diff --git "a/checkpoint-5000/trainer_state.json" "b/checkpoint-5000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-5000/trainer_state.json"
@@ -0,0 +1,30019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.737571913261543,
+  "eval_steps": 500,
+  "global_step": 5000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.9867,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 2.027,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 2.1682,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 2.14,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 2.2372,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4e-05,
+      "loss": 2.1315,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 2.1406,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.9937,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-05,
+      "loss": 2.1101,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 2.1516,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 2.1944,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8e-05,
+      "loss": 1.9559,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 1.9272,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 2.1604,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 1.9064,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 1.8924,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 1.9717,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2e-05,
+      "loss": 1.8316,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.6e-05,
+      "loss": 1.8734,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 1.7933,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.4e-05,
+      "loss": 2.0163,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 1.9915,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 1.8712,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.6e-05,
+      "loss": 1.7655,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 1.91,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010400000000000001,
+      "loss": 2.0138,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010800000000000001,
+      "loss": 1.8081,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011200000000000001,
+      "loss": 1.8079,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000116,
+      "loss": 1.882,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012,
+      "loss": 2.0069,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000124,
+      "loss": 1.8437,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012800000000000002,
+      "loss": 1.852,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000132,
+      "loss": 1.7443,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00013600000000000003,
+      "loss": 1.7793,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014,
+      "loss": 1.9135,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000144,
+      "loss": 1.8929,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000148,
+      "loss": 1.9677,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000152,
+      "loss": 2.056,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00015600000000000002,
+      "loss": 1.7649,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00016,
+      "loss": 1.89,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000164,
+      "loss": 1.8011,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000168,
+      "loss": 1.9124,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000172,
+      "loss": 1.9942,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017600000000000002,
+      "loss": 1.971,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018,
+      "loss": 1.915,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018400000000000003,
+      "loss": 1.8294,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000188,
+      "loss": 1.8736,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000192,
+      "loss": 1.7379,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000196,
+      "loss": 1.9866,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 1.9119,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020400000000000003,
+      "loss": 1.9222,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020800000000000001,
+      "loss": 1.6958,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021200000000000003,
+      "loss": 2.1636,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021600000000000002,
+      "loss": 1.928,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022000000000000003,
+      "loss": 1.8135,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022400000000000002,
+      "loss": 1.857,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 1.948,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000232,
+      "loss": 1.7749,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000236,
+      "loss": 1.8576,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00024,
+      "loss": 1.8794,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000244,
+      "loss": 1.6526,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000248,
+      "loss": 1.7652,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000252,
+      "loss": 1.7747,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00025600000000000004,
+      "loss": 1.872,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00026000000000000003,
+      "loss": 1.957,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000264,
+      "loss": 1.9778,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000268,
+      "loss": 1.6753,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00027200000000000005,
+      "loss": 1.7244,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000276,
+      "loss": 1.917,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00028,
+      "loss": 1.9036,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000284,
+      "loss": 1.7857,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000288,
+      "loss": 1.6636,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000292,
+      "loss": 1.8924,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000296,
+      "loss": 1.92,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00030000000000000003,
+      "loss": 1.8787,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000304,
+      "loss": 1.774,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000308,
+      "loss": 1.8049,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00031200000000000005,
+      "loss": 1.9016,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00031600000000000004,
+      "loss": 1.7271,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00032,
+      "loss": 1.7394,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000324,
+      "loss": 1.9497,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000328,
+      "loss": 1.7302,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000332,
+      "loss": 1.846,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000336,
+      "loss": 1.7626,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00034,
+      "loss": 1.9597,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000344,
+      "loss": 1.7221,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000348,
+      "loss": 1.8593,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035200000000000005,
+      "loss": 1.7383,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035600000000000003,
+      "loss": 1.987,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00036,
+      "loss": 1.8862,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000364,
+      "loss": 1.6766,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00036800000000000005,
+      "loss": 1.8663,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00037200000000000004,
+      "loss": 1.9523,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000376,
+      "loss": 1.7663,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00038,
+      "loss": 1.803,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000384,
+      "loss": 1.87,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000388,
+      "loss": 1.9695,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000392,
+      "loss": 1.8292,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00039600000000000003,
+      "loss": 1.8299,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004,
+      "loss": 1.8092,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00039999997787532754,
+      "loss": 1.912,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039999991150131497,
+      "loss": 1.7853,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039999980087797706,
+      "loss": 1.8775,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039999964600533815,
+      "loss": 1.786,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999994468834327,
+      "loss": 1.8799,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039999920351230456,
+      "loss": 1.872,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039999891589200773,
+      "loss": 1.7817,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999985840226058,
+      "loss": 1.9934,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999982079041722,
+      "loss": 1.7716,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999977875367901,
+      "loss": 1.8222,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999973229205525,
+      "loss": 1.8632,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999968140555623,
+      "loss": 1.7595,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039999626094193197,
+      "loss": 1.6849,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999956635797839,
+      "loss": 1.9866,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999950219692503,
+      "loss": 1.8149,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039999433611047315,
+      "loss": 1.8287,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039999360600360414,
+      "loss": 1.7472,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999928316488048,
+      "loss": 1.7806,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039999201304624643,
+      "loss": 1.9447,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999911501961102,
+      "loss": 1.8369,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000399990243098587,
+      "loss": 1.8782,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999892917538776,
+      "loss": 1.8528,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999882961621923,
+      "loss": 1.8133,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999872563237514,
+      "loss": 1.6053,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999861722387852,
+      "loss": 1.7234,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999850439075333,
+      "loss": 1.9498,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039998387133024545,
+      "loss": 1.7752,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000399982654507181,
+      "loss": 1.837,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999813934386092,
+      "loss": 1.8002,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039998008812480913,
+      "loss": 1.9792,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999787385660695,
+      "loss": 1.7035,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000399977344762689,
+      "loss": 1.7462,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039997590671497584,
+      "loss": 1.9268,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999744244232483,
+      "loss": 1.6444,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999728978878343,
+      "loss": 1.6023,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999713271090716,
+      "loss": 1.9378,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039996971208730776,
+      "loss": 1.8392,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999680528229,
+      "loss": 1.9371,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039996634931621546,
+      "loss": 1.8793,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039996460156763115,
+      "loss": 1.7281,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039996280957753356,
+      "loss": 1.8999,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999609733463193,
+      "loss": 1.7616,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999590928743946,
+      "loss": 1.8009,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039995716816217553,
+      "loss": 1.8248,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039995519921008784,
+      "loss": 1.8125,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039995318601856723,
+      "loss": 1.8755,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039995112858805915,
+      "loss": 1.8212,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039994902691901864,
+      "loss": 1.8124,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039994688101191087,
+      "loss": 2.0968,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999446908672105,
+      "loss": 1.9205,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999424564854022,
+      "loss": 1.7214,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039994017786698015,
+      "loss": 1.9117,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039993785501244865,
+      "loss": 1.8926,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999354879223215,
+      "loss": 1.7361,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039993307659712246,
+      "loss": 1.7801,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039993062103738515,
+      "loss": 1.7732,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999281212436526,
+      "loss": 1.8174,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039992557721647815,
+      "loss": 1.876,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039992298895642444,
+      "loss": 1.963,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999203564640643,
+      "loss": 1.9794,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039991767973998,
+      "loss": 1.9807,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999149587847638,
+      "loss": 1.8689,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999121935990178,
+      "loss": 1.8105,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039990938418335364,
+      "loss": 1.8673,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039990653053839295,
+      "loss": 1.7556,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039990363266476716,
+      "loss": 1.8107,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039990069056311734,
+      "loss": 1.7058,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003998977042340944,
+      "loss": 1.8396,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039989467367835917,
+      "loss": 1.8712,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039989159889658205,
+      "loss": 1.7696,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998884798894434,
+      "loss": 2.0015,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998853166576332,
+      "loss": 1.7819,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039988210920185134,
+      "loss": 1.8364,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998788575228074,
+      "loss": 1.8661,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039987556162122105,
+      "loss": 1.7113,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998722214978212,
+      "loss": 1.8855,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039986883715334697,
+      "loss": 1.8185,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039986540858854706,
+      "loss": 1.9077,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998619358041802,
+      "loss": 1.8081,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039985841880101454,
+      "loss": 1.8344,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039985485757982833,
+      "loss": 1.6801,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998512521414095,
+      "loss": 1.8112,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998476024865556,
+      "loss": 1.8427,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039984390861607424,
+      "loss": 1.8487,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998401705307825,
+      "loss": 1.8618,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998363882315077,
+      "loss": 1.7255,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039983256171908637,
+      "loss": 1.9246,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998286909943653,
+      "loss": 1.777,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039982477605820084,
+      "loss": 1.888,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998208169114591,
+      "loss": 1.8577,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998168135550161,
+      "loss": 1.802,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998127659897575,
+      "loss": 1.9086,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039980867421657887,
+      "loss": 1.8168,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998045382363855,
+      "loss": 1.844,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039980035805009236,
+      "loss": 1.6602,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039979613365862443,
+      "loss": 1.82,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039979186506291635,
+      "loss": 1.5808,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039978755226391236,
+      "loss": 1.8157,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003997831952625669,
+      "loss": 1.7821,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039977879405984375,
+      "loss": 1.7804,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003997743486567167,
+      "loss": 1.9383,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003997698590541693,
+      "loss": 1.8185,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003997653252531949,
+      "loss": 1.6616,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039976074725479654,
+      "loss": 1.8869,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003997561250599871,
+      "loss": 1.7009,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003997514586697892,
+      "loss": 1.9382,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039974674808523534,
+      "loss": 1.8235,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003997419933073676,
+      "loss": 1.771,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003997371943372381,
+      "loss": 1.715,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003997323511759085,
+      "loss": 1.6947,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039972746382445037,
+      "loss": 1.7607,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000399722532283945,
+      "loss": 1.6536,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039971755655548346,
+      "loss": 1.8122,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003997125366401667,
+      "loss": 1.8806,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003997074725391053,
+      "loss": 1.8513,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039970236425341957,
+      "loss": 1.8634,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039969721178423985,
+      "loss": 1.9743,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039969201513270617,
+      "loss": 1.743,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039968677429996806,
+      "loss": 1.6621,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003996814892871852,
+      "loss": 1.8157,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003996761600955268,
+      "loss": 1.8073,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003996707867261719,
+      "loss": 1.8465,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003996653691803095,
+      "loss": 1.6805,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000399659907459138,
+      "loss": 1.9336,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000399654401563866,
+      "loss": 1.8547,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039964885149571146,
+      "loss": 1.8622,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003996432572559025,
+      "loss": 1.7799,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039963761884567663,
+      "loss": 1.864,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039963193626628153,
+      "loss": 1.8255,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003996262095189743,
+      "loss": 1.7513,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003996204386050221,
+      "loss": 1.9518,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039961462352570157,
+      "loss": 1.8381,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003996087642822994,
+      "loss": 1.8554,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003996028608761119,
+      "loss": 1.7689,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003995969133084451,
+      "loss": 1.8383,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000399590921580615,
+      "loss": 1.7142,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039958488569394714,
+      "loss": 1.8717,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000399578805649777,
+      "loss": 1.793,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003995726814494498,
+      "loss": 1.8575,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039956651309432047,
+      "loss": 1.9538,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003995603005857538,
+      "loss": 2.0565,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003995540439251241,
+      "loss": 1.854,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039954774311381575,
+      "loss": 1.8063,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039954139815322285,
+      "loss": 1.8717,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003995350090447491,
+      "loss": 1.75,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039952857578980815,
+      "loss": 1.6838,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003995220983898233,
+      "loss": 1.7988,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003995155768462276,
+      "loss": 1.6739,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000399509011160464,
+      "loss": 1.6904,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003995024013339851,
+      "loss": 1.6439,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003994957473682533,
+      "loss": 1.869,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039948904926474083,
+      "loss": 1.8591,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039948230702492945,
+      "loss": 1.9372,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003994755206503111,
+      "loss": 1.8036,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003994686901423871,
+      "loss": 1.761,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003994618155026687,
+      "loss": 1.9229,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003994548967326768,
+      "loss": 1.8738,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039944793383394234,
+      "loss": 1.8045,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003994409268080057,
+      "loss": 1.8773,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039943387565641726,
+      "loss": 1.8563,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000399426780380737,
+      "loss": 1.827,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039941964098253466,
+      "loss": 1.7594,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039941245746339,
+      "loss": 1.8092,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039940522982489217,
+      "loss": 1.8061,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039939795806864033,
+      "loss": 1.8575,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003993906421962433,
+      "loss": 1.859,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003993832822093198,
+      "loss": 1.8997,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003993758781094981,
+      "loss": 1.768,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003993684298984163,
+      "loss": 1.8899,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003993609375777224,
+      "loss": 1.8604,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039935340114907394,
+      "loss": 1.8243,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039934582061413844,
+      "loss": 1.8511,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039933819597459306,
+      "loss": 1.7176,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039933052723212464,
+      "loss": 1.9557,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039932281438843,
+      "loss": 1.8519,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003993150574452154,
+      "loss": 1.7136,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003993072564041971,
+      "loss": 1.9016,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003992994112671011,
+      "loss": 1.8027,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039929152203566315,
+      "loss": 1.76,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003992835887116286,
+      "loss": 1.68,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003992756112967528,
+      "loss": 1.8404,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003992675897928006,
+      "loss": 1.8483,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039925952420154687,
+      "loss": 1.7165,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000399251414524776,
+      "loss": 1.6934,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039924326076428216,
+      "loss": 1.8476,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003992350629218695,
+      "loss": 1.7379,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003992268209993517,
+      "loss": 1.8374,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039921853499855225,
+      "loss": 1.8406,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003992102049213044,
+      "loss": 1.8564,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003992018307694512,
+      "loss": 1.8254,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039919341254484534,
+      "loss": 1.855,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003991849502493493,
+      "loss": 1.8129,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003991764438848354,
+      "loss": 1.8434,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039916789345318564,
+      "loss": 1.6632,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003991592989562918,
+      "loss": 1.7855,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039915066039605525,
+      "loss": 1.8238,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003991419777743874,
+      "loss": 1.9404,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039913325109320924,
+      "loss": 1.8163,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039912448035445136,
+      "loss": 1.8147,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039911566556005446,
+      "loss": 1.674,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039910680671196864,
+      "loss": 1.6375,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003990979038121539,
+      "loss": 1.8138,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039908895686258014,
+      "loss": 1.8332,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039907996586522666,
+      "loss": 1.864,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003990709308220828,
+      "loss": 1.8037,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039906185173514745,
+      "loss": 1.7727,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039905272860642937,
+      "loss": 1.8783,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039904356143794694,
+      "loss": 1.866,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003990343502317285,
+      "loss": 1.6836,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000399025094989812,
+      "loss": 1.8226,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000399015795714245,
+      "loss": 1.7026,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039900645240708503,
+      "loss": 1.9241,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039899706507039927,
+      "loss": 1.7931,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039898763370626455,
+      "loss": 1.7745,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039897815831676766,
+      "loss": 1.7543,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039896863890400493,
+      "loss": 1.8163,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039895907547008244,
+      "loss": 1.8038,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039894946801711617,
+      "loss": 1.7993,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039893981654723164,
+      "loss": 1.9959,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039893012106256436,
+      "loss": 1.7966,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039892038156525924,
+      "loss": 1.8317,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003989105980574712,
+      "loss": 1.7131,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039890077054136486,
+      "loss": 1.8636,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003988908990191144,
+      "loss": 1.8316,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000398880983492904,
+      "loss": 1.8619,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039887102396492734,
+      "loss": 1.8513,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000398861020437388,
+      "loss": 1.6909,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003988509729124991,
+      "loss": 1.7864,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003988408813924838,
+      "loss": 1.8593,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003988307458795747,
+      "loss": 1.8248,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039882056637601426,
+      "loss": 1.7911,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003988103428840547,
+      "loss": 1.7673,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039880007540595794,
+      "loss": 1.8589,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039878976394399555,
+      "loss": 1.8383,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000398779408500449,
+      "loss": 1.6251,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003987690090776093,
+      "loss": 1.7892,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039875856567777744,
+      "loss": 1.7274,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003987480783032638,
+      "loss": 1.8624,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003987375469563888,
+      "loss": 1.8218,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039872697163948243,
+      "loss": 1.8543,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003987163523548845,
+      "loss": 1.8423,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039870568910494434,
+      "loss": 1.8593,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039869498189202136,
+      "loss": 1.7345,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003986842307184843,
+      "loss": 1.8453,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039867343558671204,
+      "loss": 1.9135,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003986625964990927,
+      "loss": 1.8207,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003986517134580247,
+      "loss": 1.6494,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003986407864659156,
+      "loss": 1.79,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003986298155251831,
+      "loss": 1.6841,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003986188006382545,
+      "loss": 1.7476,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003986077418075667,
+      "loss": 1.7944,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039859663903556657,
+      "loss": 1.9451,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003985854923247105,
+      "loss": 1.965,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039857430167746454,
+      "loss": 1.734,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039856306709630475,
+      "loss": 1.8918,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003985517885837167,
+      "loss": 1.8399,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003985404661421957,
+      "loss": 1.753,
+      "step": 357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039852909977424687,
+      "loss": 1.781,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003985176894823849,
+      "loss": 1.7894,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039850623526913433,
+      "loss": 1.854,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039849473713702936,
+      "loss": 1.7999,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039848319508861387,
+      "loss": 1.8435,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003984716091264415,
+      "loss": 1.8206,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003984599792530757,
+      "loss": 1.7712,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003984483054710895,
+      "loss": 1.6843,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003984365877830656,
+      "loss": 1.7151,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003984248261915966,
+      "loss": 1.7526,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039841302069928465,
+      "loss": 1.8997,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039840117130874176,
+      "loss": 1.7865,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039838927802258946,
+      "loss": 1.8472,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003983773408434592,
+      "loss": 1.9315,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039836535977399193,
+      "loss": 1.7279,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003983533348168386,
+      "loss": 1.8795,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003983412659746595,
+      "loss": 1.8414,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003983291532501249,
+      "loss": 1.8032,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003983169966459148,
+      "loss": 1.7834,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003983047961647186,
+      "loss": 1.6585,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039829255180923574,
+      "loss": 1.8102,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039828026358217533,
+      "loss": 1.8267,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003982679314862559,
+      "loss": 1.772,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039825555552420604,
+      "loss": 1.7212,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003982431356987638,
+      "loss": 1.8234,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039823067201267713,
+      "loss": 1.8071,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039821816446870346,
+      "loss": 1.7077,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003982056130696101,
+      "loss": 1.741,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003981930178181741,
+      "loss": 1.862,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003981803787171819,
+      "loss": 1.8635,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039816769576943,
+      "loss": 1.8943,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039815496897772457,
+      "loss": 1.7162,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039814219834488114,
+      "loss": 1.8143,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039812938387372523,
+      "loss": 1.7148,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003981165255670921,
+      "loss": 1.8769,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003981036234278266,
+      "loss": 1.6237,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039809067745878317,
+      "loss": 1.8336,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039807768766282614,
+      "loss": 1.8808,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003980646540428295,
+      "loss": 1.8084,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003980515766016768,
+      "loss": 1.7587,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039803845534226143,
+      "loss": 1.9071,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003980252902674865,
+      "loss": 1.7408,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039801208138026456,
+      "loss": 1.7905,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039799882868351823,
+      "loss": 1.8778,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039798553218017943,
+      "loss": 1.8313,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039797219187319015,
+      "loss": 1.9472,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003979588077655018,
+      "loss": 1.7632,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003979453798600756,
+      "loss": 1.7821,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039793190815988233,
+      "loss": 1.8344,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003979183926679027,
+      "loss": 1.6997,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003979048333871269,
+      "loss": 1.956,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039789123032055493,
+      "loss": 1.756,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039787758347119626,
+      "loss": 1.8861,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039786389284207044,
+      "loss": 1.6501,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003978501584362063,
+      "loss": 1.8189,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003978363802566426,
+      "loss": 1.6048,
+      "step": 413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039782255830642777,
+      "loss": 1.7401,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003978086925886198,
+      "loss": 1.8954,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039779478310628645,
+      "loss": 1.8313,
+      "step": 416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039778082986250513,
+      "loss": 1.9141,
+      "step": 417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039776683286036295,
+      "loss": 1.6629,
+      "step": 418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039775279210295676,
+      "loss": 1.8725,
+      "step": 419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039773870759339297,
+      "loss": 1.6964,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003977245793347878,
+      "loss": 1.8059,
+      "step": 421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039771040733026697,
+      "loss": 1.7852,
+      "step": 422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000397696191582966,
+      "loss": 1.9568,
+      "step": 423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003976819320960303,
+      "loss": 1.8273,
+      "step": 424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039766762887261445,
+      "loss": 1.8533,
+      "step": 425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039765328191588314,
+      "loss": 1.7547,
+      "step": 426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003976388912290105,
+      "loss": 1.8107,
+      "step": 427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039762445681518053,
+      "loss": 2.0183,
+      "step": 428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003976099786775867,
+      "loss": 1.7311,
+      "step": 429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039759545681943237,
+      "loss": 1.9018,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039758089124393035,
+      "loss": 1.6872,
+      "step": 431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039756628195430324,
+      "loss": 1.7358,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039755162895378326,
+      "loss": 1.6796,
+      "step": 433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003975369322456125,
+      "loss": 1.892,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003975221918330424,
+      "loss": 1.7282,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039750740771933417,
+      "loss": 1.9827,
+      "step": 436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000397492579907759,
+      "loss": 1.6816,
+      "step": 437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003974777084015973,
+      "loss": 1.799,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039746279320413937,
+      "loss": 1.7513,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039744783431868515,
+      "loss": 1.8015,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003974328317485443,
+      "loss": 1.9568,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003974177854970361,
+      "loss": 1.7025,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003974026955674893,
+      "loss": 1.7496,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039738756196324267,
+      "loss": 1.7982,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039737238468764453,
+      "loss": 1.7978,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003973571637440526,
+      "loss": 1.8577,
+      "step": 446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039734189913583464,
+      "loss": 1.8767,
+      "step": 447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003973265908663678,
+      "loss": 1.7078,
+      "step": 448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000397311238939039,
+      "loss": 1.8624,
+      "step": 449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003972958433572448,
+      "loss": 1.8064,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039728040412439147,
+      "loss": 1.8586,
+      "step": 451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003972649212438948,
+      "loss": 1.7813,
+      "step": 452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039724939471918043,
+      "loss": 1.7999,
+      "step": 453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003972338245536835,
+      "loss": 1.8336,
+      "step": 454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039721821075084886,
+      "loss": 1.7335,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039720255331413104,
+      "loss": 1.727,
+      "step": 456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003971868522469941,
+      "loss": 1.8477,
+      "step": 457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000397171107552912,
+      "loss": 1.6951,
+      "step": 458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039715531923536807,
+      "loss": 1.6504,
+      "step": 459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039713948729785546,
+      "loss": 1.9051,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000397123611743877,
+      "loss": 1.8366,
+      "step": 461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000397107692576945,
+      "loss": 1.8481,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039709172980058164,
+      "loss": 1.8254,
+      "step": 463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003970757234183185,
+      "loss": 1.8252,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039705967343369705,
+      "loss": 1.6986,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003970435798502683,
+      "loss": 1.7434,
+      "step": 466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003970274426715928,
+      "loss": 1.7413,
+      "step": 467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039701126190124094,
+      "loss": 1.7657,
+      "step": 468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039699503754279264,
+      "loss": 1.7052,
+      "step": 469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003969787695998374,
+      "loss": 1.8257,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003969624580759746,
+      "loss": 1.8254,
+      "step": 471
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039694610297481305,
+      "loss": 1.8684,
+      "step": 472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003969297042999712,
+      "loss": 1.8561,
+      "step": 473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039691326205507717,
+      "loss": 1.7693,
+      "step": 474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039689677624376896,
+      "loss": 1.8749,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039688024686969386,
+      "loss": 1.8235,
+      "step": 476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039686367393650893,
+      "loss": 1.6995,
+      "step": 477
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003968470574478809,
+      "loss": 1.5403,
+      "step": 478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039683039740748613,
+      "loss": 1.8066,
+      "step": 479
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003968136938190106,
+      "loss": 1.8615,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039679694668614987,
+      "loss": 1.5913,
+      "step": 481
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039678015601260925,
+      "loss": 1.8386,
+      "step": 482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039676332180210364,
+      "loss": 1.8947,
+      "step": 483
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003967464440583574,
+      "loss": 1.8435,
+      "step": 484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039672952278510494,
+      "loss": 1.7903,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039671255798608975,
+      "loss": 1.8439,
+      "step": 486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039669554966506544,
+      "loss": 1.8488,
+      "step": 487
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000396678497825795,
+      "loss": 1.7086,
+      "step": 488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000396661402472051,
+      "loss": 1.792,
+      "step": 489
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039664426360761586,
+      "loss": 1.8584,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003966270812362814,
+      "loss": 1.629,
+      "step": 491
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003966098553618493,
+      "loss": 1.6584,
+      "step": 492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039659258598813055,
+      "loss": 1.6464,
+      "step": 493
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039657527311894606,
+      "loss": 1.7918,
+      "step": 494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039655791675812614,
+      "loss": 1.6825,
+      "step": 495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039654051690951093,
+      "loss": 1.7429,
+      "step": 496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039652307357695006,
+      "loss": 1.7522,
+      "step": 497
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039650558676430285,
+      "loss": 1.723,
+      "step": 498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003964880564754381,
+      "loss": 1.6514,
+      "step": 499
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003964704827142344,
+      "loss": 1.7737,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039645286548457994,
+      "loss": 1.9399,
+      "step": 501
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003964352047903723,
+      "loss": 1.7434,
+      "step": 502
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000396417500635519,
+      "loss": 1.8614,
+      "step": 503
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000396399753023937,
+      "loss": 1.7248,
+      "step": 504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039638196195955293,
+      "loss": 1.8147,
+      "step": 505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003963641274463029,
+      "loss": 1.7088,
+      "step": 506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003963462494881328,
+      "loss": 1.6859,
+      "step": 507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039632832808899807,
+      "loss": 1.8509,
+      "step": 508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003963103632528638,
+      "loss": 1.7807,
+      "step": 509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039629235498370456,
+      "loss": 1.8307,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003962743032855047,
+      "loss": 1.776,
+      "step": 511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000396256208162258,
+      "loss": 1.7817,
+      "step": 512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039623806961796806,
+      "loss": 1.7335,
+      "step": 513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000396219887656648,
+      "loss": 1.7618,
+      "step": 514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003962016622823204,
+      "loss": 1.7928,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003961833934990176,
+      "loss": 1.7305,
+      "step": 516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039616508131078154,
+      "loss": 1.8676,
+      "step": 517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039614672572166364,
+      "loss": 1.8546,
+      "step": 518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003961283267357252,
+      "loss": 1.7474,
+      "step": 519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003961098843570367,
+      "loss": 1.7562,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039609139858967873,
+      "loss": 1.7885,
+      "step": 521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039607286943774096,
+      "loss": 1.9507,
+      "step": 522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000396054296905323,
+      "loss": 1.7405,
+      "step": 523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039603568099653394,
+      "loss": 1.6542,
+      "step": 524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003960170217154926,
+      "loss": 1.9583,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003959983190663271,
+      "loss": 1.8323,
+      "step": 526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003959795730531755,
+      "loss": 1.8341,
+      "step": 527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003959607836801852,
+      "loss": 1.7341,
+      "step": 528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003959419509515133,
+      "loss": 1.913,
+      "step": 529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003959230748713265,
+      "loss": 1.8376,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003959041554438011,
+      "loss": 1.8546,
+      "step": 531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003958851926731229,
+      "loss": 1.6754,
+      "step": 532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039586618656348735,
+      "loss": 1.772,
+      "step": 533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039584713711909956,
+      "loss": 1.7421,
+      "step": 534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003958280443441741,
+      "loss": 1.7301,
+      "step": 535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039580890824293517,
+      "loss": 1.7809,
+      "step": 536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039578972881961665,
+      "loss": 1.956,
+      "step": 537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003957705060784619,
+      "loss": 1.8369,
+      "step": 538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003957512400237238,
+      "loss": 1.8083,
+      "step": 539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039573193065966497,
+      "loss": 1.9556,
+      "step": 540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039571257799055754,
+      "loss": 1.7622,
+      "step": 541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003956931820206832,
+      "loss": 1.8417,
+      "step": 542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003956737427543333,
+      "loss": 1.8039,
+      "step": 543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003956542601958087,
+      "loss": 1.6701,
+      "step": 544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003956347343494198,
+      "loss": 1.7418,
+      "step": 545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003956151652194866,
+      "loss": 1.7193,
+      "step": 546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003955955528103389,
+      "loss": 1.7874,
+      "step": 547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003955758971263157,
+      "loss": 1.7425,
+      "step": 548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039555619817176576,
+      "loss": 1.9046,
+      "step": 549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039553645595104747,
+      "loss": 1.6623,
+      "step": 550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039551667046852873,
+      "loss": 1.7682,
+      "step": 551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000395496841728587,
+      "loss": 1.8296,
+      "step": 552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003954769697356093,
+      "loss": 1.8189,
+      "step": 553
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003954570544939923,
+      "loss": 1.7958,
+      "step": 554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039543709600814214,
+      "loss": 1.7713,
+      "step": 555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003954170942824746,
+      "loss": 1.7116,
+      "step": 556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039539704932141487,
+      "loss": 1.7419,
+      "step": 557
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039537696112939804,
+      "loss": 1.9713,
+      "step": 558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039535682971086845,
+      "loss": 1.9347,
+      "step": 559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039533665507028007,
+      "loss": 1.775,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039531643721209654,
+      "loss": 1.73,
+      "step": 561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000395296176140791,
+      "loss": 1.7471,
+      "step": 562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039527587186084605,
+      "loss": 1.817,
+      "step": 563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039525552437675405,
+      "loss": 1.7526,
+      "step": 564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039523513369301675,
+      "loss": 1.7034,
+      "step": 565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039521469981414556,
+      "loss": 1.7872,
+      "step": 566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003951942227446614,
+      "loss": 1.8387,
+      "step": 567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003951737024890948,
+      "loss": 1.8827,
+      "step": 568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039515313905198567,
+      "loss": 1.7602,
+      "step": 569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039513253243788374,
+      "loss": 1.6599,
+      "step": 570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003951118826513481,
+      "loss": 1.7409,
+      "step": 571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039509118969694736,
+      "loss": 1.8524,
+      "step": 572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039507045357926,
+      "loss": 1.718,
+      "step": 573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003950496743028736,
+      "loss": 1.8281,
+      "step": 574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003950288518723856,
+      "loss": 1.7044,
+      "step": 575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003950079862924028,
+      "loss": 1.6658,
+      "step": 576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039498707756754176,
+      "loss": 1.8943,
+      "step": 577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039496612570242846,
+      "loss": 1.8424,
+      "step": 578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039494513070169833,
+      "loss": 1.8056,
+      "step": 579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003949240925699966,
+      "loss": 1.7077,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003949030113119777,
+      "loss": 1.7672,
+      "step": 581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039488188693230595,
+      "loss": 1.8221,
+      "step": 582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039486071943565495,
+      "loss": 1.6537,
+      "step": 583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039483950882670805,
+      "loss": 1.7519,
+      "step": 584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003948182551101579,
+      "loss": 1.7035,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039479695829070687,
+      "loss": 1.7506,
+      "step": 586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003947756183730668,
+      "loss": 1.7805,
+      "step": 587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003947542353619591,
+      "loss": 1.7556,
+      "step": 588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003947328092621147,
+      "loss": 1.9552,
+      "step": 589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039471134007827404,
+      "loss": 1.7991,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039468982781518705,
+      "loss": 1.8624,
+      "step": 591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039466827247761337,
+      "loss": 1.9038,
+      "step": 592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003946466740703219,
+      "loss": 1.9206,
+      "step": 593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039462503259809134,
+      "loss": 1.9064,
+      "step": 594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003946033480657097,
+      "loss": 1.7476,
+      "step": 595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003945816204779747,
+      "loss": 1.7746,
+      "step": 596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039455984983969346,
+      "loss": 1.7362,
+      "step": 597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039453803615568266,
+      "loss": 1.7308,
+      "step": 598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039451617943076845,
+      "loss": 1.6835,
+      "step": 599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003944942796697867,
+      "loss": 1.7648,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003944723368775825,
+      "loss": 1.7818,
+      "step": 601
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003944503510590107,
+      "loss": 1.7973,
+      "step": 602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039442832221893567,
+      "loss": 1.9494,
+      "step": 603
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039440625036223103,
+      "loss": 1.7235,
+      "step": 604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039438413549378027,
+      "loss": 1.6461,
+      "step": 605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039436197761847616,
+      "loss": 1.8894,
+      "step": 606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039433977674122105,
+      "loss": 1.7474,
+      "step": 607
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003943175328669269,
+      "loss": 1.7687,
+      "step": 608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039429524600051495,
+      "loss": 1.8025,
+      "step": 609
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003942729161469162,
+      "loss": 1.8003,
+      "step": 610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000394250543311071,
+      "loss": 1.8271,
+      "step": 611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003942281274979293,
+      "loss": 1.7797,
+      "step": 612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003942056687124506,
+      "loss": 1.7347,
+      "step": 613
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003941831669596037,
+      "loss": 1.7033,
+      "step": 614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003941606222443671,
+      "loss": 1.9121,
+      "step": 615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003941380345717287,
+      "loss": 1.7977,
+      "step": 616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000394115403946686,
+      "loss": 1.8368,
+      "step": 617
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000394092730374246,
+      "loss": 1.7594,
+      "step": 618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039407001385942505,
+      "loss": 1.881,
+      "step": 619
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003940472544072491,
+      "loss": 1.6392,
+      "step": 620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039402445202275373,
+      "loss": 1.8244,
+      "step": 621
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039400160671098375,
+      "loss": 1.9341,
+      "step": 622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003939787184769937,
+      "loss": 1.5803,
+      "step": 623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003939557873258475,
+      "loss": 1.9318,
+      "step": 624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039393281326261854,
+      "loss": 1.7603,
+      "step": 625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003939097962923899,
+      "loss": 1.7598,
+      "step": 626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039388673642025385,
+      "loss": 1.898,
+      "step": 627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003938636336513124,
+      "loss": 1.603,
+      "step": 628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003938404879906769,
+      "loss": 1.7666,
+      "step": 629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039381729944346837,
+      "loss": 1.7157,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039379406801481713,
+      "loss": 1.6146,
+      "step": 631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039377079370986296,
+      "loss": 1.7091,
+      "step": 632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003937474765337554,
+      "loss": 1.8471,
+      "step": 633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039372411649165316,
+      "loss": 1.824,
+      "step": 634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003937007135887246,
+      "loss": 1.6942,
+      "step": 635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003936772678301476,
+      "loss": 1.9742,
+      "step": 636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039365377922110945,
+      "loss": 1.6744,
+      "step": 637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003936302477668068,
+      "loss": 1.7982,
+      "step": 638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003936066734724461,
+      "loss": 1.7557,
+      "step": 639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000393583056343243,
+      "loss": 1.8011,
+      "step": 640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003935593963844226,
+      "loss": 1.8658,
+      "step": 641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003935356936012198,
+      "loss": 1.8232,
+      "step": 642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003935119479988786,
+      "loss": 1.6556,
+      "step": 643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003934881595826527,
+      "loss": 1.8603,
+      "step": 644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039346432835780514,
+      "loss": 1.8944,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039344045432960866,
+      "loss": 1.6288,
+      "step": 646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003934165375033451,
+      "loss": 1.9315,
+      "step": 647
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039339257788430617,
+      "loss": 1.8987,
+      "step": 648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039336857547779276,
+      "loss": 1.6569,
+      "step": 649
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003933445302891153,
+      "loss": 1.7239,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003933204423235938,
+      "loss": 1.9766,
+      "step": 651
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003932963115865576,
+      "loss": 1.8746,
+      "step": 652
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039327213808334544,
+      "loss": 1.8845,
+      "step": 653
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039324792181930584,
+      "loss": 1.8293,
+      "step": 654
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039322366279979644,
+      "loss": 1.8071,
+      "step": 655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003931993610301845,
+      "loss": 1.6595,
+      "step": 656
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003931750165158466,
+      "loss": 1.6616,
+      "step": 657
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039315062926216905,
+      "loss": 1.7848,
+      "step": 658
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003931261992745474,
+      "loss": 1.6559,
+      "step": 659
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003931017265583866,
+      "loss": 1.874,
+      "step": 660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003930772111191013,
+      "loss": 1.7907,
+      "step": 661
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003930526529621154,
+      "loss": 1.7394,
+      "step": 662
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039302805209286227,
+      "loss": 1.8215,
+      "step": 663
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003930034085167848,
+      "loss": 1.7144,
+      "step": 664
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003929787222393354,
+      "loss": 1.6354,
+      "step": 665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003929539932659757,
+      "loss": 1.8895,
+      "step": 666
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003929292216021769,
+      "loss": 1.8758,
+      "step": 667
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039290440725341975,
+      "loss": 1.7979,
+      "step": 668
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039287955022519424,
+      "loss": 1.762,
+      "step": 669
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000392854650523,
+      "loss": 1.6985,
+      "step": 670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000392829708152346,
+      "loss": 1.6646,
+      "step": 671
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039280472311875055,
+      "loss": 1.8946,
+      "step": 672
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003927796954277416,
+      "loss": 1.6791,
+      "step": 673
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003927546250848564,
+      "loss": 1.849,
+      "step": 674
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039272951209564174,
+      "loss": 1.5823,
+      "step": 675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003927043564656538,
+      "loss": 1.8089,
+      "step": 676
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039267915820045805,
+      "loss": 1.6028,
+      "step": 677
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003926539173056296,
+      "loss": 1.7725,
+      "step": 678
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039262863378675296,
+      "loss": 1.7096,
+      "step": 679
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000392603307649422,
+      "loss": 1.8553,
+      "step": 680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039257793889923995,
+      "loss": 1.6939,
+      "step": 681
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039255252754181975,
+      "loss": 1.8044,
+      "step": 682
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039252707358278336,
+      "loss": 1.7454,
+      "step": 683
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003925015770277626,
+      "loss": 1.9034,
+      "step": 684
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003924760378823984,
+      "loss": 1.7624,
+      "step": 685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039245045615234113,
+      "loss": 1.7312,
+      "step": 686
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039242483184325074,
+      "loss": 1.7426,
+      "step": 687
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039239916496079663,
+      "loss": 1.6026,
+      "step": 688
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039237345551065736,
+      "loss": 1.7428,
+      "step": 689
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003923477034985211,
+      "loss": 1.8156,
+      "step": 690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003923219089300854,
+      "loss": 1.7864,
+      "step": 691
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039229607181105725,
+      "loss": 1.7577,
+      "step": 692
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000392270192147153,
+      "loss": 1.8737,
+      "step": 693
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003922442699440985,
+      "loss": 1.7715,
+      "step": 694
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039221830520762887,
+      "loss": 1.705,
+      "step": 695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039219229794348876,
+      "loss": 1.8942,
+      "step": 696
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003921662481574322,
+      "loss": 1.8134,
+      "step": 697
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039214015585522263,
+      "loss": 1.7587,
+      "step": 698
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039211402104263285,
+      "loss": 1.7706,
+      "step": 699
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039208784372544514,
+      "loss": 2.0668,
+      "step": 700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039206162390945115,
+      "loss": 1.8216,
+      "step": 701
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039203536160045183,
+      "loss": 1.866,
+      "step": 702
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003920090568042577,
+      "loss": 1.9576,
+      "step": 703
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003919827095266887,
+      "loss": 1.6466,
+      "step": 704
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039195631977357394,
+      "loss": 1.8085,
+      "step": 705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039192988755075214,
+      "loss": 1.825,
+      "step": 706
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039190341286407125,
+      "loss": 1.6463,
+      "step": 707
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003918768957193889,
+      "loss": 1.8484,
+      "step": 708
