diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14136 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9997873697639804,
+  "eval_steps": 500,
+  "global_step": 2351,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4084507042253521e-06,
+      "loss": 2.9236,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8169014084507042e-06,
+      "loss": 2.5815,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.225352112676056e-06,
+      "loss": 2.866,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.6338028169014084e-06,
+      "loss": 2.7915,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.042253521126762e-06,
+      "loss": 2.1041,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.450704225352112e-06,
+      "loss": 1.8662,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.859154929577465e-06,
+      "loss": 2.1697,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1267605633802817e-05,
+      "loss": 2.0254,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.267605633802817e-05,
+      "loss": 2.0267,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4084507042253523e-05,
+      "loss": 1.9396,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5492957746478872e-05,
+      "loss": 1.9551,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6901408450704224e-05,
+      "loss": 1.818,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.830985915492958e-05,
+      "loss": 1.8455,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.971830985915493e-05,
+      "loss": 1.7354,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.112676056338028e-05,
+      "loss": 1.8837,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2535211267605634e-05,
+      "loss": 1.6877,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3943661971830986e-05,
+      "loss": 1.82,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.535211267605634e-05,
+      "loss": 1.8542,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.676056338028169e-05,
+      "loss": 1.6422,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8169014084507046e-05,
+      "loss": 1.789,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9577464788732395e-05,
+      "loss": 1.733,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0985915492957744e-05,
+      "loss": 1.6455,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.23943661971831e-05,
+      "loss": 1.8142,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.380281690140845e-05,
+      "loss": 1.7823,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5211267605633805e-05,
+      "loss": 1.5669,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.661971830985916e-05,
+      "loss": 1.8266,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.802816901408451e-05,
+      "loss": 1.6341,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.943661971830986e-05,
+      "loss": 1.7732,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0845070422535214e-05,
+      "loss": 1.5955,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.225352112676056e-05,
+      "loss": 1.7235,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.366197183098591e-05,
+      "loss": 1.5932,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.507042253521127e-05,
+      "loss": 1.6787,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.647887323943662e-05,
+      "loss": 1.6307,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.788732394366197e-05,
+      "loss": 1.8137,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.929577464788733e-05,
+      "loss": 1.7223,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.070422535211268e-05,
+      "loss": 1.6434,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2112676056338026e-05,
+      "loss": 1.6211,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.352112676056338e-05,
+      "loss": 2.0622,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.492957746478874e-05,
+      "loss": 2.1427,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.633802816901409e-05,
+      "loss": 1.633,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.774647887323944e-05,
+      "loss": 1.8952,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.915492957746479e-05,
+      "loss": 1.689,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.056338028169014e-05,
+      "loss": 1.5826,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.197183098591549e-05,
+      "loss": 1.934,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.338028169014085e-05,
+      "loss": 1.8921,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.47887323943662e-05,
+      "loss": 4.5167,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.619718309859155e-05,
+      "loss": 4.2504,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.76056338028169e-05,
+      "loss": 2.2087,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.901408450704226e-05,
+      "loss": 3.4768,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.042253521126761e-05,
+      "loss": 2.4177,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.183098591549297e-05,
+      "loss": 3.2291,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.323943661971832e-05,
+      "loss": 2.0211,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.464788732394367e-05,
+      "loss": 3.1958,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.605633802816902e-05,
+      "loss": 2.0205,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.746478873239437e-05,
+      "loss": 2.5708,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.887323943661972e-05,
+      "loss": 2.6817,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.028169014084508e-05,
+      "loss": 10.6328,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.169014084507043e-05,
+      "loss": 2.2963,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.309859154929578e-05,
+      "loss": 8.3243,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.450704225352113e-05,
+      "loss": 4.7019,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.591549295774647e-05,
+      "loss": 4.5669,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.732394366197182e-05,
+      "loss": 4.1873,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.873239436619719e-05,
+      "loss": 3.1368,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.014084507042254e-05,
+      "loss": 3.6138,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.15492957746479e-05,
+      "loss": 2.8718,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.295774647887325e-05,
