diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5947 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9931766489764975,
+  "global_step": 987,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 0.8405,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 0.8875,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.8105,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 0.8326,
+      "step": 4
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.8068,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.7921,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 0.7852,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.7643,
+      "step": 8
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6e-06,
+      "loss": 0.7709,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.7565,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 0.7547,
+      "step": 11
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.7551,
+      "step": 12
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 0.7409,
+      "step": 13
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.7446,
+      "step": 14
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 0.7421,
+      "step": 15
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.7197,
+      "step": 16
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1333333333333334e-05,
+      "loss": 0.6987,
+      "step": 17
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2e-05,
+      "loss": 0.6948,
+      "step": 18
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2666666666666667e-05,
+      "loss": 0.7041,
+      "step": 19
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.7211,
+      "step": 20
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4e-05,
+      "loss": 0.7248,
+      "step": 21
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 0.6965,
+      "step": 22
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5333333333333334e-05,
+      "loss": 0.7132,
+      "step": 23
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.6919,
+      "step": 24
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.7127,
+      "step": 25
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 0.6937,
+      "step": 26
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8e-05,
+      "loss": 0.7053,
+      "step": 27
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.6993,
+      "step": 28
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9333333333333333e-05,
+      "loss": 0.6823,
+      "step": 29
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.6896,
+      "step": 30
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999946117779583e-05,
+      "loss": 0.6686,
+      "step": 31
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999978447169898e-05,
+      "loss": 0.662,
+      "step": 32
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999951506350017e-05,
+      "loss": 0.6867,
+      "step": 33
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999913789608641e-05,
+      "loss": 0.652,
+      "step": 34
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998652973522228e-05,
+      "loss": 0.6772,
+      "step": 35
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998060301033363e-05,
+      "loss": 0.6734,
+      "step": 36
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997359885006716e-05,
+      "loss": 0.6656,
+      "step": 37
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996551732990278e-05,
+      "loss": 0.6682,
+      "step": 38
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995635853693057e-05,
+      "loss": 0.6552,
+      "step": 39
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999461225698498e-05,
+      "loss": 0.663,
+      "step": 40
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999348095389677e-05,
+      "loss": 0.6436,
+      "step": 41
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999224195661986e-05,
+      "loss": 0.7066,
+      "step": 42
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999089527850623e-05,
+      "loss": 0.6806,
+      "step": 43
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989440934068284e-05,
+      "loss": 0.651,
+      "step": 44
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987878938978684e-05,
+      "loss": 0.6694,
+      "step": 45
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986209310070176e-05,
+      "loss": 0.6734,
+      "step": 46
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984432065335435e-05,
+      "loss": 0.6513,
+      "step": 47
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982547223926826e-05,
+      "loss": 0.6406,
+      "step": 48
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980554806156246e-05,
+      "loss": 0.6352,
+      "step": 49
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997845483349487e-05,
+      "loss": 0.6666,
+      "step": 50
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997624732857294e-05,
+      "loss": 0.6324,
+      "step": 51
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973932315179502e-05,
+      "loss": 0.632,
+      "step": 52
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997150981826217e-05,
+      "loss": 0.6384,
+      "step": 53
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968979863926857e-05,
+      "loss": 0.6379,
+      "step": 54
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996634247943746e-05,
+      "loss": 0.6414,
+      "step": 55
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963597693215618e-05,
+      "loss": 0.6412,
+      "step": 56
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960745534840357e-05,
+      "loss": 0.6221,
+      "step": 57
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957786035047807e-05,
+      "loss": 0.6086,
+      "step": 58
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954719225730847e-05,
+      "loss": 0.6422,
+      "step": 59
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995154513993878e-05,
+      "loss": 0.6152,
+      "step": 60
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994826381187697e-05,
+      "loss": 0.6356,
+      "step": 61
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944875276906456e-05,
+      "loss": 0.6048,
+      "step": 62
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941379571543597e-05,
+      "loss": 0.6341,
+      "step": 63
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993777673345967e-05,
+      "loss": 0.6222,
+      "step": 64
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934066801480457e-05,
+      "loss": 0.6088,
+      "step": 65
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993024981558583e-05,
+      "loss": 0.6214,
+      "step": 66
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992632581690933e-05,
+      "loss": 0.6352,
+      "step": 67
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99222948477377e-05,
+      "loss": 0.6233,
+      "step": 68
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991815695151046e-05,
+      "loss": 0.6226,
+      "step": 69
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991391217281942e-05,
+      "loss": 0.5973,
+      "step": 70
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9909560557408195e-05,
+      "loss": 0.6079,
+      "step": 71
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905102152171728e-05,
+      "loss": 0.6003,
+      "step": 72
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990053700515577e-05,
+      "loss": 0.6207,
+      "step": 73
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895865165556375e-05,
+      "loss": 0.6103,
+      "step": 74
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891086683719362e-05,
+      "loss": 0.6346,
+      "step": 75
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886201611139776e-05,
+      "loss": 0.6012,
+      "step": 76
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988121000046132e-05,
+      "loss": 0.5876,
+      "step": 77
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876111905475816e-05,
+      "loss": 0.6066,
+      "step": 78
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870907381122595e-05,
+      "loss": 0.6139,
+      "step": 79
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865596483487925e-05,
+      "loss": 0.5871,
+      "step": 80
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9860179269804394e-05,
+      "loss": 0.5769,
+      "step": 81
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854655798450304e-05,
+      "loss": 0.6219,
+      "step": 82
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849026128949037e-05,
+      "loss": 0.5993,
+      "step": 83
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984329032196841e-05,
+      "loss": 0.5881,
+      "step": 84
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837448439320027e-05,
+      "loss": 0.5788,
+      "step": 85
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983150054395861e-05,
+      "loss": 0.5997,
+      "step": 86
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982544669998132e-05,
+      "loss": 0.5728,
+      "step": 87
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819286972627066e-05,
+      "loss": 0.608,
+      "step": 88
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813021428275812e-05,
+      "loss": 0.5882,
+      "step": 89
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806650134447837e-05,
+      "loss": 0.5869,
+      "step": 90
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.980017315980304e-05,
+      "loss": 0.5716,
+      "step": 91
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793590574140174e-05,
+      "loss": 0.5618,
+      "step": 92
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786902448396102e-05,
+      "loss": 0.5773,
+      "step": 93
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9780108854645045e-05,
+      "loss": 0.5599,
+      "step": 94
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977320986609778e-05,
+      "loss": 0.5968,
+      "step": 95
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976620555710087e-05,
+      "loss": 0.5644,
+      "step": 96
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759096003135863e-05,
+      "loss": 0.6087,
+      "step": 97
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751881280818467e-05,
+      "loss": 0.5824,
+      "step": 98
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9744561467897735e-05,
+      "loss": 0.6041,
+      "step": 99
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737136643255224e-05,
+      "loss": 0.5853,
+      "step": 100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972960688690414e-05,
+      "loss": 0.5921,
+      "step": 101
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972197227998848e-05,
+      "loss": 0.557,
+      "step": 102
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971423290478216e-05,
+      "loss": 0.5641,
+      "step": 103
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970638884468812e-05,
+      "loss": 0.5922,
+      "step": 104
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698440184237442e-05,
+      "loss": 0.5565,
+      "step": 105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9690387009088418e-05,
+      "loss": 0.5526,
+      "step": 106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682229406025635e-05,
+      "loss": 0.5535,
+      "step": 107
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9673967462959052e-05,
+      "loss": 0.5696,
+      "step": 108
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665601268923037e-05,
+      "loss": 0.5751,
+      "step": 109
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9657130914075406e-05,
+      "loss": 0.5625,
+      "step": 110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648556489696472e-05,
+      "loss": 0.5816,
+      "step": 111
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639878088188034e-05,
+      "loss": 0.5679,
+      "step": 112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9631095803072404e-05,
+      "loss": 0.5697,
+      "step": 113
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9622209728991382e-05,
+      "loss": 0.567,
+      "step": 114
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613219961705256e-05,
+      "loss": 0.5575,
+      "step": 115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9604126598091745e-05,
+      "loss": 0.5704,
+      "step": 116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 0.5681,
+      "step": 117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9585629474974413e-05,
+      "loss": 0.5573,
+      "step": 118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576225914803805e-05,
+      "loss": 0.5542,
+      "step": 119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566719156970095e-05,
+      "loss": 0.5612,
+      "step": 120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955710930392232e-05,
+      "loss": 0.5478,
+      "step": 121
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9547396459220526e-05,
+      "loss": 0.547,
+      "step": 122
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9537580727534643e-05,
+      "loss": 0.545,
+      "step": 123
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9527662214643355e-05,
+      "loss": 0.5575,
+      "step": 124
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951764102743296e-05,
+      "loss": 0.5503,
+      "step": 125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9507517273896224e-05,
+      "loss": 0.562,
+      "step": 126
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9497291063131208e-05,
+      "loss": 0.5685,
+      "step": 127
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486962505340098e-05,
+      "loss": 0.5501,
+      "step": 128
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9476531711828027e-05,
+      "loss": 0.5419,
+      "step": 129
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9465998795001852e-05,
+      "loss": 0.5378,
+      "step": 130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9455363868368964e-05,
+      "loss": 0.552,
+      "step": 131
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9444627046536055e-05,
+      "loss": 0.5649,