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003918503361225717,
+      "loss": 1.8845,
+      "step": 709
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000391823734079496,
+      "loss": 1.7638,
+      "step": 710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003917970895960474,
+      "loss": 1.8003,
+      "step": 711
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003917704026781209,
+      "loss": 1.7624,
+      "step": 712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003917436733316209,
+      "loss": 1.8087,
+      "step": 713
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039171690156246115,
+      "loss": 1.8672,
+      "step": 714
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039169008737656484,
+      "loss": 1.9848,
+      "step": 715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003916632307798645,
+      "loss": 1.6232,
+      "step": 716
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003916363317783021,
+      "loss": 1.7847,
+      "step": 717
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039160939037782887,
+      "loss": 1.8966,
+      "step": 718
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003915824065844057,
+      "loss": 1.6743,
+      "step": 719
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039155538040400243,
+      "loss": 1.7383,
+      "step": 720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039152831184259866,
+      "loss": 1.6931,
+      "step": 721
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003915012009061832,
+      "loss": 1.7318,
+      "step": 722
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039147404760075423,
+      "loss": 1.7002,
+      "step": 723
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003914468519323193,
+      "loss": 1.6221,
+      "step": 724
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003914196139068954,
+      "loss": 1.7545,
+      "step": 725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039139233353050894,
+      "loss": 1.7399,
+      "step": 726
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039136501080919546,
+      "loss": 1.7467,
+      "step": 727
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039133764574900013,
+      "loss": 1.8179,
+      "step": 728
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003913102383559773,
+      "loss": 1.8389,
+      "step": 729
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039128278863619076,
+      "loss": 1.6857,
+      "step": 730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039125529659571383,
+      "loss": 1.7412,
+      "step": 731
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039122776224062886,
+      "loss": 1.9313,
+      "step": 732
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003912001855770278,
+      "loss": 1.745,
+      "step": 733
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039117256661101187,
+      "loss": 1.8401,
+      "step": 734
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039114490534869175,
+      "loss": 1.8221,
+      "step": 735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039111720179618736,
+      "loss": 1.8534,
+      "step": 736
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000391089455959628,
+      "loss": 1.8352,
+      "step": 737
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039106166784515234,
+      "loss": 1.7419,
+      "step": 738
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003910338374589084,
+      "loss": 1.6659,
+      "step": 739
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039100596480705363,
+      "loss": 1.825,
+      "step": 740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003909780498957547,
+      "loss": 1.6099,
+      "step": 741
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003909500927311878,
+      "loss": 1.8521,
+      "step": 742
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003909220933195382,
+      "loss": 1.6399,
+      "step": 743
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003908940516670008,
+      "loss": 1.856,
+      "step": 744
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003908659677797796,
+      "loss": 1.9385,
+      "step": 745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003908378416640882,
+      "loss": 1.7846,
+      "step": 746
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039080967332614933,
+      "loss": 1.9374,
+      "step": 747
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003907814627721952,
+      "loss": 1.7176,
+      "step": 748
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039075321000846724,
+      "loss": 1.7,
+      "step": 749
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039072491504121636,
+      "loss": 1.8466,
+      "step": 750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039069657787670265,
+      "loss": 1.692,
+      "step": 751
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003906681985211956,
+      "loss": 1.6392,
+      "step": 752
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039063977698097427,
+      "loss": 1.8022,
+      "step": 753
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039061131326232647,
+      "loss": 1.7165,
+      "step": 754
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039058280737155004,
+      "loss": 1.8111,
+      "step": 755
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003905542593149516,
+      "loss": 1.8639,
+      "step": 756
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003905256690988473,
+      "loss": 1.656,
+      "step": 757
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003904970367295628,
+      "loss": 1.7854,
+      "step": 758
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039046836221343285,
+      "loss": 1.9154,
+      "step": 759
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003904396455568016,
+      "loss": 1.8538,
+      "step": 760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003904108867660225,
+      "loss": 1.9094,
+      "step": 761
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039038208584745826,
+      "loss": 1.8702,
+      "step": 762
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039035324280748107,
+      "loss": 1.7503,
+      "step": 763
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003903243576524724,
+      "loss": 1.7448,
+      "step": 764
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039029543038882286,
+      "loss": 1.703,
+      "step": 765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003902664610229327,
+      "loss": 1.8952,
+      "step": 766
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039023744956121105,
+      "loss": 1.7677,
+      "step": 767
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039020839601007685,
+      "loss": 1.6851,
+      "step": 768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000390179300375958,
+      "loss": 1.8955,
+      "step": 769
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003901501626652918,
+      "loss": 1.9412,
+      "step": 770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039012098288452484,
+      "loss": 1.6386,
+      "step": 771
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039009176104011316,
+      "loss": 1.818,
+      "step": 772
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039006249713852186,
+      "loss": 1.7755,
+      "step": 773
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003900331911862257,
+      "loss": 1.6705,
+      "step": 774
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003900038431897082,
+      "loss": 1.5945,
+      "step": 775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003899744531554627,
+      "loss": 1.7899,
+      "step": 776
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003899450210899918,
+      "loss": 1.847,
+      "step": 777
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00038991554699980695,
+      "loss": 1.7291,
+      "step": 778
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003898860308914293,
+      "loss": 1.7155,
+      "step": 779
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038985647277138937,
+      "loss": 1.7665,
+      "step": 780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038982687264622656,
+      "loss": 1.7511,
+      "step": 781
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003897972305224899,
+      "loss": 1.8079,
+      "step": 782
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038976754640673763,
+      "loss": 1.8026,
+      "step": 783
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038973782030553723,
+      "loss": 1.698,
+      "step": 784
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003897080522254655,
+      "loss": 1.7626,
+      "step": 785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003896782421731085,
+      "loss": 1.6517,
+      "step": 786
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003896483901550617,
+      "loss": 1.6861,
+      "step": 787
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038961849617792974,
+      "loss": 1.7257,
+      "step": 788
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003895885602483265,
+      "loss": 1.7159,
+      "step": 789
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038955858237287516,
+      "loss": 1.756,
+      "step": 790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038952856255820845,
+      "loss": 1.7573,
+      "step": 791
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003894985008109679,
+      "loss": 1.8765,
+      "step": 792
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003894683971378047,
+      "loss": 1.6799,
+      "step": 793
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038943825154537923,
+      "loss": 1.9008,
+      "step": 794
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000389408064040361,
+      "loss": 1.5727,
+      "step": 795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000389377834629429,
+      "loss": 1.753,
+      "step": 796
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003893475633192713,
+      "loss": 1.6813,
+      "step": 797
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038931725011658535,
+      "loss": 1.7802,
+      "step": 798
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003892868950280779,
+      "loss": 1.7199,
+      "step": 799
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038925649806046477,
+      "loss": 1.8527,
+      "step": 800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003892260592204713,
+      "loss": 1.7394,
+      "step": 801
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003891955785148321,
+      "loss": 1.6138,
+      "step": 802
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003891650559502907,
+      "loss": 1.7716,
+      "step": 803
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003891344915336003,
+      "loss": 1.7895,
+      "step": 804
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000389103885271523,
+      "loss": 1.6196,
+      "step": 805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003890732371708305,
+      "loss": 1.7733,
+      "step": 806
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003890425472383035,
+      "loss": 1.8074,
+      "step": 807
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000389011815480732,
+      "loss": 1.8454,
+      "step": 808
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003889810419049155,
+      "loss": 1.6665,
+      "step": 809
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003889502265176623,
+      "loss": 1.6521,
+      "step": 810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038891936932579043,
+      "loss": 1.7852,
+      "step": 811
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003888884703361268,
+      "loss": 1.8782,
+      "step": 812
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003888575295555078,
+      "loss": 1.8013,
+      "step": 813
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003888265469907788,
+      "loss": 1.6808,
+      "step": 814
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038879552264879484,
+      "loss": 1.8552,
+      "step": 815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003887644565364197,
+      "loss": 1.8003,
+      "step": 816
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038873334866052693,
+      "loss": 1.6479,
+      "step": 817
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038870219902799883,
+      "loss": 1.8324,
+      "step": 818
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038867100764572725,
+      "loss": 1.5919,
+      "step": 819
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038863977452061315,
+      "loss": 1.6592,
+      "step": 820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038860849965956674,
+      "loss": 1.7454,
+      "step": 821
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003885771830695075,
+      "loss": 1.6781,
+      "step": 822
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038854582475736414,
+      "loss": 1.8052,
+      "step": 823
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038851442473007464,
+      "loss": 1.8096,
+      "step": 824
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000388482982994586,
+      "loss": 1.7157,
+      "step": 825
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003884514995578547,
+      "loss": 1.796,
+      "step": 826
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003884199744268464,
+      "loss": 1.8746,
+      "step": 827
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003883884076085358,
+      "loss": 1.8132,
+      "step": 828
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000388356799109907,
+      "loss": 1.8604,
+      "step": 829
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038832514893795335,
+      "loss": 1.7425,
+      "step": 830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038829345709967723,
+      "loss": 1.9109,
+      "step": 831
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003882617236020905,
+      "loss": 1.6806,
+      "step": 832
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000388229948452214,
+      "loss": 1.7439,
+      "step": 833
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038819813165707784,
+      "loss": 1.7855,
+      "step": 834
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038816627322372144,
+      "loss": 1.8821,
+      "step": 835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003881343731591934,
+      "loss": 1.7434,
+      "step": 836
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038810243147055143,
+      "loss": 1.7148,
+      "step": 837
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003880704481648626,
+      "loss": 1.6571,
+      "step": 838
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038803842324920305,
+      "loss": 1.8829,
+      "step": 839
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038800635673065824,
+      "loss": 1.7589,
+      "step": 840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038797424861632276,
+      "loss": 1.7688,
+      "step": 841
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038794209891330035,
+      "loss": 1.7214,
+      "step": 842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038790990762870416,
+      "loss": 1.8685,
+      "step": 843
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003878776747696564,
+      "loss": 1.8972,
+      "step": 844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003878454003432883,
+      "loss": 1.7691,
+      "step": 845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003878130843567407,
+      "loss": 1.7816,
+      "step": 846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038778072681716326,
+      "loss": 1.8169,
+      "step": 847
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038774832773171507,
+      "loss": 1.9048,
+      "step": 848
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038771588710756423,
+      "loss": 1.6137,
+      "step": 849
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038768340495188815,
+      "loss": 1.7536,
+      "step": 850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003876508812718735,
+      "loss": 1.7821,
+      "step": 851
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003876183160747159,
+      "loss": 1.7234,
+      "step": 852
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038758570936762033,
+      "loss": 1.7202,
+      "step": 853
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038755306115780094,
+      "loss": 1.8354,
+      "step": 854
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003875203714524811,
+      "loss": 1.8125,
+      "step": 855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038748764025889316,
+      "loss": 1.8201,
+      "step": 856
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003874548675842789,
+      "loss": 1.855,
+      "step": 857
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003874220534358892,
+      "loss": 1.634,
+      "step": 858
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003873891978209839,
+      "loss": 1.7083,
+      "step": 859
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003873563007468324,
+      "loss": 1.7279,
+      "step": 860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038732336222071297,
+      "loss": 1.9583,
+      "step": 861
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038729038224991313,
+      "loss": 1.8569,
+      "step": 862
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003872573608417296,
+      "loss": 1.7852,
+      "step": 863
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038722429800346834,
+      "loss": 1.8875,
+      "step": 864
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003871911937424443,
+      "loss": 1.8448,
+      "step": 865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038715804806598177,
+      "loss": 1.7259,
+      "step": 866
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000387124860981414,
+      "loss": 1.7899,
+      "step": 867
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038709163249608366,
+      "loss": 1.8388,
+      "step": 868
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003870583626173424,
+      "loss": 1.5319,
+      "step": 869
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038702505135255105,
+      "loss": 1.813,
+      "step": 870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038699169870907963,
+      "loss": 1.9106,
+      "step": 871
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003869583046943073,
+      "loss": 1.7036,
+      "step": 872
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038692486931562233,
+      "loss": 1.7033,
+      "step": 873
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038689139258042226,
+      "loss": 1.8424,
+      "step": 874
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003868578744961137,
+      "loss": 1.8719,
+      "step": 875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038682431507011244,
+      "loss": 1.6421,
+      "step": 876
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038679071430984324,
+      "loss": 1.8485,
+      "step": 877
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003867570722227404,
+      "loss": 1.8626,
+      "step": 878
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003867233888162469,
+      "loss": 1.8194,
+      "step": 879
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003866896640978152,
+      "loss": 1.7825,
+      "step": 880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003866558980749067,
+      "loss": 1.8508,
+      "step": 881
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038662209075499216,
+      "loss": 1.8659,
+      "step": 882
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038658824214555127,
+      "loss": 1.6852,
+      "step": 883
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038655435225407286,
+      "loss": 1.6847,
+      "step": 884
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003865204210880551,
+      "loss": 1.7392,
+      "step": 885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000386486448655005,
+      "loss": 1.6964,
+      "step": 886
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038645243496243886,
+      "loss": 1.73,
+      "step": 887
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038641838001788215,
+      "loss": 1.7274,
+      "step": 888
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003863842838288695,
+      "loss": 1.7399,
+      "step": 889
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003863501464029445,
+      "loss": 2.0097,
+      "step": 890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003863159677476598,
+      "loss": 1.8165,
+      "step": 891
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038628174787057757,
+      "loss": 1.994,
+      "step": 892
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003862474867792687,
+      "loss": 1.8058,
+      "step": 893
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038621318448131336,
+      "loss": 1.7428,
+      "step": 894
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038617884098430085,
+      "loss": 1.8434,
+      "step": 895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003861444562958295,
+      "loss": 1.7079,
+      "step": 896
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038611003042350687,
+      "loss": 1.6923,
+      "step": 897
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003860755633749496,
+      "loss": 1.5725,
+      "step": 898
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003860410551577833,
+      "loss": 1.7647,
+      "step": 899
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003860065057796429,
+      "loss": 1.8055,
+      "step": 900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038597191524817226,
+      "loss": 1.747,
+      "step": 901
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038593728357102445,
+      "loss": 1.7241,
+      "step": 902
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038590261075586166,
+      "loss": 1.7833,
+      "step": 903
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003858678968103551,
+      "loss": 1.7289,
+      "step": 904
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003858331417421852,
+      "loss": 1.6386,
+      "step": 905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038579834555904115,
+      "loss": 1.6999,
+      "step": 906
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038576350826862176,
+      "loss": 1.7406,
+      "step": 907
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003857286298786346,
+      "loss": 1.9308,
+      "step": 908
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038569371039679634,
+      "loss": 1.8683,
+      "step": 909
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038565874983083283,
+      "loss": 1.7382,
+      "step": 910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038562374818847903,
+      "loss": 1.667,
+      "step": 911
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038558870547747887,
+      "loss": 1.8801,
+      "step": 912
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003855536217055854,
+      "loss": 1.8397,
+      "step": 913
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003855184968805609,
+      "loss": 1.8916,
+      "step": 914
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003854833310101765,
+      "loss": 1.6719,
+      "step": 915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038544812410221265,
+      "loss": 1.7511,
+      "step": 916
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038541287616445876,
+      "loss": 1.8214,
+      "step": 917
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038537758720471325,
+      "loss": 1.9392,
+      "step": 918
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003853422572307837,
+      "loss": 1.8301,
+      "step": 919
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003853068862504868,
+      "loss": 1.7295,
+      "step": 920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003852714742716482,
+      "loss": 1.6899,
+      "step": 921
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038523602130210267,
+      "loss": 1.7883,
+      "step": 922
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038520052734969417,
+      "loss": 1.7702,
+      "step": 923
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038516499242227556,
+      "loss": 1.6518,
+      "step": 924
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038512941652770876,
+      "loss": 1.8554,
+      "step": 925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003850937996738649,
+      "loss": 1.6414,
+      "step": 926
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038505814186862413,
+      "loss": 1.6701,
+      "step": 927
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038502244311987554,
+      "loss": 1.8782,
+      "step": 928
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003849867034355174,
+      "loss": 1.5833,
+      "step": 929
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003849509228234569,
+      "loss": 1.9115,
+      "step": 930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038491510129161057,
+      "loss": 1.8103,
+      "step": 931
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003848792388479037,
+      "loss": 1.5932,
+      "step": 932
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038484333550027074,
+      "loss": 1.7482,
+      "step": 933
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038480739125665514,
+      "loss": 1.7492,
+      "step": 934
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038477140612500955,
+      "loss": 1.854,
+      "step": 935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003847353801132955,
+      "loss": 1.8805,
+      "step": 936
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038469931322948365,
+      "loss": 1.6904,
+      "step": 937
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003846632054815537,
+      "loss": 1.7712,
+      "step": 938
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038462705687749427,
+      "loss": 1.7026,
+      "step": 939
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038459086742530316,
+      "loss": 1.8307,
+      "step": 940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003845546371329873,
+      "loss": 1.7355,
+      "step": 941
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003845183660085624,
+      "loss": 1.7913,
+      "step": 942
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003844820540600533,
+      "loss": 1.7272,
+      "step": 943
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038444570129549403,
+      "loss": 1.8121,
+      "step": 944
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003844093077229274,
+      "loss": 1.7415,
+      "step": 945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038437287335040535,
+      "loss": 1.6635,
+      "step": 946
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038433639818598894,
+      "loss": 1.7637,
+      "step": 947
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038429988223774824,
+      "loss": 1.8602,
+      "step": 948
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003842633255137622,
+      "loss": 1.735,
+      "step": 949
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003842267280221188,
+      "loss": 1.8177,
+      "step": 950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003841900897709152,
+      "loss": 1.8425,
+      "step": 951
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003841534107682576,
+      "loss": 1.713,
+      "step": 952
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038411669102226094,
+      "loss": 1.9462,
+      "step": 953
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003840799305410494,
+      "loss": 1.7878,
+      "step": 954
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038404312933275613,
+      "loss": 1.6714,
+      "step": 955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038400628740552326,
+      "loss": 1.5507,
+      "step": 956
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003839694047675019,
+      "loss": 1.8576,
+      "step": 957
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038393248142685235,
+      "loss": 1.8755,
+      "step": 958
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038389551739174363,
+      "loss": 1.8027,
+      "step": 959
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038385851267035403,
+      "loss": 1.8282,
+      "step": 960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038382146727087066,
+      "loss": 1.9122,
+      "step": 961
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038378438120148967,
+      "loss": 1.5237,
+      "step": 962
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003837472544704163,
+      "loss": 1.7402,
+      "step": 963
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003837100870858647,
+      "loss": 1.8276,
+      "step": 964
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000383672879056058,
+      "loss": 1.7938,
+      "step": 965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003836356303892283,
+      "loss": 1.6635,
+      "step": 966
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038359834109361686,
+      "loss": 1.9171,
+      "step": 967
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003835610111774738,
+      "loss": 1.7714,
+      "step": 968
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003835236406490582,
+      "loss": 1.6811,
+      "step": 969
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038348622951663817,
+      "loss": 1.9404,
+      "step": 970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003834487777884908,
+      "loss": 1.7123,
+      "step": 971
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003834112854729021,
+      "loss": 1.6184,
+      "step": 972
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038337375257816727,
+      "loss": 1.7496,
+      "step": 973
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003833361791125903,
+      "loss": 1.8543,
+      "step": 974
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038329856508448405,
+      "loss": 1.7844,
+      "step": 975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003832609105021707,
+      "loss": 1.8961,
+      "step": 976
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038322321537398105,
+      "loss": 1.7971,
+      "step": 977
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038318547970825513,
+      "loss": 1.995,
+      "step": 978
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003831477035133418,
+      "loss": 1.7378,
+      "step": 979
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038310988679759886,
+      "loss": 1.8699,
+      "step": 980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003830720295693932,
+      "loss": 1.6779,
+      "step": 981
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003830341318371006,
+      "loss": 1.8467,
+      "step": 982
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038299619360910583,
+      "loss": 1.6427,
+      "step": 983
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003829582148938025,
+      "loss": 1.6787,
+      "step": 984
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003829201956995934,
+      "loss": 1.8548,
+      "step": 985
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038288213603489003,
+      "loss": 1.8118,
+      "step": 986
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003828440359081131,
+      "loss": 1.66,
+      "step": 987
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038280589532769204,
+      "loss": 1.8314,
+      "step": 988
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003827677143020653,
+      "loss": 1.7294,
+      "step": 989
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003827294928396804,
+      "loss": 1.8218,
+      "step": 990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003826912309489937,
+      "loss": 1.9946,
+      "step": 991
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003826529286384705,
+      "loss": 1.7639,
+      "step": 992
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003826145859165851,
+      "loss": 1.85,
+      "step": 993
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038257620279182064,
+      "loss": 1.8637,
+      "step": 994
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003825377792726692,
+      "loss": 1.9304,
+      "step": 995
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038249931536763203,
+      "loss": 1.9155,
+      "step": 996
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000382460811085219,
+      "loss": 1.7579,
+      "step": 997
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003824222664339491,
+      "loss": 1.8207,
+      "step": 998
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003823836814223503,
+      "loss": 1.8678,
+      "step": 999
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038234505605895926,
+      "loss": 1.8197,
+      "step": 1000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038230639035232176,
+      "loss": 1.7391,
+      "step": 1001
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003822676843109925,
+      "loss": 1.6486,
+      "step": 1002
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038222893794353504,
+      "loss": 1.8772,
+      "step": 1003
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003821901512585219,
+      "loss": 2.0014,
+      "step": 1004
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003821513242645345,
+      "loss": 1.8011,
+      "step": 1005
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038211245697016314,
+      "loss": 1.7971,
+      "step": 1006
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038207354938400717,
+      "loss": 1.7672,
+      "step": 1007
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003820346015146747,
+      "loss": 1.7807,
+      "step": 1008
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038199561337078287,
+      "loss": 1.7486,
+      "step": 1009
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038195658496095763,
+      "loss": 1.8099,
+      "step": 1010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003819175162938339,
+      "loss": 1.7971,
+      "step": 1011
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003818784073780555,
+      "loss": 1.6976,
+      "step": 1012
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038183925822227513,
+      "loss": 1.9194,
+      "step": 1013
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038180006883515444,
+      "loss": 1.6152,
+      "step": 1014
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000381760839225364,
+      "loss": 1.8644,
+      "step": 1015
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038172156940158313,
+      "loss": 1.7561,
+      "step": 1016
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003816822593725002,
+      "loss": 1.6371,
+      "step": 1017
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038164290914681243,
+      "loss": 1.7635,
+      "step": 1018
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038160351873322597,
+      "loss": 1.6866,
+      "step": 1019
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003815640881404557,
+      "loss": 1.8289,
+      "step": 1020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038152461737722557,
+      "loss": 1.6568,
+      "step": 1021
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003814851064522684,
+      "loss": 1.7354,
+      "step": 1022
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038144555537432585,
+      "loss": 1.8734,
+      "step": 1023
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038140596415214844,
+      "loss": 1.7821,
+      "step": 1024
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038136633279449553,
+      "loss": 1.7927,
+      "step": 1025
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003813266613101356,
+      "loss": 1.7026,
+      "step": 1026
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003812869497078456,
+      "loss": 1.7643,
+      "step": 1027
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038124719799641183,
+      "loss": 1.8583,
+      "step": 1028
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003812074061846291,
+      "loss": 1.9038,
+      "step": 1029
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038116757428130113,
+      "loss": 1.76,
+      "step": 1030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003811277022952409,
+      "loss": 1.8124,
+      "step": 1031
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003810877902352696,
+      "loss": 1.9363,
+      "step": 1032
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003810478381102178,
+      "loss": 1.8008,
+      "step": 1033
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003810078459289249,
+      "loss": 1.7251,
+      "step": 1034
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003809678137002388,
+      "loss": 1.9593,
+      "step": 1035
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003809277414330167,
+      "loss": 1.687,
+      "step": 1036
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038088762913612433,
+      "loss": 1.8318,
+      "step": 1037
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003808474768184365,
+      "loss": 1.8841,
+      "step": 1038
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038080728448883666,
+      "loss": 1.6765,
+      "step": 1039
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003807670521562174,
+      "loss": 1.6513,
+      "step": 1040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003807267798294798,
+      "loss": 1.8738,
+      "step": 1041
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003806864675175341,
+      "loss": 1.8443,
+      "step": 1042
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038064611522929925,
+      "loss": 1.7169,
+      "step": 1043
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038060572297370305,
+      "loss": 1.7395,
+      "step": 1044
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038056529075968215,
+      "loss": 1.7651,
+      "step": 1045
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003805248185961821,
+      "loss": 1.7923,
+      "step": 1046
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003804843064921571,
+      "loss": 1.9696,
+      "step": 1047
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038044375445657046,
+      "loss": 1.8612,
+      "step": 1048
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038040316249839416,
+      "loss": 1.8656,
+      "step": 1049
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003803625306266089,
+      "loss": 1.6985,
+      "step": 1050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003803218588502046,
+      "loss": 1.7921,
+      "step": 1051
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038028114717817953,
+      "loss": 1.7083,
+      "step": 1052
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003802403956195411,
+      "loss": 1.6507,
+      "step": 1053
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038019960418330545,
+      "loss": 1.7636,
+      "step": 1054
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038015877287849754,
+      "loss": 1.9297,
+      "step": 1055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038011790171415123,
+      "loss": 1.6816,
+      "step": 1056
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000380076990699309,
+      "loss": 1.7119,
+      "step": 1057
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003800360398430225,
+      "loss": 1.7012,
+      "step": 1058
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003799950491543517,
+      "loss": 1.7869,
+      "step": 1059
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037995401864236586,
+      "loss": 1.783,
+      "step": 1060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037991294831614283,
+      "loss": 1.8421,
+      "step": 1061
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037987183818476915,
+      "loss": 1.7394,
+      "step": 1062
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037983068825734033,
+      "loss": 1.5947,
+      "step": 1063
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037978949854296077,
+      "loss": 1.6738,
+      "step": 1064
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037974826905074355,
+      "loss": 1.8265,
+      "step": 1065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037970699978981046,
+      "loss": 1.6617,
+      "step": 1066
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037966569076929225,
+      "loss": 1.9785,
+      "step": 1067
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003796243419983283,
+      "loss": 1.7325,
+      "step": 1068
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003795829534860671,
+      "loss": 1.6539,
+      "step": 1069
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003795415252416656,
+      "loss": 1.9021,
+      "step": 1070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003795000572742896,
+      "loss": 1.7447,
+      "step": 1071
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037945854959311386,
+      "loss": 1.8943,
+      "step": 1072
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003794170022073218,
+      "loss": 1.8136,
+      "step": 1073
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037937541512610563,
+      "loss": 1.75,
+      "step": 1074
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003793337883586663,
+      "loss": 1.7596,
+      "step": 1075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037929212191421363,
+      "loss": 1.7274,
+      "step": 1076
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037925041580196624,
+      "loss": 1.7417,
+      "step": 1077
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003792086700311514,
+      "loss": 1.6112,
+      "step": 1078
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003791668846110053,
+      "loss": 1.9272,
+      "step": 1079
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037912505955077275,
+      "loss": 1.733,
+      "step": 1080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003790831948597075,
+      "loss": 1.8467,
+      "step": 1081
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003790412905470718,
+      "loss": 1.9008,
+      "step": 1082
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000378999346622137,
+      "loss": 1.7685,
+      "step": 1083
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037895736309418305,
+      "loss": 1.6602,
+      "step": 1084
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037891533997249857,
+      "loss": 1.8033,
+      "step": 1085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003788732772663812,
+      "loss": 1.7919,
+      "step": 1086
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000378831174985137,
+      "loss": 1.9135,
+      "step": 1087
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000378789033138081,
+      "loss": 1.7735,
+      "step": 1088
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003787468517345371,
+      "loss": 1.7759,
+      "step": 1089
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037870463078383754,
+      "loss": 1.7214,
+      "step": 1090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003786623702953238,
+      "loss": 1.8756,
+      "step": 1091
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037862007027834573,
+      "loss": 1.8928,
+      "step": 1092
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003785777307422622,
+      "loss": 1.8846,
+      "step": 1093
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003785353516964405,
+      "loss": 1.7715,
+      "step": 1094
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003784929331502571,
+      "loss": 1.6682,
+      "step": 1095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037845047511309677,
+      "loss": 1.8008,
+      "step": 1096
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037840797759435327,
+      "loss": 1.8355,
+      "step": 1097
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037836544060342904,
+      "loss": 1.8787,
+      "step": 1098
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003783228641497353,
+      "loss": 1.7593,
+      "step": 1099
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037828024824269185,
+      "loss": 1.7051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003782375928917274,
+      "loss": 1.7306,
+      "step": 1101
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003781948981062793,
+      "loss": 1.7865,
+      "step": 1102
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037815216389579356,
+      "loss": 1.6461,
+      "step": 1103
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003781093902697251,
+      "loss": 1.8314,
+      "step": 1104
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003780665772375374,
+      "loss": 1.921,
+      "step": 1105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037802372480870254,
+      "loss": 1.7857,
+      "step": 1106
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037798083299270176,
+      "loss": 1.8049,
+      "step": 1107
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037793790179902457,
+      "loss": 1.9378,
+      "step": 1108
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003778949312371693,
+      "loss": 1.8001,
+      "step": 1109
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037785192131664327,
+      "loss": 1.5682,
+      "step": 1110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000377808872046962,
+      "loss": 1.7092,
+      "step": 1111
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003777657834376502,
+      "loss": 1.8647,
+      "step": 1112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037772265549824107,
+      "loss": 1.7274,
+      "step": 1113
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003776794882382765,
+      "loss": 1.8633,
+      "step": 1114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000377636281667307,
+      "loss": 1.8224,
+      "step": 1115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000377593035794892,
+      "loss": 1.7713,
+      "step": 1116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003775497506305995,
+      "loss": 1.8009,
+      "step": 1117
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003775064261840061,
+      "loss": 1.8135,
+      "step": 1118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037746306246469734,
+      "loss": 1.7458,
+      "step": 1119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003774196594822673,
+      "loss": 1.9545,
+      "step": 1120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037737621724631855,
+      "loss": 1.5973,
+      "step": 1121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037733273576646276,
+      "loss": 1.9428,
+      "step": 1122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003772892150523199,
+      "loss": 1.8241,
+      "step": 1123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037724565511351894,
+      "loss": 1.9533,
+      "step": 1124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003772020559596973,
+      "loss": 1.7265,
+      "step": 1125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037715841760050105,
+      "loss": 1.7013,
+      "step": 1126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003771147400455852,
+      "loss": 1.776,
+      "step": 1127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003770710233046132,
+      "loss": 1.7581,
+      "step": 1128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003770272673872572,
+      "loss": 1.7754,
+      "step": 1129
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037698347230319813,
+      "loss": 1.8594,
+      "step": 1130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003769396380621254,
+      "loss": 1.6504,
+      "step": 1131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003768957646737373,
+      "loss": 1.7332,
+      "step": 1132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003768518521477406,
+      "loss": 1.8879,
+      "step": 1133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037680790049385087,
+      "loss": 1.8966,
+      "step": 1134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003767639097217922,
+      "loss": 1.8303,
+      "step": 1135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037671987984129744,
+      "loss": 1.8066,
+      "step": 1136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037667581086210807,
+      "loss": 1.8422,
+      "step": 1137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003766317027939741,
+      "loss": 1.7565,
+      "step": 1138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003765875556466545,
+      "loss": 1.7913,
+      "step": 1139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003765433694299166,
+      "loss": 1.6796,
+      "step": 1140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037649914415353636,
+      "loss": 1.8136,
+      "step": 1141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003764548798272985,
+      "loss": 1.7348,
+      "step": 1142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037641057646099643,
+      "loss": 1.8239,
+      "step": 1143
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003763662340644321,
+      "loss": 1.869,
+      "step": 1144
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037632185264741604,
+      "loss": 1.7662,
+      "step": 1145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003762774322197677,
+      "loss": 1.7137,
+      "step": 1146
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037623297279131473,
+      "loss": 1.7574,
+      "step": 1147
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003761884743718938,
+      "loss": 1.6832,
+      "step": 1148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037614393697134985,
+      "loss": 1.9578,
+      "step": 1149
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003760993605995368,
+      "loss": 1.8958,
+      "step": 1150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000376054745266317,
+      "loss": 1.8257,
+      "step": 1151
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003760100909815614,
+      "loss": 1.5621,
+      "step": 1152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003759653977551496,
+      "loss": 1.9455,
+      "step": 1153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003759206655969699,
+      "loss": 1.6892,
+      "step": 1154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037587589451691905,
+      "loss": 1.7942,
+      "step": 1155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003758310845249026,
+      "loss": 1.7829,
+      "step": 1156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003757862356308346,
+      "loss": 1.6113,
+      "step": 1157
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003757413478446377,
+      "loss": 1.7483,
+      "step": 1158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037569642117624315,
+      "loss": 1.7222,
+      "step": 1159
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037565145563559085,
+      "loss": 1.7295,
+      "step": 1160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003756064512326293,
+      "loss": 1.759,
+      "step": 1161
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003755614079773155,
+      "loss": 1.6149,
+      "step": 1162
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037551632587961523,
+      "loss": 1.6974,
+      "step": 1163
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003754712049495027,
+      "loss": 1.5949,
+      "step": 1164
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003754260451969607,
+      "loss": 1.7994,
+      "step": 1165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037538084663198086,
+      "loss": 1.7736,
+      "step": 1166
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037533560926456303,
+      "loss": 1.8719,
+      "step": 1167
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000375290333104716,
+      "loss": 1.7398,
+      "step": 1168
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037524501816245674,
+      "loss": 1.7861,
+      "step": 1169
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003751996644478113,
+      "loss": 1.8517,
+      "step": 1170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003751542719708138,
+      "loss": 1.8592,
+      "step": 1171
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003751088407415073,
+      "loss": 1.8915,
+      "step": 1172
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037506337076994336,
+      "loss": 1.7445,
+      "step": 1173
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037501786206618195,
+      "loss": 1.7321,
+      "step": 1174
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003749723146402918,
+      "loss": 1.6707,
+      "step": 1175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037492672850235006,
+      "loss": 1.7017,
+      "step": 1176
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003748811036624426,
+      "loss": 1.7587,
+      "step": 1177
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003748354401306637,
+      "loss": 1.8805,
+      "step": 1178
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003747897379171163,
+      "loss": 1.7406,
+      "step": 1179
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003747439970319118,
+      "loss": 1.6862,
+      "step": 1180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003746982174851703,
+      "loss": 1.7503,
+      "step": 1181
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003746523992870204,
+      "loss": 1.8802,
+      "step": 1182
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037460654244759914,
+      "loss": 1.8094,
+      "step": 1183
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003745606469770522,
+      "loss": 1.5392,
+      "step": 1184
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037451471288553386,
+      "loss": 1.6957,
+      "step": 1185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003744687401832068,
+      "loss": 1.8146,
+      "step": 1186
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037442272888024247,
+      "loss": 1.8521,
+      "step": 1187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037437667898682065,
+      "loss": 1.6767,
+      "step": 1188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003743305905131297,
+      "loss": 1.6173,
+      "step": 1189
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037428446346936646,
+      "loss": 1.595,
+      "step": 1190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003742382978657366,
+      "loss": 1.878,
+      "step": 1191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037419209371245395,
+      "loss": 1.8501,
+      "step": 1192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003741458510197411,
+      "loss": 1.7072,
+      "step": 1193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000374099569797829,
+      "loss": 1.6475,
+      "step": 1194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003740532500569574,
+      "loss": 1.5823,
+      "step": 1195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037400689180737423,
+      "loss": 1.7405,
+      "step": 1196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037396049505933614,
+      "loss": 1.9413,
+      "step": 1197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037391405982310826,
+      "loss": 1.8405,
+      "step": 1198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037386758610896426,
+      "loss": 1.8006,
+      "step": 1199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037382107392718624,
+      "loss": 1.8576,
+      "step": 1200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037377452328806496,
+      "loss": 1.8148,
+      "step": 1201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037372793420189956,
+      "loss": 1.9463,
+      "step": 1202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037368130667899763,
+      "loss": 1.6354,
+      "step": 1203
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003736346407296755,
+      "loss": 1.9025,
+      "step": 1204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003735879363642577,
+      "loss": 1.8128,
+      "step": 1205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003735411935930776,
+      "loss": 1.6238,
+      "step": 1206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003734944124264768,
+      "loss": 1.8424,
+      "step": 1207
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037344759287480546,
+      "loss": 1.7494,
+      "step": 1208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037340073494842224,
+      "loss": 1.8493,
+      "step": 1209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003733538386576943,
+      "loss": 1.753,
+      "step": 1210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003733069040129974,
+      "loss": 1.7025,
+      "step": 1211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037325993102471546,
+      "loss": 1.7748,
+      "step": 1212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037321291970324134,
+      "loss": 1.7567,
+      "step": 1213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037316587005897605,
+      "loss": 1.8683,
+      "step": 1214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037311878210232916,
+      "loss": 1.8389,
+      "step": 1215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003730716558437187,
+      "loss": 1.9286,
+      "step": 1216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003730244912935712,
+      "loss": 1.6945,
+      "step": 1217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003729772884623217,
+      "loss": 1.781,
+      "step": 1218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037293004736041365,
+      "loss": 1.8065,
+      "step": 1219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000372882767998299,
+      "loss": 1.6522,
+      "step": 1220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003728354503864382,
+      "loss": 1.7327,
+      "step": 1221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037278809453530003,
+      "loss": 1.6261,
+      "step": 1222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037274070045536183,
+      "loss": 1.7236,
+      "step": 1223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003726932681571095,
+      "loss": 1.8231,
+      "step": 1224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000372645797651037,
+      "loss": 1.8308,
+      "step": 1225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003725982889476474,
+      "loss": 1.8109,
+      "step": 1226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003725507420574516,
+      "loss": 1.8557,
+      "step": 1227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037250315699096926,
+      "loss": 1.6892,
+      "step": 1228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037245553375872835,
+      "loss": 1.7482,
+      "step": 1229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003724078723712655,
+      "loss": 1.6938,
+      "step": 1230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003723601728391255,
+      "loss": 1.7827,
+      "step": 1231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037231243517286174,
+      "loss": 1.8789,
+      "step": 1232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003722646593830361,
+      "loss": 1.8025,
+      "step": 1233
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003722168454802188,
+      "loss": 1.7255,
+      "step": 1234
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037216899347498845,
+      "loss": 1.7843,
+      "step": 1235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037212110337793213,
+      "loss": 1.7831,
+      "step": 1236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037207317519964543,
+      "loss": 1.9191,
+      "step": 1237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003720252089507323,
+      "loss": 1.6694,
+      "step": 1238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003719772046418052,
+      "loss": 1.7753,
+      "step": 1239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037192916228348473,
+      "loss": 1.615,
+      "step": 1240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003718810818864001,
+      "loss": 1.7757,
+      "step": 1241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003718329634611892,
+      "loss": 1.8509,
+      "step": 1242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003717848070184979,
+      "loss": 1.7939,
+      "step": 1243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003717366125689806,
+      "loss": 1.8382,
+      "step": 1244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003716883801233003,
+      "loss": 1.7896,
+      "step": 1245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003716401096921281,
+      "loss": 1.6561,
+      "step": 1246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003715918012861439,
+      "loss": 1.8774,
+      "step": 1247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003715434549160356,
+      "loss": 1.7219,
+      "step": 1248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003714950705924998,
+      "loss": 1.6798,
+      "step": 1249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003714466483262412,
+      "loss": 1.8329,
+      "step": 1250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037139818812797327,
+      "loss": 1.7112,
+      "step": 1251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003713496900084175,
+      "loss": 1.6721,
+      "step": 1252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037130115397830405,
+      "loss": 1.6087,
+      "step": 1253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037125258004837134,
+      "loss": 1.5999,
+      "step": 1254
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037120396822936617,
+      "loss": 1.65,
+      "step": 1255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003711553185320437,
+      "loss": 1.7115,
+      "step": 1256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037110663096716763,
+      "loss": 1.76,
+      "step": 1257
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037105790554550983,
+      "loss": 1.7776,
+      "step": 1258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003710091422778507,
+      "loss": 1.6147,
+      "step": 1259
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003709603411749789,
+      "loss": 1.7358,
+      "step": 1260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037091150224769154,
+      "loss": 1.79,
+      "step": 1261
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037086262550679405,
+      "loss": 1.6942,
+      "step": 1262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003708137109631003,
+      "loss": 1.7101,
+      "step": 1263
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003707647586274324,
+      "loss": 1.6912,
+      "step": 1264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037071576851062095,
+      "loss": 1.8985,
+      "step": 1265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003706667406235049,
+      "loss": 1.8518,
+      "step": 1266
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037061767497693145,
+      "loss": 1.6862,
+      "step": 1267
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037056857158175615,
+      "loss": 1.8435,
+      "step": 1268
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037051943044884305,
+      "loss": 1.72,
+      "step": 1269
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037047025158906444,
+      "loss": 1.7566,
+      "step": 1270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000370421035013301,
+      "loss": 1.8772,
+      "step": 1271
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037037178073244173,
+      "loss": 1.765,
+      "step": 1272
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000370322488757384,
+      "loss": 1.8666,
+      "step": 1273
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003702731590990334,
+      "loss": 1.6133,
+      "step": 1274
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000370223791768304,
+      "loss": 1.7229,
+      "step": 1275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037017438677611826,
+      "loss": 1.867,
+      "step": 1276
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037012494413340676,
+      "loss": 1.8271,
+      "step": 1277
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037007546385110855,
+      "loss": 1.7593,
+      "step": 1278
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000370025945940171,
+      "loss": 1.5268,
+      "step": 1279
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003699763904115497,
+      "loss": 1.7109,
+      "step": 1280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003699267972762089,
+      "loss": 1.8589,
+      "step": 1281
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003698771665451206,
+      "loss": 1.8169,
+      "step": 1282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003698274982292656,
+      "loss": 1.612,
+      "step": 1283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003697777923396329,
+      "loss": 1.7612,
+      "step": 1284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036972804888721957,
+      "loss": 1.6939,
+      "step": 1285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003696782678830314,
+      "loss": 1.7288,
+      "step": 1286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036962844933808214,
+      "loss": 1.6968,
+      "step": 1287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003695785932633941,
+      "loss": 1.7489,
+      "step": 1288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036952869966999766,
+      "loss": 1.9042,
+      "step": 1289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036947876856893166,
+      "loss": 1.6613,
+      "step": 1290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036942879997124315,
+      "loss": 1.9201,
+      "step": 1291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003693787938879876,
+      "loss": 1.8066,
+      "step": 1292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003693287503302286,
+      "loss": 1.7621,
+      "step": 1293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036927866930903816,
+      "loss": 1.6454,
+      "step": 1294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003692285508354966,
+      "loss": 1.6103,
+      "step": 1295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003691783949206924,
+      "loss": 1.8363,
+      "step": 1296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036912820157572246,
+      "loss": 1.6617,
+      "step": 1297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036907797081169177,
+      "loss": 1.8131,
+      "step": 1298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003690277026397138,
+      "loss": 1.7005,
+      "step": 1299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003689773970709102,
+      "loss": 1.8988,
+      "step": 1300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003689270541164111,
+      "loss": 1.9027,
+      "step": 1301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036887667378735433,
+      "loss": 1.6751,
+      "step": 1302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003688262560948867,
+      "loss": 1.4981,
+      "step": 1303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003687758010501628,
+      "loss": 1.5878,
+      "step": 1304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003687253086643457,
+      "loss": 1.7565,
+      "step": 1305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036867477894860667,
+      "loss": 1.7145,
+      "step": 1306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036862421191412515,
+      "loss": 1.7875,
+      "step": 1307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003685736075720891,
+      "loss": 1.969,
+      "step": 1308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003685229659336945,
+      "loss": 1.9007,
+      "step": 1309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003684722870101456,
+      "loss": 1.7475,
+      "step": 1310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036842157081265497,
+      "loss": 1.6487,
+      "step": 1311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036837081735244346,
+      "loss": 1.7673,
+      "step": 1312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036832002664073996,
+      "loss": 1.6262,
+      "step": 1313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003682691986887819,
+      "loss": 1.8813,
+      "step": 1314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003682183335078147,
+      "loss": 1.7598,
+      "step": 1315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036816743110909225,
+      "loss": 1.9757,
+      "step": 1316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003681164915038763,
+      "loss": 1.8024,
+      "step": 1317
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003680655147034373,
+      "loss": 1.585,
+      "step": 1318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036801450071905357,
+      "loss": 1.7189,
+      "step": 1319
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036796344956201187,
+      "loss": 1.706,
+      "step": 1320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003679123612436071,
+      "loss": 1.7614,
+      "step": 1321
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036786123577514234,
+      "loss": 1.6353,
+      "step": 1322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036781007316792885,
+      "loss": 1.8242,
+      "step": 1323
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036775887343328635,
+      "loss": 1.641,
+      "step": 1324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036770763658254253,
+      "loss": 1.8567,
