diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,13915 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9997302109750176,
+  "eval_steps": 500,
+  "global_step": 2316,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 2.2125,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997301381692574e-05,
+      "loss": 2.3529,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997301381692574e-05,
+      "loss": 2.0437,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999460276338515e-05,
+      "loss": 1.8142,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999460276338515e-05,
+      "loss": 5.2014,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9991904145077725e-05,
+      "loss": 5.0285,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998920552677029e-05,
+      "loss": 2.2181,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998650690846287e-05,
+      "loss": 1.9741,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998380829015544e-05,
+      "loss": 1.6445,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9981109671848014e-05,
+      "loss": 1.554,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997841105354059e-05,
+      "loss": 1.5232,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975712435233165e-05,
+      "loss": 1.6303,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997301381692574e-05,
+      "loss": 1.3479,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997031519861831e-05,
+      "loss": 1.1682,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996761658031088e-05,
+      "loss": 1.3892,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996491796200345e-05,
+      "loss": 1.4308,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996221934369603e-05,
+      "loss": 1.2858,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9959520725388604e-05,
+      "loss": 1.1834,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9956822107081176e-05,
+      "loss": 1.1976,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9954123488773755e-05,
+      "loss": 1.1976,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.995142487046632e-05,
+      "loss": 1.1328,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994872625215889e-05,
+      "loss": 1.1109,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994602763385147e-05,
+      "loss": 1.2262,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994332901554404e-05,
+      "loss": 1.1031,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9940630397236615e-05,
+      "loss": 1.2434,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9937931778929194e-05,
+      "loss": 1.0342,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9935233160621766e-05,
+      "loss": 0.8919,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993253454231434e-05,
+      "loss": 0.8663,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992983592400691e-05,
+      "loss": 1.1186,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992713730569948e-05,
+      "loss": 1.091,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9924438687392055e-05,
+      "loss": 0.9284,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9921740069084634e-05,
+      "loss": 0.965,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9919041450777206e-05,
+      "loss": 1.0435,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991634283246978e-05,
+      "loss": 0.8604,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.991364421416235e-05,
+      "loss": 1.0643,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.991094559585492e-05,
+      "loss": 1.0008,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9908246977547494e-05,
+      "loss": 0.8726,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.990554835924007e-05,
+      "loss": 0.7953,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9902849740932645e-05,
+      "loss": 1.126,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.990015112262522e-05,
+      "loss": 0.8383,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9897452504317796e-05,
+      "loss": 1.0159,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.989475388601037e-05,
+      "loss": 0.9224,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.989205526770293e-05,
+      "loss": 0.9507,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.988935664939551e-05,
+      "loss": 1.017,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9886658031088084e-05,
+      "loss": 1.1385,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9883959412780656e-05,
+      "loss": 0.7967,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9881260794473235e-05,
+      "loss": 0.9508,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.987856217616581e-05,
+      "loss": 0.8141,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.987586355785838e-05,
+      "loss": 0.728,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.987316493955095e-05,
+      "loss": 0.6839,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9870466321243523e-05,
+      "loss": 0.6524,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9867767702936096e-05,
+      "loss": 0.8659,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9865069084628674e-05,
+      "loss": 0.8046,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9862370466321246e-05,
+      "loss": 0.7859,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.985967184801382e-05,
+      "loss": 0.8885,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.98569732297064e-05,
+      "loss": 1.0624,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.985427461139896e-05,
+      "loss": 0.7863,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9851575993091535e-05,
+      "loss": 0.7481,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9848877374784114e-05,
+      "loss": 0.8631,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9846178756476686e-05,
+      "loss": 1.0127,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.984348013816926e-05,
+      "loss": 1.3302,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.984078151986184e-05,
+      "loss": 0.6437,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.983808290155441e-05,
+      "loss": 0.6804,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.983538428324698e-05,
+      "loss": 1.1116,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.983268566493955e-05,
+      "loss": 0.8418,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9829987046632125e-05,
+      "loss": 0.7645,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98272884283247e-05,
+      "loss": 0.7653,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9824589810017276e-05,
+      "loss": 1.0655,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.982189119170985e-05,
+      "loss": 1.0108,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.981919257340242e-05,
+      "loss": 0.7279,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.981649395509499e-05,
+      "loss": 0.7262,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9813795336787564e-05,
+      "loss": 0.75,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9811096718480136e-05,
+      "loss": 0.743,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9808398100172715e-05,
+      "loss": 0.8266,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.980569948186529e-05,
+      "loss": 0.8379,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.980300086355786e-05,
+      "loss": 0.8079,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.980030224525044e-05,
+      "loss": 0.7824,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9797603626943004e-05,
+      "loss": 0.9083,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9794905008635576e-05,
+      "loss": 0.8015,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9792206390328155e-05,
+      "loss": 0.726,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.978950777202073e-05,
+      "loss": 0.6969,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.97868091537133e-05,
+      "loss": 0.6953,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978411053540588e-05,
+      "loss": 0.737,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978141191709845e-05,
+      "loss": 0.7533,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977871329879102e-05,
+      "loss": 0.67,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9776014680483594e-05,
+      "loss": 0.6257,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9773316062176166e-05,
+      "loss": 0.6204,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977061744386874e-05,
+      "loss": 0.6059,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976791882556132e-05,
+      "loss": 0.6396,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976522020725389e-05,
+      "loss": 0.5719,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976252158894646e-05,
+      "loss": 0.9229,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.975982297063903e-05,
+      "loss": 0.7309,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9757124352331605e-05,
+      "loss": 0.3985,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.975442573402418e-05,
+      "loss": 0.8208,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9751727115716756e-05,
+      "loss": 0.7581,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.974902849740933e-05,
+      "loss": 0.6625,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.97463298791019e-05,
+      "loss": 0.7338,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.974363126079448e-05,
+      "loss": 0.7054,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.974093264248705e-05,
+      "loss": 0.536,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.973823402417962e-05,
+      "loss": 0.7846,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9735535405872195e-05,
+      "loss": 0.5402,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.973283678756477e-05,
+      "loss": 0.6517,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9730138169257346e-05,
+      "loss": 0.493,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.972743955094992e-05,
+      "loss": 0.5854,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972474093264249e-05,
+      "loss": 0.6206,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972204231433506e-05,
+      "loss": 0.6691,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9719343696027635e-05,
+      "loss": 0.668,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971664507772021e-05,
+      "loss": 0.5883,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9713946459412786e-05,
+      "loss": 0.6845,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971124784110536e-05,
+      "loss": 0.4365,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970854922279793e-05,
+      "loss": 0.7495,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970585060449051e-05,
+      "loss": 0.5896,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970315198618308e-05,
+      "loss": 0.6586,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9700453367875646e-05,
+      "loss": 0.6317,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9697754749568225e-05,
+      "loss": 0.8567,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.96950561312608e-05,
+      "loss": 0.4546,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.969235751295337e-05,
+      "loss": 0.9692,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.968965889464595e-05,
+      "loss": 0.6291,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.968696027633852e-05,
+      "loss": 0.5538,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.968426165803109e-05,
+      "loss": 0.5545,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9681563039723664e-05,
+      "loss": 0.466,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9678864421416236e-05,
+      "loss": 0.7113,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967616580310881e-05,
+      "loss": 0.5074,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967346718480139e-05,
+      "loss": 0.6705,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967076856649396e-05,
+      "loss": 0.5684,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.966806994818653e-05,
+      "loss": 0.5592,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9665371329879104e-05,
+      "loss": 0.5241,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9662672711571676e-05,
+      "loss": 0.6117,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.965997409326425e-05,
+      "loss": 0.841,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9657275474956827e-05,
+      "loss": 0.6101,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.96545768566494e-05,
+      "loss": 0.5021,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.965187823834197e-05,
+      "loss": 0.6917,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.964917962003455e-05,
+      "loss": 0.609,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.964648100172712e-05,
+      "loss": 0.7775,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.964378238341969e-05,
+      "loss": 0.5425,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9641083765112266e-05,
+      "loss": 0.5172,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.963838514680484e-05,
+      "loss": 0.7997,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.963568652849741e-05,
+      "loss": 0.6949,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.963298791018999e-05,
+      "loss": 0.4425,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.963028929188256e-05,
+      "loss": 0.3626,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.962759067357513e-05,
+      "loss": 0.8014,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9624892055267705e-05,
+      "loss": 0.6374,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.962219343696028e-05,
+      "loss": 0.6432,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961949481865285e-05,
+      "loss": 0.9478,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961679620034543e-05,
+      "loss": 0.5038,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9614097582038e-05,
+      "loss": 0.4594,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961139896373057e-05,
+      "loss": 0.4026,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960870034542315e-05,
+      "loss": 0.6985,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9606001727115716e-05,
+      "loss": 0.4642,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960330310880829e-05,
+      "loss": 0.6817,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.960060449050087e-05,
+      "loss": 0.5242,
+      "step": 151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959790587219344e-05,
+      "loss": 0.6573,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959520725388601e-05,
+      "loss": 0.4164,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959250863557859e-05,
+      "loss": 0.3467,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958981001727116e-05,
+      "loss": 0.3921,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9587111398963735e-05,
+      "loss": 0.578,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958441278065631e-05,
+      "loss": 0.4228,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958171416234888e-05,
+      "loss": 0.5125,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957901554404145e-05,
+      "loss": 0.6059,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957631692573403e-05,
+      "loss": 0.4832,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.95736183074266e-05,
+      "loss": 0.3512,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9570919689119174e-05,
+      "loss": 0.4005,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9568221070811746e-05,
+      "loss": 0.3871,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.956552245250432e-05,
+      "loss": 0.3806,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.956282383419689e-05,
+      "loss": 0.3408,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.956012521588947e-05,
+      "loss": 0.567,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.955742659758204e-05,
+      "loss": 0.4464,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.955472797927461e-05,
+      "loss": 0.3298,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.955202936096719e-05,
+      "loss": 0.6172,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9549330742659764e-05,
+      "loss": 0.553,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.954663212435233e-05,
+      "loss": 0.383,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.954393350604491e-05,
+      "loss": 0.8125,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.954123488773748e-05,
+      "loss": 0.5586,
+      "step": 173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.953853626943005e-05,
+      "loss": 0.4229,
+      "step": 174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.953583765112263e-05,
+      "loss": 0.5972,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9533139032815203e-05,
+      "loss": 0.4871,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9530440414507776e-05,
+      "loss": 0.4129,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952774179620035e-05,
+      "loss": 0.5531,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952504317789292e-05,
+      "loss": 0.7782,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952234455958549e-05,
+      "loss": 0.6721,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951964594127807e-05,
+      "loss": 0.3938,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951694732297064e-05,
+      "loss": 0.3672,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9514248704663215e-05,
+      "loss": 0.6995,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951155008635579e-05,
+      "loss": 0.5178,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950885146804836e-05,
+      "loss": 0.3297,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950615284974093e-05,
+      "loss": 0.538,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950345423143351e-05,
+      "loss": 0.754,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950075561312608e-05,
+      "loss": 0.6441,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9498056994818654e-05,
+      "loss": 0.2859,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.949535837651123e-05,
+      "loss": 0.5134,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9492659758203805e-05,
+      "loss": 0.5244,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948996113989637e-05,
+      "loss": 0.3468,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948726252158895e-05,
+      "loss": 0.5264,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948456390328152e-05,
+      "loss": 0.68,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948186528497409e-05,
+      "loss": 0.3724,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.947916666666667e-05,
+      "loss": 0.5443,
+      "step": 196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9476468048359244e-05,
+      "loss": 0.2899,
+      "step": 197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9473769430051816e-05,
+      "loss": 0.3653,
+      "step": 198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.947107081174439e-05,
+      "loss": 0.2426,
+      "step": 199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.946837219343696e-05,
+      "loss": 0.3754,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.946567357512953e-05,
+      "loss": 0.5305,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.946297495682211e-05,
+      "loss": 0.4541,
+      "step": 202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9460276338514684e-05,
+      "loss": 0.4101,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9457577720207256e-05,
+      "loss": 0.5385,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9454879101899835e-05,
+      "loss": 0.4087,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.94521804835924e-05,
+      "loss": 0.5443,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.944948186528497e-05,
+      "loss": 0.5608,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.944678324697755e-05,
+      "loss": 0.392,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.944408462867012e-05,
+      "loss": 0.7061,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9441386010362695e-05,
+      "loss": 0.3271,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9438687392055274e-05,
+      "loss": 0.3423,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9435988773747846e-05,
+      "loss": 0.3721,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.943329015544042e-05,
+      "loss": 0.4677,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.943059153713299e-05,
+      "loss": 0.5723,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.942789291882556e-05,
+      "loss": 0.3805,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9425194300518134e-05,
+      "loss": 0.5471,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.942249568221071e-05,
+      "loss": 0.3468,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9419797063903285e-05,
+      "loss": 0.3105,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.941709844559586e-05,
+      "loss": 0.4763,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.941439982728843e-05,
+      "loss": 0.5175,
+      "step": 220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9411701208981e-05,
+      "loss": 0.4045,
+      "step": 221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9409002590673574e-05,
+      "loss": 0.304,
+      "step": 222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.940630397236615e-05,
+      "loss": 0.418,
+      "step": 223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9403605354058724e-05,
+      "loss": 0.4442,
+      "step": 224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9400906735751297e-05,
+      "loss": 0.5833,
+      "step": 225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9398208117443875e-05,
+      "loss": 0.4228,
+      "step": 226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.939550949913645e-05,
+      "loss": 0.4292,
+      "step": 227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.939281088082901e-05,
+      "loss": 0.2934,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.939011226252159e-05,
+      "loss": 0.3499,
+      "step": 229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9387413644214164e-05,
+      "loss": 0.5609,
+      "step": 230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9384715025906736e-05,
+      "loss": 0.5715,
+      "step": 231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9382016407599315e-05,
+      "loss": 0.4085,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.937931778929189e-05,
+      "loss": 0.641,
+      "step": 233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.937661917098446e-05,
+      "loss": 0.438,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.937392055267703e-05,
+      "loss": 0.4865,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.93712219343696e-05,
+      "loss": 0.3039,
+      "step": 236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9368523316062175e-05,
+      "loss": 0.3772,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9365824697754754e-05,
+      "loss": 0.3684,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9363126079447326e-05,
+      "loss": 0.4095,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.93604274611399e-05,
+      "loss": 0.47,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.935772884283247e-05,
+      "loss": 0.3676,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.935503022452504e-05,
+      "loss": 0.6223,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9352331606217614e-05,
+      "loss": 0.5087,
+      "step": 243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.934963298791019e-05,
+      "loss": 0.3485,
+      "step": 244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9346934369602765e-05,
+      "loss": 0.4089,
+      "step": 245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.934423575129534e-05,
+      "loss": 0.556,
+      "step": 246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9341537132987916e-05,
+      "loss": 0.2759,
+      "step": 247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.933883851468049e-05,
+      "loss": 0.2947,
+      "step": 248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9336139896373054e-05,
+      "loss": 0.2947,
+      "step": 249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.933344127806563e-05,
+      "loss": 0.3681,
+      "step": 250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9330742659758205e-05,
+      "loss": 0.4536,
+      "step": 251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.932804404145078e-05,
+      "loss": 0.4407,
+      "step": 252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9325345423143356e-05,
+      "loss": 0.3459,
+      "step": 253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.932264680483593e-05,
+      "loss": 0.6053,
+      "step": 254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.93199481865285e-05,
+      "loss": 0.5334,
+      "step": 255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.931724956822107e-05,
+      "loss": 0.4061,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9314550949913644e-05,
+      "loss": 0.5331,
+      "step": 257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9311852331606216e-05,
+      "loss": 0.4022,
+      "step": 258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9309153713298795e-05,
+      "loss": 0.2607,
+      "step": 259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.930645509499137e-05,
+      "loss": 0.305,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.930375647668394e-05,
+      "loss": 0.3079,
+      "step": 261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.930105785837652e-05,
+      "loss": 0.3284,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.929835924006908e-05,
+      "loss": 0.4813,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9295660621761655e-05,
+      "loss": 0.3791,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9292962003454234e-05,
+      "loss": 0.2718,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9290263385146806e-05,
+      "loss": 0.5161,
+      "step": 266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.928756476683938e-05,
+      "loss": 0.2134,
+      "step": 267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.928486614853196e-05,
+      "loss": 0.2211,
+      "step": 268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.928216753022453e-05,
+      "loss": 0.3971,
+      "step": 269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.92794689119171e-05,
+      "loss": 0.2595,
+      "step": 270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9276770293609673e-05,
+      "loss": 0.3672,
+      "step": 271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9274071675302246e-05,
+      "loss": 0.4627,
+      "step": 272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.927137305699482e-05,
+      "loss": 0.3424,
+      "step": 273
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9268674438687396e-05,
+      "loss": 0.3892,
+      "step": 274
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.926597582037997e-05,
+      "loss": 0.2731,
+      "step": 275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.926327720207254e-05,
+      "loss": 0.4607,
+      "step": 276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.926057858376511e-05,
+      "loss": 0.3027,
+      "step": 277
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9257879965457685e-05,
+      "loss": 0.2114,
+      "step": 278
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.925518134715026e-05,
+      "loss": 0.3063,
+      "step": 279
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9252482728842836e-05,
+      "loss": 0.2992,
+      "step": 280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.924978411053541e-05,
+      "loss": 0.6008,
+      "step": 281
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.924708549222798e-05,
+      "loss": 0.7236,
+      "step": 282
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.924438687392056e-05,
+      "loss": 0.3892,
+      "step": 283
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.924168825561313e-05,
+      "loss": 0.4132,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9238989637305696e-05,
+      "loss": 0.5033,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9236291018998275e-05,
+      "loss": 0.5299,
+      "step": 286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.923359240069085e-05,
+      "loss": 0.357,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.923089378238342e-05,
+      "loss": 0.5673,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9228195164076e-05,
+      "loss": 0.5696,
+      "step": 289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.922549654576857e-05,
+      "loss": 0.5464,
+      "step": 290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.922279792746114e-05,
+      "loss": 0.5129,
+      "step": 291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9220099309153714e-05,
+      "loss": 0.2375,
+      "step": 292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9217400690846286e-05,
+      "loss": 0.3563,
+      "step": 293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9214702072538865e-05,
+      "loss": 0.2936,
+      "step": 294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.921200345423144e-05,
+      "loss": 0.3163,
+      "step": 295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.920930483592401e-05,
+      "loss": 0.2301,
+      "step": 296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.920660621761659e-05,
+      "loss": 0.4927,
+      "step": 297
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9203907599309154e-05,
+      "loss": 0.3777,
+      "step": 298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9201208981001726e-05,
+      "loss": 0.4613,
+      "step": 299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9198510362694305e-05,
+      "loss": 0.3212,
+      "step": 300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.919581174438688e-05,
+      "loss": 0.295,
+      "step": 301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.919311312607945e-05,
+      "loss": 0.4462,
+      "step": 302
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.919041450777203e-05,
+      "loss": 0.257,
+      "step": 303
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.91877158894646e-05,
+      "loss": 0.3013,
+      "step": 304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.918501727115717e-05,
+      "loss": 0.3974,
+      "step": 305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9182318652849744e-05,
+      "loss": 0.3453,
+      "step": 306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9179620034542316e-05,
+      "loss": 0.3666,
+      "step": 307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.917692141623489e-05,
+      "loss": 0.336,
+      "step": 308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.917422279792747e-05,
+      "loss": 0.377,
+      "step": 309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.917152417962004e-05,
+      "loss": 0.7311,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.916882556131261e-05,
+      "loss": 0.3735,
+      "step": 311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.916612694300518e-05,
+      "loss": 0.3952,
+      "step": 312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9163428324697755e-05,
+      "loss": 0.5144,
+      "step": 313
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.916072970639033e-05,
+      "loss": 0.2203,
+      "step": 314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9158031088082906e-05,
+      "loss": 0.2859,
+      "step": 315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.915533246977548e-05,
+      "loss": 0.4005,
+      "step": 316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.915263385146805e-05,
+      "loss": 0.5704,
+      "step": 317
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.914993523316063e-05,
+      "loss": 0.4239,
+      "step": 318
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.91472366148532e-05,
+      "loss": 0.265,
+      "step": 319
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9144537996545767e-05,
+      "loss": 0.4597,
+      "step": 320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9141839378238345e-05,
+      "loss": 0.4306,
+      "step": 321
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.913914075993092e-05,
+      "loss": 0.4216,
+      "step": 322
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.913644214162349e-05,
+      "loss": 0.2565,
+      "step": 323
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.913374352331607e-05,
+      "loss": 0.2904,
+      "step": 324
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.913104490500864e-05,
+      "loss": 0.4405,
