diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,21697 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.992534218166735,
+  "global_step": 3612,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8348623853211012e-07,
+      "loss": 1.3376,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6697247706422023e-07,
+      "loss": 1.3286,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.504587155963304e-07,
+      "loss": 1.3721,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.339449541284405e-07,
+      "loss": 1.3687,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.174311926605506e-07,
+      "loss": 1.3088,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1009174311926608e-06,
+      "loss": 1.3438,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2844036697247707e-06,
+      "loss": 1.2856,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.467889908256881e-06,
+      "loss": 1.2751,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6513761467889911e-06,
+      "loss": 1.1857,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8348623853211011e-06,
+      "loss": 1.1888,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0183486238532113e-06,
+      "loss": 1.1255,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2018348623853215e-06,
+      "loss": 1.1338,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3853211009174317e-06,
+      "loss": 1.1068,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5688073394495415e-06,
+      "loss": 1.1095,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7522935779816517e-06,
+      "loss": 1.0973,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.935779816513762e-06,
+      "loss": 1.1426,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.119266055045872e-06,
+      "loss": 1.1238,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3027522935779823e-06,
+      "loss": 1.0635,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.486238532110092e-06,
+      "loss": 1.0735,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6697247706422022e-06,
+      "loss": 1.1149,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.853211009174313e-06,
+      "loss": 1.0442,
+      "step": 21
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.036697247706423e-06,
+      "loss": 1.0903,
+      "step": 22
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.220183486238532e-06,
+      "loss": 1.0465,
+      "step": 23
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.403669724770643e-06,
+      "loss": 1.1101,
+      "step": 24
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.587155963302753e-06,
+      "loss": 1.0518,
+      "step": 25
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.770642201834863e-06,
+      "loss": 1.1042,
+      "step": 26
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.954128440366973e-06,
+      "loss": 1.0774,
+      "step": 27
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.137614678899083e-06,
+      "loss": 1.0891,
+      "step": 28
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3211009174311936e-06,
+      "loss": 1.0405,
+      "step": 29
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.504587155963303e-06,
+      "loss": 1.051,
+      "step": 30
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.688073394495414e-06,
+      "loss": 1.0638,
+      "step": 31
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.871559633027524e-06,
+      "loss": 1.0717,
+      "step": 32
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.0550458715596335e-06,
+      "loss": 1.0437,
+      "step": 33
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.238532110091744e-06,
+      "loss": 1.0425,
+      "step": 34
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.422018348623854e-06,
+      "loss": 1.0045,
+      "step": 35
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.6055045871559645e-06,
+      "loss": 0.9767,
+      "step": 36
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.788990825688074e-06,
+      "loss": 0.9615,
+      "step": 37
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.972477064220184e-06,
+      "loss": 0.9988,
+      "step": 38
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.155963302752295e-06,
+      "loss": 0.9965,
+      "step": 39
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.3394495412844045e-06,
+      "loss": 1.0248,
+      "step": 40
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.522935779816515e-06,
+      "loss": 0.9773,
+      "step": 41
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.706422018348626e-06,
+      "loss": 1.0422,
+      "step": 42
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.889908256880735e-06,
+      "loss": 0.9956,
+      "step": 43
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.073394495412845e-06,
+      "loss": 1.009,
+      "step": 44
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.256880733944956e-06,
+      "loss": 0.988,
+      "step": 45
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.440366972477065e-06,
+      "loss": 0.9572,
+      "step": 46
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.623853211009175e-06,
+      "loss": 1.0161,
+      "step": 47
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.807339449541286e-06,
+      "loss": 1.0085,
+      "step": 48
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.990825688073395e-06,
+      "loss": 1.0026,
+      "step": 49
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.174311926605506e-06,
+      "loss": 1.0143,
+      "step": 50
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.357798165137616e-06,
+      "loss": 0.9557,
+      "step": 51
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.541284403669727e-06,
+      "loss": 0.9532,
+      "step": 52
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.724770642201836e-06,
+      "loss": 0.9618,
+      "step": 53
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908256880733946e-06,
+      "loss": 0.9818,
+      "step": 54
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0091743119266055e-05,
+      "loss": 0.9536,
+      "step": 55
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0275229357798166e-05,
+      "loss": 0.9714,
+      "step": 56
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0458715596330275e-05,
+      "loss": 0.9562,
+      "step": 57
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0642201834862387e-05,
+      "loss": 0.9396,
+      "step": 58
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0825688073394496e-05,
+      "loss": 0.9663,
+      "step": 59
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1009174311926607e-05,
+      "loss": 0.98,
+      "step": 60
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1192660550458716e-05,
+      "loss": 0.9655,
+      "step": 61
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1376146788990828e-05,
+      "loss": 0.9191,
+      "step": 62
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1559633027522937e-05,
+      "loss": 0.9641,
+      "step": 63
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1743119266055047e-05,
+      "loss": 0.9369,
+      "step": 64
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1926605504587156e-05,
+      "loss": 0.9916,
+      "step": 65
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2110091743119267e-05,
+      "loss": 0.9753,
+      "step": 66
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2293577981651376e-05,
+      "loss": 0.9684,
+      "step": 67
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2477064220183488e-05,
+      "loss": 0.9509,
+      "step": 68
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2660550458715597e-05,
+      "loss": 0.9237,
+      "step": 69
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2844036697247708e-05,
+      "loss": 0.9469,
+      "step": 70
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3027522935779817e-05,
+      "loss": 0.9067,
+      "step": 71
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3211009174311929e-05,
+      "loss": 0.9445,
+      "step": 72
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3394495412844038e-05,
+      "loss": 0.9258,
+      "step": 73
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3577981651376149e-05,
+      "loss": 0.9055,
+      "step": 74
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3761467889908258e-05,
+      "loss": 0.9004,
+      "step": 75
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3944954128440368e-05,
+      "loss": 0.9109,
+      "step": 76
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4128440366972477e-05,
+      "loss": 0.9314,
+      "step": 77
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.431192660550459e-05,
+      "loss": 0.9149,
+      "step": 78
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4495412844036698e-05,
+      "loss": 0.9388,
+      "step": 79
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4678899082568809e-05,
+      "loss": 0.9644,
+      "step": 80
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4862385321100918e-05,
+      "loss": 0.9097,
+      "step": 81
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.504587155963303e-05,
+      "loss": 0.9235,
+      "step": 82
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5229357798165139e-05,
+      "loss": 0.9344,
+      "step": 83
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.541284403669725e-05,
+      "loss": 0.9203,
+      "step": 84
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.559633027522936e-05,
+      "loss": 0.9038,
+      "step": 85
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.577981651376147e-05,
+      "loss": 0.9321,
+      "step": 86
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5963302752293578e-05,
+      "loss": 0.9011,
+      "step": 87
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.614678899082569e-05,
+      "loss": 0.865,
+      "step": 88
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.63302752293578e-05,
+      "loss": 0.9237,
+      "step": 89
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6513761467889912e-05,
+      "loss": 0.9292,
+      "step": 90
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.669724770642202e-05,
+      "loss": 0.9069,
+      "step": 91
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.688073394495413e-05,
+      "loss": 0.8866,
+      "step": 92
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.706422018348624e-05,
+      "loss": 0.9137,
+      "step": 93
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.724770642201835e-05,
+      "loss": 0.9177,
+      "step": 94
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.743119266055046e-05,
+      "loss": 0.9166,
+      "step": 95
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7614678899082572e-05,
+      "loss": 0.9431,
+      "step": 96
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.779816513761468e-05,
+      "loss": 0.9008,
+      "step": 97
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.798165137614679e-05,
+      "loss": 0.9443,
+      "step": 98
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.81651376146789e-05,
+      "loss": 0.9259,
+      "step": 99
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.834862385321101e-05,
+      "loss": 0.9191,
+      "step": 100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.853211009174312e-05,
+      "loss": 0.9078,
+      "step": 101
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8715596330275232e-05,
+      "loss": 0.9045,
+      "step": 102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.889908256880734e-05,
+      "loss": 0.9004,
+      "step": 103
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9082568807339454e-05,
+      "loss": 0.8903,
+      "step": 104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9266055045871563e-05,
+      "loss": 0.8895,
+      "step": 105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.944954128440367e-05,
+      "loss": 0.8967,
+      "step": 106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.963302752293578e-05,
+      "loss": 0.9287,
+      "step": 107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9816513761467893e-05,
+      "loss": 0.9301,
+      "step": 108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.9122,
+      "step": 109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9999995978487288e-05,
+      "loss": 0.9308,
+      "step": 110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9999983913952384e-05,
+      "loss": 0.9058,
+      "step": 111
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999996380640499e-05,
+      "loss": 0.9047,
+      "step": 112
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999993565586128e-05,
+      "loss": 0.9017,
+      "step": 113
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9999899462343893e-05,
+      "loss": 0.8777,
+      "step": 114
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9999855225881947e-05,
+      "loss": 0.8921,
+      "step": 115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9999802946511017e-05,
+      "loss": 0.9119,
+      "step": 116
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9999742624273148e-05,
+      "loss": 0.8914,
+      "step": 117
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999674259216862e-05,
+      "loss": 0.9082,
+      "step": 118
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999597851397145e-05,
+      "loss": 0.9316,
+      "step": 119
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999951340087545e-05,
+      "loss": 0.8778,
+      "step": 120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99994209077197e-05,
+      "loss": 0.9011,
+      "step": 121
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999320372004285e-05,
+      "loss": 0.9178,
+      "step": 122
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999211793810076e-05,
+      "loss": 0.8594,
+      "step": 123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9999095173224393e-05,
+      "loss": 0.8876,
+      "step": 124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999897051034104e-05,
+      "loss": 0.8977,
+      "step": 125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9998837805260284e-05,
+      "loss": 0.9025,
+      "step": 126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9998697058088853e-05,
+      "loss": 0.8706,
+      "step": 127
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9998548268939957e-05,
+      "loss": 0.9246,
+      "step": 128
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9998391437933266e-05,
+      "loss": 0.8789,
+      "step": 129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999822656519492e-05,
+      "loss": 0.9071,
+      "step": 130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9998053650857524e-05,
+      "loss": 0.902,
+      "step": 131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9997872695060153e-05,
+      "loss": 0.9196,
+      "step": 132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999768369794836e-05,
+      "loss": 0.8776,
+      "step": 133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9997486659674143e-05,
+      "loss": 0.89,
+      "step": 134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999728158039598e-05,
+      "loss": 0.8911,
+      "step": 135
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9997068460278833e-05,
+      "loss": 0.8951,
+      "step": 136
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99968472994941e-05,
+      "loss": 0.8712,
+      "step": 137
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9996618098219664e-05,
+      "loss": 0.892,
+      "step": 138
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9996380856639872e-05,
+      "loss": 0.8942,
+      "step": 139
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9996135574945543e-05,
+      "loss": 0.8884,
+      "step": 140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9995882253333955e-05,
+      "loss": 0.8735,
+      "step": 141
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9995620892008852e-05,
+      "loss": 0.8625,
+      "step": 142
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.999535149118045e-05,
+      "loss": 0.8884,
+      "step": 143
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.999507405106543e-05,
+      "loss": 0.8931,
+      "step": 144
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9994788571886936e-05,
+      "loss": 0.8937,
+      "step": 145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.999449505387458e-05,
+      "loss": 0.8627,
+      "step": 146
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9994193497264444e-05,
+      "loss": 0.8665,
+      "step": 147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9993883902299064e-05,
+      "loss": 0.8761,
+      "step": 148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.999356626922745e-05,
+      "loss": 0.8799,
+      "step": 149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9993240598305076e-05,
+      "loss": 0.8931,
+      "step": 150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9992906889793877e-05,
+      "loss": 0.8701,
+      "step": 151
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.999256514396226e-05,
+      "loss": 0.8715,
+      "step": 152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.999221536108509e-05,
+      "loss": 0.8854,
+      "step": 153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9991857541443697e-05,
+      "loss": 0.8865,
+      "step": 154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9991491685325883e-05,
+      "loss": 0.8864,
+      "step": 155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9991117793025895e-05,
+      "loss": 0.896,
+      "step": 156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9990735864844465e-05,
+      "loss": 0.8706,
+      "step": 157
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9990345901088774e-05,
+      "loss": 0.8615,
+      "step": 158
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9989947902072476e-05,
+      "loss": 0.8949,
+      "step": 159
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.998954186811568e-05,
+      "loss": 0.8828,
+      "step": 160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9989127799544955e-05,
+      "loss": 0.8613,
+      "step": 161
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9988705696693345e-05,
+      "loss": 0.8641,
+      "step": 162
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9988275559900347e-05,
+      "loss": 0.8822,
+      "step": 163
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9987837389511916e-05,
+      "loss": 0.8672,
+      "step": 164
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.998739118588048e-05,
+      "loss": 0.9406,
+      "step": 165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9986936949364914e-05,
+      "loss": 0.8341,
+      "step": 166
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.998647468033057e-05,
+      "loss": 0.8793,
+      "step": 167
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.998600437914925e-05,
+      "loss": 0.8926,
+      "step": 168
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9985526046199215e-05,
+      "loss": 0.8667,
+      "step": 169
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.998503968186519e-05,
+      "loss": 0.8566,
+      "step": 170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9984545286538362e-05,
+      "loss": 0.8755,
+      "step": 171
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9984042860616373e-05,
+      "loss": 0.8756,
+      "step": 172
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9983532404503325e-05,
+      "loss": 0.8508,
+      "step": 173
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.998301391860978e-05,
+      "loss": 0.8757,
+      "step": 174
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9982487403352754e-05,
+      "loss": 0.8964,
+      "step": 175
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9981952859155732e-05,
+      "loss": 0.871,
+      "step": 176
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9981410286448642e-05,
+      "loss": 0.8741,
+      "step": 177
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.998085968566788e-05,
+      "loss": 0.8533,
+      "step": 178
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9980301057256295e-05,
+      "loss": 0.8682,
+      "step": 179
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9979734401663194e-05,
+      "loss": 0.8757,
+      "step": 180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9979159719344338e-05,
+      "loss": 0.8702,
+      "step": 181
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9978577010761942e-05,
+      "loss": 0.8392,
+      "step": 182
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.997798627638469e-05,
+      "loss": 0.8596,
+      "step": 183
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9977387516687706e-05,
+      "loss": 0.8645,
+      "step": 184
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.997678073215257e-05,
+      "loss": 0.8217,
+      "step": 185
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9976165923267327e-05,
+      "loss": 0.8616,
+      "step": 186
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9975543090526463e-05,
+      "loss": 0.8579,
+      "step": 187
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9974912234430927e-05,
+      "loss": 0.8599,
+      "step": 188
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.997427335548812e-05,
+      "loss": 0.8656,
+      "step": 189
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.997362645421189e-05,
+      "loss": 0.8895,
+      "step": 190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9972971531122544e-05,
+      "loss": 0.8694,
+      "step": 191
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9972308586746838e-05,
+      "loss": 0.8743,
+      "step": 192
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.997163762161798e-05,
+      "loss": 0.8398,
+      "step": 193
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9970958636275626e-05,
+      "loss": 0.8627,
+      "step": 194
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9970271631265895e-05,
+      "loss": 0.8354,
+      "step": 195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9969576607141335e-05,
+      "loss": 0.8466,
+      "step": 196
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.996887356446096e-05,
+      "loss": 0.8783,
+      "step": 197
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9968162503790232e-05,
+      "loss": 0.88,
+      "step": 198
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9967443425701055e-05,
+      "loss": 0.8599,
+      "step": 199
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9966716330771787e-05,
+      "loss": 0.8871,
+      "step": 200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9965981219587233e-05,
+      "loss": 0.8616,
+      "step": 201
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9965238092738643e-05,
+      "loss": 0.87,
+      "step": 202
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.996448695082372e-05,
+      "loss": 0.8334,
+      "step": 203
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9963727794446605e-05,
+      "loss": 0.8213,
+      "step": 204
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9962960624217887e-05,
+      "loss": 0.8722,
+      "step": 205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.996218544075461e-05,
+      "loss": 0.8075,
+      "step": 206
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9961402244680253e-05,
+      "loss": 0.8644,
+      "step": 207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9960611036624742e-05,
+      "loss": 0.8544,
+      "step": 208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9959811817224447e-05,
+      "loss": 0.8698,
+      "step": 209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9959004587122182e-05,
+      "loss": 0.8464,
+      "step": 210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.995818934696721e-05,
+      "loss": 0.8597,
+      "step": 211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9957366097415223e-05,
+      "loss": 0.83,
+      "step": 212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9956534839128365e-05,
+      "loss": 0.8256,
+      "step": 213
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.995569557277522e-05,
+      "loss": 0.8564,
+      "step": 214
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9954848299030815e-05,
+      "loss": 0.8632,
+      "step": 215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9953993018576608e-05,
+      "loss": 0.8419,
+      "step": 216
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9953129732100507e-05,
+      "loss": 0.9016,
+      "step": 217
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9952258440296854e-05,
+      "loss": 0.8578,
+      "step": 218
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9951379143866435e-05,
+      "loss": 0.8535,
+      "step": 219
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9950491843516466e-05,
+      "loss": 0.8645,
+      "step": 220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.994959653996061e-05,
+      "loss": 0.8606,
+      "step": 221
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.994869323391895e-05,
+      "loss": 0.8237,
+      "step": 222
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9947781926118035e-05,
+      "loss": 0.8702,
+      "step": 223
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.994686261729082e-05,
+      "loss": 0.8293,
+      "step": 224
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.994593530817671e-05,
+      "loss": 0.8223,
+      "step": 225
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9944999999521545e-05,
+      "loss": 0.8813,
+      "step": 226
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.994405669207759e-05,
+      "loss": 0.868,
+      "step": 227
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9943105386603556e-05,
+      "loss": 0.8306,
+      "step": 228
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.994214608386458e-05,
+      "loss": 0.8336,
+      "step": 229
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.994117878463222e-05,
+      "loss": 0.8324,
+      "step": 230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9940203489684496e-05,
+      "loss": 0.8843,
+      "step": 231
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9939220199805828e-05,
+      "loss": 0.8594,
+      "step": 232
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.993822891578708e-05,
+      "loss": 0.847,
+      "step": 233
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9937229638425546e-05,
+      "loss": 0.8406,
+      "step": 234
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9936222368524947e-05,
+      "loss": 0.8356,
+      "step": 235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9935207106895434e-05,
+      "loss": 0.837,
+      "step": 236
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9934183854353582e-05,
+      "loss": 0.817,
+      "step": 237
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9933152611722396e-05,
+      "loss": 0.8549,
+      "step": 238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.993211337983131e-05,
+      "loss": 0.8756,
+      "step": 239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.993106615951618e-05,
+      "loss": 0.8607,
+      "step": 240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9930010951619285e-05,
+      "loss": 0.8523,
+      "step": 241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.992894775698933e-05,
+      "loss": 0.8455,
+      "step": 242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.992787657648145e-05,
+      "loss": 0.8457,
+      "step": 243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9926797410957196e-05,
+      "loss": 0.8541,
+      "step": 244
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9925710261284547e-05,
+      "loss": 0.8347,
+      "step": 245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9924615128337895e-05,
+      "loss": 0.8284,
+      "step": 246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9923512012998062e-05,
+      "loss": 0.8406,
+      "step": 247
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9922400916152283e-05,
+      "loss": 0.8429,
+      "step": 248
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9921281838694216e-05,
+      "loss": 0.8702,
+      "step": 249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9920154781523945e-05,
+      "loss": 0.8407,
+      "step": 250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9919019745547957e-05,
+      "loss": 0.8662,
+      "step": 251
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9917876731679168e-05,
+      "loss": 0.8396,
+      "step": 252
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9916725740836903e-05,
+      "loss": 0.8439,
+      "step": 253
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9915566773946913e-05,
+      "loss": 0.8326,
+      "step": 254
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9914399831941352e-05,
+      "loss": 0.8409,
+      "step": 255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.99132249157588e-05,
+      "loss": 0.8656,
+      "step": 256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9912042026344243e-05,
+      "loss": 0.8163,
+      "step": 257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9910851164649082e-05,
+      "loss": 0.848,
+      "step": 258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9909652331631128e-05,
+      "loss": 0.8237,
+      "step": 259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9908445528254602e-05,
+      "loss": 0.8097,
+      "step": 260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.990723075549015e-05,
+      "loss": 0.8596,
+      "step": 261
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.990600801431481e-05,
+      "loss": 0.829,
+      "step": 262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9904777305712036e-05,
+      "loss": 0.8193,
+      "step": 263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9903538630671687e-05,
+      "loss": 0.8499,
+      "step": 264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9902291990190043e-05,
+      "loss": 0.8418,
+      "step": 265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.990103738526977e-05,
+      "loss": 0.8334,
+      "step": 266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.989977481691995e-05,
+      "loss": 0.8264,
+      "step": 267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.989850428615608e-05,
+      "loss": 0.8604,
+      "step": 268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9897225794000036e-05,
+      "loss": 0.8212,
+      "step": 269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9895939341480127e-05,
+      "loss": 0.8569,
+      "step": 270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.989464492963104e-05,
+      "loss": 0.8695,
+      "step": 271
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.989334255949388e-05,
+      "loss": 0.8263,
+      "step": 272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9892032232116145e-05,
+      "loss": 0.8456,
+      "step": 273
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.989071394855173e-05,
+      "loss": 0.812,
+      "step": 274
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.988938770986094e-05,
+      "loss": 0.8126,
+      "step": 275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.988805351711047e-05,
+      "loss": 0.8433,
+      "step": 276
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9886711371373414e-05,
+      "loss": 0.8282,
+      "step": 277
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.988536127372926e-05,
+      "loss": 0.816,
+      "step": 278
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9884003225263902e-05,
+      "loss": 0.8315,
+      "step": 279
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9882637227069617e-05,
+      "loss": 0.832,
+      "step": 280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9881263280245083e-05,
+      "loss": 0.8033,
+      "step": 281
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.987988138589537e-05,
+      "loss": 0.8422,
+      "step": 282
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9878491545131938e-05,
+      "loss": 0.8141,
+      "step": 283
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9877093759072635e-05,
+      "loss": 0.8486,
+      "step": 284
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.987568802884171e-05,
+      "loss": 0.8033,
+      "step": 285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.987427435556979e-05,
+      "loss": 0.8502,
+      "step": 286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.98728527403939e-05,
+      "loss": 0.8271,
+      "step": 287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.987142318445745e-05,
+      "loss": 0.8528,
+      "step": 288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9869985688910233e-05,
+      "loss": 0.8271,
+      "step": 289
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9868540254908427e-05,
+      "loss": 0.839,
+      "step": 290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9867086883614603e-05,
+      "loss": 0.8337,
+      "step": 291
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.986562557619771e-05,
+      "loss": 0.8629,
+      "step": 292
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.986415633383308e-05,
+      "loss": 0.825,
+      "step": 293
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9862679157702435e-05,
+      "loss": 0.8368,
+      "step": 294
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9861194048993865e-05,
+      "loss": 0.8281,
+      "step": 295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9859701008901844e-05,
+      "loss": 0.8551,
+      "step": 296
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9858200038627232e-05,
+      "loss": 0.8447,
+      "step": 297
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9856691139377267e-05,
+      "loss": 0.8492,
+      "step": 298
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9855174312365552e-05,
+      "loss": 0.8417,
+      "step": 299
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9853649558812084e-05,
+      "loss": 0.8468,
+      "step": 300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9852116879943216e-05,
+      "loss": 0.8625,
+      "step": 301
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9850576276991697e-05,
+      "loss": 0.8129,
+      "step": 302
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9849027751196627e-05,
+      "loss": 0.8114,
+      "step": 303
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9847471303803495e-05,
+      "loss": 0.8098,
+      "step": 304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9845906936064156e-05,
+      "loss": 0.8265,
+      "step": 305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.984433464923683e-05,
+      "loss": 0.8232,
+      "step": 306
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.984275444458612e-05,
+      "loss": 0.8529,
+      "step": 307
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9841166323382982e-05,
+      "loss": 0.8463,
+      "step": 308
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9839570286904747e-05,
+      "loss": 0.8186,
+      "step": 309
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.983796633643511e-05,
+      "loss": 0.8507,
+      "step": 310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9836354473264134e-05,
+      "loss": 0.8217,
+      "step": 311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.983473469868825e-05,
+      "loss": 0.8234,
+      "step": 312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9833107014010234e-05,
+      "loss": 0.8184,
+      "step": 313
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.983147142053925e-05,
+      "loss": 0.8154,
+      "step": 314
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9829827919590804e-05,
+      "loss": 0.8379,
+      "step": 315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9828176512486767e-05,
+      "loss": 0.799,
+      "step": 316
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982651720055537e-05,
+      "loss": 0.8457,
+      "step": 317
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9824849985131208e-05,
+      "loss": 0.8368,
+      "step": 318
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9823174867555216e-05,
+      "loss": 0.8132,
+      "step": 319
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9821491849174706e-05,
+      "loss": 0.7991,
+      "step": 320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9819800931343323e-05,
+      "loss": 0.8461,
+      "step": 321
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9818102115421088e-05,
+      "loss": 0.813,
+      "step": 322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9816395402774356e-05,
+      "loss": 0.8236,
+      "step": 323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9814680794775843e-05,
+      "loss": 0.849,
+      "step": 324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9812958292804612e-05,
+      "loss": 0.8066,
+      "step": 325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9811227898246072e-05,
+      "loss": 0.8342,
+      "step": 326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9809489612491988e-05,
+      "loss": 0.823,
+      "step": 327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.980774343694047e-05,
+      "loss": 0.8231,
+      "step": 328
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9805989372995966e-05,
+      "loss": 0.835,
+      "step": 329
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9804227422069274e-05,
+      "loss": 0.83,
+      "step": 330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9802457585577537e-05,
+      "loss": 0.8354,
+      "step": 331
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9800679864944242e-05,
+      "loss": 0.8494,
+      "step": 332
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9798894261599212e-05,
+      "loss": 0.8446,
+      "step": 333
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.979710077697861e-05,
+      "loss": 0.8175,
+      "step": 334
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9795299412524948e-05,
+      "loss": 0.838,
+      "step": 335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9793490169687058e-05,
+      "loss": 0.8344,
+      "step": 336
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9791673049920122e-05,
+      "loss": 0.8356,
+      "step": 337
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.978984805468566e-05,
+      "loss": 0.8181,
+      "step": 338
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9788015185451513e-05,
+      "loss": 0.8312,
+      "step": 339
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9786174443691863e-05,
+      "loss": 0.8176,
+      "step": 340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9784325830887227e-05,
+      "loss": 0.8518,
+      "step": 341
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9782469348524443e-05,
+      "loss": 0.8336,
+      "step": 342
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9780604998096695e-05,
+      "loss": 0.8143,
+      "step": 343
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9778732781103476e-05,
+      "loss": 0.8186,
+      "step": 344
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.977685269905062e-05,
+      "loss": 0.8248,
+      "step": 345
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.977496475345027e-05,
+      "loss": 0.8293,
+      "step": 346
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.977306894582092e-05,
+      "loss": 0.8345,
+      "step": 347
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.977116527768737e-05,
+      "loss": 0.8097,
+      "step": 348
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9769253750580742e-05,
+      "loss": 0.7913,
+      "step": 349
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.976733436603848e-05,
+      "loss": 0.8475,
+      "step": 350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9765407125604352e-05,
+      "loss": 0.8153,
+      "step": 351
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9763472030828447e-05,
+      "loss": 0.8447,
+      "step": 352
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.976152908326716e-05,
+      "loss": 0.7827,
+      "step": 353
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.975957828448321e-05,
+      "loss": 0.8279,
+      "step": 354
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.975761963604563e-05,
+      "loss": 0.8175,
+      "step": 355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9755653139529768e-05,
+      "loss": 0.8416,
+      "step": 356
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.975367879651728e-05,
+      "loss": 0.8254,
+      "step": 357
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.975169660859614e-05,
+      "loss": 0.8119,
+      "step": 358
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9749706577360616e-05,
+      "loss": 0.8394,
+      "step": 359
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9747708704411307e-05,
+      "loss": 0.8186,
+      "step": 360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9745702991355098e-05,
+      "loss": 0.7856,
+      "step": 361
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9743689439805197e-05,
+      "loss": 0.8206,
+      "step": 362
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9741668051381103e-05,
+      "loss": 0.8311,
+      "step": 363
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9739638827708626e-05,
+      "loss": 0.8036,
+      "step": 364
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9737601770419873e-05,
+      "loss": 0.8177,
+      "step": 365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9735556881153258e-05,
+      "loss": 0.809,
+      "step": 366
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.973350416155349e-05,
+      "loss": 0.8367,
+      "step": 367
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9731443613271576e-05,
+      "loss": 0.8033,
+      "step": 368
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.972937523796482e-05,
+      "loss": 0.8318,
+      "step": 369
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9727299037296823e-05,
+      "loss": 0.8127,
+      "step": 370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9725215012937476e-05,
+      "loss": 0.8275,
+      "step": 371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9723123166562965e-05,
+      "loss": 0.8354,
+      "step": 372
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.972102349985577e-05,
+      "loss": 0.8158,
+      "step": 373
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.971891601450466e-05,
+      "loss": 0.8234,
+      "step": 374
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9716800712204683e-05,
+      "loss": 0.8125,
+      "step": 375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9714677594657187e-05,
+      "loss": 0.8019,
+      "step": 376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9712546663569803e-05,
+      "loss": 0.8325,
+      "step": 377
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.971040792065644e-05,
+      "loss": 0.8191,
+      "step": 378
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.97082613676373e-05,
+      "loss": 0.8163,
+      "step": 379
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9706107006238852e-05,
+      "loss": 0.8379,
+      "step": 380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9703944838193864e-05,
+      "loss": 0.8142,
+      "step": 381
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9701774865241364e-05,
+      "loss": 0.8325,
+      "step": 382
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9699597089126676e-05,
+      "loss": 0.8126,
+      "step": 383
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9697411511601384e-05,
+      "loss": 0.8014,
+      "step": 384
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9695218134423356e-05,
+      "loss": 0.8307,
+      "step": 385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9693016959356736e-05,
+      "loss": 0.8336,
+      "step": 386
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9690807988171923e-05,
+      "loss": 0.8328,
+      "step": 387
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9688591222645607e-05,
+      "loss": 0.8333,
+      "step": 388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9686366664560738e-05,
+      "loss": 0.8262,
+      "step": 389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.968413431570653e-05,
+      "loss": 0.8176,
+      "step": 390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.968189417787847e-05,
+      "loss": 0.8209,
+      "step": 391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9679646252878305e-05,
+      "loss": 0.8127,
+      "step": 392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9677390542514044e-05,
+      "loss": 0.8195,
+      "step": 393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9675127048599963e-05,
+      "loss": 0.851,
+      "step": 394
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.96728557729566e-05,
+      "loss": 0.7993,
+      "step": 395
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9670576717410742e-05,
+      "loss": 0.8187,
+      "step": 396
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.966828988379544e-05,
+      "loss": 0.8046,
+      "step": 397
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9665995273950005e-05,
+      "loss": 0.8262,
+      "step": 398
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9663692889719993e-05,
+      "loss": 0.7826,
+      "step": 399
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.966138273295722e-05,
+      "loss": 0.8406,
+      "step": 400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.965906480551974e-05,
+      "loss": 0.8184,
+      "step": 401
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9656739109271886e-05,
+      "loss": 0.7975,
+      "step": 402
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9654405646084213e-05,
+      "loss": 0.7872,
+      "step": 403
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.965206441783353e-05,
+      "loss": 0.8068,
+      "step": 404
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.964971542640289e-05,
+      "loss": 0.8051,
+      "step": 405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9647358673681603e-05,
+      "loss": 0.7943,
+      "step": 406
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.96449941615652e-05,
+      "loss": 0.8273,
+      "step": 407
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.964262189195547e-05,
+      "loss": 0.8029,
+      "step": 408
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9640241866760434e-05,
+      "loss": 0.8149,
+      "step": 409
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9637854087894354e-05,
+      "loss": 0.8162,
+      "step": 410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.963545855727773e-05,
+      "loss": 0.8076,
+      "step": 411
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.963305527683728e-05,
+      "loss": 0.8427,
+      "step": 412
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.963064424850598e-05,
+      "loss": 0.8458,
+      "step": 413
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9628225474223028e-05,
+      "loss": 0.8844,
+      "step": 414
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9625798955933844e-05,
+      "loss": 0.8278,
+      "step": 415
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.962336469559008e-05,
+      "loss": 0.7998,
+      "step": 416
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9620922695149627e-05,
+      "loss": 0.8055,
+      "step": 417
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9618472956576586e-05,
+      "loss": 0.8398,
+      "step": 418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9616015481841293e-05,
+      "loss": 0.8071,
+      "step": 419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9613550272920295e-05,
+      "loss": 0.7762,
+      "step": 420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9611077331796368e-05,
+      "loss": 0.8,
+      "step": 421
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9608596660458507e-05,
+      "loss": 0.8121,
+      "step": 422
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.960610826090192e-05,
+      "loss": 0.7905,
+      "step": 423
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9603612135128033e-05,
+      "loss": 0.798,
+      "step": 424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9601108285144487e-05,
+      "loss": 0.8186,
+      "step": 425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9598596712965136e-05,
+      "loss": 0.824,
+      "step": 426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9596077420610043e-05,
+      "loss": 0.8074,
+      "step": 427
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.959355041010548e-05,
+      "loss": 0.8193,
+      "step": 428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.959101568348393e-05,
+      "loss": 0.8165,
+      "step": 429
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9588473242784078e-05,
+      "loss": 0.8044,
+      "step": 430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.958592309005082e-05,
+      "loss": 0.7931,
+      "step": 431
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.958336522733524e-05,
+      "loss": 0.8171,
+      "step": 432
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.958079965669465e-05,
+      "loss": 0.8052,
+      "step": 433
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.957822638019253e-05,
+      "loss": 0.8257,
+      "step": 434
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.957564539989858e-05,
+      "loss": 0.8124,
+      "step": 435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9573056717888687e-05,
+      "loss": 0.8167,
+      "step": 436
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9570460336244933e-05,
+      "loss": 0.813,
+      "step": 437
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9567856257055596e-05,
+      "loss": 0.7754,
+      "step": 438
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9565244482415147e-05,
+      "loss": 0.8114,
+      "step": 439
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.956262501442424e-05,
+      "loss": 0.7906,
+      "step": 440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9559997855189716e-05,
+      "loss": 0.8328,
+      "step": 441
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.955736300682461e-05,
+      "loss": 0.7981,
+      "step": 442
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.955472047144814e-05,
+      "loss": 0.8224,
+      "step": 443
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9552070251185696e-05,
+      "loss": 0.8221,
+      "step": 444
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9549412348168864e-05,
+      "loss": 0.816,
+      "step": 445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9546746764535397e-05,
+      "loss": 0.8304,
+      "step": 446
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.954407350242924e-05,
+      "loss": 0.8147,
+      "step": 447
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.954139256400049e-05,
+      "loss": 0.8319,
+      "step": 448
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9538703951405443e-05,
+      "loss": 0.8376,
+      "step": 449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9536007666806555e-05,
+      "loss": 0.7837,
+      "step": 450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9533303712372452e-05,
+      "loss": 0.817,
+      "step": 451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9530592090277934e-05,
+      "loss": 0.8414,
+      "step": 452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9527872802703966e-05,
+      "loss": 0.8027,
+      "step": 453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9525145851837678e-05,
+      "loss": 0.8201,
+      "step": 454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.952241123987236e-05,
+      "loss": 0.8402,
+      "step": 455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9519668969007472e-05,
+      "loss": 0.8208,
+      "step": 456
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9516919041448627e-05,
+      "loss": 0.8199,
+      "step": 457
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9514161459407594e-05,
+      "loss": 0.7954,
+      "step": 458
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.951139622510231e-05,
+      "loss": 0.7856,
+      "step": 459