+      "step": 1325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003676563626270334,
+      "loss": 1.7392,
+      "step": 1326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003676050515781031,
+      "loss": 1.7124,
+      "step": 1327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036755370344710407,
+      "loss": 1.8575,
+      "step": 1328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036750231824539696,
+      "loss": 1.7078,
+      "step": 1329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036745089598435053,
+      "loss": 1.7449,
+      "step": 1330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036739943667534174,
+      "loss": 1.8876,
+      "step": 1331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003673479403297559,
+      "loss": 1.8038,
+      "step": 1332
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003672964069589864,
+      "loss": 1.6337,
+      "step": 1333
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003672448365744347,
+      "loss": 1.7082,
+      "step": 1334
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003671932291875107,
+      "loss": 1.8583,
+      "step": 1335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036714158480963233,
+      "loss": 1.8713,
+      "step": 1336
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003670899034522257,
+      "loss": 1.9001,
+      "step": 1337
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036703818512672523,
+      "loss": 1.8038,
+      "step": 1338
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003669864298445734,
+      "loss": 1.7742,
+      "step": 1339
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036693463761722076,
+      "loss": 1.7624,
+      "step": 1340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003668828084561264,
+      "loss": 1.7099,
+      "step": 1341
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036683094237275715,
+      "loss": 1.6032,
+      "step": 1342
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036677903937858835,
+      "loss": 1.755,
+      "step": 1343
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036672709948510333,
+      "loss": 1.8969,
+      "step": 1344
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003666751227037936,
+      "loss": 1.6311,
+      "step": 1345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003666231090461588,
+      "loss": 1.7446,
+      "step": 1346
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036657105852370693,
+      "loss": 1.9712,
+      "step": 1347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036651897114795386,
+      "loss": 1.6304,
+      "step": 1348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003664668469304238,
+      "loss": 1.6444,
+      "step": 1349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003664146858826491,
+      "loss": 1.7692,
+      "step": 1350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036636248801617013,
+      "loss": 1.7554,
+      "step": 1351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036631025334253565,
+      "loss": 1.8322,
+      "step": 1352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036625798187330227,
+      "loss": 1.8378,
+      "step": 1353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036620567362003484,
+      "loss": 1.7166,
+      "step": 1354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003661533285943065,
+      "loss": 1.7478,
+      "step": 1355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003661009468076985,
+      "loss": 1.5835,
+      "step": 1356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036604852827179995,
+      "loss": 1.6013,
+      "step": 1357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036599607299820833,
+      "loss": 1.757,
+      "step": 1358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036594358099852926,
+      "loss": 1.745,
+      "step": 1359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036589105228437634,
+      "loss": 1.6672,
+      "step": 1360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003658384868673715,
+      "loss": 1.7542,
+      "step": 1361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003657858847591445,
+      "loss": 1.6769,
+      "step": 1362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036573324597133356,
+      "loss": 1.6363,
+      "step": 1363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036568057051558476,
+      "loss": 1.8565,
+      "step": 1364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003656278584035523,
+      "loss": 1.6873,
+      "step": 1365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036557510964689864,
+      "loss": 1.5876,
+      "step": 1366
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036552232425729426,
+      "loss": 1.6294,
+      "step": 1367
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036546950224641773,
+      "loss": 1.6025,
+      "step": 1368
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036541664362595584,
+      "loss": 1.669,
+      "step": 1369
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036536374840760326,
+      "loss": 1.8052,
+      "step": 1370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000365310816603063,
+      "loss": 1.8722,
+      "step": 1371
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003652578482240459,
+      "loss": 1.69,
+      "step": 1372
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036520484328227116,
+      "loss": 1.7863,
+      "step": 1373
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036515180178946596,
+      "loss": 1.744,
+      "step": 1374
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003650987237573655,
+      "loss": 1.7891,
+      "step": 1375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003650456091977131,
+      "loss": 1.6512,
+      "step": 1376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003649924581222602,
+      "loss": 1.8566,
+      "step": 1377
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036493927054276635,
+      "loss": 1.6047,
+      "step": 1378
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036488604647099906,
+      "loss": 1.8603,
+      "step": 1379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003648327859187341,
+      "loss": 1.7414,
+      "step": 1380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036477948889775504,
+      "loss": 1.8048,
+      "step": 1381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003647261554198538,
+      "loss": 1.9032,
+      "step": 1382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003646727854968301,
+      "loss": 1.7945,
+      "step": 1383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000364619379140492,
+      "loss": 1.807,
+      "step": 1384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003645659363626554,
+      "loss": 1.7179,
+      "step": 1385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036451245717514436,
+      "loss": 1.6454,
+      "step": 1386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000364458941589791,
+      "loss": 1.5909,
+      "step": 1387
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036440538961843543,
+      "loss": 1.7343,
+      "step": 1388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003643518012729259,
+      "loss": 1.8327,
+      "step": 1389
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003642981765651186,
+      "loss": 1.6858,
+      "step": 1390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003642445155068778,
+      "loss": 1.8517,
+      "step": 1391
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036419081811007585,
+      "loss": 1.844,
+      "step": 1392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003641370843865932,
+      "loss": 1.8321,
+      "step": 1393
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036408331434831817,
+      "loss": 1.6329,
+      "step": 1394
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003640295080071473,
+      "loss": 1.7407,
+      "step": 1395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003639756653749849,
+      "loss": 1.8821,
+      "step": 1396
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003639217864637437,
+      "loss": 1.844,
+      "step": 1397
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003638678712853441,
+      "loss": 1.6772,
+      "step": 1398
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036381391985171465,
+      "loss": 1.8225,
+      "step": 1399
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003637599321747919,
+      "loss": 1.6648,
+      "step": 1400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003637059082665205,
+      "loss": 1.852,
+      "step": 1401
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036365184813885314,
+      "loss": 1.8308,
+      "step": 1402
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036359775180375034,
+      "loss": 1.8165,
+      "step": 1403
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003635436192731808,
+      "loss": 1.7812,
+      "step": 1404
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036348945055912103,
+      "loss": 1.673,
+      "step": 1405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003634352456735559,
+      "loss": 1.705,
+      "step": 1406
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036338100462847785,
+      "loss": 1.74,
+      "step": 1407
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036332672743588765,
+      "loss": 1.7635,
+      "step": 1408
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036327241410779394,
+      "loss": 1.6494,
+      "step": 1409
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003632180646562133,
+      "loss": 1.5169,
+      "step": 1410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003631636790931705,
+      "loss": 1.7057,
+      "step": 1411
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036310925743069804,
+      "loss": 1.8034,
+      "step": 1412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003630547996808366,
+      "loss": 1.7168,
+      "step": 1413
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036300030585563475,
+      "loss": 1.7534,
+      "step": 1414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003629457759671491,
+      "loss": 1.8023,
+      "step": 1415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036289121002744417,
+      "loss": 1.9657,
+      "step": 1416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036283660804859255,
+      "loss": 1.7284,
+      "step": 1417
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036278197004267467,
+      "loss": 1.7553,
+      "step": 1418
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003627272960217791,
+      "loss": 1.7785,
+      "step": 1419
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003626725859980022,
+      "loss": 1.7441,
+      "step": 1420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036261783998344844,
+      "loss": 1.927,
+      "step": 1421
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036256305799023023,
+      "loss": 1.6692,
+      "step": 1422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003625082400304678,
+      "loss": 1.8435,
+      "step": 1423
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036245338611628954,
+      "loss": 1.8125,
+      "step": 1424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036239849625983167,
+      "loss": 1.6979,
+      "step": 1425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036234357047323836,
+      "loss": 1.8107,
+      "step": 1426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036228860876866183,
+      "loss": 1.8166,
+      "step": 1427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003622336111582621,
+      "loss": 1.7873,
+      "step": 1428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036217857765420724,
+      "loss": 1.7998,
+      "step": 1429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003621235082686733,
+      "loss": 1.8331,
+      "step": 1430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036206840301384403,
+      "loss": 1.7286,
+      "step": 1431
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003620132619019114,
+      "loss": 1.836,
+      "step": 1432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003619580849450753,
+      "loss": 1.8915,
+      "step": 1433
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036190287215554325,
+      "loss": 1.6758,
+      "step": 1434
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036184762354553107,
+      "loss": 1.717,
+      "step": 1435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003617923391272622,
+      "loss": 1.7215,
+      "step": 1436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003617370189129682,
+      "loss": 1.6871,
+      "step": 1437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036168166291488855,
+      "loss": 1.7518,
+      "step": 1438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036162627114527044,
+      "loss": 1.8031,
+      "step": 1439
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036157084361636917,
+      "loss": 1.736,
+      "step": 1440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036151538034044796,
+      "loss": 1.6607,
+      "step": 1441
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003614598813297779,
+      "loss": 1.7426,
+      "step": 1442
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003614043465966378,
+      "loss": 1.8693,
+      "step": 1443
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003613487761533147,
+      "loss": 1.6413,
+      "step": 1444
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003612931700121033,
+      "loss": 1.8151,
+      "step": 1445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036123752818530635,
+      "loss": 1.6464,
+      "step": 1446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003611818506852343,
+      "loss": 1.7811,
+      "step": 1447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036112613752420575,
+      "loss": 1.8851,
+      "step": 1448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036107038871454694,
+      "loss": 1.8165,
+      "step": 1449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036101460426859214,
+      "loss": 1.6938,
+      "step": 1450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036095878419868357,
+      "loss": 1.7957,
+      "step": 1451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003609029285171711,
+      "loss": 1.7368,
+      "step": 1452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003608470372364127,
+      "loss": 1.6123,
+      "step": 1453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003607911103687741,
+      "loss": 1.916,
+      "step": 1454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036073514792662895,
+      "loss": 1.866,
+      "step": 1455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003606791499223588,
+      "loss": 1.7706,
+      "step": 1456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003606231163683529,
+      "loss": 1.752,
+      "step": 1457
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036056704727700865,
+      "loss": 1.8133,
+      "step": 1458
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000360510942660731,
+      "loss": 1.7715,
+      "step": 1459
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036045480253193307,
+      "loss": 1.825,
+      "step": 1460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036039862690303556,
+      "loss": 1.6501,
+      "step": 1461
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036034241578646715,
+      "loss": 1.8821,
+      "step": 1462
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003602861691946644,
+      "loss": 1.804,
+      "step": 1463
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036022988714007174,
+      "loss": 1.8797,
+      "step": 1464
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003601735696351413,
+      "loss": 1.7922,
+      "step": 1465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036011721669233315,
+      "loss": 1.8579,
+      "step": 1466
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036006082832411524,
+      "loss": 1.8309,
+      "step": 1467
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003600044045429633,
+      "loss": 1.6966,
+      "step": 1468
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035994794536136086,
+      "loss": 1.8873,
+      "step": 1469
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003598914507917994,
+      "loss": 1.7683,
+      "step": 1470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035983492084677807,
+      "loss": 1.8651,
+      "step": 1471
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000359778355538804,
+      "loss": 1.8432,
+      "step": 1472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003597217548803921,
+      "loss": 1.6583,
+      "step": 1473
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035966511888406503,
+      "loss": 1.814,
+      "step": 1474
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035960844756235337,
+      "loss": 1.8304,
+      "step": 1475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035955174092779546,
+      "loss": 1.7602,
+      "step": 1476
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035949499899293733,
+      "loss": 1.9282,
+      "step": 1477
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003594382217703331,
+      "loss": 1.6705,
+      "step": 1478
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035938140927254455,
+      "loss": 1.8359,
+      "step": 1479
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003593245615121411,
+      "loss": 1.6568,
+      "step": 1480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003592676785017003,
+      "loss": 1.7306,
+      "step": 1481
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035921076025380717,
+      "loss": 1.9175,
+      "step": 1482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035915380678105485,
+      "loss": 1.8597,
+      "step": 1483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035909681809604397,
+      "loss": 1.7314,
+      "step": 1484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035903979421138325,
+      "loss": 1.7737,
+      "step": 1485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035898273513968886,
+      "loss": 1.7369,
+      "step": 1486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035892564089358507,
+      "loss": 1.6749,
+      "step": 1487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003588685114857037,
+      "loss": 1.6353,
+      "step": 1488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003588113469286845,
+      "loss": 1.739,
+      "step": 1489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000358754147235175,
+      "loss": 1.8145,
+      "step": 1490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035869691241783024,
+      "loss": 1.6463,
+      "step": 1491
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003586396424893135,
+      "loss": 1.7943,
+      "step": 1492
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003585823374622953,
+      "loss": 1.7665,
+      "step": 1493
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003585249973494544,
+      "loss": 1.7273,
+      "step": 1494
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035846762216347705,
+      "loss": 1.7483,
+      "step": 1495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003584102119170572,
+      "loss": 1.7729,
+      "step": 1496
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003583527666228969,
+      "loss": 1.8689,
+      "step": 1497
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035829528629370555,
+      "loss": 1.9075,
+      "step": 1498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003582377709422006,
+      "loss": 1.6338,
+      "step": 1499
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000358180220581107,
+      "loss": 1.6777,
+      "step": 1500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035812263522315776,
+      "loss": 1.691,
+      "step": 1501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003580650148810933,
+      "loss": 1.6561,
+      "step": 1502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035800735956766197,
+      "loss": 1.7566,
+      "step": 1503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003579496692956199,
+      "loss": 1.7529,
+      "step": 1504
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003578919440777307,
+      "loss": 1.8339,
+      "step": 1505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035783418392676614,
+      "loss": 1.8389,
+      "step": 1506
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003577763888555052,
+      "loss": 1.7698,
+      "step": 1507
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035771855887673507,
+      "loss": 1.701,
+      "step": 1508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035766069400325025,
+      "loss": 1.8879,
+      "step": 1509
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035760279424785335,
+      "loss": 1.8094,
+      "step": 1510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003575448596233543,
+      "loss": 1.8237,
+      "step": 1511
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035748689014257105,
+      "loss": 1.7066,
+      "step": 1512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003574288858183292,
+      "loss": 1.8622,
+      "step": 1513
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035737084666346197,
+      "loss": 1.6329,
+      "step": 1514
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003573127726908103,
+      "loss": 1.5701,
+      "step": 1515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003572546639132229,
+      "loss": 1.7985,
+      "step": 1516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003571965203435562,
+      "loss": 1.7546,
+      "step": 1517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003571383419946742,
+      "loss": 1.5509,
+      "step": 1518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003570801288794487,
+      "loss": 1.7278,
+      "step": 1519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035702188101075904,
+      "loss": 1.6881,
+      "step": 1520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035696359840149255,
+      "loss": 1.7025,
+      "step": 1521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000356905281064544,
+      "loss": 1.7462,
+      "step": 1522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003568469290128159,
+      "loss": 1.7618,
+      "step": 1523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003567885422592185,
+      "loss": 1.7032,
+      "step": 1524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003567301208166696,
+      "loss": 1.7536,
+      "step": 1525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035667166469809477,
+      "loss": 1.6433,
+      "step": 1526
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035661317391642725,
+      "loss": 1.5689,
+      "step": 1527
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035655464848460795,
+      "loss": 1.8087,
+      "step": 1528
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035649608841558544,
+      "loss": 1.709,
+      "step": 1529
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003564374937223159,
+      "loss": 1.9046,
+      "step": 1530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003563788644177632,
+      "loss": 1.817,
+      "step": 1531
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035632020051489893,
+      "loss": 1.8499,
+      "step": 1532
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003562615020267023,
+      "loss": 1.9163,
+      "step": 1533
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035620276896616013,
+      "loss": 2.0344,
+      "step": 1534
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035614400134626687,
+      "loss": 1.6674,
+      "step": 1535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003560851991800248,
+      "loss": 1.9265,
+      "step": 1536
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003560263624804435,
+      "loss": 1.6647,
+      "step": 1537
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003559674912605406,
+      "loss": 1.8114,
+      "step": 1538
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000355908585533341,
+      "loss": 1.6516,
+      "step": 1539
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035584964531187754,
+      "loss": 1.8022,
+      "step": 1540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003557906706091905,
+      "loss": 1.7885,
+      "step": 1541
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003557316614383278,
+      "loss": 1.8298,
+      "step": 1542
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003556726178123451,
+      "loss": 1.8174,
+      "step": 1543
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035561353974430547,
+      "loss": 1.7775,
+      "step": 1544
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003555544272472799,
+      "loss": 1.6195,
+      "step": 1545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035549528033434677,
+      "loss": 1.7727,
+      "step": 1546
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003554360990185921,
+      "loss": 1.8178,
+      "step": 1547
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035537688331310963,
+      "loss": 1.7704,
+      "step": 1548
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003553176332310006,
+      "loss": 1.7831,
+      "step": 1549
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003552583487853739,
+      "loss": 1.8114,
+      "step": 1550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000355199029989346,
+      "loss": 1.8428,
+      "step": 1551
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035513967685604113,
+      "loss": 1.8397,
+      "step": 1552
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003550802893985908,
+      "loss": 1.7154,
+      "step": 1553
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035502086763013435,
+      "loss": 1.7132,
+      "step": 1554
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035496141156381866,
+      "loss": 1.7472,
+      "step": 1555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003549019212127981,
+      "loss": 1.7483,
+      "step": 1556
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000354842396590235,
+      "loss": 1.7986,
+      "step": 1557
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003547828377092986,
+      "loss": 1.7568,
+      "step": 1558
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035472324458316637,
+      "loss": 1.6811,
+      "step": 1559
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035466361722502306,
+      "loss": 1.9144,
+      "step": 1560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035460395564806095,
+      "loss": 1.7979,
+      "step": 1561
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035454425986548003,
+      "loss": 1.7393,
+      "step": 1562
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035448452989048773,
+      "loss": 1.8093,
+      "step": 1563
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003544247657362991,
+      "loss": 1.8365,
+      "step": 1564
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003543649674161369,
+      "loss": 1.5508,
+      "step": 1565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003543051349432313,
+      "loss": 1.8636,
+      "step": 1566
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035424526833081985,
+      "loss": 1.8589,
+      "step": 1567
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035418536759214804,
+      "loss": 1.7636,
+      "step": 1568
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035412543274046857,
+      "loss": 1.7591,
+      "step": 1569
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003540654637890419,
+      "loss": 1.7911,
+      "step": 1570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000354005460751136,
+      "loss": 1.8015,
+      "step": 1571
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035394542364002626,
+      "loss": 2.1545,
+      "step": 1572
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003538853524689958,
+      "loss": 1.7584,
+      "step": 1573
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000353825247251335,
+      "loss": 1.6779,
+      "step": 1574
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035376510800034215,
+      "loss": 1.6399,
+      "step": 1575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003537049347293227,
+      "loss": 1.7753,
+      "step": 1576
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003536447274515899,
+      "loss": 1.778,
+      "step": 1577
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003535844861804643,
+      "loss": 1.8353,
+      "step": 1578
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035352421092927424,
+      "loss": 1.8296,
+      "step": 1579
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035346390171135526,
+      "loss": 1.7863,
+      "step": 1580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035340355854005065,
+      "loss": 1.8683,
+      "step": 1581
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035334318142871113,
+      "loss": 1.8012,
+      "step": 1582
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000353282770390695,
+      "loss": 1.8108,
+      "step": 1583
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003532223254393679,
+      "loss": 1.9045,
+      "step": 1584
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003531618465881031,
+      "loss": 1.5714,
+      "step": 1585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035310133385028144,
+      "loss": 1.8212,
+      "step": 1586
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035304078723929106,
+      "loss": 1.7302,
+      "step": 1587
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003529802067685277,
+      "loss": 1.6834,
+      "step": 1588
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003529195924513947,
+      "loss": 1.8177,
+      "step": 1589
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003528589443013027,
+      "loss": 1.6969,
+      "step": 1590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003527982623316699,
+      "loss": 1.7562,
+      "step": 1591
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035273754655592193,
+      "loss": 1.6936,
+      "step": 1592
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003526767969874921,
+      "loss": 1.78,
+      "step": 1593
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000352616013639821,
+      "loss": 1.8953,
+      "step": 1594
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035255519652635666,
+      "loss": 1.7533,
+      "step": 1595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003524943456605548,
+      "loss": 1.6825,
+      "step": 1596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035243346105587835,
+      "loss": 1.7707,
+      "step": 1597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003523725427257979,
+      "loss": 1.5876,
+      "step": 1598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003523115906837914,
+      "loss": 1.7878,
+      "step": 1599
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003522506049433443,
+      "loss": 1.6445,
+      "step": 1600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003521895855179496,
+      "loss": 1.7112,
+      "step": 1601
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003521285324211075,
+      "loss": 1.6741,
+      "step": 1602
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003520674456663258,
+      "loss": 1.7501,
+      "step": 1603
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035200632526711986,
+      "loss": 1.8521,
+      "step": 1604
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035194517123701225,
+      "loss": 1.6575,
+      "step": 1605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035188398358953316,
+      "loss": 1.6585,
+      "step": 1606
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035182276233822017,
+      "loss": 1.7841,
+      "step": 1607
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003517615074966182,
+      "loss": 1.7752,
+      "step": 1608
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035170021907827986,
+      "loss": 1.8197,
+      "step": 1609
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003516388970967648,
+      "loss": 1.8378,
+      "step": 1610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035157754156564037,
+      "loss": 1.7459,
+      "step": 1611
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035151615249848136,
+      "loss": 1.5958,
+      "step": 1612
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035145472990886985,
+      "loss": 1.7998,
+      "step": 1613
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003513932738103954,
+      "loss": 1.7559,
+      "step": 1614
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003513317842166549,
+      "loss": 1.7284,
+      "step": 1615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003512702611412528,
+      "loss": 1.684,
+      "step": 1616
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035120870459780094,
+      "loss": 1.7602,
+      "step": 1617
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035114711459991837,
+      "loss": 1.9868,
+      "step": 1618
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035108549116123164,
+      "loss": 1.5582,
+      "step": 1619
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003510238342953749,
+      "loss": 1.7034,
+      "step": 1620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035096214401598953,
+      "loss": 1.5935,
+      "step": 1621
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003509004203367241,
+      "loss": 1.7443,
+      "step": 1622
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035083866327123494,
+      "loss": 1.9169,
+      "step": 1623
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003507768728331856,
+      "loss": 1.746,
+      "step": 1624
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003507150490362469,
+      "loss": 1.7771,
+      "step": 1625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003506531918940972,
+      "loss": 1.6251,
+      "step": 1626
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035059130142042224,
+      "loss": 1.7417,
+      "step": 1627
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035052937762891504,
+      "loss": 1.6939,
+      "step": 1628
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003504674205332761,
+      "loss": 1.8365,
+      "step": 1629
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035040543014721307,
+      "loss": 1.8653,
+      "step": 1630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003503434064844413,
+      "loss": 1.8034,
+      "step": 1631
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003502813495586833,
+      "loss": 1.9187,
+      "step": 1632
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003502192593836688,
+      "loss": 1.5668,
+      "step": 1633
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035015713597313523,
+      "loss": 1.7923,
+      "step": 1634
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003500949793408271,
+      "loss": 1.6871,
+      "step": 1635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003500327895004963,
+      "loss": 1.7124,
+      "step": 1636
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00034997056646590225,
+      "loss": 1.8508,
+      "step": 1637
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003499083102508116,
+      "loss": 1.701,
+      "step": 1638
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00034984602086899827,
+      "loss": 1.9549,
+      "step": 1639
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00034978369833424356,
+      "loss": 1.5959,
+      "step": 1640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00034972134266033614,
+      "loss": 1.7311,
+      "step": 1641
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003496589538610721,
+      "loss": 1.7179,
+      "step": 1642
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003495965319502546,
+      "loss": 1.6965,
+      "step": 1643
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00034953407694169437,
+      "loss": 1.801,
+      "step": 1644
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00034947158884920937,
+      "loss": 1.6146,
+      "step": 1645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00034940906768662493,
+      "loss": 1.72,
+      "step": 1646
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00034934651346777356,
+      "loss": 1.6936,
+      "step": 1647
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003492839262064952,
+      "loss": 1.7703,
+      "step": 1648
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00034922130591663717,
+      "loss": 1.765,
+      "step": 1649
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003491586526120539,
+      "loss": 1.7489,
+      "step": 1650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00034909596630660716,
+      "loss": 1.7286,
+      "step": 1651
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003490332470141663,
+      "loss": 1.8284,
+      "step": 1652
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00034897049474860763,
+      "loss": 1.65,
+      "step": 1653
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003489077095238149,
+      "loss": 1.7801,
+      "step": 1654
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00034884489135367914,
+      "loss": 1.7232,
+      "step": 1655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00034878204025209865,
+      "loss": 1.6546,
+      "step": 1656
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000348719156232979,
+      "loss": 1.7221,
+      "step": 1657
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003486562393102331,
+      "loss": 1.7321,
+      "step": 1658
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00034859328949778123,
+      "loss": 1.8199,
+      "step": 1659
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00034853030680955075,
+      "loss": 1.6659,
+      "step": 1660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003484672912594762,
+      "loss": 1.7136,
+      "step": 1661
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003484042428614998,
+      "loss": 1.8106,
+      "step": 1662
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003483411616295707,
+      "loss": 1.7933,
+      "step": 1663
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003482780475776455,
+      "loss": 1.6334,
+      "step": 1664
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034821490071968776,
+      "loss": 1.7362,
+      "step": 1665
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003481517210696688,
+      "loss": 1.6871,
+      "step": 1666
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003480885086415667,
+      "loss": 1.6552,
+      "step": 1667
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003480252634493671,
+      "loss": 1.6772,
+      "step": 1668
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034796198550706277,
+      "loss": 1.8163,
+      "step": 1669
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034789867482865375,
+      "loss": 1.8217,
+      "step": 1670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003478353314281473,
+      "loss": 1.7966,
+      "step": 1671
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034777195531955797,
+      "loss": 1.6935,
+      "step": 1672
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003477085465169076,
+      "loss": 1.6681,
+      "step": 1673
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003476451050342249,
+      "loss": 1.9224,
+      "step": 1674
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003475816308855464,
+      "loss": 1.7996,
+      "step": 1675
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034751812408491536,
+      "loss": 1.7535,
+      "step": 1676
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034745458464638255,
+      "loss": 1.7439,
+      "step": 1677
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034739101258400584,
+      "loss": 1.8175,
+      "step": 1678
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003473274079118503,
+      "loss": 1.6754,
+      "step": 1679
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034726377064398824,
+      "loss": 1.724,
+      "step": 1680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003472001007944993,
+      "loss": 1.8837,
+      "step": 1681
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003471363983774702,
+      "loss": 1.6788,
+      "step": 1682
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003470726634069948,
+      "loss": 1.7602,
+      "step": 1683
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034700889589717426,
+      "loss": 1.6202,
+      "step": 1684
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034694509586211705,
+      "loss": 1.7716,
+      "step": 1685
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003468812633159386,
+      "loss": 1.833,
+      "step": 1686
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034681739827276174,
+      "loss": 1.7629,
+      "step": 1687
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003467535007467164,
+      "loss": 1.8519,
+      "step": 1688
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003466895707519396,
+      "loss": 1.8185,
+      "step": 1689
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003466256083025757,
+      "loss": 1.7892,
+      "step": 1690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003465616134127762,
+      "loss": 1.6856,
+      "step": 1691
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034649758609669985,
+      "loss": 1.7699,
+      "step": 1692
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003464335263685122,
+      "loss": 1.7575,
+      "step": 1693
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003463694342423866,
+      "loss": 1.6917,
+      "step": 1694
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003463053097325029,
+      "loss": 1.7932,
+      "step": 1695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034624115285304875,
+      "loss": 1.7677,
+      "step": 1696
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003461769636182184,
+      "loss": 1.6178,
+      "step": 1697
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003461127420422136,
+      "loss": 1.6919,
+      "step": 1698
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003460484881392432,
+      "loss": 1.911,
+      "step": 1699
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003459842019235231,
+      "loss": 1.5402,
+      "step": 1700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003459198834092765,
+      "loss": 1.8417,
+      "step": 1701
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034585553261073357,
+      "loss": 1.9433,
+      "step": 1702
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034579114954213174,
+      "loss": 1.6901,
+      "step": 1703
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034572673421771563,
+      "loss": 1.8564,
+      "step": 1704
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003456622866517368,
+      "loss": 1.7412,
+      "step": 1705
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034559780685845413,
+      "loss": 1.6887,
+      "step": 1706
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034553329485213357,
+      "loss": 1.7668,
+      "step": 1707
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034546875064704813,
+      "loss": 1.8712,
+      "step": 1708
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003454041742574781,
+      "loss": 1.6827,
+      "step": 1709
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003453395656977107,
+      "loss": 1.7276,
+      "step": 1710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034527492498204036,
+      "loss": 1.5842,
+      "step": 1711
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034521025212476875,
+      "loss": 1.8399,
+      "step": 1712
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003451455471402044,
+      "loss": 1.672,
+      "step": 1713
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003450808100426631,
+      "loss": 1.838,
+      "step": 1714
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003450160408464678,
+      "loss": 1.85,
+      "step": 1715
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003449512395659483,
+      "loss": 1.7668,
+      "step": 1716
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003448864062154419,
+      "loss": 1.7168,
+      "step": 1717
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003448215408092927,
+      "loss": 1.6194,
+      "step": 1718
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034475664336185185,
+      "loss": 1.727,
+      "step": 1719
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034469171388747767,
+      "loss": 1.7601,
+      "step": 1720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034462675240053576,
+      "loss": 1.7198,
+      "step": 1721
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003445617589153986,
+      "loss": 1.7307,
+      "step": 1722
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034449673344644563,
+      "loss": 1.7954,
+      "step": 1723
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034443167600806373,
+      "loss": 1.5928,
+      "step": 1724
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034436658661464654,
+      "loss": 1.7159,
+      "step": 1725
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034430146528059485,
+      "loss": 1.8096,
+      "step": 1726
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003442363120203166,
+      "loss": 1.5964,
+      "step": 1727
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034417112684822674,
+      "loss": 1.6955,
+      "step": 1728
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034410590977874717,
+      "loss": 1.6374,
+      "step": 1729
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003440406608263071,
+      "loss": 1.7288,
+      "step": 1730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003439753800053426,
+      "loss": 1.7248,
+      "step": 1731
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003439100673302967,
+      "loss": 1.7844,
+      "step": 1732
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034384472281561985,
+      "loss": 1.809,
+      "step": 1733
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034377934647576914,
+      "loss": 1.6159,
+      "step": 1734
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000343713938325209,
+      "loss": 1.8423,
+      "step": 1735
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003436484983784106,
+      "loss": 1.745,
+      "step": 1736
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003435830266498524,
+      "loss": 1.6926,
+      "step": 1737
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003435175231540199,
+      "loss": 1.8702,
+      "step": 1738
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003434519879054054,
+      "loss": 1.5439,
+      "step": 1739
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034338642091850843,
+      "loss": 1.9089,
+      "step": 1740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003433208222078354,
+      "loss": 1.9519,
+      "step": 1741
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003432551917878999,
+      "loss": 1.6886,
+      "step": 1742
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034318952967322237,
+      "loss": 1.7971,
+      "step": 1743
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003431238358783304,
+      "loss": 1.5054,
+      "step": 1744
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003430581104177584,
+      "loss": 1.6638,
+      "step": 1745
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034299235330604807,
+      "loss": 1.9072,
+      "step": 1746
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034292656455774795,
+      "loss": 1.6743,
+      "step": 1747
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003428607441874135,
+      "loss": 1.8382,
+      "step": 1748
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003427948922096073,
+      "loss": 1.7824,
+      "step": 1749
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034272900863889885,
+      "loss": 1.8248,
+      "step": 1750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003426630934898647,
+      "loss": 1.9125,
+      "step": 1751
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034259714677708835,
+      "loss": 1.7244,
+      "step": 1752
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003425311685151603,
+      "loss": 1.6779,
+      "step": 1753
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003424651587186781,
+      "loss": 1.9262,
+      "step": 1754
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003423991174022461,
+      "loss": 1.7627,
+      "step": 1755
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003423330445804757,
+      "loss": 1.6917,
+      "step": 1756
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003422669402679854,
+      "loss": 1.7324,
+      "step": 1757
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003422008044794005,
+      "loss": 1.7402,
+      "step": 1758
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003421346372293534,
+      "loss": 1.8846,
+      "step": 1759
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003420684385324832,
+      "loss": 1.7387,
+      "step": 1760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003420022084034364,
+      "loss": 1.6179,
+      "step": 1761
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000341935946856866,
+      "loss": 1.6927,
+      "step": 1762
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034186965390743223,
+      "loss": 1.8427,
+      "step": 1763
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034180332956980223,
+      "loss": 2.0579,
+      "step": 1764
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034173697385864995,
+      "loss": 1.7653,
+      "step": 1765
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003416705867886564,
+      "loss": 1.7343,
+      "step": 1766
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003416041683745095,
+      "loss": 1.7173,
+      "step": 1767
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003415377186309042,
+      "loss": 1.6902,
+      "step": 1768
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034147123757254213,
+      "loss": 1.6068,
+      "step": 1769
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003414047252141322,
+      "loss": 1.8551,
+      "step": 1770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034133818157038986,
+      "loss": 1.7128,
+      "step": 1771
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003412716066560377,
+      "loss": 1.8372,
+      "step": 1772
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003412050004858053,
+      "loss": 1.7142,
+      "step": 1773
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034113836307442914,
+      "loss": 1.7099,
+      "step": 1774
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034107169443665225,
+      "loss": 1.6827,
+      "step": 1775
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003410049945872251,
+      "loss": 1.9658,
+      "step": 1776
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034093826354090464,
+      "loss": 1.699,
+      "step": 1777
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000340871501312455,
+      "loss": 1.7275,
+      "step": 1778
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034080470791664707,
+      "loss": 1.7406,
+      "step": 1779
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034073788336825867,
+      "loss": 1.7841,
+      "step": 1780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003406710276820745,
+      "loss": 1.8414,
+      "step": 1781
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034060414087288616,
+      "loss": 1.9715,
+      "step": 1782
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003405372229554922,
+      "loss": 1.7956,
+      "step": 1783
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034047027394469787,
+      "loss": 1.7585,
+      "step": 1784
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003404032938553156,
+      "loss": 1.6785,
+      "step": 1785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003403362827021643,
+      "loss": 1.7636,
+      "step": 1786
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034026924050007014,
+      "loss": 1.6882,
+      "step": 1787
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034020216726386596,
+      "loss": 1.8547,
+      "step": 1788
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003401350630083914,
+      "loss": 1.7231,
+      "step": 1789
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034006792774849315,
+      "loss": 1.7131,
+      "step": 1790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003400007614990246,
+      "loss": 1.639,
+      "step": 1791
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00033993356427484614,
+      "loss": 1.7871,
+      "step": 1792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00033986633609082485,
+      "loss": 1.6699,
+      "step": 1793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003397990769618348,
+      "loss": 1.7077,
+      "step": 1794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00033973178690275684,
+      "loss": 1.8121,
+      "step": 1795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003396644659284787,
+      "loss": 1.7195,
+      "step": 1796
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003395971140538948,
+      "loss": 1.8508,
+      "step": 1797
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003395297312939067,
+      "loss": 1.6373,
+      "step": 1798
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033946231766342254,
+      "loss": 1.6967,
+      "step": 1799
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003393948731773573,
+      "loss": 1.8437,
+      "step": 1800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033932739785063297,
+      "loss": 1.5843,
+      "step": 1801
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003392598916981782,
+      "loss": 1.6361,
+      "step": 1802
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033919235473492843,
+      "loss": 1.5782,
+      "step": 1803
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033912478697582603,
+      "loss": 1.7285,
+      "step": 1804
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033905718843582025,
+      "loss": 1.6049,
+      "step": 1805
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033898955912986696,
+      "loss": 1.6545,
+      "step": 1806
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033892189907292885,
+      "loss": 1.8178,
+      "step": 1807
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003388542082799756,
+      "loss": 1.6726,
+      "step": 1808
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033878648676598363,
+      "loss": 1.6453,
+      "step": 1809
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003387187345459359,
+      "loss": 1.6369,
+      "step": 1810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003386509516348225,
+      "loss": 1.8458,
+      "step": 1811
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033858313804764015,
+      "loss": 1.7441,
+      "step": 1812
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033851529379939243,
+      "loss": 1.6943,
+      "step": 1813
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003384474189050896,
+      "loss": 1.5658,
+      "step": 1814
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033837951337974874,
+      "loss": 1.7763,
+      "step": 1815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033831157723839385,
+      "loss": 1.8656,
+      "step": 1816
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033824361049605546,
+      "loss": 1.5815,
+      "step": 1817
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003381756131677709,
+      "loss": 1.8485,
+      "step": 1818
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033810758526858455,
+      "loss": 1.7805,
+      "step": 1819
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033803952681354737,
+      "loss": 1.8877,
+      "step": 1820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033797143781771693,
+      "loss": 1.9689,
+      "step": 1821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033790331829615773,
+      "loss": 1.8143,
+      "step": 1822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033783516826394106,
+      "loss": 1.7215,
+      "step": 1823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033776698773614475,
+      "loss": 1.8892,
+      "step": 1824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003376987767278537,
+      "loss": 1.7881,
+      "step": 1825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033763053525415925,
+      "loss": 1.6711,
+      "step": 1826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003375622633301596,
+      "loss": 1.742,
+      "step": 1827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003374939609709598,
+      "loss": 1.6847,
+      "step": 1828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003374256281916714,
+      "loss": 1.6528,
+      "step": 1829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003373572650074129,
+      "loss": 1.7498,
+      "step": 1830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003372888714333094,
+      "loss": 1.7824,
+      "step": 1831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033722044748449266,
+      "loss": 1.5613,
+      "step": 1832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003371519931761014,
+      "loss": 1.6096,
+      "step": 1833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003370835085232808,
+      "loss": 1.7017,
+      "step": 1834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033701499354118293,
+      "loss": 1.7828,
+      "step": 1835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003369464482449665,
+      "loss": 1.7672,
+      "step": 1836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033687787264979695,
+      "loss": 1.737,
+      "step": 1837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033680926677084633,
+      "loss": 1.8943,
+      "step": 1838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033674063062329347,
+      "loss": 1.8164,
+      "step": 1839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000336671964222324,
+      "loss": 1.7339,
+      "step": 1840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033660326758313007,
+      "loss": 1.7299,
+      "step": 1841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003365345407209105,
+      "loss": 1.6379,
+      "step": 1842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033646578365087107,
+      "loss": 1.6146,
+      "step": 1843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003363969963882239,
+      "loss": 1.8794,
+      "step": 1844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000336328178948188,
+      "loss": 1.4869,
+      "step": 1845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033625933134598905,
+      "loss": 1.8477,
+      "step": 1846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003361904535968593,
+      "loss": 1.7512,
+      "step": 1847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033612154571603767,
+      "loss": 1.8492,
+      "step": 1848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033605260771877,
+      "loss": 1.8091,
+      "step": 1849
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033598363962030836,
+      "loss": 1.8472,
+      "step": 1850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033591464143591186,
+      "loss": 1.8334,
+      "step": 1851
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003358456131808461,
+      "loss": 1.782,
+      "step": 1852
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033577655487038334,
+      "loss": 1.6368,
+      "step": 1853
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003357074665198025,
+      "loss": 1.754,
+      "step": 1854
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003356383481443891,
+      "loss": 1.6688,
+      "step": 1855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003355691997594355,
+      "loss": 1.4993,
+      "step": 1856
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003355000213802404,
+      "loss": 1.7421,
+      "step": 1857
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033543081302210944,
+      "loss": 1.6872,
+      "step": 1858
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033536157470035453,
+      "loss": 1.6958,
+      "step": 1859
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033529230643029466,
+      "loss": 1.7615,
+      "step": 1860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033522300822725503,
+      "loss": 1.6733,
+      "step": 1861
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003351536801065677,
+      "loss": 2.001,
+      "step": 1862
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003350843220835714,
+      "loss": 1.8624,
+      "step": 1863
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033501493417361115,
+      "loss": 1.6982,
+      "step": 1864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033494551639203895,
+      "loss": 1.7382,
+      "step": 1865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003348760687542132,
+      "loss": 1.8463,
+      "step": 1866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000334806591275499,
+      "loss": 1.874,
+      "step": 1867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000334737083971268,
+      "loss": 1.6682,
+      "step": 1868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003346675468568985,
+      "loss": 1.8209,
+      "step": 1869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003345979799477752,
+      "loss": 1.7793,
+      "step": 1870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033452838325928976,
+      "loss": 1.9424,
+      "step": 1871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003344587568068401,
+      "loss": 1.836,
+      "step": 1872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003343891006058309,
+      "loss": 1.7035,
+      "step": 1873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003343194146716734,
+      "loss": 1.7711,
+      "step": 1874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033424969901978515,
+      "loss": 1.558,
+      "step": 1875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033417995366559077,
+      "loss": 1.7025,
+      "step": 1876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033411017862452114,
+      "loss": 1.7982,
+      "step": 1877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003340403739120136,
+      "loss": 1.8778,
+      "step": 1878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003339705395435124,
+      "loss": 1.8173,
+      "step": 1879
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003339006755344682,
+      "loss": 1.7381,
+      "step": 1880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000333830781900338,
+      "loss": 1.7856,
+      "step": 1881
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003337608586565856,
+      "loss": 1.5277,
+      "step": 1882
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003336909058186814,
+      "loss": 1.8061,
+      "step": 1883
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003336209234021021,
+      "loss": 1.7612,
+      "step": 1884
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003335509114223311,
+      "loss": 1.6562,
+      "step": 1885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033348086989485833,
+      "loss": 1.7297,
+      "step": 1886
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033341079883518037,
+      "loss": 1.583,
+      "step": 1887
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003333406982588,
+      "loss": 1.7747,
+      "step": 1888
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003332705681812269,
+      "loss": 1.8446,
+      "step": 1889
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003332004086179771,
+      "loss": 1.8181,
+      "step": 1890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003331302195845731,
+      "loss": 1.9407,
+      "step": 1891
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033306000109654407,
+      "loss": 1.6876,
+      "step": 1892
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003329897531694256,
+      "loss": 1.5923,
+      "step": 1893
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003329194758187598,
+      "loss": 1.8258,
+      "step": 1894
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003328491690600953,
+      "loss": 1.7792,
+      "step": 1895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003327788329089872,
+      "loss": 1.5446,
+      "step": 1896
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033270846738099727,
+      "loss": 1.922,
+      "step": 1897
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033263807249169354,
+      "loss": 1.6979,
+      "step": 1898
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003325676482566507,
+      "loss": 1.6817,
+      "step": 1899
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003324971946914499,
+      "loss": 1.7273,
+      "step": 1900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033242671181167864,
+      "loss": 1.8522,
+      "step": 1901
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033235619963293113,
+      "loss": 1.7962,
+      "step": 1902
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033228565817080793,
+      "loss": 1.9089,
+      "step": 1903
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003322150874409162,
+      "loss": 1.7465,
+      "step": 1904
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003321444874588693,
+      "loss": 1.8121,
+      "step": 1905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003320738582402874,
+      "loss": 1.4758,
+      "step": 1906
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003320031998007969,
+      "loss": 1.8918,
+      "step": 1907
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003319325121560308,
+      "loss": 1.806,
+      "step": 1908
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003318617953216285,
+      "loss": 1.7868,
+      "step": 1909
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033179104931323585,
+      "loss": 1.7799,
+      "step": 1910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003317202741465051,
+      "loss": 1.7588,
+      "step": 1911
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003316494698370952,
+      "loss": 1.6923,
+      "step": 1912
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003315786364006713,
+      "loss": 1.7407,
+      "step": 1913
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003315077738529049,
+      "loss": 1.8134,
+      "step": 1914
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033143688220947437,
+      "loss": 1.781,
+      "step": 1915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033136596148606413,
+      "loss": 1.7402,
+      "step": 1916
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033129501169836505,
+      "loss": 1.6339,
+      "step": 1917
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003312240328620747,
+      "loss": 1.8861,
+      "step": 1918
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033115302499289685,
+      "loss": 1.8319,
+      "step": 1919
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033108198810654184,
+      "loss": 1.7066,
+      "step": 1920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003310109222187262,
+      "loss": 1.7625,
+      "step": 1921
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033093982734517315,
+      "loss": 1.6236,
+      "step": 1922
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033086870350161206,
+      "loss": 1.5948,
+      "step": 1923
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033079755070377904,
+      "loss": 1.762,
+      "step": 1924
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003307263689674163,
+      "loss": 1.7229,
+      "step": 1925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033065515830827254,
+      "loss": 1.7489,
+      "step": 1926
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033058391874210287,
+      "loss": 1.9086,
+      "step": 1927
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003305126502846689,
+      "loss": 1.7961,
+      "step": 1928
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003304413529517386,
+      "loss": 1.6996,
+      "step": 1929
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033037002675908607,
+      "loss": 1.775,
+      "step": 1930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033029867172249216,
+      "loss": 1.7898,
+      "step": 1931
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003302272878577438,
+      "loss": 1.6773,
+      "step": 1932
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003301558751806346,
+      "loss": 1.6026,
+      "step": 1933
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003300844337069643,
+      "loss": 1.6076,
+      "step": 1934
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003300129634525391,
+      "loss": 1.822,
+      "step": 1935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003299414644331715,
+      "loss": 1.7782,
+      "step": 1936
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003298699366646805,
+      "loss": 1.7169,
+      "step": 1937
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003297983801628913,
+      "loss": 1.7703,
+      "step": 1938
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003297267949436357,
+      "loss": 1.8547,