+      "step": 325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.912834628670121e-05,
+      "loss": 0.3577,
+      "step": 326
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9125647668393785e-05,
+      "loss": 0.4701,
+      "step": 327
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.912294905008636e-05,
+      "loss": 0.4019,
+      "step": 328
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.912025043177893e-05,
+      "loss": 0.3901,
+      "step": 329
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.911755181347151e-05,
+      "loss": 0.2612,
+      "step": 330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.911485319516408e-05,
+      "loss": 0.2341,
+      "step": 331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.911215457685665e-05,
+      "loss": 0.2448,
+      "step": 332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.910945595854923e-05,
+      "loss": 0.3257,
+      "step": 333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9106757340241796e-05,
+      "loss": 0.5166,
+      "step": 334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.910405872193437e-05,
+      "loss": 0.4599,
+      "step": 335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.910136010362695e-05,
+      "loss": 0.2529,
+      "step": 336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.909866148531952e-05,
+      "loss": 0.3444,
+      "step": 337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.909596286701209e-05,
+      "loss": 0.3413,
+      "step": 338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.909326424870467e-05,
+      "loss": 0.3271,
+      "step": 339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.909056563039724e-05,
+      "loss": 0.6998,
+      "step": 340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9087867012089814e-05,
+      "loss": 0.4179,
+      "step": 341
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9085168393782386e-05,
+      "loss": 0.2592,
+      "step": 342
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.908246977547496e-05,
+      "loss": 0.3958,
+      "step": 343
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.907977115716753e-05,
+      "loss": 0.2736,
+      "step": 344
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.907707253886011e-05,
+      "loss": 0.5218,
+      "step": 345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.907437392055268e-05,
+      "loss": 0.5159,
+      "step": 346
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9071675302245253e-05,
+      "loss": 0.1873,
+      "step": 347
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9068976683937826e-05,
+      "loss": 0.4474,
+      "step": 348
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.90662780656304e-05,
+      "loss": 0.3503,
+      "step": 349
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.906357944732297e-05,
+      "loss": 0.5683,
+      "step": 350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.906088082901555e-05,
+      "loss": 0.1918,
+      "step": 351
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.905818221070812e-05,
+      "loss": 0.3289,
+      "step": 352
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.905548359240069e-05,
+      "loss": 0.4423,
+      "step": 353
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.905278497409327e-05,
+      "loss": 0.4643,
+      "step": 354
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.905008635578584e-05,
+      "loss": 0.427,
+      "step": 355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.904738773747841e-05,
+      "loss": 0.2569,
+      "step": 356
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.904468911917099e-05,
+      "loss": 0.4416,
+      "step": 357
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.904199050086356e-05,
+      "loss": 0.3137,
+      "step": 358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.903929188255613e-05,
+      "loss": 0.3853,
+      "step": 359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.903659326424871e-05,
+      "loss": 0.2045,
+      "step": 360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.903389464594128e-05,
+      "loss": 0.2859,
+      "step": 361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9031196027633855e-05,
+      "loss": 0.6373,
+      "step": 362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.902849740932643e-05,
+      "loss": 0.257,
+      "step": 363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9025798791019e-05,
+      "loss": 0.3097,
+      "step": 364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.902310017271157e-05,
+      "loss": 0.4654,
+      "step": 365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.902040155440415e-05,
+      "loss": 0.4958,
+      "step": 366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.901770293609672e-05,
+      "loss": 0.3818,
+      "step": 367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9015004317789294e-05,
+      "loss": 0.3353,
+      "step": 368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9012305699481866e-05,
+      "loss": 0.3676,
+      "step": 369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.900960708117444e-05,
+      "loss": 0.3163,
+      "step": 370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.900690846286701e-05,
+      "loss": 0.2494,
+      "step": 371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.900420984455959e-05,
+      "loss": 0.3697,
+      "step": 372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.900151122625216e-05,
+      "loss": 0.419,
+      "step": 373
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8998812607944734e-05,
+      "loss": 0.4546,
+      "step": 374
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.899611398963731e-05,
+      "loss": 0.3899,
+      "step": 375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8993415371329885e-05,
+      "loss": 0.2052,
+      "step": 376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.899071675302245e-05,
+      "loss": 0.4068,
+      "step": 377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.898801813471503e-05,
+      "loss": 0.2289,
+      "step": 378
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.89853195164076e-05,
+      "loss": 0.3603,
+      "step": 379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.898262089810017e-05,
+      "loss": 0.293,
+      "step": 380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.897992227979275e-05,
+      "loss": 0.1897,
+      "step": 381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8977223661485324e-05,
+      "loss": 0.3965,
+      "step": 382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8974525043177896e-05,
+      "loss": 0.5813,
+      "step": 383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.897182642487047e-05,
+      "loss": 0.2116,
+      "step": 384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.896912780656304e-05,
+      "loss": 0.3721,
+      "step": 385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.896642918825561e-05,
+      "loss": 0.4208,
+      "step": 386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.896373056994819e-05,
+      "loss": 0.2967,
+      "step": 387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.896103195164076e-05,
+      "loss": 0.1889,
+      "step": 388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8958333333333335e-05,
+      "loss": 0.2351,
+      "step": 389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8955634715025914e-05,
+      "loss": 0.2401,
+      "step": 390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.895293609671848e-05,
+      "loss": 0.2171,
+      "step": 391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.895023747841105e-05,
+      "loss": 0.2881,
+      "step": 392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.894753886010363e-05,
+      "loss": 0.3246,
+      "step": 393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.89448402417962e-05,
+      "loss": 0.1939,
+      "step": 394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8942141623488775e-05,
+      "loss": 0.3593,
+      "step": 395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8939443005181353e-05,
+      "loss": 0.3731,
+      "step": 396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8936744386873925e-05,
+      "loss": 0.3676,
+      "step": 397
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.89340457685665e-05,
+      "loss": 0.3767,
+      "step": 398
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.893134715025907e-05,
+      "loss": 0.1649,
+      "step": 399
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.892864853195164e-05,
+      "loss": 0.1928,
+      "step": 400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8925949913644214e-05,
+      "loss": 0.4724,
+      "step": 401
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.892325129533679e-05,
+      "loss": 0.3583,
+      "step": 402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8920552677029365e-05,
+      "loss": 0.3068,
+      "step": 403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.891785405872194e-05,
+      "loss": 0.3732,
+      "step": 404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.891515544041451e-05,
+      "loss": 0.4795,
+      "step": 405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.891245682210708e-05,
+      "loss": 0.3738,
+      "step": 406
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.890975820379965e-05,
+      "loss": 0.1854,
+      "step": 407
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.890705958549223e-05,
+      "loss": 0.451,
+      "step": 408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8904360967184804e-05,
+      "loss": 0.4543,
+      "step": 409
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8901662348877376e-05,
+      "loss": 0.2541,
+      "step": 410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8898963730569955e-05,
+      "loss": 0.2486,
+      "step": 411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.889626511226253e-05,
+      "loss": 0.3211,
+      "step": 412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.889356649395509e-05,
+      "loss": 0.4354,
+      "step": 413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.889086787564767e-05,
+      "loss": 0.3461,
+      "step": 414
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.888816925734024e-05,
+      "loss": 0.239,
+      "step": 415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8885470639032815e-05,
+      "loss": 0.263,
+      "step": 416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8882772020725394e-05,
+      "loss": 0.1894,
+      "step": 417
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8880073402417966e-05,
+      "loss": 0.3614,
+      "step": 418
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.887737478411054e-05,
+      "loss": 0.1837,
+      "step": 419
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.887467616580311e-05,
+      "loss": 0.3366,
+      "step": 420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.887197754749568e-05,
+      "loss": 0.2415,
+      "step": 421
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8869278929188255e-05,
+      "loss": 0.2487,
+      "step": 422
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8866580310880834e-05,
+      "loss": 0.3652,
+      "step": 423
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8863881692573406e-05,
+      "loss": 0.665,
+      "step": 424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.886118307426598e-05,
+      "loss": 0.5027,
+      "step": 425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.885848445595855e-05,
+      "loss": 0.2974,
+      "step": 426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.885578583765112e-05,
+      "loss": 0.2855,
+      "step": 427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8853087219343694e-05,
+      "loss": 0.1593,
+      "step": 428
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.885038860103627e-05,
+      "loss": 0.4032,
+      "step": 429
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8847689982728845e-05,
+      "loss": 0.1782,
+      "step": 430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.884499136442142e-05,
+      "loss": 0.3995,
+      "step": 431
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8842292746113996e-05,
+      "loss": 0.5649,
+      "step": 432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.883959412780657e-05,
+      "loss": 0.2681,
+      "step": 433
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.883689550949913e-05,
+      "loss": 0.1981,
+      "step": 434
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.883419689119171e-05,
+      "loss": 0.3722,
+      "step": 435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8831498272884284e-05,
+      "loss": 0.1951,
+      "step": 436
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8828799654576856e-05,
+      "loss": 0.2502,
+      "step": 437
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8826101036269435e-05,
+      "loss": 0.4997,
+      "step": 438
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.882340241796201e-05,
+      "loss": 0.241,
+      "step": 439
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.882070379965458e-05,
+      "loss": 0.2273,
+      "step": 440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.881800518134715e-05,
+      "loss": 0.6973,
+      "step": 441
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8815306563039724e-05,
+      "loss": 0.2606,
+      "step": 442
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8812607944732296e-05,
+      "loss": 0.157,
+      "step": 443
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8809909326424874e-05,
+      "loss": 0.2865,
+      "step": 444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8807210708117447e-05,
+      "loss": 0.2706,
+      "step": 445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.880451208981002e-05,
+      "loss": 0.2645,
+      "step": 446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.88018134715026e-05,
+      "loss": 0.3792,
+      "step": 447
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.879911485319516e-05,
+      "loss": 0.6265,
+      "step": 448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8796416234887735e-05,
+      "loss": 0.3772,
+      "step": 449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8793717616580314e-05,
+      "loss": 0.3246,
+      "step": 450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8791018998272886e-05,
+      "loss": 0.4143,
+      "step": 451
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.878832037996546e-05,
+      "loss": 0.6993,
+      "step": 452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.878562176165804e-05,
+      "loss": 0.1939,
+      "step": 453
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.878292314335061e-05,
+      "loss": 0.1849,
+      "step": 454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.878022452504318e-05,
+      "loss": 0.3029,
+      "step": 455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.877752590673575e-05,
+      "loss": 0.2622,
+      "step": 456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8774827288428325e-05,
+      "loss": 0.4413,
+      "step": 457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.87721286701209e-05,
+      "loss": 0.3697,
+      "step": 458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8769430051813476e-05,
+      "loss": 0.2755,
+      "step": 459
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.876673143350605e-05,
+      "loss": 0.295,
+      "step": 460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.876403281519862e-05,
+      "loss": 0.186,
+      "step": 461
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.876133419689119e-05,
+      "loss": 0.3132,
+      "step": 462
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8758635578583764e-05,
+      "loss": 0.2367,
+      "step": 463
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8755936960276336e-05,
+      "loss": 0.1646,
+      "step": 464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8753238341968915e-05,
+      "loss": 0.2553,
+      "step": 465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.875053972366149e-05,
+      "loss": 0.2417,
+      "step": 466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.874784110535406e-05,
+      "loss": 0.457,
+      "step": 467
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.874514248704664e-05,
+      "loss": 0.6643,
+      "step": 468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.874244386873921e-05,
+      "loss": 0.1965,
+      "step": 469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8739745250431776e-05,
+      "loss": 0.1737,
+      "step": 470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8737046632124355e-05,
+      "loss": 0.2643,
+      "step": 471
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.873434801381693e-05,
+      "loss": 0.5332,
+      "step": 472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.87316493955095e-05,
+      "loss": 0.1467,
+      "step": 473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.872895077720208e-05,
+      "loss": 0.3124,
+      "step": 474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.872625215889465e-05,
+      "loss": 0.1777,
+      "step": 475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.872355354058722e-05,
+      "loss": 0.214,
+      "step": 476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8720854922279794e-05,
+      "loss": 0.26,
+      "step": 477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8718156303972366e-05,
+      "loss": 0.2831,
+      "step": 478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.871545768566494e-05,
+      "loss": 0.359,
+      "step": 479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.871275906735752e-05,
+      "loss": 0.4227,
+      "step": 480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.871006044905009e-05,
+      "loss": 0.3723,
+      "step": 481
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.870736183074266e-05,
+      "loss": 0.2832,
+      "step": 482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.870466321243523e-05,
+      "loss": 0.3642,
+      "step": 483
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8701964594127805e-05,
+      "loss": 0.3294,
+      "step": 484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.869926597582038e-05,
+      "loss": 0.1256,
+      "step": 485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8696567357512956e-05,
+      "loss": 0.3273,
+      "step": 486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.869386873920553e-05,
+      "loss": 0.443,
+      "step": 487
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.869117012089811e-05,
+      "loss": 0.1961,
+      "step": 488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.868847150259068e-05,
+      "loss": 0.3174,
+      "step": 489
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.868577288428325e-05,
+      "loss": 0.2417,
+      "step": 490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8683074265975823e-05,
+      "loss": 0.4424,
+      "step": 491
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8680375647668395e-05,
+      "loss": 0.2662,
+      "step": 492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.867767702936097e-05,
+      "loss": 0.321,
+      "step": 493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8674978411053546e-05,
+      "loss": 0.2994,
+      "step": 494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.867227979274612e-05,
+      "loss": 0.2048,
+      "step": 495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.866958117443869e-05,
+      "loss": 0.501,
+      "step": 496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.866688255613126e-05,
+      "loss": 0.2391,
+      "step": 497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8664183937823835e-05,
+      "loss": 0.2373,
+      "step": 498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.866148531951641e-05,
+      "loss": 0.2942,
+      "step": 499
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8658786701208986e-05,
+      "loss": 0.5687,
+      "step": 500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.865608808290156e-05,
+      "loss": 0.2318,
+      "step": 501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.865338946459413e-05,
+      "loss": 0.3261,
+      "step": 502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.865069084628671e-05,
+      "loss": 0.1648,
+      "step": 503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.864799222797928e-05,
+      "loss": 0.1505,
+      "step": 504
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8645293609671846e-05,
+      "loss": 0.4882,
+      "step": 505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8642594991364425e-05,
+      "loss": 0.3704,
+      "step": 506
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8639896373057e-05,
+      "loss": 0.3423,
+      "step": 507
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.863719775474957e-05,
+      "loss": 0.1567,
+      "step": 508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.863449913644215e-05,
+      "loss": 0.3229,
+      "step": 509
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.863180051813472e-05,
+      "loss": 0.4094,
+      "step": 510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.862910189982729e-05,
+      "loss": 0.1854,
+      "step": 511
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8626403281519864e-05,
+      "loss": 0.243,
+      "step": 512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8623704663212436e-05,
+      "loss": 0.4699,
+      "step": 513
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.862100604490501e-05,
+      "loss": 0.1853,
+      "step": 514
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.861830742659759e-05,
+      "loss": 0.3569,
+      "step": 515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.861560880829016e-05,
+      "loss": 0.3536,
+      "step": 516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.861291018998273e-05,
+      "loss": 0.2516,
+      "step": 517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.861021157167531e-05,
+      "loss": 0.3662,
+      "step": 518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8607512953367876e-05,
+      "loss": 0.1489,
+      "step": 519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.860481433506045e-05,
+      "loss": 0.1238,
+      "step": 520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8602115716753027e-05,
+      "loss": 0.2282,
+      "step": 521
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.85994170984456e-05,
+      "loss": 0.1584,
+      "step": 522
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.859671848013817e-05,
+      "loss": 0.5823,
+      "step": 523
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.859401986183075e-05,
+      "loss": 0.2749,
+      "step": 524
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.859132124352332e-05,
+      "loss": 0.2063,
+      "step": 525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8588622625215894e-05,
+      "loss": 0.5036,
+      "step": 526
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8585924006908466e-05,
+      "loss": 0.2017,
+      "step": 527
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.858322538860104e-05,
+      "loss": 0.1432,
+      "step": 528
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.858052677029361e-05,
+      "loss": 0.2872,
+      "step": 529
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.857782815198619e-05,
+      "loss": 0.1755,
+      "step": 530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.857512953367876e-05,
+      "loss": 0.2535,
+      "step": 531
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.857243091537133e-05,
+      "loss": 0.312,
+      "step": 532
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8569732297063905e-05,
+      "loss": 0.2658,
+      "step": 533
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.856703367875648e-05,
+      "loss": 0.3073,
+      "step": 534
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.856433506044905e-05,
+      "loss": 0.3161,
+      "step": 535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.856163644214163e-05,
+      "loss": 0.352,
+      "step": 536
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.85589378238342e-05,
+      "loss": 0.1435,
+      "step": 537
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.855623920552677e-05,
+      "loss": 0.1769,
+      "step": 538
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.855354058721935e-05,
+      "loss": 0.169,
+      "step": 539
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8550841968911917e-05,
+      "loss": 0.1568,
+      "step": 540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.854814335060449e-05,
+      "loss": 0.1478,
+      "step": 541
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.854544473229707e-05,
+      "loss": 0.137,
+      "step": 542
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.854274611398964e-05,
+      "loss": 0.3133,
+      "step": 543
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.854004749568221e-05,
+      "loss": 0.1769,
+      "step": 544
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.853734887737479e-05,
+      "loss": 0.141,
+      "step": 545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.853465025906736e-05,
+      "loss": 0.1645,
+      "step": 546
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8531951640759935e-05,
+      "loss": 0.4877,
+      "step": 547
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.852925302245251e-05,
+      "loss": 0.4348,
+      "step": 548
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.852655440414508e-05,
+      "loss": 0.526,
+      "step": 549
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.852385578583765e-05,
+      "loss": 0.2374,
+      "step": 550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.852115716753023e-05,
+      "loss": 0.1396,
+      "step": 551
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.85184585492228e-05,
+      "loss": 0.2896,
+      "step": 552
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8515759930915374e-05,
+      "loss": 0.1478,
+      "step": 553
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8513061312607946e-05,
+      "loss": 0.4597,
+      "step": 554
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.851036269430052e-05,
+      "loss": 0.4028,
+      "step": 555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.850766407599309e-05,
+      "loss": 0.1206,
+      "step": 556
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.850496545768567e-05,
+      "loss": 0.2342,
+      "step": 557
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.850226683937824e-05,
+      "loss": 0.2585,
+      "step": 558
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.849956822107081e-05,
+      "loss": 0.2589,
+      "step": 559
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.849686960276339e-05,
+      "loss": 0.3267,
+      "step": 560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8494170984455964e-05,
+      "loss": 0.1422,
+      "step": 561
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.849147236614853e-05,
+      "loss": 0.2434,
+      "step": 562
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.848877374784111e-05,
+      "loss": 0.6627,
+      "step": 563
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.848607512953368e-05,
+      "loss": 0.3007,
+      "step": 564
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.848337651122625e-05,
+      "loss": 0.2337,
+      "step": 565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.848067789291883e-05,
+      "loss": 0.1683,
+      "step": 566
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8477979274611403e-05,
+      "loss": 0.2028,
+      "step": 567
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8475280656303976e-05,
+      "loss": 0.1344,
+      "step": 568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.847258203799655e-05,
+      "loss": 0.5056,
+      "step": 569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.846988341968912e-05,
+      "loss": 0.2922,
+      "step": 570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.846718480138169e-05,
+      "loss": 0.1417,
+      "step": 571
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.846448618307427e-05,
+      "loss": 0.3975,
+      "step": 572
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.846178756476684e-05,
+      "loss": 0.1534,
+      "step": 573
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8459088946459415e-05,
+      "loss": 0.2176,
+      "step": 574
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8456390328151994e-05,
+      "loss": 0.122,
+      "step": 575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.845369170984456e-05,
+      "loss": 0.1266,
+      "step": 576
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.845099309153713e-05,
+      "loss": 0.4728,
+      "step": 577
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.844829447322971e-05,
+      "loss": 0.1884,
+      "step": 578
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.844559585492228e-05,
+      "loss": 0.1807,
+      "step": 579
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8442897236614854e-05,
+      "loss": 0.2498,
+      "step": 580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.844019861830743e-05,
+      "loss": 0.2415,
+      "step": 581
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8437500000000005e-05,
+      "loss": 0.1448,
+      "step": 582
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.843480138169258e-05,
+      "loss": 0.3641,
+      "step": 583
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.843210276338515e-05,
+      "loss": 0.2134,
+      "step": 584
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.842940414507772e-05,
+      "loss": 0.2932,
+      "step": 585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8426705526770293e-05,
+      "loss": 0.1387,
+      "step": 586
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.842400690846287e-05,
+      "loss": 0.2817,
+      "step": 587
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8421308290155444e-05,
+      "loss": 0.2763,
+      "step": 588
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8418609671848016e-05,
+      "loss": 0.5508,
+      "step": 589
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.841591105354059e-05,
+      "loss": 0.1837,
+      "step": 590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.841321243523316e-05,
+      "loss": 0.3023,
+      "step": 591
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.841051381692573e-05,
+      "loss": 0.3015,
+      "step": 592
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.840781519861831e-05,
+      "loss": 0.3421,
+      "step": 593
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8405116580310884e-05,
+      "loss": 0.1346,
+      "step": 594
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8402417962003456e-05,
+      "loss": 0.2419,
+      "step": 595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8399719343696035e-05,
+      "loss": 0.3406,
+      "step": 596
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.83970207253886e-05,
+      "loss": 0.2937,
+      "step": 597
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.839432210708117e-05,
+      "loss": 0.1535,
+      "step": 598
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.839162348877375e-05,
+      "loss": 0.2321,
+      "step": 599
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.838892487046632e-05,
+      "loss": 0.357,
+      "step": 600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8386226252158895e-05,
+      "loss": 0.4984,
+      "step": 601
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8383527633851474e-05,
+      "loss": 0.2466,
+      "step": 602
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8380829015544046e-05,
+      "loss": 0.1803,
+      "step": 603
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.837813039723662e-05,
+      "loss": 0.4384,
+      "step": 604
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.837543177892919e-05,
+      "loss": 0.1188,
+      "step": 605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.837273316062176e-05,
+      "loss": 0.312,
+      "step": 606
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8370034542314334e-05,
+      "loss": 0.3569,
+      "step": 607
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.836733592400691e-05,
+      "loss": 0.2786,
+      "step": 608
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8364637305699485e-05,
+      "loss": 0.1341,
+      "step": 609
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.836193868739206e-05,
+      "loss": 0.375,
+      "step": 610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.835924006908463e-05,
+      "loss": 0.249,
+      "step": 611
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.83565414507772e-05,