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.950862334075686e-05,
+      "loss": 0.813,
+      "step": 460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9505842808601477e-05,
+      "loss": 0.7996,
+      "step": 461
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.950305463087256e-05,
+      "loss": 0.7739,
+      "step": 462
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9500258809812634e-05,
+      "loss": 0.7937,
+      "step": 463
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9497455347670394e-05,
+      "loss": 0.8003,
+      "step": 464
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.949464424670067e-05,
+      "loss": 0.8206,
+      "step": 465
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9491825509164434e-05,
+      "loss": 0.8037,
+      "step": 466
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9488999137328804e-05,
+      "loss": 0.8282,
+      "step": 467
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9486165133467047e-05,
+      "loss": 0.7878,
+      "step": 468
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.948332349985855e-05,
+      "loss": 0.7972,
+      "step": 469
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.948047423878885e-05,
+      "loss": 0.8278,
+      "step": 470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.947761735254961e-05,
+      "loss": 0.8239,
+      "step": 471
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.947475284343864e-05,
+      "loss": 0.8325,
+      "step": 472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9471880713759866e-05,
+      "loss": 0.8115,
+      "step": 473
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.946900096582335e-05,
+      "loss": 0.7933,
+      "step": 474
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.946611360194528e-05,
+      "loss": 0.7784,
+      "step": 475
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9463218624447972e-05,
+      "loss": 0.8115,
+      "step": 476
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9460316035659865e-05,
+      "loss": 0.8029,
+      "step": 477
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9457405837915513e-05,
+      "loss": 0.84,
+      "step": 478
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9454488033555602e-05,
+      "loss": 0.8192,
+      "step": 479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.945156262492692e-05,
+      "loss": 0.7915,
+      "step": 480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9448629614382394e-05,
+      "loss": 0.8037,
+      "step": 481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9445689004281043e-05,
+      "loss": 0.7662,
+      "step": 482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9442740796988003e-05,
+      "loss": 0.7878,
+      "step": 483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9439784994874535e-05,
+      "loss": 0.7979,
+      "step": 484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9436821600317994e-05,
+      "loss": 0.7966,
+      "step": 485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9433850615701843e-05,
+      "loss": 0.8035,
+      "step": 486
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9430872043415657e-05,
+      "loss": 0.8141,
+      "step": 487
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9427885885855103e-05,
+      "loss": 0.7902,
+      "step": 488
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9424892145421963e-05,
+      "loss": 0.7915,
+      "step": 489
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9421890824524108e-05,
+      "loss": 0.8041,
+      "step": 490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.94188819255755e-05,
+      "loss": 0.8444,
+      "step": 491
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9415865450996217e-05,
+      "loss": 0.7748,
+      "step": 492
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9412841403212406e-05,
+      "loss": 0.7988,
+      "step": 493
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9409809784656323e-05,
+      "loss": 0.7979,
+      "step": 494
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9406770597766306e-05,
+      "loss": 0.821,
+      "step": 495
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9403723844986775e-05,
+      "loss": 0.799,
+      "step": 496
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9400669528768246e-05,
+      "loss": 0.7872,
+      "step": 497
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9397607651567316e-05,
+      "loss": 0.7981,
+      "step": 498
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9394538215846654e-05,
+      "loss": 0.819,
+      "step": 499
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9391461224075016e-05,
+      "loss": 0.8009,
+      "step": 500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.938837667872724e-05,
+      "loss": 0.8027,
+      "step": 501
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.938528458228423e-05,
+      "loss": 0.8109,
+      "step": 502
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9382184937232966e-05,
+      "loss": 0.8203,
+      "step": 503
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.93790777460665e-05,
+      "loss": 0.8016,
+      "step": 504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9375963011283957e-05,
+      "loss": 0.8262,
+      "step": 505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9372840735390523e-05,
+      "loss": 0.8065,
+      "step": 506
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9369710920897455e-05,
+      "loss": 0.7965,
+      "step": 507
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9366573570322067e-05,
+      "loss": 0.7822,
+      "step": 508
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.936342868618774e-05,
+      "loss": 0.7764,
+      "step": 509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9360276271023915e-05,
+      "loss": 0.7904,
+      "step": 510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9357116327366086e-05,
+      "loss": 0.8181,
+      "step": 511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.93539488577558e-05,
+      "loss": 0.8011,
+      "step": 512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.935077386474067e-05,
+      "loss": 0.7875,
+      "step": 513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9347591350874338e-05,
+      "loss": 0.812,
+      "step": 514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.934440131871652e-05,
+      "loss": 0.7944,
+      "step": 515
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.934120377083296e-05,
+      "loss": 0.7838,
+      "step": 516
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9337998709795454e-05,
+      "loss": 0.776,
+      "step": 517
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9334786138181846e-05,
+      "loss": 0.8029,
+      "step": 518
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9331566058576008e-05,
+      "loss": 0.8182,
+      "step": 519
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9328338473567864e-05,
+      "loss": 0.7682,
+      "step": 520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9325103385753367e-05,
+      "loss": 0.8055,
+      "step": 521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9321860797734508e-05,
+      "loss": 0.7965,
+      "step": 522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9318610712119304e-05,
+      "loss": 0.8035,
+      "step": 523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.931535313152181e-05,
+      "loss": 0.7839,
+      "step": 524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9312088058562108e-05,
+      "loss": 0.7865,
+      "step": 525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9308815495866306e-05,
+      "loss": 0.8185,
+      "step": 526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.930553544606653e-05,
+      "loss": 0.8051,
+      "step": 527
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.930224791180093e-05,
+      "loss": 0.7798,
+      "step": 528
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9298952895713684e-05,
+      "loss": 0.7841,
+      "step": 529
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9295650400454984e-05,
+      "loss": 0.7997,
+      "step": 530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9292340428681028e-05,
+      "loss": 0.8193,
+      "step": 531
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9289022983054036e-05,
+      "loss": 0.8354,
+      "step": 532
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9285698066242244e-05,
+      "loss": 0.8054,
+      "step": 533
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9282365680919887e-05,
+      "loss": 0.8103,
+      "step": 534
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.927902582976721e-05,
+      "loss": 0.8257,
+      "step": 535
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.927567851547046e-05,
+      "loss": 0.814,
+      "step": 536
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9272323740721904e-05,
+      "loss": 0.8307,
+      "step": 537
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.926896150821978e-05,
+      "loss": 0.8359,
+      "step": 538
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.926559182066835e-05,
+      "loss": 0.8125,
+      "step": 539
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.926221468077786e-05,
+      "loss": 0.8046,
+      "step": 540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9258830091264552e-05,
+      "loss": 0.7954,
+      "step": 541
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9255438054850658e-05,
+      "loss": 0.8053,
+      "step": 542
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9252038574264403e-05,
+      "loss": 0.819,
+      "step": 543
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9248631652240003e-05,
+      "loss": 0.7905,
+      "step": 544
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9245217291517644e-05,
+      "loss": 0.7697,
+      "step": 545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9241795494843508e-05,
+      "loss": 0.8086,
+      "step": 546
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.923836626496976e-05,
+      "loss": 0.7898,
+      "step": 547
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9234929604654534e-05,
+      "loss": 0.7642,
+      "step": 548
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.923148551666195e-05,
+      "loss": 0.8129,
+      "step": 549
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.922803400376209e-05,
+      "loss": 0.7949,
+      "step": 550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9224575068731015e-05,
+      "loss": 0.785,
+      "step": 551
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.922110871435076e-05,
+      "loss": 0.7906,
+      "step": 552
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.921763494340932e-05,
+      "loss": 0.7992,
+      "step": 553
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9214153758700656e-05,
+      "loss": 0.7775,
+      "step": 554
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9210665163024694e-05,
+      "loss": 0.8335,
+      "step": 555
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9207169159187327e-05,
+      "loss": 0.816,
+      "step": 556
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9203665750000388e-05,
+      "loss": 0.7771,
+      "step": 557
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.920015493828169e-05,
+      "loss": 0.7957,
+      "step": 558
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.919663672685498e-05,
+      "loss": 0.7812,
+      "step": 559
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.919311111854997e-05,
+      "loss": 0.811,
+      "step": 560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.918957811620231e-05,
+      "loss": 0.7773,
+      "step": 561
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9186037722653603e-05,
+      "loss": 0.7839,
+      "step": 562
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.91824899407514e-05,
+      "loss": 0.809,
+      "step": 563
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9178934773349192e-05,
+      "loss": 0.7961,
+      "step": 564
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9175372223306406e-05,
+      "loss": 0.8187,
+      "step": 565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.917180229348841e-05,
+      "loss": 0.774,
+      "step": 566
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9168224986766512e-05,
+      "loss": 0.7563,
+      "step": 567
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9164640306017942e-05,
+      "loss": 0.7811,
+      "step": 568
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9161048254125872e-05,
+      "loss": 0.7996,
+      "step": 569
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.91574488339794e-05,
+      "loss": 0.7959,
+      "step": 570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9153842048473542e-05,
+      "loss": 0.777,
+      "step": 571
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9150227900509253e-05,
+      "loss": 0.7739,
+      "step": 572
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9146606392993398e-05,
+      "loss": 0.8038,
+      "step": 573
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9142977528838763e-05,
+      "loss": 0.7676,
+      "step": 574
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9139341310964054e-05,
+      "loss": 0.7678,
+      "step": 575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.913569774229389e-05,
+      "loss": 0.782,
+      "step": 576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9132046825758805e-05,
+      "loss": 0.7916,
+      "step": 577
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.912838856429524e-05,
+      "loss": 0.7991,
+      "step": 578
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.912472296084554e-05,
+      "loss": 0.7793,
+      "step": 579
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.912105001835796e-05,
+      "loss": 0.7775,
+      "step": 580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9117369739786663e-05,
+      "loss": 0.8228,
+      "step": 581
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9113682128091697e-05,
+      "loss": 0.7806,
+      "step": 582
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9109987186239024e-05,
+      "loss": 0.8127,
+      "step": 583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9106284917200496e-05,
+      "loss": 0.8026,
+      "step": 584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9102575323953853e-05,
+      "loss": 0.8073,
+      "step": 585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9098858409482733e-05,
+      "loss": 0.7799,
+      "step": 586
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.909513417677666e-05,
+      "loss": 0.7891,
+      "step": 587
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.909140262883104e-05,
+      "loss": 0.7606,
+      "step": 588
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9087663768647172e-05,
+      "loss": 0.8049,
+      "step": 589
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9083917599232227e-05,
+      "loss": 0.8163,
+      "step": 590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.908016412359926e-05,
+      "loss": 0.8127,
+      "step": 591
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.90764033447672e-05,
+      "loss": 0.8124,
+      "step": 592
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.907263526576085e-05,
+      "loss": 0.7815,
+      "step": 593
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9068859889610892e-05,
+      "loss": 0.7815,
+      "step": 594
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.906507721935386e-05,
+      "loss": 0.7748,
+      "step": 595
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9061287258032176e-05,
+      "loss": 0.7839,
+      "step": 596
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.905749000869411e-05,
+      "loss": 0.792,
+      "step": 597
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9053685474393798e-05,
+      "loss": 0.7734,
+      "step": 598
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9049873658191243e-05,
+      "loss": 0.7612,
+      "step": 599
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.904605456315229e-05,
+      "loss": 0.8115,
+      "step": 600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.904222819234865e-05,
+      "loss": 0.7941,
+      "step": 601
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9038394548857886e-05,
+      "loss": 0.7959,
+      "step": 602
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9034553635763405e-05,
+      "loss": 0.8046,
+      "step": 603
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.903070545615446e-05,
+      "loss": 0.7625,
+      "step": 604
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.902685001312616e-05,
+      "loss": 0.7522,
+      "step": 605
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9022987309779436e-05,
+      "loss": 0.7512,
+      "step": 606
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.901911734922108e-05,
+      "loss": 0.7406,
+      "step": 607
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.901524013456371e-05,
+      "loss": 0.7704,
+      "step": 608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9011355668925773e-05,
+      "loss": 0.7104,
+      "step": 609
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.900746395543156e-05,
+      "loss": 0.7557,
+      "step": 610
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9003564997211184e-05,
+      "loss": 0.7413,
+      "step": 611
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.899965879740059e-05,
+      "loss": 0.7209,
+      "step": 612
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.899574535914154e-05,
+      "loss": 0.7374,
+      "step": 613
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.899182468558163e-05,
+      "loss": 0.7554,
+      "step": 614
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8987896779874257e-05,
+      "loss": 0.7296,
+      "step": 615
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8983961645178653e-05,
+      "loss": 0.7507,
+      "step": 616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8980019284659852e-05,
+      "loss": 0.7623,
+      "step": 617
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8976069701488707e-05,
+      "loss": 0.7637,
+      "step": 618
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8972112898841884e-05,
+      "loss": 0.7698,
+      "step": 619
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8968148879901837e-05,
+      "loss": 0.7303,
+      "step": 620
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8964177647856847e-05,
+      "loss": 0.7245,
+      "step": 621
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.896019920590098e-05,
+      "loss": 0.7402,
+      "step": 622
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.895621355723411e-05,
+      "loss": 0.7382,
+      "step": 623
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8952220705061903e-05,
+      "loss": 0.7581,
+      "step": 624
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.894822065259582e-05,
+      "loss": 0.7169,
+      "step": 625
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8944213403053116e-05,
+      "loss": 0.7462,
+      "step": 626
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.894019895965683e-05,
+      "loss": 0.7294,
+      "step": 627
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8936177325635786e-05,
+      "loss": 0.7719,
+      "step": 628
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8932148504224598e-05,
+      "loss": 0.7759,
+      "step": 629
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.892811249866366e-05,
+      "loss": 0.74,
+      "step": 630
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8924069312199135e-05,
+      "loss": 0.7346,
+      "step": 631
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8920018948082973e-05,
+      "loss": 0.7566,
+      "step": 632
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8915961409572888e-05,
+      "loss": 0.7322,
+      "step": 633
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8911896699932374e-05,
+      "loss": 0.7091,
+      "step": 634
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8907824822430683e-05,
+      "loss": 0.7484,
+      "step": 635
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8903745780342838e-05,
+      "loss": 0.7452,
+      "step": 636
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8899659576949623e-05,
+      "loss": 0.7435,
+      "step": 637
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8895566215537582e-05,
+      "loss": 0.7422,
+      "step": 638
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8891465699399016e-05,
+      "loss": 0.7343,
+      "step": 639
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.888735803183198e-05,
+      "loss": 0.7255,
+      "step": 640
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.888324321614028e-05,
+      "loss": 0.7358,
+      "step": 641
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8879121255633474e-05,
+      "loss": 0.7476,
+      "step": 642
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8874992153626867e-05,
+      "loss": 0.7432,
+      "step": 643
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8870855913441505e-05,
+      "loss": 0.7661,
+      "step": 644
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8866712538404173e-05,
+      "loss": 0.7255,
+      "step": 645
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8862562031847406e-05,
+      "loss": 0.7587,
+      "step": 646
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8858404397109462e-05,
+      "loss": 0.7395,
+      "step": 647
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8854239637534337e-05,
+      "loss": 0.7537,
+      "step": 648
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8850067756471756e-05,
+      "loss": 0.7554,
+      "step": 649
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.884588875727718e-05,
+      "loss": 0.7291,
+      "step": 650
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.884170264331178e-05,
+      "loss": 0.7711,
+      "step": 651
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8837509417942467e-05,
+      "loss": 0.7256,
+      "step": 652
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8833309084541853e-05,
+      "loss": 0.7528,
+      "step": 653
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8829101646488287e-05,
+      "loss": 0.7435,
+      "step": 654
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8824887107165817e-05,
+      "loss": 0.7155,
+      "step": 655
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8820665469964205e-05,
+      "loss": 0.7448,
+      "step": 656
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.881643673827893e-05,
+      "loss": 0.745,
+      "step": 657
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.881220091551117e-05,
+      "loss": 0.7374,
+      "step": 658
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8807958005067807e-05,
+      "loss": 0.7563,
+      "step": 659
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.880370801036142e-05,
+      "loss": 0.7449,
+      "step": 660
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8799450934810302e-05,
+      "loss": 0.7224,
+      "step": 661
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.879518678183842e-05,
+      "loss": 0.7268,
+      "step": 662
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.879091555487545e-05,
+      "loss": 0.7239,
+      "step": 663
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8786637257356743e-05,
+      "loss": 0.7354,
+      "step": 664
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.878235189272335e-05,
+      "loss": 0.7494,
+      "step": 665
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8778059464422e-05,
+      "loss": 0.7306,
+      "step": 666
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8773759975905098e-05,
+      "loss": 0.7433,
+      "step": 667
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8769453430630743e-05,
+      "loss": 0.7456,
+      "step": 668
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8765139832062698e-05,
+      "loss": 0.759,
+      "step": 669
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8760819183670394e-05,
+      "loss": 0.7501,
+      "step": 670
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8756491488928953e-05,
+      "loss": 0.7432,
+      "step": 671
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8752156751319135e-05,
+      "loss": 0.7282,
+      "step": 672
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.874781497432739e-05,
+      "loss": 0.7324,
+      "step": 673
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8743466161445823e-05,
+      "loss": 0.7335,
+      "step": 674
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.873911031617219e-05,
+      "loss": 0.7263,
+      "step": 675
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8734747442009904e-05,
+      "loss": 0.7304,
+      "step": 676
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8730377542468044e-05,
+      "loss": 0.7241,
+      "step": 677
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8726000621061328e-05,
+      "loss": 0.7465,
+      "step": 678
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8721616681310125e-05,
+      "loss": 0.7249,
+      "step": 679
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8717225726740448e-05,
+      "loss": 0.7292,
+      "step": 680
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.871282776088396e-05,
+      "loss": 0.7384,
+      "step": 681
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8708422787277943e-05,
+      "loss": 0.7373,
+      "step": 682
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.870401080946534e-05,
+      "loss": 0.7736,
+      "step": 683
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.869959183099471e-05,
+      "loss": 0.7338,
+      "step": 684
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.869516585542025e-05,
+      "loss": 0.7645,
+      "step": 685
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8690732886301786e-05,
+      "loss": 0.7559,
+      "step": 686
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8686292927204764e-05,
+      "loss": 0.7394,
+      "step": 687
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.868184598170025e-05,
+      "loss": 0.7104,
+      "step": 688
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8677392053364942e-05,
+      "loss": 0.7195,
+      "step": 689
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.867293114578114e-05,
+      "loss": 0.7444,
+      "step": 690
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.866846326253677e-05,
+      "loss": 0.7325,
+      "step": 691
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.866398840722535e-05,
+      "loss": 0.743,
+      "step": 692
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8659506583446028e-05,
+      "loss": 0.7489,
+      "step": 693
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8655017794803544e-05,
+      "loss": 0.7551,
+      "step": 694
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.865052204490824e-05,
+      "loss": 0.7401,
+      "step": 695
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8646019337376058e-05,
+      "loss": 0.7245,
+      "step": 696
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.864150967582854e-05,
+      "loss": 0.7233,
+      "step": 697
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8636993063892822e-05,
+      "loss": 0.7527,
+      "step": 698
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8632469505201617e-05,
+      "loss": 0.7256,
+      "step": 699
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.862793900339324e-05,
+      "loss": 0.7395,
+      "step": 700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8623401562111584e-05,
+      "loss": 0.7361,
+      "step": 701
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.861885718500613e-05,
+      "loss": 0.7218,
+      "step": 702
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.861430587573192e-05,
+      "loss": 0.7084,
+      "step": 703
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8609747637949594e-05,
+      "loss": 0.729,
+      "step": 704
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8605182475325352e-05,
+      "loss": 0.7499,
+      "step": 705
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8600610391530968e-05,
+      "loss": 0.7458,
+      "step": 706
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8596031390243777e-05,
+      "loss": 0.7465,
+      "step": 707
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.859144547514668e-05,
+      "loss": 0.7296,
+      "step": 708
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.858685264992814e-05,
+      "loss": 0.7241,
+      "step": 709
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8582252918282183e-05,
+      "loss": 0.7272,
+      "step": 710
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8577646283908384e-05,
+      "loss": 0.7362,
+      "step": 711
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8573032750511867e-05,
+      "loss": 0.7465,
+      "step": 712
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.856841232180331e-05,
+      "loss": 0.7357,
+      "step": 713
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.856378500149894e-05,
+      "loss": 0.7335,
+      "step": 714
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.855915079332051e-05,
+      "loss": 0.7121,
+      "step": 715
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.855450970099534e-05,
+      "loss": 0.7365,
+      "step": 716
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8549861728256265e-05,
+      "loss": 0.7443,
+      "step": 717
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8545206878841667e-05,
+      "loss": 0.7363,
+      "step": 718
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8540545156495442e-05,
+      "loss": 0.7437,
+      "step": 719
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8535876564967036e-05,
+      "loss": 0.7196,
+      "step": 720
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8531201108011404e-05,
+      "loss": 0.7225,
+      "step": 721
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.852651878938903e-05,
+      "loss": 0.7358,
+      "step": 722
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8521829612865914e-05,
+      "loss": 0.7186,
+      "step": 723
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.851713358221357e-05,
+      "loss": 0.7533,
+      "step": 724
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8512430701209034e-05,
+      "loss": 0.76,
+      "step": 725
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8507720973634836e-05,
+      "loss": 0.7357,
+      "step": 726
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8503004403279027e-05,
+      "loss": 0.7238,
+      "step": 727
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.849828099393516e-05,
+      "loss": 0.7423,
+      "step": 728
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8493550749402278e-05,
+      "loss": 0.7449,
+      "step": 729
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8488813673484933e-05,
+      "loss": 0.7444,
+      "step": 730
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8484069769993166e-05,
+      "loss": 0.7416,
+      "step": 731
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8479319042742512e-05,
+      "loss": 0.7559,
+      "step": 732
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8474561495553994e-05,
+      "loss": 0.7278,
+      "step": 733
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8469797132254114e-05,
+      "loss": 0.7218,
+      "step": 734
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8465025956674866e-05,
+      "loss": 0.7377,
+      "step": 735
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.846024797265372e-05,
+      "loss": 0.7456,
+      "step": 736
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8455463184033614e-05,
+      "loss": 0.7523,
+      "step": 737
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.845067159466297e-05,
+      "loss": 0.7234,
+      "step": 738
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8445873208395677e-05,
+      "loss": 0.7252,
+      "step": 739
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8441068029091087e-05,
+      "loss": 0.7155,
+      "step": 740
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8436256060614018e-05,
+      "loss": 0.729,
+      "step": 741
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.843143730683475e-05,
+      "loss": 0.7228,
+      "step": 742
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8426611771629018e-05,
+      "loss": 0.7278,
+      "step": 743
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.842177945887801e-05,
+      "loss": 0.7297,
+      "step": 744
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8416940372468372e-05,
+      "loss": 0.7108,
+      "step": 745
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8412094516292192e-05,
+      "loss": 0.7378,
+      "step": 746
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8407241894247002e-05,
+      "loss": 0.7473,
+      "step": 747
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.840238251023578e-05,
+      "loss": 0.7526,
+      "step": 748
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8397516368166938e-05,
+      "loss": 0.744,
+      "step": 749
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8392643471954336e-05,
+      "loss": 0.74,
+      "step": 750
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8387763825517246e-05,
+      "loss": 0.7454,
+      "step": 751
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8382877432780382e-05,
+      "loss": 0.7194,
+      "step": 752
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8377984297673882e-05,
+      "loss": 0.7278,
+      "step": 753
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8373084424133315e-05,
+      "loss": 0.7256,
+      "step": 754
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8368177816099655e-05,
+      "loss": 0.7278,
+      "step": 755
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.83632644775193e-05,
+      "loss": 0.7351,
+      "step": 756
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.835834441234406e-05,
+      "loss": 0.7185,
+      "step": 757
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8353417624531157e-05,
+      "loss": 0.7289,
+      "step": 758
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8348484118043215e-05,
+      "loss": 0.7402,
+      "step": 759
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8343543896848275e-05,
+      "loss": 0.7308,
+      "step": 760
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.833859696491976e-05,
+      "loss": 0.7312,
+      "step": 761
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8333643326236506e-05,
+      "loss": 0.7397,
+      "step": 762
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8328682984782734e-05,
+      "loss": 0.7146,
+      "step": 763
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8323715944548062e-05,
+      "loss": 0.7507,
+      "step": 764
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.831874220952749e-05,
+      "loss": 0.7456,
+      "step": 765
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.831376178372141e-05,
+      "loss": 0.7356,
+      "step": 766
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8308774671135587e-05,
+      "loss": 0.7256,
+      "step": 767
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.830378087578117e-05,
+      "loss": 0.7627,
+      "step": 768
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.829878040167468e-05,
+      "loss": 0.7395,
+      "step": 769
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.829377325283801e-05,
+      "loss": 0.7346,
+      "step": 770
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8288759433298428e-05,
+      "loss": 0.7303,
+      "step": 771
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8283738947088558e-05,
+      "loss": 0.7335,
+      "step": 772
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.827871179824639e-05,
+      "loss": 0.7424,
+      "step": 773
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.827367799081527e-05,
+      "loss": 0.7337,
+      "step": 774
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.826863752884391e-05,
+      "loss": 0.7169,
+      "step": 775
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.826359041638636e-05,
+      "loss": 0.7343,
+      "step": 776
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.825853665750202e-05,
+      "loss": 0.74,
+      "step": 777
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8253476256255654e-05,
+      "loss": 0.7402,
+      "step": 778
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.824840921671735e-05,
+      "loss": 0.7299,
+      "step": 779
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.824333554296254e-05,
+      "loss": 0.7231,
+      "step": 780
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8238255239071988e-05,
+      "loss": 0.7139,
+      "step": 781
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.823316830913181e-05,
+      "loss": 0.7039,
+      "step": 782
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.822807475723342e-05,
+      "loss": 0.7107,
+      "step": 783
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.822297458747358e-05,
+      "loss": 0.7246,
+      "step": 784
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8217867803954373e-05,
+      "loss": 0.7062,
+      "step": 785
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8212754410783197e-05,
+      "loss": 0.7405,
+      "step": 786
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8207634412072765e-05,
+      "loss": 0.7299,
+      "step": 787
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8202507811941107e-05,
+      "loss": 0.7237,
+      "step": 788
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.819737461451156e-05,
+      "loss": 0.7137,
+      "step": 789
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8192234823912768e-05,
+      "loss": 0.7271,
+      "step": 790
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8187088444278675e-05,
+      "loss": 0.7177,
+      "step": 791
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.818193547974853e-05,
+      "loss": 0.772,
+      "step": 792
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8176775934466876e-05,
+      "loss": 0.7103,
+      "step": 793
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8171609812583547e-05,
+      "loss": 0.7648,
+      "step": 794
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8166437118253667e-05,
+      "loss": 0.7216,
+      "step": 795
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8161257855637654e-05,
+      "loss": 0.7227,
+      "step": 796
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.815607202890119e-05,
+      "loss": 0.6947,
+      "step": 797
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8150879642215256e-05,
+      "loss": 0.7448,
+      "step": 798
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8145680699756103e-05,
+      "loss": 0.7079,
+      "step": 799
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.814047520570525e-05,
+      "loss": 0.7305,
+      "step": 800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.813526316424949e-05,
+      "loss": 0.734,
+      "step": 801
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8130044579580883e-05,
+      "loss": 0.7378,
+      "step": 802
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8124819455896753e-05,
+      "loss": 0.7449,
+      "step": 803
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8119587797399666e-05,
+      "loss": 0.7128,
+      "step": 804
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8114349608297478e-05,
+      "loss": 0.7194,
+      "step": 805
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8109104892803264e-05,
+      "loss": 0.7357,
+      "step": 806
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8103853655135366e-05,
+      "loss": 0.7294,
+      "step": 807
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8098595899517368e-05,
+      "loss": 0.7244,
+      "step": 808
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8093331630178096e-05,
+      "loss": 0.7379,
+      "step": 809
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8088060851351617e-05,
+      "loss": 0.7146,
+      "step": 810
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8082783567277226e-05,
+      "loss": 0.7314,
+      "step": 811
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8077499782199465e-05,
+      "loss": 0.7175,
+      "step": 812
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.807220950036809e-05,
+      "loss": 0.7235,
+      "step": 813
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.806691272603809e-05,
+      "loss": 0.7312,
+      "step": 814
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.806160946346967e-05,
+      "loss": 0.7701,
+      "step": 815
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8056299716928262e-05,
+      "loss": 0.741,
+      "step": 816
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8050983490684506e-05,
+      "loss": 0.7057,
+      "step": 817
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.804566078901426e-05,
+      "loss": 0.7311,
+      "step": 818
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8040331616198582e-05,
+      "loss": 0.7113,
+      "step": 819
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8034995976523745e-05,
+      "loss": 0.6962,
+      "step": 820
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8029653874281207e-05,
+      "loss": 0.7246,
+      "step": 821
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8024305313767648e-05,
+      "loss": 0.7152,
+      "step": 822
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8018950299284917e-05,
+      "loss": 0.7523,
+      "step": 823
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8013588835140073e-05,
+      "loss": 0.6957,
+      "step": 824
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8008220925645356e-05,
+      "loss": 0.752,
+      "step": 825
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8002846575118186e-05,
+      "loss": 0.7119,
+      "step": 826
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7997465787881166e-05,
+      "loss": 0.7141,
+      "step": 827
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7992078568262075e-05,
+      "loss": 0.7128,
+      "step": 828
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7986684920593874e-05,
+      "loss": 0.7083,
+      "step": 829
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.798128484921468e-05,
+      "loss": 0.7247,
+      "step": 830
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7975878358467794e-05,
+      "loss": 0.6941,
+      "step": 831
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.797046545270166e-05,
+      "loss": 0.7069,
+      "step": 832
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.79650461362699e-05,
+      "loss": 0.6956,
+      "step": 833
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7959620413531275e-05,
+      "loss": 0.7008,
+      "step": 834
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7954188288849716e-05,
+      "loss": 0.7217,
+      "step": 835
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.794874976659429e-05,
+      "loss": 0.7161,
+      "step": 836
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.794330485113922e-05,
+      "loss": 0.7335,
+      "step": 837
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.793785354686386e-05,
+      "loss": 0.704,
+      "step": 838
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7932395858152705e-05,
+      "loss": 0.7162,
+      "step": 839
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7926931789395395e-05,
+      "loss": 0.7097,
+      "step": 840
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.792146134498669e-05,
+      "loss": 0.7213,
+      "step": 841
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7915984529326483e-05,
+      "loss": 0.7095,
+      "step": 842
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.79105013468198e-05,
+      "loss": 0.7356,
+      "step": 843
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.790501180187676e-05,
+      "loss": 0.7189,
+      "step": 844
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.789951589891263e-05,
+      "loss": 0.7124,
+      "step": 845
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.789401364234778e-05,
+      "loss": 0.7325,
+      "step": 846
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7888505036607683e-05,
+      "loss": 0.7114,
+      "step": 847
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7882990086122926e-05,
+      "loss": 0.7322,
+      "step": 848
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.78774687953292e-05,
+      "loss": 0.7307,
+      "step": 849
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.78719411686673e-05,
+      "loss": 0.7245,
+      "step": 850
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7866407210583092e-05,
+      "loss": 0.7432,
+      "step": 851
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7860866925527566e-05,
+      "loss": 0.7204,
+      "step": 852
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7855320317956785e-05,
+      "loss": 0.7012,
+      "step": 853
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.78497673923319e-05,
+      "loss": 0.7124,
+      "step": 854
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.784420815311914e-05,
+      "loss": 0.7134,
+      "step": 855
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7838642604789822e-05,
+      "loss": 0.7083,
+      "step": 856
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7833070751820323e-05,
+      "loss": 0.7544,
+      "step": 857
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7827492598692103e-05,
+      "loss": 0.7155,
+      "step": 858
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.782190814989168e-05,
+      "loss": 0.7107,
+      "step": 859
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7816317409910645e-05,
+      "loss": 0.7188,
+      "step": 860
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7810720383245644e-05,
+      "loss": 0.7219,
+      "step": 861
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7805117074398376e-05,
+      "loss": 0.7245,
+      "step": 862
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7799507487875606e-05,
+      "loss": 0.7245,
+      "step": 863
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7793891628189125e-05,
+      "loss": 0.7264,
+      "step": 864
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.778826949985579e-05,
+      "loss": 0.7091,
+      "step": 865
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.77826411073975e-05,
+      "loss": 0.707,
+      "step": 866
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7777006455341178e-05,
+      "loss": 0.7112,
+      "step": 867
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7771365548218783e-05,
+      "loss": 0.7177,
+      "step": 868
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7765718390567323e-05,
+      "loss": 0.7063,
+      "step": 869
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7760064986928815e-05,
+      "loss": 0.7207,
+      "step": 870
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7754405341850308e-05,
+      "loss": 0.7168,
+      "step": 871
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7748739459883862e-05,
+      "loss": 0.6881,
+      "step": 872
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7743067345586572e-05,
+      "loss": 0.6992,
+      "step": 873
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.773738900352053e-05,
+      "loss": 0.7306,
+      "step": 874
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7731704438252834e-05,
+      "loss": 0.7041,
+      "step": 875
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.77260136543556e-05,
+      "loss": 0.7019,
+      "step": 876
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.772031665640594e-05,
+      "loss": 0.7052,
+      "step": 877
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7714613448985965e-05,
+      "loss": 0.7046,
+      "step": 878
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7708904036682774e-05,
+      "loss": 0.7218,
+      "step": 879
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.770318842408847e-05,
+      "loss": 0.7216,
+      "step": 880
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7697466615800124e-05,
+      "loss": 0.7058,
+      "step": 881
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.769173861641981e-05,
+      "loss": 0.7,
+      "step": 882
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.768600443055457e-05,
+      "loss": 0.7078,
+      "step": 883
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.768026406281642e-05,
+      "loss": 0.7106,
+      "step": 884
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7674517517822362e-05,
+      "loss": 0.7228,
+      "step": 885
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.766876480019435e-05,
+      "loss": 0.7324,
+      "step": 886
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7663005914559305e-05,
+      "loss": 0.7494,
+      "step": 887
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7657240865549122e-05,
+      "loss": 0.7301,
+      "step": 888
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.765146965780064e-05,
+      "loss": 0.6992,
+      "step": 889
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.764569229595566e-05,
+      "loss": 0.6912,
+      "step": 890
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7639908784660923e-05,
+      "loss": 0.7185,
+      "step": 891
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7634119128568126e-05,
+      "loss": 0.7153,
+      "step": 892
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7628323332333904e-05,
+      "loss": 0.7061,
+      "step": 893
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7622521400619828e-05,
+      "loss": 0.7125,
+      "step": 894
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7616713338092407e-05,
+      "loss": 0.7231,
+      "step": 895
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7610899149423085e-05,
+      "loss": 0.7111,
+      "step": 896
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7605078839288222e-05,
+      "loss": 0.704,
+      "step": 897
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7599252412369115e-05,
+      "loss": 0.7123,
+      "step": 898
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.759341987335197e-05,
+      "loss": 0.6932,
+      "step": 899
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.758758122692791e-05,
+      "loss": 0.6963,
+      "step": 900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.758173647779298e-05,
+      "loss": 0.6948,
+      "step": 901
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.757588563064812e-05,
+      "loss": 0.7329,
+      "step": 902
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.757002869019919e-05,
+      "loss": 0.6877,
+      "step": 903
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.756416566115693e-05,
+      "loss": 0.6915,
+      "step": 904
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7558296548236996e-05,
+      "loss": 0.7136,
+      "step": 905
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7552421356159935e-05,
+      "loss": 0.6907,
+      "step": 906
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.754654008965117e-05,
+      "loss": 0.7184,
+      "step": 907
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7540652753441026e-05,
+      "loss": 0.7174,
+      "step": 908
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7534759352264697e-05,
+      "loss": 0.7083,
+      "step": 909
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7528859890862263e-05,
+      "loss": 0.7031,
+      "step": 910
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.752295437397868e-05,
+      "loss": 0.6816,
+      "step": 911
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7517042806363758e-05,
+      "loss": 0.6998,
+      "step": 912
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7511125192772197e-05,
+      "loss": 0.7008,
+      "step": 913
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7505201537963546e-05,
+      "loss": 0.6958,
+      "step": 914
