2006
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1655981569853546e-05,
+      "loss": 1.4565,
+      "step": 2007
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1655760334960597e-05,
+      "loss": 1.5237,
+      "step": 2008
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.165553910006765e-05,
+      "loss": 1.5003,
+      "step": 2009
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1655317865174702e-05,
+      "loss": 1.6026,
+      "step": 2010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1655096630281753e-05,
+      "loss": 1.4743,
+      "step": 2011
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1654875395388806e-05,
+      "loss": 1.6372,
+      "step": 2012
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1654654160495858e-05,
+      "loss": 1.4341,
+      "step": 2013
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1654432925602911e-05,
+      "loss": 1.3972,
+      "step": 2014
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1654211690709962e-05,
+      "loss": 1.6396,
+      "step": 2015
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1653990455817014e-05,
+      "loss": 1.486,
+      "step": 2016
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1653769220924065e-05,
+      "loss": 1.5211,
+      "step": 2017
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1653547986031118e-05,
+      "loss": 1.6022,
+      "step": 2018
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.165332675113817e-05,
+      "loss": 1.5854,
+      "step": 2019
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1653105516245223e-05,
+      "loss": 1.5919,
+      "step": 2020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1652884281352276e-05,
+      "loss": 1.3884,
+      "step": 2021
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1652663046459327e-05,
+      "loss": 1.573,
+      "step": 2022
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.165244181156638e-05,
+      "loss": 1.4852,
+      "step": 2023
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.165222057667343e-05,
+      "loss": 1.4246,
+      "step": 2024
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1651999341780483e-05,
+      "loss": 1.4304,
+      "step": 2025
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1651778106887535e-05,
+      "loss": 1.5752,
+      "step": 2026
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1651556871994588e-05,
+      "loss": 1.6078,
+      "step": 2027
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1651335637101639e-05,
+      "loss": 1.4825,
+      "step": 2028
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1651114402208692e-05,
+      "loss": 1.5778,
+      "step": 2029
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1650893167315744e-05,
+      "loss": 1.4869,
+      "step": 2030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1650671932422795e-05,
+      "loss": 1.5216,
+      "step": 2031
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1650450697529848e-05,
+      "loss": 1.5925,
+      "step": 2032
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1650229462636899e-05,
+      "loss": 1.6947,
+      "step": 2033
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1650008227743951e-05,
+      "loss": 1.5287,
+      "step": 2034
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1649786992851004e-05,
+      "loss": 1.7078,
+      "step": 2035
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1649565757958057e-05,
+      "loss": 1.5401,
+      "step": 2036
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1649344523065109e-05,
+      "loss": 1.7957,
+      "step": 2037
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1649123288172162e-05,
+      "loss": 1.5458,
+      "step": 2038
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1648902053279213e-05,
+      "loss": 1.4783,
+      "step": 2039
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1648680818386264e-05,
+      "loss": 1.6799,
+      "step": 2040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1648459583493316e-05,
+      "loss": 1.6484,
+      "step": 2041
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1648238348600369e-05,
+      "loss": 1.4557,
+      "step": 2042
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1648017113707421e-05,
+      "loss": 1.5986,
+      "step": 2043
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1647795878814472e-05,
+      "loss": 1.584,
+      "step": 2044
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1647574643921525e-05,
+      "loss": 1.6356,
+      "step": 2045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1647353409028578e-05,
+      "loss": 1.3973,
+      "step": 2046
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1647132174135628e-05,
+      "loss": 1.5743,
+      "step": 2047
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1646910939242681e-05,
+      "loss": 1.5241,
+      "step": 2048
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1646689704349734e-05,
+      "loss": 1.4842,
+      "step": 2049
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1646468469456785e-05,
+      "loss": 1.5485,
+      "step": 2050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1646247234563837e-05,
+      "loss": 1.6474,
+      "step": 2051
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.164602599967089e-05,
+      "loss": 1.3832,
+      "step": 2052
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1645804764777942e-05,
+      "loss": 1.2946,
+      "step": 2053
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1645583529884995e-05,
+      "loss": 1.4296,
+      "step": 2054
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1645362294992046e-05,
+      "loss": 1.5888,
+      "step": 2055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1645141060099097e-05,
+      "loss": 1.5326,
+      "step": 2056
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.164491982520615e-05,
+      "loss": 1.4678,
+      "step": 2057
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1644698590313202e-05,
+      "loss": 1.6886,
+      "step": 2058
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1644477355420255e-05,
+      "loss": 1.4824,
+      "step": 2059
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1644256120527307e-05,
+      "loss": 1.6962,
+      "step": 2060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1644034885634358e-05,
+      "loss": 1.757,
+      "step": 2061
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1643813650741411e-05,
+      "loss": 1.4806,
+      "step": 2062
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1643592415848462e-05,
+      "loss": 1.508,
+      "step": 2063
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1643371180955514e-05,
+      "loss": 1.4728,
+      "step": 2064
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1643149946062567e-05,
+      "loss": 1.5802,
+      "step": 2065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1642928711169618e-05,
+      "loss": 1.395,
+      "step": 2066
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.164270747627667e-05,
+      "loss": 1.5318,
+      "step": 2067
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1642486241383723e-05,
+      "loss": 1.5826,
+      "step": 2068
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1642265006490776e-05,
+      "loss": 1.3971,
+      "step": 2069
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1642043771597828e-05,
+      "loss": 1.5474,
+      "step": 2070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.164182253670488e-05,
+      "loss": 1.7251,
+      "step": 2071
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.164160130181193e-05,
+      "loss": 1.5,
+      "step": 2072
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1641380066918983e-05,
+      "loss": 1.5672,
+      "step": 2073
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1641158832026035e-05,
+      "loss": 1.4291,
+      "step": 2074
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1640937597133088e-05,
+      "loss": 1.5715,
+      "step": 2075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.164071636224014e-05,
+      "loss": 1.4635,
+      "step": 2076
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1640495127347192e-05,
+      "loss": 1.4615,
+      "step": 2077
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1640273892454244e-05,
+      "loss": 1.6267,
+      "step": 2078
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1640052657561297e-05,
+      "loss": 1.5023,
+      "step": 2079
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1639831422668348e-05,
+      "loss": 1.6213,
+      "step": 2080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.16396101877754e-05,
+      "loss": 1.5183,
+      "step": 2081
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1639388952882453e-05,
+      "loss": 1.6415,
+      "step": 2082
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1639167717989504e-05,
+      "loss": 1.4986,
+      "step": 2083
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1638946483096556e-05,
+      "loss": 1.6715,
+      "step": 2084
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1638725248203609e-05,
+      "loss": 1.4517,
+      "step": 2085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1638504013310662e-05,
+      "loss": 1.4815,
+      "step": 2086
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1638282778417713e-05,
+      "loss": 1.7037,
+      "step": 2087
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1638061543524764e-05,
+      "loss": 1.8543,
+      "step": 2088
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1637840308631816e-05,
+      "loss": 1.5092,
+      "step": 2089
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1637619073738869e-05,
+      "loss": 1.5933,
+      "step": 2090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1637397838845921e-05,
+      "loss": 1.4895,
+      "step": 2091
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1637176603952974e-05,
+      "loss": 1.477,
+      "step": 2092
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1636955369060027e-05,
+      "loss": 1.5498,
+      "step": 2093
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1636734134167078e-05,
+      "loss": 1.4955,
+      "step": 2094
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.163651289927413e-05,
+      "loss": 1.6561,
+      "step": 2095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1636291664381181e-05,
+      "loss": 1.4273,
+      "step": 2096
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1636070429488234e-05,
+      "loss": 1.3743,
+      "step": 2097
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1635849194595286e-05,
+      "loss": 1.4706,
+      "step": 2098
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1635627959702337e-05,
+      "loss": 1.6745,
+      "step": 2099
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.163540672480939e-05,
+      "loss": 1.4651,
+      "step": 2100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1635185489916442e-05,
+      "loss": 1.5641,
+      "step": 2101
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1634964255023495e-05,
+      "loss": 1.6744,
+      "step": 2102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1634743020130546e-05,
+      "loss": 1.612,
+      "step": 2103
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1634521785237597e-05,
+      "loss": 1.3905,
+      "step": 2104
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.163430055034465e-05,
+      "loss": 1.4088,
+      "step": 2105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1634079315451702e-05,
+      "loss": 1.7409,
+      "step": 2106
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1633858080558755e-05,
+      "loss": 1.4782,
+      "step": 2107
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1633636845665807e-05,
+      "loss": 1.5567,
+      "step": 2108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.163341561077286e-05,
+      "loss": 1.5156,
+      "step": 2109
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.163319437587991e-05,
+      "loss": 1.4047,
+      "step": 2110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1632973140986963e-05,
+      "loss": 1.499,
+      "step": 2111
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1632751906094014e-05,
+      "loss": 1.5247,
+      "step": 2112
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1632530671201067e-05,
+      "loss": 1.4925,
+      "step": 2113
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.163230943630812e-05,
+      "loss": 1.5112,
+      "step": 2114
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.163208820141517e-05,
+      "loss": 1.4207,
+      "step": 2115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1631866966522223e-05,
+      "loss": 1.413,
+      "step": 2116
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1631645731629276e-05,
+      "loss": 1.4858,
+      "step": 2117
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1631424496736328e-05,
+      "loss": 1.6191,
+      "step": 2118
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.163120326184338e-05,
+      "loss": 1.5077,
+      "step": 2119
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1630982026950432e-05,
+      "loss": 1.5153,
+      "step": 2120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1630760792057483e-05,
+      "loss": 1.5273,
+      "step": 2121
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1630539557164535e-05,
+      "loss": 1.5555,
+      "step": 2122
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1630318322271588e-05,
+      "loss": 1.5934,
+      "step": 2123
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.163009708737864e-05,
+      "loss": 1.4256,
+      "step": 2124
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1629875852485693e-05,
+      "loss": 1.5723,
+      "step": 2125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1629654617592744e-05,
+      "loss": 1.4272,
+      "step": 2126
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1629433382699797e-05,
+      "loss": 1.621,
+      "step": 2127
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1629212147806848e-05,
+      "loss": 1.3723,
+      "step": 2128
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.16289909129139e-05,
+      "loss": 1.7025,
+      "step": 2129
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1628769678020953e-05,
+      "loss": 1.5123,
+      "step": 2130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1628548443128006e-05,
+      "loss": 1.461,
+      "step": 2131
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1628327208235056e-05,
+      "loss": 1.5033,
+      "step": 2132
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1628105973342109e-05,
+      "loss": 1.4931,
+      "step": 2133
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1627884738449162e-05,
+      "loss": 1.569,
+      "step": 2134
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1627663503556213e-05,
+      "loss": 1.5393,
+      "step": 2135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1627442268663265e-05,
+      "loss": 1.3924,
+      "step": 2136
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1627221033770316e-05,
+      "loss": 1.448,
+      "step": 2137
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1626999798877369e-05,
+      "loss": 1.552,
+      "step": 2138
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1626778563984421e-05,
+      "loss": 1.5395,
+      "step": 2139
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1626557329091474e-05,
+      "loss": 1.6684,
+      "step": 2140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1626336094198527e-05,
+      "loss": 1.4759,
+      "step": 2141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.162611485930558e-05,
+      "loss": 1.5407,
+      "step": 2142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.162589362441263e-05,
+      "loss": 1.4788,
+      "step": 2143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1625672389519681e-05,
+      "loss": 1.6396,
+      "step": 2144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1625451154626734e-05,
+      "loss": 1.3509,
+      "step": 2145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1625229919733786e-05,
+      "loss": 1.5084,
+      "step": 2146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1625008684840839e-05,
+      "loss": 1.3445,
+      "step": 2147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.162478744994789e-05,
+      "loss": 1.5774,
+      "step": 2148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1624566215054942e-05,
+      "loss": 1.5486,
+      "step": 2149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1624344980161995e-05,
+      "loss": 1.6482,
+      "step": 2150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1624123745269046e-05,
+      "loss": 1.6114,
+      "step": 2151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1623902510376099e-05,
+      "loss": 1.476,
+      "step": 2152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1623681275483151e-05,
+      "loss": 1.4338,
+      "step": 2153
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1623460040590202e-05,
+      "loss": 1.4733,
+      "step": 2154
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1623238805697255e-05,
+      "loss": 1.5021,
+      "step": 2155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1623017570804307e-05,
+      "loss": 1.6941,
+      "step": 2156
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.162279633591136e-05,
+      "loss": 1.4684,
+      "step": 2157
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1622575101018412e-05,
+      "loss": 1.5873,
+      "step": 2158
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1622353866125463e-05,
+      "loss": 1.6725,
+      "step": 2159
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1622132631232514e-05,
+      "loss": 1.7165,
+      "step": 2160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1621911396339567e-05,
+      "loss": 1.3502,
+      "step": 2161
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.162169016144662e-05,
+      "loss": 1.4881,
+      "step": 2162
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1621468926553672e-05,
+      "loss": 1.6079,
+      "step": 2163
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1621247691660725e-05,
+      "loss": 1.395,
+      "step": 2164
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1621026456767776e-05,
+      "loss": 1.4763,
+      "step": 2165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1620805221874828e-05,
+      "loss": 1.4603,
+      "step": 2166
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1620583986981881e-05,
+      "loss": 1.5382,
+      "step": 2167
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1620362752088932e-05,
+      "loss": 1.4238,
+      "step": 2168
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1620141517195984e-05,
+      "loss": 1.6149,
+      "step": 2169
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1619920282303035e-05,
+      "loss": 1.5312,
+      "step": 2170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1619699047410088e-05,
+      "loss": 1.5651,
+      "step": 2171
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.161947781251714e-05,
+      "loss": 1.5105,
+      "step": 2172
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1619256577624193e-05,
+      "loss": 1.5041,
+      "step": 2173
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1619035342731246e-05,
+      "loss": 1.6125,
+      "step": 2174
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1618814107838297e-05,
+      "loss": 1.5264,
+      "step": 2175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1618592872945348e-05,
+      "loss": 1.5605,
+      "step": 2176
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.16183716380524e-05,
+      "loss": 1.5766,
+      "step": 2177
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1618150403159453e-05,
+      "loss": 1.3598,
+      "step": 2178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1617929168266505e-05,
+      "loss": 1.5335,
+      "step": 2179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1617707933373558e-05,
+      "loss": 1.5038,
+      "step": 2180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1617486698480609e-05,
+      "loss": 1.4329,
+      "step": 2181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1617265463587662e-05,
+      "loss": 1.464,
+      "step": 2182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1617044228694714e-05,
+      "loss": 1.6946,
+      "step": 2183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1616822993801765e-05,
+      "loss": 1.7097,
+      "step": 2184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1616601758908818e-05,
+      "loss": 1.5423,
+      "step": 2185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.161638052401587e-05,
+      "loss": 1.6459,
+      "step": 2186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1616159289122921e-05,
+      "loss": 1.5848,
+      "step": 2187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1615938054229974e-05,
+      "loss": 1.4344,
+      "step": 2188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1615716819337027e-05,
+      "loss": 1.4699,
+      "step": 2189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1615495584444079e-05,
+      "loss": 1.6047,
+      "step": 2190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.161527434955113e-05,
+      "loss": 1.6237,
+      "step": 2191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1615053114658181e-05,
+      "loss": 1.5384,
+      "step": 2192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1614831879765234e-05,
+      "loss": 1.5035,
+      "step": 2193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1614610644872286e-05,
+      "loss": 1.6096,
+      "step": 2194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1614389409979339e-05,
+      "loss": 1.7242,
+      "step": 2195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1614168175086391e-05,
+      "loss": 1.5992,
+      "step": 2196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1613946940193444e-05,
+      "loss": 1.5038,
+      "step": 2197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1613725705300495e-05,
+      "loss": 1.6348,
+      "step": 2198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1613504470407548e-05,
+      "loss": 1.409,
+      "step": 2199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1613283235514598e-05,
+      "loss": 1.4966,
+      "step": 2200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1613062000621651e-05,
+      "loss": 1.5209,
+      "step": 2201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1612840765728704e-05,
+      "loss": 1.4829,
+      "step": 2202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1612619530835755e-05,
+      "loss": 1.5188,
+      "step": 2203
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1612398295942807e-05,
+      "loss": 1.4982,
+      "step": 2204
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.161217706104986e-05,
+      "loss": 1.6944,
+      "step": 2205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1611955826156912e-05,
+      "loss": 1.357,
+      "step": 2206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1611734591263963e-05,
+      "loss": 1.5808,
+      "step": 2207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1611513356371014e-05,
+      "loss": 1.5219,
+      "step": 2208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1611292121478067e-05,
+      "loss": 1.442,
+      "step": 2209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.161107088658512e-05,
+      "loss": 1.5086,
+      "step": 2210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1610849651692172e-05,
+      "loss": 1.4577,
+      "step": 2211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1610628416799225e-05,
+      "loss": 1.5063,
+      "step": 2212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1610407181906277e-05,
+      "loss": 1.4776,
+      "step": 2213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1610185947013328e-05,
+      "loss": 1.4394,
+      "step": 2214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1609964712120381e-05,
+      "loss": 1.5723,
+      "step": 2215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1609743477227432e-05,
+      "loss": 1.4545,
+      "step": 2216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1609522242334484e-05,
+      "loss": 1.5819,
+      "step": 2217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1609301007441537e-05,
+      "loss": 1.4935,
+      "step": 2218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1609079772548588e-05,
+      "loss": 1.4677,
+      "step": 2219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.160885853765564e-05,
+      "loss": 1.5916,
+      "step": 2220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1608637302762693e-05,
+      "loss": 1.702,
+      "step": 2221
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1608416067869746e-05,
+      "loss": 1.4946,
+      "step": 2222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1608194832976797e-05,
+      "loss": 1.4607,
+      "step": 2223
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.160797359808385e-05,
+      "loss": 1.4314,
+      "step": 2224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.16077523631909e-05,
+      "loss": 1.5486,
+      "step": 2225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1607531128297953e-05,
+      "loss": 1.5109,
+      "step": 2226
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1607309893405005e-05,
+      "loss": 1.4816,
+      "step": 2227
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1607088658512058e-05,
+      "loss": 1.5089,
+      "step": 2228
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.160686742361911e-05,
+      "loss": 1.5006,
+      "step": 2229
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1606646188726162e-05,
+      "loss": 1.4346,
+      "step": 2230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1606424953833214e-05,
+      "loss": 1.5081,
+      "step": 2231
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1606203718940265e-05,
+      "loss": 1.4355,
+      "step": 2232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1605982484047318e-05,
+      "loss": 1.5266,
+      "step": 2233
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.160576124915437e-05,
+      "loss": 1.4667,
+      "step": 2234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1605540014261423e-05,
+      "loss": 1.6039,
+      "step": 2235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1605318779368474e-05,
+      "loss": 1.6427,
+      "step": 2236
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1605097544475526e-05,
+      "loss": 1.5463,
+      "step": 2237
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1604876309582579e-05,
+      "loss": 1.585,
+      "step": 2238
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.160465507468963e-05,
+      "loss": 1.484,
+      "step": 2239
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1604433839796683e-05,
+      "loss": 1.4737,
+      "step": 2240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1604212604903734e-05,
+      "loss": 1.4941,
+      "step": 2241
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1603991370010786e-05,
+      "loss": 1.5005,
+      "step": 2242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1603770135117839e-05,
+      "loss": 1.476,
+      "step": 2243
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1603548900224891e-05,
+      "loss": 1.3945,
+      "step": 2244
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1603327665331944e-05,
+      "loss": 1.3656,
+      "step": 2245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1603106430438997e-05,
+      "loss": 1.5813,
+      "step": 2246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1602885195546048e-05,
+      "loss": 1.3966,
+      "step": 2247
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1602663960653098e-05,
+      "loss": 1.6714,
+      "step": 2248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1602442725760151e-05,
+      "loss": 1.551,
+      "step": 2249
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1602221490867204e-05,
+      "loss": 1.5672,
+      "step": 2250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1602000255974256e-05,
+      "loss": 1.6582,
+      "step": 2251
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1601779021081307e-05,
+      "loss": 1.4907,
+      "step": 2252
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.160155778618836e-05,
+      "loss": 1.5007,
+      "step": 2253
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1601336551295412e-05,
+      "loss": 1.5459,
+      "step": 2254
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1601115316402465e-05,
+      "loss": 1.4219,
+      "step": 2255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1600894081509516e-05,
+      "loss": 1.5044,
+      "step": 2256
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1600672846616569e-05,
+      "loss": 1.4176,
+      "step": 2257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.160045161172362e-05,
+      "loss": 1.4196,
+      "step": 2258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1600230376830672e-05,
+      "loss": 1.4316,
+      "step": 2259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1600009141937725e-05,
+      "loss": 1.3981,
+      "step": 2260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1599787907044777e-05,
+      "loss": 1.5127,
+      "step": 2261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.159956667215183e-05,
+      "loss": 1.4623,
+      "step": 2262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.159934543725888e-05,
+      "loss": 1.6145,
+      "step": 2263
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1599124202365932e-05,
+      "loss": 1.6525,
+      "step": 2264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1598902967472984e-05,
+      "loss": 1.3922,
+      "step": 2265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1598681732580037e-05,
+      "loss": 1.4782,
+      "step": 2266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.159846049768709e-05,
+      "loss": 1.4846,
+      "step": 2267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1598239262794142e-05,
+      "loss": 1.492,
+      "step": 2268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1598018027901193e-05,
+      "loss": 1.5959,
+      "step": 2269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1597796793008246e-05,
+      "loss": 1.4171,
+      "step": 2270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1597575558115298e-05,
+      "loss": 1.5353,
+      "step": 2271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.159735432322235e-05,
+      "loss": 1.5537,
+      "step": 2272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1597133088329402e-05,
+      "loss": 1.6307,
+      "step": 2273
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1596911853436453e-05,
+      "loss": 1.5333,
+      "step": 2274
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1596690618543505e-05,
+      "loss": 1.5278,
+      "step": 2275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1596469383650558e-05,
+      "loss": 1.6533,
+      "step": 2276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.159624814875761e-05,
+      "loss": 1.4859,
+      "step": 2277
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1596026913864663e-05,
+      "loss": 1.5106,
+      "step": 2278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1595805678971714e-05,
+      "loss": 1.4514,
+      "step": 2279
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1595584444078765e-05,
+      "loss": 1.4765,
+      "step": 2280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1595363209185818e-05,
+      "loss": 1.4672,
+      "step": 2281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.159514197429287e-05,
+      "loss": 1.5334,
+      "step": 2282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1594920739399923e-05,
+      "loss": 1.32,
+      "step": 2283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1594699504506976e-05,
+      "loss": 1.457,
+      "step": 2284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1594478269614026e-05,
+      "loss": 1.4577,
+      "step": 2285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1594257034721079e-05,
+      "loss": 1.5793,
+      "step": 2286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1594035799828132e-05,
+      "loss": 1.4258,
+      "step": 2287
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1593814564935183e-05,
+      "loss": 1.5204,
+      "step": 2288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1593593330042235e-05,
+      "loss": 1.412,
+      "step": 2289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1593372095149288e-05,
+      "loss": 1.63,
+      "step": 2290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1593150860256339e-05,
+      "loss": 1.433,
+      "step": 2291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1592929625363391e-05,
+      "loss": 1.4669,
+      "step": 2292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1592708390470444e-05,
+      "loss": 1.5266,
+      "step": 2293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1592487155577497e-05,
+      "loss": 1.5321,
+      "step": 2294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1592265920684547e-05,
+      "loss": 1.5695,
+      "step": 2295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1592044685791598e-05,
+      "loss": 1.4883,
+      "step": 2296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1591823450898651e-05,
+      "loss": 1.5313,
+      "step": 2297
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1591602216005704e-05,
+      "loss": 1.4865,
+      "step": 2298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1591380981112756e-05,
+      "loss": 1.5685,
+      "step": 2299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1591159746219809e-05,
+      "loss": 1.5998,
+      "step": 2300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.159093851132686e-05,
+      "loss": 1.591,
+      "step": 2301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1590717276433912e-05,
+      "loss": 1.5342,
+      "step": 2302
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1590496041540965e-05,
+      "loss": 1.5127,
+      "step": 2303
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1590274806648016e-05,
+      "loss": 1.5091,
+      "step": 2304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1590053571755068e-05,
+      "loss": 1.454,
+      "step": 2305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1589832336862121e-05,
+      "loss": 1.6432,
+      "step": 2306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1589611101969172e-05,
+      "loss": 1.5124,
+      "step": 2307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1589389867076225e-05,
+      "loss": 1.69,
+      "step": 2308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1589168632183277e-05,
+      "loss": 1.5397,
+      "step": 2309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.158894739729033e-05,
+      "loss": 1.5197,
+      "step": 2310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.158872616239738e-05,
+      "loss": 1.5284,
+      "step": 2311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1588504927504433e-05,
+      "loss": 1.4193,
+      "step": 2312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1588283692611484e-05,
+      "loss": 1.3815,
+      "step": 2313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1588062457718537e-05,
+      "loss": 1.5151,
+      "step": 2314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.158784122282559e-05,
+      "loss": 1.465,
+      "step": 2315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1587619987932642e-05,
+      "loss": 1.4756,
+      "step": 2316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1587398753039695e-05,
+      "loss": 1.3106,
+      "step": 2317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1587177518146746e-05,
+      "loss": 1.5648,
+      "step": 2318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1586956283253798e-05,
+      "loss": 1.6593,
+      "step": 2319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.158673504836085e-05,
+      "loss": 1.4266,
+      "step": 2320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1586513813467902e-05,
+      "loss": 1.5772,
+      "step": 2321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1586292578574954e-05,
+      "loss": 1.6671,
+      "step": 2322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1586071343682005e-05,
+      "loss": 1.5583,
+      "step": 2323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1585850108789058e-05,
+      "loss": 1.4406,
+      "step": 2324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.158562887389611e-05,
+      "loss": 1.4324,
+      "step": 2325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1585407639003163e-05,
+      "loss": 1.4953,
+      "step": 2326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1585186404110214e-05,
+      "loss": 1.5259,
+      "step": 2327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1584965169217267e-05,
+      "loss": 1.5727,
+      "step": 2328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1584743934324318e-05,
+      "loss": 1.4263,
+      "step": 2329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.158452269943137e-05,
+      "loss": 1.7916,
+      "step": 2330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1584301464538423e-05,
+      "loss": 1.4471,
+      "step": 2331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1584080229645475e-05,
+      "loss": 1.5524,
+      "step": 2332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1583858994752528e-05,
+      "loss": 1.7247,
+      "step": 2333
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1583637759859579e-05,
+      "loss": 1.4211,
+      "step": 2334
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1583416524966632e-05,
+      "loss": 1.5594,
+      "step": 2335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1583195290073683e-05,
+      "loss": 1.6638,
+      "step": 2336
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1582974055180735e-05,
+      "loss": 1.6774,
+      "step": 2337
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1582752820287788e-05,
+      "loss": 1.4223,
+      "step": 2338
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.158253158539484e-05,
+      "loss": 1.5223,
+      "step": 2339
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1582310350501891e-05,
+      "loss": 1.5061,
+      "step": 2340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1582089115608944e-05,
+      "loss": 1.6741,
+      "step": 2341
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1581867880715996e-05,
+      "loss": 1.5005,
+      "step": 2342
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1581646645823049e-05,
+      "loss": 1.443,
+      "step": 2343
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.15814254109301e-05,
+      "loss": 1.5546,
+      "step": 2344
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1581204176037151e-05,
+      "loss": 1.5484,
+      "step": 2345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1580982941144204e-05,
+      "loss": 1.6299,
+      "step": 2346
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1580761706251256e-05,
+      "loss": 1.5514,
+      "step": 2347
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1580540471358309e-05,
+      "loss": 1.3239,
+      "step": 2348
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1580319236465361e-05,
+      "loss": 1.5095,
+      "step": 2349
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1580098001572414e-05,
+      "loss": 1.4529,
+      "step": 2350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1579876766679465e-05,
+      "loss": 1.6975,
+      "step": 2351
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1579655531786516e-05,
+      "loss": 1.6607,
+      "step": 2352
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1579434296893568e-05,
+      "loss": 1.3952,
+      "step": 2353
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1579213062000621e-05,
+      "loss": 1.3496,
+      "step": 2354
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1578991827107674e-05,
+      "loss": 1.5551,
+      "step": 2355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1578770592214725e-05,
+      "loss": 1.4315,
+      "step": 2356
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1578549357321777e-05,
+      "loss": 1.4716,
+      "step": 2357
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.157832812242883e-05,
+      "loss": 1.5839,
+      "step": 2358
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1578106887535882e-05,
+      "loss": 1.4472,
+      "step": 2359
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1577885652642933e-05,
+      "loss": 1.547,
+      "step": 2360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1577664417749986e-05,
+      "loss": 1.395,
+      "step": 2361
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1577443182857037e-05,
+      "loss": 1.4793,
+      "step": 2362
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.157722194796409e-05,
+      "loss": 1.7402,
+      "step": 2363
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1577000713071142e-05,
+      "loss": 1.4586,
+      "step": 2364
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1576779478178195e-05,
+      "loss": 1.6409,
+      "step": 2365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1576558243285247e-05,
+      "loss": 1.4752,
+      "step": 2366
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1576337008392298e-05,
+      "loss": 1.4295,
+      "step": 2367
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.157611577349935e-05,
+      "loss": 1.4463,
+      "step": 2368
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1575894538606402e-05,
+      "loss": 1.4758,
+      "step": 2369
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1575673303713454e-05,
+      "loss": 1.4812,
+      "step": 2370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1575452068820507e-05,
+      "loss": 1.3368,
+      "step": 2371
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.157523083392756e-05,
+      "loss": 1.6692,
+      "step": 2372
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.157500959903461e-05,
+      "loss": 1.5596,
+      "step": 2373
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1574788364141663e-05,
+      "loss": 1.4044,
+      "step": 2374
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1574567129248716e-05,
+      "loss": 1.4182,
+      "step": 2375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1574345894355767e-05,
+      "loss": 1.4794,
+      "step": 2376
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.157412465946282e-05,
+      "loss": 1.4313,
+      "step": 2377
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.157390342456987e-05,
+      "loss": 1.4849,
+      "step": 2378
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1573682189676923e-05,
+      "loss": 1.6003,
+      "step": 2379
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1573460954783975e-05,
+      "loss": 1.6146,
+      "step": 2380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1573239719891028e-05,
+      "loss": 1.4794,
+      "step": 2381
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.157301848499808e-05,
+      "loss": 1.5522,
+      "step": 2382
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1572797250105132e-05,
+      "loss": 1.4125,
+      "step": 2383
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1572576015212182e-05,
+      "loss": 1.5399,
+      "step": 2384
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1572354780319235e-05,
+      "loss": 1.5606,
+      "step": 2385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1572133545426288e-05,
+      "loss": 1.4932,
+      "step": 2386
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.157191231053334e-05,
+      "loss": 1.6716,
+      "step": 2387
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1571691075640393e-05,
+      "loss": 1.4993,
+      "step": 2388
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1571469840747444e-05,
+      "loss": 1.5791,
+      "step": 2389
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1571248605854496e-05,
+      "loss": 1.4131,
+      "step": 2390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1571027370961549e-05,
+      "loss": 1.5131,
+      "step": 2391
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.15708061360686e-05,
+      "loss": 1.4498,
+      "step": 2392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1570584901175653e-05,
+      "loss": 1.4464,
+      "step": 2393
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1570363666282705e-05,
+      "loss": 1.5578,
+      "step": 2394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1570142431389756e-05,
+      "loss": 1.4333,
+      "step": 2395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1569921196496809e-05,
+      "loss": 1.5538,
+      "step": 2396
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1569699961603861e-05,
+      "loss": 1.503,
+      "step": 2397
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1569478726710914e-05,
+      "loss": 1.5506,
+      "step": 2398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1569257491817965e-05,
+      "loss": 1.5591,
+      "step": 2399
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1569036256925017e-05,
+      "loss": 1.5536,
+      "step": 2400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1568815022032068e-05,
+      "loss": 1.4352,
+      "step": 2401
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1568593787139121e-05,
+      "loss": 1.4833,
+      "step": 2402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1568372552246174e-05,
+      "loss": 1.6314,
+      "step": 2403
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1568151317353226e-05,
+      "loss": 1.3603,
+      "step": 2404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1567930082460277e-05,
+      "loss": 1.5776,
+      "step": 2405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.156770884756733e-05,
+      "loss": 1.4643,
+      "step": 2406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1567487612674382e-05,
+      "loss": 1.6105,
+      "step": 2407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1567266377781433e-05,
+      "loss": 1.4611,
+      "step": 2408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1567045142888486e-05,
+      "loss": 1.6094,
+      "step": 2409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1566823907995539e-05,
+      "loss": 1.5316,
+      "step": 2410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.156660267310259e-05,
+      "loss": 1.6121,
+      "step": 2411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1566381438209642e-05,
+      "loss": 1.5606,
+      "step": 2412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1566160203316695e-05,
+      "loss": 1.5163,
+      "step": 2413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1565938968423747e-05,
+      "loss": 1.6013,
+      "step": 2414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1565717733530798e-05,
+      "loss": 1.5619,
+      "step": 2415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.156549649863785e-05,
+      "loss": 1.5445,
+      "step": 2416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1565275263744902e-05,
+      "loss": 1.4027,
+      "step": 2417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1565054028851954e-05,
+      "loss": 1.353,
+      "step": 2418
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1564832793959007e-05,
+      "loss": 1.6046,
+      "step": 2419
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.156461155906606e-05,
+      "loss": 1.4563,
+      "step": 2420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1564390324173112e-05,
+      "loss": 1.569,
+      "step": 2421
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1564169089280163e-05,
+      "loss": 1.4858,
+      "step": 2422
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1563947854387216e-05,
+      "loss": 1.4468,
+      "step": 2423
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1563726619494267e-05,
+      "loss": 1.5751,
+      "step": 2424
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.156350538460132e-05,
+      "loss": 1.3779,
+      "step": 2425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1563284149708372e-05,
+      "loss": 1.6455,
+      "step": 2426
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1563062914815423e-05,
+      "loss": 1.5233,
+      "step": 2427
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1562841679922475e-05,
+      "loss": 1.4695,
+      "step": 2428
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1562620445029528e-05,
+      "loss": 1.4593,
+      "step": 2429
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.156239921013658e-05,
+      "loss": 1.7309,
+      "step": 2430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1562177975243632e-05,
+      "loss": 1.6237,
+      "step": 2431
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1561956740350684e-05,
+      "loss": 1.4248,
+      "step": 2432
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1561735505457735e-05,
+      "loss": 1.6123,
+      "step": 2433
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1561514270564788e-05,
+      "loss": 1.4094,
+      "step": 2434
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.156129303567184e-05,
+      "loss": 1.5018,
+      "step": 2435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1561071800778893e-05,
+      "loss": 1.4915,
+      "step": 2436
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1560850565885945e-05,
+      "loss": 1.6715,
+      "step": 2437
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1560629330992996e-05,
+      "loss": 1.6536,
+      "step": 2438
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1560408096100049e-05,
+      "loss": 1.5168,
+      "step": 2439
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.15601868612071e-05,
+      "loss": 1.5666,
+      "step": 2440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1559965626314153e-05,
+      "loss": 1.5165,
+      "step": 2441
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1559744391421205e-05,
+      "loss": 1.6161,
+      "step": 2442
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1559523156528258e-05,
+      "loss": 1.3796,
+      "step": 2443
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1559301921635309e-05,
+      "loss": 1.5111,
+      "step": 2444
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1559080686742361e-05,
+      "loss": 1.4286,
+      "step": 2445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1558859451849414e-05,
+      "loss": 1.5587,
+      "step": 2446
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1558638216956467e-05,
+      "loss": 1.3975,
+      "step": 2447
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1558416982063517e-05,
+      "loss": 1.5196,
+      "step": 2448
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1558195747170568e-05,
+      "loss": 1.5097,
+      "step": 2449
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1557974512277621e-05,
+      "loss": 1.4906,
+      "step": 2450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1557753277384674e-05,
+      "loss": 1.4491,
+      "step": 2451
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1557532042491726e-05,
+      "loss": 1.4785,
+      "step": 2452
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1557310807598779e-05,
+      "loss": 1.5077,
+      "step": 2453
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1557089572705831e-05,
+      "loss": 1.5166,
+      "step": 2454
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1556868337812882e-05,
+      "loss": 1.6752,
+      "step": 2455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1556647102919933e-05,
+      "loss": 1.764,
+      "step": 2456
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1556425868026986e-05,
+      "loss": 1.4303,
+      "step": 2457
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1556204633134038e-05,
+      "loss": 1.4767,
+      "step": 2458
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1555983398241091e-05,
+      "loss": 1.5558,
+      "step": 2459
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1555762163348142e-05,
+      "loss": 1.5581,
+      "step": 2460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1555540928455195e-05,
+      "loss": 1.3436,
+      "step": 2461
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1555319693562247e-05,
+      "loss": 1.5035,
+      "step": 2462
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.15550984586693e-05,
+      "loss": 1.4685,
+      "step": 2463
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.155487722377635e-05,
+      "loss": 1.6784,
+      "step": 2464
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1554655988883403e-05,
+      "loss": 1.4322,
+      "step": 2465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1554434753990454e-05,
+      "loss": 1.4626,
+      "step": 2466
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1554213519097507e-05,
+      "loss": 1.4674,
+      "step": 2467
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.155399228420456e-05,
+      "loss": 1.5209,
+      "step": 2468
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1553771049311612e-05,
+      "loss": 1.422,
+      "step": 2469
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1553549814418665e-05,
+      "loss": 1.5199,
+      "step": 2470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1553328579525716e-05,
+      "loss": 1.4716,
+      "step": 2471
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1553107344632767e-05,
+      "loss": 1.4182,
+      "step": 2472
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.155288610973982e-05,
+      "loss": 1.4902,
+      "step": 2473
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1552664874846872e-05,
+      "loss": 1.5591,
+      "step": 2474
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1552443639953924e-05,
+      "loss": 1.3694,
+      "step": 2475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1552222405060977e-05,
+      "loss": 1.5201,
+      "step": 2476
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1552001170168028e-05,
+      "loss": 1.4997,
+      "step": 2477
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.155177993527508e-05,
+      "loss": 1.4678,
+      "step": 2478
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1551558700382133e-05,
+      "loss": 1.4832,
+      "step": 2479
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1551337465489184e-05,
+      "loss": 1.6012,
+      "step": 2480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1551116230596237e-05,
+      "loss": 1.5837,
+      "step": 2481
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1550894995703288e-05,
+      "loss": 1.5554,
+      "step": 2482
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.155067376081034e-05,
+      "loss": 1.5742,
+      "step": 2483
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1550452525917393e-05,
+      "loss": 1.496,
+      "step": 2484
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1550231291024445e-05,
+      "loss": 1.5691,
+      "step": 2485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1550010056131498e-05,
+      "loss": 1.5315,
+      "step": 2486
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1549788821238549e-05,
+      "loss": 1.5624,
+      "step": 2487
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1549567586345602e-05,
+      "loss": 1.3993,
+      "step": 2488
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1549346351452653e-05,
+      "loss": 1.3291,
+      "step": 2489
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1549125116559705e-05,
+      "loss": 1.6304,
+      "step": 2490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1548903881666758e-05,
+      "loss": 1.4195,
+      "step": 2491
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.154868264677381e-05,
+      "loss": 1.6394,
+      "step": 2492
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1548461411880861e-05,
+      "loss": 1.5991,
+      "step": 2493
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1548240176987914e-05,
+      "loss": 1.3735,
+      "step": 2494
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1548018942094966e-05,
+      "loss": 1.4912,
+      "step": 2495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1547797707202017e-05,
+      "loss": 1.3895,
+      "step": 2496
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.154757647230907e-05,
+      "loss": 1.5041,
+      "step": 2497
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1547355237416121e-05,
+      "loss": 1.5189,
+      "step": 2498
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1547134002523174e-05,
+      "loss": 1.4169,
+      "step": 2499
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1546912767630226e-05,
+      "loss": 1.4942,
+      "step": 2500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1546691532737279e-05,
+      "loss": 1.5254,
+      "step": 2501
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1546470297844331e-05,
+      "loss": 1.3286,
+      "step": 2502
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1546249062951382e-05,
+      "loss": 1.4691,
+      "step": 2503
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1546027828058435e-05,
+      "loss": 1.4062,
+      "step": 2504
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1545806593165486e-05,
+      "loss": 1.5443,
+      "step": 2505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1545585358272538e-05,
+      "loss": 1.4507,
+      "step": 2506
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1545364123379591e-05,
+      "loss": 1.4596,
+      "step": 2507
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1545142888486644e-05,
+      "loss": 1.4272,
+      "step": 2508
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1544921653593695e-05,
+      "loss": 1.3661,
+      "step": 2509
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1544700418700747e-05,
+      "loss": 1.5411,
+      "step": 2510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.15444791838078e-05,
+      "loss": 1.4434,
+      "step": 2511
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.154425794891485e-05,
+      "loss": 1.5115,
+      "step": 2512
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1544036714021903e-05,
+      "loss": 1.4087,
+      "step": 2513
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1543815479128956e-05,
+      "loss": 1.5715,
+      "step": 2514
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1543594244236007e-05,
+      "loss": 1.4444,
+      "step": 2515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.154337300934306e-05,
+      "loss": 1.5693,
+      "step": 2516
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1543151774450112e-05,
+      "loss": 1.4497,
+      "step": 2517
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1542930539557165e-05,
+      "loss": 1.6624,
+      "step": 2518
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1542709304664216e-05,
+      "loss": 1.6058,
+      "step": 2519
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1542488069771268e-05,
+      "loss": 1.4316,
+      "step": 2520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1542266834878319e-05,
+      "loss": 1.6015,
+      "step": 2521
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1542045599985372e-05,
+      "loss": 1.4057,
+      "step": 2522
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1541824365092424e-05,
+      "loss": 1.4173,
+      "step": 2523
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1541603130199477e-05,
+      "loss": 1.5936,
+      "step": 2524
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.154138189530653e-05,
+      "loss": 1.5046,
+      "step": 2525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.154116066041358e-05,
+      "loss": 1.582,
+      "step": 2526
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1540939425520633e-05,
+      "loss": 1.5294,
+      "step": 2527
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1540718190627684e-05,
+      "loss": 1.5558,
+      "step": 2528
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1540496955734737e-05,
+      "loss": 1.5747,
+      "step": 2529
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.154027572084179e-05,
+      "loss": 1.3832,
+      "step": 2530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.154005448594884e-05,
+      "loss": 1.4729,
+      "step": 2531
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1539833251055893e-05,
+      "loss": 1.5286,
+      "step": 2532
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1539612016162945e-05,
+      "loss": 1.487,
+      "step": 2533
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1539390781269998e-05,
+      "loss": 1.4011,
+      "step": 2534
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.153916954637705e-05,
+      "loss": 1.5676,
+      "step": 2535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1538948311484102e-05,
+      "loss": 1.4929,
+      "step": 2536
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1538727076591152e-05,
+      "loss": 1.7189,
+      "step": 2537
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1538505841698205e-05,
+      "loss": 1.3512,
+      "step": 2538
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1538284606805258e-05,
+      "loss": 1.4606,
+      "step": 2539
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.153806337191231e-05,
+      "loss": 1.3568,
+      "step": 2540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1537842137019363e-05,
+      "loss": 1.5603,
+      "step": 2541
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1537620902126414e-05,
+      "loss": 1.5588,
+      "step": 2542
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1537399667233466e-05,
+      "loss": 1.4147,
+      "step": 2543
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1537178432340517e-05,
+      "loss": 1.4867,
+      "step": 2544
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.153695719744757e-05,
+      "loss": 1.5189,
+      "step": 2545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1536735962554623e-05,
+      "loss": 1.4887,
+      "step": 2546
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1536514727661675e-05,
+      "loss": 1.5215,
+      "step": 2547
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1536293492768726e-05,
+      "loss": 1.4992,
+      "step": 2548
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1536072257875779e-05,
+      "loss": 1.6341,
+      "step": 2549
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1535851022982831e-05,
+      "loss": 1.5076,
+      "step": 2550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1535629788089884e-05,
+      "loss": 1.5423,
+      "step": 2551
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1535408553196935e-05,
+      "loss": 1.5195,
+      "step": 2552
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1535187318303986e-05,
+      "loss": 1.5474,
+      "step": 2553
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1534966083411038e-05,
+      "loss": 1.4724,
+      "step": 2554
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1534744848518091e-05,
+      "loss": 1.5723,
+      "step": 2555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1534523613625144e-05,
+      "loss": 1.586,
+      "step": 2556
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1534302378732196e-05,
+      "loss": 1.5199,
+      "step": 2557
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1534081143839249e-05,
+      "loss": 1.3908,
+      "step": 2558
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.15338599089463e-05,
+      "loss": 1.4709,
+      "step": 2559
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.153363867405335e-05,
+      "loss": 1.4113,
+      "step": 2560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1533417439160403e-05,
+      "loss": 1.4083,
+      "step": 2561
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1533196204267456e-05,
+      "loss": 1.5916,
+      "step": 2562
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1532974969374509e-05,
+      "loss": 1.4223,
+      "step": 2563
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.153275373448156e-05,
+      "loss": 1.4281,
+      "step": 2564
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1532532499588612e-05,
+      "loss": 1.6053,
+      "step": 2565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1532311264695665e-05,
+      "loss": 1.514,
+      "step": 2566
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1532090029802717e-05,
+      "loss": 1.3212,
+      "step": 2567
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1531868794909768e-05,
+      "loss": 1.4879,
+      "step": 2568
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.153164756001682e-05,
+      "loss": 1.3684,
+      "step": 2569
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1531426325123872e-05,
+      "loss": 1.5062,
+      "step": 2570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1531205090230924e-05,
+      "loss": 1.5026,
+      "step": 2571
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1530983855337977e-05,
+      "loss": 1.4792,
+      "step": 2572
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.153076262044503e-05,
+      "loss": 1.4866,
+      "step": 2573
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1530541385552082e-05,
+      "loss": 1.5692,
+      "step": 2574
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1530320150659133e-05,
+      "loss": 1.4231,
+      "step": 2575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1530098915766186e-05,
+      "loss": 1.6321,
+      "step": 2576
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1529877680873237e-05,
+      "loss": 1.5375,
+      "step": 2577
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.152965644598029e-05,
+      "loss": 1.4991,
+      "step": 2578
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1529435211087342e-05,
+      "loss": 1.4825,
+      "step": 2579
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1529213976194394e-05,
+      "loss": 1.4279,
+      "step": 2580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1528992741301445e-05,
+      "loss": 1.4754,
+      "step": 2581
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1528771506408498e-05,
+      "loss": 1.5031,
+      "step": 2582
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.152855027151555e-05,
+      "loss": 1.442,
+      "step": 2583
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1528329036622602e-05,
+      "loss": 1.4347,
+      "step": 2584
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1528107801729654e-05,
+      "loss": 1.3891,
+      "step": 2585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1527886566836705e-05,
+      "loss": 1.4467,
+      "step": 2586
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1527665331943758e-05,
+      "loss": 1.3766,
+      "step": 2587
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.152744409705081e-05,
+      "loss": 1.555,
+      "step": 2588
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1527222862157863e-05,
+      "loss": 1.5354,
+      "step": 2589
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1527001627264915e-05,
+      "loss": 1.4574,
+      "step": 2590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1526780392371966e-05,
+      "loss": 1.565,
+      "step": 2591
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1526559157479019e-05,
+      "loss": 1.4324,
+      "step": 2592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.152633792258607e-05,
+      "loss": 1.3494,
+      "step": 2593
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1526116687693123e-05,
+      "loss": 1.5167,
+      "step": 2594
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1525895452800175e-05,
+      "loss": 1.3767,
+      "step": 2595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1525674217907228e-05,
+      "loss": 1.4394,
+      "step": 2596
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1525452983014279e-05,
+      "loss": 1.6193,
+      "step": 2597
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1525231748121331e-05,
+      "loss": 1.4077,
+      "step": 2598
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1525010513228384e-05,
+      "loss": 1.6155,
+      "step": 2599
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1524789278335435e-05,
+      "loss": 1.431,
+      "step": 2600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1524568043442487e-05,
+      "loss": 1.3047,
+      "step": 2601
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1524346808549538e-05,
+      "loss": 1.4844,
+      "step": 2602
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1524125573656591e-05,
+      "loss": 1.6817,
+      "step": 2603
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1523904338763644e-05,
+      "loss": 1.4639,
+      "step": 2604
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1523683103870696e-05,
+      "loss": 1.4806,
+      "step": 2605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1523461868977749e-05,
+      "loss": 1.4724,
+      "step": 2606
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.15232406340848e-05,
+      "loss": 1.6929,
+      "step": 2607
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1523019399191852e-05,
+      "loss": 1.5921,
+      "step": 2608
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1522798164298903e-05,
+      "loss": 1.4461,
+      "step": 2609
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1522576929405956e-05,
+      "loss": 1.5887,
+      "step": 2610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1522355694513008e-05,
+      "loss": 1.5995,
+      "step": 2611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1522134459620061e-05,
+      "loss": 1.333,
+      "step": 2612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1521913224727112e-05,
+      "loss": 1.5992,
+      "step": 2613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1521691989834165e-05,
+      "loss": 1.5764,
+      "step": 2614
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1521470754941217e-05,
+      "loss": 1.4821,
+      "step": 2615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1521249520048268e-05,
+      "loss": 1.5794,
+      "step": 2616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.152102828515532e-05,
+      "loss": 1.4487,
+      "step": 2617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1520807050262373e-05,
+      "loss": 1.3862,
+      "step": 2618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1520585815369424e-05,
+      "loss": 1.4927,
+      "step": 2619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1520364580476477e-05,
+      "loss": 1.4971,
+      "step": 2620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.152014334558353e-05,
+      "loss": 1.4378,
+      "step": 2621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1519922110690582e-05,
+      "loss": 1.5581,
+      "step": 2622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1519700875797635e-05,
+      "loss": 1.5201,
+      "step": 2623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1519479640904686e-05,
+      "loss": 1.6255,
+      "step": 2624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1519258406011737e-05,
+      "loss": 1.5144,
+      "step": 2625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.151903717111879e-05,
+      "loss": 1.5121,
+      "step": 2626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1518815936225842e-05,
+      "loss": 1.4349,
+      "step": 2627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1518594701332894e-05,
+      "loss": 1.4734,
+      "step": 2628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1518373466439947e-05,
+      "loss": 1.5176,
+      "step": 2629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1518152231546998e-05,
+      "loss": 1.3629,
+      "step": 2630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.151793099665405e-05,
+      "loss": 1.409,
+      "step": 2631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1517709761761101e-05,
+      "loss": 1.5014,
+      "step": 2632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1517488526868154e-05,
+      "loss": 1.5629,
+      "step": 2633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1517267291975207e-05,
+      "loss": 1.3709,
+      "step": 2634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1517046057082258e-05,
+      "loss": 1.5989,
+      "step": 2635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.151682482218931e-05,
+      "loss": 1.574,
+      "step": 2636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1516603587296363e-05,
+      "loss": 1.5112,
+      "step": 2637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1516382352403415e-05,
+      "loss": 1.5636,
+      "step": 2638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1516161117510468e-05,
+      "loss": 1.4123,
+      "step": 2639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1515939882617519e-05,
+      "loss": 1.5256,
+      "step": 2640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.151571864772457e-05,
+      "loss": 1.4877,
+      "step": 2641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1515497412831622e-05,
+      "loss": 1.488,
+      "step": 2642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1515276177938675e-05,
+      "loss": 1.4394,
+      "step": 2643
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1515054943045728e-05,
+      "loss": 1.3724,
+      "step": 2644
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.151483370815278e-05,
+      "loss": 1.5818,
+      "step": 2645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1514612473259831e-05,
+      "loss": 1.4722,
+      "step": 2646
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1514391238366884e-05,
+      "loss": 1.5467,
+      "step": 2647
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1514170003473935e-05,
+      "loss": 1.5974,
+      "step": 2648
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1513948768580987e-05,
+      "loss": 1.368,
+      "step": 2649
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.151372753368804e-05,
+      "loss": 1.4729,
+      "step": 2650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1513506298795093e-05,
+      "loss": 1.5409,
+      "step": 2651
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1513285063902144e-05,
+      "loss": 1.4753,
+      "step": 2652
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1513063829009196e-05,
+      "loss": 1.484,
+      "step": 2653
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1512842594116249e-05,
+      "loss": 1.4818,
+      "step": 2654
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1512621359223301e-05,
+      "loss": 1.3678,
+      "step": 2655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1512400124330352e-05,
+      "loss": 1.4719,
+      "step": 2656
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1512178889437403e-05,
+      "loss": 1.508,
+      "step": 2657
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1511957654544456e-05,
+      "loss": 1.5289,
+      "step": 2658
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1511736419651508e-05,
+      "loss": 1.4807,
+      "step": 2659
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1511515184758561e-05,
+      "loss": 1.5028,
+      "step": 2660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1511293949865614e-05,
+      "loss": 1.4405,
+      "step": 2661
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1511072714972666e-05,
+      "loss": 1.4582,
+      "step": 2662
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1510851480079717e-05,
+      "loss": 1.4353,
+      "step": 2663
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.151063024518677e-05,
+      "loss": 1.4786,
+      "step": 2664
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.151040901029382e-05,
+      "loss": 1.6714,
+      "step": 2665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1510187775400873e-05,
+      "loss": 1.3117,
+      "step": 2666
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1509966540507926e-05,
+      "loss": 1.5187,
+      "step": 2667
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1509745305614977e-05,
+      "loss": 1.4547,
+      "step": 2668
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.150952407072203e-05,
+      "loss": 1.4361,
+      "step": 2669
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1509302835829082e-05,
+      "loss": 1.5709,
+      "step": 2670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1509081600936135e-05,
+      "loss": 1.4037,
+      "step": 2671
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1508860366043186e-05,
+      "loss": 1.5805,
+      "step": 2672
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1508639131150238e-05,
+      "loss": 1.5298,
+      "step": 2673
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1508417896257289e-05,
+      "loss": 1.3917,
+      "step": 2674
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1508196661364342e-05,
+      "loss": 1.687,
+      "step": 2675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1507975426471394e-05,
+      "loss": 1.4728,
+      "step": 2676
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1507754191578447e-05,
+      "loss": 1.4442,
+      "step": 2677
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.15075329566855e-05,
+      "loss": 1.5583,
+      "step": 2678
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.150731172179255e-05,
+      "loss": 1.4844,
+      "step": 2679
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1507090486899603e-05,
+      "loss": 1.4611,
+      "step": 2680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1506869252006654e-05,
+      "loss": 1.4968,
+      "step": 2681
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1506648017113707e-05,
+      "loss": 1.4775,
+      "step": 2682
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.150642678222076e-05,
+      "loss": 1.5407,
+      "step": 2683
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1506205547327812e-05,
+      "loss": 1.4351,
+      "step": 2684
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1505984312434863e-05,
+      "loss": 1.4423,
+      "step": 2685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1505763077541915e-05,
+      "loss": 1.5498,
+      "step": 2686
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1505541842648968e-05,
+      "loss": 1.4502,
+      "step": 2687
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1505320607756019e-05,
+      "loss": 1.5143,
+      "step": 2688
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1505099372863072e-05,
+      "loss": 1.4919,
+      "step": 2689
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1504878137970122e-05,
+      "loss": 1.4225,
+      "step": 2690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1504656903077175e-05,
+      "loss": 1.4136,
+      "step": 2691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1504435668184228e-05,
+      "loss": 1.5021,
+      "step": 2692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.150421443329128e-05,
+      "loss": 1.5268,
+      "step": 2693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1503993198398333e-05,
+      "loss": 1.4402,
+      "step": 2694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1503771963505384e-05,
+      "loss": 1.4094,
+      "step": 2695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1503550728612436e-05,
+      "loss": 1.5353,
+      "step": 2696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1503329493719487e-05,
+      "loss": 1.6302,
+      "step": 2697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.150310825882654e-05,
+      "loss": 1.4517,
+      "step": 2698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1502887023933593e-05,
+      "loss": 1.3346,
+      "step": 2699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1502665789040645e-05,
+      "loss": 1.6617,
+      "step": 2700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1502444554147696e-05,
+      "loss": 1.7422,
+      "step": 2701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1502223319254749e-05,
+      "loss": 1.5998,
+      "step": 2702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1502002084361801e-05,
+      "loss": 1.4379,
+      "step": 2703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1501780849468852e-05,
+      "loss": 1.5251,
+      "step": 2704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1501559614575905e-05,
+      "loss": 1.3953,
+      "step": 2705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1501338379682956e-05,
+      "loss": 1.5032,
+      "step": 2706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1501117144790008e-05,
+      "loss": 1.5594,
+      "step": 2707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1500895909897061e-05,
+      "loss": 1.5366,
+      "step": 2708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1500674675004114e-05,
+      "loss": 1.4254,
+      "step": 2709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1500453440111166e-05,
+      "loss": 1.5036,
+      "step": 2710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1500232205218219e-05,
+      "loss": 1.4865,
+      "step": 2711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.150001097032527e-05,
+      "loss": 1.4639,
+      "step": 2712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.149978973543232e-05,
+      "loss": 1.4893,
+      "step": 2713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1499568500539373e-05,
+      "loss": 1.5011,
+      "step": 2714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1499347265646426e-05,
+      "loss": 1.4174,
+      "step": 2715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1499126030753479e-05,
+      "loss": 1.3909,
+      "step": 2716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.149890479586053e-05,
+      "loss": 1.4057,
+      "step": 2717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1498683560967582e-05,
+      "loss": 1.4528,
+      "step": 2718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1498462326074635e-05,
+      "loss": 1.5267,
+      "step": 2719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1498241091181686e-05,
+      "loss": 1.7201,
+      "step": 2720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1498019856288738e-05,
+      "loss": 1.4774,
+      "step": 2721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.149779862139579e-05,
+      "loss": 1.4777,
+      "step": 2722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1497577386502842e-05,
+      "loss": 1.3759,
+      "step": 2723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1497356151609894e-05,
+      "loss": 1.4229,
+      "step": 2724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1497134916716947e-05,
+      "loss": 1.4657,
+      "step": 2725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1496913681824e-05,
+      "loss": 1.4368,
+      "step": 2726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1496692446931052e-05,
+      "loss": 1.3254,
+      "step": 2727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1496471212038103e-05,
+      "loss": 1.4921,
+      "step": 2728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1496249977145154e-05,
+      "loss": 1.3814,
+      "step": 2729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1496028742252207e-05,
+      "loss": 1.457,
+      "step": 2730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.149580750735926e-05,
+      "loss": 1.4022,
+      "step": 2731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1495586272466312e-05,
+      "loss": 1.5378,
+      "step": 2732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1495365037573364e-05,
+      "loss": 1.6506,
+      "step": 2733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1495143802680415e-05,
+      "loss": 1.4184,
+      "step": 2734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1494922567787468e-05,
+      "loss": 1.5359,
+      "step": 2735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1494701332894519e-05,
+      "loss": 1.5394,
+      "step": 2736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1494480098001571e-05,
+      "loss": 1.5462,
+      "step": 2737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1494258863108624e-05,
+      "loss": 1.5553,
+      "step": 2738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1494037628215675e-05,
+      "loss": 1.5959,
+      "step": 2739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1493816393322728e-05,
+      "loss": 1.4965,
+      "step": 2740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.149359515842978e-05,
+      "loss": 1.4725,
+      "step": 2741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1493373923536833e-05,
+      "loss": 1.403,
+      "step": 2742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1493152688643885e-05,
+      "loss": 1.4205,
+      "step": 2743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1492931453750936e-05,
+      "loss": 1.6042,
+      "step": 2744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1492710218857987e-05,
+      "loss": 1.4225,
+      "step": 2745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.149248898396504e-05,
+      "loss": 1.5273,
+      "step": 2746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1492267749072093e-05,
+      "loss": 1.3386,
+      "step": 2747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1492046514179145e-05,
+      "loss": 1.3583,
+      "step": 2748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1491825279286198e-05,
+      "loss": 1.4562,
+      "step": 2749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1491604044393249e-05,
+      "loss": 1.5864,
+      "step": 2750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1491382809500301e-05,
+      "loss": 1.462,
+      "step": 2751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1491161574607354e-05,
+      "loss": 1.5315,
+      "step": 2752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1490940339714405e-05,
+      "loss": 1.4154,
+      "step": 2753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1490719104821457e-05,
+      "loss": 1.4905,
+      "step": 2754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.149049786992851e-05,
+      "loss": 1.4446,
+      "step": 2755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1490276635035561e-05,
+      "loss": 1.5853,
+      "step": 2756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1490055400142614e-05,
+      "loss": 1.6464,
+      "step": 2757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1489834165249666e-05,
+      "loss": 1.4145,
+      "step": 2758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1489612930356719e-05,
+      "loss": 1.403,
+      "step": 2759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.148939169546377e-05,
+      "loss": 1.4422,
+      "step": 2760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.148917046057082e-05,
+      "loss": 1.68,
+      "step": 2761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1488949225677873e-05,
+      "loss": 1.8096,
+      "step": 2762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1488727990784926e-05,
+      "loss": 1.5072,
+      "step": 2763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1488506755891978e-05,
+      "loss": 1.4262,
+      "step": 2764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1488285520999031e-05,
+      "loss": 1.523,
+      "step": 2765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1488064286106084e-05,
+      "loss": 1.4642,
+      "step": 2766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1487843051213135e-05,
+      "loss": 1.5733,
+      "step": 2767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1487621816320187e-05,
+      "loss": 1.3116,
+      "step": 2768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1487400581427238e-05,
+      "loss": 1.4393,
+      "step": 2769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.148717934653429e-05,
+      "loss": 1.3572,
+      "step": 2770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1486958111641343e-05,
+      "loss": 1.5679,
+      "step": 2771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1486736876748394e-05,
+      "loss": 1.448,
+      "step": 2772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1486515641855447e-05,
+      "loss": 1.6702,
+      "step": 2773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.14862944069625e-05,
+      "loss": 1.616,
+      "step": 2774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1486073172069552e-05,
+      "loss": 1.4775,
+      "step": 2775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1485851937176603e-05,
+      "loss": 1.5453,
+      "step": 2776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1485630702283656e-05,
+      "loss": 1.4816,
+      "step": 2777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1485409467390707e-05,
+      "loss": 1.435,
+      "step": 2778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.148518823249776e-05,
+      "loss": 1.4858,
+      "step": 2779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1484966997604812e-05,
+      "loss": 1.3801,
+      "step": 2780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1484745762711864e-05,
+      "loss": 1.4223,
+      "step": 2781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1484524527818917e-05,
+      "loss": 1.2879,
+      "step": 2782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1484303292925968e-05,
+      "loss": 1.4802,
+      "step": 2783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.148408205803302e-05,
+      "loss": 1.3417,
+      "step": 2784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1483860823140071e-05,
+      "loss": 1.458,
+      "step": 2785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1483639588247124e-05,
+      "loss": 1.4599,
+      "step": 2786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1483418353354177e-05,
+      "loss": 1.3666,
+      "step": 2787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.148319711846123e-05,
+      "loss": 1.35,
+      "step": 2788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.148297588356828e-05,
+      "loss": 1.4896,
+      "step": 2789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1482754648675333e-05,
+      "loss": 1.6547,
+      "step": 2790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1482533413782385e-05,
+      "loss": 1.4331,
+      "step": 2791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1482312178889436e-05,
+      "loss": 1.3783,
+      "step": 2792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1482090943996489e-05,
+      "loss": 1.3923,
+      "step": 2793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.148186970910354e-05,
+      "loss": 1.2919,
+      "step": 2794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1481648474210592e-05,
+      "loss": 1.4373,
+      "step": 2795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1481427239317645e-05,
+      "loss": 1.4608,
+      "step": 2796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1481206004424698e-05,
+      "loss": 1.6075,
+      "step": 2797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.148098476953175e-05,
+      "loss": 1.7267,
+      "step": 2798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1480763534638801e-05,
+      "loss": 1.5452,
+      "step": 2799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1480542299745854e-05,
+      "loss": 1.5076,
+      "step": 2800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1480321064852905e-05,
+      "loss": 1.4151,
+      "step": 2801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1480099829959957e-05,
+      "loss": 1.4796,
+      "step": 2802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.147987859506701e-05,
+      "loss": 1.4567,
+      "step": 2803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1479657360174063e-05,
+      "loss": 1.3746,
+      "step": 2804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1479436125281114e-05,
+      "loss": 1.3149,
+      "step": 2805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1479214890388166e-05,
+      "loss": 1.4389,
+      "step": 2806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1478993655495219e-05,
+      "loss": 1.5264,
+      "step": 2807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.147877242060227e-05,
+      "loss": 1.4551,
+      "step": 2808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1478551185709322e-05,
+      "loss": 1.6653,
+      "step": 2809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1478329950816373e-05,
+      "loss": 1.3735,
+      "step": 2810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1478108715923426e-05,
+      "loss": 1.5133,
+      "step": 2811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1477887481030478e-05,
+      "loss": 1.455,
+      "step": 2812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1477666246137531e-05,
+      "loss": 1.6171,
+      "step": 2813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1477445011244584e-05,
+      "loss": 1.6333,
+      "step": 2814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1477223776351636e-05,
+      "loss": 1.3608,
+      "step": 2815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1477002541458687e-05,
+      "loss": 1.5766,
+      "step": 2816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1476781306565738e-05,
+      "loss": 1.481,
+      "step": 2817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.147656007167279e-05,
+      "loss": 1.3264,
+      "step": 2818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1476338836779843e-05,
+      "loss": 1.5168,
+      "step": 2819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1476117601886896e-05,
+      "loss": 1.4538,
+      "step": 2820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1475896366993947e-05,
+      "loss": 1.4518,
+      "step": 2821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1475675132101e-05,
+      "loss": 1.309,
+      "step": 2822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1475453897208052e-05,
+      "loss": 1.5347,
+      "step": 2823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1475232662315103e-05,
+      "loss": 1.4949,
+      "step": 2824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1475011427422156e-05,
+      "loss": 1.6437,
+      "step": 2825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1474790192529208e-05,
+      "loss": 1.3237,
+      "step": 2826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1474568957636259e-05,
+      "loss": 1.5099,
+      "step": 2827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1474347722743312e-05,
+      "loss": 1.3404,
+      "step": 2828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1474126487850364e-05,
+      "loss": 1.3971,
+      "step": 2829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1473905252957417e-05,
+      "loss": 1.5713,
+      "step": 2830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.147368401806447e-05,
+      "loss": 1.5259,
+      "step": 2831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.147346278317152e-05,
+      "loss": 1.4854,
+      "step": 2832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1473241548278571e-05,
+      "loss": 1.3937,
+      "step": 2833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1473020313385624e-05,
+      "loss": 1.7346,
+      "step": 2834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1472799078492677e-05,
+      "loss": 1.5143,
+      "step": 2835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.147257784359973e-05,
+      "loss": 1.4786,
+      "step": 2836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1472356608706782e-05,
+      "loss": 1.468,
+      "step": 2837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1472135373813833e-05,
+      "loss": 1.4413,
+      "step": 2838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1471914138920885e-05,
+      "loss": 1.409,
+      "step": 2839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1471692904027936e-05,
+      "loss": 1.396,
+      "step": 2840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1471471669134989e-05,
+      "loss": 1.469,
+      "step": 2841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1471250434242042e-05,
+      "loss": 1.4954,
+      "step": 2842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1471029199349092e-05,
+      "loss": 1.3868,
+      "step": 2843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1470807964456145e-05,
+      "loss": 1.3789,
+      "step": 2844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1470586729563198e-05,
+      "loss": 1.3163,
+      "step": 2845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.147036549467025e-05,
+      "loss": 1.445,
+      "step": 2846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1470144259777303e-05,
+      "loss": 1.4819,
+      "step": 2847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1469923024884354e-05,
+      "loss": 1.5485,
+      "step": 2848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1469701789991405e-05,
+      "loss": 1.4092,
+      "step": 2849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1469480555098457e-05,
+      "loss": 1.3486,
+      "step": 2850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.146925932020551e-05,
+      "loss": 1.4941,
+      "step": 2851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1469038085312563e-05,
+      "loss": 1.595,
+      "step": 2852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1468816850419615e-05,
+      "loss": 1.4938,
+      "step": 2853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1468595615526666e-05,
+      "loss": 1.6276,
+      "step": 2854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1468374380633719e-05,
+      "loss": 1.3247,
+      "step": 2855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1468153145740771e-05,
+      "loss": 1.3369,
+      "step": 2856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1467931910847822e-05,
+      "loss": 1.4702,
+      "step": 2857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1467710675954875e-05,
+      "loss": 1.5656,
+      "step": 2858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1467489441061927e-05,
+      "loss": 1.3572,
+      "step": 2859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1467268206168978e-05,
+      "loss": 1.6035,
+      "step": 2860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1467046971276031e-05,
+      "loss": 1.371,
+      "step": 2861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1466825736383084e-05,
+      "loss": 1.4767,
+      "step": 2862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1466604501490136e-05,
+      "loss": 1.5531,
+      "step": 2863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1466383266597187e-05,
+      "loss": 1.4789,
+      "step": 2864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1466162031704238e-05,
+      "loss": 1.5744,
+      "step": 2865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.146594079681129e-05,
+      "loss": 1.4844,
+      "step": 2866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1465719561918343e-05,
+      "loss": 1.4899,
+      "step": 2867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1465498327025396e-05,
+      "loss": 1.2485,
+      "step": 2868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1465277092132448e-05,
+      "loss": 1.5027,
+      "step": 2869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1465055857239501e-05,
+      "loss": 1.5383,
+      "step": 2870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1464834622346552e-05,
+      "loss": 1.4225,
+      "step": 2871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1464613387453605e-05,
+      "loss": 1.494,
+      "step": 2872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1464392152560656e-05,
+      "loss": 1.3893,
+      "step": 2873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1464170917667708e-05,
+      "loss": 1.3859,
+      "step": 2874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.146394968277476e-05,
+      "loss": 1.2939,
+      "step": 2875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1463728447881812e-05,
+      "loss": 1.5099,
+      "step": 2876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1463507212988864e-05,
+      "loss": 1.5408,
+      "step": 2877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1463285978095917e-05,
+      "loss": 1.4873,
+      "step": 2878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.146306474320297e-05,
+      "loss": 1.4709,
+      "step": 2879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.146284350831002e-05,
+      "loss": 1.4557,
+      "step": 2880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1462622273417073e-05,
+      "loss": 1.5045,
+      "step": 2881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1462401038524124e-05,
+      "loss": 1.5493,
+      "step": 2882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1462179803631177e-05,
+      "loss": 1.3702,
+      "step": 2883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.146195856873823e-05,
+      "loss": 1.4646,
+      "step": 2884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1461737333845282e-05,
+      "loss": 1.2907,
+      "step": 2885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1461516098952334e-05,
+      "loss": 1.4918,
+      "step": 2886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1461294864059385e-05,
+      "loss": 1.3411,
+      "step": 2887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1461073629166438e-05,
+      "loss": 1.3753,
+      "step": 2888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1460852394273489e-05,
+      "loss": 1.4574,
+      "step": 2889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1460631159380541e-05,
+      "loss": 1.3574,
+      "step": 2890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1460409924487594e-05,
+      "loss": 1.5632,
+      "step": 2891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1460188689594645e-05,
+      "loss": 1.3999,
+      "step": 2892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1459967454701698e-05,
+      "loss": 1.4646,
+      "step": 2893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.145974621980875e-05,
+      "loss": 1.476,
+      "step": 2894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1459524984915803e-05,
+      "loss": 1.5284,
+      "step": 2895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1459303750022854e-05,
+      "loss": 1.4193,
+      "step": 2896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1459082515129906e-05,
+      "loss": 1.457,
+      "step": 2897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1458861280236957e-05,
+      "loss": 1.5196,
+      "step": 2898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.145864004534401e-05,
+      "loss": 1.4517,
+      "step": 2899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1458418810451063e-05,
+      "loss": 1.4245,
+      "step": 2900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1458197575558115e-05,
+      "loss": 1.4503,
+      "step": 2901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1457976340665168e-05,
+      "loss": 1.6498,
+      "step": 2902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1457755105772219e-05,
+      "loss": 1.5316,
+      "step": 2903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1457533870879271e-05,
+      "loss": 1.4972,
+      "step": 2904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1457312635986322e-05,
+      "loss": 1.5614,
+      "step": 2905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1457091401093375e-05,
+      "loss": 1.5038,
+      "step": 2906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1456870166200427e-05,
+      "loss": 1.5154,
+      "step": 2907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.145664893130748e-05,
+      "loss": 1.5031,
+      "step": 2908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1456427696414531e-05,
+      "loss": 1.366,
+      "step": 2909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1456206461521584e-05,
+      "loss": 1.6101,
+      "step": 2910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1455985226628636e-05,
+      "loss": 1.3898,
+      "step": 2911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1455763991735687e-05,
+      "loss": 1.4893,
+      "step": 2912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.145554275684274e-05,
+      "loss": 1.405,
+      "step": 2913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.145532152194979e-05,
+      "loss": 1.6129,
+      "step": 2914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1455100287056843e-05,
+      "loss": 1.4499,
+      "step": 2915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1454879052163896e-05,
+      "loss": 1.4612,
+      "step": 2916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1454657817270948e-05,
+      "loss": 1.531,
+      "step": 2917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1454436582378001e-05,
+      "loss": 1.5077,
+      "step": 2918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1454215347485054e-05,
+      "loss": 1.4883,
+      "step": 2919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1453994112592105e-05,
+      "loss": 1.4606,
+      "step": 2920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1453772877699156e-05,
+      "loss": 1.5999,
+      "step": 2921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1453551642806208e-05,
+      "loss": 1.425,
+      "step": 2922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.145333040791326e-05,
+      "loss": 1.5479,
+      "step": 2923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1453109173020313e-05,
+      "loss": 1.4909,
+      "step": 2924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1452887938127364e-05,
+      "loss": 1.664,
+      "step": 2925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1452666703234417e-05,
+      "loss": 1.5954,
+      "step": 2926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.145244546834147e-05,
+      "loss": 1.3969,
+      "step": 2927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.145222423344852e-05,
+      "loss": 1.3183,
+      "step": 2928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1452002998555573e-05,
+      "loss": 1.4837,
+      "step": 2929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1451781763662626e-05,
+      "loss": 1.5134,
+      "step": 2930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1451560528769677e-05,
+      "loss": 1.4004,
+      "step": 2931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.145133929387673e-05,
+      "loss": 1.6123,
+      "step": 2932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1451118058983782e-05,
+      "loss": 1.3102,
+      "step": 2933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1450896824090834e-05,
+      "loss": 1.4736,
+      "step": 2934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1450675589197887e-05,
+      "loss": 1.3234,
+      "step": 2935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1450454354304938e-05,
+      "loss": 1.5211,
+      "step": 2936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1450233119411989e-05,
+      "loss": 1.4829,
+      "step": 2937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1450011884519041e-05,
+      "loss": 1.4145,
+      "step": 2938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1449790649626094e-05,
+      "loss": 1.3772,
+      "step": 2939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1449569414733147e-05,
+      "loss": 1.4918,
+      "step": 2940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.14493481798402e-05,
+      "loss": 1.2851,
+      "step": 2941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.144912694494725e-05,
+      "loss": 1.3954,
+      "step": 2942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1448905710054303e-05,
+      "loss": 1.5667,
+      "step": 2943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1448684475161355e-05,
+      "loss": 1.5441,
+      "step": 2944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1448463240268406e-05,
+      "loss": 1.5119,
+      "step": 2945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1448242005375459e-05,
+      "loss": 1.6344,
+      "step": 2946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.144802077048251e-05,
+      "loss": 1.4585,
+      "step": 2947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1447799535589562e-05,
+      "loss": 1.3643,
+      "step": 2948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1447578300696615e-05,
+      "loss": 1.6601,
+      "step": 2949
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1447357065803668e-05,
+      "loss": 1.4803,
+      "step": 2950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.144713583091072e-05,
+      "loss": 1.2651,
+      "step": 2951
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1446914596017771e-05,
+      "loss": 1.4562,
+      "step": 2952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1446693361124822e-05,
+      "loss": 1.3196,
+      "step": 2953
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1446472126231875e-05,
+      "loss": 1.5395,
+      "step": 2954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1446250891338927e-05,
+      "loss": 1.4299,
+      "step": 2955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.144602965644598e-05,
+      "loss": 1.4225,
+      "step": 2956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1445808421553033e-05,
+      "loss": 1.4055,
+      "step": 2957
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1445587186660084e-05,
+      "loss": 1.4618,
+      "step": 2958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1445365951767136e-05,
+      "loss": 1.4772,
+      "step": 2959
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1445144716874189e-05,
+      "loss": 1.299,
+      "step": 2960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.144492348198124e-05,
+      "loss": 1.4769,
+      "step": 2961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1444702247088292e-05,
+      "loss": 1.4011,
+      "step": 2962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1444481012195345e-05,
+      "loss": 1.4175,
+      "step": 2963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1444259777302396e-05,
+      "loss": 1.4255,
+      "step": 2964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1444038542409448e-05,
+      "loss": 1.4819,
+      "step": 2965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1443817307516501e-05,
+      "loss": 1.461,
+      "step": 2966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1443596072623554e-05,
+      "loss": 1.4299,
+      "step": 2967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1443374837730605e-05,
+      "loss": 1.49,
+      "step": 2968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1443153602837655e-05,
+      "loss": 1.5354,
+      "step": 2969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1442932367944708e-05,
+      "loss": 1.511,
+      "step": 2970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.144271113305176e-05,
+      "loss": 1.3866,
+      "step": 2971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1442489898158813e-05,
+      "loss": 1.3968,
+      "step": 2972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1442268663265866e-05,
+      "loss": 1.2289,
+      "step": 2973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1442047428372919e-05,
+      "loss": 1.402,
+      "step": 2974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.144182619347997e-05,
+      "loss": 1.5902,
+      "step": 2975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1441604958587022e-05,
+      "loss": 1.3618,
+      "step": 2976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1441383723694073e-05,
+      "loss": 1.5001,
+      "step": 2977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1441162488801126e-05,
+      "loss": 1.6287,
+      "step": 2978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1440941253908178e-05,
+      "loss": 1.3264,
+      "step": 2979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1440720019015229e-05,
+      "loss": 1.5105,
+      "step": 2980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1440498784122282e-05,
+      "loss": 1.3936,
+      "step": 2981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1440277549229334e-05,
+      "loss": 1.413,
+      "step": 2982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1440056314336387e-05,
+      "loss": 1.4583,
+      "step": 2983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1439835079443438e-05,
+      "loss": 1.4024,
+      "step": 2984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.143961384455049e-05,
+      "loss": 1.5427,
+      "step": 2985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1439392609657541e-05,
+      "loss": 1.4583,
+      "step": 2986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1439171374764594e-05,
+      "loss": 1.3875,
+      "step": 2987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1438950139871647e-05,
+      "loss": 1.5371,
+      "step": 2988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.14387289049787e-05,
+      "loss": 1.5124,
+      "step": 2989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1438507670085752e-05,
+      "loss": 1.4748,
+      "step": 2990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1438286435192803e-05,
+      "loss": 1.5052,
+      "step": 2991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1438065200299855e-05,
+      "loss": 1.4912,
+      "step": 2992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1437843965406906e-05,
+      "loss": 1.4549,
+      "step": 2993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1437622730513959e-05,
+      "loss": 1.4133,
+      "step": 2994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1437401495621012e-05,
+      "loss": 1.5465,
+      "step": 2995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1437180260728062e-05,
+      "loss": 1.4181,
+      "step": 2996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1436959025835115e-05,
+      "loss": 1.4931,
+      "step": 2997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1436737790942168e-05,
+      "loss": 1.6139,
+      "step": 2998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.143651655604922e-05,
+      "loss": 1.5263,
+      "step": 2999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1436295321156271e-05,
+      "loss": 1.5259,
+      "step": 3000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1436074086263324e-05,
+      "loss": 1.3373,
+      "step": 3001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1435852851370375e-05,
+      "loss": 1.4766,
+      "step": 3002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1435631616477427e-05,
+      "loss": 1.5401,
+      "step": 3003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.143541038158448e-05,
+      "loss": 1.4328,
+      "step": 3004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1435189146691533e-05,
+      "loss": 1.5463,
+      "step": 3005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1434967911798585e-05,
+      "loss": 1.3846,
+      "step": 3006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1434746676905636e-05,
+      "loss": 1.3502,
+      "step": 3007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1434525442012689e-05,
+      "loss": 1.4085,
+      "step": 3008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.143430420711974e-05,
+      "loss": 1.5883,
+      "step": 3009
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1434082972226792e-05,
+      "loss": 1.4417,
+      "step": 3010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1433861737333845e-05,
+      "loss": 1.4287,
+      "step": 3011
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1433640502440897e-05,
+      "loss": 1.4118,
+      "step": 3012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1433419267547948e-05,
+      "loss": 1.6185,
+      "step": 3013
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1433198032655001e-05,
+      "loss": 1.5031,
+      "step": 3014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1432976797762054e-05,
+      "loss": 1.334,
+      "step": 3015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1432755562869105e-05,
+      "loss": 1.3596,
+      "step": 3016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1432534327976157e-05,
+      "loss": 1.4885,
+      "step": 3017
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1432313093083208e-05,
+      "loss": 1.4403,
+      "step": 3018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.143209185819026e-05,
+      "loss": 1.5307,
+      "step": 3019
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1431870623297313e-05,
+      "loss": 1.5341,
+      "step": 3020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1431649388404366e-05,
+      "loss": 1.2796,
+      "step": 3021
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1431428153511418e-05,
+      "loss": 1.525,
+      "step": 3022
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1431206918618471e-05,
+      "loss": 1.3338,
+      "step": 3023
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1430985683725522e-05,
+      "loss": 1.5116,
+      "step": 3024
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1430764448832573e-05,
+      "loss": 1.4999,
+      "step": 3025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1430543213939626e-05,
+      "loss": 1.6117,
+      "step": 3026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1430321979046678e-05,
+      "loss": 1.3715,
+      "step": 3027
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.143010074415373e-05,
+      "loss": 1.602,
+      "step": 3028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1429879509260782e-05,
+      "loss": 1.3899,
+      "step": 3029
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1429658274367834e-05,
+      "loss": 1.4368,
+      "step": 3030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1429437039474887e-05,
+      "loss": 1.4495,
+      "step": 3031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.142921580458194e-05,
+      "loss": 1.341,
+      "step": 3032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.142899456968899e-05,
+      "loss": 1.3474,
+      "step": 3033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1428773334796043e-05,
+      "loss": 1.504,
+      "step": 3034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1428552099903094e-05,
+      "loss": 1.5486,
+      "step": 3035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1428330865010147e-05,
+      "loss": 1.4792,
+      "step": 3036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.14281096301172e-05,
+      "loss": 1.4498,
+      "step": 3037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1427888395224252e-05,
+      "loss": 1.3875,
+      "step": 3038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1427667160331304e-05,
+      "loss": 1.4501,
+      "step": 3039
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1427445925438355e-05,
+      "loss": 1.5318,
+      "step": 3040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1427224690545406e-05,
+      "loss": 1.4519,
+      "step": 3041
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1427003455652459e-05,
+      "loss": 1.5022,
+      "step": 3042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1426782220759511e-05,
+      "loss": 1.5859,
+      "step": 3043
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1426560985866564e-05,
+      "loss": 1.5018,
+      "step": 3044
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1426339750973617e-05,
+      "loss": 1.4837,
+      "step": 3045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1426118516080668e-05,
+      "loss": 1.5014,
+      "step": 3046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.142589728118772e-05,
+      "loss": 1.4836,
+      "step": 3047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1425676046294773e-05,
+      "loss": 1.5727,
+      "step": 3048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1425454811401824e-05,
+      "loss": 1.3935,
+      "step": 3049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1425233576508876e-05,
+      "loss": 1.5823,
+      "step": 3050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1425012341615927e-05,
+      "loss": 1.3598,
+      "step": 3051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.142479110672298e-05,
+      "loss": 1.4392,
+      "step": 3052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1424569871830033e-05,
+      "loss": 1.4337,
+      "step": 3053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1424348636937085e-05,
+      "loss": 1.5371,
+      "step": 3054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1424127402044138e-05,
+      "loss": 1.4206,
+      "step": 3055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1423906167151189e-05,
+      "loss": 1.2806,
+      "step": 3056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.142368493225824e-05,
+      "loss": 1.4966,
+      "step": 3057
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1423463697365292e-05,
+      "loss": 1.4378,
+      "step": 3058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1423242462472345e-05,
+      "loss": 1.4136,
+      "step": 3059
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1423021227579397e-05,
+      "loss": 1.5734,
+      "step": 3060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.142279999268645e-05,
+      "loss": 1.4452,
+      "step": 3061
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1422578757793501e-05,
+      "loss": 1.5113,
+      "step": 3062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1422357522900554e-05,
+      "loss": 1.4003,
+      "step": 3063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1422136288007606e-05,
+      "loss": 1.3402,
+      "step": 3064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1421915053114657e-05,
+      "loss": 1.5719,
+      "step": 3065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.142169381822171e-05,
+      "loss": 1.4112,
+      "step": 3066
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1421472583328762e-05,
+      "loss": 1.5472,
+      "step": 3067
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1421251348435813e-05,
+      "loss": 1.5986,
+      "step": 3068
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1421030113542866e-05,
+      "loss": 1.3929,
+      "step": 3069
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1420808878649918e-05,
+      "loss": 1.4222,
+      "step": 3070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1420587643756971e-05,
+      "loss": 1.5908,
+      "step": 3071
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1420366408864022e-05,
+      "loss": 1.4323,
+      "step": 3072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1420145173971075e-05,
+      "loss": 1.5015,
+      "step": 3073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1419923939078125e-05,
+      "loss": 1.6106,
+      "step": 3074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1419702704185178e-05,
+      "loss": 1.49,
+      "step": 3075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.141948146929223e-05,
+      "loss": 1.4869,
+      "step": 3076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1419260234399283e-05,
+      "loss": 1.4984,
+      "step": 3077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1419038999506336e-05,
+      "loss": 1.4112,
+      "step": 3078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1418817764613387e-05,
+      "loss": 1.4098,
+      "step": 3079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.141859652972044e-05,
+      "loss": 1.4721,
+      "step": 3080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.141837529482749e-05,
+      "loss": 1.5141,
+      "step": 3081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1418154059934543e-05,
+      "loss": 1.3328,
+      "step": 3082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1417932825041596e-05,
+      "loss": 1.4104,
+      "step": 3083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1417711590148647e-05,
+      "loss": 1.5551,
+      "step": 3084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1417490355255699e-05,
+      "loss": 1.514,
+      "step": 3085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1417269120362752e-05,
+      "loss": 1.3903,
+      "step": 3086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1417047885469804e-05,
+      "loss": 1.465,
+      "step": 3087
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1416826650576855e-05,
+      "loss": 1.5368,
+      "step": 3088
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1416605415683908e-05,
+      "loss": 1.4279,
+      "step": 3089
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1416384180790959e-05,
+      "loss": 1.5145,
+      "step": 3090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1416162945898011e-05,
+      "loss": 1.5323,
+      "step": 3091
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1415941711005064e-05,
+      "loss": 1.5839,
+      "step": 3092
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1415720476112117e-05,
+      "loss": 1.5044,
+      "step": 3093
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.141549924121917e-05,
+      "loss": 1.3315,
+      "step": 3094
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.141527800632622e-05,
+      "loss": 1.5099,
+      "step": 3095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1415056771433273e-05,
+      "loss": 1.5354,
+      "step": 3096
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1414835536540324e-05,
+      "loss": 1.3398,
+      "step": 3097
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1414614301647376e-05,
+      "loss": 1.4307,
+      "step": 3098
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1414393066754429e-05,
+      "loss": 1.6144,
+      "step": 3099
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.141417183186148e-05,
+      "loss": 1.4778,
+      "step": 3100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1413950596968532e-05,
+      "loss": 1.501,
+      "step": 3101
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1413729362075585e-05,
+      "loss": 1.2616,
+      "step": 3102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1413508127182638e-05,
+      "loss": 1.3226,
+      "step": 3103
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1413286892289689e-05,
+      "loss": 1.3803,
+      "step": 3104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1413065657396741e-05,
+      "loss": 1.411,
+      "step": 3105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1412844422503792e-05,
+      "loss": 1.3983,
+      "step": 3106
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1412623187610845e-05,
+      "loss": 1.4944,
+      "step": 3107
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1412401952717897e-05,
+      "loss": 1.4561,
+      "step": 3108
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.141218071782495e-05,
+      "loss": 1.3675,
+      "step": 3109
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1411959482932003e-05,
+      "loss": 1.4667,
+      "step": 3110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1411738248039053e-05,
+      "loss": 1.2937,
+      "step": 3111
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1411517013146106e-05,
+      "loss": 1.5129,
+      "step": 3112
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1411295778253157e-05,
+      "loss": 1.4482,
+      "step": 3113
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.141107454336021e-05,
+      "loss": 1.3972,
+      "step": 3114
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1410853308467262e-05,
+      "loss": 1.4059,
+      "step": 3115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1410632073574315e-05,
+      "loss": 1.4242,
+      "step": 3116
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1410410838681366e-05,
+      "loss": 1.503,
+      "step": 3117
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1410189603788418e-05,
+      "loss": 1.5499,
+      "step": 3118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1409968368895471e-05,
+      "loss": 1.502,
+      "step": 3119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1409747134002524e-05,
+      "loss": 1.4536,
+      "step": 3120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1409525899109575e-05,
+      "loss": 1.3457,
+      "step": 3121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1409304664216625e-05,
+      "loss": 1.3786,
+      "step": 3122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1409083429323678e-05,
+      "loss": 1.4788,
+      "step": 3123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.140886219443073e-05,
+      "loss": 1.4974,
+      "step": 3124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1408640959537783e-05,
+      "loss": 1.4958,
+      "step": 3125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1408419724644836e-05,
+      "loss": 1.405,
+      "step": 3126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1408198489751889e-05,
+      "loss": 1.4215,
+      "step": 3127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.140797725485894e-05,
+      "loss": 1.4066,
+      "step": 3128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.140775601996599e-05,
+      "loss": 1.3859,
+      "step": 3129
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1407534785073043e-05,
+      "loss": 1.6022,
+      "step": 3130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1407313550180096e-05,
+      "loss": 1.4885,
+      "step": 3131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1407092315287148e-05,
+      "loss": 1.4062,
+      "step": 3132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1406871080394199e-05,
+      "loss": 1.457,
+      "step": 3133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1406649845501252e-05,
+      "loss": 1.4251,
+      "step": 3134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1406428610608304e-05,
+      "loss": 1.4128,
+      "step": 3135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1406207375715357e-05,
+      "loss": 1.3485,
+      "step": 3136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1405986140822408e-05,
+      "loss": 1.3823,
+      "step": 3137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.140576490592946e-05,
+      "loss": 1.4176,
+      "step": 3138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1405543671036511e-05,
+      "loss": 1.3931,
+      "step": 3139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1405322436143564e-05,
+      "loss": 1.5618,
+      "step": 3140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1405101201250617e-05,
+      "loss": 1.4402,
+      "step": 3141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.140487996635767e-05,
+      "loss": 1.5007,
+      "step": 3142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1404658731464722e-05,
+      "loss": 1.3153,
+      "step": 3143
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1404437496571773e-05,
+      "loss": 1.3253,
+      "step": 3144
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1404216261678824e-05,
+      "loss": 1.3814,
+      "step": 3145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1403995026785876e-05,
+      "loss": 1.5067,
+      "step": 3146
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1403773791892929e-05,
+      "loss": 1.436,
+      "step": 3147
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1403552556999981e-05,
+      "loss": 1.6281,
+      "step": 3148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1403331322107034e-05,
+      "loss": 1.5841,
+      "step": 3149
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1403110087214085e-05,
+      "loss": 1.632,
+      "step": 3150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1402888852321138e-05,
+      "loss": 1.4094,
+      "step": 3151
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.140266761742819e-05,
+      "loss": 1.5066,
+      "step": 3152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1402446382535241e-05,
+      "loss": 1.3777,
+      "step": 3153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1402225147642294e-05,
+      "loss": 1.3557,
+      "step": 3154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1402003912749345e-05,
+      "loss": 1.4561,
+      "step": 3155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1401782677856397e-05,
+      "loss": 1.5008,
+      "step": 3156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.140156144296345e-05,
+      "loss": 1.5189,
+      "step": 3157
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1401340208070503e-05,
+      "loss": 1.4523,
+      "step": 3158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1401118973177555e-05,
+      "loss": 1.424,
+      "step": 3159
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1400897738284606e-05,
+      "loss": 1.4882,
+      "step": 3160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1400676503391659e-05,
+      "loss": 1.2698,
+      "step": 3161
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.140045526849871e-05,
+      "loss": 1.4605,
+      "step": 3162
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1400234033605762e-05,
+      "loss": 1.503,
+      "step": 3163
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1400012798712815e-05,
+      "loss": 1.3984,
+      "step": 3164
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1399791563819867e-05,
+      "loss": 1.5015,
+      "step": 3165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1399570328926918e-05,
+      "loss": 1.5037,
+      "step": 3166
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1399349094033971e-05,
+      "loss": 1.3631,
+      "step": 3167
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1399127859141024e-05,
+      "loss": 1.3753,
+      "step": 3168
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1398906624248074e-05,
+      "loss": 1.481,
+      "step": 3169
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1398685389355127e-05,
+      "loss": 1.3951,
+      "step": 3170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.139846415446218e-05,
+      "loss": 1.4516,
+      "step": 3171
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.139824291956923e-05,
+      "loss": 1.4339,
+      "step": 3172
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1398021684676283e-05,
+      "loss": 1.3586,
+      "step": 3173
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1397800449783336e-05,
+      "loss": 1.5413,
+      "step": 3174
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1397579214890388e-05,
+      "loss": 1.4251,
+      "step": 3175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.139735797999744e-05,
+      "loss": 1.3569,
+      "step": 3176
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1397136745104492e-05,
+      "loss": 1.3912,
+      "step": 3177
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1396915510211543e-05,
+      "loss": 1.5708,
+      "step": 3178
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1396694275318596e-05,
+      "loss": 1.4366,
+      "step": 3179
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1396473040425648e-05,
+      "loss": 1.4552,
+      "step": 3180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.13962518055327e-05,
+      "loss": 1.5102,
+      "step": 3181
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1396030570639752e-05,
+      "loss": 1.5195,
+      "step": 3182
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1395809335746804e-05,
+      "loss": 1.3332,
+      "step": 3183
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1395588100853857e-05,
+      "loss": 1.4869,
+      "step": 3184
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1395366865960908e-05,
+      "loss": 1.3326,
+      "step": 3185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.139514563106796e-05,
+      "loss": 1.5805,
+      "step": 3186
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1394924396175013e-05,
+      "loss": 1.2838,
+      "step": 3187
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1394703161282064e-05,
+      "loss": 1.4362,
+      "step": 3188
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1394481926389117e-05,
+      "loss": 1.4809,
+      "step": 3189
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.139426069149617e-05,
+      "loss": 1.4247,
+      "step": 3190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1394039456603222e-05,
+      "loss": 1.3654,
+      "step": 3191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1393818221710273e-05,
+      "loss": 1.5281,
+      "step": 3192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1393596986817325e-05,
+      "loss": 1.6625,
+      "step": 3193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1393375751924376e-05,
+      "loss": 1.4856,
+      "step": 3194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1393154517031429e-05,
+      "loss": 1.4304,
+      "step": 3195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1392933282138481e-05,
+      "loss": 1.4851,
+      "step": 3196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1392712047245534e-05,
+      "loss": 1.4657,
+      "step": 3197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1392490812352587e-05,
+      "loss": 1.4858,
+      "step": 3198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1392269577459638e-05,
+      "loss": 1.6405,
+      "step": 3199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.139204834256669e-05,
+      "loss": 1.5379,
+      "step": 3200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1391827107673741e-05,
+      "loss": 1.5225,
+      "step": 3201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1391605872780794e-05,
+      "loss": 1.4964,
+      "step": 3202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1391384637887846e-05,
+      "loss": 1.3674,
+      "step": 3203
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1391163402994897e-05,
+      "loss": 1.474,
+      "step": 3204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.139094216810195e-05,
+      "loss": 1.3826,
+      "step": 3205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1390720933209002e-05,
+      "loss": 1.3987,
+      "step": 3206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1390499698316055e-05,
+      "loss": 1.3905,
+      "step": 3207
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1390278463423108e-05,
+      "loss": 1.3973,
+      "step": 3208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1390057228530159e-05,
+      "loss": 1.3712,
+      "step": 3209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.138983599363721e-05,
+      "loss": 1.4325,
+      "step": 3210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1389614758744262e-05,
+      "loss": 1.34,
+      "step": 3211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1389393523851315e-05,
+      "loss": 1.4325,
+      "step": 3212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1389172288958367e-05,
+      "loss": 1.4276,
+      "step": 3213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.138895105406542e-05,
+      "loss": 1.3943,
+      "step": 3214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1388729819172471e-05,
+      "loss": 1.4301,
+      "step": 3215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1388508584279524e-05,
+      "loss": 1.5537,
+      "step": 3216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1388287349386574e-05,
+      "loss": 1.5508,
+      "step": 3217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1388066114493627e-05,
+      "loss": 1.4283,
+      "step": 3218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.138784487960068e-05,
+      "loss": 1.4483,
+      "step": 3219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1387623644707732e-05,
+      "loss": 1.5069,
+      "step": 3220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1387402409814783e-05,
+      "loss": 1.4439,
+      "step": 3221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1387181174921836e-05,
+      "loss": 1.4536,
+      "step": 3222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1386959940028888e-05,
+      "loss": 1.5238,
+      "step": 3223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1386738705135941e-05,
+      "loss": 1.4741,
+      "step": 3224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1386517470242992e-05,
+      "loss": 1.4071,
+      "step": 3225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1386296235350043e-05,
+      "loss": 1.4235,
+      "step": 3226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1386075000457095e-05,
+      "loss": 1.558,
+      "step": 3227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1385853765564148e-05,
+      "loss": 1.2586,
+      "step": 3228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.13856325306712e-05,
+      "loss": 1.5828,
+      "step": 3229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1385411295778253e-05,
+      "loss": 1.4739,
+      "step": 3230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1385190060885306e-05,
+      "loss": 1.3744,
+      "step": 3231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1384968825992357e-05,
+      "loss": 1.3805,
+      "step": 3232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1384747591099408e-05,
+      "loss": 1.3513,
+      "step": 3233
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.138452635620646e-05,
+      "loss": 1.5517,
+      "step": 3234
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1384305121313513e-05,
+      "loss": 1.3019,
+      "step": 3235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1384083886420566e-05,
+      "loss": 1.4789,
+      "step": 3236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1383862651527617e-05,
+      "loss": 1.4011,
+      "step": 3237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1383641416634669e-05,
+      "loss": 1.3677,
+      "step": 3238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1383420181741722e-05,
+      "loss": 1.318,
+      "step": 3239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1383198946848774e-05,
+      "loss": 1.3819,
+      "step": 3240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1382977711955825e-05,
+      "loss": 1.3073,
+      "step": 3241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1382756477062878e-05,
+      "loss": 1.4338,
+      "step": 3242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1382535242169929e-05,
+      "loss": 1.4641,
+      "step": 3243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1382314007276981e-05,
+      "loss": 1.4501,
+      "step": 3244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1382092772384034e-05,
+      "loss": 1.2896,
+      "step": 3245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1381871537491087e-05,
+      "loss": 1.4216,
+      "step": 3246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.138165030259814e-05,
+      "loss": 1.4255,
+      "step": 3247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.138142906770519e-05,
+      "loss": 1.4446,
+      "step": 3248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1381207832812241e-05,
+      "loss": 1.3604,
+      "step": 3249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1380986597919294e-05,
+      "loss": 1.3584,
+      "step": 3250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1380765363026346e-05,
+      "loss": 1.3763,
+      "step": 3251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1380544128133399e-05,
+      "loss": 1.4705,
+      "step": 3252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1380322893240452e-05,
+      "loss": 1.4697,
+      "step": 3253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1380101658347502e-05,
+      "loss": 1.4688,
+      "step": 3254
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1379880423454555e-05,
+      "loss": 1.3765,
+      "step": 3255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1379659188561608e-05,
+      "loss": 1.4962,
+      "step": 3256
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1379437953668659e-05,
+      "loss": 1.389,
+      "step": 3257
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1379216718775711e-05,
+      "loss": 1.4651,
+      "step": 3258
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1378995483882762e-05,
+      "loss": 1.4454,
+      "step": 3259
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1378774248989815e-05,
+      "loss": 1.476,
+      "step": 3260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1378553014096867e-05,
+      "loss": 1.4644,
+      "step": 3261
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.137833177920392e-05,
+      "loss": 1.3446,
+      "step": 3262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1378110544310973e-05,
+      "loss": 1.4252,
+      "step": 3263
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1377889309418023e-05,
+      "loss": 1.5125,
+      "step": 3264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1377668074525076e-05,
+      "loss": 1.3417,
+      "step": 3265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1377446839632127e-05,
+      "loss": 1.4518,
+      "step": 3266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.137722560473918e-05,
+      "loss": 1.4309,
+      "step": 3267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1377004369846232e-05,
+      "loss": 1.5165,
+      "step": 3268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1376783134953285e-05,
+      "loss": 1.324,
+      "step": 3269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1376561900060336e-05,
+      "loss": 1.3812,
+      "step": 3270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1376340665167388e-05,
+      "loss": 1.5074,
+      "step": 3271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1376119430274441e-05,
+      "loss": 1.3778,
+      "step": 3272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1375898195381492e-05,
+      "loss": 1.5221,
+      "step": 3273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1375676960488545e-05,
+      "loss": 1.3342,
+      "step": 3274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1375455725595597e-05,
+      "loss": 1.4243,
+      "step": 3275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1375234490702648e-05,
+      "loss": 1.5116,
+      "step": 3276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.13750132558097e-05,
+      "loss": 1.4043,
+      "step": 3277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1374792020916753e-05,
+      "loss": 1.3608,
+      "step": 3278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1374570786023806e-05,
+      "loss": 1.4305,
+      "step": 3279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1374349551130857e-05,
+      "loss": 1.328,
+      "step": 3280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.137412831623791e-05,
+      "loss": 1.4323,
+      "step": 3281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.137390708134496e-05,
+      "loss": 1.4118,
+      "step": 3282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1373685846452013e-05,
+      "loss": 1.5212,
+      "step": 3283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1373464611559066e-05,
+      "loss": 1.3864,
+      "step": 3284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1373243376666118e-05,
+      "loss": 1.4034,
+      "step": 3285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1373022141773169e-05,
+      "loss": 1.4102,
+      "step": 3286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1372800906880222e-05,
+      "loss": 1.4571,
+      "step": 3287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1372579671987274e-05,
+      "loss": 1.4746,
+      "step": 3288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1372358437094325e-05,
+      "loss": 1.3868,
+      "step": 3289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1372137202201378e-05,
+      "loss": 1.3119,
+      "step": 3290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.137191596730843e-05,
+      "loss": 1.4901,
+      "step": 3291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1371694732415481e-05,
+      "loss": 1.4983,
+      "step": 3292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1371473497522534e-05,
+      "loss": 1.4567,
+      "step": 3293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1371252262629587e-05,
+      "loss": 1.5354,
+      "step": 3294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.137103102773664e-05,
+      "loss": 1.3693,
+      "step": 3295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1370809792843692e-05,
+      "loss": 1.4397,
+      "step": 3296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1370588557950743e-05,
+      "loss": 1.5224,
+      "step": 3297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1370367323057794e-05,
+      "loss": 1.532,
+      "step": 3298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1370146088164846e-05,
+      "loss": 1.4796,
+      "step": 3299
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1369924853271899e-05,
+      "loss": 1.4662,
+      "step": 3300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1369703618378951e-05,
+      "loss": 1.5218,
+      "step": 3301
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1369482383486004e-05,
+      "loss": 1.3682,
+      "step": 3302
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1369261148593055e-05,
+      "loss": 1.5087,
+      "step": 3303
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1369039913700108e-05,
+      "loss": 1.3982,
+      "step": 3304
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1368818678807159e-05,
+      "loss": 1.4797,
+      "step": 3305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1368597443914211e-05,
+      "loss": 1.493,
+      "step": 3306
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1368376209021264e-05,
+      "loss": 1.3619,
+      "step": 3307
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1368154974128315e-05,
+      "loss": 1.408,
+      "step": 3308
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1367933739235367e-05,
+      "loss": 1.4735,
+      "step": 3309
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.136771250434242e-05,
+      "loss": 1.4449,
+      "step": 3310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1367491269449473e-05,
+      "loss": 1.4656,
+      "step": 3311
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1367270034556525e-05,
+      "loss": 1.356,
+      "step": 3312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1367048799663576e-05,
+      "loss": 1.3684,
+      "step": 3313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1366827564770627e-05,
+      "loss": 1.3925,
+      "step": 3314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.136660632987768e-05,
+      "loss": 1.3494,
+      "step": 3315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1366385094984732e-05,
+      "loss": 1.3629,
+      "step": 3316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1366163860091785e-05,
+      "loss": 1.5683,
+      "step": 3317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1365942625198837e-05,
+      "loss": 1.3981,
+      "step": 3318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1365721390305888e-05,
+      "loss": 1.5184,
+      "step": 3319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1365500155412941e-05,
+      "loss": 1.452,
+      "step": 3320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1365278920519992e-05,
+      "loss": 1.3737,
+      "step": 3321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1365057685627044e-05,
+      "loss": 1.494,
+      "step": 3322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1364836450734097e-05,
+      "loss": 1.4261,
+      "step": 3323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.136461521584115e-05,
+      "loss": 1.4762,
+      "step": 3324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.13643939809482e-05,
+      "loss": 1.4727,
+      "step": 3325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1364172746055253e-05,
+      "loss": 1.4684,
+      "step": 3326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1363951511162306e-05,
+      "loss": 1.4295,
+      "step": 3327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1363730276269358e-05,
+      "loss": 1.3757,
+      "step": 3328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.136350904137641e-05,
+      "loss": 1.6003,
+      "step": 3329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.136328780648346e-05,
+      "loss": 1.526,
+      "step": 3330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1363066571590513e-05,
+      "loss": 1.4411,
+      "step": 3331
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1362845336697566e-05,
+      "loss": 1.4488,
+      "step": 3332
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1362624101804618e-05,
+      "loss": 1.4274,
+      "step": 3333
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.136240286691167e-05,
+      "loss": 1.5113,
+      "step": 3334
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1362181632018723e-05,
+      "loss": 1.4776,
+      "step": 3335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1361960397125774e-05,
+      "loss": 1.5244,
+      "step": 3336
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1361739162232825e-05,
+      "loss": 1.4187,
+      "step": 3337
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1361517927339878e-05,
+      "loss": 1.4809,
+      "step": 3338
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.136129669244693e-05,
+      "loss": 1.4803,
+      "step": 3339
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1361075457553983e-05,
+      "loss": 1.4992,
+      "step": 3340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1360854222661034e-05,
+      "loss": 1.5751,
+      "step": 3341
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1360632987768087e-05,
+      "loss": 1.5476,
+      "step": 3342
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.136041175287514e-05,
+      "loss": 1.3399,
+      "step": 3343
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1360190517982192e-05,
+      "loss": 1.3663,
+      "step": 3344
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1359969283089243e-05,
+      "loss": 1.4539,
+      "step": 3345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1359748048196295e-05,
+      "loss": 1.542,
+      "step": 3346
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1359526813303346e-05,
+      "loss": 1.5307,
+      "step": 3347
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1359305578410399e-05,
+      "loss": 1.3547,
+      "step": 3348
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1359084343517451e-05,
+      "loss": 1.4998,
+      "step": 3349
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1358863108624504e-05,
+      "loss": 1.45,
+      "step": 3350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1358641873731557e-05,
+      "loss": 1.4739,
+      "step": 3351
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1358420638838608e-05,
+      "loss": 1.3904,
+      "step": 3352
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.135819940394566e-05,
+      "loss": 1.3989,
+      "step": 3353
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1357978169052711e-05,
+      "loss": 1.2878,
+      "step": 3354
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1357756934159764e-05,
+      "loss": 1.4967,
+      "step": 3355
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1357535699266816e-05,
+      "loss": 1.4755,
+      "step": 3356
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1357314464373869e-05,
+      "loss": 1.4146,
+      "step": 3357
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.135709322948092e-05,
+      "loss": 1.5732,
+      "step": 3358
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1356871994587972e-05,
+      "loss": 1.492,
+      "step": 3359
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1356650759695025e-05,
+      "loss": 1.3702,
+      "step": 3360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1356429524802076e-05,
+      "loss": 1.4665,
+      "step": 3361
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1356208289909129e-05,
+      "loss": 1.2996,
+      "step": 3362
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.135598705501618e-05,
+      "loss": 1.5588,
+      "step": 3363
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1355765820123232e-05,
+      "loss": 1.5304,
+      "step": 3364
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1355544585230285e-05,
+      "loss": 1.4348,
+      "step": 3365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1355323350337337e-05,
+      "loss": 1.6406,
+      "step": 3366
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.135510211544439e-05,
+      "loss": 1.4761,
+      "step": 3367
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1354880880551441e-05,
+      "loss": 1.3684,
+      "step": 3368
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1354659645658494e-05,
+      "loss": 1.3287,
+      "step": 3369
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1354438410765544e-05,
+      "loss": 1.73,
+      "step": 3370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1354217175872597e-05,
+      "loss": 1.4577,
+      "step": 3371
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.135399594097965e-05,
+      "loss": 1.366,
+      "step": 3372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1353774706086702e-05,
+      "loss": 1.5223,
+      "step": 3373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1353553471193753e-05,
+      "loss": 1.5684,
+      "step": 3374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1353332236300806e-05,
+      "loss": 1.4115,
+      "step": 3375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1353111001407858e-05,
+      "loss": 1.5818,
+      "step": 3376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.135288976651491e-05,
+      "loss": 1.5855,
+      "step": 3377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1352668531621962e-05,
+      "loss": 1.5419,
+      "step": 3378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1352447296729013e-05,
+      "loss": 1.4312,
+      "step": 3379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1352226061836065e-05,
+      "loss": 1.4759,
+      "step": 3380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1352004826943118e-05,
+      "loss": 1.4329,
+      "step": 3381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.135178359205017e-05,
+      "loss": 1.4724,
+      "step": 3382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1351562357157223e-05,
+      "loss": 1.5792,
+      "step": 3383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1351341122264276e-05,
+      "loss": 1.5378,
+      "step": 3384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1351119887371327e-05,
+      "loss": 1.4836,
+      "step": 3385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1350898652478378e-05,