+      "step": 1939
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032965518102275157,
+      "loss": 1.6644,
+      "step": 1940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032958353841608325,
+      "loss": 1.7283,
+      "step": 1941
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003295118671394814,
+      "loss": 1.7802,
+      "step": 1942
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032944016720880324,
+      "loss": 1.791,
+      "step": 1943
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032936843863991196,
+      "loss": 1.6316,
+      "step": 1944
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032929668144867735,
+      "loss": 1.7226,
+      "step": 1945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003292248956509754,
+      "loss": 1.7702,
+      "step": 1946
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003291530812626886,
+      "loss": 1.5853,
+      "step": 1947
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032908123829970555,
+      "loss": 1.6864,
+      "step": 1948
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032900936677792125,
+      "loss": 1.7009,
+      "step": 1949
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000328937466713237,
+      "loss": 1.8318,
+      "step": 1950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032886553812156073,
+      "loss": 1.7907,
+      "step": 1951
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032879358101880606,
+      "loss": 1.7246,
+      "step": 1952
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003287215954208934,
+      "loss": 1.8426,
+      "step": 1953
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032864958134374943,
+      "loss": 1.8673,
+      "step": 1954
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003285775388033069,
+      "loss": 1.8725,
+      "step": 1955
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000328505467815505,
+      "loss": 1.8212,
+      "step": 1956
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032843336839628927,
+      "loss": 1.7711,
+      "step": 1957
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032836124056161137,
+      "loss": 1.6974,
+      "step": 1958
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003282890843274294,
+      "loss": 1.8601,
+      "step": 1959
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032821689970970777,
+      "loss": 1.706,
+      "step": 1960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032814468672441696,
+      "loss": 1.5503,
+      "step": 1961
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003280724453875339,
+      "loss": 1.7486,
+      "step": 1962
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032800017571504175,
+      "loss": 1.7722,
+      "step": 1963
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003279278777229299,
+      "loss": 1.6901,
+      "step": 1964
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003278555514271942,
+      "loss": 1.8429,
+      "step": 1965
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003277831968438364,
+      "loss": 1.8266,
+      "step": 1966
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032771081398886486,
+      "loss": 1.5982,
+      "step": 1967
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000327638402878294,
+      "loss": 1.7823,
+      "step": 1968
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032756596352814457,
+      "loss": 1.7353,
+      "step": 1969
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003274934959544434,
+      "loss": 1.7075,
+      "step": 1970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032742100017322395,
+      "loss": 1.7154,
+      "step": 1971
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032734847620052545,
+      "loss": 1.647,
+      "step": 1972
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003272759240523937,
+      "loss": 1.7006,
+      "step": 1973
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003272033437448806,
+      "loss": 1.8209,
+      "step": 1974
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003271307352940443,
+      "loss": 1.6779,
+      "step": 1975
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003270580987159492,
+      "loss": 1.7442,
+      "step": 1976
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003269854340266659,
+      "loss": 1.6925,
+      "step": 1977
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032691274124227123,
+      "loss": 1.8078,
+      "step": 1978
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003268400203788482,
+      "loss": 1.8137,
+      "step": 1979
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003267672714524861,
+      "loss": 1.8213,
+      "step": 1980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003266944944792803,
+      "loss": 1.8623,
+      "step": 1981
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032662168947533265,
+      "loss": 1.7186,
+      "step": 1982
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032654885645675086,
+      "loss": 1.8255,
+      "step": 1983
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003264759954396491,
+      "loss": 1.673,
+      "step": 1984
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003264031064401475,
+      "loss": 1.6989,
+      "step": 1985
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032633018947437266,
+      "loss": 1.7535,
+      "step": 1986
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003262572445584571,
+      "loss": 1.8201,
+      "step": 1987
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003261842717085397,
+      "loss": 1.7521,
+      "step": 1988
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003261112709407655,
+      "loss": 1.7498,
+      "step": 1989
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032603824227128556,
+      "loss": 1.6727,
+      "step": 1990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003259651857162574,
+      "loss": 1.9731,
+      "step": 1991
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003258921012918444,
+      "loss": 1.8877,
+      "step": 1992
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032581898901421635,
+      "loss": 1.7571,
+      "step": 1993
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003257458488995491,
+      "loss": 1.7152,
+      "step": 1994
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032567268096402453,
+      "loss": 1.8582,
+      "step": 1995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032559948522383093,
+      "loss": 1.8215,
+      "step": 1996
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003255262616951626,
+      "loss": 1.7961,
+      "step": 1997
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032545301039422,
+      "loss": 1.7723,
+      "step": 1998
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032537973133720964,
+      "loss": 1.8633,
+      "step": 1999
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003253064245403445,
+      "loss": 1.664,
+      "step": 2000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003252330900198432,
+      "loss": 1.7441,
+      "step": 2001
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032515972779193094,
+      "loss": 1.8134,
+      "step": 2002
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003250863378728388,
+      "loss": 1.6695,
+      "step": 2003
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032501292027880413,
+      "loss": 1.6399,
+      "step": 2004
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003249394750260703,
+      "loss": 1.7118,
+      "step": 2005
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032486600213088677,
+      "loss": 1.6273,
+      "step": 2006
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032479250160950926,
+      "loss": 1.8432,
+      "step": 2007
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032471897347819946,
+      "loss": 1.6697,
+      "step": 2008
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003246454177532253,
+      "loss": 1.7412,
+      "step": 2009
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032457183445086063,
+      "loss": 1.5673,
+      "step": 2010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003244982235873856,
+      "loss": 1.5494,
+      "step": 2011
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003244245851790864,
+      "loss": 1.8153,
+      "step": 2012
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003243509192422551,
+      "loss": 1.7135,
+      "step": 2013
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003242772257931904,
+      "loss": 1.7113,
+      "step": 2014
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003242035048481964,
+      "loss": 1.8381,
+      "step": 2015
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032412975642358375,
+      "loss": 1.7193,
+      "step": 2016
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000324055980535669,
+      "loss": 1.7164,
+      "step": 2017
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032398217720077494,
+      "loss": 1.6114,
+      "step": 2018
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032390834643523025,
+      "loss": 1.6986,
+      "step": 2019
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003238344882553697,
+      "loss": 1.7609,
+      "step": 2020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003237606026775341,
+      "loss": 1.6817,
+      "step": 2021
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003236866897180706,
+      "loss": 1.6873,
+      "step": 2022
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003236127493933321,
+      "loss": 1.7534,
+      "step": 2023
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003235387817196776,
+      "loss": 1.8136,
+      "step": 2024
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003234647867134723,
+      "loss": 1.7058,
+      "step": 2025
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032339076439108727,
+      "loss": 1.8149,
+      "step": 2026
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032331671476889975,
+      "loss": 1.8045,
+      "step": 2027
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000323242637863293,
+      "loss": 1.7454,
+      "step": 2028
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032316853369065626,
+      "loss": 1.7279,
+      "step": 2029
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003230944022673848,
+      "loss": 1.6452,
+      "step": 2030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032302024360988,
+      "loss": 1.7889,
+      "step": 2031
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003229460577345493,
+      "loss": 1.9362,
+      "step": 2032
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003228718446578059,
+      "loss": 1.7183,
+      "step": 2033
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003227976043960693,
+      "loss": 1.7026,
+      "step": 2034
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032272333696576494,
+      "loss": 1.7087,
+      "step": 2035
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003226490423833242,
+      "loss": 1.7274,
+      "step": 2036
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032257472066518454,
+      "loss": 1.9432,
+      "step": 2037
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003225003718277894,
+      "loss": 1.8579,
+      "step": 2038
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003224259958875882,
+      "loss": 1.6696,
+      "step": 2039
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003223515928610363,
+      "loss": 1.5806,
+      "step": 2040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003222771627645953,
+      "loss": 1.8637,
+      "step": 2041
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003222027056147325,
+      "loss": 1.6104,
+      "step": 2042
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003221282214279213,
+      "loss": 1.7341,
+      "step": 2043
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003220537102206411,
+      "loss": 1.791,
+      "step": 2044
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032197917200937724,
+      "loss": 1.6904,
+      "step": 2045
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003219046068106211,
+      "loss": 1.7191,
+      "step": 2046
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032183001464086995,
+      "loss": 1.6296,
+      "step": 2047
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032175539551662706,
+      "loss": 1.664,
+      "step": 2048
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032168074945440167,
+      "loss": 1.7395,
+      "step": 2049
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032160607647070905,
+      "loss": 1.764,
+      "step": 2050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003215313765820702,
+      "loss": 1.9961,
+      "step": 2051
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032145664980501234,
+      "loss": 1.701,
+      "step": 2052
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003213818961560686,
+      "loss": 1.7558,
+      "step": 2053
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003213071156517778,
+      "loss": 1.7647,
+      "step": 2054
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000321232308308685,
+      "loss": 1.8378,
+      "step": 2055
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000321157474143341,
+      "loss": 1.8143,
+      "step": 2056
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032108261317230274,
+      "loss": 1.6666,
+      "step": 2057
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032100772541213276,
+      "loss": 1.7484,
+      "step": 2058
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003209328108793999,
+      "loss": 1.7416,
+      "step": 2059
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003208578695906788,
+      "loss": 1.6978,
+      "step": 2060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003207829015625498,
+      "loss": 1.7207,
+      "step": 2061
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032070790681159943,
+      "loss": 1.7672,
+      "step": 2062
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032063288535441993,
+      "loss": 2.0049,
+      "step": 2063
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032055783720760973,
+      "loss": 1.7368,
+      "step": 2064
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003204827623877729,
+      "loss": 1.6372,
+      "step": 2065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032040766091151945,
+      "loss": 1.9412,
+      "step": 2066
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003203325327954654,
+      "loss": 1.695,
+      "step": 2067
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003202573780562326,
+      "loss": 1.7543,
+      "step": 2068
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032018219671044873,
+      "loss": 1.788,
+      "step": 2069
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032010698877474753,
+      "loss": 1.7899,
+      "step": 2070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032003175426576836,
+      "loss": 1.8932,
+      "step": 2071
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003199564932001567,
+      "loss": 1.8116,
+      "step": 2072
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003198812055945638,
+      "loss": 1.8071,
+      "step": 2073
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003198058914656468,
+      "loss": 1.7533,
+      "step": 2074
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003197305508300687,
+      "loss": 1.8424,
+      "step": 2075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003196551837044983,
+      "loss": 1.7057,
+      "step": 2076
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031957979010561045,
+      "loss": 1.7617,
+      "step": 2077
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003195043700500857,
+      "loss": 1.783,
+      "step": 2078
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003194289235546105,
+      "loss": 1.9042,
+      "step": 2079
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000319353450635877,
+      "loss": 1.5693,
+      "step": 2080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031927795131058356,
+      "loss": 1.8472,
+      "step": 2081
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000319202425595434,
+      "loss": 1.5701,
+      "step": 2082
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031912687350713824,
+      "loss": 1.75,
+      "step": 2083
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003190512950624117,
+      "loss": 1.7216,
+      "step": 2084
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003189756902779762,
+      "loss": 1.7525,
+      "step": 2085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031890005917055897,
+      "loss": 1.8698,
+      "step": 2086
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031882440175689294,
+      "loss": 1.6577,
+      "step": 2087
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003187487180537172,
+      "loss": 1.6611,
+      "step": 2088
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003186730080777765,
+      "loss": 1.6604,
+      "step": 2089
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003185972718458215,
+      "loss": 1.8525,
+      "step": 2090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003185215093746084,
+      "loss": 1.7193,
+      "step": 2091
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031844572068089966,
+      "loss": 1.7709,
+      "step": 2092
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003183699057814631,
+      "loss": 1.7324,
+      "step": 2093
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003182940646930726,
+      "loss": 1.8166,
+      "step": 2094
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003182181974325077,
+      "loss": 1.8243,
+      "step": 2095
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003181423040165538,
+      "loss": 1.7204,
+      "step": 2096
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031806638446200204,
+      "loss": 1.684,
+      "step": 2097
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003179904387856494,
+      "loss": 1.8559,
+      "step": 2098
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003179144670042987,
+      "loss": 1.8468,
+      "step": 2099
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003178384691347583,
+      "loss": 1.7464,
+      "step": 2100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003177624451938426,
+      "loss": 1.7719,
+      "step": 2101
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003176863951983715,
+      "loss": 1.7057,
+      "step": 2102
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031761031916517103,
+      "loss": 1.6337,
+      "step": 2103
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003175342171110725,
+      "loss": 1.7411,
+      "step": 2104
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003174580890529135,
+      "loss": 1.7827,
+      "step": 2105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031738193500753693,
+      "loss": 1.6105,
+      "step": 2106
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031730575499179167,
+      "loss": 1.6262,
+      "step": 2107
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003172295490225324,
+      "loss": 1.7294,
+      "step": 2108
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003171533171166193,
+      "loss": 1.7586,
+      "step": 2109
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031707705929091845,
+      "loss": 1.8016,
+      "step": 2110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031700077556230167,
+      "loss": 1.75,
+      "step": 2111
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031692446594764657,
+      "loss": 1.7696,
+      "step": 2112
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031684813046383624,
+      "loss": 1.5171,
+      "step": 2113
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031677176912775983,
+      "loss": 1.7547,
+      "step": 2114
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031669538195631193,
+      "loss": 1.7117,
+      "step": 2115
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003166189689663929,
+      "loss": 1.6944,
+      "step": 2116
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031654253017490897,
+      "loss": 1.7516,
+      "step": 2117
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031646606559877197,
+      "loss": 1.6439,
+      "step": 2118
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031638957525489936,
+      "loss": 1.7281,
+      "step": 2119
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003163130591602144,
+      "loss": 1.6504,
+      "step": 2120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003162365173316461,
+      "loss": 1.6118,
+      "step": 2121
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003161599497861291,
+      "loss": 1.653,
+      "step": 2122
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003160833565406035,
+      "loss": 1.8261,
+      "step": 2123
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031600673761201554,
+      "loss": 1.8132,
+      "step": 2124
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031593009301731685,
+      "loss": 1.8917,
+      "step": 2125
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031585342277346473,
+      "loss": 1.6661,
+      "step": 2126
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003157767268974222,
+      "loss": 1.7429,
+      "step": 2127
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031570000540615814,
+      "loss": 1.6725,
+      "step": 2128
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003156232583166468,
+      "loss": 1.5958,
+      "step": 2129
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031554648564586824,
+      "loss": 1.5605,
+      "step": 2130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003154696874108081,
+      "loss": 1.6119,
+      "step": 2131
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031539286362845787,
+      "loss": 1.649,
+      "step": 2132
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003153160143158145,
+      "loss": 1.6445,
+      "step": 2133
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003152391394898806,
+      "loss": 1.6548,
+      "step": 2134
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003151622391676645,
+      "loss": 1.6915,
+      "step": 2135
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003150853133661802,
+      "loss": 1.7782,
+      "step": 2136
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031500836210244723,
+      "loss": 1.6837,
+      "step": 2137
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003149313853934908,
+      "loss": 1.5575,
+      "step": 2138
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031485438325634186,
+      "loss": 1.6626,
+      "step": 2139
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003147773557080367,
+      "loss": 1.64,
+      "step": 2140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003147003027656175,
+      "loss": 1.7764,
+      "step": 2141
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000314623224446132,
+      "loss": 1.7143,
+      "step": 2142
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003145461207666335,
+      "loss": 1.5473,
+      "step": 2143
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031446899174418096,
+      "loss": 1.7505,
+      "step": 2144
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003143918373958389,
+      "loss": 1.7335,
+      "step": 2145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031431465773867743,
+      "loss": 1.7235,
+      "step": 2146
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031423745278977234,
+      "loss": 1.9604,
+      "step": 2147
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000314160222566205,
+      "loss": 1.5897,
+      "step": 2148
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031408296708506235,
+      "loss": 2.0507,
+      "step": 2149
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031400568636343675,
+      "loss": 1.7074,
+      "step": 2150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003139283804184265,
+      "loss": 1.7496,
+      "step": 2151
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031385104926713527,
+      "loss": 1.6388,
+      "step": 2152
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003137736929266722,
+      "loss": 1.7112,
+      "step": 2153
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003136963114141522,
+      "loss": 1.5898,
+      "step": 2154
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003136189047466957,
+      "loss": 1.829,
+      "step": 2155
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003135414729414286,
+      "loss": 1.6975,
+      "step": 2156
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031346401601548257,
+      "loss": 1.7093,
+      "step": 2157
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003133865339859945,
+      "loss": 1.5595,
+      "step": 2158
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003133090268701072,
+      "loss": 1.6826,
+      "step": 2159
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031323149468496885,
+      "loss": 1.7782,
+      "step": 2160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003131539374477331,
+      "loss": 1.7749,
+      "step": 2161
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003130763551755593,
+      "loss": 1.8205,
+      "step": 2162
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031299874788561226,
+      "loss": 1.6717,
+      "step": 2163
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003129211155950624,
+      "loss": 1.8268,
+      "step": 2164
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003128434583210855,
+      "loss": 1.8399,
+      "step": 2165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000312765776080863,
+      "loss": 1.8906,
+      "step": 2166
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003126880688915819,
+      "loss": 1.9352,
+      "step": 2167
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031261033677043457,
+      "loss": 1.7202,
+      "step": 2168
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003125325797346191,
+      "loss": 1.7867,
+      "step": 2169
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003124547978013389,
+      "loss": 1.72,
+      "step": 2170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031237699098780305,
+      "loss": 1.5987,
+      "step": 2171
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000312299159311226,
+      "loss": 1.7277,
+      "step": 2172
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003122213027888277,
+      "loss": 1.7127,
+      "step": 2173
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003121434214378337,
+      "loss": 1.582,
+      "step": 2174
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031206551527547505,
+      "loss": 1.7773,
+      "step": 2175
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003119875843189881,
+      "loss": 1.7968,
+      "step": 2176
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000311909628585615,
+      "loss": 1.803,
+      "step": 2177
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000311831648092603,
+      "loss": 1.8254,
+      "step": 2178
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003117536428572052,
+      "loss": 1.6005,
+      "step": 2179
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003116756128966799,
+      "loss": 1.7869,
+      "step": 2180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031159755822829095,
+      "loss": 1.786,
+      "step": 2181
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003115194788693079,
+      "loss": 1.705,
+      "step": 2182
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003114413748370052,
+      "loss": 1.7512,
+      "step": 2183
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031136324614866346,
+      "loss": 1.6969,
+      "step": 2184
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031128509282156815,
+      "loss": 1.7838,
+      "step": 2185
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003112069148730106,
+      "loss": 1.847,
+      "step": 2186
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031112871232028726,
+      "loss": 1.8593,
+      "step": 2187
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031105048518070036,
+      "loss": 1.6846,
+      "step": 2188
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003109722334715573,
+      "loss": 1.7778,
+      "step": 2189
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031089395721017106,
+      "loss": 1.521,
+      "step": 2190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031081565641385994,
+      "loss": 1.9404,
+      "step": 2191
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003107373310999478,
+      "loss": 1.9052,
+      "step": 2192
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031065898128576386,
+      "loss": 1.7406,
+      "step": 2193
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003105806069886427,
+      "loss": 1.7166,
+      "step": 2194
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031050220822592445,
+      "loss": 1.7871,
+      "step": 2195
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003104237850149545,
+      "loss": 1.6588,
+      "step": 2196
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003103453373730838,
+      "loss": 1.7343,
+      "step": 2197
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031026686531766855,
+      "loss": 1.816,
+      "step": 2198
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031018836886607064,
+      "loss": 1.7613,
+      "step": 2199
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031010984803565687,
+      "loss": 1.7249,
+      "step": 2200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031003130284379984,
+      "loss": 1.6767,
+      "step": 2201
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00030995273330787753,
+      "loss": 1.6741,
+      "step": 2202
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00030987413944527306,
+      "loss": 1.7595,
+      "step": 2203
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003097955212733751,
+      "loss": 1.6407,
+      "step": 2204
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030971687880957755,
+      "loss": 1.5947,
+      "step": 2205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030963821207128,
+      "loss": 1.7871,
+      "step": 2206
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030955952107588713,
+      "loss": 1.8245,
+      "step": 2207
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030948080584080896,
+      "loss": 1.806,
+      "step": 2208
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003094020663834611,
+      "loss": 1.7028,
+      "step": 2209
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003093233027212644,
+      "loss": 1.6574,
+      "step": 2210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030924451487164507,
+      "loss": 1.6796,
+      "step": 2211
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003091657028520345,
+      "loss": 1.7892,
+      "step": 2212
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030908686667986977,
+      "loss": 1.7974,
+      "step": 2213
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003090080063725931,
+      "loss": 1.8121,
+      "step": 2214
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003089291219476519,
+      "loss": 1.8421,
+      "step": 2215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003088502134224994,
+      "loss": 1.837,
+      "step": 2216
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003087712808145936,
+      "loss": 1.6128,
+      "step": 2217
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003086923241413982,
+      "loss": 1.7933,
+      "step": 2218
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000308613343420382,
+      "loss": 1.7699,
+      "step": 2219
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003085343386690194,
+      "loss": 1.732,
+      "step": 2220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003084553099047897,
+      "loss": 1.9072,
+      "step": 2221
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030837625714517794,
+      "loss": 1.9054,
+      "step": 2222
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030829718040767433,
+      "loss": 1.6786,
+      "step": 2223
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003082180797097742,
+      "loss": 1.6489,
+      "step": 2224
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030813895506897834,
+      "loss": 1.6425,
+      "step": 2225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030805980650279284,
+      "loss": 1.6192,
+      "step": 2226
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003079806340287291,
+      "loss": 1.7899,
+      "step": 2227
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003079014376643037,
+      "loss": 1.7219,
+      "step": 2228
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003078222174270386,
+      "loss": 1.7063,
+      "step": 2229
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030774297333446105,
+      "loss": 1.6725,
+      "step": 2230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003076637054041036,
+      "loss": 1.688,
+      "step": 2231
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003075844136535039,
+      "loss": 1.8319,
+      "step": 2232
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030750509810020495,
+      "loss": 1.7618,
+      "step": 2233
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030742575876175525,
+      "loss": 1.6607,
+      "step": 2234
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003073463956557082,
+      "loss": 1.5615,
+      "step": 2235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003072670087996227,
+      "loss": 1.794,
+      "step": 2236
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003071875982110628,
+      "loss": 1.9005,
+      "step": 2237
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030710816390759794,
+      "loss": 1.5293,
+      "step": 2238
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030702870590680256,
+      "loss": 1.8186,
+      "step": 2239
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030694922422625653,
+      "loss": 1.6743,
+      "step": 2240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003068697188835449,
+      "loss": 1.7703,
+      "step": 2241
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000306790189896258,
+      "loss": 1.6306,
+      "step": 2242
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030671063728199136,
+      "loss": 1.8208,
+      "step": 2243
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003066310610583457,
+      "loss": 1.8369,
+      "step": 2244
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030655146124292706,
+      "loss": 1.7201,
+      "step": 2245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030647183785334654,
+      "loss": 1.8017,
+      "step": 2246
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003063921909072206,
+      "loss": 1.8814,
+      "step": 2247
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030631252042217084,
+      "loss": 1.6655,
+      "step": 2248
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003062328264158242,
+      "loss": 1.6843,
+      "step": 2249
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003061531089058127,
+      "loss": 1.6878,
+      "step": 2250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030607336790977336,
+      "loss": 1.7302,
+      "step": 2251
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003059936034453489,
+      "loss": 1.6636,
+      "step": 2252
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003059138155301868,
+      "loss": 1.8216,
+      "step": 2253
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030583400418193984,
+      "loss": 1.7926,
+      "step": 2254
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030575416941826613,
+      "loss": 1.5939,
+      "step": 2255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003056743112568288,
+      "loss": 1.8611,
+      "step": 2256
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030559442971529624,
+      "loss": 1.7131,
+      "step": 2257
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003055145248113419,
+      "loss": 1.7386,
+      "step": 2258
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030543459656264454,
+      "loss": 1.6317,
+      "step": 2259
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000305354644986888,
+      "loss": 1.6881,
+      "step": 2260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003052746701017613,
+      "loss": 1.7415,
+      "step": 2261
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003051946719249586,
+      "loss": 1.6416,
+      "step": 2262
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030511465047417934,
+      "loss": 1.6744,
+      "step": 2263
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003050346057671279,
+      "loss": 1.7008,
+      "step": 2264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030495453782151396,
+      "loss": 1.544,
+      "step": 2265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003048744466550523,
+      "loss": 1.68,
+      "step": 2266
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003047943322854627,
+      "loss": 1.6806,
+      "step": 2267
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030471419473047034,
+      "loss": 1.8234,
+      "step": 2268
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003046340340078054,
+      "loss": 1.7955,
+      "step": 2269
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030455385013520306,
+      "loss": 1.8255,
+      "step": 2270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003044736431304039,
+      "loss": 1.8485,
+      "step": 2271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003043934130111533,
+      "loss": 1.8019,
+      "step": 2272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000304313159795202,
+      "loss": 1.517,
+      "step": 2273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003042328835003057,
+      "loss": 1.836,
+      "step": 2274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003041525841442253,
+      "loss": 1.7349,
+      "step": 2275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003040722617447268,
+      "loss": 1.7473,
+      "step": 2276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003039919163195813,
+      "loss": 1.8429,
+      "step": 2277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030391154788656486,
+      "loss": 1.882,
+      "step": 2278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030383115646345874,
+      "loss": 1.5029,
+      "step": 2279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030375074206804935,
+      "loss": 1.637,
+      "step": 2280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030367030471812814,
+      "loss": 1.6749,
+      "step": 2281
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030358984443149147,
+      "loss": 1.6641,
+      "step": 2282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030350936122594096,
+      "loss": 1.7489,
+      "step": 2283
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003034288551192834,
+      "loss": 1.8625,
+      "step": 2284
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003033483261293303,
+      "loss": 1.7366,
+      "step": 2285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030326777427389854,
+      "loss": 1.7835,
+      "step": 2286
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030318719957081,
+      "loss": 1.785,
+      "step": 2287
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003031066020378914,
+      "loss": 1.7538,
+      "step": 2288
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003030259816929749,
+      "loss": 1.7797,
+      "step": 2289
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030294533855389737,
+      "loss": 1.7425,
+      "step": 2290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003028646726385008,
+      "loss": 1.8594,
+      "step": 2291
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030278398396463237,
+      "loss": 1.7566,
+      "step": 2292
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000302703272550144,
+      "loss": 1.7785,
+      "step": 2293
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030262253841289305,
+      "loss": 1.8956,
+      "step": 2294
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030254178157074153,
+      "loss": 1.9331,
+      "step": 2295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003024610020415567,
+      "loss": 1.7571,
+      "step": 2296
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003024610020415567,
+      "loss": 1.6222,
+      "step": 2297
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003023801998432108,
+      "loss": 1.7898,
+      "step": 2298
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030229937499358096,
+      "loss": 1.6553,
+      "step": 2299
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030221852751054944,
+      "loss": 1.826,
+      "step": 2300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003021376574120035,
+      "loss": 1.7846,
+      "step": 2301
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030205676471583534,
+      "loss": 1.7511,
+      "step": 2302
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003019758494399423,
+      "loss": 1.8142,
+      "step": 2303
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003018949116022265,
+      "loss": 1.6221,
+      "step": 2304
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003018139512205953,
+      "loss": 1.5109,
+      "step": 2305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030173296831296077,
+      "loss": 1.7615,
+      "step": 2306
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003016519628972403,
+      "loss": 1.6906,
+      "step": 2307
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003015709349913558,
+      "loss": 1.7006,
+      "step": 2308
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003014898846132347,
+      "loss": 1.773,
+      "step": 2309
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030140881178080905,
+      "loss": 1.618,
+      "step": 2310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003013277165120159,
+      "loss": 1.7345,
+      "step": 2311
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030124659882479733,
+      "loss": 1.8366,
+      "step": 2312
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003011654587371004,
+      "loss": 1.6014,
+      "step": 2313
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000301084296266877,
+      "loss": 1.6743,
+      "step": 2314
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030100311143208415,
+      "loss": 1.7666,
+      "step": 2315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003009219042506837,
+      "loss": 1.7231,
+      "step": 2316
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030084067474064245,
+      "loss": 1.8799,
+      "step": 2317
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003007594229199322,
+      "loss": 1.6812,
+      "step": 2318
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003006781488065296,
+      "loss": 1.7505,
+      "step": 2319
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003005968524184164,
+      "loss": 1.7822,
+      "step": 2320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000300515533773579,
+      "loss": 1.6462,
+      "step": 2321
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030043419289000906,
+      "loss": 1.7105,
+      "step": 2322
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030035282978570276,
+      "loss": 1.8376,
+      "step": 2323
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030027144447866157,
+      "loss": 1.6605,
+      "step": 2324
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003001900369868917,
+      "loss": 1.6142,
+      "step": 2325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003001086073284043,
+      "loss": 1.7499,
+      "step": 2326
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003000271555212154,
+      "loss": 1.8937,
+      "step": 2327
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029994568158334594,
+      "loss": 1.794,
+      "step": 2328
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002998641855328218,
+      "loss": 1.7652,
+      "step": 2329
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002997826673876736,
+      "loss": 1.6811,
+      "step": 2330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002997011271659371,
+      "loss": 1.6801,
+      "step": 2331
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002996195648856528,
+      "loss": 1.7425,
+      "step": 2332
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029953798056486596,
+      "loss": 1.843,
+      "step": 2333
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002994563742216269,
+      "loss": 1.7017,
+      "step": 2334
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002993747458739909,
+      "loss": 1.676,
+      "step": 2335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002992930955400178,
+      "loss": 1.6387,
+      "step": 2336
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029921142323777254,
+      "loss": 1.8372,
+      "step": 2337
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002991297289853247,
+      "loss": 1.8247,
+      "step": 2338
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002990480128007491,
+      "loss": 1.9391,
+      "step": 2339
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000298966274702125,
+      "loss": 1.7478,
+      "step": 2340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002988845147075368,
+      "loss": 1.8239,
+      "step": 2341
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002988027328350736,
+      "loss": 1.9829,
+      "step": 2342
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002987209291028293,
+      "loss": 1.6645,
+      "step": 2343
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002986391035289028,
+      "loss": 1.8642,
+      "step": 2344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002985572561313977,
+      "loss": 1.7456,
+      "step": 2345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002984753869284225,
+      "loss": 1.6965,
+      "step": 2346
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002983934959380904,
+      "loss": 1.8397,
+      "step": 2347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029831158317851963,
+      "loss": 1.8373,
+      "step": 2348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000298229648667833,
+      "loss": 1.7215,
+      "step": 2349
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029814769242415834,
+      "loss": 1.9098,
+      "step": 2350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029806571446562823,
+      "loss": 1.681,
+      "step": 2351
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002979837148103799,
+      "loss": 1.7317,
+      "step": 2352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029790169347655574,
+      "loss": 1.7801,
+      "step": 2353
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002978196504823024,
+      "loss": 1.8464,
+      "step": 2354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002977375858457718,
+      "loss": 1.7806,
+      "step": 2355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002976554995851205,
+      "loss": 1.6938,
+      "step": 2356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002975733917185097,
+      "loss": 1.707,
+      "step": 2357
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002974912622641055,
+      "loss": 1.7436,
+      "step": 2358
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002974091112400789,
+      "loss": 1.6292,
+      "step": 2359
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029732693866460545,
+      "loss": 1.7448,
+      "step": 2360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972447445558656,
+      "loss": 1.7517,
+      "step": 2361
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002971625289320445,
+      "loss": 1.74,
+      "step": 2362
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002970802918113321,
+      "loss": 1.8153,
+      "step": 2363
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029699803321192304,
+      "loss": 1.5963,
+      "step": 2364
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029691575315201685,
+      "loss": 1.6281,
+      "step": 2365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029683345164981773,
+      "loss": 1.7254,
+      "step": 2366
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002967511287235345,
+      "loss": 1.6897,
+      "step": 2367
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000296668784391381,
+      "loss": 1.7403,
+      "step": 2368
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029658641867157554,
+      "loss": 1.8458,
+      "step": 2369
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029650403158234124,
+      "loss": 1.6575,
+      "step": 2370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000296421623141906,
+      "loss": 1.6983,
+      "step": 2371
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029633919336850254,
+      "loss": 1.7076,
+      "step": 2372
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029625674228036803,
+      "loss": 1.7919,
+      "step": 2373
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002961742698957445,
+      "loss": 1.7098,
+      "step": 2374
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002960917762328788,
+      "loss": 1.6209,
+      "step": 2375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002960092613100223,
+      "loss": 1.6519,
+      "step": 2376
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002959267251454312,
+      "loss": 1.8932,
+      "step": 2377
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029584416775736634,
+      "loss": 1.6064,
+      "step": 2378
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002957615891640932,
+      "loss": 1.6869,
+      "step": 2379
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002956789893838821,
+      "loss": 1.859,
+      "step": 2380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029559636843500805,
+      "loss": 1.8283,
+      "step": 2381
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029551372633575057,
+      "loss": 1.8691,
+      "step": 2382
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029543106310439386,
+      "loss": 1.8261,
+      "step": 2383
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029534837875922707,
+      "loss": 1.8267,
+      "step": 2384
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002952656733185438,
+      "loss": 1.6479,
+      "step": 2385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029518294680064224,
+      "loss": 1.502,
+      "step": 2386
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002951001992238255,
+      "loss": 1.8225,
+      "step": 2387
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029501743060640113,
+      "loss": 1.6595,
+      "step": 2388
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029493464096668143,
+      "loss": 1.5535,
+      "step": 2389
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002948518303229833,
+      "loss": 1.6986,
+      "step": 2390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002947689986936284,
+      "loss": 1.8524,
+      "step": 2391
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002946861460969429,
+      "loss": 1.742,
+      "step": 2392
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029460327255125777,
+      "loss": 1.6215,
+      "step": 2393
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029452037807490834,
+      "loss": 1.8717,
+      "step": 2394
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002944374626862349,
+      "loss": 1.7523,
+      "step": 2395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002943545264035821,
+      "loss": 1.8094,
+      "step": 2396
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029427156924529936,
+      "loss": 1.6575,
+      "step": 2397
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002941885912297407,
+      "loss": 1.706,
+      "step": 2398
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029410559237526474,
+      "loss": 1.7351,
+      "step": 2399
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029402257270023463,
+      "loss": 1.6914,
+      "step": 2400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029393953222301827,
+      "loss": 1.5955,
+      "step": 2401
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029385647096198805,
+      "loss": 1.7433,
+      "step": 2402
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029377338893552105,
+      "loss": 1.649,
+      "step": 2403
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002936902861619988,
+      "loss": 1.8089,
+      "step": 2404
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002936071626598077,
+      "loss": 1.8165,
+      "step": 2405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002935240184473384,
+      "loss": 1.734,
+      "step": 2406
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002934408535429863,
+      "loss": 1.8245,
+      "step": 2407
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002933576679651514,
+      "loss": 1.8506,
+      "step": 2408
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029327446173223827,
+      "loss": 1.6976,
+      "step": 2409
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029319123486265595,
+      "loss": 1.6506,
+      "step": 2410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029310798737481805,
+      "loss": 1.8129,
+      "step": 2411
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029302471928714294,
+      "loss": 1.5705,
+      "step": 2412
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002929414306180534,
+      "loss": 1.606,
+      "step": 2413
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029285812138597674,
+      "loss": 1.7254,
+      "step": 2414
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002927747916093448,
+      "loss": 1.8198,
+      "step": 2415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002926914413065941,
+      "loss": 1.9041,
+      "step": 2416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029260807049616563,
+      "loss": 1.764,
+      "step": 2417
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029252467919650476,
+      "loss": 1.866,
+      "step": 2418
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002924412674260617,
+      "loss": 1.5901,
+      "step": 2419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029235783520329095,
+      "loss": 1.5519,
+      "step": 2420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002922743825466517,
+      "loss": 1.7261,
+      "step": 2421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029219090947460756,
+      "loss": 1.8539,
+      "step": 2422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029210741600562654,
+      "loss": 1.9267,
+      "step": 2423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002920239021581815,
+      "loss": 1.5395,
+      "step": 2424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029194036795074943,
+      "loss": 1.7556,
+      "step": 2425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029185681340181206,
+      "loss": 1.6988,
+      "step": 2426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002917732385298557,
+      "loss": 1.8104,
+      "step": 2427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029168964335337077,
+      "loss": 1.6062,
+      "step": 2428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002916060278908525,
+      "loss": 1.5787,
+      "step": 2429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002915223921608007,
+      "loss": 1.7371,
+      "step": 2430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029143873618171936,
+      "loss": 1.6513,
+      "step": 2431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029135505997211704,
+      "loss": 1.6997,
+      "step": 2432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029127136355050704,
+      "loss": 1.6701,
+      "step": 2433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029118764693540666,
+      "loss": 1.7482,
+      "step": 2434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029110391014533813,
+      "loss": 1.7327,
+      "step": 2435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029102015319882777,
+      "loss": 1.7508,
+      "step": 2436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002909363761144067,
+      "loss": 1.7604,
+      "step": 2437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029085257891061023,
+      "loss": 1.6601,
+      "step": 2438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002907687616059782,
+      "loss": 1.8266,
+      "step": 2439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000290684924219055,
+      "loss": 1.7184,
+      "step": 2440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002906010667683893,
+      "loss": 1.7782,
+      "step": 2441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029051718927253434,
+      "loss": 1.7202,
+      "step": 2442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002904332917500476,
+      "loss": 1.7985,
+      "step": 2443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029034937421949135,
+      "loss": 1.8575,
+      "step": 2444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029026543669943193,
+      "loss": 1.8512,
+      "step": 2445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002901814792084403,
+      "loss": 1.6356,
+      "step": 2446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002900975017650918,
+      "loss": 1.739,
+      "step": 2447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029001350438796606,
+      "loss": 1.6639,
+      "step": 2448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00028992948709564726,
+      "loss": 1.9028,
+      "step": 2449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00028984544990672396,
+      "loss": 1.8404,
+      "step": 2450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00028976139283978916,
+      "loss": 1.7267,
+      "step": 2451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002896773159134402,
+      "loss": 1.8926,
+      "step": 2452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002895932191462788,
+      "loss": 1.7985,
+      "step": 2453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00028950910255691107,
+      "loss": 1.8033,
+      "step": 2454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002894249661639476,
+      "loss": 1.7912,
+      "step": 2455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00028934080998600326,
+      "loss": 2.0232,
+      "step": 2456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002892566340416972,
+      "loss": 1.6735,
+      "step": 2457
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00028917243834965326,
+      "loss": 1.7066,
+      "step": 2458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002890882229284994,
+      "loss": 1.5722,
+      "step": 2459
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000289003987796868,
+      "loss": 1.7343,
+      "step": 2460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002889197329733958,
+      "loss": 1.6352,
+      "step": 2461
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002888354584767239,
+      "loss": 1.6335,
+      "step": 2462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002887511643254977,
+      "loss": 1.8765,
+      "step": 2463
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002886668505383672,
+      "loss": 1.7541,
+      "step": 2464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00028858251713398636,
+      "loss": 1.7712,
+      "step": 2465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00028849816413101367,
+      "loss": 1.8574,
+      "step": 2466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00028841379154811205,
+      "loss": 1.7498,
+      "step": 2467
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00028832939940394865,
+      "loss": 1.6536,
+      "step": 2468
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00028824498771719487,
+      "loss": 1.8075,
+      "step": 2469
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002881605565065266,
+      "loss": 1.8558,
+      "step": 2470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000288076105790624,
+      "loss": 1.6706,
+      "step": 2471
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002879916355881714,
+      "loss": 1.7126,
+      "step": 2472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002879071459178576,
+      "loss": 1.753,
+      "step": 2473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002878226367983757,
+      "loss": 1.6629,
+      "step": 2474
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028773810824842303,
+      "loss": 1.8175,
+      "step": 2475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002876535602867013,
+      "loss": 1.9046,
+      "step": 2476
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002875689929319164,
+      "loss": 1.7261,
+      "step": 2477
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002874844062027787,
+      "loss": 1.5536,
+      "step": 2478
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028739980011800254,
+      "loss": 1.8208,
+      "step": 2479
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002873151746963069,
+      "loss": 1.5895,
+      "step": 2480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002872305299564149,
+      "loss": 1.773,
+      "step": 2481
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028714586591705365,
+      "loss": 1.6983,
+      "step": 2482
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028706118259695516,
+      "loss": 1.6899,
+      "step": 2483
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002869764800148551,
+      "loss": 1.7006,
+      "step": 2484
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002868917581894937,
+      "loss": 1.7316,
+      "step": 2485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002868070171396154,
+      "loss": 1.6349,
+      "step": 2486
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028672225688396885,
+      "loss": 1.7366,
+      "step": 2487
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028663747744130704,
+      "loss": 1.7095,
+      "step": 2488
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000286552678830387,
+      "loss": 1.7306,
+      "step": 2489
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002864678610699703,
+      "loss": 1.7638,
+      "step": 2490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028638302417882254,
+      "loss": 1.6404,
+      "step": 2491
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028629816817571363,
+      "loss": 1.787,
+      "step": 2492
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028621329307941757,
+      "loss": 1.7448,
+      "step": 2493
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028612839890871283,
+      "loss": 1.6568,
+      "step": 2494
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000286043485682382,
+      "loss": 1.8065,
+      "step": 2495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002859585534192116,
+      "loss": 1.5996,
+      "step": 2496
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028587360213799286,
+      "loss": 1.8257,
+      "step": 2497
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028578863185752093,
+      "loss": 1.6561,
+      "step": 2498
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028570364259659513,
+      "loss": 1.753,
+      "step": 2499
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002856186343740191,
+      "loss": 1.723,
+      "step": 2500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028553360720860065,
+      "loss": 1.7068,
+      "step": 2501
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028544856111915173,
+      "loss": 1.5513,
+      "step": 2502
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028536349612448856,
+      "loss": 1.6565,
+      "step": 2503
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002852784122434314,
+      "loss": 1.8035,
+      "step": 2504
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002851933094948049,
+      "loss": 1.6123,
+      "step": 2505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002851081878974377,
+      "loss": 1.6604,
+      "step": 2506
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028502304747016254,
+      "loss": 1.6892,
+      "step": 2507
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002849378882318167,
+      "loss": 1.7229,
+      "step": 2508
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002848527102012413,
+      "loss": 1.7638,
+      "step": 2509
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028476751339728156,
+      "loss": 1.6329,
+      "step": 2510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028468229783878724,
+      "loss": 1.7142,
+      "step": 2511
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002845970635446118,
+      "loss": 1.7309,
+      "step": 2512
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028451181053361314,
+      "loss": 1.6896,
+      "step": 2513
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002844265388246531,
+      "loss": 1.5717,
+      "step": 2514
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002843412484365979,
+      "loss": 1.9137,
+      "step": 2515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002842559393883178,
+      "loss": 1.8714,
+      "step": 2516
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000284170611698687,
+      "loss": 1.8185,
+      "step": 2517
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002840852653865841,
+      "loss": 1.666,
+      "step": 2518
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002839999004708916,
+      "loss": 1.6868,
+      "step": 2519
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028391451697049616,
+      "loss": 1.8502,
+      "step": 2520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002838291149042887,
+      "loss": 1.786,
+      "step": 2521
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028374369429116415,
+      "loss": 1.7068,
+      "step": 2522
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002836582551500215,
+      "loss": 1.7699,
+      "step": 2523
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028357279749976393,
+      "loss": 1.7207,
+      "step": 2524
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002834873213592985,
+      "loss": 1.7193,
+      "step": 2525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002834018267475369,
+      "loss": 1.7483,
+      "step": 2526
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002833163136833941,
+      "loss": 1.6955,
+      "step": 2527
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028323078218578975,
+      "loss": 1.7485,
+      "step": 2528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028314523227364753,
+      "loss": 1.905,
+      "step": 2529
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028305966396589496,
+      "loss": 1.7684,
+      "step": 2530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028297407728146374,
+      "loss": 1.8091,
+      "step": 2531
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002828884722392897,
+      "loss": 1.6072,
+      "step": 2532
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028280284885831265,
+      "loss": 1.8229,
+      "step": 2533
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002827172071574765,
+      "loss": 1.7388,
+      "step": 2534
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028263154715572913,
+      "loss": 1.7707,
+      "step": 2535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002825458688720226,
+      "loss": 1.5601,
+      "step": 2536
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000282460172325313,
+      "loss": 1.69,
+      "step": 2537
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028237445753456026,
+      "loss": 1.7986,
+      "step": 2538
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028228872451872865,
+      "loss": 1.6734,
+      "step": 2539
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028220297329678616,
+      "loss": 1.7279,
+      "step": 2540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028211720388770514,
+      "loss": 1.5336,
+      "step": 2541
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028203141631046166,
+      "loss": 1.7906,
+      "step": 2542
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028194561058403605,
+      "loss": 1.6486,
+      "step": 2543
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002818597867274124,
+      "loss": 1.5579,
+      "step": 2544
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028177394475957914,
+      "loss": 1.4407,
+      "step": 2545
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002816880846995283,
+      "loss": 1.662,
+      "step": 2546
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028160220656625633,
+      "loss": 1.6238,
+      "step": 2547
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028151631037876343,
+      "loss": 1.7286,
+      "step": 2548
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028143039615605386,
+      "loss": 1.8448,
+      "step": 2549
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028134446391713577,
+      "loss": 1.8141,
+      "step": 2550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028125851368102154,
+      "loss": 1.5945,
+      "step": 2551
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028117254546672723,
+      "loss": 1.5204,
+      "step": 2552