+      "loss": 0.1723,
+      "step": 612
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8353842832469774e-05,
+      "loss": 0.1935,
+      "step": 613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.835114421416235e-05,
+      "loss": 0.5248,
+      "step": 614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8348445595854925e-05,
+      "loss": 0.1871,
+      "step": 615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8345746977547497e-05,
+      "loss": 0.1313,
+      "step": 616
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8343048359240075e-05,
+      "loss": 0.4445,
+      "step": 617
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.834034974093265e-05,
+      "loss": 0.3744,
+      "step": 618
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.833765112262521e-05,
+      "loss": 0.3672,
+      "step": 619
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.833495250431779e-05,
+      "loss": 0.2612,
+      "step": 620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8332253886010364e-05,
+      "loss": 0.1509,
+      "step": 621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8329555267702936e-05,
+      "loss": 0.188,
+      "step": 622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8326856649395515e-05,
+      "loss": 0.2087,
+      "step": 623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.832415803108809e-05,
+      "loss": 0.5066,
+      "step": 624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.832145941278066e-05,
+      "loss": 0.1374,
+      "step": 625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.831876079447323e-05,
+      "loss": 0.1872,
+      "step": 626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.83160621761658e-05,
+      "loss": 0.2183,
+      "step": 627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8313363557858375e-05,
+      "loss": 0.2609,
+      "step": 628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8310664939550954e-05,
+      "loss": 0.1648,
+      "step": 629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8307966321243526e-05,
+      "loss": 0.4421,
+      "step": 630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.83052677029361e-05,
+      "loss": 0.171,
+      "step": 631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.830256908462868e-05,
+      "loss": 0.1526,
+      "step": 632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.829987046632124e-05,
+      "loss": 0.2547,
+      "step": 633
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8297171848013814e-05,
+      "loss": 0.2247,
+      "step": 634
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.829447322970639e-05,
+      "loss": 0.1771,
+      "step": 635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8291774611398965e-05,
+      "loss": 0.1864,
+      "step": 636
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.828907599309154e-05,
+      "loss": 0.2058,
+      "step": 637
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8286377374784116e-05,
+      "loss": 0.2958,
+      "step": 638
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.828367875647669e-05,
+      "loss": 0.6076,
+      "step": 639
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.828098013816926e-05,
+      "loss": 0.4117,
+      "step": 640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.827828151986183e-05,
+      "loss": 0.1557,
+      "step": 641
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8275582901554405e-05,
+      "loss": 0.6207,
+      "step": 642
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.827288428324698e-05,
+      "loss": 0.1913,
+      "step": 643
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8270185664939556e-05,
+      "loss": 0.559,
+      "step": 644
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.826748704663213e-05,
+      "loss": 0.1703,
+      "step": 645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.82647884283247e-05,
+      "loss": 0.3723,
+      "step": 646
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.826208981001727e-05,
+      "loss": 0.2646,
+      "step": 647
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8259391191709844e-05,
+      "loss": 0.1377,
+      "step": 648
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8256692573402416e-05,
+      "loss": 0.3667,
+      "step": 649
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8253993955094995e-05,
+      "loss": 0.4025,
+      "step": 650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.825129533678757e-05,
+      "loss": 0.1768,
+      "step": 651
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.824859671848014e-05,
+      "loss": 0.1678,
+      "step": 652
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.824589810017272e-05,
+      "loss": 0.1591,
+      "step": 653
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.824319948186528e-05,
+      "loss": 0.1722,
+      "step": 654
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8240500863557855e-05,
+      "loss": 0.4185,
+      "step": 655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8237802245250434e-05,
+      "loss": 0.2731,
+      "step": 656
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8235103626943006e-05,
+      "loss": 0.2146,
+      "step": 657
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.823240500863558e-05,
+      "loss": 0.3033,
+      "step": 658
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.822970639032816e-05,
+      "loss": 0.158,
+      "step": 659
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.822700777202073e-05,
+      "loss": 0.1905,
+      "step": 660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.82243091537133e-05,
+      "loss": 0.1523,
+      "step": 661
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8221610535405873e-05,
+      "loss": 0.2925,
+      "step": 662
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8218911917098446e-05,
+      "loss": 0.2789,
+      "step": 663
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.821621329879102e-05,
+      "loss": 0.2434,
+      "step": 664
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8213514680483597e-05,
+      "loss": 0.1792,
+      "step": 665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.821081606217617e-05,
+      "loss": 0.4847,
+      "step": 666
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.820811744386874e-05,
+      "loss": 0.24,
+      "step": 667
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.820541882556131e-05,
+      "loss": 0.1618,
+      "step": 668
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8202720207253885e-05,
+      "loss": 0.1695,
+      "step": 669
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.820002158894646e-05,
+      "loss": 0.2838,
+      "step": 670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8197322970639036e-05,
+      "loss": 0.1316,
+      "step": 671
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.819462435233161e-05,
+      "loss": 0.2524,
+      "step": 672
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.819192573402418e-05,
+      "loss": 0.1455,
+      "step": 673
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.818922711571676e-05,
+      "loss": 0.3936,
+      "step": 674
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.818652849740933e-05,
+      "loss": 0.2133,
+      "step": 675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8183829879101896e-05,
+      "loss": 0.1642,
+      "step": 676
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8181131260794475e-05,
+      "loss": 0.2128,
+      "step": 677
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.817843264248705e-05,
+      "loss": 0.1763,
+      "step": 678
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8175734024179626e-05,
+      "loss": 0.3044,
+      "step": 679
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.81730354058722e-05,
+      "loss": 0.2274,
+      "step": 680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.817033678756477e-05,
+      "loss": 0.2645,
+      "step": 681
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.816763816925734e-05,
+      "loss": 0.2068,
+      "step": 682
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8164939550949914e-05,
+      "loss": 0.2598,
+      "step": 683
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8162240932642486e-05,
+      "loss": 0.5971,
+      "step": 684
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8159542314335065e-05,
+      "loss": 0.1513,
+      "step": 685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.815684369602764e-05,
+      "loss": 0.3394,
+      "step": 686
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.815414507772021e-05,
+      "loss": 0.2444,
+      "step": 687
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.815144645941279e-05,
+      "loss": 0.152,
+      "step": 688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.814874784110536e-05,
+      "loss": 0.2769,
+      "step": 689
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8146049222797926e-05,
+      "loss": 0.2751,
+      "step": 690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8143350604490505e-05,
+      "loss": 0.1606,
+      "step": 691
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.814065198618308e-05,
+      "loss": 0.1427,
+      "step": 692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.813795336787565e-05,
+      "loss": 0.1804,
+      "step": 693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.813525474956823e-05,
+      "loss": 0.1868,
+      "step": 694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.81325561312608e-05,
+      "loss": 0.31,
+      "step": 695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.812985751295337e-05,
+      "loss": 0.2077,
+      "step": 696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8127158894645944e-05,
+      "loss": 0.2257,
+      "step": 697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8124460276338516e-05,
+      "loss": 0.1078,
+      "step": 698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.812176165803109e-05,
+      "loss": 0.1643,
+      "step": 699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.811906303972367e-05,
+      "loss": 0.2566,
+      "step": 700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.811636442141624e-05,
+      "loss": 0.2341,
+      "step": 701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.811366580310881e-05,
+      "loss": 0.224,
+      "step": 702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.811096718480138e-05,
+      "loss": 0.1694,
+      "step": 703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8108268566493955e-05,
+      "loss": 0.202,
+      "step": 704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.810556994818653e-05,
+      "loss": 0.2082,
+      "step": 705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8102871329879106e-05,
+      "loss": 0.3614,
+      "step": 706
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.810017271157168e-05,
+      "loss": 0.3875,
+      "step": 707
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.809747409326425e-05,
+      "loss": 0.1143,
+      "step": 708
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.809477547495683e-05,
+      "loss": 0.2449,
+      "step": 709
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.80920768566494e-05,
+      "loss": 0.2078,
+      "step": 710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.808937823834197e-05,
+      "loss": 0.172,
+      "step": 711
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8086679620034545e-05,
+      "loss": 0.1808,
+      "step": 712
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.808398100172712e-05,
+      "loss": 0.2613,
+      "step": 713
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.808128238341969e-05,
+      "loss": 0.336,
+      "step": 714
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.807858376511227e-05,
+      "loss": 0.5024,
+      "step": 715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.807588514680484e-05,
+      "loss": 0.1546,
+      "step": 716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.807318652849741e-05,
+      "loss": 0.2166,
+      "step": 717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8070487910189985e-05,
+      "loss": 0.2098,
+      "step": 718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.806778929188256e-05,
+      "loss": 0.2942,
+      "step": 719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.806509067357513e-05,
+      "loss": 0.1399,
+      "step": 720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.806239205526771e-05,
+      "loss": 0.1202,
+      "step": 721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.805969343696028e-05,
+      "loss": 0.1244,
+      "step": 722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.805699481865285e-05,
+      "loss": 0.1597,
+      "step": 723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.805429620034543e-05,
+      "loss": 0.24,
+      "step": 724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8051597582037996e-05,
+      "loss": 0.372,
+      "step": 725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.804889896373057e-05,
+      "loss": 0.2956,
+      "step": 726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.804620034542315e-05,
+      "loss": 0.5232,
+      "step": 727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.804350172711572e-05,
+      "loss": 0.1927,
+      "step": 728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.804080310880829e-05,
+      "loss": 0.1745,
+      "step": 729
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.803810449050087e-05,
+      "loss": 0.1662,
+      "step": 730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.803540587219344e-05,
+      "loss": 0.1932,
+      "step": 731
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8032707253886014e-05,
+      "loss": 0.4138,
+      "step": 732
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8030008635578586e-05,
+      "loss": 0.2657,
+      "step": 733
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.802731001727116e-05,
+      "loss": 0.3148,
+      "step": 734
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.802461139896373e-05,
+      "loss": 0.1811,
+      "step": 735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.802191278065631e-05,
+      "loss": 0.258,
+      "step": 736
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.801921416234888e-05,
+      "loss": 0.2835,
+      "step": 737
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8016515544041454e-05,
+      "loss": 0.2763,
+      "step": 738
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8013816925734026e-05,
+      "loss": 0.2545,
+      "step": 739
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.80111183074266e-05,
+      "loss": 0.1949,
+      "step": 740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.800841968911917e-05,
+      "loss": 0.1809,
+      "step": 741
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.800572107081175e-05,
+      "loss": 0.4839,
+      "step": 742
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.800302245250432e-05,
+      "loss": 0.1219,
+      "step": 743
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.800032383419689e-05,
+      "loss": 0.1472,
+      "step": 744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.799762521588947e-05,
+      "loss": 0.1772,
+      "step": 745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7994926597582044e-05,
+      "loss": 0.1139,
+      "step": 746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.799222797927461e-05,
+      "loss": 0.1285,
+      "step": 747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.798952936096719e-05,
+      "loss": 0.3057,
+      "step": 748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.798683074265976e-05,
+      "loss": 0.2671,
+      "step": 749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.798413212435233e-05,
+      "loss": 0.3711,
+      "step": 750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.798143350604491e-05,
+      "loss": 0.4105,
+      "step": 751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.797873488773748e-05,
+      "loss": 0.1829,
+      "step": 752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7976036269430055e-05,
+      "loss": 0.252,
+      "step": 753
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.797333765112263e-05,
+      "loss": 0.1593,
+      "step": 754
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.79706390328152e-05,
+      "loss": 0.2544,
+      "step": 755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.796794041450777e-05,
+      "loss": 0.2362,
+      "step": 756
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.796524179620035e-05,
+      "loss": 0.4601,
+      "step": 757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.796254317789292e-05,
+      "loss": 0.272,
+      "step": 758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7959844559585494e-05,
+      "loss": 0.3134,
+      "step": 759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.795714594127807e-05,
+      "loss": 0.1449,
+      "step": 760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.795444732297064e-05,
+      "loss": 0.2678,
+      "step": 761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.795174870466321e-05,
+      "loss": 0.2884,
+      "step": 762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.794905008635579e-05,
+      "loss": 0.2897,
+      "step": 763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.794635146804836e-05,
+      "loss": 0.2692,
+      "step": 764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7943652849740934e-05,
+      "loss": 0.2423,
+      "step": 765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.794095423143351e-05,
+      "loss": 0.2388,
+      "step": 766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7938255613126085e-05,
+      "loss": 0.1949,
+      "step": 767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.793555699481866e-05,
+      "loss": 0.284,
+      "step": 768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.793285837651123e-05,
+      "loss": 0.4397,
+      "step": 769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.79301597582038e-05,
+      "loss": 0.115,
+      "step": 770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.792746113989637e-05,
+      "loss": 0.3573,
+      "step": 771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.792476252158895e-05,
+      "loss": 0.2255,
+      "step": 772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7922063903281524e-05,
+      "loss": 0.6114,
+      "step": 773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7919365284974096e-05,
+      "loss": 0.2014,
+      "step": 774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.791666666666667e-05,
+      "loss": 0.3803,
+      "step": 775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.791396804835924e-05,
+      "loss": 0.1792,
+      "step": 776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.791126943005181e-05,
+      "loss": 0.2213,
+      "step": 777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.790857081174439e-05,
+      "loss": 0.4436,
+      "step": 778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.790587219343696e-05,
+      "loss": 0.2052,
+      "step": 779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7903173575129535e-05,
+      "loss": 0.2352,
+      "step": 780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7900474956822114e-05,
+      "loss": 0.1085,
+      "step": 781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.789777633851468e-05,
+      "loss": 0.3531,
+      "step": 782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.789507772020725e-05,
+      "loss": 0.1486,
+      "step": 783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.789237910189983e-05,
+      "loss": 0.1229,
+      "step": 784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.78896804835924e-05,
+      "loss": 0.264,
+      "step": 785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7886981865284975e-05,
+      "loss": 0.2011,
+      "step": 786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7884283246977553e-05,
+      "loss": 0.1358,
+      "step": 787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7881584628670126e-05,
+      "loss": 0.1518,
+      "step": 788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.78788860103627e-05,
+      "loss": 0.2676,
+      "step": 789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.787618739205527e-05,
+      "loss": 0.1469,
+      "step": 790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.787348877374784e-05,
+      "loss": 0.252,
+      "step": 791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7870790155440414e-05,
+      "loss": 0.2504,
+      "step": 792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.786809153713299e-05,
+      "loss": 0.2457,
+      "step": 793
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7865392918825565e-05,
+      "loss": 0.1319,
+      "step": 794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.786269430051814e-05,
+      "loss": 0.1356,
+      "step": 795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.785999568221071e-05,
+      "loss": 0.1669,
+      "step": 796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.785729706390328e-05,
+      "loss": 0.1628,
+      "step": 797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.785459844559585e-05,
+      "loss": 0.1773,
+      "step": 798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.785189982728843e-05,
+      "loss": 0.2189,
+      "step": 799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7849201208981004e-05,
+      "loss": 0.5451,
+      "step": 800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7846502590673576e-05,
+      "loss": 0.2617,
+      "step": 801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7843803972366155e-05,
+      "loss": 0.1183,
+      "step": 802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.784110535405873e-05,
+      "loss": 0.2436,
+      "step": 803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.783840673575129e-05,
+      "loss": 0.2373,
+      "step": 804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.783570811744387e-05,
+      "loss": 0.1526,
+      "step": 805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7833009499136443e-05,
+      "loss": 0.3848,
+      "step": 806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7830310880829015e-05,
+      "loss": 0.1346,
+      "step": 807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7827612262521594e-05,
+      "loss": 0.1315,
+      "step": 808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7824913644214166e-05,
+      "loss": 0.109,
+      "step": 809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.782221502590674e-05,
+      "loss": 0.1132,
+      "step": 810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.781951640759931e-05,
+      "loss": 0.1629,
+      "step": 811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.781681778929188e-05,
+      "loss": 0.103,
+      "step": 812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7814119170984455e-05,
+      "loss": 0.3502,
+      "step": 813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7811420552677034e-05,
+      "loss": 0.3378,
+      "step": 814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7808721934369606e-05,
+      "loss": 0.2388,
+      "step": 815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.780602331606218e-05,
+      "loss": 0.1249,
+      "step": 816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.780332469775476e-05,
+      "loss": 0.1718,
+      "step": 817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.780062607944732e-05,
+      "loss": 0.4539,
+      "step": 818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7797927461139894e-05,
+      "loss": 0.3669,
+      "step": 819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.779522884283247e-05,
+      "loss": 0.2729,
+      "step": 820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7792530224525045e-05,
+      "loss": 0.203,
+      "step": 821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.778983160621762e-05,
+      "loss": 0.3087,
+      "step": 822
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7787132987910196e-05,
+      "loss": 0.2143,
+      "step": 823
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.778443436960277e-05,
+      "loss": 0.1427,
+      "step": 824
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.778173575129534e-05,
+      "loss": 0.1334,
+      "step": 825
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.777903713298791e-05,
+      "loss": 0.3239,
+      "step": 826
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7776338514680484e-05,
+      "loss": 0.3054,
+      "step": 827
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7773639896373056e-05,
+      "loss": 0.1153,
+      "step": 828
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7770941278065635e-05,
+      "loss": 0.2161,
+      "step": 829
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.776824265975821e-05,
+      "loss": 0.4448,
+      "step": 830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.776554404145078e-05,
+      "loss": 0.2438,
+      "step": 831
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.776284542314335e-05,
+      "loss": 0.1329,
+      "step": 832
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7760146804835924e-05,
+      "loss": 0.1822,
+      "step": 833
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7757448186528496e-05,
+      "loss": 0.2352,
+      "step": 834
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7754749568221075e-05,
+      "loss": 0.245,
+      "step": 835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7752050949913647e-05,
+      "loss": 0.2542,
+      "step": 836
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.774935233160622e-05,
+      "loss": 0.543,
+      "step": 837
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.77466537132988e-05,
+      "loss": 0.1966,
+      "step": 838
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.774395509499136e-05,
+      "loss": 0.2142,
+      "step": 839
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7741256476683935e-05,
+      "loss": 0.1591,
+      "step": 840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7738557858376514e-05,
+      "loss": 0.2241,
+      "step": 841
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7735859240069086e-05,
+      "loss": 0.4073,
+      "step": 842
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.773316062176166e-05,
+      "loss": 0.3081,
+      "step": 843
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.773046200345424e-05,
+      "loss": 0.3541,
+      "step": 844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.772776338514681e-05,
+      "loss": 0.5678,
+      "step": 845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.772506476683938e-05,
+      "loss": 0.1069,
+      "step": 846
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.772236614853195e-05,
+      "loss": 0.2701,
+      "step": 847
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7719667530224525e-05,
+      "loss": 0.3149,
+      "step": 848
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.77169689119171e-05,
+      "loss": 0.6228,
+      "step": 849
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7714270293609676e-05,
+      "loss": 0.1696,
+      "step": 850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.771157167530225e-05,
+      "loss": 0.6047,
+      "step": 851
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.770887305699482e-05,
+      "loss": 0.1445,
+      "step": 852
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.770617443868739e-05,
+      "loss": 0.3049,
+      "step": 853
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7703475820379964e-05,
+      "loss": 0.8024,
+      "step": 854
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7700777202072537e-05,
+      "loss": 0.2259,
+      "step": 855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7698078583765115e-05,
+      "loss": 0.2629,
+      "step": 856
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.769537996545769e-05,
+      "loss": 0.193,
+      "step": 857
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.769268134715026e-05,
+      "loss": 0.2376,
+      "step": 858
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.768998272884284e-05,
+      "loss": 0.1361,
+      "step": 859
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.768728411053541e-05,
+      "loss": 0.2759,
+      "step": 860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7684585492227976e-05,
+      "loss": 0.1879,
+      "step": 861
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7681886873920555e-05,
+      "loss": 0.3597,
+      "step": 862
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.767918825561313e-05,
+      "loss": 0.1493,
+      "step": 863
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.76764896373057e-05,
+      "loss": 0.3159,
+      "step": 864
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.767379101899828e-05,
+      "loss": 0.6277,
+      "step": 865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.767109240069085e-05,
+      "loss": 0.1165,
+      "step": 866
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.766839378238342e-05,
+      "loss": 0.258,
+      "step": 867
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7665695164075994e-05,
+      "loss": 0.3776,
+      "step": 868
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7662996545768566e-05,
+      "loss": 0.1559,
+      "step": 869
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.766029792746114e-05,
+      "loss": 0.3251,
+      "step": 870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.765759930915372e-05,
+      "loss": 0.1168,
+      "step": 871
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.765490069084629e-05,
+      "loss": 0.2009,
+      "step": 872
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.765220207253887e-05,
+      "loss": 0.2374,
+      "step": 873
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.764950345423144e-05,
+      "loss": 0.348,
+      "step": 874
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7646804835924005e-05,
+      "loss": 0.4005,
+      "step": 875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7644106217616584e-05,
+      "loss": 0.1538,
+      "step": 876
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7641407599309156e-05,
+      "loss": 0.3093,
+      "step": 877
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.763870898100173e-05,
+      "loss": 0.1986,
+      "step": 878
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.763601036269431e-05,
+      "loss": 0.1193,
+      "step": 879
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.763331174438688e-05,
+      "loss": 0.1462,
+      "step": 880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.763061312607945e-05,
+      "loss": 0.2624,
+      "step": 881
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7627914507772023e-05,
+      "loss": 0.1821,
+      "step": 882
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7625215889464596e-05,
+      "loss": 0.1198,
+      "step": 883
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.762251727115717e-05,
+      "loss": 0.4313,
+      "step": 884
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7619818652849746e-05,
+      "loss": 0.4892,
+      "step": 885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.761712003454232e-05,
+      "loss": 0.2401,
+      "step": 886
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.761442141623489e-05,
+      "loss": 0.405,
+      "step": 887
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.761172279792746e-05,
+      "loss": 0.1645,
+      "step": 888
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7609024179620035e-05,
+      "loss": 0.3553,
+      "step": 889
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.760632556131261e-05,
+      "loss": 0.3644,
+      "step": 890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7603626943005186e-05,
+      "loss": 0.1059,
+      "step": 891
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.760092832469776e-05,
+      "loss": 0.3404,
+      "step": 892
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.759822970639033e-05,
+      "loss": 0.1462,
+      "step": 893
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.759553108808291e-05,
+      "loss": 0.2738,
+      "step": 894
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.759283246977548e-05,
+      "loss": 0.1938,
+      "step": 895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7590133851468046e-05,
+      "loss": 0.1157,
+      "step": 896
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7587435233160625e-05,
+      "loss": 0.1788,
+      "step": 897
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.75847366148532e-05,
+      "loss": 0.1235,
+      "step": 898