+      "loss": 1.4547,
+      "step": 3386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.135067741758543e-05,
+      "loss": 1.5137,
+      "step": 3387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1350456182692483e-05,
+      "loss": 1.5047,
+      "step": 3388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1350234947799536e-05,
+      "loss": 1.5091,
+      "step": 3389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1350013712906587e-05,
+      "loss": 1.3667,
+      "step": 3390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1349792478013639e-05,
+      "loss": 1.4269,
+      "step": 3391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1349571243120692e-05,
+      "loss": 1.4812,
+      "step": 3392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1349350008227743e-05,
+      "loss": 1.3488,
+      "step": 3393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1349128773334795e-05,
+      "loss": 1.4487,
+      "step": 3394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1348907538441848e-05,
+      "loss": 1.3993,
+      "step": 3395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1348686303548899e-05,
+      "loss": 1.2746,
+      "step": 3396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1348465068655951e-05,
+      "loss": 1.5708,
+      "step": 3397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1348243833763004e-05,
+      "loss": 1.4269,
+      "step": 3398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1348022598870057e-05,
+      "loss": 1.6256,
+      "step": 3399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.134780136397711e-05,
+      "loss": 1.5002,
+      "step": 3400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.134758012908416e-05,
+      "loss": 1.3574,
+      "step": 3401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1347358894191211e-05,
+      "loss": 1.3216,
+      "step": 3402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1347137659298264e-05,
+      "loss": 1.3352,
+      "step": 3403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1346916424405316e-05,
+      "loss": 1.2751,
+      "step": 3404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1346695189512369e-05,
+      "loss": 1.4407,
+      "step": 3405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1346473954619422e-05,
+      "loss": 1.4528,
+      "step": 3406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1346252719726472e-05,
+      "loss": 1.4363,
+      "step": 3407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1346031484833525e-05,
+      "loss": 1.5897,
+      "step": 3408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1345810249940576e-05,
+      "loss": 1.4792,
+      "step": 3409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1345589015047629e-05,
+      "loss": 1.4092,
+      "step": 3410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1345367780154681e-05,
+      "loss": 1.5759,
+      "step": 3411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1345146545261732e-05,
+      "loss": 1.5935,
+      "step": 3412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1344925310368785e-05,
+      "loss": 1.4824,
+      "step": 3413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1344704075475837e-05,
+      "loss": 1.405,
+      "step": 3414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.134448284058289e-05,
+      "loss": 1.685,
+      "step": 3415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1344261605689943e-05,
+      "loss": 1.4109,
+      "step": 3416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1344040370796993e-05,
+      "loss": 1.27,
+      "step": 3417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1343819135904044e-05,
+      "loss": 1.4785,
+      "step": 3418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1343597901011097e-05,
+      "loss": 1.3916,
+      "step": 3419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.134337666611815e-05,
+      "loss": 1.4813,
+      "step": 3420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1343155431225202e-05,
+      "loss": 1.3169,
+      "step": 3421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1342934196332255e-05,
+      "loss": 1.6006,
+      "step": 3422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1342712961439306e-05,
+      "loss": 1.4155,
+      "step": 3423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1342491726546358e-05,
+      "loss": 1.5425,
+      "step": 3424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.134227049165341e-05,
+      "loss": 1.4385,
+      "step": 3425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1342049256760462e-05,
+      "loss": 1.4033,
+      "step": 3426
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1341828021867515e-05,
+      "loss": 1.353,
+      "step": 3427
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1341606786974567e-05,
+      "loss": 1.5336,
+      "step": 3428
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1341385552081618e-05,
+      "loss": 1.4015,
+      "step": 3429
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.134116431718867e-05,
+      "loss": 1.4132,
+      "step": 3430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1340943082295723e-05,
+      "loss": 1.5887,
+      "step": 3431
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1340721847402776e-05,
+      "loss": 1.4955,
+      "step": 3432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1340500612509827e-05,
+      "loss": 1.4591,
+      "step": 3433
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1340279377616878e-05,
+      "loss": 1.4416,
+      "step": 3434
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.134005814272393e-05,
+      "loss": 1.3752,
+      "step": 3435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1339836907830983e-05,
+      "loss": 1.4789,
+      "step": 3436
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1339615672938036e-05,
+      "loss": 1.3907,
+      "step": 3437
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1339394438045088e-05,
+      "loss": 1.4987,
+      "step": 3438
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.133917320315214e-05,
+      "loss": 1.4683,
+      "step": 3439
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1338951968259192e-05,
+      "loss": 1.4363,
+      "step": 3440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1338730733366244e-05,
+      "loss": 1.4453,
+      "step": 3441
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1338509498473295e-05,
+      "loss": 1.4112,
+      "step": 3442
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1338288263580348e-05,
+      "loss": 1.4141,
+      "step": 3443
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.13380670286874e-05,
+      "loss": 1.4534,
+      "step": 3444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1337845793794451e-05,
+      "loss": 1.3285,
+      "step": 3445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1337624558901504e-05,
+      "loss": 1.5553,
+      "step": 3446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1337403324008557e-05,
+      "loss": 1.5121,
+      "step": 3447
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.133718208911561e-05,
+      "loss": 1.3527,
+      "step": 3448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.133696085422266e-05,
+      "loss": 1.5306,
+      "step": 3449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1336739619329713e-05,
+      "loss": 1.5303,
+      "step": 3450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1336518384436764e-05,
+      "loss": 1.5486,
+      "step": 3451
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1336297149543816e-05,
+      "loss": 1.62,
+      "step": 3452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1336075914650869e-05,
+      "loss": 1.4668,
+      "step": 3453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1335854679757921e-05,
+      "loss": 1.4822,
+      "step": 3454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1335633444864974e-05,
+      "loss": 1.4362,
+      "step": 3455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1335412209972025e-05,
+      "loss": 1.5212,
+      "step": 3456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1335190975079078e-05,
+      "loss": 1.4576,
+      "step": 3457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1334969740186129e-05,
+      "loss": 1.5062,
+      "step": 3458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1334748505293181e-05,
+      "loss": 1.452,
+      "step": 3459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1334527270400234e-05,
+      "loss": 1.644,
+      "step": 3460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1334306035507286e-05,
+      "loss": 1.4444,
+      "step": 3461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1334084800614337e-05,
+      "loss": 1.5346,
+      "step": 3462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.133386356572139e-05,
+      "loss": 1.3747,
+      "step": 3463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1333642330828443e-05,
+      "loss": 1.3857,
+      "step": 3464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1333421095935493e-05,
+      "loss": 1.643,
+      "step": 3465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1333199861042546e-05,
+      "loss": 1.3414,
+      "step": 3466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1332978626149597e-05,
+      "loss": 1.3954,
+      "step": 3467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.133275739125665e-05,
+      "loss": 1.262,
+      "step": 3468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1332536156363702e-05,
+      "loss": 1.3576,
+      "step": 3469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1332314921470755e-05,
+      "loss": 1.3157,
+      "step": 3470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1332093686577807e-05,
+      "loss": 1.4507,
+      "step": 3471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.133187245168486e-05,
+      "loss": 1.6058,
+      "step": 3472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1331651216791911e-05,
+      "loss": 1.4284,
+      "step": 3473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1331429981898962e-05,
+      "loss": 1.5031,
+      "step": 3474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1331208747006014e-05,
+      "loss": 1.4536,
+      "step": 3475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1330987512113067e-05,
+      "loss": 1.4311,
+      "step": 3476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.133076627722012e-05,
+      "loss": 1.5944,
+      "step": 3477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.133054504232717e-05,
+      "loss": 1.5286,
+      "step": 3478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1330323807434223e-05,
+      "loss": 1.4092,
+      "step": 3479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1330102572541276e-05,
+      "loss": 1.4047,
+      "step": 3480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1329881337648327e-05,
+      "loss": 1.4558,
+      "step": 3481
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.132966010275538e-05,
+      "loss": 1.4098,
+      "step": 3482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.132943886786243e-05,
+      "loss": 1.4833,
+      "step": 3483
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1329217632969483e-05,
+      "loss": 1.3122,
+      "step": 3484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1328996398076535e-05,
+      "loss": 1.5283,
+      "step": 3485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1328775163183588e-05,
+      "loss": 1.4204,
+      "step": 3486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.132855392829064e-05,
+      "loss": 1.3396,
+      "step": 3487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1328332693397693e-05,
+      "loss": 1.445,
+      "step": 3488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1328111458504744e-05,
+      "loss": 1.552,
+      "step": 3489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1327890223611795e-05,
+      "loss": 1.4065,
+      "step": 3490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1327668988718848e-05,
+      "loss": 1.5507,
+      "step": 3491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.13274477538259e-05,
+      "loss": 1.3945,
+      "step": 3492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1327226518932953e-05,
+      "loss": 1.4649,
+      "step": 3493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1327005284040004e-05,
+      "loss": 1.326,
+      "step": 3494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1326784049147057e-05,
+      "loss": 1.4624,
+      "step": 3495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1326562814254109e-05,
+      "loss": 1.3287,
+      "step": 3496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.132634157936116e-05,
+      "loss": 1.4756,
+      "step": 3497
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1326120344468213e-05,
+      "loss": 1.4654,
+      "step": 3498
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1325899109575265e-05,
+      "loss": 1.3703,
+      "step": 3499
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1325677874682316e-05,
+      "loss": 1.4486,
+      "step": 3500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1325456639789369e-05,
+      "loss": 1.4192,
+      "step": 3501
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1325235404896421e-05,
+      "loss": 1.4455,
+      "step": 3502
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1325014170003474e-05,
+      "loss": 1.4235,
+      "step": 3503
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1324792935110527e-05,
+      "loss": 1.3954,
+      "step": 3504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1324571700217578e-05,
+      "loss": 1.3548,
+      "step": 3505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1324350465324628e-05,
+      "loss": 1.291,
+      "step": 3506
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1324129230431681e-05,
+      "loss": 1.5363,
+      "step": 3507
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1323907995538734e-05,
+      "loss": 1.5345,
+      "step": 3508
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1323686760645786e-05,
+      "loss": 1.4047,
+      "step": 3509
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1323465525752839e-05,
+      "loss": 1.5787,
+      "step": 3510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.132324429085989e-05,
+      "loss": 1.4649,
+      "step": 3511
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1323023055966942e-05,
+      "loss": 1.2929,
+      "step": 3512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1322801821073993e-05,
+      "loss": 1.5116,
+      "step": 3513
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1322580586181046e-05,
+      "loss": 1.3906,
+      "step": 3514
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1322359351288099e-05,
+      "loss": 1.5106,
+      "step": 3515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.132213811639515e-05,
+      "loss": 1.378,
+      "step": 3516
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1321916881502202e-05,
+      "loss": 1.4483,
+      "step": 3517
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1321695646609255e-05,
+      "loss": 1.4869,
+      "step": 3518
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1321474411716307e-05,
+      "loss": 1.4496,
+      "step": 3519
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.132125317682336e-05,
+      "loss": 1.4394,
+      "step": 3520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1321031941930411e-05,
+      "loss": 1.4552,
+      "step": 3521
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1320810707037462e-05,
+      "loss": 1.5863,
+      "step": 3522
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1320589472144514e-05,
+      "loss": 1.4257,
+      "step": 3523
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1320368237251567e-05,
+      "loss": 1.3618,
+      "step": 3524
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.132014700235862e-05,
+      "loss": 1.4981,
+      "step": 3525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1319925767465672e-05,
+      "loss": 1.5715,
+      "step": 3526
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1319704532572723e-05,
+      "loss": 1.394,
+      "step": 3527
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1319483297679776e-05,
+      "loss": 1.5241,
+      "step": 3528
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1319262062786828e-05,
+      "loss": 1.2864,
+      "step": 3529
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.131904082789388e-05,
+      "loss": 1.4469,
+      "step": 3530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1318819593000932e-05,
+      "loss": 1.3994,
+      "step": 3531
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1318598358107985e-05,
+      "loss": 1.4703,
+      "step": 3532
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1318377123215035e-05,
+      "loss": 1.325,
+      "step": 3533
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1318155888322088e-05,
+      "loss": 1.4003,
+      "step": 3534
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.131793465342914e-05,
+      "loss": 1.4016,
+      "step": 3535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1317713418536193e-05,
+      "loss": 1.3627,
+      "step": 3536
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1317492183643244e-05,
+      "loss": 1.4366,
+      "step": 3537
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1317270948750295e-05,
+      "loss": 1.3555,
+      "step": 3538
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1317049713857348e-05,
+      "loss": 1.5105,
+      "step": 3539
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.13168284789644e-05,
+      "loss": 1.393,
+      "step": 3540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1316607244071453e-05,
+      "loss": 1.4348,
+      "step": 3541
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1316386009178506e-05,
+      "loss": 1.4864,
+      "step": 3542
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1316164774285558e-05,
+      "loss": 1.5786,
+      "step": 3543
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1315943539392609e-05,
+      "loss": 1.3131,
+      "step": 3544
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1315722304499662e-05,
+      "loss": 1.4999,
+      "step": 3545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1315501069606713e-05,
+      "loss": 1.3781,
+      "step": 3546
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1315279834713765e-05,
+      "loss": 1.4897,
+      "step": 3547
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1315058599820818e-05,
+      "loss": 1.3809,
+      "step": 3548
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1314837364927869e-05,
+      "loss": 1.388,
+      "step": 3549
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1314616130034921e-05,
+      "loss": 1.3584,
+      "step": 3550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1314394895141974e-05,
+      "loss": 1.3025,
+      "step": 3551
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1314173660249027e-05,
+      "loss": 1.4469,
+      "step": 3552
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1313952425356078e-05,
+      "loss": 1.5257,
+      "step": 3553
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.131373119046313e-05,
+      "loss": 1.3524,
+      "step": 3554
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1313509955570181e-05,
+      "loss": 1.4266,
+      "step": 3555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1313288720677234e-05,
+      "loss": 1.3874,
+      "step": 3556
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1313067485784286e-05,
+      "loss": 1.4142,
+      "step": 3557
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1312846250891339e-05,
+      "loss": 1.3489,
+      "step": 3558
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1312625015998392e-05,
+      "loss": 1.3871,
+      "step": 3559
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1312403781105442e-05,
+      "loss": 1.446,
+      "step": 3560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1312182546212495e-05,
+      "loss": 1.4283,
+      "step": 3561
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1311961311319546e-05,
+      "loss": 1.4187,
+      "step": 3562
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1311740076426599e-05,
+      "loss": 1.7176,
+      "step": 3563
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1311518841533651e-05,
+      "loss": 1.5486,
+      "step": 3564
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1311297606640704e-05,
+      "loss": 1.3451,
+      "step": 3565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1311076371747755e-05,
+      "loss": 1.4864,
+      "step": 3566
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1310855136854807e-05,
+      "loss": 1.4463,
+      "step": 3567
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.131063390196186e-05,
+      "loss": 1.51,
+      "step": 3568
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.131041266706891e-05,
+      "loss": 1.4459,
+      "step": 3569
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1310191432175963e-05,
+      "loss": 1.4882,
+      "step": 3570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1309970197283014e-05,
+      "loss": 1.5149,
+      "step": 3571
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1309748962390067e-05,
+      "loss": 1.4021,
+      "step": 3572
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.130952772749712e-05,
+      "loss": 1.4426,
+      "step": 3573
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1309306492604172e-05,
+      "loss": 1.4424,
+      "step": 3574
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1309085257711225e-05,
+      "loss": 1.3766,
+      "step": 3575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1308864022818276e-05,
+      "loss": 1.4845,
+      "step": 3576
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1308642787925328e-05,
+      "loss": 1.5334,
+      "step": 3577
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.130842155303238e-05,
+      "loss": 1.4327,
+      "step": 3578
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1308200318139432e-05,
+      "loss": 1.519,
+      "step": 3579
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1307979083246484e-05,
+      "loss": 1.262,
+      "step": 3580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1307757848353537e-05,
+      "loss": 1.4198,
+      "step": 3581
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1307536613460588e-05,
+      "loss": 1.3207,
+      "step": 3582
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.130731537856764e-05,
+      "loss": 1.4983,
+      "step": 3583
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1307094143674693e-05,
+      "loss": 1.5216,
+      "step": 3584
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1306872908781744e-05,
+      "loss": 1.6042,
+      "step": 3585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1306651673888797e-05,
+      "loss": 1.4084,
+      "step": 3586
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1306430438995848e-05,
+      "loss": 1.3598,
+      "step": 3587
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.13062092041029e-05,
+      "loss": 1.4525,
+      "step": 3588
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1305987969209953e-05,
+      "loss": 1.4722,
+      "step": 3589
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1305766734317006e-05,
+      "loss": 1.4575,
+      "step": 3590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1305545499424058e-05,
+      "loss": 1.4387,
+      "step": 3591
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.130532426453111e-05,
+      "loss": 1.3804,
+      "step": 3592
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1305103029638162e-05,
+      "loss": 1.4568,
+      "step": 3593
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1304881794745213e-05,
+      "loss": 1.4685,
+      "step": 3594
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1304660559852265e-05,
+      "loss": 1.5634,
+      "step": 3595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1304439324959318e-05,
+      "loss": 1.4275,
+      "step": 3596
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.130421809006637e-05,
+      "loss": 1.3734,
+      "step": 3597
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1303996855173421e-05,
+      "loss": 1.4029,
+      "step": 3598
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1303775620280474e-05,
+      "loss": 1.6129,
+      "step": 3599
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1303554385387527e-05,
+      "loss": 1.3316,
+      "step": 3600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1303333150494577e-05,
+      "loss": 1.4536,
+      "step": 3601
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.130311191560163e-05,
+      "loss": 1.4037,
+      "step": 3602
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1302890680708683e-05,
+      "loss": 1.4094,
+      "step": 3603
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1302669445815734e-05,
+      "loss": 1.4508,
+      "step": 3604
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1302448210922786e-05,
+      "loss": 1.4264,
+      "step": 3605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1302226976029839e-05,
+      "loss": 1.4173,
+      "step": 3606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1302005741136891e-05,
+      "loss": 1.3415,
+      "step": 3607
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1301784506243944e-05,
+      "loss": 1.5668,
+      "step": 3608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1301563271350995e-05,
+      "loss": 1.4987,
+      "step": 3609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1301342036458046e-05,
+      "loss": 1.4385,
+      "step": 3610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1301120801565099e-05,
+      "loss": 1.4725,
+      "step": 3611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1300899566672151e-05,
+      "loss": 1.4157,
+      "step": 3612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1300678331779204e-05,
+      "loss": 1.3858,
+      "step": 3613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1300457096886256e-05,
+      "loss": 1.595,
+      "step": 3614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1300235861993307e-05,
+      "loss": 1.4272,
+      "step": 3615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.130001462710036e-05,
+      "loss": 1.4444,
+      "step": 3616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1299793392207412e-05,
+      "loss": 1.3121,
+      "step": 3617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1299572157314463e-05,
+      "loss": 1.4695,
+      "step": 3618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1299350922421516e-05,
+      "loss": 1.3935,
+      "step": 3619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1299129687528567e-05,
+      "loss": 1.2979,
+      "step": 3620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.129890845263562e-05,
+      "loss": 1.3378,
+      "step": 3621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1298687217742672e-05,
+      "loss": 1.2975,
+      "step": 3622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1298465982849725e-05,
+      "loss": 1.397,
+      "step": 3623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1298244747956777e-05,
+      "loss": 1.4769,
+      "step": 3624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1298023513063828e-05,
+      "loss": 1.3762,
+      "step": 3625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.129780227817088e-05,
+      "loss": 1.5467,
+      "step": 3626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1297581043277932e-05,
+      "loss": 1.4393,
+      "step": 3627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1297359808384984e-05,
+      "loss": 1.3172,
+      "step": 3628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1297138573492037e-05,
+      "loss": 1.4373,
+      "step": 3629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.129691733859909e-05,
+      "loss": 1.6003,
+      "step": 3630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.129669610370614e-05,
+      "loss": 1.4259,
+      "step": 3631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1296474868813193e-05,
+      "loss": 1.4882,
+      "step": 3632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1296253633920246e-05,
+      "loss": 1.4754,
+      "step": 3633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1296032399027297e-05,
+      "loss": 1.3445,
+      "step": 3634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.129581116413435e-05,
+      "loss": 1.4501,
+      "step": 3635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1295589929241402e-05,
+      "loss": 1.4601,
+      "step": 3636
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1295368694348453e-05,
+      "loss": 1.4751,
+      "step": 3637
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1295147459455505e-05,
+      "loss": 1.408,
+      "step": 3638
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1294926224562558e-05,
+      "loss": 1.6779,
+      "step": 3639
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.129470498966961e-05,
+      "loss": 1.3335,
+      "step": 3640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1294483754776662e-05,
+      "loss": 1.4242,
+      "step": 3641
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1294262519883713e-05,
+      "loss": 1.4314,
+      "step": 3642
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1294041284990765e-05,
+      "loss": 1.4246,
+      "step": 3643
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1293820050097818e-05,
+      "loss": 1.2819,
+      "step": 3644
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.129359881520487e-05,
+      "loss": 1.4586,
+      "step": 3645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1293377580311923e-05,
+      "loss": 1.4388,
+      "step": 3646
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1293156345418976e-05,
+      "loss": 1.4194,
+      "step": 3647
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1292935110526027e-05,
+      "loss": 1.4677,
+      "step": 3648
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1292713875633079e-05,
+      "loss": 1.4404,
+      "step": 3649
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.129249264074013e-05,
+      "loss": 1.4028,
+      "step": 3650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1292271405847183e-05,
+      "loss": 1.5221,
+      "step": 3651
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1292050170954235e-05,
+      "loss": 1.2676,
+      "step": 3652
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1291828936061286e-05,
+      "loss": 1.3825,
+      "step": 3653
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1291607701168339e-05,
+      "loss": 1.4623,
+      "step": 3654
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1291386466275391e-05,
+      "loss": 1.3442,
+      "step": 3655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1291165231382444e-05,
+      "loss": 1.3035,
+      "step": 3656
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1290943996489495e-05,
+      "loss": 1.4393,
+      "step": 3657
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1290722761596548e-05,
+      "loss": 1.4333,
+      "step": 3658
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1290501526703598e-05,
+      "loss": 1.4018,
+      "step": 3659
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1290280291810651e-05,
+      "loss": 1.4564,
+      "step": 3660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1290059056917704e-05,
+      "loss": 1.469,
+      "step": 3661
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1289837822024756e-05,
+      "loss": 1.3149,
+      "step": 3662
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1289616587131809e-05,
+      "loss": 1.408,
+      "step": 3663
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.128939535223886e-05,
+      "loss": 1.2517,
+      "step": 3664
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1289174117345912e-05,
+      "loss": 1.3382,
+      "step": 3665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1288952882452963e-05,
+      "loss": 1.4812,
+      "step": 3666
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1288731647560016e-05,
+      "loss": 1.5088,
+      "step": 3667
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1288510412667069e-05,
+      "loss": 1.3741,
+      "step": 3668
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1288289177774121e-05,
+      "loss": 1.5943,
+      "step": 3669
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1288067942881172e-05,
+      "loss": 1.4174,
+      "step": 3670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1287846707988225e-05,
+      "loss": 1.4637,
+      "step": 3671
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1287625473095277e-05,
+      "loss": 1.501,
+      "step": 3672
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1287404238202328e-05,
+      "loss": 1.3925,
+      "step": 3673
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1287183003309381e-05,
+      "loss": 1.3091,
+      "step": 3674
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1286961768416432e-05,
+      "loss": 1.5104,
+      "step": 3675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1286740533523484e-05,
+      "loss": 1.4993,
+      "step": 3676
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1286519298630537e-05,
+      "loss": 1.493,
+      "step": 3677
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.128629806373759e-05,
+      "loss": 1.5199,
+      "step": 3678
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1286076828844642e-05,
+      "loss": 1.5881,
+      "step": 3679
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1285855593951693e-05,
+      "loss": 1.4174,
+      "step": 3680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1285634359058746e-05,
+      "loss": 1.442,
+      "step": 3681
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1285413124165797e-05,
+      "loss": 1.3531,
+      "step": 3682
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.128519188927285e-05,
+      "loss": 1.5085,
+      "step": 3683
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1284970654379902e-05,
+      "loss": 1.4695,
+      "step": 3684
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1284749419486955e-05,
+      "loss": 1.4492,
+      "step": 3685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1284528184594005e-05,
+      "loss": 1.53,
+      "step": 3686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1284306949701058e-05,
+      "loss": 1.5088,
+      "step": 3687
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.128408571480811e-05,
+      "loss": 1.3245,
+      "step": 3688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1283864479915162e-05,
+      "loss": 1.3434,
+      "step": 3689
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1283643245022214e-05,
+      "loss": 1.3235,
+      "step": 3690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1283422010129265e-05,
+      "loss": 1.3841,
+      "step": 3691
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1283200775236318e-05,
+      "loss": 1.3218,
+      "step": 3692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.128297954034337e-05,
+      "loss": 1.4032,
+      "step": 3693
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1282758305450423e-05,
+      "loss": 1.4518,
+      "step": 3694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1282537070557476e-05,
+      "loss": 1.4707,
+      "step": 3695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1282315835664528e-05,
+      "loss": 1.243,
+      "step": 3696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1282094600771579e-05,
+      "loss": 1.4403,
+      "step": 3697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.128187336587863e-05,
+      "loss": 1.4426,
+      "step": 3698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1281652130985683e-05,
+      "loss": 1.4335,
+      "step": 3699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1281430896092735e-05,
+      "loss": 1.4506,
+      "step": 3700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1281209661199788e-05,
+      "loss": 1.3098,
+      "step": 3701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1280988426306839e-05,
+      "loss": 1.4263,
+      "step": 3702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1280767191413891e-05,
+      "loss": 1.408,
+      "step": 3703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1280545956520944e-05,
+      "loss": 1.4857,
+      "step": 3704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1280324721627997e-05,
+      "loss": 1.3758,
+      "step": 3705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1280103486735048e-05,
+      "loss": 1.4548,
+      "step": 3706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.12798822518421e-05,
+      "loss": 1.4595,
+      "step": 3707
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1279661016949151e-05,
+      "loss": 1.3729,
+      "step": 3708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1279439782056204e-05,
+      "loss": 1.5064,
+      "step": 3709
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1279218547163256e-05,
+      "loss": 1.3208,
+      "step": 3710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1278997312270309e-05,
+      "loss": 1.2729,
+      "step": 3711
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1278776077377361e-05,
+      "loss": 1.347,
+      "step": 3712
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1278554842484412e-05,
+      "loss": 1.4466,
+      "step": 3713
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1278333607591463e-05,
+      "loss": 1.432,
+      "step": 3714
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1278112372698516e-05,
+      "loss": 1.446,
+      "step": 3715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1277891137805569e-05,
+      "loss": 1.3847,
+      "step": 3716
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1277669902912621e-05,
+      "loss": 1.5739,
+      "step": 3717
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1277448668019674e-05,
+      "loss": 1.329,
+      "step": 3718
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1277227433126725e-05,
+      "loss": 1.6656,
+      "step": 3719
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1277006198233777e-05,
+      "loss": 1.3306,
+      "step": 3720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.127678496334083e-05,
+      "loss": 1.4894,
+      "step": 3721
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.127656372844788e-05,
+      "loss": 1.4821,
+      "step": 3722
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1276342493554933e-05,
+      "loss": 1.3587,
+      "step": 3723
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1276121258661984e-05,
+      "loss": 1.3655,
+      "step": 3724
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1275900023769037e-05,
+      "loss": 1.3482,
+      "step": 3725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.127567878887609e-05,
+      "loss": 1.6238,
+      "step": 3726
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1275457553983142e-05,
+      "loss": 1.3483,
+      "step": 3727
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1275236319090195e-05,
+      "loss": 1.441,
+      "step": 3728
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1275015084197246e-05,
+      "loss": 1.5853,
+      "step": 3729
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1274793849304297e-05,
+      "loss": 1.6157,
+      "step": 3730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.127457261441135e-05,
+      "loss": 1.3749,
+      "step": 3731
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1274351379518402e-05,
+      "loss": 1.3564,
+      "step": 3732
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1274130144625454e-05,
+      "loss": 1.5148,
+      "step": 3733
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1273908909732507e-05,
+      "loss": 1.4002,
+      "step": 3734
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1273687674839558e-05,
+      "loss": 1.4649,
+      "step": 3735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.127346643994661e-05,
+      "loss": 1.4087,
+      "step": 3736
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1273245205053663e-05,
+      "loss": 1.5988,
+      "step": 3737
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1273023970160714e-05,
+      "loss": 1.4908,
+      "step": 3738
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1272802735267767e-05,
+      "loss": 1.3739,
+      "step": 3739
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.127258150037482e-05,
+      "loss": 1.4184,
+      "step": 3740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.127236026548187e-05,
+      "loss": 1.5014,
+      "step": 3741
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1272139030588923e-05,
+      "loss": 1.4918,
+      "step": 3742
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1271917795695976e-05,
+      "loss": 1.4768,
+      "step": 3743
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1271696560803028e-05,
+      "loss": 1.2198,
+      "step": 3744
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1271475325910079e-05,
+      "loss": 1.3773,
+      "step": 3745
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.127125409101713e-05,
+      "loss": 1.4888,
+      "step": 3746
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1271032856124183e-05,
+      "loss": 1.4521,
+      "step": 3747
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1270811621231235e-05,
+      "loss": 1.569,
+      "step": 3748
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1270590386338288e-05,
+      "loss": 1.3618,
+      "step": 3749
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.127036915144534e-05,
+      "loss": 1.3481,
+      "step": 3750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1270147916552393e-05,
+      "loss": 1.5634,
+      "step": 3751
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1269926681659444e-05,
+      "loss": 1.4632,
+      "step": 3752
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1269705446766497e-05,
+      "loss": 1.3778,
+      "step": 3753
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1269484211873547e-05,
+      "loss": 1.352,
+      "step": 3754
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.12692629769806e-05,
+      "loss": 1.5426,
+      "step": 3755
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1269041742087653e-05,
+      "loss": 1.3759,
+      "step": 3756
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1268820507194704e-05,
+      "loss": 1.5214,
+      "step": 3757
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1268599272301756e-05,
+      "loss": 1.4481,
+      "step": 3758
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1268378037408809e-05,
+      "loss": 1.4599,
+      "step": 3759
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1268156802515861e-05,
+      "loss": 1.3894,
+      "step": 3760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1267935567622912e-05,
+      "loss": 1.353,
+      "step": 3761
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1267714332729965e-05,
+      "loss": 1.4133,
+      "step": 3762
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1267493097837016e-05,
+      "loss": 1.4452,
+      "step": 3763
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1267271862944069e-05,
+      "loss": 1.3903,
+      "step": 3764
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1267050628051121e-05,
+      "loss": 1.4007,
+      "step": 3765
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1266829393158174e-05,
+      "loss": 1.3003,
+      "step": 3766
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1266608158265226e-05,
+      "loss": 1.4948,
+      "step": 3767
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1266386923372277e-05,
+      "loss": 1.4064,
+      "step": 3768
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.126616568847933e-05,
+      "loss": 1.3397,
+      "step": 3769
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.126594445358638e-05,
+      "loss": 1.4077,
+      "step": 3770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1265723218693433e-05,
+      "loss": 1.5572,
+      "step": 3771
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1265501983800486e-05,
+      "loss": 1.5003,
+      "step": 3772
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1265280748907537e-05,
+      "loss": 1.4361,
+      "step": 3773
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.126505951401459e-05,
+      "loss": 1.3687,
+      "step": 3774
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1264838279121642e-05,
+      "loss": 1.3424,
+      "step": 3775
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1264617044228695e-05,
+      "loss": 1.4581,
+      "step": 3776
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1264395809335746e-05,
+      "loss": 1.3131,
+      "step": 3777
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1264174574442798e-05,
+      "loss": 1.4209,
+      "step": 3778
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.126395333954985e-05,
+      "loss": 1.4323,
+      "step": 3779
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1263732104656902e-05,
+      "loss": 1.408,
+      "step": 3780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1263510869763954e-05,
+      "loss": 1.4981,
+      "step": 3781
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1263289634871007e-05,
+      "loss": 1.4515,
+      "step": 3782
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.126306839997806e-05,
+      "loss": 1.4942,
+      "step": 3783
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.126284716508511e-05,
+      "loss": 1.413,
+      "step": 3784
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1262625930192163e-05,
+      "loss": 1.5015,
+      "step": 3785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1262404695299214e-05,
+      "loss": 1.3269,
+      "step": 3786
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1262183460406267e-05,
+      "loss": 1.5594,
+      "step": 3787
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.126196222551332e-05,
+      "loss": 1.558,
+      "step": 3788
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1261740990620372e-05,
+      "loss": 1.5153,
+      "step": 3789
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1261519755727423e-05,
+      "loss": 1.3932,
+      "step": 3790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1261298520834475e-05,
+      "loss": 1.3465,
+      "step": 3791
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1261077285941528e-05,
+      "loss": 1.3864,
+      "step": 3792
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.126085605104858e-05,
+      "loss": 1.3353,
+      "step": 3793
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1260634816155632e-05,
+      "loss": 1.452,
+      "step": 3794
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1260413581262683e-05,
+      "loss": 1.4557,
+      "step": 3795
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1260192346369735e-05,
+      "loss": 1.4485,
+      "step": 3796
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1259971111476788e-05,
+      "loss": 1.4529,
+      "step": 3797
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.125974987658384e-05,
+      "loss": 1.4833,
+      "step": 3798
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1259528641690893e-05,
+      "loss": 1.3616,
+      "step": 3799
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1259307406797946e-05,
+      "loss": 1.2356,
+      "step": 3800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1259086171904997e-05,
+      "loss": 1.4626,
+      "step": 3801
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1258864937012047e-05,
+      "loss": 1.4243,
+      "step": 3802
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.12586437021191e-05,
+      "loss": 1.5063,
+      "step": 3803
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1258422467226153e-05,
+      "loss": 1.422,
+      "step": 3804
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1258201232333205e-05,
+      "loss": 1.4687,
+      "step": 3805
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1257979997440256e-05,
+      "loss": 1.3581,
+      "step": 3806
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1257758762547309e-05,
+      "loss": 1.2678,
+      "step": 3807
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1257537527654361e-05,
+      "loss": 1.3,
+      "step": 3808
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1257316292761414e-05,
+      "loss": 1.2963,
+      "step": 3809
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1257095057868465e-05,
+      "loss": 1.5617,
+      "step": 3810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1256873822975518e-05,
+      "loss": 1.4334,
+      "step": 3811
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1256652588082568e-05,
+      "loss": 1.449,
+      "step": 3812
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1256431353189621e-05,
+      "loss": 1.3574,
+      "step": 3813
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1256210118296674e-05,
+      "loss": 1.6236,
+      "step": 3814
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1255988883403726e-05,
+      "loss": 1.4614,
+      "step": 3815
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1255767648510779e-05,
+      "loss": 1.4364,
+      "step": 3816
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.125554641361783e-05,
+      "loss": 1.4373,
+      "step": 3817
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.125532517872488e-05,
+      "loss": 1.4301,
+      "step": 3818
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1255103943831933e-05,
+      "loss": 1.4337,
+      "step": 3819
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1254882708938986e-05,
+      "loss": 1.408,
+      "step": 3820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1254661474046039e-05,
+      "loss": 1.335,
+      "step": 3821
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1254440239153091e-05,
+      "loss": 1.4671,
+      "step": 3822
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1254219004260142e-05,
+      "loss": 1.2618,
+      "step": 3823
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1253997769367195e-05,
+      "loss": 1.6372,
+      "step": 3824
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1253776534474247e-05,
+      "loss": 1.4442,
+      "step": 3825
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1253555299581298e-05,
+      "loss": 1.3846,
+      "step": 3826
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1253334064688351e-05,
+      "loss": 1.3878,
+      "step": 3827
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1253112829795402e-05,
+      "loss": 1.464,
+      "step": 3828
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1252891594902454e-05,
+      "loss": 1.681,
+      "step": 3829
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1252670360009507e-05,
+      "loss": 1.3483,
+      "step": 3830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.125244912511656e-05,
+      "loss": 1.5503,
+      "step": 3831
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1252227890223612e-05,
+      "loss": 1.5181,
+      "step": 3832
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1252006655330663e-05,
+      "loss": 1.4533,
+      "step": 3833
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1251785420437714e-05,
+      "loss": 1.4653,
+      "step": 3834
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1251564185544767e-05,
+      "loss": 1.3925,
+      "step": 3835
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.125134295065182e-05,
+      "loss": 1.3926,
+      "step": 3836
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1251121715758872e-05,
+      "loss": 1.5356,
+      "step": 3837
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1250900480865925e-05,
+      "loss": 1.4507,
+      "step": 3838
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1250679245972975e-05,
+      "loss": 1.3397,
+      "step": 3839
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1250458011080028e-05,
+      "loss": 1.4647,
+      "step": 3840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.125023677618708e-05,
+      "loss": 1.4903,
+      "step": 3841
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1250015541294132e-05,
+      "loss": 1.3869,
+      "step": 3842
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1249794306401184e-05,
+      "loss": 1.2692,
+      "step": 3843
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1249573071508237e-05,
+      "loss": 1.3276,
+      "step": 3844
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1249351836615288e-05,
+      "loss": 1.6446,
+      "step": 3845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.124913060172234e-05,
+      "loss": 1.5969,
+      "step": 3846
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1248909366829393e-05,
+      "loss": 1.4745,
+      "step": 3847
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1248688131936446e-05,
+      "loss": 1.4399,
+      "step": 3848
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1248466897043496e-05,
+      "loss": 1.2814,
+      "step": 3849
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1248245662150549e-05,
+      "loss": 1.357,
+      "step": 3850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.12480244272576e-05,
+      "loss": 1.3649,
+      "step": 3851
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1247803192364653e-05,
+      "loss": 1.4199,
+      "step": 3852
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1247581957471705e-05,
+      "loss": 1.3969,
+      "step": 3853
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1247360722578758e-05,
+      "loss": 1.3276,
+      "step": 3854
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.124713948768581e-05,
+      "loss": 1.2678,
+      "step": 3855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1246918252792861e-05,
+      "loss": 1.3834,
+      "step": 3856
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1246697017899914e-05,
+      "loss": 1.3721,
+      "step": 3857
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1246475783006965e-05,
+      "loss": 1.3872,
+      "step": 3858
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1246254548114018e-05,
+      "loss": 1.3107,
+      "step": 3859
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.124603331322107e-05,
+      "loss": 1.4089,
+      "step": 3860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1245812078328121e-05,
+      "loss": 1.4483,
+      "step": 3861
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1245590843435174e-05,
+      "loss": 1.3463,
+      "step": 3862
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1245369608542226e-05,
+      "loss": 1.3424,
+      "step": 3863
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1245148373649279e-05,
+      "loss": 1.3375,
+      "step": 3864
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.124492713875633e-05,
+      "loss": 1.3935,
+      "step": 3865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1244705903863382e-05,
+      "loss": 1.3548,
+      "step": 3866
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1244484668970433e-05,
+      "loss": 1.4123,
+      "step": 3867
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1244263434077486e-05,
+      "loss": 1.4323,
+      "step": 3868
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1244042199184539e-05,
+      "loss": 1.4389,
+      "step": 3869
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1243820964291591e-05,
+      "loss": 1.5276,
+      "step": 3870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1243599729398644e-05,
+      "loss": 1.3441,
+      "step": 3871
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1243378494505695e-05,
+      "loss": 1.5227,
+      "step": 3872
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1243157259612747e-05,
+      "loss": 1.3509,
+      "step": 3873
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1242936024719798e-05,
+      "loss": 1.422,
+      "step": 3874
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.124271478982685e-05,
+      "loss": 1.4809,
+      "step": 3875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1242493554933903e-05,
+      "loss": 1.3926,
+      "step": 3876
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1242272320040954e-05,
+      "loss": 1.4351,
+      "step": 3877
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1242051085148007e-05,
+      "loss": 1.3451,
+      "step": 3878
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.124182985025506e-05,
+      "loss": 1.3729,
+      "step": 3879
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1241608615362112e-05,
+      "loss": 1.4465,
+      "step": 3880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1241387380469165e-05,
+      "loss": 1.3413,
+      "step": 3881
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1241166145576216e-05,
+      "loss": 1.4319,
+      "step": 3882
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1240944910683267e-05,
+      "loss": 1.4396,
+      "step": 3883
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.124072367579032e-05,
+      "loss": 1.3827,
+      "step": 3884
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1240502440897372e-05,
+      "loss": 1.3992,
+      "step": 3885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1240281206004424e-05,
+      "loss": 1.4328,
+      "step": 3886
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1240059971111477e-05,
+      "loss": 1.4927,
+      "step": 3887
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1239838736218528e-05,
+      "loss": 1.551,
+      "step": 3888
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.123961750132558e-05,
+      "loss": 1.3979,
+      "step": 3889
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1239396266432632e-05,
+      "loss": 1.4534,
+      "step": 3890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1239175031539684e-05,
+      "loss": 1.4151,
+      "step": 3891
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1238953796646737e-05,
+      "loss": 1.3587,
+      "step": 3892
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.123873256175379e-05,
+      "loss": 1.3544,
+      "step": 3893
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.123851132686084e-05,
+      "loss": 1.4364,
+      "step": 3894
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1238290091967893e-05,
+      "loss": 1.439,
+      "step": 3895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1238068857074946e-05,
+      "loss": 1.4839,
+      "step": 3896
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1237847622181998e-05,
+      "loss": 1.5078,
+      "step": 3897
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1237626387289049e-05,
+      "loss": 1.3948,
+      "step": 3898
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.12374051523961e-05,
+      "loss": 1.5452,
+      "step": 3899
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1237183917503153e-05,
+      "loss": 1.3853,
+      "step": 3900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1236962682610205e-05,
+      "loss": 1.3346,
+      "step": 3901
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1236741447717258e-05,
+      "loss": 1.4493,
+      "step": 3902
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.123652021282431e-05,
+      "loss": 1.5587,
+      "step": 3903
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1236298977931363e-05,
+      "loss": 1.615,
+      "step": 3904
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1236077743038414e-05,
+      "loss": 1.4275,
+      "step": 3905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1235856508145465e-05,
+      "loss": 1.3425,
+      "step": 3906
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1235635273252517e-05,
+      "loss": 1.4892,
+      "step": 3907
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.123541403835957e-05,
+      "loss": 1.3359,
+      "step": 3908
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1235192803466623e-05,
+      "loss": 1.5638,
+      "step": 3909
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1234971568573674e-05,
+      "loss": 1.3028,
+      "step": 3910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1234750333680726e-05,
+      "loss": 1.4201,
+      "step": 3911
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1234529098787779e-05,
+      "loss": 1.452,
+      "step": 3912
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1234307863894831e-05,
+      "loss": 1.513,
+      "step": 3913
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1234086629001882e-05,
+      "loss": 1.4711,
+      "step": 3914
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1233865394108935e-05,
+      "loss": 1.5171,
+      "step": 3915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1233644159215986e-05,
+      "loss": 1.5803,
+      "step": 3916
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1233422924323038e-05,
+      "loss": 1.3265,
+      "step": 3917
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1233201689430091e-05,
+      "loss": 1.3863,
+      "step": 3918
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1232980454537144e-05,
+      "loss": 1.4106,
+      "step": 3919
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1232759219644196e-05,
+      "loss": 1.5354,
+      "step": 3920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1232537984751247e-05,
+      "loss": 1.3714,
+      "step": 3921
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1232316749858298e-05,
+      "loss": 1.3546,
+      "step": 3922
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.123209551496535e-05,
+      "loss": 1.4915,
+      "step": 3923
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1231874280072403e-05,
+      "loss": 1.46,
+      "step": 3924
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1231653045179456e-05,
+      "loss": 1.4879,
+      "step": 3925
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1231431810286509e-05,
+      "loss": 1.4136,
+      "step": 3926
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.123121057539356e-05,
+      "loss": 1.6275,
+      "step": 3927
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1230989340500612e-05,
+      "loss": 1.3343,
+      "step": 3928
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1230768105607665e-05,
+      "loss": 1.4374,
+      "step": 3929
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1230546870714716e-05,
+      "loss": 1.4579,
+      "step": 3930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1230325635821768e-05,
+      "loss": 1.656,
+      "step": 3931
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.123010440092882e-05,
+      "loss": 1.4224,
+      "step": 3932
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1229883166035872e-05,
+      "loss": 1.5194,
+      "step": 3933
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1229661931142924e-05,
+      "loss": 1.5026,
+      "step": 3934
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1229440696249977e-05,
+      "loss": 1.3414,
+      "step": 3935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.122921946135703e-05,
+      "loss": 1.5792,
+      "step": 3936
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.122899822646408e-05,
+      "loss": 1.423,
+      "step": 3937
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1228776991571133e-05,
+      "loss": 1.4715,
+      "step": 3938
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1228555756678184e-05,
+      "loss": 1.2047,
+      "step": 3939
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1228334521785237e-05,
+      "loss": 1.5195,
+      "step": 3940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.122811328689229e-05,
+      "loss": 1.3699,
+      "step": 3941
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1227892051999342e-05,
+      "loss": 1.4892,
+      "step": 3942
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1227670817106393e-05,
+      "loss": 1.49,
+      "step": 3943
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1227449582213445e-05,
+      "loss": 1.3489,
+      "step": 3944
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1227228347320498e-05,
+      "loss": 1.2804,
+      "step": 3945
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1227007112427549e-05,
+      "loss": 1.3796,
+      "step": 3946
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1226785877534602e-05,
+      "loss": 1.3838,
+      "step": 3947
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1226564642641654e-05,
+      "loss": 1.2999,
+      "step": 3948
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1226343407748705e-05,
+      "loss": 1.325,
+      "step": 3949
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1226122172855758e-05,
+      "loss": 1.6169,
+      "step": 3950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.122590093796281e-05,
+      "loss": 1.4284,
+      "step": 3951
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1225679703069863e-05,
+      "loss": 1.3661,
+      "step": 3952
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1225458468176914e-05,
+      "loss": 1.3622,
+      "step": 3953
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1225237233283966e-05,
+      "loss": 1.4342,
+      "step": 3954
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1225015998391017e-05,
+      "loss": 1.4322,
+      "step": 3955
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.122479476349807e-05,
+      "loss": 1.5983,
+      "step": 3956
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1224573528605123e-05,
+      "loss": 1.4031,
+      "step": 3957
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1224352293712175e-05,
+      "loss": 1.3879,
+      "step": 3958
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1224131058819228e-05,
+      "loss": 1.4898,
+      "step": 3959
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1223909823926279e-05,
+      "loss": 1.575,
+      "step": 3960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1223688589033331e-05,
+      "loss": 1.204,
+      "step": 3961
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1223467354140382e-05,
+      "loss": 1.5172,
+      "step": 3962
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1223246119247435e-05,
+      "loss": 1.4592,
+      "step": 3963
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1223024884354488e-05,
+      "loss": 1.3696,
+      "step": 3964
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1222803649461538e-05,
+      "loss": 1.4459,
+      "step": 3965
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1222582414568591e-05,
+      "loss": 1.3155,
+      "step": 3966
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1222361179675644e-05,
+      "loss": 1.5179,
+      "step": 3967
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1222139944782696e-05,
+      "loss": 1.4431,
+      "step": 3968
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1221918709889749e-05,
+      "loss": 1.3035,
+      "step": 3969
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.12216974749968e-05,
+      "loss": 1.4277,
+      "step": 3970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.122147624010385e-05,
+      "loss": 1.3671,
+      "step": 3971
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1221255005210903e-05,
+      "loss": 1.4163,
+      "step": 3972
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1221033770317956e-05,
+      "loss": 1.3811,
+      "step": 3973
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1220812535425009e-05,
+      "loss": 1.4052,
+      "step": 3974
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1220591300532061e-05,
+      "loss": 1.3533,
+      "step": 3975
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1220370065639112e-05,
+      "loss": 1.3088,
+      "step": 3976
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1220148830746165e-05,
+      "loss": 1.5793,
+      "step": 3977
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1219927595853216e-05,
+      "loss": 1.4271,
+      "step": 3978
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1219706360960268e-05,
+      "loss": 1.3419,
+      "step": 3979
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1219485126067321e-05,
+      "loss": 1.4201,
+      "step": 3980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1219263891174372e-05,
+      "loss": 1.3523,
+      "step": 3981
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1219042656281424e-05,
+      "loss": 1.4009,
+      "step": 3982
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1218821421388477e-05,
+      "loss": 1.384,
+      "step": 3983
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.121860018649553e-05,
+      "loss": 1.5475,
+      "step": 3984
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1218378951602582e-05,
+      "loss": 1.4547,
+      "step": 3985
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1218157716709633e-05,
+      "loss": 1.447,
+      "step": 3986
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1217936481816684e-05,
+      "loss": 1.4593,
+      "step": 3987
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1217715246923737e-05,
+      "loss": 1.4153,
+      "step": 3988
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.121749401203079e-05,
+      "loss": 1.4802,
+      "step": 3989
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1217272777137842e-05,
+      "loss": 1.3175,
+      "step": 3990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1217051542244894e-05,
+      "loss": 1.276,
+      "step": 3991
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1216830307351945e-05,
+      "loss": 1.3958,
+      "step": 3992
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1216609072458998e-05,
+      "loss": 1.3706,
+      "step": 3993
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1216387837566049e-05,
+      "loss": 1.4383,
+      "step": 3994
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1216166602673102e-05,
+      "loss": 1.4887,
+      "step": 3995
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1215945367780154e-05,
+      "loss": 1.4733,
+      "step": 3996
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1215724132887207e-05,
+      "loss": 1.4368,
+      "step": 3997
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1215502897994258e-05,
+      "loss": 1.4024,
+      "step": 3998
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.121528166310131e-05,
+      "loss": 1.4232,
+      "step": 3999
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1215060428208363e-05,
+      "loss": 1.4547,
+      "step": 4000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1214839193315416e-05,
+      "loss": 1.512,
+      "step": 4001
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1214617958422466e-05,
+      "loss": 1.3641,
+      "step": 4002
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1214396723529517e-05,
+      "loss": 1.4343,
+      "step": 4003
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.121417548863657e-05,
+      "loss": 1.3723,
+      "step": 4004
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1213954253743623e-05,
+      "loss": 1.5984,
+      "step": 4005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1213733018850675e-05,
+      "loss": 1.4127,
+      "step": 4006
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1213511783957728e-05,
+      "loss": 1.3135,
+      "step": 4007
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.121329054906478e-05,
+      "loss": 1.3792,
+      "step": 4008
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1213069314171831e-05,
+      "loss": 1.5782,
+      "step": 4009
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1212848079278882e-05,
+      "loss": 1.4258,
+      "step": 4010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1212626844385935e-05,
+      "loss": 1.4555,
+      "step": 4011
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1212405609492987e-05,
+      "loss": 1.3082,
+      "step": 4012
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.121218437460004e-05,
+      "loss": 1.305,
+      "step": 4013
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1211963139707091e-05,
+      "loss": 1.4996,
+      "step": 4014
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1211741904814144e-05,
+      "loss": 1.3646,
+      "step": 4015
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1211520669921196e-05,
+      "loss": 1.3236,
+      "step": 4016
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1211299435028249e-05,
+      "loss": 1.3663,
+      "step": 4017
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.12110782001353e-05,
+      "loss": 1.2216,
+      "step": 4018
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1210856965242352e-05,
+      "loss": 1.3011,
+      "step": 4019
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1210635730349403e-05,
+      "loss": 1.4573,
+      "step": 4020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1210414495456456e-05,
+      "loss": 1.3866,
+      "step": 4021
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1210193260563509e-05,
+      "loss": 1.3423,
+      "step": 4022
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1209972025670561e-05,
+      "loss": 1.4245,
+      "step": 4023
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1209750790777614e-05,
+      "loss": 1.4621,
+      "step": 4024
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1209529555884665e-05,
+      "loss": 1.3122,
+      "step": 4025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1209308320991717e-05,
+      "loss": 1.4054,
+      "step": 4026
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1209087086098768e-05,
+      "loss": 1.448,
+      "step": 4027
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.120886585120582e-05,
+      "loss": 1.379,
+      "step": 4028
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1208644616312873e-05,
+      "loss": 1.3643,
+      "step": 4029
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1208423381419926e-05,
+      "loss": 1.5378,
+      "step": 4030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1208202146526977e-05,
+      "loss": 1.23,
+      "step": 4031
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.120798091163403e-05,
+      "loss": 1.4138,
+      "step": 4032
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1207759676741082e-05,
+      "loss": 1.4078,
+      "step": 4033
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1207538441848133e-05,
+      "loss": 1.4126,
+      "step": 4034
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1207317206955186e-05,
+      "loss": 1.4177,
+      "step": 4035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1207095972062237e-05,
+      "loss": 1.6323,
+      "step": 4036
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.120687473716929e-05,
+      "loss": 1.321,
+      "step": 4037
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1206653502276342e-05,
+      "loss": 1.4179,
+      "step": 4038
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1206432267383394e-05,
+      "loss": 1.393,
+      "step": 4039
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1206211032490447e-05,
+      "loss": 1.3995,
+      "step": 4040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1205989797597498e-05,
+      "loss": 1.2826,
+      "step": 4041
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.120576856270455e-05,
+      "loss": 1.3233,
+      "step": 4042
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1205547327811602e-05,
+      "loss": 1.3645,
+      "step": 4043
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1205326092918654e-05,
+      "loss": 1.5179,
+      "step": 4044
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1205104858025707e-05,
+      "loss": 1.3094,
+      "step": 4045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.120488362313276e-05,
+      "loss": 1.3886,
+      "step": 4046
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.120466238823981e-05,
+      "loss": 1.3805,
+      "step": 4047
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1204441153346863e-05,
+      "loss": 1.3865,
+      "step": 4048
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1204219918453915e-05,
+      "loss": 1.2054,
+      "step": 4049
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1203998683560966e-05,
+      "loss": 1.4272,
+      "step": 4050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1203777448668019e-05,
+      "loss": 1.4206,
+      "step": 4051
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1203556213775072e-05,
+      "loss": 1.4121,
+      "step": 4052
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1203334978882123e-05,
+      "loss": 1.477,
+      "step": 4053
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1203113743989175e-05,
+      "loss": 1.2912,
+      "step": 4054
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1202892509096228e-05,
+      "loss": 1.4263,
+      "step": 4055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.120267127420328e-05,
+      "loss": 1.5286,
+      "step": 4056
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1202450039310333e-05,
+      "loss": 1.4288,
+      "step": 4057
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1202228804417384e-05,
+      "loss": 1.3422,
+      "step": 4058
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1202007569524435e-05,
+      "loss": 1.4517,
+      "step": 4059
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1201786334631487e-05,
+      "loss": 1.4252,
+      "step": 4060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.120156509973854e-05,
+      "loss": 1.291,
+      "step": 4061
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1201343864845593e-05,
+      "loss": 1.4408,
+      "step": 4062
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1201122629952644e-05,
+      "loss": 1.3117,
+      "step": 4063
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1200901395059696e-05,
+      "loss": 1.4045,
+      "step": 4064
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1200680160166749e-05,
+      "loss": 1.4307,
+      "step": 4065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.12004589252738e-05,
+      "loss": 1.4417,
+      "step": 4066
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1200237690380852e-05,
+      "loss": 1.5975,
+      "step": 4067
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1200016455487905e-05,
+      "loss": 1.6069,
+      "step": 4068
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1199795220594956e-05,
+      "loss": 1.2458,
+      "step": 4069
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1199573985702008e-05,
+      "loss": 1.3428,
+      "step": 4070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1199352750809061e-05,
+      "loss": 1.5035,
+      "step": 4071
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1199131515916114e-05,
+      "loss": 1.4373,
+      "step": 4072
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1198910281023166e-05,
+      "loss": 1.3476,
+      "step": 4073
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1198689046130217e-05,
+      "loss": 1.4536,
+      "step": 4074
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1198467811237268e-05,
+      "loss": 1.3165,
+      "step": 4075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.119824657634432e-05,
+      "loss": 1.4681,
+      "step": 4076
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1198025341451373e-05,
+      "loss": 1.3487,
+      "step": 4077
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1197804106558426e-05,
+      "loss": 1.4636,
+      "step": 4078
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1197582871665479e-05,
+      "loss": 1.4001,
+      "step": 4079
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.119736163677253e-05,
+      "loss": 1.3847,
+      "step": 4080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1197140401879582e-05,
+      "loss": 1.4723,
+      "step": 4081
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1196919166986633e-05,
+      "loss": 1.4719,
+      "step": 4082
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1196697932093686e-05,
+      "loss": 1.5486,
+      "step": 4083
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1196476697200738e-05,
+      "loss": 1.4526,
+      "step": 4084
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.119625546230779e-05,
+      "loss": 1.408,
+      "step": 4085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1196034227414842e-05,
+      "loss": 1.4201,
+      "step": 4086
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1195812992521894e-05,
+      "loss": 1.3638,
+      "step": 4087
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1195591757628947e-05,
+      "loss": 1.3754,
+      "step": 4088
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1195370522736e-05,
+      "loss": 1.7052,
+      "step": 4089
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.119514928784305e-05,
+      "loss": 1.361,
+      "step": 4090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1194928052950101e-05,
+      "loss": 1.2073,
+      "step": 4091
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1194706818057154e-05,
+      "loss": 1.4829,
+      "step": 4092
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1194485583164207e-05,
+      "loss": 1.3877,
+      "step": 4093
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.119426434827126e-05,
+      "loss": 1.4114,
+      "step": 4094
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1194043113378312e-05,
+      "loss": 1.42,
+      "step": 4095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1193821878485363e-05,
+      "loss": 1.4795,
+      "step": 4096
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1193600643592415e-05,
+      "loss": 1.4625,
+      "step": 4097
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1193379408699466e-05,
+      "loss": 1.3969,
+      "step": 4098
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1193158173806519e-05,
+      "loss": 1.5719,
+      "step": 4099
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1192936938913572e-05,
+      "loss": 1.2408,
+      "step": 4100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1192715704020624e-05,
+      "loss": 1.4331,
+      "step": 4101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1192494469127675e-05,
+      "loss": 1.4437,
+      "step": 4102
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1192273234234728e-05,
+      "loss": 1.4773,
+      "step": 4103
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.119205199934178e-05,
+      "loss": 1.5517,
+      "step": 4104
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1191830764448833e-05,
+      "loss": 1.4312,
+      "step": 4105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1191609529555884e-05,
+      "loss": 1.4684,
+      "step": 4106
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1191388294662935e-05,
+      "loss": 1.4634,
+      "step": 4107
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1191167059769987e-05,
+      "loss": 1.4631,
+      "step": 4108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.119094582487704e-05,
+      "loss": 1.4031,
+      "step": 4109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1190724589984093e-05,
+      "loss": 1.3788,
+      "step": 4110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1190503355091145e-05,
+      "loss": 1.3413,
+      "step": 4111
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1190282120198198e-05,
+      "loss": 1.4275,
+      "step": 4112
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1190060885305249e-05,
+      "loss": 1.2151,
+      "step": 4113
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1189839650412301e-05,
+      "loss": 1.5817,
+      "step": 4114
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1189618415519352e-05,
+      "loss": 1.5082,
+      "step": 4115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1189397180626405e-05,
+      "loss": 1.428,
+      "step": 4116
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1189175945733458e-05,
+      "loss": 1.4577,
+      "step": 4117
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1188954710840508e-05,
+      "loss": 1.3073,
+      "step": 4118
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1188733475947561e-05,
+      "loss": 1.4401,
+      "step": 4119
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1188512241054614e-05,
+      "loss": 1.3089,
+      "step": 4120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1188291006161666e-05,
+      "loss": 1.3487,
+      "step": 4121
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1188069771268717e-05,
+      "loss": 1.4284,
+      "step": 4122
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.118784853637577e-05,
+      "loss": 1.5245,
+      "step": 4123
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.118762730148282e-05,
+      "loss": 1.4472,
+      "step": 4124
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1187406066589873e-05,
+      "loss": 1.4269,
+      "step": 4125
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1187184831696926e-05,
+      "loss": 1.4689,
+      "step": 4126
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1186963596803979e-05,
+      "loss": 1.3808,
+      "step": 4127
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1186742361911031e-05,
+      "loss": 1.2974,
+      "step": 4128
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1186521127018082e-05,
+      "loss": 1.345,
+      "step": 4129
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1186299892125135e-05,
+      "loss": 1.3564,
+      "step": 4130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1186078657232186e-05,
+      "loss": 1.38,
+      "step": 4131
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1185857422339238e-05,
+      "loss": 1.3956,
+      "step": 4132
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.118563618744629e-05,
+      "loss": 1.456,
+      "step": 4133
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1185414952553343e-05,
+      "loss": 1.524,
+      "step": 4134
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1185193717660394e-05,
+      "loss": 1.4278,
+      "step": 4135
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1184972482767447e-05,
+      "loss": 1.3161,
+      "step": 4136
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.11847512478745e-05,
+      "loss": 1.4028,
+      "step": 4137
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.118453001298155e-05,
+      "loss": 1.3817,
+      "step": 4138
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1184308778088603e-05,
+      "loss": 1.3435,
+      "step": 4139
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1184087543195654e-05,
+      "loss": 1.3477,
+      "step": 4140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1183866308302707e-05,
+      "loss": 1.4716,
+      "step": 4141
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.118364507340976e-05,
+      "loss": 1.2985,
+      "step": 4142
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1183423838516812e-05,
+      "loss": 1.5071,
+      "step": 4143
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1183202603623864e-05,
+      "loss": 1.3751,
+      "step": 4144
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1182981368730917e-05,
+      "loss": 1.2865,
+      "step": 4145
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1182760133837968e-05,
+      "loss": 1.2654,
+      "step": 4146
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1182538898945019e-05,
+      "loss": 1.4072,
+      "step": 4147
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1182317664052072e-05,
+      "loss": 1.4017,
+      "step": 4148
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1182096429159124e-05,
+      "loss": 1.4031,
+      "step": 4149
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1181875194266177e-05,
+      "loss": 1.3811,
+      "step": 4150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1181653959373228e-05,
+      "loss": 1.382,
+      "step": 4151
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.118143272448028e-05,
+      "loss": 1.3623,
+      "step": 4152
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1181211489587333e-05,
+      "loss": 1.3449,
+      "step": 4153
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1180990254694384e-05,
+      "loss": 1.4915,
+      "step": 4154
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1180769019801436e-05,
+      "loss": 1.5366,
+      "step": 4155
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1180547784908489e-05,
+      "loss": 1.2799,
+      "step": 4156
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.118032655001554e-05,
+      "loss": 1.3905,
+      "step": 4157
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1180105315122593e-05,
+      "loss": 1.4896,
+      "step": 4158
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1179884080229645e-05,
+      "loss": 1.3645,
+      "step": 4159
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1179662845336698e-05,
+      "loss": 1.5203,
+      "step": 4160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.117944161044375e-05,
+      "loss": 1.5171,
+      "step": 4161
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1179220375550801e-05,
+      "loss": 1.4372,
+      "step": 4162
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1178999140657852e-05,
+      "loss": 1.4644,
+      "step": 4163
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1178777905764905e-05,
+      "loss": 1.4455,
+      "step": 4164
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1178556670871957e-05,
+      "loss": 1.3585,
+      "step": 4165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.117833543597901e-05,
+      "loss": 1.3068,
+      "step": 4166
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1178114201086061e-05,
+      "loss": 1.4686,
+      "step": 4167
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1177892966193114e-05,
+      "loss": 1.7229,
+      "step": 4168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1177671731300166e-05,
+      "loss": 1.3653,
+      "step": 4169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1177450496407217e-05,
+      "loss": 1.3034,
+      "step": 4170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.117722926151427e-05,
+      "loss": 1.3504,
+      "step": 4171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1177008026621322e-05,
+      "loss": 1.29,
+      "step": 4172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1176786791728373e-05,
+      "loss": 1.4959,
+      "step": 4173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1176565556835426e-05,
+      "loss": 1.4567,
+      "step": 4174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1176344321942479e-05,
+      "loss": 1.4856,
+      "step": 4175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1176123087049531e-05,
+      "loss": 1.3824,
+      "step": 4176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1175901852156584e-05,
+      "loss": 1.2487,
+      "step": 4177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1175680617263635e-05,
+      "loss": 1.3815,
+      "step": 4178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1175459382370686e-05,
+      "loss": 1.415,
+      "step": 4179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1175238147477738e-05,
+      "loss": 1.3127,
+      "step": 4180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.117501691258479e-05,
+      "loss": 1.3915,
+      "step": 4181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1174795677691843e-05,
+      "loss": 1.4771,
+      "step": 4182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1174574442798896e-05,
+      "loss": 1.3244,
+      "step": 4183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1174353207905947e-05,
+      "loss": 1.4554,
+      "step": 4184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1174131973013e-05,
+      "loss": 1.4302,
+      "step": 4185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.117391073812005e-05,
+      "loss": 1.3652,
+      "step": 4186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1173689503227103e-05,
+      "loss": 1.4271,
+      "step": 4187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1173468268334156e-05,
+      "loss": 1.3765,
+      "step": 4188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1173247033441207e-05,
+      "loss": 1.5041,
+      "step": 4189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.117302579854826e-05,
+      "loss": 1.4077,
+      "step": 4190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1172804563655312e-05,
+      "loss": 1.4902,
+      "step": 4191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1172583328762364e-05,
+      "loss": 1.5154,
+      "step": 4192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1172362093869417e-05,
+      "loss": 1.3395,
+      "step": 4193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1172140858976468e-05,
+      "loss": 1.3716,
+      "step": 4194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1171919624083519e-05,
+      "loss": 1.3059,
+      "step": 4195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1171698389190572e-05,
+      "loss": 1.5231,
+      "step": 4196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1171477154297624e-05,
+      "loss": 1.4115,
+      "step": 4197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1171255919404677e-05,
+      "loss": 1.4202,
+      "step": 4198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.117103468451173e-05,
+      "loss": 1.4467,
+      "step": 4199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.117081344961878e-05,
+      "loss": 1.3737,
+      "step": 4200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1170592214725833e-05,
+      "loss": 1.279,
+      "step": 4201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1170370979832885e-05,
+      "loss": 1.5371,
+      "step": 4202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1170149744939936e-05,
+      "loss": 1.4677,
+      "step": 4203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1169928510046989e-05,
+      "loss": 1.3343,
+      "step": 4204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1169707275154042e-05,
+      "loss": 1.327,
+      "step": 4205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1169486040261093e-05,
+      "loss": 1.4793,
+      "step": 4206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1169264805368145e-05,
+      "loss": 1.5379,
+      "step": 4207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1169043570475198e-05,
+      "loss": 1.2825,
+      "step": 4208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.116882233558225e-05,
+      "loss": 1.3371,
+      "step": 4209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1168601100689301e-05,
+      "loss": 1.3733,
+      "step": 4210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1168379865796352e-05,
+      "loss": 1.3576,
+      "step": 4211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1168158630903405e-05,
+      "loss": 1.4088,
+      "step": 4212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1167937396010457e-05,
+      "loss": 1.3648,
+      "step": 4213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.116771616111751e-05,
+      "loss": 1.3105,
+      "step": 4214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1167494926224563e-05,
+      "loss": 1.4175,
+      "step": 4215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1167273691331615e-05,
+      "loss": 1.4956,
+      "step": 4216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1167052456438666e-05,
+      "loss": 1.6127,
+      "step": 4217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1166831221545719e-05,
+      "loss": 1.5205,
+      "step": 4218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.116660998665277e-05,
+      "loss": 1.6426,
+      "step": 4219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1166388751759822e-05,
+      "loss": 1.5276,
+      "step": 4220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1166167516866875e-05,
+      "loss": 1.3195,
+      "step": 4221
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1165946281973926e-05,
+      "loss": 1.2834,
+      "step": 4222
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1165725047080978e-05,
+      "loss": 1.5046,
+      "step": 4223
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1165503812188031e-05,
+      "loss": 1.4794,
+      "step": 4224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1165282577295084e-05,
+      "loss": 1.4877,
+      "step": 4225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1165061342402135e-05,
+      "loss": 1.2619,
+      "step": 4226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1164840107509187e-05,
+      "loss": 1.3917,
+      "step": 4227
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1164618872616238e-05,
+      "loss": 1.2642,
+      "step": 4228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.116439763772329e-05,
+      "loss": 1.3211,
+      "step": 4229
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1164176402830343e-05,
+      "loss": 1.4661,
+      "step": 4230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1163955167937396e-05,
+      "loss": 1.3557,
+      "step": 4231
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1163733933044449e-05,
+      "loss": 1.4918,
+      "step": 4232
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.11635126981515e-05,
+      "loss": 1.2714,
+      "step": 4233
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1163291463258552e-05,
+      "loss": 1.3346,
+      "step": 4234
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1163070228365603e-05,
+      "loss": 1.4089,
+      "step": 4235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1162848993472656e-05,
+      "loss": 1.4151,
+      "step": 4236
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1162627758579708e-05,
+      "loss": 1.3278,
+      "step": 4237
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1162406523686761e-05,
+      "loss": 1.297,
+      "step": 4238
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1162185288793812e-05,
+      "loss": 1.4298,
+      "step": 4239
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1161964053900864e-05,
+      "loss": 1.3918,
+      "step": 4240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1161742819007917e-05,
+      "loss": 1.3314,
+      "step": 4241
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1161521584114968e-05,
+      "loss": 1.3754,
+      "step": 4242
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.116130034922202e-05,
+      "loss": 1.4575,
+      "step": 4243
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1161079114329071e-05,
+      "loss": 1.3656,
+      "step": 4244
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1160857879436124e-05,
+      "loss": 1.4746,
+      "step": 4245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1160636644543177e-05,
+      "loss": 1.423,
+      "step": 4246
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.116041540965023e-05,
+      "loss": 1.4174,
+      "step": 4247
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1160194174757282e-05,
+      "loss": 1.3396,
+      "step": 4248
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1159972939864335e-05,
+      "loss": 1.3475,
+      "step": 4249
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1159751704971385e-05,
+      "loss": 1.3731,
+      "step": 4250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1159530470078436e-05,
+      "loss": 1.3513,
+      "step": 4251
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1159309235185489e-05,
+      "loss": 1.5077,
+      "step": 4252
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1159088000292542e-05,
+      "loss": 1.4289,
+      "step": 4253
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1158866765399594e-05,
+      "loss": 1.4735,
+      "step": 4254
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1158645530506645e-05,
+      "loss": 1.5488,
+      "step": 4255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1158424295613698e-05,
+      "loss": 1.4326,
+      "step": 4256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.115820306072075e-05,
+      "loss": 1.4499,
+      "step": 4257
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1157981825827801e-05,
+      "loss": 1.3208,
+      "step": 4258
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1157760590934854e-05,
+      "loss": 1.4666,
+      "step": 4259
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1157539356041905e-05,
+      "loss": 1.4655,
+      "step": 4260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1157318121148957e-05,
+      "loss": 1.3856,
+      "step": 4261
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.115709688625601e-05,
+      "loss": 1.5165,
+      "step": 4262
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1156875651363063e-05,
+      "loss": 1.2836,
+      "step": 4263
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1156654416470115e-05,
+      "loss": 1.3904,
+      "step": 4264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1156433181577168e-05,
+      "loss": 1.366,
+      "step": 4265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1156211946684219e-05,
+      "loss": 1.6578,
+      "step": 4266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.115599071179127e-05,
+      "loss": 1.5148,
+      "step": 4267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1155769476898322e-05,
+      "loss": 1.4143,
+      "step": 4268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1155548242005375e-05,
+      "loss": 1.2413,
+      "step": 4269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1155327007112428e-05,
+      "loss": 1.3081,
+      "step": 4270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1155105772219478e-05,
+      "loss": 1.4746,
+      "step": 4271
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1154884537326531e-05,
+      "loss": 1.5929,
+      "step": 4272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1154663302433584e-05,
+      "loss": 1.3825,
+      "step": 4273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1154442067540635e-05,
+      "loss": 1.4369,
+      "step": 4274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1154220832647687e-05,
+      "loss": 1.3806,
+      "step": 4275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.115399959775474e-05,
+      "loss": 1.2904,
+      "step": 4276
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.115377836286179e-05,
+      "loss": 1.4206,
+      "step": 4277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1153557127968843e-05,
+      "loss": 1.427,
+      "step": 4278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1153335893075896e-05,
+      "loss": 1.4881,
+      "step": 4279
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1153114658182949e-05,
+      "loss": 1.3618,
+      "step": 4280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1152893423290001e-05,
+      "loss": 1.5387,
+      "step": 4281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1152672188397052e-05,
+      "loss": 1.4156,
+      "step": 4282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1152450953504103e-05,
+      "loss": 1.3996,
+      "step": 4283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1152229718611156e-05,
+      "loss": 1.3751,
+      "step": 4284
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1152008483718208e-05,
+      "loss": 1.5487,
+      "step": 4285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.115178724882526e-05,
+      "loss": 1.4068,
+      "step": 4286
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1151566013932313e-05,
+      "loss": 1.2851,
+      "step": 4287
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1151344779039364e-05,
+      "loss": 1.4899,
+      "step": 4288
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1151123544146417e-05,
+      "loss": 1.4522,
+      "step": 4289
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.115090230925347e-05,
+      "loss": 1.5452,
+      "step": 4290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.115068107436052e-05,
+      "loss": 1.4064,
+      "step": 4291
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1150459839467573e-05,
+      "loss": 1.3776,
+      "step": 4292
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1150238604574624e-05,
+      "loss": 1.4031,
+      "step": 4293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1150017369681677e-05,
+      "loss": 1.2968,
+      "step": 4294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.114979613478873e-05,
+      "loss": 1.4306,
+      "step": 4295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1149574899895782e-05,
+      "loss": 1.2888,
+      "step": 4296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1149353665002834e-05,
+      "loss": 1.3685,
+      "step": 4297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1149132430109885e-05,
+      "loss": 1.4633,
+      "step": 4298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1148911195216936e-05,
+      "loss": 1.4679,
+      "step": 4299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1148689960323989e-05,
+      "loss": 1.3799,
+      "step": 4300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1148468725431042e-05,
+      "loss": 1.4774,
+      "step": 4301
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1148247490538094e-05,
+      "loss": 1.3875,
+      "step": 4302
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1148026255645147e-05,
+      "loss": 1.3514,
+      "step": 4303
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1147805020752198e-05,
+      "loss": 1.3246,
+      "step": 4304
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.114758378585925e-05,
+      "loss": 1.3981,
+      "step": 4305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1147362550966303e-05,
+      "loss": 1.2525,
+      "step": 4306
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1147141316073354e-05,
+      "loss": 1.3254,
+      "step": 4307
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1146920081180406e-05,
+      "loss": 1.2086,
+      "step": 4308
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1146698846287459e-05,
+      "loss": 1.5225,
+      "step": 4309
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.114647761139451e-05,
+      "loss": 1.3153,
+      "step": 4310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1146256376501563e-05,
+      "loss": 1.36,
+      "step": 4311
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1146035141608615e-05,
+      "loss": 1.4388,
+      "step": 4312
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1145813906715668e-05,
+      "loss": 1.4073,
+      "step": 4313
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1145592671822719e-05,
+      "loss": 1.4529,
+      "step": 4314
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.114537143692977e-05,
+      "loss": 1.429,
+      "step": 4315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1145150202036822e-05,
+      "loss": 1.4667,
+      "step": 4316
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1144928967143875e-05,
+      "loss": 1.3934,
+      "step": 4317
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1144707732250927e-05,
+      "loss": 1.4617,
+      "step": 4318
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.114448649735798e-05,
+      "loss": 1.399,
+      "step": 4319
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1144265262465033e-05,
+      "loss": 1.4579,
+      "step": 4320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1144044027572084e-05,
+      "loss": 1.3657,
+      "step": 4321
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1143822792679136e-05,
+      "loss": 1.4017,
+      "step": 4322
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1143601557786187e-05,
+      "loss": 1.3366,
+      "step": 4323
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.114338032289324e-05,
+      "loss": 1.2651,
+      "step": 4324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1143159088000292e-05,
+      "loss": 1.3431,
+      "step": 4325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1142937853107343e-05,
+      "loss": 1.4393,
+      "step": 4326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1142716618214396e-05,
+      "loss": 1.2819,
+      "step": 4327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1142495383321448e-05,
+      "loss": 1.4148,
+      "step": 4328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1142274148428501e-05,
+      "loss": 1.4414,
+      "step": 4329
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1142052913535552e-05,
+      "loss": 1.4697,
+      "step": 4330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1141831678642605e-05,
+      "loss": 1.3826,
+      "step": 4331
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1141610443749656e-05,
+      "loss": 1.3162,
+      "step": 4332
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1141389208856708e-05,
+      "loss": 1.4219,
+      "step": 4333
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.114116797396376e-05,
+      "loss": 1.4505,
+      "step": 4334
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1140946739070813e-05,
+      "loss": 1.3997,
+      "step": 4335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1140725504177866e-05,
+      "loss": 1.3552,
+      "step": 4336
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1140504269284917e-05,
+      "loss": 1.284,
+      "step": 4337
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.114028303439197e-05,
+      "loss": 1.4761,
+      "step": 4338
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.114006179949902e-05,
+      "loss": 1.555,
+      "step": 4339
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1139840564606073e-05,
+      "loss": 1.2603,
+      "step": 4340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1139619329713126e-05,
+      "loss": 1.4212,
+      "step": 4341
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1139398094820178e-05,
+      "loss": 1.4056,
+      "step": 4342
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.113917685992723e-05,
+      "loss": 1.3732,
+      "step": 4343
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1138955625034282e-05,
+      "loss": 1.5569,
+      "step": 4344
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1138734390141334e-05,
+      "loss": 1.394,
+      "step": 4345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1138513155248385e-05,
+      "loss": 1.4425,
+      "step": 4346
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1138291920355438e-05,
+      "loss": 1.4007,
+      "step": 4347
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1138070685462489e-05,
+      "loss": 1.4648,
+      "step": 4348
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1137849450569541e-05,
+      "loss": 1.3168,
+      "step": 4349
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1137628215676594e-05,
+      "loss": 1.4345,
+      "step": 4350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1137406980783647e-05,
+      "loss": 1.3709,
+      "step": 4351
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.11371857458907e-05,
+      "loss": 1.6096,
+      "step": 4352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1136964510997752e-05,
+      "loss": 1.4156,
+      "step": 4353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1136743276104803e-05,
+      "loss": 1.3694,
+      "step": 4354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1136522041211854e-05,
+      "loss": 1.2505,
+      "step": 4355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1136300806318906e-05,
+      "loss": 1.5191,
+      "step": 4356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1136079571425959e-05,
+      "loss": 1.356,
+      "step": 4357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1135858336533012e-05,
+      "loss": 1.4057,
+      "step": 4358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1135637101640063e-05,
+      "loss": 1.3961,
+      "step": 4359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1135415866747115e-05,
+      "loss": 1.464,
+      "step": 4360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1135194631854168e-05,
+      "loss": 1.3068,
+      "step": 4361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1134973396961219e-05,
+      "loss": 1.3745,
+      "step": 4362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1134752162068271e-05,
+      "loss": 1.435,
+      "step": 4363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1134530927175322e-05,
+      "loss": 1.3954,
+      "step": 4364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1134309692282375e-05,
+      "loss": 1.423,
+      "step": 4365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1134088457389427e-05,
+      "loss": 1.446,
+      "step": 4366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.113386722249648e-05,
+      "loss": 1.3318,
+      "step": 4367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1133645987603533e-05,
+      "loss": 1.4523,
+      "step": 4368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1133424752710585e-05,
+      "loss": 1.4181,
+      "step": 4369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1133203517817636e-05,
+      "loss": 1.3655,
+      "step": 4370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1132982282924687e-05,
+      "loss": 1.2775,
+      "step": 4371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.113276104803174e-05,
+      "loss": 1.3627,
+      "step": 4372
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1132539813138792e-05,
+      "loss": 1.2999,
+      "step": 4373
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1132318578245845e-05,
+      "loss": 1.3839,
+      "step": 4374
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1132097343352896e-05,
+      "loss": 1.4863,
+      "step": 4375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1131876108459948e-05,
+      "loss": 1.476,
+      "step": 4376
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1131654873567001e-05,
+      "loss": 1.421,
+      "step": 4377
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1131433638674054e-05,
+      "loss": 1.4225,
+      "step": 4378
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1131212403781105e-05,
+      "loss": 1.364,
+      "step": 4379
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1130991168888157e-05,
+      "loss": 1.593,
+      "step": 4380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1130769933995208e-05,
+      "loss": 1.3536,
+      "step": 4381
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.113054869910226e-05,
+      "loss": 1.3011,
+      "step": 4382
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1130327464209313e-05,
+      "loss": 1.3671,
+      "step": 4383
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1130106229316366e-05,
+      "loss": 1.3157,
+      "step": 4384
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1129884994423419e-05,
+      "loss": 1.3314,
+      "step": 4385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.112966375953047e-05,
+      "loss": 1.3627,
+      "step": 4386
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.112944252463752e-05,
+      "loss": 1.5532,
+      "step": 4387
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1129221289744573e-05,
+      "loss": 1.4505,
+      "step": 4388
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1129000054851626e-05,
+      "loss": 1.4532,
+      "step": 4389
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1128778819958678e-05,
+      "loss": 1.457,
+      "step": 4390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1128557585065731e-05,
+      "loss": 1.3995,
+      "step": 4391
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1128336350172782e-05,
+      "loss": 1.4849,
+      "step": 4392
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1128115115279834e-05,
+      "loss": 1.6035,
+      "step": 4393
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1127893880386887e-05,
+      "loss": 1.3401,
+      "step": 4394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1127672645493938e-05,
+      "loss": 1.466,
+      "step": 4395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.112745141060099e-05,
+      "loss": 1.405,
+      "step": 4396
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1127230175708041e-05,
+      "loss": 1.5094,
+      "step": 4397
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1127008940815094e-05,
+      "loss": 1.2851,
+      "step": 4398
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1126787705922147e-05,
+      "loss": 1.5051,
+      "step": 4399
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.11265664710292e-05,
+      "loss": 1.2918,
+      "step": 4400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1126345236136252e-05,
+      "loss": 1.3371,
+      "step": 4401
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1126124001243303e-05,
+      "loss": 1.4472,
+      "step": 4402
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1125902766350354e-05,
+      "loss": 1.3942,
+      "step": 4403
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1125681531457406e-05,
+      "loss": 1.2562,
+      "step": 4404
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1125460296564459e-05,
+      "loss": 1.3538,
+      "step": 4405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1125239061671512e-05,
+      "loss": 1.4599,
+      "step": 4406
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1125017826778564e-05,
+      "loss": 1.3981,
+      "step": 4407
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1124796591885615e-05,
+      "loss": 1.4858,
+      "step": 4408
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1124575356992668e-05,
+      "loss": 1.3837,
+      "step": 4409
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.112435412209972e-05,
+      "loss": 1.5269,
+      "step": 4410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1124132887206771e-05,
+      "loss": 1.4837,
+      "step": 4411
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1123911652313824e-05,
+      "loss": 1.4092,
+      "step": 4412
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1123690417420876e-05,
+      "loss": 1.3615,
+      "step": 4413
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1123469182527927e-05,
+      "loss": 1.2654,
+      "step": 4414
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.112324794763498e-05,
+      "loss": 1.3582,
+      "step": 4415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1123026712742033e-05,
+      "loss": 1.5093,
+      "step": 4416
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1122805477849085e-05,
+      "loss": 1.4739,
+      "step": 4417
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1122584242956136e-05,
+      "loss": 1.4222,
+      "step": 4418
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1122363008063187e-05,
+      "loss": 1.4104,
+      "step": 4419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.112214177317024e-05,
+      "loss": 1.3922,
+      "step": 4420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1121920538277292e-05,
+      "loss": 1.3629,
+      "step": 4421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1121699303384345e-05,
+      "loss": 1.3286,
+      "step": 4422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1121478068491397e-05,
+      "loss": 1.3072,
+      "step": 4423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.112125683359845e-05,
+      "loss": 1.4552,
+      "step": 4424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1121035598705501e-05,
+      "loss": 1.4173,
+      "step": 4425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1120814363812554e-05,
+      "loss": 1.3908,
+      "step": 4426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1120593128919605e-05,
+      "loss": 1.2465,
+      "step": 4427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1120371894026657e-05,
+      "loss": 1.2643,
+      "step": 4428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.112015065913371e-05,
+      "loss": 1.2908,
+      "step": 4429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.111992942424076e-05,
+      "loss": 1.4795,
+      "step": 4430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1119708189347813e-05,
+      "loss": 1.3809,
+      "step": 4431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1119486954454866e-05,
+      "loss": 1.5668,
+      "step": 4432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1119265719561919e-05,
+      "loss": 1.4156,
+      "step": 4433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.111904448466897e-05,
+      "loss": 1.3092,
+      "step": 4434
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1118823249776022e-05,
+      "loss": 1.429,
+      "step": 4435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1118602014883073e-05,
+      "loss": 1.4463,
+      "step": 4436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1118380779990126e-05,
+      "loss": 1.4073,
+      "step": 4437
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1118159545097178e-05,
+      "loss": 1.4481,
+      "step": 4438
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.111793831020423e-05,
+      "loss": 1.4331,
+      "step": 4439
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1117717075311283e-05,
+      "loss": 1.3757,
+      "step": 4440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1117495840418334e-05,
+      "loss": 1.4242,
+      "step": 4441
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1117274605525387e-05,
+      "loss": 1.3596,
+      "step": 4442
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1117053370632438e-05,
+      "loss": 1.3614,
+      "step": 4443
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.111683213573949e-05,
+      "loss": 1.3013,
+      "step": 4444
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1116610900846543e-05,
+      "loss": 1.3931,
+      "step": 4445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1116389665953596e-05,
+      "loss": 1.4475,
+      "step": 4446
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1116168431060647e-05,
+      "loss": 1.403,
+      "step": 4447
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.11159471961677e-05,
+      "loss": 1.3387,
+      "step": 4448
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1115725961274752e-05,
+      "loss": 1.2905,
+      "step": 4449
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1115504726381803e-05,
+      "loss": 1.2293,
+      "step": 4450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1115283491488855e-05,
+      "loss": 1.3707,
+      "step": 4451
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1115062256595906e-05,
+      "loss": 1.397,
+      "step": 4452
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1114841021702959e-05,
+      "loss": 1.4005,
+      "step": 4453
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1114619786810012e-05,
+      "loss": 1.502,
+      "step": 4454
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1114398551917064e-05,
+      "loss": 1.4845,
+      "step": 4455
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1114177317024117e-05,
+      "loss": 1.4876,
+      "step": 4456
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1113956082131168e-05,
+      "loss": 1.3434,
+      "step": 4457
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.111373484723822e-05,
+      "loss": 1.4665,
+      "step": 4458
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1113513612345271e-05,
+      "loss": 1.2652,
+      "step": 4459
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1113292377452324e-05,
+      "loss": 1.4455,
+      "step": 4460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1113071142559376e-05,
+      "loss": 1.368,
+      "step": 4461
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1112849907666429e-05,
+      "loss": 1.39,
+      "step": 4462
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.111262867277348e-05,
+      "loss": 1.4307,
+      "step": 4463
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1112407437880533e-05,
+      "loss": 1.4275,
+      "step": 4464
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1112186202987585e-05,
+      "loss": 1.4006,
+      "step": 4465
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1111964968094638e-05,
+      "loss": 1.2823,
+      "step": 4466
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1111743733201689e-05,
+      "loss": 1.3567,
+      "step": 4467
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.111152249830874e-05,
+      "loss": 1.438,
+      "step": 4468
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1111301263415792e-05,
+      "loss": 1.3285,
+      "step": 4469
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1111080028522845e-05,
+      "loss": 1.3769,
+      "step": 4470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1110858793629897e-05,
+      "loss": 1.3641,
+      "step": 4471
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.111063755873695e-05,
+      "loss": 1.5064,
+      "step": 4472
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1110416323844003e-05,
+      "loss": 1.2836,
+      "step": 4473
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1110195088951054e-05,
+      "loss": 1.4299,
+      "step": 4474
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1109973854058105e-05,
+      "loss": 1.3476,
+      "step": 4475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1109752619165157e-05,
+      "loss": 1.4331,
+      "step": 4476
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.110953138427221e-05,
+      "loss": 1.331,
+      "step": 4477
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1109310149379262e-05,
+      "loss": 1.3541,
+      "step": 4478
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1109088914486313e-05,
+      "loss": 1.3347,
+      "step": 4479
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1108867679593366e-05,
+      "loss": 1.29,
+      "step": 4480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1108646444700418e-05,
+      "loss": 1.5128,
+      "step": 4481
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1108425209807471e-05,
+      "loss": 1.5023,
+      "step": 4482
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1108203974914522e-05,
+      "loss": 1.3099,
+      "step": 4483
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1107982740021575e-05,
+      "loss": 1.2707,
+      "step": 4484
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1107761505128626e-05,
+      "loss": 1.4638,
+      "step": 4485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1107540270235678e-05,
+      "loss": 1.3374,
+      "step": 4486
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.110731903534273e-05,
+      "loss": 1.4546,
+      "step": 4487
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1107097800449783e-05,
+      "loss": 1.4472,
+      "step": 4488
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1106876565556836e-05,
+      "loss": 1.4508,
+      "step": 4489
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1106655330663887e-05,
+      "loss": 1.5705,
+      "step": 4490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1106434095770938e-05,
+      "loss": 1.3452,
+      "step": 4491
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.110621286087799e-05,
+      "loss": 1.4501,
+      "step": 4492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1105991625985043e-05,
+      "loss": 1.4265,
+      "step": 4493
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1105770391092096e-05,
+      "loss": 1.3019,
+      "step": 4494
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1105549156199148e-05,
+      "loss": 1.4759,
+      "step": 4495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.11053279213062e-05,
+      "loss": 1.5065,
+      "step": 4496
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1105106686413252e-05,
+      "loss": 1.3193,
+      "step": 4497
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1104885451520304e-05,
+      "loss": 1.5099,
+      "step": 4498
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1104664216627355e-05,
+      "loss": 1.3577,
+      "step": 4499
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1104442981734408e-05,
+      "loss": 1.265,
+      "step": 4500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1104221746841459e-05,
+      "loss": 1.3774,
+      "step": 4501
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1104000511948511e-05,
+      "loss": 1.3625,
+      "step": 4502
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1103779277055564e-05,
+      "loss": 1.3166,
+      "step": 4503
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1103558042162617e-05,
+      "loss": 1.3309,
+      "step": 4504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.110333680726967e-05,
+      "loss": 1.5809,
+      "step": 4505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.110311557237672e-05,
+      "loss": 1.3877,
+      "step": 4506
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1102894337483771e-05,
+      "loss": 1.4319,
+      "step": 4507
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1102673102590824e-05,
+      "loss": 1.373,
+      "step": 4508
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1102451867697876e-05,
+      "loss": 1.2999,
+      "step": 4509
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1102230632804929e-05,
+      "loss": 1.3191,
+      "step": 4510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1102009397911982e-05,
+      "loss": 1.3995,
+      "step": 4511
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1101788163019033e-05,
+      "loss": 1.4036,
+      "step": 4512
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1101566928126085e-05,
+      "loss": 1.3,
+      "step": 4513
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1101345693233138e-05,
+      "loss": 1.4735,
+      "step": 4514
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1101124458340189e-05,
+      "loss": 1.4805,
+      "step": 4515
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1100903223447241e-05,
+      "loss": 1.4041,
+      "step": 4516
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1100681988554294e-05,
+      "loss": 1.3253,
+      "step": 4517
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1100460753661345e-05,
+      "loss": 1.3148,
+      "step": 4518
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1100239518768397e-05,
+      "loss": 1.2615,
+      "step": 4519
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.110001828387545e-05,
+      "loss": 1.3686,
+      "step": 4520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1099797048982503e-05,
+      "loss": 1.3411,
+      "step": 4521
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1099575814089554e-05,
+      "loss": 1.2151,
+      "step": 4522
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1099354579196606e-05,
+      "loss": 1.5481,
+      "step": 4523
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1099133344303657e-05,
+      "loss": 1.4177,
+      "step": 4524
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.109891210941071e-05,
+      "loss": 1.3428,
+      "step": 4525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1098690874517762e-05,
+      "loss": 1.3678,
+      "step": 4526
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1098469639624815e-05,
+      "loss": 1.4505,
+      "step": 4527
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1098248404731868e-05,
+      "loss": 1.2939,
+      "step": 4528
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1098027169838918e-05,
+      "loss": 1.4014,
+      "step": 4529
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1097805934945971e-05,
+      "loss": 1.4672,
+      "step": 4530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1097584700053022e-05,
+      "loss": 1.4556,
+      "step": 4531
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1097363465160075e-05,
+      "loss": 1.407,
+      "step": 4532
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1097142230267127e-05,
+      "loss": 1.3404,
+      "step": 4533
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1096920995374178e-05,
+      "loss": 1.3857,
+      "step": 4534
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.109669976048123e-05,
+      "loss": 1.4213,
+      "step": 4535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1096478525588283e-05,
+      "loss": 1.4351,
+      "step": 4536
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1096257290695336e-05,
+      "loss": 1.4448,
+      "step": 4537
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1096036055802387e-05,
+      "loss": 1.5246,
+      "step": 4538
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.109581482090944e-05,
+      "loss": 1.1662,
+      "step": 4539
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.109559358601649e-05,
+      "loss": 1.4887,
+      "step": 4540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1095372351123543e-05,
+      "loss": 1.4498,
+      "step": 4541
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1095151116230596e-05,
+      "loss": 1.2694,
+      "step": 4542
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1094929881337648e-05,
+      "loss": 1.3882,
+      "step": 4543
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.10947086464447e-05,
+      "loss": 1.3511,
+      "step": 4544
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1094487411551752e-05,
+      "loss": 1.3149,
+      "step": 4545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1094266176658804e-05,
+      "loss": 1.4608,
+      "step": 4546
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1094044941765855e-05,
+      "loss": 1.4847,
+      "step": 4547
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1093823706872908e-05,
+      "loss": 1.3767,
+      "step": 4548
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.109360247197996e-05,
+      "loss": 1.3478,
+      "step": 4549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1093381237087011e-05,
+      "loss": 1.4521,
+      "step": 4550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1093160002194064e-05,
+      "loss": 1.3323,
+      "step": 4551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1092938767301117e-05,
+      "loss": 1.5774,
+      "step": 4552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.109271753240817e-05,
+      "loss": 1.235,
+      "step": 4553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1092496297515222e-05,
+      "loss": 1.2114,
+      "step": 4554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1092275062622273e-05,
+      "loss": 1.5106,
+      "step": 4555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1092053827729324e-05,
+      "loss": 1.3547,
+      "step": 4556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1091832592836376e-05,
+      "loss": 1.2712,
+      "step": 4557
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1091611357943429e-05,
+      "loss": 1.3083,
+      "step": 4558
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1091390123050482e-05,
+      "loss": 1.4732,
+      "step": 4559
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1091168888157534e-05,
+      "loss": 1.4236,
+      "step": 4560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1090947653264585e-05,
+      "loss": 1.374,
+      "step": 4561
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1090726418371638e-05,
+      "loss": 1.3518,
+      "step": 4562
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1090505183478689e-05,
+      "loss": 1.4699,
+      "step": 4563
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1090283948585741e-05,
+      "loss": 1.3184,
+      "step": 4564
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1090062713692794e-05,
+      "loss": 1.3827,
+      "step": 4565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1089841478799846e-05,
+      "loss": 1.4077,
+      "step": 4566
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1089620243906897e-05,
+      "loss": 1.5258,
+      "step": 4567
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.108939900901395e-05,
+      "loss": 1.3311,
+      "step": 4568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1089177774121003e-05,
+      "loss": 1.3131,
+      "step": 4569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1088956539228055e-05,
+      "loss": 1.3196,
+      "step": 4570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1088735304335106e-05,
+      "loss": 1.3108,
+      "step": 4571
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1088514069442157e-05,
+      "loss": 1.4316,
+      "step": 4572
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.108829283454921e-05,
+      "loss": 1.3406,
+      "step": 4573
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1088071599656262e-05,
+      "loss": 1.521,
+      "step": 4574
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1087850364763315e-05,
+      "loss": 1.2582,
+      "step": 4575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1087629129870367e-05,
+      "loss": 1.4691,
+      "step": 4576
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.108740789497742e-05,
+      "loss": 1.4737,
+      "step": 4577
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1087186660084471e-05,
+      "loss": 1.3842,
+      "step": 4578
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1086965425191522e-05,
+      "loss": 1.4706,
+      "step": 4579
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1086744190298575e-05,
+      "loss": 1.2373,
+      "step": 4580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1086522955405627e-05,
+      "loss": 1.5311,
+      "step": 4581
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.108630172051268e-05,
+      "loss": 1.2884,
+      "step": 4582
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.108608048561973e-05,
+      "loss": 1.4002,
+      "step": 4583
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1085859250726783e-05,
+      "loss": 1.3739,
+      "step": 4584
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1085638015833836e-05,
+      "loss": 1.3985,
+      "step": 4585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1085416780940889e-05,
+      "loss": 1.397,
+      "step": 4586
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.108519554604794e-05,
+      "loss": 1.4659,
+      "step": 4587
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1084974311154992e-05,
+      "loss": 1.3673,
+      "step": 4588
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1084753076262043e-05,
+      "loss": 1.4308,
+      "step": 4589
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1084531841369096e-05,
+      "loss": 1.3499,
+      "step": 4590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1084310606476148e-05,
+      "loss": 1.4172,
+      "step": 4591
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.10840893715832e-05,
+      "loss": 1.5417,
+      "step": 4592
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1083868136690253e-05,
+      "loss": 1.4346,
+      "step": 4593
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1083646901797304e-05,
+      "loss": 1.4659,
+      "step": 4594
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1083425666904355e-05,
+      "loss": 1.5187,
+      "step": 4595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1083204432011408e-05,
+      "loss": 1.2429,
+      "step": 4596
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.108298319711846e-05,
+      "loss": 1.3623,
+      "step": 4597
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1082761962225513e-05,
+      "loss": 1.3397,
+      "step": 4598
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1082540727332566e-05,
+      "loss": 1.4201,
+      "step": 4599
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1082319492439617e-05,
+      "loss": 1.5183,
+      "step": 4600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.108209825754667e-05,
+      "loss": 1.4751,
+      "step": 4601
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1081877022653722e-05,
+      "loss": 1.3637,
+      "step": 4602
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1081655787760773e-05,
+      "loss": 1.31,
+      "step": 4603
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1081434552867825e-05,
+      "loss": 1.3375,
+      "step": 4604
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1081213317974876e-05,
+      "loss": 1.3179,
+      "step": 4605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1080992083081929e-05,
+      "loss": 1.3481,
+      "step": 4606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1080770848188982e-05,
+      "loss": 1.5016,
+      "step": 4607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1080549613296034e-05,
+      "loss": 1.2497,
+      "step": 4608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1080328378403087e-05,
+      "loss": 1.5359,
+      "step": 4609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1080107143510138e-05,
+      "loss": 1.174,
+      "step": 4610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.107988590861719e-05,
+      "loss": 1.4473,
+      "step": 4611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1079664673724241e-05,
+      "loss": 1.3961,
+      "step": 4612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1079443438831294e-05,
+      "loss": 1.4967,
+      "step": 4613
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1079222203938346e-05,
+      "loss": 1.3985,
+      "step": 4614
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1079000969045399e-05,
+      "loss": 1.2412,
+      "step": 4615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.107877973415245e-05,
+      "loss": 1.3482,
+      "step": 4616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1078558499259503e-05,
+      "loss": 1.6358,
+      "step": 4617
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1078337264366555e-05,
+      "loss": 1.4611,
+      "step": 4618
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1078116029473606e-05,
+      "loss": 1.3995,
+      "step": 4619
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1077894794580659e-05,
+      "loss": 1.3386,
+      "step": 4620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1077673559687711e-05,
+      "loss": 1.3853,
+      "step": 4621
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1077452324794762e-05,
+      "loss": 1.5107,
+      "step": 4622
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1077231089901815e-05,
+      "loss": 1.4653,
+      "step": 4623
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1077009855008867e-05,
+      "loss": 1.4366,
+      "step": 4624
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.107678862011592e-05,
+      "loss": 1.3598,
+      "step": 4625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1076567385222971e-05,
+      "loss": 1.4156,
+      "step": 4626
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1076346150330024e-05,
+      "loss": 1.4408,
+      "step": 4627
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1076124915437074e-05,
+      "loss": 1.2877,
+      "step": 4628
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1075903680544127e-05,
+      "loss": 1.2864,
+      "step": 4629
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.107568244565118e-05,
+      "loss": 1.3332,
+      "step": 4630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1075461210758232e-05,
+      "loss": 1.2942,
+      "step": 4631
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1075239975865285e-05,
+      "loss": 1.3869,
+      "step": 4632
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1075018740972336e-05,
+      "loss": 1.4597,
+      "step": 4633
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1074797506079388e-05,
+      "loss": 1.2735,
+      "step": 4634
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.107457627118644e-05,
+      "loss": 1.4284,
+      "step": 4635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1074355036293492e-05,
+      "loss": 1.4796,
+      "step": 4636
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1074133801400545e-05,
+      "loss": 1.3943,
+      "step": 4637
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1073912566507596e-05,
+      "loss": 1.3845,
+      "step": 4638
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1073691331614648e-05,
+      "loss": 1.5446,
+      "step": 4639
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.10734700967217e-05,
+      "loss": 1.4626,
+      "step": 4640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1073248861828753e-05,
+      "loss": 1.3826,
+      "step": 4641
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1073027626935806e-05,
+      "loss": 1.3735,
+      "step": 4642
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1072806392042857e-05,
+      "loss": 1.5104,
+      "step": 4643
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1072585157149908e-05,
+      "loss": 1.382,
+      "step": 4644
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.107236392225696e-05,
+      "loss": 1.358,
+      "step": 4645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1072142687364013e-05,
+      "loss": 1.3983,
+      "step": 4646
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1071921452471066e-05,
+      "loss": 1.4179,
+      "step": 4647
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1071700217578118e-05,
+      "loss": 1.4103,
+      "step": 4648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.107147898268517e-05,
+      "loss": 1.3693,
+      "step": 4649
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1071257747792222e-05,
+      "loss": 1.2719,
+      "step": 4650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1071036512899273e-05,
+      "loss": 1.5456,
+      "step": 4651
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1070815278006325e-05,
+      "loss": 1.4044,
+      "step": 4652
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1070594043113378e-05,
+      "loss": 1.3863,
+      "step": 4653
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1070372808220429e-05,
+      "loss": 1.3983,
+      "step": 4654
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1070151573327481e-05,
+      "loss": 1.3433,
+      "step": 4655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1069930338434534e-05,
+      "loss": 1.2841,
+      "step": 4656
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1069709103541587e-05,
+      "loss": 1.3013,
+      "step": 4657
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.106948786864864e-05,
+      "loss": 1.4015,
+      "step": 4658
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.106926663375569e-05,
+      "loss": 1.388,
+      "step": 4659
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1069045398862741e-05,
+      "loss": 1.4598,
+      "step": 4660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1068824163969794e-05,
+      "loss": 1.4058,
+      "step": 4661
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1068602929076846e-05,
+      "loss": 1.2923,
+      "step": 4662
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1068381694183899e-05,
+      "loss": 1.4024,
+      "step": 4663
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1068160459290952e-05,
+      "loss": 1.3809,
+      "step": 4664
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1067939224398002e-05,
+      "loss": 1.3691,
+      "step": 4665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1067717989505055e-05,
+      "loss": 1.4251,
+      "step": 4666
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1067496754612106e-05,
+      "loss": 1.3822,
+      "step": 4667
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1067275519719159e-05,
+      "loss": 1.4712,
+      "step": 4668
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1067054284826211e-05,
+      "loss": 1.3449,
+      "step": 4669
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1066833049933264e-05,
+      "loss": 1.2929,
+      "step": 4670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1066611815040315e-05,
+      "loss": 1.4354,
+      "step": 4671
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1066390580147367e-05,
+      "loss": 1.6539,
+      "step": 4672
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.106616934525442e-05,
+      "loss": 1.3376,
+      "step": 4673
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1065948110361473e-05,
+      "loss": 1.4773,
+      "step": 4674
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1065726875468524e-05,
+      "loss": 1.2924,
+      "step": 4675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1065505640575574e-05,
+      "loss": 1.4026,
+      "step": 4676
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1065284405682627e-05,
+      "loss": 1.4045,
+      "step": 4677
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.106506317078968e-05,
+      "loss": 1.3351,
+      "step": 4678
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1064841935896732e-05,
+      "loss": 1.5001,
+      "step": 4679
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1064620701003785e-05,
+      "loss": 1.5866,
+      "step": 4680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1064399466110838e-05,
+      "loss": 1.3144,
+      "step": 4681
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1064178231217888e-05,
+      "loss": 1.4939,
+      "step": 4682
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.106395699632494e-05,
+      "loss": 1.3722,
+      "step": 4683
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1063735761431992e-05,
+      "loss": 1.3299,
+      "step": 4684
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1063514526539045e-05,
+      "loss": 1.2673,
+      "step": 4685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1063293291646097e-05,
+      "loss": 1.2903,
+      "step": 4686
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1063072056753148e-05,
+      "loss": 1.3572,
+      "step": 4687
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.10628508218602e-05,
+      "loss": 1.3271,
+      "step": 4688
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1062629586967253e-05,
+      "loss": 1.197,
+      "step": 4689
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1062408352074306e-05,
+      "loss": 1.5509,
+      "step": 4690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1062187117181357e-05,
+      "loss": 1.4813,
+      "step": 4691
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.106196588228841e-05,
+      "loss": 1.328,
+      "step": 4692
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.106174464739546e-05,
+      "loss": 1.4152,
+      "step": 4693
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1061523412502513e-05,
+      "loss": 1.3787,
+      "step": 4694
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1061302177609566e-05,
+      "loss": 1.378,
+      "step": 4695
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1061080942716618e-05,
+      "loss": 1.5283,
+      "step": 4696
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.106085970782367e-05,
+      "loss": 1.3659,
+      "step": 4697
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1060638472930722e-05,
+      "loss": 1.6939,
+      "step": 4698
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1060417238037774e-05,
+      "loss": 1.3515,
+      "step": 4699
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1060196003144825e-05,
+      "loss": 1.3225,
+      "step": 4700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1059974768251878e-05,
+      "loss": 1.4128,
+      "step": 4701
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.105975353335893e-05,
+      "loss": 1.5218,
+      "step": 4702
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1059532298465983e-05,
+      "loss": 1.3727,
+      "step": 4703
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1059311063573034e-05,
+      "loss": 1.4773,
+      "step": 4704
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1059089828680087e-05,
+      "loss": 1.3797,
+      "step": 4705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.105886859378714e-05,
+      "loss": 1.2024,
+      "step": 4706
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.105864735889419e-05,
+      "loss": 1.5765,
+      "step": 4707
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1058426124001243e-05,
+      "loss": 1.3052,
+      "step": 4708
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1058204889108294e-05,
+      "loss": 1.2189,
+      "step": 4709
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1057983654215346e-05,
+      "loss": 1.2276,
+      "step": 4710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1057762419322399e-05,
+      "loss": 1.4093,
+      "step": 4711
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1057541184429452e-05,
+      "loss": 1.3416,
+      "step": 4712
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1057319949536504e-05,
+      "loss": 1.4362,
+      "step": 4713
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1057098714643555e-05,
+      "loss": 1.3898,
+      "step": 4714
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1056877479750608e-05,
+      "loss": 1.4598,
+      "step": 4715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1056656244857659e-05,
+      "loss": 1.3822,
+      "step": 4716
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1056435009964711e-05,
+      "loss": 1.535,
+      "step": 4717
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1056213775071764e-05,
+      "loss": 1.3029,
+      "step": 4718
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1055992540178816e-05,
+      "loss": 1.3109,
+      "step": 4719
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1055771305285867e-05,
+      "loss": 1.2637,
+      "step": 4720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.105555007039292e-05,
+      "loss": 1.4578,
+      "step": 4721
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1055328835499973e-05,
+      "loss": 1.4636,
+      "step": 4722
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1055107600607023e-05,
+      "loss": 1.2504,
+      "step": 4723
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1054886365714076e-05,
+      "loss": 1.5063,
+      "step": 4724
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1054665130821129e-05,
+      "loss": 1.435,
+      "step": 4725
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.105444389592818e-05,
+      "loss": 1.2944,
+      "step": 4726
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1054222661035232e-05,
+      "loss": 1.4813,
+      "step": 4727
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1054001426142285e-05,
+      "loss": 1.3279,
+      "step": 4728
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1053780191249337e-05,
+      "loss": 1.5004,
+      "step": 4729
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.105355895635639e-05,
+      "loss": 1.4805,
+      "step": 4730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1053337721463441e-05,
+      "loss": 1.2807,
+      "step": 4731
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1053116486570492e-05,
+      "loss": 1.2964,
+      "step": 4732
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1052895251677545e-05,
+      "loss": 1.3542,
+      "step": 4733
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1052674016784597e-05,
+      "loss": 1.3378,
+      "step": 4734
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.105245278189165e-05,
+      "loss": 1.3378,
+      "step": 4735
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1052231546998702e-05,
+      "loss": 1.36,
+      "step": 4736
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1052010312105753e-05,
+      "loss": 1.4279,
+      "step": 4737
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1051789077212806e-05,
+      "loss": 1.4162,
+      "step": 4738
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1051567842319857e-05,
+      "loss": 1.4267,
+      "step": 4739
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.105134660742691e-05,
+      "loss": 1.3146,
+      "step": 4740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1051125372533962e-05,
+      "loss": 1.5431,
+      "step": 4741
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1050904137641013e-05,
+      "loss": 1.3161,
+      "step": 4742
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1050682902748066e-05,
+      "loss": 1.381,
+      "step": 4743
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1050461667855118e-05,
+      "loss": 1.257,
+      "step": 4744
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.105024043296217e-05,
+      "loss": 1.3355,
+      "step": 4745
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1050019198069223e-05,
+      "loss": 1.5393,
+      "step": 4746
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1049797963176274e-05,
+      "loss": 1.3432,
+      "step": 4747
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1049576728283325e-05,
+      "loss": 1.3024,
+      "step": 4748
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1049355493390378e-05,
+      "loss": 1.3439,
+      "step": 4749
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.104913425849743e-05,
+      "loss": 1.6097,
+      "step": 4750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1048913023604483e-05,
+      "loss": 1.2205,
+      "step": 4751
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1048691788711536e-05,
+      "loss": 1.5105,
+      "step": 4752
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1048470553818587e-05,
+      "loss": 1.4641,
+      "step": 4753
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.104824931892564e-05,
+      "loss": 1.4,
+      "step": 4754
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.104802808403269e-05,
+      "loss": 1.23,
+      "step": 4755
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1047806849139743e-05,
+      "loss": 1.2878,
+      "step": 4756
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1047585614246795e-05,
+      "loss": 1.4054,
+      "step": 4757
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1047364379353846e-05,
+      "loss": 1.4672,
+      "step": 4758
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1047143144460899e-05,
+      "loss": 1.5564,
+      "step": 4759
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1046921909567951e-05,
+      "loss": 1.5946,
+      "step": 4760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1046700674675004e-05,
+      "loss": 1.46,
+      "step": 4761
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1046479439782057e-05,
+      "loss": 1.3778,
+      "step": 4762