+      "loss": 2.709,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.43661971830986e-05,
+      "loss": 3.0757,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.577464788732394e-05,
+      "loss": 3.5456,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.718309859154931e-05,
+      "loss": 2.5287,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.859154929577466e-05,
+      "loss": 2.8534,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001,
+      "loss": 2.5168,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999995253538186e-05,
+      "loss": 3.1577,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999981014161752e-05,
+      "loss": 2.4658,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999957281897735e-05,
+      "loss": 3.0868,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999924056791192e-05,
+      "loss": 2.5396,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999881338905204e-05,
+      "loss": 2.903,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999829128320874e-05,
+      "loss": 2.7481,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999767425137328e-05,
+      "loss": 2.3053,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999696229471716e-05,
+      "loss": 2.1276,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999615541459207e-05,
+      "loss": 2.8166,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999525361252996e-05,
+      "loss": 2.228,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999425689024297e-05,
+      "loss": 2.6047,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999316524962345e-05,
+      "loss": 2.625,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999197869274399e-05,
+      "loss": 2.0726,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999069722185737e-05,
+      "loss": 2.3532,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998932083939656e-05,
+      "loss": 2.0942,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998784954797474e-05,
+      "loss": 2.3008,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998628335038528e-05,
+      "loss": 2.1689,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998462224960175e-05,
+      "loss": 2.346,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998286624877786e-05,
+      "loss": 2.0874,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998101535124758e-05,
+      "loss": 2.1654,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997906956052494e-05,
+      "loss": 1.9388,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997702888030423e-05,
+      "loss": 2.2317,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997489331445984e-05,
+      "loss": 1.9052,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997266286704631e-05,
+      "loss": 2.1824,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997033754229835e-05,
+      "loss": 1.8819,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996791734463077e-05,
+      "loss": 2.1797,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996540227863854e-05,
+      "loss": 2.1679,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996279234909671e-05,
+      "loss": 2.0811,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996008756096046e-05,
+      "loss": 1.9813,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995728791936504e-05,
+      "loss": 1.9379,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995439342962586e-05,
+      "loss": 1.9665,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99514040972383e-05,
+      "loss": 1.9134,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994831992787788e-05,
+      "loss": 1.93,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994514092740015e-05,
+      "loss": 2.034,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994186710184074e-05,
+      "loss": 1.8691,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993849845741524e-05,
+      "loss": 2.1198,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993503500051935e-05,
+      "loss": 1.8622,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99314767377287e-05,
+      "loss": 1.9321,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992782367579899e-05,
+      "loss": 1.9011,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992407582166581e-05,
+      "loss": 1.9676,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992023318244485e-05,
+      "loss": 1.9876,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991629576543163e-05,
+      "loss": 1.8557,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99122635781017e-05,
+      "loss": 1.9544,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990813662811051e-05,
+      "loss": 1.9241,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990391492329341e-05,
+      "loss": 1.9475,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989959847166567e-05,
+      "loss": 1.7656,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989518728142244e-05,
+      "loss": 1.9603,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989068136093873e-05,
+      "loss": 1.7672,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988608071876942e-05,
+      "loss": 1.8507,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988138536364922e-05,
+      "loss": 1.8995,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987659530449268e-05,
+      "loss": 1.834,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987171055039408e-05,
+      "loss": 1.7935,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986673111062759e-05,
+      "loss": 1.9216,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986165699464705e-05,
+      "loss": 1.7687,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985648821208616e-05,
+      "loss": 1.9045,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985122477275824e-05,
+      "loss": 1.7723,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98458666866564e-05,
+      "loss": 2.0124,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984041396395343e-05,
+      "loss": 1.8293,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983486661500174e-05,
+      "loss": 1.8396,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98292246503335e-05,
+      "loss": 1.8068,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982348808066042e-05,
+      "loss": 1.6738,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981765691687388e-05,
+      "loss": 1.9001,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981173117004484e-05,
+      "loss": 1.8076,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980571085142381e-05,