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.758203799654577e-05,
+      "loss": 0.3233,
+      "step": 899
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.757933937823835e-05,
+      "loss": 0.2296,
+      "step": 900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.757664075993092e-05,
+      "loss": 0.1142,
+      "step": 901
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.757394214162349e-05,
+      "loss": 0.3666,
+      "step": 902
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7571243523316064e-05,
+      "loss": 0.2544,
+      "step": 903
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7568544905008636e-05,
+      "loss": 0.1201,
+      "step": 904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.756584628670121e-05,
+      "loss": 0.1328,
+      "step": 905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.756314766839379e-05,
+      "loss": 0.2038,
+      "step": 906
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.756044905008636e-05,
+      "loss": 0.6083,
+      "step": 907
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.755775043177893e-05,
+      "loss": 0.1477,
+      "step": 908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.755505181347151e-05,
+      "loss": 0.1538,
+      "step": 909
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7552353195164076e-05,
+      "loss": 0.0875,
+      "step": 910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.754965457685665e-05,
+      "loss": 0.1216,
+      "step": 911
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.754695595854923e-05,
+      "loss": 0.1766,
+      "step": 912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.75442573402418e-05,
+      "loss": 0.1474,
+      "step": 913
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.754155872193437e-05,
+      "loss": 0.0993,
+      "step": 914
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.753886010362695e-05,
+      "loss": 0.1793,
+      "step": 915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.753616148531952e-05,
+      "loss": 0.2019,
+      "step": 916
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7533462867012094e-05,
+      "loss": 0.3023,
+      "step": 917
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7530764248704666e-05,
+      "loss": 0.1793,
+      "step": 918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.752806563039724e-05,
+      "loss": 0.1843,
+      "step": 919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.752536701208981e-05,
+      "loss": 0.4164,
+      "step": 920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.752266839378239e-05,
+      "loss": 0.4532,
+      "step": 921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.751996977547496e-05,
+      "loss": 0.4834,
+      "step": 922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.751727115716753e-05,
+      "loss": 0.4773,
+      "step": 923
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7514572538860105e-05,
+      "loss": 0.123,
+      "step": 924
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.751187392055268e-05,
+      "loss": 0.1112,
+      "step": 925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.750917530224525e-05,
+      "loss": 0.3669,
+      "step": 926
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.750647668393783e-05,
+      "loss": 0.2545,
+      "step": 927
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.75037780656304e-05,
+      "loss": 0.2329,
+      "step": 928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.750107944732297e-05,
+      "loss": 0.1161,
+      "step": 929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.749838082901555e-05,
+      "loss": 0.1924,
+      "step": 930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.749568221070812e-05,
+      "loss": 0.4284,
+      "step": 931
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.749298359240069e-05,
+      "loss": 0.3218,
+      "step": 932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.749028497409327e-05,
+      "loss": 0.1245,
+      "step": 933
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.748758635578584e-05,
+      "loss": 0.301,
+      "step": 934
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.748488773747841e-05,
+      "loss": 0.3472,
+      "step": 935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.748218911917099e-05,
+      "loss": 0.1129,
+      "step": 936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.747949050086356e-05,
+      "loss": 0.0985,
+      "step": 937
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7476791882556135e-05,
+      "loss": 0.1969,
+      "step": 938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.747409326424871e-05,
+      "loss": 0.1469,
+      "step": 939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.747139464594128e-05,
+      "loss": 0.1533,
+      "step": 940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.746869602763385e-05,
+      "loss": 0.1673,
+      "step": 941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.746599740932643e-05,
+      "loss": 0.1587,
+      "step": 942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7463298791019e-05,
+      "loss": 0.2929,
+      "step": 943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7460600172711574e-05,
+      "loss": 0.1882,
+      "step": 944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7457901554404146e-05,
+      "loss": 0.1767,
+      "step": 945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.745520293609672e-05,
+      "loss": 0.1152,
+      "step": 946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.745250431778929e-05,
+      "loss": 0.2075,
+      "step": 947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.744980569948187e-05,
+      "loss": 0.2919,
+      "step": 948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.744710708117444e-05,
+      "loss": 0.1314,
+      "step": 949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.744440846286701e-05,
+      "loss": 0.2328,
+      "step": 950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.744170984455959e-05,
+      "loss": 0.1609,
+      "step": 951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7439011226252164e-05,
+      "loss": 0.1134,
+      "step": 952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.743631260794473e-05,
+      "loss": 0.1205,
+      "step": 953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.743361398963731e-05,
+      "loss": 0.3327,
+      "step": 954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.743091537132988e-05,
+      "loss": 0.2489,
+      "step": 955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.742821675302245e-05,
+      "loss": 0.201,
+      "step": 956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.742551813471503e-05,
+      "loss": 0.363,
+      "step": 957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7422819516407604e-05,
+      "loss": 0.3227,
+      "step": 958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7420120898100176e-05,
+      "loss": 0.3339,
+      "step": 959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.741742227979275e-05,
+      "loss": 0.0977,
+      "step": 960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.741472366148532e-05,
+      "loss": 0.2744,
+      "step": 961
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.741202504317789e-05,
+      "loss": 0.1302,
+      "step": 962
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.740932642487047e-05,
+      "loss": 0.3997,
+      "step": 963
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.740662780656304e-05,
+      "loss": 0.4567,
+      "step": 964
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7403929188255615e-05,
+      "loss": 0.3438,
+      "step": 965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7401230569948194e-05,
+      "loss": 0.5745,
+      "step": 966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.739853195164076e-05,
+      "loss": 0.4729,
+      "step": 967
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.739583333333333e-05,
+      "loss": 0.151,
+      "step": 968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.739313471502591e-05,
+      "loss": 0.1757,
+      "step": 969
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.739043609671848e-05,
+      "loss": 0.1787,
+      "step": 970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7387737478411054e-05,
+      "loss": 0.1153,
+      "step": 971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.738503886010363e-05,
+      "loss": 0.3133,
+      "step": 972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7382340241796205e-05,
+      "loss": 0.3056,
+      "step": 973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.737964162348878e-05,
+      "loss": 0.1189,
+      "step": 974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.737694300518135e-05,
+      "loss": 0.0954,
+      "step": 975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.737424438687392e-05,
+      "loss": 0.3292,
+      "step": 976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7371545768566493e-05,
+      "loss": 0.4074,
+      "step": 977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.736884715025907e-05,
+      "loss": 0.1229,
+      "step": 978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7366148531951644e-05,
+      "loss": 0.3046,
+      "step": 979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7363449913644217e-05,
+      "loss": 0.2254,
+      "step": 980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.736075129533679e-05,
+      "loss": 0.2949,
+      "step": 981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.735805267702936e-05,
+      "loss": 0.4875,
+      "step": 982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.735535405872193e-05,
+      "loss": 0.2033,
+      "step": 983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.735265544041451e-05,
+      "loss": 0.1268,
+      "step": 984
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7349956822107084e-05,
+      "loss": 0.4902,
+      "step": 985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7347258203799656e-05,
+      "loss": 0.2251,
+      "step": 986
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7344559585492235e-05,
+      "loss": 0.1225,
+      "step": 987
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.734186096718481e-05,
+      "loss": 0.3748,
+      "step": 988
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.733916234887737e-05,
+      "loss": 0.2561,
+      "step": 989
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.733646373056995e-05,
+      "loss": 0.207,
+      "step": 990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.733376511226252e-05,
+      "loss": 0.1102,
+      "step": 991
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7331066493955095e-05,
+      "loss": 0.2215,
+      "step": 992
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7328367875647674e-05,
+      "loss": 0.4083,
+      "step": 993
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7325669257340246e-05,
+      "loss": 0.1218,
+      "step": 994
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.732297063903282e-05,
+      "loss": 0.2918,
+      "step": 995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.732027202072539e-05,
+      "loss": 0.5208,
+      "step": 996
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.731757340241796e-05,
+      "loss": 0.1181,
+      "step": 997
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7314874784110534e-05,
+      "loss": 0.1656,
+      "step": 998
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.731217616580311e-05,
+      "loss": 0.2031,
+      "step": 999
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7309477547495685e-05,
+      "loss": 0.1035,
+      "step": 1000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.730677892918826e-05,
+      "loss": 0.4487,
+      "step": 1001
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.730408031088083e-05,
+      "loss": 0.1433,
+      "step": 1002
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.73013816925734e-05,
+      "loss": 0.1075,
+      "step": 1003
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7298683074265974e-05,
+      "loss": 0.2364,
+      "step": 1004
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.729598445595855e-05,
+      "loss": 0.1466,
+      "step": 1005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7293285837651125e-05,
+      "loss": 0.1358,
+      "step": 1006
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.72905872193437e-05,
+      "loss": 0.1934,
+      "step": 1007
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7287888601036276e-05,
+      "loss": 0.5573,
+      "step": 1008
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.728518998272885e-05,
+      "loss": 0.4358,
+      "step": 1009
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.728249136442141e-05,
+      "loss": 0.1018,
+      "step": 1010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.727979274611399e-05,
+      "loss": 0.2531,
+      "step": 1011
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7277094127806564e-05,
+      "loss": 0.2999,
+      "step": 1012
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7274395509499136e-05,
+      "loss": 0.2224,
+      "step": 1013
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7271696891191715e-05,
+      "loss": 0.1197,
+      "step": 1014
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.726899827288429e-05,
+      "loss": 0.1025,
+      "step": 1015
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.726629965457686e-05,
+      "loss": 0.1187,
+      "step": 1016
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.726360103626943e-05,
+      "loss": 0.3242,
+      "step": 1017
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7260902417962e-05,
+      "loss": 0.2649,
+      "step": 1018
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7258203799654575e-05,
+      "loss": 0.1883,
+      "step": 1019
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7255505181347154e-05,
+      "loss": 0.1187,
+      "step": 1020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7252806563039726e-05,
+      "loss": 0.2276,
+      "step": 1021
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.72501079447323e-05,
+      "loss": 0.2905,
+      "step": 1022
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.724740932642488e-05,
+      "loss": 0.2203,
+      "step": 1023
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.724471070811744e-05,
+      "loss": 0.2547,
+      "step": 1024
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7242012089810015e-05,
+      "loss": 0.341,
+      "step": 1025
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.723931347150259e-05,
+      "loss": 0.4851,
+      "step": 1026
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7236614853195165e-05,
+      "loss": 0.2214,
+      "step": 1027
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.723391623488774e-05,
+      "loss": 0.2308,
+      "step": 1028
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7231217616580316e-05,
+      "loss": 0.4915,
+      "step": 1029
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.722851899827289e-05,
+      "loss": 0.1001,
+      "step": 1030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.722582037996546e-05,
+      "loss": 0.1233,
+      "step": 1031
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.722312176165803e-05,
+      "loss": 0.586,
+      "step": 1032
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7220423143350605e-05,
+      "loss": 0.175,
+      "step": 1033
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.721772452504318e-05,
+      "loss": 0.1126,
+      "step": 1034
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7215025906735756e-05,
+      "loss": 0.3034,
+      "step": 1035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.721232728842833e-05,
+      "loss": 0.2049,
+      "step": 1036
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.72096286701209e-05,
+      "loss": 0.1218,
+      "step": 1037
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.720693005181347e-05,
+      "loss": 0.3713,
+      "step": 1038
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7204231433506044e-05,
+      "loss": 0.2161,
+      "step": 1039
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7201532815198616e-05,
+      "loss": 0.2817,
+      "step": 1040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7198834196891195e-05,
+      "loss": 0.1681,
+      "step": 1041
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.719613557858377e-05,
+      "loss": 0.3948,
+      "step": 1042
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.719343696027634e-05,
+      "loss": 0.1366,
+      "step": 1043
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.719073834196892e-05,
+      "loss": 0.1044,
+      "step": 1044
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.718803972366149e-05,
+      "loss": 0.4707,
+      "step": 1045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7185341105354055e-05,
+      "loss": 0.3118,
+      "step": 1046
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7182642487046634e-05,
+      "loss": 0.29,
+      "step": 1047
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7179943868739206e-05,
+      "loss": 0.117,
+      "step": 1048
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.717724525043178e-05,
+      "loss": 0.4065,
+      "step": 1049
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.717454663212436e-05,
+      "loss": 0.3859,
+      "step": 1050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.717184801381693e-05,
+      "loss": 0.2525,
+      "step": 1051
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.71691493955095e-05,
+      "loss": 0.1003,
+      "step": 1052
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7166450777202074e-05,
+      "loss": 0.3323,
+      "step": 1053
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7163752158894646e-05,
+      "loss": 0.3196,
+      "step": 1054
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.716105354058722e-05,
+      "loss": 0.3045,
+      "step": 1055
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7158354922279797e-05,
+      "loss": 0.1355,
+      "step": 1056
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.715565630397237e-05,
+      "loss": 0.3127,
+      "step": 1057
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.715295768566494e-05,
+      "loss": 0.1695,
+      "step": 1058
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.715025906735751e-05,
+      "loss": 0.1603,
+      "step": 1059
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7147560449050085e-05,
+      "loss": 0.1094,
+      "step": 1060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.714486183074266e-05,
+      "loss": 0.5886,
+      "step": 1061
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7142163212435236e-05,
+      "loss": 0.4659,
+      "step": 1062
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.713946459412781e-05,
+      "loss": 0.286,
+      "step": 1063
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.713676597582038e-05,
+      "loss": 0.3477,
+      "step": 1064
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.713406735751296e-05,
+      "loss": 0.4756,
+      "step": 1065
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.713136873920553e-05,
+      "loss": 0.2114,
+      "step": 1066
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.71286701208981e-05,
+      "loss": 0.4273,
+      "step": 1067
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7125971502590675e-05,
+      "loss": 0.1382,
+      "step": 1068
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.712327288428325e-05,
+      "loss": 0.0999,
+      "step": 1069
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7120574265975826e-05,
+      "loss": 0.1076,
+      "step": 1070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.71178756476684e-05,
+      "loss": 0.1131,
+      "step": 1071
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.711517702936097e-05,
+      "loss": 0.1128,
+      "step": 1072
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.711247841105354e-05,
+      "loss": 0.3201,
+      "step": 1073
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7109779792746114e-05,
+      "loss": 0.1897,
+      "step": 1074
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7107081174438687e-05,
+      "loss": 0.2631,
+      "step": 1075
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7104382556131265e-05,
+      "loss": 0.0951,
+      "step": 1076
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.710168393782384e-05,
+      "loss": 0.3229,
+      "step": 1077
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.709898531951641e-05,
+      "loss": 0.2284,
+      "step": 1078
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.709628670120899e-05,
+      "loss": 0.1943,
+      "step": 1079
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.709358808290156e-05,
+      "loss": 0.2794,
+      "step": 1080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7090889464594126e-05,
+      "loss": 0.2988,
+      "step": 1081
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7088190846286705e-05,
+      "loss": 0.1359,
+      "step": 1082
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.708549222797928e-05,
+      "loss": 0.1126,
+      "step": 1083
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.708279360967185e-05,
+      "loss": 0.3016,
+      "step": 1084
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.708009499136443e-05,
+      "loss": 0.2227,
+      "step": 1085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7077396373057e-05,
+      "loss": 0.2857,
+      "step": 1086
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.707469775474957e-05,
+      "loss": 0.0985,
+      "step": 1087
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7071999136442144e-05,
+      "loss": 0.3095,
+      "step": 1088
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7069300518134716e-05,
+      "loss": 0.271,
+      "step": 1089
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.706660189982729e-05,
+      "loss": 0.3362,
+      "step": 1090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.706390328151987e-05,
+      "loss": 0.2317,
+      "step": 1091
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.706120466321244e-05,
+      "loss": 0.3788,
+      "step": 1092
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.705850604490501e-05,
+      "loss": 0.2337,
+      "step": 1093
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.705580742659759e-05,
+      "loss": 0.2594,
+      "step": 1094
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7053108808290155e-05,
+      "loss": 0.71,
+      "step": 1095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.705041018998273e-05,
+      "loss": 0.1909,
+      "step": 1096
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7047711571675306e-05,
+      "loss": 0.1834,
+      "step": 1097
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.704501295336788e-05,
+      "loss": 0.2399,
+      "step": 1098
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.704231433506045e-05,
+      "loss": 0.3186,
+      "step": 1099
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.703961571675303e-05,
+      "loss": 0.272,
+      "step": 1100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.70369170984456e-05,
+      "loss": 0.0999,
+      "step": 1101
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7034218480138173e-05,
+      "loss": 0.1273,
+      "step": 1102
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7031519861830746e-05,
+      "loss": 0.1232,
+      "step": 1103
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.702882124352332e-05,
+      "loss": 0.3203,
+      "step": 1104
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.702612262521589e-05,
+      "loss": 0.2947,
+      "step": 1105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.702342400690847e-05,
+      "loss": 0.3691,
+      "step": 1106
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.702072538860104e-05,
+      "loss": 0.3195,
+      "step": 1107
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.701802677029361e-05,
+      "loss": 0.2361,
+      "step": 1108
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7015328151986185e-05,
+      "loss": 0.354,
+      "step": 1109
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.701262953367876e-05,
+      "loss": 0.2717,
+      "step": 1110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.700993091537133e-05,
+      "loss": 0.28,
+      "step": 1111
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.700723229706391e-05,
+      "loss": 0.5061,
+      "step": 1112
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.700453367875648e-05,
+      "loss": 0.1344,
+      "step": 1113
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.700183506044905e-05,
+      "loss": 0.105,
+      "step": 1114
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.699913644214163e-05,
+      "loss": 0.1731,
+      "step": 1115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.69964378238342e-05,
+      "loss": 0.1835,
+      "step": 1116
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.699373920552677e-05,
+      "loss": 0.1044,
+      "step": 1117
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.699104058721935e-05,
+      "loss": 0.1715,
+      "step": 1118
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.698834196891192e-05,
+      "loss": 0.2146,
+      "step": 1119
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.698564335060449e-05,
+      "loss": 0.1218,
+      "step": 1120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.698294473229707e-05,
+      "loss": 0.0959,
+      "step": 1121
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.698024611398964e-05,
+      "loss": 0.1843,
+      "step": 1122
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6977547495682214e-05,
+      "loss": 0.3515,
+      "step": 1123
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6974848877374786e-05,
+      "loss": 0.2829,
+      "step": 1124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.697215025906736e-05,
+      "loss": 0.5659,
+      "step": 1125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.696945164075993e-05,
+      "loss": 0.2566,
+      "step": 1126
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.696675302245251e-05,
+      "loss": 0.1219,
+      "step": 1127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.696405440414508e-05,
+      "loss": 0.2177,
+      "step": 1128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6961355785837654e-05,
+      "loss": 0.1112,
+      "step": 1129
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6958657167530226e-05,
+      "loss": 0.5133,
+      "step": 1130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.69559585492228e-05,
+      "loss": 0.3538,
+      "step": 1131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.695325993091537e-05,
+      "loss": 0.2067,
+      "step": 1132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.695056131260795e-05,
+      "loss": 0.5317,
+      "step": 1133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.694786269430052e-05,
+      "loss": 0.2545,
+      "step": 1134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.694516407599309e-05,
+      "loss": 0.1017,
+      "step": 1135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.694246545768567e-05,
+      "loss": 0.3105,
+      "step": 1136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6939766839378244e-05,
+      "loss": 0.2375,
+      "step": 1137
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.693706822107081e-05,
+      "loss": 0.4119,
+      "step": 1138
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.693436960276339e-05,
+      "loss": 0.2326,
+      "step": 1139
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.693167098445596e-05,
+      "loss": 0.1272,
+      "step": 1140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.692897236614853e-05,
+      "loss": 0.1204,
+      "step": 1141
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.692627374784111e-05,
+      "loss": 0.2681,
+      "step": 1142
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.692357512953368e-05,
+      "loss": 0.2042,
+      "step": 1143
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6920876511226255e-05,
+      "loss": 0.1738,
+      "step": 1144
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.691817789291883e-05,
+      "loss": 0.1079,
+      "step": 1145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.69154792746114e-05,
+      "loss": 0.2466,
+      "step": 1146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.691278065630397e-05,
+      "loss": 0.2838,
+      "step": 1147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.691008203799655e-05,
+      "loss": 0.3347,
+      "step": 1148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.690738341968912e-05,
+      "loss": 0.3954,
+      "step": 1149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6904684801381694e-05,
+      "loss": 0.1547,
+      "step": 1150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.690198618307427e-05,
+      "loss": 0.0943,
+      "step": 1151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.689928756476684e-05,
+      "loss": 0.1938,
+      "step": 1152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.689658894645941e-05,
+      "loss": 0.3589,
+      "step": 1153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.689389032815199e-05,
+      "loss": 0.153,
+      "step": 1154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.689119170984456e-05,
+      "loss": 0.1899,
+      "step": 1155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6888493091537134e-05,
+      "loss": 0.223,
+      "step": 1156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.688579447322971e-05,
+      "loss": 0.135,
+      "step": 1157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6883095854922285e-05,
+      "loss": 0.2301,
+      "step": 1158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.688039723661486e-05,
+      "loss": 0.2971,
+      "step": 1159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.687769861830743e-05,
+      "loss": 0.3279,
+      "step": 1160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6875e-05,
+      "loss": 0.1085,
+      "step": 1161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.687230138169257e-05,
+      "loss": 0.3539,
+      "step": 1162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.686960276338515e-05,
+      "loss": 0.6509,
+      "step": 1163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6866904145077724e-05,
+      "loss": 0.2668,
+      "step": 1164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6864205526770296e-05,
+      "loss": 0.4415,
+      "step": 1165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.686150690846287e-05,
+      "loss": 0.2635,
+      "step": 1166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.685880829015544e-05,
+      "loss": 0.3504,
+      "step": 1167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.685610967184801e-05,
+      "loss": 0.1239,
+      "step": 1168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.685341105354059e-05,
+      "loss": 0.1975,
+      "step": 1169
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.685071243523316e-05,
+      "loss": 0.3273,
+      "step": 1170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6848013816925735e-05,
+      "loss": 0.3164,
+      "step": 1171
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6845315198618314e-05,
+      "loss": 0.3954,
+      "step": 1172
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6842616580310886e-05,
+      "loss": 0.4765,
+      "step": 1173
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.683991796200345e-05,
+      "loss": 0.3167,
+      "step": 1174
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.683721934369603e-05,
+      "loss": 0.2458,
+      "step": 1175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.68345207253886e-05,
+      "loss": 0.0842,
+      "step": 1176
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6831822107081175e-05,
+      "loss": 0.209,
+      "step": 1177
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6829123488773754e-05,
+      "loss": 0.1069,
+      "step": 1178
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6826424870466326e-05,
+      "loss": 0.1497,
+      "step": 1179
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.68237262521589e-05,
+      "loss": 0.2159,
+      "step": 1180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.682102763385147e-05,
+      "loss": 0.2134,
+      "step": 1181
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.681832901554404e-05,
+      "loss": 0.1069,
+      "step": 1182
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6815630397236614e-05,
+      "loss": 0.2191,