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7499271846702216e-05,
+      "loss": 0.6936,
+      "step": 915
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.749333612375747e-05,
+      "loss": 0.7256,
+      "step": 916
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7487394373903424e-05,
+      "loss": 0.6861,
+      "step": 917
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7481446601919052e-05,
+      "loss": 0.7365,
+      "step": 918
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.747549281258815e-05,
+      "loss": 0.7089,
+      "step": 919
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7469533010699374e-05,
+      "loss": 0.6824,
+      "step": 920
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7463567201046204e-05,
+      "loss": 0.7108,
+      "step": 921
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7457595388426955e-05,
+      "loss": 0.7145,
+      "step": 922
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7451617577644775e-05,
+      "loss": 0.6925,
+      "step": 923
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.744563377350763e-05,
+      "loss": 0.6956,
+      "step": 924
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7439643980828312e-05,
+      "loss": 0.6857,
+      "step": 925
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.743364820442442e-05,
+      "loss": 0.7164,
+      "step": 926
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7427646449118375e-05,
+      "loss": 0.6702,
+      "step": 927
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.742163871973741e-05,
+      "loss": 0.6904,
+      "step": 928
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7415625021113548e-05,
+      "loss": 0.718,
+      "step": 929
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.740960535808363e-05,
+      "loss": 0.7122,
+      "step": 930
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.740357973548928e-05,
+      "loss": 0.7227,
+      "step": 931
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.7397548158176923e-05,
+      "loss": 0.7056,
+      "step": 932
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.7391510630997775e-05,
+      "loss": 0.7106,
+      "step": 933
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.738546715880783e-05,
+      "loss": 0.687,
+      "step": 934
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.7379417746467875e-05,
+      "loss": 0.6876,
+      "step": 935
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.7373362398843462e-05,
+      "loss": 0.7104,
+      "step": 936
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.736730112080492e-05,
+      "loss": 0.705,
+      "step": 937
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.736123391722736e-05,
+      "loss": 0.696,
+      "step": 938
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.7355160792990636e-05,
+      "loss": 0.7095,
+      "step": 939
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.734908175297939e-05,
+      "loss": 0.685,
+      "step": 940
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.7342996802083e-05,
+      "loss": 0.7162,
+      "step": 941
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.7336905945195616e-05,
+      "loss": 0.6819,
+      "step": 942
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.7330809187216123e-05,
+      "loss": 0.7252,
+      "step": 943
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.7324706533048157e-05,
+      "loss": 0.6948,
+      "step": 944
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.7318597987600107e-05,
+      "loss": 0.6979,
+      "step": 945
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.7312483555785087e-05,
+      "loss": 0.7164,
+      "step": 946
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.7306363242520952e-05,
+      "loss": 0.6891,
+      "step": 947
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.7300237052730276e-05,
+      "loss": 0.729,
+      "step": 948
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.7294104991340384e-05,
+      "loss": 0.7002,
+      "step": 949
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.7287967063283296e-05,
+      "loss": 0.7117,
+      "step": 950
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.7281823273495772e-05,
+      "loss": 0.7,
+      "step": 951
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.7275673626919273e-05,
+      "loss": 0.6885,
+      "step": 952
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.7269518128499974e-05,
+      "loss": 0.6777,
+      "step": 953
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.7263356783188765e-05,
+      "loss": 0.7153,
+      "step": 954
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.7257189595941222e-05,
+      "loss": 0.7129,
+      "step": 955
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.725101657171764e-05,
+      "loss": 0.7178,
+      "step": 956
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.7244837715482987e-05,
+      "loss": 0.6814,
+      "step": 957
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.723865303220694e-05,
+      "loss": 0.7174,
+      "step": 958
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.7232462526863855e-05,
+      "loss": 0.7026,
+      "step": 959
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.722626620443277e-05,
+      "loss": 0.7218,
+      "step": 960
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.7220064069897402e-05,
+      "loss": 0.6958,
+      "step": 961
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.7213856128246146e-05,
+      "loss": 0.6984,
+      "step": 962
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.7207642384472062e-05,
+      "loss": 0.6993,
+      "step": 963
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.7201422843572883e-05,
+      "loss": 0.6926,
+      "step": 964
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.7195197510551e-05,
+      "loss": 0.7034,
+      "step": 965
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.7188966390413465e-05,
+      "loss": 0.7104,
+      "step": 966
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.7182729488171982e-05,
+      "loss": 0.7212,
+      "step": 967
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7176486808842907e-05,
+      "loss": 0.6996,
+      "step": 968
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7170238357447247e-05,
+      "loss": 0.7029,
+      "step": 969
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7163984139010643e-05,
+      "loss": 0.7092,
+      "step": 970
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.715772415856338e-05,
+      "loss": 0.6829,
+      "step": 971
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.715145842114038e-05,
+      "loss": 0.6931,
+      "step": 972
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7145186931781187e-05,
+      "loss": 0.7186,
+      "step": 973
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7138909695529975e-05,
+      "loss": 0.7065,
+      "step": 974
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7132626717435545e-05,
+      "loss": 0.7013,
+      "step": 975
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.712633800255131e-05,
+      "loss": 0.6882,
+      "step": 976
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.71200435559353e-05,
+      "loss": 0.6962,
+      "step": 977
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7113743382650156e-05,
+      "loss": 0.7103,
+      "step": 978
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7107437487763116e-05,
+      "loss": 0.6821,
+      "step": 979
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7101125876346037e-05,
+      "loss": 0.6997,
+      "step": 980
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.709480855347536e-05,
+      "loss": 0.7125,
+      "step": 981
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.708848552423212e-05,
+      "loss": 0.708,
+      "step": 982
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7082156793701954e-05,
+      "loss": 0.6887,
+      "step": 983
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.707582236697507e-05,
+      "loss": 0.6978,
+      "step": 984
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.706948224914626e-05,
+      "loss": 0.7084,
+      "step": 985
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.7063136445314912e-05,
+      "loss": 0.6725,
+      "step": 986
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.7056784960584956e-05,
+      "loss": 0.6648,
+      "step": 987
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.7050427800064912e-05,
+      "loss": 0.678,
+      "step": 988
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.7044064968867862e-05,
+      "loss": 0.6977,
+      "step": 989
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.7037696472111452e-05,
+      "loss": 0.6713,
+      "step": 990
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.703132231491787e-05,
+      "loss": 0.6873,
+      "step": 991
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.702494250241387e-05,
+      "loss": 0.6921,
+      "step": 992
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.7018557039730757e-05,
+      "loss": 0.7087,
+      "step": 993
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.7012165932004372e-05,
+      "loss": 0.6898,
+      "step": 994
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.7005769184375096e-05,
+      "loss": 0.6624,
+      "step": 995
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6999366801987856e-05,
+      "loss": 0.7067,
+      "step": 996
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.69929587899921e-05,
+      "loss": 0.6973,
+      "step": 997
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6986545153541802e-05,
+      "loss": 0.6762,
+      "step": 998
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6980125897795478e-05,
+      "loss": 0.694,
+      "step": 999
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6973701027916142e-05,
+      "loss": 0.6752,
+      "step": 1000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.696727054907134e-05,
+      "loss": 0.6816,
+      "step": 1001
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.696083446643312e-05,
+      "loss": 0.6953,
+      "step": 1002
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.695439278517804e-05,
+      "loss": 0.6603,
+      "step": 1003
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6947945510487164e-05,
+      "loss": 0.7061,
+      "step": 1004
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6941492647546044e-05,
+      "loss": 0.6927,
+      "step": 1005
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6935034201544737e-05,
+      "loss": 0.6727,
+      "step": 1006
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.692857017767779e-05,
+      "loss": 0.6636,
+      "step": 1007
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6922100581144228e-05,
+      "loss": 0.6917,
+      "step": 1008
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6915625417147575e-05,
+      "loss": 0.6908,
+      "step": 1009
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6909144690895807e-05,
+      "loss": 0.6958,
+      "step": 1010
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6902658407601404e-05,
+      "loss": 0.6779,
+      "step": 1011
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6896166572481288e-05,
+      "loss": 0.6953,
+      "step": 1012
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.688966919075687e-05,
+      "loss": 0.692,
+      "step": 1013
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6883166267653997e-05,
+      "loss": 0.6815,
+      "step": 1014
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6876657808402993e-05,
+      "loss": 0.6967,
+      "step": 1015
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6870143818238632e-05,
+      "loss": 0.6862,
+      "step": 1016
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.686362430240013e-05,
+      "loss": 0.6841,
+      "step": 1017
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.685709926613115e-05,
+      "loss": 0.6908,
+      "step": 1018
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.685056871467979e-05,
+      "loss": 0.6815,
+      "step": 1019
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.68440326532986e-05,
+      "loss": 0.6687,
+      "step": 1020
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.683749108724454e-05,
+      "loss": 0.6892,
+      "step": 1021
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.683094402177902e-05,
+      "loss": 0.6991,
+      "step": 1022
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6824391462167847e-05,
+      "loss": 0.7102,
+      "step": 1023
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6817833413681276e-05,
+      "loss": 0.6781,
+      "step": 1024
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.681126988159395e-05,
+      "loss": 0.7036,
+      "step": 1025
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6804700871184942e-05,
+      "loss": 0.6824,
+      "step": 1026
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.679812638773772e-05,
+      "loss": 0.6935,
+      "step": 1027
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6791546436540157e-05,
+      "loss": 0.6829,
+      "step": 1028
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.678496102288453e-05,
+      "loss": 0.6958,
+      "step": 1029
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.67783701520675e-05,
+      "loss": 0.6888,
+      "step": 1030
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6771773829390122e-05,
+      "loss": 0.6705,
+      "step": 1031
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6765172060157832e-05,
+      "loss": 0.7062,
+      "step": 1032
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6758564849680458e-05,
+      "loss": 0.6932,
+      "step": 1033
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6751952203272183e-05,
+      "loss": 0.6973,
+      "step": 1034
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.674533412625159e-05,
+      "loss": 0.6815,
+      "step": 1035
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6738710623941606e-05,
+      "loss": 0.6946,
+      "step": 1036
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.673208170166953e-05,
+      "loss": 0.6886,
+      "step": 1037
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6725447364767027e-05,
+      "loss": 0.6876,
+      "step": 1038
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.671880761857011e-05,
+      "loss": 0.6603,
+      "step": 1039
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.671216246841914e-05,
+      "loss": 0.7004,
+      "step": 1040
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6705511919658824e-05,
+      "loss": 0.7043,
+      "step": 1041
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.669885597763823e-05,
+      "loss": 0.7131,
+      "step": 1042
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.669219464771074e-05,
+      "loss": 0.6929,
+      "step": 1043
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.668552793523407e-05,
+      "loss": 0.6989,
+      "step": 1044
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.667885584557029e-05,
+      "loss": 0.7025,
+      "step": 1045
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.667217838408577e-05,
+      "loss": 0.6758,
+      "step": 1046
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6665495556151216e-05,
+      "loss": 0.6827,
+      "step": 1047
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.665880736714163e-05,
+      "loss": 0.6652,
+      "step": 1048
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.665211382243635e-05,
+      "loss": 0.667,
+      "step": 1049
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.664541492741901e-05,
+      "loss": 0.6894,
+      "step": 1050
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6638710687477545e-05,
+      "loss": 0.6579,
+      "step": 1051
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6632001108004194e-05,
+      "loss": 0.6715,
+      "step": 1052
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.662528619439549e-05,
+      "loss": 0.6644,
+      "step": 1053
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6618565952052254e-05,
+      "loss": 0.693,
+      "step": 1054
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6611840386379597e-05,
+      "loss": 0.6689,
+      "step": 1055
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.66051095027869e-05,
+      "loss": 0.6694,
+      "step": 1056
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.659837330668784e-05,
+      "loss": 0.6774,
+      "step": 1057
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.659163180350035e-05,
+      "loss": 0.6773,
+      "step": 1058
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.658488499864664e-05,
+      "loss": 0.6396,
+      "step": 1059
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6578132897553184e-05,
+      "loss": 0.6782,
+      "step": 1060
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.657137550565071e-05,
+      "loss": 0.6624,
+      "step": 1061
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6564612828374214e-05,
+      "loss": 0.6614,
+      "step": 1062
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6557844871162923e-05,
+      "loss": 0.6882,
+      "step": 1063
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.655107163946033e-05,
+      "loss": 0.6831,
+      "step": 1064
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6544293138714162e-05,
+      "loss": 0.6888,
+      "step": 1065
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6537509374376376e-05,
+      "loss": 0.6841,
+      "step": 1066
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.653072035190318e-05,
+      "loss": 0.6774,
+      "step": 1067
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6523926076755002e-05,
+      "loss": 0.6825,
+      "step": 1068
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6517126554396497e-05,
+      "loss": 0.6592,
+      "step": 1069
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6510321790296527e-05,
+      "loss": 0.6663,
+      "step": 1070
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.650351178992819e-05,
+      "loss": 0.6687,
+      "step": 1071
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.649669655876878e-05,
+      "loss": 0.6997,
+      "step": 1072
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6489876102299812e-05,
+      "loss": 0.6805,
+      "step": 1073
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6483050426006996e-05,
+      "loss": 0.6848,
+      "step": 1074
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6476219535380234e-05,
+      "loss": 0.6692,
+      "step": 1075
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6469383435913636e-05,
+      "loss": 0.6526,
+      "step": 1076
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6462542133105493e-05,
+      "loss": 0.6929,
+      "step": 1077
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6455695632458274e-05,
+      "loss": 0.6619,
+      "step": 1078
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.644884393947865e-05,
+      "loss": 0.6771,
+      "step": 1079
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6441987059677443e-05,
+      "loss": 0.6797,
+      "step": 1080
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6435124998569666e-05,
+      "loss": 0.6531,
+      "step": 1081
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6428257761674492e-05,
+      "loss": 0.7213,
+      "step": 1082
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6421385354515253e-05,
+      "loss": 0.6753,
+      "step": 1083
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6414507782619445e-05,
+      "loss": 0.6891,
+      "step": 1084
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.640762505151872e-05,
+      "loss": 0.6681,
+      "step": 1085
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6400737166748876e-05,
+      "loss": 0.6763,
+      "step": 1086
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.639384413384985e-05,
+      "loss": 0.6875,
+      "step": 1087
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6386945958365732e-05,
+      "loss": 0.6819,
+      "step": 1088
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6380042645844737e-05,
+      "loss": 0.6606,
+      "step": 1089
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6373134201839224e-05,
+      "loss": 0.6731,
+      "step": 1090
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6366220631905668e-05,
+      "loss": 0.6766,
+      "step": 1091
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6359301941604668e-05,
+      "loss": 0.6841,
+      "step": 1092
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6352378136500944e-05,
+      "loss": 0.6573,
+      "step": 1093
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6345449222163337e-05,
+      "loss": 0.6722,
+      "step": 1094
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6338515204164784e-05,
+      "loss": 0.6678,
+      "step": 1095
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.633157608808234e-05,
+      "loss": 0.6744,
+      "step": 1096
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6324631879497143e-05,
+      "loss": 0.6699,
+      "step": 1097
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.631768258399445e-05,
+      "loss": 0.6663,
+      "step": 1098
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6310728207163587e-05,
+      "loss": 0.6549,
+      "step": 1099
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6303768754597984e-05,
+      "loss": 0.7048,
+      "step": 1100
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.629680423189514e-05,
+      "loss": 0.6942,
+      "step": 1101
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6289834644656645e-05,
+      "loss": 0.6531,
+      "step": 1102
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.628285999848815e-05,
+      "loss": 0.674,
+      "step": 1103
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6275880298999384e-05,
+      "loss": 0.6835,
+      "step": 1104
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6268895551804136e-05,
+      "loss": 0.6805,
+      "step": 1105
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6261905762520257e-05,
+      "loss": 0.6615,
+      "step": 1106
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6254910936769654e-05,
+      "loss": 0.6511,
+      "step": 1107
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6247911080178276e-05,
+      "loss": 0.6506,
+      "step": 1108
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.624090619837613e-05,
+      "loss": 0.6578,
+      "step": 1109
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6233896296997263e-05,
+      "loss": 0.6555,
+      "step": 1110
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.622688138167975e-05,
+      "loss": 0.6549,
+      "step": 1111
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.621986145806571e-05,
+      "loss": 0.6635,
+      "step": 1112
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6212836531801288e-05,
+      "loss": 0.6794,
+      "step": 1113
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.620580660853664e-05,
+      "loss": 0.6717,
+      "step": 1114
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6198771693925967e-05,
+      "loss": 0.6746,
+      "step": 1115
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6191731793627453e-05,
+      "loss": 0.6861,
+      "step": 1116
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6184686913303314e-05,
+      "loss": 0.705,
+      "step": 1117
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6177637058619766e-05,
+      "loss": 0.6523,
+      "step": 1118
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6170582235247024e-05,
+      "loss": 0.659,
+      "step": 1119
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6163522448859302e-05,
+      "loss": 0.6537,
+      "step": 1120
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.61564577051348e-05,
+      "loss": 0.6549,
+      "step": 1121
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6149388009755714e-05,
+      "loss": 0.679,
+      "step": 1122
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6142313368408214e-05,
+      "loss": 0.665,
+      "step": 1123
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6135233786782455e-05,
+      "loss": 0.6611,
+      "step": 1124
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6128149270572562e-05,
+      "loss": 0.6716,
+      "step": 1125
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.612105982547663e-05,
+      "loss": 0.668,
+      "step": 1126
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6113965457196716e-05,
+      "loss": 0.6757,
+      "step": 1127
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.610686617143884e-05,
+      "loss": 0.6504,
+      "step": 1128
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.609976197391297e-05,
+      "loss": 0.6895,
+      "step": 1129
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.609265287033304e-05,
+      "loss": 0.6526,
+      "step": 1130
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6085538866416914e-05,
+      "loss": 0.6639,
+      "step": 1131
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6078419967886402e-05,
+      "loss": 0.6522,
+      "step": 1132
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6071296180467256e-05,
+      "loss": 0.6516,
+      "step": 1133
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6064167509889156e-05,
+      "loss": 0.6574,
+      "step": 1134
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6057033961885705e-05,
+      "loss": 0.6534,
+      "step": 1135
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6049895542194443e-05,
+      "loss": 0.6523,
+      "step": 1136
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6042752256556807e-05,
+      "loss": 0.6646,
+      "step": 1137
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6035604110718174e-05,
+      "loss": 0.6683,
+      "step": 1138
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6028451110427804e-05,
+      "loss": 0.634,
+      "step": 1139
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6021293261438877e-05,
+      "loss": 0.6639,
+      "step": 1140
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6014130569508472e-05,
+      "loss": 0.6658,
+      "step": 1141
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.600696304039756e-05,
+      "loss": 0.6681,
+      "step": 1142
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5999790679870993e-05,
+      "loss": 0.6517,
+      "step": 1143
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5992613493697536e-05,
+      "loss": 0.6501,
+      "step": 1144
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5985431487649803e-05,
+      "loss": 0.6319,
+      "step": 1145
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5978244667504307e-05,
+      "loss": 0.6646,
+      "step": 1146
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5971053039041424e-05,
+      "loss": 0.6473,
+      "step": 1147
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.59638566080454e-05,
+      "loss": 0.6547,
+      "step": 1148
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5956655380304345e-05,
+      "loss": 0.6671,
+      "step": 1149
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5949449361610218e-05,
+      "loss": 0.6493,
+      "step": 1150
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5942238557758845e-05,
+      "loss": 0.6693,
+      "step": 1151
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5935022974549887e-05,
+      "loss": 0.681,
+      "step": 1152
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5927802617786864e-05,
+      "loss": 0.6597,
+      "step": 1153
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.592057749327712e-05,
+      "loss": 0.6483,
+      "step": 1154
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5913347606831843e-05,
+      "loss": 0.6411,
+      "step": 1155
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5906112964266057e-05,
+      "loss": 0.6785,
+      "step": 1156
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.589887357139859e-05,
+      "loss": 0.647,
+      "step": 1157
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5891629434052115e-05,
+      "loss": 0.6504,
+      "step": 1158
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5884380558053104e-05,
+      "loss": 0.6677,
+      "step": 1159
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.587712694923185e-05,
+      "loss": 0.6555,
+      "step": 1160
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5869868613422444e-05,
+      "loss": 0.6704,
+      "step": 1161
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5862605556462787e-05,
+      "loss": 0.6367,
+      "step": 1162
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5855337784194576e-05,
+      "loss": 0.6504,
+      "step": 1163
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.58480653024633e-05,
+      "loss": 0.6465,
+      "step": 1164
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.584078811711823e-05,
+      "loss": 0.6587,
+      "step": 1165
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5833506234012425e-05,
+      "loss": 0.6702,
+      "step": 1166
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5826219659002722e-05,
+      "loss": 0.6554,
+      "step": 1167
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5818928397949735e-05,
+      "loss": 0.6545,
+      "step": 1168
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.581163245671784e-05,
+      "loss": 0.6479,
+      "step": 1169
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5804331841175188e-05,
+      "loss": 0.6516,
+      "step": 1170
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5797026557193674e-05,
+      "loss": 0.6722,
+      "step": 1171
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.578971661064896e-05,
+      "loss": 0.647,
+      "step": 1172
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5782402007420458e-05,
+      "loss": 0.6407,
+      "step": 1173
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5775082753391315e-05,
+      "loss": 0.6639,
+      "step": 1174
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5767758854448433e-05,
+      "loss": 0.6752,
+      "step": 1175
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5760430316482437e-05,
+      "loss": 0.6245,
+      "step": 1176
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.575309714538769e-05,
+      "loss": 0.6518,
+      "step": 1177
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5745759347062282e-05,
+      "loss": 0.6655,
+      "step": 1178
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.573841692740802e-05,
+      "loss": 0.6257,
+      "step": 1179
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5731069892330437e-05,
+      "loss": 0.6528,
+      "step": 1180
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5723718247738762e-05,
+      "loss": 0.6216,
+      "step": 1181
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.571636199954595e-05,
+      "loss": 0.6218,
+      "step": 1182
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5709001153668647e-05,
+      "loss": 0.644,
+      "step": 1183
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.57016357160272e-05,
+      "loss": 0.6316,
+      "step": 1184
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5694265692545648e-05,
+      "loss": 0.6495,
+      "step": 1185
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.568689108915172e-05,
+      "loss": 0.6431,
+      "step": 1186
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5679511911776833e-05,
+      "loss": 0.6371,
+      "step": 1187
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5672128166356073e-05,
+      "loss": 0.6439,
+      "step": 1188
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5664739858828206e-05,
+      "loss": 0.6317,
+      "step": 1189
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5657346995135668e-05,
+      "loss": 0.621,
+      "step": 1190
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5649949581224554e-05,
+      "loss": 0.6583,
+      "step": 1191
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.564254762304463e-05,
+      "loss": 0.6502,
+      "step": 1192
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.56351411265493e-05,
+      "loss": 0.6482,
+      "step": 1193
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.562773009769564e-05,
+      "loss": 0.6482,
+      "step": 1194
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.562031454244435e-05,
+      "loss": 0.634,
+      "step": 1195
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5612894466759784e-05,
+      "loss": 0.6309,
+      "step": 1196
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5605469876609925e-05,
+      "loss": 0.6276,
+      "step": 1197
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5598040777966394e-05,
+      "loss": 0.6459,
+      "step": 1198
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.559060717680443e-05,
+      "loss": 0.6724,
+      "step": 1199
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.55831690791029e-05,
+      "loss": 0.7018,
+      "step": 1200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5575726490844285e-05,
+      "loss": 0.6393,
+      "step": 1201
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5568279418014673e-05,
+      "loss": 0.6439,
+      "step": 1202
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5560827866603768e-05,
+      "loss": 0.6367,
+      "step": 1203
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5553371842604872e-05,
+      "loss": 0.6536,
+      "step": 1204
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.554591135201488e-05,
+      "loss": 0.6561,
+      "step": 1205
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.553844640083429e-05,
+      "loss": 0.5698,
+      "step": 1206
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5530976995067173e-05,
+      "loss": 0.5415,
+      "step": 1207
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5523503140721198e-05,
+      "loss": 0.5526,
+      "step": 1208
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.55160248438076e-05,
+      "loss": 0.5367,
+      "step": 1209
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.5508542110341197e-05,
+      "loss": 0.5283,
+      "step": 1210
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.550105494634037e-05,
+      "loss": 0.5273,
+      "step": 1211
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.5493563357827062e-05,
+      "loss": 0.5089,
+      "step": 1212
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.5486067350826773e-05,
+      "loss": 0.5488,
+      "step": 1213
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.5478566931368567e-05,
+      "loss": 0.5314,
+      "step": 1214
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5471062105485053e-05,
+      "loss": 0.5295,
+      "step": 1215
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.546355287921237e-05,
+      "loss": 0.5455,
+      "step": 1216
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5456039258590217e-05,
+      "loss": 0.5366,
+      "step": 1217
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5448521249661813e-05,
+      "loss": 0.5198,
+      "step": 1218
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5440998858473917e-05,
+      "loss": 0.5247,
+      "step": 1219
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5433472091076805e-05,
+      "loss": 0.5451,
+      "step": 1220
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.542594095352427e-05,
+      "loss": 0.5334,
+      "step": 1221
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.5418405451873637e-05,
+      "loss": 0.5441,
+      "step": 1222
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.541086559218572e-05,
+      "loss": 0.5272,
+      "step": 1223
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.5403321380524845e-05,
+      "loss": 0.5325,
+      "step": 1224
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.5395772822958844e-05,
+      "loss": 0.5182,
+      "step": 1225
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.5388219925559046e-05,
+      "loss": 0.538,
+      "step": 1226
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5380662694400262e-05,
+      "loss": 0.5408,
+      "step": 1227
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.537310113556079e-05,
+      "loss": 0.5371,
+      "step": 1228
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5365535255122413e-05,
+      "loss": 0.5251,
+      "step": 1229
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.535796505917039e-05,
+      "loss": 0.535,
+      "step": 1230
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5350390553793445e-05,
+      "loss": 0.5203,
+      "step": 1231
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5342811745083772e-05,
+      "loss": 0.5193,
+      "step": 1232
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.533522863913703e-05,
+      "loss": 0.5013,
+      "step": 1233
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5327641242052327e-05,
+      "loss": 0.5311,
+      "step": 1234
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.532004955993222e-05,
+      "loss": 0.541,
+      "step": 1235
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5312453598882734e-05,
+      "loss": 0.5216,
+      "step": 1236
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5304853365013312e-05,
+      "loss": 0.5341,
+      "step": 1237
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5297248864436833e-05,
+      "loss": 0.5145,
+      "step": 1238
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5289640103269626e-05,
+      "loss": 0.5369,
+      "step": 1239
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.528202708763143e-05,
+      "loss": 0.5404,
+      "step": 1240
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5274409823645428e-05,
+      "loss": 0.5138,
+      "step": 1241
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5266788317438185e-05,
+      "loss": 0.5266,
+      "step": 1242
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.525916257513971e-05,
+      "loss": 0.535,
+      "step": 1243
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5251532602883404e-05,
+      "loss": 0.5225,
+      "step": 1244
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5243898406806068e-05,
+      "loss": 0.5283,
+      "step": 1245
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5236259993047915e-05,
+      "loss": 0.5177,
+      "step": 1246
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5228617367752534e-05,
+      "loss": 0.5154,
+      "step": 1247
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5220970537066909e-05,
+      "loss": 0.5248,
+      "step": 1248
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5213319507141405e-05,
+      "loss": 0.5197,
+      "step": 1249
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.520566428412977e-05,
+      "loss": 0.53,
+      "step": 1250
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.519800487418911e-05,
+      "loss": 0.5258,
+      "step": 1251
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5190341283479916e-05,
+      "loss": 0.5261,
+      "step": 1252
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.518267351816603e-05,
+      "loss": 0.5194,
+      "step": 1253
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5175001584414655e-05,
+      "loss": 0.5069,
+      "step": 1254
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5167325488396348e-05,
+      "loss": 0.501,
+      "step": 1255
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.515964523628501e-05,
+      "loss": 0.5121,
+      "step": 1256
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.515196083425789e-05,
+      "loss": 0.5161,
+      "step": 1257
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5144272288495576e-05,
+      "loss": 0.5191,
+      "step": 1258
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5136579605181975e-05,
+      "loss": 0.512,
+      "step": 1259
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5128882790504338e-05,
+      "loss": 0.5155,
+      "step": 1260
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5121181850653232e-05,
+      "loss": 0.5308,
+      "step": 1261
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5113476791822543e-05,
+      "loss": 0.5463,
+      "step": 1262
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5105767620209469e-05,
+      "loss": 0.5262,
+      "step": 1263
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5098054342014514e-05,
+      "loss": 0.5456,
+      "step": 1264
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.509033696344149e-05,
+      "loss": 0.5256,
+      "step": 1265
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5082615490697506e-05,
+      "loss": 0.5302,
+      "step": 1266
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5074889929992957e-05,
+      "loss": 0.5359,
+      "step": 1267
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5067160287541532e-05,
+      "loss": 0.5229,
+      "step": 1268
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5059426569560205e-05,
+      "loss": 0.5191,
+      "step": 1269
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5051688782269226e-05,
+      "loss": 0.5255,
+      "step": 1270
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.504394693189211e-05,
+      "loss": 0.5191,
+      "step": 1271
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5036201024655656e-05,
+      "loss": 0.5074,
+      "step": 1272
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5028451066789913e-05,
+      "loss": 0.5259,
+      "step": 1273
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5020697064528186e-05,
+      "loss": 0.5377,
+      "step": 1274
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5012939024107049e-05,
+      "loss": 0.519,
+      "step": 1275
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5005176951766309e-05,
+      "loss": 0.5327,
+      "step": 1276
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4997410853749018e-05,
+      "loss": 0.5286,
+      "step": 1277
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.498964073630147e-05,
+      "loss": 0.5242,
+      "step": 1278
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4981866605673193e-05,
+      "loss": 0.5275,
+      "step": 1279
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4974088468116936e-05,
+      "loss": 0.5458,
+      "step": 1280
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.496630632988868e-05,
+      "loss": 0.5106,
+      "step": 1281
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4958520197247613e-05,
+      "loss": 0.5301,
+      "step": 1282
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4950730076456141e-05,
+      "loss": 0.5172,
+      "step": 1283
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4942935973779884e-05,
+      "loss": 0.5403,
+      "step": 1284
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4935137895487657e-05,
+      "loss": 0.5261,
+      "step": 1285
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.492733584785147e-05,
+      "loss": 0.5248,
+      "step": 1286
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4919529837146529e-05,
+      "loss": 0.5275,
+      "step": 1287
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4911719869651236e-05,
+      "loss": 0.5127,
+      "step": 1288
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4903905951647163e-05,
+      "loss": 0.5164,
+      "step": 1289
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4896088089419061e-05,
+      "loss": 0.5164,
+      "step": 1290
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.488826628925486e-05,
+      "loss": 0.5145,
+      "step": 1291
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4880440557445654e-05,
+      "loss": 0.5214,
+      "step": 1292
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4872610900285698e-05,
+      "loss": 0.5518,
+      "step": 1293
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4864777324072406e-05,
+      "loss": 0.5204,
+      "step": 1294
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4856939835106347e-05,
+      "loss": 0.5285,
+      "step": 1295
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4849098439691224e-05,
+      "loss": 0.5141,
+      "step": 1296
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4841253144133897e-05,
+      "loss": 0.524,
+      "step": 1297
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4833403954744356e-05,
+      "loss": 0.5186,
+      "step": 1298
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4825550877835726e-05,
+      "loss": 0.5392,
+      "step": 1299
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4817693919724256e-05,
+      "loss": 0.5101,
+      "step": 1300
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4809833086729317e-05,
+      "loss": 0.5407,
+      "step": 1301
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4801968385173393e-05,
+      "loss": 0.5386,
+      "step": 1302
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4794099821382091e-05,
+      "loss": 0.5394,
+      "step": 1303
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4786227401684113e-05,
+      "loss": 0.5387,
+      "step": 1304
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4778351132411263e-05,
+      "loss": 0.5239,
+      "step": 1305
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.477047101989845e-05,
+      "loss": 0.5076,
+      "step": 1306
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4762587070483662e-05,
+      "loss": 0.5211,
+      "step": 1307
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4754699290507988e-05,
+      "loss": 0.5267,
+      "step": 1308
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4746807686315584e-05,
+      "loss": 0.5355,
+      "step": 1309
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4738912264253692e-05,
+      "loss": 0.5356,
+      "step": 1310
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.473101303067261e-05,
+      "loss": 0.5273,
+      "step": 1311
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4723109991925722e-05,
+      "loss": 0.5082,
+      "step": 1312
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4715203154369455e-05,
+      "loss": 0.5035,
+      "step": 1313
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4707292524363305e-05,
+      "loss": 0.5297,
+      "step": 1314
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4699378108269808e-05,
+      "loss": 0.5335,
+      "step": 1315
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.469145991245455e-05,
+      "loss": 0.5175,
+      "step": 1316
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4683537943286155e-05,
+      "loss": 0.5049,
+      "step": 1317
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4675612207136283e-05,
+      "loss": 0.5122,
+      "step": 1318
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4667682710379625e-05,
+      "loss": 0.4989,
+      "step": 1319
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4659749459393894e-05,
+      "loss": 0.5203,
+      "step": 1320
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4651812460559825e-05,
+      "loss": 0.5109,
+      "step": 1321
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4643871720261166e-05,
+      "loss": 0.5125,
+      "step": 1322
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4635927244884676e-05,
+      "loss": 0.5369,
+      "step": 1323
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4627979040820116e-05,
+      "loss": 0.5226,
+      "step": 1324
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4620027114460244e-05,
+      "loss": 0.5101,
+      "step": 1325
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4612071472200818e-05,
+      "loss": 0.5222,
+      "step": 1326
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.460411212044058e-05,
+      "loss": 0.5071,
+      "step": 1327
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4596149065581255e-05,
+      "loss": 0.5154,
+      "step": 1328
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4588182314027553e-05,
+      "loss": 0.5328,
+      "step": 1329
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4580211872187147e-05,
+      "loss": 0.5593,
+      "step": 1330
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4572237746470686e-05,
+      "loss": 0.5258,
+      "step": 1331
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4564259943291782e-05,
+      "loss": 0.5308,
+      "step": 1332
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4556278469067e-05,
+      "loss": 0.5292,
+      "step": 1333
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.454829333021586e-05,
+      "loss": 0.4928,
+      "step": 1334
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.454030453316083e-05,
+      "loss": 0.5035,
+      "step": 1335
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4532312084327318e-05,
+      "loss": 0.4989,
+      "step": 1336
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4524315990143674e-05,
+      "loss": 0.502,
+      "step": 1337
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4516316257041176e-05,
+      "loss": 0.5055,
+      "step": 1338
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.450831289145403e-05,
+      "loss": 0.5345,
+      "step": 1339
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4500305899819357e-05,
+      "loss": 0.4873,
+      "step": 1340
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4492295288577211e-05,
+      "loss": 0.5137,
+      "step": 1341
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4484281064170543e-05,
+      "loss": 0.4888,
+      "step": 1342
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.447626323304521e-05,
+      "loss": 0.5176,
+      "step": 1343
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4468241801649978e-05,
+      "loss": 0.5253,
+      "step": 1344
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4460216776436505e-05,
+      "loss": 0.5133,
+      "step": 1345
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4452188163859336e-05,
+      "loss": 0.5313,
+      "step": 1346
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4444155970375906e-05,
+      "loss": 0.5164,
+      "step": 1347
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.4436120202446531e-05,
+      "loss": 0.5042,
+      "step": 1348
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.4428080866534397e-05,
+      "loss": 0.5199,
+      "step": 1349
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.442003796910556e-05,
+      "loss": 0.5054,
+      "step": 1350
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.441199151662895e-05,
+      "loss": 0.5287,
+      "step": 1351
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.4403941515576344e-05,
+      "loss": 0.5032,
+      "step": 1352
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.439588797242238e-05,
+      "loss": 0.525,
+      "step": 1353
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.438783089364454e-05,
+      "loss": 0.5085,
+      "step": 1354
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.4379770285723158e-05,
+      "loss": 0.5085,
+      "step": 1355
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.4371706155141399e-05,
+      "loss": 0.5049,
+      "step": 1356
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.4363638508385263e-05,
+      "loss": 0.5383,
+      "step": 1357
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.435556735194358e-05,
+      "loss": 0.4915,
+      "step": 1358
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.4347492692308001e-05,
+      "loss": 0.5125,
+      "step": 1359
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.4339414535972993e-05,
+      "loss": 0.4936,
+      "step": 1360
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.4331332889435842e-05,
+      "loss": 0.4786,
+      "step": 1361
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.4323247759196637e-05,
+      "loss": 0.5287,
+      "step": 1362
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.4315159151758264e-05,
+      "loss": 0.5359,
+      "step": 1363
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.4307067073626413e-05,
+      "loss": 0.5112,
+      "step": 1364
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.4298971531309565e-05,
+      "loss": 0.5267,
+      "step": 1365
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.4290872531318983e-05,