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002810865592932731,
+      "loss": 1.6669,
+      "step": 2553
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028100055517968326,
+      "loss": 1.7075,
+      "step": 2554
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028091453314498595,
+      "loss": 1.4518,
+      "step": 2555
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002808284932082131,
+      "loss": 1.5898,
+      "step": 2556
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028074243538840095,
+      "loss": 1.8477,
+      "step": 2557
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002806563597045894,
+      "loss": 1.6021,
+      "step": 2558
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002805702661758224,
+      "loss": 1.6761,
+      "step": 2559
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002804841548211479,
+      "loss": 1.8218,
+      "step": 2560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002803980256596177,
+      "loss": 1.7572,
+      "step": 2561
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028031187871028774,
+      "loss": 1.7098,
+      "step": 2562
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028022571399221756,
+      "loss": 1.7678,
+      "step": 2563
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002801395315244709,
+      "loss": 1.7517,
+      "step": 2564
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028005333132611547,
+      "loss": 1.6863,
+      "step": 2565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027996711341622265,
+      "loss": 1.642,
+      "step": 2566
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002798808778138678,
+      "loss": 1.7347,
+      "step": 2567
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002797946245381304,
+      "loss": 1.6845,
+      "step": 2568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002797083536080937,
+      "loss": 1.6079,
+      "step": 2569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027962206504284474,
+      "loss": 1.5428,
+      "step": 2570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002795357588614747,
+      "loss": 1.6619,
+      "step": 2571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027944943508307853,
+      "loss": 1.668,
+      "step": 2572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027936309372675506,
+      "loss": 1.7797,
+      "step": 2573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027927673481160703,
+      "loss": 1.5836,
+      "step": 2574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027919035835674105,
+      "loss": 1.6621,
+      "step": 2575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027910396438126766,
+      "loss": 1.6441,
+      "step": 2576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027901755290430123,
+      "loss": 1.7637,
+      "step": 2577
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027893112394496,
+      "loss": 1.6007,
+      "step": 2578
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002788446775223661,
+      "loss": 1.8098,
+      "step": 2579
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002787582136556456,
+      "loss": 1.6559,
+      "step": 2580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002786717323639282,
+      "loss": 1.6926,
+      "step": 2581
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002785852336663477,
+      "loss": 1.6371,
+      "step": 2582
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002784987175820416,
+      "loss": 1.7901,
+      "step": 2583
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027841218413015136,
+      "loss": 1.823,
+      "step": 2584
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002783256333298222,
+      "loss": 1.7698,
+      "step": 2585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002782390652002032,
+      "loss": 1.9785,
+      "step": 2586
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027815247976044723,
+      "loss": 1.72,
+      "step": 2587
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002780658770297111,
+      "loss": 1.6537,
+      "step": 2588
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002779792570271554,
+      "loss": 1.8153,
+      "step": 2589
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027789261977194445,
+      "loss": 1.7977,
+      "step": 2590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002778059652832465,
+      "loss": 1.7976,
+      "step": 2591
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002777192935802335,
+      "loss": 1.7463,
+      "step": 2592
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027763260468208136,
+      "loss": 1.8497,
+      "step": 2593
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002775458986079697,
+      "loss": 1.6784,
+      "step": 2594
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002774591753770819,
+      "loss": 1.726,
+      "step": 2595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002773724350086054,
+      "loss": 1.7032,
+      "step": 2596
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027728567752173094,
+      "loss": 1.6552,
+      "step": 2597
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027719890293565353,
+      "loss": 1.7063,
+      "step": 2598
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027711211126957156,
+      "loss": 1.6855,
+      "step": 2599
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002770253025426876,
+      "loss": 1.776,
+      "step": 2600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027693847677420774,
+      "loss": 1.8495,
+      "step": 2601
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027685163398334184,
+      "loss": 1.9084,
+      "step": 2602
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027676477418930363,
+      "loss": 1.6917,
+      "step": 2603
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027667789741131054,
+      "loss": 1.6001,
+      "step": 2604
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027659100366858385,
+      "loss": 1.646,
+      "step": 2605
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027650409298034836,
+      "loss": 1.7231,
+      "step": 2606
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002764171653658329,
+      "loss": 1.7175,
+      "step": 2607
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027633022084426993,
+      "loss": 1.6527,
+      "step": 2608
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002762432594348955,
+      "loss": 1.721,
+      "step": 2609
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027615628115694967,
+      "loss": 1.6283,
+      "step": 2610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000276069286029676,
+      "loss": 1.6781,
+      "step": 2611
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027598227407232205,
+      "loss": 1.7965,
+      "step": 2612
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002758952453041387,
+      "loss": 1.6885,
+      "step": 2613
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027580819974438103,
+      "loss": 1.6641,
+      "step": 2614
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002757211374123074,
+      "loss": 1.7767,
+      "step": 2615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002756340583271801,
+      "loss": 1.5704,
+      "step": 2616
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002755469625082651,
+      "loss": 1.7948,
+      "step": 2617
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002754598499748321,
+      "loss": 1.7461,
+      "step": 2618
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027537272074615435,
+      "loss": 1.7024,
+      "step": 2619
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027528557484150914,
+      "loss": 1.6648,
+      "step": 2620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027519841228017695,
+      "loss": 1.8016,
+      "step": 2621
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002751112330814424,
+      "loss": 1.697,
+      "step": 2622
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002750240372645935,
+      "loss": 1.8619,
+      "step": 2623
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000274936824848922,
+      "loss": 1.742,
+      "step": 2624
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027484959585372354,
+      "loss": 1.7514,
+      "step": 2625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002747623502982971,
+      "loss": 1.7154,
+      "step": 2626
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027467508820194547,
+      "loss": 1.618,
+      "step": 2627
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002745878095839753,
+      "loss": 1.8276,
+      "step": 2628
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002745005144636965,
+      "loss": 1.8154,
+      "step": 2629
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002744132028604228,
+      "loss": 1.6208,
+      "step": 2630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027432587479347167,
+      "loss": 1.5041,
+      "step": 2631
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002742385302821642,
+      "loss": 1.6559,
+      "step": 2632
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002741511693458251,
+      "loss": 1.525,
+      "step": 2633
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002740637920037825,
+      "loss": 1.796,
+      "step": 2634
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002739763982753686,
+      "loss": 1.8255,
+      "step": 2635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002738889881799189,
+      "loss": 1.8178,
+      "step": 2636
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002738015617367726,
+      "loss": 1.6307,
+      "step": 2637
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002737141189652723,
+      "loss": 1.804,
+      "step": 2638
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002736266598847648,
+      "loss": 1.6878,
+      "step": 2639
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027353918451459983,
+      "loss": 1.736,
+      "step": 2640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027345169287413114,
+      "loss": 1.6643,
+      "step": 2641
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027336418498271595,
+      "loss": 1.74,
+      "step": 2642
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002732766608597151,
+      "loss": 1.8239,
+      "step": 2643
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002731891205244931,
+      "loss": 1.5428,
+      "step": 2644
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027310156399641786,
+      "loss": 1.713,
+      "step": 2645
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027301399129486104,
+      "loss": 1.6413,
+      "step": 2646
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002729264024391978,
+      "loss": 1.959,
+      "step": 2647
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027283879744880677,
+      "loss": 1.7505,
+      "step": 2648
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002727511763430704,
+      "loss": 1.6374,
+      "step": 2649
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027266353914137453,
+      "loss": 1.7819,
+      "step": 2650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027257588586310867,
+      "loss": 1.7186,
+      "step": 2651
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027248821652766574,
+      "loss": 1.7185,
+      "step": 2652
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027240053115444227,
+      "loss": 1.6255,
+      "step": 2653
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027231282976283844,
+      "loss": 1.7271,
+      "step": 2654
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027222511237225784,
+      "loss": 1.7704,
+      "step": 2655
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002721373790021077,
+      "loss": 1.685,
+      "step": 2656
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002720496296717987,
+      "loss": 1.7239,
+      "step": 2657
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027196186440074514,
+      "loss": 1.8237,
+      "step": 2658
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002718740832083647,
+      "loss": 1.8255,
+      "step": 2659
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002717862861140788,
+      "loss": 1.8527,
+      "step": 2660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002716984731373122,
+      "loss": 1.7609,
+      "step": 2661
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027161064429749327,
+      "loss": 1.5658,
+      "step": 2662
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002715227996140538,
+      "loss": 1.7491,
+      "step": 2663
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027143493910642915,
+      "loss": 1.6562,
+      "step": 2664
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002713470627940582,
+      "loss": 1.7336,
+      "step": 2665
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002712591706963833,
+      "loss": 1.7158,
+      "step": 2666
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002711712628328502,
+      "loss": 1.7406,
+      "step": 2667
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002710833392229084,
+      "loss": 1.6897,
+      "step": 2668
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027099539988601054,
+      "loss": 1.6325,
+      "step": 2669
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000270907444841613,
+      "loss": 1.7679,
+      "step": 2670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027081947410917547,
+      "loss": 1.7516,
+      "step": 2671
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002707314877081613,
+      "loss": 1.6961,
+      "step": 2672
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002706434856580371,
+      "loss": 1.747,
+      "step": 2673
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002705554679782731,
+      "loss": 1.497,
+      "step": 2674
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027046743468834287,
+      "loss": 1.8615,
+      "step": 2675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002703793858077235,
+      "loss": 1.8812,
+      "step": 2676
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027029132135589554,
+      "loss": 1.7588,
+      "step": 2677
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000270203241352343,
+      "loss": 1.7745,
+      "step": 2678
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002701151458165532,
+      "loss": 1.7685,
+      "step": 2679
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027002703476801697,
+      "loss": 1.7491,
+      "step": 2680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026993890822622866,
+      "loss": 1.5923,
+      "step": 2681
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000269850766210686,
+      "loss": 1.7924,
+      "step": 2682
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002697626087408901,
+      "loss": 1.6684,
+      "step": 2683
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026967443583634544,
+      "loss": 1.7404,
+      "step": 2684
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026958624751656014,
+      "loss": 1.6271,
+      "step": 2685
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002694980438010454,
+      "loss": 1.7385,
+      "step": 2686
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026940982470931616,
+      "loss": 1.7497,
+      "step": 2687
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026932159026089045,
+      "loss": 1.6816,
+      "step": 2688
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026923334047528996,
+      "loss": 1.7837,
+      "step": 2689
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026914507537203965,
+      "loss": 1.7908,
+      "step": 2690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002690567949706679,
+      "loss": 1.7243,
+      "step": 2691
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002689684992907064,
+      "loss": 1.9287,
+      "step": 2692
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026888018835169035,
+      "loss": 1.7443,
+      "step": 2693
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002687918621731582,
+      "loss": 1.8558,
+      "step": 2694
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002687035207746519,
+      "loss": 1.6454,
+      "step": 2695
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002686151641757166,
+      "loss": 1.8683,
+      "step": 2696
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002685267923959009,
+      "loss": 1.7637,
+      "step": 2697
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002684384054547568,
+      "loss": 1.8187,
+      "step": 2698
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002683500033718398,
+      "loss": 1.782,
+      "step": 2699
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026826158616670827,
+      "loss": 1.7728,
+      "step": 2700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026817315385892443,
+      "loss": 1.6827,
+      "step": 2701
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002680847064680536,
+      "loss": 1.6407,
+      "step": 2702
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026799624401366436,
+      "loss": 1.4508,
+      "step": 2703
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002679077665153289,
+      "loss": 1.8062,
+      "step": 2704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002678192739926225,
+      "loss": 1.6535,
+      "step": 2705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026773076646512383,
+      "loss": 1.8651,
+      "step": 2706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002676422439524149,
+      "loss": 1.798,
+      "step": 2707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026755370647408113,
+      "loss": 1.7616,
+      "step": 2708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026746515404971096,
+      "loss": 1.5987,
+      "step": 2709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002673765866988965,
+      "loss": 1.6596,
+      "step": 2710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026728800444123283,
+      "loss": 1.7157,
+      "step": 2711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002671994072963185,
+      "loss": 1.6058,
+      "step": 2712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002671107952837555,
+      "loss": 1.6734,
+      "step": 2713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002670221684231488,
+      "loss": 1.5894,
+      "step": 2714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026693352673410687,
+      "loss": 1.696,
+      "step": 2715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002668448702362413,
+      "loss": 1.7432,
+      "step": 2716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026675619894916725,
+      "loss": 1.7311,
+      "step": 2717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026666751289250267,
+      "loss": 1.7749,
+      "step": 2718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026657881208586937,
+      "loss": 1.7033,
+      "step": 2719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026649009654889174,
+      "loss": 1.8257,
+      "step": 2720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002664013663011982,
+      "loss": 1.7465,
+      "step": 2721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002663126213624198,
+      "loss": 1.7345,
+      "step": 2722
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026622386175219104,
+      "loss": 1.7687,
+      "step": 2723
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026613508749014986,
+      "loss": 1.7618,
+      "step": 2724
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002660462985959371,
+      "loss": 1.5777,
+      "step": 2725
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026595749508919717,
+      "loss": 1.7501,
+      "step": 2726
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026586867698957735,
+      "loss": 1.7928,
+      "step": 2727
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026577984431672857,
+      "loss": 1.7208,
+      "step": 2728
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026569099709030465,
+      "loss": 1.6537,
+      "step": 2729
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026560213532996277,
+      "loss": 1.6549,
+      "step": 2730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002655132590553633,
+      "loss": 1.6018,
+      "step": 2731
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026542436828616984,
+      "loss": 1.7438,
+      "step": 2732
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002653354630420492,
+      "loss": 1.5931,
+      "step": 2733
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026524654334267127,
+      "loss": 1.6117,
+      "step": 2734
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002651576092077094,
+      "loss": 1.6527,
+      "step": 2735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026506866065683987,
+      "loss": 1.6057,
+      "step": 2736
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002649796977097422,
+      "loss": 1.8186,
+      "step": 2737
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026489072038609934,
+      "loss": 1.5935,
+      "step": 2738
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026480172870559717,
+      "loss": 1.6368,
+      "step": 2739
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026471272268792465,
+      "loss": 1.6603,
+      "step": 2740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026462370235277416,
+      "loss": 1.7517,
+      "step": 2741
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002645346677198412,
+      "loss": 1.6315,
+      "step": 2742
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002644456188088245,
+      "loss": 1.6733,
+      "step": 2743
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002643565556394255,
+      "loss": 1.7682,
+      "step": 2744
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026426747823134943,
+      "loss": 1.6624,
+      "step": 2745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026417838660430435,
+      "loss": 1.7533,
+      "step": 2746
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002640892807780014,
+      "loss": 1.7082,
+      "step": 2747
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026400016077215494,
+      "loss": 1.7352,
+      "step": 2748
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026391102660648256,
+      "loss": 1.7287,
+      "step": 2749
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026382187830070486,
+      "loss": 1.6574,
+      "step": 2750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026373271587454554,
+      "loss": 1.6886,
+      "step": 2751
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026364353934773166,
+      "loss": 1.7252,
+      "step": 2752
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026355434873999315,
+      "loss": 1.774,
+      "step": 2753
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002634651440710631,
+      "loss": 1.761,
+      "step": 2754
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002633759253606777,
+      "loss": 1.7877,
+      "step": 2755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002632866926285765,
+      "loss": 1.8776,
+      "step": 2756
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026319744589450183,
+      "loss": 1.7353,
+      "step": 2757
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002631081851781992,
+      "loss": 1.7419,
+      "step": 2758
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026301891049941724,
+      "loss": 1.8823,
+      "step": 2759
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026292962187790775,
+      "loss": 1.8333,
+      "step": 2760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002628403193334256,
+      "loss": 1.8152,
+      "step": 2761
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002627510028857286,
+      "loss": 1.8574,
+      "step": 2762
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002626616725545777,
+      "loss": 1.6431,
+      "step": 2763
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026257232835973695,
+      "loss": 1.5844,
+      "step": 2764
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002624829703209735,
+      "loss": 1.9168,
+      "step": 2765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002623935984580575,
+      "loss": 1.7528,
+      "step": 2766
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026230421279076216,
+      "loss": 1.6731,
+      "step": 2767
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026221481333886385,
+      "loss": 1.6648,
+      "step": 2768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026212540012214184,
+      "loss": 1.6574,
+      "step": 2769
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002620359731603785,
+      "loss": 1.9155,
+      "step": 2770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026194653247335934,
+      "loss": 1.678,
+      "step": 2771
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002618570780808727,
+      "loss": 1.8228,
+      "step": 2772
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026176761000271017,
+      "loss": 1.7874,
+      "step": 2773
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026167812825866616,
+      "loss": 1.6527,
+      "step": 2774
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002615886328685384,
+      "loss": 1.8862,
+      "step": 2775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002614991238521272,
+      "loss": 1.8958,
+      "step": 2776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026140960122923625,
+      "loss": 1.7283,
+      "step": 2777
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026132006501967226,
+      "loss": 1.9329,
+      "step": 2778
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002612305152432447,
+      "loss": 1.4919,
+      "step": 2779
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002611409519197662,
+      "loss": 1.6131,
+      "step": 2780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002610513750690523,
+      "loss": 1.6895,
+      "step": 2781
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002609617847109216,
+      "loss": 1.7031,
+      "step": 2782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002608721808651957,
+      "loss": 1.5634,
+      "step": 2783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002607825635516991,
+      "loss": 1.8007,
+      "step": 2784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002606929327902595,
+      "loss": 1.8172,
+      "step": 2785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002606032886007073,
+      "loss": 1.6238,
+      "step": 2786
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002605136310028759,
+      "loss": 1.7701,
+      "step": 2787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026042396001660183,
+      "loss": 1.7178,
+      "step": 2788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002603342756617246,
+      "loss": 1.7554,
+      "step": 2789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026024457795808636,
+      "loss": 1.6725,
+      "step": 2790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002601548669255326,
+      "loss": 1.7685,
+      "step": 2791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002600651425839116,
+      "loss": 1.6858,
+      "step": 2792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002599754049530745,
+      "loss": 1.7592,
+      "step": 2793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002598856540528755,
+      "loss": 1.7351,
+      "step": 2794
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002597958899031717,
+      "loss": 1.6538,
+      "step": 2795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00025970611252382306,
+      "loss": 1.8806,
+      "step": 2796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00025961632193469256,
+      "loss": 1.7457,
+      "step": 2797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00025952651815564604,
+      "loss": 1.6715,
+      "step": 2798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00025943670120655245,
+      "loss": 1.8902,
+      "step": 2799
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00025934687110728335,
+      "loss": 1.8677,
+      "step": 2800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002592570278777133,
+      "loss": 1.7187,
+      "step": 2801
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00025916717153772,
+      "loss": 1.8998,
+      "step": 2802
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00025907730210718375,
+      "loss": 1.7381,
+      "step": 2803
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00025898741960598794,
+      "loss": 1.5691,
+      "step": 2804
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00025889752405401867,
+      "loss": 1.7073,
+      "step": 2805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00025880761547116514,
+      "loss": 1.7523,
+      "step": 2806
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002587176938773193,
+      "loss": 1.65,
+      "step": 2807
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002586277592923759,
+      "loss": 1.767,
+      "step": 2808
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002585378117362329,
+      "loss": 1.762,
+      "step": 2809
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00025844785122879077,
+      "loss": 1.8756,
+      "step": 2810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00025835787778995294,
+      "loss": 1.965,
+      "step": 2811
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00025826789143962574,
+      "loss": 1.5463,
+      "step": 2812
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00025817789219771843,
+      "loss": 1.7306,
+      "step": 2813
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025808788008414304,
+      "loss": 1.5936,
+      "step": 2814
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002579978551188144,
+      "loss": 1.7248,
+      "step": 2815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002579078173216503,
+      "loss": 1.6603,
+      "step": 2816
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025781776671257123,
+      "loss": 1.7554,
+      "step": 2817
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025772770331150064,
+      "loss": 1.5599,
+      "step": 2818
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002576376271383648,
+      "loss": 1.6374,
+      "step": 2819
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025754753821309275,
+      "loss": 1.6881,
+      "step": 2820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002574574365556163,
+      "loss": 1.5793,
+      "step": 2821
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025736732218587015,
+      "loss": 1.6722,
+      "step": 2822
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002572771951237919,
+      "loss": 1.5405,
+      "step": 2823
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025718705538932185,
+      "loss": 1.6273,
+      "step": 2824
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025709690300240306,
+      "loss": 1.6482,
+      "step": 2825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002570067379829815,
+      "loss": 1.7321,
+      "step": 2826
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002569165603510059,
+      "loss": 1.5337,
+      "step": 2827
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002568263701264277,
+      "loss": 1.6343,
+      "step": 2828
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025673616732920114,
+      "loss": 1.5772,
+      "step": 2829
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002566459519792835,
+      "loss": 1.7471,
+      "step": 2830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002565557240966345,
+      "loss": 1.8353,
+      "step": 2831
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002564654837012167,
+      "loss": 1.7519,
+      "step": 2832
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025637523081299565,
+      "loss": 1.7236,
+      "step": 2833
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002562849654519394,
+      "loss": 1.6618,
+      "step": 2834
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002561946876380189,
+      "loss": 1.8958,
+      "step": 2835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002561043973912078,
+      "loss": 1.7884,
+      "step": 2836
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002560140947314824,
+      "loss": 1.7489,
+      "step": 2837
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025592377967882207,
+      "loss": 1.7399,
+      "step": 2838
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025583345225320875,
+      "loss": 1.7451,
+      "step": 2839
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025574311247462684,
+      "loss": 1.811,
+      "step": 2840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025565276036306395,
+      "loss": 1.6709,
+      "step": 2841
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002555623959385101,
+      "loss": 1.7989,
+      "step": 2842
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025547201922095813,
+      "loss": 1.8744,
+      "step": 2843
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002553816302304035,
+      "loss": 1.6179,
+      "step": 2844
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025529122898684467,
+      "loss": 1.5786,
+      "step": 2845
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002552008155102825,
+      "loss": 1.638,
+      "step": 2846
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002551103898207206,
+      "loss": 1.6542,
+      "step": 2847
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025501995193816555,
+      "loss": 1.4549,
+      "step": 2848
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025492950188262626,
+      "loss": 1.6494,
+      "step": 2849
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025483903967411466,
+      "loss": 1.7239,
+      "step": 2850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025474856533264507,
+      "loss": 1.8119,
+      "step": 2851
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002546580788782348,
+      "loss": 1.8313,
+      "step": 2852
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002545675803309035,
+      "loss": 1.6872,
+      "step": 2853
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002544770697106738,
+      "loss": 1.7396,
+      "step": 2854
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002543865470375708,
+      "loss": 1.7131,
+      "step": 2855
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002542960123316224,
+      "loss": 1.776,
+      "step": 2856
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025420546561285916,
+      "loss": 1.6363,
+      "step": 2857
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002541149069013142,
+      "loss": 1.5038,
+      "step": 2858
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002540243362170233,
+      "loss": 1.7731,
+      "step": 2859
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025393375358002496,
+      "loss": 1.6362,
+      "step": 2860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025384315901036023,
+      "loss": 1.7378,
+      "step": 2861
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025375255252807293,
+      "loss": 1.6924,
+      "step": 2862
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025366193415320945,
+      "loss": 1.8317,
+      "step": 2863
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002535713039058188,
+      "loss": 1.6965,
+      "step": 2864
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002534806618059526,
+      "loss": 1.6947,
+      "step": 2865
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025339000787366514,
+      "loss": 1.8152,
+      "step": 2866
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025329934212901334,
+      "loss": 1.6665,
+      "step": 2867
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025320866459205656,
+      "loss": 1.6655,
+      "step": 2868
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002531179752828571,
+      "loss": 1.7186,
+      "step": 2869
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002530272742214796,
+      "loss": 1.5651,
+      "step": 2870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002529365614279912,
+      "loss": 1.7167,
+      "step": 2871
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000252845836922462,
+      "loss": 1.7846,
+      "step": 2872
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002527551007249646,
+      "loss": 1.7632,
+      "step": 2873
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002526643528555739,
+      "loss": 1.8511,
+      "step": 2874
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002525735933343676,
+      "loss": 1.6872,
+      "step": 2875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002524828221814259,
+      "loss": 1.7025,
+      "step": 2876
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002523920394168317,
+      "loss": 1.7357,
+      "step": 2877
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002523012450606704,
+      "loss": 1.7015,
+      "step": 2878
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025221043913302985,
+      "loss": 1.7853,
+      "step": 2879
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025211962165400077,
+      "loss": 1.7233,
+      "step": 2880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025202879264367607,
+      "loss": 1.7998,
+      "step": 2881
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002519379521221514,
+      "loss": 1.8673,
+      "step": 2882
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002518471001095248,
+      "loss": 1.5587,
+      "step": 2883
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025175623662589727,
+      "loss": 1.7911,
+      "step": 2884
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002516653616913718,
+      "loss": 1.513,
+      "step": 2885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025157447532605425,
+      "loss": 1.523,
+      "step": 2886
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000251483577550053,
+      "loss": 1.75,
+      "step": 2887
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025139266838347885,
+      "loss": 1.6956,
+      "step": 2888
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002513017478464452,
+      "loss": 1.7661,
+      "step": 2889
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002512108159590677,
+      "loss": 1.6327,
+      "step": 2890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025111987274146504,
+      "loss": 1.6769,
+      "step": 2891
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025102891821375794,
+      "loss": 1.7276,
+      "step": 2892
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025093795239606974,
+      "loss": 1.733,
+      "step": 2893
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025084697530852637,
+      "loss": 1.7649,
+      "step": 2894
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025075598697125637,
+      "loss": 1.6731,
+      "step": 2895
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025066498740439045,
+      "loss": 1.6947,
+      "step": 2896
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000250573976628062,
+      "loss": 1.6821,
+      "step": 2897
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002504829546624068,
+      "loss": 1.7986,
+      "step": 2898
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002503919215275633,
+      "loss": 1.6122,
+      "step": 2899
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025030087724367214,
+      "loss": 1.6999,
+      "step": 2900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002502098218308766,
+      "loss": 1.5964,
+      "step": 2901
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025011875530932254,
+      "loss": 1.5642,
+      "step": 2902
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025002767769915796,
+      "loss": 1.9363,
+      "step": 2903
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024993658902053355,
+      "loss": 1.68,
+      "step": 2904
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002498454892936024,
+      "loss": 1.6517,
+      "step": 2905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002497543785385199,
+      "loss": 1.7131,
+      "step": 2906
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002496632567754441,
+      "loss": 1.7478,
+      "step": 2907
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002495721240245355,
+      "loss": 1.7489,
+      "step": 2908
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024948098030595675,
+      "loss": 1.7124,
+      "step": 2909
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024938982563987324,
+      "loss": 1.5487,
+      "step": 2910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024929866004645246,
+      "loss": 1.8324,
+      "step": 2911
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024920748354586463,
+      "loss": 1.6337,
+      "step": 2912
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002491162961582823,
+      "loss": 1.8067,
+      "step": 2913
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002490250979038803,
+      "loss": 1.6753,
+      "step": 2914
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024893388880283586,
+      "loss": 1.6272,
+      "step": 2915
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024884266887532886,
+      "loss": 1.716,
+      "step": 2916
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024875143814154135,
+      "loss": 1.733,
+      "step": 2917
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024866019662165775,
+      "loss": 1.6684,
+      "step": 2918
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002485689443358651,
+      "loss": 1.6329,
+      "step": 2919
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024847768130435255,
+      "loss": 1.6925,
+      "step": 2920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024838640754731183,
+      "loss": 1.7377,
+      "step": 2921
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002482951230849369,
+      "loss": 1.5722,
+      "step": 2922
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024820382793742403,
+      "loss": 1.6649,
+      "step": 2923
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024811252212497226,
+      "loss": 1.5922,
+      "step": 2924
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024802120566778246,
+      "loss": 1.7014,
+      "step": 2925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002479298785860582,
+      "loss": 1.7026,
+      "step": 2926
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024783854090000535,
+      "loss": 1.8106,
+      "step": 2927
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024774719262983194,
+      "loss": 1.6704,
+      "step": 2928
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024765583379574854,
+      "loss": 1.7137,
+      "step": 2929
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024756446441796803,
+      "loss": 1.6696,
+      "step": 2930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024747308451670554,
+      "loss": 1.8162,
+      "step": 2931
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002473816941121785,
+      "loss": 1.7252,
+      "step": 2932
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002472902932246069,
+      "loss": 1.6365,
+      "step": 2933
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024719888187421283,
+      "loss": 1.6148,
+      "step": 2934
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002471074600812207,
+      "loss": 1.7945,
+      "step": 2935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002470160278658573,
+      "loss": 1.795,
+      "step": 2936
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024692458524835177,
+      "loss": 1.8626,
+      "step": 2937
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002468331322489354,
+      "loss": 1.8273,
+      "step": 2938
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024674166888784184,
+      "loss": 1.6133,
+      "step": 2939
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002466501951853072,
+      "loss": 1.8218,
+      "step": 2940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002465587111615697,
+      "loss": 1.7346,
+      "step": 2941
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002464672168368698,
+      "loss": 1.7414,
+      "step": 2942
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002463757122314503,
+      "loss": 1.8135,
+      "step": 2943
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002462841973655565,
+      "loss": 1.7967,
+      "step": 2944
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002461926722594355,
+      "loss": 1.7644,
+      "step": 2945
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002461011369333371,
+      "loss": 1.7167,
+      "step": 2946
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002460095914075131,
+      "loss": 1.6717,
+      "step": 2947
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024591803570221774,
+      "loss": 1.8146,
+      "step": 2948
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002458264698377074,
+      "loss": 1.6783,
+      "step": 2949
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024573489383424057,
+      "loss": 1.5078,
+      "step": 2950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002456433077120784,
+      "loss": 1.5526,
+      "step": 2951
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002455517114914839,
+      "loss": 1.7676,
+      "step": 2952
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024546010519272234,
+      "loss": 1.7637,
+      "step": 2953
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024536848883606135,
+      "loss": 1.8066,
+      "step": 2954
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002452768624417709,
+      "loss": 1.7179,
+      "step": 2955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002451852260301229,
+      "loss": 1.7265,
+      "step": 2956
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024509357962139157,
+      "loss": 1.6093,
+      "step": 2957
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002450019232358535,
+      "loss": 1.6126,
+      "step": 2958
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024491025689378726,
+      "loss": 1.6458,
+      "step": 2959
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024481858061547383,
+      "loss": 1.8268,
+      "step": 2960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002447268944211962,
+      "loss": 1.5452,
+      "step": 2961
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024463519833123966,
+      "loss": 1.8225,
+      "step": 2962
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024454349236589166,
+      "loss": 1.604,
+      "step": 2963
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024445177654544185,
+      "loss": 1.6602,
+      "step": 2964
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024436005089018215,
+      "loss": 1.6731,
+      "step": 2965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002442683154204064,
+      "loss": 1.7382,
+      "step": 2966
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024417657015641083,
+      "loss": 1.6303,
+      "step": 2967
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024408481511849388,
+      "loss": 1.6865,
+      "step": 2968
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024399305032695595,
+      "loss": 1.7052,
+      "step": 2969
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024390127580209973,
+      "loss": 1.5688,
+      "step": 2970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024380949156422994,
+      "loss": 1.8144,
+      "step": 2971
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024371769763365366,
+      "loss": 1.7187,
+      "step": 2972
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024362589403068003,
+      "loss": 1.7669,
+      "step": 2973
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024353408077562016,
+      "loss": 1.6387,
+      "step": 2974
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024344225788878752,
+      "loss": 1.5679,
+      "step": 2975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002433504253904976,
+      "loss": 1.4543,
+      "step": 2976
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024325858330106806,
+      "loss": 1.7237,
+      "step": 2977
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002431667316408186,
+      "loss": 1.6826,
+      "step": 2978
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024307487043007114,
+      "loss": 1.7609,
+      "step": 2979
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024298299968914972,
+      "loss": 1.7442,
+      "step": 2980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002428911194383804,
+      "loss": 1.867,
+      "step": 2981
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002427992296980913,
+      "loss": 2.0102,
+      "step": 2982
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002427073304886129,
+      "loss": 1.9693,
+      "step": 2983
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002426154218302775,
+      "loss": 1.6912,
+      "step": 2984
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024252350374341953,
+      "loss": 1.6105,
+      "step": 2985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024243157624837566,
+      "loss": 1.652,
+      "step": 2986
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024233963936548452,
+      "loss": 1.7776,
+      "step": 2987
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024233963936548452,
+      "loss": 1.8523,
+      "step": 2988
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024224769311508682,
+      "loss": 1.4914,
+      "step": 2989
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024215573751752535,
+      "loss": 1.6745,
+      "step": 2990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024206377259314513,
+      "loss": 1.5334,
+      "step": 2991
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024197179836229295,
+      "loss": 1.6116,
+      "step": 2992
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024187981484531784,
+      "loss": 1.7426,
+      "step": 2993
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002417878220625708,
+      "loss": 1.7405,
+      "step": 2994
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024169582003440504,
+      "loss": 1.8009,
+      "step": 2995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002416038087811757,
+      "loss": 1.5976,
+      "step": 2996
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024151178832323983,
+      "loss": 1.7503,
+      "step": 2997
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024141975868095678,
+      "loss": 1.6757,
+      "step": 2998
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024132771987468786,
+      "loss": 1.7599,
+      "step": 2999
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002412356719247962,
+      "loss": 1.8226,
+      "step": 3000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024114361485164722,
+      "loss": 1.6389,
+      "step": 3001
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002410515486756082,
+      "loss": 1.6279,
+      "step": 3002
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024095947341704844,
+      "loss": 1.6954,
+      "step": 3003
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024086738909633932,
+      "loss": 1.7028,
+      "step": 3004
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024077529573385427,
+      "loss": 1.7459,
+      "step": 3005
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002406831933499686,
+      "loss": 1.7873,
+      "step": 3006
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002405910819650596,
+      "loss": 1.7276,
+      "step": 3007
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024049896159950666,
+      "loss": 1.7481,
+      "step": 3008
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024040683227369116,
+      "loss": 1.7827,
+      "step": 3009
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024031469400799632,
+      "loss": 1.6489,
+      "step": 3010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024022254682280742,
+      "loss": 1.7176,
+      "step": 3011
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024013039073851182,
+      "loss": 1.71,
+      "step": 3012
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024003822577549873,
+      "loss": 1.7907,
+      "step": 3013
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002399460519541593,
+      "loss": 1.7828,
+      "step": 3014
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002398538692948866,
+      "loss": 1.4706,
+      "step": 3015
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000239761677818076,
+      "loss": 1.8753,
+      "step": 3016
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002396694775441244,
+      "loss": 1.7808,
+      "step": 3017
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002395772684934307,
+      "loss": 1.5231,
+      "step": 3018
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023948505068639605,
+      "loss": 1.7732,
+      "step": 3019
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023939282414342327,
+      "loss": 1.7739,
+      "step": 3020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023930058888491714,
+      "loss": 1.718,
+      "step": 3021
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023920834493128439,
+      "loss": 1.7884,
+      "step": 3022
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023911609230293376,
+      "loss": 1.762,
+      "step": 3023
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023902383102027591,
+      "loss": 1.7647,
+      "step": 3024
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023893156110372314,
+      "loss": 1.6898,
+      "step": 3025
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023883928257369008,
+      "loss": 1.788,
+      "step": 3026
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023874699545059288,
+      "loss": 1.7465,
+      "step": 3027
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002386546997548499,
+      "loss": 1.6468,
+      "step": 3028
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023856239550688112,
+      "loss": 1.8166,
+      "step": 3029
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023847008272710868,
+      "loss": 1.6667,
+      "step": 3030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023837776143595645,
+      "loss": 1.5991,
+      "step": 3031
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023828543165385017,
+      "loss": 1.6904,
+      "step": 3032
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023819309340121755,
+      "loss": 1.5948,
+      "step": 3033
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002381007466984881,
+      "loss": 1.7908,
+      "step": 3034
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002380083915660932,
+      "loss": 1.5006,
+      "step": 3035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023791602802446615,
+      "loss": 1.7187,
+      "step": 3036
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023782365609404212,
+      "loss": 1.71,
+      "step": 3037
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023773127579525803,
+      "loss": 1.6877,
+      "step": 3038
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023763888714855273,
+      "loss": 1.791,
+      "step": 3039
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023754649017436696,
+      "loss": 1.6768,
+      "step": 3040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002374540848931432,
+      "loss": 1.6592,
+      "step": 3041
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023736167132532585,
+      "loss": 1.7232,
+      "step": 3042
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000237269249491361,
+      "loss": 1.7398,
+      "step": 3043
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023717681941169683,
+      "loss": 1.8749,
+      "step": 3044
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023708438110678316,
+      "loss": 1.6215,
+      "step": 3045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023699193459707156,
+      "loss": 1.7992,
+      "step": 3046
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023689947990301562,
+      "loss": 1.5702,
+      "step": 3047
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002368070170450706,
+      "loss": 1.7285,
+      "step": 3048
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023671454604369362,
+      "loss": 1.8529,
+      "step": 3049
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023662206691934352,
+      "loss": 1.5579,
+      "step": 3050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023652957969248108,
+      "loss": 1.6565,
+      "step": 3051
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023643708438356878,
+      "loss": 1.7053,
+      "step": 3052
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023634458101307092,
+      "loss": 1.5798,
+      "step": 3053
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002362520696014535,
+      "loss": 1.7147,
+      "step": 3054
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023615955016918446,
+      "loss": 1.7382,
+      "step": 3055
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023606702273673338,
+      "loss": 1.5938,
+      "step": 3056
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023597448732457158,
+      "loss": 1.8757,
+      "step": 3057
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023588194395317231,
+      "loss": 1.6802,
+      "step": 3058
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002357893926430105,
+      "loss": 1.8123,
+      "step": 3059
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023569683341456273,
+      "loss": 1.7688,
+      "step": 3060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023560426628830753,
+      "loss": 1.5175,
+      "step": 3061
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023551169128472495,
+      "loss": 1.6444,
+      "step": 3062
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023541910842429705,
+      "loss": 1.6394,
+      "step": 3063
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023532651772750743,
+      "loss": 1.6101,
+      "step": 3064
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002352339192148414,
+      "loss": 1.6928,
+      "step": 3065
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002351413129067862,
+      "loss": 1.744,
+      "step": 3066
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023504869882383055,
+      "loss": 1.5983,
+      "step": 3067
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023495607698646506,
+      "loss": 1.6656,
+      "step": 3068
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002348634474151821,
+      "loss": 1.7096,
+      "step": 3069
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023477081013047557,
+      "loss": 1.6411,
+      "step": 3070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023467816515284118,
+      "loss": 1.622,
+      "step": 3071
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023458551250277632,
+      "loss": 1.6889,
+      "step": 3072
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023449285220078007,
+      "loss": 1.7553,
+      "step": 3073
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023440018426735325,
+      "loss": 1.7259,
+      "step": 3074
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002343075087229983,
+      "loss": 1.612,
+      "step": 3075
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023421482558821946,
+      "loss": 1.5533,
+      "step": 3076
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023412213488352247,
+      "loss": 1.7806,
+      "step": 3077
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023402943662941493,
+      "loss": 1.6191,
+      "step": 3078
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023393673084640598,
+      "loss": 1.7007,
+      "step": 3079
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023384401755500647,
+      "loss": 1.8288,
+      "step": 3080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023375129677572894,
+      "loss": 1.754,
+      "step": 3081
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023365856852908747,
+      "loss": 1.8811,
+      "step": 3082
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023356583283559804,
+      "loss": 1.7198,
+      "step": 3083
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023347308971577796,
+      "loss": 1.8077,
+      "step": 3084
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023338033919014647,
+      "loss": 1.7482,
+      "step": 3085
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002332875812792242,
+      "loss": 1.5969,
+      "step": 3086
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023319481600353367,
+      "loss": 1.6117,
+      "step": 3087
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023310204338359883,
+      "loss": 1.7621,
+      "step": 3088
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023300926343994518,
+      "loss": 1.6776,
+      "step": 3089
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002329164761931002,
+      "loss": 1.5931,
+      "step": 3090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023282368166359265,
+      "loss": 1.6523,
+      "step": 3091
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023273087987195302,
+      "loss": 1.7935,
+      "step": 3092
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023263807083871344,
+      "loss": 1.7283,
+      "step": 3093
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023254525458440753,
+      "loss": 1.5239,
+      "step": 3094
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023245243112957073,
+      "loss": 1.6912,
+      "step": 3095
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023235960049473975,
+      "loss": 1.61,
+      "step": 3096
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023226676270045316,
+      "loss": 1.9018,
+      "step": 3097
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023217391776725102,
+      "loss": 1.5722,
+      "step": 3098
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023208106571567493,
+      "loss": 1.5686,
+      "step": 3099
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023198820656626813,
+      "loss": 1.6858,
+      "step": 3100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023189534033957537,
+      "loss": 1.7694,
+      "step": 3101
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023180246705614308,
+      "loss": 1.7325,
+      "step": 3102
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023170958673651904,
+      "loss": 1.6622,
+      "step": 3103
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023161669940125287,
+      "loss": 1.775,
+      "step": 3104
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023152380507089547,
+      "loss": 1.7465,
+      "step": 3105
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023143090376599946,
+      "loss": 1.6253,
+      "step": 3106
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023133799550711888,
+      "loss": 1.7951,
+      "step": 3107
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023124508031480945,
+      "loss": 1.7785,
+      "step": 3108
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023115215820962838,
+      "loss": 1.6981,
+      "step": 3109
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002310592292121343,
+      "loss": 1.6708,
+      "step": 3110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002309662933428875,
+      "loss": 1.6724,
+      "step": 3111
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023087335062244968,
+      "loss": 1.5062,
+      "step": 3112
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023078040107138418,
+      "loss": 1.6364,
+      "step": 3113
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023068744471025574,
+      "loss": 1.7695,
+      "step": 3114
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023059448155963064,
+      "loss": 1.5952,
+      "step": 3115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023050151164007675,
+      "loss": 1.8686,
+      "step": 3116
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023040853497216328,
+      "loss": 1.5452,
+      "step": 3117
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023031555157646105,
+      "loss": 1.6174,
+      "step": 3118
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023022256147354236,
+      "loss": 1.5463,
+      "step": 3119
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023012956468398087,
+      "loss": 1.719,
+      "step": 3120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023003656122835186,
+      "loss": 1.6278,
+      "step": 3121
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022994355112723204,
+      "loss": 1.7229,
+      "step": 3122
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022985053440119967,
+      "loss": 1.7978,
+      "step": 3123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002297575110708343,
+      "loss": 1.6204,
+      "step": 3124
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022966448115671706,
+      "loss": 1.7025,
+      "step": 3125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022957144467943054,
+      "loss": 1.7653,
+      "step": 3126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022947840165955875,
+      "loss": 1.6865,
+      "step": 3127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002293853521176871,
+      "loss": 1.728,
+      "step": 3128
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022929229607440257,
+      "loss": 1.484,
+      "step": 3129
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022919923355029344,
+      "loss": 1.8072,
+      "step": 3130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022910616456594954,
+      "loss": 1.6094,
+      "step": 3131
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000229013089141962,
+      "loss": 1.6731,
+      "step": 3132
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002289200072989236,
+      "loss": 1.5124,
+      "step": 3133
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022882691905742823,
+      "loss": 1.707,
+      "step": 3134
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002287338244380715,
+      "loss": 1.8078,
+      "step": 3135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022864072346145018,
+      "loss": 1.6673,
+      "step": 3136
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002285476161481626,
+      "loss": 1.6621,
+      "step": 3137
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022845450251880847,
+      "loss": 1.6811,
+      "step": 3138
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002283613825939888,
+      "loss": 1.4991,
+      "step": 3139
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022826825639430615,
+      "loss": 1.9231,
+      "step": 3140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022817512394036435,
+      "loss": 1.7073,
+      "step": 3141
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022808198525276868,
+      "loss": 1.6747,
+      "step": 3142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002279888403521257,
+      "loss": 1.7034,
+      "step": 3143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002278956892590435,
+      "loss": 1.7265,
+      "step": 3144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022780253199413143,
+      "loss": 1.5085,
+      "step": 3145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022770936857800016,
+      "loss": 1.756,
+      "step": 3146
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002276161990312619,
+      "loss": 1.7181,
+      "step": 3147
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022752302337452997,
+      "loss": 1.7245,
+      "step": 3148
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022742984162841935,
+      "loss": 1.6319,
+      "step": 3149
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022733665381354605,
+      "loss": 1.8226,
+      "step": 3150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022724345995052766,
+      "loss": 1.6824,
+      "step": 3151
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022715026005998298,
+      "loss": 1.834,
+      "step": 3152
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022705705416253214,
+      "loss": 1.6936,
+      "step": 3153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002269638422787967,
+      "loss": 1.7356,
+      "step": 3154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022687062442939952,
+      "loss": 1.7292,
+      "step": 3155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022677740063496463,
+      "loss": 1.7855,
+      "step": 3156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022668417091611755,
+      "loss": 1.7303,
+      "step": 3157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022659093529348514,
+      "loss": 1.6391,
+      "step": 3158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022649769378769535,
+      "loss": 1.821,
+      "step": 3159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002264044464193775,
+      "loss": 1.5838,
+      "step": 3160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022631119320916244,
+      "loss": 1.7204,
+      "step": 3161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022621793417768203,
+      "loss": 1.6204,
+      "step": 3162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022612466934556955,
+      "loss": 1.7564,
+      "step": 3163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022603139873345954,
+      "loss": 1.8335,
+      "step": 3164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002259381223619878,
+      "loss": 1.7046,
+      "step": 3165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022584484025179157,