+      "step": 1183
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.681293177892919e-05,
+      "loss": 0.3549,
+      "step": 1184
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6810233160621765e-05,
+      "loss": 0.2105,
+      "step": 1185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.680753454231434e-05,
+      "loss": 0.1066,
+      "step": 1186
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.680483592400691e-05,
+      "loss": 0.1226,
+      "step": 1187
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.680213730569948e-05,
+      "loss": 0.2895,
+      "step": 1188
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.679943868739205e-05,
+      "loss": 0.1311,
+      "step": 1189
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.679674006908463e-05,
+      "loss": 0.7485,
+      "step": 1190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6794041450777204e-05,
+      "loss": 0.4144,
+      "step": 1191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6791342832469776e-05,
+      "loss": 0.1849,
+      "step": 1192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6788644214162355e-05,
+      "loss": 0.1862,
+      "step": 1193
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.678594559585493e-05,
+      "loss": 0.215,
+      "step": 1194
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.678324697754749e-05,
+      "loss": 0.4462,
+      "step": 1195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.678054835924007e-05,
+      "loss": 0.4156,
+      "step": 1196
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6777849740932643e-05,
+      "loss": 0.1848,
+      "step": 1197
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6775151122625216e-05,
+      "loss": 0.3006,
+      "step": 1198
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6772452504317794e-05,
+      "loss": 0.3746,
+      "step": 1199
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6769753886010366e-05,
+      "loss": 0.1565,
+      "step": 1200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.676705526770294e-05,
+      "loss": 0.3365,
+      "step": 1201
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.676435664939551e-05,
+      "loss": 0.1244,
+      "step": 1202
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.676165803108808e-05,
+      "loss": 0.2924,
+      "step": 1203
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6758959412780655e-05,
+      "loss": 0.2332,
+      "step": 1204
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6756260794473234e-05,
+      "loss": 0.2394,
+      "step": 1205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6753562176165806e-05,
+      "loss": 0.33,
+      "step": 1206
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.675086355785838e-05,
+      "loss": 0.3777,
+      "step": 1207
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.674816493955096e-05,
+      "loss": 0.2219,
+      "step": 1208
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.674546632124352e-05,
+      "loss": 0.129,
+      "step": 1209
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6742767702936094e-05,
+      "loss": 0.3643,
+      "step": 1210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.674006908462867e-05,
+      "loss": 0.2485,
+      "step": 1211
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6737370466321245e-05,
+      "loss": 0.1173,
+      "step": 1212
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.673467184801382e-05,
+      "loss": 0.1638,
+      "step": 1213
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6731973229706396e-05,
+      "loss": 0.0973,
+      "step": 1214
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.672927461139897e-05,
+      "loss": 0.2545,
+      "step": 1215
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.672657599309154e-05,
+      "loss": 0.0939,
+      "step": 1216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.672387737478411e-05,
+      "loss": 0.2293,
+      "step": 1217
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6721178756476684e-05,
+      "loss": 0.1383,
+      "step": 1218
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6718480138169256e-05,
+      "loss": 0.3606,
+      "step": 1219
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6715781519861835e-05,
+      "loss": 0.1782,
+      "step": 1220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.671308290155441e-05,
+      "loss": 0.3053,
+      "step": 1221
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.671038428324698e-05,
+      "loss": 0.1934,
+      "step": 1222
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.670768566493955e-05,
+      "loss": 0.5757,
+      "step": 1223
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6704987046632124e-05,
+      "loss": 0.1196,
+      "step": 1224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6702288428324696e-05,
+      "loss": 0.3281,
+      "step": 1225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6699589810017275e-05,
+      "loss": 0.1898,
+      "step": 1226
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.669689119170985e-05,
+      "loss": 0.5511,
+      "step": 1227
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.669419257340242e-05,
+      "loss": 0.1402,
+      "step": 1228
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6691493955095e-05,
+      "loss": 0.3615,
+      "step": 1229
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.668879533678757e-05,
+      "loss": 0.1597,
+      "step": 1230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6686096718480135e-05,
+      "loss": 0.4881,
+      "step": 1231
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6683398100172714e-05,
+      "loss": 0.3169,
+      "step": 1232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6680699481865286e-05,
+      "loss": 0.3537,
+      "step": 1233
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.667800086355786e-05,
+      "loss": 0.1331,
+      "step": 1234
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.667530224525044e-05,
+      "loss": 0.0965,
+      "step": 1235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.667260362694301e-05,
+      "loss": 0.1127,
+      "step": 1236
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.666990500863558e-05,
+      "loss": 0.3022,
+      "step": 1237
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.666720639032815e-05,
+      "loss": 0.2266,
+      "step": 1238
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6664507772020725e-05,
+      "loss": 0.1518,
+      "step": 1239
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.66618091537133e-05,
+      "loss": 0.2037,
+      "step": 1240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6659110535405876e-05,
+      "loss": 0.5513,
+      "step": 1241
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.665641191709845e-05,
+      "loss": 0.1825,
+      "step": 1242
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.665371329879102e-05,
+      "loss": 0.2766,
+      "step": 1243
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.665101468048359e-05,
+      "loss": 0.2741,
+      "step": 1244
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6648316062176164e-05,
+      "loss": 0.3817,
+      "step": 1245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6645617443868737e-05,
+      "loss": 0.2808,
+      "step": 1246
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6642918825561315e-05,
+      "loss": 0.1235,
+      "step": 1247
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.664022020725389e-05,
+      "loss": 0.2775,
+      "step": 1248
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.663752158894646e-05,
+      "loss": 0.2278,
+      "step": 1249
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.663482297063904e-05,
+      "loss": 0.525,
+      "step": 1250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.663212435233161e-05,
+      "loss": 0.1833,
+      "step": 1251
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6629425734024176e-05,
+      "loss": 0.1367,
+      "step": 1252
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6626727115716755e-05,
+      "loss": 0.3107,
+      "step": 1253
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.662402849740933e-05,
+      "loss": 0.1363,
+      "step": 1254
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.66213298791019e-05,
+      "loss": 0.1454,
+      "step": 1255
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.661863126079448e-05,
+      "loss": 0.2822,
+      "step": 1256
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.661593264248705e-05,
+      "loss": 0.1298,
+      "step": 1257
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.661323402417962e-05,
+      "loss": 0.2191,
+      "step": 1258
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6610535405872194e-05,
+      "loss": 0.1672,
+      "step": 1259
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6607836787564766e-05,
+      "loss": 0.1224,
+      "step": 1260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6605138169257345e-05,
+      "loss": 0.111,
+      "step": 1261
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.660243955094992e-05,
+      "loss": 0.3853,
+      "step": 1262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.659974093264249e-05,
+      "loss": 0.2947,
+      "step": 1263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.659704231433507e-05,
+      "loss": 0.149,
+      "step": 1264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.659434369602764e-05,
+      "loss": 0.3758,
+      "step": 1265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6591645077720205e-05,
+      "loss": 0.3148,
+      "step": 1266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6588946459412784e-05,
+      "loss": 0.1013,
+      "step": 1267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6586247841105356e-05,
+      "loss": 0.3019,
+      "step": 1268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.658354922279793e-05,
+      "loss": 0.3565,
+      "step": 1269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.658085060449051e-05,
+      "loss": 0.2511,
+      "step": 1270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.657815198618308e-05,
+      "loss": 0.1986,
+      "step": 1271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.657545336787565e-05,
+      "loss": 0.1114,
+      "step": 1272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6572754749568224e-05,
+      "loss": 0.2689,
+      "step": 1273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6570056131260796e-05,
+      "loss": 0.1984,
+      "step": 1274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.656735751295337e-05,
+      "loss": 0.3947,
+      "step": 1275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6564658894645947e-05,
+      "loss": 0.2991,
+      "step": 1276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.656196027633852e-05,
+      "loss": 0.2951,
+      "step": 1277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.655926165803109e-05,
+      "loss": 0.1645,
+      "step": 1278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.655656303972367e-05,
+      "loss": 0.5967,
+      "step": 1279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6553864421416235e-05,
+      "loss": 0.1096,
+      "step": 1280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.655116580310881e-05,
+      "loss": 0.3955,
+      "step": 1281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6548467184801386e-05,
+      "loss": 0.1991,
+      "step": 1282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.654576856649396e-05,
+      "loss": 0.0927,
+      "step": 1283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.654306994818653e-05,
+      "loss": 0.1667,
+      "step": 1284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.654037132987911e-05,
+      "loss": 0.3074,
+      "step": 1285
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.653767271157168e-05,
+      "loss": 0.196,
+      "step": 1286
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.653497409326425e-05,
+      "loss": 0.3105,
+      "step": 1287
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.6532275474956825e-05,
+      "loss": 0.1922,
+      "step": 1288
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.65295768566494e-05,
+      "loss": 0.0873,
+      "step": 1289
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.652687823834197e-05,
+      "loss": 0.131,
+      "step": 1290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.652417962003455e-05,
+      "loss": 0.2832,
+      "step": 1291
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.652148100172712e-05,
+      "loss": 0.1666,
+      "step": 1292
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.651878238341969e-05,
+      "loss": 0.1779,
+      "step": 1293
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.6516083765112264e-05,
+      "loss": 0.3001,
+      "step": 1294
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.6513385146804836e-05,
+      "loss": 0.101,
+      "step": 1295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.651068652849741e-05,
+      "loss": 0.2403,
+      "step": 1296
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.650798791018999e-05,
+      "loss": 0.3161,
+      "step": 1297
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.650528929188256e-05,
+      "loss": 0.1886,
+      "step": 1298
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.650259067357513e-05,
+      "loss": 0.1058,
+      "step": 1299
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.649989205526771e-05,
+      "loss": 0.3703,
+      "step": 1300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.6497193436960276e-05,
+      "loss": 0.2275,
+      "step": 1301
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.649449481865285e-05,
+      "loss": 0.4032,
+      "step": 1302
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.649179620034543e-05,
+      "loss": 0.1281,
+      "step": 1303
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.6489097582038e-05,
+      "loss": 0.1,
+      "step": 1304
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.648639896373057e-05,
+      "loss": 0.1082,
+      "step": 1305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.648370034542315e-05,
+      "loss": 0.0936,
+      "step": 1306
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.648100172711572e-05,
+      "loss": 0.2286,
+      "step": 1307
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.6478303108808294e-05,
+      "loss": 0.3719,
+      "step": 1308
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6475604490500866e-05,
+      "loss": 0.301,
+      "step": 1309
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.647290587219344e-05,
+      "loss": 0.0986,
+      "step": 1310
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.647020725388601e-05,
+      "loss": 0.1731,
+      "step": 1311
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.646750863557859e-05,
+      "loss": 0.0832,
+      "step": 1312
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.646481001727116e-05,
+      "loss": 0.1071,
+      "step": 1313
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.646211139896373e-05,
+      "loss": 0.1159,
+      "step": 1314
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6459412780656305e-05,
+      "loss": 0.2397,
+      "step": 1315
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.645671416234888e-05,
+      "loss": 0.2724,
+      "step": 1316
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.645401554404145e-05,
+      "loss": 0.5461,
+      "step": 1317
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.645131692573403e-05,
+      "loss": 0.2275,
+      "step": 1318
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.64486183074266e-05,
+      "loss": 0.0919,
+      "step": 1319
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.644591968911917e-05,
+      "loss": 0.0905,
+      "step": 1320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.644322107081175e-05,
+      "loss": 0.1253,
+      "step": 1321
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6440522452504323e-05,
+      "loss": 0.2026,
+      "step": 1322
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.643782383419689e-05,
+      "loss": 0.4092,
+      "step": 1323
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.643512521588947e-05,
+      "loss": 0.0974,
+      "step": 1324
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.643242659758204e-05,
+      "loss": 0.1268,
+      "step": 1325
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.642972797927461e-05,
+      "loss": 0.1165,
+      "step": 1326
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.642702936096719e-05,
+      "loss": 0.6167,
+      "step": 1327
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.642433074265976e-05,
+      "loss": 0.1327,
+      "step": 1328
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6421632124352335e-05,
+      "loss": 0.5158,
+      "step": 1329
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.641893350604491e-05,
+      "loss": 0.2692,
+      "step": 1330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.641623488773748e-05,
+      "loss": 0.2247,
+      "step": 1331
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.641353626943005e-05,
+      "loss": 0.0948,
+      "step": 1332
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.641083765112263e-05,
+      "loss": 0.1332,
+      "step": 1333
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.64081390328152e-05,
+      "loss": 0.1083,
+      "step": 1334
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.6405440414507774e-05,
+      "loss": 0.3591,
+      "step": 1335
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.640274179620035e-05,
+      "loss": 0.2517,
+      "step": 1336
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.640004317789292e-05,
+      "loss": 0.3232,
+      "step": 1337
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.639734455958549e-05,
+      "loss": 0.4924,
+      "step": 1338
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.639464594127807e-05,
+      "loss": 0.4459,
+      "step": 1339
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.639194732297064e-05,
+      "loss": 0.1544,
+      "step": 1340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.638924870466321e-05,
+      "loss": 0.1598,
+      "step": 1341
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.638655008635579e-05,
+      "loss": 0.1695,
+      "step": 1342
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.6383851468048364e-05,
+      "loss": 0.0851,
+      "step": 1343
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.6381152849740936e-05,
+      "loss": 0.0886,
+      "step": 1344
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.637845423143351e-05,
+      "loss": 0.0972,
+      "step": 1345
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.637575561312608e-05,
+      "loss": 0.3344,
+      "step": 1346
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.637305699481865e-05,
+      "loss": 0.2136,
+      "step": 1347
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.637035837651123e-05,
+      "loss": 0.1732,
+      "step": 1348
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.6367659758203804e-05,
+      "loss": 0.1744,
+      "step": 1349
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.6364961139896376e-05,
+      "loss": 0.1138,
+      "step": 1350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.636226252158895e-05,
+      "loss": 0.2708,
+      "step": 1351
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.635956390328152e-05,
+      "loss": 0.4032,
+      "step": 1352
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.635686528497409e-05,
+      "loss": 0.0951,
+      "step": 1353
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.635416666666667e-05,
+      "loss": 0.276,
+      "step": 1354
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.635146804835924e-05,
+      "loss": 0.2424,
+      "step": 1355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6348769430051815e-05,
+      "loss": 0.0875,
+      "step": 1356
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6346070811744394e-05,
+      "loss": 0.1448,
+      "step": 1357
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.634337219343696e-05,
+      "loss": 0.0849,
+      "step": 1358
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.634067357512953e-05,
+      "loss": 0.113,
+      "step": 1359
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.633797495682211e-05,
+      "loss": 0.2492,
+      "step": 1360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.633527633851468e-05,
+      "loss": 0.5876,
+      "step": 1361
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6332577720207254e-05,
+      "loss": 0.1916,
+      "step": 1362
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.632987910189983e-05,
+      "loss": 0.1017,
+      "step": 1363
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6327180483592405e-05,
+      "loss": 0.289,
+      "step": 1364
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.632448186528498e-05,
+      "loss": 0.3192,
+      "step": 1365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.632178324697755e-05,
+      "loss": 0.1954,
+      "step": 1366
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.631908462867012e-05,
+      "loss": 0.2266,
+      "step": 1367
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6316386010362694e-05,
+      "loss": 0.1122,
+      "step": 1368
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.631368739205527e-05,
+      "loss": 0.1664,
+      "step": 1369
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6310988773747844e-05,
+      "loss": 0.1087,
+      "step": 1370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6308290155440417e-05,
+      "loss": 0.113,
+      "step": 1371
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.630559153713299e-05,
+      "loss": 0.149,
+      "step": 1372
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.630289291882556e-05,
+      "loss": 0.4365,
+      "step": 1373
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.630019430051813e-05,
+      "loss": 0.5907,
+      "step": 1374
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.629749568221071e-05,
+      "loss": 0.0956,
+      "step": 1375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6294797063903284e-05,
+      "loss": 0.093,
+      "step": 1376
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6292098445595856e-05,
+      "loss": 0.4399,
+      "step": 1377
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6289399827288435e-05,
+      "loss": 0.1342,
+      "step": 1378
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.628670120898101e-05,
+      "loss": 0.4029,
+      "step": 1379
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.628400259067357e-05,
+      "loss": 0.4534,
+      "step": 1380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.628130397236615e-05,
+      "loss": 0.1167,
+      "step": 1381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.627860535405872e-05,
+      "loss": 0.1194,
+      "step": 1382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6275906735751295e-05,
+      "loss": 0.1265,
+      "step": 1383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6273208117443874e-05,
+      "loss": 0.1105,
+      "step": 1384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6270509499136446e-05,
+      "loss": 0.2482,
+      "step": 1385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.626781088082902e-05,
+      "loss": 0.1494,
+      "step": 1386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.626511226252159e-05,
+      "loss": 0.3161,
+      "step": 1387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.626241364421416e-05,
+      "loss": 0.2884,
+      "step": 1388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6259715025906734e-05,
+      "loss": 0.4681,
+      "step": 1389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.625701640759931e-05,
+      "loss": 0.683,
+      "step": 1390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6254317789291885e-05,
+      "loss": 0.0866,
+      "step": 1391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.625161917098446e-05,
+      "loss": 0.1054,
+      "step": 1392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6248920552677036e-05,
+      "loss": 0.4836,
+      "step": 1393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.62462219343696e-05,
+      "loss": 0.2322,
+      "step": 1394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6243523316062174e-05,
+      "loss": 0.2468,
+      "step": 1395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.624082469775475e-05,
+      "loss": 0.1695,
+      "step": 1396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6238126079447325e-05,
+      "loss": 0.2805,
+      "step": 1397
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.62354274611399e-05,
+      "loss": 0.0849,
+      "step": 1398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6232728842832476e-05,
+      "loss": 0.386,
+      "step": 1399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.623003022452505e-05,
+      "loss": 0.2342,
+      "step": 1400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.622733160621762e-05,
+      "loss": 0.278,
+      "step": 1401
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.622463298791019e-05,
+      "loss": 0.2234,
+      "step": 1402
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6221934369602764e-05,
+      "loss": 0.2632,
+      "step": 1403
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6219235751295336e-05,
+      "loss": 0.1669,
+      "step": 1404
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6216537132987915e-05,
+      "loss": 0.4083,
+      "step": 1405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.621383851468049e-05,
+      "loss": 0.2085,
+      "step": 1406
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.621113989637306e-05,
+      "loss": 0.214,
+      "step": 1407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.620844127806563e-05,
+      "loss": 0.1889,
+      "step": 1408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.62057426597582e-05,
+      "loss": 0.1768,
+      "step": 1409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6203044041450775e-05,
+      "loss": 0.1661,
+      "step": 1410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6200345423143354e-05,
+      "loss": 0.312,
+      "step": 1411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6197646804835926e-05,
+      "loss": 0.1893,
+      "step": 1412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.61949481865285e-05,
+      "loss": 0.2844,
+      "step": 1413
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.619224956822108e-05,
+      "loss": 0.1399,
+      "step": 1414
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.618955094991364e-05,
+      "loss": 0.1698,
+      "step": 1415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6186852331606215e-05,
+      "loss": 0.2547,
+      "step": 1416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6184153713298793e-05,
+      "loss": 0.2611,
+      "step": 1417
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6181455094991366e-05,
+      "loss": 0.1355,
+      "step": 1418
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.617875647668394e-05,
+      "loss": 0.5669,
+      "step": 1419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6176057858376516e-05,
+      "loss": 0.1375,
+      "step": 1420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.617335924006909e-05,
+      "loss": 0.1314,
+      "step": 1421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.617066062176166e-05,
+      "loss": 0.1736,
+      "step": 1422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.616796200345423e-05,
+      "loss": 0.7364,
+      "step": 1423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6165263385146805e-05,
+      "loss": 0.346,
+      "step": 1424
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.616256476683938e-05,
+      "loss": 0.141,
+      "step": 1425
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6159866148531956e-05,
+      "loss": 0.0935,
+      "step": 1426
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.615716753022453e-05,
+      "loss": 0.194,
+      "step": 1427
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.61544689119171e-05,
+      "loss": 0.2013,
+      "step": 1428
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.615177029360967e-05,
+      "loss": 0.0845,
+      "step": 1429
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6149071675302244e-05,
+      "loss": 0.1274,
+      "step": 1430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6146373056994816e-05,
+      "loss": 0.21,
+      "step": 1431
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6143674438687395e-05,
+      "loss": 0.276,
+      "step": 1432
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.614097582037997e-05,
+      "loss": 0.1009,
+      "step": 1433
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.613827720207254e-05,
+      "loss": 0.1867,
+      "step": 1434
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.613557858376512e-05,
+      "loss": 0.1014,
+      "step": 1435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.613287996545769e-05,
+      "loss": 0.2421,
+      "step": 1436
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6130181347150255e-05,
+      "loss": 0.2174,
+      "step": 1437
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6127482728842834e-05,
+      "loss": 0.2008,
+      "step": 1438
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6124784110535406e-05,
+      "loss": 0.3278,
+      "step": 1439
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.612208549222798e-05,
+      "loss": 0.1472,
+      "step": 1440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.611938687392056e-05,
+      "loss": 0.1261,
+      "step": 1441
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.611668825561313e-05,
+      "loss": 0.3293,
+      "step": 1442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.61139896373057e-05,
+      "loss": 0.3934,
+      "step": 1443
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6111291018998274e-05,
+      "loss": 0.3349,
+      "step": 1444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6108592400690846e-05,
+      "loss": 0.102,
+      "step": 1445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.610589378238342e-05,
+      "loss": 0.0825,
+      "step": 1446
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6103195164076e-05,
+      "loss": 0.3655,
+      "step": 1447
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.610049654576857e-05,
+      "loss": 0.1073,
+      "step": 1448
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.609779792746114e-05,
+      "loss": 0.1284,
+      "step": 1449
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.609509930915372e-05,
+      "loss": 0.4225,
+      "step": 1450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6092400690846285e-05,
+      "loss": 0.1005,
+      "step": 1451
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6089702072538864e-05,
+      "loss": 0.1032,
+      "step": 1452
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6087003454231436e-05,
+      "loss": 0.2231,
+      "step": 1453
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.608430483592401e-05,
+      "loss": 0.4082,
+      "step": 1454
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.608160621761659e-05,
+      "loss": 0.4552,
+      "step": 1455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.607890759930916e-05,
+      "loss": 0.3061,
+      "step": 1456
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.607620898100173e-05,
+      "loss": 0.0927,
+      "step": 1457
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.60735103626943e-05,
+      "loss": 0.2406,
+      "step": 1458
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6070811744386875e-05,
+      "loss": 0.0989,
+      "step": 1459
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.606811312607945e-05,
+      "loss": 0.1315,
+      "step": 1460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6065414507772026e-05,
+      "loss": 0.1445,
+      "step": 1461
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.60627158894646e-05,
+      "loss": 0.3985,
+      "step": 1462
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.606001727115717e-05,
+      "loss": 0.1049,
+      "step": 1463
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.605731865284975e-05,
+      "loss": 0.1465,
+      "step": 1464
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6054620034542314e-05,
+      "loss": 0.176,
+      "step": 1465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6051921416234887e-05,
+      "loss": 0.1026,
+      "step": 1466
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6049222797927465e-05,