+      "step": 132
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9433788445207887e-05,
+      "loss": 0.5479,
+      "step": 133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942284818118604e-05,
+      "loss": 0.508,
+      "step": 134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9411806372367656e-05,
+      "loss": 0.5577,
+      "step": 135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940066313774417e-05,
+      "loss": 0.5447,
+      "step": 136
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9389418597400032e-05,
+      "loss": 0.5333,
+      "step": 137
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9378072872511397e-05,
+      "loss": 0.5626,
+      "step": 138
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9366626085344834e-05,
+      "loss": 0.5411,
+      "step": 139
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935507835925601e-05,
+      "loss": 0.5552,
+      "step": 140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934342981868835e-05,
+      "loss": 0.53,
+      "step": 141
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933168058917169e-05,
+      "loss": 0.5524,
+      "step": 142
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9319830797320956e-05,
+      "loss": 0.5381,
+      "step": 143
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9307880570834762e-05,
+      "loss": 0.527,
+      "step": 144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9295830038494057e-05,
+      "loss": 0.5495,
+      "step": 145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9283679330160726e-05,
+      "loss": 0.5371,
+      "step": 146
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9271428576776206e-05,
+      "loss": 0.5417,
+      "step": 147
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9259077910360045e-05,
+      "loss": 0.5412,
+      "step": 148
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9246627464008515e-05,
+      "loss": 0.5457,
+      "step": 149
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9234077371893156e-05,
+      "loss": 0.5375,
+      "step": 150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9221427769259333e-05,
+      "loss": 0.5387,
+      "step": 151
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9208678792424778e-05,
+      "loss": 0.5594,
+      "step": 152
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9195830578778133e-05,
+      "loss": 0.5245,
+      "step": 153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9182883266777447e-05,
+      "loss": 0.5446,
+      "step": 154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916983699594871e-05,
+      "loss": 0.532,
+      "step": 155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9156691906884327e-05,
+      "loss": 0.5257,
+      "step": 156
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9143448141241614e-05,
+      "loss": 0.5314,
+      "step": 157
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9130105841741274e-05,
+      "loss": 0.5304,
+      "step": 158
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911666515216585e-05,
+      "loss": 0.5219,
+      "step": 159
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9103126217358183e-05,
+      "loss": 0.5493,
+      "step": 160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9089489183219844e-05,
+      "loss": 0.5312,
+      "step": 161
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9075754196709574e-05,
+      "loss": 0.5201,
+      "step": 162
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906192140584168e-05,
+      "loss": 0.5274,
+      "step": 163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9047990959684464e-05,
+      "loss": 0.5464,
+      "step": 164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90339630083586e-05,
+      "loss": 0.5328,
+      "step": 165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9019837703035518e-05,
+      "loss": 0.5272,
+      "step": 166
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9005615195935784e-05,
+      "loss": 0.531,
+      "step": 167
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899129564032745e-05,
+      "loss": 0.5321,
+      "step": 168
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897687919052441e-05,
+      "loss": 0.5358,
+      "step": 169
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896236600188472e-05,
+      "loss": 0.531,
+      "step": 170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8947756230808955e-05,
+      "loss": 0.5327,
+      "step": 171
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8933050034738488e-05,
+      "loss": 0.5066,
+      "step": 172
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8918247572153822e-05,
+      "loss": 0.5287,
+      "step": 173
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8903349002572873e-05,
+      "loss": 0.5194,
+      "step": 174
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8888354486549238e-05,
+      "loss": 0.5188,
+      "step": 175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887326418567048e-05,
+      "loss": 0.5246,
+      "step": 176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885807826255638e-05,
+      "loss": 0.5178,
+      "step": 177
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8842796880857196e-05,
+      "loss": 0.5004,
+      "step": 178
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8827420205251875e-05,
+      "loss": 0.5108,
+      "step": 179
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8811948401446312e-05,
+      "loss": 0.5167,
+      "step": 180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879638163617153e-05,
+      "loss": 0.5353,
+      "step": 181
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8780720077181902e-05,
+      "loss": 0.4994,
+      "step": 182
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8764963893253346e-05,
+      "loss": 0.5499,
+      "step": 183
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8749113254181498e-05,
+      "loss": 0.5267,
+      "step": 184
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8733168330779882e-05,
+      "loss": 0.5149,
+      "step": 185
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8717129294878075e-05,
+      "loss": 0.5036,
+      "step": 186
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870099631931985e-05,
+      "loss": 0.5244,
+      "step": 187
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8684769577961313e-05,
+      "loss": 0.5138,
+      "step": 188
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866844924566904e-05,
+      "loss": 0.5174,
+      "step": 189
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8652035498318176e-05,
+      "loss": 0.5114,
+      "step": 190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863552851279055e-05,
+      "loss": 0.4968,
+      "step": 191
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8618928466972773e-05,
+      "loss": 0.5117,
+      "step": 192
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860223553975431e-05,
+      "loss": 0.5125,
+      "step": 193
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8585449911025552e-05,
+      "loss": 0.5157,
+      "step": 194
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8568571761675893e-05,
+      "loss": 0.5208,
+      "step": 195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8551601273591767e-05,
+      "loss": 0.5399,
+      "step": 196
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853453862965469e-05,
+      "loss": 0.5045,
+      "step": 197
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8517384013739287e-05,
+      "loss": 0.5174,
+      "step": 198
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8500137610711318e-05,
+      "loss": 0.5305,
+      "step": 199
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8482799606425682e-05,
+      "loss": 0.5183,
+      "step": 200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846537018772441e-05,
+      "loss": 0.5123,
+      "step": 201
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8447849542434662e-05,
+      "loss": 0.5142,
+      "step": 202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843023785936669e-05,
+      "loss": 0.5164,
+      "step": 203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 0.5072,
+      "step": 204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8394742140040367e-05,
+      "loss": 0.5325,
+      "step": 205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.837685848629965e-05,
+      "loss": 0.4929,
+      "step": 206
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8358884559811855e-05,
+      "loss": 0.5136,
+      "step": 207
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8340820554271998e-05,
+      "loss": 0.4984,
+      "step": 208
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8322666664345827e-05,
+      "loss": 0.5226,
+      "step": 209
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8304423085667713e-05,
+      "loss": 0.5172,
+      "step": 210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8286090014838567e-05,
+      "loss": 0.4949,
+      "step": 211
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8267667649423706e-05,
+      "loss": 0.4996,
+      "step": 212
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8249156187950717e-05,
+      "loss": 0.5119,
+      "step": 213
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8230555829907325e-05,
+      "loss": 0.4953,
+      "step": 214
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8211866775739256e-05,
+      "loss": 0.5101,
+      "step": 215
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819308922684805e-05,
+      "loss": 0.5052,
+      "step": 216
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817422338558892e-05,
+      "loss": 0.508,
+      "step": 217
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8155269455268544e-05,
+      "loss": 0.4816,
+      "step": 218
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8136227640142895e-05,
+      "loss": 0.5268,
+      "step": 219
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8117098145415026e-05,
+      "loss": 0.4916,
+      "step": 220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809788117723287e-05,
+      "loss": 0.5089,
+      "step": 221
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807857694268701e-05,
+      "loss": 0.4848,
+      "step": 222
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805918564980845e-05,
+      "loss": 0.5027,
+      "step": 223
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8039707507566372e-05,
+      "loss": 0.5017,
+      "step": 224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802014272586589e-05,
+      "loss": 0.5,
+      "step": 225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800049151554577e-05,
+      "loss": 0.5015,
+      "step": 226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79807540883762e-05,
+      "loss": 0.5064,
+      "step": 227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796093065705644e-05,
+      "loss": 0.5043,
+      "step": 228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79410214352126e-05,
+      "loss": 0.5015,
+      "step": 229
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7921026637395296e-05,
+      "loss": 0.5129,
+      "step": 230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7900946479077345e-05,
+      "loss": 0.5,
+      "step": 231
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7880781176651453e-05,
+      "loss": 0.502,
+      "step": 232
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7860530947427878e-05,
+      "loss": 0.4979,
+      "step": 233
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784019600963207e-05,
+      "loss": 0.5033,
+      "step": 234
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7819776582402364e-05,
+      "loss": 0.5076,
+      "step": 235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7799272885787565e-05,
+      "loss": 0.4962,
+      "step": 236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777868514074462e-05,
+      "loss": 0.5033,
+      "step": 237
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7758013569136207e-05,
+      "loss": 0.4922,
+      "step": 238
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7737258393728363e-05,
+      "loss": 0.4992,
+      "step": 239
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771641983818808e-05,
+      "loss": 0.4764,
+      "step": 240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769549812708088e-05,
+      "loss": 0.5142,
+      "step": 241
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7674493485868415e-05,
+      "loss": 0.4965,
+      "step": 242
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7653406140906027e-05,
+      "loss": 0.4949,
+      "step": 243
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763223631944031e-05,
+      "loss": 0.4894,
+      "step": 244
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7610984249606665e-05,
+      "loss": 0.4776,
+      "step": 245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758965016042683e-05,
+      "loss": 0.4966,
+      "step": 246
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756823428180642e-05,
+      "loss": 0.5021,
+      "step": 247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7546736844532463e-05,
+      "loss": 0.5095,
+      "step": 248
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752515808027088e-05,
+      "loss": 0.5043,
+      "step": 249
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7503498221564026e-05,
+      "loss": 0.4924,
+      "step": 250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748175750182815e-05,
+      "loss": 0.4938,
+      "step": 251
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7459936155350908e-05,
+      "loss": 0.4788,
+      "step": 252
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743803441728882e-05,
+      "loss": 0.4665,
+      "step": 253
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741605252366474e-05,
+      "loss": 0.4852,
+      "step": 254
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7393990711365312e-05,
+      "loss": 0.5024,