+      "loss": 1.8118,
+      "step": 3166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002257515524235089,
+      "loss": 1.7618,
+      "step": 3167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002256582588977797,
+      "loss": 1.7354,
+      "step": 3168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022556495969524473,
+      "loss": 1.6922,
+      "step": 3169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022547165483654616,
+      "loss": 1.5742,
+      "step": 3170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022537834434232737,
+      "loss": 1.6958,
+      "step": 3171
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000225285028233233,
+      "loss": 1.7964,
+      "step": 3172
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022519170652990895,
+      "loss": 1.6459,
+      "step": 3173
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022509837925300237,
+      "loss": 1.7237,
+      "step": 3174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022500504642316158,
+      "loss": 1.5525,
+      "step": 3175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022491170806103616,
+      "loss": 1.8376,
+      "step": 3176
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022481836418727694,
+      "loss": 1.6625,
+      "step": 3177
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022472501482253586,
+      "loss": 1.5604,
+      "step": 3178
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022463165998746622,
+      "loss": 1.7171,
+      "step": 3179
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022453829970272256,
+      "loss": 1.7824,
+      "step": 3180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022444493398896044,
+      "loss": 1.6846,
+      "step": 3181
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002243515628668367,
+      "loss": 1.829,
+      "step": 3182
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022425818635700946,
+      "loss": 1.768,
+      "step": 3183
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022416480448013797,
+      "loss": 1.726,
+      "step": 3184
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022407141725688253,
+      "loss": 1.8591,
+      "step": 3185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022397802470790494,
+      "loss": 1.5829,
+      "step": 3186
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002238846268538679,
+      "loss": 1.6777,
+      "step": 3187
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002237912237154354,
+      "loss": 1.4076,
+      "step": 3188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002236978153132726,
+      "loss": 1.5806,
+      "step": 3189
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022360440166804572,
+      "loss": 1.7147,
+      "step": 3190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002235109828004223,
+      "loss": 1.8509,
+      "step": 3191
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002234175587310709,
+      "loss": 1.7139,
+      "step": 3192
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002233241294806614,
+      "loss": 1.4825,
+      "step": 3193
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002232306950698646,
+      "loss": 1.6974,
+      "step": 3194
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022313725551935263,
+      "loss": 1.6684,
+      "step": 3195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022304381084979863,
+      "loss": 1.9226,
+      "step": 3196
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022295036108187693,
+      "loss": 1.7622,
+      "step": 3197
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002228569062362631,
+      "loss": 1.7618,
+      "step": 3198
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022276344633363353,
+      "loss": 1.6022,
+      "step": 3199
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022266998139466605,
+      "loss": 1.6631,
+      "step": 3200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022257651144003943,
+      "loss": 1.4355,
+      "step": 3201
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002224830364904336,
+      "loss": 1.7194,
+      "step": 3202
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022238955656652955,
+      "loss": 1.8016,
+      "step": 3203
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022229607168900947,
+      "loss": 1.7554,
+      "step": 3204
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022220258187855653,
+      "loss": 1.6032,
+      "step": 3205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002221090871558551,
+      "loss": 1.6488,
+      "step": 3206
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022201558754159054,
+      "loss": 1.756,
+      "step": 3207
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022192208305644933,
+      "loss": 1.5507,
+      "step": 3208
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022182857372111904,
+      "loss": 1.7864,
+      "step": 3209
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002217350595562883,
+      "loss": 1.6181,
+      "step": 3210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002216415405826468,
+      "loss": 1.5972,
+      "step": 3211
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022154801682088538,
+      "loss": 1.6684,
+      "step": 3212
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022145448829169578,
+      "loss": 1.7442,
+      "step": 3213
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022136095501577093,
+      "loss": 1.7334,
+      "step": 3214
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022126741701380476,
+      "loss": 1.6647,
+      "step": 3215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002211738743064922,
+      "loss": 1.7365,
+      "step": 3216
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022108032691452927,
+      "loss": 1.7141,
+      "step": 3217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002209867748586131,
+      "loss": 1.6552,
+      "step": 3218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022089321815944168,
+      "loss": 1.7033,
+      "step": 3219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022079965683771425,
+      "loss": 1.6874,
+      "step": 3220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022070609091413078,
+      "loss": 1.7712,
+      "step": 3221
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002206125204093926,
+      "loss": 1.6608,
+      "step": 3222
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022051894534420171,
+      "loss": 1.8063,
+      "step": 3223
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002204253657392614,
+      "loss": 1.7491,
+      "step": 3224
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002203317816152758,
+      "loss": 1.6207,
+      "step": 3225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022023819299295009,
+      "loss": 1.7538,
+      "step": 3226
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022014459989299053,
+      "loss": 1.755,
+      "step": 3227
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022005100233610415,
+      "loss": 1.7419,
+      "step": 3228
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021995740034299927,
+      "loss": 1.6211,
+      "step": 3229
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021986379393438483,
+      "loss": 1.6727,
+      "step": 3230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021977018313097105,
+      "loss": 1.7619,
+      "step": 3231
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021967656795346906,
+      "loss": 1.7171,
+      "step": 3232
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021958294842259083,
+      "loss": 1.8272,
+      "step": 3233
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021948932455904938,
+      "loss": 1.6087,
+      "step": 3234
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021939569638355867,
+      "loss": 1.7038,
+      "step": 3235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021930206391683374,
+      "loss": 1.5868,
+      "step": 3236
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021920842717959032,
+      "loss": 1.7569,
+      "step": 3237
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021911478619254533,
+      "loss": 1.7808,
+      "step": 3238
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002190211409764165,
+      "loss": 1.7347,
+      "step": 3239
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002189274915519225,
+      "loss": 1.7298,
+      "step": 3240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021883383793978305,
+      "loss": 1.5793,
+      "step": 3241
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021874018016071855,
+      "loss": 1.4935,
+      "step": 3242
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021864651823545064,
+      "loss": 1.7904,
+      "step": 3243
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002185528521847016,
+      "loss": 1.8212,
+      "step": 3244
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002184591820291948,
+      "loss": 1.8791,
+      "step": 3245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021836550778965444,
+      "loss": 1.6778,
+      "step": 3246
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021827182948680563,
+      "loss": 1.8432,
+      "step": 3247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021817814714137436,
+      "loss": 1.7233,
+      "step": 3248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021808446077408754,
+      "loss": 1.7158,
+      "step": 3249
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021799077040567309,
+      "loss": 1.8415,
+      "step": 3250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021789707605685953,
+      "loss": 1.7218,
+      "step": 3251
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021780337774837655,
+      "loss": 1.691,
+      "step": 3252
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021770967550095458,
+      "loss": 1.6185,
+      "step": 3253
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002176159693353249,
+      "loss": 1.7934,
+      "step": 3254
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021752225927221966,
+      "loss": 1.7119,
+      "step": 3255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021742854533237193,
+      "loss": 1.8139,
+      "step": 3256
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021733482753651568,
+      "loss": 1.7075,
+      "step": 3257
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002172411059053856,
+      "loss": 1.7502,
+      "step": 3258
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021714738045971722,
+      "loss": 1.6778,
+      "step": 3259
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021705365122024723,
+      "loss": 1.5608,
+      "step": 3260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002169599182077127,
+      "loss": 1.7271,
+      "step": 3261
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021686618144285178,
+      "loss": 1.6872,
+      "step": 3262
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021677244094640345,
+      "loss": 1.7633,
+      "step": 3263
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002166786967391075,
+      "loss": 1.8613,
+      "step": 3264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021658494884170456,
+      "loss": 1.7832,
+      "step": 3265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000216491197274936,
+      "loss": 1.454,
+      "step": 3266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000216397442059544,
+      "loss": 1.5658,
+      "step": 3267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002163036832162717,
+      "loss": 1.7288,
+      "step": 3268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021620992076586292,
+      "loss": 1.7119,
+      "step": 3269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021611615472906217,
+      "loss": 1.4244,
+      "step": 3270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000216022385126615,
+      "loss": 1.5922,
+      "step": 3271
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002159286119792676,
+      "loss": 1.6069,
+      "step": 3272
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000215834835307767,
+      "loss": 1.6717,
+      "step": 3273
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002157410551328609,
+      "loss": 1.6455,
+      "step": 3274
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021564727147529796,
+      "loss": 1.7274,
+      "step": 3275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021555348435582742,
+      "loss": 1.7189,
+      "step": 3276
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021545969379519942,
+      "loss": 1.806,
+      "step": 3277
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021536589981416477,
+      "loss": 1.9979,
+      "step": 3278
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021527210243347516,
+      "loss": 1.8247,
+      "step": 3279
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002151783016738829,
+      "loss": 1.6519,
+      "step": 3280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021508449755614105,
+      "loss": 1.6929,
+      "step": 3281
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021499069010100358,
+      "loss": 1.7776,
+      "step": 3282
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021489687932922498,
+      "loss": 1.5838,
+      "step": 3283
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021480306526156065,
+      "loss": 1.6744,
+      "step": 3284
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021470924791876662,
+      "loss": 1.9064,
+      "step": 3285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021461542732159963,
+      "loss": 1.7116,
+      "step": 3286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021452160349081726,
+      "loss": 1.6439,
+      "step": 3287
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002144277764471776,
+      "loss": 1.8836,
+      "step": 3288
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021433394621143972,
+      "loss": 1.8414,
+      "step": 3289
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021424011280436314,
+      "loss": 1.5304,
+      "step": 3290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002141462762467083,
+      "loss": 1.7502,
+      "step": 3291
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021405243655923614,
+      "loss": 1.6035,
+      "step": 3292
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021395859376270845,
+      "loss": 1.7958,
+      "step": 3293
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021386474787788755,
+      "loss": 1.5272,
+      "step": 3294
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021377089892553655,
+      "loss": 1.6042,
+      "step": 3295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021367704692641936,
+      "loss": 1.6682,
+      "step": 3296
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021358319190130032,
+      "loss": 1.7129,
+      "step": 3297
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021348933387094448,
+      "loss": 1.7905,
+      "step": 3298
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021339547285611777,
+      "loss": 1.7394,
+      "step": 3299
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021330160887758664,
+      "loss": 1.5988,
+      "step": 3300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000213207741956118,
+      "loss": 1.6786,
+      "step": 3301
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002131138721124797,
+      "loss": 1.6093,
+      "step": 3302
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021301999936744018,
+      "loss": 1.7383,
+      "step": 3303
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021292612374176843,
+      "loss": 1.7475,
+      "step": 3304
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021283224525623417,
+      "loss": 1.6022,
+      "step": 3305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002127383639316076,
+      "loss": 1.7157,
+      "step": 3306
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002126444797886598,
+      "loss": 1.6026,
+      "step": 3307
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002125505928481623,
+      "loss": 1.7608,
+      "step": 3308
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021245670313088715,
+      "loss": 1.953,
+      "step": 3309
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002123628106576073,
+      "loss": 1.7561,
+      "step": 3310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021226891544909605,
+      "loss": 1.8265,
+      "step": 3311
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002121750175261274,
+      "loss": 1.8667,
+      "step": 3312
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021208111690947603,
+      "loss": 1.8116,
+      "step": 3313
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021198721361991713,
+      "loss": 1.7482,
+      "step": 3314
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021189330767822643,
+      "loss": 1.7362,
+      "step": 3315
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021179939910518042,
+      "loss": 1.614,
+      "step": 3316
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021170548792155596,
+      "loss": 1.7448,
+      "step": 3317
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021161157414813065,
+      "loss": 1.7366,
+      "step": 3318
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002115176578056826,
+      "loss": 1.8324,
+      "step": 3319
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021142373891499045,
+      "loss": 1.6527,
+      "step": 3320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002113298174968335,
+      "loss": 1.7893,
+      "step": 3321
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021123589357199154,
+      "loss": 1.7349,
+      "step": 3322
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002111419671612449,
+      "loss": 1.7976,
+      "step": 3323
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021104803828537455,
+      "loss": 1.8621,
+      "step": 3324
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021095410696516195,
+      "loss": 1.7861,
+      "step": 3325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021086017322138897,
+      "loss": 1.7662,
+      "step": 3326
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021076623707483826,
+      "loss": 1.7026,
+      "step": 3327
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021067229854629288,
+      "loss": 1.6607,
+      "step": 3328
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021057835765653642,
+      "loss": 1.7583,
+      "step": 3329
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021048441442635292,
+      "loss": 1.6883,
+      "step": 3330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021039046887652712,
+      "loss": 1.5724,
+      "step": 3331
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002102965210278441,
+      "loss": 1.791,
+      "step": 3332
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021020257090108952,
+      "loss": 1.7007,
+      "step": 3333
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002101086185170495,
+      "loss": 1.8812,
+      "step": 3334
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021001466389651078,
+      "loss": 1.6407,
+      "step": 3335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020992070706026046,
+      "loss": 1.6232,
+      "step": 3336
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020982674802908616,
+      "loss": 1.6354,
+      "step": 3337
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020973278682377613,
+      "loss": 1.5366,
+      "step": 3338
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020963882346511885,
+      "loss": 1.5954,
+      "step": 3339
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002095448579739035,
+      "loss": 1.6532,
+      "step": 3340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020945089037091949,
+      "loss": 1.6704,
+      "step": 3341
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020935692067695696,
+      "loss": 1.6852,
+      "step": 3342
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002092629489128064,
+      "loss": 1.6832,
+      "step": 3343
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020916897509925874,
+      "loss": 1.8331,
+      "step": 3344
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020907499925710532,
+      "loss": 1.5019,
+      "step": 3345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020898102140713806,
+      "loss": 1.7605,
+      "step": 3346
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020888704157014928,
+      "loss": 1.7448,
+      "step": 3347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002087930597669316,
+      "loss": 1.6306,
+      "step": 3348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020869907601827828,
+      "loss": 1.6701,
+      "step": 3349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020860509034498285,
+      "loss": 1.744,
+      "step": 3350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020851110276783944,
+      "loss": 1.7878,
+      "step": 3351
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002084171133076423,
+      "loss": 1.6051,
+      "step": 3352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020832312198518644,
+      "loss": 1.7685,
+      "step": 3353
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020822912882126717,
+      "loss": 1.6166,
+      "step": 3354
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020813513383668007,
+      "loss": 1.6939,
+      "step": 3355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020804113705222123,
+      "loss": 1.7053,
+      "step": 3356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020794713848868717,
+      "loss": 1.7371,
+      "step": 3357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020785313816687473,
+      "loss": 1.7026,
+      "step": 3358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020775913610758117,
+      "loss": 1.7911,
+      "step": 3359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002076651323316042,
+      "loss": 1.4349,
+      "step": 3360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002075711268597418,
+      "loss": 1.5674,
+      "step": 3361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002074771197127924,
+      "loss": 1.6748,
+      "step": 3362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002073831109115547,
+      "loss": 1.4932,
+      "step": 3363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020728910047682793,
+      "loss": 1.4318,
+      "step": 3364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020719508842941155,
+      "loss": 1.6715,
+      "step": 3365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020710107479010538,
+      "loss": 1.7946,
+      "step": 3366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002070070595797097,
+      "loss": 1.5707,
+      "step": 3367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020691304281902505,
+      "loss": 1.5147,
+      "step": 3368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002068190245288523,
+      "loss": 1.6307,
+      "step": 3369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020672500472999268,
+      "loss": 1.5858,
+      "step": 3370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020663098344324786,
+      "loss": 1.5606,
+      "step": 3371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020653696068941967,
+      "loss": 1.5621,
+      "step": 3372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020644293648931027,
+      "loss": 1.6761,
+      "step": 3373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002063489108637223,
+      "loss": 1.6076,
+      "step": 3374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020625488383345858,
+      "loss": 1.5146,
+      "step": 3375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002061608554193223,
+      "loss": 1.663,
+      "step": 3376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020606682564211693,
+      "loss": 1.7993,
+      "step": 3377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002059727945226463,
+      "loss": 1.6789,
+      "step": 3378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020587876208171446,
+      "loss": 1.7126,
+      "step": 3379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020578472834012569,
+      "loss": 1.616,
+      "step": 3380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002056906933186848,
+      "loss": 1.5626,
+      "step": 3381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020559665703819658,
+      "loss": 1.7607,
+      "step": 3382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002055026195194664,
+      "loss": 1.66,
+      "step": 3383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020540858078329954,
+      "loss": 1.7625,
+      "step": 3384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020531454085050198,
+      "loss": 1.6745,
+      "step": 3385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002052204997418797,
+      "loss": 1.6289,
+      "step": 3386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002051264574782389,
+      "loss": 1.8016,
+      "step": 3387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020503241408038618,
+      "loss": 1.6626,
+      "step": 3388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020493836956912836,
+      "loss": 1.6698,
+      "step": 3389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020484432396527242,
+      "loss": 1.7338,
+      "step": 3390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002047502772896256,
+      "loss": 1.6031,
+      "step": 3391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002046562295629956,
+      "loss": 1.5973,
+      "step": 3392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020456218080619,
+      "loss": 1.5806,
+      "step": 3393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020446813104001686,
+      "loss": 1.6844,
+      "step": 3394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020437408028528434,
+      "loss": 1.7871,
+      "step": 3395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002042800285628009,
+      "loss": 1.6384,
+      "step": 3396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002041859758933752,
+      "loss": 1.5295,
+      "step": 3397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020409192229781594,
+      "loss": 1.7206,
+      "step": 3398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020399786779693235,
+      "loss": 1.816,
+      "step": 3399
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020390381241153357,
+      "loss": 1.4965,
+      "step": 3400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020380975616242909,
+      "loss": 1.4979,
+      "step": 3401
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002037156990704286,
+      "loss": 1.8731,
+      "step": 3402
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020362164115634177,
+      "loss": 1.7492,
+      "step": 3403
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020352758244097873,
+      "loss": 1.7806,
+      "step": 3404
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020343352294514957,
+      "loss": 1.7502,
+      "step": 3405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002033394626896648,
+      "loss": 1.6994,
+      "step": 3406
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020324540169533477,
+      "loss": 1.6593,
+      "step": 3407
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020315133998297026,
+      "loss": 1.7519,
+      "step": 3408
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020305727757338212,
+      "loss": 1.8678,
+      "step": 3409
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020296321448738135,
+      "loss": 1.6347,
+      "step": 3410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020286915074577909,
+      "loss": 1.6374,
+      "step": 3411
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020277508636938656,
+      "loss": 1.7194,
+      "step": 3412
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002026810213790153,
+      "loss": 1.7048,
+      "step": 3413
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020258695579547685,
+      "loss": 1.6664,
+      "step": 3414
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020249288963958284,
+      "loss": 1.5324,
+      "step": 3415
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020239882293214523,
+      "loss": 1.6318,
+      "step": 3416
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002023047556939759,
+      "loss": 1.7078,
+      "step": 3417
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002022106879458869,
+      "loss": 1.9372,
+      "step": 3418
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002021166197086904,
+      "loss": 1.6461,
+      "step": 3419
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020202255100319877,
+      "loss": 1.5375,
+      "step": 3420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020192848185022436,
+      "loss": 1.5455,
+      "step": 3421
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020183441227057967,
+      "loss": 1.6272,
+      "step": 3422
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020174034228507716,
+      "loss": 1.7224,
+      "step": 3423
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020164627191452975,
+      "loss": 1.6838,
+      "step": 3424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020155220117975002,
+      "loss": 1.5423,
+      "step": 3425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020145813010155082,
+      "loss": 1.7988,
+      "step": 3426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002013640587007452,
+      "loss": 1.7048,
+      "step": 3427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020126998699814604,
+      "loss": 1.8154,
+      "step": 3428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020117591501456637,
+      "loss": 1.7796,
+      "step": 3429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020108184277081934,
+      "loss": 1.7433,
+      "step": 3430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020098777028771818,
+      "loss": 1.6016,
+      "step": 3431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002008936975860761,
+      "loss": 1.6866,
+      "step": 3432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020079962468670632,
+      "loss": 1.683,
+      "step": 3433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020070555161042219,
+      "loss": 1.722,
+      "step": 3434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020061147837803711,
+      "loss": 1.7221,
+      "step": 3435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020051740501036443,
+      "loss": 1.8571,
+      "step": 3436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020042333152821756,
+      "loss": 1.8402,
+      "step": 3437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020032925795241,
+      "loss": 1.5308,
+      "step": 3438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002002351843037552,
+      "loss": 1.7032,
+      "step": 3439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020014111060306662,
+      "loss": 1.6164,
+      "step": 3440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020004703687115778,
+      "loss": 1.6375,
+      "step": 3441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019995296312884226,
+      "loss": 1.8288,
+      "step": 3442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019985888939693346,
+      "loss": 1.78,
+      "step": 3443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019976481569624488,
+      "loss": 1.68,
+      "step": 3444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019967074204759009,
+      "loss": 1.5357,
+      "step": 3445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019957666847178246,
+      "loss": 1.7564,
+      "step": 3446
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001994825949896356,
+      "loss": 1.7303,
+      "step": 3447
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019938852162196293,
+      "loss": 1.6892,
+      "step": 3448
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019929444838957783,
+      "loss": 1.856,
+      "step": 3449
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019920037531329373,
+      "loss": 1.6682,
+      "step": 3450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000199106302413924,
+      "loss": 1.4921,
+      "step": 3451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001990122297122819,
+      "loss": 1.6322,
+      "step": 3452
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019891815722918065,
+      "loss": 1.6171,
+      "step": 3453
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019882408498543368,
+      "loss": 1.5913,
+      "step": 3454
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019873001300185403,
+      "loss": 1.8826,
+      "step": 3455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019863594129925484,
+      "loss": 1.7496,
+      "step": 3456
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001985418698984492,
+      "loss": 1.6895,
+      "step": 3457
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019844779882025005,
+      "loss": 1.7472,
+      "step": 3458
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019835372808547038,
+      "loss": 1.5836,
+      "step": 3459
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001982596577149228,
+      "loss": 1.6281,
+      "step": 3460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001981655877294204,
+      "loss": 1.6771,
+      "step": 3461
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019807151814977566,
+      "loss": 1.7121,
+      "step": 3462
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019797744899680125,
+      "loss": 1.7778,
+      "step": 3463
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001978833802913096,
+      "loss": 1.5628,
+      "step": 3464
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019778931205411315,
+      "loss": 1.574,
+      "step": 3465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019769524430602418,
+      "loss": 1.8394,
+      "step": 3466
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019760117706785478,
+      "loss": 1.5664,
+      "step": 3467
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019750711036041715,
+      "loss": 1.649,
+      "step": 3468
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001974130442045232,
+      "loss": 1.6258,
+      "step": 3469
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019731897862098474,
+      "loss": 1.8043,
+      "step": 3470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001972249136306135,
+      "loss": 1.6433,
+      "step": 3471
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000197130849254221,
+      "loss": 1.7482,
+      "step": 3472
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001970367855126187,
+      "loss": 1.6857,
+      "step": 3473
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019694272242661788,
+      "loss": 1.6729,
+      "step": 3474
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019684866001702973,
+      "loss": 1.8157,
+      "step": 3475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019675459830466525,
+      "loss": 1.645,
+      "step": 3476
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019666053731033528,
+      "loss": 1.653,
+      "step": 3477
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019656647705485044,
+      "loss": 1.7855,
+      "step": 3478
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019647241755902135,
+      "loss": 1.6507,
+      "step": 3479
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019637835884365833,
+      "loss": 1.7562,
+      "step": 3480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019628430092957148,
+      "loss": 1.6917,
+      "step": 3481
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019619024383757093,
+      "loss": 1.6663,
+      "step": 3482
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019609618758846645,
+      "loss": 1.6933,
+      "step": 3483
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019600213220306772,
+      "loss": 1.9148,
+      "step": 3484
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001959080777021841,
+      "loss": 1.6237,
+      "step": 3485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001958140241066249,
+      "loss": 1.6678,
+      "step": 3486
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019571997143719917,
+      "loss": 1.5786,
+      "step": 3487
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019562591971471568,
+      "loss": 1.5989,
+      "step": 3488
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001955318689599832,
+      "loss": 1.6251,
+      "step": 3489
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019543781919381003,
+      "loss": 1.5699,
+      "step": 3490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019534377043700445,
+      "loss": 1.6627,
+      "step": 3491
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019524972271037442,
+      "loss": 1.7202,
+      "step": 3492
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019515567603472765,
+      "loss": 1.6786,
+      "step": 3493
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019506163043087174,
+      "loss": 1.5742,
+      "step": 3494
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001949675859196138,
+      "loss": 1.6653,
+      "step": 3495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019487354252176113,
+      "loss": 1.7963,
+      "step": 3496
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019477950025812034,
+      "loss": 1.6751,
+      "step": 3497
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019468545914949804,
+      "loss": 1.7682,
+      "step": 3498
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019459141921670048,
+      "loss": 1.6689,
+      "step": 3499
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001944973804805337,
+      "loss": 1.7682,
+      "step": 3500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001944033429618035,
+      "loss": 1.6917,
+      "step": 3501
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019430930668131523,
+      "loss": 1.7705,
+      "step": 3502
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019421527165987428,
+      "loss": 1.6703,
+      "step": 3503
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001941212379182856,
+      "loss": 1.6305,
+      "step": 3504
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019402720547735374,
+      "loss": 1.6344,
+      "step": 3505
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019393317435788309,
+      "loss": 1.8104,
+      "step": 3506
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019383914458067774,
+      "loss": 1.7134,
+      "step": 3507
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019374511616654147,
+      "loss": 1.8188,
+      "step": 3508
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019365108913627768,
+      "loss": 1.8337,
+      "step": 3509
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019355706351068975,
+      "loss": 1.6456,
+      "step": 3510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019346303931058037,
+      "loss": 1.7474,
+      "step": 3511
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019336901655675216,
+      "loss": 1.6877,
+      "step": 3512
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019327499527000734,
+      "loss": 1.7095,
+      "step": 3513
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019318097547114778,
+      "loss": 1.7436,
+      "step": 3514
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019308695718097503,
+      "loss": 1.7387,
+      "step": 3515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001929929404202903,
+      "loss": 1.7737,
+      "step": 3516
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019289892520989461,
+      "loss": 1.7348,
+      "step": 3517
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001928049115705885,
+      "loss": 1.8664,
+      "step": 3518
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019271089952317208,
+      "loss": 1.6055,
+      "step": 3519
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019261688908844534,
+      "loss": 1.7792,
+      "step": 3520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019252288028720767,
+      "loss": 1.6715,
+      "step": 3521
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019242887314025828,
+      "loss": 1.7039,
+      "step": 3522
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019233486766839588,
+      "loss": 1.7353,
+      "step": 3523
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019224086389241883,
+      "loss": 1.8221,
+      "step": 3524
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001921468618331253,
+      "loss": 1.6689,
+      "step": 3525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019205286151131287,
+      "loss": 1.6001,
+      "step": 3526
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019195886294777882,
+      "loss": 1.8592,
+      "step": 3527
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019186486616332,
+      "loss": 1.6468,
+      "step": 3528
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001917708711787329,
+      "loss": 1.5556,
+      "step": 3529
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019167687801481363,
+      "loss": 1.6721,
+      "step": 3530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001915828866923577,
+      "loss": 1.7297,
+      "step": 3531
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019148889723216063,
+      "loss": 1.7676,
+      "step": 3532
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019139490965501717,
+      "loss": 1.6279,
+      "step": 3533
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019130092398172177,
+      "loss": 1.6926,
+      "step": 3534
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019120694023306843,
+      "loss": 1.6447,
+      "step": 3535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019111295842985077,
+      "loss": 1.7217,
+      "step": 3536
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019101897859286196,
+      "loss": 1.7264,
+      "step": 3537
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019092500074289467,
+      "loss": 1.6807,
+      "step": 3538
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001908310249007413,
+      "loss": 1.5805,
+      "step": 3539
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019073705108719363,
+      "loss": 1.6739,
+      "step": 3540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019064307932304306,
+      "loss": 1.6716,
+      "step": 3541
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019054910962908056,
+      "loss": 1.7894,
+      "step": 3542
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001904551420260966,
+      "loss": 1.6199,
+      "step": 3543
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019036117653488123,
+      "loss": 1.7352,
+      "step": 3544
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001902672131762239,
+      "loss": 1.8426,
+      "step": 3545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019017325197091383,
+      "loss": 1.7658,
+      "step": 3546
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019007929293973956,
+      "loss": 1.6525,
+      "step": 3547
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018998533610348924,
+      "loss": 1.7052,
+      "step": 3548
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018989138148295054,
+      "loss": 1.6755,
+      "step": 3549
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018979742909891055,
+      "loss": 1.7658,
+      "step": 3550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000189703478972156,
+      "loss": 1.6031,
+      "step": 3551
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018960953112347293,
+      "loss": 1.7582,
+      "step": 3552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001895155855736471,
+      "loss": 1.6066,
+      "step": 3553
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018942164234346366,
+      "loss": 1.6028,
+      "step": 3554
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018932770145370714,
+      "loss": 1.6793,
+      "step": 3555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018923376292516176,
+      "loss": 1.7208,
+      "step": 3556
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018913982677861107,
+      "loss": 1.6738,
+      "step": 3557
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018904589303483818,
+      "loss": 1.5103,
+      "step": 3558
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018895196171462547,
+      "loss": 1.9111,
+      "step": 3559
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018885803283875513,
+      "loss": 1.6733,
+      "step": 3560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001887641064280085,
+      "loss": 1.6698,
+      "step": 3561
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018867018250316654,
+      "loss": 1.6601,
+      "step": 3562
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001885762610850096,
+      "loss": 1.6403,
+      "step": 3563
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001884823421943175,
+      "loss": 1.7121,
+      "step": 3564
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018838842585186942,
+      "loss": 1.8361,
+      "step": 3565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018829451207844403,
+      "loss": 1.7748,
+      "step": 3566
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018820060089481963,
+      "loss": 1.6542,
+      "step": 3567
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001881066923217736,
+      "loss": 1.5158,
+      "step": 3568
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018801278638008292,
+      "loss": 1.681,
+      "step": 3569
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018791888309052405,
+      "loss": 1.757,
+      "step": 3570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018782498247387267,
+      "loss": 1.58,
+      "step": 3571
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018773108455090405,
+      "loss": 1.8303,
+      "step": 3572
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018763718934239272,
+      "loss": 1.4957,
+      "step": 3573
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018754329686911282,
+      "loss": 1.5058,
+      "step": 3574
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018744940715183776,
+      "loss": 1.7124,
+      "step": 3575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018735552021134023,
+      "loss": 1.7936,
+      "step": 3576
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001872616360683924,
+      "loss": 1.5597,
+      "step": 3577
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001871677547437659,
+      "loss": 1.7235,
+      "step": 3578
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018707387625823164,
+      "loss": 1.6844,
+      "step": 3579
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018698000063255982,
+      "loss": 1.6807,
+      "step": 3580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001868861278875203,
+      "loss": 1.6333,
+      "step": 3581
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018679225804388202,
+      "loss": 1.547,
+      "step": 3582
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018669839112241344,
+      "loss": 1.5066,
+      "step": 3583
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018660452714388225,
+      "loss": 1.5654,
+      "step": 3584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018651066612905554,
+      "loss": 1.6057,
+      "step": 3585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018641680809869976,
+      "loss": 1.7318,
+      "step": 3586
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018632295307358064,
+      "loss": 1.9189,
+      "step": 3587
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018622910107446344,
+      "loss": 1.7155,
+      "step": 3588
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001861352521221125,
+      "loss": 1.7808,
+      "step": 3589
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018604140623729163,
+      "loss": 1.5583,
+      "step": 3590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001859475634407639,
+      "loss": 1.7715,
+      "step": 3591
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018585372375329177,
+      "loss": 1.6229,
+      "step": 3592
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018575988719563694,
+      "loss": 1.6678,
+      "step": 3593
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001856660537885603,
+      "loss": 1.6042,
+      "step": 3594
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001855722235528224,
+      "loss": 1.4973,
+      "step": 3595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001854783965091828,
+      "loss": 1.6238,
+      "step": 3596
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001853845726784004,
+      "loss": 1.6284,
+      "step": 3597
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018529075208123343,
+      "loss": 1.6725,
+      "step": 3598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001851969347384394,
+      "loss": 1.6866,
+      "step": 3599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018510312067077507,
+      "loss": 1.5589,
+      "step": 3600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018500930989899652,
+      "loss": 1.8507,
+      "step": 3601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018491550244385897,
+      "loss": 1.7288,
+      "step": 3602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018482169832611714,
+      "loss": 1.6522,
+      "step": 3603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018472789756652488,
+      "loss": 1.8538,
+      "step": 3604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018463410018583525,
+      "loss": 1.7357,
+      "step": 3605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018454030620480063,
+      "loss": 1.8936,
+      "step": 3606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018444651564417263,
+      "loss": 1.514,
+      "step": 3607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001843527285247021,
+      "loss": 1.685,
+      "step": 3608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018425894486713912,
+      "loss": 1.5852,
+      "step": 3609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018416516469223305,
+      "loss": 1.6774,
+      "step": 3610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018407138802073244,
+      "loss": 1.79,
+      "step": 3611
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018397761487338503,
+      "loss": 1.7156,
+      "step": 3612
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018388384527093788,
+      "loss": 1.6343,
+      "step": 3613
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018379007923413718,
+      "loss": 1.7861,
+      "step": 3614
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018369631678372836,
+      "loss": 1.5993,
+      "step": 3615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000183602557940456,
+      "loss": 1.6086,
+      "step": 3616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018350880272506405,
+      "loss": 1.608,
+      "step": 3617
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001834150511582955,
+      "loss": 1.6391,
+      "step": 3618
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018332130326089251,
+      "loss": 1.7144,
+      "step": 3619
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018322755905359657,
+      "loss": 1.6485,
+      "step": 3620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001831338185571483,
+      "loss": 1.5918,
+      "step": 3621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001830400817922874,
+      "loss": 1.6462,
+      "step": 3622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018294634877975282,
+      "loss": 1.6356,
+      "step": 3623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018285261954028274,
+      "loss": 1.7063,
+      "step": 3624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018275889409461447,
+      "loss": 1.6266,
+      "step": 3625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018266517246348437,
+      "loss": 1.5392,
+      "step": 3626
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018257145466762812,
+      "loss": 1.5917,
+      "step": 3627
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018247774072778042,
+      "loss": 1.5856,
+      "step": 3628
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001823840306646752,
+      "loss": 1.7267,
+      "step": 3629
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018229032449904546,
+      "loss": 1.786,
+      "step": 3630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018219662225162347,
+      "loss": 1.7513,
+      "step": 3631
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001821029239431405,
+      "loss": 1.7895,
+      "step": 3632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000182009229594327,
+      "loss": 1.5635,
+      "step": 3633
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018191553922591248,
+      "loss": 1.5481,
+      "step": 3634
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018182185285862571,
+      "loss": 1.7595,
+      "step": 3635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018172817051319447,
+      "loss": 1.8361,
+      "step": 3636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018163449221034555,
+      "loss": 1.6654,
+      "step": 3637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018154081797080524,
+      "loss": 1.665,
+      "step": 3638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018144714781529844,
+      "loss": 1.5614,
+      "step": 3639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018135348176454943,
+      "loss": 1.6885,
+      "step": 3640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001812598198392815,
+      "loss": 1.6832,
+      "step": 3641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018116616206021705,
+      "loss": 1.8746,
+      "step": 3642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018107250844807754,
+      "loss": 1.7233,
+      "step": 3643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018097885902358348,
+      "loss": 1.7125,
+      "step": 3644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018088521380745466,
+      "loss": 1.6465,
+      "step": 3645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018079157282040973,
+      "loss": 1.5254,
+      "step": 3646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018069793608316634,
+      "loss": 1.7839,
+      "step": 3647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018060430361644138,
+      "loss": 1.6952,
+      "step": 3648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001805106754409507,
+      "loss": 1.6472,
+      "step": 3649
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001804170515774093,
+      "loss": 1.5257,
+      "step": 3650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018032343204653096,
+      "loss": 1.7858,
+      "step": 3651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018022981686902895,
+      "loss": 1.8239,
+      "step": 3652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018013620606561519,
+      "loss": 1.6951,
+      "step": 3653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018004259965700078,
+      "loss": 1.6776,
+      "step": 3654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001799489976638959,
+      "loss": 1.6528,
+      "step": 3655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017985540010700957,
+      "loss": 1.7647,
+      "step": 3656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017976180700704996,
+      "loss": 1.7045,
+      "step": 3657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017966821838472422,
+      "loss": 1.4837,
+      "step": 3658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017957463426073863,
+      "loss": 1.8004,
+      "step": 3659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017948105465579833,
+      "loss": 1.9135,
+      "step": 3660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017938747959060747,
+      "loss": 1.7548,
+      "step": 3661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017929390908586924,
+      "loss": 1.5421,
+      "step": 3662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017920034316228582,
+      "loss": 1.6909,
+      "step": 3663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001791067818405584,
+      "loss": 1.6907,
+      "step": 3664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017901322514138693,
+      "loss": 1.6753,
+      "step": 3665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017891967308547075,
+      "loss": 1.7582,
+      "step": 3666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017882612569350786,
+      "loss": 1.5952,
+      "step": 3667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001787325829861953,
+      "loss": 1.6551,
+      "step": 3668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001786390449842291,
+      "loss": 1.6626,
+      "step": 3669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017854551170830423,
+      "loss": 1.6185,
+      "step": 3670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017845198317911466,
+      "loss": 1.7401,
+      "step": 3671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001783584594173532,
+      "loss": 1.7297,
+      "step": 3672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001782649404437117,
+      "loss": 1.6132,
+      "step": 3673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017817142627888098,
+      "loss": 1.7688,
+      "step": 3674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017807791694355072,
+      "loss": 1.6571,
+      "step": 3675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001779844124584095,
+      "loss": 1.667,
+      "step": 3676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017789091284414497,
+      "loss": 1.8199,
+      "step": 3677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017779741812144352,
+      "loss": 1.6583,
+      "step": 3678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001777039283109906,
+      "loss": 1.6978,
+      "step": 3679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017761044343347047,
+      "loss": 1.7201,
+      "step": 3680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017751696350956647,
+      "loss": 1.8245,
+      "step": 3681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017742348855996062,
+      "loss": 1.5766,
+      "step": 3682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000177330018605334,
+      "loss": 1.8581,
+      "step": 3683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017723655366636655,
+      "loss": 1.6881,
+      "step": 3684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017714309376373698,
+      "loss": 1.6321,
+      "step": 3685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001770496389181231,
+      "loss": 1.7543,
+      "step": 3686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001769561891502014,
+      "loss": 1.7112,
+      "step": 3687
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017686274448064742,
+      "loss": 1.6851,
+      "step": 3688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017676930493013544,
+      "loss": 1.5084,
+      "step": 3689
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017667587051933866,
+      "loss": 1.5135,
+      "step": 3690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001765824412689291,
+      "loss": 1.7996,
+      "step": 3691
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017648901719957775,
+      "loss": 1.6667,
+      "step": 3692
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017639559833195436,
+      "loss": 1.6528,
+      "step": 3693
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017630218468672745,
+      "loss": 1.538,
+      "step": 3694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017620877628456464,
+      "loss": 1.6274,
+      "step": 3695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017611537314613215,
+      "loss": 1.7577,
+      "step": 3696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001760219752920951,
+      "loss": 1.7007,
+      "step": 3697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001759285827431175,
+      "loss": 1.6449,
+      "step": 3698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017583519551986213,
+      "loss": 1.5096,
+      "step": 3699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001757418136429906,
+      "loss": 1.6715,
+      "step": 3700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017564843713316334,
+      "loss": 1.8116,
+      "step": 3701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001755550660110396,
+      "loss": 1.6577,
+      "step": 3702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001754617002972775,
+      "loss": 1.7302,
+      "step": 3703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001753683400125338,
+      "loss": 1.7943,
+      "step": 3704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017527498517746421,
+      "loss": 1.7165,
+      "step": 3705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017518163581272317,
+      "loss": 1.7664,
+      "step": 3706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017508829193896394,
+      "loss": 1.6438,
+      "step": 3707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001749949535768384,
+      "loss": 1.794,
+      "step": 3708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017490162074699768,
+      "loss": 1.7149,
+      "step": 3709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017480829347009107,
+      "loss": 1.6027,
+      "step": 3710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017471497176676707,
+      "loss": 1.7148,
+      "step": 3711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001746216556576727,
+      "loss": 1.7758,
+      "step": 3712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017452834516345394,
+      "loss": 1.649,
+      "step": 3713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017443504030475535,
+      "loss": 1.8308,
+      "step": 3714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001743417411022203,
+      "loss": 1.6431,
+      "step": 3715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001742484475764911,
+      "loss": 1.6688,
+      "step": 3716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001741551597482085,
+      "loss": 1.6421,
+      "step": 3717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017406187763801222,
+      "loss": 1.7424,
+      "step": 3718
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001739686012665405,
+      "loss": 1.6269,
+      "step": 3719
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017387533065443052,
+      "loss": 1.6691,
+      "step": 3720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017378206582231807,
+      "loss": 1.571,
+      "step": 3721
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017368880679083756,
+      "loss": 1.7353,
+      "step": 3722
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001735955535806225,
+      "loss": 1.8588,
+      "step": 3723
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017350230621230472,
+      "loss": 1.7388,
+      "step": 3724
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001734090647065149,
+      "loss": 1.5401,
+      "step": 3725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001733158290838825,
+      "loss": 1.6466,
+      "step": 3726
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017322259936503542,
+      "loss": 1.6625,
+      "step": 3727
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017312937557060058,
+      "loss": 1.5899,
+      "step": 3728
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001730361577212033,
+      "loss": 1.5508,
+      "step": 3729
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017294294583746788,
+      "loss": 1.8108,
+      "step": 3730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017284973994001707,
+      "loss": 1.6601,
+      "step": 3731
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017275654004947238,
+      "loss": 1.7437,
+      "step": 3732
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017266334618645397,
+      "loss": 1.8229,
+      "step": 3733
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001725701583715807,
+      "loss": 1.7992,
+      "step": 3734
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017247697662547008,
+      "loss": 1.7918,
+      "step": 3735
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017238380096873813,
+      "loss": 1.5437,
+      "step": 3736
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017229063142199986,
+      "loss": 1.7633,
+      "step": 3737
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017219746800586862,
+      "loss": 1.5654,
+      "step": 3738
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017210431074095653,
+      "loss": 1.7682,
+      "step": 3739
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001720111596478743,
+      "loss": 1.5802,
+      "step": 3740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017191801474723137,
+      "loss": 1.613,
+      "step": 3741
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001718248760596357,
+      "loss": 1.791,
+      "step": 3742
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017173174360569387,
+      "loss": 1.7321,
+      "step": 3743
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017163861740601122,
+      "loss": 1.6406,
+      "step": 3744
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017154549748119157,
+      "loss": 1.7007,
+      "step": 3745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017145238385183744,
+      "loss": 1.7539,
+      "step": 3746
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017135927653854987,
+      "loss": 1.657,
+      "step": 3747
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017126617556192857,
+      "loss": 1.7308,
+      "step": 3748
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001711730809425718,
+      "loss": 1.556,
+      "step": 3749
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017107999270107643,
+      "loss": 1.5569,
+      "step": 3750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017098691085803798,
+      "loss": 1.6564,
+      "step": 3751
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001708938354340505,
+      "loss": 1.6595,
+      "step": 3752
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001708007664497066,
+      "loss": 1.8333,
+      "step": 3753
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001707077039255975,
+      "loss": 1.742,
+      "step": 3754
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017061464788231293,
+      "loss": 1.6576,
+      "step": 3755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017052159834044132,
+      "loss": 1.5764,
+      "step": 3756
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001704285553205695,
+      "loss": 1.8141,
+      "step": 3757
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017033551884328293,
+      "loss": 1.7375,
+      "step": 3758
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001702424889291657,
+      "loss": 1.8571,
+      "step": 3759
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017014946559880037,
+      "loss": 1.7386,
+      "step": 3760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017005644887276797,
+      "loss": 1.7738,
+      "step": 3761
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001699634387716482,
+      "loss": 1.7545,
+      "step": 3762
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016987043531601923,
+      "loss": 1.6802,
+      "step": 3763
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016977743852645777,
+      "loss": 1.6668,
+      "step": 3764
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000169684448423539,
+      "loss": 1.9679,
+      "step": 3765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016959146502783676,
+      "loss": 1.613,
+      "step": 3766
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001694984883599233,
+      "loss": 1.819,
+      "step": 3767
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016940551844036938,
+      "loss": 1.6844,
+      "step": 3768
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016931255528974433,
+      "loss": 1.6557,
+      "step": 3769
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001692195989286159,
+      "loss": 1.5945,
+      "step": 3770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001691266493775504,
+      "loss": 1.6748,
+      "step": 3771
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016903370665711257,
+      "loss": 1.6592,
+      "step": 3772
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016894077078786574,
+      "loss": 1.7025,
+      "step": 3773
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016884784179037167,
+      "loss": 1.6613,
+      "step": 3774
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016875491968519057,
+      "loss": 1.6383,
+      "step": 3775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016866200449288117,
+      "loss": 1.679,
+      "step": 3776
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016856909623400065,
+      "loss": 1.5669,
+      "step": 3777
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001684761949291046,
+      "loss": 1.7882,
+      "step": 3778
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016838330059874712,
+      "loss": 1.7655,
+      "step": 3779
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016829041326348098,
+      "loss": 1.7013,