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1046258204889108e-05,
+      "loss": 1.2696,
+      "step": 4763
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1046036969996159e-05,
+      "loss": 1.3844,
+      "step": 4764
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1045815735103211e-05,
+      "loss": 1.4226,
+      "step": 4765
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1045594500210264e-05,
+      "loss": 1.3797,
+      "step": 4766
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1045373265317316e-05,
+      "loss": 1.3564,
+      "step": 4767
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1045152030424369e-05,
+      "loss": 1.2563,
+      "step": 4768
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.104493079553142e-05,
+      "loss": 1.4714,
+      "step": 4769
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1044709560638473e-05,
+      "loss": 1.4128,
+      "step": 4770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1044488325745523e-05,
+      "loss": 1.5056,
+      "step": 4771
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1044267090852576e-05,
+      "loss": 1.2855,
+      "step": 4772
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1044045855959629e-05,
+      "loss": 1.2752,
+      "step": 4773
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1043824621066681e-05,
+      "loss": 1.2093,
+      "step": 4774
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1043603386173732e-05,
+      "loss": 1.3554,
+      "step": 4775
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1043382151280785e-05,
+      "loss": 1.4046,
+      "step": 4776
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1043160916387837e-05,
+      "loss": 1.2857,
+      "step": 4777
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.104293968149489e-05,
+      "loss": 1.3346,
+      "step": 4778
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1042718446601941e-05,
+      "loss": 1.3603,
+      "step": 4779
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1042497211708992e-05,
+      "loss": 1.3295,
+      "step": 4780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1042275976816044e-05,
+      "loss": 1.4537,
+      "step": 4781
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1042054741923097e-05,
+      "loss": 1.3048,
+      "step": 4782
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.104183350703015e-05,
+      "loss": 1.3069,
+      "step": 4783
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1041612272137202e-05,
+      "loss": 1.2536,
+      "step": 4784
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1041391037244255e-05,
+      "loss": 1.4001,
+      "step": 4785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1041169802351306e-05,
+      "loss": 1.3409,
+      "step": 4786
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1040948567458358e-05,
+      "loss": 1.2926,
+      "step": 4787
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.104072733256541e-05,
+      "loss": 1.3005,
+      "step": 4788
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1040506097672462e-05,
+      "loss": 1.3508,
+      "step": 4789
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1040284862779515e-05,
+      "loss": 1.2985,
+      "step": 4790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1040063627886566e-05,
+      "loss": 1.3035,
+      "step": 4791
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1039842392993618e-05,
+      "loss": 1.4386,
+      "step": 4792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.103962115810067e-05,
+      "loss": 1.3392,
+      "step": 4793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1039399923207723e-05,
+      "loss": 1.4055,
+      "step": 4794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1039178688314774e-05,
+      "loss": 1.3119,
+      "step": 4795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1038957453421827e-05,
+      "loss": 1.4296,
+      "step": 4796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1038736218528878e-05,
+      "loss": 1.4885,
+      "step": 4797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.103851498363593e-05,
+      "loss": 1.283,
+      "step": 4798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1038293748742983e-05,
+      "loss": 1.4447,
+      "step": 4799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1038072513850036e-05,
+      "loss": 1.3086,
+      "step": 4800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1037851278957088e-05,
+      "loss": 1.3621,
+      "step": 4801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.103763004406414e-05,
+      "loss": 1.2981,
+      "step": 4802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1037408809171192e-05,
+      "loss": 1.4989,
+      "step": 4803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1037187574278243e-05,
+      "loss": 1.3058,
+      "step": 4804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1036966339385295e-05,
+      "loss": 1.3629,
+      "step": 4805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1036745104492348e-05,
+      "loss": 1.4038,
+      "step": 4806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.10365238695994e-05,
+      "loss": 1.4005,
+      "step": 4807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1036302634706451e-05,
+      "loss": 1.417,
+      "step": 4808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1036081399813504e-05,
+      "loss": 1.4789,
+      "step": 4809
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1035860164920557e-05,
+      "loss": 1.2411,
+      "step": 4810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1035638930027608e-05,
+      "loss": 1.4836,
+      "step": 4811
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.103541769513466e-05,
+      "loss": 1.3214,
+      "step": 4812
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1035196460241711e-05,
+      "loss": 1.4219,
+      "step": 4813
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1034975225348764e-05,
+      "loss": 1.2421,
+      "step": 4814
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1034753990455816e-05,
+      "loss": 1.5072,
+      "step": 4815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1034532755562869e-05,
+      "loss": 1.3812,
+      "step": 4816
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1034311520669922e-05,
+      "loss": 1.456,
+      "step": 4817
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1034090285776972e-05,
+      "loss": 1.3743,
+      "step": 4818
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1033869050884025e-05,
+      "loss": 1.3329,
+      "step": 4819
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1033647815991076e-05,
+      "loss": 1.4167,
+      "step": 4820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1033426581098129e-05,
+      "loss": 1.3712,
+      "step": 4821
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1033205346205181e-05,
+      "loss": 1.2654,
+      "step": 4822
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1032984111312234e-05,
+      "loss": 1.4277,
+      "step": 4823
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1032762876419285e-05,
+      "loss": 1.4721,
+      "step": 4824
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1032541641526337e-05,
+      "loss": 1.2522,
+      "step": 4825
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.103232040663339e-05,
+      "loss": 1.3876,
+      "step": 4826
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1032099171740441e-05,
+      "loss": 1.4176,
+      "step": 4827
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1031877936847494e-05,
+      "loss": 1.4715,
+      "step": 4828
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1031656701954546e-05,
+      "loss": 1.4406,
+      "step": 4829
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1031435467061597e-05,
+      "loss": 1.371,
+      "step": 4830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.103121423216865e-05,
+      "loss": 1.3975,
+      "step": 4831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1030992997275702e-05,
+      "loss": 1.3772,
+      "step": 4832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1030771762382755e-05,
+      "loss": 1.3801,
+      "step": 4833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1030550527489807e-05,
+      "loss": 1.4162,
+      "step": 4834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1030329292596858e-05,
+      "loss": 1.3818,
+      "step": 4835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.103010805770391e-05,
+      "loss": 1.4624,
+      "step": 4836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1029886822810962e-05,
+      "loss": 1.3781,
+      "step": 4837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1029665587918015e-05,
+      "loss": 1.3841,
+      "step": 4838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1029444353025067e-05,
+      "loss": 1.3944,
+      "step": 4839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.102922311813212e-05,
+      "loss": 1.4625,
+      "step": 4840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.102900188323917e-05,
+      "loss": 1.4026,
+      "step": 4841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1028780648346223e-05,
+      "loss": 1.3628,
+      "step": 4842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1028559413453274e-05,
+      "loss": 1.3921,
+      "step": 4843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1028338178560327e-05,
+      "loss": 1.4268,
+      "step": 4844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.102811694366738e-05,
+      "loss": 1.3017,
+      "step": 4845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.102789570877443e-05,
+      "loss": 1.3399,
+      "step": 4846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1027674473881483e-05,
+      "loss": 1.3313,
+      "step": 4847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1027453238988536e-05,
+      "loss": 1.2722,
+      "step": 4848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1027232004095588e-05,
+      "loss": 1.2513,
+      "step": 4849
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.102701076920264e-05,
+      "loss": 1.3151,
+      "step": 4850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1026789534309692e-05,
+      "loss": 1.2516,
+      "step": 4851
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1026568299416743e-05,
+      "loss": 1.3185,
+      "step": 4852
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1026347064523795e-05,
+      "loss": 1.3017,
+      "step": 4853
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1026125829630848e-05,
+      "loss": 1.2839,
+      "step": 4854
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.10259045947379e-05,
+      "loss": 1.3342,
+      "step": 4855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1025683359844953e-05,
+      "loss": 1.472,
+      "step": 4856
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1025462124952004e-05,
+      "loss": 1.3417,
+      "step": 4857
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1025240890059057e-05,
+      "loss": 1.3579,
+      "step": 4858
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1025019655166108e-05,
+      "loss": 1.3571,
+      "step": 4859
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.102479842027316e-05,
+      "loss": 1.3076,
+      "step": 4860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1024577185380213e-05,
+      "loss": 1.3459,
+      "step": 4861
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1024355950487264e-05,
+      "loss": 1.5001,
+      "step": 4862
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1024134715594316e-05,
+      "loss": 1.4328,
+      "step": 4863
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1023913480701369e-05,
+      "loss": 1.5889,
+      "step": 4864
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1023692245808422e-05,
+      "loss": 1.4321,
+      "step": 4865
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1023471010915474e-05,
+      "loss": 1.4596,
+      "step": 4866
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1023249776022525e-05,
+      "loss": 1.4231,
+      "step": 4867
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1023028541129576e-05,
+      "loss": 1.513,
+      "step": 4868
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1022807306236629e-05,
+      "loss": 1.1864,
+      "step": 4869
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1022586071343681e-05,
+      "loss": 1.4711,
+      "step": 4870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1022364836450734e-05,
+      "loss": 1.2894,
+      "step": 4871
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1022143601557786e-05,
+      "loss": 1.4118,
+      "step": 4872
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1021922366664837e-05,
+      "loss": 1.2763,
+      "step": 4873
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.102170113177189e-05,
+      "loss": 1.4393,
+      "step": 4874
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1021479896878943e-05,
+      "loss": 1.4235,
+      "step": 4875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1021258661985993e-05,
+      "loss": 1.4357,
+      "step": 4876
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1021037427093046e-05,
+      "loss": 1.2192,
+      "step": 4877
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1020816192200099e-05,
+      "loss": 1.3881,
+      "step": 4878
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.102059495730715e-05,
+      "loss": 1.3561,
+      "step": 4879
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1020373722414202e-05,
+      "loss": 1.4761,
+      "step": 4880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1020152487521255e-05,
+      "loss": 1.515,
+      "step": 4881
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1019931252628307e-05,
+      "loss": 1.3585,
+      "step": 4882
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1019710017735358e-05,
+      "loss": 1.4741,
+      "step": 4883
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.101948878284241e-05,
+      "loss": 1.3572,
+      "step": 4884
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1019267547949462e-05,
+      "loss": 1.3719,
+      "step": 4885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1019046313056515e-05,
+      "loss": 1.3662,
+      "step": 4886
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1018825078163567e-05,
+      "loss": 1.3575,
+      "step": 4887
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.101860384327062e-05,
+      "loss": 1.3961,
+      "step": 4888
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1018382608377672e-05,
+      "loss": 1.362,
+      "step": 4889
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1018161373484723e-05,
+      "loss": 1.4119,
+      "step": 4890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1017940138591776e-05,
+      "loss": 1.3972,
+      "step": 4891
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1017718903698827e-05,
+      "loss": 1.3766,
+      "step": 4892
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.101749766880588e-05,
+      "loss": 1.4099,
+      "step": 4893
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1017276433912932e-05,
+      "loss": 1.2679,
+      "step": 4894
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1017055199019983e-05,
+      "loss": 1.3495,
+      "step": 4895
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1016833964127036e-05,
+      "loss": 1.3191,
+      "step": 4896
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1016612729234088e-05,
+      "loss": 1.3514,
+      "step": 4897
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.101639149434114e-05,
+      "loss": 1.23,
+      "step": 4898
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1016170259448192e-05,
+      "loss": 1.3612,
+      "step": 4899
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1015949024555244e-05,
+      "loss": 1.4296,
+      "step": 4900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1015727789662295e-05,
+      "loss": 1.5718,
+      "step": 4901
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1015506554769348e-05,
+      "loss": 1.253,
+      "step": 4902
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.10152853198764e-05,
+      "loss": 1.3482,
+      "step": 4903
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1015064084983453e-05,
+      "loss": 1.3843,
+      "step": 4904
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1014842850090506e-05,
+      "loss": 1.4189,
+      "step": 4905
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1014621615197557e-05,
+      "loss": 1.3764,
+      "step": 4906
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.101440038030461e-05,
+      "loss": 1.3731,
+      "step": 4907
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.101417914541166e-05,
+      "loss": 1.4388,
+      "step": 4908
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1013957910518713e-05,
+      "loss": 1.3455,
+      "step": 4909
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1013736675625765e-05,
+      "loss": 1.3792,
+      "step": 4910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1013515440732818e-05,
+      "loss": 1.3551,
+      "step": 4911
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1013294205839869e-05,
+      "loss": 1.3853,
+      "step": 4912
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1013072970946921e-05,
+      "loss": 1.3347,
+      "step": 4913
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1012851736053974e-05,
+      "loss": 1.4612,
+      "step": 4914
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1012630501161025e-05,
+      "loss": 1.3965,
+      "step": 4915
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1012409266268078e-05,
+      "loss": 1.3887,
+      "step": 4916
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1012188031375129e-05,
+      "loss": 1.4067,
+      "step": 4917
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1011966796482181e-05,
+      "loss": 1.47,
+      "step": 4918
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1011745561589234e-05,
+      "loss": 1.4104,
+      "step": 4919
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1011524326696286e-05,
+      "loss": 1.304,
+      "step": 4920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1011303091803339e-05,
+      "loss": 1.3713,
+      "step": 4921
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1011081856910392e-05,
+      "loss": 1.3831,
+      "step": 4922
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1010860622017443e-05,
+      "loss": 1.4874,
+      "step": 4923
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1010639387124493e-05,
+      "loss": 1.2731,
+      "step": 4924
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1010418152231546e-05,
+      "loss": 1.2541,
+      "step": 4925
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1010196917338599e-05,
+      "loss": 1.4002,
+      "step": 4926
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1009975682445651e-05,
+      "loss": 1.4246,
+      "step": 4927
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1009754447552702e-05,
+      "loss": 1.3715,
+      "step": 4928
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1009533212659755e-05,
+      "loss": 1.2832,
+      "step": 4929
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1009311977766807e-05,
+      "loss": 1.3856,
+      "step": 4930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1009090742873858e-05,
+      "loss": 1.4174,
+      "step": 4931
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1008869507980911e-05,
+      "loss": 1.3525,
+      "step": 4932
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1008648273087964e-05,
+      "loss": 1.4574,
+      "step": 4933
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1008427038195014e-05,
+      "loss": 1.4811,
+      "step": 4934
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1008205803302067e-05,
+      "loss": 1.4037,
+      "step": 4935
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.100798456840912e-05,
+      "loss": 1.4036,
+      "step": 4936
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1007763333516172e-05,
+      "loss": 1.318,
+      "step": 4937
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1007542098623225e-05,
+      "loss": 1.2977,
+      "step": 4938
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1007320863730276e-05,
+      "loss": 1.393,
+      "step": 4939
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1007099628837327e-05,
+      "loss": 1.4655,
+      "step": 4940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.100687839394438e-05,
+      "loss": 1.4413,
+      "step": 4941
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1006657159051432e-05,
+      "loss": 1.4469,
+      "step": 4942
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1006435924158485e-05,
+      "loss": 1.4039,
+      "step": 4943
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1006214689265536e-05,
+      "loss": 1.2865,
+      "step": 4944
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1005993454372588e-05,
+      "loss": 1.4292,
+      "step": 4945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.100577221947964e-05,
+      "loss": 1.4474,
+      "step": 4946
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1005550984586692e-05,
+      "loss": 1.542,
+      "step": 4947
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1005329749693744e-05,
+      "loss": 1.3601,
+      "step": 4948
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1005108514800797e-05,
+      "loss": 1.43,
+      "step": 4949
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1004887279907848e-05,
+      "loss": 1.269,
+      "step": 4950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.10046660450149e-05,
+      "loss": 1.361,
+      "step": 4951
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1004444810121953e-05,
+      "loss": 1.312,
+      "step": 4952
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1004223575229006e-05,
+      "loss": 1.4228,
+      "step": 4953
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1004002340336058e-05,
+      "loss": 1.4417,
+      "step": 4954
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.100378110544311e-05,
+      "loss": 1.3839,
+      "step": 4955
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.100355987055016e-05,
+      "loss": 1.2437,
+      "step": 4956
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1003338635657213e-05,
+      "loss": 1.5467,
+      "step": 4957
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1003117400764265e-05,
+      "loss": 1.4576,
+      "step": 4958
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1002896165871318e-05,
+      "loss": 1.524,
+      "step": 4959
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.100267493097837e-05,
+      "loss": 1.2634,
+      "step": 4960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1002453696085421e-05,
+      "loss": 1.4143,
+      "step": 4961
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1002232461192474e-05,
+      "loss": 1.4751,
+      "step": 4962
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1002011226299527e-05,
+      "loss": 1.467,
+      "step": 4963
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1001789991406578e-05,
+      "loss": 1.294,
+      "step": 4964
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.100156875651363e-05,
+      "loss": 1.3697,
+      "step": 4965
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1001347521620681e-05,
+      "loss": 1.5393,
+      "step": 4966
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1001126286727734e-05,
+      "loss": 1.3951,
+      "step": 4967
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1000905051834786e-05,
+      "loss": 1.3236,
+      "step": 4968
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1000683816941839e-05,
+      "loss": 1.5759,
+      "step": 4969
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1000462582048892e-05,
+      "loss": 1.3823,
+      "step": 4970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1000241347155942e-05,
+      "loss": 1.4444,
+      "step": 4971
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1000020112262993e-05,
+      "loss": 1.3831,
+      "step": 4972
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0999798877370046e-05,
+      "loss": 1.2547,
+      "step": 4973
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0999577642477099e-05,
+      "loss": 1.1933,
+      "step": 4974
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0999356407584151e-05,
+      "loss": 1.4942,
+      "step": 4975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0999135172691204e-05,
+      "loss": 1.366,
+      "step": 4976
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0998913937798255e-05,
+      "loss": 1.4684,
+      "step": 4977
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0998692702905307e-05,
+      "loss": 1.5001,
+      "step": 4978
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.099847146801236e-05,
+      "loss": 1.4095,
+      "step": 4979
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0998250233119411e-05,
+      "loss": 1.5537,
+      "step": 4980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0998028998226464e-05,
+      "loss": 1.2582,
+      "step": 4981
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0997807763333516e-05,
+      "loss": 1.3846,
+      "step": 4982
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0997586528440567e-05,
+      "loss": 1.5998,
+      "step": 4983
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.099736529354762e-05,
+      "loss": 1.3048,
+      "step": 4984
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0997144058654672e-05,
+      "loss": 1.5397,
+      "step": 4985
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0996922823761725e-05,
+      "loss": 1.3529,
+      "step": 4986
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0996701588868776e-05,
+      "loss": 1.3306,
+      "step": 4987
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0996480353975827e-05,
+      "loss": 1.4189,
+      "step": 4988
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.099625911908288e-05,
+      "loss": 1.3641,
+      "step": 4989
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0996037884189932e-05,
+      "loss": 1.3574,
+      "step": 4990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0995816649296985e-05,
+      "loss": 1.2619,
+      "step": 4991
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0995595414404037e-05,
+      "loss": 1.3501,
+      "step": 4992
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.099537417951109e-05,
+      "loss": 1.3587,
+      "step": 4993
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.099515294461814e-05,
+      "loss": 1.3204,
+      "step": 4994
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0994931709725193e-05,
+      "loss": 1.4074,
+      "step": 4995
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0994710474832244e-05,
+      "loss": 1.3922,
+      "step": 4996
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0994489239939297e-05,
+      "loss": 1.4239,
+      "step": 4997
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.099426800504635e-05,
+      "loss": 1.3349,
+      "step": 4998
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.09940467701534e-05,
+      "loss": 1.4382,
+      "step": 4999
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0993825535260453e-05,
+      "loss": 1.5516,
+      "step": 5000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0993604300367506e-05,
+      "loss": 1.3667,
+      "step": 5001
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0993383065474558e-05,
+      "loss": 1.2908,
+      "step": 5002
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0993161830581609e-05,
+      "loss": 1.4541,
+      "step": 5003
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0992940595688662e-05,
+      "loss": 1.43,
+      "step": 5004
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0992719360795713e-05,
+      "loss": 1.373,
+      "step": 5005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0992498125902765e-05,
+      "loss": 1.3538,
+      "step": 5006
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0992276891009818e-05,
+      "loss": 1.4119,
+      "step": 5007
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.099205565611687e-05,
+      "loss": 1.3672,
+      "step": 5008
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0991834421223923e-05,
+      "loss": 1.4244,
+      "step": 5009
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0991613186330974e-05,
+      "loss": 1.6171,
+      "step": 5010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0991391951438027e-05,
+      "loss": 1.4206,
+      "step": 5011
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0991170716545078e-05,
+      "loss": 1.3777,
+      "step": 5012
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.099094948165213e-05,
+      "loss": 1.2982,
+      "step": 5013
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0990728246759183e-05,
+      "loss": 1.3858,
+      "step": 5014
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0990507011866235e-05,
+      "loss": 1.4562,
+      "step": 5015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0990285776973286e-05,
+      "loss": 1.3918,
+      "step": 5016
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0990064542080339e-05,
+      "loss": 1.3449,
+      "step": 5017
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0989843307187392e-05,
+      "loss": 1.3677,
+      "step": 5018
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0989622072294442e-05,
+      "loss": 1.3587,
+      "step": 5019
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0989400837401495e-05,
+      "loss": 1.3539,
+      "step": 5020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0989179602508546e-05,
+      "loss": 1.3373,
+      "step": 5021
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0988958367615599e-05,
+      "loss": 1.3564,
+      "step": 5022
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0988737132722651e-05,
+      "loss": 1.336,
+      "step": 5023
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0988515897829704e-05,
+      "loss": 1.4244,
+      "step": 5024
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0988294662936756e-05,
+      "loss": 1.3491,
+      "step": 5025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0988073428043809e-05,
+      "loss": 1.3013,
+      "step": 5026
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.098785219315086e-05,
+      "loss": 1.3439,
+      "step": 5027
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0987630958257911e-05,
+      "loss": 1.3864,
+      "step": 5028
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0987409723364963e-05,
+      "loss": 1.4237,
+      "step": 5029
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0987188488472016e-05,
+      "loss": 1.3107,
+      "step": 5030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0986967253579069e-05,
+      "loss": 1.3692,
+      "step": 5031
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.098674601868612e-05,
+      "loss": 1.3735,
+      "step": 5032
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0986524783793172e-05,
+      "loss": 1.4131,
+      "step": 5033
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0986303548900225e-05,
+      "loss": 1.4798,
+      "step": 5034
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0986082314007276e-05,
+      "loss": 1.4458,
+      "step": 5035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0985861079114328e-05,
+      "loss": 1.3189,
+      "step": 5036
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0985639844221381e-05,
+      "loss": 1.403,
+      "step": 5037
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0985418609328432e-05,
+      "loss": 1.3782,
+      "step": 5038
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0985197374435485e-05,
+      "loss": 1.3948,
+      "step": 5039
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0984976139542537e-05,
+      "loss": 1.3143,
+      "step": 5040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.098475490464959e-05,
+      "loss": 1.4965,
+      "step": 5041
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0984533669756642e-05,
+      "loss": 1.3967,
+      "step": 5042
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0984312434863693e-05,
+      "loss": 1.3845,
+      "step": 5043
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0984091199970744e-05,
+      "loss": 1.2618,
+      "step": 5044
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0983869965077797e-05,
+      "loss": 1.4117,
+      "step": 5045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.098364873018485e-05,
+      "loss": 1.1591,
+      "step": 5046
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0983427495291902e-05,
+      "loss": 1.3952,
+      "step": 5047
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0983206260398953e-05,
+      "loss": 1.358,
+      "step": 5048
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0982985025506006e-05,
+      "loss": 1.3893,
+      "step": 5049
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0982763790613058e-05,
+      "loss": 1.4556,
+      "step": 5050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.098254255572011e-05,
+      "loss": 1.3481,
+      "step": 5051
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0982321320827162e-05,
+      "loss": 1.451,
+      "step": 5052
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0982100085934214e-05,
+      "loss": 1.4264,
+      "step": 5053
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0981878851041265e-05,
+      "loss": 1.3978,
+      "step": 5054
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0981657616148318e-05,
+      "loss": 1.3562,
+      "step": 5055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.098143638125537e-05,
+      "loss": 1.3692,
+      "step": 5056
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0981215146362423e-05,
+      "loss": 1.289,
+      "step": 5057
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0980993911469476e-05,
+      "loss": 1.446,
+      "step": 5058
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0980772676576527e-05,
+      "loss": 1.5352,
+      "step": 5059
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0980551441683577e-05,
+      "loss": 1.3877,
+      "step": 5060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.098033020679063e-05,
+      "loss": 1.4336,
+      "step": 5061
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0980108971897683e-05,
+      "loss": 1.3606,
+      "step": 5062
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0979887737004735e-05,
+      "loss": 1.4321,
+      "step": 5063
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0979666502111788e-05,
+      "loss": 1.4095,
+      "step": 5064
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0979445267218839e-05,
+      "loss": 1.3554,
+      "step": 5065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0979224032325891e-05,
+      "loss": 1.3287,
+      "step": 5066
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0979002797432944e-05,
+      "loss": 1.4202,
+      "step": 5067
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0978781562539995e-05,
+      "loss": 1.2943,
+      "step": 5068
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0978560327647048e-05,
+      "loss": 1.3888,
+      "step": 5069
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0978339092754099e-05,
+      "loss": 1.4369,
+      "step": 5070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0978117857861151e-05,
+      "loss": 1.4883,
+      "step": 5071
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0977896622968204e-05,
+      "loss": 1.3084,
+      "step": 5072
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0977675388075256e-05,
+      "loss": 1.277,
+      "step": 5073
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0977454153182309e-05,
+      "loss": 1.3458,
+      "step": 5074
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.097723291828936e-05,
+      "loss": 1.3461,
+      "step": 5075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.097701168339641e-05,
+      "loss": 1.4621,
+      "step": 5076
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0976790448503463e-05,
+      "loss": 1.2801,
+      "step": 5077
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0976569213610516e-05,
+      "loss": 1.3299,
+      "step": 5078
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0976347978717569e-05,
+      "loss": 1.542,
+      "step": 5079
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0976126743824621e-05,
+      "loss": 1.4811,
+      "step": 5080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0975905508931672e-05,
+      "loss": 1.3721,
+      "step": 5081
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0975684274038725e-05,
+      "loss": 1.2845,
+      "step": 5082
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0975463039145777e-05,
+      "loss": 1.3069,
+      "step": 5083
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0975241804252828e-05,
+      "loss": 1.4619,
+      "step": 5084
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0975020569359881e-05,
+      "loss": 1.4335,
+      "step": 5085
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0974799334466934e-05,
+      "loss": 1.2985,
+      "step": 5086
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0974578099573984e-05,
+      "loss": 1.2701,
+      "step": 5087
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0974356864681037e-05,
+      "loss": 1.3273,
+      "step": 5088
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.097413562978809e-05,
+      "loss": 1.3386,
+      "step": 5089
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0973914394895142e-05,
+      "loss": 1.3503,
+      "step": 5090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0973693160002193e-05,
+      "loss": 1.2988,
+      "step": 5091
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0973471925109244e-05,
+      "loss": 1.2851,
+      "step": 5092
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0973250690216297e-05,
+      "loss": 1.3534,
+      "step": 5093
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.097302945532335e-05,
+      "loss": 1.3887,
+      "step": 5094
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0972808220430402e-05,
+      "loss": 1.4277,
+      "step": 5095
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0972586985537455e-05,
+      "loss": 1.448,
+      "step": 5096
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0972365750644507e-05,
+      "loss": 1.4585,
+      "step": 5097
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0972144515751558e-05,
+      "loss": 1.2581,
+      "step": 5098
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.097192328085861e-05,
+      "loss": 1.3291,
+      "step": 5099
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0971702045965662e-05,
+      "loss": 1.3593,
+      "step": 5100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0971480811072714e-05,
+      "loss": 1.407,
+      "step": 5101
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0971259576179767e-05,
+      "loss": 1.4734,
+      "step": 5102
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0971038341286818e-05,
+      "loss": 1.3542,
+      "step": 5103
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.097081710639387e-05,
+      "loss": 1.4377,
+      "step": 5104
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0970595871500923e-05,
+      "loss": 1.2861,
+      "step": 5105
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0970374636607976e-05,
+      "loss": 1.4821,
+      "step": 5106
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0970153401715027e-05,
+      "loss": 1.3637,
+      "step": 5107
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0969932166822079e-05,
+      "loss": 1.4955,
+      "step": 5108
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.096971093192913e-05,
+      "loss": 1.243,
+      "step": 5109
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0969489697036183e-05,
+      "loss": 1.4194,
+      "step": 5110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0969268462143235e-05,
+      "loss": 1.3765,
+      "step": 5111
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0969047227250288e-05,
+      "loss": 1.5384,
+      "step": 5112
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.096882599235734e-05,
+      "loss": 1.4059,
+      "step": 5113
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0968604757464391e-05,
+      "loss": 1.4206,
+      "step": 5114
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0968383522571444e-05,
+      "loss": 1.5255,
+      "step": 5115
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0968162287678495e-05,
+      "loss": 1.4178,
+      "step": 5116
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0967941052785548e-05,
+      "loss": 1.3402,
+      "step": 5117
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.09677198178926e-05,
+      "loss": 1.4009,
+      "step": 5118
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0967498582999653e-05,
+      "loss": 1.3448,
+      "step": 5119
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0967277348106704e-05,
+      "loss": 1.2914,
+      "step": 5120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0967056113213756e-05,
+      "loss": 1.3024,
+      "step": 5121
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0966834878320809e-05,
+      "loss": 1.2501,
+      "step": 5122
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.096661364342786e-05,
+      "loss": 1.3253,
+      "step": 5123
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0966392408534912e-05,
+      "loss": 1.3726,
+      "step": 5124
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0966171173641963e-05,
+      "loss": 1.3827,
+      "step": 5125
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0965949938749016e-05,
+      "loss": 1.3549,
+      "step": 5126
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0965728703856069e-05,
+      "loss": 1.2944,
+      "step": 5127
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0965507468963121e-05,
+      "loss": 1.2824,
+      "step": 5128
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0965286234070174e-05,
+      "loss": 1.4489,
+      "step": 5129
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0965064999177226e-05,
+      "loss": 1.3726,
+      "step": 5130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0964843764284277e-05,
+      "loss": 1.5058,
+      "step": 5131
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0964622529391328e-05,
+      "loss": 1.266,
+      "step": 5132
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0964401294498381e-05,
+      "loss": 1.3452,
+      "step": 5133
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0964180059605433e-05,
+      "loss": 1.4241,
+      "step": 5134
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0963958824712486e-05,
+      "loss": 1.379,
+      "step": 5135
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0963737589819537e-05,
+      "loss": 1.3714,
+      "step": 5136
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.096351635492659e-05,
+      "loss": 1.3969,
+      "step": 5137
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0963295120033642e-05,
+      "loss": 1.3568,
+      "step": 5138
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0963073885140695e-05,
+      "loss": 1.4221,
+      "step": 5139
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0962852650247746e-05,
+      "loss": 1.4593,
+      "step": 5140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0962631415354797e-05,
+      "loss": 1.3036,
+      "step": 5141
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.096241018046185e-05,
+      "loss": 1.468,
+      "step": 5142
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0962188945568902e-05,
+      "loss": 1.4889,
+      "step": 5143
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0961967710675955e-05,
+      "loss": 1.395,
+      "step": 5144
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0961746475783007e-05,
+      "loss": 1.4634,
+      "step": 5145
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.096152524089006e-05,
+      "loss": 1.3307,
+      "step": 5146
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.096130400599711e-05,
+      "loss": 1.3508,
+      "step": 5147
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0961082771104162e-05,
+      "loss": 1.3004,
+      "step": 5148
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0960861536211214e-05,
+      "loss": 1.399,
+      "step": 5149
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0960640301318267e-05,
+      "loss": 1.3889,
+      "step": 5150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.096041906642532e-05,
+      "loss": 1.3349,
+      "step": 5151
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.096019783153237e-05,
+      "loss": 1.4349,
+      "step": 5152
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0959976596639423e-05,
+      "loss": 1.4795,
+      "step": 5153
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0959755361746476e-05,
+      "loss": 1.3494,
+      "step": 5154
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0959534126853528e-05,
+      "loss": 1.4329,
+      "step": 5155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0959312891960579e-05,
+      "loss": 1.3105,
+      "step": 5156
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0959091657067632e-05,
+      "loss": 1.3288,
+      "step": 5157
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0958870422174683e-05,
+      "loss": 1.3592,
+      "step": 5158
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0958649187281735e-05,
+      "loss": 1.6036,
+      "step": 5159
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0958427952388788e-05,
+      "loss": 1.3874,
+      "step": 5160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.095820671749584e-05,
+      "loss": 1.5334,
+      "step": 5161
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0957985482602893e-05,
+      "loss": 1.3236,
+      "step": 5162
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0957764247709944e-05,
+      "loss": 1.3151,
+      "step": 5163
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0957543012816995e-05,
+      "loss": 1.2234,
+      "step": 5164
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0957321777924048e-05,
+      "loss": 1.5279,
+      "step": 5165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.09571005430311e-05,
+      "loss": 1.2433,
+      "step": 5166
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0956879308138153e-05,
+      "loss": 1.3212,
+      "step": 5167
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0956658073245205e-05,
+      "loss": 1.3798,
+      "step": 5168
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0956436838352256e-05,
+      "loss": 1.3864,
+      "step": 5169
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0956215603459309e-05,
+      "loss": 1.4103,
+      "step": 5170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0955994368566361e-05,
+      "loss": 1.3081,
+      "step": 5171
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0955773133673412e-05,
+      "loss": 1.3948,
+      "step": 5172
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0955551898780465e-05,
+      "loss": 1.4159,
+      "step": 5173
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0955330663887516e-05,
+      "loss": 1.3444,
+      "step": 5174
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0955109428994569e-05,
+      "loss": 1.3986,
+      "step": 5175
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0954888194101621e-05,
+      "loss": 1.2343,
+      "step": 5176
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0954666959208674e-05,
+      "loss": 1.3011,
+      "step": 5177
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0954445724315726e-05,
+      "loss": 1.3885,
+      "step": 5178
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0954224489422777e-05,
+      "loss": 1.4046,
+      "step": 5179
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0954003254529828e-05,
+      "loss": 1.4239,
+      "step": 5180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.095378201963688e-05,
+      "loss": 1.3794,
+      "step": 5181
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0953560784743933e-05,
+      "loss": 1.3902,
+      "step": 5182
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0953339549850986e-05,
+      "loss": 1.3787,
+      "step": 5183
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0953118314958039e-05,
+      "loss": 1.3752,
+      "step": 5184
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.095289708006509e-05,
+      "loss": 1.359,
+      "step": 5185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0952675845172142e-05,
+      "loss": 1.4138,
+      "step": 5186
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0952454610279195e-05,
+      "loss": 1.3409,
+      "step": 5187
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0952233375386246e-05,
+      "loss": 1.3094,
+      "step": 5188
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0952012140493298e-05,
+      "loss": 1.3378,
+      "step": 5189
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0951790905600351e-05,
+      "loss": 1.4128,
+      "step": 5190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0951569670707402e-05,
+      "loss": 1.3498,
+      "step": 5191
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0951348435814454e-05,
+      "loss": 1.2625,
+      "step": 5192
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0951127200921507e-05,
+      "loss": 1.4163,
+      "step": 5193
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.095090596602856e-05,
+      "loss": 1.4742,
+      "step": 5194
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.095068473113561e-05,
+      "loss": 1.3438,
+      "step": 5195
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0950463496242663e-05,
+      "loss": 1.4231,
+      "step": 5196
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0950242261349714e-05,
+      "loss": 1.5566,
+      "step": 5197
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0950021026456767e-05,
+      "loss": 1.4184,
+      "step": 5198
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.094979979156382e-05,
+      "loss": 1.4377,
+      "step": 5199
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0949578556670872e-05,
+      "loss": 1.2829,
+      "step": 5200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0949357321777925e-05,
+      "loss": 1.447,
+      "step": 5201
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0949136086884976e-05,
+      "loss": 1.3896,
+      "step": 5202
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0948914851992028e-05,
+      "loss": 1.3407,
+      "step": 5203
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0948693617099079e-05,
+      "loss": 1.2885,
+      "step": 5204
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0948472382206132e-05,
+      "loss": 1.3785,
+      "step": 5205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0948251147313184e-05,
+      "loss": 1.4734,
+      "step": 5206
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0948029912420235e-05,
+      "loss": 1.5518,
+      "step": 5207
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0947808677527288e-05,
+      "loss": 1.4612,
+      "step": 5208
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.094758744263434e-05,
+      "loss": 1.4929,
+      "step": 5209
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0947366207741393e-05,
+      "loss": 1.3221,
+      "step": 5210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0947144972848444e-05,
+      "loss": 1.4047,
+      "step": 5211
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0946923737955497e-05,
+      "loss": 1.4764,
+      "step": 5212
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0946702503062547e-05,
+      "loss": 1.5297,
+      "step": 5213
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.09464812681696e-05,
+      "loss": 1.4445,
+      "step": 5214
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0946260033276653e-05,
+      "loss": 1.3548,
+      "step": 5215
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0946038798383705e-05,
+      "loss": 1.2975,
+      "step": 5216
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0945817563490758e-05,
+      "loss": 1.4394,
+      "step": 5217
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0945596328597809e-05,
+      "loss": 1.3088,
+      "step": 5218
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0945375093704861e-05,
+      "loss": 1.2951,
+      "step": 5219
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0945153858811912e-05,
+      "loss": 1.5695,
+      "step": 5220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0944932623918965e-05,
+      "loss": 1.4021,
+      "step": 5221
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0944711389026018e-05,
+      "loss": 1.4916,
+      "step": 5222
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.094449015413307e-05,
+      "loss": 1.4548,
+      "step": 5223
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0944268919240121e-05,
+      "loss": 1.4828,
+      "step": 5224
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0944047684347174e-05,
+      "loss": 1.5345,
+      "step": 5225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0943826449454226e-05,
+      "loss": 1.55,
+      "step": 5226
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0943605214561277e-05,
+      "loss": 1.462,
+      "step": 5227
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.094338397966833e-05,
+      "loss": 1.3736,
+      "step": 5228
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.094316274477538e-05,
+      "loss": 1.3345,
+      "step": 5229
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0942941509882433e-05,
+      "loss": 1.4682,
+      "step": 5230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0942720274989486e-05,
+      "loss": 1.4756,
+      "step": 5231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0942499040096539e-05,
+      "loss": 1.2857,
+      "step": 5232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0942277805203591e-05,
+      "loss": 1.3432,
+      "step": 5233
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0942056570310644e-05,
+      "loss": 1.2858,
+      "step": 5234
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0941835335417695e-05,
+      "loss": 1.3068,
+      "step": 5235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0941614100524746e-05,
+      "loss": 1.3213,
+      "step": 5236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0941392865631798e-05,
+      "loss": 1.3956,
+      "step": 5237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0941171630738851e-05,
+      "loss": 1.3897,
+      "step": 5238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0940950395845904e-05,
+      "loss": 1.2551,
+      "step": 5239
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0940729160952954e-05,
+      "loss": 1.2663,
+      "step": 5240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0940507926060007e-05,
+      "loss": 1.2951,
+      "step": 5241
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.094028669116706e-05,
+      "loss": 1.3592,
+      "step": 5242
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0940065456274112e-05,
+      "loss": 1.4001,
+      "step": 5243
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0939844221381163e-05,
+      "loss": 1.3268,
+      "step": 5244
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0939622986488214e-05,
+      "loss": 1.2703,
+      "step": 5245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0939401751595267e-05,
+      "loss": 1.4101,
+      "step": 5246
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.093918051670232e-05,
+      "loss": 1.3506,
+      "step": 5247
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0938959281809372e-05,
+      "loss": 1.3645,
+      "step": 5248
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0938738046916425e-05,
+      "loss": 1.3915,
+      "step": 5249
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0938516812023477e-05,
+      "loss": 1.3519,
+      "step": 5250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0938295577130528e-05,
+      "loss": 1.411,
+      "step": 5251
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0938074342237579e-05,
+      "loss": 1.3915,
+      "step": 5252
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0937853107344632e-05,
+      "loss": 1.3182,
+      "step": 5253
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0937631872451684e-05,
+      "loss": 1.4503,
+      "step": 5254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0937410637558737e-05,
+      "loss": 1.429,
+      "step": 5255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0937189402665788e-05,
+      "loss": 1.4359,
+      "step": 5256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.093696816777284e-05,
+      "loss": 1.4243,
+      "step": 5257
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0936746932879893e-05,
+      "loss": 1.3802,
+      "step": 5258
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0936525697986946e-05,
+      "loss": 1.3086,
+      "step": 5259
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0936304463093997e-05,
+      "loss": 1.446,
+      "step": 5260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0936083228201049e-05,
+      "loss": 1.2553,
+      "step": 5261
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.09358619933081e-05,
+      "loss": 1.321,
+      "step": 5262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0935640758415153e-05,
+      "loss": 1.2806,
+      "step": 5263
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0935419523522205e-05,
+      "loss": 1.2891,
+      "step": 5264
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0935198288629258e-05,
+      "loss": 1.2834,
+      "step": 5265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.093497705373631e-05,
+      "loss": 1.342,
+      "step": 5266
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0934755818843361e-05,
+      "loss": 1.3606,
+      "step": 5267
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0934534583950412e-05,
+      "loss": 1.3634,
+      "step": 5268
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0934313349057465e-05,
+      "loss": 1.3352,
+      "step": 5269
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0934092114164518e-05,
+      "loss": 1.3181,
+      "step": 5270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.093387087927157e-05,
+      "loss": 1.3952,
+      "step": 5271
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0933649644378623e-05,
+      "loss": 1.3881,
+      "step": 5272
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0933428409485674e-05,
+      "loss": 1.3456,
+      "step": 5273
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0933207174592726e-05,
+      "loss": 1.3541,
+      "step": 5274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0932985939699779e-05,
+      "loss": 1.2863,
+      "step": 5275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.093276470480683e-05,
+      "loss": 1.3568,
+      "step": 5276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0932543469913882e-05,
+      "loss": 1.293,
+      "step": 5277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0932322235020933e-05,
+      "loss": 1.3658,
+      "step": 5278
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0932101000127986e-05,
+      "loss": 1.3001,
+      "step": 5279
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0931879765235039e-05,
+      "loss": 1.3357,
+      "step": 5280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0931658530342091e-05,
+      "loss": 1.3467,
+      "step": 5281
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0931437295449144e-05,
+      "loss": 1.3951,
+      "step": 5282
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0931216060556195e-05,
+      "loss": 1.3544,
+      "step": 5283
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0930994825663247e-05,
+      "loss": 1.5386,
+      "step": 5284
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0930773590770298e-05,
+      "loss": 1.3892,
+      "step": 5285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0930552355877351e-05,
+      "loss": 1.2476,
+      "step": 5286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0930331120984403e-05,
+      "loss": 1.327,
+      "step": 5287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0930109886091456e-05,
+      "loss": 1.3108,
+      "step": 5288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0929888651198507e-05,
+      "loss": 1.3026,
+      "step": 5289
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.092966741630556e-05,
+      "loss": 1.4127,
+      "step": 5290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0929446181412612e-05,
+      "loss": 1.3448,
+      "step": 5291
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0929224946519663e-05,
+      "loss": 1.5294,
+      "step": 5292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0929003711626716e-05,
+      "loss": 1.4428,
+      "step": 5293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0928782476733768e-05,
+      "loss": 1.3724,
+      "step": 5294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.092856124184082e-05,
+      "loss": 1.3457,
+      "step": 5295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0928340006947872e-05,
+      "loss": 1.258,
+      "step": 5296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0928118772054925e-05,
+      "loss": 1.4903,
+      "step": 5297
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0927897537161977e-05,
+      "loss": 1.2861,
+      "step": 5298
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0927676302269028e-05,
+      "loss": 1.4102,
+      "step": 5299
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.092745506737608e-05,
+      "loss": 1.2344,
+      "step": 5300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0927233832483132e-05,
+      "loss": 1.3791,
+      "step": 5301
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0927012597590184e-05,
+      "loss": 1.4994,
+      "step": 5302
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0926791362697237e-05,
+      "loss": 1.3263,
+      "step": 5303
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.092657012780429e-05,
+      "loss": 1.2884,
+      "step": 5304
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0926348892911342e-05,
+      "loss": 1.3389,
+      "step": 5305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0926127658018393e-05,
+      "loss": 1.3898,
+      "step": 5306
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0925906423125446e-05,
+      "loss": 1.2799,
+      "step": 5307
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0925685188232496e-05,
+      "loss": 1.337,
+      "step": 5308
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0925463953339549e-05,
+      "loss": 1.2426,
+      "step": 5309
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0925242718446602e-05,
+      "loss": 1.3939,
+      "step": 5310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0925021483553653e-05,
+      "loss": 1.4116,
+      "step": 5311
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0924800248660705e-05,
+      "loss": 1.2323,
+      "step": 5312
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0924579013767758e-05,
+      "loss": 1.3142,
+      "step": 5313
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.092435777887481e-05,
+      "loss": 1.4419,
+      "step": 5314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0924136543981861e-05,
+      "loss": 1.3133,
+      "step": 5315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0923915309088914e-05,
+      "loss": 1.2381,
+      "step": 5316
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0923694074195965e-05,
+      "loss": 1.3721,
+      "step": 5317
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0923472839303018e-05,
+      "loss": 1.3736,
+      "step": 5318
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.092325160441007e-05,
+      "loss": 1.3354,
+      "step": 5319
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0923030369517123e-05,
+      "loss": 1.3493,
+      "step": 5320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0922809134624175e-05,
+      "loss": 1.3177,
+      "step": 5321
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0922587899731226e-05,
+      "loss": 1.354,
+      "step": 5322
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0922366664838279e-05,
+      "loss": 1.4083,
+      "step": 5323
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.092214542994533e-05,
+      "loss": 1.4309,
+      "step": 5324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0921924195052382e-05,
+      "loss": 1.3772,
+      "step": 5325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0921702960159435e-05,
+      "loss": 1.3179,
+      "step": 5326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0921481725266488e-05,
+      "loss": 1.3606,
+      "step": 5327
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0921260490373539e-05,
+      "loss": 1.3,
+      "step": 5328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0921039255480591e-05,
+      "loss": 1.3865,
+      "step": 5329
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0920818020587644e-05,
+      "loss": 1.3883,
+      "step": 5330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0920596785694696e-05,
+      "loss": 1.4163,
+      "step": 5331
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0920375550801747e-05,
+      "loss": 1.2712,
+      "step": 5332
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0920154315908798e-05,
+      "loss": 1.3433,
+      "step": 5333
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.091993308101585e-05,
+      "loss": 1.3126,
+      "step": 5334
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0919711846122903e-05,
+      "loss": 1.3589,
+      "step": 5335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0919490611229956e-05,
+      "loss": 1.4548,
+      "step": 5336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0919269376337009e-05,
+      "loss": 1.4807,
+      "step": 5337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.091904814144406e-05,
+      "loss": 1.3791,
+      "step": 5338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0918826906551112e-05,
+      "loss": 1.3489,
+      "step": 5339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0918605671658163e-05,
+      "loss": 1.4736,
+      "step": 5340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0918384436765216e-05,
+      "loss": 1.2927,
+      "step": 5341
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0918163201872268e-05,
+      "loss": 1.5098,
+      "step": 5342
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0917941966979321e-05,
+      "loss": 1.3116,
+      "step": 5343
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0917720732086372e-05,
+      "loss": 1.4654,
+      "step": 5344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0917499497193424e-05,
+      "loss": 1.4537,
+      "step": 5345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0917278262300477e-05,
+      "loss": 1.349,
+      "step": 5346
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.091705702740753e-05,
+      "loss": 1.3183,
+      "step": 5347
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.091683579251458e-05,
+      "loss": 1.4034,
+      "step": 5348
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0916614557621632e-05,
+      "loss": 1.3627,
+      "step": 5349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0916393322728684e-05,
+      "loss": 1.2935,
+      "step": 5350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0916172087835737e-05,
+      "loss": 1.3923,
+      "step": 5351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.091595085294279e-05,
+      "loss": 1.4386,
+      "step": 5352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0915729618049842e-05,
+      "loss": 1.3037,
+      "step": 5353
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0915508383156895e-05,
+      "loss": 1.2881,
+      "step": 5354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0915287148263946e-05,
+      "loss": 1.415,
+      "step": 5355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0915065913370996e-05,
+      "loss": 1.4005,
+      "step": 5356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0914844678478049e-05,
+      "loss": 1.4749,
+      "step": 5357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0914623443585102e-05,
+      "loss": 1.3146,
+      "step": 5358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0914402208692154e-05,
+      "loss": 1.3308,
+      "step": 5359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0914180973799205e-05,
+      "loss": 1.375,
+      "step": 5360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0913959738906258e-05,
+      "loss": 1.1312,
+      "step": 5361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.091373850401331e-05,
+      "loss": 1.4226,
+      "step": 5362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0913517269120363e-05,
+      "loss": 1.3027,
+      "step": 5363
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0913296034227414e-05,
+      "loss": 1.3739,
+      "step": 5364
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0913074799334467e-05,
+      "loss": 1.4209,
+      "step": 5365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0912853564441517e-05,
+      "loss": 1.4843,
+      "step": 5366
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.091263232954857e-05,
+      "loss": 1.5793,
+      "step": 5367
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0912411094655623e-05,
+      "loss": 1.2727,
+      "step": 5368
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0912189859762675e-05,
+      "loss": 1.3273,
+      "step": 5369
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0911968624869728e-05,
+      "loss": 1.6024,
+      "step": 5370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0911747389976779e-05,
+      "loss": 1.4197,
+      "step": 5371
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0911526155083831e-05,
+      "loss": 1.4134,
+      "step": 5372
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0911304920190882e-05,
+      "loss": 1.2816,
+      "step": 5373
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0911083685297935e-05,
+      "loss": 1.4676,
+      "step": 5374
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0910862450404988e-05,
+      "loss": 1.3735,
+      "step": 5375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.091064121551204e-05,
+      "loss": 1.4864,
+      "step": 5376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0910419980619091e-05,
+      "loss": 1.3644,
+      "step": 5377
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0910198745726144e-05,
+      "loss": 1.5217,
+      "step": 5378
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0909977510833196e-05,
+      "loss": 1.4439,
+      "step": 5379
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0909756275940247e-05,
+      "loss": 1.2224,
+      "step": 5380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.09095350410473e-05,
+      "loss": 1.215,
+      "step": 5381
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.090931380615435e-05,
+      "loss": 1.3307,
+      "step": 5382
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0909092571261403e-05,
+      "loss": 1.3615,
+      "step": 5383
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0908871336368456e-05,
+      "loss": 1.3298,
+      "step": 5384
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0908650101475509e-05,
+      "loss": 1.4233,
+      "step": 5385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0908428866582561e-05,
+      "loss": 1.3184,
+      "step": 5386
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0908207631689612e-05,
+      "loss": 1.2927,
+      "step": 5387
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0907986396796665e-05,
+      "loss": 1.3028,
+      "step": 5388
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0907765161903716e-05,
+      "loss": 1.3789,
+      "step": 5389
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0907543927010768e-05,
+      "loss": 1.2707,
+      "step": 5390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0907322692117821e-05,
+      "loss": 1.3918,
+      "step": 5391
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0907101457224874e-05,
+      "loss": 1.3686,
+      "step": 5392
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0906880222331924e-05,
+      "loss": 1.2871,
+      "step": 5393
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0906658987438977e-05,
+      "loss": 1.3089,
+      "step": 5394
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.090643775254603e-05,
+      "loss": 1.3105,
+      "step": 5395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.090621651765308e-05,
+      "loss": 1.4709,
+      "step": 5396
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0905995282760133e-05,
+      "loss": 1.3251,
+      "step": 5397
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0905774047867186e-05,
+      "loss": 1.2506,
+      "step": 5398
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0905552812974237e-05,
+      "loss": 1.2613,
+      "step": 5399
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.090533157808129e-05,
+      "loss": 1.3616,
+      "step": 5400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0905110343188342e-05,
+      "loss": 1.3831,
+      "step": 5401
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0904889108295395e-05,
+      "loss": 1.3475,
+      "step": 5402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0904667873402445e-05,
+      "loss": 1.3733,
+      "step": 5403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0904446638509498e-05,
+      "loss": 1.3272,
+      "step": 5404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0904225403616549e-05,
+      "loss": 1.282,
+      "step": 5405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0904004168723602e-05,
+      "loss": 1.3288,
+      "step": 5406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0903782933830654e-05,
+      "loss": 1.319,
+      "step": 5407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0903561698937707e-05,
+      "loss": 1.3552,
+      "step": 5408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.090334046404476e-05,
+      "loss": 1.3044,
+      "step": 5409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.090311922915181e-05,
+      "loss": 1.5408,
+      "step": 5410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0902897994258863e-05,
+      "loss": 1.2794,
+      "step": 5411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0902676759365914e-05,
+      "loss": 1.2998,
+      "step": 5412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0902455524472967e-05,
+      "loss": 1.5349,
+      "step": 5413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0902234289580019e-05,
+      "loss": 1.3586,
+      "step": 5414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.090201305468707e-05,
+      "loss": 1.3258,
+      "step": 5415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0901791819794123e-05,
+      "loss": 1.293,
+      "step": 5416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0901570584901175e-05,
+      "loss": 1.4679,
+      "step": 5417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0901349350008228e-05,
+      "loss": 1.2567,
+      "step": 5418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.090112811511528e-05,
+      "loss": 1.4409,
+      "step": 5419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0900906880222331e-05,
+      "loss": 1.3188,
+      "step": 5420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0900685645329382e-05,
+      "loss": 1.3659,
+      "step": 5421
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0900464410436435e-05,
+      "loss": 1.2801,
+      "step": 5422
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0900243175543488e-05,
+      "loss": 1.4862,
+      "step": 5423
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.090002194065054e-05,
+      "loss": 1.4338,
+      "step": 5424
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0899800705757593e-05,
+      "loss": 1.402,
+      "step": 5425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0899579470864644e-05,
+      "loss": 1.4223,
+      "step": 5426
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0899358235971696e-05,
+      "loss": 1.2427,
+      "step": 5427
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0899137001078747e-05,
+      "loss": 1.3294,
+      "step": 5428
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.08989157661858e-05,
+      "loss": 1.4275,
+      "step": 5429
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0898694531292852e-05,
+      "loss": 1.263,
+      "step": 5430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0898473296399903e-05,
+      "loss": 1.2881,
+      "step": 5431
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0898252061506956e-05,
+      "loss": 1.3866,
+      "step": 5432
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0898030826614009e-05,
+      "loss": 1.322,
+      "step": 5433
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0897809591721061e-05,
+      "loss": 1.3093,
+      "step": 5434
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0897588356828114e-05,
+      "loss": 1.4789,
+      "step": 5435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0897367121935165e-05,
+      "loss": 1.3519,
+      "step": 5436
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0897145887042216e-05,
+      "loss": 1.3074,
+      "step": 5437
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0896924652149268e-05,
+      "loss": 1.3033,
+      "step": 5438
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0896703417256321e-05,
+      "loss": 1.3182,
+      "step": 5439
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0896482182363373e-05,
+      "loss": 1.5724,
+      "step": 5440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0896260947470426e-05,
+      "loss": 1.4547,
+      "step": 5441
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0896039712577477e-05,
+      "loss": 1.3603,
+      "step": 5442
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.089581847768453e-05,
+      "loss": 1.3912,
+      "step": 5443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.089559724279158e-05,
+      "loss": 1.3122,
+      "step": 5444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0895376007898633e-05,
+      "loss": 1.4688,
+      "step": 5445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0895154773005686e-05,
+      "loss": 1.4177,
+      "step": 5446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0894933538112738e-05,
+      "loss": 1.2214,
+      "step": 5447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.089471230321979e-05,
+      "loss": 1.2854,
+      "step": 5448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0894491068326842e-05,
+      "loss": 1.3012,
+      "step": 5449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0894269833433895e-05,
+      "loss": 1.3766,
+      "step": 5450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0894048598540947e-05,
+      "loss": 1.4151,
+      "step": 5451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0893827363647998e-05,
+      "loss": 1.3706,
+      "step": 5452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0893606128755049e-05,
+      "loss": 1.3487,
+      "step": 5453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0893384893862102e-05,
+      "loss": 1.3106,
+      "step": 5454
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0893163658969154e-05,
+      "loss": 1.3166,
+      "step": 5455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0892942424076207e-05,
+      "loss": 1.2339,
+      "step": 5456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.089272118918326e-05,
+      "loss": 1.4833,
+      "step": 5457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0892499954290312e-05,
+      "loss": 1.5717,
+      "step": 5458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0892278719397363e-05,
+      "loss": 1.4866,
+      "step": 5459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0892057484504416e-05,
+      "loss": 1.2963,
+      "step": 5460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0891836249611466e-05,
+      "loss": 1.4526,
+      "step": 5461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0891615014718519e-05,
+      "loss": 1.3495,
+      "step": 5462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0891393779825572e-05,
+      "loss": 1.3242,
+      "step": 5463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0891172544932623e-05,
+      "loss": 1.3329,
+      "step": 5464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0890951310039675e-05,
+      "loss": 1.4096,
+      "step": 5465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0890730075146728e-05,
+      "loss": 1.4003,
+      "step": 5466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.089050884025378e-05,
+      "loss": 1.409,
+      "step": 5467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0890287605360831e-05,
+      "loss": 1.345,
+      "step": 5468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0890066370467884e-05,
+      "loss": 1.3692,
+      "step": 5469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0889845135574935e-05,
+      "loss": 1.3151,
+      "step": 5470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0889623900681987e-05,
+      "loss": 1.4586,
+      "step": 5471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.088940266578904e-05,
+      "loss": 1.3232,
+      "step": 5472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0889181430896093e-05,
+      "loss": 1.3812,
+      "step": 5473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0888960196003145e-05,
+      "loss": 1.3597,
+      "step": 5474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0888738961110196e-05,
+      "loss": 1.46,
+      "step": 5475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0888517726217249e-05,
+      "loss": 1.3078,
+      "step": 5476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.08882964913243e-05,
+      "loss": 1.4046,
+      "step": 5477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0888075256431352e-05,
+      "loss": 1.4027,
+      "step": 5478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0887854021538405e-05,
+      "loss": 1.4071,
+      "step": 5479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0887632786645458e-05,
+      "loss": 1.1996,
+      "step": 5480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0887411551752509e-05,
+      "loss": 1.3307,
+      "step": 5481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0887190316859561e-05,
+      "loss": 1.405,
+      "step": 5482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0886969081966614e-05,
+      "loss": 1.3379,
+      "step": 5483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0886747847073665e-05,
+      "loss": 1.43,
+      "step": 5484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0886526612180717e-05,
+      "loss": 1.4551,
+      "step": 5485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0886305377287768e-05,
+      "loss": 1.3504,
+      "step": 5486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.088608414239482e-05,
+      "loss": 1.5186,
+      "step": 5487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0885862907501873e-05,
+      "loss": 1.2731,
+      "step": 5488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0885641672608926e-05,
+      "loss": 1.2933,
+      "step": 5489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0885420437715979e-05,
+      "loss": 1.4802,
+      "step": 5490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.088519920282303e-05,
+      "loss": 1.3591,
+      "step": 5491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0884977967930082e-05,
+      "loss": 1.3498,
+      "step": 5492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0884756733037133e-05,
+      "loss": 1.4811,
+      "step": 5493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0884535498144186e-05,
+      "loss": 1.5673,
+      "step": 5494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0884314263251238e-05,
+      "loss": 1.3776,
+      "step": 5495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0884093028358291e-05,
+      "loss": 1.3367,
+      "step": 5496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0883871793465342e-05,
+      "loss": 1.4746,
+      "step": 5497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0883650558572394e-05,
+      "loss": 1.3066,
+      "step": 5498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0883429323679447e-05,
+      "loss": 1.304,
+      "step": 5499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0883208088786498e-05,
+      "loss": 1.2777,
+      "step": 5500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.088298685389355e-05,
+      "loss": 1.4898,
+      "step": 5501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0882765619000603e-05,
+      "loss": 1.4759,
+      "step": 5502
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0882544384107654e-05,
+      "loss": 1.4402,
+      "step": 5503
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0882323149214707e-05,
+      "loss": 1.4226,
+      "step": 5504
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.088210191432176e-05,
+      "loss": 1.4176,
+      "step": 5505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0881880679428812e-05,
+      "loss": 1.3498,
+      "step": 5506
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0881659444535865e-05,
+      "loss": 1.2575,
+      "step": 5507
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0881438209642915e-05,
+      "loss": 1.4905,
+      "step": 5508
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0881216974749966e-05,
+      "loss": 1.3807,
+      "step": 5509
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0880995739857019e-05,
+      "loss": 1.4735,
+      "step": 5510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0880774504964072e-05,
+      "loss": 1.3981,
+      "step": 5511
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0880553270071124e-05,
+      "loss": 1.3653,
+      "step": 5512
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0880332035178177e-05,
+      "loss": 1.2964,
+      "step": 5513
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0880110800285228e-05,
+      "loss": 1.3132,
+      "step": 5514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.087988956539228e-05,
+      "loss": 1.448,
+      "step": 5515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0879668330499331e-05,
+      "loss": 1.3564,
+      "step": 5516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0879447095606384e-05,
+      "loss": 1.2621,
+      "step": 5517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0879225860713437e-05,
+      "loss": 1.4334,
+      "step": 5518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0879004625820487e-05,
+      "loss": 1.347,
+      "step": 5519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.087878339092754e-05,
+      "loss": 1.4333,
+      "step": 5520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0878562156034593e-05,
+      "loss": 1.311,
+      "step": 5521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0878340921141645e-05,
+      "loss": 1.498,
+      "step": 5522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0878119686248698e-05,
+      "loss": 1.4357,
+      "step": 5523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0877898451355749e-05,
+      "loss": 1.3755,
+      "step": 5524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.08776772164628e-05,
+      "loss": 1.4238,
+      "step": 5525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0877455981569852e-05,
+      "loss": 1.3656,
+      "step": 5526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0877234746676905e-05,
+      "loss": 1.2517,
+      "step": 5527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0877013511783958e-05,
+      "loss": 1.2337,
+      "step": 5528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.087679227689101e-05,
+      "loss": 1.3242,
+      "step": 5529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0876571041998061e-05,
+      "loss": 1.3261,
+      "step": 5530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0876349807105114e-05,
+      "loss": 1.4824,
+      "step": 5531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0876128572212165e-05,
+      "loss": 1.3066,
+      "step": 5532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0875907337319217e-05,
+      "loss": 1.3392,
+      "step": 5533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.087568610242627e-05,
+      "loss": 1.5061,
+      "step": 5534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.087546486753332e-05,
+      "loss": 1.2934,
+      "step": 5535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0875243632640373e-05,
+      "loss": 1.4538,
+      "step": 5536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0875022397747426e-05,
+      "loss": 1.2857,
+      "step": 5537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0874801162854479e-05,
+      "loss": 1.3701,
+      "step": 5538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0874579927961531e-05,
+      "loss": 1.2174,
+      "step": 5539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0874358693068582e-05,
+      "loss": 1.3921,
+      "step": 5540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0874137458175633e-05,
+      "loss": 1.3702,
+      "step": 5541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0873916223282686e-05,
+      "loss": 1.4367,
+      "step": 5542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0873694988389738e-05,
+      "loss": 1.5523,
+      "step": 5543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0873473753496791e-05,
+      "loss": 1.363,
+      "step": 5544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0873252518603844e-05,
+      "loss": 1.3405,
+      "step": 5545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0873031283710894e-05,
+      "loss": 1.3454,
+      "step": 5546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0872810048817947e-05,
+      "loss": 1.4569,
+      "step": 5547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0872588813925e-05,
+      "loss": 1.3633,
+      "step": 5548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.087236757903205e-05,
+      "loss": 1.5032,
+      "step": 5549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0872146344139103e-05,
+      "loss": 1.2687,
+      "step": 5550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0871925109246156e-05,
+      "loss": 1.3319,
+      "step": 5551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0871703874353207e-05,
+      "loss": 1.377,
+      "step": 5552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.087148263946026e-05,
+      "loss": 1.1922,
+      "step": 5553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0871261404567312e-05,
+      "loss": 1.248,
+      "step": 5554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0871040169674365e-05,
+      "loss": 1.3541,
+      "step": 5555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0870818934781415e-05,
+      "loss": 1.3676,
+      "step": 5556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0870597699888466e-05,
+      "loss": 1.4772,
+      "step": 5557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0870376464995519e-05,
+      "loss": 1.3947,
+      "step": 5558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0870155230102572e-05,
+      "loss": 1.323,
+      "step": 5559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0869933995209624e-05,
+      "loss": 1.4378,
+      "step": 5560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0869712760316677e-05,
+      "loss": 1.221,
+      "step": 5561
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.086949152542373e-05,
+      "loss": 1.3237,
+      "step": 5562
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.086927029053078e-05,
+      "loss": 1.3848,
+      "step": 5563
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0869049055637833e-05,
+      "loss": 1.3909,
+      "step": 5564
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0868827820744884e-05,
+      "loss": 1.3999,
+      "step": 5565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0868606585851936e-05,
+      "loss": 1.3867,
+      "step": 5566
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0868385350958989e-05,
+      "loss": 1.4656,
+      "step": 5567
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.086816411606604e-05,
+      "loss": 1.2546,
+      "step": 5568
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0867942881173093e-05,
+      "loss": 1.2074,
+      "step": 5569
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0867721646280145e-05,
+      "loss": 1.4091,
+      "step": 5570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0867500411387198e-05,
+      "loss": 1.4948,
+      "step": 5571
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0867279176494249e-05,
+      "loss": 1.3806,
+      "step": 5572
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0867057941601301e-05,
+      "loss": 1.2687,
+      "step": 5573
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0866836706708352e-05,
+      "loss": 1.2343,
+      "step": 5574
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0866615471815405e-05,
+      "loss": 1.3225,
+      "step": 5575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0866394236922458e-05,
+      "loss": 1.3415,
+      "step": 5576
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.086617300202951e-05,
+      "loss": 1.2068,
+      "step": 5577
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0865951767136563e-05,
+      "loss": 1.3745,
+      "step": 5578
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0865730532243614e-05,
+      "loss": 1.3834,
+      "step": 5579
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0865509297350666e-05,
+      "loss": 1.3583,
+      "step": 5580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0865288062457717e-05,
+      "loss": 1.4067,
+      "step": 5581
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.086506682756477e-05,
+      "loss": 1.5105,
+      "step": 5582
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0864845592671822e-05,
+      "loss": 1.3164,
+      "step": 5583
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0864624357778875e-05,
+      "loss": 1.3358,
+      "step": 5584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0864403122885926e-05,
+      "loss": 1.4518,
+      "step": 5585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0864181887992979e-05,
+      "loss": 1.3251,
+      "step": 5586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0863960653100031e-05,
+      "loss": 1.3972,
+      "step": 5587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0863739418207082e-05,
+      "loss": 1.2872,
+      "step": 5588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0863518183314135e-05,
+      "loss": 1.4472,
+      "step": 5589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0863296948421186e-05,
+      "loss": 1.315,
+      "step": 5590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0863075713528238e-05,
+      "loss": 1.4567,
+      "step": 5591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.086285447863529e-05,
+      "loss": 1.3354,
+      "step": 5592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0862633243742343e-05,
+      "loss": 1.2872,
+      "step": 5593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0862412008849396e-05,
+      "loss": 1.3434,
+      "step": 5594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0862190773956449e-05,
+      "loss": 1.2969,
+      "step": 5595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.08619695390635e-05,
+      "loss": 1.2724,
+      "step": 5596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.086174830417055e-05,
+      "loss": 1.298,
+      "step": 5597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0861527069277603e-05,
+      "loss": 1.3415,
+      "step": 5598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0861305834384656e-05,
+      "loss": 1.4462,
+      "step": 5599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0861084599491708e-05,
+      "loss": 1.3378,
+      "step": 5600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.086086336459876e-05,
+      "loss": 1.4569,
+      "step": 5601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0860642129705812e-05,
+      "loss": 1.4168,
+      "step": 5602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0860420894812864e-05,
+      "loss": 1.5121,
+      "step": 5603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0860199659919915e-05,
+      "loss": 1.3386,
+      "step": 5604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0859978425026968e-05,
+      "loss": 1.3423,
+      "step": 5605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.085975719013402e-05,
+      "loss": 1.428,
+      "step": 5606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0859535955241072e-05,
+      "loss": 1.1507,
+      "step": 5607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0859314720348124e-05,
+      "loss": 1.4309,
+      "step": 5608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0859093485455177e-05,
+      "loss": 1.3603,
+      "step": 5609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.085887225056223e-05,
+      "loss": 1.3879,
+      "step": 5610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0858651015669282e-05,
+      "loss": 1.3877,
+      "step": 5611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0858429780776333e-05,
+      "loss": 1.3098,
+      "step": 5612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0858208545883384e-05,
+      "loss": 1.2474,
+      "step": 5613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0857987310990436e-05,
+      "loss": 1.4249,
+      "step": 5614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0857766076097489e-05,
+      "loss": 1.4836,
+      "step": 5615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0857544841204542e-05,
+      "loss": 1.3365,
+      "step": 5616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0857323606311594e-05,
+      "loss": 1.5178,
+      "step": 5617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0857102371418645e-05,
+      "loss": 1.3726,
+      "step": 5618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0856881136525698e-05,
+      "loss": 1.2437,
+      "step": 5619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0856659901632749e-05,
+      "loss": 1.3792,
+      "step": 5620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0856438666739801e-05,
+      "loss": 1.2406,
+      "step": 5621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0856217431846854e-05,
+      "loss": 1.3121,
+      "step": 5622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0855996196953905e-05,
+      "loss": 1.4052,
+      "step": 5623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0855774962060957e-05,
+      "loss": 1.4065,
+      "step": 5624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.085555372716801e-05,
+      "loss": 1.3924,
+      "step": 5625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0855332492275063e-05,
+      "loss": 1.3135,
+      "step": 5626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0855111257382115e-05,
+      "loss": 1.4604,
+      "step": 5627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0854890022489166e-05,
+      "loss": 1.3936,
+      "step": 5628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0854668787596217e-05,
+      "loss": 1.3869,
+      "step": 5629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.085444755270327e-05,
+      "loss": 1.4378,
+      "step": 5630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0854226317810322e-05,
+      "loss": 1.3974,
+      "step": 5631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0854005082917375e-05,
+      "loss": 1.3875,
+      "step": 5632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0853783848024428e-05,
+      "loss": 1.3452,
+      "step": 5633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0853562613131479e-05,
+      "loss": 1.3955,
+      "step": 5634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0853341378238531e-05,
+      "loss": 1.3008,
+      "step": 5635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0853120143345582e-05,
+      "loss": 1.4026,
+      "step": 5636
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0852898908452635e-05,
+      "loss": 1.4176,
+      "step": 5637
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0852677673559687e-05,
+      "loss": 1.2443,
+      "step": 5638
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0852456438666738e-05,
+      "loss": 1.4934,
+      "step": 5639
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.085223520377379e-05,
+      "loss": 1.4392,
+      "step": 5640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0852013968880843e-05,
+      "loss": 1.4215,
+      "step": 5641
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0851792733987896e-05,
+      "loss": 1.4311,
+      "step": 5642
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0851571499094949e-05,
+      "loss": 1.3706,
+      "step": 5643
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0851350264202e-05,
+      "loss": 1.2703,
+      "step": 5644
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.085112902930905e-05,
+      "loss": 1.3347,
+      "step": 5645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0850907794416103e-05,
+      "loss": 1.3305,
+      "step": 5646
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0850686559523156e-05,
+      "loss": 1.2629,
+      "step": 5647
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0850465324630208e-05,
+      "loss": 1.2917,
+      "step": 5648
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0850244089737261e-05,
+      "loss": 1.2149,
+      "step": 5649
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0850022854844312e-05,
+      "loss": 1.5219,
+      "step": 5650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0849801619951364e-05,
+      "loss": 1.4282,
+      "step": 5651
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0849580385058417e-05,
+      "loss": 1.282,
+      "step": 5652
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0849359150165468e-05,
+      "loss": 1.3985,
+      "step": 5653
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.084913791527252e-05,
+      "loss": 1.4002,
+      "step": 5654
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0848916680379573e-05,
+      "loss": 1.4515,
+      "step": 5655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0848695445486624e-05,
+      "loss": 1.3279,
+      "step": 5656
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0848474210593677e-05,
+      "loss": 1.3828,
+      "step": 5657
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.084825297570073e-05,
+      "loss": 1.4977,
+      "step": 5658
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0848031740807782e-05,
+      "loss": 1.3109,
+      "step": 5659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0847810505914833e-05,
+      "loss": 1.3498,
+      "step": 5660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0847589271021884e-05,
+      "loss": 1.5364,
+      "step": 5661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0847368036128936e-05,
+      "loss": 1.4199,
+      "step": 5662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0847146801235989e-05,
+      "loss": 1.3366,
+      "step": 5663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0846925566343042e-05,
+      "loss": 1.499,
+      "step": 5664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0846704331450094e-05,
+      "loss": 1.2873,
+      "step": 5665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0846483096557147e-05,
+      "loss": 1.3171,
+      "step": 5666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0846261861664198e-05,
+      "loss": 1.3682,
+      "step": 5667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.084604062677125e-05,
+      "loss": 1.2855,
+      "step": 5668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0845819391878301e-05,
+      "loss": 1.34,
+      "step": 5669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0845598156985354e-05,
+      "loss": 1.3043,
+      "step": 5670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0845376922092407e-05,
+      "loss": 1.2582,
+      "step": 5671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0845155687199457e-05,
+      "loss": 1.4126,
+      "step": 5672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.084493445230651e-05,
+      "loss": 1.314,
+      "step": 5673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0844713217413563e-05,
+      "loss": 1.3606,
+      "step": 5674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0844491982520615e-05,
+      "loss": 1.2777,
+      "step": 5675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0844270747627666e-05,
+      "loss": 1.2038,
+      "step": 5676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0844049512734719e-05,
+      "loss": 1.423,
+      "step": 5677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.084382827784177e-05,
+      "loss": 1.3458,
+      "step": 5678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0843607042948822e-05,
+      "loss": 1.3429,
+      "step": 5679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0843385808055875e-05,
+      "loss": 1.3664,
+      "step": 5680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0843164573162928e-05,
+      "loss": 1.2931,
+      "step": 5681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.084294333826998e-05,
+      "loss": 1.3514,
+      "step": 5682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0842722103377031e-05,
+      "loss": 1.338,
+      "step": 5683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0842500868484084e-05,
+      "loss": 1.4063,
+      "step": 5684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0842279633591135e-05,
+      "loss": 1.3036,
+      "step": 5685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0842058398698187e-05,
+      "loss": 1.4117,
+      "step": 5686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.084183716380524e-05,
+      "loss": 1.2984,
+      "step": 5687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0841615928912292e-05,
+      "loss": 1.4101,
+      "step": 5688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0841394694019343e-05,
+      "loss": 1.459,
+      "step": 5689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0841173459126396e-05,
+      "loss": 1.4634,
+      "step": 5690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0840952224233449e-05,
+      "loss": 1.4464,
+      "step": 5691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.08407309893405e-05,
+      "loss": 1.395,
+      "step": 5692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0840509754447552e-05,
+      "loss": 1.3204,
+      "step": 5693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0840288519554603e-05,
+      "loss": 1.1999,
+      "step": 5694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0840067284661656e-05,
+      "loss": 1.3682,
+      "step": 5695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0839846049768708e-05,
+      "loss": 1.2727,
+      "step": 5696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0839624814875761e-05,
+      "loss": 1.2749,
+      "step": 5697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0839403579982813e-05,
+      "loss": 1.4569,
+      "step": 5698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0839182345089866e-05,
+      "loss": 1.3228,
+      "step": 5699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0838961110196917e-05,
+      "loss": 1.3569,
+      "step": 5700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0838739875303968e-05,
+      "loss": 1.2458,
+      "step": 5701
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.083851864041102e-05,
+      "loss": 1.4127,
+      "step": 5702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0838297405518073e-05,
+      "loss": 1.4413,
+      "step": 5703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0838076170625126e-05,
+      "loss": 1.3306,
+      "step": 5704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0837854935732177e-05,
+      "loss": 1.4374,
+      "step": 5705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.083763370083923e-05,
+      "loss": 1.4352,
+      "step": 5706
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0837412465946282e-05,
+      "loss": 1.2683,
+      "step": 5707
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0837191231053333e-05,
+      "loss": 1.4799,
+      "step": 5708
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0836969996160385e-05,
+      "loss": 1.4154,
+      "step": 5709
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0836748761267438e-05,
+      "loss": 1.2628,
+      "step": 5710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0836527526374489e-05,
+      "loss": 1.3596,
+      "step": 5711
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0836306291481542e-05,
+      "loss": 1.3215,
+      "step": 5712
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0836085056588594e-05,
+      "loss": 1.2674,
+      "step": 5713
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0835863821695647e-05,
+      "loss": 1.4132,
+      "step": 5714
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.08356425868027e-05,
+      "loss": 1.3837,
+      "step": 5715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.083542135190975e-05,
+      "loss": 1.3556,
+      "step": 5716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0835200117016801e-05,
+      "loss": 1.3427,
+      "step": 5717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0834978882123854e-05,
+      "loss": 1.3563,
+      "step": 5718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0834757647230906e-05,
+      "loss": 1.4183,
+      "step": 5719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0834536412337959e-05,
+      "loss": 1.5915,
+      "step": 5720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0834315177445012e-05,
+      "loss": 1.4147,
+      "step": 5721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0834093942552063e-05,
+      "loss": 1.4619,
+      "step": 5722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0833872707659115e-05,
+      "loss": 1.318,
+      "step": 5723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0833651472766166e-05,
+      "loss": 1.3329,
+      "step": 5724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0833430237873219e-05,
+      "loss": 1.3297,
+      "step": 5725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0833209002980271e-05,
+      "loss": 1.3479,
+      "step": 5726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0832987768087322e-05,
+      "loss": 1.247,
+      "step": 5727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0832766533194375e-05,
+      "loss": 1.4037,
+      "step": 5728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0832545298301428e-05,
+      "loss": 1.4652,
+      "step": 5729
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.083232406340848e-05,
+      "loss": 1.3023,
+      "step": 5730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0832102828515533e-05,
+      "loss": 1.3782,
+      "step": 5731
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0831881593622584e-05,
+      "loss": 1.3431,
+      "step": 5732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0831660358729635e-05,
+      "loss": 1.3603,
+      "step": 5733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0831439123836687e-05,
+      "loss": 1.3665,
+      "step": 5734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.083121788894374e-05,
+      "loss": 1.3546,
+      "step": 5735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0830996654050792e-05,
+      "loss": 1.3575,
+      "step": 5736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0830775419157845e-05,
+      "loss": 1.2785,
+      "step": 5737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0830554184264896e-05,
+      "loss": 1.3667,
+      "step": 5738
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0830332949371949e-05,
+      "loss": 1.3766,
+      "step": 5739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0830111714479001e-05,
+      "loss": 1.3487,
+      "step": 5740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0829890479586052e-05,
+      "loss": 1.3252,
+      "step": 5741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0829669244693105e-05,
+      "loss": 1.4701,
+      "step": 5742
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0829448009800156e-05,
+      "loss": 1.2569,
+      "step": 5743
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0829226774907208e-05,
+      "loss": 1.3629,
+      "step": 5744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.082900554001426e-05,
+      "loss": 1.3548,
+      "step": 5745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0828784305121313e-05,
+      "loss": 1.4733,
+      "step": 5746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0828563070228366e-05,
+      "loss": 1.3983,
+      "step": 5747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0828341835335417e-05,
+      "loss": 1.3984,
+      "step": 5748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0828120600442468e-05,
+      "loss": 1.3786,
+      "step": 5749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.082789936554952e-05,
+      "loss": 1.4023,
+      "step": 5750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0827678130656573e-05,
+      "loss": 1.3819,
+      "step": 5751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0827456895763626e-05,
+      "loss": 1.4362,
+      "step": 5752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0827235660870678e-05,
+      "loss": 1.2183,
+      "step": 5753
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.082701442597773e-05,
+      "loss": 1.3079,
+      "step": 5754
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0826793191084782e-05,
+      "loss": 1.3802,
+      "step": 5755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0826571956191834e-05,
+      "loss": 1.422,
+      "step": 5756
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0826350721298885e-05,
+      "loss": 1.4334,
+      "step": 5757
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0826129486405938e-05,
+      "loss": 1.2267,
+      "step": 5758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.082590825151299e-05,
+      "loss": 1.3924,
+      "step": 5759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0825687016620042e-05,
+      "loss": 1.2198,
+      "step": 5760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0825465781727094e-05,
+      "loss": 1.3526,
+      "step": 5761
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0825244546834147e-05,
+      "loss": 1.3548,
+      "step": 5762
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.08250233119412e-05,
+      "loss": 1.4234,
+      "step": 5763
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.082480207704825e-05,
+      "loss": 1.3757,
+      "step": 5764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0824580842155301e-05,
+      "loss": 1.3167,
+      "step": 5765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0824359607262354e-05,
+      "loss": 1.499,
+      "step": 5766
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0824138372369406e-05,
+      "loss": 1.3179,
+      "step": 5767
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0823917137476459e-05,
+      "loss": 1.2436,
+      "step": 5768
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0823695902583512e-05,
+      "loss": 1.3352,
+      "step": 5769
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0823474667690564e-05,
+      "loss": 1.3767,
+      "step": 5770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0823253432797615e-05,
+      "loss": 1.4347,
+      "step": 5771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0823032197904668e-05,
+      "loss": 1.4032,
+      "step": 5772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0822810963011719e-05,
+      "loss": 1.3882,
+      "step": 5773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0822589728118771e-05,
+      "loss": 1.2146,
+      "step": 5774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0822368493225824e-05,
+      "loss": 1.5081,
+      "step": 5775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0822147258332875e-05,
+      "loss": 1.2667,
+      "step": 5776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0821926023439927e-05,
+      "loss": 1.4172,
+      "step": 5777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.082170478854698e-05,
+      "loss": 1.35,
+      "step": 5778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0821483553654033e-05,
+      "loss": 1.3611,
+      "step": 5779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0821262318761084e-05,
+      "loss": 1.3185,
+      "step": 5780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0821041083868136e-05,
+      "loss": 1.4075,
+      "step": 5781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0820819848975187e-05,
+      "loss": 1.3481,
+      "step": 5782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.082059861408224e-05,
+      "loss": 1.285,
+      "step": 5783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0820377379189292e-05,
+      "loss": 1.3171,
+      "step": 5784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0820156144296345e-05,
+      "loss": 1.3963,
+      "step": 5785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0819934909403398e-05,
+      "loss": 1.306,
+      "step": 5786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0819713674510449e-05,
+      "loss": 1.4356,
+      "step": 5787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0819492439617501e-05,
+      "loss": 1.4671,
+      "step": 5788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0819271204724552e-05,
+      "loss": 1.3692,
+      "step": 5789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0819049969831605e-05,
+      "loss": 1.259,
+      "step": 5790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0818828734938657e-05,
+      "loss": 1.349,
+      "step": 5791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.081860750004571e-05,
+      "loss": 1.2673,
+      "step": 5792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.081838626515276e-05,
+      "loss": 1.4802,
+      "step": 5793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0818165030259813e-05,
+      "loss": 1.315,
+      "step": 5794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0817943795366866e-05,
+      "loss": 1.236,
+      "step": 5795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0817722560473917e-05,
+      "loss": 1.3278,
+      "step": 5796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.081750132558097e-05,
+      "loss": 1.4719,
+      "step": 5797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.081728009068802e-05,
+      "loss": 1.3527,
+      "step": 5798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0817058855795073e-05,
+      "loss": 1.2734,
+      "step": 5799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0816837620902126e-05,
+      "loss": 1.343,
+      "step": 5800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0816616386009178e-05,
+      "loss": 1.2439,
+      "step": 5801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0816395151116231e-05,
+      "loss": 1.4098,
+      "step": 5802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0816173916223284e-05,
+      "loss": 1.3209,
+      "step": 5803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0815952681330334e-05,
+      "loss": 1.4445,
+      "step": 5804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0815731446437385e-05,
+      "loss": 1.2621,
+      "step": 5805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0815510211544438e-05,
+      "loss": 1.2642,
+      "step": 5806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.081528897665149e-05,
+      "loss": 1.3885,
+      "step": 5807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0815067741758543e-05,
+      "loss": 1.1993,
+      "step": 5808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0814846506865594e-05,
+      "loss": 1.2729,
+      "step": 5809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0814625271972647e-05,
+      "loss": 1.2797,
+      "step": 5810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.08144040370797e-05,
+      "loss": 1.3735,
+      "step": 5811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.081418280218675e-05,
+      "loss": 1.3876,
+      "step": 5812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0813961567293803e-05,
+      "loss": 1.4091,
+      "step": 5813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0813740332400855e-05,
+      "loss": 1.3142,
+      "step": 5814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0813519097507906e-05,
+      "loss": 1.4084,
+      "step": 5815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0813297862614959e-05,
+      "loss": 1.3346,
+      "step": 5816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0813076627722012e-05,
+      "loss": 1.409,
+      "step": 5817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0812855392829064e-05,
+      "loss": 1.4305,
+      "step": 5818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0812634157936117e-05,
+      "loss": 1.3504,
+      "step": 5819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0812412923043168e-05,
+      "loss": 1.1796,
+      "step": 5820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0812191688150219e-05,
+      "loss": 1.3622,
+      "step": 5821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0811970453257271e-05,
+      "loss": 1.3156,
+      "step": 5822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0811749218364324e-05,
+      "loss": 1.5376,
+      "step": 5823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0811527983471377e-05,
+      "loss": 1.3809,
+      "step": 5824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0811306748578427e-05,
+      "loss": 1.2515,
+      "step": 5825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.081108551368548e-05,
+      "loss": 1.346,
+      "step": 5826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0810864278792533e-05,
+      "loss": 1.4259,
+      "step": 5827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0810643043899585e-05,
+      "loss": 1.3425,
+      "step": 5828
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0810421809006636e-05,
+      "loss": 1.414,
+      "step": 5829
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0810200574113689e-05,
+      "loss": 1.3175,
+      "step": 5830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.080997933922074e-05,
+      "loss": 1.3241,
+      "step": 5831
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0809758104327792e-05,
+      "loss": 1.4846,
+      "step": 5832
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0809536869434845e-05,
+      "loss": 1.1914,
+      "step": 5833
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0809315634541898e-05,
+      "loss": 1.2938,
+      "step": 5834
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.080909439964895e-05,
+      "loss": 1.2895,
+      "step": 5835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0808873164756001e-05,
+      "loss": 1.3789,
+      "step": 5836
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0808651929863052e-05,
+      "loss": 1.3434,
+      "step": 5837
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0808430694970105e-05,
+      "loss": 1.429,
+      "step": 5838
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0808209460077157e-05,
+      "loss": 1.4025,
+      "step": 5839
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.080798822518421e-05,
+      "loss": 1.5246,
+      "step": 5840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0807766990291262e-05,
+      "loss": 1.3451,
+      "step": 5841
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0807545755398313e-05,
+      "loss": 1.334,
+      "step": 5842
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0807324520505366e-05,
+      "loss": 1.305,
+      "step": 5843
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0807103285612419e-05,
+      "loss": 1.3497,
+      "step": 5844
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.080688205071947e-05,
+      "loss": 1.2651,
+      "step": 5845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0806660815826522e-05,
+      "loss": 1.3716,
+      "step": 5846
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0806439580933573e-05,
+      "loss": 1.3621,
+      "step": 5847
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0806218346040626e-05,
+      "loss": 1.2795,
+      "step": 5848
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0805997111147678e-05,
+      "loss": 1.4628,
+      "step": 5849
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0805775876254731e-05,
+      "loss": 1.3326,
+      "step": 5850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0805554641361783e-05,
+      "loss": 1.2926,
+      "step": 5851
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0805333406468834e-05,
+      "loss": 1.4391,
+      "step": 5852
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0805112171575885e-05,
+      "loss": 1.3468,
+      "step": 5853
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0804890936682938e-05,
+      "loss": 1.2945,
+      "step": 5854
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.080466970178999e-05,
+      "loss": 1.3842,
+      "step": 5855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0804448466897043e-05,
+      "loss": 1.3373,
+      "step": 5856
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0804227232004096e-05,
+      "loss": 1.332,
+      "step": 5857
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0804005997111147e-05,
+      "loss": 1.3549,
+      "step": 5858
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.08037847622182e-05,
+      "loss": 1.343,
+      "step": 5859
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0803563527325252e-05,
+      "loss": 1.311,
+      "step": 5860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0803342292432303e-05,
+      "loss": 1.4663,
+      "step": 5861
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0803121057539355e-05,
+      "loss": 1.405,
+      "step": 5862
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0802899822646408e-05,
+      "loss": 1.3541,
+      "step": 5863
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0802678587753459e-05,
+      "loss": 1.3004,
+      "step": 5864
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0802457352860512e-05,
+      "loss": 1.3863,
+      "step": 5865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0802236117967564e-05,
+      "loss": 1.4645,
+      "step": 5866
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0802014883074617e-05,
+      "loss": 1.3995,
+      "step": 5867
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0801793648181668e-05,
+      "loss": 1.3857,
+      "step": 5868
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.080157241328872e-05,
+      "loss": 1.3975,
+      "step": 5869
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0801351178395771e-05,
+      "loss": 1.415,
+      "step": 5870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0801129943502824e-05,
+      "loss": 1.4116,
+      "step": 5871
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0800908708609876e-05,
+      "loss": 1.2305,
+      "step": 5872
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0800687473716929e-05,
+      "loss": 1.2531,
+      "step": 5873
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0800466238823982e-05,
+      "loss": 1.361,
+      "step": 5874
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0800245003931033e-05,
+      "loss": 1.2715,
+      "step": 5875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0800023769038085e-05,
+      "loss": 1.494,
+      "step": 5876
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0799802534145136e-05,
+      "loss": 1.3833,
+      "step": 5877
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0799581299252189e-05,
+      "loss": 1.324,
+      "step": 5878
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0799360064359241e-05,
+      "loss": 1.3209,
+      "step": 5879
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0799138829466292e-05,
+      "loss": 1.3516,
+      "step": 5880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0798917594573345e-05,
+      "loss": 1.5128,
+      "step": 5881
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0798696359680398e-05,
+      "loss": 1.3946,
+      "step": 5882
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.079847512478745e-05,
+      "loss": 1.5234,
+      "step": 5883
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0798253889894501e-05,
+      "loss": 1.3429,
+      "step": 5884
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0798032655001554e-05,
+      "loss": 1.4821,
+      "step": 5885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0797811420108605e-05,
+      "loss": 1.4043,
+      "step": 5886
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0797590185215657e-05,
+      "loss": 1.268,
+      "step": 5887
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.079736895032271e-05,
+      "loss": 1.4658,
+      "step": 5888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0797147715429762e-05,
+      "loss": 1.3925,
+      "step": 5889
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0796926480536815e-05,
+      "loss": 1.4045,
+      "step": 5890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0796705245643866e-05,
+      "loss": 1.4116,
+      "step": 5891
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0796484010750919e-05,
+      "loss": 1.3418,
+      "step": 5892
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.079626277585797e-05,
+      "loss": 1.2145,
+      "step": 5893
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0796041540965022e-05,
+      "loss": 1.3823,
+      "step": 5894
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0795820306072075e-05,
+      "loss": 1.4305,
+      "step": 5895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0795599071179127e-05,
+      "loss": 1.3891,
+      "step": 5896
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0795377836286178e-05,
+      "loss": 1.3635,
+      "step": 5897
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.079515660139323e-05,
+      "loss": 1.1772,
+      "step": 5898
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0794935366500283e-05,
+      "loss": 1.4166,
+      "step": 5899
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0794714131607334e-05,
+      "loss": 1.5124,
+      "step": 5900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0794492896714387e-05,
+      "loss": 1.3596,
+      "step": 5901
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0794271661821438e-05,
+      "loss": 1.3725,
+      "step": 5902
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.079405042692849e-05,
+      "loss": 1.4045,
+      "step": 5903
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0793829192035543e-05,
+      "loss": 1.3519,
+      "step": 5904
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0793607957142596e-05,
+      "loss": 1.2592,
+      "step": 5905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0793386722249648e-05,
+      "loss": 1.5062,
+      "step": 5906
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0793165487356701e-05,
+      "loss": 1.3916,
+      "step": 5907
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0792944252463752e-05,
+      "loss": 1.222,
+      "step": 5908
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0792723017570803e-05,
+      "loss": 1.3832,
+      "step": 5909
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0792501782677855e-05,
+      "loss": 1.2849,
+      "step": 5910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0792280547784908e-05,
+      "loss": 1.3594,
+      "step": 5911
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.079205931289196e-05,
+      "loss": 1.3665,
+      "step": 5912
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0791838077999012e-05,
+      "loss": 1.3636,
+      "step": 5913
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0791616843106064e-05,
+      "loss": 1.3873,
+      "step": 5914
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0791395608213117e-05,
+      "loss": 1.3767,
+      "step": 5915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.079117437332017e-05,
+      "loss": 1.1844,
+      "step": 5916
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.079095313842722e-05,
+      "loss": 1.4962,
+      "step": 5917
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0790731903534273e-05,
+      "loss": 1.2354,
+      "step": 5918
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0790510668641324e-05,
+      "loss": 1.4129,
+      "step": 5919
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0790289433748376e-05,
+      "loss": 1.5483,
+      "step": 5920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0790068198855429e-05,
+      "loss": 1.3693,
+      "step": 5921
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0789846963962482e-05,
+      "loss": 1.3667,
+      "step": 5922
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0789625729069534e-05,
+      "loss": 1.3101,
+      "step": 5923
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0789404494176585e-05,
+      "loss": 1.3872,
+      "step": 5924
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0789183259283636e-05,
+      "loss": 1.2683,
+      "step": 5925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0788962024390689e-05,
+      "loss": 1.4209,
+      "step": 5926
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0788740789497741e-05,
+      "loss": 1.3075,
+      "step": 5927
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0788519554604794e-05,
+      "loss": 1.2054,
+      "step": 5928
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0788298319711845e-05,
+      "loss": 1.2188,
+      "step": 5929
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0788077084818897e-05,
+      "loss": 1.2789,
+      "step": 5930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.078785584992595e-05,
+      "loss": 1.3875,
+      "step": 5931
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0787634615033003e-05,
+      "loss": 1.2531,
+      "step": 5932
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0787413380140054e-05,
+      "loss": 1.3299,
+      "step": 5933
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0787192145247106e-05,
+      "loss": 1.2902,
+      "step": 5934
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0786970910354157e-05,
+      "loss": 1.4542,
+      "step": 5935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.078674967546121e-05,
+      "loss": 1.4452,
+      "step": 5936
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0786528440568262e-05,
+      "loss": 1.5187,
+      "step": 5937
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0786307205675315e-05,
+      "loss": 1.4075,
+      "step": 5938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0786085970782368e-05,
+      "loss": 1.4006,
+      "step": 5939
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0785864735889418e-05,
+      "loss": 1.3038,
+      "step": 5940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.078564350099647e-05,
+      "loss": 1.3429,
+      "step": 5941
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0785422266103522e-05,
+      "loss": 1.406,
+      "step": 5942
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0785201031210575e-05,
+      "loss": 1.3118,
+      "step": 5943
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0784979796317627e-05,
+      "loss": 1.4091,
+      "step": 5944
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.078475856142468e-05,
+      "loss": 1.2075,
+      "step": 5945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.078453732653173e-05,
+      "loss": 1.4252,
+      "step": 5946
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0784316091638783e-05,
+      "loss": 1.3171,
+      "step": 5947
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0784094856745836e-05,
+      "loss": 1.2953,
+      "step": 5948
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0783873621852887e-05,
+      "loss": 1.368,
+      "step": 5949
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.078365238695994e-05,
+      "loss": 1.3473,
+      "step": 5950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.078343115206699e-05,
+      "loss": 1.3313,
+      "step": 5951
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0783209917174043e-05,
+      "loss": 1.2435,
+      "step": 5952
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0782988682281096e-05,
+      "loss": 1.3311,
+      "step": 5953
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0782767447388148e-05,
+      "loss": 1.3737,
+      "step": 5954
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0782546212495201e-05,
+      "loss": 1.4163,
+      "step": 5955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0782324977602252e-05,
+      "loss": 1.4696,
+      "step": 5956
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0782103742709304e-05,
+      "loss": 1.3595,
+      "step": 5957
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0781882507816355e-05,
+      "loss": 1.3216,
+      "step": 5958
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0781661272923408e-05,
+      "loss": 1.2185,
+      "step": 5959
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.078144003803046e-05,
+      "loss": 1.4682,
+      "step": 5960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0781218803137513e-05,
+      "loss": 1.3193,
+      "step": 5961
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0780997568244564e-05,
+      "loss": 1.4223,
+      "step": 5962
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0780776333351617e-05,
+      "loss": 1.2064,
+      "step": 5963
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.078055509845867e-05,
+      "loss": 1.2446,
+      "step": 5964
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.078033386356572e-05,
+      "loss": 1.3811,
+      "step": 5965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0780112628672773e-05,
+      "loss": 1.4852,
+      "step": 5966
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0779891393779825e-05,
+      "loss": 1.4085,
+      "step": 5967
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0779670158886876e-05,
+      "loss": 1.3074,
+      "step": 5968
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0779448923993929e-05,
+      "loss": 1.2752,
+      "step": 5969
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0779227689100982e-05,
+      "loss": 1.4079,
+      "step": 5970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0779006454208034e-05,
+      "loss": 1.2378,
+      "step": 5971
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0778785219315085e-05,
+      "loss": 1.2179,
+      "step": 5972
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0778563984422138e-05,
+      "loss": 1.3137,
+      "step": 5973
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0778342749529189e-05,
+      "loss": 1.4285,
+      "step": 5974
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0778121514636241e-05,
+      "loss": 1.3744,
+      "step": 5975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0777900279743294e-05,
+      "loss": 1.3505,
+      "step": 5976
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0777679044850346e-05,
+      "loss": 1.4467,
+      "step": 5977
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0777457809957399e-05,
+      "loss": 1.324,
+      "step": 5978
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.077723657506445e-05,
+      "loss": 1.466,
+      "step": 5979
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0777015340171503e-05,
+      "loss": 1.3986,
+      "step": 5980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0776794105278554e-05,
+      "loss": 1.2822,
+      "step": 5981
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0776572870385606e-05,
+      "loss": 1.2505,
+      "step": 5982
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0776351635492659e-05,
+      "loss": 1.252,
+      "step": 5983
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.077613040059971e-05,
+      "loss": 1.4364,
+      "step": 5984
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0775909165706762e-05,
+      "loss": 1.418,
+      "step": 5985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0775687930813815e-05,
+      "loss": 1.3319,
+      "step": 5986
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0775466695920868e-05,
+      "loss": 1.4798,
+      "step": 5987
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0775245461027918e-05,
+      "loss": 1.3124,
+      "step": 5988
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0775024226134971e-05,
+      "loss": 1.4486,
+      "step": 5989
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0774802991242022e-05,
+      "loss": 1.3075,
+      "step": 5990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0774581756349075e-05,
+      "loss": 1.3596,
+      "step": 5991
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0774360521456127e-05,
+      "loss": 1.233,
+      "step": 5992
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.077413928656318e-05,
+      "loss": 1.4751,
+      "step": 5993
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0773918051670232e-05,
+      "loss": 1.3954,
+      "step": 5994
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0773696816777283e-05,
+      "loss": 1.3666,
+      "step": 5995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0773475581884336e-05,
+      "loss": 1.3506,
+      "step": 5996
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0773254346991387e-05,
+      "loss": 1.3715,
+      "step": 5997
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.077303311209844e-05,
+      "loss": 1.3352,
+      "step": 5998
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0772811877205492e-05,
+      "loss": 1.2139,
+      "step": 5999
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0772590642312545e-05,
+      "loss": 1.3017,
+      "step": 6000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0772369407419596e-05,
+      "loss": 1.3059,
+      "step": 6001
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0772148172526648e-05,
+      "loss": 1.3295,
+      "step": 6002
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0771926937633701e-05,
+      "loss": 1.3427,
+      "step": 6003
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0771705702740753e-05,
+      "loss": 1.2222,
+      "step": 6004
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0771484467847804e-05,
+      "loss": 1.1529,
+      "step": 6005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0771263232954855e-05,
+      "loss": 1.335,
+      "step": 6006
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0771041998061908e-05,
+      "loss": 1.2699,
+      "step": 6007
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.077082076316896e-05,
+      "loss": 1.394,
+      "step": 6008
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0770599528276013e-05,
+      "loss": 1.2107,
+      "step": 6009
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0770378293383066e-05,
+      "loss": 1.3334,
+      "step": 6010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0770157058490118e-05,
+      "loss": 1.2947,
+      "step": 6011
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.076993582359717e-05,
+      "loss": 1.28,
+      "step": 6012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.076971458870422e-05,
+      "loss": 1.285,
+      "step": 6013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0769493353811273e-05,
+      "loss": 1.3992,
+      "step": 6014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0769272118918325e-05,
+      "loss": 1.4237,
+      "step": 6015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0769050884025378e-05,
+      "loss": 1.4104,
+      "step": 6016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0768829649132429e-05,
+      "loss": 1.3765,
+      "step": 6017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0768608414239482e-05,
+      "loss": 1.4356,
+      "step": 6018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0768387179346534e-05,
+      "loss": 1.3308,
+      "step": 6019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0768165944453587e-05,
+      "loss": 1.3326,
+      "step": 6020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0767944709560638e-05,
+      "loss": 1.3183,
+      "step": 6021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0767723474667689e-05,
+      "loss": 1.3259,
+      "step": 6022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0767502239774741e-05,
+      "loss": 1.2606,
+      "step": 6023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0767281004881794e-05,
+      "loss": 1.3362,
+      "step": 6024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0767059769988846e-05,
+      "loss": 1.374,
+      "step": 6025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0766838535095899e-05,
+      "loss": 1.3331,
+      "step": 6026
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0766617300202952e-05,
+      "loss": 1.4188,
+      "step": 6027
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0766396065310003e-05,
+      "loss": 1.3236,
+      "step": 6028
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0766174830417054e-05,
+      "loss": 1.414,
+      "step": 6029
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0765953595524106e-05,
+      "loss": 1.473,
+      "step": 6030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0765732360631159e-05,
+      "loss": 1.3042,
+      "step": 6031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0765511125738211e-05,
+      "loss": 1.2702,
+      "step": 6032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0765289890845262e-05,
+      "loss": 1.3166,
+      "step": 6033
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0765068655952315e-05,
+      "loss": 1.4103,
+      "step": 6034
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0764847421059367e-05,
+      "loss": 1.4159,
+      "step": 6035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.076462618616642e-05,
+      "loss": 1.3373,
+      "step": 6036
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0764404951273471e-05,
+      "loss": 1.2588,
+      "step": 6037
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0764183716380524e-05,
+      "loss": 1.4275,
+      "step": 6038
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0763962481487575e-05,
+      "loss": 1.2418,
+      "step": 6039
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0763741246594627e-05,
+      "loss": 1.2341,
+      "step": 6040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.076352001170168e-05,
+      "loss": 1.4067,
+      "step": 6041
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0763298776808732e-05,
+      "loss": 1.4359,
+      "step": 6042
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0763077541915785e-05,
+      "loss": 1.2826,
+      "step": 6043
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0762856307022836e-05,
+      "loss": 1.3103,
+      "step": 6044
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0762635072129887e-05,
+      "loss": 1.3108,
+      "step": 6045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.076241383723694e-05,
+      "loss": 1.3889,
+      "step": 6046
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0762192602343992e-05,
+      "loss": 1.2729,
+      "step": 6047
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0761971367451045e-05,
+      "loss": 1.4255,
+      "step": 6048
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0761750132558097e-05,
+      "loss": 1.3594,
+      "step": 6049
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0761528897665148e-05,
+      "loss": 1.371,
+      "step": 6050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.07613076627722e-05,
+      "loss": 1.3278,
+      "step": 6051
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0761086427879253e-05,
+      "loss": 1.3758,
+      "step": 6052
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0760865192986304e-05,
+      "loss": 1.3142,
+      "step": 6053
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0760643958093357e-05,
+      "loss": 1.4088,
+      "step": 6054
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0760422723200408e-05,
+      "loss": 1.3456,
+      "step": 6055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.076020148830746e-05,
+      "loss": 1.387,
+      "step": 6056
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0759980253414513e-05,
+      "loss": 1.3405,
+      "step": 6057
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0759759018521566e-05,
+      "loss": 1.3866,
+      "step": 6058
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0759537783628618e-05,
+      "loss": 1.3404,
+      "step": 6059
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.075931654873567e-05,
+      "loss": 1.3192,
+      "step": 6060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0759095313842722e-05,
+      "loss": 1.2946,
+      "step": 6061
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0758874078949773e-05,
+      "loss": 1.3816,
+      "step": 6062
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0758652844056825e-05,
+      "loss": 1.2353,
+      "step": 6063
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0758431609163878e-05,
+      "loss": 1.2724,
+      "step": 6064
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.075821037427093e-05,
+      "loss": 1.2853,
+      "step": 6065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0757989139377982e-05,
+      "loss": 1.3089,
+      "step": 6066
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0757767904485034e-05,
+      "loss": 1.2579,
+      "step": 6067
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0757546669592087e-05,
+      "loss": 1.4537,
+      "step": 6068
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0757325434699138e-05,
+      "loss": 1.2449,
+      "step": 6069
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.075710419980619e-05,
+      "loss": 1.3918,
+      "step": 6070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0756882964913243e-05,
+      "loss": 1.3517,
+      "step": 6071
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0756661730020294e-05,
+      "loss": 1.1967,
+      "step": 6072
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0756440495127346e-05,
+      "loss": 1.2761,
+      "step": 6073
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0756219260234399e-05,
+      "loss": 1.3336,
+      "step": 6074
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0755998025341452e-05,
+      "loss": 1.4259,
+      "step": 6075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0755776790448503e-05,
+      "loss": 1.2788,
+      "step": 6076
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0755555555555555e-05,
+      "loss": 1.3853,
+      "step": 6077
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0755334320662606e-05,
+      "loss": 1.3389,
+      "step": 6078
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0755113085769659e-05,
+      "loss": 1.3001,
+      "step": 6079
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0754891850876711e-05,
+      "loss": 1.4038,
+      "step": 6080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0754670615983764e-05,
+      "loss": 1.2227,
+      "step": 6081
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0754449381090817e-05,
+      "loss": 1.4237,
+      "step": 6082
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0754228146197867e-05,
+      "loss": 1.5045,
+      "step": 6083
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.075400691130492e-05,
+      "loss": 1.3718,
+      "step": 6084
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0753785676411971e-05,
+      "loss": 1.4331,
+      "step": 6085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0753564441519024e-05,
+      "loss": 1.3299,
+      "step": 6086
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0753343206626076e-05,
+      "loss": 1.3154,
+      "step": 6087
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0753121971733127e-05,
+      "loss": 1.1877,
+      "step": 6088
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.075290073684018e-05,
+      "loss": 1.2773,
+      "step": 6089
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0752679501947232e-05,
+      "loss": 1.265,
+      "step": 6090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0752458267054285e-05,
+      "loss": 1.3085,
+      "step": 6091
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0752237032161338e-05,
+      "loss": 1.3305,
+      "step": 6092
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0752015797268388e-05,
+      "loss": 1.3683,
+      "step": 6093
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.075179456237544e-05,
+      "loss": 1.3087,
+      "step": 6094
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0751573327482492e-05,
+      "loss": 1.3287,
+      "step": 6095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0751352092589545e-05,
+      "loss": 1.3715,
+      "step": 6096
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0751130857696597e-05,
+      "loss": 1.3351,
+      "step": 6097
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.075090962280365e-05,
+      "loss": 1.3112,
+      "step": 6098
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.07506883879107e-05,
+      "loss": 1.2918,
+      "step": 6099
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0750467153017753e-05,
+      "loss": 1.3311,
+      "step": 6100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0750245918124804e-05,
+      "loss": 1.4617,
+      "step": 6101
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0750024683231857e-05,
+      "loss": 1.3721,
+      "step": 6102
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.074980344833891e-05,
+      "loss": 1.2562,
+      "step": 6103
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0749582213445962e-05,
+      "loss": 1.549,
+      "step": 6104
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0749360978553013e-05,
+      "loss": 1.2587,
+      "step": 6105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0749139743660066e-05,
+      "loss": 1.3831,
+      "step": 6106
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0748918508767118e-05,
+      "loss": 1.2671,
+      "step": 6107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0748697273874171e-05,
+      "loss": 1.3507,
+      "step": 6108
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0748476038981222e-05,
+      "loss": 1.3564,
+      "step": 6109
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0748254804088273e-05,
+      "loss": 1.3026,
+      "step": 6110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0748033569195325e-05,
+      "loss": 1.4389,
+      "step": 6111
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0747812334302378e-05,
+      "loss": 1.3135,
+      "step": 6112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.074759109940943e-05,
+      "loss": 1.2998,
+      "step": 6113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0747369864516483e-05,
+      "loss": 1.2708,
+      "step": 6114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0747148629623536e-05,
+      "loss": 1.2413,
+      "step": 6115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0746927394730587e-05,
+      "loss": 1.3242,
+      "step": 6116
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0746706159837638e-05,
+      "loss": 1.3093,
+      "step": 6117
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.074648492494469e-05,
+      "loss": 1.4624,
+      "step": 6118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0746263690051743e-05,
+      "loss": 1.333,
+      "step": 6119
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0746042455158795e-05,
+      "loss": 1.3337,
+      "step": 6120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0745821220265846e-05,
+      "loss": 1.367,
+      "step": 6121
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0745599985372899e-05,
+      "loss": 1.2775,
+      "step": 6122
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0745378750479952e-05,
+      "loss": 1.4428,
+      "step": 6123
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0745157515587004e-05,
+      "loss": 1.1755,
+      "step": 6124
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0744936280694055e-05,
+      "loss": 1.3085,
+      "step": 6125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0744715045801106e-05,
+      "loss": 1.3849,
+      "step": 6126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0744493810908159e-05,
+      "loss": 1.2865,
+      "step": 6127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0744272576015211e-05,
+      "loss": 1.266,
+      "step": 6128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0744051341122264e-05,
+      "loss": 1.4467,
+      "step": 6129
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0743830106229316e-05,
+      "loss": 1.3816,
+      "step": 6130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0743608871336369e-05,
+      "loss": 1.4744,
+      "step": 6131
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.074338763644342e-05,
+      "loss": 1.3703,
+      "step": 6132
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0743166401550471e-05,
+      "loss": 1.3442,
+      "step": 6133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0742945166657524e-05,
+      "loss": 1.3507,
+      "step": 6134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0742723931764576e-05,
+      "loss": 1.4124,
+      "step": 6135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0742502696871629e-05,
+      "loss": 1.2789,
+      "step": 6136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.074228146197868e-05,
+      "loss": 1.2513,
+      "step": 6137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0742060227085732e-05,
+      "loss": 1.3814,
+      "step": 6138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0741838992192785e-05,
+      "loss": 1.2356,
+      "step": 6139
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0741617757299838e-05,