+      "step": 255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7371849218138427e-05,
+      "loss": 0.4659,
+      "step": 256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734962828259065e-05,
+      "loss": 0.5019,
+      "step": 257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7327328144184648e-05,
+      "loss": 0.5015,
+      "step": 258
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7304949043236612e-05,
+      "loss": 0.4861,
+      "step": 259
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728249122091368e-05,
+      "loss": 0.4918,
+      "step": 260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725995491923131e-05,
+      "loss": 0.4815,
+      "step": 261
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.72373403810507e-05,
+      "loss": 0.4847,
+      "step": 262
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7214647850076165e-05,
+      "loss": 0.4926,
+      "step": 263
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7191877570852482e-05,
+      "loss": 0.4928,
+      "step": 264
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7169029788762302e-05,
+      "loss": 0.4729,
+      "step": 265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7146104750023466e-05,
+      "loss": 0.4918,
+      "step": 266
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712310270168637e-05,
+      "loss": 0.506,
+      "step": 267
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710002389163131e-05,
+      "loss": 0.4757,
+      "step": 268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7076868568565785e-05,
+      "loss": 0.4883,
+      "step": 269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7053636982021844e-05,
+      "loss": 0.4985,
+      "step": 270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7030329382353374e-05,
+      "loss": 0.4918,
+      "step": 271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7006946020733426e-05,
+      "loss": 0.494,
+      "step": 272
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6983487149151486e-05,
+      "loss": 0.4749,
+      "step": 273
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695995302041077e-05,
+      "loss": 0.4805,
+      "step": 274
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.693634388812551e-05,
+      "loss": 0.4913,
+      "step": 275
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6912660006718186e-05,
+      "loss": 0.5004,
+      "step": 276
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688890163141683e-05,
+      "loss": 0.4698,
+      "step": 277
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6865069018252246e-05,
+      "loss": 0.4982,
+      "step": 278
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684116242405525e-05,
+      "loss": 0.476,
+      "step": 279
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6817182106453928e-05,
+      "loss": 0.4719,
+      "step": 280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6793128323870823e-05,
+      "loss": 0.4953,
+      "step": 281
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676900133552018e-05,
+      "loss": 0.4862,
+      "step": 282
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6744801401405138e-05,
+      "loss": 0.4943,
+      "step": 283
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6720528782314944e-05,
+      "loss": 0.4971,
+      "step": 284
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6696183739822108e-05,
+      "loss": 0.4861,
+      "step": 285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6671766536279624e-05,
+      "loss": 0.5098,
+      "step": 286
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6647277434818122e-05,
+      "loss": 0.4934,
+      "step": 287
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6622716699343032e-05,
+      "loss": 0.4783,
+      "step": 288
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659808459453175e-05,
+      "loss": 0.4965,
+      "step": 289
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6573381385830773e-05,
+      "loss": 0.4608,
+      "step": 290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6548607339452853e-05,
+      "loss": 0.4697,
+      "step": 291
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6523762722374108e-05,
+      "loss": 0.4767,
+      "step": 292
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6498847802331167e-05,
+      "loss": 0.4829,
+      "step": 293
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647386284781828e-05,
+      "loss": 0.4993,
+      "step": 294
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6448808128084398e-05,
+      "loss": 0.4845,
+      "step": 295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6423683913130323e-05,
+      "loss": 0.4736,
+      "step": 296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6398490473705742e-05,
+      "loss": 0.4979,
+      "step": 297
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.637322808130635e-05,
+      "loss": 0.4672,
+      "step": 298
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634789700817091e-05,
+      "loss": 0.4906,
+      "step": 299
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6322497527278308e-05,
+      "loss": 0.4894,
+      "step": 300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629702991234463e-05,
+      "loss": 0.5026,
+      "step": 301
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627149443782021e-05,
+      "loss": 0.4657,
+      "step": 302
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6245891378886655e-05,
+      "loss": 0.4702,
+      "step": 303
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6220221011453903e-05,
+      "loss": 0.4546,
+      "step": 304
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6194483612157232e-05,
+      "loss": 0.4795,
+      "step": 305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6168679458354284e-05,
+      "loss": 0.4767,
+      "step": 306
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6142808828122082e-05,
+      "loss": 0.4831,
+      "step": 307
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.611687200025403e-05,
+      "loss": 0.4859,
+      "step": 308
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6090869254256892e-05,
+      "loss": 0.446,
+      "step": 309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6064800870347816e-05,
+      "loss": 0.4865,
+      "step": 310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603866712945128e-05,
+      "loss": 0.4613,
+      "step": 311
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6012468313196086e-05,
+      "loss": 0.4775,
+      "step": 312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.59862047039123e-05,
+      "loss": 0.4814,
+      "step": 313
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5959876584628256e-05,
+      "loss": 0.4683,
+      "step": 314
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593348423906745e-05,
+      "loss": 0.4896,
+      "step": 315
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590702795164551e-05,
+      "loss": 0.4615,
+      "step": 316
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5880508007467158e-05,
+      "loss": 0.4533,
+      "step": 317
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.585392469232307e-05,
+      "loss": 0.4639,
+      "step": 318
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5827278292686868e-05,
+      "loss": 0.4761,
+      "step": 319
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5800569095711983e-05,
+      "loss": 0.4789,
+      "step": 320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5773797389228583e-05,
+      "loss": 0.4693,
+      "step": 321
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5746963461740464e-05,
+      "loss": 0.494,
+      "step": 322
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572006760242195e-05,
+      "loss": 0.4607,
+      "step": 323
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5693110101114763e-05,
+      "loss": 0.4728,
+      "step": 324
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5666091248324904e-05,
+      "loss": 0.4557,
+      "step": 325
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5639011335219537e-05,
+      "loss": 0.485,
+      "step": 326
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5611870653623826e-05,
+      "loss": 0.4765,
+      "step": 327
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.558466949601781e-05,
+      "loss": 0.4864,
+      "step": 328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5557408155533243e-05,
+      "loss": 0.4927,
+      "step": 329
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5530086925950435e-05,
+      "loss": 0.4567,
+      "step": 330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.550270610169509e-05,
+      "loss": 0.4232,
+      "step": 331
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547526597783513e-05,
+      "loss": 0.3976,
+      "step": 332
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5447766850077517e-05,
+      "loss": 0.4,
+      "step": 333
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542020901476506e-05,
+      "loss": 0.4187,
+      "step": 334
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5392592768873237e-05,
+      "loss": 0.4039,
+      "step": 335
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5364918410006967e-05,
+      "loss": 0.4147,
+      "step": 336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533718623639744e-05,
+      "loss": 0.3807,
+      "step": 337
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530939654689887e-05,
+      "loss": 0.414,
+      "step": 338
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5281549640985295e-05,
+      "loss": 0.4227,
+      "step": 339
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5253645818747333e-05,
+      "loss": 0.4077,
+      "step": 340
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5225685380888973e-05,
+      "loss": 0.3964,
+      "step": 341
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5197668628724302e-05,
+      "loss": 0.3985,
+      "step": 342
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.516959586417429e-05,
+      "loss": 0.401,
+      "step": 343
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5141467389763509e-05,
+      "loss": 0.4098,
+      "step": 344
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5113283508616895e-05,
+      "loss": 0.4055,
+      "step": 345
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5085044524456462e-05,
+      "loss": 0.4204,
+      "step": 346
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.505675074159805e-05,
+      "loss": 0.3954,
+      "step": 347
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5028402464948023e-05,
+      "loss": 0.4094,
+      "step": 348
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.3859,
+      "step": 349
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4971543652831562e-05,
+      "loss": 0.4026,
+      "step": 350
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4943033730100936e-05,
+      "loss": 0.4027,
+      "step": 351
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4914470539043713e-05,
+      "loss": 0.4298,
+      "step": 352
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4885854387469521e-05,
+      "loss": 0.4177,
+      "step": 353
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4857185583758722e-05,
+      "loss": 0.4222,
+      "step": 354
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4828464436859074e-05,
+      "loss": 0.4018,
+      "step": 355
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4799691256282414e-05,
+      "loss": 0.4178,
+      "step": 356
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4770866352101308e-05,
+      "loss": 0.4158,
+      "step": 357
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4741990034945728e-05,
+      "loss": 0.4007,
+      "step": 358
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4713062615999687e-05,
+      "loss": 0.4006,
+      "step": 359
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4684084406997903e-05,
+      "loss": 0.4393,
+      "step": 360
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4655055720222422e-05,
+      "loss": 0.4072,
+      "step": 361
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4625976868499266e-05,
+      "loss": 0.4215,
+      "step": 362
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4596848165195052e-05,
+      "loss": 0.4061,
+      "step": 363
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4567669924213623e-05,
+      "loss": 0.398,
+      "step": 364
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4538442459992661e-05,
+      "loss": 0.4085,
+      "step": 365
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4509166087500305e-05,
+      "loss": 0.3997,
+      "step": 366
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4479841122231735e-05,
+      "loss": 0.4034,
+      "step": 367
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4450467880205813e-05,
+      "loss": 0.393,
+      "step": 368
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4421046677961627e-05,
+      "loss": 0.4098,
+      "step": 369
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.439157783255513e-05,
+      "loss": 0.4156,
+      "step": 370
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4362061661555675e-05,
+      "loss": 0.3915,
+      "step": 371
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4332498483042639e-05,
+      "loss": 0.3921,
+      "step": 372
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.430288861560195e-05,
+      "loss": 0.405,
+      "step": 373
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4273232378322699e-05,
+      "loss": 0.4048,
+      "step": 374
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4243530090793667e-05,
+      "loss": 0.3986,
+      "step": 375
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.421378207309989e-05,
+      "loss": 0.3924,
+      "step": 376
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4183988645819223e-05,
+      "loss": 0.4012,
+      "step": 377
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4154150130018867e-05,
+      "loss": 0.4113,
+      "step": 378