+      "loss": 0.514,
+      "step": 1366
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.4282770080168713e-05,
+      "loss": 0.5176,
+      "step": 1367
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.4274664184375582e-05,
+      "loss": 0.516,
+      "step": 1368
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.4266554850459173e-05,
+      "loss": 0.5177,
+      "step": 1369
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.4258442084941852e-05,
+      "loss": 0.5089,
+      "step": 1370
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.4250325894348733e-05,
+      "loss": 0.5096,
+      "step": 1371
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.4242206285207693e-05,
+      "loss": 0.4973,
+      "step": 1372
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.423408326404935e-05,
+      "loss": 0.4958,
+      "step": 1373
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.4225956837407069e-05,
+      "loss": 0.5031,
+      "step": 1374
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.421782701181696e-05,
+      "loss": 0.5199,
+      "step": 1375
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.4209693793817865e-05,
+      "loss": 0.4985,
+      "step": 1376
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.4201557189951345e-05,
+      "loss": 0.5019,
+      "step": 1377
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.4193417206761695e-05,
+      "loss": 0.5029,
+      "step": 1378
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.418527385079592e-05,
+      "loss": 0.4884,
+      "step": 1379
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.4177127128603748e-05,
+      "loss": 0.4984,
+      "step": 1380
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.4168977046737604e-05,
+      "loss": 0.5081,
+      "step": 1381
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.416082361175262e-05,
+      "loss": 0.4997,
+      "step": 1382
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.4152666830206629e-05,
+      "loss": 0.5036,
+      "step": 1383
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.4144506708660147e-05,
+      "loss": 0.5078,
+      "step": 1384
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.413634325367638e-05,
+      "loss": 0.4916,
+      "step": 1385
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.4128176471821218e-05,
+      "loss": 0.4962,
+      "step": 1386
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.4120006369663222e-05,
+      "loss": 0.5146,
+      "step": 1387
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.4111832953773625e-05,
+      "loss": 0.4994,
+      "step": 1388
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.410365623072633e-05,
+      "loss": 0.5197,
+      "step": 1389
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.4095476207097897e-05,
+      "loss": 0.5002,
+      "step": 1390
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.4087292889467536e-05,
+      "loss": 0.4792,
+      "step": 1391
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.4079106284417113e-05,
+      "loss": 0.5529,
+      "step": 1392
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.4070916398531128e-05,
+      "loss": 0.5265,
+      "step": 1393
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.4062723238396737e-05,
+      "loss": 0.5234,
+      "step": 1394
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.4054526810603715e-05,
+      "loss": 0.5229,
+      "step": 1395
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.404632712174447e-05,
+      "loss": 0.4993,
+      "step": 1396
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4038124178414032e-05,
+      "loss": 0.492,
+      "step": 1397
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4029917987210048e-05,
+      "loss": 0.502,
+      "step": 1398
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4021708554732785e-05,
+      "loss": 0.5086,
+      "step": 1399
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4013495887585102e-05,
+      "loss": 0.494,
+      "step": 1400
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4005279992372472e-05,
+      "loss": 0.51,
+      "step": 1401
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.3997060875702962e-05,
+      "loss": 0.4937,
+      "step": 1402
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.3988838544187223e-05,
+      "loss": 0.5081,
+      "step": 1403
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.3980613004438502e-05,
+      "loss": 0.5261,
+      "step": 1404
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.3972384263072623e-05,
+      "loss": 0.4795,
+      "step": 1405
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.396415232670798e-05,
+      "loss": 0.4948,
+      "step": 1406
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.395591720196554e-05,
+      "loss": 0.5227,
+      "step": 1407
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.3947678895468834e-05,
+      "loss": 0.514,
+      "step": 1408
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.3939437413843957e-05,
+      "loss": 0.5056,
+      "step": 1409
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.3931192763719552e-05,
+      "loss": 0.5117,
+      "step": 1410
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.3922944951726811e-05,
+      "loss": 0.5076,
+      "step": 1411
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.391469398449947e-05,
+      "loss": 0.5131,
+      "step": 1412
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.3906439868673805e-05,
+      "loss": 0.5139,
+      "step": 1413
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.389818261088862e-05,
+      "loss": 0.4998,
+      "step": 1414
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.388992221778525e-05,
+      "loss": 0.5181,
+      "step": 1415
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.3881658696007547e-05,
+      "loss": 0.4982,
+      "step": 1416
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.3873392052201886e-05,
+      "loss": 0.5178,
+      "step": 1417
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.3865122293017147e-05,
+      "loss": 0.5007,
+      "step": 1418
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.3856849425104728e-05,
+      "loss": 0.5267,
+      "step": 1419
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3848573455118503e-05,
+      "loss": 0.4899,
+      "step": 1420
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3840294389714865e-05,
+      "loss": 0.5016,
+      "step": 1421
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3832012235552683e-05,
+      "loss": 0.5184,
+      "step": 1422
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3823726999293317e-05,
+      "loss": 0.494,
+      "step": 1423
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3815438687600602e-05,
+      "loss": 0.4783,
+      "step": 1424
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.380714730714085e-05,
+      "loss": 0.4963,
+      "step": 1425
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3798852864582841e-05,
+      "loss": 0.4852,
+      "step": 1426
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3790555366597813e-05,
+      "loss": 0.4966,
+      "step": 1427
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3782254819859464e-05,
+      "loss": 0.5089,
+      "step": 1428
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3773951231043947e-05,
+      "loss": 0.5087,
+      "step": 1429
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.376564460682986e-05,
+      "loss": 0.4971,
+      "step": 1430
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.375733495389824e-05,
+      "loss": 0.5051,
+      "step": 1431
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.374902227893256e-05,
+      "loss": 0.5077,
+      "step": 1432
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.3740706588618733e-05,
+      "loss": 0.4953,
+      "step": 1433
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.3732387889645086e-05,
+      "loss": 0.5012,
+      "step": 1434
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.3724066188702368e-05,
+      "loss": 0.4921,
+      "step": 1435
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.3715741492483743e-05,
+      "loss": 0.5212,
+      "step": 1436
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.3707413807684788e-05,
+      "loss": 0.48,
+      "step": 1437
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3699083141003483e-05,
+      "loss": 0.5136,
+      "step": 1438
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3690749499140204e-05,
+      "loss": 0.4914,
+      "step": 1439
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3682412888797715e-05,
+      "loss": 0.5049,
+      "step": 1440
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3674073316681178e-05,
+      "loss": 0.5055,
+      "step": 1441
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.366573078949813e-05,
+      "loss": 0.496,
+      "step": 1442
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3657385313958486e-05,
+      "loss": 0.5126,
+      "step": 1443
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.364903689677454e-05,
+      "loss": 0.492,
+      "step": 1444
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3640685544660934e-05,
+      "loss": 0.5035,
+      "step": 1445
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3632331264334688e-05,
+      "loss": 0.504,
+      "step": 1446
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3623974062515171e-05,
+      "loss": 0.5327,
+      "step": 1447
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3615613945924103e-05,
+      "loss": 0.5073,
+      "step": 1448
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3607250921285544e-05,
+      "loss": 0.4845,
+      "step": 1449
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.3598884995325894e-05,
+      "loss": 0.4814,
+      "step": 1450
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.3590516174773895e-05,
+      "loss": 0.5005,
+      "step": 1451
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.3582144466360607e-05,
+      "loss": 0.4888,
+      "step": 1452
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.3573769876819415e-05,
+      "loss": 0.512,
+      "step": 1453
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.3565392412886028e-05,
+      "loss": 0.5002,
+      "step": 1454
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.3557012081298453e-05,
+      "loss": 0.4967,
+      "step": 1455
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.3548628888797019e-05,
+      "loss": 0.4996,
+      "step": 1456
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.3540242842124347e-05,
+      "loss": 0.5096,
+      "step": 1457
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.3531853948025354e-05,
+      "loss": 0.488,
+      "step": 1458
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.3523462213247251e-05,
+      "loss": 0.5068,
+      "step": 1459
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.3515067644539532e-05,
+      "loss": 0.4917,
+      "step": 1460
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.3506670248653967e-05,
+      "loss": 0.5052,
+      "step": 1461
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.3498270032344607e-05,
+      "loss": 0.5026,
+      "step": 1462
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.3489867002367766e-05,
+      "loss": 0.5038,
+      "step": 1463
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.348146116548202e-05,
+      "loss": 0.5046,
+      "step": 1464
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.3473052528448203e-05,
+      "loss": 0.4833,
+      "step": 1465
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.3464641098029412e-05,
+      "loss": 0.5164,
+      "step": 1466
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.3456226880990972e-05,
+      "loss": 0.494,
+      "step": 1467
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.3447809884100468e-05,
+      "loss": 0.4887,
+      "step": 1468
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.3439390114127705e-05,
+      "loss": 0.5,
+      "step": 1469
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.3430967577844727e-05,
+      "loss": 0.4983,
+      "step": 1470
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.3422542282025805e-05,
+      "loss": 0.5038,
+      "step": 1471
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.3414114233447424e-05,
+      "loss": 0.4937,
+      "step": 1472
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.3405683438888281e-05,
+      "loss": 0.5042,
+      "step": 1473
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.339724990512929e-05,
+      "loss": 0.5054,
+      "step": 1474
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.338881363895356e-05,
+      "loss": 0.5004,
+      "step": 1475
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3380374647146406e-05,
+      "loss": 0.4937,
+      "step": 1476
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.337193293649533e-05,
+      "loss": 0.4998,
+      "step": 1477
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3363488513790017e-05,
+      "loss": 0.5107,
+      "step": 1478
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.335504138582234e-05,
+      "loss": 0.4838,
+      "step": 1479
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3346591559386348e-05,
+      "loss": 0.5189,
+      "step": 1480
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3338139041278254e-05,
+      "loss": 0.489,
+      "step": 1481
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3329683838296442e-05,
+      "loss": 0.5277,
+      "step": 1482
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.332122595724145e-05,
+      "loss": 0.4979,
+      "step": 1483
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3312765404915979e-05,
+      "loss": 0.4893,
+      "step": 1484
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3304302188124868e-05,
+      "loss": 0.502,
+      "step": 1485
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3295836313675107e-05,
+      "loss": 0.4576,
+      "step": 1486
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3287367788375818e-05,
+      "loss": 0.4823,
+      "step": 1487
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3278896619038258e-05,
+      "loss": 0.5003,
+      "step": 1488
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3270422812475807e-05,
+      "loss": 0.4852,
+      "step": 1489
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3261946375503976e-05,
+      "loss": 0.4774,
+      "step": 1490
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3253467314940381e-05,
+      "loss": 0.4762,
+      "step": 1491
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.3244985637604754e-05,
+      "loss": 0.4933,
+      "step": 1492
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.3236501350318926e-05,
+      "loss": 0.5068,
+      "step": 1493
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.3228014459906831e-05,
+      "loss": 0.4756,
+      "step": 1494
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.32195249731945e-05,
+      "loss": 0.4684,
+      "step": 1495
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.3211032897010046e-05,
+      "loss": 0.5043,
+      "step": 1496
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.320253823818367e-05,
+      "loss": 0.5009,
+      "step": 1497
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.3194041003547643e-05,
+      "loss": 0.5026,
+      "step": 1498
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.3185541199936316e-05,
+      "loss": 0.5129,
+      "step": 1499
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.3177038834186107e-05,
+      "loss": 0.4885,
+      "step": 1500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.3168533913135481e-05,
+      "loss": 0.4809,
+      "step": 1501
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.3160026443624974e-05,
+      "loss": 0.5078,
+      "step": 1502
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.3151516432497164e-05,
+      "loss": 0.5111,
+      "step": 1503
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.314300388659667e-05,
+      "loss": 0.4702,
+      "step": 1504
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3134488812770161e-05,
+      "loss": 0.4806,
+      "step": 1505
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3125971217866332e-05,
+      "loss": 0.4931,
+      "step": 1506
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3117451108735905e-05,
+      "loss": 0.4822,
+      "step": 1507
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.310892849223162e-05,
+      "loss": 0.4863,
+      "step": 1508
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3100403375208246e-05,
+      "loss": 0.4905,
+      "step": 1509
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3091875764522556e-05,
+      "loss": 0.4947,
+      "step": 1510
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3083345667033325e-05,
+      "loss": 0.4617,
+      "step": 1511
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3074813089601335e-05,
+      "loss": 0.4654,
+      "step": 1512
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3066278039089355e-05,
+      "loss": 0.4758,
+      "step": 1513
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3057740522362159e-05,
+      "loss": 0.4877,
+      "step": 1514
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.304920054628648e-05,
+      "loss": 0.4881,
+      "step": 1515
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3040658117731051e-05,
+      "loss": 0.4774,
+      "step": 1516
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3032113243566566e-05,
+      "loss": 0.4783,
+      "step": 1517
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.302356593066569e-05,
+      "loss": 0.4843,
+      "step": 1518
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3015016185903046e-05,
+      "loss": 0.4896,
+      "step": 1519
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.300646401615522e-05,
+      "loss": 0.4531,
+      "step": 1520
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2997909428300742e-05,
+      "loss": 0.5045,
+      "step": 1521
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2989352429220085e-05,
+      "loss": 0.4672,
+      "step": 1522
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2980793025795672e-05,
+      "loss": 0.4485,
+      "step": 1523
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.297223122491185e-05,
+      "loss": 0.4606,
+      "step": 1524
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2963667033454892e-05,
+      "loss": 0.4888,
+      "step": 1525
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2955100458313006e-05,
+      "loss": 0.4671,
+      "step": 1526
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2946531506376308e-05,
+      "loss": 0.4759,
+      "step": 1527
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2937960184536825e-05,
+      "loss": 0.4712,
+      "step": 1528
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2929386499688499e-05,
+      "loss": 0.489,
+      "step": 1529
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.292081045872716e-05,
+      "loss": 0.4852,
+      "step": 1530
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2912232068550543e-05,
+      "loss": 0.4833,
+      "step": 1531
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2903651336058268e-05,
+      "loss": 0.483,
+      "step": 1532
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2895068268151841e-05,
+      "loss": 0.4837,
+      "step": 1533
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2886482871734644e-05,
+      "loss": 0.5092,
+      "step": 1534
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2877895153711935e-05,
+      "loss": 0.4852,
+      "step": 1535
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2869305120990841e-05,
+      "loss": 0.4595,
+      "step": 1536
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2860712780480333e-05,
+      "loss": 0.4752,
+      "step": 1537
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2852118139091268e-05,
+      "loss": 0.4744,
+      "step": 1538
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.284352120373633e-05,
+      "loss": 0.4781,
+      "step": 1539
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2834921981330059e-05,
+      "loss": 0.4706,
+      "step": 1540
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.282632047878883e-05,
+      "loss": 0.5077,
+      "step": 1541
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2817716703030851e-05,
+      "loss": 0.4578,
+      "step": 1542
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2809110660976165e-05,
+      "loss": 0.4595,
+      "step": 1543
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2800502359546636e-05,
+      "loss": 0.5007,
+      "step": 1544
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2791891805665934e-05,
+      "loss": 0.4633,
+      "step": 1545
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2783279006259554e-05,
+      "loss": 0.4884,
+      "step": 1546
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2774663968254792e-05,
+      "loss": 0.4583,
+      "step": 1547
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2766046698580745e-05,
+      "loss": 0.4636,
+      "step": 1548
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2757427204168306e-05,
+      "loss": 0.4707,
+      "step": 1549
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2748805491950159e-05,
+      "loss": 0.4785,
+      "step": 1550
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2740181568860761e-05,
+      "loss": 0.4556,
+      "step": 1551
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.273155544183636e-05,
+      "loss": 0.4716,
+      "step": 1552
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.2722927117814977e-05,
+      "loss": 0.4692,
+      "step": 1553
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.2714296603736384e-05,
+      "loss": 0.4786,
+      "step": 1554
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.2705663906542131e-05,
+      "loss": 0.4741,
+      "step": 1555
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.2697029033175522e-05,
+      "loss": 0.4842,
+      "step": 1556
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.2688391990581602e-05,
+      "loss": 0.5033,
+      "step": 1557
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.2679752785707171e-05,
+      "loss": 0.494,
+      "step": 1558
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.2671111425500757e-05,
+      "loss": 0.4885,
+      "step": 1559
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.2662467916912636e-05,
+      "loss": 0.4615,
+      "step": 1560
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.2653822266894794e-05,
+      "loss": 0.4523,
+      "step": 1561
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.2645174482400957e-05,
+      "loss": 0.4998,
+      "step": 1562
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.263652457038656e-05,
+      "loss": 0.4552,
+      "step": 1563
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.2627872537808745e-05,
+      "loss": 0.4838,
+      "step": 1564
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.261921839162637e-05,
+      "loss": 0.4702,
+      "step": 1565
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.2610562138799977e-05,
+      "loss": 0.4665,
+      "step": 1566
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.2601903786291822e-05,
+      "loss": 0.5164,
+      "step": 1567
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.2593243341065837e-05,
+      "loss": 0.4683,
+      "step": 1568
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.2584580810087637e-05,
+      "loss": 0.4894,
+      "step": 1569
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.257591620032452e-05,
+      "loss": 0.4679,
+      "step": 1570
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.2567249518745452e-05,
+      "loss": 0.4832,
+      "step": 1571
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.255858077232107e-05,
+      "loss": 0.4345,
+      "step": 1572
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.2549909968023668e-05,
+      "loss": 0.4642,
+      "step": 1573
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.2541237112827197e-05,
+      "loss": 0.4689,
+      "step": 1574
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.2532562213707248e-05,
+      "loss": 0.4621,
+      "step": 1575
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.2523885277641074e-05,
+      "loss": 0.4855,
+      "step": 1576
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2515206311607558e-05,
+      "loss": 0.4865,
+      "step": 1577
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2506525322587207e-05,
+      "loss": 0.45,
+      "step": 1578
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2497842317562165e-05,
+      "loss": 0.4771,
+      "step": 1579
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2489157303516198e-05,
+      "loss": 0.4638,
+      "step": 1580
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2480470287434683e-05,
+      "loss": 0.458,
+      "step": 1581
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2471781276304607e-05,
+      "loss": 0.4908,
+      "step": 1582
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2463090277114564e-05,
+      "loss": 0.4678,
+      "step": 1583
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2454397296854752e-05,
+      "loss": 0.4527,
+      "step": 1584
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.244570234251695e-05,
+      "loss": 0.4778,
+      "step": 1585
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2437005421094537e-05,
+      "loss": 0.4437,
+      "step": 1586
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2428306539582472e-05,
+      "loss": 0.4804,
+      "step": 1587
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2419605704977278e-05,
+      "loss": 0.4639,
+      "step": 1588
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.2410902924277064e-05,
+      "loss": 0.4644,
+      "step": 1589
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.2402198204481494e-05,
+      "loss": 0.4523,
+      "step": 1590
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.2393491552591806e-05,
+      "loss": 0.4748,
+      "step": 1591
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.2384782975610774e-05,
+      "loss": 0.4838,
+      "step": 1592
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.2376072480542731e-05,
+      "loss": 0.4725,
+      "step": 1593
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.236736007439355e-05,
+      "loss": 0.4615,
+      "step": 1594
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.2358645764170643e-05,
+      "loss": 0.4706,
+      "step": 1595
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.234992955688295e-05,
+      "loss": 0.453,
+      "step": 1596
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.234121145954094e-05,
+      "loss": 0.4768,
+      "step": 1597
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.23324914791566e-05,
+      "loss": 0.4595,
+      "step": 1598
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.2323769622743433e-05,
+      "loss": 0.4772,
+      "step": 1599
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.2315045897316446e-05,
+      "loss": 0.4602,
+      "step": 1600
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2306320309892161e-05,
+      "loss": 0.4691,
+      "step": 1601
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.229759286748859e-05,
+      "loss": 0.4637,
+      "step": 1602
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.228886357712523e-05,
+      "loss": 0.4686,
+      "step": 1603
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2280132445823075e-05,
+      "loss": 0.452,
+      "step": 1604
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2271399480604595e-05,
+      "loss": 0.4501,
+      "step": 1605
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.226266468849374e-05,
+      "loss": 0.4479,
+      "step": 1606
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2253928076515921e-05,
+      "loss": 0.4558,
+      "step": 1607
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.224518965169802e-05,
+      "loss": 0.441,
+      "step": 1608
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2236449421068373e-05,
+      "loss": 0.4513,
+      "step": 1609
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2227707391656772e-05,
+      "loss": 0.4509,
+      "step": 1610
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2218963570494449e-05,
+      "loss": 0.4631,
+      "step": 1611
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2210217964614084e-05,
+      "loss": 0.4692,
+      "step": 1612
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2201470581049792e-05,
+      "loss": 0.4261,
+      "step": 1613
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2192721426837112e-05,
+      "loss": 0.4428,
+      "step": 1614
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2183970509013015e-05,
+      "loss": 0.4595,
+      "step": 1615
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2175217834615884e-05,
+      "loss": 0.4761,
+      "step": 1616
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.216646341068552e-05,
+      "loss": 0.4421,
+      "step": 1617
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2157707244263122e-05,
+      "loss": 0.4401,
+      "step": 1618
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2148949342391302e-05,
+      "loss": 0.4568,
+      "step": 1619
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2140189712114066e-05,
+      "loss": 0.4529,
+      "step": 1620
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.21314283604768e-05,
+      "loss": 0.4468,
+      "step": 1621
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2122665294526284e-05,
+      "loss": 0.4337,
+      "step": 1622
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2113900521310674e-05,
+      "loss": 0.437,
+      "step": 1623
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2105134047879501e-05,
+      "loss": 0.4569,
+      "step": 1624
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.209636588128366e-05,
+      "loss": 0.4636,
+      "step": 1625
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.208759602857541e-05,
+      "loss": 0.4549,
+      "step": 1626
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2078824496808369e-05,
+      "loss": 0.4581,
+      "step": 1627
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2070051293037493e-05,
+      "loss": 0.4722,
+      "step": 1628
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2061276424319104e-05,
+      "loss": 0.4379,
+      "step": 1629
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2052499897710843e-05,
+      "loss": 0.4646,
+      "step": 1630
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2043721720271699e-05,
+      "loss": 0.4613,
+      "step": 1631
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2034941899061978e-05,
+      "loss": 0.4536,
+      "step": 1632
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2026160441143311e-05,
+      "loss": 0.4664,
+      "step": 1633
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.201737735357865e-05,
+      "loss": 0.4396,
+      "step": 1634
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2008592643432255e-05,
+      "loss": 0.4658,
+      "step": 1635
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1999806317769688e-05,
+      "loss": 0.4413,
+      "step": 1636
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.1991018383657817e-05,
+      "loss": 0.4547,
+      "step": 1637
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.1982228848164799e-05,
+      "loss": 0.4436,
+      "step": 1638
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.1973437718360075e-05,
+      "loss": 0.4447,
+      "step": 1639
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.1964645001314379e-05,
+      "loss": 0.4666,
+      "step": 1640
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.1955850704099712e-05,
+      "loss": 0.4556,
+      "step": 1641
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.1947054833789349e-05,
+      "loss": 0.48,
+      "step": 1642
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1938257397457832e-05,
+      "loss": 0.4557,
+      "step": 1643
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1929458402180964e-05,
+      "loss": 0.4323,
+      "step": 1644
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1920657855035796e-05,
+      "loss": 0.4681,
+      "step": 1645
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1911855763100634e-05,
+      "loss": 0.4443,
+      "step": 1646
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1903052133455018e-05,
+      "loss": 0.4607,
+      "step": 1647
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1894246973179732e-05,
+      "loss": 0.4406,
+      "step": 1648
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1885440289356791e-05,
+      "loss": 0.4353,
+      "step": 1649
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1876632089069431e-05,
+      "loss": 0.4301,
+      "step": 1650
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.186782237940211e-05,
+      "loss": 0.4557,
+      "step": 1651
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.18590111674405e-05,
+      "loss": 0.4592,
+      "step": 1652
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1850198460271483e-05,
+      "loss": 0.4184,
+      "step": 1653
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1841384264983135e-05,
+      "loss": 0.4418,
+      "step": 1654
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1832568588664746e-05,
+      "loss": 0.4438,
+      "step": 1655
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1823751438406779e-05,
+      "loss": 0.4457,
+      "step": 1656
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1814932821300891e-05,
+      "loss": 0.4778,
+      "step": 1657
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1806112744439922e-05,
+      "loss": 0.4445,
+      "step": 1658
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1797291214917882e-05,
+      "loss": 0.4638,
+      "step": 1659
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1788468239829949e-05,
+      "loss": 0.4631,
+      "step": 1660
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.177964382627246e-05,
+      "loss": 0.4564,
+      "step": 1661
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1770817981342915e-05,
+      "loss": 0.4337,
+      "step": 1662
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1761990712139968e-05,
+      "loss": 0.4537,
+      "step": 1663
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1753162025763409e-05,
+      "loss": 0.4499,
+      "step": 1664
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1744331929314179e-05,
+      "loss": 0.4525,
+      "step": 1665
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.173550042989434e-05,
+      "loss": 0.4469,
+      "step": 1666
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.172666753460709e-05,
+      "loss": 0.4511,
+      "step": 1667
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1717833250556759e-05,
+      "loss": 0.4274,
+      "step": 1668
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1708997584848772e-05,
+      "loss": 0.4215,
+      "step": 1669
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.170016054458968e-05,
+      "loss": 0.4451,
+      "step": 1670
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1691322136887139e-05,
+      "loss": 0.4286,
+      "step": 1671
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1682482368849909e-05,
+      "loss": 0.4445,
+      "step": 1672
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1673641247587825e-05,
+      "loss": 0.4426,
+      "step": 1673
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1664798780211834e-05,
+      "loss": 0.4476,
+      "step": 1674
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1655954973833955e-05,
+      "loss": 0.4342,
+      "step": 1675
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1647109835567272e-05,
+      "loss": 0.4385,
+      "step": 1676
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1638263372525965e-05,
+      "loss": 0.4326,
+      "step": 1677
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.162941559182526e-05,
+      "loss": 0.4414,
+      "step": 1678
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.1620566500581451e-05,
+      "loss": 0.4378,
+      "step": 1679
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.1611716105911885e-05,
+      "loss": 0.4596,
+      "step": 1680
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.1602864414934958e-05,
+      "loss": 0.4536,
+      "step": 1681
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.1594011434770107e-05,
+      "loss": 0.442,
+      "step": 1682
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.1585157172537807e-05,
+      "loss": 0.4458,
+      "step": 1683
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.1576301635359562e-05,
+      "loss": 0.4476,
+      "step": 1684
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1567444830357902e-05,
+      "loss": 0.4633,
+      "step": 1685
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1558586764656379e-05,
+      "loss": 0.4442,
+      "step": 1686
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1549727445379559e-05,
+      "loss": 0.4602,
+      "step": 1687
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1540866879653015e-05,
+      "loss": 0.4153,
+      "step": 1688
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1532005074603322e-05,
+      "loss": 0.4214,
+      "step": 1689
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.152314203735805e-05,
+      "loss": 0.4326,
+      "step": 1690
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1514277775045768e-05,
+      "loss": 0.4695,
+      "step": 1691
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1505412294796019e-05,
+      "loss": 0.4578,
+      "step": 1692
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1496545603739336e-05,
+      "loss": 0.4186,
+      "step": 1693
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1487677709007216e-05,
+      "loss": 0.4435,
+      "step": 1694
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1478808617732134e-05,
+      "loss": 0.4311,
+      "step": 1695
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1469938337047517e-05,
+      "loss": 0.4332,
+      "step": 1696
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1461066874087762e-05,
+      "loss": 0.4366,
+      "step": 1697
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1452194235988203e-05,
+      "loss": 0.4404,
+      "step": 1698
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.144332042988513e-05,
+      "loss": 0.459,
+      "step": 1699
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1434445462915762e-05,
+      "loss": 0.4429,
+      "step": 1700
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1425569342218261e-05,
+      "loss": 0.4199,
+      "step": 1701
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1416692074931716e-05,
+      "loss": 0.436,
+      "step": 1702
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1407813668196133e-05,
+      "loss": 0.4543,
+      "step": 1703
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1398934129152436e-05,
+      "loss": 0.4469,
+      "step": 1704
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.139005346494246e-05,
+      "loss": 0.4402,
+      "step": 1705
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1381171682708951e-05,
+      "loss": 0.4433,
+      "step": 1706
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1372288789595545e-05,
+      "loss": 0.4431,
+      "step": 1707
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1363404792746775e-05,
+      "loss": 0.4469,
+      "step": 1708
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1354519699308065e-05,
+      "loss": 0.4332,
+      "step": 1709
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1345633516425714e-05,
+      "loss": 0.4386,
+      "step": 1710
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1336746251246908e-05,
+      "loss": 0.4268,
+      "step": 1711
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.132785791091969e-05,
+      "loss": 0.4268,
+      "step": 1712
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.131896850259298e-05,
+      "loss": 0.4202,
+      "step": 1713
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.131007803341655e-05,
+      "loss": 0.4352,
+      "step": 1714
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1301186510541024e-05,
+      "loss": 0.4296,
+      "step": 1715
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.129229394111788e-05,
+      "loss": 0.4484,
+      "step": 1716
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1283400332299434e-05,
+      "loss": 0.4191,
+      "step": 1717
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.127450569123884e-05,
+      "loss": 0.4576,
+      "step": 1718
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1265610025090075e-05,
+      "loss": 0.4338,
+      "step": 1719
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.125671334100795e-05,
+      "loss": 0.4078,
+      "step": 1720
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1247815646148088e-05,
+      "loss": 0.4255,
+      "step": 1721
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1238916947666932e-05,
+      "loss": 0.4669,
+      "step": 1722
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1230017252721722e-05,
+      "loss": 0.4246,
+      "step": 1723
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.122111656847051e-05,
+      "loss": 0.4189,
+      "step": 1724
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.121221490207213e-05,
+      "loss": 0.4191,
+      "step": 1725
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1203312260686232e-05,
+      "loss": 0.4291,
+      "step": 1726
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1194408651473219e-05,
+      "loss": 0.4474,
+      "step": 1727
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1185504081594292e-05,
+      "loss": 0.4149,
+      "step": 1728
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1176598558211413e-05,
+      "loss": 0.4294,
+      "step": 1729
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1167692088487328e-05,
+      "loss": 0.4387,
+      "step": 1730
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1158784679585523e-05,
+      "loss": 0.4257,
+      "step": 1731
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1149876338670258e-05,
+      "loss": 0.4221,
+      "step": 1732
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.1140967072906528e-05,
+      "loss": 0.4536,
+      "step": 1733
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.1132056889460078e-05,
+      "loss": 0.403,
+      "step": 1734
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.1123145795497396e-05,
+      "loss": 0.4519,
+      "step": 1735
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.1114233798185694e-05,
+      "loss": 0.4341,
+      "step": 1736
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.1105320904692916e-05,
+      "loss": 0.4079,
+      "step": 1737
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.1096407122187722e-05,
+      "loss": 0.4391,
+      "step": 1738
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1087492457839492e-05,
+      "loss": 0.4258,
+      "step": 1739
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1078576918818314e-05,
+      "loss": 0.4186,
+      "step": 1740
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.106966051229498e-05,
+      "loss": 0.4225,
+      "step": 1741
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1060743245440975e-05,
+      "loss": 0.3912,
+      "step": 1742
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1051825125428478e-05,
+      "loss": 0.427,
+      "step": 1743
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1042906159430358e-05,
+      "loss": 0.4055,
+      "step": 1744
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1033986354620165e-05,
+      "loss": 0.433,
+      "step": 1745
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1025065718172117e-05,
+      "loss": 0.4171,
+      "step": 1746
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1016144257261106e-05,
+      "loss": 0.424,
+      "step": 1747
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1007221979062683e-05,
+      "loss": 0.4081,
+      "step": 1748
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0998298890753063e-05,
+      "loss": 0.4206,
+      "step": 1749
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0989374999509109e-05,
+      "loss": 0.4213,
+      "step": 1750
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0980450312508324e-05,
+      "loss": 0.4233,
+      "step": 1751
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.097152483692886e-05,
+      "loss": 0.4334,
+      "step": 1752
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.09625985799495e-05,
+      "loss": 0.415,
+      "step": 1753
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0953671548749653e-05,
+      "loss": 0.4005,
+      "step": 1754
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0944743750509358e-05,
+      "loss": 0.419,
+      "step": 1755
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.093581519240926e-05,
+      "loss": 0.4274,
+      "step": 1756
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0926885881630624e-05,
+      "loss": 0.4406,
+      "step": 1757
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0917955825355316e-05,
+      "loss": 0.4296,
+      "step": 1758
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0909025030765804e-05,
+      "loss": 0.4215,
+      "step": 1759
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0900093505045148e-05,
+      "loss": 0.4326,
+      "step": 1760
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0891161255376998e-05,
+      "loss": 0.4139,
+      "step": 1761
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0882228288945585e-05,
+      "loss": 0.4155,
+      "step": 1762
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0873294612935714e-05,
+      "loss": 0.4128,
+      "step": 1763
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0864360234532767e-05,
+      "loss": 0.4353,
+      "step": 1764
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0855425160922685e-05,
+      "loss": 0.4414,
+      "step": 1765
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.084648939929197e-05,
+      "loss": 0.4385,
+      "step": 1766
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0837552956827682e-05,
+      "loss": 0.4131,
+      "step": 1767
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0828615840717416e-05,
+      "loss": 0.4261,
+      "step": 1768
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0819678058149326e-05,
+      "loss": 0.4199,
+      "step": 1769
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0810739616312092e-05,
+      "loss": 0.4092,
+      "step": 1770
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0801800522394921e-05,
+      "loss": 0.4022,
+      "step": 1771
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0792860783587547e-05,
+      "loss": 0.4089,
+      "step": 1772
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0783920407080233e-05,
+      "loss": 0.4091,
+      "step": 1773
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0774979400063741e-05,
+      "loss": 0.4009,
+      "step": 1774
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0766037769729347e-05,
+      "loss": 0.417,
+      "step": 1775
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.075709552326883e-05,
+      "loss": 0.4236,
+      "step": 1776
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0748152667874456e-05,
+      "loss": 0.4086,
+      "step": 1777
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0739209210738989e-05,
+      "loss": 0.4171,
+      "step": 1778
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0730265159055675e-05,
+      "loss": 0.4258,
+      "step": 1779
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0721320520018238e-05,
+      "loss": 0.4128,
+      "step": 1780
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.071237530082087e-05,
+      "loss": 0.4323,
+      "step": 1781
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0703429508658234e-05,
+      "loss": 0.3966,
+      "step": 1782
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0694483150725458e-05,
+      "loss": 0.3921,
+      "step": 1783
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0685536234218122e-05,
+      "loss": 0.3814,
+      "step": 1784
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0676588766332248e-05,
+      "loss": 0.411,
+      "step": 1785
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0667640754264303e-05,
+      "loss": 0.409,
+      "step": 1786
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0658692205211207e-05,
+      "loss": 0.4078,
+      "step": 1787
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0649743126370296e-05,
+      "loss": 0.3901,
+      "step": 1788
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0640793524939334e-05,
+      "loss": 0.4026,
+      "step": 1789
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0631843408116513e-05,
+      "loss": 0.4001,
+      "step": 1790
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0622892783100431e-05,
+      "loss": 0.4234,
+      "step": 1791
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.06139416570901e-05,
+      "loss": 0.4151,
+      "step": 1792
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0604990037284932e-05,
+      "loss": 0.4139,
+      "step": 1793
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0596037930884741e-05,
+      "loss": 0.3943,
+      "step": 1794
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0587085345089725e-05,
+      "loss": 0.4038,
+      "step": 1795
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0578132287100471e-05,
+      "loss": 0.4166,
+      "step": 1796
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0569178764117951e-05,
+      "loss": 0.4271,
+      "step": 1797
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0560224783343505e-05,
+      "loss": 0.4442,
+      "step": 1798
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0551270351978842e-05,
+      "loss": 0.4071,
+      "step": 1799
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0542315477226033e-05,
+      "loss": 0.3951,
+      "step": 1800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0533360166287503e-05,
+      "loss": 0.4222,
+      "step": 1801
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0524404426366038e-05,
+      "loss": 0.4172,
+      "step": 1802
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0515448264664759e-05,
+      "loss": 0.4153,
+      "step": 1803
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0506491688387128e-05,
+      "loss": 0.399,
+      "step": 1804
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0497534704736947e-05,
+      "loss": 0.4023,
+      "step": 1805
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0488577320918336e-05,
+      "loss": 0.4071,
+      "step": 1806
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0479619544135747e-05,
+      "loss": 0.4066,
+      "step": 1807
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0470661381593936e-05,
+      "loss": 0.4036,
+      "step": 1808
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0461702840497979e-05,
+      "loss": 0.323,
+      "step": 1809
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0452743928053255e-05,
+      "loss": 0.2951,
+      "step": 1810
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0443784651465435e-05,
+      "loss": 0.308,
+      "step": 1811
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.0434825017940494e-05,
+      "loss": 0.2872,
+      "step": 1812
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.0425865034684685e-05,
+      "loss": 0.3078,
+      "step": 1813
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.041690470890455e-05,
+      "loss": 0.287,
+      "step": 1814
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.0407944047806891e-05,
+      "loss": 0.3067,
+      "step": 1815