+      "loss": 1.3897,
+      "step": 6140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0741396522406888e-05,
+      "loss": 1.2491,
+      "step": 6141
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0741175287513941e-05,
+      "loss": 1.4148,
+      "step": 6142
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0740954052620992e-05,
+      "loss": 1.4302,
+      "step": 6143
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0740732817728045e-05,
+      "loss": 1.5123,
+      "step": 6144
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0740511582835097e-05,
+      "loss": 1.3721,
+      "step": 6145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.074029034794215e-05,
+      "loss": 1.2983,
+      "step": 6146
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0740069113049202e-05,
+      "loss": 1.4063,
+      "step": 6147
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0739847878156253e-05,
+      "loss": 1.4178,
+      "step": 6148
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0739626643263306e-05,
+      "loss": 1.3174,
+      "step": 6149
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0739405408370357e-05,
+      "loss": 1.28,
+      "step": 6150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.073918417347741e-05,
+      "loss": 1.3611,
+      "step": 6151
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0738962938584462e-05,
+      "loss": 1.2797,
+      "step": 6152
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0738741703691515e-05,
+      "loss": 1.2747,
+      "step": 6153
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0738520468798566e-05,
+      "loss": 1.3337,
+      "step": 6154
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0738299233905618e-05,
+      "loss": 1.3672,
+      "step": 6155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.073807799901267e-05,
+      "loss": 1.1365,
+      "step": 6156
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0737856764119722e-05,
+      "loss": 1.3178,
+      "step": 6157
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0737635529226774e-05,
+      "loss": 1.2842,
+      "step": 6158
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0737414294333825e-05,
+      "loss": 1.3524,
+      "step": 6159
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0737193059440878e-05,
+      "loss": 1.3463,
+      "step": 6160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.073697182454793e-05,
+      "loss": 1.3047,
+      "step": 6161
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0736750589654983e-05,
+      "loss": 1.3558,
+      "step": 6162
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0736529354762036e-05,
+      "loss": 1.3949,
+      "step": 6163
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0736308119869087e-05,
+      "loss": 1.4277,
+      "step": 6164
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.073608688497614e-05,
+      "loss": 1.408,
+      "step": 6165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.073586565008319e-05,
+      "loss": 1.4051,
+      "step": 6166
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0735644415190243e-05,
+      "loss": 1.381,
+      "step": 6167
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0735423180297295e-05,
+      "loss": 1.4133,
+      "step": 6168
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0735201945404348e-05,
+      "loss": 1.4705,
+      "step": 6169
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0734980710511399e-05,
+      "loss": 1.432,
+      "step": 6170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0734759475618452e-05,
+      "loss": 1.4399,
+      "step": 6171
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0734538240725504e-05,
+      "loss": 1.2916,
+      "step": 6172
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0734317005832555e-05,
+      "loss": 1.3254,
+      "step": 6173
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0734095770939608e-05,
+      "loss": 1.2463,
+      "step": 6174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.073387453604666e-05,
+      "loss": 1.315,
+      "step": 6175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0733653301153711e-05,
+      "loss": 1.3286,
+      "step": 6176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0733432066260764e-05,
+      "loss": 1.4038,
+      "step": 6177
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0733210831367816e-05,
+      "loss": 1.3714,
+      "step": 6178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0732989596474869e-05,
+      "loss": 1.2924,
+      "step": 6179
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0732768361581922e-05,
+      "loss": 1.3403,
+      "step": 6180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0732547126688973e-05,
+      "loss": 1.3401,
+      "step": 6181
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0732325891796023e-05,
+      "loss": 1.3506,
+      "step": 6182
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0732104656903076e-05,
+      "loss": 1.3514,
+      "step": 6183
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0731883422010129e-05,
+      "loss": 1.3783,
+      "step": 6184
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0731662187117181e-05,
+      "loss": 1.2221,
+      "step": 6185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0731440952224234e-05,
+      "loss": 1.297,
+      "step": 6186
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0731219717331285e-05,
+      "loss": 1.4989,
+      "step": 6187
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0730998482438337e-05,
+      "loss": 1.4489,
+      "step": 6188
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0730777247545388e-05,
+      "loss": 1.2638,
+      "step": 6189
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0730556012652441e-05,
+      "loss": 1.3617,
+      "step": 6190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0730334777759494e-05,
+      "loss": 1.2959,
+      "step": 6191
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0730113542866545e-05,
+      "loss": 1.4377,
+      "step": 6192
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0729892307973597e-05,
+      "loss": 1.2453,
+      "step": 6193
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.072967107308065e-05,
+      "loss": 1.4054,
+      "step": 6194
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0729449838187702e-05,
+      "loss": 1.2803,
+      "step": 6195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0729228603294755e-05,
+      "loss": 1.2868,
+      "step": 6196
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0729007368401806e-05,
+      "loss": 1.3293,
+      "step": 6197
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0728786133508857e-05,
+      "loss": 1.5788,
+      "step": 6198
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.072856489861591e-05,
+      "loss": 1.3171,
+      "step": 6199
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0728343663722962e-05,
+      "loss": 1.2714,
+      "step": 6200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0728122428830015e-05,
+      "loss": 1.3994,
+      "step": 6201
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0727901193937067e-05,
+      "loss": 1.2584,
+      "step": 6202
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0727679959044118e-05,
+      "loss": 1.3904,
+      "step": 6203
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.072745872415117e-05,
+      "loss": 1.3456,
+      "step": 6204
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0727237489258222e-05,
+      "loss": 1.4997,
+      "step": 6205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0727016254365274e-05,
+      "loss": 1.3057,
+      "step": 6206
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0726795019472327e-05,
+      "loss": 1.3994,
+      "step": 6207
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.072657378457938e-05,
+      "loss": 1.3387,
+      "step": 6208
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.072635254968643e-05,
+      "loss": 1.3036,
+      "step": 6209
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0726131314793483e-05,
+      "loss": 1.4235,
+      "step": 6210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0725910079900536e-05,
+      "loss": 1.4567,
+      "step": 6211
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0725688845007588e-05,
+      "loss": 1.3664,
+      "step": 6212
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.072546761011464e-05,
+      "loss": 1.2418,
+      "step": 6213
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.072524637522169e-05,
+      "loss": 1.3899,
+      "step": 6214
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0725025140328743e-05,
+      "loss": 1.2239,
+      "step": 6215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0724803905435795e-05,
+      "loss": 1.41,
+      "step": 6216
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0724582670542848e-05,
+      "loss": 1.3652,
+      "step": 6217
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.07243614356499e-05,
+      "loss": 1.4423,
+      "step": 6218
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0724140200756952e-05,
+      "loss": 1.307,
+      "step": 6219
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0723918965864004e-05,
+      "loss": 1.3557,
+      "step": 6220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0723697730971055e-05,
+      "loss": 1.3145,
+      "step": 6221
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0723476496078108e-05,
+      "loss": 1.4855,
+      "step": 6222
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.072325526118516e-05,
+      "loss": 1.3198,
+      "step": 6223
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0723034026292213e-05,
+      "loss": 1.3201,
+      "step": 6224
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0722812791399264e-05,
+      "loss": 1.3314,
+      "step": 6225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0722591556506316e-05,
+      "loss": 1.2536,
+      "step": 6226
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0722370321613369e-05,
+      "loss": 1.3883,
+      "step": 6227
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0722149086720422e-05,
+      "loss": 1.2072,
+      "step": 6228
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0721927851827473e-05,
+      "loss": 1.3384,
+      "step": 6229
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0721706616934523e-05,
+      "loss": 1.281,
+      "step": 6230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0721485382041576e-05,
+      "loss": 1.4852,
+      "step": 6231
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0721264147148629e-05,
+      "loss": 1.2453,
+      "step": 6232
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0721042912255681e-05,
+      "loss": 1.4091,
+      "step": 6233
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0720821677362734e-05,
+      "loss": 1.3103,
+      "step": 6234
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0720600442469787e-05,
+      "loss": 1.4713,
+      "step": 6235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0720379207576837e-05,
+      "loss": 1.4007,
+      "step": 6236
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.072015797268389e-05,
+      "loss": 1.3752,
+      "step": 6237
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0719936737790941e-05,
+      "loss": 1.3572,
+      "step": 6238
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0719715502897994e-05,
+      "loss": 1.3621,
+      "step": 6239
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0719494268005046e-05,
+      "loss": 1.238,
+      "step": 6240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0719273033112097e-05,
+      "loss": 1.4412,
+      "step": 6241
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.071905179821915e-05,
+      "loss": 1.169,
+      "step": 6242
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0718830563326202e-05,
+      "loss": 1.3222,
+      "step": 6243
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0718609328433255e-05,
+      "loss": 1.3846,
+      "step": 6244
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0718388093540306e-05,
+      "loss": 1.3881,
+      "step": 6245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0718166858647358e-05,
+      "loss": 1.4224,
+      "step": 6246
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.071794562375441e-05,
+      "loss": 1.3969,
+      "step": 6247
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0717724388861462e-05,
+      "loss": 1.3107,
+      "step": 6248
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0717503153968515e-05,
+      "loss": 1.2178,
+      "step": 6249
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0717281919075567e-05,
+      "loss": 1.3718,
+      "step": 6250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.071706068418262e-05,
+      "loss": 1.3673,
+      "step": 6251
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.071683944928967e-05,
+      "loss": 1.3381,
+      "step": 6252
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0716618214396723e-05,
+      "loss": 1.3539,
+      "step": 6253
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0716396979503774e-05,
+      "loss": 1.3927,
+      "step": 6254
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0716175744610827e-05,
+      "loss": 1.5512,
+      "step": 6255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.071595450971788e-05,
+      "loss": 1.1763,
+      "step": 6256
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0715733274824932e-05,
+      "loss": 1.4468,
+      "step": 6257
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0715512039931983e-05,
+      "loss": 1.3882,
+      "step": 6258
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0715290805039036e-05,
+      "loss": 1.3075,
+      "step": 6259
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0715069570146088e-05,
+      "loss": 1.2867,
+      "step": 6260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.071484833525314e-05,
+      "loss": 1.2364,
+      "step": 6261
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0714627100360192e-05,
+      "loss": 1.4274,
+      "step": 6262
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0714405865467243e-05,
+      "loss": 1.3219,
+      "step": 6263
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0714184630574295e-05,
+      "loss": 1.304,
+      "step": 6264
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0713963395681348e-05,
+      "loss": 1.3511,
+      "step": 6265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.07137421607884e-05,
+      "loss": 1.3423,
+      "step": 6266
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0713520925895453e-05,
+      "loss": 1.2753,
+      "step": 6267
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0713299691002506e-05,
+      "loss": 1.2625,
+      "step": 6268
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0713078456109557e-05,
+      "loss": 1.3285,
+      "step": 6269
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0712857221216608e-05,
+      "loss": 1.271,
+      "step": 6270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.071263598632366e-05,
+      "loss": 1.4305,
+      "step": 6271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0712414751430713e-05,
+      "loss": 1.4439,
+      "step": 6272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0712193516537765e-05,
+      "loss": 1.2226,
+      "step": 6273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0711972281644816e-05,
+      "loss": 1.2706,
+      "step": 6274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0711751046751869e-05,
+      "loss": 1.3691,
+      "step": 6275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0711529811858922e-05,
+      "loss": 1.295,
+      "step": 6276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0711308576965972e-05,
+      "loss": 1.3636,
+      "step": 6277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0711087342073025e-05,
+      "loss": 1.2779,
+      "step": 6278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0710866107180078e-05,
+      "loss": 1.1778,
+      "step": 6279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0710644872287129e-05,
+      "loss": 1.3176,
+      "step": 6280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0710423637394181e-05,
+      "loss": 1.4306,
+      "step": 6281
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0710202402501234e-05,
+      "loss": 1.1839,
+      "step": 6282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0709981167608286e-05,
+      "loss": 1.3495,
+      "step": 6283
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0709759932715339e-05,
+      "loss": 1.3991,
+      "step": 6284
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.070953869782239e-05,
+      "loss": 1.3223,
+      "step": 6285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0709317462929441e-05,
+      "loss": 1.2651,
+      "step": 6286
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0709096228036494e-05,
+      "loss": 1.218,
+      "step": 6287
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0708874993143546e-05,
+      "loss": 1.2769,
+      "step": 6288
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0708653758250599e-05,
+      "loss": 1.3176,
+      "step": 6289
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0708432523357651e-05,
+      "loss": 1.2671,
+      "step": 6290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0708211288464702e-05,
+      "loss": 1.3272,
+      "step": 6291
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0707990053571755e-05,
+      "loss": 1.1935,
+      "step": 6292
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0707768818678806e-05,
+      "loss": 1.2615,
+      "step": 6293
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0707547583785858e-05,
+      "loss": 1.4304,
+      "step": 6294
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0707326348892911e-05,
+      "loss": 1.3267,
+      "step": 6295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0707105113999962e-05,
+      "loss": 1.3775,
+      "step": 6296
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0706883879107015e-05,
+      "loss": 1.2191,
+      "step": 6297
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0706662644214067e-05,
+      "loss": 1.2835,
+      "step": 6298
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.070644140932112e-05,
+      "loss": 1.3211,
+      "step": 6299
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0706220174428172e-05,
+      "loss": 1.3243,
+      "step": 6300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0705998939535223e-05,
+      "loss": 1.3018,
+      "step": 6301
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0705777704642274e-05,
+      "loss": 1.2616,
+      "step": 6302
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0705556469749327e-05,
+      "loss": 1.3255,
+      "step": 6303
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.070533523485638e-05,
+      "loss": 1.4627,
+      "step": 6304
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0705113999963432e-05,
+      "loss": 1.2935,
+      "step": 6305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0704892765070485e-05,
+      "loss": 1.4645,
+      "step": 6306
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0704671530177536e-05,
+      "loss": 1.2959,
+      "step": 6307
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0704450295284588e-05,
+      "loss": 1.4649,
+      "step": 6308
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0704229060391639e-05,
+      "loss": 1.3999,
+      "step": 6309
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0704007825498692e-05,
+      "loss": 1.4542,
+      "step": 6310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0703786590605744e-05,
+      "loss": 1.3602,
+      "step": 6311
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0703565355712795e-05,
+      "loss": 1.2121,
+      "step": 6312
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0703344120819848e-05,
+      "loss": 1.4257,
+      "step": 6313
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.07031228859269e-05,
+      "loss": 1.215,
+      "step": 6314
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0702901651033953e-05,
+      "loss": 1.4994,
+      "step": 6315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0702680416141006e-05,
+      "loss": 1.3909,
+      "step": 6316
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0702459181248057e-05,
+      "loss": 1.404,
+      "step": 6317
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0702237946355108e-05,
+      "loss": 1.3586,
+      "step": 6318
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.070201671146216e-05,
+      "loss": 1.3172,
+      "step": 6319
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0701795476569213e-05,
+      "loss": 1.3394,
+      "step": 6320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0701574241676265e-05,
+      "loss": 1.3823,
+      "step": 6321
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0701353006783318e-05,
+      "loss": 1.3888,
+      "step": 6322
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0701131771890369e-05,
+      "loss": 1.2888,
+      "step": 6323
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0700910536997422e-05,
+      "loss": 1.2589,
+      "step": 6324
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0700689302104474e-05,
+      "loss": 1.2928,
+      "step": 6325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0700468067211525e-05,
+      "loss": 1.2671,
+      "step": 6326
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0700246832318578e-05,
+      "loss": 1.2074,
+      "step": 6327
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.070002559742563e-05,
+      "loss": 1.3192,
+      "step": 6328
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0699804362532681e-05,
+      "loss": 1.2381,
+      "step": 6329
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0699583127639734e-05,
+      "loss": 1.3695,
+      "step": 6330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0699361892746786e-05,
+      "loss": 1.3079,
+      "step": 6331
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0699140657853839e-05,
+      "loss": 1.4394,
+      "step": 6332
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.069891942296089e-05,
+      "loss": 1.4855,
+      "step": 6333
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0698698188067941e-05,
+      "loss": 1.2334,
+      "step": 6334
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0698476953174993e-05,
+      "loss": 1.2595,
+      "step": 6335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0698255718282046e-05,
+      "loss": 1.4051,
+      "step": 6336
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0698034483389099e-05,
+      "loss": 1.2809,
+      "step": 6337
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0697813248496151e-05,
+      "loss": 1.392,
+      "step": 6338
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0697592013603204e-05,
+      "loss": 1.3321,
+      "step": 6339
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0697370778710255e-05,
+      "loss": 1.1529,
+      "step": 6340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0697149543817307e-05,
+      "loss": 1.4041,
+      "step": 6341
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0696928308924358e-05,
+      "loss": 1.2791,
+      "step": 6342
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0696707074031411e-05,
+      "loss": 1.4443,
+      "step": 6343
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0696485839138464e-05,
+      "loss": 1.437,
+      "step": 6344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0696264604245515e-05,
+      "loss": 1.249,
+      "step": 6345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0696043369352567e-05,
+      "loss": 1.3152,
+      "step": 6346
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.069582213445962e-05,
+      "loss": 1.3697,
+      "step": 6347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0695600899566672e-05,
+      "loss": 1.4123,
+      "step": 6348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0695379664673723e-05,
+      "loss": 1.3202,
+      "step": 6349
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0695158429780776e-05,
+      "loss": 1.2924,
+      "step": 6350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0694937194887827e-05,
+      "loss": 1.2838,
+      "step": 6351
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.069471595999488e-05,
+      "loss": 1.3399,
+      "step": 6352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0694494725101932e-05,
+      "loss": 1.3343,
+      "step": 6353
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0694273490208985e-05,
+      "loss": 1.3504,
+      "step": 6354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0694052255316037e-05,
+      "loss": 1.3007,
+      "step": 6355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0693831020423088e-05,
+      "loss": 1.2918,
+      "step": 6356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.069360978553014e-05,
+      "loss": 1.4587,
+      "step": 6357
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0693388550637192e-05,
+      "loss": 1.1566,
+      "step": 6358
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0693167315744244e-05,
+      "loss": 1.3712,
+      "step": 6359
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0692946080851297e-05,
+      "loss": 1.3676,
+      "step": 6360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.069272484595835e-05,
+      "loss": 1.4095,
+      "step": 6361
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.06925036110654e-05,
+      "loss": 1.5119,
+      "step": 6362
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0692282376172453e-05,
+      "loss": 1.2364,
+      "step": 6363
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0692061141279506e-05,
+      "loss": 1.3699,
+      "step": 6364
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0691839906386557e-05,
+      "loss": 1.2532,
+      "step": 6365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.069161867149361e-05,
+      "loss": 1.2825,
+      "step": 6366
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.069139743660066e-05,
+      "loss": 1.3115,
+      "step": 6367
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0691176201707713e-05,
+      "loss": 1.3707,
+      "step": 6368
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0690954966814765e-05,
+      "loss": 1.3857,
+      "step": 6369
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0690733731921818e-05,
+      "loss": 1.3144,
+      "step": 6370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.069051249702887e-05,
+      "loss": 1.3153,
+      "step": 6371
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0690291262135923e-05,
+      "loss": 1.3145,
+      "step": 6372
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0690070027242974e-05,
+      "loss": 1.2935,
+      "step": 6373
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0689848792350025e-05,
+      "loss": 1.3042,
+      "step": 6374
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0689627557457078e-05,
+      "loss": 1.4193,
+      "step": 6375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.068940632256413e-05,
+      "loss": 1.5207,
+      "step": 6376
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0689185087671183e-05,
+      "loss": 1.3526,
+      "step": 6377
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0688963852778234e-05,
+      "loss": 1.2883,
+      "step": 6378
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0688742617885286e-05,
+      "loss": 1.2948,
+      "step": 6379
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0688521382992339e-05,
+      "loss": 1.4895,
+      "step": 6380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.068830014809939e-05,
+      "loss": 1.3015,
+      "step": 6381
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0688078913206443e-05,
+      "loss": 1.4539,
+      "step": 6382
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0687857678313495e-05,
+      "loss": 1.2735,
+      "step": 6383
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0687636443420546e-05,
+      "loss": 1.3668,
+      "step": 6384
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0687415208527599e-05,
+      "loss": 1.3144,
+      "step": 6385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0687193973634651e-05,
+      "loss": 1.3373,
+      "step": 6386
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0686972738741704e-05,
+      "loss": 1.3818,
+      "step": 6387
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0686751503848757e-05,
+      "loss": 1.3598,
+      "step": 6388
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0686530268955807e-05,
+      "loss": 1.3601,
+      "step": 6389
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0686309034062858e-05,
+      "loss": 1.2962,
+      "step": 6390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0686087799169911e-05,
+      "loss": 1.3305,
+      "step": 6391
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0685866564276964e-05,
+      "loss": 1.2898,
+      "step": 6392
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0685645329384016e-05,
+      "loss": 1.3524,
+      "step": 6393
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0685424094491069e-05,
+      "loss": 1.3094,
+      "step": 6394
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.068520285959812e-05,
+      "loss": 1.1993,
+      "step": 6395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0684981624705172e-05,
+      "loss": 1.2635,
+      "step": 6396
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0684760389812223e-05,
+      "loss": 1.3358,
+      "step": 6397
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0684539154919276e-05,
+      "loss": 1.3472,
+      "step": 6398
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0684317920026328e-05,
+      "loss": 1.3098,
+      "step": 6399
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.068409668513338e-05,
+      "loss": 1.398,
+      "step": 6400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0683875450240432e-05,
+      "loss": 1.1655,
+      "step": 6401
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0683654215347485e-05,
+      "loss": 1.303,
+      "step": 6402
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0683432980454537e-05,
+      "loss": 1.3509,
+      "step": 6403
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.068321174556159e-05,
+      "loss": 1.2964,
+      "step": 6404
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.068299051066864e-05,
+      "loss": 1.4457,
+      "step": 6405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0682769275775692e-05,
+      "loss": 1.2736,
+      "step": 6406
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0682548040882744e-05,
+      "loss": 1.4962,
+      "step": 6407
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0682326805989797e-05,
+      "loss": 1.3008,
+      "step": 6408
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.068210557109685e-05,
+      "loss": 1.4766,
+      "step": 6409
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0681884336203902e-05,
+      "loss": 1.449,
+      "step": 6410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0681663101310953e-05,
+      "loss": 1.3151,
+      "step": 6411
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0681441866418006e-05,
+      "loss": 1.3573,
+      "step": 6412
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0681220631525058e-05,
+      "loss": 1.2851,
+      "step": 6413
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.068099939663211e-05,
+      "loss": 1.4574,
+      "step": 6414
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0680778161739162e-05,
+      "loss": 1.3145,
+      "step": 6415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0680556926846213e-05,
+      "loss": 1.2814,
+      "step": 6416
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0680335691953265e-05,
+      "loss": 1.2136,
+      "step": 6417
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0680114457060318e-05,
+      "loss": 1.3622,
+      "step": 6418
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.067989322216737e-05,
+      "loss": 1.3563,
+      "step": 6419
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0679671987274423e-05,
+      "loss": 1.2605,
+      "step": 6420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0679450752381474e-05,
+      "loss": 1.3231,
+      "step": 6421
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0679229517488525e-05,
+      "loss": 1.4482,
+      "step": 6422
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0679008282595578e-05,
+      "loss": 1.3122,
+      "step": 6423
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.067878704770263e-05,
+      "loss": 1.2977,
+      "step": 6424
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0678565812809683e-05,
+      "loss": 1.2296,
+      "step": 6425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0678344577916735e-05,
+      "loss": 1.3592,
+      "step": 6426
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0678123343023786e-05,
+      "loss": 1.392,
+      "step": 6427
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0677902108130839e-05,
+      "loss": 1.2839,
+      "step": 6428
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0677680873237892e-05,
+      "loss": 1.4581,
+      "step": 6429
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0677459638344942e-05,
+      "loss": 1.3055,
+      "step": 6430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0677238403451995e-05,
+      "loss": 1.4311,
+      "step": 6431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0677017168559048e-05,
+      "loss": 1.2104,
+      "step": 6432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0676795933666099e-05,
+      "loss": 1.3822,
+      "step": 6433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0676574698773151e-05,
+      "loss": 1.3762,
+      "step": 6434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0676353463880204e-05,
+      "loss": 1.3133,
+      "step": 6435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0676132228987256e-05,
+      "loss": 1.2438,
+      "step": 6436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0675910994094307e-05,
+      "loss": 1.3824,
+      "step": 6437
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0675689759201358e-05,
+      "loss": 1.3077,
+      "step": 6438
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0675468524308411e-05,
+      "loss": 1.3777,
+      "step": 6439
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0675247289415464e-05,
+      "loss": 1.3227,
+      "step": 6440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0675026054522516e-05,
+      "loss": 1.3971,
+      "step": 6441
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0674804819629569e-05,
+      "loss": 1.473,
+      "step": 6442
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0674583584736621e-05,
+      "loss": 1.4732,
+      "step": 6443
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0674362349843672e-05,
+      "loss": 1.1977,
+      "step": 6444
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0674141114950725e-05,
+      "loss": 1.5011,
+      "step": 6445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0673919880057776e-05,
+      "loss": 1.3408,
+      "step": 6446
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0673698645164828e-05,
+      "loss": 1.279,
+      "step": 6447
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0673477410271881e-05,
+      "loss": 1.2942,
+      "step": 6448
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0673256175378932e-05,
+      "loss": 1.3098,
+      "step": 6449
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0673034940485985e-05,
+      "loss": 1.3112,
+      "step": 6450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0672813705593037e-05,
+      "loss": 1.2994,
+      "step": 6451
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.067259247070009e-05,
+      "loss": 1.7217,
+      "step": 6452
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.067237123580714e-05,
+      "loss": 1.3341,
+      "step": 6453
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0672150000914193e-05,
+      "loss": 1.2359,
+      "step": 6454
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0671928766021244e-05,
+      "loss": 1.2622,
+      "step": 6455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0671707531128297e-05,
+      "loss": 1.4249,
+      "step": 6456
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.067148629623535e-05,
+      "loss": 1.3648,
+      "step": 6457
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0671265061342402e-05,
+      "loss": 1.3648,
+      "step": 6458
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0671043826449455e-05,
+      "loss": 1.325,
+      "step": 6459
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0670822591556506e-05,
+      "loss": 1.2739,
+      "step": 6460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0670601356663558e-05,
+      "loss": 1.3504,
+      "step": 6461
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0670380121770609e-05,
+      "loss": 1.3647,
+      "step": 6462
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0670158886877662e-05,
+      "loss": 1.3935,
+      "step": 6463
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0669937651984714e-05,
+      "loss": 1.2941,
+      "step": 6464
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0669716417091767e-05,
+      "loss": 1.4376,
+      "step": 6465
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0669495182198818e-05,
+      "loss": 1.3602,
+      "step": 6466
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.066927394730587e-05,
+      "loss": 1.4787,
+      "step": 6467
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0669052712412923e-05,
+      "loss": 1.3378,
+      "step": 6468
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0668831477519974e-05,
+      "loss": 1.353,
+      "step": 6469
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0668610242627027e-05,
+      "loss": 1.383,
+      "step": 6470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0668389007734078e-05,
+      "loss": 1.3176,
+      "step": 6471
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.066816777284113e-05,
+      "loss": 1.3326,
+      "step": 6472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0667946537948183e-05,
+      "loss": 1.3276,
+      "step": 6473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0667725303055235e-05,
+      "loss": 1.3991,
+      "step": 6474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0667504068162288e-05,
+      "loss": 1.2356,
+      "step": 6475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.066728283326934e-05,
+      "loss": 1.2956,
+      "step": 6476
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0667061598376392e-05,
+      "loss": 1.2936,
+      "step": 6477
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0666840363483442e-05,
+      "loss": 1.5605,
+      "step": 6478
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0666619128590495e-05,
+      "loss": 1.3315,
+      "step": 6479
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0666397893697548e-05,
+      "loss": 1.5049,
+      "step": 6480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.06661766588046e-05,
+      "loss": 1.2083,
+      "step": 6481
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0665955423911651e-05,
+      "loss": 1.3215,
+      "step": 6482
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0665734189018704e-05,
+      "loss": 1.388,
+      "step": 6483
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0665512954125756e-05,
+      "loss": 1.4256,
+      "step": 6484
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0665291719232807e-05,
+      "loss": 1.3773,
+      "step": 6485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.066507048433986e-05,
+      "loss": 1.3682,
+      "step": 6486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0664849249446913e-05,
+      "loss": 1.2998,
+      "step": 6487
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0664628014553963e-05,
+      "loss": 1.4394,
+      "step": 6488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0664406779661016e-05,
+      "loss": 1.4249,
+      "step": 6489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0664185544768069e-05,
+      "loss": 1.3898,
+      "step": 6490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0663964309875121e-05,
+      "loss": 1.2758,
+      "step": 6491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0663743074982174e-05,
+      "loss": 1.2686,
+      "step": 6492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0663521840089225e-05,
+      "loss": 1.3687,
+      "step": 6493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0663300605196276e-05,
+      "loss": 1.3652,
+      "step": 6494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0663079370303328e-05,
+      "loss": 1.3005,
+      "step": 6495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0662858135410381e-05,
+      "loss": 1.225,
+      "step": 6496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0662636900517434e-05,
+      "loss": 1.2336,
+      "step": 6497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0662415665624486e-05,
+      "loss": 1.2504,
+      "step": 6498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0662194430731537e-05,
+      "loss": 1.3663,
+      "step": 6499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.066197319583859e-05,
+      "loss": 1.3105,
+      "step": 6500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0661751960945642e-05,
+      "loss": 1.2811,
+      "step": 6501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0661530726052693e-05,
+      "loss": 1.2423,
+      "step": 6502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0661309491159746e-05,
+      "loss": 1.3976,
+      "step": 6503
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0661088256266797e-05,
+      "loss": 1.2335,
+      "step": 6504
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.066086702137385e-05,
+      "loss": 1.2942,
+      "step": 6505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0660645786480902e-05,
+      "loss": 1.4196,
+      "step": 6506
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0660424551587955e-05,
+      "loss": 1.3921,
+      "step": 6507
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0660203316695007e-05,
+      "loss": 1.3842,
+      "step": 6508
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0659982081802058e-05,
+      "loss": 1.4276,
+      "step": 6509
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0659760846909109e-05,
+      "loss": 1.4276,
+      "step": 6510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0659539612016162e-05,
+      "loss": 1.4302,
+      "step": 6511
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0659318377123214e-05,
+      "loss": 1.3141,
+      "step": 6512
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0659097142230267e-05,
+      "loss": 1.2957,
+      "step": 6513
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.065887590733732e-05,
+      "loss": 1.3589,
+      "step": 6514
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.065865467244437e-05,
+      "loss": 1.3472,
+      "step": 6515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0658433437551423e-05,
+      "loss": 1.3938,
+      "step": 6516
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0658212202658476e-05,
+      "loss": 1.3681,
+      "step": 6517
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0657990967765527e-05,
+      "loss": 1.3605,
+      "step": 6518
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.065776973287258e-05,
+      "loss": 1.2519,
+      "step": 6519
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.065754849797963e-05,
+      "loss": 1.428,
+      "step": 6520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0657327263086683e-05,
+      "loss": 1.4101,
+      "step": 6521
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0657106028193735e-05,
+      "loss": 1.3892,
+      "step": 6522
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0656884793300788e-05,
+      "loss": 1.3173,
+      "step": 6523
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.065666355840784e-05,
+      "loss": 1.3397,
+      "step": 6524
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0656442323514891e-05,
+      "loss": 1.2734,
+      "step": 6525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0656221088621942e-05,
+      "loss": 1.3263,
+      "step": 6526
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0655999853728995e-05,
+      "loss": 1.3899,
+      "step": 6527
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0655778618836048e-05,
+      "loss": 1.42,
+      "step": 6528
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.06555573839431e-05,
+      "loss": 1.4086,
+      "step": 6529
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0655336149050153e-05,
+      "loss": 1.3718,
+      "step": 6530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0655114914157204e-05,
+      "loss": 1.4952,
+      "step": 6531
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0654893679264256e-05,
+      "loss": 1.3312,
+      "step": 6532
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0654672444371309e-05,
+      "loss": 1.27,
+      "step": 6533
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.065445120947836e-05,
+      "loss": 1.3164,
+      "step": 6534
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0654229974585413e-05,
+      "loss": 1.3592,
+      "step": 6535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0654008739692465e-05,
+      "loss": 1.3069,
+      "step": 6536
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0653787504799516e-05,
+      "loss": 1.3281,
+      "step": 6537
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0653566269906569e-05,
+      "loss": 1.1801,
+      "step": 6538
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0653345035013621e-05,
+      "loss": 1.434,
+      "step": 6539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0653123800120674e-05,
+      "loss": 1.2387,
+      "step": 6540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0652902565227725e-05,
+      "loss": 1.2867,
+      "step": 6541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0652681330334776e-05,
+      "loss": 1.2672,
+      "step": 6542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0652460095441828e-05,
+      "loss": 1.1672,
+      "step": 6543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0652238860548881e-05,
+      "loss": 1.4799,
+      "step": 6544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0652017625655934e-05,
+      "loss": 1.5225,
+      "step": 6545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0651796390762986e-05,
+      "loss": 1.2825,
+      "step": 6546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0651575155870039e-05,
+      "loss": 1.2502,
+      "step": 6547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.065135392097709e-05,
+      "loss": 1.3067,
+      "step": 6548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0651132686084142e-05,
+      "loss": 1.4202,
+      "step": 6549
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0650911451191193e-05,
+      "loss": 1.1654,
+      "step": 6550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0650690216298246e-05,
+      "loss": 1.3649,
+      "step": 6551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0650468981405298e-05,
+      "loss": 1.3478,
+      "step": 6552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.065024774651235e-05,
+      "loss": 1.4575,
+      "step": 6553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0650026511619402e-05,
+      "loss": 1.4595,
+      "step": 6554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0649805276726455e-05,
+      "loss": 1.2791,
+      "step": 6555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0649584041833507e-05,
+      "loss": 1.4051,
+      "step": 6556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0649362806940558e-05,
+      "loss": 1.4195,
+      "step": 6557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.064914157204761e-05,
+      "loss": 1.2685,
+      "step": 6558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0648920337154662e-05,
+      "loss": 1.2975,
+      "step": 6559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0648699102261714e-05,
+      "loss": 1.2628,
+      "step": 6560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0648477867368767e-05,
+      "loss": 1.3025,
+      "step": 6561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.064825663247582e-05,
+      "loss": 1.282,
+      "step": 6562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0648035397582872e-05,
+      "loss": 1.3486,
+      "step": 6563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0647814162689923e-05,
+      "loss": 1.2322,
+      "step": 6564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0647592927796976e-05,
+      "loss": 1.2858,
+      "step": 6565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0647371692904027e-05,
+      "loss": 1.3406,
+      "step": 6566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0647150458011079e-05,
+      "loss": 1.3284,
+      "step": 6567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0646929223118132e-05,
+      "loss": 1.3457,
+      "step": 6568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0646707988225184e-05,
+      "loss": 1.4163,
+      "step": 6569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0646486753332235e-05,
+      "loss": 1.3879,
+      "step": 6570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0646265518439288e-05,
+      "loss": 1.3369,
+      "step": 6571
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.064604428354634e-05,
+      "loss": 1.3447,
+      "step": 6572
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0645823048653391e-05,
+      "loss": 1.3326,
+      "step": 6573
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0645601813760444e-05,
+      "loss": 1.3526,
+      "step": 6574
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0645380578867495e-05,
+      "loss": 1.2982,
+      "step": 6575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0645159343974548e-05,
+      "loss": 1.4236,
+      "step": 6576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.06449381090816e-05,
+      "loss": 1.3125,
+      "step": 6577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0644716874188653e-05,
+      "loss": 1.2861,
+      "step": 6578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0644495639295705e-05,
+      "loss": 1.3017,
+      "step": 6579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0644274404402758e-05,
+      "loss": 1.2961,
+      "step": 6580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0644053169509809e-05,
+      "loss": 1.2788,
+      "step": 6581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.064383193461686e-05,
+      "loss": 1.252,
+      "step": 6582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0643610699723912e-05,
+      "loss": 1.3414,
+      "step": 6583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0643389464830965e-05,
+      "loss": 1.4301,
+      "step": 6584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0643168229938018e-05,
+      "loss": 1.2235,
+      "step": 6585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0642946995045069e-05,
+      "loss": 1.3423,
+      "step": 6586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0642725760152121e-05,
+      "loss": 1.2192,
+      "step": 6587
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0642504525259174e-05,
+      "loss": 1.3292,
+      "step": 6588
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0642283290366226e-05,
+      "loss": 1.2706,
+      "step": 6589
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0642062055473277e-05,
+      "loss": 1.3621,
+      "step": 6590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.064184082058033e-05,
+      "loss": 1.3913,
+      "step": 6591
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0641619585687381e-05,
+      "loss": 1.3593,
+      "step": 6592
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0641398350794434e-05,
+      "loss": 1.3622,
+      "step": 6593
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0641177115901486e-05,
+      "loss": 1.3161,
+      "step": 6594
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0640955881008539e-05,
+      "loss": 1.2644,
+      "step": 6595
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0640734646115591e-05,
+      "loss": 1.367,
+      "step": 6596
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0640513411222642e-05,
+      "loss": 1.4018,
+      "step": 6597
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0640292176329693e-05,
+      "loss": 1.3318,
+      "step": 6598
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0640070941436746e-05,
+      "loss": 1.3063,
+      "step": 6599
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0639849706543798e-05,
+      "loss": 1.3921,
+      "step": 6600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0639628471650851e-05,
+      "loss": 1.292,
+      "step": 6601
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0639407236757904e-05,
+      "loss": 1.3566,
+      "step": 6602
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0639186001864955e-05,
+      "loss": 1.2551,
+      "step": 6603
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0638964766972007e-05,
+      "loss": 1.2666,
+      "step": 6604
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.063874353207906e-05,
+      "loss": 1.4501,
+      "step": 6605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.063852229718611e-05,
+      "loss": 1.4005,
+      "step": 6606
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0638301062293163e-05,
+      "loss": 1.3946,
+      "step": 6607
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0638079827400214e-05,
+      "loss": 1.2838,
+      "step": 6608
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0637858592507267e-05,
+      "loss": 1.4063,
+      "step": 6609
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.063763735761432e-05,
+      "loss": 1.3577,
+      "step": 6610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0637416122721372e-05,
+      "loss": 1.3807,
+      "step": 6611
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0637194887828425e-05,
+      "loss": 1.4432,
+      "step": 6612
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0636973652935476e-05,
+      "loss": 1.313,
+      "step": 6613
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0636752418042526e-05,
+      "loss": 1.3214,
+      "step": 6614
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0636531183149579e-05,
+      "loss": 1.4979,
+      "step": 6615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0636309948256632e-05,
+      "loss": 1.3859,
+      "step": 6616
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0636088713363684e-05,
+      "loss": 1.4729,
+      "step": 6617
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0635867478470737e-05,
+      "loss": 1.5933,
+      "step": 6618
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0635646243577788e-05,
+      "loss": 1.3409,
+      "step": 6619
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.063542500868484e-05,
+      "loss": 1.263,
+      "step": 6620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0635203773791893e-05,
+      "loss": 1.3072,
+      "step": 6621
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0634982538898944e-05,
+      "loss": 1.4082,
+      "step": 6622
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0634761304005997e-05,
+      "loss": 1.3317,
+      "step": 6623
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0634540069113048e-05,
+      "loss": 1.3103,
+      "step": 6624
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.06343188342201e-05,
+      "loss": 1.3091,
+      "step": 6625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0634097599327153e-05,
+      "loss": 1.2631,
+      "step": 6626
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0633876364434205e-05,
+      "loss": 1.3929,
+      "step": 6627
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0633655129541258e-05,
+      "loss": 1.2211,
+      "step": 6628
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0633433894648309e-05,
+      "loss": 1.2994,
+      "step": 6629
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.063321265975536e-05,
+      "loss": 1.3572,
+      "step": 6630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0632991424862412e-05,
+      "loss": 1.303,
+      "step": 6631
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0632770189969465e-05,
+      "loss": 1.2756,
+      "step": 6632
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0632548955076518e-05,
+      "loss": 1.3194,
+      "step": 6633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.063232772018357e-05,
+      "loss": 1.4218,
+      "step": 6634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0632106485290621e-05,
+      "loss": 1.4107,
+      "step": 6635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0631885250397674e-05,
+      "loss": 1.3203,
+      "step": 6636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0631664015504726e-05,
+      "loss": 1.4519,
+      "step": 6637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0631442780611777e-05,
+      "loss": 1.3739,
+      "step": 6638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.063122154571883e-05,
+      "loss": 1.4321,
+      "step": 6639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0631000310825883e-05,
+      "loss": 1.3094,
+      "step": 6640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0630779075932933e-05,
+      "loss": 1.3023,
+      "step": 6641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0630557841039986e-05,
+      "loss": 1.3153,
+      "step": 6642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0630336606147039e-05,
+      "loss": 1.365,
+      "step": 6643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0630115371254091e-05,
+      "loss": 1.2767,
+      "step": 6644
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0629894136361142e-05,
+      "loss": 1.2578,
+      "step": 6645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0629672901468195e-05,
+      "loss": 1.3898,
+      "step": 6646
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0629451666575246e-05,
+      "loss": 1.3807,
+      "step": 6647
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0629230431682298e-05,
+      "loss": 1.3513,
+      "step": 6648
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0629009196789351e-05,
+      "loss": 1.3068,
+      "step": 6649
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0628787961896404e-05,
+      "loss": 1.2194,
+      "step": 6650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0628566727003456e-05,
+      "loss": 1.4453,
+      "step": 6651
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0628345492110507e-05,
+      "loss": 1.2361,
+      "step": 6652
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.062812425721756e-05,
+      "loss": 1.4355,
+      "step": 6653
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.062790302232461e-05,
+      "loss": 1.2566,
+      "step": 6654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0627681787431663e-05,
+      "loss": 1.3497,
+      "step": 6655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0627460552538716e-05,
+      "loss": 1.3443,
+      "step": 6656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0627239317645767e-05,
+      "loss": 1.5503,
+      "step": 6657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.062701808275282e-05,
+      "loss": 1.2691,
+      "step": 6658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0626796847859872e-05,
+      "loss": 1.3256,
+      "step": 6659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0626575612966925e-05,
+      "loss": 1.3756,
+      "step": 6660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0626354378073976e-05,
+      "loss": 1.2352,
+      "step": 6661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0626133143181028e-05,
+      "loss": 1.2439,
+      "step": 6662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0625911908288079e-05,
+      "loss": 1.3323,
+      "step": 6663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0625690673395132e-05,
+      "loss": 1.3528,
+      "step": 6664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0625469438502184e-05,
+      "loss": 1.2921,
+      "step": 6665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0625248203609237e-05,
+      "loss": 1.2414,
+      "step": 6666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.062502696871629e-05,
+      "loss": 1.3269,
+      "step": 6667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.062480573382334e-05,
+      "loss": 1.4042,
+      "step": 6668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0624584498930393e-05,
+      "loss": 1.4391,
+      "step": 6669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0624363264037444e-05,
+      "loss": 1.295,
+      "step": 6670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0624142029144497e-05,
+      "loss": 1.3878,
+      "step": 6671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.062392079425155e-05,
+      "loss": 1.227,
+      "step": 6672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0623699559358602e-05,
+      "loss": 1.2021,
+      "step": 6673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0623478324465653e-05,
+      "loss": 1.3344,
+      "step": 6674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0623257089572705e-05,
+      "loss": 1.3256,
+      "step": 6675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0623035854679758e-05,
+      "loss": 1.2924,
+      "step": 6676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.062281461978681e-05,
+      "loss": 1.467,
+      "step": 6677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0622593384893861e-05,
+      "loss": 1.2903,
+      "step": 6678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0622372150000912e-05,
+      "loss": 1.4313,
+      "step": 6679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0622150915107965e-05,
+      "loss": 1.3725,
+      "step": 6680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0621929680215018e-05,
+      "loss": 1.3308,
+      "step": 6681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.062170844532207e-05,
+      "loss": 1.4696,
+      "step": 6682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0621487210429123e-05,
+      "loss": 1.3452,
+      "step": 6683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0621265975536175e-05,
+      "loss": 1.4585,
+      "step": 6684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0621044740643226e-05,
+      "loss": 1.3879,
+      "step": 6685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0620823505750277e-05,
+      "loss": 1.3125,
+      "step": 6686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.062060227085733e-05,
+      "loss": 1.3431,
+      "step": 6687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0620381035964383e-05,
+      "loss": 1.4313,
+      "step": 6688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0620159801071435e-05,
+      "loss": 1.4105,
+      "step": 6689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0619938566178486e-05,
+      "loss": 1.424,
+      "step": 6690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0619717331285539e-05,
+      "loss": 1.3032,
+      "step": 6691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0619496096392591e-05,
+      "loss": 1.3174,
+      "step": 6692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0619274861499644e-05,
+      "loss": 1.3466,
+      "step": 6693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0619053626606695e-05,
+      "loss": 1.4059,
+      "step": 6694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0618832391713747e-05,
+      "loss": 1.2099,
+      "step": 6695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0618611156820798e-05,
+      "loss": 1.2808,
+      "step": 6696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0618389921927851e-05,
+      "loss": 1.4228,
+      "step": 6697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0618168687034904e-05,
+      "loss": 1.3424,
+      "step": 6698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0617947452141956e-05,
+      "loss": 1.3568,
+      "step": 6699
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0617726217249009e-05,
+      "loss": 1.2367,
+      "step": 6700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.061750498235606e-05,
+      "loss": 1.3406,
+      "step": 6701
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.061728374746311e-05,
+      "loss": 1.5161,
+      "step": 6702
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0617062512570163e-05,
+      "loss": 1.3371,
+      "step": 6703
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0616841277677216e-05,
+      "loss": 1.3038,
+      "step": 6704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0616620042784268e-05,
+      "loss": 1.349,
+      "step": 6705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.061639880789132e-05,
+      "loss": 1.2522,
+      "step": 6706
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0616177572998372e-05,
+      "loss": 1.3993,
+      "step": 6707
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0615956338105425e-05,
+      "loss": 1.2587,
+      "step": 6708
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0615735103212477e-05,
+      "loss": 1.3922,
+      "step": 6709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0615513868319528e-05,
+      "loss": 1.3171,
+      "step": 6710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.061529263342658e-05,
+      "loss": 1.2985,
+      "step": 6711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0615071398533632e-05,
+      "loss": 1.3287,
+      "step": 6712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0614850163640684e-05,
+      "loss": 1.4014,
+      "step": 6713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0614628928747737e-05,
+      "loss": 1.3343,
+      "step": 6714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.061440769385479e-05,
+      "loss": 1.3981,
+      "step": 6715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0614186458961842e-05,
+      "loss": 1.4496,
+      "step": 6716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0613965224068893e-05,
+      "loss": 1.3143,
+      "step": 6717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0613743989175944e-05,
+      "loss": 1.2476,
+      "step": 6718
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0613522754282997e-05,
+      "loss": 1.3504,
+      "step": 6719
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0613301519390049e-05,
+      "loss": 1.3585,
+      "step": 6720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0613080284497102e-05,
+      "loss": 1.3916,
+      "step": 6721
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0612859049604154e-05,
+      "loss": 1.2869,
+      "step": 6722
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0612637814711205e-05,
+      "loss": 1.4995,
+      "step": 6723
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0612416579818258e-05,
+      "loss": 1.2595,
+      "step": 6724
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.061219534492531e-05,
+      "loss": 1.2882,
+      "step": 6725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0611974110032361e-05,
+      "loss": 1.3351,
+      "step": 6726
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0611752875139414e-05,
+      "loss": 1.2843,
+      "step": 6727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0611531640246465e-05,
+      "loss": 1.3081,
+      "step": 6728
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0611310405353518e-05,
+      "loss": 1.2691,
+      "step": 6729
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.061108917046057e-05,
+      "loss": 1.3816,
+      "step": 6730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0610867935567623e-05,
+      "loss": 1.2489,
+      "step": 6731
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0610646700674675e-05,
+      "loss": 1.4324,
+      "step": 6732
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0610425465781726e-05,
+      "loss": 1.4399,
+      "step": 6733
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0610204230888779e-05,
+      "loss": 1.3784,
+      "step": 6734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.060998299599583e-05,
+      "loss": 1.231,
+      "step": 6735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0609761761102882e-05,
+      "loss": 1.2178,
+      "step": 6736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0609540526209935e-05,
+      "loss": 1.5034,
+      "step": 6737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0609319291316988e-05,
+      "loss": 1.4935,
+      "step": 6738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0609098056424039e-05,
+      "loss": 1.293,
+      "step": 6739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0608876821531091e-05,
+      "loss": 1.3106,
+      "step": 6740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0608655586638144e-05,
+      "loss": 1.3888,
+      "step": 6741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0608434351745195e-05,
+      "loss": 1.4256,
+      "step": 6742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0608213116852247e-05,
+      "loss": 1.3161,
+      "step": 6743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.06079918819593e-05,
+      "loss": 1.2813,
+      "step": 6744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0607770647066351e-05,
+      "loss": 1.2767,
+      "step": 6745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0607549412173403e-05,
+      "loss": 1.4168,
+      "step": 6746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0607328177280456e-05,
+      "loss": 1.3157,
+      "step": 6747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0607106942387509e-05,
+      "loss": 1.4317,
+      "step": 6748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.060688570749456e-05,
+      "loss": 1.2677,
+      "step": 6749
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0606664472601612e-05,
+      "loss": 1.3367,
+      "step": 6750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0606443237708663e-05,
+      "loss": 1.3349,
+      "step": 6751
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0606222002815716e-05,
+      "loss": 1.3793,
+      "step": 6752
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0606000767922768e-05,
+      "loss": 1.2883,
+      "step": 6753
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0605779533029821e-05,
+      "loss": 1.4526,
+      "step": 6754
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0605558298136874e-05,
+      "loss": 1.3268,
+      "step": 6755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0605337063243925e-05,
+      "loss": 1.3619,
+      "step": 6756
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0605115828350977e-05,
+      "loss": 1.3605,
+      "step": 6757
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0604894593458028e-05,
+      "loss": 1.2736,
+      "step": 6758
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.060467335856508e-05,
+      "loss": 1.2293,
+      "step": 6759
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0604452123672133e-05,
+      "loss": 1.3095,
+      "step": 6760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0604230888779184e-05,
+      "loss": 1.3053,
+      "step": 6761
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0604009653886237e-05,
+      "loss": 1.2802,
+      "step": 6762
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.060378841899329e-05,
+      "loss": 1.4717,
+      "step": 6763
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0603567184100342e-05,
+      "loss": 1.276,
+      "step": 6764
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0603345949207395e-05,
+      "loss": 1.4859,
+      "step": 6765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0603124714314446e-05,
+      "loss": 1.3453,
+      "step": 6766
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0602903479421496e-05,
+      "loss": 1.3192,
+      "step": 6767
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0602682244528549e-05,
+      "loss": 1.2354,
+      "step": 6768
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0602461009635602e-05,
+      "loss": 1.3482,
+      "step": 6769
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0602239774742654e-05,
+      "loss": 1.1688,
+      "step": 6770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0602018539849707e-05,
+      "loss": 1.3991,
+      "step": 6771
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0601797304956758e-05,
+      "loss": 1.3941,
+      "step": 6772
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.060157607006381e-05,
+      "loss": 1.3898,
+      "step": 6773
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0601354835170861e-05,
+      "loss": 1.3236,
+      "step": 6774
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0601133600277914e-05,
+      "loss": 1.3506,
+      "step": 6775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0600912365384967e-05,
+      "loss": 1.27,
+      "step": 6776
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.060069113049202e-05,
+      "loss": 1.2692,
+      "step": 6777
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.060046989559907e-05,
+      "loss": 1.3802,
+      "step": 6778
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0600248660706123e-05,
+      "loss": 1.3792,
+      "step": 6779
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0600027425813175e-05,
+      "loss": 1.5122,
+      "step": 6780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0599806190920228e-05,
+      "loss": 1.3386,
+      "step": 6781
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0599584956027279e-05,
+      "loss": 1.3635,
+      "step": 6782
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.059936372113433e-05,
+      "loss": 1.3404,
+      "step": 6783
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0599142486241382e-05,
+      "loss": 1.3465,
+      "step": 6784
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0598921251348435e-05,
+      "loss": 1.284,
+      "step": 6785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0598700016455488e-05,
+      "loss": 1.5408,
+      "step": 6786
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.059847878156254e-05,
+      "loss": 1.3596,
+      "step": 6787
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0598257546669593e-05,
+      "loss": 1.4605,
+      "step": 6788
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0598036311776644e-05,
+      "loss": 1.2347,
+      "step": 6789
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0597815076883695e-05,
+      "loss": 1.2838,
+      "step": 6790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0597593841990747e-05,
+      "loss": 1.3763,
+      "step": 6791
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.05973726070978e-05,
+      "loss": 1.3326,
+      "step": 6792
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0597151372204853e-05,
+      "loss": 1.2281,
+      "step": 6793
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0596930137311903e-05,
+      "loss": 1.4125,
+      "step": 6794
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0596708902418956e-05,
+      "loss": 1.38,
+      "step": 6795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0596487667526009e-05,
+      "loss": 1.2117,
+      "step": 6796
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0596266432633061e-05,
+      "loss": 1.2865,
+      "step": 6797
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0596045197740112e-05,
+      "loss": 1.2703,
+      "step": 6798
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0595823962847165e-05,
+      "loss": 1.4622,
+      "step": 6799
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0595602727954216e-05,
+      "loss": 1.2763,
+      "step": 6800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0595381493061268e-05,
+      "loss": 1.4204,
+      "step": 6801
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0595160258168321e-05,
+      "loss": 1.4083,
+      "step": 6802
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0594939023275374e-05,
+      "loss": 1.3351,
+      "step": 6803
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0594717788382426e-05,
+      "loss": 1.2536,
+      "step": 6804
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0594496553489477e-05,
+      "loss": 1.3631,
+      "step": 6805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0594275318596528e-05,
+      "loss": 1.2298,
+      "step": 6806
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.059405408370358e-05,
+      "loss": 1.394,
+      "step": 6807
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0593832848810633e-05,
+      "loss": 1.2964,
+      "step": 6808
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0593611613917686e-05,
+      "loss": 1.3399,
+      "step": 6809
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0593390379024737e-05,
+      "loss": 1.3018,
+      "step": 6810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.059316914413179e-05,
+      "loss": 1.3682,
+      "step": 6811
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0592947909238842e-05,
+      "loss": 1.2683,
+      "step": 6812
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0592726674345895e-05,
+      "loss": 1.324,
+      "step": 6813
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0592505439452946e-05,
+      "loss": 1.2244,
+      "step": 6814
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0592284204559998e-05,
+      "loss": 1.3244,
+      "step": 6815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0592062969667049e-05,
+      "loss": 1.2204,
+      "step": 6816
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0591841734774102e-05,
+      "loss": 1.255,
+      "step": 6817
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0591620499881154e-05,
+      "loss": 1.3334,
+      "step": 6818
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0591399264988207e-05,
+      "loss": 1.2767,
+      "step": 6819
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.059117803009526e-05,
+      "loss": 1.2919,
+      "step": 6820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.059095679520231e-05,
+      "loss": 1.3997,
+      "step": 6821
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0590735560309363e-05,
+      "loss": 1.2677,
+      "step": 6822
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0590514325416414e-05,
+      "loss": 1.3366,
+      "step": 6823
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0590293090523467e-05,
+      "loss": 1.3427,
+      "step": 6824
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.059007185563052e-05,
+      "loss": 1.2726,
+      "step": 6825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0589850620737572e-05,
+      "loss": 1.2894,
+      "step": 6826
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0589629385844623e-05,
+      "loss": 1.352,
+      "step": 6827
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0589408150951675e-05,
+      "loss": 1.2967,
+      "step": 6828
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0589186916058728e-05,
+      "loss": 1.3086,
+      "step": 6829
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0588965681165779e-05,
+      "loss": 1.3775,
+      "step": 6830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0588744446272831e-05,
+      "loss": 1.339,
+      "step": 6831
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0588523211379882e-05,
+      "loss": 1.3345,
+      "step": 6832
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0588301976486935e-05,
+      "loss": 1.3717,
+      "step": 6833
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0588080741593988e-05,
+      "loss": 1.2346,
+      "step": 6834
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.058785950670104e-05,
+      "loss": 1.3435,
+      "step": 6835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0587638271808093e-05,
+      "loss": 1.3245,
+      "step": 6836
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0587417036915144e-05,
+      "loss": 1.2486,
+      "step": 6837
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0587195802022196e-05,
+      "loss": 1.5894,
+      "step": 6838
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0586974567129247e-05,
+      "loss": 1.2743,
+      "step": 6839
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.05867533322363e-05,
+      "loss": 1.3425,
+      "step": 6840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0586532097343352e-05,
+      "loss": 1.3586,
+      "step": 6841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0586310862450405e-05,
+      "loss": 1.2814,
+      "step": 6842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0586089627557456e-05,
+      "loss": 1.2329,
+      "step": 6843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0585868392664509e-05,
+      "loss": 1.4343,
+      "step": 6844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0585647157771561e-05,
+      "loss": 1.2798,
+      "step": 6845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0585425922878612e-05,
+      "loss": 1.4679,
+      "step": 6846
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0585204687985665e-05,
+      "loss": 1.3409,
+      "step": 6847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0584983453092717e-05,
+      "loss": 1.3209,
+      "step": 6848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0584762218199768e-05,
+      "loss": 1.2353,
+      "step": 6849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0584540983306821e-05,
+      "loss": 1.2955,
+      "step": 6850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0584319748413874e-05,
+      "loss": 1.3387,
+      "step": 6851
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0584098513520926e-05,
+      "loss": 1.2665,
+      "step": 6852
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0583877278627979e-05,
+      "loss": 1.2806,
+      "step": 6853
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.058365604373503e-05,
+      "loss": 1.4391,
+      "step": 6854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.058343480884208e-05,
+      "loss": 1.2784,
+      "step": 6855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0583213573949133e-05,
+      "loss": 1.2604,
+      "step": 6856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0582992339056186e-05,
+      "loss": 1.3579,
+      "step": 6857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0582771104163238e-05,
+      "loss": 1.3054,
+      "step": 6858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0582549869270291e-05,
+      "loss": 1.461,
+      "step": 6859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0582328634377342e-05,
+      "loss": 1.4125,
+      "step": 6860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0582107399484395e-05,
+      "loss": 1.3379,
+      "step": 6861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0581886164591445e-05,
+      "loss": 1.4321,
+      "step": 6862
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0581664929698498e-05,
+      "loss": 1.4295,
+      "step": 6863
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.058144369480555e-05,
+      "loss": 1.2645,
+      "step": 6864
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0581222459912602e-05,
+      "loss": 1.2588,
+      "step": 6865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0581001225019654e-05,
+      "loss": 1.2228,
+      "step": 6866
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0580779990126707e-05,
+      "loss": 1.323,
+      "step": 6867
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.058055875523376e-05,
+      "loss": 1.274,
+      "step": 6868
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0580337520340812e-05,
+      "loss": 1.2885,
+      "step": 6869
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0580116285447863e-05,
+      "loss": 1.3678,
+      "step": 6870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0579895050554914e-05,
+      "loss": 1.3816,
+      "step": 6871
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0579673815661967e-05,
+      "loss": 1.2092,
+      "step": 6872
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0579452580769019e-05,
+      "loss": 1.3102,
+      "step": 6873
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0579231345876072e-05,
+      "loss": 1.2009,
+      "step": 6874
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0579010110983124e-05,
+      "loss": 1.3976,
+      "step": 6875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0578788876090175e-05,
+      "loss": 1.1772,
+      "step": 6876
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0578567641197228e-05,
+      "loss": 1.3315,
+      "step": 6877
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0578346406304279e-05,
+      "loss": 1.4309,
+      "step": 6878
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0578125171411331e-05,
+      "loss": 1.3052,
+      "step": 6879
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0577903936518384e-05,
+      "loss": 1.4813,
+      "step": 6880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0577682701625437e-05,
+      "loss": 1.3718,
+      "step": 6881
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0577461466732488e-05,
+      "loss": 1.3061,
+      "step": 6882
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.057724023183954e-05,
+      "loss": 1.4223,
+      "step": 6883
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0577018996946593e-05,
+      "loss": 1.4212,
+      "step": 6884
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0576797762053645e-05,
+      "loss": 1.4999,
+      "step": 6885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0576576527160696e-05,
+      "loss": 1.2836,
+      "step": 6886
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0576355292267747e-05,
+      "loss": 1.2726,
+      "step": 6887
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.05761340573748e-05,
+      "loss": 1.1471,
+      "step": 6888
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0575912822481852e-05,
+      "loss": 1.3133,
+      "step": 6889
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0575691587588905e-05,
+      "loss": 1.5237,
+      "step": 6890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0575470352695958e-05,
+      "loss": 1.439,
+      "step": 6891
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.057524911780301e-05,
+      "loss": 1.3193,
+      "step": 6892
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0575027882910061e-05,
+      "loss": 1.2431,
+      "step": 6893
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0574806648017112e-05,
+      "loss": 1.3918,
+      "step": 6894
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0574585413124165e-05,
+      "loss": 1.2484,
+      "step": 6895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0574364178231217e-05,
+      "loss": 1.2527,
+      "step": 6896
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.057414294333827e-05,
+      "loss": 1.6012,
+      "step": 6897
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0573921708445321e-05,
+      "loss": 1.2102,
+      "step": 6898
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0573700473552373e-05,
+      "loss": 1.4685,
+      "step": 6899
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0573479238659426e-05,
+      "loss": 1.3128,
+      "step": 6900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0573258003766479e-05,
+      "loss": 1.3183,
+      "step": 6901
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.057303676887353e-05,
+      "loss": 1.3789,
+      "step": 6902
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.057281553398058e-05,
+      "loss": 1.2715,
+      "step": 6903
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0572594299087633e-05,
+      "loss": 1.3174,
+      "step": 6904
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0572373064194686e-05,
+      "loss": 1.4777,
+      "step": 6905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0572151829301738e-05,
+      "loss": 1.3334,
+      "step": 6906
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0571930594408791e-05,
+      "loss": 1.2568,
+      "step": 6907
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0571709359515844e-05,
+      "loss": 1.333,
+      "step": 6908
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0571488124622895e-05,
+      "loss": 1.256,
+      "step": 6909
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0571266889729947e-05,
+      "loss": 1.2792,
+      "step": 6910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0571045654836998e-05,
+      "loss": 1.293,
+      "step": 6911
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.057082441994405e-05,
+      "loss": 1.1474,
+      "step": 6912
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0570603185051103e-05,
+      "loss": 1.2384,
+      "step": 6913
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0570381950158154e-05,
+      "loss": 1.2227,
+      "step": 6914
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0570160715265207e-05,
+      "loss": 1.2834,
+      "step": 6915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.056993948037226e-05,
+      "loss": 1.3244,
+      "step": 6916
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0569718245479312e-05,
+      "loss": 1.3434,
+      "step": 6917
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0569497010586363e-05,
+      "loss": 1.2592,
+      "step": 6918
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0569275775693416e-05,
+      "loss": 1.3883,
+      "step": 6919
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0569054540800466e-05,
+      "loss": 1.3824,
+      "step": 6920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0568833305907519e-05,
+      "loss": 1.2327,
+      "step": 6921
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0568612071014572e-05,
+      "loss": 1.2215,
+      "step": 6922
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0568390836121624e-05,
+      "loss": 1.4343,
+      "step": 6923
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0568169601228677e-05,
+      "loss": 1.4823,
+      "step": 6924
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0567948366335728e-05,
+      "loss": 1.3701,
+      "step": 6925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.056772713144278e-05,
+      "loss": 1.3868,
+      "step": 6926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0567505896549831e-05,
+      "loss": 1.4633,
+      "step": 6927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0567284661656884e-05,
+      "loss": 1.4052,
+      "step": 6928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0567063426763937e-05,
+      "loss": 1.3035,
+      "step": 6929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.056684219187099e-05,
+      "loss": 1.2985,
+      "step": 6930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.056662095697804e-05,
+      "loss": 1.3442,
+      "step": 6931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0566399722085093e-05,
+      "loss": 1.3561,
+      "step": 6932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0566178487192145e-05,
+      "loss": 1.2968,
+      "step": 6933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0565957252299196e-05,
+      "loss": 1.44,
+      "step": 6934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0565736017406249e-05,
+      "loss": 1.2682,
+      "step": 6935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.05655147825133e-05,
+      "loss": 1.4685,
+      "step": 6936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0565293547620352e-05,
+      "loss": 1.2562,
+      "step": 6937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0565072312727405e-05,
+      "loss": 1.3095,
+      "step": 6938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0564851077834458e-05,
+      "loss": 1.3607,
+      "step": 6939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.056462984294151e-05,
+      "loss": 1.302,
+      "step": 6940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0564408608048563e-05,
+      "loss": 1.3358,
+      "step": 6941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0564187373155614e-05,
+      "loss": 1.2261,
+      "step": 6942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0563966138262665e-05,
+      "loss": 1.3868,
+      "step": 6943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0563744903369717e-05,
+      "loss": 1.344,
+      "step": 6944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.056352366847677e-05,
+      "loss": 1.1638,
+      "step": 6945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0563302433583823e-05,
+      "loss": 1.3547,
+      "step": 6946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0563081198690873e-05,
+      "loss": 1.3476,
+      "step": 6947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0562859963797926e-05,
+      "loss": 1.3555,
+      "step": 6948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0562638728904979e-05,
+      "loss": 1.3451,
+      "step": 6949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.056241749401203e-05,
+      "loss": 1.3488,
+      "step": 6950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0562196259119082e-05,
+      "loss": 1.2983,
+      "step": 6951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0561975024226135e-05,
+      "loss": 1.2316,
+      "step": 6952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0561753789333186e-05,
+      "loss": 1.2529,
+      "step": 6953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0561532554440238e-05,
+      "loss": 1.3054,
+      "step": 6954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0561311319547291e-05,
+      "loss": 1.3648,
+      "step": 6955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0561090084654344e-05,
+      "loss": 1.2764,
+      "step": 6956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0560868849761396e-05,
+      "loss": 1.2672,
+      "step": 6957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0560647614868447e-05,
+      "loss": 1.4195,
+      "step": 6958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0560426379975498e-05,
+      "loss": 1.1927,
+      "step": 6959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.056020514508255e-05,
+      "loss": 1.2776,
+      "step": 6960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0559983910189603e-05,
+      "loss": 1.2362,
+      "step": 6961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0559762675296656e-05,
+      "loss": 1.3166,
+      "step": 6962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0559541440403708e-05,
+      "loss": 1.3692,
+      "step": 6963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.055932020551076e-05,
+      "loss": 1.2634,
+      "step": 6964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0559098970617812e-05,
+      "loss": 1.1661,
+      "step": 6965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0558877735724863e-05,
+      "loss": 1.3439,
+      "step": 6966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0558656500831916e-05,
+      "loss": 1.3499,
+      "step": 6967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0558435265938968e-05,
+      "loss": 1.4053,
+      "step": 6968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0558214031046019e-05,
+      "loss": 1.2585,
+      "step": 6969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0557992796153072e-05,
+      "loss": 1.2523,
+      "step": 6970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0557771561260124e-05,
+      "loss": 1.3414,
+      "step": 6971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0557550326367177e-05,
+      "loss": 1.3711,
+      "step": 6972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.055732909147423e-05,
+      "loss": 1.401,
+      "step": 6973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.055710785658128e-05,
+      "loss": 1.4293,
+      "step": 6974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0556886621688331e-05,
+      "loss": 1.1923,
+      "step": 6975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0556665386795384e-05,
+      "loss": 1.1952,
+      "step": 6976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0556444151902437e-05,
+      "loss": 1.3599,
+      "step": 6977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.055622291700949e-05,
+      "loss": 1.1854,
+      "step": 6978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0556001682116542e-05,
+      "loss": 1.3531,
+      "step": 6979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0555780447223593e-05,
+      "loss": 1.3279,
+      "step": 6980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0555559212330645e-05,
+      "loss": 1.3442,
+      "step": 6981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0555337977437696e-05,
+      "loss": 1.2472,
+      "step": 6982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0555116742544749e-05,
+      "loss": 1.3563,
+      "step": 6983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0554895507651801e-05,
+      "loss": 1.2154,
+      "step": 6984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0554674272758854e-05,
+      "loss": 1.3857,
+      "step": 6985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0554453037865905e-05,
+      "loss": 1.3248,
+      "step": 6986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0554231802972958e-05,
+      "loss": 1.4079,
+      "step": 6987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.055401056808001e-05,
+      "loss": 1.3828,
+      "step": 6988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0553789333187063e-05,
+      "loss": 1.1888,
+      "step": 6989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0553568098294114e-05,
+      "loss": 1.3426,
+      "step": 6990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0553346863401165e-05,
+      "loss": 1.3748,
+      "step": 6991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0553125628508217e-05,
+      "loss": 1.2903,
+      "step": 6992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.055290439361527e-05,
+      "loss": 1.3461,
+      "step": 6993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0552683158722322e-05,
+      "loss": 1.304,
+      "step": 6994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0552461923829375e-05,
+      "loss": 1.3261,
+      "step": 6995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0552240688936426e-05,
+      "loss": 1.393,
+      "step": 6996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0552019454043479e-05,
+      "loss": 1.4472,
+      "step": 6997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0551798219150531e-05,
+      "loss": 1.4235,
+      "step": 6998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0551576984257582e-05,
+      "loss": 1.4768,
+      "step": 6999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0551355749364635e-05,
+      "loss": 1.3414,
+      "step": 7000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0551134514471687e-05,
+      "loss": 1.2642,
+      "step": 7001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0550913279578738e-05,
+      "loss": 1.3367,
+      "step": 7002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0550692044685791e-05,
+      "loss": 1.3331,
+      "step": 7003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0550470809792844e-05,
+      "loss": 1.2956,
+      "step": 7004
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0550249574899896e-05,
+      "loss": 1.4013,
+      "step": 7005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0550028340006947e-05,
+      "loss": 1.333,
+      "step": 7006
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0549807105113998e-05,
+      "loss": 1.4352,
+      "step": 7007
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.054958587022105e-05,
+      "loss": 1.1556,
+      "step": 7008
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0549364635328103e-05,
+      "loss": 1.2066,
+      "step": 7009
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0549143400435156e-05,
+      "loss": 1.3532,
+      "step": 7010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0548922165542208e-05,
+      "loss": 1.3025,
+      "step": 7011
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0548700930649261e-05,
+      "loss": 1.3255,
+      "step": 7012
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0548479695756312e-05,
+      "loss": 1.2078,
+      "step": 7013
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0548258460863365e-05,
+      "loss": 1.3756,
+      "step": 7014
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0548037225970415e-05,
+      "loss": 1.2068,
+      "step": 7015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0547815991077468e-05,
+      "loss": 1.4811,
+      "step": 7016
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.054759475618452e-05,
+      "loss": 1.375,
+      "step": 7017
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0547373521291572e-05,
+      "loss": 1.4196,
+      "step": 7018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0547152286398624e-05,
+      "loss": 1.3277,
+      "step": 7019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0546931051505677e-05,
+      "loss": 1.2737,
+      "step": 7020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.054670981661273e-05,
+      "loss": 1.3649,
+      "step": 7021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.054648858171978e-05,
+      "loss": 1.3256,
+      "step": 7022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0546267346826833e-05,
+      "loss": 1.2513,
+      "step": 7023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0546046111933884e-05,
+      "loss": 1.4252,
+      "step": 7024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0545824877040937e-05,
+      "loss": 1.2799,
+      "step": 7025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0545603642147989e-05,
+      "loss": 1.1939,
+      "step": 7026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0545382407255042e-05,
+      "loss": 1.3643,
+      "step": 7027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0545161172362094e-05,
+      "loss": 1.2787,
+      "step": 7028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0544939937469145e-05,
+      "loss": 1.2065,
+      "step": 7029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0544718702576198e-05,
+      "loss": 1.3174,
+      "step": 7030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0544497467683249e-05,
+      "loss": 1.3401,
+      "step": 7031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0544276232790301e-05,
+      "loss": 1.3457,
+      "step": 7032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0544054997897354e-05,
+      "loss": 1.3737,
+      "step": 7033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0543833763004407e-05,
+      "loss": 1.3606,
+      "step": 7034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0543612528111458e-05,
+      "loss": 1.3184,
+      "step": 7035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.054339129321851e-05,
+      "loss": 1.261,
+      "step": 7036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0543170058325563e-05,
+      "loss": 1.4317,
+      "step": 7037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0542948823432614e-05,
+      "loss": 1.3439,
+      "step": 7038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0542727588539666e-05,
+      "loss": 1.3384,
+      "step": 7039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0542506353646717e-05,
+      "loss": 1.2845,
+      "step": 7040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.054228511875377e-05,
+      "loss": 1.2354,
+      "step": 7041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0542063883860822e-05,
+      "loss": 1.252,
+      "step": 7042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0541842648967875e-05,
+      "loss": 1.255,
+      "step": 7043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0541621414074928e-05,
+      "loss": 1.2526,
+      "step": 7044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.054140017918198e-05,
+      "loss": 1.3818,
+      "step": 7045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0541178944289031e-05,
+      "loss": 1.4345,
+      "step": 7046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0540957709396082e-05,
+      "loss": 1.3587,
+      "step": 7047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0540736474503135e-05,
+      "loss": 1.3231,
+      "step": 7048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0540515239610187e-05,
+      "loss": 1.2174,
+      "step": 7049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.054029400471724e-05,
+      "loss": 1.3675,
+      "step": 7050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0540072769824291e-05,
+      "loss": 1.2885,
+      "step": 7051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0539851534931343e-05,
+      "loss": 1.3211,
+      "step": 7052
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0539630300038396e-05,
+      "loss": 1.2475,
+      "step": 7053
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0539409065145447e-05,
+      "loss": 1.3377,
+      "step": 7054
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.05391878302525e-05,
+      "loss": 1.2824,
+      "step": 7055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0538966595359552e-05,
+      "loss": 1.3819,
+      "step": 7056
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0538745360466603e-05,
+      "loss": 1.4565,
+      "step": 7057
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0538524125573656e-05,
+      "loss": 1.3675,
+      "step": 7058
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0538302890680708e-05,
+      "loss": 1.3315,
+      "step": 7059
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0538081655787761e-05,
+      "loss": 1.4014,
+      "step": 7060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0537860420894814e-05,
+      "loss": 1.3454,
+      "step": 7061
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0537639186001865e-05,
+      "loss": 1.3844,
+      "step": 7062
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0537417951108915e-05,
+      "loss": 1.2755,
+      "step": 7063
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0537196716215968e-05,
+      "loss": 1.3923,
+      "step": 7064
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.053697548132302e-05,
+      "loss": 1.2858,
+      "step": 7065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0536754246430073e-05,
+      "loss": 1.3483,
+      "step": 7066
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0536533011537126e-05,
+      "loss": 1.3062,
+      "step": 7067
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0536311776644177e-05,
+      "loss": 1.2987,
+      "step": 7068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.053609054175123e-05,
+      "loss": 1.3248,
+      "step": 7069
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.053586930685828e-05,
+      "loss": 1.3115,
+      "step": 7070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0535648071965333e-05,
+      "loss": 1.3438,
+      "step": 7071
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0535426837072386e-05,
+      "loss": 1.4056,
+      "step": 7072
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0535205602179436e-05,
+      "loss": 1.2989,
+      "step": 7073
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0534984367286489e-05,
+      "loss": 1.3207,
+      "step": 7074
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0534763132393542e-05,
+      "loss": 1.4393,
+      "step": 7075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0534541897500594e-05,
+      "loss": 1.2093,
+      "step": 7076
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0534320662607647e-05,
+      "loss": 1.3081,
+      "step": 7077
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0534099427714698e-05,
+      "loss": 1.2715,
+      "step": 7078
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0533878192821749e-05,
+      "loss": 1.4342,
+      "step": 7079
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0533656957928801e-05,
+      "loss": 1.2171,
+      "step": 7080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0533435723035854e-05,
+      "loss": 1.2537,
+      "step": 7081
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0533214488142907e-05,
+      "loss": 1.4205,
+      "step": 7082
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.053299325324996e-05,
+      "loss": 1.3251,
+      "step": 7083
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.053277201835701e-05,
+      "loss": 1.3475,
+      "step": 7084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0532550783464063e-05,
+      "loss": 1.1819,
+      "step": 7085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0532329548571115e-05,
+      "loss": 1.2732,
+      "step": 7086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0532108313678166e-05,
+      "loss": 1.3081,
+      "step": 7087
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0531887078785219e-05,
+      "loss": 1.3779,
+      "step": 7088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0531665843892271e-05,
+      "loss": 1.363,
+      "step": 7089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0531444608999322e-05,
+      "loss": 1.3041,
+      "step": 7090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0531223374106375e-05,
+      "loss": 1.3456,
+      "step": 7091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0531002139213428e-05,
+      "loss": 1.479,
+      "step": 7092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.053078090432048e-05,
+      "loss": 1.4844,
+      "step": 7093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0530559669427531e-05,
+      "loss": 1.347,
+      "step": 7094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0530338434534582e-05,
+      "loss": 1.3914,
+      "step": 7095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0530117199641635e-05,
+      "loss": 1.2275,
+      "step": 7096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0529895964748687e-05,
+      "loss": 1.3584,
+      "step": 7097
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.052967472985574e-05,
+      "loss": 1.3489,
+      "step": 7098
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0529453494962793e-05,
+      "loss": 1.2161,
+      "step": 7099
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0529232260069843e-05,
+      "loss": 1.3082,
+      "step": 7100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0529011025176896e-05,
+      "loss": 1.4208,
+      "step": 7101
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0528789790283949e-05,
+      "loss": 1.4428,
+      "step": 7102
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0528568555391e-05,
+      "loss": 1.2277,
+      "step": 7103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0528347320498052e-05,
+      "loss": 1.3814,
+      "step": 7104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0528126085605105e-05,
+      "loss": 1.4411,
+      "step": 7105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0527904850712156e-05,
+      "loss": 1.3236,
+      "step": 7106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0527683615819208e-05,
+      "loss": 1.2999,
+      "step": 7107
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0527462380926261e-05,
+      "loss": 1.3474,
+      "step": 7108
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0527241146033314e-05,
+      "loss": 1.2626,
+      "step": 7109
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0527019911140364e-05,
+      "loss": 1.3256,
+      "step": 7110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0526798676247415e-05,
+      "loss": 1.32,
+      "step": 7111
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0526577441354468e-05,
+      "loss": 1.4546,
+      "step": 7112
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.052635620646152e-05,
+      "loss": 1.3182,
+      "step": 7113
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0526134971568573e-05,
+      "loss": 1.3981,
+      "step": 7114
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0525913736675626e-05,
+      "loss": 1.2706,
+      "step": 7115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0525692501782678e-05,
+      "loss": 1.3717,
+      "step": 7116
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.052547126688973e-05,
+      "loss": 1.2111,
+      "step": 7117
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0525250031996782e-05,
+      "loss": 1.391,
+      "step": 7118
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0525028797103833e-05,
+      "loss": 1.3501,
+      "step": 7119
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0524807562210885e-05,
+      "loss": 1.2622,
+      "step": 7120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0524586327317938e-05,
+      "loss": 1.3314,
+      "step": 7121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0524365092424989e-05,
+      "loss": 1.2349,
+      "step": 7122
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0524143857532042e-05,
+      "loss": 1.3088,
+      "step": 7123
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0523922622639094e-05,
+      "loss": 1.4021,
+      "step": 7124
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0523701387746147e-05,
+      "loss": 1.3005,
+      "step": 7125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0523480152853198e-05,
+      "loss": 1.4875,
+      "step": 7126
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.052325891796025e-05,
+      "loss": 1.383,
+      "step": 7127
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0523037683067301e-05,
+      "loss": 1.2403,
+      "step": 7128
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0522816448174354e-05,
+      "loss": 1.2452,
+      "step": 7129
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0522595213281407e-05,
+      "loss": 1.4175,
+      "step": 7130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0522373978388459e-05,
+      "loss": 1.2934,
+      "step": 7131
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0522152743495512e-05,
+      "loss": 1.3386,
+      "step": 7132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0521931508602563e-05,
+      "loss": 1.297,
+      "step": 7133
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0521710273709615e-05,
+      "loss": 1.3321,
+      "step": 7134
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0521489038816666e-05,
+      "loss": 1.4385,
+      "step": 7135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0521267803923719e-05,
+      "loss": 1.4271,
+      "step": 7136
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0521046569030771e-05,
+      "loss": 1.2774,
+      "step": 7137
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0520825334137824e-05,
+      "loss": 1.3036,
+      "step": 7138
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0520604099244875e-05,
+      "loss": 1.2676,
+      "step": 7139
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0520382864351928e-05,
+      "loss": 1.2495,
+      "step": 7140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.052016162945898e-05,
+      "loss": 1.3637,
+      "step": 7141
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0519940394566031e-05,
+      "loss": 1.2757,
+      "step": 7142
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0519719159673084e-05,
+      "loss": 1.3563,
+      "step": 7143
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0519497924780135e-05,
+      "loss": 1.4077,
+      "step": 7144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0519276689887187e-05,
+      "loss": 1.3088,
+      "step": 7145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.051905545499424e-05,
+      "loss": 1.3318,
+      "step": 7146
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0518834220101292e-05,
+      "loss": 1.2888,
+      "step": 7147
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0518612985208345e-05,
+      "loss": 1.3908,
+      "step": 7148
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0518391750315398e-05,
+      "loss": 1.3259,
+      "step": 7149
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0518170515422449e-05,
+      "loss": 1.3183,
+      "step": 7150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.05179492805295e-05,
+      "loss": 1.3057,
+      "step": 7151
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0517728045636552e-05,
+      "loss": 1.3215,
+      "step": 7152
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0517506810743605e-05,
+      "loss": 1.2875,
+      "step": 7153
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0517285575850657e-05,
+      "loss": 1.2715,
+      "step": 7154
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0517064340957708e-05,
+      "loss": 1.3559,
+      "step": 7155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0516843106064761e-05,
+      "loss": 1.3629,
+      "step": 7156
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0516621871171813e-05,
+      "loss": 1.5219,
+      "step": 7157
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0516400636278864e-05,
+      "loss": 1.3689,
+      "step": 7158
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0516179401385917e-05,
+      "loss": 1.2187,
+      "step": 7159
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.051595816649297e-05,
+      "loss": 1.2221,
+      "step": 7160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.051573693160002e-05,
+      "loss": 1.351,
+      "step": 7161
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0515515696707073e-05,
+      "loss": 1.4086,
+      "step": 7162
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0515294461814126e-05,
+      "loss": 1.3868,
+      "step": 7163
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0515073226921178e-05,
+      "loss": 1.3259,
+      "step": 7164
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0514851992028231e-05,
+      "loss": 1.3016,
+      "step": 7165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0514630757135282e-05,
+      "loss": 1.4442,
+      "step": 7166
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0514409522242333e-05,
+      "loss": 1.4416,
+      "step": 7167
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0514188287349385e-05,
+      "loss": 1.2796,
+      "step": 7168
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0513967052456438e-05,
+      "loss": 1.4661,
+      "step": 7169
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.051374581756349e-05,
+      "loss": 1.3003,
+      "step": 7170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0513524582670543e-05,
+      "loss": 1.3172,
+      "step": 7171
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0513303347777594e-05,
+      "loss": 1.3632,
+      "step": 7172
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0513082112884647e-05,
+      "loss": 1.3445,
+      "step": 7173
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.05128608779917e-05,
+      "loss": 1.3395,
+      "step": 7174
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.051263964309875e-05,
+      "loss": 1.3797,
+      "step": 7175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0512418408205803e-05,
+      "loss": 1.2546,
+      "step": 7176
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0512197173312854e-05,
+      "loss": 1.2222,
+      "step": 7177
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0511975938419906e-05,
+      "loss": 1.3119,
+      "step": 7178
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0511754703526959e-05,
+      "loss": 1.2355,
+      "step": 7179
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0511533468634012e-05,
+      "loss": 1.3948,
+      "step": 7180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0511312233741064e-05,
+      "loss": 1.3089,
+      "step": 7181
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0511090998848115e-05,
+      "loss": 1.2774,
+      "step": 7182
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0510869763955166e-05,
+      "loss": 1.2824,
+      "step": 7183
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0510648529062219e-05,
+      "loss": 1.3567,
+      "step": 7184
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0510427294169271e-05,
+      "loss": 1.4766,
+      "step": 7185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0510206059276324e-05,
+      "loss": 1.3707,
+      "step": 7186
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0509984824383377e-05,
+      "loss": 1.143,
+      "step": 7187
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0509763589490428e-05,
+      "loss": 1.3938,
+      "step": 7188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.050954235459748e-05,
+      "loss": 1.2475,
+      "step": 7189
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0509321119704533e-05,
+      "loss": 1.4179,
+      "step": 7190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0509099884811584e-05,
+      "loss": 1.4078,
+      "step": 7191
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0508878649918636e-05,
+      "loss": 1.3182,
+      "step": 7192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0508657415025687e-05,
+      "loss": 1.2727,
+      "step": 7193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.050843618013274e-05,
+      "loss": 1.3537,
+      "step": 7194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0508214945239792e-05,
+      "loss": 1.3618,
+      "step": 7195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0507993710346845e-05,
+      "loss": 1.2363,
+      "step": 7196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0507772475453898e-05,
+      "loss": 1.315,
+      "step": 7197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0507551240560949e-05,
+      "loss": 1.2808,
+      "step": 7198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0507330005668e-05,
+      "loss": 1.3539,
+      "step": 7199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0507108770775052e-05,
+      "loss": 1.3628,
+      "step": 7200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0506887535882105e-05,
+      "loss": 1.3498,
+      "step": 7201
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0506666300989157e-05,
+      "loss": 1.3143,
+      "step": 7202
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.050644506609621e-05,
+      "loss": 1.2768,
+      "step": 7203
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.050622383120326e-05,
+      "loss": 1.2799,
+      "step": 7204
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0506002596310313e-05,
+      "loss": 1.2255,
+      "step": 7205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0505781361417366e-05,
+      "loss": 1.4597,
+      "step": 7206
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0505560126524417e-05,
+      "loss": 1.1315,
+      "step": 7207
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.050533889163147e-05,
+      "loss": 1.3605,
+      "step": 7208
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0505117656738522e-05,
+      "loss": 1.3871,
+      "step": 7209
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0504896421845573e-05,
+      "loss": 1.3682,
+      "step": 7210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0504675186952626e-05,
+      "loss": 1.3012,
+      "step": 7211
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0504453952059678e-05,
+      "loss": 1.3288,
+      "step": 7212
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0504232717166731e-05,
+      "loss": 1.3235,
+      "step": 7213
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0504011482273782e-05,
+      "loss": 1.2649,
+      "step": 7214
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0503790247380833e-05,
+      "loss": 1.3573,
+      "step": 7215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0503569012487885e-05,
+      "loss": 1.4122,
+      "step": 7216
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0503347777594938e-05,
+      "loss": 1.3007,
+      "step": 7217
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.050312654270199e-05,
+      "loss": 1.2686,
+      "step": 7218
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0502905307809043e-05,
+      "loss": 1.4812,
+      "step": 7219
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0502684072916096e-05,
+      "loss": 1.3395,
+      "step": 7220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0502462838023147e-05,
+      "loss": 1.3489,
+      "step": 7221
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.05022416031302e-05,
+      "loss": 1.4099,
+      "step": 7222
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.050202036823725e-05,
+      "loss": 1.4072,
+      "step": 7223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0501799133344303e-05,
+      "loss": 1.2088,
+      "step": 7224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0501577898451356e-05,
+      "loss": 1.4656,
+      "step": 7225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0501356663558406e-05,
+      "loss": 1.2923,
+      "step": 7226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0501135428665459e-05,
+      "loss": 1.354,
+      "step": 7227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0500914193772512e-05,
+      "loss": 1.369,
+      "step": 7228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0500692958879564e-05,
+      "loss": 1.3849,
+      "step": 7229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0500471723986615e-05,
+      "loss": 1.3191,
+      "step": 7230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0500250489093668e-05,
+      "loss": 1.3908,
+      "step": 7231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0500029254200719e-05,
+      "loss": 1.4219,
+      "step": 7232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0499808019307771e-05,
+      "loss": 1.3257,
+      "step": 7233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0499586784414824e-05,
+      "loss": 1.3086,
+      "step": 7234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0499365549521877e-05,
+      "loss": 1.315,
+      "step": 7235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.049914431462893e-05,
+      "loss": 1.3197,
+      "step": 7236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.049892307973598e-05,
+      "loss": 1.2747,
+      "step": 7237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0498701844843033e-05,
+      "loss": 1.3482,
+      "step": 7238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0498480609950084e-05,
+      "loss": 1.1896,
+      "step": 7239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0498259375057136e-05,
+      "loss": 1.2488,
+      "step": 7240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0498038140164189e-05,
+      "loss": 1.3984,
+      "step": 7241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0497816905271241e-05,
+      "loss": 1.3512,
+      "step": 7242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0497595670378292e-05,
+      "loss": 1.2932,
+      "step": 7243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0497374435485345e-05,
+      "loss": 1.32,
+      "step": 7244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0497153200592398e-05,
+      "loss": 1.0832,
+      "step": 7245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0496931965699449e-05,
+      "loss": 1.1788,
+      "step": 7246
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0496710730806501e-05,
+      "loss": 1.2922,
+      "step": 7247
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0496489495913552e-05,
+      "loss": 1.1477,
+      "step": 7248
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0496268261020605e-05,
+      "loss": 1.2027,
+      "step": 7249
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0496047026127657e-05,
+      "loss": 1.296,
+      "step": 7250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.049582579123471e-05,
+      "loss": 1.2432,
+      "step": 7251
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0495604556341762e-05,
+      "loss": 1.3715,
+      "step": 7252
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0495383321448815e-05,
+      "loss": 1.3022,
+      "step": 7253
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0495162086555866e-05,
+      "loss": 1.4969,
+      "step": 7254
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0494940851662917e-05,
+      "loss": 1.2741,
+      "step": 7255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.049471961676997e-05,
+      "loss": 1.3111,
+      "step": 7256
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0494498381877022e-05,
+      "loss": 1.3472,
+      "step": 7257
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0494277146984075e-05,
+      "loss": 1.3518,
+      "step": 7258
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0494055912091126e-05,
+      "loss": 1.33,
+      "step": 7259
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0493834677198178e-05,
+      "loss": 1.4124,
+      "step": 7260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0493613442305231e-05,
+      "loss": 1.2594,
+      "step": 7261
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0493392207412284e-05,
+      "loss": 1.347,
+      "step": 7262
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0493170972519334e-05,
+      "loss": 1.301,
+      "step": 7263
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0492949737626387e-05,
+      "loss": 1.4035,
+      "step": 7264
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0492728502733438e-05,
+      "loss": 1.2504,
+      "step": 7265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.049250726784049e-05,
+      "loss": 1.333,
+      "step": 7266
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0492286032947543e-05,
+      "loss": 1.3715,
+      "step": 7267
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0492064798054596e-05,
+      "loss": 1.3027,
+      "step": 7268
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0491843563161648e-05,
+      "loss": 1.4723,
+      "step": 7269
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.04916223282687e-05,
+      "loss": 1.353,
+      "step": 7270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.049140109337575e-05,
+      "loss": 1.2484,
+      "step": 7271
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0491179858482803e-05,
+      "loss": 1.331,
+      "step": 7272
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0490958623589855e-05,
+      "loss": 1.3724,
+      "step": 7273
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0490737388696908e-05,
+      "loss": 1.2352,
+      "step": 7274
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.049051615380396e-05,
+      "loss": 1.2415,
+      "step": 7275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0490294918911012e-05,
+      "loss": 1.4309,
+      "step": 7276
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0490073684018064e-05,
+      "loss": 1.3743,
+      "step": 7277
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0489852449125117e-05,
+      "loss": 1.4805,
+      "step": 7278
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0489631214232168e-05,
+      "loss": 1.4447,
+      "step": 7279
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.048940997933922e-05,
+      "loss": 1.1776,
+      "step": 7280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0489188744446271e-05,
+      "loss": 1.3454,
+      "step": 7281
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0488967509553324e-05,
+      "loss": 1.2869,
+      "step": 7282
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0488746274660377e-05,
+      "loss": 1.248,
+      "step": 7283
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0488525039767429e-05,
+      "loss": 1.3254,
+      "step": 7284
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0488303804874482e-05,
+      "loss": 1.2375,
+      "step": 7285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0488082569981533e-05,
+      "loss": 1.3505,
+      "step": 7286
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0487861335088584e-05,
+      "loss": 1.2832,
+      "step": 7287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0487640100195636e-05,
+      "loss": 1.2256,
+      "step": 7288
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0487418865302689e-05,
+      "loss": 1.2777,
+      "step": 7289
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0487197630409741e-05,
+      "loss": 1.2301,
+      "step": 7290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0486976395516794e-05,
+      "loss": 1.4454,
+      "step": 7291
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0486755160623845e-05,
+      "loss": 1.2765,
+      "step": 7292
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0486533925730898e-05,
+      "loss": 1.2631,
+      "step": 7293
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.048631269083795e-05,
+      "loss": 1.3052,
+      "step": 7294
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0486091455945001e-05,
+      "loss": 1.458,
+      "step": 7295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0485870221052054e-05,
+      "loss": 1.2966,
+      "step": 7296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0485648986159105e-05,
+      "loss": 1.2828,
+      "step": 7297
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0485427751266157e-05,
+      "loss": 1.3012,
+      "step": 7298
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.048520651637321e-05,
+      "loss": 1.3273,
+      "step": 7299
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0484985281480262e-05,
+      "loss": 1.3804,
+      "step": 7300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0484764046587315e-05,
+      "loss": 1.305,
+      "step": 7301
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0484542811694366e-05,
+      "loss": 1.3912,
+      "step": 7302
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0484321576801417e-05,
+      "loss": 1.2365,
+      "step": 7303
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.048410034190847e-05,
+      "loss": 1.2418,
+      "step": 7304
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0483879107015522e-05,
+      "loss": 1.3477,
+      "step": 7305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0483657872122575e-05,
+      "loss": 1.4086,
+      "step": 7306
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0483436637229627e-05,
+      "loss": 1.3699,
+      "step": 7307
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0483215402336678e-05,
+      "loss": 1.3157,
+      "step": 7308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0482994167443731e-05,
+      "loss": 1.3622,
+      "step": 7309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0482772932550783e-05,
+      "loss": 1.3889,
+      "step": 7310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0482551697657834e-05,
+      "loss": 1.2283,
+      "step": 7311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0482330462764887e-05,
+      "loss": 1.299,
+      "step": 7312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.048210922787194e-05,
+      "loss": 1.3261,
+      "step": 7313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.048188799297899e-05,
+      "loss": 1.3765,
+      "step": 7314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0481666758086043e-05,
+      "loss": 1.3127,
+      "step": 7315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0481445523193096e-05,
+      "loss": 1.2399,
+      "step": 7316
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0481224288300148e-05,
+      "loss": 1.3599,
+      "step": 7317
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.04810030534072e-05,
+      "loss": 1.2818,
+      "step": 7318
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0480781818514252e-05,
+      "loss": 1.2565,
+      "step": 7319
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0480560583621303e-05,
+      "loss": 1.2301,
+      "step": 7320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0480339348728355e-05,
+      "loss": 1.3158,
+      "step": 7321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0480118113835408e-05,
+      "loss": 1.2398,
+      "step": 7322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.047989687894246e-05,
+      "loss": 1.4958,
+      "step": 7323
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0479675644049513e-05,
+      "loss": 1.4652,
+      "step": 7324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0479454409156564e-05,
+      "loss": 1.2798,
+      "step": 7325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0479233174263617e-05,
+      "loss": 1.2661,
+      "step": 7326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0479011939370668e-05,
+      "loss": 1.1778,
+      "step": 7327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.047879070447772e-05,
+      "loss": 1.4054,
+      "step": 7328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0478569469584773e-05,
+      "loss": 1.2777,
+      "step": 7329
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0478348234691824e-05,
+      "loss": 1.4223,
+      "step": 7330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0478126999798876e-05,
+      "loss": 1.2201,
+      "step": 7331
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0477905764905929e-05,
+      "loss": 1.3191,
+      "step": 7332
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0477684530012982e-05,
+      "loss": 1.3165,
+      "step": 7333
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0477463295120033e-05,
+      "loss": 1.2805,
+      "step": 7334
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0477242060227085e-05,
+      "loss": 1.3675,
+      "step": 7335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0477020825334136e-05,
+      "loss": 1.2818,
+      "step": 7336
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0476799590441189e-05,
+      "loss": 1.3445,
+      "step": 7337
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0476578355548241e-05,
+      "loss": 1.4575,
+      "step": 7338
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0476357120655294e-05,
+      "loss": 1.4122,
+      "step": 7339
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0476135885762347e-05,
+      "loss": 1.4122,
+      "step": 7340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0475914650869398e-05,
+      "loss": 1.47,
+      "step": 7341
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.047569341597645e-05,
+      "loss": 1.3066,
+      "step": 7342
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0475472181083501e-05,
+      "loss": 1.3626,
+      "step": 7343
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0475250946190554e-05,
+      "loss": 1.5228,
+      "step": 7344
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0475029711297606e-05,
+      "loss": 1.3211,
+      "step": 7345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0474808476404659e-05,
+      "loss": 1.3593,
+      "step": 7346
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.047458724151171e-05,
+      "loss": 1.3047,
+      "step": 7347
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0474366006618762e-05,
+      "loss": 1.2723,
+      "step": 7348
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0474144771725815e-05,
+      "loss": 1.2834,
+      "step": 7349
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0473923536832868e-05,
+      "loss": 1.4011,
+      "step": 7350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0473702301939919e-05,
+      "loss": 1.2252,
+      "step": 7351
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.047348106704697e-05,
+      "loss": 1.338,
+      "step": 7352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0473259832154022e-05,
+      "loss": 1.3103,
+      "step": 7353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0473038597261075e-05,
+      "loss": 1.2472,
+      "step": 7354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0472817362368127e-05,
+      "loss": 1.2633,
+      "step": 7355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.047259612747518e-05,
+      "loss": 1.2677,
+      "step": 7356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0472374892582233e-05,
+      "loss": 1.3515,
+      "step": 7357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0472153657689283e-05,
+      "loss": 1.3619,
+      "step": 7358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0471932422796334e-05,
+      "loss": 1.2999,
+      "step": 7359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0471711187903387e-05,
+      "loss": 1.368,
+      "step": 7360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.047148995301044e-05,
+      "loss": 1.286,
+      "step": 7361
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0471268718117492e-05,
+      "loss": 1.5364,
+      "step": 7362
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0471047483224543e-05,
+      "loss": 1.3582,
+      "step": 7363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0470826248331596e-05,
+      "loss": 1.3281,
+      "step": 7364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0470605013438648e-05,
+      "loss": 1.3374,
+      "step": 7365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0470383778545701e-05,
+      "loss": 1.4315,
+      "step": 7366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0470162543652752e-05,
+      "loss": 1.3046,
+      "step": 7367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0469941308759804e-05,
+      "loss": 1.3526,
+      "step": 7368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0469720073866855e-05,
+      "loss": 1.28,
+      "step": 7369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0469498838973908e-05,
+      "loss": 1.2959,
+      "step": 7370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.046927760408096e-05,
+      "loss": 1.2801,
+      "step": 7371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0469056369188013e-05,
+      "loss": 1.3618,
+      "step": 7372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0468835134295066e-05,
+      "loss": 1.4197,
+      "step": 7373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0468613899402117e-05,
+      "loss": 1.1924,
+      "step": 7374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0468392664509168e-05,
+      "loss": 1.3705,
+      "step": 7375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.046817142961622e-05,
+      "loss": 1.2922,
+      "step": 7376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0467950194723273e-05,
+      "loss": 1.3591,
+      "step": 7377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0467728959830326e-05,
+      "loss": 1.3374,
+      "step": 7378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0467507724937378e-05,
+      "loss": 1.3495,
+      "step": 7379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0467286490044429e-05,
+      "loss": 1.2971,
+      "step": 7380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0467065255151482e-05,
+      "loss": 1.333,
+      "step": 7381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0466844020258534e-05,
+      "loss": 1.2493,
+      "step": 7382
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0466622785365585e-05,
+      "loss": 1.3321,
+      "step": 7383
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0466401550472638e-05,
+      "loss": 1.2404,
+      "step": 7384
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0466180315579689e-05,
+      "loss": 1.3769,
+      "step": 7385
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0465959080686741e-05,
+      "loss": 1.3797,
+      "step": 7386
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0465737845793794e-05,
+      "loss": 1.3034,
+      "step": 7387
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0465516610900847e-05,
+      "loss": 1.2962,
+      "step": 7388
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.04652953760079e-05,
+      "loss": 1.2191,
+      "step": 7389
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.046507414111495e-05,
+      "loss": 1.4837,
+      "step": 7390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0464852906222001e-05,
+      "loss": 1.4147,
+      "step": 7391
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0464631671329054e-05,
+      "loss": 1.3492,
+      "step": 7392
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0464410436436106e-05,
+      "loss": 1.3796,
+      "step": 7393
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0464189201543159e-05,
+      "loss": 1.3716,
+      "step": 7394
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0463967966650211e-05,
+      "loss": 1.3562,
+      "step": 7395
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0463746731757262e-05,
+      "loss": 1.3067,
+      "step": 7396
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0463525496864315e-05,
+      "loss": 1.2325,
+      "step": 7397
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0463304261971368e-05,
+      "loss": 1.268,
+      "step": 7398
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0463083027078419e-05,
+      "loss": 1.2521,
+      "step": 7399
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0462861792185471e-05,
+      "loss": 1.348,
+      "step": 7400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0462640557292522e-05,
+      "loss": 1.4207,
+      "step": 7401
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0462419322399575e-05,
+      "loss": 1.4422,
+      "step": 7402
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0462198087506627e-05,
+      "loss": 1.3677,
+      "step": 7403
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.046197685261368e-05,
+      "loss": 1.2738,
+      "step": 7404
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0461755617720732e-05,
+      "loss": 1.3043,
+      "step": 7405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0461534382827783e-05,
+      "loss": 1.2429,
+      "step": 7406
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0461313147934836e-05,
+      "loss": 1.4437,
+      "step": 7407
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0461091913041887e-05,
+      "loss": 1.2199,
+      "step": 7408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.046087067814894e-05,
+      "loss": 1.292,
+      "step": 7409
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0460649443255992e-05,
+      "loss": 1.2421,
+      "step": 7410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0460428208363045e-05,
+      "loss": 1.2963,
+      "step": 7411
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0460206973470096e-05,
+      "loss": 1.376,
+      "step": 7412
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0459985738577148e-05,
+      "loss": 1.3109,
+      "step": 7413
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0459764503684201e-05,
+      "loss": 1.293,
+      "step": 7414
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0459543268791252e-05,
+      "loss": 1.3408,
+      "step": 7415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0459322033898304e-05,
+      "loss": 1.2852,
+      "step": 7416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0459100799005357e-05,
+      "loss": 1.3097,
+      "step": 7417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0458879564112408e-05,
+      "loss": 1.2987,
+      "step": 7418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.045865832921946e-05,
+      "loss": 1.3214,
+      "step": 7419
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0458437094326513e-05,
+      "loss": 1.2709,
+      "step": 7420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0458215859433566e-05,
+      "loss": 1.2584,
+      "step": 7421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0457994624540617e-05,
+      "loss": 1.3289,
+      "step": 7422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.045777338964767e-05,
+      "loss": 1.3321,
+      "step": 7423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.045755215475472e-05,
+      "loss": 1.246,
+      "step": 7424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0457330919861773e-05,
+      "loss": 1.3769,
+      "step": 7425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0457109684968825e-05,
+      "loss": 1.2098,
+      "step": 7426
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0456888450075878e-05,
+      "loss": 1.2339,
+      "step": 7427
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.045666721518293e-05,
+      "loss": 1.1598,
+      "step": 7428
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0456445980289982e-05,
+      "loss": 1.3702,
+      "step": 7429
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0456224745397034e-05,
+      "loss": 1.2595,
+      "step": 7430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0456003510504085e-05,
+      "loss": 1.3576,
+      "step": 7431
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0455782275611138e-05,
+      "loss": 1.3054,
+      "step": 7432
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.045556104071819e-05,
+      "loss": 1.3892,
+      "step": 7433
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0455339805825241e-05,
+      "loss": 1.3152,
+      "step": 7434
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0455118570932294e-05,
+      "loss": 1.3701,
+      "step": 7435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0454897336039347e-05,
+      "loss": 1.3065,
+      "step": 7436
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0454676101146399e-05,
+      "loss": 1.3114,
+      "step": 7437
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0454454866253452e-05,
+      "loss": 1.3643,
+      "step": 7438
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0454233631360503e-05,
+      "loss": 1.29,
+      "step": 7439
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0454012396467554e-05,
+      "loss": 1.2895,
+      "step": 7440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0453791161574606e-05,
+      "loss": 1.3331,
+      "step": 7441
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0453569926681659e-05,
+      "loss": 1.3759,
+      "step": 7442
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0453348691788711e-05,
+      "loss": 1.3681,
+      "step": 7443
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0453127456895764e-05,
+      "loss": 1.2017,
+      "step": 7444
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0452906222002815e-05,
+      "loss": 1.156,
+      "step": 7445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0452684987109868e-05,
+      "loss": 1.4255,
+      "step": 7446
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0452463752216918e-05,
+      "loss": 1.2827,
+      "step": 7447
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0452242517323971e-05,
+      "loss": 1.3465,
+      "step": 7448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0452021282431024e-05,
+      "loss": 1.4465,
+      "step": 7449
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0451800047538076e-05,
+      "loss": 1.3454,
+      "step": 7450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0451578812645127e-05,
+      "loss": 1.278,
+      "step": 7451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.045135757775218e-05,
+      "loss": 1.2604,
+      "step": 7452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0451136342859232e-05,
+      "loss": 1.3199,
+      "step": 7453
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0450915107966285e-05,
+      "loss": 1.1893,
+      "step": 7454
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0450693873073336e-05,
+      "loss": 1.3304,
+      "step": 7455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0450472638180387e-05,
+      "loss": 1.3419,
+      "step": 7456
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.045025140328744e-05,
+      "loss": 1.2703,
+      "step": 7457
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0450030168394492e-05,
+      "loss": 1.351,
+      "step": 7458
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0449808933501545e-05,
+      "loss": 1.2864,
+      "step": 7459
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0449587698608597e-05,
+      "loss": 1.2952,
+      "step": 7460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.044936646371565e-05,
+      "loss": 1.2988,
+      "step": 7461
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0449145228822701e-05,
+      "loss": 1.2731,
+      "step": 7462
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0448923993929752e-05,
+      "loss": 1.2466,
+      "step": 7463
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0448702759036804e-05,
+      "loss": 1.3075,
+      "step": 7464
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0448481524143857e-05,
+      "loss": 1.2986,
+      "step": 7465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.044826028925091e-05,
+      "loss": 1.2805,
+      "step": 7466
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.044803905435796e-05,
+      "loss": 1.2793,
+      "step": 7467
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0447817819465013e-05,
+      "loss": 1.4721,
+      "step": 7468
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0447596584572066e-05,
+      "loss": 1.3008,
+      "step": 7469
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0447375349679118e-05,
+      "loss": 1.3314,
+      "step": 7470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.044715411478617e-05,
+      "loss": 1.2385,
+      "step": 7471
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0446932879893222e-05,
+      "loss": 1.2351,
+      "step": 7472
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0446711645000273e-05,
+      "loss": 1.3771,
+      "step": 7473
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0446490410107325e-05,
+      "loss": 1.4067,
+      "step": 7474
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0446269175214378e-05,
+      "loss": 1.4509,
+      "step": 7475
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.044604794032143e-05,
+      "loss": 1.3789,
+      "step": 7476
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0445826705428483e-05,
+      "loss": 1.2979,
+      "step": 7477
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0445605470535534e-05,
+      "loss": 1.3704,
+      "step": 7478
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0445384235642585e-05,
+      "loss": 1.3523,
+      "step": 7479
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0445163000749638e-05,
+      "loss": 1.3513,
+      "step": 7480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.044494176585669e-05,
+      "loss": 1.3614,
+      "step": 7481
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0444720530963743e-05,
+      "loss": 1.3263,
+      "step": 7482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0444499296070796e-05,
+      "loss": 1.1996,
+      "step": 7483
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0444278061177846e-05,
+      "loss": 1.3139,
+      "step": 7484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0444056826284899e-05,
+      "loss": 1.3487,
+      "step": 7485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0443835591391952e-05,
+      "loss": 1.3564,
+      "step": 7486
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0443614356499003e-05,
+      "loss": 1.3367,
+      "step": 7487
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0443393121606055e-05,
+      "loss": 1.3431,
+      "step": 7488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0443171886713106e-05,
+      "loss": 1.4765,
+      "step": 7489
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0442950651820159e-05,
+      "loss": 1.3153,
+      "step": 7490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0442729416927211e-05,
+      "loss": 1.3285,
+      "step": 7491
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0442508182034264e-05,
+      "loss": 1.3782,
+      "step": 7492
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0442286947141317e-05,
+      "loss": 1.3117,
+      "step": 7493
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0442065712248367e-05,
+      "loss": 1.3908,
+      "step": 7494
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.044184447735542e-05,
+      "loss": 1.3316,
+      "step": 7495
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0441623242462471e-05,
+      "loss": 1.2531,
+      "step": 7496
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0441402007569524e-05,
+      "loss": 1.4586,
+      "step": 7497
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0441180772676576e-05,
+      "loss": 1.2307,
+      "step": 7498
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0440959537783629e-05,
+      "loss": 1.2133,
+      "step": 7499
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.044073830289068e-05,
+      "loss": 1.3011,
+      "step": 7500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0440517067997732e-05,
+      "loss": 1.2248,
+      "step": 7501
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0440295833104785e-05,
+      "loss": 1.2475,
+      "step": 7502
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0440074598211836e-05,
+      "loss": 1.2151,
+      "step": 7503
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0439853363318889e-05,
+      "loss": 1.2406,
+      "step": 7504
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.043963212842594e-05,
+      "loss": 1.316,
+      "step": 7505
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0439410893532992e-05,
+      "loss": 1.2415,
+      "step": 7506
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0439189658640045e-05,
+      "loss": 1.4973,
+      "step": 7507
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0438968423747097e-05,
+      "loss": 1.272,
+      "step": 7508
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.043874718885415e-05,
+      "loss": 1.3158,
+      "step": 7509
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.04385259539612e-05,
+      "loss": 1.2575,
+      "step": 7510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0438304719068253e-05,
+      "loss": 1.2635,
+      "step": 7511
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0438083484175304e-05,
+      "loss": 1.3515,
+      "step": 7512
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0437862249282357e-05,
+      "loss": 1.4012,
+      "step": 7513
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.043764101438941e-05,
+      "loss": 1.4111,
+      "step": 7514
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0437419779496462e-05,
+      "loss": 1.4069,
+      "step": 7515
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0437198544603513e-05,
+      "loss": 1.3044,
+      "step": 7516
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0436977309710566e-05,
+      "loss": 1.332,