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4124266847251917e-05,
+      "loss": 0.4122,
+      "step": 379
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.40943391195539e-05,
+      "loss": 0.4156,
+      "step": 380
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.40643672694393e-05,
+      "loss": 0.4101,
+      "step": 381
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4034351619898088e-05,
+      "loss": 0.4041,
+      "step": 382
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4004292494392227e-05,
+      "loss": 0.4066,
+      "step": 383
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3974190216852203e-05,
+      "loss": 0.3939,
+      "step": 384
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.394404511167353e-05,
+      "loss": 0.3974,
+      "step": 385
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3913857503713246e-05,
+      "loss": 0.3997,
+      "step": 386
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.388362771828642e-05,
+      "loss": 0.3927,
+      "step": 387
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3853356081162646e-05,
+      "loss": 0.3971,
+      "step": 388
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.382304291856253e-05,
+      "loss": 0.4073,
+      "step": 389
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3792688557154166e-05,
+      "loss": 0.3977,
+      "step": 390
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3762293324049645e-05,
+      "loss": 0.3857,
+      "step": 391
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3731857546801486e-05,
+      "loss": 0.4137,
+      "step": 392
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3701381553399147e-05,
+      "loss": 0.3973,
+      "step": 393
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3670865672265462e-05,
+      "loss": 0.3863,
+      "step": 394
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3640310232253122e-05,
+      "loss": 0.4055,
+      "step": 395
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3609715562641116e-05,
+      "loss": 0.3966,
+      "step": 396
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3579081993131187e-05,
+      "loss": 0.3977,
+      "step": 397
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3548409853844292e-05,
+      "loss": 0.3788,
+      "step": 398
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3517699475317016e-05,
+      "loss": 0.3862,
+      "step": 399
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3486951188498044e-05,
+      "loss": 0.4015,
+      "step": 400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3456165324744564e-05,
+      "loss": 0.388,
+      "step": 401
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3425342215818718e-05,
+      "loss": 0.3904,
+      "step": 402
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3394482193884013e-05,
+      "loss": 0.4096,
+      "step": 403
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3363585591501751e-05,
+      "loss": 0.3992,
+      "step": 404
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3332652741627445e-05,
+      "loss": 0.3957,
+      "step": 405
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3301683977607213e-05,
+      "loss": 0.3971,
+      "step": 406
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3270679633174219e-05,
+      "loss": 0.3995,
+      "step": 407
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3239640042445037e-05,
+      "loss": 0.403,
+      "step": 408
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3208565539916089e-05,
+      "loss": 0.407,
+      "step": 409
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3177456460460007e-05,
+      "loss": 0.4065,
+      "step": 410
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3146313139322051e-05,
+      "loss": 0.3925,
+      "step": 411
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3115135912116479e-05,
+      "loss": 0.4011,
+      "step": 412
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3083925114822938e-05,
+      "loss": 0.4089,
+      "step": 413
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3052681083782837e-05,
+      "loss": 0.3798,
+      "step": 414
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3021404155695728e-05,
+      "loss": 0.3958,
+      "step": 415
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2990094667615682e-05,
+      "loss": 0.4148,
+      "step": 416
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2958752956947645e-05,
+      "loss": 0.405,
+      "step": 417
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.292737936144381e-05,
+      "loss": 0.3856,
+      "step": 418
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2895974219199974e-05,
+      "loss": 0.3908,
+      "step": 419
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2864537868651894e-05,
+      "loss": 0.3928,
+      "step": 420
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2833070648571648e-05,
+      "loss": 0.4035,
+      "step": 421
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2801572898063971e-05,
+      "loss": 0.4084,
+      "step": 422
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2770044956562613e-05,
+      "loss": 0.4076,
+      "step": 423
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2738487163826671e-05,
+      "loss": 0.4072,
+      "step": 424
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.270689985993693e-05,
+      "loss": 0.4046,
+      "step": 425
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2675283385292212e-05,
+      "loss": 0.3917,
+      "step": 426
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2643638080605677e-05,
+      "loss": 0.3905,
+      "step": 427
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2611964286901195e-05,
+      "loss": 0.4112,
+      "step": 428
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2580262345509621e-05,
+      "loss": 0.4173,
+      "step": 429
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2548532598065164e-05,
+      "loss": 0.375,
+      "step": 430
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2516775386501662e-05,
+      "loss": 0.3846,
+      "step": 431
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.248499105304894e-05,
+      "loss": 0.4067,
+      "step": 432
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2453179940229086e-05,
+      "loss": 0.3915,
+      "step": 433
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2421342390852775e-05,
+      "loss": 0.4039,
+      "step": 434
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2389478748015584e-05,
+      "loss": 0.3867,
+      "step": 435
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2357589355094275e-05,
+      "loss": 0.3865,
+      "step": 436
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2325674555743106e-05,
+      "loss": 0.401,
+      "step": 437
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2293734693890131e-05,
+      "loss": 0.3851,
+      "step": 438
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2261770113733485e-05,
+      "loss": 0.399,
+      "step": 439
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2229781159737675e-05,
+      "loss": 0.4125,
+      "step": 440
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2197768176629876e-05,
+      "loss": 0.4072,
+      "step": 441
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2165731509396218e-05,
+      "loss": 0.391,
+      "step": 442
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2133671503278044e-05,
+      "loss": 0.4054,
+      "step": 443
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2101588503768224e-05,
+      "loss": 0.4015,
+      "step": 444
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2069482856607402e-05,
+      "loss": 0.3811,
+      "step": 445
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2037354907780299e-05,
+      "loss": 0.4115,
+      "step": 446
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2005205003511948e-05,
+      "loss": 0.3824,
+      "step": 447
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1973033490264e-05,
+      "loss": 0.3879,
+      "step": 448
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1940840714730972e-05,
+      "loss": 0.3956,
+      "step": 449
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1908627023836504e-05,
+      "loss": 0.391,
+      "step": 450
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1876392764729636e-05,
+      "loss": 0.413,
+      "step": 451
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1844138284781062e-05,
+      "loss": 0.4131,
+      "step": 452
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1811863931579377e-05,
+      "loss": 0.3758,
+      "step": 453
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1779570052927344e-05,
+      "loss": 0.3712,
+      "step": 454
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1747256996838145e-05,
+      "loss": 0.3989,
+      "step": 455
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1714925111531619e-05,
+      "loss": 0.3971,
+      "step": 456
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.168257474543052e-05,
+      "loss": 0.3915,
+      "step": 457
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1650206247156759e-05,
+      "loss": 0.3827,
+      "step": 458
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.161781996552765e-05,
+      "loss": 0.3795,
+      "step": 459
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1585416249552144e-05,
+      "loss": 0.3978,
+      "step": 460
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1552995448427073e-05,
+      "loss": 0.3913,
+      "step": 461
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.152055791153339e-05,
+      "loss": 0.3867,
+      "step": 462
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1488103988432402e-05,
+      "loss": 0.3812,
+      "step": 463
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1455634028861984e-05,
+      "loss": 0.3735,
+      "step": 464
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1423148382732854e-05,
+      "loss": 0.4087,
+      "step": 465
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.139064740012475e-05,
+      "loss": 0.3857,
+      "step": 466
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1358131431282702e-05,
+      "loss": 0.3779,
+      "step": 467
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1325600826613221e-05,
+      "loss": 0.4071,
+      "step": 468
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1293055936680557e-05,
+      "loss": 0.3963,
+      "step": 469
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1260497112202895e-05,
+      "loss": 0.3906,
+      "step": 470
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1227924704048585e-05,
+      "loss": 0.3746,
+      "step": 471
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1195339063232362e-05,
+      "loss": 0.4035,
+      "step": 472
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1162740540911562e-05,
+      "loss": 0.3853,
+      "step": 473
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1130129488382341e-05,
+      "loss": 0.4082,
+      "step": 474
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1097506257075878e-05,
+      "loss": 0.415,
+      "step": 475
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1064871198554605e-05,
+      "loss": 0.3598,
+      "step": 476
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1032224664508406e-05,
+      "loss": 0.3931,
+      "step": 477
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0999567006750824e-05,
+      "loss": 0.3834,
+      "step": 478
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.096689857721529e-05,
+      "loss": 0.3946,
+      "step": 479
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.09342197279513e-05,
+      "loss": 0.3923,
+      "step": 480
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0901530811120655e-05,
+      "loss": 0.3579,
+      "step": 481
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.086883217899363e-05,
+      "loss": 0.3984,
+      "step": 482
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0836124183945209e-05,
+      "loss": 0.3822,
+      "step": 483
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0803407178451273e-05,
+      "loss": 0.3771,
+      "step": 484
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0770681515084804e-05,
+      "loss": 0.4052,
+      "step": 485
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.073794754651208e-05,
+      "loss": 0.3939,
+      "step": 486
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0705205625488884e-05,
+      "loss": 0.4013,
+      "step": 487
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0672456104856697e-05,
+      "loss": 0.3831,
+      "step": 488
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0639699337538897e-05,
+      "loss": 0.3709,
+      "step": 489
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0606935676536956e-05,
+      "loss": 0.3991,
+      "step": 490
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0574165474926634e-05,
+      "loss": 0.3669,
+      "step": 491
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0541389085854177e-05,
+      "loss": 0.4039,
+      "step": 492
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.050860686253251e-05,
+      "loss": 0.3852,
+      "step": 493
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0475819158237426e-05,
+      "loss": 0.3965,
+      "step": 494
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0443026326303789e-05,
+      "loss": 0.3638,
+      "step": 495
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.041022872012172e-05,
+      "loss": 0.3857,
+      "step": 496
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0377426693132786e-05,
+      "loss": 0.3871,
+      "step": 497
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0344620598826199e-05,
+      "loss": 0.4076,
+      "step": 498
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0311810790734995e-05,
+      "loss": 0.3882,
+      "step": 499
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0278997622432245e-05,
+      "loss": 0.4068,
+      "step": 500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0246181447527213e-05,