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.03989830585988e-05,
+      "loss": 0.2938,
+      "step": 1816
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.0390021748487623e-05,
+      "loss": 0.3028,
+      "step": 1817
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.0381060124680958e-05,
+      "loss": 0.2967,
+      "step": 1818
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.0372098194386667e-05,
+      "loss": 0.3079,
+      "step": 1819
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.0363135964812851e-05,
+      "loss": 0.3043,
+      "step": 1820
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.035417344316786e-05,
+      "loss": 0.2914,
+      "step": 1821
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.0345210636660263e-05,
+      "loss": 0.3018,
+      "step": 1822
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.0336247552498874e-05,
+      "loss": 0.304,
+      "step": 1823
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.0327284197892726e-05,
+      "loss": 0.29,
+      "step": 1824
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.0318320580051063e-05,
+      "loss": 0.2791,
+      "step": 1825
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.0309356706183348e-05,
+      "loss": 0.2922,
+      "step": 1826
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.0300392583499248e-05,
+      "loss": 0.2917,
+      "step": 1827
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.0291428219208633e-05,
+      "loss": 0.3051,
+      "step": 1828
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.0282463620521554e-05,
+      "loss": 0.2781,
+      "step": 1829
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.0273498794648271e-05,
+      "loss": 0.2767,
+      "step": 1830
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.0264533748799213e-05,
+      "loss": 0.2991,
+      "step": 1831
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.0255568490184985e-05,
+      "loss": 0.3108,
+      "step": 1832
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.0246603026016372e-05,
+      "loss": 0.3057,
+      "step": 1833
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.0237637363504318e-05,
+      "loss": 0.3221,
+      "step": 1834
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.0228671509859929e-05,
+      "loss": 0.3097,
+      "step": 1835
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.0219705472294463e-05,
+      "loss": 0.2913,
+      "step": 1836
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.0210739258019326e-05,
+      "loss": 0.2861,
+      "step": 1837
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.0201772874246067e-05,
+      "loss": 0.3048,
+      "step": 1838
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.0192806328186373e-05,
+      "loss": 0.2969,
+      "step": 1839
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.0183839627052059e-05,
+      "loss": 0.2771,
+      "step": 1840
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.0174872778055066e-05,
+      "loss": 0.296,
+      "step": 1841
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.0165905788407455e-05,
+      "loss": 0.2934,
+      "step": 1842
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.0156938665321397e-05,
+      "loss": 0.2808,
+      "step": 1843
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.0147971416009165e-05,
+      "loss": 0.3057,
+      "step": 1844
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.0139004047683152e-05,
+      "loss": 0.2826,
+      "step": 1845
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.013003656755583e-05,
+      "loss": 0.2867,
+      "step": 1846
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.0121068982839765e-05,
+      "loss": 0.2946,
+      "step": 1847
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.011210130074761e-05,
+      "loss": 0.2914,
+      "step": 1848
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.0103133528492091e-05,
+      "loss": 0.2782,
+      "step": 1849
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.0094165673286015e-05,
+      "loss": 0.2931,
+      "step": 1850
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.008519774234225e-05,
+      "loss": 0.3026,
+      "step": 1851
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.0076229742873722e-05,
+      "loss": 0.287,
+      "step": 1852
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.0067261682093419e-05,
+      "loss": 0.294,
+      "step": 1853
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.0058293567214372e-05,
+      "loss": 0.2943,
+      "step": 1854
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.0049325405449663e-05,
+      "loss": 0.2782,
+      "step": 1855
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.0040357204012405e-05,
+      "loss": 0.2818,
+      "step": 1856
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.0031388970115748e-05,
+      "loss": 0.2905,
+      "step": 1857
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.0022420710972856e-05,
+      "loss": 0.286,
+      "step": 1858
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.0013452433796931e-05,
+      "loss": 0.294,
+      "step": 1859
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.000448414580118e-05,
+      "loss": 0.2918,
+      "step": 1860
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.995515854198821e-06,
+      "loss": 0.2856,
+      "step": 1861
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.98654756620307e-06,
+      "loss": 0.2874,
+      "step": 1862
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.977579289027147e-06,
+      "loss": 0.2729,
+      "step": 1863
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.968611029884256e-06,
+      "loss": 0.2961,
+      "step": 1864
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.959642795987596e-06,
+      "loss": 0.2908,
+      "step": 1865
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.950674594550339e-06,
+      "loss": 0.2833,
+      "step": 1866
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.941706432785628e-06,
+      "loss": 0.3113,
+      "step": 1867
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.932738317906584e-06,
+      "loss": 0.2832,
+      "step": 1868
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.923770257126281e-06,
+      "loss": 0.2979,
+      "step": 1869
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.914802257657755e-06,
+      "loss": 0.2798,
+      "step": 1870
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.905834326713988e-06,
+      "loss": 0.291,
+      "step": 1871
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.896866471507912e-06,
+      "loss": 0.2851,
+      "step": 1872
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.887898699252393e-06,
+      "loss": 0.2908,
+      "step": 1873
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.878931017160238e-06,
+      "loss": 0.2881,
+      "step": 1874
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.869963432444173e-06,
+      "loss": 0.2946,
+      "step": 1875
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.860995952316851e-06,
+      "loss": 0.2809,
+      "step": 1876
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.852028583990836e-06,
+      "loss": 0.2827,
+      "step": 1877
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.843061334678607e-06,
+      "loss": 0.2999,
+      "step": 1878
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.834094211592547e-06,
+      "loss": 0.2991,
+      "step": 1879
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.825127221944934e-06,
+      "loss": 0.2668,
+      "step": 1880
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.81616037294794e-06,
+      "loss": 0.2666,
+      "step": 1881
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.807193671813627e-06,
+      "loss": 0.2893,
+      "step": 1882
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.798227125753936e-06,
+      "loss": 0.2893,
+      "step": 1883
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.789260741980679e-06,
+      "loss": 0.2825,
+      "step": 1884
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.780294527705543e-06,
+      "loss": 0.2771,
+      "step": 1885
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.771328490140076e-06,
+      "loss": 0.2895,
+      "step": 1886
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.762362636495684e-06,
+      "loss": 0.2898,
+      "step": 1887
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.753396973983631e-06,
+      "loss": 0.2659,
+      "step": 1888
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.744431509815017e-06,
+      "loss": 0.2931,
+      "step": 1889
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.73546625120079e-06,
+      "loss": 0.3076,
+      "step": 1890
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.726501205351732e-06,
+      "loss": 0.2796,
+      "step": 1891
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.717536379478448e-06,
+      "loss": 0.2747,
+      "step": 1892
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.70857178079137e-06,
+      "loss": 0.2773,
+      "step": 1893
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.699607416500752e-06,
+      "loss": 0.2924,
+      "step": 1894
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.690643293816654e-06,
+      "loss": 0.275,
+      "step": 1895
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.681679419948939e-06,
+      "loss": 0.2776,
+      "step": 1896
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.67271580210728e-06,
+      "loss": 0.2992,
+      "step": 1897
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.66375244750113e-06,
+      "loss": 0.2682,
+      "step": 1898
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.654789363339742e-06,
+      "loss": 0.2853,
+      "step": 1899
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.645826556832147e-06,
+      "loss": 0.2689,
+      "step": 1900
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.63686403518715e-06,
+      "loss": 0.2667,
+      "step": 1901
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.627901805613335e-06,
+      "loss": 0.2791,
+      "step": 1902
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.618939875319045e-06,
+      "loss": 0.2873,
+      "step": 1903
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.609978251512382e-06,
+      "loss": 0.256,
+      "step": 1904
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.601016941401202e-06,
+      "loss": 0.2734,
+      "step": 1905
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.59205595219311e-06,
+      "loss": 0.2724,
+      "step": 1906
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.583095291095454e-06,
+      "loss": 0.2722,
+      "step": 1907
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.574134965315315e-06,
+      "loss": 0.2845,
+      "step": 1908
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.565174982059506e-06,
+      "loss": 0.2682,
+      "step": 1909
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.556215348534565e-06,
+      "loss": 0.2933,
+      "step": 1910
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.547256071946749e-06,
+      "loss": 0.2779,
+      "step": 1911
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.538297159502026e-06,
+      "loss": 0.2769,
+      "step": 1912
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.52933861840607e-06,
+      "loss": 0.268,
+      "step": 1913
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.52038045586426e-06,
+      "loss": 0.2705,
+      "step": 1914
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.511422679081667e-06,
+      "loss": 0.3074,
+      "step": 1915
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.502465295263056e-06,
+      "loss": 0.2637,
+      "step": 1916
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.493508311612874e-06,
+      "loss": 0.3004,
+      "step": 1917
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.484551735335246e-06,
+      "loss": 0.302,
+      "step": 1918
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.475595573633967e-06,
+      "loss": 0.3042,
+      "step": 1919
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.4666398337125e-06,
+      "loss": 0.2939,
+      "step": 1920
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.457684522773972e-06,
+      "loss": 0.275,
+      "step": 1921
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.44872964802116e-06,
+      "loss": 0.2992,
+      "step": 1922
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.439775216656495e-06,
+      "loss": 0.2839,
+      "step": 1923
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.430821235882049e-06,
+      "loss": 0.2715,
+      "step": 1924
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.421867712899529e-06,
+      "loss": 0.3012,
+      "step": 1925
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.412914654910277e-06,
+      "loss": 0.2783,
+      "step": 1926
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.403962069115264e-06,
+      "loss": 0.2874,
+      "step": 1927
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.395009962715071e-06,
+      "loss": 0.2712,
+      "step": 1928
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.386058342909904e-06,
+      "loss": 0.2741,
+      "step": 1929
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.377107216899572e-06,
+      "loss": 0.2632,
+      "step": 1930
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.36815659188349e-06,
+      "loss": 0.2687,
+      "step": 1931
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.359206475060668e-06,
+      "loss": 0.2818,
+      "step": 1932
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.350256873629708e-06,
+      "loss": 0.2588,
+      "step": 1933
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.341307794788796e-06,
+      "loss": 0.2604,
+      "step": 1934
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.332359245735698e-06,
+      "loss": 0.3,
+      "step": 1935
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.323411233667757e-06,
+      "loss": 0.2798,
+      "step": 1936
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.31446376578188e-06,
+      "loss": 0.2781,
+      "step": 1937
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.305516849274542e-06,
+      "loss": 0.2822,
+      "step": 1938
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.296570491341766e-06,
+      "loss": 0.2921,
+      "step": 1939
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.287624699179134e-06,
+      "loss": 0.2776,
+      "step": 1940
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.27867947998177e-06,
+      "loss": 0.2838,
+      "step": 1941
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.26973484094433e-06,
+      "loss": 0.29,
+      "step": 1942
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.260790789261016e-06,
+      "loss": 0.2727,
+      "step": 1943
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.251847332125547e-06,
+      "loss": 0.2814,
+      "step": 1944
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.242904476731173e-06,
+      "loss": 0.2634,
+      "step": 1945
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.233962230270655e-06,
+      "loss": 0.2804,
+      "step": 1946
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.225020599936262e-06,
+      "loss": 0.2936,
+      "step": 1947
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.21607959291977e-06,
+      "loss": 0.282,
+      "step": 1948
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.207139216412455e-06,
+      "loss": 0.2729,
+      "step": 1949
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.198199477605084e-06,
+      "loss": 0.2805,
+      "step": 1950
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.189260383687911e-06,
+      "loss": 0.2735,
+      "step": 1951
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.180321941850674e-06,
+      "loss": 0.2921,
+      "step": 1952
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.171384159282585e-06,
+      "loss": 0.279,
+      "step": 1953
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.162447043172321e-06,
+      "loss": 0.2762,
+      "step": 1954
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.153510600708031e-06,
+      "loss": 0.2808,
+      "step": 1955
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.14457483907732e-06,
+      "loss": 0.2941,
+      "step": 1956
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.135639765467238e-06,
+      "loss": 0.275,
+      "step": 1957
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.126705387064291e-06,
+      "loss": 0.2762,
+      "step": 1958
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.117771711054418e-06,
+      "loss": 0.2725,
+      "step": 1959
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.108838744623005e-06,
+      "loss": 0.2869,
+      "step": 1960
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.099906494954855e-06,
+      "loss": 0.278,
+      "step": 1961
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.0909749692342e-06,
+      "loss": 0.2784,
+      "step": 1962
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.082044174644688e-06,
+      "loss": 0.2698,
+      "step": 1963
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.073114118369378e-06,
+      "loss": 0.2794,
+      "step": 1964
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.064184807590741e-06,
+      "loss": 0.2639,
+      "step": 1965
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.055256249490643e-06,
+      "loss": 0.2877,
+      "step": 1966
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.046328451250348e-06,
+      "loss": 0.2784,
+      "step": 1967
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.037401420050502e-06,
+      "loss": 0.2851,
+      "step": 1968
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.028475163071142e-06,
+      "loss": 0.2791,
+      "step": 1969
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.01954968749168e-06,
+      "loss": 0.2724,
+      "step": 1970
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.010625000490896e-06,
+      "loss": 0.2798,
+      "step": 1971
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.001701109246942e-06,
+      "loss": 0.2783,
+      "step": 1972
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.992778020937319e-06,
+      "loss": 0.2902,
+      "step": 1973
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.983855742738898e-06,
+      "loss": 0.2853,
+      "step": 1974
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.974934281827886e-06,
+      "loss": 0.2902,
+      "step": 1975
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.966013645379838e-06,
+      "loss": 0.2839,
+      "step": 1976
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.957093840569644e-06,
+      "loss": 0.2646,
+      "step": 1977
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.948174874571526e-06,
+      "loss": 0.2787,
+      "step": 1978
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.939256754559029e-06,
+      "loss": 0.2624,
+      "step": 1979
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.930339487705022e-06,
+      "loss": 0.2814,
+      "step": 1980
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.921423081181686e-06,
+      "loss": 0.2825,
+      "step": 1981
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.91250754216051e-06,
+      "loss": 0.2833,
+      "step": 1982
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.90359287781228e-06,
+      "loss": 0.268,
+      "step": 1983
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.894679095307086e-06,
+      "loss": 0.2828,
+      "step": 1984
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.885766201814311e-06,
+      "loss": 0.2622,
+      "step": 1985
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.876854204502609e-06,
+      "loss": 0.2747,
+      "step": 1986
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.867943110539924e-06,
+      "loss": 0.2808,
+      "step": 1987
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.859032927093477e-06,
+      "loss": 0.2736,
+      "step": 1988
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.850123661329745e-06,
+      "loss": 0.2736,
+      "step": 1989
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.841215320414478e-06,
+      "loss": 0.2718,
+      "step": 1990
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.832307911512676e-06,
+      "loss": 0.2703,
+      "step": 1991
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.823401441788589e-06,
+      "loss": 0.2822,
+      "step": 1992
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.814495918405712e-06,
+      "loss": 0.2793,
+      "step": 1993
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.805591348526783e-06,
+      "loss": 0.2814,
+      "step": 1994
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.796687739313771e-06,
+      "loss": 0.2954,
+      "step": 1995
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.787785097927869e-06,
+      "loss": 0.2795,
+      "step": 1996
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.778883431529495e-06,
+      "loss": 0.2813,
+      "step": 1997
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.769982747278282e-06,
+      "loss": 0.2847,
+      "step": 1998
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.76108305233307e-06,
+      "loss": 0.2873,
+      "step": 1999
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.752184353851917e-06,
+      "loss": 0.2688,
+      "step": 2000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.743286658992056e-06,
+      "loss": 0.2834,
+      "step": 2001
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.734389974909929e-06,
+      "loss": 0.2739,
+      "step": 2002
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.725494308761165e-06,
+      "loss": 0.2886,
+      "step": 2003
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.716599667700568e-06,
+      "loss": 0.2821,
+      "step": 2004
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.707706058882122e-06,
+      "loss": 0.2589,
+      "step": 2005
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.698813489458979e-06,
+      "loss": 0.2708,
+      "step": 2006
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.689921966583452e-06,
+      "loss": 0.2839,
+      "step": 2007
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.681031497407022e-06,
+      "loss": 0.2822,
+      "step": 2008
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.672142089080311e-06,
+      "loss": 0.3012,
+      "step": 2009
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.663253748753095e-06,
+      "loss": 0.2946,
+      "step": 2010
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.654366483574288e-06,
+      "loss": 0.2552,
+      "step": 2011
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.645480300691938e-06,
+      "loss": 0.3023,
+      "step": 2012
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.636595207253225e-06,
+      "loss": 0.2498,
+      "step": 2013
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.627711210404462e-06,
+      "loss": 0.2905,
+      "step": 2014
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.618828317291055e-06,
+      "loss": 0.2729,
+      "step": 2015
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.609946535057541e-06,
+      "loss": 0.2726,
+      "step": 2016
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.601065870847569e-06,
+      "loss": 0.2686,
+      "step": 2017
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.592186331803872e-06,
+      "loss": 0.2825,
+      "step": 2018
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.583307925068287e-06,
+      "loss": 0.2988,
+      "step": 2019
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.57443065778174e-06,
+      "loss": 0.2609,
+      "step": 2020
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.565554537084241e-06,
+      "loss": 0.2919,
+      "step": 2021
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.556679570114873e-06,
+      "loss": 0.2841,
+      "step": 2022
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.547805764011799e-06,
+      "loss": 0.2833,
+      "step": 2023
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.538933125912241e-06,
+      "loss": 0.2912,
+      "step": 2024
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.530061662952484e-06,
+      "loss": 0.2859,
+      "step": 2025
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.521191382267871e-06,
+      "loss": 0.2833,
+      "step": 2026
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.512322290992785e-06,
+      "loss": 0.2769,
+      "step": 2027
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.503454396260666e-06,
+      "loss": 0.2558,
+      "step": 2028
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.494587705203985e-06,
+      "loss": 0.2777,
+      "step": 2029
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.485722224954237e-06,
+      "loss": 0.2963,
+      "step": 2030
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.476857962641951e-06,
+      "loss": 0.2843,
+      "step": 2031
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.467994925396681e-06,
+      "loss": 0.2916,
+      "step": 2032
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.459133120346989e-06,
+      "loss": 0.256,
+      "step": 2033
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.450272554620444e-06,
+      "loss": 0.2791,
+      "step": 2034
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.441413235343624e-06,
+      "loss": 0.2676,
+      "step": 2035
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.4325551696421e-06,
+      "loss": 0.2847,
+      "step": 2036
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.423698364640443e-06,
+      "loss": 0.288,
+      "step": 2037
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.414842827462196e-06,
+      "loss": 0.2763,
+      "step": 2038
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.405988565229894e-06,
+      "loss": 0.2691,
+      "step": 2039
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.397135585065044e-06,
+      "loss": 0.2678,
+      "step": 2040
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.388283894088117e-06,
+      "loss": 0.2812,
+      "step": 2041
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.37943349941855e-06,
+      "loss": 0.2638,
+      "step": 2042
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.370584408174747e-06,
+      "loss": 0.2825,
+      "step": 2043
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.361736627474042e-06,
+      "loss": 0.2968,
+      "step": 2044
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.352890164432731e-06,
+      "loss": 0.2705,
+      "step": 2045
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.344045026166052e-06,
+      "loss": 0.2723,
+      "step": 2046
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.335201219788169e-06,
+      "loss": 0.2877,
+      "step": 2047
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.326358752412176e-06,
+      "loss": 0.2746,
+      "step": 2048
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.317517631150096e-06,
+      "loss": 0.2741,
+      "step": 2049
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.308677863112863e-06,
+      "loss": 0.2749,
+      "step": 2050
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.299839455410322e-06,
+      "loss": 0.2943,
+      "step": 2051
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.291002415151232e-06,
+      "loss": 0.2763,
+      "step": 2052
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.282166749443246e-06,
+      "loss": 0.277,
+      "step": 2053
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.27333246539291e-06,
+      "loss": 0.2598,
+      "step": 2054
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.264499570105663e-06,
+      "loss": 0.2866,
+      "step": 2055
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.255668070685823e-06,
+      "loss": 0.2574,
+      "step": 2056
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.24683797423659e-06,
+      "loss": 0.272,
+      "step": 2057
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.238009287860037e-06,
+      "loss": 0.2902,
+      "step": 2058
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.229182018657086e-06,
+      "loss": 0.2698,
+      "step": 2059
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.220356173727545e-06,
+      "loss": 0.2765,
+      "step": 2060
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.211531760170058e-06,
+      "loss": 0.2745,
+      "step": 2061
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.202708785082122e-06,
+      "loss": 0.2856,
+      "step": 2062
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.19388725556008e-06,
+      "loss": 0.2838,
+      "step": 2063
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.185067178699112e-06,
+      "loss": 0.2856,
+      "step": 2064
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.176248561593225e-06,
+      "loss": 0.2762,
+      "step": 2065
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.167431411335258e-06,
+      "loss": 0.2667,
+      "step": 2066
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.158615735016866e-06,
+      "loss": 0.2919,
+      "step": 2067
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.149801539728522e-06,
+      "loss": 0.2896,
+      "step": 2068
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.140988832559501e-06,
+      "loss": 0.28,
+      "step": 2069
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.13217762059789e-06,
+      "loss": 0.2763,
+      "step": 2070
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.123367910930569e-06,
+      "loss": 0.2666,
+      "step": 2071
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.114559710643209e-06,
+      "loss": 0.2739,
+      "step": 2072
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.105753026820271e-06,
+      "loss": 0.2817,
+      "step": 2073
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.096947866544984e-06,
+      "loss": 0.2846,
+      "step": 2074
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.08814423689937e-06,
+      "loss": 0.3018,
+      "step": 2075
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.079342144964206e-06,
+      "loss": 0.2859,
+      "step": 2076
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.07054159781904e-06,
+      "loss": 0.2755,
+      "step": 2077
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.06174260254217e-06,
+      "loss": 0.2617,
+      "step": 2078
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.052945166210655e-06,
+      "loss": 0.2753,
+      "step": 2079
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.044149295900291e-06,
+      "loss": 0.2574,
+      "step": 2080
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.035354998685625e-06,
+      "loss": 0.2705,
+      "step": 2081
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.026562281639929e-06,
+      "loss": 0.2622,
+      "step": 2082
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.017771151835205e-06,
+      "loss": 0.2793,
+      "step": 2083
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.008981616342184e-06,
+      "loss": 0.2601,
+      "step": 2084
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.00019368223031e-06,
+      "loss": 0.2853,
+      "step": 2085
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.991407356567747e-06,
+      "loss": 0.2779,
+      "step": 2086
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.982622646421355e-06,
+      "loss": 0.2701,
+      "step": 2087
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.973839558856694e-06,
+      "loss": 0.2705,
+      "step": 2088
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.965058100938027e-06,
+      "loss": 0.2766,
+      "step": 2089
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.956278279728306e-06,
+      "loss": 0.2609,
+      "step": 2090
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.94750010228916e-06,
+      "loss": 0.2806,
+      "step": 2091
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.9387235756809e-06,
+      "loss": 0.2625,
+      "step": 2092
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.929948706962508e-06,
+      "loss": 0.2815,
+      "step": 2093
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.921175503191636e-06,
+      "loss": 0.2714,
+      "step": 2094
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.912403971424593e-06,
+      "loss": 0.2586,
+      "step": 2095
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.903634118716343e-06,
+      "loss": 0.2683,
+      "step": 2096
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.894865952120504e-06,
+      "loss": 0.2851,
+      "step": 2097
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.886099478689328e-06,
+      "loss": 0.283,
+      "step": 2098
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.877334705473718e-06,
+      "loss": 0.2655,
+      "step": 2099
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.868571639523201e-06,
+      "loss": 0.2776,
+      "step": 2100
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.859810287885936e-06,
+      "loss": 0.2626,
+      "step": 2101
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.8510506576087e-06,
+      "loss": 0.2773,
+      "step": 2102
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.84229275573688e-06,
+      "loss": 0.2892,
+      "step": 2103
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.833536589314485e-06,
+      "loss": 0.2531,
+      "step": 2104
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.824782165384119e-06,
+      "loss": 0.2534,
+      "step": 2105
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.816029490986988e-06,
+      "loss": 0.2765,
+      "step": 2106
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.807278573162891e-06,
+      "loss": 0.2791,
+      "step": 2107
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.79852941895021e-06,
+      "loss": 0.2663,
+      "step": 2108
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.789782035385918e-06,
+      "loss": 0.2703,
+      "step": 2109
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.781036429505555e-06,
+      "loss": 0.2814,
+      "step": 2110
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.772292608343233e-06,
+      "loss": 0.2813,
+      "step": 2111
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.76355057893163e-06,
+      "loss": 0.2695,
+      "step": 2112
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.754810348301984e-06,
+      "loss": 0.2927,
+      "step": 2113
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.74607192348408e-06,
+      "loss": 0.2654,
+      "step": 2114
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.737335311506262e-06,
+      "loss": 0.2941,
+      "step": 2115
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.72860051939541e-06,
+      "loss": 0.2598,
+      "step": 2116
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.71986755417693e-06,
+      "loss": 0.2805,
+      "step": 2117
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.711136422874775e-06,
+      "loss": 0.2503,
+      "step": 2118
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.702407132511415e-06,
+      "loss": 0.2645,
+      "step": 2119
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.69367969010784e-06,
+      "loss": 0.2701,
+      "step": 2120
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.684954102683555e-06,
+      "loss": 0.2788,
+      "step": 2121
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.676230377256569e-06,
+      "loss": 0.2841,
+      "step": 2122
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.667508520843401e-06,
+      "loss": 0.2776,
+      "step": 2123
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.658788540459063e-06,
+      "loss": 0.2614,
+      "step": 2124
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.650070443117051e-06,
+      "loss": 0.2717,
+      "step": 2125
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.641354235829358e-06,
+      "loss": 0.2623,
+      "step": 2126
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.632639925606451e-06,
+      "loss": 0.2785,
+      "step": 2127
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.623927519457269e-06,
+      "loss": 0.2577,
+      "step": 2128
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.615217024389228e-06,
+      "loss": 0.278,
+      "step": 2129
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.606508447408196e-06,
+      "loss": 0.2916,
+      "step": 2130
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.597801795518508e-06,
+      "loss": 0.2793,
+      "step": 2131
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.589097075722942e-06,
+      "loss": 0.2679,
+      "step": 2132
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.580394295022728e-06,
+      "loss": 0.25,
+      "step": 2133
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.571693460417534e-06,
+      "loss": 0.2678,
+      "step": 2134
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.5629945789054645e-06,
+      "loss": 0.2698,
+      "step": 2135
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.554297657483053e-06,
+      "loss": 0.2638,
+      "step": 2136
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.5456027031452505e-06,
+      "loss": 0.2578,
+      "step": 2137
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.536909722885437e-06,
+      "loss": 0.2727,
+      "step": 2138
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.528218723695397e-06,
+      "loss": 0.2534,
+      "step": 2139
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.519529712565321e-06,
+      "loss": 0.2842,
+      "step": 2140
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.510842696483804e-06,
+      "loss": 0.2737,
+      "step": 2141
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.502157682437835e-06,
+      "loss": 0.2642,
+      "step": 2142
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.493474677412795e-06,
+      "loss": 0.2548,
+      "step": 2143
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.484793688392445e-06,
+      "loss": 0.2435,
+      "step": 2144
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.476114722358924e-06,
+      "loss": 0.2843,
+      "step": 2145
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.4674377862927536e-06,
+      "loss": 0.2655,
+      "step": 2146
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.458762887172809e-06,
+      "loss": 0.2753,
+      "step": 2147
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.450090031976336e-06,
+      "loss": 0.2693,
+      "step": 2148
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.441419227678933e-06,
+      "loss": 0.2603,
+      "step": 2149
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.432750481254552e-06,
+      "loss": 0.2672,
+      "step": 2150
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.424083799675483e-06,
+      "loss": 0.2676,
+      "step": 2151
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.4154191899123665e-06,
+      "loss": 0.2419,
+      "step": 2152
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.4067566589341665e-06,
+      "loss": 0.2485,
+      "step": 2153
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.398096213708179e-06,
+      "loss": 0.2653,
+      "step": 2154
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.389437861200024e-06,
+      "loss": 0.2602,
+      "step": 2155
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.380781608373635e-06,
+      "loss": 0.2745,
+      "step": 2156
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.372127462191254e-06,
+      "loss": 0.2796,
+      "step": 2157
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.363475429613441e-06,
+      "loss": 0.2718,
+      "step": 2158
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.354825517599043e-06,
+      "loss": 0.2699,
+      "step": 2159
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.34617773310521e-06,
+      "loss": 0.264,
+      "step": 2160
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.337532083087371e-06,
+      "loss": 0.2768,
+      "step": 2161
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.328888574499247e-06,
+      "loss": 0.2772,
+      "step": 2162
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.320247214292835e-06,
+      "loss": 0.2589,
+      "step": 2163
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.311608009418403e-06,
+      "loss": 0.2775,
+      "step": 2164
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.302970966824481e-06,
+      "loss": 0.2734,
+      "step": 2165
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.29433609345787e-06,
+      "loss": 0.2705,
+      "step": 2166
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.285703396263619e-06,
+      "loss": 0.2619,
+      "step": 2167
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.277072882185027e-06,
+      "loss": 0.2654,
+      "step": 2168
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.268444558163641e-06,
+      "loss": 0.2692,
+      "step": 2169
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.259818431139242e-06,
+      "loss": 0.2577,
+      "step": 2170
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.251194508049844e-06,
+      "loss": 0.2625,
+      "step": 2171
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.242572795831693e-06,
+      "loss": 0.2531,
+      "step": 2172
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.2339533014192555e-06,
+      "loss": 0.2675,
+      "step": 2173
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.2253360317452095e-06,
+      "loss": 0.2704,
+      "step": 2174
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.216720993740452e-06,
+      "loss": 0.2777,
+      "step": 2175
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.208108194334071e-06,
+      "loss": 0.2599,
+      "step": 2176
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.19949764045337e-06,
+      "loss": 0.2701,
+      "step": 2177
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.190889339023838e-06,
+      "loss": 0.2559,
+      "step": 2178
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.1822832969691524e-06,
+      "loss": 0.2549,
+      "step": 2179
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.173679521211173e-06,
+      "loss": 0.2753,
+      "step": 2180
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.1650780186699445e-06,
+      "loss": 0.2477,
+      "step": 2181
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.156478796263673e-06,
+      "loss": 0.2666,
+      "step": 2182
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.147881860908734e-06,
+      "loss": 0.2822,
+      "step": 2183
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.139287219519668e-06,
+      "loss": 0.2545,
+      "step": 2184
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.130694879009163e-06,
+      "loss": 0.2771,
+      "step": 2185
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.122104846288065e-06,
+      "loss": 0.2676,
+      "step": 2186
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.113517128265356e-06,
+      "loss": 0.2642,
+      "step": 2187
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.104931731848159e-06,
+      "loss": 0.2651,
+      "step": 2188
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.096348663941736e-06,
+      "loss": 0.2745,
+      "step": 2189
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.087767931449461e-06,
+      "loss": 0.2593,
+      "step": 2190
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.079189541272844e-06,
+      "loss": 0.2776,
+      "step": 2191
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.070613500311507e-06,
+      "loss": 0.2656,
+      "step": 2192
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.062039815463178e-06,
+      "loss": 0.2569,
+      "step": 2193
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.053468493623695e-06,
+      "loss": 0.2649,
+      "step": 2194
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.044899541686996e-06,
+      "loss": 0.2495,
+      "step": 2195
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.03633296654511e-06,
+      "loss": 0.2762,
+      "step": 2196
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.027768775088155e-06,
+      "loss": 0.2502,
+      "step": 2197
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.01920697420433e-06,
+      "loss": 0.2578,
+      "step": 2198
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.0106475707799156e-06,
+      "loss": 0.2507,
+      "step": 2199
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.002090571699258e-06,
+      "loss": 0.2556,
+      "step": 2200
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.9935359838447805e-06,
+      "loss": 0.2751,
+      "step": 2201
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.984983814096954e-06,
+      "loss": 0.2708,
+      "step": 2202
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.9764340693343125e-06,
+      "loss": 0.2685,
+      "step": 2203
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.967886756433437e-06,
+      "loss": 0.2734,
+      "step": 2204
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.959341882268953e-06,
+      "loss": 0.2458,
+      "step": 2205
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.950799453713524e-06,
+      "loss": 0.2672,
+      "step": 2206
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.942259477637847e-06,
+      "loss": 0.2547,
+      "step": 2207
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.933721960910648e-06,
+      "loss": 0.2607,
+      "step": 2208
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.925186910398669e-06,
+      "loss": 0.263,
+      "step": 2209
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.916654332966679e-06,
+      "loss": 0.2717,
+      "step": 2210
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.908124235477448e-06,
+      "loss": 0.2466,
+      "step": 2211
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.899596624791756e-06,
+      "loss": 0.2931,
+      "step": 2212
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.8910715077683835e-06,
+      "loss": 0.2818,
+      "step": 2213
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.882548891264098e-06,
+      "loss": 0.2615,
+      "step": 2214
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.874028782133668e-06,
+      "loss": 0.2682,
+      "step": 2215
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.865511187229839e-06,
+      "loss": 0.2419,
+      "step": 2216
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.85699611340333e-06,
+      "loss": 0.2902,
+      "step": 2217
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.848483567502839e-06,
+      "loss": 0.255,
+      "step": 2218
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.8399735563750306e-06,
+      "loss": 0.2457,
+      "step": 2219
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.831466086864523e-06,
+      "loss": 0.2409,
+      "step": 2220
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.8229611658139e-06,
+      "loss": 0.2425,
+      "step": 2221
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.814458800063686e-06,
+      "loss": 0.2595,
+      "step": 2222
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.8059589964523584e-06,
+      "loss": 0.2611,
+      "step": 2223
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.797461761816334e-06,
+      "loss": 0.2649,
+      "step": 2224
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.788967102989956e-06,
+      "loss": 0.2769,
+      "step": 2225
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.780475026805504e-06,
+      "loss": 0.2801,
+      "step": 2226
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.771985540093172e-06,
+      "loss": 0.2625,
+      "step": 2227
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.763498649681077e-06,
+      "loss": 0.2528,
+      "step": 2228
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.755014362395249e-06,
+      "loss": 0.2665,
+      "step": 2229
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.746532685059619e-06,
+      "loss": 0.2659,
+      "step": 2230
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.738053624496024e-06,
+      "loss": 0.2526,
+      "step": 2231
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.7295771875241925e-06,
+      "loss": 0.244,
+      "step": 2232
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.7211033809617465e-06,
+      "loss": 0.2582,
+      "step": 2233
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.712632211624187e-06,
+      "loss": 0.2548,
+      "step": 2234
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.704163686324898e-06,
+      "loss": 0.2491,
+      "step": 2235
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.695697811875137e-06,
+      "loss": 0.2639,
+      "step": 2236
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.687234595084024e-06,
+      "loss": 0.2462,
+      "step": 2237
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.678774042758551e-06,
+      "loss": 0.2674,
+      "step": 2238
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.6703161617035625e-06,
+      "loss": 0.27,
+      "step": 2239
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.66186095872175e-06,
+      "loss": 0.2692,
+      "step": 2240
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.653408440613656e-06,
+      "loss": 0.249,
+      "step": 2241
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.6449586141776625e-06,
+      "loss": 0.2509,
+      "step": 2242
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.636511486209985e-06,
+      "loss": 0.251,
+      "step": 2243
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.628067063504672e-06,
+      "loss": 0.2444,
+      "step": 2244
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.619625352853594e-06,
+      "loss": 0.2591,
+      "step": 2245
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.61118636104644e-06,
+      "loss": 0.2451,
+      "step": 2246
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.602750094870713e-06,
+      "loss": 0.2534,
+      "step": 2247
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.5943165611117244e-06,
+      "loss": 0.2649,
+      "step": 2248
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.5858857665525824e-06,
+      "loss": 0.2504,
+      "step": 2249
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.5774577179742e-06,
+      "loss": 0.2677,
+      "step": 2250
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.5690324221552774e-06,
+      "loss": 0.2511,
+      "step": 2251
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.560609885872298e-06,
+      "loss": 0.2523,
+      "step": 2252
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.552190115899535e-06,
+      "loss": 0.2535,
+      "step": 2253
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.543773119009029e-06,
+      "loss": 0.2553,
+      "step": 2254
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.535358901970592e-06,
+      "loss": 0.2395,
+      "step": 2255
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.526947471551799e-06,
+      "loss": 0.2565,
+      "step": 2256
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.518538834517982e-06,
+      "loss": 0.2559,
+      "step": 2257
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.510132997632237e-06,
+      "loss": 0.2565,
+      "step": 2258
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.501729967655395e-06,
+      "loss": 0.2577,
+      "step": 2259
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.4933297513460335e-06,
+      "loss": 0.2492,
+      "step": 2260
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.484932355460469e-06,
+      "loss": 0.2501,
+      "step": 2261
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.4765377867527525e-06,
+      "loss": 0.2663,
+      "step": 2262
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.46814605197465e-06,
+      "loss": 0.2507,
+      "step": 2263
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.459757157875658e-06,
+      "loss": 0.2487,
+      "step": 2264
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.451371111202986e-06,
+      "loss": 0.253,
+      "step": 2265
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.442987918701549e-06,
+      "loss": 0.2476,
+      "step": 2266