+      "loss": 0.3466,
+      "step": 1467
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.604652417962004e-05,
+      "loss": 0.1402,
+      "step": 1468
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.604382556131261e-05,
+      "loss": 0.215,
+      "step": 1469
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.604112694300519e-05,
+      "loss": 0.1315,
+      "step": 1470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.603842832469776e-05,
+      "loss": 0.4193,
+      "step": 1471
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6035729706390326e-05,
+      "loss": 0.4101,
+      "step": 1472
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6033031088082905e-05,
+      "loss": 0.1021,
+      "step": 1473
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.603033246977548e-05,
+      "loss": 0.1647,
+      "step": 1474
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.602763385146805e-05,
+      "loss": 0.4825,
+      "step": 1475
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.602493523316063e-05,
+      "loss": 0.6388,
+      "step": 1476
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.60222366148532e-05,
+      "loss": 0.3709,
+      "step": 1477
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.601953799654577e-05,
+      "loss": 0.2978,
+      "step": 1478
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6016839378238344e-05,
+      "loss": 0.2539,
+      "step": 1479
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6014140759930916e-05,
+      "loss": 0.1154,
+      "step": 1480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.601144214162349e-05,
+      "loss": 0.1006,
+      "step": 1481
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.600874352331607e-05,
+      "loss": 0.2075,
+      "step": 1482
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.600604490500864e-05,
+      "loss": 0.1593,
+      "step": 1483
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.600334628670121e-05,
+      "loss": 0.3391,
+      "step": 1484
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.600064766839379e-05,
+      "loss": 0.2195,
+      "step": 1485
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.5997949050086355e-05,
+      "loss": 0.2989,
+      "step": 1486
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.599525043177893e-05,
+      "loss": 0.3045,
+      "step": 1487
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.5992551813471506e-05,
+      "loss": 0.1474,
+      "step": 1488
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.598985319516408e-05,
+      "loss": 0.4451,
+      "step": 1489
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.598715457685665e-05,
+      "loss": 0.1893,
+      "step": 1490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.598445595854923e-05,
+      "loss": 0.0881,
+      "step": 1491
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.59817573402418e-05,
+      "loss": 0.1119,
+      "step": 1492
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.5979058721934374e-05,
+      "loss": 0.3653,
+      "step": 1493
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.5976360103626946e-05,
+      "loss": 0.1236,
+      "step": 1494
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.597366148531952e-05,
+      "loss": 0.2542,
+      "step": 1495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.597096286701209e-05,
+      "loss": 0.3011,
+      "step": 1496
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.596826424870467e-05,
+      "loss": 0.2515,
+      "step": 1497
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.596556563039724e-05,
+      "loss": 0.284,
+      "step": 1498
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.596286701208981e-05,
+      "loss": 0.1946,
+      "step": 1499
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5960168393782385e-05,
+      "loss": 0.4446,
+      "step": 1500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.595746977547496e-05,
+      "loss": 0.2553,
+      "step": 1501
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.595477115716753e-05,
+      "loss": 0.09,
+      "step": 1502
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.595207253886011e-05,
+      "loss": 0.2666,
+      "step": 1503
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.594937392055268e-05,
+      "loss": 0.1583,
+      "step": 1504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.594667530224525e-05,
+      "loss": 0.2867,
+      "step": 1505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.594397668393783e-05,
+      "loss": 0.0903,
+      "step": 1506
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.59412780656304e-05,
+      "loss": 0.1589,
+      "step": 1507
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.593857944732297e-05,
+      "loss": 0.281,
+      "step": 1508
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.593588082901555e-05,
+      "loss": 0.1408,
+      "step": 1509
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.593318221070812e-05,
+      "loss": 0.1151,
+      "step": 1510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.593048359240069e-05,
+      "loss": 0.2462,
+      "step": 1511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.592778497409327e-05,
+      "loss": 0.3192,
+      "step": 1512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.592508635578584e-05,
+      "loss": 0.6519,
+      "step": 1513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5922387737478414e-05,
+      "loss": 0.1078,
+      "step": 1514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5919689119170986e-05,
+      "loss": 0.4744,
+      "step": 1515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.591699050086356e-05,
+      "loss": 0.5016,
+      "step": 1516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.591429188255613e-05,
+      "loss": 0.1836,
+      "step": 1517
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.591159326424871e-05,
+      "loss": 0.2074,
+      "step": 1518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.590889464594128e-05,
+      "loss": 0.2172,
+      "step": 1519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5906196027633854e-05,
+      "loss": 0.2892,
+      "step": 1520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.590349740932643e-05,
+      "loss": 0.2578,
+      "step": 1521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5900798791019e-05,
+      "loss": 0.3135,
+      "step": 1522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.589810017271157e-05,
+      "loss": 0.3563,
+      "step": 1523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.589540155440415e-05,
+      "loss": 0.3692,
+      "step": 1524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.589270293609672e-05,
+      "loss": 0.3246,
+      "step": 1525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.589000431778929e-05,
+      "loss": 0.2055,
+      "step": 1526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.588730569948187e-05,
+      "loss": 0.091,
+      "step": 1527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5884607081174444e-05,
+      "loss": 0.4041,
+      "step": 1528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5881908462867016e-05,
+      "loss": 0.1552,
+      "step": 1529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.587920984455959e-05,
+      "loss": 0.1074,
+      "step": 1530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.587651122625216e-05,
+      "loss": 0.1627,
+      "step": 1531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.587381260794473e-05,
+      "loss": 0.1868,
+      "step": 1532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.587111398963731e-05,
+      "loss": 0.2184,
+      "step": 1533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.586841537132988e-05,
+      "loss": 0.0913,
+      "step": 1534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5865716753022455e-05,
+      "loss": 0.4999,
+      "step": 1535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.586301813471503e-05,
+      "loss": 0.2784,
+      "step": 1536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.58603195164076e-05,
+      "loss": 0.294,
+      "step": 1537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.585762089810017e-05,
+      "loss": 0.4341,
+      "step": 1538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.585492227979275e-05,
+      "loss": 0.3445,
+      "step": 1539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.585222366148532e-05,
+      "loss": 0.2434,
+      "step": 1540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5849525043177895e-05,
+      "loss": 0.2058,
+      "step": 1541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5846826424870473e-05,
+      "loss": 0.1066,
+      "step": 1542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.584412780656304e-05,
+      "loss": 0.4896,
+      "step": 1543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.584142918825561e-05,
+      "loss": 0.1667,
+      "step": 1544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.583873056994819e-05,
+      "loss": 0.1101,
+      "step": 1545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.583603195164076e-05,
+      "loss": 0.0815,
+      "step": 1546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.3726,
+      "step": 1547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.583063471502591e-05,
+      "loss": 0.2847,
+      "step": 1548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5827936096718485e-05,
+      "loss": 0.0891,
+      "step": 1549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.582523747841106e-05,
+      "loss": 0.1756,
+      "step": 1550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.582253886010363e-05,
+      "loss": 0.2517,
+      "step": 1551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.58198402417962e-05,
+      "loss": 0.2712,
+      "step": 1552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.581714162348877e-05,
+      "loss": 0.2087,
+      "step": 1553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.581444300518135e-05,
+      "loss": 0.5086,
+      "step": 1554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5811744386873924e-05,
+      "loss": 0.1378,
+      "step": 1555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5809045768566496e-05,
+      "loss": 0.4204,
+      "step": 1556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.580634715025907e-05,
+      "loss": 0.3147,
+      "step": 1557
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.580364853195164e-05,
+      "loss": 0.1107,
+      "step": 1558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.580094991364421e-05,
+      "loss": 0.1147,
+      "step": 1559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.579825129533679e-05,
+      "loss": 0.3997,
+      "step": 1560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.579555267702936e-05,
+      "loss": 0.0955,
+      "step": 1561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5792854058721935e-05,
+      "loss": 0.1259,
+      "step": 1562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5790155440414514e-05,
+      "loss": 0.0964,
+      "step": 1563
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5787456822107086e-05,
+      "loss": 0.2563,
+      "step": 1564
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.578475820379965e-05,
+      "loss": 0.1179,
+      "step": 1565
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.578205958549223e-05,
+      "loss": 0.1035,
+      "step": 1566
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.57793609671848e-05,
+      "loss": 0.1736,
+      "step": 1567
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5776662348877375e-05,
+      "loss": 0.0979,
+      "step": 1568
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5773963730569954e-05,
+      "loss": 0.2723,
+      "step": 1569
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5771265112262526e-05,
+      "loss": 0.2467,
+      "step": 1570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.57685664939551e-05,
+      "loss": 0.0972,
+      "step": 1571
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.576586787564767e-05,
+      "loss": 0.1962,
+      "step": 1572
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.576316925734024e-05,
+      "loss": 0.118,
+      "step": 1573
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5760470639032814e-05,
+      "loss": 0.0947,
+      "step": 1574
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.575777202072539e-05,
+      "loss": 0.3386,
+      "step": 1575
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5755073402417965e-05,
+      "loss": 0.1006,
+      "step": 1576
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.575237478411054e-05,
+      "loss": 0.3249,
+      "step": 1577
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5749676165803116e-05,
+      "loss": 0.1749,
+      "step": 1578
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.574697754749568e-05,
+      "loss": 0.1329,
+      "step": 1579
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.574427892918825e-05,
+      "loss": 0.2484,
+      "step": 1580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.574158031088083e-05,
+      "loss": 0.1072,
+      "step": 1581
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5738881692573404e-05,
+      "loss": 0.3278,
+      "step": 1582
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5736183074265976e-05,
+      "loss": 0.1364,
+      "step": 1583
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5733484455958555e-05,
+      "loss": 0.8653,
+      "step": 1584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.573078583765113e-05,
+      "loss": 0.3231,
+      "step": 1585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.57280872193437e-05,
+      "loss": 0.0839,
+      "step": 1586
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.572538860103627e-05,
+      "loss": 0.2947,
+      "step": 1587
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5722689982728844e-05,
+      "loss": 0.0894,
+      "step": 1588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5719991364421416e-05,
+      "loss": 0.2412,
+      "step": 1589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5717292746113994e-05,
+      "loss": 0.1502,
+      "step": 1590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5714594127806567e-05,
+      "loss": 0.1152,
+      "step": 1591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.571189550949914e-05,
+      "loss": 0.2816,
+      "step": 1592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.570919689119171e-05,
+      "loss": 0.1045,
+      "step": 1593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.570649827288428e-05,
+      "loss": 0.28,
+      "step": 1594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5703799654576855e-05,
+      "loss": 0.0841,
+      "step": 1595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5701101036269434e-05,
+      "loss": 0.0947,
+      "step": 1596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5698402417962006e-05,
+      "loss": 0.1541,
+      "step": 1597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.569570379965458e-05,
+      "loss": 0.0887,
+      "step": 1598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.569300518134716e-05,
+      "loss": 0.2025,
+      "step": 1599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.569030656303972e-05,
+      "loss": 0.2464,
+      "step": 1600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5687607944732294e-05,
+      "loss": 0.1471,
+      "step": 1601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.568490932642487e-05,
+      "loss": 0.097,
+      "step": 1602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5682210708117445e-05,
+      "loss": 0.1929,
+      "step": 1603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.567951208981002e-05,
+      "loss": 0.1115,
+      "step": 1604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5676813471502596e-05,
+      "loss": 0.1922,
+      "step": 1605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.567411485319517e-05,
+      "loss": 0.1733,
+      "step": 1606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.567141623488774e-05,
+      "loss": 0.2929,
+      "step": 1607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.566871761658031e-05,
+      "loss": 0.2947,
+      "step": 1608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5666018998272884e-05,
+      "loss": 0.2722,
+      "step": 1609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5663320379965456e-05,
+      "loss": 0.1971,
+      "step": 1610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5660621761658035e-05,
+      "loss": 0.1621,
+      "step": 1611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.565792314335061e-05,
+      "loss": 0.1612,
+      "step": 1612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.565522452504318e-05,
+      "loss": 0.1889,
+      "step": 1613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.565252590673575e-05,
+      "loss": 0.0808,
+      "step": 1614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5649827288428324e-05,
+      "loss": 0.1053,
+      "step": 1615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5647128670120896e-05,
+      "loss": 0.1294,
+      "step": 1616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5644430051813475e-05,
+      "loss": 0.1132,
+      "step": 1617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.564173143350605e-05,
+      "loss": 0.102,
+      "step": 1618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.563903281519862e-05,
+      "loss": 0.1482,
+      "step": 1619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.56363341968912e-05,
+      "loss": 0.331,
+      "step": 1620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.563363557858377e-05,
+      "loss": 0.1746,
+      "step": 1621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5630936960276335e-05,
+      "loss": 0.1947,
+      "step": 1622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5628238341968914e-05,
+      "loss": 0.0856,
+      "step": 1623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5625539723661486e-05,
+      "loss": 0.4667,
+      "step": 1624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.562284110535406e-05,
+      "loss": 0.1359,
+      "step": 1625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.562014248704664e-05,
+      "loss": 0.2935,
+      "step": 1626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.561744386873921e-05,
+      "loss": 0.3338,
+      "step": 1627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.561474525043178e-05,
+      "loss": 0.0895,
+      "step": 1628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.561204663212435e-05,
+      "loss": 0.2861,
+      "step": 1629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5609348013816925e-05,
+      "loss": 0.0947,
+      "step": 1630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.56066493955095e-05,
+      "loss": 0.1596,
+      "step": 1631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5603950777202076e-05,
+      "loss": 0.1886,
+      "step": 1632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.560125215889465e-05,
+      "loss": 0.2519,
+      "step": 1633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.559855354058722e-05,
+      "loss": 0.1859,
+      "step": 1634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.55958549222798e-05,
+      "loss": 0.1658,
+      "step": 1635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5593156303972365e-05,
+      "loss": 0.0771,
+      "step": 1636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.559045768566494e-05,
+      "loss": 0.1269,
+      "step": 1637
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5587759067357516e-05,
+      "loss": 0.23,
+      "step": 1638
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.558506044905009e-05,
+      "loss": 0.0988,
+      "step": 1639
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.558236183074266e-05,
+      "loss": 0.2935,
+      "step": 1640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.557966321243524e-05,
+      "loss": 0.1589,
+      "step": 1641
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.557696459412781e-05,
+      "loss": 0.2902,
+      "step": 1642
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.557426597582038e-05,
+      "loss": 0.1553,
+      "step": 1643
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5571567357512955e-05,
+      "loss": 0.199,
+      "step": 1644
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.556886873920553e-05,
+      "loss": 0.2204,
+      "step": 1645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5566170120898106e-05,
+      "loss": 0.2949,
+      "step": 1646
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.556347150259068e-05,
+      "loss": 0.1007,
+      "step": 1647
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.556077288428325e-05,
+      "loss": 0.1791,
+      "step": 1648
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.555807426597582e-05,
+      "loss": 0.0946,
+      "step": 1649
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5555375647668394e-05,
+      "loss": 0.0957,
+      "step": 1650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5552677029360966e-05,
+      "loss": 0.1295,
+      "step": 1651
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5549978411053545e-05,
+      "loss": 0.3053,
+      "step": 1652
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.554727979274612e-05,
+      "loss": 0.0893,
+      "step": 1653
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.554458117443869e-05,
+      "loss": 0.2286,
+      "step": 1654
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.554188255613127e-05,
+      "loss": 0.1824,
+      "step": 1655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.553918393782384e-05,
+      "loss": 0.3238,
+      "step": 1656
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5536485319516405e-05,
+      "loss": 0.1006,
+      "step": 1657
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5533786701208984e-05,
+      "loss": 0.4,
+      "step": 1658
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5531088082901556e-05,
+      "loss": 0.2908,
+      "step": 1659
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.552838946459413e-05,
+      "loss": 0.1293,
+      "step": 1660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.552569084628671e-05,
+      "loss": 0.4209,
+      "step": 1661
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.552299222797928e-05,
+      "loss": 0.486,
+      "step": 1662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.552029360967185e-05,
+      "loss": 0.3071,
+      "step": 1663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5517594991364424e-05,
+      "loss": 0.2409,
+      "step": 1664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5514896373056996e-05,
+      "loss": 0.1277,
+      "step": 1665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.551219775474957e-05,
+      "loss": 0.2544,
+      "step": 1666
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.550949913644215e-05,
+      "loss": 0.4082,
+      "step": 1667
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.550680051813472e-05,
+      "loss": 0.221,
+      "step": 1668
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.550410189982729e-05,
+      "loss": 0.2119,
+      "step": 1669
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.550140328151987e-05,
+      "loss": 0.0899,
+      "step": 1670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5498704663212435e-05,
+      "loss": 0.1259,
+      "step": 1671
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.549600604490501e-05,
+      "loss": 0.1418,
+      "step": 1672
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5493307426597586e-05,
+      "loss": 0.2214,
+      "step": 1673
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.549060880829016e-05,
+      "loss": 0.2663,
+      "step": 1674
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.548791018998273e-05,
+      "loss": 0.2254,
+      "step": 1675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.548521157167531e-05,
+      "loss": 0.3611,
+      "step": 1676
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.548251295336788e-05,
+      "loss": 0.1125,
+      "step": 1677
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.547981433506045e-05,
+      "loss": 0.2615,
+      "step": 1678
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5477115716753025e-05,
+      "loss": 0.1129,
+      "step": 1679
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.54744170984456e-05,
+      "loss": 0.4358,
+      "step": 1680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.547171848013817e-05,
+      "loss": 0.0883,
+      "step": 1681
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.546901986183075e-05,
+      "loss": 0.1172,
+      "step": 1682
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.546632124352332e-05,
+      "loss": 0.0887,
+      "step": 1683
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.546362262521589e-05,
+      "loss": 0.3768,
+      "step": 1684
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5460924006908464e-05,
+      "loss": 0.0982,
+      "step": 1685
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5458225388601037e-05,
+      "loss": 0.2557,
+      "step": 1686
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.545552677029361e-05,
+      "loss": 0.2566,
+      "step": 1687
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.545282815198619e-05,
+      "loss": 0.1468,
+      "step": 1688
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.545012953367876e-05,
+      "loss": 0.0925,
+      "step": 1689
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.544743091537133e-05,
+      "loss": 0.2014,
+      "step": 1690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.544473229706391e-05,
+      "loss": 0.2861,
+      "step": 1691
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.544203367875648e-05,
+      "loss": 0.4424,
+      "step": 1692
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.543933506044905e-05,
+      "loss": 0.1293,
+      "step": 1693
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.543663644214163e-05,
+      "loss": 0.3105,
+      "step": 1694
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.54339378238342e-05,
+      "loss": 0.218,
+      "step": 1695
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.543123920552677e-05,
+      "loss": 0.1537,
+      "step": 1696
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.542854058721935e-05,
+      "loss": 0.5119,
+      "step": 1697
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.542584196891192e-05,
+      "loss": 0.2633,
+      "step": 1698
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5423143350604494e-05,
+      "loss": 0.073,
+      "step": 1699
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5420444732297066e-05,
+      "loss": 0.3607,
+      "step": 1700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.541774611398964e-05,
+      "loss": 0.1008,
+      "step": 1701
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.541504749568221e-05,
+      "loss": 0.113,
+      "step": 1702
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.541234887737479e-05,
+      "loss": 0.0789,
+      "step": 1703
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.540965025906736e-05,
+      "loss": 0.0856,
+      "step": 1704
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.540695164075993e-05,
+      "loss": 0.1261,
+      "step": 1705
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5404253022452505e-05,
+      "loss": 0.2241,
+      "step": 1706
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.540155440414508e-05,
+      "loss": 0.0824,
+      "step": 1707
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.539885578583765e-05,
+      "loss": 0.2532,
+      "step": 1708
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.539615716753023e-05,
+      "loss": 0.1245,
+      "step": 1709
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.53934585492228e-05,
+      "loss": 0.0805,
+      "step": 1710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.539075993091537e-05,
+      "loss": 0.0992,
+      "step": 1711
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.538806131260795e-05,
+      "loss": 0.2982,
+      "step": 1712
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5385362694300523e-05,
+      "loss": 0.5432,
+      "step": 1713
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.538266407599309e-05,
+      "loss": 0.3725,
+      "step": 1714
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.537996545768567e-05,
+      "loss": 0.1513,
+      "step": 1715
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.537726683937824e-05,
+      "loss": 0.3132,
+      "step": 1716
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.537456822107081e-05,
+      "loss": 0.2767,
+      "step": 1717
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.537186960276339e-05,
+      "loss": 0.1019,
+      "step": 1718
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.536917098445596e-05,
+      "loss": 0.0881,
+      "step": 1719
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5366472366148535e-05,
+      "loss": 0.2853,
+      "step": 1720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.536377374784111e-05,
+      "loss": 0.1506,
+      "step": 1721
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.536107512953368e-05,
+      "loss": 0.3643,
+      "step": 1722
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.535837651122625e-05,
+      "loss": 0.1113,
+      "step": 1723
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.535567789291883e-05,
+      "loss": 0.0957,
+      "step": 1724
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.53529792746114e-05,
+      "loss": 0.3374,
+      "step": 1725
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5350280656303974e-05,
+      "loss": 0.337,
+      "step": 1726
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.534758203799655e-05,
+      "loss": 0.2201,
+      "step": 1727
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.534488341968912e-05,
+      "loss": 0.1002,
+      "step": 1728
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.534218480138169e-05,
+      "loss": 0.4509,
+      "step": 1729
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.533948618307427e-05,
+      "loss": 0.0835,
+      "step": 1730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.533678756476684e-05,
+      "loss": 0.1888,
+      "step": 1731
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5334088946459413e-05,
+      "loss": 0.2026,
+      "step": 1732
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.533139032815199e-05,
+      "loss": 0.0897,
+      "step": 1733
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5328691709844564e-05,
+      "loss": 0.3332,
+      "step": 1734
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5325993091537136e-05,
+      "loss": 0.7781,
+      "step": 1735
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.532329447322971e-05,
+      "loss": 0.3543,
+      "step": 1736
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.532059585492228e-05,
+      "loss": 0.5832,
+      "step": 1737
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.531789723661485e-05,
+      "loss": 0.2267,
+      "step": 1738
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.531519861830743e-05,
+      "loss": 0.0832,
+      "step": 1739
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5312500000000004e-05,
+      "loss": 0.2054,
+      "step": 1740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5309801381692576e-05,
+      "loss": 0.2775,
+      "step": 1741
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.530710276338515e-05,
+      "loss": 0.3715,
+      "step": 1742
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.530440414507772e-05,
+      "loss": 0.2464,
+      "step": 1743
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.530170552677029e-05,
+      "loss": 0.467,
+      "step": 1744
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.529900690846287e-05,
+      "loss": 0.1014,
+      "step": 1745
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.529630829015544e-05,
+      "loss": 0.108,
+      "step": 1746
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5293609671848015e-05,
+      "loss": 0.2993,
+      "step": 1747
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5290911053540594e-05,
+      "loss": 0.4492,
+      "step": 1748
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5288212435233166e-05,
+      "loss": 0.5136,
+      "step": 1749
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.528551381692573e-05,
+      "loss": 0.1752,
+      "step": 1750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.528281519861831e-05,