+      "loss": 0.3751,
+      "step": 501
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0213362619661581e-05,
+      "loss": 0.3886,
+      "step": 502
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0180541492505605e-05,
+      "loss": 0.4093,
+      "step": 503
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0147718419754335e-05,
+      "loss": 0.3692,
+      "step": 504
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.011489375512377e-05,
+      "loss": 0.3914,
+      "step": 505
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0082067852347077e-05,
+      "loss": 0.3986,
+      "step": 506
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0049241065170765e-05,
+      "loss": 0.387,
+      "step": 507
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0016413747350866e-05,
+      "loss": 0.3784,
+      "step": 508
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.983586252649137e-06,
+      "loss": 0.3867,
+      "step": 509
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.950758934829242e-06,
+      "loss": 0.3898,
+      "step": 510
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.917932147652923e-06,
+      "loss": 0.3793,
+      "step": 511
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.885106244876233e-06,
+      "loss": 0.3681,
+      "step": 512
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.85228158024567e-06,
+      "loss": 0.3799,
+      "step": 513
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.819458507494395e-06,
+      "loss": 0.3972,
+      "step": 514
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.786637380338422e-06,
+      "loss": 0.3786,
+      "step": 515
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.75381855247279e-06,
+      "loss": 0.3758,
+      "step": 516
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.72100237756776e-06,
+      "loss": 0.3819,
+      "step": 517
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.688189209265005e-06,
+      "loss": 0.3961,
+      "step": 518
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.655379401173804e-06,
+      "loss": 0.3722,
+      "step": 519
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.622573306867219e-06,
+      "loss": 0.3787,
+      "step": 520
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.589771279878283e-06,
+      "loss": 0.3904,
+      "step": 521
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.556973673696214e-06,
+      "loss": 0.3908,
+      "step": 522
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.524180841762577e-06,
+      "loss": 0.384,
+      "step": 523
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.491393137467497e-06,
+      "loss": 0.3784,
+      "step": 524
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.458610914145826e-06,
+      "loss": 0.3703,
+      "step": 525
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.425834525073368e-06,
+      "loss": 0.386,
+      "step": 526
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.393064323463048e-06,
+      "loss": 0.3832,
+      "step": 527
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.360300662461104e-06,
+      "loss": 0.3772,
+      "step": 528
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.327543895143307e-06,
+      "loss": 0.3729,
+      "step": 529
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.29479437451112e-06,
+      "loss": 0.3669,
+      "step": 530
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.262052453487924e-06,
+      "loss": 0.371,
+      "step": 531
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.229318484915198e-06,
+      "loss": 0.3844,
+      "step": 532
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.196592821548729e-06,
+      "loss": 0.3775,
+      "step": 533
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.163875816054795e-06,
+      "loss": 0.3737,
+      "step": 534
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.131167821006372e-06,
+      "loss": 0.4066,
+      "step": 535
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.098469188879348e-06,
+      "loss": 0.3785,
+      "step": 536
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.065780272048703e-06,
+      "loss": 0.3725,
+      "step": 537
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.033101422784713e-06,
+      "loss": 0.3843,
+      "step": 538
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.000432993249176e-06,
+      "loss": 0.3697,
+      "step": 539
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.967775335491596e-06,
+      "loss": 0.3895,
+      "step": 540
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.935128801445399e-06,
+      "loss": 0.3821,
+      "step": 541
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.902493742924122e-06,
+      "loss": 0.3935,
+      "step": 542
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.869870511617662e-06,
+      "loss": 0.3931,
+      "step": 543
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.83725945908844e-06,
+      "loss": 0.3807,
+      "step": 544
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.804660936767643e-06,
+      "loss": 0.395,
+      "step": 545
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.772075295951416e-06,
+      "loss": 0.3625,
+      "step": 546
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.739502887797108e-06,
+      "loss": 0.3892,
+      "step": 547
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.706944063319445e-06,
+      "loss": 0.3807,
+      "step": 548
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.674399173386779e-06,
+      "loss": 0.3852,
+      "step": 549
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.641868568717301e-06,
+      "loss": 0.3778,
+      "step": 550
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.609352599875253e-06,
+      "loss": 0.3926,
+      "step": 551
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.576851617267151e-06,
+      "loss": 0.3875,
+      "step": 552
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.544365971138016e-06,
+      "loss": 0.3844,
+      "step": 553
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.511896011567602e-06,
+      "loss": 0.3959,
+      "step": 554
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.479442088466612e-06,
+      "loss": 0.3884,
+      "step": 555
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.447004551572929e-06,
+      "loss": 0.38,
+      "step": 556
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.41458375044786e-06,
+      "loss": 0.3888,
+      "step": 557
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.382180034472353e-06,
+      "loss": 0.3647,
+      "step": 558
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.349793752843246e-06,
+      "loss": 0.3894,
+      "step": 559
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.317425254569484e-06,
+      "loss": 0.3631,
+      "step": 560
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.285074888468385e-06,
+      "loss": 0.3663,
+      "step": 561
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.252743003161858e-06,
+      "loss": 0.3836,
+      "step": 562
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.22042994707266e-06,
+      "loss": 0.3755,
+      "step": 563
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.188136068420627e-06,
+      "loss": 0.3668,
+      "step": 564
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.155861715218943e-06,
+      "loss": 0.3849,
+      "step": 565
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.123607235270366e-06,
+      "loss": 0.3869,
+      "step": 566
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.091372976163496e-06,
+      "loss": 0.3704,
+      "step": 567
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.059159285269031e-06,
+      "loss": 0.3878,
+      "step": 568
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.026966509736001e-06,
+      "loss": 0.3881,
+      "step": 569
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.994794996488055e-06,
+      "loss": 0.3751,
+      "step": 570
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.962645092219705e-06,
+      "loss": 0.3835,
+      "step": 571
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.9305171433926e-06,
+      "loss": 0.3724,
+      "step": 572
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.898411496231781e-06,
+      "loss": 0.3851,
+      "step": 573
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.866328496721958e-06,
+      "loss": 0.3818,
+      "step": 574
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.834268490603785e-06,
+      "loss": 0.3707,
+      "step": 575
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.802231823370126e-06,
+      "loss": 0.3767,
+      "step": 576
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.77021884026233e-06,
+      "loss": 0.3818,
+      "step": 577
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.73822988626652e-06,
+      "loss": 0.3751,
+      "step": 578
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.706265306109872e-06,
+      "loss": 0.3693,
+      "step": 579
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.674325444256899e-06,
+      "loss": 0.3737,
+      "step": 580
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.642410644905726e-06,
+      "loss": 0.3822,
+      "step": 581
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.6105212519844195e-06,
+      "loss": 0.343,
+      "step": 582
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.578657609147226e-06,
+      "loss": 0.3694,
+      "step": 583
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.54682005977092e-06,
+      "loss": 0.3765,
+      "step": 584
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.51500894695106e-06,
+      "loss": 0.3702,
+      "step": 585
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.483224613498339e-06,
+      "loss": 0.3912,
+      "step": 586
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.45146740193484e-06,
+      "loss": 0.3745,
+      "step": 587
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.419737654490379e-06,
+      "loss": 0.3928,
+      "step": 588
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.388035713098807e-06,
+      "loss": 0.3672,
+      "step": 589
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.356361919394324e-06,
+      "loss": 0.3873,
+      "step": 590
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.324716614707794e-06,
+      "loss": 0.3746,
+      "step": 591
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2931001400630715e-06,
+      "loss": 0.3678,
+      "step": 592
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.261512836173332e-06,
+      "loss": 0.3769,
+      "step": 593
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.229955043437391e-06,
+      "loss": 0.3659,
+      "step": 594
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.198427101936029e-06,
+      "loss": 0.3772,
+      "step": 595
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.166929351428353e-06,
+      "loss": 0.3784,
+      "step": 596
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.135462131348107e-06,
+      "loss": 0.3749,
+      "step": 597
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.104025780800031e-06,
+      "loss": 0.3942,
+      "step": 598
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.072620638556193e-06,
+      "loss": 0.3779,
+      "step": 599
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.041247043052357e-06,
+      "loss": 0.389,
+      "step": 600
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.00990533238432e-06,
+      "loss": 0.3725,
+      "step": 601
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.978595844304272e-06,
+      "loss": 0.3539,
+      "step": 602
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.947318916217167e-06,
+      "loss": 0.3664,
+      "step": 603
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.916074885177066e-06,
+      "loss": 0.3756,
+      "step": 604
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.884864087883523e-06,
+      "loss": 0.3634,
+      "step": 605
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.853686860677949e-06,
+      "loss": 0.3672,
+      "step": 606
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.822543539539996e-06,
+      "loss": 0.3718,
+      "step": 607
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.791434460083916e-06,
+      "loss": 0.383,
+      "step": 608
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.760359957554964e-06,
+      "loss": 0.3726,
+      "step": 609
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.729320366825785e-06,
+      "loss": 0.3648,
+      "step": 610
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.698316022392789e-06,
+      "loss": 0.3683,
+      "step": 611
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.66734725837256e-06,
+      "loss": 0.3792,
+      "step": 612
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.636414408498249e-06,
+      "loss": 0.3738,
+      "step": 613
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6055178061159886e-06,
+      "loss": 0.3614,
+      "step": 614
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.574657784181287e-06,
+      "loss": 0.3561,
+      "step": 615
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.543834675255437e-06,
+      "loss": 0.3894,
+      "step": 616
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.513048811501959e-06,
+      "loss": 0.3706,
+      "step": 617
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.482300524682985e-06,
+      "loss": 0.3732,
+      "step": 618
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.451590146155714e-06,
+      "loss": 0.3701,
+      "step": 619
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.420918006868813e-06,
+      "loss": 0.3813,
+      "step": 620
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.390284437358889e-06,
+      "loss": 0.383,
+      "step": 621
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3596897677468824e-06,
+      "loss": 0.3802,
+      "step": 622
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.329134327734541e-06,
+      "loss": 0.3726,
+      "step": 623