+      "step": 7517
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0436756074817618e-05,
+      "loss": 1.4328,
+      "step": 7518
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.043653483992467e-05,
+      "loss": 1.3039,
+      "step": 7519
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0436313605031722e-05,
+      "loss": 1.168,
+      "step": 7520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0436092370138774e-05,
+      "loss": 1.2244,
+      "step": 7521
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0435871135245825e-05,
+      "loss": 1.2879,
+      "step": 7522
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0435649900352878e-05,
+      "loss": 1.2809,
+      "step": 7523
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.043542866545993e-05,
+      "loss": 1.3221,
+      "step": 7524
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0435207430566983e-05,
+      "loss": 1.3051,
+      "step": 7525
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0434986195674036e-05,
+      "loss": 1.2339,
+      "step": 7526
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0434764960781087e-05,
+      "loss": 1.3332,
+      "step": 7527
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0434543725888138e-05,
+      "loss": 1.3627,
+      "step": 7528
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.043432249099519e-05,
+      "loss": 1.3625,
+      "step": 7529
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0434101256102243e-05,
+      "loss": 1.2733,
+      "step": 7530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0433880021209296e-05,
+      "loss": 1.2958,
+      "step": 7531
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0433658786316348e-05,
+      "loss": 1.3372,
+      "step": 7532
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0433437551423399e-05,
+      "loss": 1.2737,
+      "step": 7533
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0433216316530452e-05,
+      "loss": 1.3574,
+      "step": 7534
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0432995081637503e-05,
+      "loss": 1.2129,
+      "step": 7535
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0432773846744555e-05,
+      "loss": 1.2837,
+      "step": 7536
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0432552611851608e-05,
+      "loss": 1.2796,
+      "step": 7537
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0432331376958659e-05,
+      "loss": 1.2711,
+      "step": 7538
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0432110142065711e-05,
+      "loss": 1.3759,
+      "step": 7539
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0431888907172764e-05,
+      "loss": 1.3813,
+      "step": 7540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0431667672279817e-05,
+      "loss": 1.1962,
+      "step": 7541
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0431446437386869e-05,
+      "loss": 1.2199,
+      "step": 7542
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.043122520249392e-05,
+      "loss": 1.3756,
+      "step": 7543
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0431003967600971e-05,
+      "loss": 1.3466,
+      "step": 7544
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0430782732708024e-05,
+      "loss": 1.2641,
+      "step": 7545
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0430561497815076e-05,
+      "loss": 1.2036,
+      "step": 7546
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0430340262922129e-05,
+      "loss": 1.2152,
+      "step": 7547
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0430119028029181e-05,
+      "loss": 1.2574,
+      "step": 7548
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0429897793136232e-05,
+      "loss": 1.2303,
+      "step": 7549
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0429676558243285e-05,
+      "loss": 1.2057,
+      "step": 7550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0429455323350336e-05,
+      "loss": 1.3992,
+      "step": 7551
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0429234088457388e-05,
+      "loss": 1.3248,
+      "step": 7552
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0429012853564441e-05,
+      "loss": 1.1723,
+      "step": 7553
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0428791618671494e-05,
+      "loss": 1.241,
+      "step": 7554
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0428570383778545e-05,
+      "loss": 1.4844,
+      "step": 7555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0428349148885597e-05,
+      "loss": 1.2076,
+      "step": 7556
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.042812791399265e-05,
+      "loss": 1.4613,
+      "step": 7557
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0427906679099702e-05,
+      "loss": 1.356,
+      "step": 7558
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0427685444206753e-05,
+      "loss": 1.3458,
+      "step": 7559
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0427464209313804e-05,
+      "loss": 1.3371,
+      "step": 7560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0427242974420857e-05,
+      "loss": 1.1914,
+      "step": 7561
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.042702173952791e-05,
+      "loss": 1.222,
+      "step": 7562
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0426800504634962e-05,
+      "loss": 1.4114,
+      "step": 7563
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0426579269742015e-05,
+      "loss": 1.445,
+      "step": 7564
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0426358034849067e-05,
+      "loss": 1.403,
+      "step": 7565
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0426136799956118e-05,
+      "loss": 1.3439,
+      "step": 7566
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.042591556506317e-05,
+      "loss": 1.4286,
+      "step": 7567
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0425694330170222e-05,
+      "loss": 1.3494,
+      "step": 7568
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0425473095277274e-05,
+      "loss": 1.3533,
+      "step": 7569
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0425251860384327e-05,
+      "loss": 1.3217,
+      "step": 7570
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0425030625491378e-05,
+      "loss": 1.3879,
+      "step": 7571
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.042480939059843e-05,
+      "loss": 1.2309,
+      "step": 7572
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0424588155705483e-05,
+      "loss": 1.4005,
+      "step": 7573
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0424366920812536e-05,
+      "loss": 1.4051,
+      "step": 7574
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0424145685919587e-05,
+      "loss": 1.2889,
+      "step": 7575
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.042392445102664e-05,
+      "loss": 1.3747,
+      "step": 7576
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.042370321613369e-05,
+      "loss": 1.2285,
+      "step": 7577
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0423481981240743e-05,
+      "loss": 1.1828,
+      "step": 7578
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0423260746347795e-05,
+      "loss": 1.3186,
+      "step": 7579
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0423039511454848e-05,
+      "loss": 1.2547,
+      "step": 7580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.04228182765619e-05,
+      "loss": 1.3327,
+      "step": 7581
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0422597041668952e-05,
+      "loss": 1.3117,
+      "step": 7582
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0422375806776004e-05,
+      "loss": 1.3201,
+      "step": 7583
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0422154571883055e-05,
+      "loss": 1.2806,
+      "step": 7584
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0421933336990108e-05,
+      "loss": 1.3935,
+      "step": 7585
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.042171210209716e-05,
+      "loss": 1.4051,
+      "step": 7586
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0421490867204211e-05,
+      "loss": 1.3907,
+      "step": 7587
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0421269632311264e-05,
+      "loss": 1.2921,
+      "step": 7588
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0421048397418316e-05,
+      "loss": 1.3215,
+      "step": 7589
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0420827162525369e-05,
+      "loss": 1.2716,
+      "step": 7590
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.042060592763242e-05,
+      "loss": 1.3215,
+      "step": 7591
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0420384692739473e-05,
+      "loss": 1.2372,
+      "step": 7592
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0420163457846524e-05,
+      "loss": 1.2607,
+      "step": 7593
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0419942222953576e-05,
+      "loss": 1.4014,
+      "step": 7594
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0419720988060629e-05,
+      "loss": 1.3031,
+      "step": 7595
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0419499753167681e-05,
+      "loss": 1.2816,
+      "step": 7596
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0419278518274734e-05,
+      "loss": 1.2903,
+      "step": 7597
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0419057283381785e-05,
+      "loss": 1.3739,
+      "step": 7598
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0418836048488838e-05,
+      "loss": 1.3494,
+      "step": 7599
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0418614813595888e-05,
+      "loss": 1.2812,
+      "step": 7600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0418393578702941e-05,
+      "loss": 1.2711,
+      "step": 7601
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0418172343809994e-05,
+      "loss": 1.3226,
+      "step": 7602
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0417951108917046e-05,
+      "loss": 1.3674,
+      "step": 7603
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0417729874024097e-05,
+      "loss": 1.2887,
+      "step": 7604
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.041750863913115e-05,
+      "loss": 1.3746,
+      "step": 7605
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0417287404238202e-05,
+      "loss": 1.27,
+      "step": 7606
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0417066169345253e-05,
+      "loss": 1.3402,
+      "step": 7607
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0416844934452306e-05,
+      "loss": 1.2228,
+      "step": 7608
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0416623699559357e-05,
+      "loss": 1.132,
+      "step": 7609
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.041640246466641e-05,
+      "loss": 1.2452,
+      "step": 7610
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0416181229773462e-05,
+      "loss": 1.2677,
+      "step": 7611
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0415959994880515e-05,
+      "loss": 1.2725,
+      "step": 7612
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0415738759987567e-05,
+      "loss": 1.3439,
+      "step": 7613
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0415517525094618e-05,
+      "loss": 1.4036,
+      "step": 7614
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0415296290201671e-05,
+      "loss": 1.2393,
+      "step": 7615
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0415075055308722e-05,
+      "loss": 1.4084,
+      "step": 7616
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0414853820415774e-05,
+      "loss": 1.3114,
+      "step": 7617
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0414632585522827e-05,
+      "loss": 1.2486,
+      "step": 7618
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.041441135062988e-05,
+      "loss": 1.3204,
+      "step": 7619
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.041419011573693e-05,
+      "loss": 1.4163,
+      "step": 7620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0413968880843983e-05,
+      "loss": 1.275,
+      "step": 7621
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0413747645951036e-05,
+      "loss": 1.4434,
+      "step": 7622
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0413526411058087e-05,
+      "loss": 1.4297,
+      "step": 7623
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.041330517616514e-05,
+      "loss": 1.3243,
+      "step": 7624
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0413083941272192e-05,
+      "loss": 1.2711,
+      "step": 7625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0412862706379243e-05,
+      "loss": 1.2486,
+      "step": 7626
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0412641471486295e-05,
+      "loss": 1.4229,
+      "step": 7627
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0412420236593348e-05,
+      "loss": 1.2473,
+      "step": 7628
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.04121990017004e-05,
+      "loss": 1.3184,
+      "step": 7629
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0411977766807453e-05,
+      "loss": 1.2111,
+      "step": 7630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0411756531914504e-05,
+      "loss": 1.1859,
+      "step": 7631
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0411535297021555e-05,
+      "loss": 1.2715,
+      "step": 7632
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0411314062128608e-05,
+      "loss": 1.2608,
+      "step": 7633
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.041109282723566e-05,
+      "loss": 1.3959,
+      "step": 7634
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0410871592342713e-05,
+      "loss": 1.1964,
+      "step": 7635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0410650357449766e-05,
+      "loss": 1.2698,
+      "step": 7636
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0410429122556816e-05,
+      "loss": 1.2648,
+      "step": 7637
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0410207887663869e-05,
+      "loss": 1.2691,
+      "step": 7638
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.040998665277092e-05,
+      "loss": 1.2433,
+      "step": 7639
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0409765417877973e-05,
+      "loss": 1.4341,
+      "step": 7640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0409544182985025e-05,
+      "loss": 1.2144,
+      "step": 7641
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0409322948092076e-05,
+      "loss": 1.3098,
+      "step": 7642
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0409101713199129e-05,
+      "loss": 1.3838,
+      "step": 7643
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0408880478306181e-05,
+      "loss": 1.2502,
+      "step": 7644
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0408659243413234e-05,
+      "loss": 1.3241,
+      "step": 7645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0408438008520287e-05,
+      "loss": 1.211,
+      "step": 7646
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0408216773627337e-05,
+      "loss": 1.2722,
+      "step": 7647
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0407995538734388e-05,
+      "loss": 1.2967,
+      "step": 7648
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0407774303841441e-05,
+      "loss": 1.3858,
+      "step": 7649
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0407553068948494e-05,
+      "loss": 1.3859,
+      "step": 7650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0407331834055546e-05,
+      "loss": 1.2879,
+      "step": 7651
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0407110599162599e-05,
+      "loss": 1.342,
+      "step": 7652
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.040688936426965e-05,
+      "loss": 1.4316,
+      "step": 7653
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0406668129376702e-05,
+      "loss": 1.3429,
+      "step": 7654
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0406446894483753e-05,
+      "loss": 1.3129,
+      "step": 7655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0406225659590806e-05,
+      "loss": 1.2905,
+      "step": 7656
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0406004424697859e-05,
+      "loss": 1.2534,
+      "step": 7657
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0405783189804911e-05,
+      "loss": 1.3697,
+      "step": 7658
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0405561954911962e-05,
+      "loss": 1.3374,
+      "step": 7659
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0405340720019015e-05,
+      "loss": 1.1565,
+      "step": 7660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0405119485126067e-05,
+      "loss": 1.2475,
+      "step": 7661
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.040489825023312e-05,
+      "loss": 1.4257,
+      "step": 7662
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.040467701534017e-05,
+      "loss": 1.2412,
+      "step": 7663
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0404455780447222e-05,
+      "loss": 1.272,
+      "step": 7664
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0404234545554274e-05,
+      "loss": 1.2201,
+      "step": 7665
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0404013310661327e-05,
+      "loss": 1.1987,
+      "step": 7666
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.040379207576838e-05,
+      "loss": 1.215,
+      "step": 7667
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0403570840875432e-05,
+      "loss": 1.1943,
+      "step": 7668
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0403349605982485e-05,
+      "loss": 1.3456,
+      "step": 7669
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0403128371089536e-05,
+      "loss": 1.3428,
+      "step": 7670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0402907136196588e-05,
+      "loss": 1.36,
+      "step": 7671
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.040268590130364e-05,
+      "loss": 1.2179,
+      "step": 7672
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0402464666410692e-05,
+      "loss": 1.3394,
+      "step": 7673
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0402243431517744e-05,
+      "loss": 1.377,
+      "step": 7674
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0402022196624795e-05,
+      "loss": 1.3348,
+      "step": 7675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0401800961731848e-05,
+      "loss": 1.3442,
+      "step": 7676
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.04015797268389e-05,
+      "loss": 1.2905,
+      "step": 7677
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0401358491945953e-05,
+      "loss": 1.2488,
+      "step": 7678
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0401137257053004e-05,
+      "loss": 1.2679,
+      "step": 7679
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0400916022160057e-05,
+      "loss": 1.2718,
+      "step": 7680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0400694787267108e-05,
+      "loss": 1.3733,
+      "step": 7681
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.040047355237416e-05,
+      "loss": 1.3275,
+      "step": 7682
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0400252317481213e-05,
+      "loss": 1.4006,
+      "step": 7683
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0400031082588265e-05,
+      "loss": 1.3649,
+      "step": 7684
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0399809847695318e-05,
+      "loss": 1.178,
+      "step": 7685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0399588612802369e-05,
+      "loss": 1.2942,
+      "step": 7686
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0399367377909422e-05,
+      "loss": 1.1345,
+      "step": 7687
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0399146143016473e-05,
+      "loss": 1.3787,
+      "step": 7688
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0398924908123525e-05,
+      "loss": 1.1884,
+      "step": 7689
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0398703673230578e-05,
+      "loss": 1.2356,
+      "step": 7690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0398482438337629e-05,
+      "loss": 1.291,
+      "step": 7691
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0398261203444681e-05,
+      "loss": 1.3121,
+      "step": 7692
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0398039968551734e-05,
+      "loss": 1.2493,
+      "step": 7693
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0397818733658787e-05,
+      "loss": 1.2447,
+      "step": 7694
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0397597498765837e-05,
+      "loss": 1.2579,
+      "step": 7695
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.039737626387289e-05,
+      "loss": 1.3397,
+      "step": 7696
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0397155028979941e-05,
+      "loss": 1.1789,
+      "step": 7697
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0396933794086994e-05,
+      "loss": 1.4322,
+      "step": 7698
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0396712559194046e-05,
+      "loss": 1.2266,
+      "step": 7699
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0396491324301099e-05,
+      "loss": 1.3317,
+      "step": 7700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0396270089408151e-05,
+      "loss": 1.204,
+      "step": 7701
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0396048854515202e-05,
+      "loss": 1.1825,
+      "step": 7702
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0395827619622255e-05,
+      "loss": 1.2623,
+      "step": 7703
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0395606384729306e-05,
+      "loss": 1.3698,
+      "step": 7704
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0395385149836358e-05,
+      "loss": 1.2608,
+      "step": 7705
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0395163914943411e-05,
+      "loss": 1.4243,
+      "step": 7706
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0394942680050464e-05,
+      "loss": 1.3715,
+      "step": 7707
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0394721445157515e-05,
+      "loss": 1.318,
+      "step": 7708
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0394500210264567e-05,
+      "loss": 1.328,
+      "step": 7709
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.039427897537162e-05,
+      "loss": 1.3755,
+      "step": 7710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.039405774047867e-05,
+      "loss": 1.2816,
+      "step": 7711
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0393836505585723e-05,
+      "loss": 1.3422,
+      "step": 7712
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0393615270692774e-05,
+      "loss": 1.2849,
+      "step": 7713
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0393394035799827e-05,
+      "loss": 1.3111,
+      "step": 7714
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.039317280090688e-05,
+      "loss": 1.3397,
+      "step": 7715
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0392951566013932e-05,
+      "loss": 1.3038,
+      "step": 7716
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0392730331120985e-05,
+      "loss": 1.2905,
+      "step": 7717
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0392509096228037e-05,
+      "loss": 1.2975,
+      "step": 7718
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0392287861335088e-05,
+      "loss": 1.2795,
+      "step": 7719
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.039206662644214e-05,
+      "loss": 1.2936,
+      "step": 7720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0391845391549192e-05,
+      "loss": 1.2665,
+      "step": 7721
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0391624156656244e-05,
+      "loss": 1.4616,
+      "step": 7722
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0391402921763297e-05,
+      "loss": 1.3511,
+      "step": 7723
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0391181686870348e-05,
+      "loss": 1.3107,
+      "step": 7724
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.03909604519774e-05,
+      "loss": 1.1781,
+      "step": 7725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0390739217084453e-05,
+      "loss": 1.4227,
+      "step": 7726
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0390517982191504e-05,
+      "loss": 1.3339,
+      "step": 7727
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0390296747298557e-05,
+      "loss": 1.1622,
+      "step": 7728
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.039007551240561e-05,
+      "loss": 1.3666,
+      "step": 7729
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.038985427751266e-05,
+      "loss": 1.3585,
+      "step": 7730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0389633042619713e-05,
+      "loss": 1.3091,
+      "step": 7731
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0389411807726765e-05,
+      "loss": 1.4258,
+      "step": 7732
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0389190572833818e-05,
+      "loss": 1.3295,
+      "step": 7733
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.038896933794087e-05,
+      "loss": 1.2275,
+      "step": 7734
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0388748103047922e-05,
+      "loss": 1.4128,
+      "step": 7735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0388526868154973e-05,
+      "loss": 1.4338,
+      "step": 7736
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0388305633262025e-05,
+      "loss": 1.4686,
+      "step": 7737
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0388084398369078e-05,
+      "loss": 1.3599,
+      "step": 7738
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.038786316347613e-05,
+      "loss": 1.4184,
+      "step": 7739
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0387641928583183e-05,
+      "loss": 1.4137,
+      "step": 7740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0387420693690234e-05,
+      "loss": 1.2802,
+      "step": 7741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0387199458797286e-05,
+      "loss": 1.2738,
+      "step": 7742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0386978223904337e-05,
+      "loss": 1.3645,
+      "step": 7743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.038675698901139e-05,
+      "loss": 1.3284,
+      "step": 7744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0386535754118443e-05,
+      "loss": 1.2654,
+      "step": 7745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0386314519225494e-05,
+      "loss": 1.3378,
+      "step": 7746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0386093284332546e-05,
+      "loss": 1.3164,
+      "step": 7747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0385872049439599e-05,
+      "loss": 1.386,
+      "step": 7748
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0385650814546651e-05,
+      "loss": 1.3157,
+      "step": 7749
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0385429579653704e-05,
+      "loss": 1.1244,
+      "step": 7750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0385208344760755e-05,
+      "loss": 1.3642,
+      "step": 7751
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0384987109867806e-05,
+      "loss": 1.147,
+      "step": 7752
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0384765874974858e-05,
+      "loss": 1.2611,
+      "step": 7753
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0384544640081911e-05,
+      "loss": 1.2844,
+      "step": 7754
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0384323405188964e-05,
+      "loss": 1.3248,
+      "step": 7755
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0384102170296016e-05,
+      "loss": 1.2674,
+      "step": 7756
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0383880935403067e-05,
+      "loss": 1.3664,
+      "step": 7757
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.038365970051012e-05,
+      "loss": 1.3353,
+      "step": 7758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0383438465617172e-05,
+      "loss": 1.3567,
+      "step": 7759
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0383217230724223e-05,
+      "loss": 1.2935,
+      "step": 7760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0382995995831276e-05,
+      "loss": 1.2703,
+      "step": 7761
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0382774760938329e-05,
+      "loss": 1.4872,
+      "step": 7762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.038255352604538e-05,
+      "loss": 1.176,
+      "step": 7763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0382332291152432e-05,
+      "loss": 1.3271,
+      "step": 7764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0382111056259485e-05,
+      "loss": 1.2095,
+      "step": 7765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0381889821366537e-05,
+      "loss": 1.2956,
+      "step": 7766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0381668586473588e-05,
+      "loss": 1.2436,
+      "step": 7767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0381447351580639e-05,
+      "loss": 1.324,
+      "step": 7768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0381226116687692e-05,
+      "loss": 1.3254,
+      "step": 7769
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0381004881794744e-05,
+      "loss": 1.3348,
+      "step": 7770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0380783646901797e-05,
+      "loss": 1.3621,
+      "step": 7771
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.038056241200885e-05,
+      "loss": 1.4435,
+      "step": 7772
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0380341177115902e-05,
+      "loss": 1.4886,
+      "step": 7773
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0380119942222953e-05,
+      "loss": 1.3984,
+      "step": 7774
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0379898707330006e-05,
+      "loss": 1.2935,
+      "step": 7775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0379677472437057e-05,
+      "loss": 1.2022,
+      "step": 7776
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.037945623754411e-05,
+      "loss": 1.2959,
+      "step": 7777
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0379235002651162e-05,
+      "loss": 1.3687,
+      "step": 7778
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0379013767758213e-05,
+      "loss": 1.2659,
+      "step": 7779
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0378792532865265e-05,
+      "loss": 1.2833,
+      "step": 7780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0378571297972318e-05,
+      "loss": 1.3171,
+      "step": 7781
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.037835006307937e-05,
+      "loss": 1.3412,
+      "step": 7782
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0378128828186422e-05,
+      "loss": 1.2594,
+      "step": 7783
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0377907593293472e-05,
+      "loss": 1.2007,
+      "step": 7784
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0377686358400525e-05,
+      "loss": 1.3862,
+      "step": 7785
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0377465123507578e-05,
+      "loss": 1.4109,
+      "step": 7786
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.037724388861463e-05,
+      "loss": 1.411,
+      "step": 7787
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0377022653721683e-05,
+      "loss": 1.1906,
+      "step": 7788
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0376801418828736e-05,
+      "loss": 1.2594,
+      "step": 7789
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0376580183935786e-05,
+      "loss": 1.3718,
+      "step": 7790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0376358949042839e-05,
+      "loss": 1.2452,
+      "step": 7791
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.037613771414989e-05,
+      "loss": 1.3554,
+      "step": 7792
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0375916479256943e-05,
+      "loss": 1.3265,
+      "step": 7793
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0375695244363995e-05,
+      "loss": 1.3739,
+      "step": 7794
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0375474009471046e-05,
+      "loss": 1.2821,
+      "step": 7795
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0375252774578099e-05,
+      "loss": 1.3192,
+      "step": 7796
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0375031539685151e-05,
+      "loss": 1.3534,
+      "step": 7797
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0374810304792204e-05,
+      "loss": 1.315,
+      "step": 7798
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0374589069899255e-05,
+      "loss": 1.3531,
+      "step": 7799
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0374367835006307e-05,
+      "loss": 1.2014,
+      "step": 7800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0374146600113358e-05,
+      "loss": 1.3642,
+      "step": 7801
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0373925365220411e-05,
+      "loss": 1.3089,
+      "step": 7802
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0373704130327464e-05,
+      "loss": 1.3534,
+      "step": 7803
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0373482895434516e-05,
+      "loss": 1.3719,
+      "step": 7804
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0373261660541569e-05,
+      "loss": 1.406,
+      "step": 7805
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.037304042564862e-05,
+      "loss": 1.3598,
+      "step": 7806
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0372819190755672e-05,
+      "loss": 1.3298,
+      "step": 7807
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0372597955862723e-05,
+      "loss": 1.3673,
+      "step": 7808
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0372376720969776e-05,
+      "loss": 1.3978,
+      "step": 7809
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0372155486076829e-05,
+      "loss": 1.2567,
+      "step": 7810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0371934251183881e-05,
+      "loss": 1.3036,
+      "step": 7811
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0371713016290932e-05,
+      "loss": 1.3811,
+      "step": 7812
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0371491781397985e-05,
+      "loss": 1.2341,
+      "step": 7813
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0371270546505037e-05,
+      "loss": 1.301,
+      "step": 7814
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0371049311612088e-05,
+      "loss": 1.4632,
+      "step": 7815
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.037082807671914e-05,
+      "loss": 1.3688,
+      "step": 7816
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0370606841826192e-05,
+      "loss": 1.319,
+      "step": 7817
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0370385606933244e-05,
+      "loss": 1.3996,
+      "step": 7818
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0370164372040297e-05,
+      "loss": 1.3626,
+      "step": 7819
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.036994313714735e-05,
+      "loss": 1.3824,
+      "step": 7820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0369721902254402e-05,
+      "loss": 1.39,
+      "step": 7821
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0369500667361455e-05,
+      "loss": 1.3866,
+      "step": 7822
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0369279432468506e-05,
+      "loss": 1.3273,
+      "step": 7823
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0369058197575557e-05,
+      "loss": 1.2221,
+      "step": 7824
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.036883696268261e-05,
+      "loss": 1.2204,
+      "step": 7825
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0368615727789662e-05,
+      "loss": 1.3088,
+      "step": 7826
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0368394492896714e-05,
+      "loss": 1.3295,
+      "step": 7827
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0368173258003765e-05,
+      "loss": 1.428,
+      "step": 7828
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0367952023110818e-05,
+      "loss": 1.4292,
+      "step": 7829
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.036773078821787e-05,
+      "loss": 1.259,
+      "step": 7830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0367509553324921e-05,
+      "loss": 1.2959,
+      "step": 7831
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0367288318431974e-05,
+      "loss": 1.2711,
+      "step": 7832
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0367067083539027e-05,
+      "loss": 1.2745,
+      "step": 7833
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0366845848646078e-05,
+      "loss": 1.3194,
+      "step": 7834
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.036662461375313e-05,
+      "loss": 1.2445,
+      "step": 7835
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0366403378860183e-05,
+      "loss": 1.3405,
+      "step": 7836
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0366182143967235e-05,
+      "loss": 1.2991,
+      "step": 7837
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0365960909074288e-05,
+      "loss": 1.2333,
+      "step": 7838
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0365739674181339e-05,
+      "loss": 1.3124,
+      "step": 7839
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.036551843928839e-05,
+      "loss": 1.3012,
+      "step": 7840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0365297204395443e-05,
+      "loss": 1.2717,
+      "step": 7841
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0365075969502495e-05,
+      "loss": 1.3102,
+      "step": 7842
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0364854734609548e-05,
+      "loss": 1.3154,
+      "step": 7843
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.03646334997166e-05,
+      "loss": 1.201,
+      "step": 7844
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0364412264823651e-05,
+      "loss": 1.2035,
+      "step": 7845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0364191029930704e-05,
+      "loss": 1.2395,
+      "step": 7846
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0363969795037757e-05,
+      "loss": 1.2467,
+      "step": 7847
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0363748560144807e-05,
+      "loss": 1.2711,
+      "step": 7848
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.036352732525186e-05,
+      "loss": 1.2249,
+      "step": 7849
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0363306090358911e-05,
+      "loss": 1.2431,
+      "step": 7850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0363084855465964e-05,
+      "loss": 1.3444,
+      "step": 7851
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0362863620573016e-05,
+      "loss": 1.2946,
+      "step": 7852
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0362642385680069e-05,
+      "loss": 1.4284,
+      "step": 7853
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0362421150787121e-05,
+      "loss": 1.3868,
+      "step": 7854
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0362199915894172e-05,
+      "loss": 1.4138,
+      "step": 7855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0361978681001223e-05,
+      "loss": 1.3479,
+      "step": 7856
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0361757446108276e-05,
+      "loss": 1.314,
+      "step": 7857
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0361536211215328e-05,
+      "loss": 1.2379,
+      "step": 7858
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0361314976322381e-05,
+      "loss": 1.4188,
+      "step": 7859
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0361093741429434e-05,
+      "loss": 1.0905,
+      "step": 7860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0360872506536485e-05,
+      "loss": 1.2627,
+      "step": 7861
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0360651271643537e-05,
+      "loss": 1.4321,
+      "step": 7862
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.036043003675059e-05,
+      "loss": 1.4043,
+      "step": 7863
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.036020880185764e-05,
+      "loss": 1.2265,
+      "step": 7864
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0359987566964693e-05,
+      "loss": 1.247,
+      "step": 7865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0359766332071746e-05,
+      "loss": 1.321,
+      "step": 7866
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0359545097178797e-05,
+      "loss": 1.218,
+      "step": 7867
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.035932386228585e-05,
+      "loss": 1.2562,
+      "step": 7868
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0359102627392902e-05,
+      "loss": 1.3936,
+      "step": 7869
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0358881392499955e-05,
+      "loss": 1.4183,
+      "step": 7870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0358660157607006e-05,
+      "loss": 1.3493,
+      "step": 7871
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0358438922714057e-05,
+      "loss": 1.4329,
+      "step": 7872
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.035821768782111e-05,
+      "loss": 1.2388,
+      "step": 7873
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0357996452928162e-05,
+      "loss": 1.3186,
+      "step": 7874
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0357775218035214e-05,
+      "loss": 1.28,
+      "step": 7875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0357553983142267e-05,
+      "loss": 1.3055,
+      "step": 7876
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0357332748249318e-05,
+      "loss": 1.3923,
+      "step": 7877
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.035711151335637e-05,
+      "loss": 1.4806,
+      "step": 7878
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0356890278463423e-05,
+      "loss": 1.4378,
+      "step": 7879
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0356669043570474e-05,
+      "loss": 1.316,
+      "step": 7880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0356447808677527e-05,
+      "loss": 1.4524,
+      "step": 7881
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.035622657378458e-05,
+      "loss": 1.3259,
+      "step": 7882
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.035600533889163e-05,
+      "loss": 1.2453,
+      "step": 7883
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0355784103998683e-05,
+      "loss": 1.3786,
+      "step": 7884
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0355562869105735e-05,
+      "loss": 1.4957,
+      "step": 7885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0355341634212788e-05,
+      "loss": 1.365,
+      "step": 7886
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0355120399319839e-05,
+      "loss": 1.3246,
+      "step": 7887
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.035489916442689e-05,
+      "loss": 1.4174,
+      "step": 7888
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0354677929533942e-05,
+      "loss": 1.3561,
+      "step": 7889
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0354456694640995e-05,
+      "loss": 1.3032,
+      "step": 7890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0354235459748048e-05,
+      "loss": 1.3127,
+      "step": 7891
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.03540142248551e-05,
+      "loss": 1.37,
+      "step": 7892
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0353792989962153e-05,
+      "loss": 1.2375,
+      "step": 7893
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0353571755069204e-05,
+      "loss": 1.4138,
+      "step": 7894
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0353350520176256e-05,
+      "loss": 1.3826,
+      "step": 7895
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0353129285283307e-05,
+      "loss": 1.3072,
+      "step": 7896
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.035290805039036e-05,
+      "loss": 1.2566,
+      "step": 7897
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0352686815497413e-05,
+      "loss": 1.3839,
+      "step": 7898
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0352465580604464e-05,
+      "loss": 1.2221,
+      "step": 7899
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0352244345711516e-05,
+      "loss": 1.3031,
+      "step": 7900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0352023110818569e-05,
+      "loss": 1.2689,
+      "step": 7901
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0351801875925621e-05,
+      "loss": 1.3755,
+      "step": 7902
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0351580641032672e-05,
+      "loss": 1.3696,
+      "step": 7903
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0351359406139725e-05,
+      "loss": 1.4416,
+      "step": 7904
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0351138171246776e-05,
+      "loss": 1.3577,
+      "step": 7905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0350916936353828e-05,
+      "loss": 1.3926,
+      "step": 7906
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0350695701460881e-05,
+      "loss": 1.2522,
+      "step": 7907
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0350474466567934e-05,
+      "loss": 1.2816,
+      "step": 7908
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0350253231674986e-05,
+      "loss": 1.2167,
+      "step": 7909
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0350031996782037e-05,
+      "loss": 1.1509,
+      "step": 7910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.034981076188909e-05,
+      "loss": 1.2327,
+      "step": 7911
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.034958952699614e-05,
+      "loss": 1.4273,
+      "step": 7912
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0349368292103193e-05,
+      "loss": 1.2819,
+      "step": 7913
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0349147057210246e-05,
+      "loss": 1.3844,
+      "step": 7914
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0348925822317299e-05,
+      "loss": 1.2987,
+      "step": 7915
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.034870458742435e-05,
+      "loss": 1.254,
+      "step": 7916
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0348483352531402e-05,
+      "loss": 1.3447,
+      "step": 7917
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0348262117638455e-05,
+      "loss": 1.3508,
+      "step": 7918
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0348040882745506e-05,
+      "loss": 1.2259,
+      "step": 7919
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0347819647852558e-05,
+      "loss": 1.2378,
+      "step": 7920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0347598412959609e-05,
+      "loss": 1.2514,
+      "step": 7921
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0347377178066662e-05,
+      "loss": 1.2467,
+      "step": 7922
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0347155943173714e-05,
+      "loss": 1.2662,
+      "step": 7923
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0346934708280767e-05,
+      "loss": 1.2334,
+      "step": 7924
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.034671347338782e-05,
+      "loss": 1.2318,
+      "step": 7925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0346492238494872e-05,
+      "loss": 1.272,
+      "step": 7926
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0346271003601923e-05,
+      "loss": 1.387,
+      "step": 7927
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0346049768708974e-05,
+      "loss": 1.2911,
+      "step": 7928
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.0345828533816027e-05,
+      "loss": 1.3209,
+      "step": 7929
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.034560729892308e-05,
+      "loss": 1.2742,
+      "step": 7930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0345386064030132e-05,
+      "loss": 1.34,
+      "step": 7931
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0345164829137183e-05,
+      "loss": 1.2818,
+      "step": 7932
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0344943594244235e-05,
+      "loss": 1.3809,
+      "step": 7933
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0344722359351288e-05,
+      "loss": 1.2587,
+      "step": 7934
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.034450112445834e-05,
+      "loss": 1.3691,
+      "step": 7935
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0344279889565392e-05,
+      "loss": 1.2051,
+      "step": 7936
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0344058654672444e-05,
+      "loss": 1.3362,
+      "step": 7937
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0343837419779495e-05,
+      "loss": 1.3983,
+      "step": 7938
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0343616184886548e-05,
+      "loss": 1.3698,
+      "step": 7939
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.03433949499936e-05,
+      "loss": 1.2738,
+      "step": 7940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0343173715100653e-05,
+      "loss": 1.2898,
+      "step": 7941
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0342952480207706e-05,
+      "loss": 1.3377,
+      "step": 7942
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0342731245314756e-05,
+      "loss": 1.282,
+      "step": 7943
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0342510010421807e-05,
+      "loss": 1.4098,
+      "step": 7944
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.034228877552886e-05,
+      "loss": 1.2738,
+      "step": 7945
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0342067540635913e-05,
+      "loss": 1.2717,
+      "step": 7946
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0341846305742965e-05,
+      "loss": 1.2942,
+      "step": 7947
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0341625070850018e-05,
+      "loss": 1.3092,
+      "step": 7948
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0341403835957069e-05,
+      "loss": 1.2578,
+      "step": 7949
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0341182601064121e-05,
+      "loss": 1.19,
+      "step": 7950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0340961366171174e-05,
+      "loss": 1.3159,
+      "step": 7951
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0340740131278225e-05,
+      "loss": 1.3304,
+      "step": 7952
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0340518896385277e-05,
+      "loss": 1.2787,
+      "step": 7953
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0340297661492328e-05,
+      "loss": 1.2165,
+      "step": 7954
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0340076426599381e-05,
+      "loss": 1.3274,
+      "step": 7955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0339855191706434e-05,
+      "loss": 1.153,
+      "step": 7956
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0339633956813486e-05,
+      "loss": 1.2291,
+      "step": 7957
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0339412721920539e-05,
+      "loss": 1.295,
+      "step": 7958
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.033919148702759e-05,
+      "loss": 1.2017,
+      "step": 7959
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.033897025213464e-05,
+      "loss": 1.2877,
+      "step": 7960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0338749017241693e-05,
+      "loss": 1.377,
+      "step": 7961
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0338527782348746e-05,
+      "loss": 1.3929,
+      "step": 7962
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0338306547455798e-05,
+      "loss": 1.3087,
+      "step": 7963
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0338085312562851e-05,
+      "loss": 1.4001,
+      "step": 7964
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0337864077669902e-05,
+      "loss": 1.2305,
+      "step": 7965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0337642842776955e-05,
+      "loss": 1.2905,
+      "step": 7966
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0337421607884007e-05,
+      "loss": 1.2786,
+      "step": 7967
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0337200372991058e-05,
+      "loss": 1.2469,
+      "step": 7968
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.033697913809811e-05,
+      "loss": 1.2299,
+      "step": 7969
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0336757903205163e-05,
+      "loss": 1.2013,
+      "step": 7970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0336536668312214e-05,
+      "loss": 1.2843,
+      "step": 7971
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0336315433419267e-05,
+      "loss": 1.3032,
+      "step": 7972
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.033609419852632e-05,
+      "loss": 1.3519,
+      "step": 7973
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0335872963633372e-05,
+      "loss": 1.4014,
+      "step": 7974
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0335651728740423e-05,
+      "loss": 1.2857,
+      "step": 7975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0335430493847474e-05,
+      "loss": 1.2303,
+      "step": 7976
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0335209258954527e-05,
+      "loss": 1.1673,
+      "step": 7977
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.033498802406158e-05,
+      "loss": 1.2793,
+      "step": 7978
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0334766789168632e-05,
+      "loss": 1.2893,
+      "step": 7979
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0334545554275684e-05,
+      "loss": 1.3728,
+      "step": 7980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0334324319382735e-05,
+      "loss": 1.2996,
+      "step": 7981
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0334103084489788e-05,
+      "loss": 1.2991,
+      "step": 7982
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.033388184959684e-05,
+      "loss": 1.2036,
+      "step": 7983
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0333660614703891e-05,
+      "loss": 1.3748,
+      "step": 7984
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0333439379810944e-05,
+      "loss": 1.3748,
+      "step": 7985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0333218144917997e-05,
+      "loss": 1.3309,
+      "step": 7986
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0332996910025048e-05,
+      "loss": 1.3554,
+      "step": 7987
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.03327756751321e-05,
+      "loss": 1.3616,
+      "step": 7988
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0332554440239153e-05,
+      "loss": 1.3276,
+      "step": 7989
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0332333205346205e-05,
+      "loss": 1.2723,
+      "step": 7990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0332111970453256e-05,
+      "loss": 1.3401,
+      "step": 7991
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0331890735560309e-05,
+      "loss": 1.2135,
+      "step": 7992
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.033166950066736e-05,
+      "loss": 1.3482,
+      "step": 7993
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0331448265774413e-05,
+      "loss": 1.2716,
+      "step": 7994
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0331227030881465e-05,
+      "loss": 1.2985,
+      "step": 7995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0331005795988518e-05,
+      "loss": 1.2228,
+      "step": 7996
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.033078456109557e-05,
+      "loss": 1.2304,
+      "step": 7997
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0330563326202621e-05,
+      "loss": 1.1836,
+      "step": 7998
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0330342091309674e-05,
+      "loss": 1.3488,
+      "step": 7999
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0330120856416725e-05,
+      "loss": 1.3626,
+      "step": 8000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0329899621523777e-05,
+      "loss": 1.3516,
+      "step": 8001
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.032967838663083e-05,
+      "loss": 1.3921,
+      "step": 8002
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0329457151737881e-05,
+      "loss": 1.3092,
+      "step": 8003
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0329235916844934e-05,
+      "loss": 1.2977,
+      "step": 8004
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0329014681951986e-05,
+      "loss": 1.2518,
+      "step": 8005
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0328793447059039e-05,
+      "loss": 1.3474,
+      "step": 8006
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.032857221216609e-05,
+      "loss": 1.2901,
+      "step": 8007
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0328350977273142e-05,
+      "loss": 1.3144,
+      "step": 8008
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0328129742380193e-05,
+      "loss": 1.2839,
+      "step": 8009
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0327908507487246e-05,
+      "loss": 1.2994,
+      "step": 8010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0327687272594298e-05,
+      "loss": 1.3734,
+      "step": 8011
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0327466037701351e-05,
+      "loss": 1.5486,
+      "step": 8012
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0327244802808404e-05,
+      "loss": 1.494,
+      "step": 8013
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0327023567915455e-05,
+      "loss": 1.3014,
+      "step": 8014
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0326802333022507e-05,
+      "loss": 1.304,
+      "step": 8015
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0326581098129558e-05,
+      "loss": 1.2463,
+      "step": 8016
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.032635986323661e-05,
+      "loss": 1.269,
+      "step": 8017
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0326138628343663e-05,
+      "loss": 1.2427,
+      "step": 8018
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0325917393450716e-05,
+      "loss": 1.3581,
+      "step": 8019
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0325696158557767e-05,
+      "loss": 1.3541,
+      "step": 8020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.032547492366482e-05,
+      "loss": 1.2639,
+      "step": 8021
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0325253688771872e-05,
+      "loss": 1.1697,
+      "step": 8022
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0325032453878923e-05,
+      "loss": 1.2805,
+      "step": 8023
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0324811218985976e-05,
+      "loss": 1.276,
+      "step": 8024
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0324589984093027e-05,
+      "loss": 1.2807,
+      "step": 8025
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.032436874920008e-05,
+      "loss": 1.3788,
+      "step": 8026
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0324147514307132e-05,
+      "loss": 1.214,
+      "step": 8027
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0323926279414184e-05,
+      "loss": 1.2856,
+      "step": 8028
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0323705044521237e-05,
+      "loss": 1.2083,
+      "step": 8029
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.032348380962829e-05,
+      "loss": 1.2598,
+      "step": 8030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.032326257473534e-05,
+      "loss": 1.2131,
+      "step": 8031
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0323041339842391e-05,
+      "loss": 1.2815,
+      "step": 8032
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0322820104949444e-05,
+      "loss": 1.2694,
+      "step": 8033
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0322598870056497e-05,
+      "loss": 1.3215,
+      "step": 8034
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.032237763516355e-05,
+      "loss": 1.3142,
+      "step": 8035
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.03221564002706e-05,
+      "loss": 1.2152,
+      "step": 8036
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0321935165377653e-05,
+      "loss": 1.3696,
+      "step": 8037
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0321713930484705e-05,
+      "loss": 1.1466,
+      "step": 8038
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0321492695591758e-05,
+      "loss": 1.5053,
+      "step": 8039
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0321271460698809e-05,
+      "loss": 1.3061,
+      "step": 8040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0321050225805862e-05,
+      "loss": 1.2885,
+      "step": 8041
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0320828990912912e-05,
+      "loss": 1.2562,
+      "step": 8042
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0320607756019965e-05,
+      "loss": 1.401,
+      "step": 8043
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0320386521127018e-05,
+      "loss": 1.4787,
+      "step": 8044
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.032016528623407e-05,
+      "loss": 1.3637,
+      "step": 8045
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0319944051341123e-05,
+      "loss": 1.3199,
+      "step": 8046
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0319722816448174e-05,
+      "loss": 1.3001,
+      "step": 8047
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0319501581555225e-05,
+      "loss": 1.1801,
+      "step": 8048
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0319280346662277e-05,
+      "loss": 1.2384,
+      "step": 8049
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.031905911176933e-05,
+      "loss": 1.3798,
+      "step": 8050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0318837876876383e-05,
+      "loss": 1.4787,
+      "step": 8051
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0318616641983435e-05,
+      "loss": 1.2718,
+      "step": 8052
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0318395407090486e-05,
+      "loss": 1.3089,
+      "step": 8053
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0318174172197539e-05,
+      "loss": 1.3269,
+      "step": 8054
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0317952937304591e-05,
+      "loss": 1.2547,
+      "step": 8055
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0317731702411642e-05,
+      "loss": 1.2874,
+      "step": 8056
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0317510467518695e-05,
+      "loss": 1.3955,
+      "step": 8057
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0317289232625746e-05,
+      "loss": 1.3687,
+      "step": 8058
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0317067997732798e-05,
+      "loss": 1.4708,
+      "step": 8059
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0316846762839851e-05,
+      "loss": 1.3355,
+      "step": 8060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0316625527946904e-05,
+      "loss": 1.2738,
+      "step": 8061
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0316404293053956e-05,
+      "loss": 1.4913,
+      "step": 8062
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0316183058161007e-05,
+      "loss": 1.4084,
+      "step": 8063
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0315961823268058e-05,
+      "loss": 1.3167,
+      "step": 8064
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.031574058837511e-05,
+      "loss": 1.2062,
+      "step": 8065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0315519353482163e-05,
+      "loss": 1.308,
+      "step": 8066
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0315298118589216e-05,
+      "loss": 1.1797,
+      "step": 8067
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0315076883696269e-05,
+      "loss": 1.3366,
+      "step": 8068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.031485564880332e-05,
+      "loss": 1.485,
+      "step": 8069
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0314634413910372e-05,
+      "loss": 1.3997,
+      "step": 8070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0314413179017425e-05,
+      "loss": 1.3031,
+      "step": 8071
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0314191944124476e-05,
+      "loss": 1.3492,
+      "step": 8072
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0313970709231528e-05,
+      "loss": 1.251,
+      "step": 8073
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0313749474338579e-05,
+      "loss": 1.3231,
+      "step": 8074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0313528239445632e-05,
+      "loss": 1.4129,
+      "step": 8075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0313307004552684e-05,
+      "loss": 1.3111,
+      "step": 8076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0313085769659737e-05,
+      "loss": 1.3379,
+      "step": 8077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.031286453476679e-05,
+      "loss": 1.3588,
+      "step": 8078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.031264329987384e-05,
+      "loss": 1.1934,
+      "step": 8079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0312422064980893e-05,
+      "loss": 1.3157,
+      "step": 8080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0312200830087944e-05,
+      "loss": 1.3148,
+      "step": 8081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0311979595194997e-05,
+      "loss": 1.3527,
+      "step": 8082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.031175836030205e-05,
+      "loss": 1.3725,
+      "step": 8083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0311537125409102e-05,
+      "loss": 1.3999,
+      "step": 8084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0311315890516153e-05,
+      "loss": 1.343,
+      "step": 8085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0311094655623205e-05,
+      "loss": 1.2149,
+      "step": 8086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0310873420730258e-05,
+      "loss": 1.2504,
+      "step": 8087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0310652185837309e-05,
+      "loss": 1.3629,
+      "step": 8088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0310430950944362e-05,
+      "loss": 1.302,
+      "step": 8089
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0310209716051414e-05,
+      "loss": 1.4256,
+      "step": 8090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0309988481158465e-05,
+      "loss": 1.217,
+      "step": 8091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0309767246265518e-05,
+      "loss": 1.3163,
+      "step": 8092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.030954601137257e-05,
+      "loss": 1.2114,
+      "step": 8093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0309324776479623e-05,
+      "loss": 1.4282,
+      "step": 8094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0309103541586674e-05,
+      "loss": 1.1969,
+      "step": 8095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0308882306693726e-05,
+      "loss": 1.3153,
+      "step": 8096
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0308661071800777e-05,
+      "loss": 1.2845,
+      "step": 8097
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.030843983690783e-05,
+      "loss": 1.2995,
+      "step": 8098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0308218602014883e-05,
+      "loss": 1.3663,
+      "step": 8099
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0307997367121935e-05,
+      "loss": 1.2478,
+      "step": 8100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0307776132228988e-05,
+      "loss": 1.3953,
+      "step": 8101
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0307554897336039e-05,
+      "loss": 1.2692,
+      "step": 8102
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0307333662443091e-05,
+      "loss": 1.3049,
+      "step": 8103
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0307112427550142e-05,
+      "loss": 1.2465,
+      "step": 8104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0306891192657195e-05,
+      "loss": 1.3175,
+      "step": 8105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0306669957764247e-05,
+      "loss": 1.4055,
+      "step": 8106
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0306448722871298e-05,
+      "loss": 1.2719,
+      "step": 8107
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0306227487978351e-05,
+      "loss": 1.3251,
+      "step": 8108
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0306006253085404e-05,
+      "loss": 1.2474,
+      "step": 8109
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0305785018192456e-05,
+      "loss": 1.4246,
+      "step": 8110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0305563783299507e-05,
+      "loss": 1.214,
+      "step": 8111
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.030534254840656e-05,
+      "loss": 1.4717,
+      "step": 8112
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.030512131351361e-05,
+      "loss": 1.2188,
+      "step": 8113
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0304900078620663e-05,
+      "loss": 1.2185,
+      "step": 8114
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0304678843727716e-05,
+      "loss": 1.3909,
+      "step": 8115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0304457608834768e-05,
+      "loss": 1.4049,
+      "step": 8116
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0304236373941821e-05,
+      "loss": 1.3758,
+      "step": 8117
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0304015139048872e-05,
+      "loss": 1.2227,
+      "step": 8118
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0303793904155925e-05,
+      "loss": 1.3026,
+      "step": 8119
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0303572669262976e-05,
+      "loss": 1.2199,
+      "step": 8120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0303351434370028e-05,
+      "loss": 1.3338,
+      "step": 8121
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.030313019947708e-05,
+      "loss": 1.2849,
+      "step": 8122
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0302908964584133e-05,
+      "loss": 1.2505,
+      "step": 8123
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0302687729691184e-05,
+      "loss": 1.4367,
+      "step": 8124
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0302466494798237e-05,
+      "loss": 1.2537,
+      "step": 8125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.030224525990529e-05,
+      "loss": 1.3349,
+      "step": 8126
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0302024025012342e-05,
+      "loss": 1.3385,
+      "step": 8127
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0301802790119393e-05,
+      "loss": 1.4111,
+      "step": 8128
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0301581555226444e-05,
+      "loss": 1.3483,
+      "step": 8129
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0301360320333497e-05,
+      "loss": 1.3877,
+      "step": 8130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.030113908544055e-05,
+      "loss": 1.2183,
+      "step": 8131
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0300917850547602e-05,
+      "loss": 1.3523,
+      "step": 8132
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0300696615654654e-05,
+      "loss": 1.2315,
+      "step": 8133
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0300475380761707e-05,
+      "loss": 1.3968,
+      "step": 8134
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0300254145868758e-05,
+      "loss": 1.1704,
+      "step": 8135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0300032910975809e-05,
+      "loss": 1.2277,
+      "step": 8136
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0299811676082861e-05,
+      "loss": 1.2735,
+      "step": 8137
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0299590441189914e-05,
+      "loss": 1.3259,
+      "step": 8138
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0299369206296967e-05,
+      "loss": 1.3226,
+      "step": 8139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0299147971404018e-05,
+      "loss": 1.3179,
+      "step": 8140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.029892673651107e-05,
+      "loss": 1.3137,
+      "step": 8141
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0298705501618123e-05,
+      "loss": 1.3254,
+      "step": 8142
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0298484266725175e-05,
+      "loss": 1.2279,
+      "step": 8143
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0298263031832226e-05,
+      "loss": 1.252,
+      "step": 8144
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0298041796939279e-05,
+      "loss": 1.2622,
+      "step": 8145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.029782056204633e-05,
+      "loss": 1.4015,
+      "step": 8146
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0297599327153383e-05,
+      "loss": 1.3318,
+      "step": 8147
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0297378092260435e-05,
+      "loss": 1.351,
+      "step": 8148
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0297156857367488e-05,
+      "loss": 1.36,
+      "step": 8149
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.029693562247454e-05,
+      "loss": 1.2783,
+      "step": 8150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0296714387581591e-05,
+      "loss": 1.3364,
+      "step": 8151
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0296493152688642e-05,
+      "loss": 1.3124,
+      "step": 8152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0296271917795695e-05,
+      "loss": 1.3557,
+      "step": 8153
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0296050682902747e-05,
+      "loss": 1.4261,
+      "step": 8154
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.02958294480098e-05,
+      "loss": 1.3321,
+      "step": 8155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0295608213116853e-05,
+      "loss": 1.3562,
+      "step": 8156
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0295386978223904e-05,
+      "loss": 1.229,
+      "step": 8157
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0295165743330956e-05,
+      "loss": 1.2812,
+      "step": 8158
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0294944508438009e-05,
+      "loss": 1.2008,
+      "step": 8159
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.029472327354506e-05,
+      "loss": 1.3905,
+      "step": 8160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0294502038652112e-05,
+      "loss": 1.3585,
+      "step": 8161
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0294280803759163e-05,
+      "loss": 1.3308,
+      "step": 8162
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0294059568866216e-05,
+      "loss": 1.2926,
+      "step": 8163
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0293838333973268e-05,
+      "loss": 1.3181,
+      "step": 8164
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0293617099080321e-05,
+      "loss": 1.291,
+      "step": 8165
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0293395864187374e-05,
+      "loss": 1.2176,
+      "step": 8166
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0293174629294425e-05,
+      "loss": 1.3314,
+      "step": 8167
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0292953394401477e-05,
+      "loss": 1.3302,
+      "step": 8168
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0292732159508528e-05,
+      "loss": 1.3708,
+      "step": 8169
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.029251092461558e-05,
+      "loss": 1.3243,
+      "step": 8170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0292289689722633e-05,
+      "loss": 1.3213,
+      "step": 8171
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0292068454829686e-05,
+      "loss": 1.3142,
+      "step": 8172
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0291847219936737e-05,
+      "loss": 1.3037,
+      "step": 8173
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.029162598504379e-05,
+      "loss": 1.28,
+      "step": 8174
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0291404750150842e-05,
+      "loss": 1.4692,
+      "step": 8175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0291183515257893e-05,
+      "loss": 1.4372,
+      "step": 8176
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0290962280364946e-05,
+      "loss": 1.3307,
+      "step": 8177
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0290741045471997e-05,
+      "loss": 1.4218,
+      "step": 8178
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0290519810579049e-05,
+      "loss": 1.4041,
+      "step": 8179
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0290298575686102e-05,
+      "loss": 1.2905,
+      "step": 8180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0290077340793154e-05,
+      "loss": 1.4574,
+      "step": 8181
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0289856105900207e-05,
+      "loss": 1.3364,
+      "step": 8182
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0289634871007258e-05,
+      "loss": 1.2337,
+      "step": 8183
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.028941363611431e-05,
+      "loss": 1.2924,
+      "step": 8184
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0289192401221361e-05,
+      "loss": 1.3094,
+      "step": 8185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0288971166328414e-05,
+      "loss": 1.2374,
+      "step": 8186
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0288749931435467e-05,
+      "loss": 1.3875,
+      "step": 8187
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.028852869654252e-05,
+      "loss": 1.4103,
+      "step": 8188
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.028830746164957e-05,
+      "loss": 1.4337,
+      "step": 8189
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0288086226756623e-05,
+      "loss": 1.3239,
+      "step": 8190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0287864991863675e-05,
+      "loss": 1.5224,
+      "step": 8191
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0287643756970726e-05,
+      "loss": 1.2779,
+      "step": 8192
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0287422522077779e-05,
+      "loss": 1.3792,
+      "step": 8193
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0287201287184832e-05,
+      "loss": 1.304,
+      "step": 8194
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0286980052291882e-05,
+      "loss": 1.2571,
+      "step": 8195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0286758817398935e-05,
+      "loss": 1.4505,
+      "step": 8196
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0286537582505988e-05,
+      "loss": 1.3061,
+      "step": 8197
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.028631634761304e-05,
+      "loss": 1.2306,
+      "step": 8198
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0286095112720091e-05,
+      "loss": 1.3993,
+      "step": 8199
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0285873877827144e-05,
+      "loss": 1.3328,
+      "step": 8200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0285652642934195e-05,
+      "loss": 1.2737,
+      "step": 8201
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0285431408041247e-05,
+      "loss": 1.3277,
+      "step": 8202
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.02852101731483e-05,
+      "loss": 1.2948,
+      "step": 8203
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0284988938255353e-05,
+      "loss": 1.3125,
+      "step": 8204
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0284767703362405e-05,
+      "loss": 1.2547,
+      "step": 8205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0284546468469456e-05,
+      "loss": 1.3701,
+      "step": 8206
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0284325233576509e-05,
+      "loss": 1.3219,
+      "step": 8207
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.028410399868356e-05,
+      "loss": 1.261,
+      "step": 8208
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0283882763790612e-05,
+      "loss": 1.1814,
+      "step": 8209
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0283661528897665e-05,
+      "loss": 1.162,
+      "step": 8210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0283440294004716e-05,
+      "loss": 1.2812,
+      "step": 8211
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0283219059111768e-05,
+      "loss": 1.201,
+      "step": 8212
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0282997824218821e-05,
+      "loss": 1.2211,
+      "step": 8213
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0282776589325874e-05,
+      "loss": 1.1825,
+      "step": 8214
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0282555354432926e-05,
+      "loss": 1.294,
+      "step": 8215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0282334119539977e-05,
+      "loss": 1.2702,
+      "step": 8216
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0282112884647028e-05,
+      "loss": 1.2644,
+      "step": 8217
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.028189164975408e-05,
+      "loss": 1.1979,
+      "step": 8218
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0281670414861133e-05,
+      "loss": 1.3607,
+      "step": 8219
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0281449179968186e-05,
+      "loss": 1.2813,
+      "step": 8220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0281227945075239e-05,
+      "loss": 1.327,
+      "step": 8221
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.028100671018229e-05,
+      "loss": 1.2961,
+      "step": 8222
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0280785475289342e-05,
+      "loss": 1.3291,
+      "step": 8223
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0280564240396393e-05,
+      "loss": 1.1479,
+      "step": 8224
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0280343005503446e-05,
+      "loss": 1.2617,
+      "step": 8225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0280121770610498e-05,
+      "loss": 1.4007,
+      "step": 8226
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.027990053571755e-05,
+      "loss": 1.1843,
+      "step": 8227
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0279679300824602e-05,
+      "loss": 1.3979,
+      "step": 8228
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0279458065931654e-05,
+      "loss": 1.3552,
+      "step": 8229
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0279236831038707e-05,
+      "loss": 1.3727,
+      "step": 8230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.027901559614576e-05,
+      "loss": 1.288,
+      "step": 8231
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.027879436125281e-05,
+      "loss": 1.2393,
+      "step": 8232
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0278573126359861e-05,
+      "loss": 1.3167,
+      "step": 8233
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0278351891466914e-05,
+      "loss": 1.2168,
+      "step": 8234
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0278130656573967e-05,
+      "loss": 1.3483,
+      "step": 8235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.027790942168102e-05,
+      "loss": 1.3247,
+      "step": 8236
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0277688186788072e-05,
+      "loss": 1.1962,
+      "step": 8237
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0277466951895124e-05,
+      "loss": 1.188,
+      "step": 8238
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0277245717002175e-05,
+      "loss": 1.2174,
+      "step": 8239
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0277024482109226e-05,
+      "loss": 1.2714,
+      "step": 8240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0276803247216279e-05,
+      "loss": 1.2805,
+      "step": 8241
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0276582012323332e-05,
+      "loss": 1.3124,
+      "step": 8242
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0276360777430384e-05,
+      "loss": 1.2749,
+      "step": 8243
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0276139542537435e-05,
+      "loss": 1.3618,
+      "step": 8244
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0275918307644488e-05,
+      "loss": 1.3194,
+      "step": 8245
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.027569707275154e-05,
+      "loss": 1.3624,
+      "step": 8246
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0275475837858593e-05,
+      "loss": 1.4038,
+      "step": 8247
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0275254602965644e-05,
+      "loss": 1.2795,
+      "step": 8248
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0275033368072696e-05,
+      "loss": 1.2304,
+      "step": 8249
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0274812133179747e-05,
+      "loss": 1.3211,
+      "step": 8250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.02745908982868e-05,
+      "loss": 1.1684,
+      "step": 8251
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0274369663393853e-05,
+      "loss": 1.2925,
+      "step": 8252
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0274148428500905e-05,
+      "loss": 1.4166,
+      "step": 8253
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0273927193607958e-05,
+      "loss": 1.2795,
+      "step": 8254
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0273705958715009e-05,
+      "loss": 1.3466,
+      "step": 8255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0273484723822061e-05,
+      "loss": 1.3296,
+      "step": 8256
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0273263488929112e-05,
+      "loss": 1.3418,
+      "step": 8257
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0273042254036165e-05,
+      "loss": 1.2798,
+      "step": 8258
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0272821019143217e-05,
+      "loss": 1.2629,
+      "step": 8259
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.027259978425027e-05,
+      "loss": 1.2834,
+      "step": 8260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0272378549357321e-05,
+      "loss": 1.3649,
+      "step": 8261
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0272157314464374e-05,
+      "loss": 1.3102,
+      "step": 8262
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0271936079571426e-05,
+      "loss": 1.3331,
+      "step": 8263
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0271714844678477e-05,
+      "loss": 1.2518,
+      "step": 8264
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.027149360978553e-05,
+      "loss": 1.2262,
+      "step": 8265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.027127237489258e-05,
+      "loss": 1.2987,
+      "step": 8266
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0271051139999633e-05,
+      "loss": 1.2915,
+      "step": 8267
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0270829905106686e-05,
+      "loss": 1.2736,
+      "step": 8268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0270608670213738e-05,
+      "loss": 1.2638,
+      "step": 8269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0270387435320791e-05,
+      "loss": 1.113,
+      "step": 8270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0270166200427842e-05,
+      "loss": 1.325,
+      "step": 8271
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0269944965534895e-05,
+      "loss": 1.174,
+      "step": 8272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0269723730641946e-05,
+      "loss": 1.2863,
+      "step": 8273
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0269502495748998e-05,
+      "loss": 1.46,
+      "step": 8274
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.026928126085605e-05,
+      "loss": 1.4315,
+      "step": 8275
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0269060025963103e-05,
+      "loss": 1.311,
+      "step": 8276
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0268838791070154e-05,
+      "loss": 1.4676,
+      "step": 8277
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0268617556177207e-05,
+      "loss": 1.2059,
+      "step": 8278
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.026839632128426e-05,
+      "loss": 1.185,
+      "step": 8279
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.026817508639131e-05,
+      "loss": 1.2691,
+      "step": 8280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0267953851498363e-05,
+      "loss": 1.3528,
+      "step": 8281
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0267732616605414e-05,
+      "loss": 1.1442,
+      "step": 8282
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0267511381712467e-05,
+      "loss": 1.2016,
+      "step": 8283
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.026729014681952e-05,
+      "loss": 1.31,
+      "step": 8284
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0267068911926572e-05,
+      "loss": 1.2509,
+      "step": 8285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0266847677033624e-05,
+      "loss": 1.2494,
+      "step": 8286
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0266626442140675e-05,
+      "loss": 1.2702,
+      "step": 8287
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0266405207247728e-05,
+      "loss": 1.2311,
+      "step": 8288
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0266183972354779e-05,
+      "loss": 1.2842,
+      "step": 8289
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0265962737461831e-05,
+      "loss": 1.418,
+      "step": 8290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0265741502568884e-05,
+      "loss": 1.3337,
+      "step": 8291
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0265520267675937e-05,
+      "loss": 1.2132,
+      "step": 8292
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0265299032782988e-05,
+      "loss": 1.2974,
+      "step": 8293
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.026507779789004e-05,
+      "loss": 1.2243,
+      "step": 8294
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0264856562997093e-05,
+      "loss": 1.4093,
+      "step": 8295
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0264635328104144e-05,
+      "loss": 1.1982,
+      "step": 8296
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0264414093211196e-05,
+      "loss": 1.2161,
+      "step": 8297
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0264192858318249e-05,
+      "loss": 1.3971,
+      "step": 8298
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.02639716234253e-05,
+      "loss": 1.258,
+      "step": 8299
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0263750388532352e-05,
+      "loss": 1.1916,
+      "step": 8300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0263529153639405e-05,
+      "loss": 1.1001,
+      "step": 8301
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0263307918746458e-05,
+      "loss": 1.32,
+      "step": 8302
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.026308668385351e-05,
+      "loss": 1.2894,
+      "step": 8303
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0262865448960561e-05,
+      "loss": 1.2575,
+      "step": 8304
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0262644214067612e-05,
+      "loss": 1.3085,
+      "step": 8305
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0262422979174665e-05,
+      "loss": 1.335,
+      "step": 8306
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0262201744281717e-05,
+      "loss": 1.2265,
+      "step": 8307
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.026198050938877e-05,
+      "loss": 1.1785,
+      "step": 8308
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0261759274495823e-05,
+      "loss": 1.2077,
+      "step": 8309
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0261538039602874e-05,
+      "loss": 1.2611,
+      "step": 8310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0261316804709926e-05,
+      "loss": 1.2974,
+      "step": 8311
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0261095569816977e-05,
+      "loss": 1.3618,
+      "step": 8312
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.026087433492403e-05,
+      "loss": 1.262,
+      "step": 8313
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0260653100031082e-05,
+      "loss": 1.2523,
+      "step": 8314
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0260431865138133e-05,
+      "loss": 1.2468,
+      "step": 8315
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0260210630245186e-05,
+      "loss": 1.3581,
+      "step": 8316
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0259989395352238e-05,
+      "loss": 1.39,
+      "step": 8317
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0259768160459291e-05,
+      "loss": 1.2858,
+      "step": 8318
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0259546925566344e-05,
+      "loss": 1.3178,
+      "step": 8319
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0259325690673395e-05,
+      "loss": 1.2453,
+      "step": 8320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0259104455780445e-05,
+      "loss": 1.2574,
+      "step": 8321
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0258883220887498e-05,
+      "loss": 1.2963,
+      "step": 8322
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.025866198599455e-05,
+      "loss": 1.2473,
+      "step": 8323
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0258440751101603e-05,
+      "loss": 1.3516,
+      "step": 8324
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0258219516208656e-05,
+      "loss": 1.2193,
+      "step": 8325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0257998281315707e-05,
+      "loss": 1.3371,
+      "step": 8326
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.025777704642276e-05,
+      "loss": 1.2794,
+      "step": 8327
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.025755581152981e-05,
+      "loss": 1.3878,
+      "step": 8328
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0257334576636863e-05,
+      "loss": 1.2527,
+      "step": 8329
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0257113341743916e-05,
+      "loss": 1.4679,
+      "step": 8330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0256892106850968e-05,
+      "loss": 1.2181,
+      "step": 8331
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0256670871958019e-05,
+      "loss": 1.2385,
+      "step": 8332
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0256449637065072e-05,
+      "loss": 1.2279,
+      "step": 8333
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0256228402172124e-05,
+      "loss": 1.3817,
+      "step": 8334
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0256007167279177e-05,
+      "loss": 1.2415,
+      "step": 8335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0255785932386228e-05,
+      "loss": 1.3852,
+      "step": 8336
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0255564697493279e-05,
+      "loss": 1.274,
+      "step": 8337
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0255343462600331e-05,
+      "loss": 1.3035,
+      "step": 8338
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0255122227707384e-05,
+      "loss": 1.3407,
+      "step": 8339
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0254900992814437e-05,
+      "loss": 1.4387,
+      "step": 8340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.025467975792149e-05,
+      "loss": 1.2401,
+      "step": 8341
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0254458523028542e-05,
+      "loss": 1.3656,
+      "step": 8342
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0254237288135593e-05,
+      "loss": 1.2806,
+      "step": 8343
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0254016053242645e-05,
+      "loss": 1.3427,
+      "step": 8344
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0253794818349696e-05,
+      "loss": 1.3824,
+      "step": 8345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0253573583456749e-05,
+      "loss": 1.271,
+      "step": 8346
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0253352348563802e-05,
+      "loss": 1.1938,
+      "step": 8347
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0253131113670852e-05,
+      "loss": 1.1853,
+      "step": 8348
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0252909878777905e-05,
+      "loss": 1.4304,
+      "step": 8349
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0252688643884958e-05,
+      "loss": 1.3456,
+      "step": 8350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.025246740899201e-05,
+      "loss": 1.3185,
+      "step": 8351
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0252246174099061e-05,
+      "loss": 1.3207,
+      "step": 8352
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0252024939206114e-05,
+      "loss": 1.2574,
+      "step": 8353
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0251803704313165e-05,
+      "loss": 1.4024,
+      "step": 8354
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0251582469420217e-05,
+      "loss": 1.3026,
+      "step": 8355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.025136123452727e-05,
+      "loss": 1.277,
+      "step": 8356
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0251139999634323e-05,
+      "loss": 1.1712,
+      "step": 8357
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0250918764741375e-05,
+      "loss": 1.2891,
+      "step": 8358
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0250697529848426e-05,
+      "loss": 1.4934,
+      "step": 8359
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0250476294955479e-05,
+      "loss": 1.3891,
+      "step": 8360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.025025506006253e-05,
+      "loss": 1.1711,
+      "step": 8361
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0250033825169582e-05,
+      "loss": 1.2501,
+      "step": 8362
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0249812590276635e-05,
+      "loss": 1.2972,
+      "step": 8363
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0249591355383687e-05,
+      "loss": 1.4163,
+      "step": 8364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0249370120490738e-05,
+      "loss": 1.5662,
+      "step": 8365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0249148885597791e-05,
+      "loss": 1.3787,
+      "step": 8366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0248927650704844e-05,
+      "loss": 1.2321,
+      "step": 8367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0248706415811895e-05,
+      "loss": 1.3007,
+      "step": 8368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0248485180918947e-05,
+      "loss": 1.3124,
+      "step": 8369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0248263946025998e-05,
+      "loss": 1.2377,
+      "step": 8370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.024804271113305e-05,
+      "loss": 1.3524,
+      "step": 8371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0247821476240103e-05,
+      "loss": 1.3145,
+      "step": 8372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0247600241347156e-05,
+      "loss": 1.2968,
+      "step": 8373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0247379006454209e-05,
+      "loss": 1.3495,
+      "step": 8374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.024715777156126e-05,
+      "loss": 1.2897,
+      "step": 8375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0246936536668312e-05,
+      "loss": 1.1767,
+      "step": 8376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0246715301775363e-05,
+      "loss": 1.3777,
+      "step": 8377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0246494066882416e-05,
+      "loss": 1.2418,
+      "step": 8378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0246272831989468e-05,
+      "loss": 1.3194,
+      "step": 8379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.024605159709652e-05,
+      "loss": 1.2663,
+      "step": 8380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0245830362203572e-05,
+      "loss": 1.359,
+      "step": 8381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0245609127310624e-05,
+      "loss": 1.2369,
+      "step": 8382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0245387892417677e-05,
+      "loss": 1.2016,
+      "step": 8383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0245166657524728e-05,
+      "loss": 1.343,
+      "step": 8384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.024494542263178e-05,
+      "loss": 1.2557,
+      "step": 8385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0244724187738831e-05,
+      "loss": 1.2621,
+      "step": 8386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0244502952845884e-05,
+      "loss": 1.2132,
+      "step": 8387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0244281717952937e-05,
+      "loss": 1.2632,
+      "step": 8388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.024406048305999e-05,
+      "loss": 1.3453,
+      "step": 8389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0243839248167042e-05,
+      "loss": 1.3573,
+      "step": 8390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0243618013274094e-05,
+      "loss": 1.3223,
+      "step": 8391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0243396778381145e-05,
+      "loss": 1.3553,
+      "step": 8392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0243175543488196e-05,
+      "loss": 1.4953,
+      "step": 8393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0242954308595249e-05,
+      "loss": 1.2946,
+      "step": 8394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0242733073702301e-05,
+      "loss": 1.3835,
+      "step": 8395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0242511838809354e-05,
+      "loss": 1.2313,
+      "step": 8396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0242290603916405e-05,
+      "loss": 1.3601,
+      "step": 8397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0242069369023458e-05,
+      "loss": 1.3812,
+      "step": 8398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.024184813413051e-05,
+      "loss": 1.2343,
+      "step": 8399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0241626899237561e-05,
+      "loss": 1.421,
+      "step": 8400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0241405664344614e-05,
+      "loss": 1.3219,
+      "step": 8401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0241184429451666e-05,
+      "loss": 1.4213,
+      "step": 8402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0240963194558717e-05,
+      "loss": 1.3041,
+      "step": 8403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.024074195966577e-05,
+      "loss": 1.2991,
+      "step": 8404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0240520724772823e-05,
+      "loss": 1.3892,
+      "step": 8405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0240299489879875e-05,
+      "loss": 1.2904,
+      "step": 8406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0240078254986928e-05,
+      "loss": 1.2942,
+      "step": 8407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0239857020093979e-05,
+      "loss": 1.2718,
+      "step": 8408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.023963578520103e-05,
+      "loss": 1.4236,
+      "step": 8409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0239414550308082e-05,
+      "loss": 1.2256,
+      "step": 8410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0239193315415135e-05,
+      "loss": 1.2265,
+      "step": 8411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0238972080522187e-05,
+      "loss": 1.4015,
+      "step": 8412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.023875084562924e-05,
+      "loss": 1.3085,
+      "step": 8413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0238529610736291e-05,
+      "loss": 1.1329,
+      "step": 8414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0238308375843344e-05,
+      "loss": 1.3301,
+      "step": 8415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0238087140950394e-05,
+      "loss": 1.3387,
+      "step": 8416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0237865906057447e-05,
+      "loss": 1.2696,
+      "step": 8417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.02376446711645e-05,
+      "loss": 1.3718,
+      "step": 8418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.023742343627155e-05,
+      "loss": 1.3,
+      "step": 8419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0237202201378603e-05,
+      "loss": 1.2215,
+      "step": 8420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0236980966485656e-05,
+      "loss": 1.3735,
+      "step": 8421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0236759731592708e-05,
+      "loss": 1.3827,
+      "step": 8422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0236538496699761e-05,
+      "loss": 1.2375,
+      "step": 8423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0236317261806812e-05,
+      "loss": 1.192,
+      "step": 8424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0236096026913863e-05,
+      "loss": 1.1597,
+      "step": 8425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0235874792020916e-05,
+      "loss": 1.3395,
+      "step": 8426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0235653557127968e-05,
+      "loss": 1.3555,
+      "step": 8427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.023543232223502e-05,
+      "loss": 1.2959,
+      "step": 8428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0235211087342073e-05,
+      "loss": 1.3008,
+      "step": 8429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0234989852449124e-05,
+      "loss": 1.3453,
+      "step": 8430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0234768617556177e-05,
+      "loss": 1.3358,
+      "step": 8431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0234547382663228e-05,
+      "loss": 1.3567,
+      "step": 8432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.023432614777028e-05,
+      "loss": 1.203,
+      "step": 8433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0234104912877333e-05,
+      "loss": 1.2829,
+      "step": 8434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0233883677984386e-05,
+      "loss": 1.3553,
+      "step": 8435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0233662443091437e-05,
+      "loss": 1.322,
+      "step": 8436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.023344120819849e-05,
+      "loss": 1.4084,
+      "step": 8437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0233219973305542e-05,
+      "loss": 1.2768,
+      "step": 8438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0232998738412594e-05,
+      "loss": 1.3838,
+      "step": 8439
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0232777503519645e-05,
+      "loss": 1.4019,
+      "step": 8440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0232556268626696e-05,
+      "loss": 1.3418,
+      "step": 8441
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0232335033733749e-05,
+      "loss": 1.2214,
+      "step": 8442
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0232113798840801e-05,
+      "loss": 1.248,
+      "step": 8443
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0231892563947854e-05,
+      "loss": 1.1421,
+      "step": 8444
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0231671329054907e-05,
+      "loss": 1.2563,
+      "step": 8445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.023145009416196e-05,
+      "loss": 1.3924,
+      "step": 8446
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.023122885926901e-05,
+      "loss": 1.3382,
+      "step": 8447
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0231007624376063e-05,
+      "loss": 1.3762,
+      "step": 8448
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0230786389483114e-05,
+      "loss": 1.3416,
+      "step": 8449
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0230565154590166e-05,
+      "loss": 1.2144,
+      "step": 8450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0230343919697219e-05,
+      "loss": 1.2802,
+      "step": 8451
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.023012268480427e-05,
+      "loss": 1.1515,
+      "step": 8452
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0229901449911322e-05,
+      "loss": 1.2544,
+      "step": 8453
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0229680215018375e-05,
+      "loss": 1.3199,
+      "step": 8454
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0229458980125428e-05,
+      "loss": 1.3455,
+      "step": 8455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0229237745232479e-05,
+      "loss": 1.2905,
+      "step": 8456
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0229016510339531e-05,
+      "loss": 1.3818,
+      "step": 8457
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0228795275446582e-05,
+      "loss": 1.2973,
+      "step": 8458
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0228574040553635e-05,
+      "loss": 1.3356,
+      "step": 8459
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0228352805660687e-05,
+      "loss": 1.237,
+      "step": 8460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.022813157076774e-05,
+      "loss": 1.296,
+      "step": 8461
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0227910335874793e-05,
+      "loss": 1.3561,
+      "step": 8462
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0227689100981844e-05,
+      "loss": 1.3419,
+      "step": 8463
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0227467866088896e-05,
+      "loss": 1.2677,
+      "step": 8464
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0227246631195947e-05,
+      "loss": 1.3915,
+      "step": 8465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0227025396303e-05,
+      "loss": 1.2284,
+      "step": 8466
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0226804161410052e-05,
+      "loss": 1.3611,
+      "step": 8467
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0226582926517103e-05,
+      "loss": 1.2757,
+      "step": 8468
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0226361691624156e-05,
+      "loss": 1.3121,
+      "step": 8469
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0226140456731208e-05,
+      "loss": 1.2831,
+      "step": 8470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0225919221838261e-05,
+      "loss": 1.4005,
+      "step": 8471
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0225697986945312e-05,
+      "loss": 1.282,
+      "step": 8472
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0225476752052365e-05,
+      "loss": 1.2788,
+      "step": 8473
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0225255517159415e-05,
+      "loss": 1.2063,
+      "step": 8474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0225034282266468e-05,
+      "loss": 1.3714,
+      "step": 8475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.022481304737352e-05,
+      "loss": 1.3246,
+      "step": 8476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0224591812480573e-05,
+      "loss": 1.3423,
+      "step": 8477
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0224370577587626e-05,
+      "loss": 1.3368,
+      "step": 8478
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0224149342694677e-05,
+      "loss": 1.4333,
+      "step": 8479
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.022392810780173e-05,
+      "loss": 1.2549,
+      "step": 8480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.022370687290878e-05,
+      "loss": 1.2597,
+      "step": 8481
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0223485638015833e-05,
+      "loss": 1.3642,
+      "step": 8482
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0223264403122886e-05,
+      "loss": 1.305,
+      "step": 8483
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0223043168229938e-05,
+      "loss": 1.2735,
+      "step": 8484
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0222821933336989e-05,
+      "loss": 1.4329,
+      "step": 8485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0222600698444042e-05,
+      "loss": 1.2386,
+      "step": 8486
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0222379463551094e-05,
+      "loss": 1.2999,
+      "step": 8487
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0222158228658145e-05,
+      "loss": 1.2418,
+      "step": 8488
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0221936993765198e-05,
+      "loss": 1.2378,
+      "step": 8489
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0221715758872249e-05,
+      "loss": 1.2618,
+      "step": 8490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0221494523979301e-05,
+      "loss": 1.4482,
+      "step": 8491
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0221273289086354e-05,
+      "loss": 1.4165,
+      "step": 8492
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0221052054193407e-05,
+      "loss": 1.3106,
+      "step": 8493
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.022083081930046e-05,
+      "loss": 1.2578,
+      "step": 8494
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0220609584407512e-05,
+      "loss": 1.1961,
+      "step": 8495
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0220388349514563e-05,
+      "loss": 1.3001,
+      "step": 8496
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0220167114621614e-05,
+      "loss": 1.2749,
+      "step": 8497
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0219945879728666e-05,
+      "loss": 1.2495,
+      "step": 8498
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0219724644835719e-05,
+      "loss": 1.3097,
+      "step": 8499
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0219503409942772e-05,
+      "loss": 1.2184,
+      "step": 8500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0219282175049822e-05,
+      "loss": 1.4687,
+      "step": 8501
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0219060940156875e-05,
+      "loss": 1.3498,
+      "step": 8502
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0218839705263928e-05,
+      "loss": 1.2241,
+      "step": 8503
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0218618470370979e-05,
+      "loss": 1.3642,
+      "step": 8504
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0218397235478031e-05,
+      "loss": 1.184,
+      "step": 8505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0218176000585084e-05,
+      "loss": 1.2082,
+      "step": 8506
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0217954765692135e-05,
+      "loss": 1.2672,
+      "step": 8507
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0217733530799187e-05,
+      "loss": 1.2566,
+      "step": 8508
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.021751229590624e-05,
+      "loss": 1.2007,
+      "step": 8509
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0217291061013293e-05,
+      "loss": 1.281,
+      "step": 8510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0217069826120345e-05,
+      "loss": 1.3586,
+      "step": 8511
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0216848591227396e-05,
+      "loss": 1.2193,
+      "step": 8512
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0216627356334447e-05,
+      "loss": 1.4112,
+      "step": 8513
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.02164061214415e-05,
+      "loss": 1.2984,
+      "step": 8514
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0216184886548552e-05,
+      "loss": 1.2855,
+      "step": 8515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0215963651655605e-05,
+      "loss": 1.2293,
+      "step": 8516
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0215742416762657e-05,
+      "loss": 1.4718,
+      "step": 8517
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0215521181869708e-05,
+      "loss": 1.3522,
+      "step": 8518
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0215299946976761e-05,
+      "loss": 1.2759,
+      "step": 8519
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0215078712083812e-05,
+      "loss": 1.3618,
+      "step": 8520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0214857477190865e-05,
+      "loss": 1.2163,
+      "step": 8521
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0214636242297917e-05,
+      "loss": 1.2632,
+      "step": 8522
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0214415007404968e-05,
+      "loss": 1.2086,
+      "step": 8523
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.021419377251202e-05,
+      "loss": 1.3693,
+      "step": 8524
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0213972537619073e-05,
+      "loss": 1.3225,
+      "step": 8525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0213751302726126e-05,
+      "loss": 1.2193,
+      "step": 8526
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0213530067833178e-05,
+      "loss": 1.3249,
+      "step": 8527
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.021330883294023e-05,
+      "loss": 1.3724,
+      "step": 8528
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.021308759804728e-05,
+      "loss": 1.1339,
+      "step": 8529
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0212866363154333e-05,
+      "loss": 1.3565,
+      "step": 8530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0212645128261386e-05,
+      "loss": 1.269,
+      "step": 8531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0212423893368438e-05,
+      "loss": 1.3196,
+      "step": 8532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.021220265847549e-05,
+      "loss": 1.1609,
+      "step": 8533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0211981423582542e-05,
+      "loss": 1.2611,
+      "step": 8534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0211760188689594e-05,
+      "loss": 1.2838,
+      "step": 8535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0211538953796647e-05,
+      "loss": 1.2467,
+      "step": 8536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0211317718903698e-05,
+      "loss": 1.2477,
+      "step": 8537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.021109648401075e-05,
+      "loss": 1.2179,
+      "step": 8538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0210875249117803e-05,
+      "loss": 1.2771,
+      "step": 8539
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0210654014224854e-05,
+      "loss": 1.3006,
+      "step": 8540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0210432779331907e-05,
+      "loss": 1.3638,
+      "step": 8541
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.021021154443896e-05,
+      "loss": 1.3038,
+      "step": 8542
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0209990309546012e-05,
+      "loss": 1.3478,
+      "step": 8543
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0209769074653063e-05,
+      "loss": 1.2292,
+      "step": 8544
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0209547839760114e-05,
+      "loss": 1.3956,
+      "step": 8545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0209326604867166e-05,
+      "loss": 1.258,
+      "step": 8546
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0209105369974219e-05,
+      "loss": 1.2005,
+      "step": 8547
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0208884135081271e-05,
+      "loss": 1.3866,
+      "step": 8548
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0208662900188324e-05,
+      "loss": 1.3014,
+      "step": 8549
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0208441665295377e-05,
+      "loss": 1.2547,
+      "step": 8550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0208220430402428e-05,
+      "loss": 1.279,
+      "step": 8551
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.020799919550948e-05,
+      "loss": 1.2827,
+      "step": 8552
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0207777960616531e-05,
+      "loss": 1.2271,
+      "step": 8553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0207556725723584e-05,
+      "loss": 1.3503,
+      "step": 8554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0207335490830636e-05,
+      "loss": 1.3202,
+      "step": 8555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0207114255937687e-05,
+      "loss": 1.3105,
+      "step": 8556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.020689302104474e-05,
+      "loss": 1.4407,
+      "step": 8557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0206671786151793e-05,
+      "loss": 1.3908,
+      "step": 8558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0206450551258845e-05,
+      "loss": 1.243,
+      "step": 8559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0206229316365896e-05,
+      "loss": 1.4379,
+      "step": 8560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0206008081472949e-05,
+      "loss": 1.2402,
+      "step": 8561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.020578684658e-05,
+      "loss": 1.3359,
+      "step": 8562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0205565611687052e-05,
+      "loss": 1.3478,
+      "step": 8563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0205344376794105e-05,
+      "loss": 1.4029,
+      "step": 8564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0205123141901157e-05,
+      "loss": 1.3112,
+      "step": 8565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.020490190700821e-05,
+      "loss": 1.3281,
+      "step": 8566
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0204680672115261e-05,
+      "loss": 1.1792,
+      "step": 8567
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0204459437222314e-05,
+      "loss": 1.2714,
+      "step": 8568
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0204238202329364e-05,
+      "loss": 1.2987,
+      "step": 8569
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0204016967436417e-05,
+      "loss": 1.2459,
+      "step": 8570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.020379573254347e-05,
+      "loss": 1.2767,
+      "step": 8571
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.020357449765052e-05,
+      "loss": 1.336,
+      "step": 8572
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0203353262757573e-05,
+      "loss": 1.3207,
+      "step": 8573
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0203132027864626e-05,
+      "loss": 1.2286,
+      "step": 8574
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0202910792971678e-05,
+      "loss": 1.3758,
+      "step": 8575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.020268955807873e-05,
+      "loss": 1.4169,
+      "step": 8576
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0202468323185782e-05,
+      "loss": 1.2704,
+      "step": 8577
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0202247088292833e-05,
+      "loss": 1.2975,
+      "step": 8578
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0202025853399886e-05,
+      "loss": 1.3523,
+      "step": 8579
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0201804618506938e-05,
+      "loss": 1.3618,
+      "step": 8580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.020158338361399e-05,
+      "loss": 1.4329,
+      "step": 8581
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0201362148721043e-05,
+      "loss": 1.2393,
+      "step": 8582
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0201140913828094e-05,
+      "loss": 1.3456,
+      "step": 8583
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0200919678935147e-05,
+      "loss": 1.2936,
+      "step": 8584
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0200698444042198e-05,
+      "loss": 1.3238,
+      "step": 8585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.020047720914925e-05,
+      "loss": 1.1706,
+      "step": 8586
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0200255974256303e-05,
+      "loss": 1.3836,
+      "step": 8587
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0200034739363356e-05,
+      "loss": 1.1965,
+      "step": 8588
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0199813504470407e-05,
+      "loss": 1.3345,
+      "step": 8589
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0199592269577459e-05,
+      "loss": 1.3174,
+      "step": 8590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0199371034684512e-05,
+      "loss": 1.429,
+      "step": 8591
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0199149799791563e-05,
+      "loss": 1.3589,
+      "step": 8592
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0198928564898615e-05,
+      "loss": 1.3937,
+      "step": 8593
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0198707330005666e-05,
+      "loss": 1.3164,
+      "step": 8594
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0198486095112719e-05,
+      "loss": 1.5082,
+      "step": 8595
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0198264860219771e-05,
+      "loss": 1.3516,
+      "step": 8596
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0198043625326824e-05,
+      "loss": 1.1337,
+      "step": 8597
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0197822390433877e-05,
+      "loss": 1.2752,
+      "step": 8598
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.019760115554093e-05,
+      "loss": 1.3057,
+      "step": 8599
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.019737992064798e-05,
+      "loss": 1.3672,
+      "step": 8600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0197158685755031e-05,
+      "loss": 1.249,
+      "step": 8601
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0196937450862084e-05,
+      "loss": 1.2921,
+      "step": 8602
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0196716215969136e-05,
+      "loss": 1.321,
+      "step": 8603
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0196494981076189e-05,
+      "loss": 1.2594,
+      "step": 8604
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.019627374618324e-05,
+      "loss": 1.241,
+      "step": 8605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0196052511290292e-05,
+      "loss": 1.2878,
+      "step": 8606
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0195831276397345e-05,
+      "loss": 1.3401,
+      "step": 8607
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0195610041504396e-05,
+      "loss": 1.2461,
+      "step": 8608
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0195388806611449e-05,
+      "loss": 1.2414,
+      "step": 8609
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0195167571718501e-05,
+      "loss": 1.1897,
+      "step": 8610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0194946336825552e-05,
+      "loss": 1.2444,
+      "step": 8611
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0194725101932605e-05,
+      "loss": 1.3755,
+      "step": 8612
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0194503867039657e-05,
+      "loss": 1.3495,
+      "step": 8613
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.019428263214671e-05,
+      "loss": 1.2757,
+      "step": 8614
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0194061397253763e-05,
+      "loss": 1.3286,
+      "step": 8615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0193840162360814e-05,
+      "loss": 1.4316,
+      "step": 8616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0193618927467864e-05,
+      "loss": 1.2097,
+      "step": 8617
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0193397692574917e-05,
+      "loss": 1.3093,
+      "step": 8618
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.019317645768197e-05,
+      "loss": 1.3988,
+      "step": 8619
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0192955222789022e-05,
+      "loss": 1.3331,
+      "step": 8620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0192733987896075e-05,
+      "loss": 1.2313,
+      "step": 8621
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0192512753003126e-05,
+      "loss": 1.2878,
+      "step": 8622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0192291518110178e-05,
+      "loss": 1.1797,
+      "step": 8623
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0192070283217231e-05,
+      "loss": 1.2128,
+      "step": 8624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0191849048324282e-05,
+      "loss": 1.2144,
+      "step": 8625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0191627813431335e-05,
+      "loss": 1.2186,
+      "step": 8626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0191406578538385e-05,
+      "loss": 1.2647,
+      "step": 8627
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0191185343645438e-05,
+      "loss": 1.3699,
+      "step": 8628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.019096410875249e-05,
+      "loss": 1.4229,
+      "step": 8629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0190742873859543e-05,
+      "loss": 1.4788,
+      "step": 8630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0190521638966596e-05,
+      "loss": 1.4238,
+      "step": 8631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0190300404073647e-05,
+      "loss": 1.426,
+      "step": 8632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0190079169180698e-05,
+      "loss": 1.3682,
+      "step": 8633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.018985793428775e-05,
+      "loss": 1.3556,
+      "step": 8634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0189636699394803e-05,
+      "loss": 1.4468,
+      "step": 8635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0189415464501856e-05,
+      "loss": 1.3632,
+      "step": 8636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0189194229608908e-05,
+      "loss": 1.1247,
+      "step": 8637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0188972994715959e-05,
+      "loss": 1.3482,
+      "step": 8638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0188751759823012e-05,
+      "loss": 1.3958,
+      "step": 8639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0188530524930064e-05,
+      "loss": 1.3674,
+      "step": 8640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0188309290037115e-05,
+      "loss": 1.2019,
+      "step": 8641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0188088055144168e-05,
+      "loss": 1.3121,
+      "step": 8642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.018786682025122e-05,
+      "loss": 1.2131,
+      "step": 8643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0187645585358271e-05,
+      "loss": 1.3715,
+      "step": 8644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0187424350465324e-05,
+      "loss": 1.4401,
+      "step": 8645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0187203115572377e-05,
+      "loss": 1.3927,
+      "step": 8646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.018698188067943e-05,
+      "loss": 1.3671,
+      "step": 8647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.018676064578648e-05,
+      "loss": 1.344,
+      "step": 8648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0186539410893531e-05,
+      "loss": 1.2949,
+      "step": 8649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0186318176000584e-05,
+      "loss": 1.1843,
+      "step": 8650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0186096941107636e-05,
+      "loss": 1.3215,
+      "step": 8651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0185875706214689e-05,
+      "loss": 1.2889,
+      "step": 8652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0185654471321742e-05,
+      "loss": 1.4858,
+      "step": 8653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0185433236428794e-05,
+      "loss": 1.3072,
+      "step": 8654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0185212001535845e-05,
+      "loss": 1.2246,
+      "step": 8655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0184990766642898e-05,
+      "loss": 1.2727,
+      "step": 8656
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0184769531749949e-05,
+      "loss": 1.2336,
+      "step": 8657
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0184548296857001e-05,
+      "loss": 1.2353,
+      "step": 8658
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0184327061964054e-05,
+      "loss": 1.2304,
+      "step": 8659
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0184105827071105e-05,
+      "loss": 1.3205,
+      "step": 8660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0183884592178157e-05,
+      "loss": 1.2349,
+      "step": 8661
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.018366335728521e-05,
+      "loss": 1.1934,
+      "step": 8662
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0183442122392263e-05,
+      "loss": 1.3197,
+      "step": 8663
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0183220887499313e-05,
+      "loss": 1.2667,
+      "step": 8664
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0182999652606366e-05,
+      "loss": 1.2584,
+      "step": 8665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0182778417713417e-05,
+      "loss": 1.2484,
+      "step": 8666
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.018255718282047e-05,
+      "loss": 1.3067,
+      "step": 8667
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0182335947927522e-05,
+      "loss": 1.3618,
+      "step": 8668
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0182114713034575e-05,
+      "loss": 1.2986,
+      "step": 8669
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0181893478141627e-05,
+      "loss": 1.2873,
+      "step": 8670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0181672243248678e-05,
+      "loss": 1.2791,
+      "step": 8671
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0181451008355731e-05,
+      "loss": 1.2724,
+      "step": 8672
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0181229773462782e-05,
+      "loss": 1.4366,
+      "step": 8673
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0181008538569834e-05,
+      "loss": 1.2292,
+      "step": 8674
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0180787303676887e-05,
+      "loss": 1.3289,
+      "step": 8675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0180566068783938e-05,
+      "loss": 1.3209,
+      "step": 8676
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.018034483389099e-05,
+      "loss": 1.3277,
+      "step": 8677
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0180123598998043e-05,
+      "loss": 1.2925,
+      "step": 8678
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0179902364105096e-05,
+      "loss": 1.3398,
+      "step": 8679
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0179681129212147e-05,
+      "loss": 1.3318,
+      "step": 8680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.01794598943192e-05,
+      "loss": 1.2813,
+      "step": 8681
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.017923865942625e-05,
+      "loss": 1.3114,
+      "step": 8682
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0179017424533303e-05,
+      "loss": 1.2339,
+      "step": 8683
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0178796189640356e-05,
+      "loss": 1.3215,
+      "step": 8684
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0178574954747408e-05,
+      "loss": 1.3881,
+      "step": 8685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.017835371985446e-05,
+      "loss": 1.2546,
+      "step": 8686
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0178132484961512e-05,
+      "loss": 1.2785,
+      "step": 8687
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0177911250068564e-05,
+      "loss": 1.3427,
+      "step": 8688
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0177690015175615e-05,
+      "loss": 1.3405,
+      "step": 8689
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0177468780282668e-05,
+      "loss": 1.1663,
+      "step": 8690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.017724754538972e-05,
+      "loss": 1.2493,
+      "step": 8691
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0177026310496773e-05,
+      "loss": 1.2902,
+      "step": 8692
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0176805075603824e-05,
+      "loss": 1.3424,
+      "step": 8693
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0176583840710877e-05,
+      "loss": 1.3481,
+      "step": 8694
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.017636260581793e-05,
+      "loss": 1.3128,
+      "step": 8695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.017614137092498e-05,
+      "loss": 1.319,
+      "step": 8696
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0175920136032033e-05,
+      "loss": 1.2563,
+      "step": 8697
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0175698901139084e-05,
+      "loss": 1.2871,
+      "step": 8698
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0175477666246136e-05,
+      "loss": 1.304,
+      "step": 8699
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0175256431353189e-05,
+      "loss": 1.306,
+      "step": 8700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0175035196460241e-05,
+      "loss": 1.2432,
+      "step": 8701
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0174813961567294e-05,
+      "loss": 1.2684,
+      "step": 8702
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0174592726674347e-05,
+      "loss": 1.3114,
+      "step": 8703
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0174371491781398e-05,
+      "loss": 1.3579,
+      "step": 8704
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0174150256888449e-05,
+      "loss": 1.2905,
+      "step": 8705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0173929021995501e-05,
+      "loss": 1.3278,
+      "step": 8706
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0173707787102554e-05,
+      "loss": 1.167,
+      "step": 8707
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0173486552209606e-05,
+      "loss": 1.2921,
+      "step": 8708
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0173265317316657e-05,
+      "loss": 1.3663,
+      "step": 8709
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.017304408242371e-05,
+      "loss": 1.2759,
+      "step": 8710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0172822847530763e-05,
+      "loss": 1.2422,
+      "step": 8711
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0172601612637815e-05,
+      "loss": 1.2684,
+      "step": 8712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0172380377744866e-05,
+      "loss": 1.279,
+      "step": 8713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0172159142851919e-05,
+      "loss": 1.3786,
+      "step": 8714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.017193790795897e-05,
+      "loss": 1.2297,
+      "step": 8715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0171716673066022e-05,
+      "loss": 1.3592,
+      "step": 8716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0171495438173075e-05,
+      "loss": 1.2246,
+      "step": 8717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0171274203280127e-05,
+      "loss": 1.4362,
+      "step": 8718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.017105296838718e-05,
+      "loss": 1.2375,
+      "step": 8719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0170831733494231e-05,
+      "loss": 1.2726,
+      "step": 8720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0170610498601282e-05,
+      "loss": 1.2487,
+      "step": 8721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0170389263708334e-05,
+      "loss": 1.3243,
+      "step": 8722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0170168028815387e-05,
+      "loss": 1.2202,
+      "step": 8723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.016994679392244e-05,
+      "loss": 1.356,
+      "step": 8724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0169725559029492e-05,
+      "loss": 1.2607,
+      "step": 8725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0169504324136543e-05,
+      "loss": 1.3083,
+      "step": 8726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0169283089243596e-05,
+      "loss": 1.2483,
+      "step": 8727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0169061854350648e-05,
+      "loss": 1.2882,
+      "step": 8728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.01688406194577e-05,
+      "loss": 1.1889,
+      "step": 8729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0168619384564752e-05,
+      "loss": 1.2797,
+      "step": 8730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0168398149671803e-05,
+      "loss": 1.2636,
+      "step": 8731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0168176914778855e-05,
+      "loss": 1.2824,
+      "step": 8732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0167955679885908e-05,
+      "loss": 1.2723,
+      "step": 8733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.016773444499296e-05,
+      "loss": 1.2718,
+      "step": 8734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0167513210100013e-05,
+      "loss": 1.3686,
+      "step": 8735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0167291975207064e-05,
+      "loss": 1.2786,
+      "step": 8736
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0167070740314115e-05,
+      "loss": 1.3079,
+      "step": 8737
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0166849505421168e-05,
+      "loss": 1.2957,
+      "step": 8738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.016662827052822e-05,
+      "loss": 1.2149,
+      "step": 8739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0166407035635273e-05,
+      "loss": 1.4586,
+      "step": 8740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0166185800742326e-05,
+      "loss": 1.3108,
+      "step": 8741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0165964565849377e-05,
+      "loss": 1.2951,
+      "step": 8742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0165743330956429e-05,
+      "loss": 1.2864,
+      "step": 8743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0165522096063482e-05,
+      "loss": 1.3453,
+      "step": 8744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0165300861170533e-05,
+      "loss": 1.3825,
+      "step": 8745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0165079626277585e-05,
+      "loss": 1.29,
+      "step": 8746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0164858391384638e-05,
+      "loss": 1.3148,
+      "step": 8747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0164637156491689e-05,
+      "loss": 1.561,
+      "step": 8748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0164415921598741e-05,
+      "loss": 1.2827,
+      "step": 8749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0164194686705794e-05,
+      "loss": 1.3683,
+      "step": 8750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0163973451812847e-05,
+      "loss": 1.3562,
+      "step": 8751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0163752216919898e-05,
+      "loss": 1.3304,
+      "step": 8752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.016353098202695e-05,
+      "loss": 1.1923,
+      "step": 8753
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0163309747134001e-05,
+      "loss": 1.3964,
+      "step": 8754
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0163088512241054e-05,
+      "loss": 1.3731,
+      "step": 8755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0162867277348106e-05,
+      "loss": 1.2965,
+      "step": 8756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0162646042455159e-05,
+      "loss": 1.2531,
+      "step": 8757
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.016242480756221e-05,
+      "loss": 1.3014,
+      "step": 8758
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0162203572669262e-05,
+      "loss": 1.3867,
+      "step": 8759
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0161982337776315e-05,
+      "loss": 1.3683,
+      "step": 8760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0161761102883366e-05,
+      "loss": 1.2298,
+      "step": 8761
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0161539867990419e-05,
+      "loss": 1.2654,
+      "step": 8762
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0161318633097471e-05,
+      "loss": 1.3162,
+      "step": 8763
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0161097398204522e-05,
+      "loss": 1.1744,
+      "step": 8764
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0160876163311575e-05,
+      "loss": 1.3688,
+      "step": 8765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0160654928418627e-05,
+      "loss": 1.3486,
+      "step": 8766
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.016043369352568e-05,
+      "loss": 1.3584,
+      "step": 8767
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0160212458632731e-05,
+      "loss": 1.3355,
+      "step": 8768
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0159991223739783e-05,
+      "loss": 1.3639,
+      "step": 8769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0159769988846834e-05,
+      "loss": 1.2856,
+      "step": 8770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0159548753953887e-05,
+      "loss": 1.2364,
+      "step": 8771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.015932751906094e-05,
+      "loss": 1.2703,
+      "step": 8772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0159106284167992e-05,
+      "loss": 1.2056,
+      "step": 8773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0158885049275045e-05,
+      "loss": 1.2095,
+      "step": 8774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0158663814382096e-05,
+      "loss": 1.3193,
+      "step": 8775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0158442579489148e-05,
+      "loss": 1.2976,
+      "step": 8776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.01582213445962e-05,
+      "loss": 1.3295,
+      "step": 8777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0158000109703252e-05,
+      "loss": 1.2513,
+      "step": 8778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0157778874810305e-05,
+      "loss": 1.264,
+      "step": 8779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0157557639917355e-05,
+      "loss": 1.2336,
+      "step": 8780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0157336405024408e-05,
+      "loss": 1.1679,
+      "step": 8781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.015711517013146e-05,
+      "loss": 1.2964,
+      "step": 8782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0156893935238513e-05,
+      "loss": 1.4217,
+      "step": 8783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0156672700345564e-05,
+      "loss": 1.1998,
+      "step": 8784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0156451465452617e-05,
+      "loss": 1.2534,
+      "step": 8785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0156230230559668e-05,
+      "loss": 1.3525,
+      "step": 8786
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.015600899566672e-05,
+      "loss": 1.2724,
+      "step": 8787
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0155787760773773e-05,
+      "loss": 1.186,
+      "step": 8788
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0155566525880826e-05,
+      "loss": 1.2777,
+      "step": 8789
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0155345290987878e-05,
+      "loss": 1.2624,
+      "step": 8790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0155124056094929e-05,
+      "loss": 1.267,
+      "step": 8791
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0154902821201982e-05,
+      "loss": 1.3659,
+      "step": 8792
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0154681586309033e-05,
+      "loss": 1.2472,
+      "step": 8793
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0154460351416085e-05,
+      "loss": 1.4177,
+      "step": 8794
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0154239116523138e-05,
+      "loss": 1.3895,
+      "step": 8795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.015401788163019e-05,
+      "loss": 1.3608,
+      "step": 8796
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0153796646737241e-05,
+      "loss": 1.2735,
+      "step": 8797
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0153575411844294e-05,
+      "loss": 1.3188,
+      "step": 8798
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0153354176951347e-05,
+      "loss": 1.2936,
+      "step": 8799
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.01531329420584e-05,
+      "loss": 1.2582,
+      "step": 8800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.015291170716545e-05,
+      "loss": 1.2787,
+      "step": 8801
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0152690472272501e-05,
+      "loss": 1.2231,
+      "step": 8802
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0152469237379554e-05,
+      "loss": 1.166,
+      "step": 8803
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0152248002486606e-05,
+      "loss": 1.2146,
+      "step": 8804
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0152026767593659e-05,
+      "loss": 1.3713,
+      "step": 8805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0151805532700711e-05,
+      "loss": 1.2147,
+      "step": 8806
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0151584297807764e-05,
+      "loss": 1.4035,
+      "step": 8807
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0151363062914815e-05,
+      "loss": 1.2718,
+      "step": 8808
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0151141828021866e-05,
+      "loss": 1.2243,
+      "step": 8809
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0150920593128919e-05,
+      "loss": 1.235,
+      "step": 8810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0150699358235971e-05,
+      "loss": 1.4275,
+      "step": 8811
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0150478123343024e-05,
+      "loss": 1.2479,
+      "step": 8812
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0150256888450075e-05,
+      "loss": 1.2212,
+      "step": 8813
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0150035653557127e-05,
+      "loss": 1.408,
+      "step": 8814
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.014981441866418e-05,
+      "loss": 1.3907,
+      "step": 8815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0149593183771233e-05,
+      "loss": 1.2511,
+      "step": 8816
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0149371948878283e-05,
+      "loss": 1.2107,
+      "step": 8817
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0149150713985336e-05,
+      "loss": 1.2272,
+      "step": 8818
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0148929479092387e-05,
+      "loss": 1.3801,
+      "step": 8819
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.014870824419944e-05,
+      "loss": 1.171,
+      "step": 8820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0148487009306492e-05,
+      "loss": 1.221,
+      "step": 8821
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0148265774413545e-05,
+      "loss": 1.1812,
+      "step": 8822
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0148044539520597e-05,
+      "loss": 1.279,
+      "step": 8823
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0147823304627648e-05,
+      "loss": 1.248,
+      "step": 8824
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.01476020697347e-05,
+      "loss": 1.2358,
+      "step": 8825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0147380834841752e-05,
+      "loss": 1.3469,
+      "step": 8826
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0147159599948804e-05,
+      "loss": 1.3087,
+      "step": 8827
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0146938365055857e-05,
+      "loss": 1.2589,
+      "step": 8828
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.014671713016291e-05,
+      "loss": 1.2126,
+      "step": 8829
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.014649589526996e-05,
+      "loss": 1.347,
+      "step": 8830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0146274660377013e-05,
+      "loss": 1.4309,
+      "step": 8831
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0146053425484066e-05,
+      "loss": 1.3664,
+      "step": 8832
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0145832190591117e-05,
+      "loss": 1.3472,
+      "step": 8833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.014561095569817e-05,
+      "loss": 1.153,
+      "step": 8834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.014538972080522e-05,
+      "loss": 1.3219,
+      "step": 8835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0145168485912273e-05,
+      "loss": 1.3964,
+      "step": 8836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0144947251019326e-05,
+      "loss": 1.467,
+      "step": 8837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0144726016126378e-05,
+      "loss": 1.4374,
+      "step": 8838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.014450478123343e-05,
+      "loss": 1.2284,
+      "step": 8839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0144283546340482e-05,
+      "loss": 1.3704,
+      "step": 8840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0144062311447533e-05,
+      "loss": 1.2504,
+      "step": 8841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0143841076554585e-05,
+      "loss": 1.1874,
+      "step": 8842
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0143619841661638e-05,
+      "loss": 1.2935,
+      "step": 8843
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.014339860676869e-05,
+      "loss": 1.3479,
+      "step": 8844
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0143177371875743e-05,
+      "loss": 1.3117,
+      "step": 8845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0142956136982794e-05,
+      "loss": 1.4255,
+      "step": 8846
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0142734902089847e-05,
+      "loss": 1.4248,
+      "step": 8847
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.01425136671969e-05,
+      "loss": 1.2481,
+      "step": 8848
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.014229243230395e-05,
+      "loss": 1.2489,
+      "step": 8849
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0142071197411003e-05,
+      "loss": 1.1731,
+      "step": 8850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0141849962518055e-05,
+      "loss": 1.286,
+      "step": 8851
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0141628727625106e-05,
+      "loss": 1.2614,
+      "step": 8852
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0141407492732159e-05,
+      "loss": 1.223,
+      "step": 8853
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0141186257839211e-05,
+      "loss": 1.2064,
+      "step": 8854
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0140965022946264e-05,
+      "loss": 1.2744,
+      "step": 8855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0140743788053315e-05,
+      "loss": 1.2141,
+      "step": 8856
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0140522553160368e-05,
+      "loss": 1.3509,
+      "step": 8857
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0140301318267419e-05,
+      "loss": 1.3005,
+      "step": 8858
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0140080083374471e-05,
+      "loss": 1.3844,
+      "step": 8859
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0139858848481524e-05,
+      "loss": 1.3205,
+      "step": 8860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0139637613588576e-05,
+      "loss": 1.2732,
+      "step": 8861
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0139416378695627e-05,
+      "loss": 1.1939,
+      "step": 8862
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.013919514380268e-05,
+      "loss": 1.2181,
+      "step": 8863
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0138973908909732e-05,
+      "loss": 1.3613,
+      "step": 8864
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0138752674016783e-05,
+      "loss": 1.2983,
+      "step": 8865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0138531439123836e-05,
+      "loss": 1.3036,
+      "step": 8866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0138310204230889e-05,
+      "loss": 1.4016,
+      "step": 8867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.013808896933794e-05,
+      "loss": 1.2633,
+      "step": 8868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0137867734444992e-05,
+      "loss": 1.3485,
+      "step": 8869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0137646499552045e-05,
+      "loss": 1.2515,
+      "step": 8870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0137425264659097e-05,
+      "loss": 1.3054,
+      "step": 8871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0137204029766148e-05,
+      "loss": 1.3119,
+      "step": 8872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0136982794873201e-05,
+      "loss": 1.231,
+      "step": 8873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0136761559980252e-05,
+      "loss": 1.3543,
+      "step": 8874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0136540325087304e-05,
+      "loss": 1.229,
+      "step": 8875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0136319090194357e-05,
+      "loss": 1.1796,
+      "step": 8876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.013609785530141e-05,
+      "loss": 1.2782,
+      "step": 8877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0135876620408462e-05,
+      "loss": 1.2452,
+      "step": 8878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0135655385515513e-05,
+      "loss": 1.2896,
+      "step": 8879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0135434150622566e-05,
+      "loss": 1.1941,
+      "step": 8880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0135212915729617e-05,
+      "loss": 1.2463,
+      "step": 8881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.013499168083667e-05,
+      "loss": 1.2265,
+      "step": 8882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0134770445943722e-05,
+      "loss": 1.4558,
+      "step": 8883
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0134549211050773e-05,
+      "loss": 1.2421,
+      "step": 8884
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0134327976157825e-05,
+      "loss": 1.2414,
+      "step": 8885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0134106741264878e-05,
+      "loss": 1.2654,
+      "step": 8886
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.013388550637193e-05,
+      "loss": 1.4046,
+      "step": 8887
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0133664271478983e-05,
+      "loss": 1.2884,
+      "step": 8888
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0133443036586034e-05,
+      "loss": 1.2335,
+      "step": 8889
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0133221801693085e-05,
+      "loss": 1.377,
+      "step": 8890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0133000566800138e-05,
+      "loss": 1.2256,
+      "step": 8891
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.013277933190719e-05,
+      "loss": 1.3559,
+      "step": 8892
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0132558097014243e-05,
+      "loss": 1.4275,
+      "step": 8893