+      "step": 3780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016819753294385696,
+      "loss": 1.6023,
+      "step": 3781
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016810465966042465,
+      "loss": 1.68,
+      "step": 3782
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001680117934337319,
+      "loss": 1.743,
+      "step": 3783
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016791893428432512,
+      "loss": 1.5717,
+      "step": 3784
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016782608223274905,
+      "loss": 1.6605,
+      "step": 3785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001677332372995468,
+      "loss": 1.7118,
+      "step": 3786
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016764039950526024,
+      "loss": 1.6662,
+      "step": 3787
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001675475688704293,
+      "loss": 1.5542,
+      "step": 3788
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001674547454155925,
+      "loss": 1.5678,
+      "step": 3789
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016736192916128663,
+      "loss": 1.6359,
+      "step": 3790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016726912012804705,
+      "loss": 1.7963,
+      "step": 3791
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016717631833640745,
+      "loss": 1.8207,
+      "step": 3792
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001670835238068998,
+      "loss": 1.4832,
+      "step": 3793
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001669907365600548,
+      "loss": 1.6565,
+      "step": 3794
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016689795661640124,
+      "loss": 1.7798,
+      "step": 3795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016680518399646635,
+      "loss": 1.6702,
+      "step": 3796
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016671241872077584,
+      "loss": 1.6784,
+      "step": 3797
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001666196608098536,
+      "loss": 1.617,
+      "step": 3798
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001665269102842221,
+      "loss": 1.7438,
+      "step": 3799
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016643416716440198,
+      "loss": 1.7516,
+      "step": 3800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001663414314709125,
+      "loss": 1.5853,
+      "step": 3801
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016624870322427108,
+      "loss": 1.6743,
+      "step": 3802
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016615598244499355,
+      "loss": 1.6018,
+      "step": 3803
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016606326915359406,
+      "loss": 1.6411,
+      "step": 3804
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001659705633705851,
+      "loss": 1.6982,
+      "step": 3805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001658778651164776,
+      "loss": 1.6509,
+      "step": 3806
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016578517441178056,
+      "loss": 1.7915,
+      "step": 3807
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016569249127700172,
+      "loss": 1.4705,
+      "step": 3808
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016559981573264677,
+      "loss": 1.5907,
+      "step": 3809
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016550714779921995,
+      "loss": 1.6731,
+      "step": 3810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016541448749722375,
+      "loss": 1.5746,
+      "step": 3811
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001653218348471589,
+      "loss": 1.7512,
+      "step": 3812
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016522918986952448,
+      "loss": 1.6402,
+      "step": 3813
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001651365525848179,
+      "loss": 1.5801,
+      "step": 3814
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001650439230135349,
+      "loss": 1.7761,
+      "step": 3815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016495130117616947,
+      "loss": 1.645,
+      "step": 3816
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016485868709321385,
+      "loss": 1.6941,
+      "step": 3817
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016476608078515862,
+      "loss": 1.7038,
+      "step": 3818
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016467348227249264,
+      "loss": 1.9026,
+      "step": 3819
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016458089157570297,
+      "loss": 1.8071,
+      "step": 3820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016448830871527504,
+      "loss": 1.8001,
+      "step": 3821
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001643957337116925,
+      "loss": 1.4896,
+      "step": 3822
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001643031665854373,
+      "loss": 1.6505,
+      "step": 3823
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016421060735698955,
+      "loss": 1.8316,
+      "step": 3824
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001641180560468277,
+      "loss": 1.7625,
+      "step": 3825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016402551267542847,
+      "loss": 1.6417,
+      "step": 3826
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016393297726326672,
+      "loss": 1.7122,
+      "step": 3827
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016384044983081556,
+      "loss": 1.6129,
+      "step": 3828
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001637479303985465,
+      "loss": 1.5925,
+      "step": 3829
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016365541898692913,
+      "loss": 1.7123,
+      "step": 3830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016356291561643124,
+      "loss": 1.578,
+      "step": 3831
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000163470420307519,
+      "loss": 1.7856,
+      "step": 3832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016337793308065655,
+      "loss": 1.6174,
+      "step": 3833
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016328545395630648,
+      "loss": 1.7285,
+      "step": 3834
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001631929829549295,
+      "loss": 1.6172,
+      "step": 3835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016310052009698442,
+      "loss": 1.7266,
+      "step": 3836
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016300806540292848,
+      "loss": 1.4922,
+      "step": 3837
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016291561889321692,
+      "loss": 1.6995,
+      "step": 3838
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001628231805883032,
+      "loss": 1.787,
+      "step": 3839
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016273075050863902,
+      "loss": 1.7694,
+      "step": 3840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016263832867467423,
+      "loss": 1.6285,
+      "step": 3841
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016254591510685685,
+      "loss": 1.6496,
+      "step": 3842
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016245350982563306,
+      "loss": 1.8033,
+      "step": 3843
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001623611128514473,
+      "loss": 1.7068,
+      "step": 3844
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016226872420474202,
+      "loss": 1.6097,
+      "step": 3845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016217634390595795,
+      "loss": 1.7817,
+      "step": 3846
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001620839719755339,
+      "loss": 1.7735,
+      "step": 3847
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016199160843390686,
+      "loss": 1.7143,
+      "step": 3848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000161899253301512,
+      "loss": 1.8393,
+      "step": 3849
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016180690659878247,
+      "loss": 1.6384,
+      "step": 3850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016171456834614985,
+      "loss": 1.7242,
+      "step": 3851
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016162223856404362,
+      "loss": 1.5543,
+      "step": 3852
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016152991727289137,
+      "loss": 1.6558,
+      "step": 3853
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016143760449311892,
+      "loss": 1.6619,
+      "step": 3854
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001613453002451502,
+      "loss": 1.6561,
+      "step": 3855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001612530045494072,
+      "loss": 1.7858,
+      "step": 3856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016116071742630994,
+      "loss": 1.7349,
+      "step": 3857
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016106843889627682,
+      "loss": 1.837,
+      "step": 3858
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001609761689797241,
+      "loss": 1.6907,
+      "step": 3859
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016088390769706626,
+      "loss": 1.6969,
+      "step": 3860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016079165506871566,
+      "loss": 1.6144,
+      "step": 3861
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016069941111508296,
+      "loss": 1.7497,
+      "step": 3862
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016060717585657683,
+      "loss": 1.6269,
+      "step": 3863
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016051494931360394,
+      "loss": 1.7669,
+      "step": 3864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016042273150656928,
+      "loss": 1.561,
+      "step": 3865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016033052245587566,
+      "loss": 1.6912,
+      "step": 3866
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000160238322181924,
+      "loss": 1.7228,
+      "step": 3867
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001601461307051134,
+      "loss": 1.8597,
+      "step": 3868
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001600539480458408,
+      "loss": 1.6388,
+      "step": 3869
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015996177422450137,
+      "loss": 1.7857,
+      "step": 3870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015986960926148817,
+      "loss": 1.562,
+      "step": 3871
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015977745317719257,
+      "loss": 1.6999,
+      "step": 3872
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015968530599200372,
+      "loss": 1.628,
+      "step": 3873
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001595931677263089,
+      "loss": 1.7358,
+      "step": 3874
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015950103840049336,
+      "loss": 1.7952,
+      "step": 3875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015940891803494045,
+      "loss": 1.5446,
+      "step": 3876
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001593168066500315,
+      "loss": 1.6055,
+      "step": 3877
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015922470426614572,
+      "loss": 1.7879,
+      "step": 3878
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015913261090366065,
+      "loss": 1.5521,
+      "step": 3879
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015904052658295158,
+      "loss": 1.6959,
+      "step": 3880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015894845132439188,
+      "loss": 1.6298,
+      "step": 3881
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015885638514835282,
+      "loss": 1.5385,
+      "step": 3882
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015876432807520384,
+      "loss": 1.6012,
+      "step": 3883
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015867228012531222,
+      "loss": 1.7709,
+      "step": 3884
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001585802413190432,
+      "loss": 1.7103,
+      "step": 3885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001584882116767602,
+      "loss": 1.4825,
+      "step": 3886
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015839619121882435,
+      "loss": 1.5618,
+      "step": 3887
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015830417996559498,
+      "loss": 1.6544,
+      "step": 3888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015821217793742922,
+      "loss": 1.8,
+      "step": 3889
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015812018515468226,
+      "loss": 1.6667,
+      "step": 3890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015802820163770713,
+      "loss": 1.7898,
+      "step": 3891
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001579362274068549,
+      "loss": 1.7317,
+      "step": 3892
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015784426248247461,
+      "loss": 1.6113,
+      "step": 3893
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001577523068849132,
+      "loss": 1.6988,
+      "step": 3894
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015766036063451553,
+      "loss": 1.5355,
+      "step": 3895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015756842375162439,
+      "loss": 1.6465,
+      "step": 3896
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015747649625658051,
+      "loss": 1.753,
+      "step": 3897
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001573845781697226,
+      "loss": 1.5979,
+      "step": 3898
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015729266951138712,
+      "loss": 1.7814,
+      "step": 3899
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015720077030190868,
+      "loss": 1.7811,
+      "step": 3900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015710888056161965,
+      "loss": 1.7432,
+      "step": 3901
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015701700031085032,
+      "loss": 1.6296,
+      "step": 3902
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015692512956992888,
+      "loss": 1.794,
+      "step": 3903
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015683326835918147,
+      "loss": 1.7391,
+      "step": 3904
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015674141669893204,
+      "loss": 1.71,
+      "step": 3905
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015664957460950245,
+      "loss": 1.6887,
+      "step": 3906
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015655774211121253,
+      "loss": 1.6526,
+      "step": 3907
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001564659192243799,
+      "loss": 1.7562,
+      "step": 3908
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015637410596932005,
+      "loss": 1.5703,
+      "step": 3909
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001562823023663464,
+      "loss": 1.4714,
+      "step": 3910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001561905084357701,
+      "loss": 1.8114,
+      "step": 3911
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001560987241979004,
+      "loss": 1.7821,
+      "step": 3912
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015600694967304415,
+      "loss": 1.762,
+      "step": 3913
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015591518488150617,
+      "loss": 1.72,
+      "step": 3914
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015582342984358916,
+      "loss": 1.6285,
+      "step": 3915
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015573168457959366,
+      "loss": 1.7916,
+      "step": 3916
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015563994910981795,
+      "loss": 1.6249,
+      "step": 3917
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001555482234545582,
+      "loss": 1.7417,
+      "step": 3918
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001554565076341084,
+      "loss": 1.5843,
+      "step": 3919
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015536480166876044,
+      "loss": 1.5469,
+      "step": 3920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015527310557880383,
+      "loss": 1.703,
+      "step": 3921
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015518141938452624,
+      "loss": 1.6534,
+      "step": 3922
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015508974310621279,
+      "loss": 1.5231,
+      "step": 3923
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015499807676414658,
+      "loss": 1.637,
+      "step": 3924
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001549064203786085,
+      "loss": 1.8051,
+      "step": 3925
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001548147739698772,
+      "loss": 1.6812,
+      "step": 3926
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015472313755822921,
+      "loss": 1.748,
+      "step": 3927
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015463151116393864,
+      "loss": 1.7377,
+      "step": 3928
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001545398948072777,
+      "loss": 1.5566,
+      "step": 3929
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015444828850851619,
+      "loss": 1.6635,
+      "step": 3930
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015435669228792167,
+      "loss": 1.7338,
+      "step": 3931
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015426510616575945,
+      "loss": 1.6605,
+      "step": 3932
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015417353016229272,
+      "loss": 1.7191,
+      "step": 3933
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015408196429778233,
+      "loss": 1.6188,
+      "step": 3934
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015399040859248687,
+      "loss": 1.5223,
+      "step": 3935
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001538988630666629,
+      "loss": 1.8333,
+      "step": 3936
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001538073277405645,
+      "loss": 1.7444,
+      "step": 3937
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015371580263444357,
+      "loss": 1.7499,
+      "step": 3938
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015362428776854973,
+      "loss": 1.6976,
+      "step": 3939
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001535327831631303,
+      "loss": 1.742,
+      "step": 3940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001534412888384304,
+      "loss": 1.7179,
+      "step": 3941
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015334980481469275,
+      "loss": 1.7309,
+      "step": 3942
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015325833111215812,
+      "loss": 1.6823,
+      "step": 3943
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015316686775106462,
+      "loss": 1.829,
+      "step": 3944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015307541475164828,
+      "loss": 1.5933,
+      "step": 3945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001529839721341427,
+      "loss": 1.6096,
+      "step": 3946
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015289253991877932,
+      "loss": 1.7923,
+      "step": 3947
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015280111812578724,
+      "loss": 1.741,
+      "step": 3948
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015270970677539307,
+      "loss": 1.7607,
+      "step": 3949
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015261830588782146,
+      "loss": 1.7591,
+      "step": 3950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001525269154832945,
+      "loss": 1.6634,
+      "step": 3951
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000152435535582032,
+      "loss": 1.6827,
+      "step": 3952
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015234416620425148,
+      "loss": 1.611,
+      "step": 3953
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001522528073701681,
+      "loss": 1.7012,
+      "step": 3954
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015216145909999472,
+      "loss": 1.761,
+      "step": 3955
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015207012141394178,
+      "loss": 1.65,
+      "step": 3956
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015197879433221756,
+      "loss": 1.6352,
+      "step": 3957
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001518874778750278,
+      "loss": 1.7081,
+      "step": 3958
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015179617206257596,
+      "loss": 1.7408,
+      "step": 3959
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001517048769150632,
+      "loss": 1.5687,
+      "step": 3960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015161359245268824,
+      "loss": 1.6746,
+      "step": 3961
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001515223186956475,
+      "loss": 1.7347,
+      "step": 3962
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015143105566413492,
+      "loss": 1.6771,
+      "step": 3963
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015133980337834224,
+      "loss": 1.7928,
+      "step": 3964
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001512485618584587,
+      "loss": 1.7139,
+      "step": 3965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015115733112467116,
+      "loss": 1.6153,
+      "step": 3966
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015106611119716415,
+      "loss": 1.6403,
+      "step": 3967
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001509749020961198,
+      "loss": 1.7294,
+      "step": 3968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015088370384171778,
+      "loss": 1.5862,
+      "step": 3969
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015079251645413536,
+      "loss": 1.6618,
+      "step": 3970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015070133995354756,
+      "loss": 1.7447,
+      "step": 3971
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015061017436012684,
+      "loss": 1.6517,
+      "step": 3972
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015051901969404327,
+      "loss": 1.7993,
+      "step": 3973
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015042787597546454,
+      "loss": 1.6079,
+      "step": 3974
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015033674322455592,
+      "loss": 1.6019,
+      "step": 3975
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015024562146148017,
+      "loss": 1.678,
+      "step": 3976
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001501545107063977,
+      "loss": 1.7168,
+      "step": 3977
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001500634109794665,
+      "loss": 1.7463,
+      "step": 3978
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001499723223008421,
+      "loss": 1.7612,
+      "step": 3979
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014988124469067753,
+      "loss": 1.5439,
+      "step": 3980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014979017816912343,
+      "loss": 1.7452,
+      "step": 3981
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014969912275632793,
+      "loss": 1.6816,
+      "step": 3982
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014960807847243682,
+      "loss": 1.5676,
+      "step": 3983
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014951704533759324,
+      "loss": 1.709,
+      "step": 3984
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001494260233719381,
+      "loss": 1.7048,
+      "step": 3985
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014933501259560962,
+      "loss": 1.5792,
+      "step": 3986
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001492440130287437,
+      "loss": 1.7306,
+      "step": 3987
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014915302469147362,
+      "loss": 1.6388,
+      "step": 3988
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014906204760393034,
+      "loss": 1.5687,
+      "step": 3989
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001489710817862422,
+      "loss": 1.8389,
+      "step": 3990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014888012725853498,
+      "loss": 1.6817,
+      "step": 3991
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014878918404093226,
+      "loss": 1.7476,
+      "step": 3992
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014869825215355488,
+      "loss": 1.6785,
+      "step": 3993
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014860733161652117,
+      "loss": 1.6818,
+      "step": 3994
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014851642244994703,
+      "loss": 1.4718,
+      "step": 3995
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014842552467394577,
+      "loss": 1.4172,
+      "step": 3996
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014833463830862828,
+      "loss": 1.6319,
+      "step": 3997
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014824376337410283,
+      "loss": 1.6955,
+      "step": 3998
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014815289989047518,
+      "loss": 1.6735,
+      "step": 3999
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014806204787784868,
+      "loss": 1.8226,
+      "step": 4000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014797120735632398,
+      "loss": 1.5954,
+      "step": 4001
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001478803783459993,
+      "loss": 1.594,
+      "step": 4002
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014778956086697017,
+      "loss": 1.739,
+      "step": 4003
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001476987549393297,
+      "loss": 1.6224,
+      "step": 4004
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014760796058316839,
+      "loss": 1.7411,
+      "step": 4005
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014751717781857413,
+      "loss": 1.5782,
+      "step": 4006
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014742640666563246,
+      "loss": 1.9473,
+      "step": 4007
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014733564714442617,
+      "loss": 1.4968,
+      "step": 4008
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014724489927503542,
+      "loss": 1.601,
+      "step": 4009
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014715416307753797,
+      "loss": 1.7732,
+      "step": 4010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014706343857200883,
+      "loss": 1.6121,
+      "step": 4011
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014697272577852055,
+      "loss": 1.6525,
+      "step": 4012
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014688202471714292,
+      "loss": 1.6828,
+      "step": 4013
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001467913354079434,
+      "loss": 1.5331,
+      "step": 4014
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001467006578709867,
+      "loss": 1.5904,
+      "step": 4015
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014660999212633485,
+      "loss": 1.7232,
+      "step": 4016
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014651933819404743,
+      "loss": 1.6127,
+      "step": 4017
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014642869609418123,
+      "loss": 1.8082,
+      "step": 4018
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014633806584679062,
+      "loss": 1.5863,
+      "step": 4019
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014624744747192704,
+      "loss": 1.8597,
+      "step": 4020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014615684098963979,
+      "loss": 1.613,
+      "step": 4021
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001460662464199751,
+      "loss": 1.585,
+      "step": 4022
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014597566378297673,
+      "loss": 1.6306,
+      "step": 4023
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014588509309868583,
+      "loss": 1.6741,
+      "step": 4024
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014579453438714083,
+      "loss": 1.8019,
+      "step": 4025
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001457039876683776,
+      "loss": 1.631,
+      "step": 4026
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001456134529624292,
+      "loss": 1.6822,
+      "step": 4027
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014552293028932627,
+      "loss": 1.6158,
+      "step": 4028
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014543241966909657,
+      "loss": 1.7164,
+      "step": 4029
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001453419211217653,
+      "loss": 1.6281,
+      "step": 4030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014525143466735497,
+      "loss": 1.6396,
+      "step": 4031
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001451609603258854,
+      "loss": 1.6129,
+      "step": 4032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014507049811737378,
+      "loss": 1.6679,
+      "step": 4033
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001449800480618345,
+      "loss": 1.6472,
+      "step": 4034
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014488961017927943,
+      "loss": 1.7998,
+      "step": 4035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001447991844897176,
+      "loss": 1.8147,
+      "step": 4036
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014470877101315538,
+      "loss": 1.8575,
+      "step": 4037
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014461836976959654,
+      "loss": 1.6102,
+      "step": 4038
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014452798077904197,
+      "loss": 1.625,
+      "step": 4039
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014443760406148996,
+      "loss": 1.7526,
+      "step": 4040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014434723963693607,
+      "loss": 1.5441,
+      "step": 4041
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014425688752537315,
+      "loss": 1.6391,
+      "step": 4042
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014416654774679133,
+      "loss": 1.6833,
+      "step": 4043
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014407622032117792,
+      "loss": 1.6702,
+      "step": 4044
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014398590526851766,
+      "loss": 1.567,
+      "step": 4045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014389560260879235,
+      "loss": 1.6884,
+      "step": 4046
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014380531236198122,
+      "loss": 1.6138,
+      "step": 4047
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014371503454806066,
+      "loss": 1.5257,
+      "step": 4048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001436247691870044,
+      "loss": 1.7083,
+      "step": 4049
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001435345162987833,
+      "loss": 1.7437,
+      "step": 4050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014344427590336556,
+      "loss": 1.9665,
+      "step": 4051
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014335404802071655,
+      "loss": 1.5301,
+      "step": 4052
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014326383267079885,
+      "loss": 1.7462,
+      "step": 4053
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001431736298735724,
+      "loss": 1.7714,
+      "step": 4054
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014308343964899414,
+      "loss": 1.5211,
+      "step": 4055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001429932620170185,
+      "loss": 1.8166,
+      "step": 4056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014290309699759694,
+      "loss": 1.6243,
+      "step": 4057
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014281294461067817,
+      "loss": 1.7519,
+      "step": 4058
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001427228048762081,
+      "loss": 1.7504,
+      "step": 4059
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014263267781412987,
+      "loss": 1.7038,
+      "step": 4060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001425425634443838,
+      "loss": 1.7446,
+      "step": 4061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014245246178690727,
+      "loss": 1.7653,
+      "step": 4062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001423623728616352,
+      "loss": 1.7723,
+      "step": 4063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014227229668849938,
+      "loss": 1.6568,
+      "step": 4064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014218223328742885,
+      "loss": 1.5526,
+      "step": 4065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001420921826783498,
+      "loss": 1.5646,
+      "step": 4066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014200214488118565,
+      "loss": 1.6087,
+      "step": 4067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014191211991585704,
+      "loss": 1.668,
+      "step": 4068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014182210780228156,
+      "loss": 1.5887,
+      "step": 4069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014173210856037428,
+      "loss": 1.7061,
+      "step": 4070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001416421222100471,
+      "loss": 1.6819,
+      "step": 4071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014155214877120928,
+      "loss": 1.5917,
+      "step": 4072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014146218826376715,
+      "loss": 1.7417,
+      "step": 4073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014137224070762412,
+      "loss": 1.6556,
+      "step": 4074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014128230612268081,
+      "loss": 1.6784,
+      "step": 4075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014119238452883496,
+      "loss": 1.608,
+      "step": 4076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014110247594598135,
+      "loss": 1.6466,
+      "step": 4077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001410125803940121,
+      "loss": 1.7322,
+      "step": 4078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014092269789281627,
+      "loss": 1.7129,
+      "step": 4079
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014083282846228003,
+      "loss": 1.6496,
+      "step": 4080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014074297212228676,
+      "loss": 1.6766,
+      "step": 4081
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014065312889271678,
+      "loss": 1.8497,
+      "step": 4082
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014056329879344768,
+      "loss": 1.7482,
+      "step": 4083
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014047348184435395,
+      "loss": 1.7584,
+      "step": 4084
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001403836780653075,
+      "loss": 1.6763,
+      "step": 4085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014029388747617699,
+      "loss": 1.7161,
+      "step": 4086
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014020411009682836,
+      "loss": 1.7801,
+      "step": 4087
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014011434594712454,
+      "loss": 1.5082,
+      "step": 4088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014002459504692554,
+      "loss": 1.5808,
+      "step": 4089
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00013993485741608844,
+      "loss": 1.8052,
+      "step": 4090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00013984513307446737,
+      "loss": 1.6708,
+      "step": 4091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00013975542204191363,
+      "loss": 1.8493,
+      "step": 4092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00013966572433827547,
+      "loss": 1.7331,
+      "step": 4093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00013957603998339816,
+      "loss": 1.7174,
+      "step": 4094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00013948636899712414,
+      "loss": 1.7221,
+      "step": 4095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001393967113992928,
+      "loss": 1.6295,
+      "step": 4096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00013930706720974054,
+      "loss": 1.7681,
+      "step": 4097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00013921743644830086,
+      "loss": 1.6097,
+      "step": 4098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001391278191348043,
+      "loss": 1.7801,
+      "step": 4099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00013903821528907844,
+      "loss": 1.7029,
+      "step": 4100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00013894862493094777,
+      "loss": 1.7299,
+      "step": 4101
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013885904808023388,
+      "loss": 1.8923,
+      "step": 4102
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013876948475675534,
+      "loss": 1.6911,
+      "step": 4103
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013867993498032776,
+      "loss": 1.7039,
+      "step": 4104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013859039877076368,
+      "loss": 1.7673,
+      "step": 4105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013850087614787282,
+      "loss": 1.5532,
+      "step": 4106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001384113671314617,
+      "loss": 1.6937,
+      "step": 4107
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013832187174133383,
+      "loss": 1.8529,
+      "step": 4108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001382323899972899,
+      "loss": 1.7513,
+      "step": 4109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013814292191912734,
+      "loss": 1.4289,
+      "step": 4110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013805346752664073,
+      "loss": 1.6652,
+      "step": 4111
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013796402683962148,
+      "loss": 1.5702,
+      "step": 4112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001378745998778582,
+      "loss": 1.732,
+      "step": 4113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001377851866611362,
+      "loss": 1.6129,
+      "step": 4114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001376957872092379,
+      "loss": 1.6679,
+      "step": 4115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013760640154194257,
+      "loss": 1.7868,
+      "step": 4116
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001375170296790266,
+      "loss": 1.5779,
+      "step": 4117
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013742767164026315,
+      "loss": 1.5982,
+      "step": 4118
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001373383274454224,
+      "loss": 1.6694,
+      "step": 4119
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001372489971142715,
+      "loss": 1.7002,
+      "step": 4120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013715968066657446,
+      "loss": 1.6621,
+      "step": 4121
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013707037812209225,
+      "loss": 1.6866,
+      "step": 4122
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001369810895005828,
+      "loss": 1.6254,
+      "step": 4123
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001368918148218009,
+      "loss": 1.7647,
+      "step": 4124
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013680255410549827,
+      "loss": 1.7637,
+      "step": 4125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001367133073714235,
+      "loss": 1.7081,
+      "step": 4126
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013662407463932227,
+      "loss": 1.6076,
+      "step": 4127
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013653485592893697,
+      "loss": 1.762,
+      "step": 4128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013644565126000693,
+      "loss": 1.7541,
+      "step": 4129
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013635646065226839,
+      "loss": 1.6995,
+      "step": 4130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013626728412545448,
+      "loss": 1.6225,
+      "step": 4131
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013617812169929524,
+      "loss": 1.6962,
+      "step": 4132
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013608897339351743,
+      "loss": 1.6735,
+      "step": 4133
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013599983922784505,
+      "loss": 1.7076,
+      "step": 4134
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013591071922199865,
+      "loss": 1.6231,
+      "step": 4135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001358216133956957,
+      "loss": 1.7813,
+      "step": 4136
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001357325217686506,
+      "loss": 1.8397,
+      "step": 4137
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013564344436057453,
+      "loss": 1.8431,
+      "step": 4138
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013555438119117565,
+      "loss": 1.5871,
+      "step": 4139
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013546533228015877,
+      "loss": 1.5993,
+      "step": 4140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013537629764722584,
+      "loss": 1.7251,
+      "step": 4141
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001352872773120754,
+      "loss": 1.8669,
+      "step": 4142
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013519827129440293,
+      "loss": 1.665,
+      "step": 4143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013510927961390073,
+      "loss": 1.7537,
+      "step": 4144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001350203022902578,
+      "loss": 1.8529,
+      "step": 4145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013493133934316023,
+      "loss": 1.5546,
+      "step": 4146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013484239079229063,
+      "loss": 1.5319,
+      "step": 4147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001347534566573287,
+      "loss": 1.7547,
+      "step": 4148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013466453695795084,
+      "loss": 1.6145,
+      "step": 4149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013457563171383018,
+      "loss": 1.5142,
+      "step": 4150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013448674094463674,
+      "loss": 1.8191,
+      "step": 4151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013439786467003733,
+      "loss": 1.5198,
+      "step": 4152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013430900290969542,
+      "loss": 1.7793,
+      "step": 4153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013422015568327153,
+      "loss": 1.6401,
+      "step": 4154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013413132301042264,
+      "loss": 1.601,
+      "step": 4155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001340425049108029,
+      "loss": 1.6082,
+      "step": 4156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013395370140406292,
+      "loss": 1.6804,
+      "step": 4157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001338649125098502,
+      "loss": 1.6311,
+      "step": 4158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013377613824780898,
+      "loss": 1.6219,
+      "step": 4159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013368737863758026,
+      "loss": 1.5817,
+      "step": 4160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001335986336988019,
+      "loss": 1.7396,
+      "step": 4161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013350990345110822,
+      "loss": 1.6014,
+      "step": 4162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013342118791413073,
+      "loss": 1.6561,
+      "step": 4163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013333248710749733,
+      "loss": 1.8762,
+      "step": 4164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013324380105083283,
+      "loss": 1.7378,
+      "step": 4165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001331551297637587,
+      "loss": 1.794,
+      "step": 4166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013306647326589318,
+      "loss": 1.6116,
+      "step": 4167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013297783157685124,
+      "loss": 1.6169,
+      "step": 4168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013288920471624452,
+      "loss": 1.6949,
+      "step": 4169
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001328005927036815,
+      "loss": 1.7108,
+      "step": 4170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013271199555876724,
+      "loss": 1.5767,
+      "step": 4171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013262341330110358,
+      "loss": 1.6459,
+      "step": 4172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013253484595028906,
+      "loss": 1.6956,
+      "step": 4173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013244629352591894,
+      "loss": 1.8315,
+      "step": 4174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001323577560475851,
+      "loss": 1.7702,
+      "step": 4175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013226923353487616,
+      "loss": 1.7114,
+      "step": 4176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013218072600737753,
+      "loss": 1.819,
+      "step": 4177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013209223348467113,
+      "loss": 1.6313,
+      "step": 4178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013200375598633566,
+      "loss": 1.8114,
+      "step": 4179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013191529353194647,
+      "loss": 1.6319,
+      "step": 4180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013182684614107562,
+      "loss": 1.7298,
+      "step": 4181
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013173841383329178,
+      "loss": 1.6924,
+      "step": 4182
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013164999662816027,
+      "loss": 1.5773,
+      "step": 4183
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013156159454524318,
+      "loss": 1.5629,
+      "step": 4184
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013147320760409915,
+      "loss": 1.7238,
+      "step": 4185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013138483582428352,
+      "loss": 1.7594,
+      "step": 4186
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001312964792253482,
+      "loss": 1.5976,
+      "step": 4187
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013120813782684185,
+      "loss": 1.6607,
+      "step": 4188
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013111981164830972,
+      "loss": 1.5207,
+      "step": 4189
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013103150070929362,
+      "loss": 1.4984,
+      "step": 4190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013094320502933212,
+      "loss": 1.6689,
+      "step": 4191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013085492462796037,
+      "loss": 1.6474,
+      "step": 4192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001307666595247101,
+      "loss": 1.6981,
+      "step": 4193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001306784097391096,
+      "loss": 1.9185,
+      "step": 4194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013059017529068397,
+      "loss": 1.7889,
+      "step": 4195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013050195619895467,
+      "loss": 1.7033,
+      "step": 4196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001304137524834399,
+      "loss": 1.6185,
+      "step": 4197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013032556416365458,
+      "loss": 1.7319,
+      "step": 4198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013023739125910997,
+      "loss": 1.7319,
+      "step": 4199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013014923378931406,
+      "loss": 1.6448,
+      "step": 4200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001300610917737714,
+      "loss": 1.7356,
+      "step": 4201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012997296523198314,
+      "loss": 1.7441,
+      "step": 4202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012988485418344693,
+      "loss": 1.7487,
+      "step": 4203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012979675864765704,
+      "loss": 1.6347,
+      "step": 4204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012970867864410445,
+      "loss": 1.6673,
+      "step": 4205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012962061419227654,
+      "loss": 1.7471,
+      "step": 4206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012953256531165715,
+      "loss": 1.6789,
+      "step": 4207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012944453202172696,
+      "loss": 1.7428,
+      "step": 4208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012935651434196291,
+      "loss": 1.7558,
+      "step": 4209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012926851229183876,
+      "loss": 1.7631,
+      "step": 4210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012918052589082447,
+      "loss": 1.7141,
+      "step": 4211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012909255515838704,
+      "loss": 1.5404,
+      "step": 4212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012900460011398948,
+      "loss": 1.7552,
+      "step": 4213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012891666077709162,
+      "loss": 1.8362,
+      "step": 4214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012882873716714983,
+      "loss": 1.761,
+      "step": 4215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001287408293036168,
+      "loss": 1.5561,
+      "step": 4216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012865293720594188,
+      "loss": 1.6907,
+      "step": 4217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012856506089357087,
+      "loss": 1.7334,
+      "step": 4218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012847720038594622,
+      "loss": 1.5092,
+      "step": 4219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012838935570250678,
+      "loss": 1.8634,
+      "step": 4220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012830152686268783,
+      "loss": 1.6322,
+      "step": 4221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012821371388592123,
+      "loss": 1.7091,
+      "step": 4222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012812591679163538,
+      "loss": 1.5219,
+      "step": 4223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012803813559925497,
+      "loss": 1.7601,
+      "step": 4224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001279503703282013,
+      "loss": 1.6105,
+      "step": 4225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001278626209978923,
+      "loss": 1.7529,
+      "step": 4226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012777488762774215,
+      "loss": 1.6255,
+      "step": 4227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012768717023716158,
+      "loss": 1.6465,
+      "step": 4228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012759946884555775,
+      "loss": 1.6793,
+      "step": 4229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001275117834723343,
+      "loss": 1.5892,
+      "step": 4230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001274241141368914,
+      "loss": 1.8042,
+      "step": 4231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012733646085862551,
+      "loss": 1.5212,
+      "step": 4232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012724882365692958,
+      "loss": 1.5418,
+      "step": 4233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012716120255119325,
+      "loss": 1.6734,
+      "step": 4234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012707359756080225,
+      "loss": 1.6402,
+      "step": 4235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012698600870513898,
+      "loss": 1.825,
+      "step": 4236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012689843600358213,
+      "loss": 1.6438,
+      "step": 4237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012681087947550692,
+      "loss": 1.6553,
+      "step": 4238
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001267233391402849,
+      "loss": 1.6824,
+      "step": 4239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012663581501728405,
+      "loss": 1.5906,
+      "step": 4240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012654830712586888,
+      "loss": 1.6572,
+      "step": 4241
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012646081548540024,
+      "loss": 1.5399,
+      "step": 4242
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012637334011523528,
+      "loss": 1.7351,
+      "step": 4243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001262858810347277,
+      "loss": 1.5226,
+      "step": 4244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012619843826322752,
+      "loss": 1.6895,
+      "step": 4245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012611101182008114,
+      "loss": 1.8536,
+      "step": 4246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012602360172463137,
+      "loss": 1.6322,
+      "step": 4247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012593620799621747,
+      "loss": 1.5943,
+      "step": 4248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012584883065417497,
+      "loss": 1.6395,
+      "step": 4249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012576146971783584,
+      "loss": 1.8766,
+      "step": 4250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012567412520652838,
+      "loss": 1.7756,
+      "step": 4251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012558679713957731,
+      "loss": 1.7544,
+      "step": 4252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012549948553630365,
+      "loss": 1.613,
+      "step": 4253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012541219041602476,
+      "loss": 1.5076,
+      "step": 4254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012532491179805447,
+      "loss": 1.5072,
+      "step": 4255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012523764970170294,
+      "loss": 1.7121,
+      "step": 4256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001251504041462765,
+      "loss": 1.8431,
+      "step": 4257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000125063175151078,
+      "loss": 1.6695,
+      "step": 4258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001249759627354066,
+      "loss": 1.7373,
+      "step": 4259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001248887669185577,
+      "loss": 1.7635,
+      "step": 4260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001248015877198231,
+      "loss": 1.6944,
+      "step": 4261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012471442515849093,
+      "loss": 1.5992,
+      "step": 4262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012462727925384564,
+      "loss": 1.6138,
+      "step": 4263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012454015002516799,
+      "loss": 1.6177,
+      "step": 4264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012445303749173499,
+      "loss": 1.5014,
+      "step": 4265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012436594167282,
+      "loss": 1.6922,
+      "step": 4266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012427886258769267,
+      "loss": 1.6909,
+      "step": 4267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012419180025561896,
+      "loss": 1.5111,
+      "step": 4268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012410475469586127,
+      "loss": 1.6402,
+      "step": 4269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012401772592767797,
+      "loss": 1.7014,
+      "step": 4270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012393071397032398,
+      "loss": 1.6495,
+      "step": 4271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001238437188430504,
+      "loss": 1.714,
+      "step": 4272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012375674056510456,
+      "loss": 1.6346,
+      "step": 4273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001236697791557302,
+      "loss": 1.5149,
+      "step": 4274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001235828346341671,
+      "loss": 1.738,
+      "step": 4275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012349590701965163,
+      "loss": 1.5509,
+      "step": 4276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012340899633141622,
+      "loss": 1.7182,
+      "step": 4277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012332210258868948,
+      "loss": 1.746,
+      "step": 4278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012323522581069644,
+      "loss": 1.5621,
+      "step": 4279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012314836601665823,
+      "loss": 1.6669,
+      "step": 4280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012306152322579236,
+      "loss": 1.686,
+      "step": 4281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001229746974573124,
+      "loss": 1.6268,
+      "step": 4282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012288788873042843,
+      "loss": 1.6603,
+      "step": 4283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012280109706434654,
+      "loss": 1.7132,
+      "step": 4284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012271432247826908,
+      "loss": 1.5388,
+      "step": 4285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001226275649913947,
+      "loss": 1.625,
+      "step": 4286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001225408246229181,
+      "loss": 1.6606,
+      "step": 4287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012245410139203035,
+      "loss": 1.7649,
+      "step": 4288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012236739531791863,
+      "loss": 1.6419,
+      "step": 4289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001222807064197665,
+      "loss": 1.778,
+      "step": 4290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012219403471675356,
+      "loss": 1.6385,
+      "step": 4291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001221073802280556,
+      "loss": 1.6514,
+      "step": 4292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012202074297284465,
+      "loss": 1.6765,
+      "step": 4293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012193412297028894,
+      "loss": 1.824,
+      "step": 4294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012184752023955281,
+      "loss": 1.7888,
+      "step": 4295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012176093479979683,
+      "loss": 1.4818,
+      "step": 4296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001216743666701778,
+      "loss": 1.5906,
+      "step": 4297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012158781586984864,
+      "loss": 1.677,
+      "step": 4298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012150128241795842,
+      "loss": 1.6538,
+      "step": 4299
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012141476633365237,
+      "loss": 1.6752,
+      "step": 4300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012132826763607186,
+      "loss": 1.6352,
+      "step": 4301
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012124178634435449,
+      "loss": 1.5177,
+      "step": 4302
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012115532247763389,
+      "loss": 1.7718,
+      "step": 4303
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012106887605504002,
+      "loss": 1.6457,
+      "step": 4304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012098244709569879,
+      "loss": 1.5962,
+      "step": 4305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012089603561873237,
+      "loss": 1.7569,
+      "step": 4306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012080964164325897,
+      "loss": 1.5579,
+      "step": 4307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000120723265188393,
+      "loss": 1.622,
+      "step": 4308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012063690627324498,
+      "loss": 1.5637,
+      "step": 4309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001205505649169215,
+      "loss": 1.7365,
+      "step": 4310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012046424113852528,
+      "loss": 1.6649,
+      "step": 4311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012037793495715526,
+      "loss": 1.7492,
+      "step": 4312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012029164639190638,
+      "loss": 1.6706,
+      "step": 4313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012020537546186964,
+      "loss": 1.5708,
+      "step": 4314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012011912218613224,
+      "loss": 1.8715,
+      "step": 4315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012003288658377747,
+      "loss": 1.6348,
+      "step": 4316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001199466686738846,
+      "loss": 1.5851,
+      "step": 4317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011986046847552908,
+      "loss": 1.8108,
+      "step": 4318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011977428600778246,
+      "loss": 1.7242,
+      "step": 4319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011968812128971232,
+      "loss": 1.6364,
+      "step": 4320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011960197434038231,
+      "loss": 1.7087,
+      "step": 4321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011951584517885215,
+      "loss": 1.6028,
+      "step": 4322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011942973382417766,
+      "loss": 1.6921,
+      "step": 4323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011934364029541067,
+      "loss": 1.6376,
+      "step": 4324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011925756461159907,
+      "loss": 1.7816,
+      "step": 4325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001191715067917869,
+      "loss": 1.7306,
+      "step": 4326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001190854668550141,
+      "loss": 1.6005,
+      "step": 4327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011899944482031675,
+      "loss": 1.7739,
+      "step": 4328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011891344070672699,
+      "loss": 1.6932,
+      "step": 4329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011882745453327287,
+      "loss": 1.6464,
+      "step": 4330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011874148631897857,
+      "loss": 1.8473,
+      "step": 4331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011865553608286424,
+      "loss": 1.6861,
+      "step": 4332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001185696038439462,
+      "loss": 1.6709,
+      "step": 4333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011848368962123659,
+      "loss": 1.5742,
+      "step": 4334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011839779343374369,
+      "loss": 1.5893,
+      "step": 4335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011831191530047174,
+      "loss": 1.6173,
+      "step": 4336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011822605524042096,
+      "loss": 1.7142,
+      "step": 4337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011814021327258764,
+      "loss": 1.7167,
+      "step": 4338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011805438941596398,
+      "loss": 1.5434,
+      "step": 4339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011796858368953836,
+      "loss": 1.7036,
+      "step": 4340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001178827961122949,
+      "loss": 1.8202,
+      "step": 4341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011779702670321387,
+      "loss": 1.7685,
+      "step": 4342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011771127548127143,
+      "loss": 1.8388,
+      "step": 4343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001176255424654398,
+      "loss": 1.6172,
+      "step": 4344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001175398276746871,
+      "loss": 1.4126,
+      "step": 4345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011745413112797736,
+      "loss": 1.6854,
+      "step": 4346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011736845284427085,
+      "loss": 1.7603,
+      "step": 4347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011728279284252357,
+      "loss": 1.5987,
+      "step": 4348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001171971511416874,
+      "loss": 1.7076,
+      "step": 4349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011711152776071038,
+      "loss": 1.6585,
+      "step": 4350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011702592271853632,
+      "loss": 1.6975,
+      "step": 4351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011694033603410515,
+      "loss": 1.6042,
+      "step": 4352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011685476772635249,
+      "loss": 1.4995,
+      "step": 4353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011676921781421029,
+      "loss": 1.7877,
+      "step": 4354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011668368631660596,
+      "loss": 1.5401,
+      "step": 4355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011659817325246325,
+      "loss": 1.8078,
+      "step": 4356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011651267864070145,
+      "loss": 1.882,
+      "step": 4357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011642720250023617,
+      "loss": 1.6159,
+      "step": 4358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011634174484997853,
+      "loss": 1.7666,
+      "step": 4359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011625630570883586,
+      "loss": 1.7064,
+      "step": 4360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011617088509571134,
+      "loss": 1.4699,
+      "step": 4361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011608548302950388,
+      "loss": 1.6124,
+      "step": 4362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011600009952910853,
+      "loss": 1.7564,
+      "step": 4363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011591473461341596,
+      "loss": 1.6289,
+      "step": 4364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011582938830131307,
+      "loss": 1.5874,
+      "step": 4365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011574406061168224,
+      "loss": 1.634,
+      "step": 4366
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011565875156340204,
+      "loss": 1.795,
+      "step": 4367
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011557346117534693,
+      "loss": 1.586,
+      "step": 4368
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011548818946638693,
+      "loss": 1.6238,
+      "step": 4369
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011540293645538831,
+      "loss": 1.5597,
+      "step": 4370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011531770216121285,
+      "loss": 1.6447,
+      "step": 4371
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001152324866027185,
+      "loss": 1.803,
+      "step": 4372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011514728979875879,
+      "loss": 1.7431,
+      "step": 4373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011506211176818331,
+      "loss": 1.64,
+      "step": 4374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001149769525298375,
+      "loss": 1.8091,
+      "step": 4375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011489181210256238,
+      "loss": 1.7608,
+      "step": 4376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011480669050519518,
+      "loss": 1.7336,
+      "step": 4377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011472158775656861,
+      "loss": 1.5913,
+      "step": 4378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001146365038755115,
+      "loss": 1.6189,
+      "step": 4379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011455143888084827,
+      "loss": 1.5632,
+      "step": 4380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011446639279139933,
+      "loss": 1.5732,
+      "step": 4381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011438136562598085,
+      "loss": 1.5454,
+      "step": 4382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011429635740340488,
+      "loss": 1.6969,
+      "step": 4383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011421136814247914,
+      "loss": 1.5212,
+      "step": 4384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011412639786200716,
+      "loss": 1.5673,
+      "step": 4385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011404144658078848,
+      "loss": 1.8013,
+      "step": 4386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011395651431761813,
+      "loss": 1.707,
+      "step": 4387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011387160109128725,
+      "loss": 1.715,
+      "step": 4388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001137867069205824,
+      "loss": 1.5916,
+      "step": 4389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011370183182428644,
+      "loss": 1.5794,
+      "step": 4390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011361697582117745,
+      "loss": 1.588,
+      "step": 4391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011353213893002974,
+      "loss": 1.8062,
+      "step": 4392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011344732116961308,
+      "loss": 1.7273,
+      "step": 4393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011336252255869306,
+      "loss": 1.6616,
+      "step": 4394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011327774311603128,