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.298618446600856e-06,
+      "loss": 0.3855,
+      "step": 624
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.268142453198518e-06,
+      "loss": 0.3697,
+      "step": 625
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.237706675950359e-06,
+      "loss": 0.3552,
+      "step": 626
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.207311442845834e-06,
+      "loss": 0.3929,
+      "step": 627
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.176957081437474e-06,
+      "loss": 0.3748,
+      "step": 628
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.146643918837358e-06,
+      "loss": 0.3737,
+      "step": 629
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.116372281713581e-06,
+      "loss": 0.3613,
+      "step": 630
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.086142496286756e-06,
+      "loss": 0.379,
+      "step": 631
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.055954888326473e-06,
+      "loss": 0.3559,
+      "step": 632
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.025809783147803e-06,
+      "loss": 0.3638,
+      "step": 633
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.995707505607777e-06,
+      "loss": 0.3802,
+      "step": 634
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.965648380101916e-06,
+      "loss": 0.3563,
+      "step": 635
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.935632730560702e-06,
+      "loss": 0.3498,
+      "step": 636
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9056608804461005e-06,
+      "loss": 0.3893,
+      "step": 637
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.875733152748086e-06,
+      "loss": 0.3649,
+      "step": 638
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.845849869981137e-06,
+      "loss": 0.3785,
+      "step": 639
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.816011354180782e-06,
+      "loss": 0.3817,
+      "step": 640
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.78621792690011e-06,
+      "loss": 0.3811,
+      "step": 641
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7564699092063345e-06,
+      "loss": 0.3702,
+      "step": 642
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.726767621677305e-06,
+      "loss": 0.3859,
+      "step": 643
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.697111384398051e-06,
+      "loss": 0.3674,
+      "step": 644
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.667501516957365e-06,
+      "loss": 0.3675,
+      "step": 645
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.637938338444325e-06,
+      "loss": 0.3364,
+      "step": 646
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.608422167444877e-06,
+      "loss": 0.3819,
+      "step": 647
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.578953322038372e-06,
+      "loss": 0.366,
+      "step": 648
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.54953211979419e-06,
+      "loss": 0.3594,
+      "step": 649
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.520158877768265e-06,
+      "loss": 0.4048,
+      "step": 650
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.4908339124997005e-06,
+      "loss": 0.3804,
+      "step": 651
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4615575400073405e-06,
+      "loss": 0.372,
+      "step": 652
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.432330075786381e-06,
+      "loss": 0.3957,
+      "step": 653
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.403151834804951e-06,
+      "loss": 0.3722,
+      "step": 654
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.374023131500738e-06,
+      "loss": 0.3686,
+      "step": 655
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.344944279777579e-06,
+      "loss": 0.3591,
+      "step": 656
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3159155930021e-06,
+      "loss": 0.3797,
+      "step": 657
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.286937384000315e-06,
+      "loss": 0.3717,
+      "step": 658
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.258009965054276e-06,
+      "loss": 0.3903,
+      "step": 659
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.229133647898696e-06,
+      "loss": 0.3355,
+      "step": 660
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.20030874371759e-06,
+      "loss": 0.3172,
+      "step": 661
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.171535563140928e-06,
+      "loss": 0.3218,
+      "step": 662
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1428144162412815e-06,
+      "loss": 0.3293,
+      "step": 663
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.114145612530484e-06,
+      "loss": 0.3094,
+      "step": 664
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.085529460956293e-06,
+      "loss": 0.31,
+      "step": 665
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.05696626989907e-06,
+      "loss": 0.3165,
+      "step": 666
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.028456347168443e-06,
+      "loss": 0.3047,
+      "step": 667
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.3347,
+      "step": 668
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.971597535051977e-06,
+      "loss": 0.3227,
+      "step": 669
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.943249258401955e-06,
+      "loss": 0.3137,
+      "step": 670
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.914955475543543e-06,
+      "loss": 0.3068,
+      "step": 671
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.886716491383111e-06,
+      "loss": 0.3144,
+      "step": 672
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.858532610236491e-06,
+      "loss": 0.3109,
+      "step": 673
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8304041358257155e-06,
+      "loss": 0.2974,
+      "step": 674
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.802331371275703e-06,
+      "loss": 0.3105,
+      "step": 675
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.774314619111029e-06,
+      "loss": 0.3317,
+      "step": 676
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.746354181252668e-06,
+      "loss": 0.2933,
+      "step": 677
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.718450359014713e-06,
+      "loss": 0.3255,
+      "step": 678
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.690603453101134e-06,
+      "loss": 0.3239,
+      "step": 679
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.662813763602562e-06,
+      "loss": 0.288,
+      "step": 680
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.635081589993033e-06,
+      "loss": 0.306,
+      "step": 681
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.60740723112677e-06,
+      "loss": 0.2912,
+      "step": 682
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.579790985234939e-06,
+      "loss": 0.3014,
+      "step": 683
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.552233149922483e-06,
+      "loss": 0.3136,
+      "step": 684
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.524734022164871e-06,
+      "loss": 0.3185,
+      "step": 685
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.4972938983049156e-06,
+      "loss": 0.3091,
+      "step": 686
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.4699130740495675e-06,
+      "loss": 0.3196,
+      "step": 687
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.44259184446676e-06,
+      "loss": 0.3106,
+      "step": 688
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.415330503982193e-06,
+      "loss": 0.3072,
+      "step": 689
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.388129346376177e-06,
+      "loss": 0.3092,
+      "step": 690
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.360988664780465e-06,
+      "loss": 0.3253,
+      "step": 691
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3339087516750965e-06,
+      "loss": 0.3187,
+      "step": 692
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.306889898885241e-06,
+      "loss": 0.3034,
+      "step": 693
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.279932397578052e-06,
+      "loss": 0.3122,
+      "step": 694
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.253036538259538e-06,
+      "loss": 0.3145,
+      "step": 695
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.226202610771419e-06,
+      "loss": 0.3165,
+      "step": 696
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.19943090428802e-06,
+      "loss": 0.2856,
+      "step": 697
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1727217073131345e-06,
+      "loss": 0.3018,
+      "step": 698
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.146075307676932e-06,
+      "loss": 0.3038,
+      "step": 699
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.119491992532847e-06,
+      "loss": 0.3221,
+      "step": 700
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.092972048354491e-06,
+      "loss": 0.2965,
+      "step": 701
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0665157609325565e-06,
+      "loss": 0.3141,
+      "step": 702
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.040123415371748e-06,
+      "loss": 0.3103,
+      "step": 703
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.013795296087697e-06,
+      "loss": 0.3067,
+      "step": 704
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.98753168680392e-06,
+      "loss": 0.3146,
+      "step": 705
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.961332870548723e-06,
+      "loss": 0.2966,
+      "step": 706
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.935199129652187e-06,
+      "loss": 0.3119,
+      "step": 707
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.909130745743108e-06,
+      "loss": 0.3093,
+      "step": 708
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.883127999745977e-06,
+      "loss": 0.3174,
+      "step": 709
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.857191171877921e-06,
+      "loss": 0.3004,
+      "step": 710
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.831320541645717e-06,
+      "loss": 0.3032,
+      "step": 711
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8055163878427703e-06,
+      "loss": 0.3031,
+      "step": 712
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7797789885461023e-06,
+      "loss": 0.3068,
+      "step": 713
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7541086211133504e-06,
+      "loss": 0.3088,
+      "step": 714
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7285055621797938e-06,
+      "loss": 0.3236,
+      "step": 715
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7029700876553722e-06,
+      "loss": 0.2985,
+      "step": 716
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.677502472721699e-06,
+      "loss": 0.3168,
+      "step": 717
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.652102991829093e-06,
+      "loss": 0.3136,
+      "step": 718
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6267719186936502e-06,
+      "loss": 0.3248,
+      "step": 719
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6015095262942578e-06,
+      "loss": 0.3013,
+      "step": 720
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5763160868696823e-06,
+      "loss": 0.3071,
+      "step": 721
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5511918719156014e-06,
+      "loss": 0.3119,
+      "step": 722
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5261371521817247e-06,
+      "loss": 0.2995,
+      "step": 723
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5011521976688335e-06,
+      "loss": 0.3152,
+      "step": 724
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4762372776258945e-06,
+      "loss": 0.3068,
+      "step": 725
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4513926605471504e-06,
+      "loss": 0.306,
+      "step": 726
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.426618614169227e-06,
+      "loss": 0.3028,
+      "step": 727
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4019154054682503e-06,
+      "loss": 0.3176,
+      "step": 728
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.377283300656967e-06,
+      "loss": 0.2922,
+      "step": 729
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3527225651818784e-06,
+      "loss": 0.3081,
+      "step": 730
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.328233463720377e-06,
+      "loss": 0.2857,
+      "step": 731
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3038162601778944e-06,
+      "loss": 0.295,
+      "step": 732
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2794712176850597e-06,
+      "loss": 0.2973,
+      "step": 733
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.255198598594862e-06,
+      "loss": 0.3282,
+      "step": 734
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.230998664479823e-06,
+      "loss": 0.3017,
+      "step": 735
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.206871676129181e-06,
+      "loss": 0.2966,
+      "step": 736
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.182817893546075e-06,
+      "loss": 0.3024,
+      "step": 737
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.158837575944751e-06,
+      "loss": 0.3,
+      "step": 738
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1349309817477534e-06,
+      "loss": 0.3098,
+      "step": 739
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1110983685831718e-06,
+      "loss": 0.304,
+      "step": 740
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.087339993281816e-06,
+      "loss": 0.2929,
+      "step": 741
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0636561118744943e-06,
+      "loss": 0.3078,
+      "step": 742
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0400469795892274e-06,
+      "loss": 0.3125,
+      "step": 743
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0165128508485166e-06,
+      "loss": 0.2997,
+      "step": 744
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9930539792665767e-06,
+      "loss": 0.3265,
+      "step": 745
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.969670617646626e-06,