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.434607587113977e-06,
+      "loss": 0.2612,
+      "step": 2267
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.426230123180587e-06,
+      "loss": 0.2484,
+      "step": 2268
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.417855533639397e-06,
+      "loss": 0.2559,
+      "step": 2269
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.409483825226108e-06,
+      "loss": 0.2339,
+      "step": 2270
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.401115004674108e-06,
+      "loss": 0.2593,
+      "step": 2271
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.392749078714459e-06,
+      "loss": 0.2621,
+      "step": 2272
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.3843860540759e-06,
+      "loss": 0.2388,
+      "step": 2273
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.37602593748483e-06,
+      "loss": 0.2508,
+      "step": 2274
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.367668735665314e-06,
+      "loss": 0.2465,
+      "step": 2275
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.35931445533907e-06,
+      "loss": 0.2564,
+      "step": 2276
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.350963103225467e-06,
+      "loss": 0.2567,
+      "step": 2277
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.3426146860415175e-06,
+      "loss": 0.2474,
+      "step": 2278
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.334269210501876e-06,
+      "loss": 0.2435,
+      "step": 2279
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.325926683318825e-06,
+      "loss": 0.2668,
+      "step": 2280
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.317587111202288e-06,
+      "loss": 0.2657,
+      "step": 2281
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.309250500859801e-06,
+      "loss": 0.2648,
+      "step": 2282
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.30091685899652e-06,
+      "loss": 0.2451,
+      "step": 2283
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.2925861923152136e-06,
+      "loss": 0.2627,
+      "step": 2284
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.284258507516261e-06,
+      "loss": 0.2546,
+      "step": 2285
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.275933811297635e-06,
+      "loss": 0.2629,
+      "step": 2286
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.267612110354917e-06,
+      "loss": 0.2467,
+      "step": 2287
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.2592934113812684e-06,
+      "loss": 0.2691,
+      "step": 2288
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.250977721067439e-06,
+      "loss": 0.2604,
+      "step": 2289
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.242665046101763e-06,
+      "loss": 0.2608,
+      "step": 2290
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.234355393170142e-06,
+      "loss": 0.2495,
+      "step": 2291
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.226048768956058e-06,
+      "loss": 0.2687,
+      "step": 2292
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.217745180140541e-06,
+      "loss": 0.2602,
+      "step": 2293
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.209444633402193e-06,
+      "loss": 0.2605,
+      "step": 2294
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.201147135417162e-06,
+      "loss": 0.2529,
+      "step": 2295
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.192852692859152e-06,
+      "loss": 0.2407,
+      "step": 2296
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.1845613123994e-06,
+      "loss": 0.2537,
+      "step": 2297
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.176273000706687e-06,
+      "loss": 0.2457,
+      "step": 2298
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.167987764447321e-06,
+      "loss": 0.2445,
+      "step": 2299
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.159705610285138e-06,
+      "loss": 0.2565,
+      "step": 2300
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.151426544881498e-06,
+      "loss": 0.2698,
+      "step": 2301
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.143150574895276e-06,
+      "loss": 0.2438,
+      "step": 2302
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.134877706982851e-06,
+      "loss": 0.2688,
+      "step": 2303
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.126607947798115e-06,
+      "loss": 0.2545,
+      "step": 2304
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.118341303992454e-06,
+      "loss": 0.252,
+      "step": 2305
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.110077782214756e-06,
+      "loss": 0.2621,
+      "step": 2306
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.1018173891113866e-06,
+      "loss": 0.2558,
+      "step": 2307
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.093560131326201e-06,
+      "loss": 0.2776,
+      "step": 2308
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.085306015500532e-06,
+      "loss": 0.2505,
+      "step": 2309
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.077055048273193e-06,
+      "loss": 0.2561,
+      "step": 2310
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.0688072362804516e-06,
+      "loss": 0.2429,
+      "step": 2311
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.060562586156046e-06,
+      "loss": 0.2536,
+      "step": 2312
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.052321104531169e-06,
+      "loss": 0.2433,
+      "step": 2313
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.044082798034465e-06,
+      "loss": 0.2421,
+      "step": 2314
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.035847673292023e-06,
+      "loss": 0.2676,
+      "step": 2315
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.027615736927379e-06,
+      "loss": 0.264,
+      "step": 2316
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.019386995561498e-06,
+      "loss": 0.2358,
+      "step": 2317
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.011161455812779e-06,
+      "loss": 0.2567,
+      "step": 2318
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.0029391242970415e-06,
+      "loss": 0.2578,
+      "step": 2319
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.994720007627528e-06,
+      "loss": 0.2421,
+      "step": 2320
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.986504112414904e-06,
+      "loss": 0.2484,
+      "step": 2321
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.978291445267221e-06,
+      "loss": 0.2545,
+      "step": 2322
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.9700820127899524e-06,
+      "loss": 0.2418,
+      "step": 2323
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.961875821585971e-06,
+      "loss": 0.2595,
+      "step": 2324
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.953672878255534e-06,
+      "loss": 0.2513,
+      "step": 2325
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.945473189396289e-06,
+      "loss": 0.249,
+      "step": 2326
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.937276761603266e-06,
+      "loss": 0.2246,
+      "step": 2327
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.929083601468875e-06,
+      "loss": 0.2597,
+      "step": 2328
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.920893715582891e-06,
+      "loss": 0.2743,
+      "step": 2329
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.912707110532466e-06,
+      "loss": 0.2472,
+      "step": 2330
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.904523792902105e-06,
+      "loss": 0.2541,
+      "step": 2331
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.89634376927367e-06,
+      "loss": 0.2574,
+      "step": 2332
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.888167046226376e-06,
+      "loss": 0.2429,
+      "step": 2333
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.879993630336782e-06,
+      "loss": 0.2463,
+      "step": 2334
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.871823528178788e-06,
+      "loss": 0.2507,
+      "step": 2335
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.863656746323625e-06,
+      "loss": 0.2433,
+      "step": 2336
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.855493291339858e-06,
+      "loss": 0.2552,
+      "step": 2337
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.847333169793374e-06,
+      "loss": 0.2494,
+      "step": 2338
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.8391763882473805e-06,
+      "loss": 0.2633,
+      "step": 2339
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.831022953262401e-06,
+      "loss": 0.2522,
+      "step": 2340
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.822872871396255e-06,
+      "loss": 0.2594,
+      "step": 2341
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.814726149204082e-06,
+      "loss": 0.2642,
+      "step": 2342
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.806582793238309e-06,
+      "loss": 0.2406,
+      "step": 2343
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.7984428100486564e-06,
+      "loss": 0.2529,
+      "step": 2344
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.790306206182137e-06,
+      "loss": 0.22,
+      "step": 2345
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.782172988183039e-06,
+      "loss": 0.2471,
+      "step": 2346
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.774043162592931e-06,
+      "loss": 0.2541,
+      "step": 2347
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.765916735950654e-06,
+      "loss": 0.2602,
+      "step": 2348
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.75779371479231e-06,
+      "loss": 0.2563,
+      "step": 2349
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.749674105651268e-06,
+      "loss": 0.2482,
+      "step": 2350
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.7415579150581515e-06,
+      "loss": 0.2629,
+      "step": 2351
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.733445149540831e-06,
+      "loss": 0.2527,
+      "step": 2352
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.7253358156244234e-06,
+      "loss": 0.2497,
+      "step": 2353
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.717229919831289e-06,
+      "loss": 0.2572,
+      "step": 2354
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.709127468681021e-06,
+      "loss": 0.2563,
+      "step": 2355
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.701028468690438e-06,
+      "loss": 0.2377,
+      "step": 2356
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.69293292637359e-06,
+      "loss": 0.2438,
+      "step": 2357
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.684840848241739e-06,
+      "loss": 0.229,
+      "step": 2358
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.676752240803368e-06,
+      "loss": 0.2542,
+      "step": 2359
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.668667110564161e-06,
+      "loss": 0.2491,
+      "step": 2360
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.660585464027006e-06,
+      "loss": 0.2368,
+      "step": 2361
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.652507307692001e-06,
+      "loss": 0.2358,
+      "step": 2362
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.644432648056421e-06,
+      "loss": 0.2572,
+      "step": 2363
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.636361491614738e-06,
+      "loss": 0.2547,
+      "step": 2364
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.628293844858605e-06,
+      "loss": 0.2621,
+      "step": 2365
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.620229714276846e-06,
+      "loss": 0.2426,
+      "step": 2366
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.612169106355463e-06,
+      "loss": 0.267,
+      "step": 2367
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.6041120275776264e-06,
+      "loss": 0.2458,
+      "step": 2368
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.5960584844236565e-06,
+      "loss": 0.2222,
+      "step": 2369
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.5880084833710505e-06,
+      "loss": 0.2487,
+      "step": 2370
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.579962030894439e-06,
+      "loss": 0.2525,
+      "step": 2371
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.571919133465605e-06,
+      "loss": 0.2274,
+      "step": 2372
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.5638797975534705e-06,
+      "loss": 0.2566,
+      "step": 2373
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.555844029624095e-06,
+      "loss": 0.2525,
+      "step": 2374
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.547811836140667e-06,
+      "loss": 0.2461,
+      "step": 2375
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.539783223563499e-06,
+      "loss": 0.2626,
+      "step": 2376
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.531758198350026e-06,
+      "loss": 0.2646,
+      "step": 2377
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.523736766954792e-06,
+      "loss": 0.2419,
+      "step": 2378
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.515718935829461e-06,
+      "loss": 0.2433,
+      "step": 2379
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.507704711422791e-06,
+      "loss": 0.2399,
+      "step": 2380
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.4996941001806436e-06,
+      "loss": 0.2486,
+      "step": 2381
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.491687108545974e-06,
+      "loss": 0.2558,
+      "step": 2382
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.483683742958827e-06,
+      "loss": 0.2605,
+      "step": 2383
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.475684009856328e-06,
+      "loss": 0.2335,
+      "step": 2384
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.4676879156726836e-06,
+      "loss": 0.265,
+      "step": 2385
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.459695466839174e-06,
+      "loss": 0.2387,
+      "step": 2386
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.4517066697841435e-06,
+      "loss": 0.2417,
+      "step": 2387
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.4437215309330034e-06,
+      "loss": 0.2413,
+      "step": 2388
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.435740056708219e-06,
+      "loss": 0.2674,
+      "step": 2389
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.4277622535293136e-06,
+      "loss": 0.2451,
+      "step": 2390
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.419788127812854e-06,
+      "loss": 0.2365,
+      "step": 2391
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.411817685972449e-06,
+      "loss": 0.2484,
+      "step": 2392
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.403850934418746e-06,
+      "loss": 0.2428,
+      "step": 2393
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.395887879559426e-06,
+      "loss": 0.2657,
+      "step": 2394
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.387928527799188e-06,
+      "loss": 0.2601,
+      "step": 2395
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.3799728855397615e-06,
+      "loss": 0.2274,
+      "step": 2396
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.37202095917989e-06,
+      "loss": 0.25,
+      "step": 2397
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.364072755115325e-06,
+      "loss": 0.2435,
+      "step": 2398
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.356128279738835e-06,
+      "loss": 0.2295,
+      "step": 2399
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.348187539440175e-06,
+      "loss": 0.2363,
+      "step": 2400
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.340250540606106e-06,
+      "loss": 0.2354,
+      "step": 2401
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.332317289620376e-06,
+      "loss": 0.232,
+      "step": 2402
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.324387792863719e-06,
+      "loss": 0.2758,
+      "step": 2403
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.316462056713848e-06,
+      "loss": 0.2478,
+      "step": 2404
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.308540087545453e-06,
+      "loss": 0.2386,
+      "step": 2405
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.3006218917301945e-06,
+      "loss": 0.2451,
+      "step": 2406
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.292707475636698e-06,
+      "loss": 0.2557,
+      "step": 2407
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.284796845630547e-06,
+      "loss": 0.2207,
+      "step": 2408
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.276890008074283e-06,
+      "loss": 0.2378,
+      "step": 2409
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.268986969327393e-06,
+      "loss": 0.2495,
+      "step": 2410
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.261087735746313e-06,
+      "loss": 0.2144,
+      "step": 2411
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.253192313684419e-06,
+      "loss": 0.1863,
+      "step": 2412
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.245300709492014e-06,
+      "loss": 0.1788,
+      "step": 2413
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.237412929516339e-06,
+      "loss": 0.1877,
+      "step": 2414
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.229528980101554e-06,
+      "loss": 0.1822,
+      "step": 2415
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.22164886758874e-06,
+      "loss": 0.1814,
+      "step": 2416
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.213772598315892e-06,
+      "loss": 0.1866,
+      "step": 2417
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.205900178617909e-06,
+      "loss": 0.1721,
+      "step": 2418
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.1980316148266065e-06,
+      "loss": 0.1753,
+      "step": 2419
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.190166913270684e-06,
+      "loss": 0.1746,
+      "step": 2420
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.182306080275744e-06,
+      "loss": 0.1797,
+      "step": 2421
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.174449122164274e-06,
+      "loss": 0.1809,
+      "step": 2422
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.166596045255647e-06,
+      "loss": 0.1825,
+      "step": 2423
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.158746855866108e-06,
+      "loss": 0.1752,
+      "step": 2424
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.150901560308782e-06,
+      "loss": 0.1739,
+      "step": 2425
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.14306016489366e-06,
+      "loss": 0.1708,
+      "step": 2426
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.135222675927594e-06,
+      "loss": 0.1813,
+      "step": 2427
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.127389099714302e-06,
+      "loss": 0.1708,
+      "step": 2428
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.119559442554347e-06,
+      "loss": 0.1932,
+      "step": 2429
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.111733710745142e-06,
+      "loss": 0.1771,
+      "step": 2430
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.103911910580942e-06,
+      "loss": 0.1819,
+      "step": 2431
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.096094048352842e-06,
+      "loss": 0.174,
+      "step": 2432
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.0882801303487664e-06,
+      "loss": 0.1784,
+      "step": 2433
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.080470162853473e-06,
+      "loss": 0.1878,
+      "step": 2434
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.072664152148534e-06,
+      "loss": 0.1859,
+      "step": 2435
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.064862104512347e-06,
+      "loss": 0.179,
+      "step": 2436
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.057064026220118e-06,
+      "loss": 0.1792,
+      "step": 2437
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.04926992354386e-06,
+      "loss": 0.1768,
+      "step": 2438
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.0414798027523916e-06,
+      "loss": 0.1836,
+      "step": 2439
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.033693670111324e-06,
+      "loss": 0.1841,
+      "step": 2440
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.025911531883067e-06,
+      "loss": 0.1716,
+      "step": 2441
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.018133394326811e-06,
+      "loss": 0.1736,
+      "step": 2442
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.0103592636985325e-06,
+      "loss": 0.1783,
+      "step": 2443
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.0025891462509864e-06,
+      "loss": 0.1784,
+      "step": 2444
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.994823048233695e-06,
+      "loss": 0.1723,
+      "step": 2445
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.9870609758929545e-06,
+      "loss": 0.1795,
+      "step": 2446
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.979302935471813e-06,
+      "loss": 0.1709,
+      "step": 2447
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.97154893321009e-06,
+      "loss": 0.1794,
+      "step": 2448
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.963798975344345e-06,
+      "loss": 0.1669,
+      "step": 2449
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.9560530681078886e-06,
+      "loss": 0.178,
+      "step": 2450
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.948311217730776e-06,
+      "loss": 0.1807,
+      "step": 2451
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.940573430439799e-06,
+      "loss": 0.1891,
+      "step": 2452
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.932839712458472e-06,
+      "loss": 0.1815,
+      "step": 2453
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.92511007000705e-06,
+      "loss": 0.1877,
+      "step": 2454
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.917384509302498e-06,
+      "loss": 0.1819,
+      "step": 2455
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.909663036558511e-06,
+      "loss": 0.1775,
+      "step": 2456
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.901945657985486e-06,
+      "loss": 0.1799,
+      "step": 2457
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.894232379790534e-06,
+      "loss": 0.1675,
+      "step": 2458
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.8865232081774585e-06,
+      "loss": 0.1698,
+      "step": 2459
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.878818149346769e-06,
+      "loss": 0.1723,
+      "step": 2460
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.871117209495664e-06,
+      "loss": 0.1833,
+      "step": 2461
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.863420394818027e-06,
+      "loss": 0.1684,
+      "step": 2462
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.855727711504428e-06,
+      "loss": 0.183,
+      "step": 2463
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.84803916574211e-06,
+      "loss": 0.1815,
+      "step": 2464
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.840354763714991e-06,
+      "loss": 0.1842,
+      "step": 2465
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.832674511603656e-06,
+      "loss": 0.19,
+      "step": 2466
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.824998415585348e-06,
+      "loss": 0.1823,
+      "step": 2467
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.817326481833974e-06,
+      "loss": 0.1813,
+      "step": 2468
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.809658716520088e-06,
+      "loss": 0.1757,
+      "step": 2469
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.801995125810893e-06,
+      "loss": 0.1771,
+      "step": 2470
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.794335715870235e-06,
+      "loss": 0.1824,
+      "step": 2471
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.786680492858598e-06,
+      "loss": 0.1698,
+      "step": 2472
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.7790294629330956e-06,
+      "loss": 0.1752,
+      "step": 2473
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.771382632247471e-06,
+      "loss": 0.1818,
+      "step": 2474
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.7637400069520865e-06,
+      "loss": 0.1773,
+      "step": 2475
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.756101593193931e-06,
+      "loss": 0.178,
+      "step": 2476
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.7484673971166e-06,
+      "loss": 0.1799,
+      "step": 2477
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.740837424860291e-06,
+      "loss": 0.1732,
+      "step": 2478
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.7332116825618154e-06,
+      "loss": 0.1749,
+      "step": 2479
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.725590176354576e-06,
+      "loss": 0.1795,
+      "step": 2480
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.71797291236857e-06,
+      "loss": 0.1732,
+      "step": 2481
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.710359896730379e-06,
+      "loss": 0.1674,
+      "step": 2482
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.702751135563173e-06,
+      "loss": 0.1902,
+      "step": 2483
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.695146634986692e-06,
+      "loss": 0.1716,
+      "step": 2484
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.687546401117266e-06,
+      "loss": 0.1761,
+      "step": 2485
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.679950440067778e-06,
+      "loss": 0.1866,
+      "step": 2486
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.672358757947677e-06,
+      "loss": 0.1797,
+      "step": 2487
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.664771360862973e-06,
+      "loss": 0.1803,
+      "step": 2488
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.65718825491623e-06,
+      "loss": 0.1733,
+      "step": 2489
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.6496094462065575e-06,
+      "loss": 0.1814,
+      "step": 2490
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.642034940829612e-06,
+      "loss": 0.1656,
+      "step": 2491
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.634464744877587e-06,
+      "loss": 0.1772,
+      "step": 2492
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.626898864439211e-06,
+      "loss": 0.1888,
+      "step": 2493
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.619337305599739e-06,
+      "loss": 0.1706,
+      "step": 2494
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.611780074440951e-06,
+      "loss": 0.1652,
+      "step": 2495
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.604227177041156e-06,
+      "loss": 0.1811,
+      "step": 2496
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.596678619475159e-06,
+      "loss": 0.1766,
+      "step": 2497
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.5891344078142865e-06,
+      "loss": 0.1767,
+      "step": 2498
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.581594548126368e-06,
+      "loss": 0.1839,
+      "step": 2499
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.574059046475731e-06,
+      "loss": 0.1767,
+      "step": 2500
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.566527908923199e-06,
+      "loss": 0.1796,
+      "step": 2501
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.559001141526087e-06,
+      "loss": 0.1864,
+      "step": 2502
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.5514787503381895e-06,
+      "loss": 0.1841,
+      "step": 2503
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.543960741409784e-06,
+      "loss": 0.1816,
+      "step": 2504
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.536447120787631e-06,
+      "loss": 0.1728,
+      "step": 2505
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.52893789451495e-06,
+      "loss": 0.1703,
+      "step": 2506
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.521433068631432e-06,
+      "loss": 0.1757,
+      "step": 2507
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.513932649173226e-06,
+      "loss": 0.1742,
+      "step": 2508
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.50643664217294e-06,
+      "loss": 0.1759,
+      "step": 2509
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.498945053659631e-06,
+      "loss": 0.1698,
+      "step": 2510
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.491457889658807e-06,
+      "loss": 0.17,
+      "step": 2511
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.483975156192404e-06,
+      "loss": 0.1693,
+      "step": 2512
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.476496859278805e-06,
+      "loss": 0.1701,
+      "step": 2513
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.46902300493283e-06,
+      "loss": 0.1805,
+      "step": 2514
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.461553599165714e-06,
+      "loss": 0.1877,
+      "step": 2515
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.454088647985122e-06,
+      "loss": 0.1758,
+      "step": 2516
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.446628157395132e-06,
+      "loss": 0.1859,
+      "step": 2517
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.439172133396235e-06,
+      "loss": 0.1729,
+      "step": 2518
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.43172058198533e-06,
+      "loss": 0.1707,
+      "step": 2519
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.424273509155719e-06,
+      "loss": 0.1711,
+      "step": 2520
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.416830920897103e-06,
+      "loss": 0.1825,
+      "step": 2521
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.409392823195572e-06,
+      "loss": 0.1638,
+      "step": 2522
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.401959222033609e-06,
+      "loss": 0.1688,
+      "step": 2523
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.394530123390073e-06,
+      "loss": 0.171,
+      "step": 2524
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.387105533240219e-06,
+      "loss": 0.1847,
+      "step": 2525
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.379685457555653e-06,
+      "loss": 0.1836,
+      "step": 2526
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.3722699023043634e-06,
+      "loss": 0.187,
+      "step": 2527
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.364858873450702e-06,
+      "loss": 0.1723,
+      "step": 2528
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.3574523769553755e-06,
+      "loss": 0.1951,
+      "step": 2529
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.35005041877545e-06,
+      "loss": 0.1719,
+      "step": 2530
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.342653004864338e-06,
+      "loss": 0.1801,
+      "step": 2531
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.335260141171799e-06,
+      "loss": 0.1891,
+      "step": 2532
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.327871833643928e-06,
+      "loss": 0.1814,
+      "step": 2533
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.320488088223168e-06,
+      "loss": 0.1839,
+      "step": 2534
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.3131089108482795e-06,
+      "loss": 0.1721,
+      "step": 2535
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.305734307454354e-06,
+      "loss": 0.1857,
+      "step": 2536
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.2983642839728024e-06,
+      "loss": 0.1811,
+      "step": 2537
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.290998846331355e-06,
+      "loss": 0.1742,
+      "step": 2538
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.283638000454051e-06,
+      "loss": 0.183,
+      "step": 2539
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.2762817522612435e-06,
+      "loss": 0.1692,
+      "step": 2540
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.26893010766957e-06,
+      "loss": 0.1668,
+      "step": 2541
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.261583072591981e-06,
+      "loss": 0.1817,
+      "step": 2542
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.25424065293772e-06,
+      "loss": 0.1671,
+      "step": 2543
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.246902854612313e-06,
+      "loss": 0.1678,
+      "step": 2544
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.239569683517566e-06,
+      "loss": 0.17,
+      "step": 2545
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.23224114555157e-06,
+      "loss": 0.1795,
+      "step": 2546
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2249172466086864e-06,
+      "loss": 0.186,
+      "step": 2547
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2175979925795454e-06,
+      "loss": 0.1718,
+      "step": 2548
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.210283389351042e-06,
+      "loss": 0.1732,
+      "step": 2549
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.202973442806328e-06,
+      "loss": 0.1699,
+      "step": 2550
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.195668158824816e-06,
+      "loss": 0.1663,
+      "step": 2551
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.188367543282161e-06,
+      "loss": 0.1872,
+      "step": 2552
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.181071602050265e-06,
+      "loss": 0.1741,
+      "step": 2553
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.173780340997281e-06,
+      "loss": 0.1755,
+      "step": 2554
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.16649376598758e-06,
+      "loss": 0.1706,
+      "step": 2555
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.159211882881775e-06,
+      "loss": 0.1916,
+      "step": 2556
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.151934697536705e-06,
+      "loss": 0.1869,
+      "step": 2557
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.144662215805426e-06,
+      "loss": 0.1779,
+      "step": 2558
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.137394443537216e-06,
+      "loss": 0.1785,
+      "step": 2559
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.13013138657756e-06,
+      "loss": 0.1794,
+      "step": 2560
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.122873050768156e-06,
+      "loss": 0.1734,
+      "step": 2561
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.115619441946898e-06,
+      "loss": 0.1791,
+      "step": 2562
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.108370565947887e-06,
+      "loss": 0.1714,
+      "step": 2563
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.10112642860141e-06,
+      "loss": 0.1836,
+      "step": 2564
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.093887035733945e-06,
+      "loss": 0.1801,
+      "step": 2565
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.086652393168155e-06,
+      "loss": 0.1818,
+      "step": 2566
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.079422506722881e-06,
+      "loss": 0.1728,
+      "step": 2567
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.072197382213139e-06,
+      "loss": 0.1777,
+      "step": 2568
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.0649770254501176e-06,
+      "loss": 0.1837,
+      "step": 2569
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.057761442241158e-06,
+      "loss": 0.1895,
+      "step": 2570
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.050550638389784e-06,
+      "loss": 0.1791,
+      "step": 2571
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.043344619695658e-06,
+      "loss": 0.1826,
+      "step": 2572
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.036143391954601e-06,
+      "loss": 0.1729,
+      "step": 2573
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.028946960958578e-06,
+      "loss": 0.1723,
+      "step": 2574
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.021755332495695e-06,
+      "loss": 0.1769,
+      "step": 2575
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.0145685123502e-06,
+      "loss": 0.1716,
+      "step": 2576
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.007386506302469e-06,
+      "loss": 0.1865,
+      "step": 2577
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.000209320129008e-06,
+      "loss": 0.1802,
+      "step": 2578
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.993036959602447e-06,
+      "loss": 0.1836,
+      "step": 2579
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.985869430491531e-06,
+      "loss": 0.1807,
+      "step": 2580
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.978706738561126e-06,
+      "loss": 0.188,
+      "step": 2581
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.971548889572197e-06,
+      "loss": 0.1721,
+      "step": 2582
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.96439588928183e-06,
+      "loss": 0.1805,
+      "step": 2583
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.957247743443195e-06,
+      "loss": 0.1827,
+      "step": 2584
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.9501044578055614e-06,
+      "loss": 0.1826,
+      "step": 2585
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.942966038114298e-06,
+      "loss": 0.1694,
+      "step": 2586
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.935832490110849e-06,
+      "loss": 0.1637,
+      "step": 2587
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.928703819532748e-06,
+      "loss": 0.1736,
+      "step": 2588
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.921580032113602e-06,
+      "loss": 0.1772,
+      "step": 2589
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.914461133583088e-06,
+      "loss": 0.1792,
+      "step": 2590
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.9073471296669605e-06,
+      "loss": 0.1789,
+      "step": 2591
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.90023802608703e-06,
+      "loss": 0.1816,
+      "step": 2592
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.893133828561163e-06,
+      "loss": 0.177,
+      "step": 2593
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.886034542803285e-06,
+      "loss": 0.1746,
+      "step": 2594
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.878940174523371e-06,
+      "loss": 0.182,
+      "step": 2595
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.871850729427439e-06,
+      "loss": 0.1734,
+      "step": 2596
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.864766213217546e-06,
+      "loss": 0.1759,
+      "step": 2597
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.85768663159179e-06,
+      "loss": 0.1757,
+      "step": 2598
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.850611990244288e-06,
+      "loss": 0.1737,
+      "step": 2599
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.843542294865203e-06,
+      "loss": 0.1795,
+      "step": 2600
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.8364775511407015e-06,
+      "loss": 0.1754,
+      "step": 2601
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.829417764752978e-06,
+      "loss": 0.1744,
+      "step": 2602
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.822362941380237e-06,
+      "loss": 0.1785,
+      "step": 2603
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.815313086696689e-06,
+      "loss": 0.1617,
+      "step": 2604
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.8082682063725506e-06,
+      "loss": 0.1758,
+      "step": 2605
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.8012283060740376e-06,
+      "loss": 0.1764,
+      "step": 2606
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.7941933914633587e-06,
+      "loss": 0.1713,
+      "step": 2607
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.7871634681987156e-06,
+      "loss": 0.1658,
+      "step": 2608
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.7801385419342904e-06,
+      "loss": 0.1723,
+      "step": 2609
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.7731186183202482e-06,
+      "loss": 0.1734,
+      "step": 2610
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.7661037030027383e-06,
+      "loss": 0.1721,
+      "step": 2611
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.75909380162387e-06,
+      "loss": 0.1725,
+      "step": 2612
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.7520889198217293e-06,
+      "loss": 0.1874,
+      "step": 2613
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.7450890632303516e-06,
+      "loss": 0.1783,
+      "step": 2614
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.738094237479746e-06,
+      "loss": 0.1711,
+      "step": 2615
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.731104448195867e-06,
+      "loss": 0.175,
+      "step": 2616
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.724119701000619e-06,
+      "loss": 0.1788,
+      "step": 2617
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.717140001511854e-06,
+      "loss": 0.1776,
+      "step": 2618
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.710165355343357e-06,
+      "loss": 0.1677,
+      "step": 2619
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.7031957681048604e-06,
+      "loss": 0.1709,
+      "step": 2620
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.696231245402019e-06,
+      "loss": 0.1771,
+      "step": 2621
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.6892717928364143e-06,
+      "loss": 0.1734,
+      "step": 2622
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.682317416005553e-06,
+      "loss": 0.1845,
+      "step": 2623
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.6753681205028568e-06,
+      "loss": 0.1855,
+      "step": 2624
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.6684239119176644e-06,
+      "loss": 0.181,
+      "step": 2625
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.6614847958352173e-06,
+      "loss": 0.1826,
+      "step": 2626
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.6545507778366697e-06,
+      "loss": 0.1786,
+      "step": 2627
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.647621863499058e-06,
+      "loss": 0.189,
+      "step": 2628
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.640698058395338e-06,
+      "loss": 0.1726,
+      "step": 2629
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.633779368094338e-06,
+      "loss": 0.1813,
+      "step": 2630
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.626865798160779e-06,
+      "loss": 0.1695,
+      "step": 2631
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.6199573541552647e-06,
+      "loss": 0.1829,
+      "step": 2632
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.613054041634272e-06,
+      "loss": 0.189,
+      "step": 2633
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.6061558661501528e-06,
+      "loss": 0.1815,
+      "step": 2634
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.599262833251128e-06,
+      "loss": 0.1845,
+      "step": 2635
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.592374948481282e-06,
+      "loss": 0.1779,
+      "step": 2636
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.585492217380556e-06,
+      "loss": 0.1722,
+      "step": 2637
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.57861464548475e-06,
+      "loss": 0.1726,
+      "step": 2638
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5717422383255094e-06,
+      "loss": 0.1704,
+      "step": 2639
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5648750014303345e-06,
+      "loss": 0.1804,
+      "step": 2640
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5580129403225573e-06,
+      "loss": 0.1828,
+      "step": 2641
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.551156060521356e-06,
+      "loss": 0.1739,
+      "step": 2642
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5443043675417288e-06,
+      "loss": 0.1746,
+      "step": 2643
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.5374578668945137e-06,
+      "loss": 0.1773,
+      "step": 2644
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.530616564086368e-06,
+      "loss": 0.1706,
+      "step": 2645
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.5237804646197684e-06,
+      "loss": 0.1733,
+      "step": 2646
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.5169495739930084e-06,
+      "loss": 0.1756,
+      "step": 2647
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.5101238977001873e-06,
+      "loss": 0.1718,
+      "step": 2648
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.5033034412312193e-06,
+      "loss": 0.1783,
+      "step": 2649
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.4964882100718133e-06,
+      "loss": 0.1809,
+      "step": 2650
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.4896782097034755e-06,
+      "loss": 0.1784,
+      "step": 2651
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.482873445603506e-06,
+      "loss": 0.1777,
+      "step": 2652
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.476073923244997e-06,
+      "loss": 0.1766,
+      "step": 2653
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.4692796480968182e-06,
+      "loss": 0.1834,
+      "step": 2654
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.4624906256236247e-06,
+      "loss": 0.1735,
+      "step": 2655
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.455706861285847e-06,
+      "loss": 0.175,
+      "step": 2656
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.4489283605396727e-06,
+      "loss": 0.1772,
+      "step": 2657
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.442155128837079e-06,
+      "loss": 0.1726,
+      "step": 2658
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.4353871716257904e-06,
+      "loss": 0.1716,
+      "step": 2659
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.4286244943492908e-06,
+      "loss": 0.1707,
+      "step": 2660
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.4218671024468185e-06,
+      "loss": 0.1841,
+      "step": 2661
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.4151150013533616e-06,
+      "loss": 0.1788,
+      "step": 2662
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.408368196499653e-06,
+      "loss": 0.173,
+      "step": 2663
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.4016266933121633e-06,
+      "loss": 0.1728,
+      "step": 2664
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.3948904972131012e-06,
+      "loss": 0.1786,
+      "step": 2665
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.3881596136204075e-06,
+      "loss": 0.1693,
+      "step": 2666
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.381434047947748e-06,
+      "loss": 0.1668,
+      "step": 2667
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.3747138056045093e-06,
+      "loss": 0.1727,
+      "step": 2668
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.367998891995805e-06,
+      "loss": 0.1792,
+      "step": 2669
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.3612893125224555e-06,
+      "loss": 0.1752,
+      "step": 2670
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.354585072580995e-06,
+      "loss": 0.176,
+      "step": 2671
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.347886177563654e-06,
+      "loss": 0.1813,
+      "step": 2672
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.3411926328583754e-06,
+      "loss": 0.1723,
+      "step": 2673
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.3345044438487915e-06,
+      "loss": 0.1874,
+      "step": 2674
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.327821615914233e-06,
+      "loss": 0.1806,
+      "step": 2675
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.321144154429713e-06,
+      "loss": 0.1779,
+      "step": 2676
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.3144720647659288e-06,
+      "loss": 0.1731,
+      "step": 2677
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.307805352289265e-06,
+      "loss": 0.1688,
+      "step": 2678
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.301144022361772e-06,
+      "loss": 0.1755,
+      "step": 2679
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.294488080341175e-06,
+      "loss": 0.1724,
+      "step": 2680
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.2878375315808643e-06,
+      "loss": 0.1681,
+      "step": 2681
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.281192381429894e-06,
+      "loss": 0.1733,
+      "step": 2682
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.274552635232975e-06,
+      "loss": 0.1709,
+      "step": 2683
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.2679182983304714e-06,
+      "loss": 0.1647,
+      "step": 2684
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.2612893760583974e-06,
+      "loss": 0.1742,
+      "step": 2685
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.254665873748414e-06,
+      "loss": 0.1748,
+      "step": 2686
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.248047796727818e-06,
+      "loss": 0.1747,
+      "step": 2687
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.2414351503195472e-06,
+      "loss": 0.184,
+      "step": 2688
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.23482793984217e-06,
+      "loss": 0.1796,
+      "step": 2689
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.2282261706098816e-06,
+      "loss": 0.1678,
+      "step": 2690
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.221629847932504e-06,
+      "loss": 0.1799,
+      "step": 2691
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.2150389771154734e-06,
+      "loss": 0.1731,
+      "step": 2692
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.2084535634598445e-06,
+      "loss": 0.1752,
+      "step": 2693
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.2018736122622852e-06,
+      "loss": 0.1819,
+      "step": 2694
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.1952991288150637e-06,
+      "loss": 0.1711,
+      "step": 2695
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.1887301184060513e-06,
+      "loss": 0.1844,
+      "step": 2696
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.182166586318727e-06,
+      "loss": 0.1542,
+      "step": 2697
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.1756085378321524e-06,
+      "loss": 0.1714,
+      "step": 2698
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.169055978220983e-06,
+      "loss": 0.1695,
+      "step": 2699
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.162508912755462e-06,
+      "loss": 0.179,
+      "step": 2700
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.155967346701405e-06,
+      "loss": 0.1775,
+      "step": 2701
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.1494312853202125e-06,
+      "loss": 0.1757,
+      "step": 2702
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.1429007338688568e-06,
+      "loss": 0.172,
+      "step": 2703
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.136375697599875e-06,
+      "loss": 0.1854,
+      "step": 2704
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.1298561817613683e-06,
+      "loss": 0.1729,
+      "step": 2705
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.1233421915970075e-06,
+      "loss": 0.1684,
+      "step": 2706
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.1168337323460053e-06,
+      "loss": 0.1854,
+      "step": 2707
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.110330809243134e-06,
+      "loss": 0.1697,
+      "step": 2708
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.103833427518712e-06,
+      "loss": 0.1782,
+      "step": 2709
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.0973415923985973e-06,
+      "loss": 0.1889,
+      "step": 2710
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.0908553091041913e-06,
+      "loss": 0.1818,
+      "step": 2711
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.0843745828524285e-06,
+      "loss": 0.1671,
+      "step": 2712
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.0778994188557722e-06,
+      "loss": 0.1689,
+      "step": 2713
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.0714298223222138e-06,
+      "loss": 0.1806,
+      "step": 2714
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.064965798455266e-06,
+      "loss": 0.1697,
+      "step": 2715
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.05850735245396e-06,
+      "loss": 0.185,
+      "step": 2716