+      "loss": 0.0871,
+      "step": 1751
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.528011658031088e-05,
+      "loss": 0.1472,
+      "step": 1752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5277417962003454e-05,
+      "loss": 0.3684,
+      "step": 1753
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.527471934369603e-05,
+      "loss": 0.3413,
+      "step": 1754
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5272020725388605e-05,
+      "loss": 0.2062,
+      "step": 1755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.526932210708118e-05,
+      "loss": 0.3459,
+      "step": 1756
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.526662348877375e-05,
+      "loss": 0.2154,
+      "step": 1757
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.526392487046632e-05,
+      "loss": 0.1012,
+      "step": 1758
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5261226252158894e-05,
+      "loss": 0.2515,
+      "step": 1759
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.525852763385147e-05,
+      "loss": 0.4757,
+      "step": 1760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5255829015544045e-05,
+      "loss": 0.2125,
+      "step": 1761
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.525313039723662e-05,
+      "loss": 0.2435,
+      "step": 1762
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.525043177892919e-05,
+      "loss": 0.2174,
+      "step": 1763
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.524773316062176e-05,
+      "loss": 0.0976,
+      "step": 1764
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.524503454231433e-05,
+      "loss": 0.291,
+      "step": 1765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.524233592400691e-05,
+      "loss": 0.1775,
+      "step": 1766
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5239637305699484e-05,
+      "loss": 0.2667,
+      "step": 1767
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5236938687392056e-05,
+      "loss": 0.3341,
+      "step": 1768
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5234240069084635e-05,
+      "loss": 0.1459,
+      "step": 1769
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.523154145077721e-05,
+      "loss": 0.2994,
+      "step": 1770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.522884283246977e-05,
+      "loss": 0.1824,
+      "step": 1771
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.522614421416235e-05,
+      "loss": 0.2072,
+      "step": 1772
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.522344559585492e-05,
+      "loss": 0.0753,
+      "step": 1773
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5220746977547495e-05,
+      "loss": 0.202,
+      "step": 1774
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5218048359240074e-05,
+      "loss": 0.2577,
+      "step": 1775
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5215349740932646e-05,
+      "loss": 0.338,
+      "step": 1776
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.521265112262522e-05,
+      "loss": 0.09,
+      "step": 1777
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.520995250431779e-05,
+      "loss": 0.1579,
+      "step": 1778
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.520725388601036e-05,
+      "loss": 0.1505,
+      "step": 1779
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5204555267702934e-05,
+      "loss": 0.1707,
+      "step": 1780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.520185664939551e-05,
+      "loss": 0.4934,
+      "step": 1781
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5199158031088085e-05,
+      "loss": 0.0946,
+      "step": 1782
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.519645941278066e-05,
+      "loss": 0.3456,
+      "step": 1783
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5193760794473236e-05,
+      "loss": 0.2722,
+      "step": 1784
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.51910621761658e-05,
+      "loss": 0.0805,
+      "step": 1785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5188363557858374e-05,
+      "loss": 0.186,
+      "step": 1786
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.518566493955095e-05,
+      "loss": 0.1333,
+      "step": 1787
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5182966321243525e-05,
+      "loss": 0.1286,
+      "step": 1788
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.51802677029361e-05,
+      "loss": 0.2844,
+      "step": 1789
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5177569084628676e-05,
+      "loss": 0.29,
+      "step": 1790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.517487046632125e-05,
+      "loss": 0.7638,
+      "step": 1791
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.517217184801382e-05,
+      "loss": 0.096,
+      "step": 1792
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.516947322970639e-05,
+      "loss": 0.1464,
+      "step": 1793
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5166774611398964e-05,
+      "loss": 0.0791,
+      "step": 1794
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5164075993091536e-05,
+      "loss": 0.1013,
+      "step": 1795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5161377374784115e-05,
+      "loss": 0.2563,
+      "step": 1796
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.515867875647669e-05,
+      "loss": 0.1583,
+      "step": 1797
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.515598013816926e-05,
+      "loss": 0.1204,
+      "step": 1798
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.515328151986183e-05,
+      "loss": 0.0977,
+      "step": 1799
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.51505829015544e-05,
+      "loss": 0.2108,
+      "step": 1800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5147884283246975e-05,
+      "loss": 0.2213,
+      "step": 1801
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5145185664939554e-05,
+      "loss": 0.1445,
+      "step": 1802
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5142487046632126e-05,
+      "loss": 0.2237,
+      "step": 1803
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.51397884283247e-05,
+      "loss": 0.3899,
+      "step": 1804
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.513708981001728e-05,
+      "loss": 0.2301,
+      "step": 1805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.513439119170985e-05,
+      "loss": 0.3372,
+      "step": 1806
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5131692573402415e-05,
+      "loss": 0.0971,
+      "step": 1807
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5128993955094993e-05,
+      "loss": 0.2256,
+      "step": 1808
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5126295336787566e-05,
+      "loss": 0.1316,
+      "step": 1809
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.512359671848014e-05,
+      "loss": 0.0813,
+      "step": 1810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5120898100172717e-05,
+      "loss": 0.3247,
+      "step": 1811
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.511819948186529e-05,
+      "loss": 0.0743,
+      "step": 1812
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.511550086355786e-05,
+      "loss": 0.238,
+      "step": 1813
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.511280224525043e-05,
+      "loss": 0.0971,
+      "step": 1814
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5110103626943005e-05,
+      "loss": 0.091,
+      "step": 1815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.510740500863558e-05,
+      "loss": 0.1396,
+      "step": 1816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5104706390328156e-05,
+      "loss": 0.1785,
+      "step": 1817
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.510200777202073e-05,
+      "loss": 0.1634,
+      "step": 1818
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.50993091537133e-05,
+      "loss": 0.3011,
+      "step": 1819
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.509661053540587e-05,
+      "loss": 0.1186,
+      "step": 1820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5093911917098444e-05,
+      "loss": 0.1013,
+      "step": 1821
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5091213298791016e-05,
+      "loss": 0.3286,
+      "step": 1822
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5088514680483595e-05,
+      "loss": 0.124,
+      "step": 1823
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.508581606217617e-05,
+      "loss": 0.1039,
+      "step": 1824
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.508311744386874e-05,
+      "loss": 0.1634,
+      "step": 1825
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.508041882556132e-05,
+      "loss": 0.2774,
+      "step": 1826
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.507772020725389e-05,
+      "loss": 0.1584,
+      "step": 1827
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5075021588946456e-05,
+      "loss": 0.2404,
+      "step": 1828
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5072322970639034e-05,
+      "loss": 0.3386,
+      "step": 1829
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5069624352331606e-05,
+      "loss": 0.7029,
+      "step": 1830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.506692573402418e-05,
+      "loss": 0.3026,
+      "step": 1831
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.506422711571676e-05,
+      "loss": 0.2242,
+      "step": 1832
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.506152849740933e-05,
+      "loss": 0.2642,
+      "step": 1833
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.50588298791019e-05,
+      "loss": 0.3141,
+      "step": 1834
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5056131260794474e-05,
+      "loss": 0.1649,
+      "step": 1835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5053432642487046e-05,
+      "loss": 0.1576,
+      "step": 1836
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5050734024179625e-05,
+      "loss": 0.3764,
+      "step": 1837
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.50480354058722e-05,
+      "loss": 0.1322,
+      "step": 1838
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.504533678756477e-05,
+      "loss": 0.2429,
+      "step": 1839
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.504263816925735e-05,
+      "loss": 0.0897,
+      "step": 1840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.503993955094992e-05,
+      "loss": 0.0737,
+      "step": 1841
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5037240932642485e-05,
+      "loss": 0.1819,
+      "step": 1842
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5034542314335064e-05,
+      "loss": 0.256,
+      "step": 1843
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5031843696027636e-05,
+      "loss": 0.1273,
+      "step": 1844
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.502914507772021e-05,
+      "loss": 0.1877,
+      "step": 1845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.502644645941279e-05,
+      "loss": 0.2688,
+      "step": 1846
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.502374784110536e-05,
+      "loss": 0.1504,
+      "step": 1847
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.502104922279793e-05,
+      "loss": 0.2458,
+      "step": 1848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.50183506044905e-05,
+      "loss": 0.4037,
+      "step": 1849
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5015651986183075e-05,
+      "loss": 0.0934,
+      "step": 1850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.501295336787565e-05,
+      "loss": 0.1874,
+      "step": 1851
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5010254749568226e-05,
+      "loss": 0.0817,
+      "step": 1852
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.50075561312608e-05,
+      "loss": 0.1211,
+      "step": 1853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.500485751295337e-05,
+      "loss": 0.2248,
+      "step": 1854
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.500215889464595e-05,
+      "loss": 0.0783,
+      "step": 1855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4999460276338515e-05,
+      "loss": 0.0904,
+      "step": 1856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.499676165803109e-05,
+      "loss": 0.3726,
+      "step": 1857
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4994063039723665e-05,
+      "loss": 0.2611,
+      "step": 1858
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.499136442141624e-05,
+      "loss": 0.2514,
+      "step": 1859
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.498866580310881e-05,
+      "loss": 0.1621,
+      "step": 1860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.498596718480139e-05,
+      "loss": 0.5909,
+      "step": 1861
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.498326856649396e-05,
+      "loss": 0.1166,
+      "step": 1862
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.498056994818653e-05,
+      "loss": 0.3851,
+      "step": 1863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4977871329879105e-05,
+      "loss": 0.0786,
+      "step": 1864
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.497517271157168e-05,
+      "loss": 0.1613,
+      "step": 1865
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.497247409326425e-05,
+      "loss": 0.5949,
+      "step": 1866
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.496977547495683e-05,
+      "loss": 0.1777,
+      "step": 1867
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.49670768566494e-05,
+      "loss": 0.0911,
+      "step": 1868
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.496437823834197e-05,
+      "loss": 0.2138,
+      "step": 1869
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4961679620034544e-05,
+      "loss": 0.0772,
+      "step": 1870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4958981001727116e-05,
+      "loss": 0.0812,
+      "step": 1871
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.495628238341969e-05,
+      "loss": 0.2649,
+      "step": 1872
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.495358376511227e-05,
+      "loss": 0.106,
+      "step": 1873
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.495088514680484e-05,
+      "loss": 0.1949,
+      "step": 1874
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.494818652849741e-05,
+      "loss": 0.2651,
+      "step": 1875
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.494548791018999e-05,
+      "loss": 0.1084,
+      "step": 1876
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.494278929188256e-05,
+      "loss": 0.2124,
+      "step": 1877
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.494009067357513e-05,
+      "loss": 0.4155,
+      "step": 1878
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4937392055267706e-05,
+      "loss": 0.1987,
+      "step": 1879
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.493469343696028e-05,
+      "loss": 0.1035,
+      "step": 1880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.493199481865285e-05,
+      "loss": 0.2289,
+      "step": 1881
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.492929620034543e-05,
+      "loss": 0.1261,
+      "step": 1882
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4926597582038e-05,
+      "loss": 0.3879,
+      "step": 1883
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4923898963730574e-05,
+      "loss": 0.1601,
+      "step": 1884
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4921200345423146e-05,
+      "loss": 0.1351,
+      "step": 1885
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.491850172711572e-05,
+      "loss": 0.1949,
+      "step": 1886
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.491580310880829e-05,
+      "loss": 0.2789,
+      "step": 1887
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.491310449050087e-05,
+      "loss": 0.2075,
+      "step": 1888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.491040587219344e-05,
+      "loss": 0.1748,
+      "step": 1889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.490770725388601e-05,
+      "loss": 0.3089,
+      "step": 1890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4905008635578585e-05,
+      "loss": 0.0926,
+      "step": 1891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.490231001727116e-05,
+      "loss": 0.1746,
+      "step": 1892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.489961139896373e-05,
+      "loss": 0.5534,
+      "step": 1893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.489691278065631e-05,
+      "loss": 0.2189,
+      "step": 1894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.489421416234888e-05,
+      "loss": 0.1429,
+      "step": 1895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.489151554404145e-05,
+      "loss": 0.2663,
+      "step": 1896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.488881692573403e-05,
+      "loss": 0.2236,
+      "step": 1897
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.48861183074266e-05,
+      "loss": 0.2426,
+      "step": 1898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.488341968911917e-05,
+      "loss": 0.0842,
+      "step": 1899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.488072107081175e-05,
+      "loss": 0.3212,
+      "step": 1900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.487802245250432e-05,
+      "loss": 0.1853,
+      "step": 1901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.487532383419689e-05,
+      "loss": 0.1158,
+      "step": 1902
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.487262521588947e-05,
+      "loss": 0.0848,
+      "step": 1903
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.486992659758204e-05,
+      "loss": 0.3545,
+      "step": 1904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4867227979274614e-05,
+      "loss": 0.2832,
+      "step": 1905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4864529360967187e-05,
+      "loss": 0.3249,
+      "step": 1906
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.486183074265976e-05,
+      "loss": 0.1456,
+      "step": 1907
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.485913212435233e-05,
+      "loss": 0.1234,
+      "step": 1908
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.485643350604491e-05,
+      "loss": 0.2648,
+      "step": 1909
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.485373488773748e-05,
+      "loss": 0.3891,
+      "step": 1910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4851036269430054e-05,
+      "loss": 0.2443,
+      "step": 1911
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.484833765112263e-05,
+      "loss": 0.1823,
+      "step": 1912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.48456390328152e-05,
+      "loss": 0.1549,
+      "step": 1913
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.484294041450777e-05,
+      "loss": 0.2341,
+      "step": 1914
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.484024179620035e-05,
+      "loss": 0.0813,
+      "step": 1915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.483754317789292e-05,
+      "loss": 0.1371,
+      "step": 1916
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.483484455958549e-05,
+      "loss": 0.0731,
+      "step": 1917
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.483214594127807e-05,
+      "loss": 0.2758,
+      "step": 1918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4829447322970644e-05,
+      "loss": 0.0917,
+      "step": 1919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4826748704663216e-05,
+      "loss": 0.0855,
+      "step": 1920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.482405008635579e-05,
+      "loss": 0.1291,
+      "step": 1921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.482135146804836e-05,
+      "loss": 0.0893,
+      "step": 1922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.481865284974093e-05,
+      "loss": 0.2704,
+      "step": 1923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.481595423143351e-05,
+      "loss": 0.2187,
+      "step": 1924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.481325561312608e-05,
+      "loss": 0.1177,
+      "step": 1925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4810556994818655e-05,
+      "loss": 0.3396,
+      "step": 1926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.480785837651123e-05,
+      "loss": 0.2158,
+      "step": 1927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.48051597582038e-05,
+      "loss": 0.2064,
+      "step": 1928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.480246113989637e-05,
+      "loss": 0.3791,
+      "step": 1929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.479976252158895e-05,
+      "loss": 0.2448,
+      "step": 1930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.479706390328152e-05,
+      "loss": 0.1494,
+      "step": 1931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4794365284974095e-05,
+      "loss": 0.1985,
+      "step": 1932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4791666666666673e-05,
+      "loss": 0.1659,
+      "step": 1933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4788968048359246e-05,
+      "loss": 0.2878,
+      "step": 1934
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.478626943005181e-05,
+      "loss": 0.0946,
+      "step": 1935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.478357081174439e-05,
+      "loss": 0.1544,
+      "step": 1936
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.478087219343696e-05,
+      "loss": 0.2237,
+      "step": 1937
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4778173575129534e-05,
+      "loss": 0.1382,
+      "step": 1938
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.477547495682211e-05,
+      "loss": 0.2139,
+      "step": 1939
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4772776338514685e-05,
+      "loss": 0.1261,
+      "step": 1940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.477007772020726e-05,
+      "loss": 0.0877,
+      "step": 1941
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.476737910189983e-05,
+      "loss": 0.1691,
+      "step": 1942
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.47646804835924e-05,
+      "loss": 0.1753,
+      "step": 1943
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.476198186528497e-05,
+      "loss": 0.1353,
+      "step": 1944
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.475928324697755e-05,
+      "loss": 0.1573,
+      "step": 1945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4756584628670124e-05,
+      "loss": 0.2778,
+      "step": 1946
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4753886010362696e-05,
+      "loss": 0.1734,
+      "step": 1947
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.475118739205527e-05,
+      "loss": 0.1228,
+      "step": 1948
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.474848877374784e-05,
+      "loss": 0.1384,
+      "step": 1949
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.474579015544041e-05,
+      "loss": 0.4211,
+      "step": 1950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.474309153713299e-05,
+      "loss": 0.1059,
+      "step": 1951
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4740392918825563e-05,
+      "loss": 0.2468,
+      "step": 1952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4737694300518135e-05,
+      "loss": 0.1485,
+      "step": 1953
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4734995682210714e-05,
+      "loss": 0.0975,
+      "step": 1954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4732297063903286e-05,
+      "loss": 0.2809,
+      "step": 1955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.472959844559585e-05,
+      "loss": 0.1873,
+      "step": 1956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.472689982728843e-05,
+      "loss": 0.0821,
+      "step": 1957
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4724201208981e-05,
+      "loss": 0.2967,
+      "step": 1958
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4721502590673575e-05,
+      "loss": 0.1298,
+      "step": 1959
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4718803972366154e-05,
+      "loss": 0.3165,
+      "step": 1960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4716105354058726e-05,
+      "loss": 0.1937,
+      "step": 1961
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.47134067357513e-05,
+      "loss": 0.22,
+      "step": 1962
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.471070811744387e-05,
+      "loss": 0.1326,
+      "step": 1963
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.470800949913644e-05,
+      "loss": 0.1958,
+      "step": 1964
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4705310880829014e-05,
+      "loss": 0.3647,
+      "step": 1965
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.470261226252159e-05,
+      "loss": 0.0845,
+      "step": 1966
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4699913644214165e-05,
+      "loss": 0.1117,
+      "step": 1967
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.469721502590674e-05,
+      "loss": 0.1656,
+      "step": 1968
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4694516407599316e-05,
+      "loss": 0.4254,
+      "step": 1969
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.469181778929188e-05,
+      "loss": 0.2862,
+      "step": 1970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.468911917098445e-05,
+      "loss": 0.3956,
+      "step": 1971
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.468642055267703e-05,
+      "loss": 0.2769,
+      "step": 1972
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4683721934369604e-05,
+      "loss": 0.2738,
+      "step": 1973
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4681023316062176e-05,
+      "loss": 0.1362,
+      "step": 1974
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4678324697754755e-05,
+      "loss": 0.0897,
+      "step": 1975
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.467562607944733e-05,
+      "loss": 0.1764,
+      "step": 1976
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.46729274611399e-05,
+      "loss": 0.1315,
+      "step": 1977
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.467022884283247e-05,
+      "loss": 0.0797,
+      "step": 1978
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4667530224525044e-05,
+      "loss": 0.2669,
+      "step": 1979
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4664831606217616e-05,
+      "loss": 0.1431,
+      "step": 1980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4662132987910195e-05,
+      "loss": 0.0981,
+      "step": 1981
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4659434369602767e-05,
+      "loss": 0.1947,
+      "step": 1982
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.465673575129534e-05,
+      "loss": 0.1357,
+      "step": 1983
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.465403713298791e-05,
+      "loss": 0.1971,
+      "step": 1984
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.465133851468048e-05,
+      "loss": 0.2502,
+      "step": 1985
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4648639896373055e-05,
+      "loss": 0.0897,
+      "step": 1986
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4645941278065634e-05,
+      "loss": 0.0933,
+      "step": 1987
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4643242659758206e-05,
+      "loss": 0.1722,
+      "step": 1988
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.464054404145078e-05,
+      "loss": 0.0785,
+      "step": 1989
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.463784542314336e-05,
+      "loss": 0.2192,
+      "step": 1990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.463514680483593e-05,
+      "loss": 0.0773,
+      "step": 1991
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4632448186528494e-05,
+      "loss": 0.3104,
+      "step": 1992
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.462974956822107e-05,
+      "loss": 0.4815,
+      "step": 1993
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4627050949913645e-05,
+      "loss": 0.5066,
+      "step": 1994
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.462435233160622e-05,
+      "loss": 0.294,
+      "step": 1995
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4621653713298796e-05,
+      "loss": 0.4206,
+      "step": 1996
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.461895509499137e-05,
+      "loss": 0.1915,
+      "step": 1997
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.461625647668394e-05,
+      "loss": 0.2976,
+      "step": 1998
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.461355785837651e-05,
+      "loss": 0.3442,
+      "step": 1999
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4610859240069084e-05,
+      "loss": 0.1853,
+      "step": 2000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4608160621761657e-05,
+      "loss": 0.1403,
+      "step": 2001
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4605462003454235e-05,
+      "loss": 0.0816,
+      "step": 2002
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.460276338514681e-05,
+      "loss": 0.1407,
+      "step": 2003
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.460006476683938e-05,
+      "loss": 0.1048,
+      "step": 2004
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.459736614853195e-05,
+      "loss": 0.2574,
+      "step": 2005
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4594667530224524e-05,
+      "loss": 0.1993,
+      "step": 2006
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4591968911917096e-05,
+      "loss": 0.0862,
+      "step": 2007
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4589270293609675e-05,
+      "loss": 0.1814,
+      "step": 2008
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.458657167530225e-05,
+      "loss": 0.4546,
+      "step": 2009
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.458387305699482e-05,
+      "loss": 0.2175,
+      "step": 2010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.45811744386874e-05,
+      "loss": 0.0836,
+      "step": 2011
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.457847582037997e-05,
+      "loss": 0.1446,
+      "step": 2012
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4575777202072535e-05,
+      "loss": 0.3536,
+      "step": 2013
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4573078583765114e-05,
+      "loss": 0.3222,
+      "step": 2014
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4570379965457686e-05,
+      "loss": 0.2781,
+      "step": 2015
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.456768134715026e-05,
+      "loss": 0.4506,
+      "step": 2016
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.456498272884284e-05,
+      "loss": 0.3773,
+      "step": 2017
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.456228411053541e-05,
+      "loss": 0.1043,
+      "step": 2018
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.455958549222798e-05,
+      "loss": 0.1526,
+      "step": 2019
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.455688687392055e-05,
+      "loss": 0.41,
+      "step": 2020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4554188255613125e-05,
+      "loss": 0.1635,
+      "step": 2021
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.45514896373057e-05,
+      "loss": 0.2992,
+      "step": 2022
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4548791018998276e-05,
+      "loss": 0.1003,
+      "step": 2023
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.454609240069085e-05,
+      "loss": 0.1001,
+      "step": 2024
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.454339378238342e-05,
+      "loss": 0.3107,
+      "step": 2025
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4540695164076e-05,
+      "loss": 0.2745,
+      "step": 2026
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4537996545768565e-05,
+      "loss": 0.1718,
+      "step": 2027
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.453529792746114e-05,
+      "loss": 0.0873,
+      "step": 2028
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4532599309153716e-05,
+      "loss": 0.0882,
+      "step": 2029
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.452990069084629e-05,
+      "loss": 0.2113,
+      "step": 2030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4527202072538867e-05,
+      "loss": 0.1716,
+      "step": 2031
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.452450345423144e-05,
+      "loss": 0.3672,
+      "step": 2032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.452180483592401e-05,
+      "loss": 0.2353,
+      "step": 2033
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.451910621761658e-05,
+      "loss": 0.1952,