+      "loss": 0.317,
+      "step": 746
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.946363017978159e-06,
+      "loss": 0.3157,
+      "step": 747
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9231314314342185e-06,
+      "loss": 0.3188,
+      "step": 748
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.899976108368695e-06,
+      "loss": 0.2959,
+      "step": 749
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.87689729831363e-06,
+      "loss": 0.3046,
+      "step": 750
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.853895249976537e-06,
+      "loss": 0.3204,
+      "step": 751
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8309702112377034e-06,
+      "loss": 0.3122,
+      "step": 752
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8081224291475216e-06,
+      "loss": 0.3064,
+      "step": 753
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7853521499238377e-06,
+      "loss": 0.2813,
+      "step": 754
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7626596189492983e-06,
+      "loss": 0.2939,
+      "step": 755
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.740045080768694e-06,
+      "loss": 0.313,
+      "step": 756
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7175087790863232e-06,
+      "loss": 0.3233,
+      "step": 757
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.69505095676339e-06,
+      "loss": 0.3289,
+      "step": 758
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.672671855815355e-06,
+      "loss": 0.3157,
+      "step": 759
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6503717174093548e-06,
+      "loss": 0.3013,
+      "step": 760
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6281507818615737e-06,
+      "loss": 0.3033,
+      "step": 761
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.606009288634689e-06,
+      "loss": 0.3171,
+      "step": 762
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5839474763352635e-06,
+      "loss": 0.3056,
+      "step": 763
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5619655827111812e-06,
+      "loss": 0.3011,
+      "step": 764
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.540063844649092e-06,
+      "loss": 0.3268,
+      "step": 765
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5182424981718512e-06,
+      "loss": 0.3108,
+      "step": 766
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.496501778435977e-06,
+      "loss": 0.3168,
+      "step": 767
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.474841919729122e-06,
+      "loss": 0.3043,
+      "step": 768
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.453263155467541e-06,
+      "loss": 0.2806,
+      "step": 769
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4317657181935827e-06,
+      "loss": 0.3154,
+      "step": 770
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.410349839573175e-06,
+      "loss": 0.2989,
+      "step": 771
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3890157503933376e-06,
+      "loss": 0.3193,
+      "step": 772
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3677636805596905e-06,
+      "loss": 0.3077,
+      "step": 773
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.346593859093974e-06,
+      "loss": 0.3018,
+      "step": 774
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.325506514131588e-06,
+      "loss": 0.2842,
+      "step": 775
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3045018729191236e-06,
+      "loss": 0.3075,
+      "step": 776
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2835801618119247e-06,
+      "loss": 0.3136,
+      "step": 777
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2627416062716366e-06,
+      "loss": 0.3104,
+      "step": 778
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.241986430863796e-06,
+      "loss": 0.3358,
+      "step": 779
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2213148592553847e-06,
+      "loss": 0.3122,
+      "step": 780
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2007271142124374e-06,
+      "loss": 0.3021,
+      "step": 781
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1802234175976377e-06,
+      "loss": 0.2994,
+      "step": 782
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.159803990367931e-06,
+      "loss": 0.2944,
+      "step": 783
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1394690525721275e-06,
+      "loss": 0.2988,
+      "step": 784
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1192188233485458e-06,
+      "loss": 0.2946,
+      "step": 785
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0990535209226547e-06,
+      "loss": 0.3073,
+      "step": 786
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0789733626047083e-06,
+      "loss": 0.3033,
+      "step": 787
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.058978564787404e-06,
+      "loss": 0.3297,
+      "step": 788
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0390693429435626e-06,
+      "loss": 0.302,
+      "step": 789
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.019245911623805e-06,
+      "loss": 0.3211,
+      "step": 790
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9995084844542324e-06,
+      "loss": 0.3076,
+      "step": 791
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.979857274134115e-06,
+      "loss": 0.2997,
+      "step": 792
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.960292492433631e-06,
+      "loss": 0.3092,
+      "step": 793
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9408143501915534e-06,
+      "loss": 0.3047,
+      "step": 794
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9214230573129944e-06,
+      "loss": 0.3043,
+      "step": 795
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9021188227671318e-06,
+      "loss": 0.3086,
+      "step": 796
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8829018545849753e-06,
+      "loss": 0.3133,
+      "step": 797
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8637723598571078e-06,
+      "loss": 0.3071,
+      "step": 798
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8447305447314578e-06,
+      "loss": 0.3038,
+      "step": 799
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8257766144110823e-06,
+      "loss": 0.3133,
+      "step": 800
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8069107731519507e-06,
+      "loss": 0.2934,
+      "step": 801
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7881332242607463e-06,
+      "loss": 0.3139,
+      "step": 802
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7694441700926767e-06,
+      "loss": 0.2906,
+      "step": 803
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7508438120492864e-06,
+      "loss": 0.3054,
+      "step": 804
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7323323505762958e-06,
+      "loss": 0.302,
+      "step": 805
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7139099851614328e-06,
+      "loss": 0.299,
+      "step": 806
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6955769143322898e-06,
+      "loss": 0.3005,
+      "step": 807
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6773333356541787e-06,
+      "loss": 0.3101,
+      "step": 808
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.659179445728004e-06,
+      "loss": 0.2918,
+      "step": 809
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.641115440188148e-06,
+      "loss": 0.3078,
+      "step": 810
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6231415137003536e-06,
+      "loss": 0.3078,
+      "step": 811
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6052578599596358e-06,
+      "loss": 0.3194,
+      "step": 812
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.587464671688187e-06,
+      "loss": 0.3039,
+      "step": 813
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.569762140633313e-06,
+      "loss": 0.3024,
+      "step": 814
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5521504575653412e-06,
+      "loss": 0.307,
+      "step": 815
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5346298122755932e-06,
+      "loss": 0.3003,
+      "step": 816
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5172003935743206e-06,
+      "loss": 0.3089,
+      "step": 817
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4998623892886877e-06,
+      "loss": 0.3124,
+      "step": 818
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4826159862607182e-06,
+      "loss": 0.3094,
+      "step": 819
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4654613703453125e-06,
+      "loss": 0.3201,
+      "step": 820
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4483987264082333e-06,
+      "loss": 0.3076,
+      "step": 821
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4314282383241097e-06,
+      "loss": 0.297,
+      "step": 822
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4145500889744523e-06,
+      "loss": 0.3084,
+      "step": 823
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3977644602456942e-06,
+      "loss": 0.3017,
+      "step": 824
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3810715330272285e-06,
+      "loss": 0.2972,
+      "step": 825
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3644714872094533e-06,
+      "loss": 0.2925,
+      "step": 826
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3479645016818265e-06,
+      "loss": 0.2947,
+      "step": 827
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3315507543309624e-06,
+      "loss": 0.2866,
+      "step": 828
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.315230422038688e-06,
+      "loss": 0.3154,
+      "step": 829
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2990036806801543e-06,
+      "loss": 0.3079,
+      "step": 830
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2828707051219257e-06,
+      "loss": 0.3153,
+      "step": 831
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2668316692201188e-06,
+      "loss": 0.322,
+      "step": 832
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2508867458185037e-06,
+      "loss": 0.2948,
+      "step": 833
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2350361067466554e-06,
+      "loss": 0.3029,
+      "step": 834
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2192799228181008e-06,
+      "loss": 0.314,
+      "step": 835
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.203618363828475e-06,
+      "loss": 0.3105,
+      "step": 836
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1880515985536911e-06,
+      "loss": 0.3033,
+      "step": 837
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1725797947481266e-06,
+      "loss": 0.2881,
+      "step": 838
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.157203119142809e-06,
+      "loss": 0.3216,
+      "step": 839
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1419217374436231e-06,
+      "loss": 0.304,
+      "step": 840
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1267358143295249e-06,
+      "loss": 0.3115,
+      "step": 841
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1116455134507665e-06,
+      "loss": 0.2995,
+      "step": 842
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0966509974271289e-06,
+      "loss": 0.3216,
+      "step": 843
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0817524278461777e-06,
+      "loss": 0.3245,
+      "step": 844
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0669499652615144e-06,
+      "loss": 0.3232,
+      "step": 845
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0522437691910493e-06,
+      "loss": 0.3004,
+      "step": 846
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0376339981152828e-06,
+      "loss": 0.3095,
+      "step": 847
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.023120809475594e-06,
+      "loss": 0.3061,
+      "step": 848
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.008704359672551e-06,
+      "loss": 0.3117,
+      "step": 849
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.943848040642168e-07,
+      "loss": 0.302,
+      "step": 850
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.801622969644842e-07,
+      "loss": 0.3002,
+      "step": 851
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.660369916414015e-07,
+      "loss": 0.3066,
+      "step": 852
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.520090403155391e-07,
+      "loss": 0.2911,
+      "step": 853
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.380785941583237e-07,
+      "loss": 0.3088,
+      "step": 854
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.242458032904311e-07,
+      "loss": 0.2985,
+      "step": 855
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.105108167801568e-07,
+      "loss": 0.2983,
+      "step": 856
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.968737826418206e-07,
+      "loss": 0.3039,
+      "step": 857
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.833348478341519e-07,
+      "loss": 0.3108,
+      "step": 858
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.698941582587261e-07,
+      "loss": 0.2923,
+      "step": 859
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.565518587583865e-07,
+      "loss": 0.2967,
+      "step": 860
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.433080931156767e-07,
+      "loss": 0.3037,
+      "step": 861
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.301630040512931e-07,
+      "loss": 0.2957,
+      "step": 862
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.171167332225538e-07,
+      "loss": 0.3011,
+      "step": 863
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.041694212218698e-07,
+      "loss": 0.3037,
+      "step": 864
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.913212075752241e-07,
+      "loss": 0.2929,
+      "step": 865
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.785722307406685e-07,
+      "loss": 0.3015,
+      "step": 866
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.659226281068444e-07,
+      "loss": 0.3011,
+      "step": 867
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.533725359914856e-07,
+      "loss": 0.3168,
+      "step": 868