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0132336862121296e-05,
+      "loss": 1.3658,
+      "step": 8894
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0132115627228347e-05,
+      "loss": 1.3063,
+      "step": 8895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0131894392335399e-05,
+      "loss": 1.2291,
+      "step": 8896
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.013167315744245e-05,
+      "loss": 1.4198,
+      "step": 8897
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0131451922549503e-05,
+      "loss": 1.1896,
+      "step": 8898
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0131230687656555e-05,
+      "loss": 1.3679,
+      "step": 8899
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0131009452763608e-05,
+      "loss": 1.3169,
+      "step": 8900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0130788217870659e-05,
+      "loss": 1.1965,
+      "step": 8901
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0130566982977711e-05,
+      "loss": 1.2848,
+      "step": 8902
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0130345748084764e-05,
+      "loss": 1.3535,
+      "step": 8903
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0130124513191817e-05,
+      "loss": 1.4356,
+      "step": 8904
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0129903278298868e-05,
+      "loss": 1.3015,
+      "step": 8905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0129682043405918e-05,
+      "loss": 1.2036,
+      "step": 8906
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0129460808512971e-05,
+      "loss": 1.2943,
+      "step": 8907
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0129239573620024e-05,
+      "loss": 1.2754,
+      "step": 8908
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0129018338727076e-05,
+      "loss": 1.2352,
+      "step": 8909
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0128797103834129e-05,
+      "loss": 1.2707,
+      "step": 8910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0128575868941182e-05,
+      "loss": 1.2484,
+      "step": 8911
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0128354634048232e-05,
+      "loss": 1.2026,
+      "step": 8912
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0128133399155283e-05,
+      "loss": 1.5328,
+      "step": 8913
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0127912164262336e-05,
+      "loss": 1.2481,
+      "step": 8914
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0127690929369389e-05,
+      "loss": 1.2225,
+      "step": 8915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0127469694476441e-05,
+      "loss": 1.3069,
+      "step": 8916
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0127248459583492e-05,
+      "loss": 1.3259,
+      "step": 8917
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0127027224690545e-05,
+      "loss": 1.3013,
+      "step": 8918
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0126805989797597e-05,
+      "loss": 1.308,
+      "step": 8919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.012658475490465e-05,
+      "loss": 1.1889,
+      "step": 8920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0126363520011701e-05,
+      "loss": 1.4485,
+      "step": 8921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0126142285118753e-05,
+      "loss": 1.2635,
+      "step": 8922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0125921050225804e-05,
+      "loss": 1.3392,
+      "step": 8923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0125699815332857e-05,
+      "loss": 1.2524,
+      "step": 8924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.012547858043991e-05,
+      "loss": 1.3589,
+      "step": 8925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0125257345546962e-05,
+      "loss": 1.2164,
+      "step": 8926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0125036110654015e-05,
+      "loss": 1.3937,
+      "step": 8927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0124814875761066e-05,
+      "loss": 1.3178,
+      "step": 8928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0124593640868117e-05,
+      "loss": 1.2999,
+      "step": 8929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.012437240597517e-05,
+      "loss": 1.4129,
+      "step": 8930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0124151171082222e-05,
+      "loss": 1.2035,
+      "step": 8931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0123929936189275e-05,
+      "loss": 1.261,
+      "step": 8932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0123708701296327e-05,
+      "loss": 1.3364,
+      "step": 8933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0123487466403378e-05,
+      "loss": 1.3139,
+      "step": 8934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.012326623151043e-05,
+      "loss": 1.1515,
+      "step": 8935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0123044996617483e-05,
+      "loss": 1.2317,
+      "step": 8936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0122823761724534e-05,
+      "loss": 1.2938,
+      "step": 8937
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0122602526831587e-05,
+      "loss": 1.2522,
+      "step": 8938
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0122381291938638e-05,
+      "loss": 1.2771,
+      "step": 8939
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.012216005704569e-05,
+      "loss": 1.3455,
+      "step": 8940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0121938822152743e-05,
+      "loss": 1.3248,
+      "step": 8941
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0121717587259796e-05,
+      "loss": 1.2175,
+      "step": 8942
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0121496352366848e-05,
+      "loss": 1.3778,
+      "step": 8943
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0121275117473899e-05,
+      "loss": 1.2929,
+      "step": 8944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0121053882580952e-05,
+      "loss": 1.2834,
+      "step": 8945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0120832647688003e-05,
+      "loss": 1.2561,
+      "step": 8946
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0120611412795055e-05,
+      "loss": 1.3959,
+      "step": 8947
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0120390177902108e-05,
+      "loss": 1.362,
+      "step": 8948
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.012016894300916e-05,
+      "loss": 1.2937,
+      "step": 8949
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0119947708116211e-05,
+      "loss": 1.1793,
+      "step": 8950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0119726473223264e-05,
+      "loss": 1.3666,
+      "step": 8951
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0119505238330317e-05,
+      "loss": 1.353,
+      "step": 8952
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0119284003437368e-05,
+      "loss": 1.2738,
+      "step": 8953
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.011906276854442e-05,
+      "loss": 1.2809,
+      "step": 8954
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0118841533651471e-05,
+      "loss": 1.2943,
+      "step": 8955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0118620298758524e-05,
+      "loss": 1.2548,
+      "step": 8956
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0118399063865576e-05,
+      "loss": 1.2804,
+      "step": 8957
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0118177828972629e-05,
+      "loss": 1.1863,
+      "step": 8958
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0117956594079681e-05,
+      "loss": 1.1838,
+      "step": 8959
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0117735359186732e-05,
+      "loss": 1.2815,
+      "step": 8960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0117514124293785e-05,
+      "loss": 1.3197,
+      "step": 8961
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0117292889400836e-05,
+      "loss": 1.3969,
+      "step": 8962
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0117071654507889e-05,
+      "loss": 1.2561,
+      "step": 8963
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0116850419614941e-05,
+      "loss": 1.2864,
+      "step": 8964
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0116629184721994e-05,
+      "loss": 1.2765,
+      "step": 8965
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0116407949829045e-05,
+      "loss": 1.3358,
+      "step": 8966
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0116186714936097e-05,
+      "loss": 1.3382,
+      "step": 8967
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.011596548004315e-05,
+      "loss": 1.3437,
+      "step": 8968
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.01157442451502e-05,
+      "loss": 1.3093,
+      "step": 8969
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0115523010257253e-05,
+      "loss": 1.3202,
+      "step": 8970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0115301775364306e-05,
+      "loss": 1.241,
+      "step": 8971
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0115080540471357e-05,
+      "loss": 1.3656,
+      "step": 8972
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.011485930557841e-05,
+      "loss": 1.2141,
+      "step": 8973
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0114638070685462e-05,
+      "loss": 1.4499,
+      "step": 8974
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0114416835792515e-05,
+      "loss": 1.3726,
+      "step": 8975
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0114195600899567e-05,
+      "loss": 1.2552,
+      "step": 8976
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0113974366006618e-05,
+      "loss": 1.3013,
+      "step": 8977
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.011375313111367e-05,
+      "loss": 1.5741,
+      "step": 8978
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0113531896220722e-05,
+      "loss": 1.1421,
+      "step": 8979
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0113310661327774e-05,
+      "loss": 1.1974,
+      "step": 8980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0113089426434827e-05,
+      "loss": 1.333,
+      "step": 8981
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.011286819154188e-05,
+      "loss": 1.4257,
+      "step": 8982
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.011264695664893e-05,
+      "loss": 1.3012,
+      "step": 8983
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0112425721755983e-05,
+      "loss": 1.3647,
+      "step": 8984
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0112204486863034e-05,
+      "loss": 1.3981,
+      "step": 8985
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0111983251970087e-05,
+      "loss": 1.3562,
+      "step": 8986
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.011176201707714e-05,
+      "loss": 1.3369,
+      "step": 8987
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.011154078218419e-05,
+      "loss": 1.2836,
+      "step": 8988
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0111319547291243e-05,
+      "loss": 1.3332,
+      "step": 8989
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0111098312398296e-05,
+      "loss": 1.2986,
+      "step": 8990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0110877077505348e-05,
+      "loss": 1.3028,
+      "step": 8991
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.01106558426124e-05,
+      "loss": 1.1518,
+      "step": 8992
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0110434607719452e-05,
+      "loss": 1.2618,
+      "step": 8993
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0110213372826503e-05,
+      "loss": 1.3427,
+      "step": 8994
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0109992137933555e-05,
+      "loss": 1.2681,
+      "step": 8995
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0109770903040608e-05,
+      "loss": 1.2931,
+      "step": 8996
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.010954966814766e-05,
+      "loss": 1.2709,
+      "step": 8997
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0109328433254713e-05,
+      "loss": 1.2884,
+      "step": 8998
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0109107198361764e-05,
+      "loss": 1.257,
+      "step": 8999
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0108885963468817e-05,
+      "loss": 1.5944,
+      "step": 9000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0108664728575867e-05,
+      "loss": 1.2969,
+      "step": 9001
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.010844349368292e-05,
+      "loss": 1.3652,
+      "step": 9002
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0108222258789973e-05,
+      "loss": 1.3249,
+      "step": 9003
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0108001023897025e-05,
+      "loss": 1.2947,
+      "step": 9004
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0107779789004076e-05,
+      "loss": 1.2203,
+      "step": 9005
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0107558554111129e-05,
+      "loss": 1.4395,
+      "step": 9006
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0107337319218181e-05,
+      "loss": 1.4257,
+      "step": 9007
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0107116084325234e-05,
+      "loss": 1.278,
+      "step": 9008
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0106894849432285e-05,
+      "loss": 1.3254,
+      "step": 9009
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0106673614539336e-05,
+      "loss": 1.3064,
+      "step": 9010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0106452379646388e-05,
+      "loss": 1.2286,
+      "step": 9011
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0106231144753441e-05,
+      "loss": 1.2375,
+      "step": 9012
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0106009909860494e-05,
+      "loss": 1.2174,
+      "step": 9013
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0105788674967546e-05,
+      "loss": 1.3922,
+      "step": 9014
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0105567440074599e-05,
+      "loss": 1.2606,
+      "step": 9015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.010534620518165e-05,
+      "loss": 1.2622,
+      "step": 9016
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.01051249702887e-05,
+      "loss": 1.3416,
+      "step": 9017
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0104903735395753e-05,
+      "loss": 1.2792,
+      "step": 9018
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0104682500502806e-05,
+      "loss": 1.2754,
+      "step": 9019
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0104461265609859e-05,
+      "loss": 1.3292,
+      "step": 9020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.010424003071691e-05,
+      "loss": 1.1041,
+      "step": 9021
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0104018795823962e-05,
+      "loss": 1.4305,
+      "step": 9022
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0103797560931015e-05,
+      "loss": 1.2099,
+      "step": 9023
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0103576326038067e-05,
+      "loss": 1.3265,
+      "step": 9024
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0103355091145118e-05,
+      "loss": 1.3712,
+      "step": 9025
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0103133856252171e-05,
+      "loss": 1.3525,
+      "step": 9026
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0102912621359222e-05,
+      "loss": 1.3243,
+      "step": 9027
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0102691386466274e-05,
+      "loss": 1.2563,
+      "step": 9028
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0102470151573327e-05,
+      "loss": 1.1767,
+      "step": 9029
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.010224891668038e-05,
+      "loss": 1.1934,
+      "step": 9030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0102027681787432e-05,
+      "loss": 1.2068,
+      "step": 9031
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0101806446894483e-05,
+      "loss": 1.3112,
+      "step": 9032
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0101585212001536e-05,
+      "loss": 1.2831,
+      "step": 9033
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0101363977108587e-05,
+      "loss": 1.2618,
+      "step": 9034
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.010114274221564e-05,
+      "loss": 1.314,
+      "step": 9035
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0100921507322692e-05,
+      "loss": 1.2373,
+      "step": 9036
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0100700272429745e-05,
+      "loss": 1.3097,
+      "step": 9037
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0100479037536795e-05,
+      "loss": 1.2719,
+      "step": 9038
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0100257802643848e-05,
+      "loss": 1.2713,
+      "step": 9039
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.01000365677509e-05,
+      "loss": 1.2108,
+      "step": 9040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0099815332857952e-05,
+      "loss": 1.4604,
+      "step": 9041
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0099594097965004e-05,
+      "loss": 1.3993,
+      "step": 9042
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0099372863072055e-05,
+      "loss": 1.3114,
+      "step": 9043
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0099151628179108e-05,
+      "loss": 1.2597,
+      "step": 9044
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.009893039328616e-05,
+      "loss": 1.3271,
+      "step": 9045
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0098709158393213e-05,
+      "loss": 1.3358,
+      "step": 9046
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0098487923500266e-05,
+      "loss": 1.2044,
+      "step": 9047
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0098266688607317e-05,
+      "loss": 1.2553,
+      "step": 9048
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0098045453714369e-05,
+      "loss": 1.1868,
+      "step": 9049
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.009782421882142e-05,
+      "loss": 1.3982,
+      "step": 9050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0097602983928473e-05,
+      "loss": 1.3575,
+      "step": 9051
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0097381749035525e-05,
+      "loss": 1.369,
+      "step": 9052
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0097160514142578e-05,
+      "loss": 1.304,
+      "step": 9053
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0096939279249629e-05,
+      "loss": 1.3005,
+      "step": 9054
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0096718044356681e-05,
+      "loss": 1.3106,
+      "step": 9055
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0096496809463734e-05,
+      "loss": 1.3127,
+      "step": 9056
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0096275574570785e-05,
+      "loss": 1.2784,
+      "step": 9057
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0096054339677838e-05,
+      "loss": 1.2922,
+      "step": 9058
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0095833104784888e-05,
+      "loss": 1.1438,
+      "step": 9059
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0095611869891941e-05,
+      "loss": 1.3834,
+      "step": 9060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0095390634998994e-05,
+      "loss": 1.3495,
+      "step": 9061
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0095169400106046e-05,
+      "loss": 1.2796,
+      "step": 9062
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0094948165213099e-05,
+      "loss": 1.2894,
+      "step": 9063
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0094726930320152e-05,
+      "loss": 1.2318,
+      "step": 9064
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0094505695427202e-05,
+      "loss": 1.1942,
+      "step": 9065
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0094284460534253e-05,
+      "loss": 1.1791,
+      "step": 9066
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0094063225641306e-05,
+      "loss": 1.2065,
+      "step": 9067
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0093841990748359e-05,
+      "loss": 1.242,
+      "step": 9068
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0093620755855411e-05,
+      "loss": 1.301,
+      "step": 9069
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0093399520962462e-05,
+      "loss": 1.3402,
+      "step": 9070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0093178286069515e-05,
+      "loss": 1.2342,
+      "step": 9071
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0092957051176567e-05,
+      "loss": 1.2575,
+      "step": 9072
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0092735816283618e-05,
+      "loss": 1.3751,
+      "step": 9073
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0092514581390671e-05,
+      "loss": 1.39,
+      "step": 9074
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0092293346497723e-05,
+      "loss": 1.3586,
+      "step": 9075
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0092072111604774e-05,
+      "loss": 1.2323,
+      "step": 9076
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0091850876711827e-05,
+      "loss": 1.2689,
+      "step": 9077
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.009162964181888e-05,
+      "loss": 1.3236,
+      "step": 9078
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0091408406925932e-05,
+      "loss": 1.1186,
+      "step": 9079
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0091187172032985e-05,
+      "loss": 1.3646,
+      "step": 9080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0090965937140036e-05,
+      "loss": 1.2756,
+      "step": 9081
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0090744702247087e-05,
+      "loss": 1.2457,
+      "step": 9082
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.009052346735414e-05,
+      "loss": 1.1838,
+      "step": 9083
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0090302232461192e-05,
+      "loss": 1.2934,
+      "step": 9084
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0090080997568245e-05,
+      "loss": 1.4336,
+      "step": 9085
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0089859762675297e-05,
+      "loss": 1.1615,
+      "step": 9086
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0089638527782348e-05,
+      "loss": 1.2892,
+      "step": 9087
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.00894172928894e-05,
+      "loss": 1.2468,
+      "step": 9088
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0089196057996452e-05,
+      "loss": 1.2015,
+      "step": 9089
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0088974823103504e-05,
+      "loss": 1.4038,
+      "step": 9090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0088753588210557e-05,
+      "loss": 1.452,
+      "step": 9091
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0088532353317608e-05,
+      "loss": 1.3546,
+      "step": 9092
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.008831111842466e-05,
+      "loss": 1.3228,
+      "step": 9093
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0088089883531713e-05,
+      "loss": 1.2674,
+      "step": 9094
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0087868648638766e-05,
+      "loss": 1.3385,
+      "step": 9095
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0087647413745818e-05,
+      "loss": 1.3417,
+      "step": 9096
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0087426178852869e-05,
+      "loss": 1.3308,
+      "step": 9097
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.008720494395992e-05,
+      "loss": 1.2444,
+      "step": 9098
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0086983709066973e-05,
+      "loss": 1.4356,
+      "step": 9099
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0086762474174025e-05,
+      "loss": 1.169,
+      "step": 9100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0086541239281078e-05,
+      "loss": 1.2942,
+      "step": 9101
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.008632000438813e-05,
+      "loss": 1.2979,
+      "step": 9102
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0086098769495181e-05,
+      "loss": 1.1825,
+      "step": 9103
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0085877534602234e-05,
+      "loss": 1.2789,
+      "step": 9104
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0085656299709285e-05,
+      "loss": 1.291,
+      "step": 9105
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0085435064816337e-05,
+      "loss": 1.2915,
+      "step": 9106
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.008521382992339e-05,
+      "loss": 1.2903,
+      "step": 9107
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0084992595030443e-05,
+      "loss": 1.2196,
+      "step": 9108
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0084771360137494e-05,
+      "loss": 1.2951,
+      "step": 9109
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0084550125244546e-05,
+      "loss": 1.4431,
+      "step": 9110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0084328890351599e-05,
+      "loss": 1.3824,
+      "step": 9111
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0084107655458651e-05,
+      "loss": 1.4,
+      "step": 9112
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0083886420565702e-05,
+      "loss": 1.3724,
+      "step": 9113
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0083665185672753e-05,
+      "loss": 1.3742,
+      "step": 9114
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0083443950779806e-05,
+      "loss": 1.1853,
+      "step": 9115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0083222715886859e-05,
+      "loss": 1.2299,
+      "step": 9116
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0083001480993911e-05,
+      "loss": 1.2832,
+      "step": 9117
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0082780246100964e-05,
+      "loss": 1.2572,
+      "step": 9118
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0082559011208016e-05,
+      "loss": 1.3835,
+      "step": 9119
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0082337776315067e-05,
+      "loss": 1.3264,
+      "step": 9120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.008211654142212e-05,
+      "loss": 1.3263,
+      "step": 9121
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.008189530652917e-05,
+      "loss": 1.2764,
+      "step": 9122
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0081674071636223e-05,
+      "loss": 1.3009,
+      "step": 9123
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0081452836743276e-05,
+      "loss": 1.2584,
+      "step": 9124
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0081231601850327e-05,
+      "loss": 1.3463,
+      "step": 9125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.008101036695738e-05,
+      "loss": 1.2506,
+      "step": 9126
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0080789132064432e-05,
+      "loss": 1.4149,
+      "step": 9127
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0080567897171485e-05,
+      "loss": 1.4047,
+      "step": 9128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0080346662278536e-05,
+      "loss": 1.1221,
+      "step": 9129
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0080125427385588e-05,
+      "loss": 1.1651,
+      "step": 9130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.007990419249264e-05,
+      "loss": 1.2604,
+      "step": 9131
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0079682957599692e-05,
+      "loss": 1.2761,
+      "step": 9132
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0079461722706744e-05,
+      "loss": 1.2624,
+      "step": 9133
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0079240487813797e-05,
+      "loss": 1.2146,
+      "step": 9134
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.007901925292085e-05,
+      "loss": 1.3876,
+      "step": 9135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.00787980180279e-05,
+      "loss": 1.2227,
+      "step": 9136
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0078576783134953e-05,
+      "loss": 1.2313,
+      "step": 9137
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0078355548242004e-05,
+      "loss": 1.2253,
+      "step": 9138
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0078134313349057e-05,
+      "loss": 1.2551,
+      "step": 9139
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.007791307845611e-05,
+      "loss": 1.2614,
+      "step": 9140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0077691843563162e-05,
+      "loss": 1.2532,
+      "step": 9141
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0077470608670213e-05,
+      "loss": 1.1806,
+      "step": 9142
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0077249373777265e-05,
+      "loss": 1.2478,
+      "step": 9143
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0077028138884318e-05,
+      "loss": 1.4741,
+      "step": 9144
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0076806903991369e-05,
+      "loss": 1.3366,
+      "step": 9145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0076585669098422e-05,
+      "loss": 1.3192,
+      "step": 9146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0076364434205473e-05,
+      "loss": 1.3843,
+      "step": 9147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0076143199312525e-05,
+      "loss": 1.1752,
+      "step": 9148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0075921964419578e-05,
+      "loss": 1.3847,
+      "step": 9149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.007570072952663e-05,
+      "loss": 1.2318,
+      "step": 9150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0075479494633683e-05,
+      "loss": 1.2527,
+      "step": 9151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0075258259740734e-05,
+      "loss": 1.3077,
+      "step": 9152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0075037024847787e-05,
+      "loss": 1.1624,
+      "step": 9153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0074815789954837e-05,
+      "loss": 1.2278,
+      "step": 9154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.007459455506189e-05,
+      "loss": 1.1849,
+      "step": 9155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0074373320168943e-05,
+      "loss": 1.1756,
+      "step": 9156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0074152085275995e-05,
+      "loss": 1.4836,
+      "step": 9157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0073930850383046e-05,
+      "loss": 1.3167,
+      "step": 9158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0073709615490099e-05,
+      "loss": 1.2708,
+      "step": 9159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0073488380597151e-05,
+      "loss": 1.3931,
+      "step": 9160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0073267145704202e-05,
+      "loss": 1.2743,
+      "step": 9161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0073045910811255e-05,
+      "loss": 1.2845,
+      "step": 9162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0072824675918306e-05,
+      "loss": 1.4282,
+      "step": 9163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0072603441025358e-05,
+      "loss": 1.1742,
+      "step": 9164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0072382206132411e-05,
+      "loss": 1.1989,
+      "step": 9165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0072160971239464e-05,
+      "loss": 1.3112,
+      "step": 9166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0071939736346516e-05,
+      "loss": 1.3643,
+      "step": 9167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0071718501453569e-05,
+      "loss": 1.4132,
+      "step": 9168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.007149726656062e-05,
+      "loss": 1.2185,
+      "step": 9169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.007127603166767e-05,
+      "loss": 1.3903,
+      "step": 9170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0071054796774723e-05,
+      "loss": 1.2126,
+      "step": 9171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0070833561881776e-05,
+      "loss": 1.277,
+      "step": 9172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0070612326988829e-05,
+      "loss": 1.3156,
+      "step": 9173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.007039109209588e-05,
+      "loss": 1.3318,
+      "step": 9174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0070169857202932e-05,
+      "loss": 1.3773,
+      "step": 9175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0069948622309985e-05,
+      "loss": 1.2525,
+      "step": 9176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0069727387417036e-05,
+      "loss": 1.3017,
+      "step": 9177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0069506152524088e-05,
+      "loss": 1.2248,
+      "step": 9178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0069284917631141e-05,
+      "loss": 1.3365,
+      "step": 9179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0069063682738192e-05,
+      "loss": 1.271,
+      "step": 9180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0068842447845244e-05,
+      "loss": 1.2232,
+      "step": 9181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0068621212952297e-05,
+      "loss": 1.2678,
+      "step": 9182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.006839997805935e-05,
+      "loss": 1.287,
+      "step": 9183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0068178743166402e-05,
+      "loss": 1.2442,
+      "step": 9184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0067957508273453e-05,
+      "loss": 1.2874,
+      "step": 9185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0067736273380504e-05,
+      "loss": 1.291,
+      "step": 9186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0067515038487557e-05,
+      "loss": 1.2185,
+      "step": 9187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.006729380359461e-05,
+      "loss": 1.2542,
+      "step": 9188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0067072568701662e-05,
+      "loss": 1.3348,
+      "step": 9189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0066851333808715e-05,
+      "loss": 1.3923,
+      "step": 9190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0066630098915765e-05,
+      "loss": 1.2665,
+      "step": 9191
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0066408864022818e-05,
+      "loss": 1.3408,
+      "step": 9192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0066187629129869e-05,
+      "loss": 1.2192,
+      "step": 9193
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0065966394236922e-05,
+      "loss": 1.2626,
+      "step": 9194
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0065745159343974e-05,
+      "loss": 1.379,
+      "step": 9195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0065523924451025e-05,
+      "loss": 1.2385,
+      "step": 9196
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0065302689558078e-05,
+      "loss": 1.2223,
+      "step": 9197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.006508145466513e-05,
+      "loss": 1.3484,
+      "step": 9198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0064860219772183e-05,
+      "loss": 1.316,
+      "step": 9199
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0064638984879236e-05,
+      "loss": 1.136,
+      "step": 9200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0064417749986286e-05,
+      "loss": 1.325,
+      "step": 9201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0064196515093337e-05,
+      "loss": 1.2725,
+      "step": 9202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.006397528020039e-05,
+      "loss": 1.2811,
+      "step": 9203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0063754045307443e-05,
+      "loss": 1.3923,
+      "step": 9204
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0063532810414495e-05,
+      "loss": 1.2167,
+      "step": 9205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0063311575521548e-05,
+      "loss": 1.1816,
+      "step": 9206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0063090340628599e-05,
+      "loss": 1.3803,
+      "step": 9207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0062869105735651e-05,
+      "loss": 1.1991,
+      "step": 9208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0062647870842704e-05,
+      "loss": 1.26,
+      "step": 9209
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0062426635949755e-05,
+      "loss": 1.3729,
+      "step": 9210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0062205401056808e-05,
+      "loss": 1.341,
+      "step": 9211
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006198416616386e-05,
+      "loss": 1.4367,
+      "step": 9212
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0061762931270911e-05,
+      "loss": 1.3882,
+      "step": 9213
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0061541696377964e-05,
+      "loss": 1.2375,
+      "step": 9214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0061320461485016e-05,
+      "loss": 1.2994,
+      "step": 9215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0061099226592069e-05,
+      "loss": 1.31,
+      "step": 9216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006087799169912e-05,
+      "loss": 1.2183,
+      "step": 9217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006065675680617e-05,
+      "loss": 1.4234,
+      "step": 9218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0060435521913223e-05,
+      "loss": 1.269,
+      "step": 9219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0060214287020276e-05,
+      "loss": 1.4242,
+      "step": 9220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0059993052127329e-05,
+      "loss": 1.4047,
+      "step": 9221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0059771817234381e-05,
+      "loss": 1.3152,
+      "step": 9222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0059550582341434e-05,
+      "loss": 1.2684,
+      "step": 9223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0059329347448485e-05,
+      "loss": 1.2502,
+      "step": 9224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0059108112555537e-05,
+      "loss": 1.2496,
+      "step": 9225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0058886877662588e-05,
+      "loss": 1.248,
+      "step": 9226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005866564276964e-05,
+      "loss": 1.3451,
+      "step": 9227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0058444407876693e-05,
+      "loss": 1.1748,
+      "step": 9228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0058223172983744e-05,
+      "loss": 1.0972,
+      "step": 9229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0058001938090797e-05,
+      "loss": 1.3518,
+      "step": 9230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005778070319785e-05,
+      "loss": 1.3357,
+      "step": 9231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0057559468304902e-05,
+      "loss": 1.2894,
+      "step": 9232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0057338233411953e-05,
+      "loss": 1.218,
+      "step": 9233
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0057116998519006e-05,
+      "loss": 1.26,
+      "step": 9234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0056895763626057e-05,
+      "loss": 1.3363,
+      "step": 9235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005667452873311e-05,
+      "loss": 1.1593,
+      "step": 9236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0056453293840162e-05,
+      "loss": 1.1657,
+      "step": 9237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0056232058947214e-05,
+      "loss": 1.2192,
+      "step": 9238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0056010824054267e-05,
+      "loss": 1.2788,
+      "step": 9239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0055789589161318e-05,
+      "loss": 1.3079,
+      "step": 9240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005556835426837e-05,
+      "loss": 1.3625,
+      "step": 9241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0055347119375422e-05,
+      "loss": 1.2629,
+      "step": 9242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0055125884482474e-05,
+      "loss": 1.2798,
+      "step": 9243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0054904649589527e-05,
+      "loss": 1.3299,
+      "step": 9244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005468341469658e-05,
+      "loss": 1.2782,
+      "step": 9245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005446217980363e-05,
+      "loss": 1.314,
+      "step": 9246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0054240944910683e-05,
+      "loss": 1.3337,
+      "step": 9247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0054019710017736e-05,
+      "loss": 1.2579,
+      "step": 9248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0053798475124786e-05,
+      "loss": 1.3803,
+      "step": 9249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0053577240231839e-05,
+      "loss": 1.334,
+      "step": 9250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005335600533889e-05,
+      "loss": 1.3055,
+      "step": 9251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0053134770445943e-05,
+      "loss": 1.2199,
+      "step": 9252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0052913535552995e-05,
+      "loss": 1.1663,
+      "step": 9253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0052692300660048e-05,
+      "loss": 1.2435,
+      "step": 9254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.00524710657671e-05,
+      "loss": 1.2558,
+      "step": 9255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0052249830874151e-05,
+      "loss": 1.338,
+      "step": 9256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0052028595981204e-05,
+      "loss": 1.3508,
+      "step": 9257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0051807361088255e-05,
+      "loss": 1.2199,
+      "step": 9258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0051586126195307e-05,
+      "loss": 1.2069,
+      "step": 9259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005136489130236e-05,
+      "loss": 1.2702,
+      "step": 9260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0051143656409413e-05,
+      "loss": 1.3615,
+      "step": 9261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0050922421516464e-05,
+      "loss": 1.298,
+      "step": 9262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0050701186623516e-05,
+      "loss": 1.2273,
+      "step": 9263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0050479951730569e-05,
+      "loss": 1.3877,
+      "step": 9264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005025871683762e-05,
+      "loss": 1.2269,
+      "step": 9265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0050037481944672e-05,
+      "loss": 1.2545,
+      "step": 9266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0049816247051723e-05,
+      "loss": 1.32,
+      "step": 9267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0049595012158776e-05,
+      "loss": 1.2818,
+      "step": 9268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0049373777265829e-05,
+      "loss": 1.4316,
+      "step": 9269
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0049152542372881e-05,
+      "loss": 1.3886,
+      "step": 9270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0048931307479934e-05,
+      "loss": 1.338,
+      "step": 9271
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0048710072586986e-05,
+      "loss": 1.4662,
+      "step": 9272
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0048488837694037e-05,
+      "loss": 1.3002,
+      "step": 9273
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0048267602801088e-05,
+      "loss": 1.3012,
+      "step": 9274
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004804636790814e-05,
+      "loss": 1.2743,
+      "step": 9275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0047825133015193e-05,
+      "loss": 1.1743,
+      "step": 9276
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0047603898122246e-05,
+      "loss": 1.3924,
+      "step": 9277
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0047382663229297e-05,
+      "loss": 1.3494,
+      "step": 9278
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004716142833635e-05,
+      "loss": 1.3681,
+      "step": 9279
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0046940193443402e-05,
+      "loss": 1.1868,
+      "step": 9280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0046718958550453e-05,
+      "loss": 1.2647,
+      "step": 9281
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0046497723657506e-05,
+      "loss": 1.2545,
+      "step": 9282
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0046276488764558e-05,
+      "loss": 1.2606,
+      "step": 9283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004605525387161e-05,
+      "loss": 1.3375,
+      "step": 9284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0045834018978662e-05,
+      "loss": 1.3863,
+      "step": 9285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0045612784085714e-05,
+      "loss": 1.3889,
+      "step": 9286
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0045391549192767e-05,
+      "loss": 1.3079,
+      "step": 9287
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004517031429982e-05,
+      "loss": 1.3515,
+      "step": 9288
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004494907940687e-05,
+      "loss": 1.2108,
+      "step": 9289
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0044727844513922e-05,
+      "loss": 1.4075,
+      "step": 9290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0044506609620974e-05,
+      "loss": 1.1834,
+      "step": 9291
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0044285374728027e-05,
+      "loss": 1.3237,
+      "step": 9292
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004406413983508e-05,
+      "loss": 1.3385,
+      "step": 9293
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0043842904942132e-05,
+      "loss": 1.2326,
+      "step": 9294
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0043621670049183e-05,
+      "loss": 1.3245,
+      "step": 9295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0043400435156235e-05,
+      "loss": 1.3191,
+      "step": 9296
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0043179200263288e-05,
+      "loss": 1.3026,
+      "step": 9297
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0042957965370339e-05,
+      "loss": 1.2158,
+      "step": 9298
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0042736730477392e-05,
+      "loss": 1.3769,
+      "step": 9299
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0042515495584443e-05,
+      "loss": 1.3777,
+      "step": 9300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0042294260691495e-05,
+      "loss": 1.2286,
+      "step": 9301
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0042073025798548e-05,
+      "loss": 1.2989,
+      "step": 9302
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.00418517909056e-05,
+      "loss": 1.1133,
+      "step": 9303
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0041630556012653e-05,
+      "loss": 1.3338,
+      "step": 9304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0041409321119704e-05,
+      "loss": 1.2416,
+      "step": 9305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0041188086226755e-05,
+      "loss": 1.3614,
+      "step": 9306
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0040966851333807e-05,
+      "loss": 1.2768,
+      "step": 9307
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004074561644086e-05,
+      "loss": 1.3473,
+      "step": 9308
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0040524381547913e-05,
+      "loss": 1.3398,
+      "step": 9309
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0040303146654965e-05,
+      "loss": 1.2936,
+      "step": 9310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0040081911762016e-05,
+      "loss": 1.2039,
+      "step": 9311
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0039860676869069e-05,
+      "loss": 1.4201,
+      "step": 9312
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0039639441976121e-05,
+      "loss": 1.311,
+      "step": 9313
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0039418207083172e-05,
+      "loss": 1.3471,
+      "step": 9314
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0039196972190225e-05,
+      "loss": 1.3213,
+      "step": 9315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0038975737297278e-05,
+      "loss": 1.2745,
+      "step": 9316
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0038754502404328e-05,
+      "loss": 1.3551,
+      "step": 9317
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0038533267511381e-05,
+      "loss": 1.1167,
+      "step": 9318
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0038312032618434e-05,
+      "loss": 1.3802,
+      "step": 9319
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0038090797725486e-05,
+      "loss": 1.2946,
+      "step": 9320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0037869562832537e-05,
+      "loss": 1.1649,
+      "step": 9321
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0037648327939588e-05,
+      "loss": 1.3418,
+      "step": 9322
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003742709304664e-05,
+      "loss": 1.2956,
+      "step": 9323
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0037205858153693e-05,
+      "loss": 1.2837,
+      "step": 9324
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0036984623260746e-05,
+      "loss": 1.3441,
+      "step": 9325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0036763388367799e-05,
+      "loss": 1.196,
+      "step": 9326
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0036542153474851e-05,
+      "loss": 1.1386,
+      "step": 9327
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0036320918581902e-05,
+      "loss": 1.3831,
+      "step": 9328
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0036099683688955e-05,
+      "loss": 1.3534,
+      "step": 9329
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0035878448796006e-05,
+      "loss": 1.0903,
+      "step": 9330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0035657213903058e-05,
+      "loss": 1.306,
+      "step": 9331
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0035435979010111e-05,
+      "loss": 1.3724,
+      "step": 9332
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0035214744117162e-05,
+      "loss": 1.3117,
+      "step": 9333
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0034993509224214e-05,
+      "loss": 1.1532,
+      "step": 9334
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0034772274331267e-05,
+      "loss": 1.4239,
+      "step": 9335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003455103943832e-05,
+      "loss": 1.2383,
+      "step": 9336
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003432980454537e-05,
+      "loss": 1.2738,
+      "step": 9337
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0034108569652423e-05,
+      "loss": 1.2651,
+      "step": 9338
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0033887334759474e-05,
+      "loss": 1.4417,
+      "step": 9339
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0033666099866527e-05,
+      "loss": 1.2411,
+      "step": 9340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003344486497358e-05,
+      "loss": 1.3008,
+      "step": 9341
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0033223630080632e-05,
+      "loss": 1.2707,
+      "step": 9342
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0033002395187685e-05,
+      "loss": 1.2935,
+      "step": 9343
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0032781160294735e-05,
+      "loss": 1.2266,
+      "step": 9344
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0032559925401788e-05,
+      "loss": 1.3837,
+      "step": 9345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0032338690508839e-05,
+      "loss": 1.3179,
+      "step": 9346
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0032117455615892e-05,
+      "loss": 1.3517,
+      "step": 9347
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0031896220722944e-05,
+      "loss": 1.1257,
+      "step": 9348
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0031674985829995e-05,
+      "loss": 1.282,
+      "step": 9349
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0031453750937048e-05,
+      "loss": 1.2112,
+      "step": 9350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.00312325160441e-05,
+      "loss": 1.2349,
+      "step": 9351
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0031011281151153e-05,
+      "loss": 1.2817,
+      "step": 9352
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0030790046258204e-05,
+      "loss": 1.1513,
+      "step": 9353
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0030568811365256e-05,
+      "loss": 1.328,
+      "step": 9354
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0030347576472307e-05,
+      "loss": 1.193,
+      "step": 9355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003012634157936e-05,
+      "loss": 1.3915,
+      "step": 9356
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0029905106686413e-05,
+      "loss": 1.2587,
+      "step": 9357
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0029683871793465e-05,
+      "loss": 1.2903,
+      "step": 9358
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0029462636900518e-05,
+      "loss": 1.3118,
+      "step": 9359
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0029241402007569e-05,
+      "loss": 1.3131,
+      "step": 9360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0029020167114621e-05,
+      "loss": 1.2463,
+      "step": 9361
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0028798932221672e-05,
+      "loss": 1.3012,
+      "step": 9362
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0028577697328725e-05,
+      "loss": 1.2919,
+      "step": 9363
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0028356462435778e-05,
+      "loss": 1.1577,
+      "step": 9364
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002813522754283e-05,
+      "loss": 1.262,
+      "step": 9365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0027913992649881e-05,
+      "loss": 1.3908,
+      "step": 9366
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0027692757756934e-05,
+      "loss": 1.2959,
+      "step": 9367
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0027471522863986e-05,
+      "loss": 1.2618,
+      "step": 9368
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0027250287971037e-05,
+      "loss": 1.3087,
+      "step": 9369
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002702905307809e-05,
+      "loss": 1.3253,
+      "step": 9370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002680781818514e-05,
+      "loss": 1.1638,
+      "step": 9371
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0026586583292193e-05,
+      "loss": 1.362,
+      "step": 9372
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0026365348399246e-05,
+      "loss": 1.271,
+      "step": 9373
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0026144113506299e-05,
+      "loss": 1.2983,
+      "step": 9374
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0025922878613351e-05,
+      "loss": 1.3715,
+      "step": 9375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0025701643720404e-05,
+      "loss": 1.3236,
+      "step": 9376
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0025480408827455e-05,
+      "loss": 1.221,
+      "step": 9377
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0025259173934506e-05,
+      "loss": 1.3061,
+      "step": 9378
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0025037939041558e-05,
+      "loss": 1.2329,
+      "step": 9379
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002481670414861e-05,
+      "loss": 1.2703,
+      "step": 9380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0024595469255663e-05,
+      "loss": 1.307,
+      "step": 9381
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0024374234362714e-05,
+      "loss": 1.3358,
+      "step": 9382
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0024152999469767e-05,
+      "loss": 1.2302,
+      "step": 9383
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002393176457682e-05,
+      "loss": 1.2593,
+      "step": 9384
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0023710529683872e-05,
+      "loss": 1.2262,
+      "step": 9385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0023489294790923e-05,
+      "loss": 1.2133,
+      "step": 9386
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0023268059897976e-05,
+      "loss": 1.3025,
+      "step": 9387
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0023046825005027e-05,
+      "loss": 1.2463,
+      "step": 9388
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.002282559011208e-05,
+      "loss": 1.2839,
+      "step": 9389
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0022604355219132e-05,
+      "loss": 1.3774,
+      "step": 9390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0022383120326184e-05,
+      "loss": 1.3586,
+      "step": 9391
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0022161885433237e-05,
+      "loss": 1.2551,
+      "step": 9392
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0021940650540288e-05,
+      "loss": 1.2834,
+      "step": 9393
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0021719415647339e-05,
+      "loss": 1.2632,
+      "step": 9394
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0021498180754392e-05,
+      "loss": 1.3403,
+      "step": 9395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0021276945861444e-05,
+      "loss": 1.2765,
+      "step": 9396
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0021055710968497e-05,
+      "loss": 1.2845,
+      "step": 9397
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.002083447607555e-05,
+      "loss": 1.3409,
+      "step": 9398
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.00206132411826e-05,
+      "loss": 1.3463,
+      "step": 9399
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0020392006289653e-05,
+      "loss": 1.2942,
+      "step": 9400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0020170771396706e-05,
+      "loss": 1.3218,
+      "step": 9401
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0019949536503756e-05,
+      "loss": 1.3275,
+      "step": 9402
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0019728301610809e-05,
+      "loss": 1.3663,
+      "step": 9403
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.001950706671786e-05,
+      "loss": 1.2045,
+      "step": 9404
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0019285831824913e-05,
+      "loss": 1.349,
+      "step": 9405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0019064596931965e-05,
+      "loss": 1.2228,
+      "step": 9406
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0018843362039018e-05,
+      "loss": 1.3678,
+      "step": 9407
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.001862212714607e-05,
+      "loss": 1.236,
+      "step": 9408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0018400892253121e-05,
+      "loss": 1.2637,
+      "step": 9409
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0018179657360172e-05,
+      "loss": 1.375,
+      "step": 9410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0017958422467225e-05,
+      "loss": 1.1901,
+      "step": 9411
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0017737187574277e-05,
+      "loss": 1.2909,
+      "step": 9412
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.001751595268133e-05,
+      "loss": 1.401,
+      "step": 9413
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0017294717788383e-05,
+      "loss": 1.2802,
+      "step": 9414
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0017073482895434e-05,
+      "loss": 1.2762,
+      "step": 9415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0016852248002486e-05,
+      "loss": 1.4386,
+      "step": 9416
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0016631013109539e-05,
+      "loss": 1.1416,
+      "step": 9417
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.001640977821659e-05,
+      "loss": 1.4077,
+      "step": 9418
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0016188543323642e-05,
+      "loss": 1.2156,
+      "step": 9419
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0015967308430695e-05,
+      "loss": 1.2525,
+      "step": 9420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0015746073537746e-05,
+      "loss": 1.2746,
+      "step": 9421
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0015524838644799e-05,
+      "loss": 1.1976,
+      "step": 9422
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0015303603751851e-05,
+      "loss": 1.3192,
+      "step": 9423
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0015082368858904e-05,
+      "loss": 1.2514,
+      "step": 9424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0014861133965955e-05,
+      "loss": 1.3483,
+      "step": 9425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0014639899073006e-05,
+      "loss": 1.3189,
+      "step": 9426
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0014418664180058e-05,
+      "loss": 1.328,
+      "step": 9427
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.001419742928711e-05,
+      "loss": 1.3716,
+      "step": 9428
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0013976194394163e-05,
+      "loss": 1.2581,
+      "step": 9429
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0013754959501216e-05,
+      "loss": 1.2853,
+      "step": 9430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0013533724608269e-05,
+      "loss": 1.2726,
+      "step": 9431
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.001331248971532e-05,
+      "loss": 1.3316,
+      "step": 9432
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0013091254822372e-05,
+      "loss": 1.1747,
+      "step": 9433
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0012870019929423e-05,
+      "loss": 1.2427,
+      "step": 9434
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0012648785036476e-05,
+      "loss": 1.2288,
+      "step": 9435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0012427550143528e-05,
+      "loss": 1.3318,
+      "step": 9436
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.001220631525058e-05,
+      "loss": 1.3871,
+      "step": 9437
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0011985080357632e-05,
+      "loss": 1.3753,
+      "step": 9438
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0011763845464684e-05,
+      "loss": 1.2643,
+      "step": 9439
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0011542610571737e-05,
+      "loss": 1.2519,
+      "step": 9440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0011321375678788e-05,
+      "loss": 1.2435,
+      "step": 9441
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.001110014078584e-05,
+      "loss": 1.4041,
+      "step": 9442
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0010878905892891e-05,
+      "loss": 1.2871,
+      "step": 9443
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0010657670999944e-05,
+      "loss": 1.4569,
+      "step": 9444
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0010436436106997e-05,
+      "loss": 1.2746,
+      "step": 9445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.001021520121405e-05,
+      "loss": 1.3373,
+      "step": 9446
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0009993966321102e-05,
+      "loss": 1.2874,
+      "step": 9447
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0009772731428153e-05,
+      "loss": 1.3753,
+      "step": 9448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0009551496535205e-05,
+      "loss": 1.309,
+      "step": 9449
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0009330261642256e-05,
+      "loss": 1.2941,
+      "step": 9450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0009109026749309e-05,
+      "loss": 1.3425,
+      "step": 9451
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0008887791856362e-05,
+      "loss": 1.2843,
+      "step": 9452
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0008666556963413e-05,
+      "loss": 1.385,
+      "step": 9453
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0008445322070465e-05,
+      "loss": 1.17,
+      "step": 9454
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0008224087177518e-05,
+      "loss": 1.2233,
+      "step": 9455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.000800285228457e-05,
+      "loss": 1.3595,
+      "step": 9456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0007781617391621e-05,
+      "loss": 1.228,
+      "step": 9457
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0007560382498674e-05,
+      "loss": 1.2031,
+      "step": 9458
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0007339147605725e-05,
+      "loss": 1.2595,
+      "step": 9459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0007117912712777e-05,
+      "loss": 1.2884,
+      "step": 9460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.000689667781983e-05,
+      "loss": 1.3877,
+      "step": 9461
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0006675442926883e-05,
+      "loss": 1.3777,
+      "step": 9462
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0006454208033935e-05,
+      "loss": 1.4276,
+      "step": 9463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0006232973140986e-05,
+      "loss": 1.3921,
+      "step": 9464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0006011738248039e-05,
+      "loss": 1.2922,
+      "step": 9465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.000579050335509e-05,
+      "loss": 1.2865,
+      "step": 9466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0005569268462142e-05,
+      "loss": 1.2796,
+      "step": 9467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0005348033569195e-05,
+      "loss": 1.372,
+      "step": 9468
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0005126798676248e-05,
+      "loss": 1.2214,
+      "step": 9469
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0004905563783298e-05,
+      "loss": 1.1696,
+      "step": 9470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0004684328890351e-05,
+      "loss": 1.1377,
+      "step": 9471
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0004463093997404e-05,
+      "loss": 1.4171,
+      "step": 9472
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0004241859104456e-05,
+      "loss": 1.2856,
+      "step": 9473
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0004020624211507e-05,
+      "loss": 1.2903,
+      "step": 9474
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0003799389318558e-05,
+      "loss": 1.3288,
+      "step": 9475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.000357815442561e-05,
+      "loss": 1.2921,
+      "step": 9476
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0003356919532663e-05,
+      "loss": 1.2391,
+      "step": 9477
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0003135684639716e-05,
+      "loss": 1.1153,
+      "step": 9478
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0002914449746769e-05,
+      "loss": 1.2659,
+      "step": 9479
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0002693214853821e-05,
+      "loss": 1.1455,
+      "step": 9480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0002471979960872e-05,
+      "loss": 1.1474,
+      "step": 9481
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0002250745067923e-05,
+      "loss": 1.2366,
+      "step": 9482
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0002029510174976e-05,
+      "loss": 1.2818,
+      "step": 9483
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0001808275282028e-05,
+      "loss": 1.2618,
+      "step": 9484
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0001587040389081e-05,
+      "loss": 1.2847,
+      "step": 9485
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0001365805496132e-05,
+      "loss": 1.2641,
+      "step": 9486
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0001144570603184e-05,
+      "loss": 1.2988,
+      "step": 9487
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0000923335710237e-05,
+      "loss": 1.2403,
+      "step": 9488
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.000070210081729e-05,
+      "loss": 1.4033,
+      "step": 9489
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.000048086592434e-05,
+      "loss": 1.2299,
+      "step": 9490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0000259631031393e-05,
+      "loss": 1.3403,
+      "step": 9491
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0000038396138444e-05,
+      "loss": 1.3952,
+      "step": 9492
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.999817161245497e-06,
+      "loss": 1.3044,
+      "step": 9493
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99959592635255e-06,
+      "loss": 1.467,
+      "step": 9494
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.999374691459602e-06,
+      "loss": 1.349,
+      "step": 9495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.999153456566655e-06,
+      "loss": 1.2341,
+      "step": 9496
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.998932221673705e-06,
+      "loss": 1.2699,
+      "step": 9497
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.998710986780756e-06,
+      "loss": 1.2886,
+      "step": 9498
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.998489751887809e-06,
+      "loss": 1.2884,
+      "step": 9499
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.998268516994862e-06,
+      "loss": 1.3133,
+      "step": 9500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.998047282101914e-06,
+      "loss": 1.2302,
+      "step": 9501
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.997826047208967e-06,
+      "loss": 1.2799,
+      "step": 9502
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.997604812316018e-06,
+      "loss": 1.2839,
+      "step": 9503
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99738357742307e-06,
+      "loss": 1.2043,
+      "step": 9504
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.997162342530123e-06,
+      "loss": 1.3592,
+      "step": 9505
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.996941107637174e-06,
+      "loss": 1.3088,
+      "step": 9506
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.996719872744226e-06,
+      "loss": 1.3124,
+      "step": 9507
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.996498637851277e-06,
+      "loss": 1.4604,
+      "step": 9508
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99627740295833e-06,
+      "loss": 1.327,
+      "step": 9509
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.996056168065383e-06,
+      "loss": 1.2977,
+      "step": 9510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.995834933172435e-06,
+      "loss": 1.348,
+      "step": 9511
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.995613698279488e-06,
+      "loss": 1.374,
+      "step": 9512
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.995392463386539e-06,
+      "loss": 1.2803,
+      "step": 9513
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99517122849359e-06,
+      "loss": 1.2985,
+      "step": 9514
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.994949993600642e-06,
+      "loss": 1.1799,
+      "step": 9515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.994728758707695e-06,
+      "loss": 1.3565,
+      "step": 9516
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.994507523814747e-06,
+      "loss": 1.22,
+      "step": 9517
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9942862889218e-06,
+      "loss": 1.222,
+      "step": 9518
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.994065054028851e-06,
+      "loss": 1.264,
+      "step": 9519
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.993843819135904e-06,
+      "loss": 1.2673,
+      "step": 9520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.993622584242956e-06,
+      "loss": 1.2678,
+      "step": 9521
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.993401349350007e-06,
+      "loss": 1.237,
+      "step": 9522
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99318011445706e-06,
+      "loss": 1.3697,
+      "step": 9523
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.992958879564112e-06,
+      "loss": 1.316,
+      "step": 9524
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.992737644671163e-06,
+      "loss": 1.3502,
+      "step": 9525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.992516409778216e-06,
+      "loss": 1.3747,
+      "step": 9526
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.992295174885269e-06,
+      "loss": 1.1101,
+      "step": 9527
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.992073939992321e-06,
+      "loss": 1.3344,
+      "step": 9528
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.991852705099372e-06,
+      "loss": 1.2739,
+      "step": 9529
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.991631470206425e-06,
+      "loss": 1.24,
+      "step": 9530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.991410235313476e-06,
+      "loss": 1.2729,
+      "step": 9531
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.991189000420528e-06,
+      "loss": 1.1644,
+      "step": 9532
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99096776552758e-06,
+      "loss": 1.3273,
+      "step": 9533
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.990746530634633e-06,
+      "loss": 1.2274,
+      "step": 9534
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.990525295741686e-06,
+      "loss": 1.2914,
+      "step": 9535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.990304060848737e-06,
+      "loss": 1.2683,
+      "step": 9536
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99008282595579e-06,
+      "loss": 1.3954,
+      "step": 9537
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98986159106284e-06,
+      "loss": 1.3197,
+      "step": 9538
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.989640356169893e-06,
+      "loss": 1.3464,
+      "step": 9539
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.989419121276946e-06,
+      "loss": 1.2751,
+      "step": 9540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.989197886383997e-06,
+      "loss": 1.2859,
+      "step": 9541
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98897665149105e-06,
+      "loss": 1.4608,
+      "step": 9542
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.988755416598102e-06,
+      "loss": 1.436,
+      "step": 9543
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.988534181705154e-06,
+      "loss": 1.3036,
+      "step": 9544
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.988312946812205e-06,
+      "loss": 1.2749,
+      "step": 9545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.988091711919258e-06,
+      "loss": 1.1932,
+      "step": 9546
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987870477026309e-06,
+      "loss": 1.4886,
+      "step": 9547
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987649242133362e-06,
+      "loss": 1.2573,
+      "step": 9548
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987428007240414e-06,
+      "loss": 1.3721,
+      "step": 9549
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987206772347467e-06,
+      "loss": 1.3623,
+      "step": 9550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98698553745452e-06,
+      "loss": 1.1966,
+      "step": 9551
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98676430256157e-06,
+      "loss": 1.1899,
+      "step": 9552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.986543067668623e-06,
+      "loss": 1.3286,
+      "step": 9553
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.986321832775674e-06,
+      "loss": 1.1831,
+      "step": 9554
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.986100597882726e-06,
+      "loss": 1.2318,
+      "step": 9555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.985879362989779e-06,
+      "loss": 1.3086,
+      "step": 9556
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98565812809683e-06,
+      "loss": 1.3031,
+      "step": 9557
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.985436893203883e-06,
+      "loss": 1.2631,
+      "step": 9558
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.985215658310935e-06,
+      "loss": 1.1827,
+      "step": 9559
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.984994423417988e-06,
+      "loss": 1.2622,
+      "step": 9560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98477318852504e-06,
+      "loss": 1.3911,
+      "step": 9561
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.984551953632091e-06,
+      "loss": 1.2793,
+      "step": 9562
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.984330718739142e-06,
+      "loss": 1.1638,
+      "step": 9563
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.984109483846195e-06,
+      "loss": 1.1116,
+      "step": 9564
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.983888248953247e-06,
+      "loss": 1.2007,
+      "step": 9565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9836670140603e-06,
+      "loss": 1.2091,
+      "step": 9566
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.983445779167353e-06,
+      "loss": 1.2264,
+      "step": 9567
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.983224544274404e-06,
+      "loss": 1.3258,
+      "step": 9568
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.983003309381456e-06,
+      "loss": 1.2696,
+      "step": 9569
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.982782074488507e-06,
+      "loss": 1.3918,
+      "step": 9570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98256083959556e-06,
+      "loss": 1.3104,
+      "step": 9571
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.982339604702612e-06,
+      "loss": 1.3545,
+      "step": 9572
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.982118369809665e-06,
+      "loss": 1.4048,
+      "step": 9573
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.981897134916716e-06,
+      "loss": 1.4196,
+      "step": 9574
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.981675900023768e-06,
+      "loss": 1.4466,
+      "step": 9575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.981454665130821e-06,
+      "loss": 1.3869,
+      "step": 9576
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.981233430237874e-06,
+      "loss": 1.5107,
+      "step": 9577
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.981012195344925e-06,
+      "loss": 1.4207,
+      "step": 9578
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.980790960451976e-06,
+      "loss": 1.3066,
+      "step": 9579
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.980569725559028e-06,
+      "loss": 1.2936,
+      "step": 9580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.98034849066608e-06,
+      "loss": 1.2246,
+      "step": 9581
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.980127255773133e-06,
+      "loss": 1.2145,
+      "step": 9582
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.979906020880186e-06,
+      "loss": 1.2002,
+      "step": 9583
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.979684785987239e-06,
+      "loss": 1.3055,
+      "step": 9584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.97946355109429e-06,
+      "loss": 1.295,
+      "step": 9585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.97924231620134e-06,
+      "loss": 1.2451,
+      "step": 9586
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.979021081308393e-06,
+      "loss": 1.302,
+      "step": 9587
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.978799846415446e-06,
+      "loss": 1.393,
+      "step": 9588
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.978578611522498e-06,
+      "loss": 1.2257,
+      "step": 9589
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.97835737662955e-06,
+      "loss": 1.2214,
+      "step": 9590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.978136141736602e-06,
+      "loss": 1.2127,
+      "step": 9591
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.977914906843654e-06,
+      "loss": 1.4146,
+      "step": 9592
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.977693671950707e-06,
+      "loss": 1.281,
+      "step": 9593
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.977472437057758e-06,
+      "loss": 1.3315,
+      "step": 9594
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.97725120216481e-06,
+      "loss": 1.3871,
+      "step": 9595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.977029967271861e-06,
+      "loss": 1.214,
+      "step": 9596
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.976808732378914e-06,
+      "loss": 1.3385,
+      "step": 9597
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.976587497485967e-06,
+      "loss": 1.2433,
+      "step": 9598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.97636626259302e-06,
+      "loss": 1.3493,
+      "step": 9599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.976145027700072e-06,
+      "loss": 1.3022,
+      "step": 9600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.975923792807123e-06,
+      "loss": 1.2741,
+      "step": 9601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.975702557914174e-06,
+      "loss": 1.191,
+      "step": 9602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.975481323021226e-06,
+      "loss": 1.2619,
+      "step": 9603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.975260088128279e-06,
+      "loss": 1.2435,
+      "step": 9604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.975038853235332e-06,
+      "loss": 1.2576,
+      "step": 9605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.974817618342384e-06,
+      "loss": 1.3174,
+      "step": 9606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.974596383449435e-06,
+      "loss": 1.3199,
+      "step": 9607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.974375148556488e-06,
+      "loss": 1.308,
+      "step": 9608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.97415391366354e-06,
+      "loss": 1.2989,
+      "step": 9609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.973932678770591e-06,
+      "loss": 1.2915,
+      "step": 9610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.973711443877644e-06,
+      "loss": 1.3772,
+      "step": 9611
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.973490208984695e-06,
+      "loss": 1.3316,
+      "step": 9612
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.973268974091747e-06,
+      "loss": 1.2757,
+      "step": 9613
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.9730477391988e-06,
+      "loss": 1.2779,
+      "step": 9614
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.972826504305853e-06,
+      "loss": 1.2043,
+      "step": 9615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.972605269412905e-06,
+      "loss": 1.2242,
+      "step": 9616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.972384034519956e-06,
+      "loss": 1.3123,
+      "step": 9617
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.972162799627009e-06,
+      "loss": 1.354,
+      "step": 9618
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.97194156473406e-06,
+      "loss": 1.4518,
+      "step": 9619
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.971720329841112e-06,
+      "loss": 1.1021,
+      "step": 9620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.971499094948165e-06,
+      "loss": 1.2496,
+      "step": 9621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.971277860055218e-06,
+      "loss": 1.3208,
+      "step": 9622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.971056625162268e-06,
+      "loss": 1.2805,
+      "step": 9623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.970835390269321e-06,
+      "loss": 1.2422,
+      "step": 9624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.970614155376374e-06,
+      "loss": 1.1528,
+      "step": 9625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.970392920483425e-06,
+      "loss": 1.4159,
+      "step": 9626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.970171685590477e-06,
+      "loss": 1.34,
+      "step": 9627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.96995045069753e-06,
+      "loss": 1.3903,
+      "step": 9628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.96972921580458e-06,
+      "loss": 1.2682,
+      "step": 9629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.969507980911633e-06,
+      "loss": 1.3337,
+      "step": 9630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.969286746018686e-06,
+      "loss": 1.3935,
+      "step": 9631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.969065511125739e-06,
+      "loss": 1.1894,
+      "step": 9632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.96884427623279e-06,
+      "loss": 1.4158,
+      "step": 9633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.968623041339842e-06,
+      "loss": 1.3346,
+      "step": 9634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.968401806446893e-06,
+      "loss": 1.1814,
+      "step": 9635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.968180571553946e-06,
+      "loss": 1.3947,
+      "step": 9636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.967959336660998e-06,
+      "loss": 1.2833,
+      "step": 9637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.967738101768051e-06,
+      "loss": 1.1367,
+      "step": 9638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.967516866875102e-06,
+      "loss": 1.1888,
+      "step": 9639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.967295631982154e-06,
+      "loss": 1.1522,
+      "step": 9640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.967074397089207e-06,
+      "loss": 1.2338,
+      "step": 9641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.966853162196258e-06,
+      "loss": 1.3471,
+      "step": 9642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.96663192730331e-06,
+      "loss": 1.2438,
+      "step": 9643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.966410692410363e-06,
+      "loss": 1.1859,
+      "step": 9644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.966189457517414e-06,
+      "loss": 1.2869,
+      "step": 9645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.965968222624467e-06,
+      "loss": 1.2763,
+      "step": 9646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.96574698773152e-06,
+      "loss": 1.365,
+      "step": 9647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.965525752838572e-06,
+      "loss": 1.2037,
+      "step": 9648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.965304517945624e-06,
+      "loss": 1.1328,
+      "step": 9649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.965083283052675e-06,
+      "loss": 1.2238,
+      "step": 9650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.964862048159726e-06,
+      "loss": 1.3538,
+      "step": 9651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.964640813266779e-06,
+      "loss": 1.2025,
+      "step": 9652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.964419578373832e-06,
+      "loss": 1.2098,
+      "step": 9653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.964198343480884e-06,
+      "loss": 1.3741,
+      "step": 9654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.963977108587937e-06,
+      "loss": 1.1888,
+      "step": 9655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.963755873694988e-06,
+      "loss": 1.2788,
+      "step": 9656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.96353463880204e-06,
+      "loss": 1.2969,
+      "step": 9657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.963313403909091e-06,
+      "loss": 1.2665,
+      "step": 9658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.963092169016144e-06,
+      "loss": 1.2369,
+      "step": 9659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.962870934123196e-06,
+      "loss": 1.2569,
+      "step": 9660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.962649699230247e-06,
+      "loss": 1.2343,
+      "step": 9661
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.9624284643373e-06,
+      "loss": 1.3305,
+      "step": 9662
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.962207229444353e-06,
+      "loss": 1.3011,
+      "step": 9663
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.961985994551405e-06,
+      "loss": 1.2064,
+      "step": 9664
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.961764759658458e-06,
+      "loss": 1.2737,
+      "step": 9665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.961543524765509e-06,
+      "loss": 1.2928,
+      "step": 9666
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.96132228987256e-06,
+      "loss": 1.2835,
+      "step": 9667
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.961101054979612e-06,
+      "loss": 1.3635,
+      "step": 9668
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.960879820086665e-06,
+      "loss": 1.2931,
+      "step": 9669
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.960658585193717e-06,
+      "loss": 1.1861,
+      "step": 9670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.96043735030077e-06,
+      "loss": 1.3153,
+      "step": 9671
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.960216115407821e-06,
+      "loss": 1.2043,
+      "step": 9672
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.959994880514874e-06,
+      "loss": 1.4329,
+      "step": 9673
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.959773645621925e-06,
+      "loss": 1.1383,
+      "step": 9674
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.959552410728977e-06,
+      "loss": 1.2848,
+      "step": 9675
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.95933117583603e-06,
+      "loss": 1.4439,
+      "step": 9676
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.959109940943082e-06,
+      "loss": 1.4317,
+      "step": 9677
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.958888706050133e-06,
+      "loss": 1.2328,
+      "step": 9678
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.958667471157186e-06,
+      "loss": 1.3022,
+      "step": 9679
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.958446236264239e-06,
+      "loss": 1.3088,
+      "step": 9680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.958225001371291e-06,
+      "loss": 1.2664,
+      "step": 9681
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.958003766478342e-06,
+      "loss": 1.2073,
+      "step": 9682
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.957782531585393e-06,
+      "loss": 1.2433,
+      "step": 9683
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.957561296692446e-06,
+      "loss": 1.3798,
+      "step": 9684
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.957340061799498e-06,
+      "loss": 1.3035,
+      "step": 9685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.95711882690655e-06,
+      "loss": 1.2271,
+      "step": 9686
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.956897592013603e-06,
+      "loss": 1.1499,
+      "step": 9687
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.956676357120656e-06,
+      "loss": 1.2009,
+      "step": 9688
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.956455122227707e-06,
+      "loss": 1.1907,
+      "step": 9689
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.956233887334758e-06,
+      "loss": 1.2691,
+      "step": 9690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.95601265244181e-06,
+      "loss": 1.3032,
+      "step": 9691
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.955791417548863e-06,
+      "loss": 1.3579,
+      "step": 9692
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.955570182655916e-06,
+      "loss": 1.4964,
+      "step": 9693
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.955348947762967e-06,
+      "loss": 1.2982,
+      "step": 9694
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.95512771287002e-06,
+      "loss": 1.2688,
+      "step": 9695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.954906477977072e-06,
+      "loss": 1.2493,
+      "step": 9696
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.954685243084124e-06,
+      "loss": 1.1914,
+      "step": 9697
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.954464008191175e-06,
+      "loss": 1.1662,
+      "step": 9698
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.954242773298228e-06,
+      "loss": 1.2741,
+      "step": 9699
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.954021538405279e-06,
+      "loss": 1.3487,
+      "step": 9700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.953800303512332e-06,
+      "loss": 1.2345,
+      "step": 9701
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.953579068619384e-06,
+      "loss": 1.3798,
+      "step": 9702
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.953357833726437e-06,
+      "loss": 1.1979,
+      "step": 9703
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.95313659883349e-06,
+      "loss": 1.3406,
+      "step": 9704
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.95291536394054e-06,
+      "loss": 1.2897,
+      "step": 9705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.952694129047593e-06,
+      "loss": 1.2451,
+      "step": 9706
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.952472894154644e-06,
+      "loss": 1.2278,
+      "step": 9707
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.952251659261696e-06,
+      "loss": 1.3194,
+      "step": 9708
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.952030424368749e-06,
+      "loss": 1.3315,
+      "step": 9709
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.951809189475802e-06,
+      "loss": 1.2053,
+      "step": 9710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.951587954582853e-06,
+      "loss": 1.2271,
+      "step": 9711
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.951366719689905e-06,
+      "loss": 1.3575,
+      "step": 9712
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.951145484796958e-06,
+      "loss": 1.329,
+      "step": 9713
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.950924249904009e-06,
+      "loss": 1.3051,
+      "step": 9714
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.950703015011061e-06,
+      "loss": 1.2877,
+      "step": 9715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.950481780118112e-06,
+      "loss": 1.2644,
+      "step": 9716
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.950260545225165e-06,
+      "loss": 1.3075,
+      "step": 9717
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.950039310332217e-06,
+      "loss": 1.2706,
+      "step": 9718
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.94981807543927e-06,
+      "loss": 1.1993,
+      "step": 9719
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.949596840546323e-06,
+      "loss": 1.2466,
+      "step": 9720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.949375605653374e-06,
+      "loss": 1.2067,
+      "step": 9721
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.949154370760426e-06,
+      "loss": 1.3253,
+      "step": 9722
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.948933135867477e-06,
+      "loss": 1.3319,
+      "step": 9723
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.94871190097453e-06,
+      "loss": 1.2798,
+      "step": 9724
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.948490666081582e-06,
+      "loss": 1.3688,
+      "step": 9725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.948269431188635e-06,
+      "loss": 1.2583,
+      "step": 9726
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.948048196295686e-06,
+      "loss": 1.2855,
+      "step": 9727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.947826961402738e-06,
+      "loss": 1.2075,
+      "step": 9728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.947605726509791e-06,
+      "loss": 1.3272,
+      "step": 9729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.947384491616842e-06,
+      "loss": 1.2723,
+      "step": 9730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.947163256723895e-06,
+      "loss": 1.2182,
+      "step": 9731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.946942021830947e-06,
+      "loss": 1.3791,
+      "step": 9732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.946720786937998e-06,
+      "loss": 1.3586,
+      "step": 9733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.94649955204505e-06,
+      "loss": 1.3352,
+      "step": 9734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.946278317152103e-06,
+      "loss": 1.325,
+      "step": 9735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.946057082259156e-06,
+      "loss": 1.3796,
+      "step": 9736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.945835847366209e-06,
+      "loss": 1.2965,
+      "step": 9737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.94561461247326e-06,
+      "loss": 1.3039,
+      "step": 9738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.94539337758031e-06,
+      "loss": 1.3128,
+      "step": 9739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.945172142687363e-06,
+      "loss": 1.1816,
+      "step": 9740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.944950907794416e-06,
+      "loss": 1.3055,
+      "step": 9741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.944729672901468e-06,
+      "loss": 1.2936,
+      "step": 9742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.94450843800852e-06,
+      "loss": 1.1665,
+      "step": 9743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.944287203115572e-06,
+      "loss": 1.2974,
+      "step": 9744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.944065968222624e-06,
+      "loss": 1.2128,
+      "step": 9745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.943844733329675e-06,
+      "loss": 1.1825,
+      "step": 9746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.943623498436728e-06,
+      "loss": 1.3154,
+      "step": 9747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.94340226354378e-06,
+      "loss": 1.323,
+      "step": 9748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.943181028650831e-06,
+      "loss": 1.3381,
+      "step": 9749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.942959793757884e-06,
+      "loss": 1.1919,
+      "step": 9750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.942738558864937e-06,
+      "loss": 1.1929,
+      "step": 9751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.94251732397199e-06,
+      "loss": 1.3723,
+      "step": 9752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.942296089079042e-06,
+      "loss": 1.3805,
+      "step": 9753
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.942074854186093e-06,
+      "loss": 1.3249,
+      "step": 9754
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.941853619293144e-06,
+      "loss": 1.2624,
+      "step": 9755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.941632384400196e-06,
+      "loss": 1.4016,
+      "step": 9756
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.941411149507249e-06,
+      "loss": 1.4411,
+      "step": 9757
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.941189914614302e-06,
+      "loss": 1.2637,
+      "step": 9758
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.940968679721354e-06,
+      "loss": 1.3105,
+      "step": 9759
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.940747444828405e-06,
+      "loss": 1.3232,
+      "step": 9760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.940526209935458e-06,
+      "loss": 1.2929,
+      "step": 9761
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.940304975042509e-06,
+      "loss": 1.4129,
+      "step": 9762
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.940083740149561e-06,
+      "loss": 1.2539,
+      "step": 9763
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.939862505256614e-06,
+      "loss": 1.2546,
+      "step": 9764
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.939641270363665e-06,
+      "loss": 1.1978,
+      "step": 9765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.939420035470717e-06,
+      "loss": 1.2933,
+      "step": 9766
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.93919880057777e-06,
+      "loss": 1.2091,
+      "step": 9767
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.938977565684823e-06,
+      "loss": 1.4329,
+      "step": 9768
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.938756330791875e-06,
+      "loss": 1.2382,
+      "step": 9769
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.938535095898926e-06,
+      "loss": 1.2464,
+      "step": 9770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.938313861005977e-06,
+      "loss": 1.3327,
+      "step": 9771
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.93809262611303e-06,
+      "loss": 1.1479,
+      "step": 9772
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.937871391220082e-06,
+      "loss": 1.2691,
+      "step": 9773
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.937650156327135e-06,
+      "loss": 1.3263,
+      "step": 9774
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.937428921434188e-06,
+      "loss": 1.3737,
+      "step": 9775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.937207686541238e-06,
+      "loss": 1.2438,
+      "step": 9776
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.936986451648291e-06,
+      "loss": 1.2061,
+      "step": 9777
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.936765216755342e-06,
+      "loss": 1.2606,
+      "step": 9778
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.936543981862395e-06,
+      "loss": 1.2806,
+      "step": 9779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.936322746969447e-06,
+      "loss": 1.393,
+      "step": 9780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.9361015120765e-06,
+      "loss": 1.3044,
+      "step": 9781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.93588027718355e-06,
+      "loss": 1.264,
+      "step": 9782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.935659042290603e-06,
+      "loss": 1.2775,
+      "step": 9783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.935437807397656e-06,
+      "loss": 1.1975,
+      "step": 9784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.935216572504709e-06,
+      "loss": 1.1783,
+      "step": 9785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.93499533761176e-06,
+      "loss": 1.1406,
+      "step": 9786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.93477410271881e-06,
+      "loss": 1.3282,
+      "step": 9787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.934552867825863e-06,
+      "loss": 1.1719,
+      "step": 9788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.934331632932916e-06,
+      "loss": 1.3882,
+      "step": 9789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.934110398039968e-06,
+      "loss": 1.4687,
+      "step": 9790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.93388916314702e-06,
+      "loss": 1.2508,
+      "step": 9791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.933667928254073e-06,
+      "loss": 1.3144,
+      "step": 9792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.933446693361124e-06,
+      "loss": 1.3698,
+      "step": 9793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.933225458468177e-06,
+      "loss": 1.1638,
+      "step": 9794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.933004223575228e-06,
+      "loss": 1.3398,
+      "step": 9795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.93278298868228e-06,
+      "loss": 1.2568,
+      "step": 9796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.932561753789333e-06,
+      "loss": 1.2387,
+      "step": 9797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.932340518896384e-06,
+      "loss": 1.0615,
+      "step": 9798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.932119284003437e-06,
+      "loss": 1.225,
+      "step": 9799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.93189804911049e-06,
+      "loss": 1.2548,
+      "step": 9800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.931676814217542e-06,
+      "loss": 1.3053,
+      "step": 9801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.931455579324593e-06,
+      "loss": 1.338,
+      "step": 9802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.931234344431645e-06,
+      "loss": 1.1685,
+      "step": 9803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.931013109538696e-06,
+      "loss": 1.2695,
+      "step": 9804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.930791874645749e-06,
+      "loss": 1.2678,
+      "step": 9805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.930570639752802e-06,
+      "loss": 1.3045,
+      "step": 9806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.930349404859854e-06,
+      "loss": 1.1906,
+      "step": 9807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.930128169966907e-06,
+      "loss": 1.3586,
+      "step": 9808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.929906935073958e-06,
+      "loss": 1.3281,
+      "step": 9809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.92968570018101e-06,
+      "loss": 1.3311,
+      "step": 9810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.929464465288061e-06,
+      "loss": 1.3472,
+      "step": 9811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.929243230395114e-06,
+      "loss": 1.2991,
+      "step": 9812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.929021995502166e-06,
+      "loss": 1.188,
+      "step": 9813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.928800760609219e-06,
+      "loss": 1.2306,
+      "step": 9814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.92857952571627e-06,
+      "loss": 1.2902,
+      "step": 9815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.928358290823323e-06,
+      "loss": 1.2925,
+      "step": 9816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.928137055930375e-06,
+      "loss": 1.2376,
+      "step": 9817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.927915821037426e-06,
+      "loss": 1.3364,
+      "step": 9818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.927694586144479e-06,
+      "loss": 1.1878,
+      "step": 9819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.92747335125153e-06,
+      "loss": 1.3533,
+      "step": 9820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.927252116358582e-06,
+      "loss": 1.157,
+      "step": 9821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.927030881465635e-06,
+      "loss": 1.4374,
+      "step": 9822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.926809646572687e-06,
+      "loss": 1.3649,
+      "step": 9823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.92658841167974e-06,
+      "loss": 1.2474,
+      "step": 9824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.926367176786793e-06,
+      "loss": 1.2273,
+      "step": 9825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.926145941893844e-06,
+      "loss": 1.4108,
+      "step": 9826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.925924707000895e-06,
+      "loss": 1.342,
+      "step": 9827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.925703472107947e-06,
+      "loss": 1.3715,
+      "step": 9828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.925482237215e-06,
+      "loss": 1.2983,
+      "step": 9829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.925261002322052e-06,
+      "loss": 1.1855,
+      "step": 9830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.925039767429103e-06,
+      "loss": 1.2222,
+      "step": 9831
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.924818532536156e-06,
+      "loss": 1.3469,
+      "step": 9832
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.924597297643209e-06,
+      "loss": 1.3565,
+      "step": 9833
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.92437606275026e-06,
+      "loss": 1.2847,
+      "step": 9834
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.924154827857312e-06,
+      "loss": 1.2216,
+      "step": 9835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.923933592964363e-06,
+      "loss": 1.2364,
+      "step": 9836
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.923712358071416e-06,
+      "loss": 1.3579,
+      "step": 9837
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.923491123178468e-06,
+      "loss": 1.4565,
+      "step": 9838
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.92326988828552e-06,
+      "loss": 1.3586,
+      "step": 9839
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.923048653392573e-06,
+      "loss": 1.2769,
+      "step": 9840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.922827418499626e-06,
+      "loss": 1.3507,
+      "step": 9841
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.922606183606677e-06,
+      "loss": 1.1928,
+      "step": 9842
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.922384948713728e-06,
+      "loss": 1.2217,
+      "step": 9843
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.92216371382078e-06,
+      "loss": 1.2326,
+      "step": 9844
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.921942478927833e-06,
+      "loss": 1.3482,
+      "step": 9845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.921721244034886e-06,
+      "loss": 1.236,
+      "step": 9846
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.921500009141937e-06,
+      "loss": 1.4004,
+      "step": 9847
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.92127877424899e-06,
+      "loss": 1.3738,
+      "step": 9848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.921057539356042e-06,
+      "loss": 1.2774,
+      "step": 9849
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.920836304463093e-06,
+      "loss": 1.1484,
+      "step": 9850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.920615069570145e-06,
+      "loss": 1.3406,
+      "step": 9851
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.920393834677198e-06,
+      "loss": 1.3264,
+      "step": 9852
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.920172599784249e-06,
+      "loss": 1.2342,
+      "step": 9853
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.919951364891301e-06,
+      "loss": 1.2213,
+      "step": 9854
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.919730129998354e-06,
+      "loss": 1.1746,
+      "step": 9855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.919508895105407e-06,
+      "loss": 1.2721,
+      "step": 9856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.91928766021246e-06,
+      "loss": 1.2076,
+      "step": 9857
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.91906642531951e-06,
+      "loss": 1.2273,
+      "step": 9858
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.918845190426561e-06,
+      "loss": 1.3718,
+      "step": 9859
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.918623955533614e-06,
+      "loss": 1.2193,
+      "step": 9860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.918402720640666e-06,
+      "loss": 1.383,
+      "step": 9861
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.918181485747719e-06,
+      "loss": 1.2845,
+      "step": 9862
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.917960250854772e-06,
+      "loss": 1.3384,
+      "step": 9863
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.917739015961823e-06,
+      "loss": 1.256,
+      "step": 9864
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.917517781068875e-06,
+      "loss": 1.4383,
+      "step": 9865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.917296546175926e-06,
+      "loss": 1.2849,
+      "step": 9866
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.917075311282979e-06,
+      "loss": 1.3625,
+      "step": 9867
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.916854076390031e-06,
+      "loss": 1.4288,
+      "step": 9868
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.916632841497082e-06,
+      "loss": 1.2999,
+      "step": 9869
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.916411606604135e-06,
+      "loss": 1.39,
+      "step": 9870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.916190371711187e-06,
+      "loss": 1.213,
+      "step": 9871
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.91596913681824e-06,
+      "loss": 1.1776,
+      "step": 9872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.915747901925293e-06,
+      "loss": 1.2405,
+      "step": 9873
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.915526667032344e-06,
+      "loss": 1.411,
+      "step": 9874
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.915305432139394e-06,
+      "loss": 1.3727,
+      "step": 9875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.915084197246447e-06,
+      "loss": 1.1321,
+      "step": 9876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.9148629623535e-06,
+      "loss": 1.1914,
+      "step": 9877
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.914641727460552e-06,
+      "loss": 1.1886,
+      "step": 9878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.914420492567605e-06,
+      "loss": 1.3327,
+      "step": 9879
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.914199257674656e-06,
+      "loss": 1.2141,
+      "step": 9880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.913978022781708e-06,
+      "loss": 1.2228,
+      "step": 9881
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.913756787888761e-06,
+      "loss": 1.3181,
+      "step": 9882
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.913535552995812e-06,
+      "loss": 1.4217,
+      "step": 9883
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.913314318102865e-06,
+      "loss": 1.3232,
+      "step": 9884
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.913093083209917e-06,
+      "loss": 1.2641,
+      "step": 9885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.912871848316968e-06,
+      "loss": 1.3308,
+      "step": 9886
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.91265061342402e-06,
+      "loss": 1.23,
+      "step": 9887
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.912429378531073e-06,
+      "loss": 1.1678,
+      "step": 9888
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.912208143638126e-06,
+      "loss": 1.2861,
+      "step": 9889
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.911986908745177e-06,
+      "loss": 1.2817,
+      "step": 9890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.911765673852228e-06,
+      "loss": 1.1403,
+      "step": 9891
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.91154443895928e-06,
+      "loss": 1.281,
+      "step": 9892
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.911323204066333e-06,
+      "loss": 1.3028,
+      "step": 9893
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.911101969173386e-06,
+      "loss": 1.2838,
+      "step": 9894
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.910880734280438e-06,
+      "loss": 1.2973,
+      "step": 9895
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.910659499387491e-06,
+      "loss": 1.2361,
+      "step": 9896
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.910438264494542e-06,
+      "loss": 1.3184,
+      "step": 9897
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.910217029601594e-06,
+      "loss": 1.3041,
+      "step": 9898
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.909995794708645e-06,
+      "loss": 1.2779,
+      "step": 9899
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.909774559815698e-06,
+      "loss": 1.2867,
+      "step": 9900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.90955332492275e-06,
+      "loss": 1.2393,
+      "step": 9901
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.909332090029801e-06,
+      "loss": 1.344,
+      "step": 9902
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.909110855136854e-06,
+      "loss": 1.2209,
+      "step": 9903
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.908889620243907e-06,
+      "loss": 1.2624,
+      "step": 9904
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.90866838535096e-06,
+      "loss": 1.2675,
+      "step": 9905
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.90844715045801e-06,
+      "loss": 1.2872,
+      "step": 9906
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.908225915565063e-06,
+      "loss": 1.3332,
+      "step": 9907
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.908004680672114e-06,
+      "loss": 1.3157,
+      "step": 9908
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.907783445779166e-06,
+      "loss": 1.3715,
+      "step": 9909
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.907562210886219e-06,
+      "loss": 1.2301,
+      "step": 9910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.907340975993272e-06,
+      "loss": 1.2168,
+      "step": 9911
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.907119741100324e-06,
+      "loss": 1.1574,
+      "step": 9912
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.906898506207375e-06,
+      "loss": 1.3227,
+      "step": 9913
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.906677271314428e-06,
+      "loss": 1.3016,
+      "step": 9914
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.906456036421479e-06,
+      "loss": 1.2981,
+      "step": 9915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.906234801528531e-06,
+      "loss": 1.2582,
+      "step": 9916
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.906013566635584e-06,
+      "loss": 1.2878,
+      "step": 9917
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.905792331742636e-06,
+      "loss": 1.4647,
+      "step": 9918
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.905571096849687e-06,
+      "loss": 1.2146,
+      "step": 9919
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.90534986195674e-06,
+      "loss": 1.2591,
+      "step": 9920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.905128627063793e-06,
+      "loss": 1.2422,
+      "step": 9921
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.904907392170844e-06,
+      "loss": 1.3569,
+      "step": 9922
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.904686157277896e-06,
+      "loss": 1.2286,
+      "step": 9923
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.904464922384947e-06,
+      "loss": 1.407,
+      "step": 9924
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.904243687492e-06,
+      "loss": 1.2458,
+      "step": 9925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.904022452599052e-06,
+      "loss": 1.1561,
+      "step": 9926
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.903801217706105e-06,
+      "loss": 1.2469,
+      "step": 9927
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.903579982813158e-06,
+      "loss": 1.2332,
+      "step": 9928
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.90335874792021e-06,
+      "loss": 1.1482,
+      "step": 9929
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.903137513027261e-06,
+      "loss": 1.2853,
+      "step": 9930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.902916278134312e-06,
+      "loss": 1.3109,
+      "step": 9931
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.902695043241365e-06,
+      "loss": 1.2989,
+      "step": 9932
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.902473808348417e-06,
+      "loss": 1.3507,
+      "step": 9933
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.90225257345547e-06,
+      "loss": 1.2583,
+      "step": 9934
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.90203133856252e-06,
+      "loss": 1.3623,
+      "step": 9935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.901810103669573e-06,
+      "loss": 1.4989,
+      "step": 9936
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.901588868776626e-06,
+      "loss": 1.2938,
+      "step": 9937
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.901367633883677e-06,
+      "loss": 1.2551,
+      "step": 9938
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.90114639899073e-06,
+      "loss": 1.211,
+      "step": 9939
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.90092516409778e-06,
+      "loss": 1.3651,
+      "step": 9940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.900703929204833e-06,
+      "loss": 1.3695,
+      "step": 9941
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.900482694311886e-06,
+      "loss": 1.2827,
+      "step": 9942
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.900261459418938e-06,
+      "loss": 1.3084,
+      "step": 9943
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.90004022452599e-06,
+      "loss": 1.3024,
+      "step": 9944
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.899818989633043e-06,
+      "loss": 1.3311,
+      "step": 9945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.899597754740094e-06,
+      "loss": 1.2072,
+      "step": 9946
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.899376519847145e-06,
+      "loss": 1.3214,
+      "step": 9947
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.899155284954198e-06,
+      "loss": 1.2326,
+      "step": 9948
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.89893405006125e-06,
+      "loss": 1.2234,
+      "step": 9949
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.898712815168303e-06,
+      "loss": 1.3361,
+      "step": 9950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.898491580275354e-06,
+      "loss": 1.281,
+      "step": 9951
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.898270345382407e-06,
+      "loss": 1.2054,
+      "step": 9952
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.89804911048946e-06,
+      "loss": 1.297,
+      "step": 9953
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.89782787559651e-06,
+      "loss": 1.2673,
+      "step": 9954
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.897606640703563e-06,
+      "loss": 1.3444,
+      "step": 9955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.897385405810615e-06,
+      "loss": 1.2034,
+      "step": 9956
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.897164170917666e-06,
+      "loss": 1.2894,
+      "step": 9957
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.896942936024719e-06,
+      "loss": 1.2938,
+      "step": 9958
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.896721701131772e-06,
+      "loss": 1.276,
+      "step": 9959
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.896500466238824e-06,
+      "loss": 1.3385,
+      "step": 9960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.896279231345877e-06,
+      "loss": 1.2628,
+      "step": 9961
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.896057996452928e-06,
+      "loss": 1.3058,
+      "step": 9962
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.895836761559979e-06,
+      "loss": 1.2519,
+      "step": 9963
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.895615526667031e-06,
+      "loss": 1.4296,
+      "step": 9964
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.895394291774084e-06,
+      "loss": 1.2465,
+      "step": 9965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.895173056881136e-06,
+      "loss": 1.2299,
+      "step": 9966
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.894951821988189e-06,
+      "loss": 1.2663,
+      "step": 9967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.89473058709524e-06,
+      "loss": 1.4094,
+      "step": 9968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.894509352202293e-06,
+      "loss": 1.279,
+      "step": 9969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.894288117309345e-06,
+      "loss": 1.2655,
+      "step": 9970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.894066882416396e-06,
+      "loss": 1.2014,
+      "step": 9971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.893845647523449e-06,
+      "loss": 1.1542,
+      "step": 9972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.8936244126305e-06,
+      "loss": 1.348,
+      "step": 9973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.893403177737552e-06,
+      "loss": 1.3328,
+      "step": 9974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.893181942844605e-06,
+      "loss": 1.2488,
+      "step": 9975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.892960707951657e-06,
+      "loss": 1.2133,
+      "step": 9976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.89273947305871e-06,
+      "loss": 1.2209,
+      "step": 9977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.892518238165761e-06,
+      "loss": 1.2809,
+      "step": 9978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.892297003272812e-06,
+      "loss": 1.187,
+      "step": 9979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.892075768379865e-06,
+      "loss": 1.3066,
+      "step": 9980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.891854533486917e-06,
+      "loss": 1.2025,
+      "step": 9981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.89163329859397e-06,
+      "loss": 1.1591,
+      "step": 9982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.891412063701022e-06,
+      "loss": 1.3249,
+      "step": 9983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.891190828808073e-06,
+      "loss": 1.3543,
+      "step": 9984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.890969593915126e-06,
+      "loss": 1.2845,
+      "step": 9985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.890748359022178e-06,
+      "loss": 1.2229,
+      "step": 9986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.89052712412923e-06,
+      "loss": 1.2452,
+      "step": 9987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.890305889236282e-06,
+      "loss": 1.2752,
+      "step": 9988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.890084654343335e-06,
+      "loss": 1.2077,
+      "step": 9989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.889863419450386e-06,
+      "loss": 1.3951,
+      "step": 9990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.889642184557438e-06,
+      "loss": 1.396,
+      "step": 9991
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.88942094966449e-06,
+      "loss": 1.1981,
+      "step": 9992
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.889199714771543e-06,
+      "loss": 1.367,
+      "step": 9993
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.888978479878594e-06,
+      "loss": 1.1713,
+      "step": 9994
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.888757244985645e-06,
+      "loss": 1.2955,
+      "step": 9995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.888536010092698e-06,
+      "loss": 1.2272,
+      "step": 9996
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.88831477519975e-06,
+      "loss": 1.3207,
+      "step": 9997
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.888093540306803e-06,
+      "loss": 1.1765,
+      "step": 9998
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.887872305413856e-06,
+      "loss": 1.3178,
+      "step": 9999
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.887651070520908e-06,
+      "loss": 1.34,
+      "step": 10000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.88742983562796e-06,
+      "loss": 1.1464,
+      "step": 10001
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.887208600735012e-06,
+      "loss": 1.4469,
+      "step": 10002
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.886987365842063e-06,
+      "loss": 1.255,
+      "step": 10003
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.886766130949115e-06,
+      "loss": 1.2667,
+      "step": 10004
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.886544896056168e-06,
+      "loss": 1.2866,
+      "step": 10005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.886323661163219e-06,
+      "loss": 1.171,
+      "step": 10006
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.886102426270271e-06,
+      "loss": 1.2537,
+      "step": 10007
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.885881191377324e-06,
+      "loss": 1.2062,
+      "step": 10008
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.885659956484377e-06,
+      "loss": 1.2759,
+      "step": 10009
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.885438721591428e-06,
+      "loss": 1.2295,
+      "step": 10010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.88521748669848e-06,
+      "loss": 1.1793,
+      "step": 10011
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.884996251805531e-06,
+      "loss": 1.278,
+      "step": 10012
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.884775016912584e-06,
+      "loss": 1.3616,
+      "step": 10013
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.884553782019636e-06,
+      "loss": 1.2977,
+      "step": 10014
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.884332547126689e-06,
+      "loss": 1.185,
+      "step": 10015
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.884111312233742e-06,
+      "loss": 1.2744,
+      "step": 10016
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.883890077340793e-06,
+      "loss": 1.3779,
+      "step": 10017
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.883668842447845e-06,
+      "loss": 1.2832,
+      "step": 10018
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.883447607554896e-06,
+      "loss": 1.2682,
+      "step": 10019
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.883226372661949e-06,
+      "loss": 1.3667,
+      "step": 10020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.883005137769001e-06,
+      "loss": 1.2465,
+      "step": 10021
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.882783902876054e-06,
+      "loss": 1.3324,
+      "step": 10022
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.882562667983105e-06,
+      "loss": 1.2539,
+      "step": 10023
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.882341433090157e-06,
+      "loss": 1.3927,
+      "step": 10024
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.88212019819721e-06,
+      "loss": 1.185,
+      "step": 10025
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.881898963304261e-06,
+      "loss": 1.1929,
+      "step": 10026
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.881677728411314e-06,
+      "loss": 1.2884,
+      "step": 10027
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.881456493518364e-06,
+      "loss": 1.1218,
+      "step": 10028
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.881235258625417e-06,
+      "loss": 1.2698,
+      "step": 10029
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.88101402373247e-06,
+      "loss": 1.2787,
+      "step": 10030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.880792788839522e-06,
+      "loss": 1.3441,
+      "step": 10031
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.880571553946575e-06,
+      "loss": 1.3883,
+      "step": 10032
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.880350319053626e-06,
+      "loss": 1.423,
+      "step": 10033
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.880129084160678e-06,
+      "loss": 1.205,
+      "step": 10034
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.87990784926773e-06,
+      "loss": 1.2796,
+      "step": 10035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.879686614374782e-06,
+      "loss": 1.221,
+      "step": 10036
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.879465379481835e-06,
+      "loss": 1.3346,
+      "step": 10037
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.879244144588887e-06,
+      "loss": 1.2793,
+      "step": 10038
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.879022909695938e-06,
+      "loss": 1.3421,
+      "step": 10039
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.87880167480299e-06,
+      "loss": 1.1659,
+      "step": 10040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.878580439910043e-06,
+      "loss": 1.2976,
+      "step": 10041
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.878359205017094e-06,
+      "loss": 1.1514,
+      "step": 10042
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.878137970124147e-06,
+      "loss": 1.273,
+      "step": 10043
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.877916735231198e-06,
+      "loss": 1.3492,
+      "step": 10044
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.87769550033825e-06,
+      "loss": 1.2428,
+      "step": 10045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.877474265445303e-06,
+      "loss": 1.3989,
+      "step": 10046
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.877253030552356e-06,
+      "loss": 1.3155,
+      "step": 10047
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.877031795659408e-06,
+      "loss": 1.2991,
+      "step": 10048
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.876810560766461e-06,
+      "loss": 1.2525,
+      "step": 10049
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.876589325873512e-06,
+      "loss": 1.3164,
+      "step": 10050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.876368090980563e-06,
+      "loss": 1.2307,
+      "step": 10051
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.876146856087615e-06,
+      "loss": 1.2725,
+      "step": 10052
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.875925621194668e-06,
+      "loss": 1.1557,
+      "step": 10053
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.87570438630172e-06,
+      "loss": 1.2568,
+      "step": 10054
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.875483151408771e-06,
+      "loss": 1.2759,
+      "step": 10055
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.875261916515824e-06,
+      "loss": 1.2442,
+      "step": 10056
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.875040681622877e-06,
+      "loss": 1.3083,
+      "step": 10057
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.87481944672993e-06,
+      "loss": 1.1402,
+      "step": 10058
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.87459821183698e-06,
+      "loss": 1.1868,
+      "step": 10059
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.874376976944033e-06,
+      "loss": 1.2759,
+      "step": 10060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.874155742051084e-06,
+      "loss": 1.235,
+      "step": 10061
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.873934507158136e-06,
+      "loss": 1.3454,
+      "step": 10062
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.873713272265189e-06,
+      "loss": 1.2989,
+      "step": 10063
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.873492037372242e-06,
+      "loss": 1.3234,
+      "step": 10064
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.873270802479294e-06,
+      "loss": 1.205,
+      "step": 10065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.873049567586345e-06,
+      "loss": 1.3646,
+      "step": 10066
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.872828332693396e-06,
+      "loss": 1.4192,
+      "step": 10067
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.872607097800449e-06,
+      "loss": 1.2063,
+      "step": 10068
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.872385862907501e-06,
+      "loss": 1.4298,
+      "step": 10069
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.872164628014554e-06,
+      "loss": 1.2765,
+      "step": 10070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.871943393121606e-06,
+      "loss": 1.2983,
+      "step": 10071
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.871722158228657e-06,
+      "loss": 1.3103,
+      "step": 10072
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.87150092333571e-06,
+      "loss": 1.2525,
+      "step": 10073
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.871279688442763e-06,
+      "loss": 1.2051,
+      "step": 10074
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.871058453549814e-06,
+      "loss": 1.235,
+      "step": 10075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.870837218656866e-06,
+      "loss": 1.3105,
+      "step": 10076
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.870615983763917e-06,
+      "loss": 1.255,
+      "step": 10077
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.87039474887097e-06,
+      "loss": 1.298,
+      "step": 10078
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.870173513978022e-06,
+      "loss": 1.1317,
+      "step": 10079
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.869952279085075e-06,
+      "loss": 1.4338,
+      "step": 10080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.869731044192127e-06,
+      "loss": 1.2174,
+      "step": 10081
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.869509809299178e-06,
+      "loss": 1.3956,
+      "step": 10082
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.86928857440623e-06,
+      "loss": 1.3324,
+      "step": 10083
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.869067339513282e-06,
+      "loss": 1.2215,
+      "step": 10084
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.868846104620335e-06,
+      "loss": 1.2747,
+      "step": 10085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.868624869727387e-06,
+      "loss": 1.2294,
+      "step": 10086
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.86840363483444e-06,
+      "loss": 1.1815,
+      "step": 10087
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.86818239994149e-06,
+      "loss": 1.3183,
+      "step": 10088
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.867961165048543e-06,
+      "loss": 1.2422,
+      "step": 10089
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.867739930155596e-06,
+      "loss": 1.1434,
+      "step": 10090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.867518695262647e-06,
+      "loss": 1.3307,
+      "step": 10091
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.8672974603697e-06,
+      "loss": 1.2853,
+      "step": 10092
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.867076225476752e-06,
+      "loss": 1.246,
+      "step": 10093
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.866854990583803e-06,
+      "loss": 1.2913,
+      "step": 10094
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.866633755690856e-06,
+      "loss": 1.4121,
+      "step": 10095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.866412520797908e-06,
+      "loss": 1.1968,
+      "step": 10096
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.86619128590496e-06,
+      "loss": 1.287,
+      "step": 10097
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.865970051012012e-06,
+      "loss": 1.3501,
+      "step": 10098
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.865748816119063e-06,
+      "loss": 1.2487,
+      "step": 10099
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.865527581226115e-06,
+      "loss": 1.2158,
+      "step": 10100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.865306346333168e-06,
+      "loss": 1.3017,
+      "step": 10101
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.86508511144022e-06,
+      "loss": 1.3576,
+      "step": 10102
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.864863876547273e-06,
+      "loss": 1.2608,
+      "step": 10103
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.864642641654326e-06,
+      "loss": 1.263,
+      "step": 10104
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.864421406761377e-06,
+      "loss": 1.1754,
+      "step": 10105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.86420017186843e-06,
+      "loss": 1.2073,
+      "step": 10106
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.86397893697548e-06,
+      "loss": 1.4888,
+      "step": 10107
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.863757702082533e-06,
+      "loss": 1.2574,
+      "step": 10108
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.863536467189585e-06,
+      "loss": 1.1834,
+      "step": 10109
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.863315232296636e-06,
+      "loss": 1.1916,
+      "step": 10110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.863093997403689e-06,
+      "loss": 1.1461,
+      "step": 10111
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.862872762510742e-06,
+      "loss": 1.2089,
+      "step": 10112
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.862651527617794e-06,
+      "loss": 1.2931,
+      "step": 10113
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.862430292724845e-06,
+      "loss": 1.328,
+      "step": 10114
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.862209057831898e-06,
+      "loss": 1.2331,
+      "step": 10115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.861987822938949e-06,
+      "loss": 1.3288,
+      "step": 10116
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.861766588046001e-06,
+      "loss": 1.3206,
+      "step": 10117
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.861545353153054e-06,
+      "loss": 1.3215,
+      "step": 10118
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.861324118260106e-06,
+      "loss": 1.3314,
+      "step": 10119
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.861102883367159e-06,
+      "loss": 1.3725,
+      "step": 10120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.86088164847421e-06,
+      "loss": 1.3162,
+      "step": 10121
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.860660413581263e-06,
+      "loss": 1.1618,
+      "step": 10122
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.860439178688313e-06,
+      "loss": 1.2984,
+      "step": 10123
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.860217943795366e-06,
+      "loss": 1.3279,
+      "step": 10124
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.859996708902419e-06,
+      "loss": 1.1851,
+      "step": 10125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.859775474009471e-06,
+      "loss": 1.144,
+      "step": 10126
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.859554239116522e-06,
+      "loss": 1.2791,
+      "step": 10127
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.859333004223575e-06,
+      "loss": 1.347,
+      "step": 10128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.859111769330627e-06,
+      "loss": 1.2873,
+      "step": 10129
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.858890534437678e-06,
+      "loss": 1.2202,
+      "step": 10130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.858669299544731e-06,
+      "loss": 1.2093,
+      "step": 10131
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.858448064651782e-06,
+      "loss": 1.2223,
+      "step": 10132
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.858226829758835e-06,
+      "loss": 1.4782,
+      "step": 10133
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.858005594865887e-06,
+      "loss": 1.2631,
+      "step": 10134
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.85778435997294e-06,
+      "loss": 1.3724,
+      "step": 10135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.857563125079992e-06,
+      "loss": 1.1366,
+      "step": 10136
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.857341890187043e-06,
+      "loss": 1.1907,
+      "step": 10137
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.857120655294096e-06,
+      "loss": 1.2025,
+      "step": 10138
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.856899420401147e-06,
+      "loss": 1.3017,
+      "step": 10139
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.8566781855082e-06,
+      "loss": 1.3244,
+      "step": 10140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.856456950615252e-06,
+      "loss": 1.2175,
+      "step": 10141
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.856235715722305e-06,
+      "loss": 1.3224,
+      "step": 10142
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.856014480829356e-06,
+      "loss": 1.4146,
+      "step": 10143
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.855793245936408e-06,
+      "loss": 1.2714,
+      "step": 10144
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.85557201104346e-06,
+      "loss": 1.2207,
+      "step": 10145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.855350776150513e-06,
+      "loss": 1.3835,
+      "step": 10146
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.855129541257564e-06,
+      "loss": 1.3764,
+      "step": 10147
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.854908306364615e-06,
+      "loss": 1.3572,
+      "step": 10148
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.854687071471668e-06,
+      "loss": 1.1465,
+      "step": 10149
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.85446583657872e-06,
+      "loss": 1.2926,
+      "step": 10150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.854244601685773e-06,
+      "loss": 1.2769,
+      "step": 10151
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.854023366792826e-06,
+      "loss": 1.4056,
+      "step": 10152
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.853802131899878e-06,
+      "loss": 1.3484,
+      "step": 10153
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.85358089700693e-06,
+      "loss": 1.2842,
+      "step": 10154
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.85335966211398e-06,
+      "loss": 1.3234,
+      "step": 10155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.853138427221033e-06,
+      "loss": 1.1722,
+      "step": 10156
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.852917192328085e-06,
+      "loss": 1.3407,
+      "step": 10157
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.852695957435138e-06,
+      "loss": 1.3908,
+      "step": 10158
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.852474722542189e-06,
+      "loss": 1.2772,
+      "step": 10159
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.852253487649241e-06,
+      "loss": 1.219,
+      "step": 10160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.852032252756294e-06,
+      "loss": 1.2973,
+      "step": 10161
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.851811017863347e-06,
+      "loss": 1.3887,
+      "step": 10162
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.851589782970398e-06,
+      "loss": 1.2061,
+      "step": 10163
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.85136854807745e-06,
+      "loss": 1.4059,
+      "step": 10164
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.851147313184501e-06,
+      "loss": 1.2896,
+      "step": 10165
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.850926078291554e-06,
+      "loss": 1.2054,
+      "step": 10166
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.850704843398606e-06,
+      "loss": 1.203,
+      "step": 10167
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.850483608505659e-06,
+      "loss": 1.2606,
+      "step": 10168
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.850262373612712e-06,
+      "loss": 1.241,
+      "step": 10169
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.850041138719763e-06,
+      "loss": 1.2995,
+      "step": 10170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.849819903826813e-06,
+      "loss": 1.2771,
+      "step": 10171
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.849598668933866e-06,
+      "loss": 1.2593,
+      "step": 10172
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.849377434040919e-06,
+      "loss": 1.2557,
+      "step": 10173
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.849156199147971e-06,
+      "loss": 1.221,
+      "step": 10174
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.848934964255024e-06,
+      "loss": 1.2965,
+      "step": 10175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.848713729362075e-06,
+      "loss": 1.2257,
+      "step": 10176
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.848492494469127e-06,
+      "loss": 1.3336,
+      "step": 10177
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.84827125957618e-06,
+      "loss": 1.365,
+      "step": 10178
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.848050024683231e-06,
+      "loss": 1.2472,
+      "step": 10179
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.847828789790284e-06,
+      "loss": 1.2209,
+      "step": 10180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.847607554897334e-06,
+      "loss": 1.3032,
+      "step": 10181
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.847386320004387e-06,
+      "loss": 1.3233,
+      "step": 10182
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.84716508511144e-06,
+      "loss": 1.1667,
+      "step": 10183
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.846943850218492e-06,
+      "loss": 1.4735,
+      "step": 10184
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.846722615325545e-06,
+      "loss": 1.4442,
+      "step": 10185
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.846501380432596e-06,
+      "loss": 1.2456,
+      "step": 10186
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.846280145539647e-06,
+      "loss": 1.2554,
+      "step": 10187
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.8460589106467e-06,
+      "loss": 1.2382,
+      "step": 10188
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.845837675753752e-06,
+      "loss": 1.3413,
+      "step": 10189
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.845616440860805e-06,
+      "loss": 1.1387,
+      "step": 10190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.845395205967857e-06,
+      "loss": 1.3451,
+      "step": 10191
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.845173971074908e-06,
+      "loss": 1.2804,
+      "step": 10192
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.84495273618196e-06,
+      "loss": 1.2196,
+      "step": 10193
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.844731501289013e-06,
+      "loss": 1.2756,
+      "step": 10194
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.844510266396064e-06,
+      "loss": 1.2683,
+      "step": 10195
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.844289031503117e-06,
+      "loss": 1.1867,
+      "step": 10196
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.84406779661017e-06,
+      "loss": 1.1951,
+      "step": 10197
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.84384656171722e-06,
+      "loss": 1.217,
+      "step": 10198
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.843625326824273e-06,
+      "loss": 1.3311,
+      "step": 10199
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.843404091931326e-06,
+      "loss": 1.3764,
+      "step": 10200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.843182857038378e-06,
+      "loss": 1.3002,
+      "step": 10201
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.842961622145429e-06,
+      "loss": 1.1942,
+      "step": 10202
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.842740387252482e-06,
+      "loss": 1.2175,
+      "step": 10203
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.842519152359533e-06,
+      "loss": 1.1565,
+      "step": 10204
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.842297917466585e-06,
+      "loss": 1.4079,
+      "step": 10205
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.842076682573638e-06,
+      "loss": 1.2318,
+      "step": 10206
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.84185544768069e-06,
+      "loss": 1.3391,
+      "step": 10207
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.841634212787743e-06,
+      "loss": 1.1891,
+      "step": 10208
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.841412977894794e-06,
+      "loss": 1.3508,
+      "step": 10209
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.841191743001847e-06,
+      "loss": 1.2365,
+      "step": 10210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.840970508108898e-06,
+      "loss": 1.4381,
+      "step": 10211
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.84074927321595e-06,
+      "loss": 1.368,
+      "step": 10212
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.840528038323003e-06,
+      "loss": 1.2692,
+      "step": 10213
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.840306803430054e-06,
+      "loss": 1.324,
+      "step": 10214
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.840085568537106e-06,
+      "loss": 1.209,
+      "step": 10215
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.839864333644159e-06,
+      "loss": 1.2325,
+      "step": 10216
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.839643098751212e-06,
+      "loss": 1.1724,
+      "step": 10217
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.839421863858262e-06,
+      "loss": 1.2162,
+      "step": 10218
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.839200628965315e-06,
+      "loss": 1.1878,
+      "step": 10219
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.838979394072366e-06,
+      "loss": 1.2226,
+      "step": 10220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.838758159179419e-06,
+      "loss": 1.2841,
+      "step": 10221
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.838536924286471e-06,
+      "loss": 1.3769,
+      "step": 10222
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.838315689393524e-06,
+      "loss": 1.2974,
+      "step": 10223
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.838094454500576e-06,
+      "loss": 1.3629,
+      "step": 10224
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.837873219607627e-06,
+      "loss": 1.3351,
+      "step": 10225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.83765198471468e-06,
+      "loss": 1.2276,
+      "step": 10226
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.837430749821731e-06,
+      "loss": 1.2178,
+      "step": 10227
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.837209514928784e-06,
+      "loss": 1.3179,
+      "step": 10228
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.836988280035836e-06,
+      "loss": 1.3315,
+      "step": 10229
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.836767045142887e-06,
+      "loss": 1.1378,
+      "step": 10230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.83654581024994e-06,
+      "loss": 1.3336,
+      "step": 10231
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.836324575356992e-06,
+      "loss": 1.3077,
+      "step": 10232
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.836103340464045e-06,
+      "loss": 1.3691,
+      "step": 10233
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.835882105571097e-06,
+      "loss": 1.3392,
+      "step": 10234
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.835660870678148e-06,
+      "loss": 1.1592,
+      "step": 10235
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.8354396357852e-06,
+      "loss": 1.3572,
+      "step": 10236
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.835218400892252e-06,
+      "loss": 1.3896,
+      "step": 10237
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.834997165999305e-06,
+      "loss": 1.3341,
+      "step": 10238
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.834775931106357e-06,
+      "loss": 1.229,
+      "step": 10239
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.83455469621341e-06,
+      "loss": 1.2435,
+      "step": 10240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.83433346132046e-06,
+      "loss": 1.1633,
+      "step": 10241
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.834112226427513e-06,
+      "loss": 1.3856,
+      "step": 10242
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.833890991534564e-06,
+      "loss": 1.2648,
+      "step": 10243
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.833669756641617e-06,
+      "loss": 1.1984,
+      "step": 10244
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.83344852174867e-06,
+      "loss": 1.2452,
+      "step": 10245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.833227286855722e-06,
+      "loss": 1.204,
+      "step": 10246
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.833006051962773e-06,
+      "loss": 1.3598,
+      "step": 10247
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.832784817069826e-06,
+      "loss": 1.258,
+      "step": 10248
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.832563582176878e-06,
+      "loss": 1.1358,
+      "step": 10249
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.83234234728393e-06,
+      "loss": 1.2949,
+      "step": 10250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.832121112390982e-06,
+      "loss": 1.1604,
+      "step": 10251
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.831899877498033e-06,
+      "loss": 1.2802,
+      "step": 10252
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.831678642605085e-06,
+      "loss": 1.2148,
+      "step": 10253
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.831457407712138e-06,
+      "loss": 1.3808,
+      "step": 10254
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.83123617281919e-06,
+      "loss": 1.2736,
+      "step": 10255
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.831014937926243e-06,
+      "loss": 1.1235,
+      "step": 10256
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.830793703033296e-06,
+      "loss": 1.2293,
+      "step": 10257
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.830572468140347e-06,
+      "loss": 1.2142,
+      "step": 10258
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.830351233247398e-06,
+      "loss": 1.3445,
+      "step": 10259
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.83012999835445e-06,
+      "loss": 1.3154,
+      "step": 10260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.829908763461503e-06,
+      "loss": 1.1653,
+      "step": 10261
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.829687528568555e-06,
+      "loss": 1.228,
+      "step": 10262
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.829466293675606e-06,
+      "loss": 1.3288,
+      "step": 10263
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.829245058782659e-06,
+      "loss": 1.1888,
+      "step": 10264
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.829023823889712e-06,
+      "loss": 1.3909,
+      "step": 10265
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.828802588996764e-06,
+      "loss": 1.3672,
+      "step": 10266
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.828581354103815e-06,
+      "loss": 1.2468,
+      "step": 10267
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.828360119210868e-06,
+      "loss": 1.3329,
+      "step": 10268
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.828138884317919e-06,
+      "loss": 1.2506,
+      "step": 10269