+      "step": 2034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4516407599309155e-05,
+      "loss": 0.1917,
+      "step": 2035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.451370898100173e-05,
+      "loss": 0.2631,
+      "step": 2036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4511010362694306e-05,
+      "loss": 0.3266,
+      "step": 2037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.450831174438688e-05,
+      "loss": 0.1857,
+      "step": 2038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.450561312607945e-05,
+      "loss": 0.0947,
+      "step": 2039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.450291450777203e-05,
+      "loss": 0.1686,
+      "step": 2040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4500215889464594e-05,
+      "loss": 0.127,
+      "step": 2041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4497517271157166e-05,
+      "loss": 0.4092,
+      "step": 2042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4494818652849745e-05,
+      "loss": 0.1134,
+      "step": 2043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.449212003454232e-05,
+      "loss": 0.0944,
+      "step": 2044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.448942141623489e-05,
+      "loss": 0.1443,
+      "step": 2045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.448672279792747e-05,
+      "loss": 0.167,
+      "step": 2046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.448402417962004e-05,
+      "loss": 0.3016,
+      "step": 2047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.448132556131261e-05,
+      "loss": 0.4139,
+      "step": 2048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4478626943005184e-05,
+      "loss": 0.072,
+      "step": 2049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4475928324697756e-05,
+      "loss": 0.2719,
+      "step": 2050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.447322970639033e-05,
+      "loss": 0.2845,
+      "step": 2051
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.447053108808291e-05,
+      "loss": 0.2757,
+      "step": 2052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.446783246977548e-05,
+      "loss": 0.1829,
+      "step": 2053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.446513385146805e-05,
+      "loss": 0.2628,
+      "step": 2054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4462435233160624e-05,
+      "loss": 0.1651,
+      "step": 2055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4459736614853196e-05,
+      "loss": 0.0857,
+      "step": 2056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.445703799654577e-05,
+      "loss": 0.7479,
+      "step": 2057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.445433937823835e-05,
+      "loss": 0.1,
+      "step": 2058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.445164075993092e-05,
+      "loss": 0.278,
+      "step": 2059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.444894214162349e-05,
+      "loss": 0.122,
+      "step": 2060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.444624352331607e-05,
+      "loss": 0.3172,
+      "step": 2061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4443544905008635e-05,
+      "loss": 0.3773,
+      "step": 2062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.444084628670121e-05,
+      "loss": 0.0828,
+      "step": 2063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4438147668393786e-05,
+      "loss": 0.0854,
+      "step": 2064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.443544905008636e-05,
+      "loss": 0.3596,
+      "step": 2065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.443275043177893e-05,
+      "loss": 0.0746,
+      "step": 2066
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.443005181347151e-05,
+      "loss": 0.1224,
+      "step": 2067
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.442735319516408e-05,
+      "loss": 0.3505,
+      "step": 2068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.442465457685665e-05,
+      "loss": 0.0916,
+      "step": 2069
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4421955958549225e-05,
+      "loss": 0.2112,
+      "step": 2070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.44192573402418e-05,
+      "loss": 0.3032,
+      "step": 2071
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.441655872193437e-05,
+      "loss": 0.129,
+      "step": 2072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.441386010362695e-05,
+      "loss": 0.0786,
+      "step": 2073
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.441116148531952e-05,
+      "loss": 0.1245,
+      "step": 2074
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.440846286701209e-05,
+      "loss": 0.2242,
+      "step": 2075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4405764248704665e-05,
+      "loss": 0.0822,
+      "step": 2076
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4403065630397237e-05,
+      "loss": 0.4759,
+      "step": 2077
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.440036701208981e-05,
+      "loss": 0.3211,
+      "step": 2078
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.439766839378239e-05,
+      "loss": 0.4349,
+      "step": 2079
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.439496977547496e-05,
+      "loss": 0.0882,
+      "step": 2080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.439227115716753e-05,
+      "loss": 0.2386,
+      "step": 2081
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.438957253886011e-05,
+      "loss": 0.3054,
+      "step": 2082
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.438687392055268e-05,
+      "loss": 0.1678,
+      "step": 2083
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.438417530224525e-05,
+      "loss": 0.2709,
+      "step": 2084
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.438147668393783e-05,
+      "loss": 0.1232,
+      "step": 2085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.43787780656304e-05,
+      "loss": 0.1905,
+      "step": 2086
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.437607944732297e-05,
+      "loss": 0.1799,
+      "step": 2087
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.437338082901555e-05,
+      "loss": 0.0912,
+      "step": 2088
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.437068221070812e-05,
+      "loss": 0.1971,
+      "step": 2089
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4367983592400694e-05,
+      "loss": 0.2096,
+      "step": 2090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4365284974093266e-05,
+      "loss": 0.3728,
+      "step": 2091
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.436258635578584e-05,
+      "loss": 0.0933,
+      "step": 2092
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.435988773747841e-05,
+      "loss": 0.087,
+      "step": 2093
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.435718911917099e-05,
+      "loss": 0.1272,
+      "step": 2094
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.435449050086356e-05,
+      "loss": 0.4163,
+      "step": 2095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.435179188255613e-05,
+      "loss": 0.1567,
+      "step": 2096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.434909326424871e-05,
+      "loss": 0.323,
+      "step": 2097
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.434639464594128e-05,
+      "loss": 0.0866,
+      "step": 2098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.434369602763385e-05,
+      "loss": 0.1318,
+      "step": 2099
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.434099740932643e-05,
+      "loss": 0.4524,
+      "step": 2100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4338298791019e-05,
+      "loss": 0.2623,
+      "step": 2101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.433560017271157e-05,
+      "loss": 0.3098,
+      "step": 2102
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.433290155440415e-05,
+      "loss": 0.4955,
+      "step": 2103
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4330202936096724e-05,
+      "loss": 0.0833,
+      "step": 2104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4327504317789296e-05,
+      "loss": 0.1244,
+      "step": 2105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.432480569948187e-05,
+      "loss": 0.1337,
+      "step": 2106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.432210708117444e-05,
+      "loss": 0.0938,
+      "step": 2107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.431940846286701e-05,
+      "loss": 0.4148,
+      "step": 2108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.431670984455959e-05,
+      "loss": 0.3078,
+      "step": 2109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.431401122625216e-05,
+      "loss": 0.1575,
+      "step": 2110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4311312607944735e-05,
+      "loss": 0.1863,
+      "step": 2111
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.430861398963731e-05,
+      "loss": 0.1833,
+      "step": 2112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.430591537132988e-05,
+      "loss": 0.1367,
+      "step": 2113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.430321675302245e-05,
+      "loss": 0.0938,
+      "step": 2114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.430051813471503e-05,
+      "loss": 0.0956,
+      "step": 2115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.42978195164076e-05,
+      "loss": 0.3972,
+      "step": 2116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4295120898100174e-05,
+      "loss": 0.1879,
+      "step": 2117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.429242227979275e-05,
+      "loss": 0.3045,
+      "step": 2118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.428972366148532e-05,
+      "loss": 0.1008,
+      "step": 2119
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.428702504317789e-05,
+      "loss": 0.0821,
+      "step": 2120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.428432642487047e-05,
+      "loss": 0.0732,
+      "step": 2121
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.428162780656304e-05,
+      "loss": 0.0795,
+      "step": 2122
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4278929188255613e-05,
+      "loss": 0.1337,
+      "step": 2123
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.427623056994819e-05,
+      "loss": 0.2771,
+      "step": 2124
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4273531951640764e-05,
+      "loss": 0.2597,
+      "step": 2125
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4270833333333337e-05,
+      "loss": 0.2518,
+      "step": 2126
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.426813471502591e-05,
+      "loss": 0.0701,
+      "step": 2127
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.426543609671848e-05,
+      "loss": 0.2851,
+      "step": 2128
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.426273747841105e-05,
+      "loss": 0.2759,
+      "step": 2129
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.426003886010363e-05,
+      "loss": 0.1017,
+      "step": 2130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4257340241796204e-05,
+      "loss": 0.1158,
+      "step": 2131
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4254641623488776e-05,
+      "loss": 0.0814,
+      "step": 2132
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.425194300518135e-05,
+      "loss": 0.2458,
+      "step": 2133
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.424924438687392e-05,
+      "loss": 0.1112,
+      "step": 2134
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.424654576856649e-05,
+      "loss": 0.0852,
+      "step": 2135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.424384715025907e-05,
+      "loss": 0.2107,
+      "step": 2136
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.424114853195164e-05,
+      "loss": 0.3519,
+      "step": 2137
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4238449913644215e-05,
+      "loss": 0.3001,
+      "step": 2138
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4235751295336794e-05,
+      "loss": 0.0918,
+      "step": 2139
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4233052677029366e-05,
+      "loss": 0.1325,
+      "step": 2140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.423035405872193e-05,
+      "loss": 0.2866,
+      "step": 2141
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.422765544041451e-05,
+      "loss": 0.2004,
+      "step": 2142
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.422495682210708e-05,
+      "loss": 0.4555,
+      "step": 2143
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4222258203799654e-05,
+      "loss": 0.2007,
+      "step": 2144
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.421955958549223e-05,
+      "loss": 0.2727,
+      "step": 2145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4216860967184805e-05,
+      "loss": 0.1483,
+      "step": 2146
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.421416234887738e-05,
+      "loss": 0.9444,
+      "step": 2147
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.421146373056995e-05,
+      "loss": 0.2107,
+      "step": 2148
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.420876511226252e-05,
+      "loss": 0.069,
+      "step": 2149
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4206066493955094e-05,
+      "loss": 0.1244,
+      "step": 2150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.420336787564767e-05,
+      "loss": 0.2657,
+      "step": 2151
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4200669257340245e-05,
+      "loss": 0.0732,
+      "step": 2152
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.419797063903282e-05,
+      "loss": 0.0909,
+      "step": 2153
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4195272020725396e-05,
+      "loss": 0.3367,
+      "step": 2154
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.419257340241796e-05,
+      "loss": 0.0802,
+      "step": 2155
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.418987478411053e-05,
+      "loss": 0.2257,
+      "step": 2156
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.418717616580311e-05,
+      "loss": 0.2364,
+      "step": 2157
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4184477547495684e-05,
+      "loss": 0.181,
+      "step": 2158
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4181778929188256e-05,
+      "loss": 0.2359,
+      "step": 2159
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4179080310880835e-05,
+      "loss": 0.0929,
+      "step": 2160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.417638169257341e-05,
+      "loss": 0.0752,
+      "step": 2161
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.417368307426598e-05,
+      "loss": 0.4853,
+      "step": 2162
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.417098445595855e-05,
+      "loss": 0.1052,
+      "step": 2163
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.416828583765112e-05,
+      "loss": 0.4002,
+      "step": 2164
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4165587219343695e-05,
+      "loss": 0.2911,
+      "step": 2165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4162888601036274e-05,
+      "loss": 0.245,
+      "step": 2166
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4160189982728846e-05,
+      "loss": 0.4192,
+      "step": 2167
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.415749136442142e-05,
+      "loss": 0.0997,
+      "step": 2168
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.415479274611399e-05,
+      "loss": 0.1978,
+      "step": 2169
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.415209412780656e-05,
+      "loss": 0.0996,
+      "step": 2170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4149395509499135e-05,
+      "loss": 0.3613,
+      "step": 2171
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4146696891191713e-05,
+      "loss": 0.3094,
+      "step": 2172
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4143998272884285e-05,
+      "loss": 0.2476,
+      "step": 2173
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.414129965457686e-05,
+      "loss": 0.1707,
+      "step": 2174
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4138601036269436e-05,
+      "loss": 0.2101,
+      "step": 2175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4135902417962e-05,
+      "loss": 0.5819,
+      "step": 2176
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4133203799654574e-05,
+      "loss": 0.3297,
+      "step": 2177
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.413050518134715e-05,
+      "loss": 0.079,
+      "step": 2178
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4127806563039725e-05,
+      "loss": 0.0918,
+      "step": 2179
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.41251079447323e-05,
+      "loss": 0.0766,
+      "step": 2180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4122409326424876e-05,
+      "loss": 0.3261,
+      "step": 2181
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.411971070811745e-05,
+      "loss": 0.1223,
+      "step": 2182
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.411701208981002e-05,
+      "loss": 0.0912,
+      "step": 2183
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.411431347150259e-05,
+      "loss": 0.152,
+      "step": 2184
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4111614853195164e-05,
+      "loss": 0.1076,
+      "step": 2185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4108916234887736e-05,
+      "loss": 0.4913,
+      "step": 2186
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4106217616580315e-05,
+      "loss": 0.0688,
+      "step": 2187
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.410351899827289e-05,
+      "loss": 0.0842,
+      "step": 2188
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.410082037996546e-05,
+      "loss": 0.1251,
+      "step": 2189
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.409812176165803e-05,
+      "loss": 0.2241,
+      "step": 2190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.40954231433506e-05,
+      "loss": 0.1241,
+      "step": 2191
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4092724525043175e-05,
+      "loss": 0.1134,
+      "step": 2192
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4090025906735754e-05,
+      "loss": 0.379,
+      "step": 2193
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4087327288428326e-05,
+      "loss": 0.0985,
+      "step": 2194
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.40846286701209e-05,
+      "loss": 0.1453,
+      "step": 2195
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.408193005181348e-05,
+      "loss": 0.5169,
+      "step": 2196
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.407923143350605e-05,
+      "loss": 0.1548,
+      "step": 2197
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4076532815198615e-05,
+      "loss": 0.1378,
+      "step": 2198
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4073834196891194e-05,
+      "loss": 0.3064,
+      "step": 2199
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4071135578583766e-05,
+      "loss": 0.084,
+      "step": 2200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.406843696027634e-05,
+      "loss": 0.0787,
+      "step": 2201
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4065738341968917e-05,
+      "loss": 0.0905,
+      "step": 2202
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.406303972366149e-05,
+      "loss": 0.1926,
+      "step": 2203
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.406034110535406e-05,
+      "loss": 0.1918,
+      "step": 2204
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.405764248704663e-05,
+      "loss": 0.1469,
+      "step": 2205
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4054943868739205e-05,
+      "loss": 0.1977,
+      "step": 2206
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.405224525043178e-05,
+      "loss": 0.1907,
+      "step": 2207
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4049546632124356e-05,
+      "loss": 0.3202,
+      "step": 2208
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.404684801381693e-05,
+      "loss": 0.1087,
+      "step": 2209
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.40441493955095e-05,
+      "loss": 0.1224,
+      "step": 2210
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.404145077720208e-05,
+      "loss": 0.2467,
+      "step": 2211
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4038752158894644e-05,
+      "loss": 0.3926,
+      "step": 2212
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4036053540587216e-05,
+      "loss": 0.1016,
+      "step": 2213
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4033354922279795e-05,
+      "loss": 0.21,
+      "step": 2214
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.403065630397237e-05,
+      "loss": 0.1139,
+      "step": 2215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.402795768566494e-05,
+      "loss": 0.1307,
+      "step": 2216
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.402525906735752e-05,
+      "loss": 0.21,
+      "step": 2217
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.402256044905009e-05,
+      "loss": 0.4184,
+      "step": 2218
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.401986183074266e-05,
+      "loss": 0.4175,
+      "step": 2219
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4017163212435234e-05,
+      "loss": 0.4387,
+      "step": 2220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4014464594127807e-05,
+      "loss": 0.1887,
+      "step": 2221
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.401176597582038e-05,
+      "loss": 0.3286,
+      "step": 2222
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.400906735751296e-05,
+      "loss": 0.1148,
+      "step": 2223
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.400636873920553e-05,
+      "loss": 0.0839,
+      "step": 2224
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.400367012089811e-05,
+      "loss": 0.2106,
+      "step": 2225
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4000971502590674e-05,
+      "loss": 0.1881,
+      "step": 2226
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3998272884283246e-05,
+      "loss": 0.1587,
+      "step": 2227
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3995574265975825e-05,
+      "loss": 0.2125,
+      "step": 2228
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.39928756476684e-05,
+      "loss": 0.3223,
+      "step": 2229
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.399017702936097e-05,
+      "loss": 0.2301,
+      "step": 2230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.398747841105355e-05,
+      "loss": 0.0958,
+      "step": 2231
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.398477979274612e-05,
+      "loss": 0.1973,
+      "step": 2232
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3982081174438685e-05,
+      "loss": 0.3235,
+      "step": 2233
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3979382556131264e-05,
+      "loss": 0.5021,
+      "step": 2234
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3976683937823836e-05,
+      "loss": 0.1058,
+      "step": 2235
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.397398531951641e-05,
+      "loss": 0.1727,
+      "step": 2236
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.397128670120899e-05,
+      "loss": 0.2336,
+      "step": 2237
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.396858808290156e-05,
+      "loss": 0.2358,
+      "step": 2238
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.396588946459413e-05,
+      "loss": 0.2341,
+      "step": 2239
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.39631908462867e-05,
+      "loss": 0.0854,
+      "step": 2240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3960492227979275e-05,
+      "loss": 0.1766,
+      "step": 2241
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.395779360967185e-05,
+      "loss": 0.1719,
+      "step": 2242
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3955094991364426e-05,
+      "loss": 0.0761,
+      "step": 2243
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3952396373057e-05,
+      "loss": 0.1166,
+      "step": 2244
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.394969775474957e-05,
+      "loss": 0.1243,
+      "step": 2245
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.394699913644215e-05,
+      "loss": 0.1088,
+      "step": 2246
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3944300518134715e-05,
+      "loss": 0.196,
+      "step": 2247
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.394160189982729e-05,
+      "loss": 0.4023,
+      "step": 2248
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3938903281519866e-05,
+      "loss": 0.2718,
+      "step": 2249
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.393620466321244e-05,
+      "loss": 0.3492,
+      "step": 2250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.393350604490501e-05,
+      "loss": 0.1936,
+      "step": 2251
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.393080742659759e-05,
+      "loss": 0.2715,
+      "step": 2252
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.392810880829016e-05,
+      "loss": 0.3352,
+      "step": 2253
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.392541018998273e-05,
+      "loss": 0.1982,
+      "step": 2254
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3922711571675305e-05,
+      "loss": 0.072,
+      "step": 2255
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.392001295336788e-05,
+      "loss": 0.099,
+      "step": 2256
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.391731433506045e-05,
+      "loss": 0.3632,
+      "step": 2257
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.391461571675303e-05,
+      "loss": 0.0822,
+      "step": 2258
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.39119170984456e-05,
+      "loss": 0.1232,
+      "step": 2259
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.390921848013817e-05,
+      "loss": 0.2556,
+      "step": 2260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3906519861830744e-05,
+      "loss": 0.2099,
+      "step": 2261
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3903821243523316e-05,
+      "loss": 0.4445,
+      "step": 2262
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.390112262521589e-05,
+      "loss": 0.2101,
+      "step": 2263
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.389842400690847e-05,
+      "loss": 0.1999,
+      "step": 2264
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.389572538860104e-05,
+      "loss": 0.1734,
+      "step": 2265
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.389302677029361e-05,
+      "loss": 0.4066,
+      "step": 2266
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.389032815198619e-05,
+      "loss": 0.2743,
+      "step": 2267
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.388762953367876e-05,
+      "loss": 0.5798,
+      "step": 2268
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.388493091537133e-05,
+      "loss": 0.3847,
+      "step": 2269
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3882232297063906e-05,
+      "loss": 0.0772,
+      "step": 2270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.387953367875648e-05,
+      "loss": 0.4449,
+      "step": 2271
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.387683506044905e-05,
+      "loss": 0.5338,
+      "step": 2272
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.387413644214163e-05,
+      "loss": 0.2583,
+      "step": 2273
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.38714378238342e-05,
+      "loss": 0.067,
+      "step": 2274
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3868739205526774e-05,
+      "loss": 0.186,
+      "step": 2275
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3866040587219346e-05,
+      "loss": 0.1954,
+      "step": 2276
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.386334196891192e-05,
+      "loss": 0.1376,
+      "step": 2277
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.386064335060449e-05,
+      "loss": 0.1866,
+      "step": 2278
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.385794473229707e-05,
+      "loss": 0.4158,
+      "step": 2279
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.385524611398964e-05,
+      "loss": 0.3746,
+      "step": 2280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.385254749568221e-05,
+      "loss": 0.2394,
+      "step": 2281
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.384984887737479e-05,
+      "loss": 0.0733,
+      "step": 2282
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.384715025906736e-05,
+      "loss": 0.0839,
+      "step": 2283
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.384445164075993e-05,
+      "loss": 0.0985,
+      "step": 2284
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.384175302245251e-05,
+      "loss": 0.4293,
+      "step": 2285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.383905440414508e-05,
+      "loss": 0.0743,
+      "step": 2286
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.383635578583765e-05,
+      "loss": 0.444,
+      "step": 2287
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.383365716753023e-05,
+      "loss": 0.2279,
+      "step": 2288
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.38309585492228e-05,
+      "loss": 0.3299,
+      "step": 2289
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3828259930915375e-05,
+      "loss": 0.2073,
+      "step": 2290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.382556131260795e-05,
+      "loss": 0.1966,
+      "step": 2291
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.382286269430052e-05,
+      "loss": 0.1957,
+      "step": 2292
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.382016407599309e-05,
+      "loss": 0.1506,
+      "step": 2293
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.381746545768567e-05,
+      "loss": 0.1964,
+      "step": 2294
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.381476683937824e-05,
+      "loss": 0.5625,
+      "step": 2295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3812068221070815e-05,
+      "loss": 0.3418,
+      "step": 2296
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3809369602763387e-05,
+      "loss": 0.2344,
+      "step": 2297
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.380667098445596e-05,
+      "loss": 0.072,
+      "step": 2298
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.380397236614853e-05,
+      "loss": 0.3818,
+      "step": 2299
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.380127374784111e-05,
+      "loss": 0.0779,
+      "step": 2300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.379857512953368e-05,
+      "loss": 0.0859,
+      "step": 2301
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3795876511226254e-05,
+      "loss": 0.0749,
+      "step": 2302
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.379317789291883e-05,
+      "loss": 0.1544,
+      "step": 2303
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.37904792746114e-05,
+      "loss": 0.118,
+      "step": 2304
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.378778065630397e-05,
+      "loss": 0.107,
+      "step": 2305
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.378508203799655e-05,
+      "loss": 0.0681,
+      "step": 2306
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.378238341968912e-05,
+      "loss": 0.1363,
+      "step": 2307
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.377968480138169e-05,
+      "loss": 0.0642,
+      "step": 2308
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.377698618307427e-05,
+      "loss": 0.0966,
+      "step": 2309
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3774287564766844e-05,
+      "loss": 0.1124,
+      "step": 2310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3771588946459416e-05,
+      "loss": 0.1205,
+      "step": 2311
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.376889032815199e-05,
+      "loss": 0.1403,
+      "step": 2312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.376619170984456e-05,
+      "loss": 0.0671,
+      "step": 2313
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.376349309153713e-05,
+      "loss": 0.1919,
+      "step": 2314
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.376079447322971e-05,
+      "loss": 0.0644,
+      "step": 2315
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.375809585492228e-05,
+      "loss": 0.0962,
+      "step": 2316
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 18528,
+  "num_train_epochs": 8,
+  "save_steps": 32,
+  "total_flos": 1.5966523269316608e+18,
+  "trial_name": null,
+  "trial_params": null
+}