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.409220896399582e-07,
+      "loss": 0.3053,
+      "step": 869
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.285714232237972e-07,
+      "loss": 0.295,
+      "step": 870
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.163206698392744e-07,
+      "loss": 0.3091,
+      "step": 871
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.04169961505945e-07,
+      "loss": 0.2892,
+      "step": 872
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.9211942916524e-07,
+      "loss": 0.3035,
+      "step": 873
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.801692026790462e-07,
+      "loss": 0.3,
+      "step": 874
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.683194108283109e-07,
+      "loss": 0.3107,
+      "step": 875
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.565701813116543e-07,
+      "loss": 0.291,
+      "step": 876
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.449216407439906e-07,
+      "loss": 0.3084,
+      "step": 877
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.333739146551665e-07,
+      "loss": 0.3084,
+      "step": 878
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.219271274886052e-07,
+      "loss": 0.2949,
+      "step": 879
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.105814025999701e-07,
+      "loss": 0.2983,
+      "step": 880
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.993368622558304e-07,
+      "loss": 0.2997,
+      "step": 881
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.881936276323463e-07,
+      "loss": 0.3267,
+      "step": 882
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.771518188139624e-07,
+      "loss": 0.291,
+      "step": 883
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.662115547921144e-07,
+      "loss": 0.28,
+      "step": 884
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.553729534639463e-07,
+      "loss": 0.3076,
+      "step": 885
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.446361316310389e-07,
+      "loss": 0.3108,
+      "step": 886
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.340012049981491e-07,
+      "loss": 0.2943,
+      "step": 887
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.234682881719766e-07,
+      "loss": 0.3076,
+      "step": 888
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.130374946599037e-07,
+      "loss": 0.3124,
+      "step": 889
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.027089368687965e-07,
+      "loss": 0.306,
+      "step": 890
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.924827261037779e-07,
+      "loss": 0.296,
+      "step": 891
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.823589725670419e-07,
+      "loss": 0.2993,
+      "step": 892
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.723377853566469e-07,
+      "loss": 0.2958,
+      "step": 893
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.6241927246535645e-07,
+      "loss": 0.2989,
+      "step": 894
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5260354077947486e-07,
+      "loss": 0.3176,
+      "step": 895
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.428906960776835e-07,
+      "loss": 0.3214,
+      "step": 896
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.332808430299085e-07,
+      "loss": 0.2984,
+      "step": 897
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.237740851961947e-07,
+      "loss": 0.3044,
+      "step": 898
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1437052502558693e-07,
+      "loss": 0.2923,
+      "step": 899
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0507026385502747e-07,
+      "loss": 0.302,
+      "step": 900
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.958734019082555e-07,
+      "loss": 0.2997,
+      "step": 901
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.867800382947451e-07,
+      "loss": 0.3162,
+      "step": 902
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7779027100861785e-07,
+      "loss": 0.3017,
+      "step": 903
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.689041969275997e-07,
+      "loss": 0.2917,
+      "step": 904
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.601219118119681e-07,
+      "loss": 0.2976,
+      "step": 905
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5144351030353077e-07,
+      "loss": 0.2814,
+      "step": 906
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4286908592459535e-07,
+      "loss": 0.3118,
+      "step": 907
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.3439873107696695e-07,
+      "loss": 0.3026,
+      "step": 908
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.260325370409501e-07,
+      "loss": 0.3153,
+      "step": 909
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.1777059397436693e-07,
+      "loss": 0.3141,
+      "step": 910
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.0961299091158636e-07,
+      "loss": 0.3118,
+      "step": 911
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0155981576255986e-07,
+      "loss": 0.316,
+      "step": 912
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9361115531188145e-07,
+      "loss": 0.3081,
+      "step": 913
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.8576709521784374e-07,
+      "loss": 0.3069,
+      "step": 914
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.7802772001152224e-07,
+      "loss": 0.3072,
+      "step": 915
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.703931130958626e-07,
+      "loss": 0.3084,
+      "step": 916
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.628633567447769e-07,
+      "loss": 0.3084,
+      "step": 917
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.5543853210226566e-07,
+      "loss": 0.313,
+      "step": 918
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.48118719181536e-07,
+      "loss": 0.2964,
+      "step": 919
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4090399686414135e-07,
+      "loss": 0.2937,
+      "step": 920
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3379444289913344e-07,
+      "loss": 0.3243,
+      "step": 921
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2679013390222383e-07,
+      "loss": 0.2996,
+      "step": 922
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1989114535495793e-07,
+      "loss": 0.2971,
+      "step": 923
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.13097551603898e-07,
+      "loss": 0.3011,
+      "step": 924
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0640942585982816e-07,
+      "loss": 0.3128,
+      "step": 925
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9982684019696053e-07,
+      "loss": 0.3033,
+      "step": 926
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9334986555216374e-07,
+      "loss": 0.3119,
+      "step": 927
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.869785717241901e-07,
+      "loss": 0.2998,
+      "step": 928
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8071302737293294e-07,
+      "loss": 0.2948,
+      "step": 929
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7455330001868054e-07,
+      "loss": 0.3181,
+      "step": 930
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6849945604138997e-07,
+      "loss": 0.314,
+      "step": 931
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6255156067997325e-07,
+      "loss": 0.2938,
+      "step": 932
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.567096780315891e-07,
+      "loss": 0.3021,
+      "step": 933
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.509738710509634e-07,
+      "loss": 0.2966,
+      "step": 934
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4534420154969754e-07,
+      "loss": 0.3087,
+      "step": 935
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3982073019560782e-07,
+      "loss": 0.3098,
+      "step": 936
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3440351651207828e-07,
+      "loss": 0.3155,
+      "step": 937
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.2909261887740664e-07,
+      "loss": 0.3154,
+      "step": 938
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2388809452418716e-07,
+      "loss": 0.3032,
+      "step": 939
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1878999953868098e-07,
+      "loss": 0.306,
+      "step": 940
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1379838886022787e-07,
+      "loss": 0.312,
+      "step": 941
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0891331628063884e-07,
+      "loss": 0.3194,
+      "step": 942
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.041348344436277e-07,
+      "loss": 0.3241,
+      "step": 943
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.946299484423272e-08,
+      "loss": 0.3021,
+      "step": 944
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.489784782827582e-08,
+      "loss": 0.3036,
+      "step": 945
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.04394425918076e-08,
+      "loss": 0.2876,
+      "step": 946
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.608782718058207e-08,
+      "loss": 0.297,
+      "step": 947
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.18430484895405e-08,
+      "loss": 0.3099,
+      "step": 948
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.77051522623018e-08,
+      "loss": 0.3043,
+      "step": 949
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.367418309067398e-08,
+      "loss": 0.2954,
+      "step": 950
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.975018441417126e-08,
+      "loss": 0.3107,
+      "step": 951
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.593319851954549e-08,
+      "loss": 0.306,
+      "step": 952
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.222326654033106e-08,
+      "loss": 0.2865,
+      "step": 953
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.862042845640403e-08,
+      "loss": 0.2995,
+      "step": 954
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5124723093546994e-08,
+      "loss": 0.3023,
+      "step": 955
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.1736188123032715e-08,
+      "loss": 0.2969,
+      "step": 956
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.845486006121891e-08,
+      "loss": 0.2908,
+      "step": 957
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.528077426915412e-08,
+      "loss": 0.3072,
+      "step": 958
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.221396495219687e-08,
+      "loss": 0.293,
+      "step": 959
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.925446515964604e-08,
+      "loss": 0.3085,
+      "step": 960
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.64023067843855e-08,
+      "loss": 0.2931,
+      "step": 961
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3657520562541125e-08,
+      "loss": 0.2933,
+      "step": 962
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.1020136073146575e-08,
+      "loss": 0.2884,
+      "step": 963
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.8490181737830247e-08,
+      "loss": 0.3083,
+      "step": 964
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.606768482050215e-08,
+      "loss": 0.2887,
+      "step": 965
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.3752671427065276e-08,
+      "loss": 0.2973,
+      "step": 966
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.1545166505132497e-08,
+      "loss": 0.3066,
+      "step": 967
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.944519384375676e-08,
+      "loss": 0.2982,
+      "step": 968
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7452776073175748e-08,
+      "loss": 0.3079,
+      "step": 969
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.556793466456874e-08,
+      "loss": 0.2929,
+      "step": 970
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.3790689929823464e-08,
+      "loss": 0.3081,
+      "step": 971
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2121061021318492e-08,
+      "loss": 0.3178,
+      "step": 972
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0559065931716738e-08,
+      "loss": 0.2997,
+      "step": 973
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.104721493770063e-09,
+      "loss": 0.292,
+      "step": 974
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.758043380140523e-09,
+      "loss": 0.2966,
+      "step": 975
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.5190461032305085e-09,
+      "loss": 0.3,
+      "step": 976
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.387743015022873e-09,
+      "loss": 0.3035,
+      "step": 977
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.364146306943262e-09,
+      "loss": 0.2918,
+      "step": 978
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.4482670097246707e-09,
+      "loss": 0.2997,
+      "step": 979
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.6401149932875346e-09,
+      "loss": 0.289,
+      "step": 980
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9396989666398137e-09,
+      "loss": 0.3049,
+      "step": 981
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3470264777726283e-09,
+      "loss": 0.3086,
+      "step": 982
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.621039135892073e-10,
+      "loss": 0.306,
+      "step": 983
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.849364998305017e-10,
+      "loss": 0.2971,
+      "step": 984
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1552830101967402e-10,
+      "loss": 0.3283,
+      "step": 985
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.388222041990965e-11,
+      "loss": 0.3053,
+      "step": 986
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0,
+      "loss": 0.2978,
+      "step": 987
+    },
+    {
+      "epoch": 2.99,
+      "step": 987,
+      "total_flos": 2.513466573025444e+18,
+      "train_loss": 0.3876540904530638,
+      "train_runtime": 48412.9608,
+      "train_samples_per_second": 5.231,
+      "train_steps_per_second": 0.02
+    }
+  ],
+  "max_steps": 987,
+  "num_train_epochs": 3,
+  "total_flos": 2.513466573025444e+18,
+  "trial_name": null,
+  "trial_params": null
+}