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.827917649424971e-06,
+      "loss": 1.2325,
+      "step": 10270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.827696414532024e-06,
+      "loss": 1.4022,
+      "step": 10271
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.827475179639076e-06,
+      "loss": 1.2346,
+      "step": 10272
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.827253944746129e-06,
+      "loss": 1.2287,
+      "step": 10273
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.82703270985318e-06,
+      "loss": 1.2652,
+      "step": 10274
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.82681147496023e-06,
+      "loss": 1.3088,
+      "step": 10275
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.826590240067283e-06,
+      "loss": 1.2883,
+      "step": 10276
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.826369005174336e-06,
+      "loss": 1.2671,
+      "step": 10277
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.826147770281389e-06,
+      "loss": 1.2885,
+      "step": 10278
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.825926535388441e-06,
+      "loss": 1.1961,
+      "step": 10279
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.825705300495492e-06,
+      "loss": 1.3795,
+      "step": 10280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.825484065602545e-06,
+      "loss": 1.1944,
+      "step": 10281
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.825262830709597e-06,
+      "loss": 1.1656,
+      "step": 10282
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.825041595816648e-06,
+      "loss": 1.393,
+      "step": 10283
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.824820360923701e-06,
+      "loss": 1.2079,
+      "step": 10284
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.824599126030752e-06,
+      "loss": 1.3202,
+      "step": 10285
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.824377891137804e-06,
+      "loss": 1.3597,
+      "step": 10286
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.824156656244857e-06,
+      "loss": 1.2959,
+      "step": 10287
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.82393542135191e-06,
+      "loss": 1.275,
+      "step": 10288
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.823714186458962e-06,
+      "loss": 1.1801,
+      "step": 10289
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.823492951566013e-06,
+      "loss": 1.2755,
+      "step": 10290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.823271716673066e-06,
+      "loss": 1.3063,
+      "step": 10291
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.823050481780117e-06,
+      "loss": 1.2486,
+      "step": 10292
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.82282924688717e-06,
+      "loss": 1.1948,
+      "step": 10293
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.822608011994222e-06,
+      "loss": 1.2918,
+      "step": 10294
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.822386777101275e-06,
+      "loss": 1.4255,
+      "step": 10295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.822165542208326e-06,
+      "loss": 1.4083,
+      "step": 10296
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.821944307315378e-06,
+      "loss": 1.2325,
+      "step": 10297
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.82172307242243e-06,
+      "loss": 1.195,
+      "step": 10298
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.821501837529482e-06,
+      "loss": 1.2816,
+      "step": 10299
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.821280602636534e-06,
+      "loss": 1.2001,
+      "step": 10300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.821059367743587e-06,
+      "loss": 1.2842,
+      "step": 10301
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.820838132850638e-06,
+      "loss": 1.1829,
+      "step": 10302
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.82061689795769e-06,
+      "loss": 1.2391,
+      "step": 10303
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.820395663064743e-06,
+      "loss": 1.3387,
+      "step": 10304
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.820174428171796e-06,
+      "loss": 1.1479,
+      "step": 10305
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.819953193278847e-06,
+      "loss": 1.2781,
+      "step": 10306
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.8197319583859e-06,
+      "loss": 1.1796,
+      "step": 10307
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.81951072349295e-06,
+      "loss": 1.2341,
+      "step": 10308
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.819289488600003e-06,
+      "loss": 1.2857,
+      "step": 10309
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.819068253707055e-06,
+      "loss": 1.2819,
+      "step": 10310
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.818847018814108e-06,
+      "loss": 1.1637,
+      "step": 10311
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.81862578392116e-06,
+      "loss": 1.2449,
+      "step": 10312
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.818404549028211e-06,
+      "loss": 1.253,
+      "step": 10313
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.818183314135264e-06,
+      "loss": 1.2943,
+      "step": 10314
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.817962079242315e-06,
+      "loss": 1.3379,
+      "step": 10315
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.817740844349368e-06,
+      "loss": 1.2783,
+      "step": 10316
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.81751960945642e-06,
+      "loss": 1.3087,
+      "step": 10317
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.817298374563471e-06,
+      "loss": 1.3516,
+      "step": 10318
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.817077139670524e-06,
+      "loss": 1.1813,
+      "step": 10319
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.816855904777576e-06,
+      "loss": 1.2051,
+      "step": 10320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.816634669884629e-06,
+      "loss": 1.3097,
+      "step": 10321
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.816413434991682e-06,
+      "loss": 1.3145,
+      "step": 10322
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.816192200098732e-06,
+      "loss": 1.2743,
+      "step": 10323
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.815970965205783e-06,
+      "loss": 1.1801,
+      "step": 10324
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.815749730312836e-06,
+      "loss": 1.2883,
+      "step": 10325
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.815528495419889e-06,
+      "loss": 1.3643,
+      "step": 10326
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.815307260526941e-06,
+      "loss": 1.2928,
+      "step": 10327
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.815086025633994e-06,
+      "loss": 1.2051,
+      "step": 10328
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.814864790741045e-06,
+      "loss": 1.2466,
+      "step": 10329
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.814643555848097e-06,
+      "loss": 1.3095,
+      "step": 10330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.814422320955148e-06,
+      "loss": 1.256,
+      "step": 10331
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.814201086062201e-06,
+      "loss": 1.2817,
+      "step": 10332
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.813979851169254e-06,
+      "loss": 1.2177,
+      "step": 10333
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.813758616276304e-06,
+      "loss": 1.2934,
+      "step": 10334
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.813537381383357e-06,
+      "loss": 1.1118,
+      "step": 10335
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.81331614649041e-06,
+      "loss": 1.3558,
+      "step": 10336
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.813094911597462e-06,
+      "loss": 1.3329,
+      "step": 10337
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.812873676704515e-06,
+      "loss": 1.3574,
+      "step": 10338
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.812652441811566e-06,
+      "loss": 1.3552,
+      "step": 10339
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.812431206918617e-06,
+      "loss": 1.2048,
+      "step": 10340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.81220997202567e-06,
+      "loss": 1.3621,
+      "step": 10341
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.811988737132722e-06,
+      "loss": 1.15,
+      "step": 10342
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.811767502239775e-06,
+      "loss": 1.1824,
+      "step": 10343
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.811546267346827e-06,
+      "loss": 1.3401,
+      "step": 10344
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.811325032453878e-06,
+      "loss": 1.3771,
+      "step": 10345
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.81110379756093e-06,
+      "loss": 1.1797,
+      "step": 10346
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.810882562667982e-06,
+      "loss": 1.3254,
+      "step": 10347
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.810661327775034e-06,
+      "loss": 1.4535,
+      "step": 10348
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.810440092882087e-06,
+      "loss": 1.253,
+      "step": 10349
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.81021885798914e-06,
+      "loss": 1.3558,
+      "step": 10350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.80999762309619e-06,
+      "loss": 1.172,
+      "step": 10351
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.809776388203243e-06,
+      "loss": 1.1867,
+      "step": 10352
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.809555153310296e-06,
+      "loss": 1.2036,
+      "step": 10353
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.809333918417348e-06,
+      "loss": 1.2343,
+      "step": 10354
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.809112683524399e-06,
+      "loss": 1.3019,
+      "step": 10355
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.80889144863145e-06,
+      "loss": 1.3697,
+      "step": 10356
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.808670213738503e-06,
+      "loss": 1.3202,
+      "step": 10357
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.808448978845555e-06,
+      "loss": 1.2105,
+      "step": 10358
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.808227743952608e-06,
+      "loss": 1.264,
+      "step": 10359
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.80800650905966e-06,
+      "loss": 1.2396,
+      "step": 10360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.807785274166713e-06,
+      "loss": 1.3121,
+      "step": 10361
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.807564039273764e-06,
+      "loss": 1.1661,
+      "step": 10362
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.807342804380815e-06,
+      "loss": 1.1602,
+      "step": 10363
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.807121569487868e-06,
+      "loss": 1.3324,
+      "step": 10364
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.80690033459492e-06,
+      "loss": 1.0943,
+      "step": 10365
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.806679099701973e-06,
+      "loss": 1.2252,
+      "step": 10366
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.806457864809024e-06,
+      "loss": 1.3032,
+      "step": 10367
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.806236629916076e-06,
+      "loss": 1.2753,
+      "step": 10368
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.806015395023129e-06,
+      "loss": 1.2347,
+      "step": 10369
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.805794160130182e-06,
+      "loss": 1.2673,
+      "step": 10370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.805572925237232e-06,
+      "loss": 1.2921,
+      "step": 10371
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.805351690344285e-06,
+      "loss": 1.3841,
+      "step": 10372
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.805130455451336e-06,
+      "loss": 1.1435,
+      "step": 10373
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.804909220558389e-06,
+      "loss": 1.3626,
+      "step": 10374
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.804687985665441e-06,
+      "loss": 1.245,
+      "step": 10375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.804466750772494e-06,
+      "loss": 1.3838,
+      "step": 10376
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.804245515879546e-06,
+      "loss": 1.192,
+      "step": 10377
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.804024280986597e-06,
+      "loss": 1.2169,
+      "step": 10378
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.80380304609365e-06,
+      "loss": 1.2486,
+      "step": 10379
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.803581811200701e-06,
+      "loss": 1.3523,
+      "step": 10380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.803360576307753e-06,
+      "loss": 1.257,
+      "step": 10381
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.803139341414806e-06,
+      "loss": 1.2598,
+      "step": 10382
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.802918106521859e-06,
+      "loss": 1.102,
+      "step": 10383
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.80269687162891e-06,
+      "loss": 1.2716,
+      "step": 10384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.802475636735962e-06,
+      "loss": 1.2832,
+      "step": 10385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.802254401843015e-06,
+      "loss": 1.3025,
+      "step": 10386
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.802033166950066e-06,
+      "loss": 1.2912,
+      "step": 10387
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.801811932057118e-06,
+      "loss": 1.2427,
+      "step": 10388
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.80159069716417e-06,
+      "loss": 1.2338,
+      "step": 10389
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.801369462271222e-06,
+      "loss": 1.2842,
+      "step": 10390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.801148227378275e-06,
+      "loss": 1.2661,
+      "step": 10391
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.800926992485327e-06,
+      "loss": 1.3307,
+      "step": 10392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.80070575759238e-06,
+      "loss": 1.2935,
+      "step": 10393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.80048452269943e-06,
+      "loss": 1.2961,
+      "step": 10394
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.800263287806483e-06,
+      "loss": 1.2292,
+      "step": 10395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.800042052913534e-06,
+      "loss": 1.3261,
+      "step": 10396
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.799820818020587e-06,
+      "loss": 1.2951,
+      "step": 10397
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.79959958312764e-06,
+      "loss": 1.4766,
+      "step": 10398
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.799378348234692e-06,
+      "loss": 1.1688,
+      "step": 10399
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.799157113341743e-06,
+      "loss": 1.3027,
+      "step": 10400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.798935878448796e-06,
+      "loss": 1.3175,
+      "step": 10401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.798714643555848e-06,
+      "loss": 1.2053,
+      "step": 10402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.798493408662899e-06,
+      "loss": 1.2498,
+      "step": 10403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.798272173769952e-06,
+      "loss": 1.237,
+      "step": 10404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.798050938877004e-06,
+      "loss": 1.3503,
+      "step": 10405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.797829703984055e-06,
+      "loss": 1.171,
+      "step": 10406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.797608469091108e-06,
+      "loss": 1.2463,
+      "step": 10407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.79738723419816e-06,
+      "loss": 1.351,
+      "step": 10408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.797165999305213e-06,
+      "loss": 1.3314,
+      "step": 10409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.796944764412264e-06,
+      "loss": 1.2551,
+      "step": 10410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.796723529519317e-06,
+      "loss": 1.3121,
+      "step": 10411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.796502294626368e-06,
+      "loss": 1.2681,
+      "step": 10412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.79628105973342e-06,
+      "loss": 1.2894,
+      "step": 10413
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.796059824840473e-06,
+      "loss": 1.2799,
+      "step": 10414
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.795838589947525e-06,
+      "loss": 1.2479,
+      "step": 10415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.795617355054578e-06,
+      "loss": 1.3318,
+      "step": 10416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.795396120161629e-06,
+      "loss": 1.3127,
+      "step": 10417
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.795174885268681e-06,
+      "loss": 1.2511,
+      "step": 10418
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.794953650375732e-06,
+      "loss": 1.3749,
+      "step": 10419
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.794732415482785e-06,
+      "loss": 1.2571,
+      "step": 10420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.794511180589838e-06,
+      "loss": 1.2773,
+      "step": 10421
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.794289945696889e-06,
+      "loss": 1.2829,
+      "step": 10422
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.794068710803941e-06,
+      "loss": 1.3595,
+      "step": 10423
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.793847475910994e-06,
+      "loss": 1.3077,
+      "step": 10424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.793626241018046e-06,
+      "loss": 1.3191,
+      "step": 10425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.793405006125099e-06,
+      "loss": 1.2795,
+      "step": 10426
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.79318377123215e-06,
+      "loss": 1.2849,
+      "step": 10427
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.7929625363392e-06,
+      "loss": 1.4356,
+      "step": 10428
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.792741301446253e-06,
+      "loss": 1.2799,
+      "step": 10429
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.792520066553306e-06,
+      "loss": 1.3737,
+      "step": 10430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.792298831660359e-06,
+      "loss": 1.2471,
+      "step": 10431
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.792077596767411e-06,
+      "loss": 1.3009,
+      "step": 10432
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.791856361874462e-06,
+      "loss": 1.2784,
+      "step": 10433
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.791635126981515e-06,
+      "loss": 1.3417,
+      "step": 10434
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.791413892088566e-06,
+      "loss": 1.3125,
+      "step": 10435
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.791192657195618e-06,
+      "loss": 1.164,
+      "step": 10436
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.790971422302671e-06,
+      "loss": 1.2544,
+      "step": 10437
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.790750187409722e-06,
+      "loss": 1.3676,
+      "step": 10438
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.790528952516774e-06,
+      "loss": 1.2244,
+      "step": 10439
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.790307717623827e-06,
+      "loss": 1.2122,
+      "step": 10440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.79008648273088e-06,
+      "loss": 1.482,
+      "step": 10441
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.789865247837932e-06,
+      "loss": 1.2142,
+      "step": 10442
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.789644012944983e-06,
+      "loss": 1.3423,
+      "step": 10443
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.789422778052034e-06,
+      "loss": 1.2084,
+      "step": 10444
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.789201543159087e-06,
+      "loss": 1.1629,
+      "step": 10445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.78898030826614e-06,
+      "loss": 1.3436,
+      "step": 10446
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.788759073373192e-06,
+      "loss": 1.2426,
+      "step": 10447
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.788537838480245e-06,
+      "loss": 1.1677,
+      "step": 10448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.788316603587296e-06,
+      "loss": 1.3735,
+      "step": 10449
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.788095368694348e-06,
+      "loss": 1.2203,
+      "step": 10450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.787874133801399e-06,
+      "loss": 1.2806,
+      "step": 10451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.787652898908452e-06,
+      "loss": 1.3015,
+      "step": 10452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.787431664015504e-06,
+      "loss": 1.2689,
+      "step": 10453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.787210429122557e-06,
+      "loss": 1.3391,
+      "step": 10454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.786989194229608e-06,
+      "loss": 1.24,
+      "step": 10455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.78676795933666e-06,
+      "loss": 1.3075,
+      "step": 10456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.786546724443713e-06,
+      "loss": 1.1539,
+      "step": 10457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.786325489550766e-06,
+      "loss": 1.2926,
+      "step": 10458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.786104254657817e-06,
+      "loss": 1.2277,
+      "step": 10459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.785883019764867e-06,
+      "loss": 1.3271,
+      "step": 10460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.78566178487192e-06,
+      "loss": 1.3079,
+      "step": 10461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.785440549978973e-06,
+      "loss": 1.2539,
+      "step": 10462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.785219315086025e-06,
+      "loss": 1.178,
+      "step": 10463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.784998080193078e-06,
+      "loss": 1.1493,
+      "step": 10464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.78477684530013e-06,
+      "loss": 1.2801,
+      "step": 10465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.784555610407181e-06,
+      "loss": 1.307,
+      "step": 10466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.784334375514234e-06,
+      "loss": 1.3509,
+      "step": 10467
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.784113140621285e-06,
+      "loss": 1.3538,
+      "step": 10468
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.783891905728338e-06,
+      "loss": 1.3198,
+      "step": 10469
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.78367067083539e-06,
+      "loss": 1.3207,
+      "step": 10470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.783449435942441e-06,
+      "loss": 1.152,
+      "step": 10471
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.783228201049494e-06,
+      "loss": 1.2448,
+      "step": 10472
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.783006966156546e-06,
+      "loss": 1.326,
+      "step": 10473
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.782785731263599e-06,
+      "loss": 1.3346,
+      "step": 10474
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.78256449637065e-06,
+      "loss": 1.2064,
+      "step": 10475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.782343261477702e-06,
+      "loss": 1.3371,
+      "step": 10476
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.782122026584753e-06,
+      "loss": 1.2744,
+      "step": 10477
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.781900791691806e-06,
+      "loss": 1.1849,
+      "step": 10478
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.781679556798859e-06,
+      "loss": 1.2897,
+      "step": 10479
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.781458321905911e-06,
+      "loss": 1.2339,
+      "step": 10480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.781237087012964e-06,
+      "loss": 1.4052,
+      "step": 10481
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.781015852120015e-06,
+      "loss": 1.3149,
+      "step": 10482
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.780794617227067e-06,
+      "loss": 1.4624,
+      "step": 10483
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.780573382334118e-06,
+      "loss": 1.2728,
+      "step": 10484
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.780352147441171e-06,
+      "loss": 1.2107,
+      "step": 10485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.780130912548224e-06,
+      "loss": 1.2028,
+      "step": 10486
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.779909677655276e-06,
+      "loss": 1.3143,
+      "step": 10487
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.779688442762327e-06,
+      "loss": 1.3313,
+      "step": 10488
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.77946720786938e-06,
+      "loss": 1.2899,
+      "step": 10489
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.779245972976432e-06,
+      "loss": 1.217,
+      "step": 10490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.779024738083483e-06,
+      "loss": 1.2111,
+      "step": 10491
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.778803503190536e-06,
+      "loss": 1.2939,
+      "step": 10492
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.778582268297587e-06,
+      "loss": 1.171,
+      "step": 10493
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.77836103340464e-06,
+      "loss": 1.3032,
+      "step": 10494
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.778139798511692e-06,
+      "loss": 1.3151,
+      "step": 10495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.777918563618745e-06,
+      "loss": 1.1519,
+      "step": 10496
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.777697328725797e-06,
+      "loss": 1.3567,
+      "step": 10497
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.777476093832848e-06,
+      "loss": 1.246,
+      "step": 10498
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.7772548589399e-06,
+      "loss": 1.2824,
+      "step": 10499
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.777033624046952e-06,
+      "loss": 1.1667,
+      "step": 10500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.776812389154004e-06,
+      "loss": 1.2704,
+      "step": 10501
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.776591154261057e-06,
+      "loss": 1.1713,
+      "step": 10502
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.77636991936811e-06,
+      "loss": 1.268,
+      "step": 10503
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.77614868447516e-06,
+      "loss": 1.237,
+      "step": 10504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.775927449582213e-06,
+      "loss": 1.271,
+      "step": 10505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.775706214689266e-06,
+      "loss": 1.2426,
+      "step": 10506
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.775484979796317e-06,
+      "loss": 1.1825,
+      "step": 10507
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.775263744903369e-06,
+      "loss": 1.2952,
+      "step": 10508
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.775042510010422e-06,
+      "loss": 1.2669,
+      "step": 10509
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.774821275117473e-06,
+      "loss": 1.2392,
+      "step": 10510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.774600040224525e-06,
+      "loss": 1.3291,
+      "step": 10511
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.774378805331578e-06,
+      "loss": 1.3417,
+      "step": 10512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.77415757043863e-06,
+      "loss": 1.2764,
+      "step": 10513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.773936335545683e-06,
+      "loss": 1.2464,
+      "step": 10514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.773715100652734e-06,
+      "loss": 1.16,
+      "step": 10515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.773493865759785e-06,
+      "loss": 1.3412,
+      "step": 10516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.773272630866838e-06,
+      "loss": 1.4013,
+      "step": 10517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.77305139597389e-06,
+      "loss": 1.2276,
+      "step": 10518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.772830161080943e-06,
+      "loss": 1.2739,
+      "step": 10519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.772608926187994e-06,
+      "loss": 1.3429,
+      "step": 10520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.772387691295046e-06,
+      "loss": 1.1382,
+      "step": 10521
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.772166456402099e-06,
+      "loss": 1.277,
+      "step": 10522
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.77194522150915e-06,
+      "loss": 1.2613,
+      "step": 10523
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.771723986616202e-06,
+      "loss": 1.3332,
+      "step": 10524
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.771502751723255e-06,
+      "loss": 1.2895,
+      "step": 10525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.771281516830306e-06,
+      "loss": 1.1929,
+      "step": 10526
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.771060281937359e-06,
+      "loss": 1.2329,
+      "step": 10527
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.770839047044411e-06,
+      "loss": 1.1469,
+      "step": 10528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.770617812151464e-06,
+      "loss": 1.305,
+      "step": 10529
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.770396577258516e-06,
+      "loss": 1.0961,
+      "step": 10530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.770175342365567e-06,
+      "loss": 1.2522,
+      "step": 10531
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.769954107472618e-06,
+      "loss": 1.3014,
+      "step": 10532
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.769732872579671e-06,
+      "loss": 1.1053,
+      "step": 10533
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.769511637686723e-06,
+      "loss": 1.2554,
+      "step": 10534
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.769290402793776e-06,
+      "loss": 1.2325,
+      "step": 10535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.769069167900829e-06,
+      "loss": 1.3298,
+      "step": 10536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.76884793300788e-06,
+      "loss": 1.2275,
+      "step": 10537
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.768626698114932e-06,
+      "loss": 1.2727,
+      "step": 10538
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.768405463221983e-06,
+      "loss": 1.3348,
+      "step": 10539
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.768184228329036e-06,
+      "loss": 1.3004,
+      "step": 10540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.767962993436088e-06,
+      "loss": 1.3466,
+      "step": 10541
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.76774175854314e-06,
+      "loss": 1.2651,
+      "step": 10542
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.767520523650192e-06,
+      "loss": 1.2808,
+      "step": 10543
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.767299288757245e-06,
+      "loss": 1.4051,
+      "step": 10544
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.767078053864297e-06,
+      "loss": 1.249,
+      "step": 10545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.76685681897135e-06,
+      "loss": 1.2564,
+      "step": 10546
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.7666355840784e-06,
+      "loss": 1.2983,
+      "step": 10547
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.766414349185452e-06,
+      "loss": 1.1403,
+      "step": 10548
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.766193114292504e-06,
+      "loss": 1.2989,
+      "step": 10549
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.765971879399557e-06,
+      "loss": 1.3074,
+      "step": 10550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.76575064450661e-06,
+      "loss": 1.1839,
+      "step": 10551
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.765529409613662e-06,
+      "loss": 1.2973,
+      "step": 10552
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.765308174720713e-06,
+      "loss": 1.2357,
+      "step": 10553
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.765086939827766e-06,
+      "loss": 1.3387,
+      "step": 10554
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.764865704934818e-06,
+      "loss": 1.265,
+      "step": 10555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.764644470041869e-06,
+      "loss": 1.3481,
+      "step": 10556
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.764423235148922e-06,
+      "loss": 1.4204,
+      "step": 10557
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.764202000255974e-06,
+      "loss": 1.3638,
+      "step": 10558
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.763980765363025e-06,
+      "loss": 1.3239,
+      "step": 10559
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.763759530470078e-06,
+      "loss": 1.2909,
+      "step": 10560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.76353829557713e-06,
+      "loss": 1.3689,
+      "step": 10561
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.763317060684183e-06,
+      "loss": 1.1864,
+      "step": 10562
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.763095825791234e-06,
+      "loss": 1.2012,
+      "step": 10563
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.762874590898285e-06,
+      "loss": 1.3051,
+      "step": 10564
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.762653356005338e-06,
+      "loss": 1.2917,
+      "step": 10565
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.76243212111239e-06,
+      "loss": 1.2403,
+      "step": 10566
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.762210886219443e-06,
+      "loss": 1.3171,
+      "step": 10567
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.761989651326495e-06,
+      "loss": 1.2825,
+      "step": 10568
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.761768416433548e-06,
+      "loss": 1.3406,
+      "step": 10569
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.761547181540599e-06,
+      "loss": 1.2997,
+      "step": 10570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.761325946647651e-06,
+      "loss": 1.0825,
+      "step": 10571
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.761104711754702e-06,
+      "loss": 1.3799,
+      "step": 10572
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.760883476861755e-06,
+      "loss": 1.3002,
+      "step": 10573
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.760662241968808e-06,
+      "loss": 1.1889,
+      "step": 10574
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.760441007075859e-06,
+      "loss": 1.1748,
+      "step": 10575
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.760219772182911e-06,
+      "loss": 1.3559,
+      "step": 10576
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.759998537289964e-06,
+      "loss": 1.2235,
+      "step": 10577
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.759777302397016e-06,
+      "loss": 1.1553,
+      "step": 10578
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.759556067504067e-06,
+      "loss": 1.3636,
+      "step": 10579
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.75933483261112e-06,
+      "loss": 1.2129,
+      "step": 10580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.75911359771817e-06,
+      "loss": 1.2697,
+      "step": 10581
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.758892362825223e-06,
+      "loss": 1.2012,
+      "step": 10582
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.758671127932276e-06,
+      "loss": 1.3393,
+      "step": 10583
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.758449893039329e-06,
+      "loss": 1.1628,
+      "step": 10584
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.758228658146381e-06,
+      "loss": 1.3067,
+      "step": 10585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.758007423253432e-06,
+      "loss": 1.3518,
+      "step": 10586
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.757786188360485e-06,
+      "loss": 1.3159,
+      "step": 10587
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.757564953467536e-06,
+      "loss": 1.2449,
+      "step": 10588
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.757343718574588e-06,
+      "loss": 1.1757,
+      "step": 10589
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.757122483681641e-06,
+      "loss": 1.1615,
+      "step": 10590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.756901248788694e-06,
+      "loss": 1.2688,
+      "step": 10591
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.756680013895744e-06,
+      "loss": 1.3433,
+      "step": 10592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.756458779002797e-06,
+      "loss": 1.3457,
+      "step": 10593
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.75623754410985e-06,
+      "loss": 1.2239,
+      "step": 10594
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.7560163092169e-06,
+      "loss": 1.4351,
+      "step": 10595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.755795074323953e-06,
+      "loss": 1.1698,
+      "step": 10596
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.755573839431004e-06,
+      "loss": 1.1449,
+      "step": 10597
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.755352604538057e-06,
+      "loss": 1.2304,
+      "step": 10598
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.75513136964511e-06,
+      "loss": 1.2345,
+      "step": 10599
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.754910134752162e-06,
+      "loss": 1.3015,
+      "step": 10600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.754688899859215e-06,
+      "loss": 1.2541,
+      "step": 10601
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.754467664966267e-06,
+      "loss": 1.3172,
+      "step": 10602
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.754246430073318e-06,
+      "loss": 1.2851,
+      "step": 10603
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.754025195180369e-06,
+      "loss": 1.2832,
+      "step": 10604
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.753803960287422e-06,
+      "loss": 1.2564,
+      "step": 10605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.753582725394474e-06,
+      "loss": 1.2644,
+      "step": 10606
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.753361490501527e-06,
+      "loss": 1.4186,
+      "step": 10607
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.753140255608578e-06,
+      "loss": 1.2338,
+      "step": 10608
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.75291902071563e-06,
+      "loss": 1.2675,
+      "step": 10609
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.752697785822683e-06,
+      "loss": 1.2681,
+      "step": 10610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.752476550929734e-06,
+      "loss": 1.293,
+      "step": 10611
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.752255316036787e-06,
+      "loss": 1.2672,
+      "step": 10612
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.75203408114384e-06,
+      "loss": 1.2433,
+      "step": 10613
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.75181284625089e-06,
+      "loss": 1.2533,
+      "step": 10614
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.751591611357943e-06,
+      "loss": 1.3732,
+      "step": 10615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.751370376464995e-06,
+      "loss": 1.2335,
+      "step": 10616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.751149141572048e-06,
+      "loss": 1.3359,
+      "step": 10617
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.7509279066791e-06,
+      "loss": 1.2574,
+      "step": 10618
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.750706671786151e-06,
+      "loss": 1.3376,
+      "step": 10619
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.750485436893202e-06,
+      "loss": 1.2405,
+      "step": 10620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.750264202000255e-06,
+      "loss": 1.3569,
+      "step": 10621
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.750042967107308e-06,
+      "loss": 1.2118,
+      "step": 10622
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.74982173221436e-06,
+      "loss": 1.2065,
+      "step": 10623
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.749600497321411e-06,
+      "loss": 1.3034,
+      "step": 10624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.749379262428464e-06,
+      "loss": 1.2414,
+      "step": 10625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.749158027535516e-06,
+      "loss": 1.2387,
+      "step": 10626
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.748936792642567e-06,
+      "loss": 1.195,
+      "step": 10627
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.74871555774962e-06,
+      "loss": 1.2455,
+      "step": 10628
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.748494322856672e-06,
+      "loss": 1.2064,
+      "step": 10629
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.748273087963723e-06,
+      "loss": 1.2835,
+      "step": 10630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.748051853070776e-06,
+      "loss": 1.1955,
+      "step": 10631
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.747830618177829e-06,
+      "loss": 1.3826,
+      "step": 10632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.747609383284881e-06,
+      "loss": 1.285,
+      "step": 10633
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.747388148391934e-06,
+      "loss": 1.2622,
+      "step": 10634
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.747166913498985e-06,
+      "loss": 1.4563,
+      "step": 10635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.746945678606036e-06,
+      "loss": 1.2022,
+      "step": 10636
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.746724443713088e-06,
+      "loss": 1.2684,
+      "step": 10637
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.746503208820141e-06,
+      "loss": 1.3113,
+      "step": 10638
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.746281973927194e-06,
+      "loss": 1.3419,
+      "step": 10639
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.746060739034246e-06,
+      "loss": 1.2118,
+      "step": 10640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.745839504141297e-06,
+      "loss": 1.2061,
+      "step": 10641
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.74561826924835e-06,
+      "loss": 1.2096,
+      "step": 10642
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.745397034355402e-06,
+      "loss": 1.2984,
+      "step": 10643
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.745175799462453e-06,
+      "loss": 1.243,
+      "step": 10644
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.744954564569506e-06,
+      "loss": 1.1875,
+      "step": 10645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.744733329676557e-06,
+      "loss": 1.2135,
+      "step": 10646
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.74451209478361e-06,
+      "loss": 1.3929,
+      "step": 10647
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.744290859890662e-06,
+      "loss": 1.3482,
+      "step": 10648
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.744069624997715e-06,
+      "loss": 1.3468,
+      "step": 10649
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.743848390104767e-06,
+      "loss": 1.3085,
+      "step": 10650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.743627155211818e-06,
+      "loss": 1.2062,
+      "step": 10651
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.743405920318869e-06,
+      "loss": 1.1565,
+      "step": 10652
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.743184685425922e-06,
+      "loss": 1.3104,
+      "step": 10653
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.742963450532974e-06,
+      "loss": 1.2469,
+      "step": 10654
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.742742215640027e-06,
+      "loss": 1.4471,
+      "step": 10655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.74252098074708e-06,
+      "loss": 1.2358,
+      "step": 10656
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.74229974585413e-06,
+      "loss": 1.2914,
+      "step": 10657
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.742078510961183e-06,
+      "loss": 1.2964,
+      "step": 10658
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.741857276068236e-06,
+      "loss": 1.3495,
+      "step": 10659
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.741636041175286e-06,
+      "loss": 1.3457,
+      "step": 10660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.741414806282339e-06,
+      "loss": 1.1747,
+      "step": 10661
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.741193571389392e-06,
+      "loss": 1.161,
+      "step": 10662
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.740972336496443e-06,
+      "loss": 1.2777,
+      "step": 10663
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.740751101603495e-06,
+      "loss": 1.2365,
+      "step": 10664
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.740529866710548e-06,
+      "loss": 1.2215,
+      "step": 10665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.7403086318176e-06,
+      "loss": 1.1965,
+      "step": 10666
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.740087396924651e-06,
+      "loss": 1.2521,
+      "step": 10667
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.739866162031702e-06,
+      "loss": 1.191,
+      "step": 10668
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.739644927138755e-06,
+      "loss": 1.1306,
+      "step": 10669
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.739423692245808e-06,
+      "loss": 1.379,
+      "step": 10670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.73920245735286e-06,
+      "loss": 1.1358,
+      "step": 10671
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.738981222459913e-06,
+      "loss": 1.2396,
+      "step": 10672
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.738759987566965e-06,
+      "loss": 1.2624,
+      "step": 10673
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.738538752674016e-06,
+      "loss": 1.3035,
+      "step": 10674
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.738317517781069e-06,
+      "loss": 1.2496,
+      "step": 10675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.73809628288812e-06,
+      "loss": 1.2476,
+      "step": 10676
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.737875047995172e-06,
+      "loss": 1.2393,
+      "step": 10677
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.737653813102225e-06,
+      "loss": 1.3031,
+      "step": 10678
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.737432578209276e-06,
+      "loss": 1.334,
+      "step": 10679
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.737211343316329e-06,
+      "loss": 1.1719,
+      "step": 10680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.736990108423381e-06,
+      "loss": 1.1852,
+      "step": 10681
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.736768873530434e-06,
+      "loss": 1.2685,
+      "step": 10682
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.736547638637485e-06,
+      "loss": 1.1821,
+      "step": 10683
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.736326403744537e-06,
+      "loss": 1.3797,
+      "step": 10684
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.736105168851588e-06,
+      "loss": 1.3194,
+      "step": 10685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.735883933958641e-06,
+      "loss": 1.305,
+      "step": 10686
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.735662699065693e-06,
+      "loss": 1.3272,
+      "step": 10687
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.735441464172746e-06,
+      "loss": 1.348,
+      "step": 10688
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.735220229279799e-06,
+      "loss": 1.2735,
+      "step": 10689
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.73499899438685e-06,
+      "loss": 1.1521,
+      "step": 10690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.734777759493902e-06,
+      "loss": 1.1218,
+      "step": 10691
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.734556524600953e-06,
+      "loss": 1.2823,
+      "step": 10692
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.734335289708006e-06,
+      "loss": 1.2774,
+      "step": 10693
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.734114054815058e-06,
+      "loss": 1.3131,
+      "step": 10694
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.733892819922111e-06,
+      "loss": 1.3521,
+      "step": 10695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.733671585029162e-06,
+      "loss": 1.2886,
+      "step": 10696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.733450350136215e-06,
+      "loss": 1.1792,
+      "step": 10697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.733229115243267e-06,
+      "loss": 1.2555,
+      "step": 10698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.733007880350318e-06,
+      "loss": 1.2532,
+      "step": 10699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.73278664545737e-06,
+      "loss": 1.3208,
+      "step": 10700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.732565410564422e-06,
+      "loss": 1.3017,
+      "step": 10701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.732344175671474e-06,
+      "loss": 1.4218,
+      "step": 10702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.732122940778527e-06,
+      "loss": 1.3255,
+      "step": 10703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.73190170588558e-06,
+      "loss": 1.3949,
+      "step": 10704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.731680470992632e-06,
+      "loss": 1.2585,
+      "step": 10705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.731459236099685e-06,
+      "loss": 1.3388,
+      "step": 10706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.731238001206736e-06,
+      "loss": 1.2454,
+      "step": 10707
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.731016766313786e-06,
+      "loss": 1.3626,
+      "step": 10708
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.730795531420839e-06,
+      "loss": 1.2228,
+      "step": 10709
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.730574296527892e-06,
+      "loss": 1.297,
+      "step": 10710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.730353061634944e-06,
+      "loss": 1.2853,
+      "step": 10711
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.730131826741995e-06,
+      "loss": 1.3156,
+      "step": 10712
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.729910591849048e-06,
+      "loss": 1.2536,
+      "step": 10713
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.7296893569561e-06,
+      "loss": 1.3277,
+      "step": 10714
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.729468122063151e-06,
+      "loss": 1.4249,
+      "step": 10715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.729246887170204e-06,
+      "loss": 1.2485,
+      "step": 10716
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.729025652277255e-06,
+      "loss": 1.3133,
+      "step": 10717
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.728804417384307e-06,
+      "loss": 1.2713,
+      "step": 10718
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.72858318249136e-06,
+      "loss": 1.28,
+      "step": 10719
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.728361947598413e-06,
+      "loss": 1.3244,
+      "step": 10720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.728140712705465e-06,
+      "loss": 1.2215,
+      "step": 10721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.727919477812518e-06,
+      "loss": 1.2583,
+      "step": 10722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.727698242919569e-06,
+      "loss": 1.183,
+      "step": 10723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.72747700802662e-06,
+      "loss": 1.2328,
+      "step": 10724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.727255773133672e-06,
+      "loss": 1.3557,
+      "step": 10725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.727034538240725e-06,
+      "loss": 1.2475,
+      "step": 10726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.726813303347778e-06,
+      "loss": 1.1448,
+      "step": 10727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.726592068454829e-06,
+      "loss": 1.2259,
+      "step": 10728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.726370833561881e-06,
+      "loss": 1.2675,
+      "step": 10729
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.726149598668934e-06,
+      "loss": 1.1727,
+      "step": 10730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.725928363775986e-06,
+      "loss": 1.2298,
+      "step": 10731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.725707128883037e-06,
+      "loss": 1.3736,
+      "step": 10732
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.72548589399009e-06,
+      "loss": 1.2521,
+      "step": 10733
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.72526465909714e-06,
+      "loss": 1.2211,
+      "step": 10734
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.725043424204193e-06,
+      "loss": 1.2912,
+      "step": 10735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.724822189311246e-06,
+      "loss": 1.2784,
+      "step": 10736
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.724600954418299e-06,
+      "loss": 1.2631,
+      "step": 10737
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.724379719525351e-06,
+      "loss": 1.3035,
+      "step": 10738
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.724158484632402e-06,
+      "loss": 1.2479,
+      "step": 10739
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.723937249739453e-06,
+      "loss": 1.2536,
+      "step": 10740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.723716014846506e-06,
+      "loss": 1.2263,
+      "step": 10741
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.723494779953558e-06,
+      "loss": 1.1216,
+      "step": 10742
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.723273545060611e-06,
+      "loss": 1.2875,
+      "step": 10743
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.723052310167664e-06,
+      "loss": 1.3631,
+      "step": 10744
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.722831075274714e-06,
+      "loss": 1.3126,
+      "step": 10745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.722609840381767e-06,
+      "loss": 1.4875,
+      "step": 10746
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.72238860548882e-06,
+      "loss": 1.3284,
+      "step": 10747
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.72216737059587e-06,
+      "loss": 1.2605,
+      "step": 10748
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.721946135702923e-06,
+      "loss": 1.3115,
+      "step": 10749
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.721724900809974e-06,
+      "loss": 1.1906,
+      "step": 10750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.721503665917027e-06,
+      "loss": 1.2492,
+      "step": 10751
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.72128243102408e-06,
+      "loss": 1.2056,
+      "step": 10752
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.721061196131132e-06,
+      "loss": 1.2641,
+      "step": 10753
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.720839961238185e-06,
+      "loss": 1.2827,
+      "step": 10754
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.720618726345235e-06,
+      "loss": 1.171,
+      "step": 10755
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.720397491452286e-06,
+      "loss": 1.2048,
+      "step": 10756
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.720176256559339e-06,
+      "loss": 1.3405,
+      "step": 10757
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.719955021666392e-06,
+      "loss": 1.2525,
+      "step": 10758
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.719733786773444e-06,
+      "loss": 1.2653,
+      "step": 10759
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.719512551880497e-06,
+      "loss": 1.2016,
+      "step": 10760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.719291316987548e-06,
+      "loss": 1.1905,
+      "step": 10761
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.7190700820946e-06,
+      "loss": 1.2969,
+      "step": 10762
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.718848847201653e-06,
+      "loss": 1.2571,
+      "step": 10763
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.718627612308704e-06,
+      "loss": 1.1922,
+      "step": 10764
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.718406377415757e-06,
+      "loss": 1.3094,
+      "step": 10765
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.718185142522809e-06,
+      "loss": 1.2612,
+      "step": 10766
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.71796390762986e-06,
+      "loss": 1.1943,
+      "step": 10767
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.717742672736913e-06,
+      "loss": 1.3096,
+      "step": 10768
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.717521437843965e-06,
+      "loss": 1.1933,
+      "step": 10769
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.717300202951018e-06,
+      "loss": 1.2602,
+      "step": 10770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.717078968058069e-06,
+      "loss": 1.2872,
+      "step": 10771
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.71685773316512e-06,
+      "loss": 1.2945,
+      "step": 10772
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.716636498272172e-06,
+      "loss": 1.2049,
+      "step": 10773
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.716415263379225e-06,
+      "loss": 1.2353,
+      "step": 10774
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.716194028486278e-06,
+      "loss": 1.2771,
+      "step": 10775
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.71597279359333e-06,
+      "loss": 1.342,
+      "step": 10776
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.715751558700383e-06,
+      "loss": 1.1504,
+      "step": 10777
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.715530323807434e-06,
+      "loss": 1.2692,
+      "step": 10778
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.715309088914486e-06,
+      "loss": 1.3221,
+      "step": 10779
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.715087854021537e-06,
+      "loss": 1.21,
+      "step": 10780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.71486661912859e-06,
+      "loss": 1.2331,
+      "step": 10781
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.714645384235642e-06,
+      "loss": 1.1362,
+      "step": 10782
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.714424149342693e-06,
+      "loss": 1.2789,
+      "step": 10783
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.714202914449746e-06,
+      "loss": 1.317,
+      "step": 10784
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.713981679556799e-06,
+      "loss": 1.2187,
+      "step": 10785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.713760444663851e-06,
+      "loss": 1.2621,
+      "step": 10786
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.713539209770902e-06,
+      "loss": 1.2306,
+      "step": 10787
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.713317974877955e-06,
+      "loss": 1.2023,
+      "step": 10788
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.713096739985006e-06,
+      "loss": 1.3239,
+      "step": 10789
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.712875505092058e-06,
+      "loss": 1.2011,
+      "step": 10790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.712654270199111e-06,
+      "loss": 1.1447,
+      "step": 10791
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.712433035306163e-06,
+      "loss": 1.2801,
+      "step": 10792
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.712211800413216e-06,
+      "loss": 1.2568,
+      "step": 10793
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.711990565520267e-06,
+      "loss": 1.2376,
+      "step": 10794
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.71176933062732e-06,
+      "loss": 1.2505,
+      "step": 10795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.71154809573437e-06,
+      "loss": 1.1658,
+      "step": 10796
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.711326860841423e-06,
+      "loss": 1.1908,
+      "step": 10797
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.711105625948476e-06,
+      "loss": 1.1685,
+      "step": 10798
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.710884391055528e-06,
+      "loss": 1.3398,
+      "step": 10799
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.71066315616258e-06,
+      "loss": 1.2702,
+      "step": 10800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.710441921269632e-06,
+      "loss": 1.253,
+      "step": 10801
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.710220686376685e-06,
+      "loss": 1.27,
+      "step": 10802
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.709999451483735e-06,
+      "loss": 1.2697,
+      "step": 10803
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.709778216590788e-06,
+      "loss": 1.3741,
+      "step": 10804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.709556981697839e-06,
+      "loss": 1.3052,
+      "step": 10805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.709335746804892e-06,
+      "loss": 1.3667,
+      "step": 10806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.709114511911944e-06,
+      "loss": 1.2101,
+      "step": 10807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.708893277018997e-06,
+      "loss": 1.2062,
+      "step": 10808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.70867204212605e-06,
+      "loss": 1.2336,
+      "step": 10809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.708450807233102e-06,
+      "loss": 1.1692,
+      "step": 10810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.708229572340153e-06,
+      "loss": 1.2331,
+      "step": 10811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.708008337447204e-06,
+      "loss": 1.2762,
+      "step": 10812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.707787102554256e-06,
+      "loss": 1.2354,
+      "step": 10813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.707565867661309e-06,
+      "loss": 1.2476,
+      "step": 10814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.707344632768362e-06,
+      "loss": 1.2711,
+      "step": 10815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.707123397875413e-06,
+      "loss": 1.2993,
+      "step": 10816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.706902162982465e-06,
+      "loss": 1.2104,
+      "step": 10817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.706680928089518e-06,
+      "loss": 1.1915,
+      "step": 10818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.706459693196569e-06,
+      "loss": 1.277,
+      "step": 10819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.706238458303621e-06,
+      "loss": 1.1532,
+      "step": 10820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.706017223410672e-06,
+      "loss": 1.23,
+      "step": 10821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.705795988517725e-06,
+      "loss": 1.093,
+      "step": 10822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.705574753624778e-06,
+      "loss": 1.1876,
+      "step": 10823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.70535351873183e-06,
+      "loss": 1.4158,
+      "step": 10824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.705132283838883e-06,
+      "loss": 1.1957,
+      "step": 10825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.704911048945935e-06,
+      "loss": 1.228,
+      "step": 10826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.704689814052986e-06,
+      "loss": 1.2608,
+      "step": 10827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.704468579160037e-06,
+      "loss": 1.1651,
+      "step": 10828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.70424734426709e-06,
+      "loss": 1.2394,
+      "step": 10829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.704026109374142e-06,
+      "loss": 1.2022,
+      "step": 10830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.703804874481195e-06,
+      "loss": 1.2334,
+      "step": 10831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.703583639588246e-06,
+      "loss": 1.2705,
+      "step": 10832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.703362404695299e-06,
+      "loss": 1.1949,
+      "step": 10833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.703141169802351e-06,
+      "loss": 1.3072,
+      "step": 10834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.702919934909404e-06,
+      "loss": 1.2668,
+      "step": 10835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.702698700016455e-06,
+      "loss": 1.2704,
+      "step": 10836
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.702477465123507e-06,
+      "loss": 1.4273,
+      "step": 10837
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.702256230230558e-06,
+      "loss": 1.4807,
+      "step": 10838
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.70203499533761e-06,
+      "loss": 1.2507,
+      "step": 10839
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.701813760444663e-06,
+      "loss": 1.3687,
+      "step": 10840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.701592525551716e-06,
+      "loss": 1.3051,
+      "step": 10841
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.701371290658769e-06,
+      "loss": 1.2557,
+      "step": 10842
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.70115005576582e-06,
+      "loss": 1.321,
+      "step": 10843
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.70092882087287e-06,
+      "loss": 1.3124,
+      "step": 10844
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.700707585979923e-06,
+      "loss": 1.345,
+      "step": 10845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.700486351086976e-06,
+      "loss": 1.3172,
+      "step": 10846
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.700265116194028e-06,
+      "loss": 1.3506,
+      "step": 10847
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.700043881301081e-06,
+      "loss": 1.2528,
+      "step": 10848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.699822646408132e-06,
+      "loss": 1.3024,
+      "step": 10849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.699601411515184e-06,
+      "loss": 1.3445,
+      "step": 10850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.699380176622237e-06,
+      "loss": 1.2851,
+      "step": 10851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.699158941729288e-06,
+      "loss": 1.262,
+      "step": 10852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.69893770683634e-06,
+      "loss": 1.2223,
+      "step": 10853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.698716471943392e-06,
+      "loss": 1.3738,
+      "step": 10854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.698495237050444e-06,
+      "loss": 1.2467,
+      "step": 10855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.698274002157497e-06,
+      "loss": 1.2977,
+      "step": 10856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.69805276726455e-06,
+      "loss": 1.2442,
+      "step": 10857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.697831532371602e-06,
+      "loss": 1.1779,
+      "step": 10858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.697610297478653e-06,
+      "loss": 1.14,
+      "step": 10859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.697389062585704e-06,
+      "loss": 1.315,
+      "step": 10860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.697167827692756e-06,
+      "loss": 1.3351,
+      "step": 10861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.696946592799809e-06,
+      "loss": 1.3042,
+      "step": 10862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.696725357906862e-06,
+      "loss": 1.2648,
+      "step": 10863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.696504123013914e-06,
+      "loss": 1.272,
+      "step": 10864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.696282888120965e-06,
+      "loss": 1.3828,
+      "step": 10865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.696061653228018e-06,
+      "loss": 1.3279,
+      "step": 10866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.69584041833507e-06,
+      "loss": 1.3176,
+      "step": 10867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.695619183442121e-06,
+      "loss": 1.3085,
+      "step": 10868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.695397948549174e-06,
+      "loss": 1.2651,
+      "step": 10869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.695176713656227e-06,
+      "loss": 1.3364,
+      "step": 10870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.694955478763277e-06,
+      "loss": 1.2246,
+      "step": 10871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.69473424387033e-06,
+      "loss": 1.4744,
+      "step": 10872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.694513008977383e-06,
+      "loss": 1.2395,
+      "step": 10873
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.694291774084435e-06,
+      "loss": 1.3366,
+      "step": 10874
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.694070539191486e-06,
+      "loss": 1.1769,
+      "step": 10875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.693849304298539e-06,
+      "loss": 1.4041,
+      "step": 10876
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.69362806940559e-06,
+      "loss": 1.2804,
+      "step": 10877
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.693406834512642e-06,
+      "loss": 1.2796,
+      "step": 10878
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.693185599619695e-06,
+      "loss": 1.1708,
+      "step": 10879
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.692964364726748e-06,
+      "loss": 1.276,
+      "step": 10880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.6927431298338e-06,
+      "loss": 1.2391,
+      "step": 10881
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.692521894940851e-06,
+      "loss": 1.2746,
+      "step": 10882
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.692300660047904e-06,
+      "loss": 1.2944,
+      "step": 10883
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.692079425154955e-06,
+      "loss": 1.3005,
+      "step": 10884
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.691858190262007e-06,
+      "loss": 1.2616,
+      "step": 10885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.69163695536906e-06,
+      "loss": 1.2642,
+      "step": 10886
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.69141572047611e-06,
+      "loss": 1.3514,
+      "step": 10887
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.691194485583163e-06,
+      "loss": 1.2435,
+      "step": 10888
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.690973250690216e-06,
+      "loss": 1.3696,
+      "step": 10889
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.690752015797269e-06,
+      "loss": 1.1931,
+      "step": 10890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.69053078090432e-06,
+      "loss": 1.2156,
+      "step": 10891
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.690309546011372e-06,
+      "loss": 1.4151,
+      "step": 10892
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.690088311118423e-06,
+      "loss": 1.3128,
+      "step": 10893
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.689867076225476e-06,
+      "loss": 1.2092,
+      "step": 10894
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.689645841332528e-06,
+      "loss": 1.2275,
+      "step": 10895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.689424606439581e-06,
+      "loss": 1.2912,
+      "step": 10896
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.689203371546634e-06,
+      "loss": 1.2471,
+      "step": 10897
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.688982136653684e-06,
+      "loss": 1.2566,
+      "step": 10898
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.688760901760737e-06,
+      "loss": 1.2831,
+      "step": 10899
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.688539666867788e-06,
+      "loss": 1.2033,
+      "step": 10900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.68831843197484e-06,
+      "loss": 1.2661,
+      "step": 10901
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.688097197081893e-06,
+      "loss": 1.3256,
+      "step": 10902
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.687875962188946e-06,
+      "loss": 1.2658,
+      "step": 10903
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.687654727295997e-06,
+      "loss": 1.2671,
+      "step": 10904
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.68743349240305e-06,
+      "loss": 1.2813,
+      "step": 10905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.687212257510102e-06,
+      "loss": 1.1804,
+      "step": 10906
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.686991022617153e-06,
+      "loss": 1.2335,
+      "step": 10907
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.686769787724205e-06,
+      "loss": 1.1769,
+      "step": 10908
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.686548552831256e-06,
+      "loss": 1.1967,
+      "step": 10909
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.686327317938309e-06,
+      "loss": 1.2908,
+      "step": 10910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.686106083045362e-06,
+      "loss": 1.2686,
+      "step": 10911
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.685884848152414e-06,
+      "loss": 1.1769,
+      "step": 10912
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.685663613259467e-06,
+      "loss": 1.2062,
+      "step": 10913
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.685442378366518e-06,
+      "loss": 1.153,
+      "step": 10914
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.68522114347357e-06,
+      "loss": 1.2567,
+      "step": 10915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.684999908580621e-06,
+      "loss": 1.1359,
+      "step": 10916
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.684778673687674e-06,
+      "loss": 1.2478,
+      "step": 10917
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.684557438794727e-06,
+      "loss": 1.3063,
+      "step": 10918
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.684336203901779e-06,
+      "loss": 1.2656,
+      "step": 10919
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.68411496900883e-06,
+      "loss": 1.2645,
+      "step": 10920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.683893734115883e-06,
+      "loss": 1.2459,
+      "step": 10921
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.683672499222935e-06,
+      "loss": 1.3486,
+      "step": 10922
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.683451264329988e-06,
+      "loss": 1.2345,
+      "step": 10923
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.683230029437039e-06,
+      "loss": 1.2695,
+      "step": 10924
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.68300879454409e-06,
+      "loss": 1.2413,
+      "step": 10925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.682787559651142e-06,
+      "loss": 1.3161,
+      "step": 10926
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.682566324758195e-06,
+      "loss": 1.2624,
+      "step": 10927
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.682345089865248e-06,
+      "loss": 1.2818,
+      "step": 10928
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.6821238549723e-06,
+      "loss": 1.2394,
+      "step": 10929
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.681902620079353e-06,
+      "loss": 1.3627,
+      "step": 10930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.681681385186404e-06,
+      "loss": 1.257,
+      "step": 10931
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.681460150293455e-06,
+      "loss": 1.2833,
+      "step": 10932
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.681238915400507e-06,
+      "loss": 1.3217,
+      "step": 10933
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.68101768050756e-06,
+      "loss": 1.1734,
+      "step": 10934
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.680796445614612e-06,
+      "loss": 1.32,
+      "step": 10935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.680575210721663e-06,
+      "loss": 1.1239,
+      "step": 10936
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.680353975828716e-06,
+      "loss": 1.3322,
+      "step": 10937
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.680132740935769e-06,
+      "loss": 1.1967,
+      "step": 10938
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.679911506042821e-06,
+      "loss": 1.2589,
+      "step": 10939
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.679690271149872e-06,
+      "loss": 1.183,
+      "step": 10940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.679469036256925e-06,
+      "loss": 1.3008,
+      "step": 10941
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.679247801363976e-06,
+      "loss": 1.2673,
+      "step": 10942
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.679026566471028e-06,
+      "loss": 1.278,
+      "step": 10943
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.678805331578081e-06,
+      "loss": 1.2375,
+      "step": 10944
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.678584096685133e-06,
+      "loss": 1.1234,
+      "step": 10945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.678362861792186e-06,
+      "loss": 1.3246,
+      "step": 10946
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.678141626899237e-06,
+      "loss": 1.2651,
+      "step": 10947
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.677920392006288e-06,
+      "loss": 1.2319,
+      "step": 10948
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.67769915711334e-06,
+      "loss": 1.1709,
+      "step": 10949
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.677477922220393e-06,
+      "loss": 1.2666,
+      "step": 10950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.677256687327446e-06,
+      "loss": 1.3043,
+      "step": 10951
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.677035452434498e-06,
+      "loss": 1.2913,
+      "step": 10952
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.67681421754155e-06,
+      "loss": 1.2443,
+      "step": 10953
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.676592982648602e-06,
+      "loss": 1.225,
+      "step": 10954
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.676371747755655e-06,
+      "loss": 1.3032,
+      "step": 10955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.676150512862705e-06,
+      "loss": 1.3319,
+      "step": 10956
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.675929277969758e-06,
+      "loss": 1.2085,
+      "step": 10957
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.675708043076809e-06,
+      "loss": 1.2961,
+      "step": 10958
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.675486808183862e-06,
+      "loss": 1.171,
+      "step": 10959
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.675265573290914e-06,
+      "loss": 1.244,
+      "step": 10960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.675044338397967e-06,
+      "loss": 1.3246,
+      "step": 10961
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.67482310350502e-06,
+      "loss": 1.2335,
+      "step": 10962
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.67460186861207e-06,
+      "loss": 1.2267,
+      "step": 10963
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.674380633719123e-06,
+      "loss": 1.2054,
+      "step": 10964
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.674159398826174e-06,
+      "loss": 1.3252,
+      "step": 10965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.673938163933226e-06,
+      "loss": 1.1796,
+      "step": 10966
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.673716929040279e-06,
+      "loss": 1.2387,
+      "step": 10967
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.673495694147332e-06,
+      "loss": 1.3146,
+      "step": 10968
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.673274459254383e-06,
+      "loss": 1.339,
+      "step": 10969
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.673053224361435e-06,
+      "loss": 1.3004,
+      "step": 10970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.672831989468488e-06,
+      "loss": 1.3637,
+      "step": 10971
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.672610754575539e-06,
+      "loss": 1.1971,
+      "step": 10972
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.672389519682591e-06,
+      "loss": 1.3147,
+      "step": 10973
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.672168284789644e-06,
+      "loss": 1.1976,
+      "step": 10974
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.671947049896695e-06,
+      "loss": 1.299,
+      "step": 10975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.671725815003748e-06,
+      "loss": 1.362,
+      "step": 10976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.6715045801108e-06,
+      "loss": 1.2897,
+      "step": 10977
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.671283345217853e-06,
+      "loss": 1.2976,
+      "step": 10978
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.671062110324904e-06,
+      "loss": 1.2491,
+      "step": 10979
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.670840875431956e-06,
+      "loss": 1.3086,
+      "step": 10980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.670619640539007e-06,
+      "loss": 1.3181,
+      "step": 10981
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.67039840564606e-06,
+      "loss": 1.3031,
+      "step": 10982
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.670177170753112e-06,
+      "loss": 1.3024,
+      "step": 10983
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.669955935860165e-06,
+      "loss": 1.2768,
+      "step": 10984
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.669734700967218e-06,
+      "loss": 1.3106,
+      "step": 10985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.669513466074269e-06,
+      "loss": 1.2171,
+      "step": 10986
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.669292231181321e-06,
+      "loss": 1.2193,
+      "step": 10987
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.669070996288372e-06,
+      "loss": 1.2171,
+      "step": 10988
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.668849761395425e-06,
+      "loss": 1.3734,
+      "step": 10989
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.668628526502477e-06,
+      "loss": 1.3173,
+      "step": 10990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.668407291609528e-06,
+      "loss": 1.1375,
+      "step": 10991
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.66818605671658e-06,
+      "loss": 1.2887,
+      "step": 10992
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.667964821823633e-06,
+      "loss": 1.2462,
+      "step": 10993
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.667743586930686e-06,
+      "loss": 1.3317,
+      "step": 10994
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.667522352037737e-06,
+      "loss": 1.342,
+      "step": 10995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.66730111714479e-06,
+      "loss": 1.2172,
+      "step": 10996
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.66707988225184e-06,
+      "loss": 1.2825,
+      "step": 10997
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.666858647358893e-06,
+      "loss": 1.2051,
+      "step": 10998
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.666637412465946e-06,
+      "loss": 1.3062,
+      "step": 10999
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.666416177572998e-06,
+      "loss": 1.2031,
+      "step": 11000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.666194942680051e-06,
+      "loss": 1.2966,
+      "step": 11001
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.665973707787102e-06,
+      "loss": 1.277,
+      "step": 11002
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.665752472894154e-06,
+      "loss": 1.2481,
+      "step": 11003
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.665531238001205e-06,
+      "loss": 1.3164,
+      "step": 11004
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.665310003108258e-06,
+      "loss": 1.2593,
+      "step": 11005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.66508876821531e-06,
+      "loss": 1.3215,
+      "step": 11006
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.664867533322363e-06,
+      "loss": 1.2298,
+      "step": 11007
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.664646298429414e-06,
+      "loss": 1.3376,
+      "step": 11008
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.664425063536467e-06,
+      "loss": 1.1784,
+      "step": 11009
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.66420382864352e-06,
+      "loss": 1.2515,
+      "step": 11010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.663982593750572e-06,
+      "loss": 1.3046,
+      "step": 11011
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.663761358857623e-06,
+      "loss": 1.2433,
+      "step": 11012
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.663540123964674e-06,
+      "loss": 1.3892,
+      "step": 11013
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.663318889071726e-06,
+      "loss": 1.2273,
+      "step": 11014
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.663097654178779e-06,
+      "loss": 1.2557,
+      "step": 11015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.662876419285832e-06,
+      "loss": 1.3223,
+      "step": 11016
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.662655184392884e-06,
+      "loss": 1.2353,
+      "step": 11017
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.662433949499935e-06,
+      "loss": 1.3839,
+      "step": 11018
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.662212714606988e-06,
+      "loss": 1.315,
+      "step": 11019
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.661991479714039e-06,
+      "loss": 1.1613,
+      "step": 11020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.661770244821091e-06,
+      "loss": 1.3709,
+      "step": 11021
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.661549009928144e-06,
+      "loss": 1.2316,
+      "step": 11022
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.661327775035197e-06,
+      "loss": 1.2253,
+      "step": 11023
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.661106540142247e-06,
+      "loss": 1.2804,
+      "step": 11024
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.6608853052493e-06,
+      "loss": 1.2027,
+      "step": 11025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.660664070356353e-06,
+      "loss": 1.3246,
+      "step": 11026
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.660442835463405e-06,
+      "loss": 1.3226,
+      "step": 11027
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.660221600570456e-06,
+      "loss": 1.2525,
+      "step": 11028
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.660000365677507e-06,
+      "loss": 1.2271,
+      "step": 11029
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.65977913078456e-06,
+      "loss": 1.2532,
+      "step": 11030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.659557895891612e-06,
+      "loss": 1.3129,
+      "step": 11031
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.659336660998665e-06,
+      "loss": 1.2512,
+      "step": 11032
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.659115426105718e-06,
+      "loss": 1.2486,
+      "step": 11033
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.65889419121277e-06,
+      "loss": 1.1969,
+      "step": 11034
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.658672956319821e-06,
+      "loss": 1.2146,
+      "step": 11035
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.658451721426872e-06,
+      "loss": 1.2387,
+      "step": 11036
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.658230486533925e-06,
+      "loss": 1.4362,
+      "step": 11037
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.658009251640977e-06,
+      "loss": 1.224,
+      "step": 11038
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.65778801674803e-06,
+      "loss": 1.252,
+      "step": 11039
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.65756678185508e-06,
+      "loss": 1.3019,
+      "step": 11040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.657345546962133e-06,
+      "loss": 1.2327,
+      "step": 11041
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.657124312069186e-06,
+      "loss": 1.1938,
+      "step": 11042
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.656903077176239e-06,
+      "loss": 1.2231,
+      "step": 11043
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.65668184228329e-06,
+      "loss": 1.204,
+      "step": 11044
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.656460607390342e-06,
+      "loss": 1.1515,
+      "step": 11045
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.656239372497393e-06,
+      "loss": 1.283,
+      "step": 11046
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.656018137604446e-06,
+      "loss": 1.2095,
+      "step": 11047
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.655796902711498e-06,
+      "loss": 1.3341,
+      "step": 11048
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.655575667818551e-06,
+      "loss": 1.3035,
+      "step": 11049
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.655354432925604e-06,
+      "loss": 1.2903,
+      "step": 11050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.655133198032654e-06,
+      "loss": 1.2103,
+      "step": 11051
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.654911963139707e-06,
+      "loss": 1.3353,
+      "step": 11052
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.654690728246758e-06,
+      "loss": 1.3929,
+      "step": 11053
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.65446949335381e-06,
+      "loss": 1.2942,
+      "step": 11054
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.654248258460863e-06,
+      "loss": 1.1666,
+      "step": 11055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.654027023567916e-06,
+      "loss": 1.2809,
+      "step": 11056
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.653805788674967e-06,
+      "loss": 1.2259,
+      "step": 11057
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.65358455378202e-06,
+      "loss": 1.2364,
+      "step": 11058
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.653363318889072e-06,
+      "loss": 1.2268,
+      "step": 11059
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.653142083996123e-06,
+      "loss": 1.1407,
+      "step": 11060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.652920849103175e-06,
+      "loss": 1.2032,
+      "step": 11061
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.652699614210226e-06,
+      "loss": 1.4117,
+      "step": 11062
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.652478379317279e-06,
+      "loss": 1.1333,
+      "step": 11063
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.652257144424332e-06,
+      "loss": 1.1957,
+      "step": 11064
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.652035909531384e-06,
+      "loss": 1.21,
+      "step": 11065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.651814674638437e-06,
+      "loss": 1.2528,
+      "step": 11066
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.651593439745488e-06,
+      "loss": 1.3194,
+      "step": 11067
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.65137220485254e-06,
+      "loss": 1.313,
+      "step": 11068
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.651150969959591e-06,
+      "loss": 1.281,
+      "step": 11069
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.650929735066644e-06,
+      "loss": 1.2736,
+      "step": 11070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.650708500173697e-06,
+      "loss": 1.1833,
+      "step": 11071
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.650487265280749e-06,
+      "loss": 1.1721,
+      "step": 11072
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.6502660303878e-06,
+      "loss": 1.2359,
+      "step": 11073
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.650044795494853e-06,
+      "loss": 1.1727,
+      "step": 11074
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.649823560601905e-06,
+      "loss": 1.229,
+      "step": 11075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.649602325708956e-06,
+      "loss": 1.2138,
+      "step": 11076
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.649381090816009e-06,
+      "loss": 1.1713,
+      "step": 11077
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.649159855923061e-06,
+      "loss": 1.2412,
+      "step": 11078
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.648938621030112e-06,
+      "loss": 1.3892,
+      "step": 11079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.648717386137165e-06,
+      "loss": 1.1862,
+      "step": 11080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.648496151244218e-06,
+      "loss": 1.2798,
+      "step": 11081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.64827491635127e-06,
+      "loss": 1.2329,
+      "step": 11082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.648053681458321e-06,
+      "loss": 1.3818,
+      "step": 11083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.647832446565374e-06,
+      "loss": 1.1688,
+      "step": 11084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.647611211672425e-06,
+      "loss": 1.2417,
+      "step": 11085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.647389976779477e-06,
+      "loss": 1.3004,
+      "step": 11086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.64716874188653e-06,
+      "loss": 1.1806,
+      "step": 11087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.646947506993582e-06,
+      "loss": 1.4342,
+      "step": 11088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.646726272100635e-06,
+      "loss": 1.3178,
+      "step": 11089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.646505037207686e-06,
+      "loss": 1.3095,
+      "step": 11090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.646283802314739e-06,
+      "loss": 1.1866,
+      "step": 11091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.64606256742179e-06,
+      "loss": 1.2632,
+      "step": 11092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.645841332528842e-06,
+      "loss": 1.1789,
+      "step": 11093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.645620097635895e-06,
+      "loss": 1.2883,
+      "step": 11094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.645398862742946e-06,
+      "loss": 1.1901,
+      "step": 11095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.645177627849998e-06,
+      "loss": 1.1734,
+      "step": 11096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.644956392957051e-06,
+      "loss": 1.1058,
+      "step": 11097
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.644735158064103e-06,
+      "loss": 1.3247,
+      "step": 11098
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.644513923171156e-06,
+      "loss": 1.2684,
+      "step": 11099
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.644292688278207e-06,
+      "loss": 1.2431,
+      "step": 11100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.644071453385258e-06,
+      "loss": 1.1601,
+      "step": 11101
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.64385021849231e-06,
+      "loss": 1.1746,
+      "step": 11102
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.643628983599363e-06,
+      "loss": 1.2816,
+      "step": 11103
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.643407748706416e-06,
+      "loss": 1.3619,
+      "step": 11104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.643186513813468e-06,
+      "loss": 1.4235,
+      "step": 11105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.64296527892052e-06,
+      "loss": 1.2706,
+      "step": 11106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.642744044027572e-06,
+      "loss": 1.1553,
+      "step": 11107
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.642522809134623e-06,
+      "loss": 1.1568,
+      "step": 11108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.642301574241675e-06,
+      "loss": 1.2778,
+      "step": 11109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.642080339348728e-06,
+      "loss": 1.2133,
+      "step": 11110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.641859104455779e-06,
+      "loss": 1.2288,
+      "step": 11111
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.641637869562832e-06,
+      "loss": 1.193,
+      "step": 11112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.641416634669884e-06,
+      "loss": 1.2413,
+      "step": 11113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.641195399776937e-06,
+      "loss": 1.3164,
+      "step": 11114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.64097416488399e-06,
+      "loss": 1.2046,
+      "step": 11115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.64075292999104e-06,
+      "loss": 1.3364,
+      "step": 11116
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.640531695098091e-06,
+      "loss": 1.2671,
+      "step": 11117
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.640310460205144e-06,
+      "loss": 1.1988,
+      "step": 11118
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.640089225312196e-06,
+      "loss": 1.2992,
+      "step": 11119
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.639867990419249e-06,
+      "loss": 1.2343,
+      "step": 11120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.639646755526302e-06,
+      "loss": 1.2558,
+      "step": 11121
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.639425520633353e-06,
+      "loss": 1.4931,
+      "step": 11122
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.639204285740405e-06,
+      "loss": 1.2188,
+      "step": 11123
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.638983050847456e-06,
+      "loss": 1.1381,
+      "step": 11124
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.638761815954509e-06,
+      "loss": 1.1214,
+      "step": 11125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.638540581061561e-06,
+      "loss": 1.1742,
+      "step": 11126
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.638319346168614e-06,
+      "loss": 1.2131,
+      "step": 11127
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.638098111275665e-06,
+      "loss": 1.2844,
+      "step": 11128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.637876876382717e-06,
+      "loss": 1.2279,
+      "step": 11129
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.63765564148977e-06,
+      "loss": 1.258,
+      "step": 11130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.637434406596823e-06,
+      "loss": 1.3612,
+      "step": 11131
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.637213171703874e-06,
+      "loss": 1.1926,
+      "step": 11132
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.636991936810925e-06,
+      "loss": 1.2657,
+      "step": 11133
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.636770701917977e-06,
+      "loss": 1.2548,
+      "step": 11134
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.63654946702503e-06,
+      "loss": 1.2595,
+      "step": 11135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.636328232132082e-06,
+      "loss": 1.2937,
+      "step": 11136
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.636106997239135e-06,
+      "loss": 1.2582,
+      "step": 11137
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.635885762346188e-06,
+      "loss": 1.2125,
+      "step": 11138
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.635664527453239e-06,
+      "loss": 1.213,
+      "step": 11139
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.635443292560291e-06,
+      "loss": 1.2329,
+      "step": 11140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.635222057667342e-06,
+      "loss": 1.2277,
+      "step": 11141
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.635000822774395e-06,
+      "loss": 1.23,
+      "step": 11142
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.634779587881447e-06,
+      "loss": 1.2049,
+      "step": 11143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.634558352988498e-06,
+      "loss": 1.2663,
+      "step": 11144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.63433711809555e-06,
+      "loss": 1.2187,
+      "step": 11145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.634115883202603e-06,
+      "loss": 1.2323,
+      "step": 11146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.633894648309656e-06,
+      "loss": 1.1983,
+      "step": 11147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.633673413416707e-06,
+      "loss": 1.2342,
+      "step": 11148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.63345217852376e-06,
+      "loss": 1.2958,
+      "step": 11149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.63323094363081e-06,
+      "loss": 1.1683,
+      "step": 11150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.633009708737863e-06,
+      "loss": 1.227,
+      "step": 11151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.632788473844916e-06,
+      "loss": 1.2229,
+      "step": 11152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.632567238951968e-06,
+      "loss": 1.2254,
+      "step": 11153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.632346004059021e-06,
+      "loss": 1.1385,
+      "step": 11154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.632124769166072e-06,
+      "loss": 1.1983,
+      "step": 11155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.631903534273124e-06,
+      "loss": 1.3336,
+      "step": 11156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.631682299380175e-06,
+      "loss": 1.229,
+      "step": 11157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.631461064487228e-06,
+      "loss": 1.1646,
+      "step": 11158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.63123982959428e-06,
+      "loss": 1.2874,
+      "step": 11159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.631018594701333e-06,
+      "loss": 1.2765,
+      "step": 11160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.630797359808384e-06,
+      "loss": 1.2035,
+      "step": 11161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.630576124915437e-06,
+      "loss": 1.38,
+      "step": 11162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.63035489002249e-06,
+      "loss": 1.2364,
+      "step": 11163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.63013365512954e-06,
+      "loss": 1.3376,
+      "step": 11164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.629912420236593e-06,
+      "loss": 1.2247,
+      "step": 11165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.629691185343644e-06,
+      "loss": 1.2846,
+      "step": 11166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.629469950450696e-06,
+      "loss": 1.1655,
+      "step": 11167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.629248715557749e-06,
+      "loss": 1.1923,
+      "step": 11168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.629027480664802e-06,
+      "loss": 1.2886,
+      "step": 11169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.628806245771854e-06,
+      "loss": 1.2814,
+      "step": 11170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.628585010878905e-06,
+      "loss": 1.3933,
+      "step": 11171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.628363775985958e-06,
+      "loss": 1.3005,
+      "step": 11172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.628142541093009e-06,
+      "loss": 1.3117,
+      "step": 11173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.627921306200061e-06,
+      "loss": 1.2309,
+      "step": 11174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.627700071307114e-06,
+      "loss": 1.3907,
+      "step": 11175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.627478836414167e-06,
+      "loss": 1.1801,
+      "step": 11176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.627257601521217e-06,
+      "loss": 1.2479,
+      "step": 11177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.62703636662827e-06,
+      "loss": 1.2086,
+      "step": 11178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.626815131735323e-06,
+      "loss": 1.3012,
+      "step": 11179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.626593896842374e-06,
+      "loss": 1.248,
+      "step": 11180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.626372661949426e-06,
+      "loss": 1.2451,
+      "step": 11181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.626151427056479e-06,
+      "loss": 1.253,
+      "step": 11182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.62593019216353e-06,
+      "loss": 1.2922,
+      "step": 11183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.625708957270582e-06,
+      "loss": 1.2372,
+      "step": 11184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.625487722377635e-06,
+      "loss": 1.3358,
+      "step": 11185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.625266487484688e-06,
+      "loss": 1.287,
+      "step": 11186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.62504525259174e-06,
+      "loss": 1.2168,
+      "step": 11187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.624824017698791e-06,
+      "loss": 1.2872,
+      "step": 11188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.624602782805842e-06,
+      "loss": 1.2062,
+      "step": 11189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.624381547912895e-06,
+      "loss": 1.2554,
+      "step": 11190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.624160313019947e-06,
+      "loss": 1.1835,
+      "step": 11191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.623939078127e-06,
+      "loss": 1.2197,
+      "step": 11192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.623717843234052e-06,
+      "loss": 1.2283,
+      "step": 11193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.623496608341103e-06,
+      "loss": 1.1806,
+      "step": 11194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.623275373448156e-06,
+      "loss": 1.3332,
+      "step": 11195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.623054138555207e-06,
+      "loss": 1.1678,
+      "step": 11196
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.62283290366226e-06,
+      "loss": 1.2899,
+      "step": 11197
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.622611668769312e-06,
+      "loss": 1.2072,
+      "step": 11198
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.622390433876363e-06,
+      "loss": 1.1945,
+      "step": 11199
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.622169198983416e-06,
+      "loss": 1.153,
+      "step": 11200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.621947964090468e-06,
+      "loss": 1.2896,
+      "step": 11201
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.621726729197521e-06,
+      "loss": 1.2168,
+      "step": 11202
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.621505494304574e-06,
+      "loss": 1.2365,
+      "step": 11203
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.621284259411624e-06,
+      "loss": 1.1975,
+      "step": 11204
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.621063024518675e-06,
+      "loss": 1.1351,
+      "step": 11205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.620841789625728e-06,
+      "loss": 1.271,
+      "step": 11206
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.62062055473278e-06,
+      "loss": 1.2558,
+      "step": 11207
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.620399319839833e-06,
+      "loss": 1.2502,
+      "step": 11208
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.620178084946886e-06,
+      "loss": 1.3073,
+      "step": 11209
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.619956850053937e-06,
+      "loss": 1.209,
+      "step": 11210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.61973561516099e-06,
+      "loss": 1.4029,
+      "step": 11211
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.61951438026804e-06,
+      "loss": 1.342,
+      "step": 11212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.619293145375093e-06,
+      "loss": 1.4123,
+      "step": 11213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.619071910482145e-06,
+      "loss": 1.273,
+      "step": 11214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.618850675589196e-06,
+      "loss": 1.1993,
+      "step": 11215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.618629440696249e-06,
+      "loss": 1.1925,
+      "step": 11216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.618408205803302e-06,
+      "loss": 1.362,
+      "step": 11217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.618186970910354e-06,
+      "loss": 1.2647,
+      "step": 11218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.617965736017407e-06,
+      "loss": 1.1952,
+      "step": 11219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.617744501124458e-06,
+      "loss": 1.1434,
+      "step": 11220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.617523266231509e-06,
+      "loss": 1.171,
+      "step": 11221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.617302031338561e-06,
+      "loss": 1.145,
+      "step": 11222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.617080796445614e-06,
+      "loss": 1.1694,
+      "step": 11223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.616859561552666e-06,
+      "loss": 1.2437,
+      "step": 11224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.616638326659719e-06,
+      "loss": 1.4238,
+      "step": 11225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.61641709176677e-06,
+      "loss": 1.2062,
+      "step": 11226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.616195856873823e-06,
+      "loss": 1.2274,
+      "step": 11227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.615974621980874e-06,
+      "loss": 1.2851,
+      "step": 11228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.615753387087926e-06,
+      "loss": 1.2447,
+      "step": 11229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.615532152194979e-06,
+      "loss": 1.3215,
+      "step": 11230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.615310917302031e-06,
+      "loss": 1.2639,
+      "step": 11231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.615089682409082e-06,
+      "loss": 1.2589,
+      "step": 11232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.614868447516135e-06,
+      "loss": 1.2025,
+      "step": 11233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.614647212623188e-06,
+      "loss": 1.3398,
+      "step": 11234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.61442597773024e-06,
+      "loss": 1.3492,
+      "step": 11235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.614204742837291e-06,
+      "loss": 1.2605,
+      "step": 11236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.613983507944342e-06,
+      "loss": 1.1845,
+      "step": 11237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.613762273051395e-06,
+      "loss": 1.3431,
+      "step": 11238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.613541038158447e-06,
+      "loss": 1.1984,
+      "step": 11239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.6133198032655e-06,
+      "loss": 1.2881,
+      "step": 11240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.613098568372552e-06,
+      "loss": 1.2648,
+      "step": 11241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.612877333479605e-06,
+      "loss": 1.2109,
+      "step": 11242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.612656098586656e-06,
+      "loss": 1.2415,
+      "step": 11243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.612434863693709e-06,
+      "loss": 1.1513,
+      "step": 11244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.61221362880076e-06,
+      "loss": 1.3241,
+      "step": 11245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.611992393907812e-06,
+      "loss": 1.1997,
+      "step": 11246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.611771159014865e-06,
+      "loss": 1.2318,
+      "step": 11247
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.611549924121916e-06,
+      "loss": 1.4123,
+      "step": 11248
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.611328689228968e-06,
+      "loss": 1.2669,
+      "step": 11249
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.61110745433602e-06,
+      "loss": 1.1825,
+      "step": 11250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.610886219443073e-06,
+      "loss": 1.3741,
+      "step": 11251
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.610664984550124e-06,
+      "loss": 1.2977,
+      "step": 11252
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.610443749657177e-06,
+      "loss": 1.2931,
+      "step": 11253
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.610222514764228e-06,
+      "loss": 1.2057,
+      "step": 11254
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.61000127987128e-06,
+      "loss": 1.1134,
+      "step": 11255
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.609780044978333e-06,
+      "loss": 1.2602,
+      "step": 11256
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.609558810085386e-06,
+      "loss": 1.1813,
+      "step": 11257
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.609337575192438e-06,
+      "loss": 1.3295,
+      "step": 11258
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.60911634029949e-06,
+      "loss": 1.2851,
+      "step": 11259
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.608895105406542e-06,
+      "loss": 1.144,
+      "step": 11260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.608673870513593e-06,
+      "loss": 1.2183,
+      "step": 11261
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.608452635620645e-06,
+      "loss": 1.1611,
+      "step": 11262
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.608231400727698e-06,
+      "loss": 1.1762,
+      "step": 11263
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.60801016583475e-06,
+      "loss": 1.1713,
+      "step": 11264
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.607788930941802e-06,
+      "loss": 1.1876,
+      "step": 11265
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.607567696048854e-06,
+      "loss": 1.3484,
+      "step": 11266
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.607346461155907e-06,
+      "loss": 1.2125,
+      "step": 11267
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.607125226262958e-06,
+      "loss": 1.2363,
+      "step": 11268
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.60690399137001e-06,
+      "loss": 1.2709,
+      "step": 11269
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.606682756477061e-06,
+      "loss": 1.1474,
+      "step": 11270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.606461521584114e-06,
+      "loss": 1.2616,
+      "step": 11271
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.606240286691166e-06,
+      "loss": 1.2809,
+      "step": 11272
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.606019051798219e-06,
+      "loss": 1.4087,
+      "step": 11273
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.605797816905272e-06,
+      "loss": 1.3304,
+      "step": 11274
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.605576582012324e-06,
+      "loss": 1.3415,
+      "step": 11275
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.605355347119375e-06,
+      "loss": 1.2405,
+      "step": 11276
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.605134112226426e-06,
+      "loss": 1.2201,
+      "step": 11277
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.604912877333479e-06,
+      "loss": 1.2136,
+      "step": 11278
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.604691642440531e-06,
+      "loss": 1.1077,
+      "step": 11279
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.604470407547584e-06,
+      "loss": 1.2031,
+      "step": 11280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.604249172654635e-06,
+      "loss": 1.1924,
+      "step": 11281
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.604027937761687e-06,
+      "loss": 1.2953,
+      "step": 11282
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.60380670286874e-06,
+      "loss": 1.3836,
+      "step": 11283
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.603585467975791e-06,
+      "loss": 1.1969,
+      "step": 11284
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.603364233082844e-06,
+      "loss": 1.2715,
+      "step": 11285
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.603142998189896e-06,
+      "loss": 1.2325,
+      "step": 11286
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.602921763296947e-06,
+      "loss": 1.2551,
+      "step": 11287
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.602700528404e-06,
+      "loss": 1.2438,
+      "step": 11288
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.602479293511052e-06,
+      "loss": 1.2146,
+      "step": 11289
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.602258058618105e-06,
+      "loss": 1.2465,
+      "step": 11290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.602036823725158e-06,
+      "loss": 1.3051,
+      "step": 11291
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.601815588832209e-06,
+      "loss": 1.2582,
+      "step": 11292
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.60159435393926e-06,
+      "loss": 1.3367,
+      "step": 11293
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.601373119046312e-06,
+      "loss": 1.2485,
+      "step": 11294
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.601151884153365e-06,
+      "loss": 1.3803,
+      "step": 11295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.600930649260417e-06,
+      "loss": 1.3156,
+      "step": 11296
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.60070941436747e-06,
+      "loss": 1.2691,
+      "step": 11297
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.60048817947452e-06,
+      "loss": 1.2914,
+      "step": 11298
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.600266944581573e-06,
+      "loss": 1.2357,
+      "step": 11299
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.600045709688624e-06,
+      "loss": 1.246,
+      "step": 11300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.599824474795677e-06,
+      "loss": 1.1323,
+      "step": 11301
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.59960323990273e-06,
+      "loss": 1.2058,
+      "step": 11302
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.59938200500978e-06,
+      "loss": 1.3014,
+      "step": 11303
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.599160770116833e-06,
+      "loss": 1.1407,
+      "step": 11304
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.598939535223886e-06,
+      "loss": 1.2287,
+      "step": 11305
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.598718300330938e-06,
+      "loss": 1.3049,
+      "step": 11306
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.598497065437991e-06,
+      "loss": 1.1941,
+      "step": 11307
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.598275830545042e-06,
+      "loss": 1.3101,
+      "step": 11308
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.598054595652093e-06,
+      "loss": 1.32,
+      "step": 11309
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.597833360759145e-06,
+      "loss": 1.3321,
+      "step": 11310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.597612125866198e-06,
+      "loss": 1.1045,
+      "step": 11311
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.59739089097325e-06,
+      "loss": 1.2115,
+      "step": 11312
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.597169656080303e-06,
+      "loss": 1.2558,
+      "step": 11313
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.596948421187354e-06,
+      "loss": 1.2543,
+      "step": 11314
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.596727186294407e-06,
+      "loss": 1.184,
+      "step": 11315
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.596505951401458e-06,
+      "loss": 1.1658,
+      "step": 11316
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.59628471650851e-06,
+      "loss": 1.2767,
+      "step": 11317
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.596063481615563e-06,
+      "loss": 1.2903,
+      "step": 11318
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.595842246722614e-06,
+      "loss": 1.2431,
+      "step": 11319
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.595621011829666e-06,
+      "loss": 1.4147,
+      "step": 11320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.595399776936719e-06,
+      "loss": 1.2532,
+      "step": 11321
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.595178542043772e-06,
+      "loss": 1.2106,
+      "step": 11322
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.594957307150824e-06,
+      "loss": 1.1643,
+      "step": 11323
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.594736072257875e-06,
+      "loss": 1.3713,
+      "step": 11324
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.594514837364926e-06,
+      "loss": 1.2537,
+      "step": 11325
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.594293602471979e-06,
+      "loss": 1.1943,
+      "step": 11326
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.594072367579031e-06,
+      "loss": 1.3402,
+      "step": 11327
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.593851132686084e-06,
+      "loss": 1.3329,
+      "step": 11328
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.593629897793137e-06,
+      "loss": 1.1926,
+      "step": 11329
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.593408662900187e-06,
+      "loss": 1.297,
+      "step": 11330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.59318742800724e-06,
+      "loss": 1.1265,
+      "step": 11331
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.592966193114293e-06,
+      "loss": 1.3015,
+      "step": 11332
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.592744958221344e-06,
+      "loss": 1.1916,
+      "step": 11333
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.592523723328396e-06,
+      "loss": 1.2697,
+      "step": 11334
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.592302488435449e-06,
+      "loss": 1.2306,
+      "step": 11335
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.5920812535425e-06,
+      "loss": 1.3622,
+      "step": 11336
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.591860018649552e-06,
+      "loss": 1.3499,
+      "step": 11337
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.591638783756605e-06,
+      "loss": 1.2938,
+      "step": 11338
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.591417548863658e-06,
+      "loss": 1.1544,
+      "step": 11339
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.591196313970708e-06,
+      "loss": 1.3564,
+      "step": 11340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.59097507907776e-06,
+      "loss": 1.234,
+      "step": 11341
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.590753844184812e-06,
+      "loss": 1.3097,
+      "step": 11342
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.590532609291865e-06,
+      "loss": 1.2193,
+      "step": 11343
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.590311374398917e-06,
+      "loss": 1.3598,
+      "step": 11344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.59009013950597e-06,
+      "loss": 1.3278,
+      "step": 11345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.589868904613022e-06,
+      "loss": 1.3747,
+      "step": 11346
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.589647669720073e-06,
+      "loss": 1.1674,
+      "step": 11347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.589426434827126e-06,
+      "loss": 1.2034,
+      "step": 11348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.589205199934177e-06,
+      "loss": 1.2801,
+      "step": 11349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.58898396504123e-06,
+      "loss": 1.1487,
+      "step": 11350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.588762730148282e-06,
+      "loss": 1.1585,
+      "step": 11351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.588541495255333e-06,
+      "loss": 1.2117,
+      "step": 11352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.588320260362386e-06,
+      "loss": 1.1285,
+      "step": 11353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.588099025469438e-06,
+      "loss": 1.2492,
+      "step": 11354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.587877790576491e-06,
+      "loss": 1.245,
+      "step": 11355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.587656555683542e-06,
+      "loss": 1.3022,
+      "step": 11356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.587435320790594e-06,
+      "loss": 1.1623,
+      "step": 11357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.587214085897645e-06,
+      "loss": 1.3684,
+      "step": 11358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.586992851004698e-06,
+      "loss": 1.2779,
+      "step": 11359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.58677161611175e-06,
+      "loss": 1.2466,
+      "step": 11360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.586550381218803e-06,
+      "loss": 1.3199,
+      "step": 11361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.586329146325856e-06,
+      "loss": 1.2491,
+      "step": 11362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.586107911432907e-06,
+      "loss": 1.1743,
+      "step": 11363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.58588667653996e-06,
+      "loss": 1.2896,
+      "step": 11364
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.58566544164701e-06,
+      "loss": 1.136,
+      "step": 11365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.585444206754063e-06,
+      "loss": 1.3357,
+      "step": 11366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.585222971861115e-06,
+      "loss": 1.2386,
+      "step": 11367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.585001736968168e-06,
+      "loss": 1.2125,
+      "step": 11368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.584780502075219e-06,
+      "loss": 1.3019,
+      "step": 11369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.584559267182272e-06,
+      "loss": 1.2614,
+      "step": 11370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.584338032289324e-06,
+      "loss": 1.2568,
+      "step": 11371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.584116797396375e-06,
+      "loss": 1.3504,
+      "step": 11372
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.583895562503428e-06,
+      "loss": 1.268,
+      "step": 11373
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.583674327610479e-06,
+      "loss": 1.265,
+      "step": 11374
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.583453092717531e-06,
+      "loss": 1.2047,
+      "step": 11375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.583231857824584e-06,
+      "loss": 1.2452,
+      "step": 11376
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.583010622931636e-06,
+      "loss": 1.3366,
+      "step": 11377
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.582789388038689e-06,
+      "loss": 1.2638,
+      "step": 11378
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.582568153145742e-06,
+      "loss": 1.2127,
+      "step": 11379
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.582346918252793e-06,
+      "loss": 1.2522,
+      "step": 11380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.582125683359844e-06,
+      "loss": 1.3721,
+      "step": 11381
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.581904448466896e-06,
+      "loss": 1.0905,
+      "step": 11382
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.581683213573949e-06,
+      "loss": 1.1754,
+      "step": 11383
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.581461978681001e-06,
+      "loss": 1.2078,
+      "step": 11384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.581240743788052e-06,
+      "loss": 1.232,
+      "step": 11385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.581019508895105e-06,
+      "loss": 1.2566,
+      "step": 11386
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.580798274002158e-06,
+      "loss": 1.2596,
+      "step": 11387
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.580577039109208e-06,
+      "loss": 1.2702,
+      "step": 11388
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.580355804216261e-06,
+      "loss": 1.2077,
+      "step": 11389
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.580134569323314e-06,
+      "loss": 1.2441,
+      "step": 11390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.579913334430365e-06,
+      "loss": 1.2729,
+      "step": 11391
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.579692099537417e-06,
+      "loss": 1.3268,
+      "step": 11392
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.57947086464447e-06,
+      "loss": 1.1995,
+      "step": 11393
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.579249629751522e-06,
+      "loss": 1.1201,
+      "step": 11394
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.579028394858575e-06,
+      "loss": 1.3913,
+      "step": 11395
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.578807159965626e-06,
+      "loss": 1.339,
+      "step": 11396
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.578585925072677e-06,
+      "loss": 1.305,
+      "step": 11397
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.57836469017973e-06,
+      "loss": 1.3196,
+      "step": 11398
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.578143455286782e-06,
+      "loss": 1.3183,
+      "step": 11399
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.577922220393835e-06,
+      "loss": 1.3423,
+      "step": 11400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.577700985500886e-06,
+      "loss": 1.3405,
+      "step": 11401
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.577479750607938e-06,
+      "loss": 1.2018,
+      "step": 11402
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.57725851571499e-06,
+      "loss": 1.2059,
+      "step": 11403
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.577037280822042e-06,
+      "loss": 1.2565,
+      "step": 11404
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.576816045929094e-06,
+      "loss": 1.2741,
+      "step": 11405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.576594811036147e-06,
+      "loss": 1.3359,
+      "step": 11406
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.576373576143198e-06,
+      "loss": 1.225,
+      "step": 11407
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.57615234125025e-06,
+      "loss": 1.2957,
+      "step": 11408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.575931106357303e-06,
+      "loss": 1.2764,
+      "step": 11409
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.575709871464356e-06,
+      "loss": 1.3265,
+      "step": 11410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.575488636571408e-06,
+      "loss": 1.3554,
+      "step": 11411
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.57526740167846e-06,
+      "loss": 1.2319,
+      "step": 11412
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.57504616678551e-06,
+      "loss": 1.2557,
+      "step": 11413
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.574824931892563e-06,
+      "loss": 1.2283,
+      "step": 11414
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.574603696999615e-06,
+      "loss": 1.3414,
+      "step": 11415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.574382462106668e-06,
+      "loss": 1.2209,
+      "step": 11416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.57416122721372e-06,
+      "loss": 1.2179,
+      "step": 11417
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.573939992320772e-06,
+      "loss": 1.2027,
+      "step": 11418
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.573718757427824e-06,
+      "loss": 1.3066,
+      "step": 11419
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.573497522534877e-06,
+      "loss": 1.3118,
+      "step": 11420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.573276287641928e-06,
+      "loss": 1.2264,
+      "step": 11421
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.57305505274898e-06,
+      "loss": 1.311,
+      "step": 11422
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.572833817856031e-06,
+      "loss": 1.3696,
+      "step": 11423
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.572612582963084e-06,
+      "loss": 1.1773,
+      "step": 11424
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.572391348070136e-06,
+      "loss": 1.2285,
+      "step": 11425
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.572170113177189e-06,
+      "loss": 1.2966,
+      "step": 11426
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.571948878284242e-06,
+      "loss": 1.2935,
+      "step": 11427
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.571727643391293e-06,
+      "loss": 1.2696,
+      "step": 11428
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.571506408498343e-06,
+      "loss": 1.1861,
+      "step": 11429
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.571285173605396e-06,
+      "loss": 1.3997,
+      "step": 11430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.571063938712449e-06,
+      "loss": 1.3656,
+      "step": 11431
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.570842703819501e-06,
+      "loss": 1.1329,
+      "step": 11432
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.570621468926554e-06,
+      "loss": 1.3697,
+      "step": 11433
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.570400234033605e-06,
+      "loss": 1.2173,
+      "step": 11434
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.570178999140657e-06,
+      "loss": 1.2791,
+      "step": 11435
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.56995776424771e-06,
+      "loss": 1.1998,
+      "step": 11436
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.569736529354761e-06,
+      "loss": 1.3101,
+      "step": 11437
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.569515294461814e-06,
+      "loss": 1.2143,
+      "step": 11438
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.569294059568866e-06,
+      "loss": 1.2003,
+      "step": 11439
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.569072824675917e-06,
+      "loss": 1.1962,
+      "step": 11440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.56885158978297e-06,
+      "loss": 1.2564,
+      "step": 11441
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.568630354890022e-06,
+      "loss": 1.1312,
+      "step": 11442
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.568409119997075e-06,
+      "loss": 1.3255,
+      "step": 11443
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.568187885104126e-06,
+      "loss": 1.2472,
+      "step": 11444
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.567966650211177e-06,
+      "loss": 1.2215,
+      "step": 11445
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.56774541531823e-06,
+      "loss": 1.4166,
+      "step": 11446
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.567524180425282e-06,
+      "loss": 1.2374,
+      "step": 11447
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.567302945532335e-06,
+      "loss": 1.2837,
+      "step": 11448
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.567081710639387e-06,
+      "loss": 1.2772,
+      "step": 11449
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.56686047574644e-06,
+      "loss": 1.1828,
+      "step": 11450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.56663924085349e-06,
+      "loss": 1.169,
+      "step": 11451
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.566418005960543e-06,
+      "loss": 1.2039,
+      "step": 11452
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.566196771067594e-06,
+      "loss": 1.1404,
+      "step": 11453
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.565975536174647e-06,
+      "loss": 1.3901,
+      "step": 11454
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.5657543012817e-06,
+      "loss": 1.3167,
+      "step": 11455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.56553306638875e-06,
+      "loss": 1.2013,
+      "step": 11456
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.565311831495803e-06,
+      "loss": 1.2268,
+      "step": 11457
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.565090596602856e-06,
+      "loss": 1.4648,
+      "step": 11458
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.564869361709908e-06,
+      "loss": 1.3318,
+      "step": 11459
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.56464812681696e-06,
+      "loss": 1.4266,
+      "step": 11460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.564426891924012e-06,
+      "loss": 1.3032,
+      "step": 11461
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.564205657031063e-06,
+      "loss": 1.2451,
+      "step": 11462
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.563984422138115e-06,
+      "loss": 1.2074,
+      "step": 11463
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.563763187245168e-06,
+      "loss": 1.2864,
+      "step": 11464
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.56354195235222e-06,
+      "loss": 1.4516,
+      "step": 11465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.563320717459273e-06,
+      "loss": 1.2727,
+      "step": 11466
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.563099482566324e-06,
+      "loss": 1.3662,
+      "step": 11467
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.562878247673377e-06,
+      "loss": 1.2844,
+      "step": 11468
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.562657012780428e-06,
+      "loss": 1.257,
+      "step": 11469
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.56243577788748e-06,
+      "loss": 1.0505,
+      "step": 11470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.562214542994533e-06,
+      "loss": 1.2551,
+      "step": 11471
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.561993308101585e-06,
+      "loss": 1.3487,
+      "step": 11472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.561772073208636e-06,
+      "loss": 1.2931,
+      "step": 11473
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.561550838315689e-06,
+      "loss": 1.1876,
+      "step": 11474
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.561329603422742e-06,
+      "loss": 1.2642,
+      "step": 11475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.561108368529793e-06,
+      "loss": 1.3138,
+      "step": 11476
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.560887133636845e-06,
+      "loss": 1.1868,
+      "step": 11477
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.560665898743896e-06,
+      "loss": 1.3086,
+      "step": 11478
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.560444663850949e-06,
+      "loss": 1.3455,
+      "step": 11479
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.560223428958001e-06,
+      "loss": 1.3261,
+      "step": 11480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.560002194065054e-06,
+      "loss": 1.1958,
+      "step": 11481
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.559780959172107e-06,
+      "loss": 1.1545,
+      "step": 11482
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.559559724279159e-06,
+      "loss": 1.2105,
+      "step": 11483
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.55933848938621e-06,
+      "loss": 1.2605,
+      "step": 11484
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.559117254493261e-06,
+      "loss": 1.2031,
+      "step": 11485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.558896019600314e-06,
+      "loss": 1.2152,
+      "step": 11486
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.558674784707366e-06,
+      "loss": 1.2629,
+      "step": 11487
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.558453549814419e-06,
+      "loss": 1.1845,
+      "step": 11488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.55823231492147e-06,
+      "loss": 1.1796,
+      "step": 11489
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.558011080028522e-06,
+      "loss": 1.2644,
+      "step": 11490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.557789845135575e-06,
+      "loss": 1.2187,
+      "step": 11491
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.557568610242626e-06,
+      "loss": 1.2021,
+      "step": 11492
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.557347375349678e-06,
+      "loss": 1.2875,
+      "step": 11493
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.557126140456731e-06,
+      "loss": 1.2807,
+      "step": 11494
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.556904905563782e-06,
+      "loss": 1.2253,
+      "step": 11495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.556683670670835e-06,
+      "loss": 1.4337,
+      "step": 11496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.556462435777887e-06,
+      "loss": 1.3038,
+      "step": 11497
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.55624120088494e-06,
+      "loss": 1.2027,
+      "step": 11498
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.556019965991992e-06,
+      "loss": 1.1697,
+      "step": 11499
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.555798731099043e-06,
+      "loss": 1.177,
+      "step": 11500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.555577496206094e-06,
+      "loss": 1.214,
+      "step": 11501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.555356261313147e-06,
+      "loss": 1.1639,
+      "step": 11502
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.5551350264202e-06,
+      "loss": 1.1951,
+      "step": 11503
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.554913791527252e-06,
+      "loss": 1.2081,
+      "step": 11504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.554692556634303e-06,
+      "loss": 1.236,
+      "step": 11505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.554471321741356e-06,
+      "loss": 1.3843,
+      "step": 11506
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.554250086848408e-06,
+      "loss": 1.2598,
+      "step": 11507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.554028851955461e-06,
+      "loss": 1.4326,
+      "step": 11508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.553807617062512e-06,
+      "loss": 1.2421,
+      "step": 11509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.553586382169564e-06,
+      "loss": 1.2579,
+      "step": 11510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.553365147276615e-06,
+      "loss": 1.207,
+      "step": 11511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.553143912383668e-06,
+      "loss": 1.415,
+      "step": 11512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.55292267749072e-06,
+      "loss": 1.2857,
+      "step": 11513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.552701442597773e-06,
+      "loss": 1.2882,
+      "step": 11514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.552480207704826e-06,
+      "loss": 1.2129,
+      "step": 11515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.552258972811877e-06,
+      "loss": 1.2653,
+      "step": 11516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.552037737918928e-06,
+      "loss": 1.3015,
+      "step": 11517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.55181650302598e-06,
+      "loss": 1.2391,
+      "step": 11518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.551595268133033e-06,
+      "loss": 1.2112,
+      "step": 11519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.551374033240085e-06,
+      "loss": 1.2566,
+      "step": 11520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.551152798347138e-06,
+      "loss": 1.2666,
+      "step": 11521
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.550931563454189e-06,
+      "loss": 1.2255,
+      "step": 11522
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.550710328561242e-06,
+      "loss": 1.1856,
+      "step": 11523
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.550489093668294e-06,
+      "loss": 1.3016,
+      "step": 11524
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.550267858775345e-06,
+      "loss": 1.3073,
+      "step": 11525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.550046623882398e-06,
+      "loss": 1.2389,
+      "step": 11526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.549825388989449e-06,
+      "loss": 1.2173,
+      "step": 11527
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.549604154096501e-06,
+      "loss": 1.2396,
+      "step": 11528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.549382919203554e-06,
+      "loss": 1.2325,
+      "step": 11529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.549161684310606e-06,
+      "loss": 1.1507,
+      "step": 11530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.548940449417659e-06,
+      "loss": 1.2521,
+      "step": 11531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.54871921452471e-06,
+      "loss": 1.2631,
+      "step": 11532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.548497979631761e-06,
+      "loss": 1.1855,
+      "step": 11533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.548276744738814e-06,
+      "loss": 1.331,
+      "step": 11534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.548055509845866e-06,
+      "loss": 1.281,
+      "step": 11535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.547834274952919e-06,
+      "loss": 1.1852,
+      "step": 11536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.547613040059971e-06,
+      "loss": 1.2244,
+      "step": 11537
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.547391805167022e-06,
+      "loss": 1.4107,
+      "step": 11538
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.547170570274075e-06,
+      "loss": 1.3759,
+      "step": 11539
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.546949335381128e-06,
+      "loss": 1.197,
+      "step": 11540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.546728100488178e-06,
+      "loss": 1.2483,
+      "step": 11541
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.546506865595231e-06,
+      "loss": 1.2668,
+      "step": 11542
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.546285630702284e-06,
+      "loss": 1.2376,
+      "step": 11543
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.546064395809335e-06,
+      "loss": 1.3138,
+      "step": 11544
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.545843160916387e-06,
+      "loss": 1.2703,
+      "step": 11545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.54562192602344e-06,
+      "loss": 1.3144,
+      "step": 11546
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.545400691130492e-06,
+      "loss": 1.1055,
+      "step": 11547
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.545179456237543e-06,
+      "loss": 1.1867,
+      "step": 11548
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.544958221344596e-06,
+      "loss": 1.2906,
+      "step": 11549
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.544736986451647e-06,
+      "loss": 1.2725,
+      "step": 11550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.5445157515587e-06,
+      "loss": 1.1781,
+      "step": 11551
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.544294516665752e-06,
+      "loss": 1.2945,
+      "step": 11552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.544073281772805e-06,
+      "loss": 1.2444,
+      "step": 11553
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.543852046879857e-06,
+      "loss": 1.1322,
+      "step": 11554
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.543630811986908e-06,
+      "loss": 1.2218,
+      "step": 11555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.54340957709396e-06,
+      "loss": 1.2873,
+      "step": 11556
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.543188342201012e-06,
+      "loss": 1.2208,
+      "step": 11557
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.542967107308064e-06,
+      "loss": 1.1781,
+      "step": 11558
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.542745872415117e-06,
+      "loss": 1.2842,
+      "step": 11559
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.542524637522168e-06,
+      "loss": 1.3975,
+      "step": 11560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.54230340262922e-06,
+      "loss": 1.2307,
+      "step": 11561
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.542082167736273e-06,
+      "loss": 1.205,
+      "step": 11562
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.541860932843326e-06,
+      "loss": 1.2984,
+      "step": 11563
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.541639697950377e-06,
+      "loss": 1.2272,
+      "step": 11564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.54141846305743e-06,
+      "loss": 1.2589,
+      "step": 11565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.54119722816448e-06,
+      "loss": 1.331,
+      "step": 11566
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.540975993271533e-06,
+      "loss": 1.2837,
+      "step": 11567
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.540754758378585e-06,
+      "loss": 1.1601,
+      "step": 11568
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.540533523485638e-06,
+      "loss": 1.2785,
+      "step": 11569
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.54031228859269e-06,
+      "loss": 1.4115,
+      "step": 11570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.540091053699742e-06,
+      "loss": 1.2685,
+      "step": 11571
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.539869818806794e-06,
+      "loss": 1.3055,
+      "step": 11572
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.539648583913845e-06,
+      "loss": 1.2743,
+      "step": 11573
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.539427349020898e-06,
+      "loss": 1.1152,
+      "step": 11574
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.53920611412795e-06,
+      "loss": 1.2231,
+      "step": 11575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.538984879235003e-06,
+      "loss": 1.3119,
+      "step": 11576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.538763644342054e-06,
+      "loss": 1.21,
+      "step": 11577
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.538542409449106e-06,
+      "loss": 1.1635,
+      "step": 11578
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.538321174556159e-06,
+      "loss": 1.2602,
+      "step": 11579
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.53809993966321e-06,
+      "loss": 1.3038,
+      "step": 11580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.537878704770263e-06,
+      "loss": 1.3006,
+      "step": 11581
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.537657469877313e-06,
+      "loss": 1.2326,
+      "step": 11582
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.537436234984366e-06,
+      "loss": 1.2271,
+      "step": 11583
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.537215000091419e-06,
+      "loss": 1.2341,
+      "step": 11584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.536993765198471e-06,
+      "loss": 1.2724,
+      "step": 11585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.536772530305524e-06,
+      "loss": 1.244,
+      "step": 11586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.536551295412577e-06,
+      "loss": 1.3395,
+      "step": 11587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.536330060519627e-06,
+      "loss": 1.2526,
+      "step": 11588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.536108825626678e-06,
+      "loss": 1.2331,
+      "step": 11589
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.535887590733731e-06,
+      "loss": 1.2302,
+      "step": 11590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.535666355840784e-06,
+      "loss": 1.286,
+      "step": 11591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.535445120947836e-06,
+      "loss": 1.4882,
+      "step": 11592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.535223886054887e-06,
+      "loss": 1.1736,
+      "step": 11593
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.53500265116194e-06,
+      "loss": 1.1009,
+      "step": 11594
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.534781416268992e-06,
+      "loss": 1.4347,
+      "step": 11595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.534560181376045e-06,
+      "loss": 1.1214,
+      "step": 11596
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.534338946483096e-06,
+      "loss": 1.2045,
+      "step": 11597
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.534117711590147e-06,
+      "loss": 1.252,
+      "step": 11598
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.5338964766972e-06,
+      "loss": 1.3919,
+      "step": 11599
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.533675241804252e-06,
+      "loss": 1.3307,
+      "step": 11600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.533454006911305e-06,
+      "loss": 1.2346,
+      "step": 11601
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.533232772018357e-06,
+      "loss": 1.1838,
+      "step": 11602
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.53301153712541e-06,
+      "loss": 1.315,
+      "step": 11603
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.53279030223246e-06,
+      "loss": 1.2587,
+      "step": 11604
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.532569067339512e-06,
+      "loss": 1.3545,
+      "step": 11605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.532347832446564e-06,
+      "loss": 1.2449,
+      "step": 11606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.532126597553617e-06,
+      "loss": 1.2897,
+      "step": 11607
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.53190536266067e-06,
+      "loss": 1.2205,
+      "step": 11608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.53168412776772e-06,
+      "loss": 1.2184,
+      "step": 11609
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.531462892874773e-06,
+      "loss": 1.1246,
+      "step": 11610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.531241657981826e-06,
+      "loss": 1.4445,
+      "step": 11611
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.531020423088878e-06,
+      "loss": 1.3217,
+      "step": 11612
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.53079918819593e-06,
+      "loss": 1.2974,
+      "step": 11613
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.530577953302982e-06,
+      "loss": 1.1905,
+      "step": 11614
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.530356718410033e-06,
+      "loss": 1.0721,
+      "step": 11615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.530135483517085e-06,
+      "loss": 1.2763,
+      "step": 11616
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.529914248624138e-06,
+      "loss": 1.337,
+      "step": 11617
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.52969301373119e-06,
+      "loss": 1.1645,
+      "step": 11618
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.529471778838243e-06,
+      "loss": 1.2025,
+      "step": 11619
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.529250543945294e-06,
+      "loss": 1.1375,
+      "step": 11620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.529029309052345e-06,
+      "loss": 1.1945,
+      "step": 11621
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.528808074159398e-06,
+      "loss": 1.3,
+      "step": 11622
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.52858683926645e-06,
+      "loss": 1.2933,
+      "step": 11623
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.528365604373503e-06,
+      "loss": 1.1436,
+      "step": 11624
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.528144369480555e-06,
+      "loss": 1.2384,
+      "step": 11625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.527923134587606e-06,
+      "loss": 1.3709,
+      "step": 11626
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.527701899694659e-06,
+      "loss": 1.2786,
+      "step": 11627
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.527480664801712e-06,
+      "loss": 1.1526,
+      "step": 11628
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.527259429908763e-06,
+      "loss": 1.2099,
+      "step": 11629
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.527038195015815e-06,
+      "loss": 1.3842,
+      "step": 11630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.526816960122866e-06,
+      "loss": 1.2542,
+      "step": 11631
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.526595725229919e-06,
+      "loss": 1.4557,
+      "step": 11632
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.526374490336971e-06,
+      "loss": 1.1982,
+      "step": 11633
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.526153255444024e-06,
+      "loss": 1.2778,
+      "step": 11634
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.525932020551076e-06,
+      "loss": 1.3135,
+      "step": 11635
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.525710785658127e-06,
+      "loss": 1.2794,
+      "step": 11636