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.0520544895128403e-06,
+      "loss": 0.1752,
+      "step": 2717
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.0456072148219606e-06,
+      "loss": 0.17,
+      "step": 2718
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.0391655335668813e-06,
+      "loss": 0.1868,
+      "step": 2719
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.032729450928662e-06,
+      "loss": 0.1704,
+      "step": 2720
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.02629897208386e-06,
+      "loss": 0.1825,
+      "step": 2721
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.0198741022045275e-06,
+      "loss": 0.172,
+      "step": 2722
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.0134548464582016e-06,
+      "loss": 0.1726,
+      "step": 2723
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.007041210007907e-06,
+      "loss": 0.1743,
+      "step": 2724
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.0006331980121452e-06,
+      "loss": 0.1691,
+      "step": 2725
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.994230815624902e-06,
+      "loss": 0.1804,
+      "step": 2726
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.9878340679956275e-06,
+      "loss": 0.1664,
+      "step": 2727
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.981442960269242e-06,
+      "loss": 0.1733,
+      "step": 2728
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.975057497586131e-06,
+      "loss": 0.1775,
+      "step": 2729
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.9686776850821354e-06,
+      "loss": 0.1722,
+      "step": 2730
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.9623035278885547e-06,
+      "loss": 0.1883,
+      "step": 2731
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.955935031132141e-06,
+      "loss": 0.1696,
+      "step": 2732
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.9495721999350923e-06,
+      "loss": 0.1832,
+      "step": 2733
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9432150394150462e-06,
+      "loss": 0.1661,
+      "step": 2734
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.936863554685091e-06,
+      "loss": 0.1684,
+      "step": 2735
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9305177508537376e-06,
+      "loss": 0.1703,
+      "step": 2736
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.924177633024933e-06,
+      "loss": 0.174,
+      "step": 2737
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9178432062980477e-06,
+      "loss": 0.1798,
+      "step": 2738
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9115144757678803e-06,
+      "loss": 0.1808,
+      "step": 2739
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.9051914465246436e-06,
+      "loss": 0.1824,
+      "step": 2740
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.8988741236539663e-06,
+      "loss": 0.1692,
+      "step": 2741
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.892562512236886e-06,
+      "loss": 0.1776,
+      "step": 2742
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.8862566173498495e-06,
+      "loss": 0.1811,
+      "step": 2743
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.879956444064703e-06,
+      "loss": 0.1656,
+      "step": 2744
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.8736619974486935e-06,
+      "loss": 0.176,
+      "step": 2745
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8673732825644587e-06,
+      "loss": 0.1632,
+      "step": 2746
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.861090304470028e-06,
+      "loss": 0.1647,
+      "step": 2747
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8548130682188167e-06,
+      "loss": 0.1758,
+      "step": 2748
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8485415788596226e-06,
+      "loss": 0.1695,
+      "step": 2749
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8422758414366216e-06,
+      "loss": 0.1778,
+      "step": 2750
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8360158609893594e-06,
+      "loss": 0.1723,
+      "step": 2751
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.8297616425527563e-06,
+      "loss": 0.1732,
+      "step": 2752
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.8235131911570956e-06,
+      "loss": 0.1739,
+      "step": 2753
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.8172705118280196e-06,
+      "loss": 0.1771,
+      "step": 2754
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.8110336095865365e-06,
+      "loss": 0.1634,
+      "step": 2755
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.804802489449001e-06,
+      "loss": 0.1685,
+      "step": 2756
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.798577156427118e-06,
+      "loss": 0.1649,
+      "step": 2757
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7923576155279385e-06,
+      "loss": 0.1723,
+      "step": 2758
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.786143871753858e-06,
+      "loss": 0.1745,
+      "step": 2759
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.779935930102602e-06,
+      "loss": 0.1709,
+      "step": 2760
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.773733795567234e-06,
+      "loss": 0.1651,
+      "step": 2761
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.767537473136149e-06,
+      "loss": 0.1695,
+      "step": 2762
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7613469677930604e-06,
+      "loss": 0.1733,
+      "step": 2763
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7551622845170144e-06,
+      "loss": 0.1741,
+      "step": 2764
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7489834282823634e-06,
+      "loss": 0.1631,
+      "step": 2765
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.742810404058779e-06,
+      "loss": 0.1824,
+      "step": 2766
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.736643216811239e-06,
+      "loss": 0.1672,
+      "step": 2767
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7304818715000268e-06,
+      "loss": 0.166,
+      "step": 2768
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.72432637308073e-06,
+      "loss": 0.1656,
+      "step": 2769
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7181767265042304e-06,
+      "loss": 0.1734,
+      "step": 2770
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.712032936716705e-06,
+      "loss": 0.1747,
+      "step": 2771
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.705895008659619e-06,
+      "loss": 0.166,
+      "step": 2772
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.699762947269724e-06,
+      "loss": 0.1836,
+      "step": 2773
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.693636757479053e-06,
+      "loss": 0.1747,
+      "step": 2774
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.6875164442149147e-06,
+      "loss": 0.1712,
+      "step": 2775
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.681402012399894e-06,
+      "loss": 0.1797,
+      "step": 2776
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.675293466951844e-06,
+      "loss": 0.1704,
+      "step": 2777
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.669190812783882e-06,
+      "loss": 0.1655,
+      "step": 2778
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.663094054804388e-06,
+      "loss": 0.1843,
+      "step": 2779
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.657003197917002e-06,
+      "loss": 0.1722,
+      "step": 2780
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.650918247020614e-06,
+      "loss": 0.181,
+      "step": 2781
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6448392070093666e-06,
+      "loss": 0.1714,
+      "step": 2782
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6387660827726436e-06,
+      "loss": 0.1754,
+      "step": 2783
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6326988791950802e-06,
+      "loss": 0.1725,
+      "step": 2784
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.626637601156541e-06,
+      "loss": 0.1767,
+      "step": 2785
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6205822535321267e-06,
+      "loss": 0.1779,
+      "step": 2786
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.614532841192169e-06,
+      "loss": 0.1706,
+      "step": 2787
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6084893690022285e-06,
+      "loss": 0.1836,
+      "step": 2788
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6024518418230803e-06,
+      "loss": 0.1715,
+      "step": 2789
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.596420264510725e-06,
+      "loss": 0.1626,
+      "step": 2790
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.5903946419163716e-06,
+      "loss": 0.1666,
+      "step": 2791
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.5843749788864516e-06,
+      "loss": 0.1773,
+      "step": 2792
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.578361280262591e-06,
+      "loss": 0.1791,
+      "step": 2793
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.5723535508816245e-06,
+      "loss": 0.1805,
+      "step": 2794
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.5663517955755824e-06,
+      "loss": 0.1746,
+      "step": 2795
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.5603560191716914e-06,
+      "loss": 0.1696,
+      "step": 2796
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.5543662264923707e-06,
+      "loss": 0.1649,
+      "step": 2797
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.5483824223552266e-06,
+      "loss": 0.1726,
+      "step": 2798
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.5424046115730473e-06,
+      "loss": 0.1788,
+      "step": 2799
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5364327989537995e-06,
+      "loss": 0.1753,
+      "step": 2800
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5304669893006295e-06,
+      "loss": 0.1658,
+      "step": 2801
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.524507187411852e-06,
+      "loss": 0.1698,
+      "step": 2802
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.518553398080952e-06,
+      "loss": 0.1678,
+      "step": 2803
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.512605626096576e-06,
+      "loss": 0.1683,
+      "step": 2804
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5066638762425334e-06,
+      "loss": 0.1763,
+      "step": 2805
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.500728153297788e-06,
+      "loss": 0.1678,
+      "step": 2806
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4947984620364563e-06,
+      "loss": 0.1717,
+      "step": 2807
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4888748072278046e-06,
+      "loss": 0.1815,
+      "step": 2808
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.482957193636244e-06,
+      "loss": 0.1786,
+      "step": 2809
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.477045626021326e-06,
+      "loss": 0.1866,
+      "step": 2810
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.471140109137736e-06,
+      "loss": 0.1776,
+      "step": 2811
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.465240647735303e-06,
+      "loss": 0.1629,
+      "step": 2812
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.459347246558974e-06,
+      "loss": 0.1806,
+      "step": 2813
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.453459910348829e-06,
+      "loss": 0.164,
+      "step": 2814
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.4475786438400663e-06,
+      "loss": 0.174,
+      "step": 2815
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.4417034517630024e-06,
+      "loss": 0.1685,
+      "step": 2816
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.435834338843074e-06,
+      "loss": 0.1674,
+      "step": 2817
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.429971309800817e-06,
+      "loss": 0.1745,
+      "step": 2818
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.424114369351883e-06,
+      "loss": 0.177,
+      "step": 2819
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4182635222070227e-06,
+      "loss": 0.166,
+      "step": 2820
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4124187730720916e-06,
+      "loss": 0.1873,
+      "step": 2821
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4065801266480336e-06,
+      "loss": 0.1739,
+      "step": 2822
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.400747587630887e-06,
+      "loss": 0.1619,
+      "step": 2823
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.3949211607117785e-06,
+      "loss": 0.181,
+      "step": 2824
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.3891008505769163e-06,
+      "loss": 0.1655,
+      "step": 2825
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.3832866619075936e-06,
+      "loss": 0.1646,
+      "step": 2826
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.3774785993801743e-06,
+      "loss": 0.1643,
+      "step": 2827
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.3716766676660985e-06,
+      "loss": 0.1726,
+      "step": 2828
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.365880871431876e-06,
+      "loss": 0.1737,
+      "step": 2829
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.360091215339079e-06,
+      "loss": 0.1713,
+      "step": 2830
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3543077040443405e-06,
+      "loss": 0.1719,
+      "step": 2831
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.348530342199361e-06,
+      "loss": 0.1806,
+      "step": 2832
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3427591344508803e-06,
+      "loss": 0.1718,
+      "step": 2833
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.336994085440697e-06,
+      "loss": 0.1767,
+      "step": 2834
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3312351998056547e-06,
+      "loss": 0.1697,
+      "step": 2835
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.3254824821776412e-06,
+      "loss": 0.1707,
+      "step": 2836
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.3197359371835802e-06,
+      "loss": 0.1743,
+      "step": 2837
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.313995569445434e-06,
+      "loss": 0.1748,
+      "step": 2838
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.3082613835801928e-06,
+      "loss": 0.1706,
+      "step": 2839
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.3025333841998764e-06,
+      "loss": 0.1755,
+      "step": 2840
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.2968115759115327e-06,
+      "loss": 0.1792,
+      "step": 2841
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.291095963317226e-06,
+      "loss": 0.1739,
+      "step": 2842
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2853865510140372e-06,
+      "loss": 0.168,
+      "step": 2843
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.27968334359406e-06,
+      "loss": 0.1692,
+      "step": 2844
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.273986345644401e-06,
+      "loss": 0.1659,
+      "step": 2845
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.26829556174717e-06,
+      "loss": 0.1781,
+      "step": 2846
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2626109964794764e-06,
+      "loss": 0.1646,
+      "step": 2847
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.256932654413431e-06,
+      "loss": 0.1759,
+      "step": 2848
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.251260540116137e-06,
+      "loss": 0.1749,
+      "step": 2849
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2455946581496958e-06,
+      "loss": 0.1567,
+      "step": 2850
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2399350130711874e-06,
+      "loss": 0.1699,
+      "step": 2851
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2342816094326793e-06,
+      "loss": 0.1815,
+      "step": 2852
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2286344517812175e-06,
+      "loss": 0.1653,
+      "step": 2853
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.222993544658827e-06,
+      "loss": 0.1743,
+      "step": 2854
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.2173588926025023e-06,
+      "loss": 0.1748,
+      "step": 2855
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.211730500144209e-06,
+      "loss": 0.1677,
+      "step": 2856
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.206108371810877e-06,
+      "loss": 0.1639,
+      "step": 2857
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.2004925121243994e-06,
+      "loss": 0.1717,
+      "step": 2858
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.1948829256016248e-06,
+      "loss": 0.1736,
+      "step": 2859
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.1892796167543562e-06,
+      "loss": 0.1729,
+      "step": 2860
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.183682590089358e-06,
+      "loss": 0.1711,
+      "step": 2861
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.178091850108324e-06,
+      "loss": 0.1739,
+      "step": 2862
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.1725074013079016e-06,
+      "loss": 0.1639,
+      "step": 2863
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.1669292481796813e-06,
+      "loss": 0.1675,
+      "step": 2864
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.161357395210182e-06,
+      "loss": 0.181,
+      "step": 2865
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.15579184688086e-06,
+      "loss": 0.1625,
+      "step": 2866
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.1502326076681023e-06,
+      "loss": 0.1734,
+      "step": 2867
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.144679682043217e-06,
+      "loss": 0.1648,
+      "step": 2868
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.139133074472434e-06,
+      "loss": 0.1693,
+      "step": 2869
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.1335927894169085e-06,
+      "loss": 0.1659,
+      "step": 2870
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.1280588313327042e-06,
+      "loss": 0.1675,
+      "step": 2871
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.122531204670798e-06,
+      "loss": 0.1799,
+      "step": 2872
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.117009913877074e-06,
+      "loss": 0.1734,
+      "step": 2873
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.1114949633923186e-06,
+      "loss": 0.1734,
+      "step": 2874
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.1059863576522253e-06,
+      "loss": 0.1682,
+      "step": 2875
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.1004841010873734e-06,
+      "loss": 0.1782,
+      "step": 2876
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.0949881981232454e-06,
+      "loss": 0.1715,
+      "step": 2877
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.0894986531802064e-06,
+      "loss": 0.1691,
+      "step": 2878
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.084015470673516e-06,
+      "loss": 0.1671,
+      "step": 2879
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.0785386550133114e-06,
+      "loss": 0.1765,
+      "step": 2880
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.0730682106046072e-06,
+      "loss": 0.1673,
+      "step": 2881
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.067604141847297e-06,
+      "loss": 0.1725,
+      "step": 2882
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.0621464531361444e-06,
+      "loss": 0.1716,
+      "step": 2883
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.0566951488607823e-06,
+      "loss": 0.1709,
+      "step": 2884
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0512502334057105e-06,
+      "loss": 0.1582,
+      "step": 2885
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0458117111502863e-06,
+      "loss": 0.1705,
+      "step": 2886
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.040379586468727e-06,
+      "loss": 0.1786,
+      "step": 2887
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0349538637301047e-06,
+      "loss": 0.1617,
+      "step": 2888
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0295345472983397e-06,
+      "loss": 0.1776,
+      "step": 2889
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0241216415322106e-06,
+      "loss": 0.1684,
+      "step": 2890
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.018715150785321e-06,
+      "loss": 0.1696,
+      "step": 2891
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0133150794061307e-06,
+      "loss": 0.1802,
+      "step": 2892
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.007921431737928e-06,
+      "loss": 0.1706,
+      "step": 2893
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0025342121188395e-06,
+      "loss": 0.1784,
+      "step": 2894
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.997153424881819e-06,
+      "loss": 0.1642,
+      "step": 2895
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.9917790743546474e-06,
+      "loss": 0.1681,
+      "step": 2896
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.986411164859928e-06,
+      "loss": 0.1719,
+      "step": 2897
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9810497007150818e-06,
+      "loss": 0.1689,
+      "step": 2898
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9756946862323534e-06,
+      "loss": 0.1634,
+      "step": 2899
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.970346125718793e-06,
+      "loss": 0.172,
+      "step": 2900
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9650040234762592e-06,
+      "loss": 0.1655,
+      "step": 2901
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9596683838014196e-06,
+      "loss": 0.1751,
+      "step": 2902
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.9543392109857416e-06,
+      "loss": 0.1674,
+      "step": 2903
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.9490165093154956e-06,
+      "loss": 0.1746,
+      "step": 2904
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.9437002830717433e-06,
+      "loss": 0.1674,
+      "step": 2905
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.9383905365303323e-06,
+      "loss": 0.1727,
+      "step": 2906
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.9330872739619144e-06,
+      "loss": 0.1617,
+      "step": 2907
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.927790499631912e-06,
+      "loss": 0.1722,
+      "step": 2908
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.922500217800536e-06,
+      "loss": 0.1785,
+      "step": 2909
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.917216432722774e-06,
+      "loss": 0.167,
+      "step": 2910
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.911939148648385e-06,
+      "loss": 0.1626,
+      "step": 2911
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.9066683698219057e-06,
+      "loss": 0.1786,
+      "step": 2912
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.9014041004826345e-06,
+      "loss": 0.1789,
+      "step": 2913
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.8961463448646367e-06,
+      "loss": 0.1695,
+      "step": 2914
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.8908951071967386e-06,
+      "loss": 0.163,
+      "step": 2915
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.8856503917025259e-06,
+      "loss": 0.1751,
+      "step": 2916
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.8804122026003312e-06,
+      "loss": 0.1529,
+      "step": 2917
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.87518054410325e-06,
+      "loss": 0.1565,
+      "step": 2918
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.8699554204191184e-06,
+      "loss": 0.1783,
+      "step": 2919
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.8647368357505125e-06,
+      "loss": 0.1664,
+      "step": 2920
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.8595247942947536e-06,
+      "loss": 0.174,
+      "step": 2921
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.854319300243901e-06,
+      "loss": 0.1668,
+      "step": 2922
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.8491203577847473e-06,
+      "loss": 0.1637,
+      "step": 2923
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.8439279710988144e-06,
+      "loss": 0.1747,
+      "step": 2924
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.838742144362352e-06,
+      "loss": 0.1697,
+      "step": 2925
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.8335628817463324e-06,
+      "loss": 0.1698,
+      "step": 2926
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.8283901874164533e-06,
+      "loss": 0.1715,
+      "step": 2927
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.8232240655331246e-06,
+      "loss": 0.1682,
+      "step": 2928
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.8180645202514702e-06,
+      "loss": 0.1669,
+      "step": 2929
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.8129115557213262e-06,
+      "loss": 0.1691,
+      "step": 2930
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.807765176087235e-06,
+      "loss": 0.1722,
+      "step": 2931
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.8026253854884424e-06,
+      "loss": 0.1729,
+      "step": 2932
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.7974921880588946e-06,
+      "loss": 0.178,
+      "step": 2933
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.7923655879272395e-06,
+      "loss": 0.1684,
+      "step": 2934
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.787245589216806e-06,
+      "loss": 0.1812,
+      "step": 2935
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.7821321960456294e-06,
+      "loss": 0.1632,
+      "step": 2936
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.777025412526423e-06,
+      "loss": 0.168,
+      "step": 2937
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.7719252427665856e-06,
+      "loss": 0.1662,
+      "step": 2938
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.766831690868196e-06,
+      "loss": 0.17,
+      "step": 2939
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7617447609280114e-06,
+      "loss": 0.1749,
+      "step": 2940
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7566644570374636e-06,
+      "loss": 0.1621,
+      "step": 2941
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7515907832826517e-06,
+      "loss": 0.1653,
+      "step": 2942
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7465237437443471e-06,
+      "loss": 0.1719,
+      "step": 2943
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.74146334249798e-06,
+      "loss": 0.1579,
+      "step": 2944
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7364095836136452e-06,
+      "loss": 0.182,
+      "step": 2945
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7313624711560916e-06,
+      "loss": 0.1683,
+      "step": 2946
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7263220091847288e-06,
+      "loss": 0.1701,
+      "step": 2947
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7212882017536137e-06,
+      "loss": 0.1661,
+      "step": 2948
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7162610529114453e-06,
+      "loss": 0.1591,
+      "step": 2949
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.711240566701574e-06,
+      "loss": 0.1656,
+      "step": 2950
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.706226747161992e-06,
+      "loss": 0.1665,
+      "step": 2951
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.7012195983253245e-06,
+      "loss": 0.1674,
+      "step": 2952
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.6962191242188342e-06,
+      "loss": 0.1773,
+      "step": 2953
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.6912253288644166e-06,
+      "loss": 0.1767,
+      "step": 2954
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.6862382162785906e-06,
+      "loss": 0.1717,
+      "step": 2955
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.6812577904725092e-06,
+      "loss": 0.1679,
+      "step": 2956
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6762840554519389e-06,
+      "loss": 0.1779,
+      "step": 2957
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6713170152172665e-06,
+      "loss": 0.1634,
+      "step": 2958
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.666356673763495e-06,
+      "loss": 0.1716,
+      "step": 2959
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6614030350802413e-06,
+      "loss": 0.1614,
+      "step": 2960
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6564561031517278e-06,
+      "loss": 0.1687,
+      "step": 2961
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6515158819567867e-06,
+      "loss": 0.1667,
+      "step": 2962
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6465823754688503e-06,
+      "loss": 0.175,
+      "step": 2963
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6416555876559436e-06,
+      "loss": 0.1723,
+      "step": 2964
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6367355224807035e-06,
+      "loss": 0.1785,
+      "step": 2965
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6318221839003478e-06,
+      "loss": 0.1762,
+      "step": 2966
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6269155758666865e-06,
+      "loss": 0.1788,
+      "step": 2967
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6220157023261173e-06,
+      "loss": 0.1695,
+      "step": 2968
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6171225672196212e-06,
+      "loss": 0.1692,
+      "step": 2969
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6122361744827585e-06,
+      "loss": 0.1678,
+      "step": 2970
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6073565280456681e-06,
+      "loss": 0.1628,
+      "step": 2971
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6024836318330628e-06,
+      "loss": 0.1624,
+      "step": 2972
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5976174897642227e-06,
+      "loss": 0.1714,
+      "step": 2973
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5927581057530006e-06,
+      "loss": 0.1649,
+      "step": 2974
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5879054837078089e-06,
+      "loss": 0.1695,
+      "step": 2975
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5830596275316268e-06,
+      "loss": 0.1647,
+      "step": 2976
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5782205411219887e-06,
+      "loss": 0.1696,
+      "step": 2977
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5733882283709845e-06,
+      "loss": 0.1694,
+      "step": 2978
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5685626931652531e-06,
+      "loss": 0.1713,
+      "step": 2979
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5637439393859844e-06,
+      "loss": 0.1685,
+      "step": 2980
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5589319709089167e-06,
+      "loss": 0.1674,
+      "step": 2981
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5541267916043268e-06,
+      "loss": 0.1592,
+      "step": 2982
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5493284053370329e-06,
+      "loss": 0.1671,
+      "step": 2983
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5445368159663888e-06,
+      "loss": 0.18,
+      "step": 2984
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.539752027346284e-06,
+      "loss": 0.1752,
+      "step": 2985
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5349740433251347e-06,
+      "loss": 0.1566,
+      "step": 2986
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.530202867745887e-06,
+      "loss": 0.1609,
+      "step": 2987
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5254385044460085e-06,
+      "loss": 0.1629,
+      "step": 2988
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5206809572574889e-06,
+      "loss": 0.1741,
+      "step": 2989
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5159302300068346e-06,
+      "loss": 0.1682,
+      "step": 2990
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5111863265150684e-06,
+      "loss": 0.171,
+      "step": 2991
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5064492505977234e-06,
+      "loss": 0.1624,
+      "step": 2992
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5017190060648434e-06,
+      "loss": 0.1707,
+      "step": 2993
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4969955967209749e-06,
+      "loss": 0.1784,
+      "step": 2994
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4922790263651665e-06,
+      "loss": 0.1631,
+      "step": 2995
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4875692987909708e-06,
+      "loss": 0.1829,
+      "step": 2996
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4828664177864327e-06,
+      "loss": 0.1712,
+      "step": 2997
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.478170387134089e-06,
+      "loss": 0.1653,
+      "step": 2998
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.473481210610973e-06,
+      "loss": 0.1673,
+      "step": 2999
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4687988919885977e-06,
+      "loss": 0.164,
+      "step": 3000
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4641234350329658e-06,
+      "loss": 0.1687,
+      "step": 3001
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4594548435045597e-06,
+      "loss": 0.1667,
+      "step": 3002
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4547931211583377e-06,
+      "loss": 0.1633,
+      "step": 3003
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.450138271743734e-06,
+      "loss": 0.1785,
+      "step": 3004
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4454902990046594e-06,
+      "loss": 0.1674,
+      "step": 3005
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4408492066794888e-06,
+      "loss": 0.1671,
+      "step": 3006
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4362149985010665e-06,
+      "loss": 0.1704,
+      "step": 3007
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.431587678196693e-06,
+      "loss": 0.1769,
+      "step": 3008
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.426967249488137e-06,
+      "loss": 0.1776,
+      "step": 3009
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4223537160916201e-06,
+      "loss": 0.175,
+      "step": 3010
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4177470817178196e-06,
+      "loss": 0.1591,
+      "step": 3011
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4131473500718596e-06,
+      "loss": 0.1624,
+      "step": 3012
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4085545248533227e-06,
+      "loss": 0.1595,
+      "step": 3013
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4039686097562255e-06,
+      "loss": 0.1612,
+      "step": 3014
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.3993896084690327e-06,
+      "loss": 0.1384,
+      "step": 3015
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.3948175246746465e-06,
+      "loss": 0.1439,
+      "step": 3016
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.3902523620504049e-06,
+      "loss": 0.1357,
+      "step": 3017
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.38569412426808e-06,
+      "loss": 0.1392,
+      "step": 3018
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.3811428149938745e-06,
+      "loss": 0.1358,
+      "step": 3019
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.376598437888418e-06,
+      "loss": 0.1411,
+      "step": 3020
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.3720609966067633e-06,
+      "loss": 0.1322,
+      "step": 3021
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.367530494798387e-06,
+      "loss": 0.144,
+      "step": 3022
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.363006936107183e-06,
+      "loss": 0.1451,
+      "step": 3023
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.358490324171461e-06,
+      "loss": 0.1421,
+      "step": 3024
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3539806626239438e-06,
+      "loss": 0.1342,
+      "step": 3025
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3494779550917637e-06,
+      "loss": 0.1373,
+      "step": 3026
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3449822051964589e-06,
+      "loss": 0.1422,
+      "step": 3027
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3404934165539741e-06,
+      "loss": 0.1419,
+      "step": 3028
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.336011592774652e-06,
+      "loss": 0.1375,
+      "step": 3029
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.331536737463235e-06,
+      "loss": 0.134,
+      "step": 3030
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3270688542188615e-06,
+      "loss": 0.1372,
+      "step": 3031
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3226079466350584e-06,
+      "loss": 0.1399,
+      "step": 3032
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.31815401829975e-06,
+      "loss": 0.1378,
+      "step": 3033
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3137070727952384e-06,
+      "loss": 0.1397,
+      "step": 3034
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.309267113698215e-06,
+      "loss": 0.1388,
+      "step": 3035
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.3048341445797519e-06,
+      "loss": 0.1374,
+      "step": 3036
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.3004081690052929e-06,
+      "loss": 0.1396,
+      "step": 3037
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.2959891905346634e-06,
+      "loss": 0.1312,
+      "step": 3038
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.2915772127220594e-06,
+      "loss": 0.1416,
+      "step": 3039
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.2871722391160458e-06,
+      "loss": 0.1448,
+      "step": 3040
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2827742732595516e-06,
+      "loss": 0.1365,
+      "step": 3041
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2783833186898753e-06,
+      "loss": 0.1409,
+      "step": 3042
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2739993789386729e-06,
+      "loss": 0.1359,
+      "step": 3043
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2696224575319572e-06,
+      "loss": 0.1338,
+      "step": 3044
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2652525579900976e-06,
+      "loss": 0.1412,
+      "step": 3045
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.260889683827814e-06,
+      "loss": 0.1343,
+      "step": 3046
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2565338385541792e-06,
+      "loss": 0.134,
+      "step": 3047
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2521850256726086e-06,
+      "loss": 0.138,
+      "step": 3048
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2478432486808655e-06,
+      "loss": 0.1369,
+      "step": 3049
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.243508511071051e-06,
+      "loss": 0.1374,
+      "step": 3050
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2391808163296049e-06,
+      "loss": 0.1378,
+      "step": 3051
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.234860167937304e-06,
+      "loss": 0.1423,
+      "step": 3052
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2305465693692576e-06,
+      "loss": 0.1425,
+      "step": 3053
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2262400240949023e-06,
+      "loss": 0.139,
+      "step": 3054
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2219405355780046e-06,
+      "loss": 0.1392,
+      "step": 3055
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2176481072766544e-06,
+      "loss": 0.14,
+      "step": 3056
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2133627426432604e-06,
+      "loss": 0.141,
+      "step": 3057
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2090844451245553e-06,
+      "loss": 0.1442,
+      "step": 3058
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.2048132181615823e-06,
+      "loss": 0.1342,
+      "step": 3059
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.200549065189701e-06,
+      "loss": 0.1405,
+      "step": 3060
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.1962919896385783e-06,
+      "loss": 0.132,
+      "step": 3061
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.192041994932196e-06,
+      "loss": 0.1337,
+      "step": 3062
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.187799084488832e-06,
+      "loss": 0.1397,
+      "step": 3063
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.1835632617210713e-06,
+      "loss": 0.1383,
+      "step": 3064
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1793345300357973e-06,
+      "loss": 0.1401,
+      "step": 3065
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.175112892834188e-06,
+      "loss": 0.1457,
+      "step": 3066
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1708983535117169e-06,
+      "loss": 0.1345,
+      "step": 3067
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1666909154581484e-06,
+      "loss": 0.1362,
+      "step": 3068
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1624905820575383e-06,
+      "loss": 0.135,
+      "step": 3069
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1582973566882205e-06,
+      "loss": 0.1432,
+      "step": 3070
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.154111242722823e-06,
+      "loss": 0.1317,
+      "step": 3071
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1499322435282444e-06,
+      "loss": 0.1441,
+      "step": 3072
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1457603624656654e-06,
+      "loss": 0.1397,
+      "step": 3073
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1415956028905395e-06,
+      "loss": 0.135,
+      "step": 3074
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1374379681525938e-06,
+      "loss": 0.1481,
+      "step": 3075
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1332874615958256e-06,
+      "loss": 0.1341,
+      "step": 3076
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1291440865584968e-06,
+      "loss": 0.1323,
+      "step": 3077
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.125007846373134e-06,
+      "loss": 0.1356,
+      "step": 3078
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.120878744366527e-06,
+      "loss": 0.1372,
+      "step": 3079
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1167567838597226e-06,
+      "loss": 0.1385,
+      "step": 3080
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1126419681680233e-06,
+      "loss": 0.1404,
+      "step": 3081
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1085343006009864e-06,
+      "loss": 0.1388,
+      "step": 3082
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1044337844624198e-06,
+      "loss": 0.1455,
+      "step": 3083
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.1003404230503788e-06,
+      "loss": 0.1458,
+      "step": 3084
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.0962542196571636e-06,
+      "loss": 0.1419,
+      "step": 3085
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.092175177569319e-06,
+      "loss": 0.1454,
+      "step": 3086
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.088103300067629e-06,
+      "loss": 0.141,
+      "step": 3087
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.0840385904271133e-06,
+      "loss": 0.141,
+      "step": 3088
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.0799810519170307e-06,
+      "loss": 0.1424,
+      "step": 3089
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0759306878008669e-06,
+      "loss": 0.1427,
+      "step": 3090
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0718875013363427e-06,
+      "loss": 0.1415,
+      "step": 3091
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0678514957754026e-06,
+      "loss": 0.1334,
+      "step": 3092
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0638226743642155e-06,
+      "loss": 0.135,
+      "step": 3093
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0598010403431747e-06,
+      "loss": 0.1421,
+      "step": 3094
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0557865969468873e-06,
+      "loss": 0.1376,
+      "step": 3095
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.0517793474041827e-06,
+      "loss": 0.1393,
+      "step": 3096
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.0477792949381004e-06,
+      "loss": 0.1318,
+      "step": 3097
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.043786442765894e-06,
+      "loss": 0.1387,
+      "step": 3098
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.0398007940990219e-06,
+      "loss": 0.139,
+      "step": 3099
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.0358223521431554e-06,
+      "loss": 0.1439,
+      "step": 3100
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.031851120098164e-06,
+      "loss": 0.1409,
+      "step": 3101
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.0278871011581193e-06,
+      "loss": 0.1403,
+      "step": 3102
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.023930298511293e-06,
+      "loss": 0.1407,
+      "step": 3103
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.0199807153401498e-06,
+      "loss": 0.1394,
+      "step": 3104
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.0160383548213504e-06,
+      "loss": 0.1275,
+      "step": 3105
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.0121032201257463e-06,
+      "loss": 0.139,
+      "step": 3106
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.0081753144183747e-06,
+      "loss": 0.1371,
+      "step": 3107
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.0042546408584597e-06,
+      "loss": 0.141,
+      "step": 3108
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.000341202599412e-06,
+      "loss": 0.136,
+      "step": 3109
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.96435002788817e-07,
+      "loss": 0.1366,
+      "step": 3110
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.925360445684418e-07,
+      "loss": 0.1363,
+      "step": 3111
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.886443310742288e-07,
+      "loss": 0.1419,
+      "step": 3112
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.847598654362934e-07,
+      "loss": 0.1402,
+      "step": 3113
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.808826507789215e-07,
+      "loss": 0.1391,
+      "step": 3114
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.770126902205646e-07,
+      "loss": 0.136,
+      "step": 3115
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.731499868738448e-07,
+      "loss": 0.1427,
+      "step": 3116
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.692945438455415e-07,
+      "loss": 0.1371,
+      "step": 3117
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.654463642365985e-07,
+      "loss": 0.1333,
+      "step": 3118
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.616054511421158e-07,
+      "loss": 0.1378,
+      "step": 3119
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.577718076513508e-07,
+      "loss": 0.1411,
+      "step": 3120
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.53945436847713e-07,
+      "loss": 0.1406,
+      "step": 3121
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.501263418087603e-07,
+      "loss": 0.1353,
+      "step": 3122
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.463145256062023e-07,
+      "loss": 0.1349,
+      "step": 3123
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.425099913058933e-07,
+      "loss": 0.1321,
+      "step": 3124
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.387127419678266e-07,
+      "loss": 0.1339,
+      "step": 3125
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.349227806461414e-07,
+      "loss": 0.1452,
+      "step": 3126
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.311401103891104e-07,
+      "loss": 0.138,
+      "step": 3127
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.273647342391501e-07,
+      "loss": 0.1357,
+      "step": 3128
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.235966552328024e-07,
+      "loss": 0.1408,
+      "step": 3129
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.198358764007431e-07,
+      "loss": 0.1346,
+      "step": 3130
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.16082400767776e-07,
+      "loss": 0.1394,
+      "step": 3131
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.123362313528316e-07,
+      "loss": 0.1383,
+      "step": 3132
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.085973711689622e-07,
+      "loss": 0.1398,
+      "step": 3133
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.048658232233443e-07,
+      "loss": 0.1364,
+      "step": 3134
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.011415905172694e-07,
+      "loss": 0.1333,
+      "step": 3135
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 8.974246760461491e-07,
+      "loss": 0.1325,
+      "step": 3136
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 8.937150827995067e-07,
+      "loss": 0.1362,
+      "step": 3137
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 8.900128137609776e-07,
+      "loss": 0.1381,
+      "step": 3138
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 8.863178719083054e-07,
+      "loss": 0.1375,
+      "step": 3139
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 8.826302602133419e-07,
+      "loss": 0.148,
+      "step": 3140
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 8.789499816420422e-07,
+      "loss": 0.1379,
+      "step": 3141
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 8.75277039154464e-07,
+      "loss": 0.1393,
+      "step": 3142
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 8.716114357047645e-07,
+      "loss": 0.1419,
+      "step": 3143
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.679531742411972e-07,
+      "loss": 0.1365,
+      "step": 3144
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.643022577061111e-07,
+      "loss": 0.1377,
+      "step": 3145
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.606586890359481e-07,
+      "loss": 0.1389,
+      "step": 3146
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.570224711612385e-07,
+      "loss": 0.1398,
+      "step": 3147
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.533936070066041e-07,
+      "loss": 0.1389,
+      "step": 3148
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.497720994907488e-07,
+      "loss": 0.1408,
+      "step": 3149
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.461579515264584e-07,
+      "loss": 0.1402,
+      "step": 3150
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.425511660206043e-07,
+      "loss": 0.1374,
+      "step": 3151
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.389517458741303e-07,
+      "loss": 0.1339,
+      "step": 3152
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.35359693982063e-07,
+      "loss": 0.1361,
+      "step": 3153
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.317750132334934e-07,
+      "loss": 0.1341,
+      "step": 3154
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.281977065115932e-07,
+      "loss": 0.1386,
+      "step": 3155
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.246277766935962e-07,
+      "loss": 0.1458,
+      "step": 3156
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.210652266508102e-07,
+      "loss": 0.1306,
+      "step": 3157
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.175100592486008e-07,
+      "loss": 0.1317,
+      "step": 3158
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.139622773463985e-07,
+      "loss": 0.132,
+      "step": 3159
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.10421883797694e-07,
+      "loss": 0.1363,
+      "step": 3160
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.068888814500342e-07,
+      "loss": 0.1388,
+      "step": 3161
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 8.033632731450213e-07,
+      "loss": 0.1425,
+      "step": 3162
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.998450617183118e-07,
+      "loss": 0.1425,
+      "step": 3163
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.96334249999613e-07,
+      "loss": 0.1352,
+      "step": 3164
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.928308408126772e-07,
+      "loss": 0.1363,
+      "step": 3165
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.893348369753074e-07,
+      "loss": 0.1391,
+      "step": 3166