+      "loss": 1.6194,
+      "step": 4395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001131929828603846,
+      "loss": 1.7909,
+      "step": 4396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011310824181050634,
+      "loss": 1.7705,
+      "step": 4397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011302351998514491,
+      "loss": 1.4525,
+      "step": 4398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011293881740304488,
+      "loss": 1.6983,
+      "step": 4399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011285413408294631,
+      "loss": 1.7375,
+      "step": 4400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011276947004358525,
+      "loss": 1.6217,
+      "step": 4401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011268482530369322,
+      "loss": 1.7079,
+      "step": 4402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011260019988199744,
+      "loss": 1.4938,
+      "step": 4403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011251559379722138,
+      "loss": 1.5518,
+      "step": 4404
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011243100706808358,
+      "loss": 1.6106,
+      "step": 4405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011234643971329874,
+      "loss": 1.7965,
+      "step": 4406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011226189175157695,
+      "loss": 1.7451,
+      "step": 4407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011217736320162435,
+      "loss": 1.6052,
+      "step": 4408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011209285408214241,
+      "loss": 1.5435,
+      "step": 4409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011200836441182861,
+      "loss": 1.5778,
+      "step": 4410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011192389420937605,
+      "loss": 1.6161,
+      "step": 4411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011183944349347335,
+      "loss": 1.5359,
+      "step": 4412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011175501228280513,
+      "loss": 1.6065,
+      "step": 4413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011167060059605136,
+      "loss": 1.802,
+      "step": 4414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011158620845188799,
+      "loss": 1.6032,
+      "step": 4415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011150183586898633,
+      "loss": 1.5934,
+      "step": 4416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011141748286601367,
+      "loss": 1.626,
+      "step": 4417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011133314946163285,
+      "loss": 1.6022,
+      "step": 4418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011124883567450224,
+      "loss": 1.6439,
+      "step": 4419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011116454152327615,
+      "loss": 1.4832,
+      "step": 4420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011108026702660424,
+      "loss": 1.5117,
+      "step": 4421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011099601220313207,
+      "loss": 1.7248,
+      "step": 4422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011091177707150062,
+      "loss": 1.5205,
+      "step": 4423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011082756165034672,
+      "loss": 1.524,
+      "step": 4424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011074336595830281,
+      "loss": 1.7069,
+      "step": 4425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001106591900139968,
+      "loss": 1.7443,
+      "step": 4426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011057503383605245,
+      "loss": 1.5425,
+      "step": 4427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011049089744308894,
+      "loss": 1.6333,
+      "step": 4428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011040678085372124,
+      "loss": 1.5434,
+      "step": 4429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011032268408655981,
+      "loss": 1.8235,
+      "step": 4430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011023860716021079,
+      "loss": 1.6625,
+      "step": 4431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011015455009327604,
+      "loss": 1.6882,
+      "step": 4432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011007051290435275,
+      "loss": 1.7099,
+      "step": 4433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010998649561203405,
+      "loss": 1.7015,
+      "step": 4434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010990249823490825,
+      "loss": 1.673,
+      "step": 4435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010981852079155974,
+      "loss": 1.5661,
+      "step": 4436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010973456330056806,
+      "loss": 1.6798,
+      "step": 4437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010965062578050864,
+      "loss": 1.4483,
+      "step": 4438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001095667082499524,
+      "loss": 1.7962,
+      "step": 4439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010948281072746572,
+      "loss": 1.6799,
+      "step": 4440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010939893323161076,
+      "loss": 1.6705,
+      "step": 4441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010931507578094504,
+      "loss": 1.6674,
+      "step": 4442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010923123839402185,
+      "loss": 1.6996,
+      "step": 4443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010914742108938983,
+      "loss": 1.5609,
+      "step": 4444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010906362388559333,
+      "loss": 1.6639,
+      "step": 4445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010897984680117228,
+      "loss": 1.6835,
+      "step": 4446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010889608985466193,
+      "loss": 1.8593,
+      "step": 4447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001088123530645934,
+      "loss": 1.4772,
+      "step": 4448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010872863644949305,
+      "loss": 1.7615,
+      "step": 4449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000108644940027883,
+      "loss": 1.6206,
+      "step": 4450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010856126381828072,
+      "loss": 1.5321,
+      "step": 4451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010847760783919931,
+      "loss": 1.5751,
+      "step": 4452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010839397210914746,
+      "loss": 1.5734,
+      "step": 4453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001083103566466293,
+      "loss": 1.6108,
+      "step": 4454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010822676147014443,
+      "loss": 1.6555,
+      "step": 4455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010814318659818794,
+      "loss": 1.4606,
+      "step": 4456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010805963204925064,
+      "loss": 1.7096,
+      "step": 4457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010797609784181857,
+      "loss": 1.4392,
+      "step": 4458
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010789258399437345,
+      "loss": 1.5363,
+      "step": 4459
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010780909052539244,
+      "loss": 1.6232,
+      "step": 4460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001077256174533483,
+      "loss": 1.5756,
+      "step": 4461
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000107642164796709,
+      "loss": 1.6727,
+      "step": 4462
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010755873257393836,
+      "loss": 1.8055,
+      "step": 4463
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010747532080349534,
+      "loss": 1.7275,
+      "step": 4464
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001073919295038345,
+      "loss": 1.6795,
+      "step": 4465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010730855869340602,
+      "loss": 1.6005,
+      "step": 4466
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010722520839065517,
+      "loss": 1.7689,
+      "step": 4467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010714187861402329,
+      "loss": 1.6126,
+      "step": 4468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010705856938194658,
+      "loss": 1.4976,
+      "step": 4469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010697528071285705,
+      "loss": 1.659,
+      "step": 4470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010689201262518194,
+      "loss": 1.7047,
+      "step": 4471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010680876513734415,
+      "loss": 1.6001,
+      "step": 4472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010672553826776184,
+      "loss": 1.788,
+      "step": 4473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010664233203484856,
+      "loss": 1.7297,
+      "step": 4474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001065591464570137,
+      "loss": 1.6742,
+      "step": 4475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001064759815526616,
+      "loss": 1.6807,
+      "step": 4476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010639283734019233,
+      "loss": 1.5177,
+      "step": 4477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010630971383800115,
+      "loss": 1.712,
+      "step": 4478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010622661106447901,
+      "loss": 1.6607,
+      "step": 4479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000106143529038012,
+      "loss": 1.588,
+      "step": 4480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010606046777698173,
+      "loss": 1.6795,
+      "step": 4481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010597742729976543,
+      "loss": 1.7618,
+      "step": 4482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010589440762473528,
+      "loss": 1.7646,
+      "step": 4483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010581140877025935,
+      "loss": 1.7318,
+      "step": 4484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010572843075470063,
+      "loss": 1.705,
+      "step": 4485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010564547359641794,
+      "loss": 1.6426,
+      "step": 4486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010556253731376511,
+      "loss": 1.6006,
+      "step": 4487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010547962192509162,
+      "loss": 1.8377,
+      "step": 4488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010539672744874227,
+      "loss": 1.6354,
+      "step": 4489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010531385390305705,
+      "loss": 1.6257,
+      "step": 4490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010523100130637162,
+      "loss": 1.6688,
+      "step": 4491
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010514816967701673,
+      "loss": 1.6757,
+      "step": 4492
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010506535903331867,
+      "loss": 1.6974,
+      "step": 4493
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010498256939359894,
+      "loss": 1.5584,
+      "step": 4494
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010489980077617452,
+      "loss": 1.627,
+      "step": 4495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001048170531993578,
+      "loss": 1.5523,
+      "step": 4496
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010473432668145622,
+      "loss": 1.7442,
+      "step": 4497
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010465162124077297,
+      "loss": 1.7411,
+      "step": 4498
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010456893689560612,
+      "loss": 1.6526,
+      "step": 4499
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010448627366424952,
+      "loss": 1.5144,
+      "step": 4500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010440363156499198,
+      "loss": 1.5913,
+      "step": 4501
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010432101061611785,
+      "loss": 1.6608,
+      "step": 4502
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010423841083590684,
+      "loss": 1.642,
+      "step": 4503
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010415583224263372,
+      "loss": 1.6624,
+      "step": 4504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010407327485456888,
+      "loss": 1.703,
+      "step": 4505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010399073868997776,
+      "loss": 1.7472,
+      "step": 4506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010390822376712128,
+      "loss": 1.68,
+      "step": 4507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010382573010425554,
+      "loss": 1.7673,
+      "step": 4508
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000103743257719632,
+      "loss": 1.6064,
+      "step": 4509
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010366080663149751,
+      "loss": 1.6831,
+      "step": 4510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010357837685809397,
+      "loss": 1.5525,
+      "step": 4511
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010349596841765883,
+      "loss": 1.6599,
+      "step": 4512
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010341358132842453,
+      "loss": 1.7227,
+      "step": 4513
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010333121560861907,
+      "loss": 1.5782,
+      "step": 4514
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010324887127646552,
+      "loss": 1.7197,
+      "step": 4515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010316654835018229,
+      "loss": 1.6851,
+      "step": 4516
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001030842468479832,
+      "loss": 1.6965,
+      "step": 4517
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010300196678807696,
+      "loss": 1.852,
+      "step": 4518
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010291970818866801,
+      "loss": 1.6845,
+      "step": 4519
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010283747106795556,
+      "loss": 1.6036,
+      "step": 4520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010275525544413449,
+      "loss": 1.6056,
+      "step": 4521
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010267306133539458,
+      "loss": 1.5741,
+      "step": 4522
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010259088875992109,
+      "loss": 1.5978,
+      "step": 4523
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010250873773589444,
+      "loss": 1.7078,
+      "step": 4524
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010242660828149033,
+      "loss": 1.7436,
+      "step": 4525
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001023445004148796,
+      "loss": 1.4804,
+      "step": 4526
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001022624141542282,
+      "loss": 1.5109,
+      "step": 4527
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010218034951769766,
+      "loss": 1.761,
+      "step": 4528
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010209830652344435,
+      "loss": 1.7827,
+      "step": 4529
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010201628518962003,
+      "loss": 1.657,
+      "step": 4530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010193428553437172,
+      "loss": 1.5641,
+      "step": 4531
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010185230757584165,
+      "loss": 1.5376,
+      "step": 4532
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000101770351332167,
+      "loss": 1.7988,
+      "step": 4533
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010168841682148045,
+      "loss": 1.8528,
+      "step": 4534
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001016065040619097,
+      "loss": 1.5964,
+      "step": 4535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001015246130715776,
+      "loss": 1.6253,
+      "step": 4536
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010144274386860236,
+      "loss": 1.6203,
+      "step": 4537
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010136089647109722,
+      "loss": 1.5488,
+      "step": 4538
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010127907089717075,
+      "loss": 1.6833,
+      "step": 4539
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010119726716492644,
+      "loss": 1.4571,
+      "step": 4540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010111548529246323,
+      "loss": 1.6406,
+      "step": 4541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010103372529787498,
+      "loss": 1.7868,
+      "step": 4542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010095198719925099,
+      "loss": 1.6533,
+      "step": 4543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010087027101467538,
+      "loss": 1.7638,
+      "step": 4544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010078857676222751,
+      "loss": 1.7476,
+      "step": 4545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010070690445998226,
+      "loss": 1.7864,
+      "step": 4546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010062525412600909,
+      "loss": 1.8074,
+      "step": 4547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010054362577837308,
+      "loss": 1.5664,
+      "step": 4548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010046201943513405,
+      "loss": 1.649,
+      "step": 4549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010038043511434727,
+      "loss": 1.5112,
+      "step": 4550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001002988728340629,
+      "loss": 1.674,
+      "step": 4551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010021733261232636,
+      "loss": 1.5735,
+      "step": 4552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010013581446717827,
+      "loss": 1.7092,
+      "step": 4553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010005431841665407,
+      "loss": 1.709,
+      "step": 4554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.997284447878465e-05,
+      "loss": 1.692,
+      "step": 4555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.989139267159573e-05,
+      "loss": 1.5629,
+      "step": 4556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.980996301310836e-05,
+      "loss": 1.7266,
+      "step": 4557
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.972855552133846e-05,
+      "loss": 1.544,
+      "step": 4558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.964717021429726e-05,
+      "loss": 1.6823,
+      "step": 4559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.956580710999104e-05,
+      "loss": 1.6369,
+      "step": 4560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.948446622642101e-05,
+      "loss": 1.6612,
+      "step": 4561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.940314758158367e-05,
+      "loss": 1.6415,
+      "step": 4562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.932185119347037e-05,
+      "loss": 1.692,
+      "step": 4563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.924057708006786e-05,
+      "loss": 1.8535,
+      "step": 4564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.915932525935758e-05,
+      "loss": 1.8264,
+      "step": 4565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.90780957493163e-05,
+      "loss": 1.7406,
+      "step": 4566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.899688856791589e-05,
+      "loss": 1.6612,
+      "step": 4567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.891570373312302e-05,
+      "loss": 1.6475,
+      "step": 4568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.883454126289968e-05,
+      "loss": 1.5344,
+      "step": 4569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.875340117520269e-05,
+      "loss": 1.8248,
+      "step": 4570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.867228348798417e-05,
+      "loss": 1.564,
+      "step": 4571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.859118821919098e-05,
+      "loss": 1.6471,
+      "step": 4572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.851011538676526e-05,
+      "loss": 1.7473,
+      "step": 4573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.842906500864422e-05,
+      "loss": 1.7324,
+      "step": 4574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.834803710275978e-05,
+      "loss": 1.7621,
+      "step": 4575
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.826703168703928e-05,
+      "loss": 1.6712,
+      "step": 4576
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.818604877940476e-05,
+      "loss": 1.5615,
+      "step": 4577
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.810508839777357e-05,
+      "loss": 1.6814,
+      "step": 4578
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.802415056005776e-05,
+      "loss": 1.7842,
+      "step": 4579
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.794323528416465e-05,
+      "loss": 1.638,
+      "step": 4580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.786234258799657e-05,
+      "loss": 1.465,
+      "step": 4581
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.77814724894506e-05,
+      "loss": 1.8147,
+      "step": 4582
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.770062500641913e-05,
+      "loss": 1.6759,
+      "step": 4583
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.761980015678924e-05,
+      "loss": 1.6459,
+      "step": 4584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.753899795844333e-05,
+      "loss": 1.6792,
+      "step": 4585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.745821842925849e-05,
+      "loss": 1.602,
+      "step": 4586
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.737746158710697e-05,
+      "loss": 1.6899,
+      "step": 4587
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.729672744985603e-05,
+      "loss": 1.7539,
+      "step": 4588
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.72160160353677e-05,
+      "loss": 1.7753,
+      "step": 4589
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.713532736149929e-05,
+      "loss": 1.4995,
+      "step": 4590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.705466144610269e-05,
+      "loss": 1.7893,
+      "step": 4591
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.697401830702518e-05,
+      "loss": 1.7913,
+      "step": 4592
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.689339796210862e-05,
+      "loss": 1.836,
+      "step": 4593
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.681280042919006e-05,
+      "loss": 1.8311,
+      "step": 4594
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.673222572610143e-05,
+      "loss": 1.7764,
+      "step": 4595
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.665167387066973e-05,
+      "loss": 1.5783,
+      "step": 4596
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.657114488071672e-05,
+      "loss": 1.5876,
+      "step": 4597
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.649063877405906e-05,
+      "loss": 1.7558,
+      "step": 4598
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.641015556850866e-05,
+      "loss": 1.5828,
+      "step": 4599
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.632969528187197e-05,
+      "loss": 1.6238,
+      "step": 4600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.624925793195065e-05,
+      "loss": 1.7721,
+      "step": 4601
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.616884353654123e-05,
+      "loss": 1.5871,
+      "step": 4602
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.60884521134352e-05,
+      "loss": 1.6516,
+      "step": 4603
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.600808368041872e-05,
+      "loss": 1.552,
+      "step": 4604
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.592773825527322e-05,
+      "loss": 1.4244,
+      "step": 4605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.584741585577478e-05,
+      "loss": 1.8131,
+      "step": 4606
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.576711649969437e-05,
+      "loss": 1.5201,
+      "step": 4607
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.568684020479806e-05,
+      "loss": 1.6394,
+      "step": 4608
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.560658698884671e-05,
+      "loss": 1.5614,
+      "step": 4609
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.552635686959616e-05,
+      "loss": 1.713,
+      "step": 4610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.544614986479692e-05,
+      "loss": 1.6388,
+      "step": 4611
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.536596599219465e-05,
+      "loss": 1.6394,
+      "step": 4612
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.528580526952964e-05,
+      "loss": 1.6607,
+      "step": 4613
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.520566771453734e-05,
+      "loss": 1.8648,
+      "step": 4614
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.512555334494777e-05,
+      "loss": 1.5349,
+      "step": 4615
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.504546217848602e-05,
+      "loss": 1.8281,
+      "step": 4616
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.496539423287212e-05,
+      "loss": 1.4792,
+      "step": 4617
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.488534952582064e-05,
+      "loss": 1.4949,
+      "step": 4618
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.480532807504139e-05,
+      "loss": 1.6189,
+      "step": 4619
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.47253298982387e-05,
+      "loss": 1.6139,
+      "step": 4620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.464535501311207e-05,
+      "loss": 1.6117,
+      "step": 4621
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.456540343735549e-05,
+      "loss": 1.5691,
+      "step": 4622
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.448547518865809e-05,
+      "loss": 1.6716,
+      "step": 4623
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.440557028470381e-05,
+      "loss": 1.4991,
+      "step": 4624
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.432568874317118e-05,
+      "loss": 1.6287,
+      "step": 4625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.424583058173389e-05,
+      "loss": 1.7102,
+      "step": 4626
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.416599581806016e-05,
+      "loss": 1.6533,
+      "step": 4627
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.408618446981328e-05,
+      "loss": 1.6851,
+      "step": 4628
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.400639655465116e-05,
+      "loss": 1.5565,
+      "step": 4629
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.392663209022663e-05,
+      "loss": 1.6132,
+      "step": 4630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.384689109418741e-05,
+      "loss": 1.7677,
+      "step": 4631
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.376717358417581e-05,
+      "loss": 1.5127,
+      "step": 4632
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.368747957782919e-05,
+      "loss": 1.7823,
+      "step": 4633
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.360780909277947e-05,
+      "loss": 1.6117,
+      "step": 4634
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.352816214665357e-05,
+      "loss": 1.5493,
+      "step": 4635
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.344853875707303e-05,
+      "loss": 1.6296,
+      "step": 4636
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.336893894165433e-05,
+      "loss": 1.5053,
+      "step": 4637
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.32893627180087e-05,
+      "loss": 1.7584,
+      "step": 4638
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.320981010374202e-05,
+      "loss": 1.6403,
+      "step": 4639
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.313028111645516e-05,
+      "loss": 1.7198,
+      "step": 4640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.30507757737435e-05,
+      "loss": 1.7474,
+      "step": 4641
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.297129409319754e-05,
+      "loss": 1.6246,
+      "step": 4642
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.28918360924021e-05,
+      "loss": 1.6647,
+      "step": 4643
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.281240178893718e-05,
+      "loss": 1.5764,
+      "step": 4644
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.273299120037736e-05,
+      "loss": 1.6376,
+      "step": 4645
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.265360434429182e-05,
+      "loss": 1.7476,
+      "step": 4646
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.257424123824483e-05,
+      "loss": 1.7686,
+      "step": 4647
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.249490189979507e-05,
+      "loss": 1.7504,
+      "step": 4648
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.241558634649623e-05,
+      "loss": 1.7232,
+      "step": 4649
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.233629459589648e-05,
+      "loss": 1.6603,
+      "step": 4650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.225702666553893e-05,
+      "loss": 1.7357,
+      "step": 4651
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.217778257296144e-05,
+      "loss": 1.7034,
+      "step": 4652
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.209856233569633e-05,
+      "loss": 1.6168,
+      "step": 4653
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.201936597127099e-05,
+      "loss": 1.7189,
+      "step": 4654
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.19401934972072e-05,
+      "loss": 1.6009,
+      "step": 4655
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.186104493102175e-05,
+      "loss": 1.6664,
+      "step": 4656
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.178192029022587e-05,
+      "loss": 1.6591,
+      "step": 4657
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.17028195923257e-05,
+      "loss": 1.7217,
+      "step": 4658
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.162374285482208e-05,
+      "loss": 1.5775,
+      "step": 4659
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.15446900952103e-05,
+      "loss": 1.5376,
+      "step": 4660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.146566133098073e-05,
+      "loss": 1.653,
+      "step": 4661
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.138665657961802e-05,
+      "loss": 1.755,
+      "step": 4662
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.130767585860191e-05,
+      "loss": 1.5822,
+      "step": 4663
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.122871918540647e-05,
+      "loss": 1.6938,
+      "step": 4664
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.114978657750063e-05,
+      "loss": 1.752,
+      "step": 4665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.107087805234804e-05,
+      "loss": 1.7439,
+      "step": 4666
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.099199362740698e-05,
+      "loss": 1.8376,
+      "step": 4667
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.091313332013031e-05,
+      "loss": 1.7121,
+      "step": 4668
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.083429714796554e-05,
+      "loss": 1.7488,
+      "step": 4669
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.075548512835508e-05,
+      "loss": 1.6638,
+      "step": 4670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.067669727873564e-05,
+      "loss": 1.7101,
+      "step": 4671
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.05979336165389e-05,
+      "loss": 1.6821,
+      "step": 4672
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.051919415919101e-05,
+      "loss": 1.7644,
+      "step": 4673
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.044047892411294e-05,
+      "loss": 1.6734,
+      "step": 4674
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.036178792872002e-05,
+      "loss": 1.7366,
+      "step": 4675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.028312119042249e-05,
+      "loss": 1.56,
+      "step": 4676
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.020447872662505e-05,
+      "loss": 1.7507,
+      "step": 4677
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.012586055472703e-05,
+      "loss": 1.5919,
+      "step": 4678
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.004726669212249e-05,
+      "loss": 1.696,
+      "step": 4679
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.996869715620011e-05,
+      "loss": 1.7675,
+      "step": 4680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.989015196434316e-05,
+      "loss": 1.5371,
+      "step": 4681
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.98116311339294e-05,
+      "loss": 1.7459,
+      "step": 4682
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.973313468233144e-05,
+      "loss": 1.6836,
+      "step": 4683
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.965466262691619e-05,
+      "loss": 1.6372,
+      "step": 4684
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.957621498504554e-05,
+      "loss": 1.647,
+      "step": 4685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.949779177407558e-05,
+      "loss": 1.7093,
+      "step": 4686
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.941939301135728e-05,
+      "loss": 1.7213,
+      "step": 4687
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.934101871423619e-05,
+      "loss": 1.4896,
+      "step": 4688
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.92626689000522e-05,
+      "loss": 1.7143,
+      "step": 4689
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.91843435861401e-05,
+      "loss": 1.579,
+      "step": 4690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.910604278982897e-05,
+      "loss": 1.6756,
+      "step": 4691
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.902776652844276e-05,
+      "loss": 1.5745,
+      "step": 4692
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.894951481929968e-05,
+      "loss": 1.515,
+      "step": 4693
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.887128767971272e-05,
+      "loss": 1.6098,
+      "step": 4694
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.879308512698947e-05,
+      "loss": 1.7656,
+      "step": 4695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.871490717843185e-05,
+      "loss": 1.6097,
+      "step": 4696
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.863675385133663e-05,
+      "loss": 1.8023,
+      "step": 4697
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.855862516299478e-05,
+      "loss": 1.6651,
+      "step": 4698
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.848052113069224e-05,
+      "loss": 1.6381,
+      "step": 4699
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.840244177170904e-05,
+      "loss": 1.6763,
+      "step": 4700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.832438710332009e-05,
+      "loss": 1.538,
+      "step": 4701
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.824635714279485e-05,
+      "loss": 1.6884,
+      "step": 4702
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.816835190739698e-05,
+      "loss": 1.6311,
+      "step": 4703
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.809037141438508e-05,
+      "loss": 1.5898,
+      "step": 4704
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.801241568101188e-05,
+      "loss": 1.6448,
+      "step": 4705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.7934484724525e-05,
+      "loss": 1.6892,
+      "step": 4706
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.785657856216631e-05,
+      "loss": 1.7692,
+      "step": 4707
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.777869721117229e-05,
+      "loss": 1.5418,
+      "step": 4708
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.770084068877407e-05,
+      "loss": 1.5951,
+      "step": 4709
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.762300901219695e-05,
+      "loss": 1.6539,
+      "step": 4710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.754520219866111e-05,
+      "loss": 1.7986,
+      "step": 4711
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.746742026538091e-05,
+      "loss": 1.3896,
+      "step": 4712
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.738966322956548e-05,
+      "loss": 1.5659,
+      "step": 4713
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.731193110841818e-05,
+      "loss": 1.7541,
+      "step": 4714
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.723422391913703e-05,
+      "loss": 1.5589,
+      "step": 4715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.71565416789146e-05,
+      "loss": 1.5456,
+      "step": 4716
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.707888440493765e-05,
+      "loss": 1.612,
+      "step": 4717
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.700125211438779e-05,
+      "loss": 1.6708,
+      "step": 4718
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.692364482444073e-05,
+      "loss": 1.5768,
+      "step": 4719
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.684606255226697e-05,
+      "loss": 1.4952,
+      "step": 4720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.67685053150312e-05,
+      "loss": 1.5089,
+      "step": 4721
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.669097312989279e-05,
+      "loss": 1.7466,
+      "step": 4722
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.661346601400553e-05,
+      "loss": 1.6806,
+      "step": 4723
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.653598398451748e-05,
+      "loss": 1.6799,
+      "step": 4724
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.645852705857145e-05,
+      "loss": 1.6844,
+      "step": 4725
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.638109525330436e-05,
+      "loss": 1.683,
+      "step": 4726
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.630368858584788e-05,
+      "loss": 1.637,
+      "step": 4727
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.622630707332786e-05,
+      "loss": 1.5674,
+      "step": 4728
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.614895073286478e-05,
+      "loss": 1.5352,
+      "step": 4729
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.607161958157354e-05,
+      "loss": 1.5525,
+      "step": 4730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.599431363656327e-05,
+      "loss": 1.6203,
+      "step": 4731
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.591703291493778e-05,
+      "loss": 1.6564,
+      "step": 4732
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.583977743379504e-05,
+      "loss": 1.6546,
+      "step": 4733
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.576254721022774e-05,
+      "loss": 1.7833,
+      "step": 4734
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.568534226132263e-05,
+      "loss": 1.7409,
+      "step": 4735
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.560816260416116e-05,
+      "loss": 1.6724,
+      "step": 4736
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.553100825581903e-05,
+      "loss": 1.6391,
+      "step": 4737
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.545387923336652e-05,
+      "loss": 1.7873,
+      "step": 4738
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.537677555386807e-05,
+      "loss": 1.6583,
+      "step": 4739
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.529969723438253e-05,
+      "loss": 1.4448,
+      "step": 4740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.52226442919634e-05,
+      "loss": 1.6529,
+      "step": 4741
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.514561674365824e-05,
+      "loss": 1.7536,
+      "step": 4742
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.50686146065092e-05,
+      "loss": 1.7237,
+      "step": 4743
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.499163789755276e-05,
+      "loss": 1.7751,
+      "step": 4744
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.491468663381982e-05,
+      "loss": 1.7782,
+      "step": 4745
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.48377608323355e-05,
+      "loss": 1.766,
+      "step": 4746
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.476086051011947e-05,
+      "loss": 1.7782,
+      "step": 4747
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.468398568418563e-05,
+      "loss": 1.6245,
+      "step": 4748
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.46071363715422e-05,
+      "loss": 1.4437,
+      "step": 4749
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.453031258919192e-05,
+      "loss": 1.6586,
+      "step": 4750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.44535143541318e-05,
+      "loss": 1.7163,
+      "step": 4751
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.437674168335325e-05,
+      "loss": 1.5617,
+      "step": 4752
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.429999459384186e-05,
+      "loss": 1.6075,
+      "step": 4753
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.422327310257781e-05,
+      "loss": 1.551,
+      "step": 4754
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.414657722653532e-05,
+      "loss": 1.6391,
+      "step": 4755
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.406990698268325e-05,
+      "loss": 1.5878,
+      "step": 4756
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.399326238798449e-05,
+      "loss": 1.5054,
+      "step": 4757
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.391664345939647e-05,
+      "loss": 1.586,
+      "step": 4758
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.384005021387101e-05,
+      "loss": 1.5379,
+      "step": 4759
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.376348266835387e-05,
+      "loss": 1.672,
+      "step": 4760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.368694083978561e-05,
+      "loss": 1.5876,
+      "step": 4761
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.361042474510066e-05,
+      "loss": 1.6347,
+      "step": 4762
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.353393440122809e-05,
+      "loss": 1.6439,
+      "step": 4763
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.345746982509103e-05,
+      "loss": 1.4473,
+      "step": 4764
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.33810310336071e-05,
+      "loss": 1.6031,
+      "step": 4765
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.330461804368814e-05,
+      "loss": 1.5143,
+      "step": 4766
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.322823087224021e-05,
+      "loss": 1.535,
+      "step": 4767
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.315186953616378e-05,
+      "loss": 1.6589,
+      "step": 4768
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.307553405235349e-05,
+      "loss": 1.7807,
+      "step": 4769
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.299922443769839e-05,
+      "loss": 1.5272,
+      "step": 4770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.292294070908163e-05,
+      "loss": 1.8175,
+      "step": 4771
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.284668288338076e-05,
+      "loss": 1.6637,
+      "step": 4772
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.27704509774677e-05,
+      "loss": 1.5464,
+      "step": 4773
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.269424500820832e-05,
+      "loss": 1.6815,
+      "step": 4774
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.261806499246314e-05,
+      "loss": 1.7232,
+      "step": 4775
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.254191094708654e-05,
+      "loss": 1.7899,
+      "step": 4776
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.246578288892752e-05,
+      "loss": 1.7495,
+      "step": 4777
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.238968083482903e-05,
+      "loss": 1.5863,
+      "step": 4778
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.231360480162847e-05,
+      "loss": 1.8252,
+      "step": 4779
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.223755480615747e-05,
+      "loss": 1.7638,
+      "step": 4780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.21615308652417e-05,
+      "loss": 1.682,
+      "step": 4781
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.208553299570136e-05,
+      "loss": 1.5501,
+      "step": 4782
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.200956121435057e-05,
+      "loss": 1.5843,
+      "step": 4783
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.1933615537998e-05,
+      "loss": 1.5631,
+      "step": 4784
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.185769598344626e-05,
+      "loss": 1.7378,
+      "step": 4785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.178180256749233e-05,
+      "loss": 1.7372,
+      "step": 4786
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.170593530692745e-05,
+      "loss": 1.8464,
+      "step": 4787
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.16300942185369e-05,
+      "loss": 1.722,
+      "step": 4788
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.15542793191004e-05,
+      "loss": 1.7003,
+      "step": 4789
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.147849062539159e-05,
+      "loss": 1.3903,
+      "step": 4790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.140272815417863e-05,
+      "loss": 1.3476,
+      "step": 4791
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.132699192222354e-05,
+      "loss": 1.7011,
+      "step": 4792
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.125128194628285e-05,
+      "loss": 1.605,
+      "step": 4793
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.117559824310715e-05,
+      "loss": 1.7047,
+      "step": 4794
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.10999408294411e-05,
+      "loss": 1.7484,
+      "step": 4795
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.102430972202381e-05,
+      "loss": 1.3614,
+      "step": 4796
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.094870493758825e-05,
+      "loss": 1.5753,
+      "step": 4797
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.087312649286189e-05,
+      "loss": 1.7056,
+      "step": 4798
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.079757440456606e-05,
+      "loss": 1.5949,
+      "step": 4799
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.072204868941646e-05,
+      "loss": 1.6708,
+      "step": 4800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.064654936412304e-05,
+      "loss": 1.7237,
+      "step": 4801
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.057107644538957e-05,
+      "loss": 1.6781,
+      "step": 4802
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.049562994991437e-05,
+      "loss": 1.5892,
+      "step": 4803
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.042020989438957e-05,
+      "loss": 1.5809,
+      "step": 4804
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.034481629550176e-05,
+      "loss": 1.7087,
+      "step": 4805
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.026944916993139e-05,
+      "loss": 1.6385,
+      "step": 4806
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.019410853435326e-05,
+      "loss": 1.5275,
+      "step": 4807
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.011879440543622e-05,
+      "loss": 1.4799,
+      "step": 4808
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.004350679984336e-05,
+      "loss": 1.6985,
+      "step": 4809
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.996824573423174e-05,
+      "loss": 1.6553,
+      "step": 4810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.989301122525254e-05,
+      "loss": 1.6774,
+      "step": 4811
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.981780328955133e-05,
+      "loss": 1.6843,
+      "step": 4812
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.974262194376745e-05,
+      "loss": 1.5098,
+      "step": 4813
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.966746720453459e-05,
+      "loss": 1.6799,
+      "step": 4814
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.959233908848051e-05,
+      "loss": 1.7975,
+      "step": 4815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.951723761222714e-05,
+      "loss": 1.6815,
+      "step": 4816
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.944216279239025e-05,
+      "loss": 1.6351,
+      "step": 4817
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.936711464558009e-05,
+      "loss": 1.9147,
+      "step": 4818
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.92920931884007e-05,
+      "loss": 1.7957,
+      "step": 4819
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.921709843745029e-05,
+      "loss": 1.4683,
+      "step": 4820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.914213040932126e-05,
+      "loss": 1.7066,
+      "step": 4821
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.906718912060007e-05,
+      "loss": 1.6327,
+      "step": 4822
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.899227458786726e-05,
+      "loss": 1.5879,
+      "step": 4823
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.891738682769732e-05,
+      "loss": 1.5991,
+      "step": 4824
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.884252585665903e-05,
+      "loss": 1.7104,
+      "step": 4825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.876769169131501e-05,
+      "loss": 1.6662,
+      "step": 4826
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.869288434822224e-05,
+      "loss": 1.6044,
+      "step": 4827
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.861810384393142e-05,
+      "loss": 1.6724,
+      "step": 4828
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.85433501949876e-05,
+      "loss": 1.6007,
+      "step": 4829
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.846862341792982e-05,
+      "loss": 1.5748,
+      "step": 4830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.839392352929098e-05,
+      "loss": 1.7409,
+      "step": 4831
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.831925054559837e-05,
+      "loss": 1.7379,
+      "step": 4832
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.824460448337299e-05,
+      "loss": 1.5805,
+      "step": 4833
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.816998535913016e-05,
+      "loss": 1.6855,
+      "step": 4834
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.809539318937897e-05,
+      "loss": 1.7002,
+      "step": 4835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.80208279906228e-05,
+      "loss": 1.6347,
+      "step": 4836
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.794628977935898e-05,
+      "loss": 1.6997,
+      "step": 4837
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.787177857207874e-05,
+      "loss": 1.6726,
+      "step": 4838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.779729438526759e-05,
+      "loss": 1.7049,
+      "step": 4839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.772283723540474e-05,
+      "loss": 1.6758,
+      "step": 4840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.764840713896374e-05,
+      "loss": 1.6391,
+      "step": 4841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.757400411241187e-05,
+      "loss": 1.7725,
+      "step": 4842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.749962817221062e-05,
+      "loss": 1.6171,
+      "step": 4843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.742527933481552e-05,
+      "loss": 1.6737,
+      "step": 4844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.735095761667581e-05,
+      "loss": 1.7048,
+      "step": 4845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.727666303423512e-05,
+      "loss": 1.7201,
+      "step": 4846
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.720239560393073e-05,
+      "loss": 1.6348,
+      "step": 4847
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.712815534219417e-05,
+      "loss": 1.8529,
+      "step": 4848
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.705394226545077e-05,
+      "loss": 1.6316,
+      "step": 4849
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.697975639011998e-05,
+      "loss": 1.611,
+      "step": 4850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.690559773261523e-05,
+      "loss": 1.5338,
+      "step": 4851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.683146630934377e-05,
+      "loss": 1.6098,
+      "step": 4852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.675736213670705e-05,
+      "loss": 1.5061,
+      "step": 4853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.668328523110027e-05,
+      "loss": 1.7394,
+      "step": 4854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.66092356089128e-05,
+      "loss": 1.6303,
+      "step": 4855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.653521328652775e-05,
+      "loss": 1.5544,
+      "step": 4856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.64612182803224e-05,
+      "loss": 1.6884,
+      "step": 4857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.638725060666796e-05,
+      "loss": 1.6255,
+      "step": 4858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.631331028192941e-05,
+      "loss": 1.7071,
+      "step": 4859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.623939732246592e-05,
+      "loss": 1.7119,
+      "step": 4860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.616551174463038e-05,
+      "loss": 1.5591,
+      "step": 4861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.609165356476986e-05,
+      "loss": 1.6119,
+      "step": 4862
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.601782279922509e-05,
+      "loss": 1.7885,
+      "step": 4863
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.594401946433096e-05,
+      "loss": 1.6723,
+      "step": 4864
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.58702435764163e-05,
+      "loss": 1.6808,
+      "step": 4865
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.579649515180362e-05,
+      "loss": 1.7456,
+      "step": 4866
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.572277420680968e-05,
+      "loss": 1.6805,
+      "step": 4867
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.564908075774486e-05,
+      "loss": 1.4072,
+      "step": 4868
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.557541482091371e-05,
+      "loss": 1.552,
+      "step": 4869
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.550177641261444e-05,
+      "loss": 1.6631,
+      "step": 4870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.542816554913939e-05,
+      "loss": 1.4785,
+      "step": 4871
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.535458224677478e-05,
+      "loss": 1.7449,
+      "step": 4872
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.528102652180058e-05,
+      "loss": 1.6819,
+      "step": 4873
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.520749839049081e-05,
+      "loss": 1.6942,
+      "step": 4874
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.513399786911325e-05,
+      "loss": 1.6517,
+      "step": 4875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.50605249739298e-05,
+      "loss": 1.6167,
+      "step": 4876
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.498707972119592e-05,
+      "loss": 1.5353,
+      "step": 4877
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.49136621271612e-05,
+      "loss": 1.564,
+      "step": 4878
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.484027220806907e-05,
+      "loss": 1.5592,
+      "step": 4879
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.476690998015686e-05,
+      "loss": 1.5412,
+      "step": 4880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.469357545965563e-05,
+      "loss": 1.6047,
+      "step": 4881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.462026866279039e-05,
+      "loss": 1.5876,
+      "step": 4882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.454698960578012e-05,
+      "loss": 1.6752,
+      "step": 4883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.447373830483748e-05,
+      "loss": 1.6711,
+      "step": 4884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.44005147761691e-05,
+      "loss": 1.4842,
+      "step": 4885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.432731903597545e-05,
+      "loss": 1.6119,
+      "step": 4886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.425415110045096e-05,
+      "loss": 1.6855,
+      "step": 4887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.418101098578363e-05,
+      "loss": 1.7287,
+      "step": 4888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.41078987081556e-05,
+      "loss": 1.5061,
+      "step": 4889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.403481428374266e-05,
+      "loss": 1.5211,
+      "step": 4890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.396175772871446e-05,
+      "loss": 1.4737,
+      "step": 4891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.388872905923454e-05,
+      "loss": 1.8717,
+      "step": 4892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.381572829146029e-05,
+      "loss": 1.6541,
+      "step": 4893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.374275544154293e-05,
+      "loss": 1.6881,
+      "step": 4894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.366981052562736e-05,
+      "loss": 1.4826,
+      "step": 4895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.359689355985253e-05,
+      "loss": 1.7081,
+      "step": 4896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.352400456035094e-05,
+      "loss": 1.5743,
+      "step": 4897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.345114354324918e-05,
+      "loss": 1.5192,
+      "step": 4898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.337831052466737e-05,
+      "loss": 1.6082,
+      "step": 4899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.330550552071964e-05,
+      "loss": 1.7096,
+      "step": 4900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.323272854751395e-05,
+      "loss": 1.7195,
+      "step": 4901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.315997962115183e-05,
+      "loss": 1.7857,
+      "step": 4902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.308725875772884e-05,
+      "loss": 1.7462,
+      "step": 4903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.301456597333412e-05,
+      "loss": 1.6169,
+      "step": 4904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.294190128405086e-05,
+      "loss": 1.5507,
+      "step": 4905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.286926470595572e-05,
+      "loss": 1.551,
+      "step": 4906
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.27966562551194e-05,
+      "loss": 1.7934,
+      "step": 4907
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.272407594760635e-05,
+      "loss": 1.569,
+      "step": 4908
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.265152379947457e-05,
+      "loss": 1.5926,
+      "step": 4909
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.257899982677612e-05,
+      "loss": 1.7362,
+      "step": 4910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.25065040455566e-05,
+      "loss": 1.6547,
+      "step": 4911
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.243403647185552e-05,
+      "loss": 1.9265,
+      "step": 4912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.236159712170602e-05,
+      "loss": 1.5729,
+      "step": 4913
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.228918601113512e-05,
+      "loss": 1.8644,
+      "step": 4914
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.221680315616362e-05,
+      "loss": 1.4587,
+      "step": 4915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.214444857280582e-05,
+      "loss": 1.6476,
+      "step": 4916
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.207212227707013e-05,
+      "loss": 1.6945,
+      "step": 4917
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.199982428495828e-05,
+      "loss": 1.4538,
+      "step": 4918
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.192755461246619e-05,
+      "loss": 1.5705,
+      "step": 4919
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.18553132755831e-05,
+      "loss": 1.748,
+      "step": 4920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.178310029029227e-05,
+      "loss": 1.6486,
+      "step": 4921
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.171091567257062e-05,
+      "loss": 1.4553,
+      "step": 4922
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.163875943838865e-05,
+      "loss": 1.5303,
+      "step": 4923
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.15666316037108e-05,
+      "loss": 1.4854,
+      "step": 4924
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.149453218449502e-05,
+      "loss": 1.5817,
+      "step": 4925
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.142246119669318e-05,
+      "loss": 1.6001,
+      "step": 4926
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.13504186562506e-05,
+      "loss": 1.7317,
+      "step": 4927
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.127840457910658e-05,
+      "loss": 1.6285,
+      "step": 4928
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.1206418981194e-05,
+      "loss": 1.5665,
+      "step": 4929
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.113446187843933e-05,
+      "loss": 1.6216,
+      "step": 4930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.1062533286763e-05,
+      "loss": 1.6589,
+      "step": 4931
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.099063322207881e-05,
+      "loss": 1.5715,
+      "step": 4932
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.091876170029457e-05,
+      "loss": 1.6609,
+      "step": 4933
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.08469187373114e-05,
+      "loss": 1.7406,
+      "step": 4934
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.077510434902459e-05,
+      "loss": 1.6629,
+      "step": 4935
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.070331855132271e-05,
+      "loss": 1.5239,
+      "step": 4936
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.063156136008809e-05,
+      "loss": 1.6526,
+      "step": 4937
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.055983279119684e-05,
+      "loss": 1.7704,
+      "step": 4938
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.048813286051861e-05,
+      "loss": 1.7759,
+      "step": 4939
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.041646158391687e-05,
+      "loss": 1.7439,
+      "step": 4940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.034481897724852e-05,
+      "loss": 1.5989,
+      "step": 4941
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.027320505636434e-05,
+      "loss": 1.4862,
+      "step": 4942
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.020161983710874e-05,
+      "loss": 1.6627,
+      "step": 4943
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.013006333531955e-05,
+      "loss": 1.6727,
+      "step": 4944
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.005853556682857e-05,
+      "loss": 1.7569,
+      "step": 4945
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.998703654746097e-05,
+      "loss": 1.6651,
+      "step": 4946
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.991556629303578e-05,
+      "loss": 1.6442,
+      "step": 4947
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.984412481936542e-05,
+      "loss": 1.6468,
+      "step": 4948
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.977271214225618e-05,
+      "loss": 1.752,
+      "step": 4949
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.970132827750784e-05,
+      "loss": 1.6005,
+      "step": 4950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.962997324091393e-05,
+      "loss": 1.7786,
+      "step": 4951
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.955864704826149e-05,
+      "loss": 1.6897,
+      "step": 4952
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.948734971533108e-05,
+      "loss": 1.9344,
+      "step": 4953
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.941608125789718e-05,
+      "loss": 1.51,
+      "step": 4954
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.934484169172755e-05,
+      "loss": 1.4182,
+      "step": 4955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.927363103258376e-05,
+      "loss": 1.7939,
+      "step": 4956
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.920244929622095e-05,
+      "loss": 1.5745,
+      "step": 4957
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.913129649838795e-05,
+      "loss": 1.4545,
+      "step": 4958
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.90601726548269e-05,
+      "loss": 1.8378,
+      "step": 4959
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.898907778127384e-05,
+      "loss": 1.5849,
+      "step": 4960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.891801189345826e-05,
+      "loss": 1.606,
+      "step": 4961
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.884697500710318e-05,
+      "loss": 1.56,
+      "step": 4962
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.877596713792532e-05,
+      "loss": 1.6846,
+      "step": 4963
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.870498830163492e-05,
+      "loss": 1.7439,
+      "step": 4964
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.863403851393594e-05,
+      "loss": 1.6387,
+      "step": 4965
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.856311779052563e-05,
+      "loss": 1.6617,
+      "step": 4966
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.849222614709509e-05,
+      "loss": 1.6794,
+      "step": 4967
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.842136359932873e-05,
+      "loss": 1.7147,
+      "step": 4968
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.835053016290482e-05,
+      "loss": 1.6966,
+      "step": 4969
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.827972585349486e-05,
+      "loss": 1.5878,
+      "step": 4970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.820895068676414e-05,
+      "loss": 1.5479,
+      "step": 4971
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.813820467837153e-05,
+      "loss": 1.712,
+      "step": 4972
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.806748784396918e-05,
+      "loss": 1.7339,
+      "step": 4973
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.799680019920311e-05,
+      "loss": 1.6201,
+      "step": 4974
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.792614175971261e-05,
+      "loss": 1.7698,
+      "step": 4975
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.785551254113073e-05,
+      "loss": 1.5584,
+      "step": 4976
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.778491255908385e-05,
+      "loss": 1.4848,
+      "step": 4977
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.771434182919205e-05,
+      "loss": 1.5038,
+      "step": 4978
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.764380036706891e-05,
+      "loss": 1.7038,
+      "step": 4979
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.75732881883214e-05,
+      "loss": 1.6436,
+      "step": 4980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.75028053085502e-05,
+      "loss": 1.7248,
+      "step": 4981
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.743235174334935e-05,
+      "loss": 1.5854,
+      "step": 4982
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.736192750830652e-05,
+      "loss": 1.6544,
+      "step": 4983
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.729153261900277e-05,
+      "loss": 1.6016,
+      "step": 4984
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.722116709101278e-05,
+      "loss": 1.4654,
+      "step": 4985
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.715083093990475e-05,
+      "loss": 1.7633,
+      "step": 4986
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.708052418124022e-05,
+      "loss": 1.6312,
+      "step": 4987
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.701024683057446e-05,
+      "loss": 1.7288,
+      "step": 4988
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.693999890345595e-05,
+      "loss": 1.6427,
+      "step": 4989
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.686978041542696e-05,
+      "loss": 1.6772,
+      "step": 4990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.679959138202294e-05,
+      "loss": 1.5896,
+      "step": 4991
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.672943181877309e-05,
+      "loss": 1.7846,
+      "step": 4992
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.665930174120003e-05,
+      "loss": 1.5749,
+      "step": 4993
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.658920116481969e-05,
+      "loss": 1.6201,
+      "step": 4994
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.65191301051417e-05,
+      "loss": 1.7142,
+      "step": 4995
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.644908857766894e-05,
+      "loss": 1.873,
+      "step": 4996
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6379076597898e-05,
+      "loss": 1.6486,
+      "step": 4997
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.630909418131866e-05,
+      "loss": 1.7585,
+      "step": 4998
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.623914134341438e-05,
+      "loss": 1.477,
+      "step": 4999
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.616921809966205e-05,
+      "loss": 1.7329,
+      "step": 5000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 6779,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 1.1430643571946947e+18,
+  "trial_name": null,
+  "trial_params": null
+}