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.525489550765178e-06,
+      "loss": 1.2387,
+      "step": 11637
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.525268315872231e-06,
+      "loss": 1.2468,
+      "step": 11638
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.525047080979284e-06,
+      "loss": 1.2238,
+      "step": 11639
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.524825846086336e-06,
+      "loss": 1.2593,
+      "step": 11640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.524604611193389e-06,
+      "loss": 1.2481,
+      "step": 11641
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.52438337630044e-06,
+      "loss": 1.2416,
+      "step": 11642
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.524162141407492e-06,
+      "loss": 1.2906,
+      "step": 11643
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.523940906514545e-06,
+      "loss": 1.2946,
+      "step": 11644
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.523719671621596e-06,
+      "loss": 1.2299,
+      "step": 11645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.523498436728648e-06,
+      "loss": 1.2973,
+      "step": 11646
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.523277201835701e-06,
+      "loss": 1.1756,
+      "step": 11647
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.523055966942752e-06,
+      "loss": 1.1813,
+      "step": 11648
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.522834732049805e-06,
+      "loss": 1.2589,
+      "step": 11649
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.522613497156857e-06,
+      "loss": 1.147,
+      "step": 11650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.52239226226391e-06,
+      "loss": 1.1567,
+      "step": 11651
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.52217102737096e-06,
+      "loss": 1.2541,
+      "step": 11652
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.521949792478013e-06,
+      "loss": 1.3747,
+      "step": 11653
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.521728557585064e-06,
+      "loss": 1.2747,
+      "step": 11654
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.521507322692117e-06,
+      "loss": 1.1888,
+      "step": 11655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.52128608779917e-06,
+      "loss": 1.1759,
+      "step": 11656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.521064852906222e-06,
+      "loss": 1.2124,
+      "step": 11657
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.520843618013275e-06,
+      "loss": 1.3606,
+      "step": 11658
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.520622383120326e-06,
+      "loss": 1.245,
+      "step": 11659
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.520401148227378e-06,
+      "loss": 1.2782,
+      "step": 11660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.52017991333443e-06,
+      "loss": 1.2873,
+      "step": 11661
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.519958678441482e-06,
+      "loss": 1.1376,
+      "step": 11662
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.519737443548534e-06,
+      "loss": 1.2627,
+      "step": 11663
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.519516208655585e-06,
+      "loss": 1.2481,
+      "step": 11664
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.519294973762638e-06,
+      "loss": 1.2415,
+      "step": 11665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.51907373886969e-06,
+      "loss": 1.1717,
+      "step": 11666
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.518852503976743e-06,
+      "loss": 1.3922,
+      "step": 11667
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.518631269083794e-06,
+      "loss": 1.2772,
+      "step": 11668
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.518410034190847e-06,
+      "loss": 1.2407,
+      "step": 11669
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.518188799297898e-06,
+      "loss": 1.2914,
+      "step": 11670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.51796756440495e-06,
+      "loss": 1.3201,
+      "step": 11671
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.517746329512003e-06,
+      "loss": 1.2506,
+      "step": 11672
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.517525094619055e-06,
+      "loss": 1.2351,
+      "step": 11673
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.517303859726108e-06,
+      "loss": 1.2154,
+      "step": 11674
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.517082624833159e-06,
+      "loss": 1.2823,
+      "step": 11675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.516861389940212e-06,
+      "loss": 1.282,
+      "step": 11676
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.516640155047262e-06,
+      "loss": 1.3454,
+      "step": 11677
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.516418920154315e-06,
+      "loss": 1.1603,
+      "step": 11678
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.516197685261368e-06,
+      "loss": 1.1411,
+      "step": 11679
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.51597645036842e-06,
+      "loss": 1.2077,
+      "step": 11680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.515755215475471e-06,
+      "loss": 1.26,
+      "step": 11681
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.515533980582524e-06,
+      "loss": 1.1359,
+      "step": 11682
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.515312745689576e-06,
+      "loss": 1.099,
+      "step": 11683
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.515091510796629e-06,
+      "loss": 1.2668,
+      "step": 11684
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.51487027590368e-06,
+      "loss": 1.2971,
+      "step": 11685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.514649041010731e-06,
+      "loss": 1.1105,
+      "step": 11686
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.514427806117784e-06,
+      "loss": 1.1086,
+      "step": 11687
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.514206571224836e-06,
+      "loss": 1.2309,
+      "step": 11688
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.513985336331889e-06,
+      "loss": 1.2139,
+      "step": 11689
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.513764101438941e-06,
+      "loss": 1.2237,
+      "step": 11690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.513542866545994e-06,
+      "loss": 1.2682,
+      "step": 11691
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.513321631653045e-06,
+      "loss": 1.1478,
+      "step": 11692
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.513100396760096e-06,
+      "loss": 1.184,
+      "step": 11693
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.512879161867148e-06,
+      "loss": 1.3698,
+      "step": 11694
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.512657926974201e-06,
+      "loss": 1.2597,
+      "step": 11695
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.512436692081254e-06,
+      "loss": 1.3962,
+      "step": 11696
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.512215457188305e-06,
+      "loss": 1.2746,
+      "step": 11697
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.511994222295357e-06,
+      "loss": 1.3371,
+      "step": 11698
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.51177298740241e-06,
+      "loss": 1.3099,
+      "step": 11699
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.511551752509462e-06,
+      "loss": 1.2873,
+      "step": 11700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.511330517616513e-06,
+      "loss": 1.2895,
+      "step": 11701
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.511109282723564e-06,
+      "loss": 1.1825,
+      "step": 11702
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.510888047830617e-06,
+      "loss": 1.139,
+      "step": 11703
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.51066681293767e-06,
+      "loss": 1.373,
+      "step": 11704
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.510445578044722e-06,
+      "loss": 1.2865,
+      "step": 11705
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.510224343151775e-06,
+      "loss": 1.2612,
+      "step": 11706
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.510003108258827e-06,
+      "loss": 1.3292,
+      "step": 11707
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.509781873365878e-06,
+      "loss": 1.3557,
+      "step": 11708
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.509560638472929e-06,
+      "loss": 1.2135,
+      "step": 11709
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.509339403579982e-06,
+      "loss": 1.2408,
+      "step": 11710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.509118168687034e-06,
+      "loss": 1.2237,
+      "step": 11711
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.508896933794087e-06,
+      "loss": 1.1236,
+      "step": 11712
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.508675698901138e-06,
+      "loss": 1.3454,
+      "step": 11713
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.50845446400819e-06,
+      "loss": 1.3753,
+      "step": 11714
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.508233229115243e-06,
+      "loss": 1.383,
+      "step": 11715
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.508011994222296e-06,
+      "loss": 1.3216,
+      "step": 11716
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.507790759329347e-06,
+      "loss": 1.2474,
+      "step": 11717
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.5075695244364e-06,
+      "loss": 1.2564,
+      "step": 11718
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.50734828954345e-06,
+      "loss": 1.2673,
+      "step": 11719
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.507127054650503e-06,
+      "loss": 1.1574,
+      "step": 11720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.506905819757555e-06,
+      "loss": 1.2562,
+      "step": 11721
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.506684584864608e-06,
+      "loss": 1.4351,
+      "step": 11722
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.50646334997166e-06,
+      "loss": 1.2952,
+      "step": 11723
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.506242115078712e-06,
+      "loss": 1.2988,
+      "step": 11724
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.506020880185762e-06,
+      "loss": 1.2257,
+      "step": 11725
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.505799645292815e-06,
+      "loss": 1.1583,
+      "step": 11726
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.505578410399868e-06,
+      "loss": 1.1121,
+      "step": 11727
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.50535717550692e-06,
+      "loss": 1.2718,
+      "step": 11728
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.505135940613973e-06,
+      "loss": 1.3396,
+      "step": 11729
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.504914705721024e-06,
+      "loss": 1.2701,
+      "step": 11730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.504693470828076e-06,
+      "loss": 1.2815,
+      "step": 11731
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.504472235935129e-06,
+      "loss": 1.1098,
+      "step": 11732
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.50425100104218e-06,
+      "loss": 1.214,
+      "step": 11733
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.504029766149233e-06,
+      "loss": 1.3711,
+      "step": 11734
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.503808531256283e-06,
+      "loss": 1.3435,
+      "step": 11735
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.503587296363336e-06,
+      "loss": 1.427,
+      "step": 11736
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.503366061470389e-06,
+      "loss": 1.2819,
+      "step": 11737
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.503144826577441e-06,
+      "loss": 1.3108,
+      "step": 11738
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.502923591684494e-06,
+      "loss": 1.2185,
+      "step": 11739
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.502702356791545e-06,
+      "loss": 1.2431,
+      "step": 11740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.502481121898597e-06,
+      "loss": 1.3534,
+      "step": 11741
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.502259887005648e-06,
+      "loss": 1.1717,
+      "step": 11742
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.502038652112701e-06,
+      "loss": 1.1616,
+      "step": 11743
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.501817417219754e-06,
+      "loss": 1.2693,
+      "step": 11744
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.501596182326806e-06,
+      "loss": 1.3321,
+      "step": 11745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.501374947433857e-06,
+      "loss": 1.298,
+      "step": 11746
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.50115371254091e-06,
+      "loss": 1.2849,
+      "step": 11747
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.500932477647962e-06,
+      "loss": 1.3534,
+      "step": 11748
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.500711242755013e-06,
+      "loss": 1.216,
+      "step": 11749
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.500490007862066e-06,
+      "loss": 1.2762,
+      "step": 11750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.500268772969118e-06,
+      "loss": 1.2663,
+      "step": 11751
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.50004753807617e-06,
+      "loss": 1.3728,
+      "step": 11752
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.499826303183222e-06,
+      "loss": 1.2584,
+      "step": 11753
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.499605068290275e-06,
+      "loss": 1.1641,
+      "step": 11754
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.499383833397327e-06,
+      "loss": 1.2963,
+      "step": 11755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.499162598504378e-06,
+      "loss": 1.1696,
+      "step": 11756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.49894136361143e-06,
+      "loss": 1.3302,
+      "step": 11757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.498720128718482e-06,
+      "loss": 1.1803,
+      "step": 11758
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.498498893825534e-06,
+      "loss": 1.2514,
+      "step": 11759
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.498277658932587e-06,
+      "loss": 1.2443,
+      "step": 11760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.49805642403964e-06,
+      "loss": 1.2584,
+      "step": 11761
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.497835189146692e-06,
+      "loss": 1.163,
+      "step": 11762
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.497613954253743e-06,
+      "loss": 1.2389,
+      "step": 11763
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.497392719360796e-06,
+      "loss": 1.283,
+      "step": 11764
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.497171484467847e-06,
+      "loss": 1.3421,
+      "step": 11765
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.4969502495749e-06,
+      "loss": 1.152,
+      "step": 11766
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.496729014681952e-06,
+      "loss": 1.1747,
+      "step": 11767
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.496507779789003e-06,
+      "loss": 1.2299,
+      "step": 11768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.496286544896055e-06,
+      "loss": 1.3075,
+      "step": 11769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.496065310003108e-06,
+      "loss": 1.1627,
+      "step": 11770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.49584407511016e-06,
+      "loss": 1.3231,
+      "step": 11771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.495622840217213e-06,
+      "loss": 1.2638,
+      "step": 11772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.495401605324264e-06,
+      "loss": 1.2327,
+      "step": 11773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.495180370431315e-06,
+      "loss": 1.2602,
+      "step": 11774
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.494959135538368e-06,
+      "loss": 1.313,
+      "step": 11775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.49473790064542e-06,
+      "loss": 1.268,
+      "step": 11776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.494516665752473e-06,
+      "loss": 1.2018,
+      "step": 11777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.494295430859525e-06,
+      "loss": 1.2762,
+      "step": 11778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.494074195966576e-06,
+      "loss": 1.2286,
+      "step": 11779
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.493852961073629e-06,
+      "loss": 1.2074,
+      "step": 11780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.49363172618068e-06,
+      "loss": 1.2317,
+      "step": 11781
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.493410491287733e-06,
+      "loss": 1.2496,
+      "step": 11782
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.493189256394785e-06,
+      "loss": 1.2503,
+      "step": 11783
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.492968021501838e-06,
+      "loss": 1.2669,
+      "step": 11784
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.492746786608889e-06,
+      "loss": 1.215,
+      "step": 11785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.492525551715941e-06,
+      "loss": 1.226,
+      "step": 11786
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.492304316822994e-06,
+      "loss": 1.23,
+      "step": 11787
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.492083081930046e-06,
+      "loss": 1.37,
+      "step": 11788
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.491861847037097e-06,
+      "loss": 1.2511,
+      "step": 11789
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.491640612144148e-06,
+      "loss": 1.2293,
+      "step": 11790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.491419377251201e-06,
+      "loss": 1.1903,
+      "step": 11791
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.491198142358254e-06,
+      "loss": 1.2318,
+      "step": 11792
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.490976907465306e-06,
+      "loss": 1.1959,
+      "step": 11793
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.490755672572359e-06,
+      "loss": 1.2099,
+      "step": 11794
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.49053443767941e-06,
+      "loss": 1.2955,
+      "step": 11795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.490313202786462e-06,
+      "loss": 1.3245,
+      "step": 11796
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.490091967893513e-06,
+      "loss": 1.3107,
+      "step": 11797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.489870733000566e-06,
+      "loss": 1.2798,
+      "step": 11798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.489649498107618e-06,
+      "loss": 1.2833,
+      "step": 11799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.489428263214671e-06,
+      "loss": 1.2398,
+      "step": 11800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.489207028321722e-06,
+      "loss": 1.2267,
+      "step": 11801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.488985793428775e-06,
+      "loss": 1.2878,
+      "step": 11802
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.488764558535827e-06,
+      "loss": 1.1798,
+      "step": 11803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.48854332364288e-06,
+      "loss": 1.1419,
+      "step": 11804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.48832208874993e-06,
+      "loss": 1.24,
+      "step": 11805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.488100853856982e-06,
+      "loss": 1.2491,
+      "step": 11806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.487879618964034e-06,
+      "loss": 1.1255,
+      "step": 11807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.487658384071087e-06,
+      "loss": 1.2829,
+      "step": 11808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.48743714917814e-06,
+      "loss": 1.4047,
+      "step": 11809
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.487215914285192e-06,
+      "loss": 1.2851,
+      "step": 11810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.486994679392245e-06,
+      "loss": 1.2695,
+      "step": 11811
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.486773444499296e-06,
+      "loss": 1.1882,
+      "step": 11812
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.486552209606347e-06,
+      "loss": 1.2654,
+      "step": 11813
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.486330974713399e-06,
+      "loss": 1.2968,
+      "step": 11814
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.486109739820452e-06,
+      "loss": 1.2045,
+      "step": 11815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.485888504927504e-06,
+      "loss": 1.4741,
+      "step": 11816
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.485667270034555e-06,
+      "loss": 1.209,
+      "step": 11817
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.485446035141608e-06,
+      "loss": 1.2201,
+      "step": 11818
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.48522480024866e-06,
+      "loss": 1.2104,
+      "step": 11819
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.485003565355713e-06,
+      "loss": 1.2461,
+      "step": 11820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.484782330462764e-06,
+      "loss": 1.3429,
+      "step": 11821
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.484561095569817e-06,
+      "loss": 1.3032,
+      "step": 11822
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.484339860676868e-06,
+      "loss": 1.275,
+      "step": 11823
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.48411862578392e-06,
+      "loss": 1.1771,
+      "step": 11824
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.483897390890973e-06,
+      "loss": 1.2266,
+      "step": 11825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.483676155998025e-06,
+      "loss": 1.3444,
+      "step": 11826
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.483454921105078e-06,
+      "loss": 1.2544,
+      "step": 11827
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.483233686212129e-06,
+      "loss": 1.3314,
+      "step": 11828
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.483012451319182e-06,
+      "loss": 1.2719,
+      "step": 11829
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.482791216426232e-06,
+      "loss": 1.2385,
+      "step": 11830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.482569981533285e-06,
+      "loss": 1.2388,
+      "step": 11831
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.482348746640338e-06,
+      "loss": 1.2238,
+      "step": 11832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.48212751174739e-06,
+      "loss": 1.1739,
+      "step": 11833
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.481906276854441e-06,
+      "loss": 1.1607,
+      "step": 11834
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.481685041961494e-06,
+      "loss": 1.236,
+      "step": 11835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.481463807068546e-06,
+      "loss": 1.1952,
+      "step": 11836
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.481242572175597e-06,
+      "loss": 1.2057,
+      "step": 11837
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.48102133728265e-06,
+      "loss": 1.2593,
+      "step": 11838
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.480800102389701e-06,
+      "loss": 1.2873,
+      "step": 11839
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.480578867496753e-06,
+      "loss": 1.235,
+      "step": 11840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.480357632603806e-06,
+      "loss": 1.2447,
+      "step": 11841
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.480136397710859e-06,
+      "loss": 1.2751,
+      "step": 11842
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.479915162817911e-06,
+      "loss": 1.1916,
+      "step": 11843
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.479693927924962e-06,
+      "loss": 1.2222,
+      "step": 11844
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.479472693032015e-06,
+      "loss": 1.2542,
+      "step": 11845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.479251458139066e-06,
+      "loss": 1.2696,
+      "step": 11846
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.479030223246118e-06,
+      "loss": 1.3094,
+      "step": 11847
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.478808988353171e-06,
+      "loss": 1.3502,
+      "step": 11848
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.478587753460224e-06,
+      "loss": 1.1219,
+      "step": 11849
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.478366518567275e-06,
+      "loss": 1.1952,
+      "step": 11850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.478145283674327e-06,
+      "loss": 1.2603,
+      "step": 11851
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.47792404878138e-06,
+      "loss": 1.2459,
+      "step": 11852
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.47770281388843e-06,
+      "loss": 1.3323,
+      "step": 11853
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.477481578995483e-06,
+      "loss": 1.239,
+      "step": 11854
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.477260344102536e-06,
+      "loss": 1.2888,
+      "step": 11855
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.477039109209587e-06,
+      "loss": 1.1489,
+      "step": 11856
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.47681787431664e-06,
+      "loss": 1.1882,
+      "step": 11857
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.476596639423692e-06,
+      "loss": 1.2336,
+      "step": 11858
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.476375404530745e-06,
+      "loss": 1.257,
+      "step": 11859
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.476154169637797e-06,
+      "loss": 1.307,
+      "step": 11860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.475932934744848e-06,
+      "loss": 1.3889,
+      "step": 11861
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.475711699851899e-06,
+      "loss": 1.1838,
+      "step": 11862
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.475490464958952e-06,
+      "loss": 1.1142,
+      "step": 11863
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.475269230066004e-06,
+      "loss": 1.214,
+      "step": 11864
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.475047995173057e-06,
+      "loss": 1.2566,
+      "step": 11865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.47482676028011e-06,
+      "loss": 1.1272,
+      "step": 11866
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.47460552538716e-06,
+      "loss": 1.2737,
+      "step": 11867
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.474384290494213e-06,
+      "loss": 1.2461,
+      "step": 11868
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.474163055601264e-06,
+      "loss": 1.1223,
+      "step": 11869
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.473941820708317e-06,
+      "loss": 1.3571,
+      "step": 11870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.47372058581537e-06,
+      "loss": 1.3085,
+      "step": 11871
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.47349935092242e-06,
+      "loss": 1.3356,
+      "step": 11872
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.473278116029473e-06,
+      "loss": 1.1302,
+      "step": 11873
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.473056881136525e-06,
+      "loss": 1.194,
+      "step": 11874
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.472835646243578e-06,
+      "loss": 1.2184,
+      "step": 11875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.47261441135063e-06,
+      "loss": 1.3782,
+      "step": 11876
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.472393176457682e-06,
+      "loss": 1.2567,
+      "step": 11877
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.472171941564732e-06,
+      "loss": 1.3962,
+      "step": 11878
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.471950706671785e-06,
+      "loss": 1.1548,
+      "step": 11879
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.471729471778838e-06,
+      "loss": 1.2088,
+      "step": 11880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.47150823688589e-06,
+      "loss": 1.2431,
+      "step": 11881
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.471287001992943e-06,
+      "loss": 1.2146,
+      "step": 11882
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.471065767099994e-06,
+      "loss": 1.3196,
+      "step": 11883
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.470844532207046e-06,
+      "loss": 1.2725,
+      "step": 11884
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.470623297314097e-06,
+      "loss": 1.1674,
+      "step": 11885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.47040206242115e-06,
+      "loss": 1.273,
+      "step": 11886
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.470180827528203e-06,
+      "loss": 1.1622,
+      "step": 11887
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.469959592635253e-06,
+      "loss": 1.3002,
+      "step": 11888
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.469738357742306e-06,
+      "loss": 1.194,
+      "step": 11889
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.469517122849359e-06,
+      "loss": 1.2665,
+      "step": 11890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.469295887956411e-06,
+      "loss": 1.1781,
+      "step": 11891
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.469074653063464e-06,
+      "loss": 1.2419,
+      "step": 11892
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.468853418170515e-06,
+      "loss": 1.2557,
+      "step": 11893
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.468632183277566e-06,
+      "loss": 1.1799,
+      "step": 11894
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.468410948384618e-06,
+      "loss": 1.2074,
+      "step": 11895
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.468189713491671e-06,
+      "loss": 1.3387,
+      "step": 11896
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.467968478598724e-06,
+      "loss": 1.2476,
+      "step": 11897
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.467747243705776e-06,
+      "loss": 1.3494,
+      "step": 11898
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.467526008812827e-06,
+      "loss": 1.2239,
+      "step": 11899
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.46730477391988e-06,
+      "loss": 1.2037,
+      "step": 11900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.46708353902693e-06,
+      "loss": 1.3239,
+      "step": 11901
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.466862304133983e-06,
+      "loss": 1.3397,
+      "step": 11902
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.466641069241036e-06,
+      "loss": 1.2582,
+      "step": 11903
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.466419834348088e-06,
+      "loss": 1.2317,
+      "step": 11904
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.46619859945514e-06,
+      "loss": 1.3086,
+      "step": 11905
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.465977364562192e-06,
+      "loss": 1.2229,
+      "step": 11906
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.465756129669245e-06,
+      "loss": 1.2754,
+      "step": 11907
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.465534894776297e-06,
+      "loss": 1.0911,
+      "step": 11908
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.465313659883348e-06,
+      "loss": 1.1954,
+      "step": 11909
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.465092424990399e-06,
+      "loss": 1.2922,
+      "step": 11910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.464871190097452e-06,
+      "loss": 1.1319,
+      "step": 11911
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.464649955204504e-06,
+      "loss": 1.2564,
+      "step": 11912
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.464428720311557e-06,
+      "loss": 1.1677,
+      "step": 11913
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.46420748541861e-06,
+      "loss": 1.2745,
+      "step": 11914
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.463986250525662e-06,
+      "loss": 1.2213,
+      "step": 11915
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.463765015632713e-06,
+      "loss": 1.2008,
+      "step": 11916
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.463543780739766e-06,
+      "loss": 1.2805,
+      "step": 11917
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.463322545846817e-06,
+      "loss": 1.2254,
+      "step": 11918
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.46310131095387e-06,
+      "loss": 1.3152,
+      "step": 11919
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.462880076060922e-06,
+      "loss": 1.1978,
+      "step": 11920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.462658841167973e-06,
+      "loss": 1.2244,
+      "step": 11921
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.462437606275025e-06,
+      "loss": 1.1847,
+      "step": 11922
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.462216371382078e-06,
+      "loss": 1.3183,
+      "step": 11923
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.46199513648913e-06,
+      "loss": 1.2471,
+      "step": 11924
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.461773901596181e-06,
+      "loss": 1.3604,
+      "step": 11925
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.461552666703234e-06,
+      "loss": 1.183,
+      "step": 11926
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.461331431810285e-06,
+      "loss": 1.3465,
+      "step": 11927
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.461110196917338e-06,
+      "loss": 1.2383,
+      "step": 11928
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.46088896202439e-06,
+      "loss": 1.1984,
+      "step": 11929
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.460667727131443e-06,
+      "loss": 1.3102,
+      "step": 11930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.460446492238495e-06,
+      "loss": 1.3455,
+      "step": 11931
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.460225257345546e-06,
+      "loss": 1.2244,
+      "step": 11932
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.460004022452599e-06,
+      "loss": 1.3405,
+      "step": 11933
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.45978278755965e-06,
+      "loss": 1.26,
+      "step": 11934
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.459561552666702e-06,
+      "loss": 1.3332,
+      "step": 11935
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.459340317773755e-06,
+      "loss": 1.3089,
+      "step": 11936
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.459119082880808e-06,
+      "loss": 1.3741,
+      "step": 11937
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.458897847987859e-06,
+      "loss": 1.2738,
+      "step": 11938
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.458676613094911e-06,
+      "loss": 1.1668,
+      "step": 11939
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.458455378201964e-06,
+      "loss": 1.213,
+      "step": 11940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.458234143309015e-06,
+      "loss": 1.2101,
+      "step": 11941
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.458012908416067e-06,
+      "loss": 1.1486,
+      "step": 11942
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.457791673523118e-06,
+      "loss": 1.2029,
+      "step": 11943
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.457570438630171e-06,
+      "loss": 1.244,
+      "step": 11944
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.457349203737224e-06,
+      "loss": 1.3472,
+      "step": 11945
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.457127968844276e-06,
+      "loss": 1.2131,
+      "step": 11946
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.456906733951329e-06,
+      "loss": 1.2813,
+      "step": 11947
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.456685499058381e-06,
+      "loss": 1.2779,
+      "step": 11948
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.456464264165432e-06,
+      "loss": 1.3241,
+      "step": 11949
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.456243029272483e-06,
+      "loss": 1.2491,
+      "step": 11950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.456021794379536e-06,
+      "loss": 1.1355,
+      "step": 11951
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.455800559486588e-06,
+      "loss": 1.1648,
+      "step": 11952
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.455579324593641e-06,
+      "loss": 1.3376,
+      "step": 11953
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.455358089700692e-06,
+      "loss": 1.1817,
+      "step": 11954
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.455136854807745e-06,
+      "loss": 1.235,
+      "step": 11955
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.454915619914797e-06,
+      "loss": 1.2433,
+      "step": 11956
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.454694385021848e-06,
+      "loss": 1.2648,
+      "step": 11957
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.4544731501289e-06,
+      "loss": 1.2023,
+      "step": 11958
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.454251915235953e-06,
+      "loss": 1.1831,
+      "step": 11959
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.454030680343004e-06,
+      "loss": 1.1784,
+      "step": 11960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.453809445450057e-06,
+      "loss": 1.2982,
+      "step": 11961
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.45358821055711e-06,
+      "loss": 1.2135,
+      "step": 11962
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.453366975664162e-06,
+      "loss": 1.248,
+      "step": 11963
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.453145740771215e-06,
+      "loss": 1.2964,
+      "step": 11964
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.452924505878266e-06,
+      "loss": 1.1692,
+      "step": 11965
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.452703270985317e-06,
+      "loss": 1.3571,
+      "step": 11966
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.452482036092369e-06,
+      "loss": 1.2823,
+      "step": 11967
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.452260801199422e-06,
+      "loss": 1.2684,
+      "step": 11968
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.452039566306474e-06,
+      "loss": 1.2877,
+      "step": 11969
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.451818331413527e-06,
+      "loss": 1.1603,
+      "step": 11970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.451597096520578e-06,
+      "loss": 1.2589,
+      "step": 11971
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.45137586162763e-06,
+      "loss": 1.1238,
+      "step": 11972
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.451154626734681e-06,
+      "loss": 1.2715,
+      "step": 11973
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.450933391841734e-06,
+      "loss": 1.4484,
+      "step": 11974
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.450712156948787e-06,
+      "loss": 1.2096,
+      "step": 11975
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.450490922055838e-06,
+      "loss": 1.3043,
+      "step": 11976
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.45026968716289e-06,
+      "loss": 1.2384,
+      "step": 11977
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.450048452269943e-06,
+      "loss": 1.2069,
+      "step": 11978
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.449827217376995e-06,
+      "loss": 1.2315,
+      "step": 11979
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.449605982484048e-06,
+      "loss": 1.1642,
+      "step": 11980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.449384747591099e-06,
+      "loss": 1.1988,
+      "step": 11981
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.44916351269815e-06,
+      "loss": 1.1961,
+      "step": 11982
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.448942277805202e-06,
+      "loss": 1.2059,
+      "step": 11983
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.448721042912255e-06,
+      "loss": 1.279,
+      "step": 11984
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.448499808019308e-06,
+      "loss": 1.2154,
+      "step": 11985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.44827857312636e-06,
+      "loss": 1.2115,
+      "step": 11986
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.448057338233411e-06,
+      "loss": 1.2572,
+      "step": 11987
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.447836103340464e-06,
+      "loss": 1.2679,
+      "step": 11988
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.447614868447515e-06,
+      "loss": 1.1567,
+      "step": 11989
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.447393633554567e-06,
+      "loss": 1.2087,
+      "step": 11990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.44717239866162e-06,
+      "loss": 1.2367,
+      "step": 11991
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.446951163768671e-06,
+      "loss": 1.1221,
+      "step": 11992
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.446729928875723e-06,
+      "loss": 1.2251,
+      "step": 11993
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.446508693982776e-06,
+      "loss": 1.2754,
+      "step": 11994
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.446287459089829e-06,
+      "loss": 1.218,
+      "step": 11995
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.446066224196881e-06,
+      "loss": 1.3057,
+      "step": 11996
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.445844989303932e-06,
+      "loss": 1.3567,
+      "step": 11997
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.445623754410983e-06,
+      "loss": 1.2183,
+      "step": 11998
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.445402519518036e-06,
+      "loss": 1.3346,
+      "step": 11999
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.445181284625088e-06,
+      "loss": 1.2359,
+      "step": 12000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.444960049732141e-06,
+      "loss": 1.2227,
+      "step": 12001
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.444738814839194e-06,
+      "loss": 1.1648,
+      "step": 12002
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.444517579946245e-06,
+      "loss": 1.3534,
+      "step": 12003
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.444296345053297e-06,
+      "loss": 1.1799,
+      "step": 12004
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.44407511016035e-06,
+      "loss": 1.2004,
+      "step": 12005
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.4438538752674e-06,
+      "loss": 1.3613,
+      "step": 12006
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.443632640374453e-06,
+      "loss": 1.2557,
+      "step": 12007
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.443411405481506e-06,
+      "loss": 1.3469,
+      "step": 12008
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.443190170588557e-06,
+      "loss": 1.2379,
+      "step": 12009
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.44296893569561e-06,
+      "loss": 1.2004,
+      "step": 12010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.442747700802662e-06,
+      "loss": 1.2423,
+      "step": 12011
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.442526465909715e-06,
+      "loss": 1.2671,
+      "step": 12012
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.442305231016766e-06,
+      "loss": 1.4208,
+      "step": 12013
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.442083996123816e-06,
+      "loss": 1.1703,
+      "step": 12014
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.441862761230869e-06,
+      "loss": 1.1934,
+      "step": 12015
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.441641526337922e-06,
+      "loss": 1.1757,
+      "step": 12016
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.441420291444974e-06,
+      "loss": 1.3192,
+      "step": 12017
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.441199056552027e-06,
+      "loss": 1.2088,
+      "step": 12018
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.44097782165908e-06,
+      "loss": 1.3149,
+      "step": 12019
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.44075658676613e-06,
+      "loss": 1.2114,
+      "step": 12020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.440535351873183e-06,
+      "loss": 1.109,
+      "step": 12021
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.440314116980234e-06,
+      "loss": 1.1956,
+      "step": 12022
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.440092882087287e-06,
+      "loss": 1.2025,
+      "step": 12023
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.43987164719434e-06,
+      "loss": 1.2972,
+      "step": 12024
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.43965041230139e-06,
+      "loss": 1.2429,
+      "step": 12025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.439429177408443e-06,
+      "loss": 1.2491,
+      "step": 12026
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.439207942515495e-06,
+      "loss": 1.2785,
+      "step": 12027
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.438986707622548e-06,
+      "loss": 1.2924,
+      "step": 12028
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.438765472729599e-06,
+      "loss": 1.2747,
+      "step": 12029
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.438544237836651e-06,
+      "loss": 1.2145,
+      "step": 12030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.438323002943702e-06,
+      "loss": 1.2936,
+      "step": 12031
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.438101768050755e-06,
+      "loss": 1.2812,
+      "step": 12032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.437880533157808e-06,
+      "loss": 1.2394,
+      "step": 12033
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.43765929826486e-06,
+      "loss": 1.1305,
+      "step": 12034
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.437438063371913e-06,
+      "loss": 1.1734,
+      "step": 12035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.437216828478964e-06,
+      "loss": 1.209,
+      "step": 12036
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.436995593586016e-06,
+      "loss": 1.1856,
+      "step": 12037
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.436774358693067e-06,
+      "loss": 1.3058,
+      "step": 12038
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.43655312380012e-06,
+      "loss": 1.274,
+      "step": 12039
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.436331888907173e-06,
+      "loss": 1.1857,
+      "step": 12040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.436110654014225e-06,
+      "loss": 1.207,
+      "step": 12041
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.435889419121276e-06,
+      "loss": 1.0912,
+      "step": 12042
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.435668184228329e-06,
+      "loss": 1.1921,
+      "step": 12043
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.435446949335381e-06,
+      "loss": 1.2287,
+      "step": 12044
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.435225714442432e-06,
+      "loss": 1.1657,
+      "step": 12045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.435004479549485e-06,
+      "loss": 1.2374,
+      "step": 12046
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.434783244656536e-06,
+      "loss": 1.2335,
+      "step": 12047
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.434562009763588e-06,
+      "loss": 1.1485,
+      "step": 12048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.434340774870641e-06,
+      "loss": 1.187,
+      "step": 12049
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.434119539977694e-06,
+      "loss": 1.2289,
+      "step": 12050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.433898305084746e-06,
+      "loss": 1.2635,
+      "step": 12051
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.433677070191799e-06,
+      "loss": 1.364,
+      "step": 12052
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.43345583529885e-06,
+      "loss": 1.3078,
+      "step": 12053
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.4332346004059e-06,
+      "loss": 1.2044,
+      "step": 12054
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.433013365512953e-06,
+      "loss": 1.31,
+      "step": 12055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.432792130620006e-06,
+      "loss": 1.2052,
+      "step": 12056
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.432570895727058e-06,
+      "loss": 1.2054,
+      "step": 12057
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.43234966083411e-06,
+      "loss": 1.1886,
+      "step": 12058
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.432128425941162e-06,
+      "loss": 1.2945,
+      "step": 12059
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.431907191048215e-06,
+      "loss": 1.3059,
+      "step": 12060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.431685956155266e-06,
+      "loss": 1.4096,
+      "step": 12061
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.431464721262318e-06,
+      "loss": 1.1237,
+      "step": 12062
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.43124348636937e-06,
+      "loss": 1.2342,
+      "step": 12063
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.431022251476422e-06,
+      "loss": 1.1521,
+      "step": 12064
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.430801016583474e-06,
+      "loss": 1.3442,
+      "step": 12065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.430579781690527e-06,
+      "loss": 1.3097,
+      "step": 12066
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.43035854679758e-06,
+      "loss": 1.3089,
+      "step": 12067
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.430137311904632e-06,
+      "loss": 1.2582,
+      "step": 12068
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.429916077011683e-06,
+      "loss": 1.2946,
+      "step": 12069
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.429694842118734e-06,
+      "loss": 1.3208,
+      "step": 12070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.429473607225787e-06,
+      "loss": 1.2191,
+      "step": 12071
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.42925237233284e-06,
+      "loss": 1.2,
+      "step": 12072
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.429031137439892e-06,
+      "loss": 1.2737,
+      "step": 12073
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.428809902546944e-06,
+      "loss": 1.3504,
+      "step": 12074
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.428588667653995e-06,
+      "loss": 1.1605,
+      "step": 12075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.428367432761048e-06,
+      "loss": 1.2667,
+      "step": 12076
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.428146197868099e-06,
+      "loss": 1.1605,
+      "step": 12077
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.427924962975151e-06,
+      "loss": 1.2404,
+      "step": 12078
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.427703728082204e-06,
+      "loss": 1.169,
+      "step": 12079
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.427482493189255e-06,
+      "loss": 1.2689,
+      "step": 12080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.427261258296308e-06,
+      "loss": 1.2241,
+      "step": 12081
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.42704002340336e-06,
+      "loss": 1.2169,
+      "step": 12082
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.426818788510413e-06,
+      "loss": 1.1995,
+      "step": 12083
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.426597553617465e-06,
+      "loss": 1.1839,
+      "step": 12084
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.426376318724516e-06,
+      "loss": 1.3185,
+      "step": 12085
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.426155083831567e-06,
+      "loss": 1.2871,
+      "step": 12086
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.42593384893862e-06,
+      "loss": 1.0899,
+      "step": 12087
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.425712614045672e-06,
+      "loss": 1.1485,
+      "step": 12088
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.425491379152725e-06,
+      "loss": 1.1687,
+      "step": 12089
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.425270144259778e-06,
+      "loss": 1.2738,
+      "step": 12090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.425048909366829e-06,
+      "loss": 1.2862,
+      "step": 12091
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.424827674473881e-06,
+      "loss": 1.1913,
+      "step": 12092
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.424606439580934e-06,
+      "loss": 1.3359,
+      "step": 12093
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.424385204687985e-06,
+      "loss": 1.2314,
+      "step": 12094
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.424163969795037e-06,
+      "loss": 1.2233,
+      "step": 12095
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.423942734902088e-06,
+      "loss": 1.189,
+      "step": 12096
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.423721500009141e-06,
+      "loss": 1.3175,
+      "step": 12097
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.423500265116194e-06,
+      "loss": 1.2507,
+      "step": 12098
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.423279030223246e-06,
+      "loss": 1.2093,
+      "step": 12099
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.423057795330299e-06,
+      "loss": 1.3167,
+      "step": 12100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.42283656043735e-06,
+      "loss": 1.3804,
+      "step": 12101
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.4226153255444e-06,
+      "loss": 1.2335,
+      "step": 12102
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.422394090651453e-06,
+      "loss": 1.3373,
+      "step": 12103
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.422172855758506e-06,
+      "loss": 1.2297,
+      "step": 12104
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.421951620865558e-06,
+      "loss": 1.2391,
+      "step": 12105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.421730385972611e-06,
+      "loss": 1.1913,
+      "step": 12106
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.421509151079662e-06,
+      "loss": 1.2815,
+      "step": 12107
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.421287916186715e-06,
+      "loss": 1.2238,
+      "step": 12108
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.421066681293767e-06,
+      "loss": 1.2383,
+      "step": 12109
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.420845446400818e-06,
+      "loss": 1.1874,
+      "step": 12110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.42062421150787e-06,
+      "loss": 1.1866,
+      "step": 12111
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.420402976614923e-06,
+      "loss": 1.3338,
+      "step": 12112
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.420181741721974e-06,
+      "loss": 1.3285,
+      "step": 12113
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.419960506829027e-06,
+      "loss": 1.1456,
+      "step": 12114
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.41973927193608e-06,
+      "loss": 1.2292,
+      "step": 12115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.419518037043132e-06,
+      "loss": 1.339,
+      "step": 12116
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.419296802150183e-06,
+      "loss": 1.2271,
+      "step": 12117
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.419075567257234e-06,
+      "loss": 1.2955,
+      "step": 12118
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.418854332364287e-06,
+      "loss": 1.1608,
+      "step": 12119
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.418633097471339e-06,
+      "loss": 1.1115,
+      "step": 12120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.418411862578392e-06,
+      "loss": 1.1677,
+      "step": 12121
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.418190627685444e-06,
+      "loss": 1.1777,
+      "step": 12122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.417969392792497e-06,
+      "loss": 1.2587,
+      "step": 12123
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.417748157899548e-06,
+      "loss": 1.2507,
+      "step": 12124
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.4175269230066e-06,
+      "loss": 1.3616,
+      "step": 12125
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.417305688113651e-06,
+      "loss": 1.2982,
+      "step": 12126
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.417084453220704e-06,
+      "loss": 1.2498,
+      "step": 12127
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.416863218327757e-06,
+      "loss": 1.2879,
+      "step": 12128
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.416641983434808e-06,
+      "loss": 1.2332,
+      "step": 12129
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.41642074854186e-06,
+      "loss": 1.2653,
+      "step": 12130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.416199513648913e-06,
+      "loss": 1.3062,
+      "step": 12131
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.415978278755965e-06,
+      "loss": 1.2402,
+      "step": 12132
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.415757043863016e-06,
+      "loss": 1.153,
+      "step": 12133
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.415535808970069e-06,
+      "loss": 1.2729,
+      "step": 12134
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.41531457407712e-06,
+      "loss": 1.1784,
+      "step": 12135
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.415093339184172e-06,
+      "loss": 1.1604,
+      "step": 12136
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.414872104291225e-06,
+      "loss": 1.1835,
+      "step": 12137
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.414650869398278e-06,
+      "loss": 1.2918,
+      "step": 12138
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.41442963450533e-06,
+      "loss": 1.2464,
+      "step": 12139
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.414208399612381e-06,
+      "loss": 1.204,
+      "step": 12140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.413987164719434e-06,
+      "loss": 1.3159,
+      "step": 12141
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.413765929826485e-06,
+      "loss": 1.2176,
+      "step": 12142
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.413544694933537e-06,
+      "loss": 1.3401,
+      "step": 12143
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.41332346004059e-06,
+      "loss": 1.2945,
+      "step": 12144
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.413102225147643e-06,
+      "loss": 1.3013,
+      "step": 12145
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.412880990254693e-06,
+      "loss": 1.2084,
+      "step": 12146
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.412659755361746e-06,
+      "loss": 1.176,
+      "step": 12147
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.412438520468799e-06,
+      "loss": 1.1972,
+      "step": 12148
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.41221728557585e-06,
+      "loss": 1.184,
+      "step": 12149
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.411996050682902e-06,
+      "loss": 1.392,
+      "step": 12150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.411774815789953e-06,
+      "loss": 1.2126,
+      "step": 12151
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.411553580897006e-06,
+      "loss": 1.2071,
+      "step": 12152
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.411332346004058e-06,
+      "loss": 1.1251,
+      "step": 12153
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.411111111111111e-06,
+      "loss": 1.3076,
+      "step": 12154
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.410889876218164e-06,
+      "loss": 1.293,
+      "step": 12155
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.410668641325216e-06,
+      "loss": 1.3524,
+      "step": 12156
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.410447406432267e-06,
+      "loss": 1.2753,
+      "step": 12157
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.410226171539318e-06,
+      "loss": 1.1357,
+      "step": 12158
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.41000493664637e-06,
+      "loss": 1.1939,
+      "step": 12159
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.409783701753423e-06,
+      "loss": 1.1323,
+      "step": 12160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.409562466860476e-06,
+      "loss": 1.207,
+      "step": 12161
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.409341231967527e-06,
+      "loss": 1.3432,
+      "step": 12162
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.40911999707458e-06,
+      "loss": 1.2118,
+      "step": 12163
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.408898762181632e-06,
+      "loss": 1.1619,
+      "step": 12164
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.408677527288683e-06,
+      "loss": 1.3,
+      "step": 12165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.408456292395736e-06,
+      "loss": 1.2876,
+      "step": 12166
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.408235057502788e-06,
+      "loss": 1.2289,
+      "step": 12167
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.408013822609839e-06,
+      "loss": 1.2796,
+      "step": 12168
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.407792587716892e-06,
+      "loss": 1.2874,
+      "step": 12169
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.407571352823944e-06,
+      "loss": 1.1797,
+      "step": 12170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.407350117930997e-06,
+      "loss": 1.2706,
+      "step": 12171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.40712888303805e-06,
+      "loss": 1.2523,
+      "step": 12172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.4069076481451e-06,
+      "loss": 1.2339,
+      "step": 12173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.406686413252151e-06,
+      "loss": 1.2959,
+      "step": 12174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.406465178359204e-06,
+      "loss": 1.2945,
+      "step": 12175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.406243943466257e-06,
+      "loss": 1.2854,
+      "step": 12176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.40602270857331e-06,
+      "loss": 1.1587,
+      "step": 12177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.405801473680362e-06,
+      "loss": 1.2027,
+      "step": 12178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.405580238787413e-06,
+      "loss": 1.2489,
+      "step": 12179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.405359003894465e-06,
+      "loss": 1.1668,
+      "step": 12180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.405137769001518e-06,
+      "loss": 1.2582,
+      "step": 12181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.404916534108569e-06,
+      "loss": 1.1982,
+      "step": 12182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.404695299215621e-06,
+      "loss": 1.1195,
+      "step": 12183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.404474064322672e-06,
+      "loss": 1.2759,
+      "step": 12184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.404252829429725e-06,
+      "loss": 1.3006,
+      "step": 12185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.404031594536778e-06,
+      "loss": 1.1745,
+      "step": 12186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.40381035964383e-06,
+      "loss": 1.1456,
+      "step": 12187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.403589124750883e-06,
+      "loss": 1.248,
+      "step": 12188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.403367889857934e-06,
+      "loss": 1.2075,
+      "step": 12189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.403146654964985e-06,
+      "loss": 1.1836,
+      "step": 12190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.402925420072037e-06,
+      "loss": 1.3446,
+      "step": 12191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.40270418517909e-06,
+      "loss": 1.3082,
+      "step": 12192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.402482950286143e-06,
+      "loss": 1.3391,
+      "step": 12193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.402261715393195e-06,
+      "loss": 1.2348,
+      "step": 12194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.402040480500246e-06,
+      "loss": 1.1939,
+      "step": 12195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.401819245607299e-06,
+      "loss": 1.2031,
+      "step": 12196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.401598010714351e-06,
+      "loss": 1.2931,
+      "step": 12197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.401376775821402e-06,
+      "loss": 1.3153,
+      "step": 12198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.401155540928455e-06,
+      "loss": 1.1274,
+      "step": 12199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.400934306035506e-06,
+      "loss": 1.2725,
+      "step": 12200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.400713071142558e-06,
+      "loss": 1.2545,
+      "step": 12201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.400491836249611e-06,
+      "loss": 1.3302,
+      "step": 12202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.400270601356664e-06,
+      "loss": 1.2475,
+      "step": 12203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.400049366463716e-06,
+      "loss": 1.2076,
+      "step": 12204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.399828131570767e-06,
+      "loss": 1.2591,
+      "step": 12205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.399606896677818e-06,
+      "loss": 1.2981,
+      "step": 12206
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.39938566178487e-06,
+      "loss": 1.2251,
+      "step": 12207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.399164426891923e-06,
+      "loss": 1.2841,
+      "step": 12208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.398943191998976e-06,
+      "loss": 1.27,
+      "step": 12209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.398721957106028e-06,
+      "loss": 1.225,
+      "step": 12210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.39850072221308e-06,
+      "loss": 1.2926,
+      "step": 12211
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.398279487320132e-06,
+      "loss": 1.2555,
+      "step": 12212
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.398058252427185e-06,
+      "loss": 1.3458,
+      "step": 12213
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.397837017534236e-06,
+      "loss": 1.3045,
+      "step": 12214
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.397615782641288e-06,
+      "loss": 1.2959,
+      "step": 12215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.39739454774834e-06,
+      "loss": 1.2775,
+      "step": 12216
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.397173312855392e-06,
+      "loss": 1.1651,
+      "step": 12217
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.396952077962444e-06,
+      "loss": 1.3154,
+      "step": 12218
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.396730843069497e-06,
+      "loss": 1.2635,
+      "step": 12219
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.39650960817655e-06,
+      "loss": 1.235,
+      "step": 12220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.3962883732836e-06,
+      "loss": 1.2381,
+      "step": 12221
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.396067138390651e-06,
+      "loss": 1.2575,
+      "step": 12222
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.395845903497704e-06,
+      "loss": 1.2168,
+      "step": 12223
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.395624668604757e-06,
+      "loss": 1.2328,
+      "step": 12224
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.395403433711809e-06,
+      "loss": 1.1774,
+      "step": 12225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.395182198818862e-06,
+      "loss": 1.1522,
+      "step": 12226
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.394960963925914e-06,
+      "loss": 1.2911,
+      "step": 12227
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.394739729032965e-06,
+      "loss": 1.3081,
+      "step": 12228
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.394518494140018e-06,
+      "loss": 1.2829,
+      "step": 12229
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.394297259247069e-06,
+      "loss": 1.2412,
+      "step": 12230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.394076024354121e-06,
+      "loss": 1.2123,
+      "step": 12231
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.393854789461174e-06,
+      "loss": 1.2321,
+      "step": 12232
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.393633554568225e-06,
+      "loss": 1.4089,
+      "step": 12233
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.393412319675278e-06,
+      "loss": 1.2782,
+      "step": 12234
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.39319108478233e-06,
+      "loss": 1.161,
+      "step": 12235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.392969849889383e-06,
+      "loss": 1.1337,
+      "step": 12236
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.392748614996434e-06,
+      "loss": 1.4202,
+      "step": 12237
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.392527380103486e-06,
+      "loss": 1.2261,
+      "step": 12238
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.392306145210537e-06,
+      "loss": 1.2762,
+      "step": 12239
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.39208491031759e-06,
+      "loss": 1.1682,
+      "step": 12240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.391863675424642e-06,
+      "loss": 1.2903,
+      "step": 12241
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.391642440531695e-06,
+      "loss": 1.2777,
+      "step": 12242
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.391421205638748e-06,
+      "loss": 1.2652,
+      "step": 12243
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.391199970745799e-06,
+      "loss": 1.2192,
+      "step": 12244
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.390978735852851e-06,
+      "loss": 1.2667,
+      "step": 12245
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.390757500959902e-06,
+      "loss": 1.2397,
+      "step": 12246
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.390536266066955e-06,
+      "loss": 1.3047,
+      "step": 12247
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.390315031174007e-06,
+      "loss": 1.2436,
+      "step": 12248
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.39009379628106e-06,
+      "loss": 1.2911,
+      "step": 12249
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.389872561388111e-06,
+      "loss": 1.284,
+      "step": 12250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.389651326495164e-06,
+      "loss": 1.159,
+      "step": 12251
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.389430091602216e-06,
+      "loss": 1.2757,
+      "step": 12252
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.389208856709267e-06,
+      "loss": 1.2023,
+      "step": 12253
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.38898762181632e-06,
+      "loss": 1.1868,
+      "step": 12254
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.38876638692337e-06,
+      "loss": 1.1842,
+      "step": 12255
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.388545152030423e-06,
+      "loss": 1.2026,
+      "step": 12256
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.388323917137476e-06,
+      "loss": 1.4407,
+      "step": 12257
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.388102682244528e-06,
+      "loss": 1.2445,
+      "step": 12258
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.387881447351581e-06,
+      "loss": 1.3127,
+      "step": 12259
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.387660212458634e-06,
+      "loss": 1.4097,
+      "step": 12260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.387438977565685e-06,
+      "loss": 1.3364,
+      "step": 12261
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.387217742672735e-06,
+      "loss": 1.3031,
+      "step": 12262
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.386996507779788e-06,
+      "loss": 1.2754,
+      "step": 12263
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.38677527288684e-06,
+      "loss": 1.3225,
+      "step": 12264
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.386554037993893e-06,
+      "loss": 1.2308,
+      "step": 12265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.386332803100944e-06,
+      "loss": 1.2507,
+      "step": 12266
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.386111568207997e-06,
+      "loss": 1.2252,
+      "step": 12267
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.38589033331505e-06,
+      "loss": 1.2295,
+      "step": 12268
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.385669098422102e-06,
+      "loss": 1.2802,
+      "step": 12269
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.385447863529153e-06,
+      "loss": 1.3297,
+      "step": 12270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.385226628636206e-06,
+      "loss": 1.147,
+      "step": 12271
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.385005393743256e-06,
+      "loss": 1.269,
+      "step": 12272
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.384784158850309e-06,
+      "loss": 1.3155,
+      "step": 12273
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.384562923957362e-06,
+      "loss": 1.1788,
+      "step": 12274
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.384341689064414e-06,
+      "loss": 1.2277,
+      "step": 12275
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.384120454171467e-06,
+      "loss": 1.2402,
+      "step": 12276
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.383899219278518e-06,
+      "loss": 1.282,
+      "step": 12277
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.383677984385569e-06,
+      "loss": 1.3668,
+      "step": 12278
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.383456749492621e-06,
+      "loss": 1.197,
+      "step": 12279
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.383235514599674e-06,
+      "loss": 1.3202,
+      "step": 12280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.383014279706727e-06,
+      "loss": 1.2223,
+      "step": 12281
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.382793044813778e-06,
+      "loss": 1.1855,
+      "step": 12282
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.38257180992083e-06,
+      "loss": 1.3143,
+      "step": 12283
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.382350575027883e-06,
+      "loss": 1.2888,
+      "step": 12284
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.382129340134935e-06,
+      "loss": 1.2848,
+      "step": 12285
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.381908105241986e-06,
+      "loss": 1.2255,
+      "step": 12286
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.381686870349039e-06,
+      "loss": 1.267,
+      "step": 12287
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.38146563545609e-06,
+      "loss": 1.1081,
+      "step": 12288
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.381244400563142e-06,
+      "loss": 1.2504,
+      "step": 12289
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.381023165670195e-06,
+      "loss": 1.219,
+      "step": 12290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.380801930777248e-06,
+      "loss": 1.2576,
+      "step": 12291
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.3805806958843e-06,
+      "loss": 1.1363,
+      "step": 12292
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.380359460991351e-06,
+      "loss": 1.2901,
+      "step": 12293
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.380138226098402e-06,
+      "loss": 1.1814,
+      "step": 12294
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.379916991205455e-06,
+      "loss": 1.2278,
+      "step": 12295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.379695756312507e-06,
+      "loss": 1.3347,
+      "step": 12296
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.37947452141956e-06,
+      "loss": 1.3037,
+      "step": 12297
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.379253286526613e-06,
+      "loss": 1.3492,
+      "step": 12298
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.379032051633663e-06,
+      "loss": 1.2596,
+      "step": 12299
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.378810816740716e-06,
+      "loss": 1.295,
+      "step": 12300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.378589581847769e-06,
+      "loss": 1.4085,
+      "step": 12301
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.37836834695482e-06,
+      "loss": 1.1363,
+      "step": 12302
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.378147112061872e-06,
+      "loss": 1.2617,
+      "step": 12303
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.377925877168923e-06,
+      "loss": 1.2336,
+      "step": 12304
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.377704642275976e-06,
+      "loss": 1.1333,
+      "step": 12305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.377483407383028e-06,
+      "loss": 1.1649,
+      "step": 12306
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.377262172490081e-06,
+      "loss": 1.2237,
+      "step": 12307
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.377040937597134e-06,
+      "loss": 1.2726,
+      "step": 12308
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.376819702704184e-06,
+      "loss": 1.206,
+      "step": 12309
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.376598467811235e-06,
+      "loss": 1.3053,
+      "step": 12310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.376377232918288e-06,
+      "loss": 1.336,
+      "step": 12311
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.37615599802534e-06,
+      "loss": 1.2364,
+      "step": 12312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.375934763132393e-06,
+      "loss": 1.2466,
+      "step": 12313
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.375713528239446e-06,
+      "loss": 1.3685,
+      "step": 12314
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.375492293346497e-06,
+      "loss": 1.1311,
+      "step": 12315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.37527105845355e-06,
+      "loss": 1.2378,
+      "step": 12316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.375049823560602e-06,
+      "loss": 1.1923,
+      "step": 12317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.374828588667653e-06,
+      "loss": 1.3201,
+      "step": 12318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.374607353774706e-06,
+      "loss": 1.2287,
+      "step": 12319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.374386118881758e-06,
+      "loss": 1.2813,
+      "step": 12320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.374164883988809e-06,
+      "loss": 1.2225,
+      "step": 12321
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.373943649095862e-06,
+      "loss": 1.3395,
+      "step": 12322
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.373722414202914e-06,
+      "loss": 1.2773,
+      "step": 12323
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.373501179309967e-06,
+      "loss": 1.2488,
+      "step": 12324
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.373279944417018e-06,
+      "loss": 1.2343,
+      "step": 12325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.37305870952407e-06,
+      "loss": 1.3361,
+      "step": 12326
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.372837474631121e-06,
+      "loss": 1.2418,
+      "step": 12327
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.372616239738174e-06,
+      "loss": 1.1244,
+      "step": 12328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.372395004845227e-06,
+      "loss": 1.3892,
+      "step": 12329
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.37217376995228e-06,
+      "loss": 1.2342,
+      "step": 12330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.371952535059332e-06,
+      "loss": 1.335,
+      "step": 12331
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.371731300166383e-06,
+      "loss": 1.3117,
+      "step": 12332
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.371510065273435e-06,
+      "loss": 1.1536,
+      "step": 12333
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.371288830380486e-06,
+      "loss": 1.3705,
+      "step": 12334
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.371067595487539e-06,
+      "loss": 1.1833,
+      "step": 12335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.370846360594591e-06,
+      "loss": 1.2241,
+      "step": 12336
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.370625125701642e-06,
+      "loss": 1.2188,
+      "step": 12337
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.370403890808695e-06,
+      "loss": 1.1576,
+      "step": 12338
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.370182655915748e-06,
+      "loss": 1.263,
+      "step": 12339
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.3699614210228e-06,
+      "loss": 1.2487,
+      "step": 12340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.369740186129851e-06,
+      "loss": 1.1983,
+      "step": 12341
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.369518951236904e-06,
+      "loss": 1.2304,
+      "step": 12342
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.369297716343955e-06,
+      "loss": 1.1863,
+      "step": 12343
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.369076481451007e-06,
+      "loss": 1.3598,
+      "step": 12344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.36885524655806e-06,
+      "loss": 1.1941,
+      "step": 12345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.368634011665112e-06,
+      "loss": 1.3393,
+      "step": 12346
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.368412776772165e-06,
+      "loss": 1.2907,
+      "step": 12347
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.368191541879216e-06,
+      "loss": 1.2888,
+      "step": 12348
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.367970306986269e-06,
+      "loss": 1.2234,
+      "step": 12349
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.36774907209332e-06,
+      "loss": 1.1561,
+      "step": 12350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.367527837200372e-06,
+      "loss": 1.3158,
+      "step": 12351
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.367306602307425e-06,
+      "loss": 1.312,
+      "step": 12352
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.367085367414477e-06,
+      "loss": 1.2173,
+      "step": 12353
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.366864132521528e-06,
+      "loss": 1.2476,
+      "step": 12354
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.366642897628581e-06,
+      "loss": 1.2873,
+      "step": 12355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.366421662735634e-06,
+      "loss": 1.311,
+      "step": 12356
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.366200427842686e-06,
+      "loss": 1.4294,
+      "step": 12357
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.365979192949737e-06,
+      "loss": 1.2048,
+      "step": 12358
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.365757958056788e-06,
+      "loss": 1.3055,
+      "step": 12359
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.36553672316384e-06,
+      "loss": 1.1948,
+      "step": 12360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.365315488270893e-06,
+      "loss": 1.2952,
+      "step": 12361
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.365094253377946e-06,
+      "loss": 1.18,
+      "step": 12362
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.364873018484998e-06,
+      "loss": 1.3459,
+      "step": 12363
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.364651783592051e-06,
+      "loss": 1.234,
+      "step": 12364
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.364430548699102e-06,
+      "loss": 1.2636,
+      "step": 12365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.364209313806153e-06,
+      "loss": 1.3656,
+      "step": 12366
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.363988078913205e-06,
+      "loss": 1.2489,
+      "step": 12367
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.363766844020258e-06,
+      "loss": 1.2046,
+      "step": 12368
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.36354560912731e-06,
+      "loss": 1.322,
+      "step": 12369
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.363324374234362e-06,
+      "loss": 1.1799,
+      "step": 12370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.363103139341414e-06,
+      "loss": 1.2348,
+      "step": 12371
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.362881904448467e-06,
+      "loss": 1.3042,
+      "step": 12372
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.36266066955552e-06,
+      "loss": 1.1858,
+      "step": 12373
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.36243943466257e-06,
+      "loss": 1.2409,
+      "step": 12374
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.362218199769623e-06,
+      "loss": 1.224,
+      "step": 12375
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.361996964876674e-06,
+      "loss": 1.2151,
+      "step": 12376
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.361775729983727e-06,
+      "loss": 1.2197,
+      "step": 12377
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.361554495090779e-06,
+      "loss": 1.2184,
+      "step": 12378
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.361333260197832e-06,
+      "loss": 1.1993,
+      "step": 12379
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.361112025304884e-06,
+      "loss": 1.2861,
+      "step": 12380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.360890790411935e-06,
+      "loss": 1.309,
+      "step": 12381
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.360669555518986e-06,
+      "loss": 1.2094,
+      "step": 12382
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.360448320626039e-06,
+      "loss": 1.1775,
+      "step": 12383
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.360227085733091e-06,
+      "loss": 1.3108,
+      "step": 12384
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.360005850840144e-06,
+      "loss": 1.2434,
+      "step": 12385
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.359784615947195e-06,
+      "loss": 1.2211,
+      "step": 12386
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.359563381054248e-06,
+      "loss": 1.2665,
+      "step": 12387
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.3593421461613e-06,
+      "loss": 1.1926,
+      "step": 12388
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.359120911268353e-06,
+      "loss": 1.2631,
+      "step": 12389
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.358899676375404e-06,
+      "loss": 1.4234,
+      "step": 12390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.358678441482456e-06,
+      "loss": 1.2269,
+      "step": 12391
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.358457206589507e-06,
+      "loss": 1.2926,
+      "step": 12392
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.35823597169656e-06,
+      "loss": 1.3653,
+      "step": 12393
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.358014736803612e-06,
+      "loss": 1.3254,
+      "step": 12394
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.357793501910665e-06,
+      "loss": 1.294,
+      "step": 12395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.357572267017718e-06,
+      "loss": 1.0852,
+      "step": 12396
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.357351032124769e-06,
+      "loss": 1.3076,
+      "step": 12397
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.35712979723182e-06,
+      "loss": 1.1285,
+      "step": 12398
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.356908562338872e-06,
+      "loss": 1.2368,
+      "step": 12399
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.356687327445925e-06,
+      "loss": 1.1927,
+      "step": 12400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.356466092552977e-06,
+      "loss": 1.2064,
+      "step": 12401
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.35624485766003e-06,
+      "loss": 1.2168,
+      "step": 12402
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.356023622767081e-06,
+      "loss": 1.2445,
+      "step": 12403
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.355802387874133e-06,
+      "loss": 1.3059,
+      "step": 12404
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.355581152981186e-06,
+      "loss": 1.3468,
+      "step": 12405
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.355359918088237e-06,
+      "loss": 1.2748,
+      "step": 12406
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.35513868319529e-06,
+      "loss": 1.1921,
+      "step": 12407
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.35491744830234e-06,
+      "loss": 1.2261,
+      "step": 12408
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.354696213409393e-06,
+      "loss": 1.1797,
+      "step": 12409
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.354474978516446e-06,
+      "loss": 1.288,
+      "step": 12410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.354253743623498e-06,
+      "loss": 1.2356,
+      "step": 12411
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.354032508730551e-06,
+      "loss": 1.1742,
+      "step": 12412
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.353811273837602e-06,
+      "loss": 1.233,
+      "step": 12413
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.353590038944655e-06,
+      "loss": 1.1608,
+      "step": 12414
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.353368804051705e-06,
+      "loss": 1.1275,
+      "step": 12415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.353147569158758e-06,
+      "loss": 1.2675,
+      "step": 12416
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.35292633426581e-06,
+      "loss": 1.2875,
+      "step": 12417
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.352705099372863e-06,
+      "loss": 1.2718,
+      "step": 12418
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.352483864479914e-06,
+      "loss": 1.282,
+      "step": 12419
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.352262629586967e-06,
+      "loss": 1.2292,
+      "step": 12420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.35204139469402e-06,
+      "loss": 1.2201,
+      "step": 12421
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.35182015980107e-06,
+      "loss": 1.2363,
+      "step": 12422
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.351598924908123e-06,
+      "loss": 1.343,
+      "step": 12423
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.351377690015176e-06,
+      "loss": 1.2595,
+      "step": 12424
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.351156455122226e-06,
+      "loss": 1.1982,
+      "step": 12425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.350935220229279e-06,
+      "loss": 1.291,
+      "step": 12426
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.350713985336332e-06,
+      "loss": 1.2232,
+      "step": 12427
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.350492750443384e-06,
+      "loss": 1.2171,
+      "step": 12428
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.350271515550435e-06,
+      "loss": 1.1527,
+      "step": 12429
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.350050280657488e-06,
+      "loss": 1.222,
+      "step": 12430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.349829045764539e-06,
+      "loss": 1.1155,
+      "step": 12431
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.349607810871591e-06,
+      "loss": 1.2214,
+      "step": 12432
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.349386575978644e-06,
+      "loss": 1.2495,
+      "step": 12433
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.349165341085697e-06,
+      "loss": 1.2367,
+      "step": 12434
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.34894410619275e-06,
+      "loss": 1.1797,
+      "step": 12435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.3487228712998e-06,
+      "loss": 1.3176,
+      "step": 12436
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.348501636406853e-06,
+      "loss": 1.2797,
+      "step": 12437
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.348280401513904e-06,
+      "loss": 1.2277,
+      "step": 12438
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.348059166620956e-06,
+      "loss": 1.1557,
+      "step": 12439
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.347837931728009e-06,
+      "loss": 1.2114,
+      "step": 12440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.34761669683506e-06,
+      "loss": 1.1405,
+      "step": 12441
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.347395461942112e-06,
+      "loss": 1.3025,
+      "step": 12442
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.347174227049165e-06,
+      "loss": 1.3186,
+      "step": 12443
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.346952992156218e-06,
+      "loss": 1.2141,
+      "step": 12444
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.34673175726327e-06,
+      "loss": 1.3188,
+      "step": 12445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.346510522370321e-06,
+      "loss": 1.2305,
+      "step": 12446
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.346289287477372e-06,
+      "loss": 1.184,
+      "step": 12447
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.346068052584425e-06,
+      "loss": 1.2939,
+      "step": 12448
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.345846817691477e-06,
+      "loss": 1.2112,
+      "step": 12449
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.34562558279853e-06,
+      "loss": 1.166,
+      "step": 12450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.345404347905583e-06,
+      "loss": 1.1995,
+      "step": 12451
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.345183113012633e-06,
+      "loss": 1.3019,
+      "step": 12452
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.344961878119686e-06,
+      "loss": 1.2524,
+      "step": 12453
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.344740643226737e-06,
+      "loss": 1.2306,
+      "step": 12454
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.34451940833379e-06,
+      "loss": 1.194,
+      "step": 12455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.344298173440842e-06,
+      "loss": 1.1046,
+      "step": 12456
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.344076938547895e-06,
+      "loss": 1.2489,
+      "step": 12457
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.343855703654946e-06,
+      "loss": 1.2638,
+      "step": 12458
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.343634468761998e-06,
+      "loss": 1.1914,
+      "step": 12459
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.343413233869051e-06,
+      "loss": 1.2261,
+      "step": 12460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.343191998976104e-06,
+      "loss": 1.1858,
+      "step": 12461
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.342970764083154e-06,
+      "loss": 1.3445,
+      "step": 12462
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.342749529190205e-06,
+      "loss": 1.2622,
+      "step": 12463
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.342528294297258e-06,
+      "loss": 1.2368,
+      "step": 12464
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.34230705940431e-06,
+      "loss": 1.2348,
+      "step": 12465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.342085824511363e-06,
+      "loss": 1.2796,
+      "step": 12466
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.341864589618416e-06,
+      "loss": 1.2924,
+      "step": 12467
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.341643354725468e-06,
+      "loss": 1.1579,
+      "step": 12468
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.34142211983252e-06,
+      "loss": 1.3397,
+      "step": 12469
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.34120088493957e-06,
+      "loss": 1.3469,
+      "step": 12470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.340979650046623e-06,
+      "loss": 1.1653,
+      "step": 12471
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.340758415153676e-06,
+      "loss": 1.275,
+      "step": 12472
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.340537180260728e-06,
+      "loss": 1.3743,
+      "step": 12473
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.340315945367779e-06,
+      "loss": 1.2475,
+      "step": 12474
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.340094710474832e-06,
+      "loss": 1.1457,
+      "step": 12475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.339873475581884e-06,
+      "loss": 1.2572,
+      "step": 12476
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.339652240688937e-06,
+      "loss": 1.2476,
+      "step": 12477
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.339431005795988e-06,
+      "loss": 1.3885,
+      "step": 12478
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.339209770903039e-06,
+      "loss": 1.2788,
+      "step": 12479
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.338988536010091e-06,
+      "loss": 1.237,
+      "step": 12480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.338767301117144e-06,
+      "loss": 1.2092,
+      "step": 12481
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.338546066224197e-06,
+      "loss": 1.2926,
+      "step": 12482
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.33832483133125e-06,
+      "loss": 1.2839,
+      "step": 12483
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.338103596438302e-06,
+      "loss": 1.2668,
+      "step": 12484
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.337882361545353e-06,
+      "loss": 1.1695,
+      "step": 12485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.337661126652404e-06,
+      "loss": 1.2275,
+      "step": 12486
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.337439891759456e-06,
+      "loss": 1.3046,
+      "step": 12487
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.337218656866509e-06,
+      "loss": 1.1884,
+      "step": 12488
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.336997421973561e-06,
+      "loss": 1.1206,
+      "step": 12489
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.336776187080612e-06,
+      "loss": 1.1887,
+      "step": 12490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.336554952187665e-06,
+      "loss": 1.2045,
+      "step": 12491
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.336333717294718e-06,
+      "loss": 1.2176,
+      "step": 12492
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.33611248240177e-06,
+      "loss": 1.218,
+      "step": 12493
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.335891247508821e-06,
+      "loss": 1.1717,
+      "step": 12494
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.335670012615874e-06,
+      "loss": 1.2291,
+      "step": 12495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.335448777722925e-06,
+      "loss": 1.339,
+      "step": 12496
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.335227542829977e-06,
+      "loss": 1.3889,
+      "step": 12497
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.33500630793703e-06,
+      "loss": 1.2243,
+      "step": 12498
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.334785073044082e-06,
+      "loss": 1.2021,
+      "step": 12499
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.334563838151135e-06,
+      "loss": 1.2683,
+      "step": 12500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.334342603258186e-06,
+      "loss": 1.1501,
+      "step": 12501
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.334121368365239e-06,
+      "loss": 1.2294,
+      "step": 12502
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.33390013347229e-06,
+      "loss": 1.2585,
+      "step": 12503
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.333678898579342e-06,
+      "loss": 1.3218,
+      "step": 12504
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.333457663686395e-06,
+      "loss": 1.3427,
+      "step": 12505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.333236428793447e-06,
+      "loss": 1.2259,
+      "step": 12506
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.333015193900498e-06,
+      "loss": 1.2707,
+      "step": 12507
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.332793959007551e-06,
+      "loss": 1.2749,
+      "step": 12508
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.332572724114604e-06,
+      "loss": 1.1639,
+      "step": 12509
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.332351489221654e-06,
+      "loss": 1.1826,
+      "step": 12510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.332130254328707e-06,
+      "loss": 1.2088,
+      "step": 12511
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.331909019435758e-06,
+      "loss": 1.3289,
+      "step": 12512
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.33168778454281e-06,
+      "loss": 1.2442,
+      "step": 12513
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.331466549649863e-06,
+      "loss": 1.1688,
+      "step": 12514
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.331245314756916e-06,
+      "loss": 1.2866,
+      "step": 12515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.331024079863968e-06,
+      "loss": 1.1752,
+      "step": 12516
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.33080284497102e-06,
+      "loss": 1.1263,
+      "step": 12517
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.330581610078072e-06,
+      "loss": 1.2122,
+      "step": 12518
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.330360375185123e-06,
+      "loss": 1.2267,
+      "step": 12519
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.330139140292175e-06,
+      "loss": 1.2743,
+      "step": 12520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.329917905399228e-06,
+      "loss": 1.2349,
+      "step": 12521
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.32969667050628e-06,
+      "loss": 1.2315,
+      "step": 12522
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.329475435613332e-06,
+      "loss": 1.1986,
+      "step": 12523
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.329254200720384e-06,
+      "loss": 1.2545,
+      "step": 12524
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.329032965827437e-06,
+      "loss": 1.1351,
+      "step": 12525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.328811730934488e-06,
+      "loss": 1.2453,
+      "step": 12526
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.32859049604154e-06,
+      "loss": 1.1598,
+      "step": 12527
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.328369261148593e-06,
+      "loss": 1.2203,
+      "step": 12528
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.328148026255644e-06,
+      "loss": 1.2891,
+      "step": 12529
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.327926791362697e-06,
+      "loss": 1.1826,
+      "step": 12530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.327705556469749e-06,
+      "loss": 1.2899,
+      "step": 12531
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.327484321576802e-06,
+      "loss": 1.2016,
+      "step": 12532
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.327263086683854e-06,
+      "loss": 1.3183,
+      "step": 12533
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.327041851790905e-06,
+      "loss": 1.2642,
+      "step": 12534
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.326820616897956e-06,
+      "loss": 1.1875,
+      "step": 12535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.326599382005009e-06,
+      "loss": 1.2403,
+      "step": 12536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.326378147112061e-06,
+      "loss": 1.3567,
+      "step": 12537
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.326156912219114e-06,
+      "loss": 1.2997,
+      "step": 12538
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.325935677326167e-06,
+      "loss": 1.2058,
+      "step": 12539
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.325714442433218e-06,
+      "loss": 1.2131,
+      "step": 12540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.32549320754027e-06,
+      "loss": 1.265,
+      "step": 12541
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.325271972647321e-06,
+      "loss": 1.2298,
+      "step": 12542
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.325050737754374e-06,
+      "loss": 1.1893,
+      "step": 12543
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.324829502861426e-06,
+      "loss": 1.4018,
+      "step": 12544
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.324608267968477e-06,
+      "loss": 1.2609,
+      "step": 12545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.32438703307553e-06,
+      "loss": 1.3997,
+      "step": 12546
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.324165798182582e-06,
+      "loss": 1.1255,
+      "step": 12547
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.323944563289635e-06,
+      "loss": 1.2084,
+      "step": 12548
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.323723328396688e-06,
+      "loss": 1.2932,
+      "step": 12549
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.323502093503739e-06,
+      "loss": 1.2501,
+      "step": 12550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.32328085861079e-06,
+      "loss": 1.2927,
+      "step": 12551
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.323059623717842e-06,
+      "loss": 1.2457,
+      "step": 12552
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.322838388824895e-06,
+      "loss": 1.2001,
+      "step": 12553
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.322617153931947e-06,
+      "loss": 1.2872,
+      "step": 12554
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.322395919039e-06,
+      "loss": 1.1298,
+      "step": 12555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.322174684146051e-06,
+      "loss": 1.1567,
+      "step": 12556
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.321953449253103e-06,
+      "loss": 1.3356,
+      "step": 12557
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.321732214360154e-06,
+      "loss": 1.3932,
+      "step": 12558
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.321510979467207e-06,
+      "loss": 1.1798,
+      "step": 12559
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.32128974457426e-06,
+      "loss": 1.2324,
+      "step": 12560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.321068509681312e-06,
+      "loss": 1.3095,
+      "step": 12561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.320847274788363e-06,
+      "loss": 1.3562,
+      "step": 12562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.320626039895416e-06,
+      "loss": 1.2668,
+      "step": 12563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.320404805002468e-06,
+      "loss": 1.2153,
+      "step": 12564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.320183570109521e-06,
+      "loss": 1.1845,
+      "step": 12565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.319962335216572e-06,
+      "loss": 1.2974,
+      "step": 12566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.319741100323623e-06,
+      "loss": 1.2846,
+      "step": 12567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.319519865430675e-06,
+      "loss": 1.2757,
+      "step": 12568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.319298630537728e-06,
+      "loss": 1.2644,
+      "step": 12569
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.31907739564478e-06,
+      "loss": 1.1338,
+      "step": 12570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.318856160751833e-06,
+      "loss": 1.2963,
+      "step": 12571
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.318634925858886e-06,
+      "loss": 1.1722,
+      "step": 12572
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.318413690965937e-06,
+      "loss": 1.2675,
+      "step": 12573
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.318192456072988e-06,
+      "loss": 1.2514,
+      "step": 12574
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.31797122118004e-06,
+      "loss": 1.3265,
+      "step": 12575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.317749986287093e-06,
+      "loss": 1.1625,
+      "step": 12576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.317528751394146e-06,
+      "loss": 1.4136,
+      "step": 12577
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.317307516501196e-06,
+      "loss": 1.2841,
+      "step": 12578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.317086281608249e-06,
+      "loss": 1.1884,
+      "step": 12579
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.316865046715302e-06,
+      "loss": 1.2594,
+      "step": 12580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.316643811822354e-06,
+      "loss": 1.1919,
+      "step": 12581
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.316422576929405e-06,
+      "loss": 1.2945,
+      "step": 12582
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.316201342036456e-06,
+      "loss": 1.2067,
+      "step": 12583
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.315980107143509e-06,
+      "loss": 1.2212,
+      "step": 12584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.315758872250561e-06,
+      "loss": 1.2428,
+      "step": 12585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.315537637357614e-06,
+      "loss": 1.1573,
+      "step": 12586
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.315316402464667e-06,
+      "loss": 1.2486,
+      "step": 12587
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.31509516757172e-06,
+      "loss": 1.2585,
+      "step": 12588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.31487393267877e-06,
+      "loss": 1.2539,
+      "step": 12589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.314652697785823e-06,
+      "loss": 1.3177,
+      "step": 12590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.314431462892874e-06,
+      "loss": 1.3043,
+      "step": 12591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.314210227999926e-06,
+      "loss": 1.3443,
+      "step": 12592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.313988993106979e-06,
+      "loss": 1.2287,
+      "step": 12593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.31376775821403e-06,
+      "loss": 1.3623,
+      "step": 12594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.313546523321082e-06,
+      "loss": 1.3199,
+      "step": 12595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.313325288428135e-06,
+      "loss": 1.2669,
+      "step": 12596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.313104053535188e-06,
+      "loss": 1.2279,
+      "step": 12597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.312882818642239e-06,
+      "loss": 1.2104,
+      "step": 12598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.312661583749291e-06,
+      "loss": 1.122,
+      "step": 12599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.312440348856342e-06,
+      "loss": 1.2431,
+      "step": 12600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.312219113963395e-06,
+      "loss": 1.2171,
+      "step": 12601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.311997879070447e-06,
+      "loss": 1.2396,
+      "step": 12602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.3117766441775e-06,
+      "loss": 1.2855,
+      "step": 12603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.311555409284553e-06,
+      "loss": 1.1454,
+      "step": 12604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.311334174391603e-06,
+      "loss": 1.2349,
+      "step": 12605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.311112939498656e-06,
+      "loss": 1.2734,
+      "step": 12606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.310891704605707e-06,
+      "loss": 1.2467,
+      "step": 12607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.31067046971276e-06,
+      "loss": 1.297,
+      "step": 12608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.310449234819812e-06,
+      "loss": 1.3314,
+      "step": 12609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.310227999926865e-06,
+      "loss": 1.1835,
+      "step": 12610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.310006765033916e-06,
+      "loss": 1.1692,
+      "step": 12611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.309785530140968e-06,
+      "loss": 1.0548,
+      "step": 12612
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.309564295248021e-06,
+      "loss": 1.19,
+      "step": 12613
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.309343060355072e-06,
+      "loss": 1.2431,
+      "step": 12614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.309121825462124e-06,
+      "loss": 1.3119,
+      "step": 12615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.308900590569175e-06,
+      "loss": 1.3526,
+      "step": 12616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.308679355676228e-06,
+      "loss": 1.359,
+      "step": 12617
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.30845812078328e-06,
+      "loss": 1.3411,
+      "step": 12618
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.308236885890333e-06,
+      "loss": 1.2866,
+      "step": 12619
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.308015650997386e-06,
+      "loss": 1.2478,
+      "step": 12620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.307794416104438e-06,
+      "loss": 1.2057,
+      "step": 12621
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.30757318121149e-06,
+      "loss": 1.129,
+      "step": 12622
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.30735194631854e-06,
+      "loss": 1.2537,
+      "step": 12623
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.307130711425593e-06,
+      "loss": 1.2407,
+      "step": 12624
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.306909476532646e-06,
+      "loss": 1.1934,
+      "step": 12625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.306688241639698e-06,
+      "loss": 1.2078,
+      "step": 12626
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.306467006746749e-06,
+      "loss": 1.2282,
+      "step": 12627
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.306245771853802e-06,
+      "loss": 1.3524,
+      "step": 12628
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.306024536960854e-06,
+      "loss": 1.2485,
+      "step": 12629
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.305803302067905e-06,
+      "loss": 1.3045,
+      "step": 12630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.305582067174958e-06,
+      "loss": 1.2112,
+      "step": 12631
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.30536083228201e-06,
+      "loss": 1.3062,
+      "step": 12632
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.305139597389061e-06,
+      "loss": 1.1445,
+      "step": 12633
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.304918362496114e-06,
+      "loss": 1.2145,
+      "step": 12634
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.304697127603167e-06,
+      "loss": 1.2165,
+      "step": 12635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.30447589271022e-06,
+      "loss": 1.4629,
+      "step": 12636
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.304254657817272e-06,
+      "loss": 1.1795,
+      "step": 12637
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.304033422924323e-06,
+      "loss": 1.2354,
+      "step": 12638
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.303812188031374e-06,
+      "loss": 1.1098,
+      "step": 12639
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.303590953138426e-06,
+      "loss": 1.1758,
+      "step": 12640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.303369718245479e-06,
+      "loss": 1.2103,
+      "step": 12641
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.303148483352531e-06,
+      "loss": 1.3112,
+      "step": 12642
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.302927248459584e-06,
+      "loss": 1.239,
+      "step": 12643
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.302706013566635e-06,
+      "loss": 1.3019,
+      "step": 12644
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.302484778673688e-06,
+      "loss": 1.3072,
+      "step": 12645
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.302263543780738e-06,
+      "loss": 1.2346,
+      "step": 12646
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.302042308887791e-06,
+      "loss": 1.2232,
+      "step": 12647
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.301821073994844e-06,
+      "loss": 1.1834,
+      "step": 12648
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.301599839101895e-06,
+      "loss": 1.2871,
+      "step": 12649
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.301378604208947e-06,
+      "loss": 1.2949,
+      "step": 12650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.301157369316e-06,
+      "loss": 1.2228,
+      "step": 12651
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.300936134423052e-06,
+      "loss": 1.1193,
+      "step": 12652
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.300714899530105e-06,
+      "loss": 1.2695,
+      "step": 12653
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.300493664637156e-06,
+      "loss": 1.2243,
+      "step": 12654
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.300272429744207e-06,
+      "loss": 1.2682,
+      "step": 12655
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.30005119485126e-06,
+      "loss": 1.2304,
+      "step": 12656
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.299829959958312e-06,
+      "loss": 1.2316,
+      "step": 12657
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.299608725065365e-06,
+      "loss": 1.2305,
+      "step": 12658
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.299387490172417e-06,
+      "loss": 1.1573,
+      "step": 12659
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.299166255279468e-06,
+      "loss": 1.3189,
+      "step": 12660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.298945020386521e-06,
+      "loss": 1.2129,
+      "step": 12661
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.298723785493572e-06,
+      "loss": 1.2791,
+      "step": 12662
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.298502550600624e-06,
+      "loss": 1.3527,
+      "step": 12663
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.298281315707677e-06,
+      "loss": 1.1944,
+      "step": 12664
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.29806008081473e-06,
+      "loss": 1.2176,
+      "step": 12665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.29783884592178e-06,
+      "loss": 1.2378,
+      "step": 12666
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.297617611028833e-06,
+      "loss": 1.1828,
+      "step": 12667
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.297396376135886e-06,
+      "loss": 1.2648,
+      "step": 12668
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.297175141242938e-06,
+      "loss": 1.3492,
+      "step": 12669
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.29695390634999e-06,
+      "loss": 1.2049,
+      "step": 12670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.29673267145704e-06,
+      "loss": 1.2073,
+      "step": 12671
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.296511436564093e-06,
+      "loss": 1.0573,
+      "step": 12672
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.296290201671145e-06,
+      "loss": 1.0996,
+      "step": 12673
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.296068966778198e-06,
+      "loss": 1.3357,
+      "step": 12674
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.29584773188525e-06,
+      "loss": 1.166,
+      "step": 12675
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.295626496992302e-06,
+      "loss": 1.1472,
+      "step": 12676
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.295405262099354e-06,
+      "loss": 1.2571,
+      "step": 12677
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.295184027206407e-06,
+      "loss": 1.2752,
+      "step": 12678
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.294962792313458e-06,
+      "loss": 1.1965,
+      "step": 12679
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.29474155742051e-06,
+      "loss": 1.2349,
+      "step": 12680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.294520322527563e-06,
+      "loss": 1.3134,
+      "step": 12681
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.294299087634614e-06,
+      "loss": 1.3182,
+      "step": 12682
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.294077852741666e-06,
+      "loss": 1.2618,
+      "step": 12683
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.293856617848719e-06,
+      "loss": 1.2815,
+      "step": 12684
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.293635382955772e-06,
+      "loss": 1.2336,
+      "step": 12685
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.293414148062823e-06,
+      "loss": 1.2196,
+      "step": 12686
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.293192913169874e-06,
+      "loss": 1.1755,
+      "step": 12687
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.292971678276926e-06,
+      "loss": 1.1633,
+      "step": 12688
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.292750443383979e-06,
+      "loss": 1.308,
+      "step": 12689
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.292529208491031e-06,
+      "loss": 1.1351,
+      "step": 12690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.292307973598084e-06,
+      "loss": 1.3467,
+      "step": 12691
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.292086738705137e-06,
+      "loss": 1.2776,
+      "step": 12692
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.291865503812188e-06,
+      "loss": 1.1981,
+      "step": 12693
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.29164426891924e-06,
+      "loss": 1.3873,
+      "step": 12694
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.291423034026291e-06,
+      "loss": 1.3141,
+      "step": 12695
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.291201799133344e-06,
+      "loss": 1.2544,
+      "step": 12696
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.290980564240396e-06,
+      "loss": 1.2114,
+      "step": 12697
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.290759329347447e-06,
+      "loss": 1.243,
+      "step": 12698
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.2905380944545e-06,
+      "loss": 1.2458,
+      "step": 12699
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.290316859561552e-06,
+      "loss": 1.1198,
+      "step": 12700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.290095624668605e-06,
+      "loss": 1.3041,
+      "step": 12701
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.289874389775656e-06,
+      "loss": 1.1756,
+      "step": 12702
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.289653154882709e-06,
+      "loss": 1.3268,
+      "step": 12703
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.28943191998976e-06,
+      "loss": 1.2839,
+      "step": 12704
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.289210685096812e-06,
+      "loss": 1.205,
+      "step": 12705
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.288989450203865e-06,
+      "loss": 1.206,
+      "step": 12706
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.288768215310917e-06,
+      "loss": 1.3093,
+      "step": 12707
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.28854698041797e-06,
+      "loss": 1.1561,
+      "step": 12708
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.288325745525021e-06,
+      "loss": 1.2171,
+      "step": 12709
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.288104510632073e-06,
+      "loss": 1.2942,
+      "step": 12710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.287883275739124e-06,
+      "loss": 1.2116,
+      "step": 12711
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.287662040846177e-06,
+      "loss": 1.2223,
+      "step": 12712
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.28744080595323e-06,
+      "loss": 1.1357,
+      "step": 12713
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.287219571060282e-06,
+      "loss": 1.1351,
+      "step": 12714
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.286998336167333e-06,
+      "loss": 1.1336,
+      "step": 12715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.286777101274386e-06,
+      "loss": 1.2518,
+      "step": 12716
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.286555866381438e-06,
+      "loss": 1.1757,
+      "step": 12717
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.28633463148849e-06,
+      "loss": 1.3303,
+      "step": 12718
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.286113396595542e-06,
+      "loss": 1.1514,
+      "step": 12719
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.285892161702593e-06,
+      "loss": 1.2148,
+      "step": 12720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.285670926809645e-06,
+      "loss": 1.2105,
+      "step": 12721
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.285449691916698e-06,
+      "loss": 1.3619,
+      "step": 12722
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.28522845702375e-06,
+      "loss": 1.2741,
+      "step": 12723
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.285007222130803e-06,
+      "loss": 1.3944,
+      "step": 12724
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.284785987237856e-06,
+      "loss": 1.2442,
+      "step": 12725
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.284564752344907e-06,
+      "loss": 1.3184,
+      "step": 12726
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.284343517451958e-06,
+      "loss": 1.1496,
+      "step": 12727
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.28412228255901e-06,
+      "loss": 1.1847,
+      "step": 12728
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.283901047666063e-06,
+      "loss": 1.1942,
+      "step": 12729
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.283679812773116e-06,
+      "loss": 1.2198,
+      "step": 12730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.283458577880166e-06,
+      "loss": 1.1732,
+      "step": 12731
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.283237342987219e-06,
+      "loss": 1.234,
+      "step": 12732
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.283016108094272e-06,
+      "loss": 1.1828,
+      "step": 12733
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.282794873201323e-06,
+      "loss": 1.1945,
+      "step": 12734
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.282573638308375e-06,
+      "loss": 1.1812,
+      "step": 12735
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.282352403415428e-06,
+      "loss": 1.3813,
+      "step": 12736
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.282131168522479e-06,
+      "loss": 1.2366,
+      "step": 12737
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.281909933629531e-06,
+      "loss": 1.1729,
+      "step": 12738
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.281688698736584e-06,
+      "loss": 1.3223,
+      "step": 12739
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.281467463843637e-06,
+      "loss": 1.2232,
+      "step": 12740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.28124622895069e-06,
+      "loss": 1.2567,
+      "step": 12741
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.28102499405774e-06,
+      "loss": 1.3387,
+      "step": 12742
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.280803759164791e-06,
+      "loss": 1.2332,
+      "step": 12743
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.280582524271844e-06,
+      "loss": 1.1313,
+      "step": 12744
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.280361289378896e-06,
+      "loss": 1.2891,
+      "step": 12745
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.280140054485949e-06,
+      "loss": 1.3037,
+      "step": 12746
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.279918819593001e-06,
+      "loss": 1.1842,
+      "step": 12747
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.279697584700052e-06,
+      "loss": 1.2123,
+      "step": 12748
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.279476349807105e-06,
+      "loss": 1.2068,
+      "step": 12749
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.279255114914156e-06,
+      "loss": 1.1897,
+      "step": 12750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.279033880021209e-06,
+      "loss": 1.3017,
+      "step": 12751
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.278812645128261e-06,
+      "loss": 1.3939,
+      "step": 12752
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.278591410235312e-06,
+      "loss": 1.3437,
+      "step": 12753
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.278370175342365e-06,
+      "loss": 1.284,
+      "step": 12754
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.278148940449417e-06,
+      "loss": 1.2021,
+      "step": 12755
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.27792770555647e-06,
+      "loss": 1.1443,
+      "step": 12756
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.277706470663523e-06,
+      "loss": 1.2348,
+      "step": 12757
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.277485235770573e-06,
+      "loss": 1.2538,
+      "step": 12758
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.277264000877624e-06,
+      "loss": 1.1375,
+      "step": 12759
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.277042765984677e-06,
+      "loss": 1.3356,
+      "step": 12760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.27682153109173e-06,
+      "loss": 1.2257,
+      "step": 12761
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.276600296198782e-06,
+      "loss": 1.3143,
+      "step": 12762
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.276379061305835e-06,
+      "loss": 1.21,
+      "step": 12763
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.276157826412886e-06,
+      "loss": 1.3653,
+      "step": 12764
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.275936591519938e-06,
+      "loss": 1.2804,
+      "step": 12765
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.275715356626991e-06,
+      "loss": 1.2093,
+      "step": 12766
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.275494121734042e-06,
+      "loss": 1.2535,
+      "step": 12767
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.275272886841094e-06,
+      "loss": 1.3112,
+      "step": 12768
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.275051651948145e-06,
+      "loss": 1.2653,
+      "step": 12769
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.274830417055198e-06,
+      "loss": 1.1511,
+      "step": 12770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.27460918216225e-06,
+      "loss": 1.3074,
+      "step": 12771
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.274387947269303e-06,
+      "loss": 1.2396,
+      "step": 12772
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.274166712376356e-06,
+      "loss": 1.2636,
+      "step": 12773
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.273945477483407e-06,
+      "loss": 1.1748,
+      "step": 12774
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.273724242590458e-06,
+      "loss": 1.2392,
+      "step": 12775
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.27350300769751e-06,
+      "loss": 1.104,
+      "step": 12776
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.273281772804563e-06,
+      "loss": 1.1696,
+      "step": 12777
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.273060537911615e-06,
+      "loss": 1.1869,
+      "step": 12778
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.272839303018668e-06,
+      "loss": 1.2071,
+      "step": 12779
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.272618068125719e-06,
+      "loss": 1.2081,
+      "step": 12780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.272396833232772e-06,
+      "loss": 1.1545,
+      "step": 12781
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.272175598339824e-06,
+      "loss": 1.2886,
+      "step": 12782
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.271954363446875e-06,
+      "loss": 1.2437,
+      "step": 12783
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.271733128553928e-06,
+      "loss": 1.151,
+      "step": 12784
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.27151189366098e-06,
+      "loss": 1.1879,
+      "step": 12785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.271290658768031e-06,
+      "loss": 1.1398,
+      "step": 12786
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.271069423875084e-06,
+      "loss": 1.3167,
+      "step": 12787
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.270848188982137e-06,
+      "loss": 1.2255,
+      "step": 12788
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.270626954089189e-06,
+      "loss": 1.2108,
+      "step": 12789
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.27040571919624e-06,
+      "loss": 1.3549,
+      "step": 12790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.270184484303291e-06,
+      "loss": 1.3028,
+      "step": 12791
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.269963249410344e-06,
+      "loss": 1.3362,
+      "step": 12792
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.269742014517396e-06,
+      "loss": 1.1938,
+      "step": 12793
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.269520779624449e-06,
+      "loss": 1.1901,
+      "step": 12794
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.269299544731501e-06,
+      "loss": 1.3034,
+      "step": 12795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.269078309838554e-06,
+      "loss": 1.2935,
+      "step": 12796
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.268857074945605e-06,
+      "loss": 1.2916,
+      "step": 12797
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.268635840052658e-06,
+      "loss": 1.2347,
+      "step": 12798
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.268414605159708e-06,
+      "loss": 1.2593,
+      "step": 12799
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.268193370266761e-06,
+      "loss": 1.2297,
+      "step": 12800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.267972135373814e-06,
+      "loss": 1.1996,
+      "step": 12801
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.267750900480865e-06,
+      "loss": 1.2519,
+      "step": 12802
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.267529665587917e-06,
+      "loss": 1.2412,
+      "step": 12803
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.26730843069497e-06,
+      "loss": 1.3034,
+      "step": 12804
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.267087195802022e-06,
+      "loss": 1.3086,
+      "step": 12805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.266865960909073e-06,
+      "loss": 1.2211,
+      "step": 12806
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.266644726016126e-06,
+      "loss": 1.2916,
+      "step": 12807
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.266423491123177e-06,
+      "loss": 1.3105,
+      "step": 12808
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.26620225623023e-06,
+      "loss": 1.3526,
+      "step": 12809
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.265981021337282e-06,
+      "loss": 1.3298,
+      "step": 12810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.265759786444335e-06,
+      "loss": 1.2838,
+      "step": 12811
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.265538551551387e-06,
+      "loss": 1.1806,
+      "step": 12812
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.265317316658438e-06,
+      "loss": 1.2755,
+      "step": 12813
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.265096081765491e-06,
+      "loss": 1.2544,
+      "step": 12814
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.264874846872542e-06,
+      "loss": 1.1487,
+      "step": 12815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.264653611979594e-06,
+      "loss": 1.4222,
+      "step": 12816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.264432377086647e-06,
+      "loss": 1.2586,
+      "step": 12817
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.2642111421937e-06,
+      "loss": 1.2681,
+      "step": 12818
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.26398990730075e-06,
+      "loss": 1.2257,
+      "step": 12819
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.263768672407803e-06,
+      "loss": 1.2525,
+      "step": 12820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.263547437514856e-06,
+      "loss": 1.1604,
+      "step": 12821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.263326202621907e-06,
+      "loss": 1.1668,
+      "step": 12822
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.26310496772896e-06,
+      "loss": 1.2973,
+      "step": 12823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.26288373283601e-06,
+      "loss": 1.1134,
+      "step": 12824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.262662497943063e-06,
+      "loss": 1.2654,
+      "step": 12825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.262441263050115e-06,
+      "loss": 1.1844,
+      "step": 12826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.262220028157168e-06,
+      "loss": 1.4092,
+      "step": 12827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.26199879326422e-06,
+      "loss": 1.3606,
+      "step": 12828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.261777558371273e-06,
+      "loss": 1.2269,
+      "step": 12829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.261556323478324e-06,
+      "loss": 1.0811,
+      "step": 12830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.261335088585375e-06,
+      "loss": 1.2599,
+      "step": 12831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.261113853692428e-06,
+      "loss": 1.2858,
+      "step": 12832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.26089261879948e-06,
+      "loss": 1.2974,
+      "step": 12833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.260671383906533e-06,
+      "loss": 1.1621,
+      "step": 12834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.260450149013584e-06,
+      "loss": 1.2742,
+      "step": 12835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.260228914120636e-06,
+      "loss": 1.2579,
+      "step": 12836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.260007679227689e-06,
+      "loss": 1.2635,
+      "step": 12837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.25978644433474e-06,
+      "loss": 1.2532,
+      "step": 12838
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.259565209441793e-06,
+      "loss": 1.2217,
+      "step": 12839
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.259343974548845e-06,
+      "loss": 1.2075,
+      "step": 12840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.259122739655896e-06,
+      "loss": 1.2242,
+      "step": 12841
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.258901504762949e-06,
+      "loss": 1.1954,
+      "step": 12842
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.258680269870001e-06,
+      "loss": 1.1813,
+      "step": 12843
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.258459034977054e-06,
+      "loss": 1.1383,
+      "step": 12844
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.258237800084107e-06,
+      "loss": 1.296,
+      "step": 12845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.258016565191158e-06,
+      "loss": 1.2329,
+      "step": 12846
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.257795330298208e-06,
+      "loss": 1.2524,
+      "step": 12847
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.257574095405261e-06,
+      "loss": 1.2482,
+      "step": 12848
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.257352860512314e-06,
+      "loss": 1.1822,
+      "step": 12849
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.257131625619366e-06,
+      "loss": 1.1903,
+      "step": 12850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.256910390726419e-06,
+      "loss": 1.2061,
+      "step": 12851
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.25668915583347e-06,
+      "loss": 1.1909,
+      "step": 12852
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.256467920940522e-06,
+      "loss": 1.1684,
+      "step": 12853
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.256246686047575e-06,
+      "loss": 1.2018,
+      "step": 12854
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.256025451154626e-06,
+      "loss": 1.301,
+      "step": 12855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.255804216261679e-06,
+      "loss": 1.1604,
+      "step": 12856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.25558298136873e-06,
+      "loss": 1.1577,
+      "step": 12857
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.255361746475782e-06,
+      "loss": 1.1852,
+      "step": 12858
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.255140511582835e-06,
+      "loss": 1.2086,
+      "step": 12859
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.254919276689887e-06,
+      "loss": 1.231,
+      "step": 12860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.25469804179694e-06,
+      "loss": 1.2471,
+      "step": 12861
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.25447680690399e-06,
+      "loss": 1.1148,
+      "step": 12862
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.254255572011042e-06,
+      "loss": 1.292,
+      "step": 12863
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.254034337118094e-06,
+      "loss": 1.2843,
+      "step": 12864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.253813102225147e-06,
+      "loss": 1.2366,
+      "step": 12865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.2535918673322e-06,
+      "loss": 1.2041,
+      "step": 12866
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.253370632439252e-06,
+      "loss": 1.2435,
+      "step": 12867
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.253149397546303e-06,
+      "loss": 1.251,
+      "step": 12868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.252928162653356e-06,
+      "loss": 1.1342,
+      "step": 12869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.252706927760408e-06,
+      "loss": 1.3162,
+      "step": 12870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.25248569286746e-06,
+      "loss": 1.1762,
+      "step": 12871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.252264457974512e-06,
+      "loss": 1.254,
+      "step": 12872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.252043223081563e-06,
+      "loss": 1.1918,
+      "step": 12873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.251821988188615e-06,
+      "loss": 1.1505,
+      "step": 12874
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.251600753295668e-06,
+      "loss": 1.4077,
+      "step": 12875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.25137951840272e-06,
+      "loss": 1.167,
+      "step": 12876
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.251158283509773e-06,
+      "loss": 1.3592,
+      "step": 12877
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.250937048616824e-06,
+      "loss": 1.3417,
+      "step": 12878
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.250715813723875e-06,
+      "loss": 1.1883,
+      "step": 12879
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.250494578830928e-06,
+      "loss": 1.2783,
+      "step": 12880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.25027334393798e-06,
+      "loss": 1.2258,
+      "step": 12881
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.250052109045033e-06,
+      "loss": 1.2564,
+      "step": 12882
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.249830874152086e-06,
+      "loss": 1.2405,
+      "step": 12883
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.249609639259136e-06,
+      "loss": 1.2652,
+      "step": 12884
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.249388404366189e-06,
+      "loss": 1.2695,
+      "step": 12885
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.249167169473242e-06,
+      "loss": 1.185,
+      "step": 12886
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.248945934580293e-06,
+      "loss": 1.2079,
+      "step": 12887
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.248724699687345e-06,
+      "loss": 1.1907,
+      "step": 12888
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.248503464794398e-06,
+      "loss": 1.2746,
+      "step": 12889
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.248282229901449e-06,
+      "loss": 1.1593,
+      "step": 12890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.248060995008501e-06,
+      "loss": 1.3543,
+      "step": 12891
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.247839760115554e-06,
+      "loss": 1.1644,
+      "step": 12892
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.247618525222607e-06,
+      "loss": 1.3244,
+      "step": 12893
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.247397290329657e-06,
+      "loss": 1.2468,
+      "step": 12894
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.247176055436708e-06,
+      "loss": 1.2364,
+      "step": 12895
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.246954820543761e-06,
+      "loss": 1.1648,
+      "step": 12896
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.246733585650814e-06,
+      "loss": 1.282,
+      "step": 12897
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.246512350757866e-06,
+      "loss": 1.3422,
+      "step": 12898
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.246291115864919e-06,
+      "loss": 1.2186,
+      "step": 12899
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.246069880971971e-06,
+      "loss": 1.2912,
+      "step": 12900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.245848646079022e-06,
+      "loss": 1.1717,
+      "step": 12901
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.245627411186075e-06,
+      "loss": 1.3144,
+      "step": 12902
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.245406176293126e-06,
+      "loss": 1.4075,
+      "step": 12903
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.245184941400179e-06,
+      "loss": 1.1419,
+      "step": 12904
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.244963706507231e-06,
+      "loss": 1.186,
+      "step": 12905
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.244742471614282e-06,
+      "loss": 1.2032,
+      "step": 12906
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.244521236721335e-06,
+      "loss": 1.1313,
+      "step": 12907
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.244300001828387e-06,
+      "loss": 1.2734,
+      "step": 12908
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.24407876693544e-06,
+      "loss": 1.2281,
+      "step": 12909
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.24385753204249e-06,
+      "loss": 1.1994,
+      "step": 12910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.243636297149543e-06,
+      "loss": 1.1863,
+      "step": 12911
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.243415062256594e-06,
+      "loss": 1.2229,
+      "step": 12912
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.243193827363647e-06,
+      "loss": 1.3727,
+      "step": 12913
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.2429725924707e-06,
+      "loss": 1.2225,
+      "step": 12914
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.242751357577752e-06,
+      "loss": 1.3483,
+      "step": 12915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.242530122684805e-06,
+      "loss": 1.1519,
+      "step": 12916
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.242308887791856e-06,
+      "loss": 1.1542,
+      "step": 12917
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.242087652898908e-06,
+      "loss": 1.1573,
+      "step": 12918
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.24186641800596e-06,
+      "loss": 1.2626,
+      "step": 12919
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.241645183113012e-06,
+      "loss": 1.1361,
+      "step": 12920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.241423948220064e-06,
+      "loss": 1.2477,
+      "step": 12921
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.241202713327117e-06,
+      "loss": 1.2302,
+      "step": 12922
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.240981478434168e-06,
+      "loss": 1.3682,
+      "step": 12923
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.24076024354122e-06,
+      "loss": 1.3662,
+      "step": 12924
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.240539008648273e-06,
+      "loss": 1.2295,
+      "step": 12925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.240317773755324e-06,
+      "loss": 1.2138,
+      "step": 12926
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.240096538862377e-06,
+      "loss": 1.2032,
+      "step": 12927
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.239875303969428e-06,
+      "loss": 1.2208,
+      "step": 12928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.23965406907648e-06,
+      "loss": 1.2631,
+      "step": 12929
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.239432834183533e-06,
+      "loss": 1.3907,
+      "step": 12930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.239211599290585e-06,
+      "loss": 1.2491,
+      "step": 12931
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.238990364397638e-06,
+      "loss": 1.2208,
+      "step": 12932
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.23876912950469e-06,
+      "loss": 1.2488,
+      "step": 12933
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.238547894611742e-06,
+      "loss": 1.3232,
+      "step": 12934
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.238326659718793e-06,
+      "loss": 1.2516,
+      "step": 12935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.238105424825845e-06,
+      "loss": 1.296,
+      "step": 12936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.237884189932898e-06,
+      "loss": 1.221,
+      "step": 12937
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.23766295503995e-06,
+      "loss": 1.273,
+      "step": 12938
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.237441720147001e-06,
+      "loss": 1.2093,
+      "step": 12939
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.237220485254054e-06,
+      "loss": 1.1097,
+      "step": 12940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.236999250361107e-06,
+      "loss": 1.2485,
+      "step": 12941
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.236778015468159e-06,
+      "loss": 1.3534,
+      "step": 12942
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.23655678057521e-06,
+      "loss": 1.1172,
+      "step": 12943
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.236335545682263e-06,
+      "loss": 1.1031,
+      "step": 12944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.236114310789314e-06,
+      "loss": 1.3249,
+      "step": 12945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.235893075896366e-06,
+      "loss": 1.2782,
+      "step": 12946
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.235671841003419e-06,
+      "loss": 1.2697,
+      "step": 12947
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.235450606110471e-06,
+      "loss": 1.2986,
+      "step": 12948
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.235229371217524e-06,
+      "loss": 1.2126,
+      "step": 12949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.235008136324575e-06,
+      "loss": 1.2394,
+      "step": 12950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.234786901431626e-06,
+      "loss": 1.3387,
+      "step": 12951
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.234565666538678e-06,
+      "loss": 1.1709,
+      "step": 12952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.234344431645731e-06,
+      "loss": 1.3145,
+      "step": 12953
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.234123196752784e-06,
+      "loss": 1.2115,
+      "step": 12954
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.233901961859836e-06,
+      "loss": 1.2714,
+      "step": 12955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.233680726966887e-06,
+      "loss": 1.3134,
+      "step": 12956
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.23345949207394e-06,
+      "loss": 1.3914,
+      "step": 12957
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.233238257180992e-06,
+      "loss": 1.1923,
+      "step": 12958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.233017022288043e-06,
+      "loss": 1.1577,
+      "step": 12959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.232795787395096e-06,
+      "loss": 1.263,
+      "step": 12960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.232574552502147e-06,
+      "loss": 1.2563,
+      "step": 12961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.2323533176092e-06,
+      "loss": 1.2046,
+      "step": 12962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.232132082716252e-06,
+      "loss": 1.3221,
+      "step": 12963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.231910847823305e-06,
+      "loss": 1.241,
+      "step": 12964
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.231689612930357e-06,
+      "loss": 1.3134,
+      "step": 12965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.231468378037408e-06,
+      "loss": 1.2767,
+      "step": 12966
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.23124714314446e-06,
+      "loss": 1.2675,
+      "step": 12967
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.231025908251512e-06,
+      "loss": 1.1686,
+      "step": 12968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.230804673358564e-06,
+      "loss": 1.3338,
+      "step": 12969
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.230583438465617e-06,
+      "loss": 1.414,
+      "step": 12970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.23036220357267e-06,
+      "loss": 1.3394,
+      "step": 12971
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.23014096867972e-06,
+      "loss": 1.1892,
+      "step": 12972
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.229919733786773e-06,
+      "loss": 1.2987,
+      "step": 12973
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.229698498893826e-06,
+      "loss": 1.1419,
+      "step": 12974
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.229477264000877e-06,
+      "loss": 1.2944,
+      "step": 12975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.22925602910793e-06,
+      "loss": 1.3396,
+      "step": 12976
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.22903479421498e-06,
+      "loss": 1.2872,
+      "step": 12977
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.228813559322033e-06,
+      "loss": 1.1572,
+      "step": 12978
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.228592324429085e-06,
+      "loss": 1.3449,
+      "step": 12979
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.228371089536138e-06,
+      "loss": 1.3507,
+      "step": 12980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.22814985464319e-06,
+      "loss": 1.2509,
+      "step": 12981
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.227928619750242e-06,
+      "loss": 1.321,
+      "step": 12982
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.227707384857292e-06,
+      "loss": 1.2698,
+      "step": 12983
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.227486149964345e-06,
+      "loss": 1.3026,
+      "step": 12984
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.227264915071398e-06,
+      "loss": 1.1882,
+      "step": 12985
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.22704368017845e-06,
+      "loss": 1.1684,
+      "step": 12986
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.226822445285503e-06,
+      "loss": 1.304,
+      "step": 12987
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.226601210392554e-06,
+      "loss": 1.3299,
+      "step": 12988
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.226379975499606e-06,
+      "loss": 1.2734,
+      "step": 12989
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.226158740606659e-06,
+      "loss": 1.2882,
+      "step": 12990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.22593750571371e-06,
+      "loss": 1.2751,
+      "step": 12991
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.225716270820763e-06,
+      "loss": 1.1597,
+      "step": 12992
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.225495035927815e-06,
+      "loss": 1.1994,
+      "step": 12993
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.225273801034866e-06,
+      "loss": 1.2846,
+      "step": 12994
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.225052566141919e-06,
+      "loss": 1.3078,
+      "step": 12995
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.224831331248971e-06,
+      "loss": 1.2427,
+      "step": 12996
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.224610096356024e-06,
+      "loss": 1.1365,
+      "step": 12997
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.224388861463075e-06,
+      "loss": 1.2767,
+      "step": 12998
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.224167626570128e-06,
+      "loss": 1.1312,
+      "step": 12999
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.223946391677178e-06,
+      "loss": 1.3485,
+      "step": 13000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 54693,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 2.85132034317312e+18,
+  "trial_name": null,
+  "trial_params": null
+}