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.858462412993472e-07,
+      "loss": 0.1451,
+      "step": 3167
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.823650565906837e-07,
+      "loss": 0.1398,
+      "step": 3168
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.788912856492437e-07,
+      "loss": 0.1362,
+      "step": 3169
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.754249312689876e-07,
+      "loss": 0.1447,
+      "step": 3170
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.719659962379133e-07,
+      "loss": 0.142,
+      "step": 3171
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.685144833380531e-07,
+      "loss": 0.1461,
+      "step": 3172
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.650703953454664e-07,
+      "loss": 0.1321,
+      "step": 3173
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.616337350302416e-07,
+      "loss": 0.1374,
+      "step": 3174
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.582045051564935e-07,
+      "loss": 0.1389,
+      "step": 3175
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.547827084823589e-07,
+      "loss": 0.1407,
+      "step": 3176
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.513683477600009e-07,
+      "loss": 0.1369,
+      "step": 3177
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.479614257355972e-07,
+      "loss": 0.1333,
+      "step": 3178
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.445619451493424e-07,
+      "loss": 0.1341,
+      "step": 3179
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.411699087354496e-07,
+      "loss": 0.1454,
+      "step": 3180
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.377853192221407e-07,
+      "loss": 0.1342,
+      "step": 3181
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.344081793316516e-07,
+      "loss": 0.1401,
+      "step": 3182
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.310384917802216e-07,
+      "loss": 0.1403,
+      "step": 3183
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.276762592781006e-07,
+      "loss": 0.1414,
+      "step": 3184
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.243214845295387e-07,
+      "loss": 0.1393,
+      "step": 3185
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.209741702327933e-07,
+      "loss": 0.1441,
+      "step": 3186
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.176343190801161e-07,
+      "loss": 0.1381,
+      "step": 3187
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.143019337577573e-07,
+      "loss": 0.1385,
+      "step": 3188
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.10977016945964e-07,
+      "loss": 0.1354,
+      "step": 3189
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.076595713189749e-07,
+      "loss": 0.1347,
+      "step": 3190
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.043495995450189e-07,
+      "loss": 0.1377,
+      "step": 3191
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.010471042863165e-07,
+      "loss": 0.1435,
+      "step": 3192
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.977520881990718e-07,
+      "loss": 0.143,
+      "step": 3193
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.944645539334749e-07,
+      "loss": 0.1416,
+      "step": 3194
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.911845041336984e-07,
+      "loss": 0.1373,
+      "step": 3195
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.879119414378921e-07,
+      "loss": 0.1424,
+      "step": 3196
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.846468684781916e-07,
+      "loss": 0.1366,
+      "step": 3197
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.813892878806994e-07,
+      "loss": 0.1393,
+      "step": 3198
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.781392022654965e-07,
+      "loss": 0.1398,
+      "step": 3199
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.748966142466362e-07,
+      "loss": 0.1361,
+      "step": 3200
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.716615264321391e-07,
+      "loss": 0.1342,
+      "step": 3201
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.684339414239948e-07,
+      "loss": 0.1396,
+      "step": 3202
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.65213861818158e-07,
+      "loss": 0.1359,
+      "step": 3203
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.620012902045481e-07,
+      "loss": 0.1407,
+      "step": 3204
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.587962291670414e-07,
+      "loss": 0.1373,
+      "step": 3205
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.555986812834814e-07,
+      "loss": 0.1399,
+      "step": 3206
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.52408649125661e-07,
+      "loss": 0.1413,
+      "step": 3207
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.492261352593321e-07,
+      "loss": 0.143,
+      "step": 3208
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.460511422441984e-07,
+      "loss": 0.1377,
+      "step": 3209
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.428836726339138e-07,
+      "loss": 0.1456,
+      "step": 3210
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.39723728976086e-07,
+      "loss": 0.1407,
+      "step": 3211
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.365713138122609e-07,
+      "loss": 0.1337,
+      "step": 3212
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.334264296779357e-07,
+      "loss": 0.1381,
+      "step": 3213
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.302890791025473e-07,
+      "loss": 0.135,
+      "step": 3214
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.271592646094793e-07,
+      "loss": 0.1351,
+      "step": 3215
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.240369887160458e-07,
+      "loss": 0.1474,
+      "step": 3216
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.209222539335014e-07,
+      "loss": 0.1387,
+      "step": 3217
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.178150627670366e-07,
+      "loss": 0.1333,
+      "step": 3218
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.147154177157721e-07,
+      "loss": 0.1357,
+      "step": 3219
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.116233212727607e-07,
+      "loss": 0.1345,
+      "step": 3220
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.085387759249839e-07,
+      "loss": 0.1421,
+      "step": 3221
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.054617841533484e-07,
+      "loss": 0.144,
+      "step": 3222
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.023923484326866e-07,
+      "loss": 0.1393,
+      "step": 3223
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 5.993304712317538e-07,
+      "loss": 0.1367,
+      "step": 3224
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 5.962761550132245e-07,
+      "loss": 0.1376,
+      "step": 3225
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 5.93229402233697e-07,
+      "loss": 0.1399,
+      "step": 3226
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 5.901902153436779e-07,
+      "loss": 0.1343,
+      "step": 3227
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.871585967875948e-07,
+      "loss": 0.1379,
+      "step": 3228
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.841345490037853e-07,
+      "loss": 0.1344,
+      "step": 3229
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.811180744244993e-07,
+      "loss": 0.1329,
+      "step": 3230
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.781091754758961e-07,
+      "loss": 0.1364,
+      "step": 3231
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.751078545780386e-07,
+      "loss": 0.1393,
+      "step": 3232
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.72114114144896e-07,
+      "loss": 0.1497,
+      "step": 3233
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.691279565843455e-07,
+      "loss": 0.1422,
+      "step": 3234
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.66149384298158e-07,
+      "loss": 0.1364,
+      "step": 3235
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.631783996820083e-07,
+      "loss": 0.1444,
+      "step": 3236
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.602150051254662e-07,
+      "loss": 0.1391,
+      "step": 3237
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.572592030119983e-07,
+      "loss": 0.136,
+      "step": 3238
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.543109957189619e-07,
+      "loss": 0.137,
+      "step": 3239
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.513703856176112e-07,
+      "loss": 0.1453,
+      "step": 3240
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.484373750730809e-07,
+      "loss": 0.1339,
+      "step": 3241
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.45511966444402e-07,
+      "loss": 0.1398,
+      "step": 3242
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.425941620844888e-07,
+      "loss": 0.1383,
+      "step": 3243
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.396839643401375e-07,
+      "loss": 0.1438,
+      "step": 3244
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.367813755520279e-07,
+      "loss": 0.1421,
+      "step": 3245
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.338863980547204e-07,
+      "loss": 0.1408,
+      "step": 3246
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.309990341766513e-07,
+      "loss": 0.1382,
+      "step": 3247
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.281192862401352e-07,
+      "loss": 0.1378,
+      "step": 3248
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.252471565613615e-07,
+      "loss": 0.1422,
+      "step": 3249
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.223826474503901e-07,
+      "loss": 0.1397,
+      "step": 3250
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.195257612111537e-07,
+      "loss": 0.1425,
+      "step": 3251
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.166765001414542e-07,
+      "loss": 0.1398,
+      "step": 3252
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.138348665329562e-07,
+      "loss": 0.136,
+      "step": 3253
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.11000862671196e-07,
+      "loss": 0.1381,
+      "step": 3254
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.081744908355713e-07,
+      "loss": 0.143,
+      "step": 3255
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.053557532993358e-07,
+      "loss": 0.1395,
+      "step": 3256
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.025446523296096e-07,
+      "loss": 0.1389,
+      "step": 3257
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.997411901873695e-07,
+      "loss": 0.1381,
+      "step": 3258
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.969453691274451e-07,
+      "loss": 0.1393,
+      "step": 3259
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.941571913985221e-07,
+      "loss": 0.1364,
+      "step": 3260
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.913766592431412e-07,
+      "loss": 0.139,
+      "step": 3261
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.886037748976891e-07,
+      "loss": 0.1402,
+      "step": 3262
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.858385405924071e-07,
+      "loss": 0.1397,
+      "step": 3263
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.830809585513762e-07,
+      "loss": 0.1341,
+      "step": 3264
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.803310309925302e-07,
+      "loss": 0.1359,
+      "step": 3265
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.775887601276408e-07,
+      "loss": 0.1441,
+      "step": 3266
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.7485414816232254e-07,
+      "loss": 0.1367,
+      "step": 3267
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.7212719729603397e-07,
+      "loss": 0.1328,
+      "step": 3268
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.6940790972206605e-07,
+      "loss": 0.1393,
+      "step": 3269
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.666962876275494e-07,
+      "loss": 0.1345,
+      "step": 3270
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.6399233319344703e-07,
+      "loss": 0.1373,
+      "step": 3271
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.6129604859455836e-07,
+      "loss": 0.1465,
+      "step": 3272
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.5860743599951186e-07,
+      "loss": 0.1338,
+      "step": 3273
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.5592649757076445e-07,
+      "loss": 0.1344,
+      "step": 3274
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.5325323546460355e-07,
+      "loss": 0.1373,
+      "step": 3275
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.5058765183113807e-07,
+      "loss": 0.1301,
+      "step": 3276
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.479297488143053e-07,
+      "loss": 0.1391,
+      "step": 3277
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.4527952855186405e-07,
+      "loss": 0.1357,
+      "step": 3278
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.4263699317539156e-07,
+      "loss": 0.1465,
+      "step": 3279
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.4000214481028645e-07,
+      "loss": 0.1432,
+      "step": 3280
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.3737498557576364e-07,
+      "loss": 0.1312,
+      "step": 3281
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.347555175848539e-07,
+      "loss": 0.1359,
+      "step": 3282
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.3214374294440306e-07,
+      "loss": 0.1411,
+      "step": 3283
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.2953966375506864e-07,
+      "loss": 0.1361,
+      "step": 3284
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.269432821113162e-07,
+      "loss": 0.1383,
+      "step": 3285
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.2435460010142316e-07,
+      "loss": 0.132,
+      "step": 3286
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.217736198074718e-07,
+      "loss": 0.1359,
+      "step": 3287
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.1920034330535265e-07,
+      "loss": 0.1359,
+      "step": 3288
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.1663477266475907e-07,
+      "loss": 0.1369,
+      "step": 3289
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.140769099491837e-07,
+      "loss": 0.1438,
+      "step": 3290
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.1152675721592206e-07,
+      "loss": 0.1376,
+      "step": 3291
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.0898431651607117e-07,
+      "loss": 0.1378,
+      "step": 3292
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.064495898945209e-07,
+      "loss": 0.1406,
+      "step": 3293
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.0392257938995927e-07,
+      "loss": 0.1487,
+      "step": 3294
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.014032870348661e-07,
+      "loss": 0.1366,
+      "step": 3295
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.9889171485551383e-07,
+      "loss": 0.1382,
+      "step": 3296
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.963878648719688e-07,
+      "loss": 0.1376,
+      "step": 3297
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.9389173909808234e-07,
+      "loss": 0.1309,
+      "step": 3298
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.914033395414973e-07,
+      "loss": 0.1423,
+      "step": 3299
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.88922668203634e-07,
+      "loss": 0.1378,
+      "step": 3300
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.8644972707970743e-07,
+      "loss": 0.1323,
+      "step": 3301
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.8398451815870984e-07,
+      "loss": 0.1307,
+      "step": 3302
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.8152704342341416e-07,
+      "loss": 0.131,
+      "step": 3303
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.7907730485037374e-07,
+      "loss": 0.1407,
+      "step": 3304
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.7663530440992025e-07,
+      "loss": 0.1369,
+      "step": 3305
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.742010440661603e-07,
+      "loss": 0.1413,
+      "step": 3306
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.717745257769745e-07,
+      "loss": 0.1462,
+      "step": 3307
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.693557514940194e-07,
+      "loss": 0.1315,
+      "step": 3308
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.66944723162721e-07,
+      "loss": 0.1349,
+      "step": 3309
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.645414427222749e-07,
+      "loss": 0.1337,
+      "step": 3310
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.6214591210564586e-07,
+      "loss": 0.1395,
+      "step": 3311
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.59758133239565e-07,
+      "loss": 0.1374,
+      "step": 3312
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.5737810804453155e-07,
+      "loss": 0.1432,
+      "step": 3313
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.5500583843480207e-07,
+      "loss": 0.1405,
+      "step": 3314
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.526413263184003e-07,
+      "loss": 0.1277,
+      "step": 3315
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.5028457359711053e-07,
+      "loss": 0.1366,
+      "step": 3316
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.4793558216647315e-07,
+      "loss": 0.1388,
+      "step": 3317
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.4559435391578913e-07,
+      "loss": 0.142,
+      "step": 3318
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.432608907281143e-07,
+      "loss": 0.1398,
+      "step": 3319
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.409351944802575e-07,
+      "loss": 0.1389,
+      "step": 3320
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.386172670427845e-07,
+      "loss": 0.1449,
+      "step": 3321
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.3630711028000974e-07,
+      "loss": 0.1378,
+      "step": 3322
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.340047260499968e-07,
+      "loss": 0.1317,
+      "step": 3323
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.3171011620455993e-07,
+      "loss": 0.1394,
+      "step": 3324
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.294232825892596e-07,
+      "loss": 0.1446,
+      "step": 3325
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.271442270434022e-07,
+      "loss": 0.1371,
+      "step": 3326
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.248729514000382e-07,
+      "loss": 0.1373,
+      "step": 3327
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.226094574859584e-07,
+      "loss": 0.1366,
+      "step": 3328
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.2035374712169997e-07,
+      "loss": 0.1343,
+      "step": 3329
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.1810582212153273e-07,
+      "loss": 0.1329,
+      "step": 3330
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.1586568429347155e-07,
+      "loss": 0.1385,
+      "step": 3331
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.13633335439264e-07,
+      "loss": 0.1346,
+      "step": 3332
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.114087773543939e-07,
+      "loss": 0.1392,
+      "step": 3333
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.091920118280789e-07,
+      "loss": 0.1379,
+      "step": 3334
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.0698304064326835e-07,
+      "loss": 0.1349,
+      "step": 3335
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.0478186557664436e-07,
+      "loss": 0.138,
+      "step": 3336
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.0258848839861744e-07,
+      "loss": 0.1364,
+      "step": 3337
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.004029108733264e-07,
+      "loss": 0.1398,
+      "step": 3338
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.9822513475863735e-07,
+      "loss": 0.1375,
+      "step": 3339
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.9605516180613914e-07,
+      "loss": 0.139,
+      "step": 3340
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.93892993761149e-07,
+      "loss": 0.1391,
+      "step": 3341
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.917386323627047e-07,
+      "loss": 0.1425,
+      "step": 3342
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.895920793435614e-07,
+      "loss": 0.1295,
+      "step": 3343
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.874533364302001e-07,
+      "loss": 0.1295,
+      "step": 3344
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.85322405342815e-07,
+      "loss": 0.1409,
+      "step": 3345
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.831992877953205e-07,
+      "loss": 0.1359,
+      "step": 3346
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.810839854953451e-07,
+      "loss": 0.1376,
+      "step": 3347
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.7897650014423016e-07,
+      "loss": 0.1342,
+      "step": 3348
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.768768334370353e-07,
+      "loss": 0.1308,
+      "step": 3349
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.7478498706252523e-07,
+      "loss": 0.1354,
+      "step": 3350
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.727009627031785e-07,
+      "loss": 0.1359,
+      "step": 3351
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.7062476203518097e-07,
+      "loss": 0.1427,
+      "step": 3352
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.6855638672842465e-07,
+      "loss": 0.1419,
+      "step": 3353
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.6649583844651104e-07,
+      "loss": 0.1421,
+      "step": 3354
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.6444311884674224e-07,
+      "loss": 0.1387,
+      "step": 3355
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.6239822958012863e-07,
+      "loss": 0.138,
+      "step": 3356
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.603611722913768e-07,
+      "loss": 0.1309,
+      "step": 3357
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.5833194861889954e-07,
+      "loss": 0.1388,
+      "step": 3358
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.5631056019480574e-07,
+      "loss": 0.1396,
+      "step": 3359
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.542970086449026e-07,
+      "loss": 0.1421,
+      "step": 3360
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.522912955886958e-07,
+      "loss": 0.1353,
+      "step": 3361
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.5029342263938385e-07,
+      "loss": 0.1424,
+      "step": 3362
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.483033914038635e-07,
+      "loss": 0.1332,
+      "step": 3363
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.4632120348272e-07,
+      "loss": 0.1346,
+      "step": 3364
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.443468604702326e-07,
+      "loss": 0.1348,
+      "step": 3365
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.4238036395437094e-07,
+      "loss": 0.1352,
+      "step": 3366
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.4042171551679206e-07,
+      "loss": 0.1349,
+      "step": 3367
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.384709167328425e-07,
+      "loss": 0.1343,
+      "step": 3368
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.3652796917155496e-07,
+      "loss": 0.1433,
+      "step": 3369
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.3459287439564825e-07,
+      "loss": 0.1423,
+      "step": 3370
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.3266563396152075e-07,
+      "loss": 0.1364,
+      "step": 3371
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.3074624941926137e-07,
+      "loss": 0.1402,
+      "step": 3372
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.2883472231263194e-07,
+      "loss": 0.1373,
+      "step": 3373
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.2693105417907922e-07,
+      "loss": 0.1376,
+      "step": 3374
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.2503524654972965e-07,
+      "loss": 0.1379,
+      "step": 3375
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.2314730094938564e-07,
+      "loss": 0.1401,
+      "step": 3376
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.2126721889652592e-07,
+      "loss": 0.1366,
+      "step": 3377
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.1939500190330642e-07,
+      "loss": 0.1417,
+      "step": 3378
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1753065147555598e-07,
+      "loss": 0.1387,
+      "step": 3379
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1567416911277506e-07,
+      "loss": 0.1428,
+      "step": 3380
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1382555630813929e-07,
+      "loss": 0.1341,
+      "step": 3381
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.119848145484904e-07,
+      "loss": 0.1343,
+      "step": 3382
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1015194531434303e-07,
+      "loss": 0.138,
+      "step": 3383
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.0832695007987792e-07,
+      "loss": 0.1374,
+      "step": 3384
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.0650983031294424e-07,
+      "loss": 0.1399,
+      "step": 3385
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.0470058747505516e-07,
+      "loss": 0.1376,
+      "step": 3386
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.0289922302138997e-07,
+      "loss": 0.1413,
+      "step": 3387
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.0110573840078974e-07,
+      "loss": 0.134,
+      "step": 3388
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.993201350557594e-07,
+      "loss": 0.1339,
+      "step": 3389
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.975424144224636e-07,
+      "loss": 0.1375,
+      "step": 3390
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.9577257793072956e-07,
+      "loss": 0.1356,
+      "step": 3391
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.9401062700403873e-07,
+      "loss": 0.1345,
+      "step": 3392
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.922565630595341e-07,
+      "loss": 0.1367,
+      "step": 3393
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.9051038750801275e-07,
+      "loss": 0.1308,
+      "step": 3394
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.88772101753929e-07,
+      "loss": 0.1398,
+      "step": 3395
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.8704170719539116e-07,
+      "loss": 0.1315,
+      "step": 3396
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.8531920522415813e-07,
+      "loss": 0.1412,
+      "step": 3397
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.8360459722564394e-07,
+      "loss": 0.1293,
+      "step": 3398
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.8189788457891211e-07,
+      "loss": 0.1333,
+      "step": 3399
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.8019906865667568e-07,
+      "loss": 0.1354,
+      "step": 3400
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.7850815082529728e-07,
+      "loss": 0.1403,
+      "step": 3401
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.768251324447856e-07,
+      "loss": 0.1384,
+      "step": 3402
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.7515001486879569e-07,
+      "loss": 0.1358,
+      "step": 3403
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.7348279944463087e-07,
+      "loss": 0.1368,
+      "step": 3404
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.718234875132363e-07,
+      "loss": 0.1386,
+      "step": 3405
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.7017208040919886e-07,
+      "loss": 0.1334,
+      "step": 3406
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.6852857946075164e-07,
+      "loss": 0.138,
+      "step": 3407
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.668929859897661e-07,
+      "loss": 0.1407,
+      "step": 3408
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.6526530131175333e-07,
+      "loss": 0.1369,
+      "step": 3409
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.6364552673586608e-07,
+      "loss": 0.1369,
+      "step": 3410
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.6203366356489114e-07,
+      "loss": 0.1332,
+      "step": 3411
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.604297130952559e-07,
+      "loss": 0.1428,
+      "step": 3412
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.5883367661702064e-07,
+      "loss": 0.1421,
+      "step": 3413
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.57245555413883e-07,
+      "loss": 0.1338,
+      "step": 3414
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.5566535076317003e-07,
+      "loss": 0.1331,
+      "step": 3415
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.5409306393584734e-07,
+      "loss": 0.1412,
+      "step": 3416
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.525286961965078e-07,
+      "loss": 0.1334,
+      "step": 3417
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.5097224880337603e-07,
+      "loss": 0.134,
+      "step": 3418
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.4942372300830733e-07,
+      "loss": 0.1383,
+      "step": 3419
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.478831200567854e-07,
+      "loss": 0.1339,
+      "step": 3420
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4635044118792018e-07,
+      "loss": 0.14,
+      "step": 3421
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4482568763444894e-07,
+      "loss": 0.1332,
+      "step": 3422
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4330886062273619e-07,
+      "loss": 0.1355,
+      "step": 3423
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.417999613727683e-07,
+      "loss": 0.1369,
+      "step": 3424
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.402989910981578e-07,
+      "loss": 0.1356,
+      "step": 3425
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.388059510061379e-07,
+      "loss": 0.136,
+      "step": 3426
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.3732084229756582e-07,
+      "loss": 0.1376,
+      "step": 3427
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.3584366616691712e-07,
+      "loss": 0.1414,
+      "step": 3428
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.3437442380228926e-07,
+      "loss": 0.1405,
+      "step": 3429
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.3291311638539805e-07,
+      "loss": 0.1386,
+      "step": 3430
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.3145974509157555e-07,
+      "loss": 0.138,
+      "step": 3431
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.3001431108977115e-07,
+      "loss": 0.1381,
+      "step": 3432
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.2857681554255263e-07,
+      "loss": 0.1342,
+      "step": 3433
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.2714725960610187e-07,
+      "loss": 0.1371,
+      "step": 3434
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.257256444302113e-07,
+      "loss": 0.1358,
+      "step": 3435
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.2431197115829297e-07,
+      "loss": 0.1428,
+      "step": 3436
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.2290624092736735e-07,
+      "loss": 0.1424,
+      "step": 3437
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.215084548680656e-07,
+      "loss": 0.143,
+      "step": 3438
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.2011861410463067e-07,
+      "loss": 0.1361,
+      "step": 3439
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.1873671975491608e-07,
+      "loss": 0.1355,
+      "step": 3440
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.1736277293038279e-07,
+      "loss": 0.1384,
+      "step": 3441
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.1599677473609904e-07,
+      "loss": 0.1336,
+      "step": 3442
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.1463872627074046e-07,
+      "loss": 0.1348,
+      "step": 3443
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.1328862862658885e-07,
+      "loss": 0.1333,
+      "step": 3444
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.1194648288953336e-07,
+      "loss": 0.1414,
+      "step": 3445
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.1061229013906161e-07,
+      "loss": 0.138,
+      "step": 3446
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.0928605144827186e-07,
+      "loss": 0.1326,
+      "step": 3447
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.0796776788385865e-07,
+      "loss": 0.1406,
+      "step": 3448
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.066574405061216e-07,
+      "loss": 0.138,
+      "step": 3449
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.0535507036896098e-07,
+      "loss": 0.1375,
+      "step": 3450
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.0406065851987556e-07,
+      "loss": 0.1451,
+      "step": 3451
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.0277420599996479e-07,
+      "loss": 0.1386,
+      "step": 3452
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.0149571384392431e-07,
+      "loss": 0.1393,
+      "step": 3453
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.0022518308005046e-07,
+      "loss": 0.1394,
+      "step": 3454
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 9.89626147302325e-08,
+      "loss": 0.1328,
+      "step": 3455
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 9.77080098099592e-08,
+      "loss": 0.1372,
+      "step": 3456
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 9.64613693283123e-08,
+      "loss": 0.1374,
+      "step": 3457
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 9.522269428796638e-08,
+      "loss": 0.1327,
+      "step": 3458
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 9.39919856851923e-08,
+      "loss": 0.1368,
+      "step": 3459
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 9.276924450985159e-08,
+      "loss": 0.1413,
+      "step": 3460
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 9.155447174539867e-08,
+      "loss": 0.1421,
+      "step": 3461
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 9.034766836887643e-08,
+      "loss": 0.1398,
+      "step": 3462
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 8.914883535092178e-08,
+      "loss": 0.139,
+      "step": 3463
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 8.795797365575898e-08,
+      "loss": 0.1354,
+      "step": 3464
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 8.677508424120074e-08,
+      "loss": 0.1322,
+      "step": 3465
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 8.560016805864823e-08,
+      "loss": 0.1422,
+      "step": 3466
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 8.44332260530889e-08,
+      "loss": 0.132,
+      "step": 3467
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 8.327425916309861e-08,
+      "loss": 0.1375,
+      "step": 3468
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.212326832083506e-08,
+      "loss": 0.1367,
+      "step": 3469
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.098025445204549e-08,
+      "loss": 0.1419,
+      "step": 3470
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.984521847605786e-08,
+      "loss": 0.1439,
+      "step": 3471
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.871816130578414e-08,
+      "loss": 0.141,
+      "step": 3472
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.75990838477203e-08,
+      "loss": 0.1342,
+      "step": 3473
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.648798700194194e-08,
+      "loss": 0.135,
+      "step": 3474
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.53848716621075e-08,
+      "loss": 0.1322,
+      "step": 3475
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.42897387154562e-08,
+      "loss": 0.1341,
+      "step": 3476
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.320258904280453e-08,
+      "loss": 0.1332,
+      "step": 3477
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.212342351855194e-08,
+      "loss": 0.1334,
+      "step": 3478
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.105224301067193e-08,
+      "loss": 0.1402,
+      "step": 3479
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.998904838071973e-08,
+      "loss": 0.1363,
+      "step": 3480
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.893384048382468e-08,
+      "loss": 0.1352,
+      "step": 3481
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.788662016869229e-08,
+      "loss": 0.1385,
+      "step": 3482
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.68473882776044e-08,
+      "loss": 0.1409,
+      "step": 3483
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.581614564641903e-08,
+      "loss": 0.1381,
+      "step": 3484
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.479289310456716e-08,
+      "loss": 0.1403,
+      "step": 3485
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.377763147505378e-08,
+      "loss": 0.1378,
+      "step": 3486
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.27703615744557e-08,
+      "loss": 0.1385,
+      "step": 3487
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.177108421292266e-08,
+      "loss": 0.1383,
+      "step": 3488
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.077980019417506e-08,
+      "loss": 0.1376,
+      "step": 3489
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 5.979651031550737e-08,
+      "loss": 0.1409,
+      "step": 3490
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 5.8821215367779185e-08,
+      "loss": 0.1366,
+      "step": 3491
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 5.785391613542524e-08,
+      "loss": 0.132,
+      "step": 3492
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.689461339644653e-08,
+      "loss": 0.1315,
+      "step": 3493
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.594330792241143e-08,
+      "loss": 0.1345,
+      "step": 3494
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.5000000478457886e-08,
+      "loss": 0.1464,
+      "step": 3495
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.4064691823291215e-08,
+      "loss": 0.1426,
+      "step": 3496
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.313738270918189e-08,
+      "loss": 0.1332,
+      "step": 3497
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.221807388196665e-08,
+      "loss": 0.1354,
+      "step": 3498
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.1306766081048456e-08,
+      "loss": 0.139,
+      "step": 3499
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.040346003939434e-08,
+      "loss": 0.134,
+      "step": 3500
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.950815648353646e-08,
+      "loss": 0.1385,
+      "step": 3501
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.862085613356771e-08,
+      "loss": 0.1399,
+      "step": 3502
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.77415597031472e-08,
+      "loss": 0.1377,
+      "step": 3503
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.687026789949589e-08,
+      "loss": 0.1417,
+      "step": 3504
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.6006981423395434e-08,
+      "loss": 0.1395,
+      "step": 3505
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.51517009691893e-08,
+      "loss": 0.1323,
+      "step": 3506
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.430442722478168e-08,
+      "loss": 0.1352,
+      "step": 3507
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.346516087163744e-08,
+      "loss": 0.1379,
+      "step": 3508
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.263390258477995e-08,
+      "loss": 0.136,
+      "step": 3509
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.1810653032793306e-08,
+      "loss": 0.1385,
+      "step": 3510
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.0995412877817833e-08,
+      "loss": 0.1408,
+      "step": 3511
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.018818277555347e-08,
+      "loss": 0.1376,
+      "step": 3512
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.938896337525977e-08,
+      "loss": 0.1294,
+      "step": 3513
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.859775531974808e-08,
+      "loss": 0.1299,
+      "step": 3514
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.781455924539046e-08,
+      "loss": 0.1378,
+      "step": 3515
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.703937578211303e-08,
+      "loss": 0.1336,
+      "step": 3516
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.6272205553397056e-08,
+      "loss": 0.1391,
+      "step": 3517
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.551304917628229e-08,
+      "loss": 0.1358,
+      "step": 3518
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.4761907261356976e-08,
+      "loss": 0.1364,
+      "step": 3519
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.401878041276785e-08,
+      "loss": 0.1354,
+      "step": 3520
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.328366922821458e-08,
+      "loss": 0.1347,
+      "step": 3521
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.255657429894643e-08,
+      "loss": 0.1301,
+      "step": 3522
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.183749620977006e-08,
+      "loss": 0.1388,
+      "step": 3523
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.112643553904171e-08,
+      "loss": 0.143,
+      "step": 3524
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.0423392858668354e-08,
+      "loss": 0.1393,
+      "step": 3525
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.9728368734108783e-08,
+      "loss": 0.1383,
+      "step": 3526
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.9041363724373605e-08,
+      "loss": 0.1338,
+      "step": 3527
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.836237838202194e-08,
+      "loss": 0.1334,
+      "step": 3528
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.7691413253163602e-08,
+      "loss": 0.1348,
+      "step": 3529
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.7028468877458026e-08,
+      "loss": 0.1383,
+      "step": 3530
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.637354578811202e-08,
+      "loss": 0.1365,
+      "step": 3531
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.5726644511883115e-08,
+      "loss": 0.1353,
+      "step": 3532
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.5087765569075106e-08,
+      "loss": 0.1352,
+      "step": 3533
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.445690947354029e-08,
+      "loss": 0.1381,
+      "step": 3534
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.3834076732677237e-08,
+      "loss": 0.1369,
+      "step": 3535
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.3219267847431893e-08,
+      "loss": 0.1394,
+      "step": 3536
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.2612483312296484e-08,
+      "loss": 0.1393,
+      "step": 3537
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.2013723615310622e-08,
+      "loss": 0.1314,
+      "step": 3538
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.142298923805686e-08,
+      "loss": 0.1474,
+      "step": 3539
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.0840280655665123e-08,
+      "loss": 0.1337,
+      "step": 3540
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.02655983368083e-08,
+      "loss": 0.1362,
+      "step": 3541
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.9698942743706652e-08,
+      "loss": 0.1399,
+      "step": 3542
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.9140314332122266e-08,
+      "loss": 0.1406,
+      "step": 3543
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.858971355136019e-08,
+      "loss": 0.1357,
+      "step": 3544
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.8047140844270615e-08,
+      "loss": 0.1352,
+      "step": 3545
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.7512596647246694e-08,
+      "loss": 0.1362,
+      "step": 3546
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.6986081390223397e-08,
+      "loss": 0.1386,
+      "step": 3547
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.6467595496677534e-08,
+      "loss": 0.1365,
+      "step": 3548
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.595713938362886e-08,
+      "loss": 0.1308,
+      "step": 3549
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.545471346164007e-08,
+      "loss": 0.1314,
+      "step": 3550
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.4960318134811248e-08,
+      "loss": 0.137,
+      "step": 3551
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.4473953800787643e-08,
+      "loss": 0.1356,
+      "step": 3552
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.3995620850753011e-08,
+      "loss": 0.1355,
+      "step": 3553
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.3525319669430715e-08,
+      "loss": 0.132,
+      "step": 3554
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.306305063508595e-08,
+      "loss": 0.1334,
+      "step": 3555
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.2608814119523526e-08,
+      "loss": 0.1418,
+      "step": 3556
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.2162610488086757e-08,
+      "loss": 0.137,
+      "step": 3557
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.1724440099656342e-08,
+      "loss": 0.1437,
+      "step": 3558
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.129430330665704e-08,
+      "loss": 0.136,
+      "step": 3559
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.0872200455046556e-08,
+      "loss": 0.1403,
+      "step": 3560
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.0458131884323318e-08,
+      "loss": 0.1418,
+      "step": 3561
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.0052097927525372e-08,
+      "loss": 0.145,
+      "step": 3562
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 9.654098911225928e-09,
+      "loss": 0.1358,
+      "step": 3563
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 9.264135155537812e-09,
+      "loss": 0.1385,
+      "step": 3564
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 8.8822069741068e-09,
+      "loss": 0.1424,
+      "step": 3565
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 8.50831467412161e-09,
+      "loss": 0.1343,
+      "step": 3566
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 8.142458556303911e-09,
+      "loss": 0.1376,
+      "step": 3567
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.784638914912768e-09,
+      "loss": 0.1346,
+      "step": 3568
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.4348560377424105e-09,
+      "loss": 0.1367,
+      "step": 3569
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.093110206125575e-09,
+      "loss": 0.1385,
+      "step": 3570
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.759401694927947e-09,
+      "loss": 0.1455,
+      "step": 3571
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.433730772553715e-09,
+      "loss": 0.145,
+      "step": 3572
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.116097700938906e-09,
+      "loss": 0.137,
+      "step": 3573
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 5.806502735559161e-09,
+      "loss": 0.1352,
+      "step": 3574
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 5.504946125419741e-09,
+      "loss": 0.1369,
+      "step": 3575
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 5.21142811306552e-09,
+      "loss": 0.1382,
+      "step": 3576
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.9259489345721e-09,
+      "loss": 0.1353,
+      "step": 3577
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.648508819552477e-09,
+      "loss": 0.1397,
+      "step": 3578
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.379107991151488e-09,
+      "loss": 0.1313,
+      "step": 3579
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.117746666049138e-09,
+      "loss": 0.1499,
+      "step": 3580
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.8644250544594975e-09,
+      "loss": 0.1401,
+      "step": 3581
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.6191433601295845e-09,
+      "loss": 0.136,
+      "step": 3582
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.3819017803404797e-09,
+      "loss": 0.1427,
+      "step": 3583
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.152700505905104e-09,
+      "loss": 0.1364,
+      "step": 3584
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.9315397211715504e-09,
+      "loss": 0.1296,
+      "step": 3585
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.7184196040186408e-09,
+      "loss": 0.135,
+      "step": 3586
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.5133403258614797e-09,
+      "loss": 0.138,
+      "step": 3587
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.3163020516436817e-09,
+      "loss": 0.1359,
+      "step": 3588
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.127304939846253e-09,
+      "loss": 0.1373,
+      "step": 3589
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.9463491424775994e-09,
+      "loss": 0.1319,
+      "step": 3590
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.7734348050812977e-09,
+      "loss": 0.1439,
+      "step": 3591
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.6085620667349866e-09,
+      "loss": 0.1338,
+      "step": 3592
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.451731060043704e-09,
+      "loss": 0.1395,
+      "step": 3593
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.3029419111476594e-09,
+      "loss": 0.1452,
+      "step": 3594
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.1621947397189026e-09,
+      "loss": 0.1373,
+      "step": 3595
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.0294896589602144e-09,
+      "loss": 0.139,
+      "step": 3596
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 9.048267756073259e-10,
+      "loss": 0.1405,
+      "step": 3597
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.882061899255888e-10,
+      "loss": 0.1386,
+      "step": 3598
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.79627995715526e-10,
+      "loss": 0.1342,
+      "step": 3599
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 5.7909228030395e-10,
+      "loss": 0.1389,
+      "step": 3600
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.86599124555065e-10,
+      "loss": 0.1399,
+      "step": 3601
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.021486028582544e-10,
+      "loss": 0.1413,
+      "step": 3602
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.2574078314029366e-10,
+      "loss": 0.1376,
+      "step": 3603
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.573757268531374e-10,
+      "loss": 0.1351,
+      "step": 3604
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.9705348898502175e-10,
+      "loss": 0.1357,
+      "step": 3605
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.4477411805380316e-10,
+      "loss": 0.1297,
+      "step": 3606
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.0053765610584798e-10,
+      "loss": 0.1361,
+      "step": 3607
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.434413872158374e-11,
+      "loss": 0.1412,
+      "step": 3608
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.6193595012168345e-11,
+      "loss": 0.133,
+      "step": 3609
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.6086047618379952e-11,
+      "loss": 0.1315,
+      "step": 3610
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.02151271283735e-12,
+      "loss": 0.1334,
+      "step": 3611
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0,
+      "loss": 0.1393,
+      "step": 3612
+    },
+    {
+      "epoch": 5.99,
+      "step": 3612,
+      "total_flos": 9545592376033280.0,
+      "train_loss": 0.4382430633173168,
+      "train_runtime": 64967.3342,
+      "train_samples_per_second": 28.491,
+      "train_steps_per_second": 0.056
+    }
+  ],
+  "max_steps": 3612,
+  "num_train_epochs": 6,
+  "total_flos": 9545592376033280.0,
+  "trial_name": null,
+  "trial_params": null
+}