+      "loss": 1.8295,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979959597244088e-05,
+      "loss": 1.7842,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979338654470569e-05,
+      "loss": 1.9185,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978708258000737e-05,
+      "loss": 1.7716,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978068409031449e-05,
+      "loss": 1.8532,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977419108777514e-05,
+      "loss": 1.651,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976760358471686e-05,
+      "loss": 1.7622,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976092159364656e-05,
+      "loss": 1.7809,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975414512725057e-05,
+      "loss": 1.8399,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974727419839459e-05,
+      "loss": 1.7889,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974030882012367e-05,
+      "loss": 1.8085,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973324900566213e-05,
+      "loss": 1.9424,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972609476841367e-05,
+      "loss": 1.7412,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97188461219612e-05,
+      "loss": 1.8473,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97115030800669e-05,
+      "loss": 1.8044,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970406565667212e-05,
+      "loss": 1.7999,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969653386589748e-05,
+      "loss": 1.7142,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968890772204271e-05,
+      "loss": 1.8592,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.968118723958668e-05,
+      "loss": 1.7114,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967337243318739e-05,
+      "loss": 1.8653,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966546331768191e-05,
+      "loss": 1.7548,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965745990808637e-05,
+      "loss": 1.8328,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96493622195959e-05,
+      "loss": 1.8734,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96411702675847e-05,
+      "loss": 1.7676,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963288406760582e-05,
+      "loss": 1.7316,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962450363539136e-05,
+      "loss": 1.8106,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961602898685226e-05,
+      "loss": 1.6182,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960746013807835e-05,
+      "loss": 1.7275,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959879710533835e-05,
+      "loss": 1.7654,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959003990507972e-05,
+      "loss": 1.7455,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958118855392876e-05,
+      "loss": 1.7166,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957224306869053e-05,
+      "loss": 1.8019,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956320346634876e-05,
+      "loss": 1.7263,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955406976406593e-05,
+      "loss": 1.6946,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954484197918315e-05,
+      "loss": 1.8705,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953552012922012e-05,
+      "loss": 1.7575,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952610423187516e-05,
+      "loss": 1.7242,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95165943050252e-05,
+      "loss": 1.7863,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950699036672559e-05,
+      "loss": 1.7971,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949729243521024e-05,
+      "loss": 1.6965,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94875005288915e-05,
+      "loss": 1.7734,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947761466636014e-05,
+      "loss": 1.9362,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946763486638528e-05,
+      "loss": 1.7111,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945756114791445e-05,
+      "loss": 1.7852,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944739353007344e-05,
+      "loss": 1.7385,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943713203216632e-05,
+      "loss": 1.6909,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942677667367541e-05,
+      "loss": 1.7783,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941632747426129e-05,
+      "loss": 1.7274,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940578445376258e-05,
+      "loss": 1.6975,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939514763219615e-05,
+      "loss": 1.7308,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938441702975689e-05,
+      "loss": 1.7851,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937359266681775e-05,
+      "loss": 1.7223,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936267456392971e-05,
+      "loss": 1.5831,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93516627418217e-05,
+      "loss": 1.7606,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934055722140061e-05,
+      "loss": 1.8564,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93293580237512e-05,
+      "loss": 1.7919,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931806517013612e-05,
+      "loss": 1.7403,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930667868199579e-05,
+      "loss": 1.6688,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929519858094843e-05,
+      "loss": 1.7205,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928362488878996e-05,
+      "loss": 1.75,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927195762749405e-05,
+      "loss": 1.6975,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926019681921196e-05,
+      "loss": 1.7852,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92483424862726e-05,
+      "loss": 1.765,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923639465118239e-05,
+      "loss": 1.6558,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922435333662536e-05,
+      "loss": 1.9242,
+      "step": 199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.921221856546293e-05,
+      "loss": 1.7685,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9199990360734e-05,
+      "loss": 1.6635,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918766874565487e-05,
+      "loss": 1.7696,
+      "step": 202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917525374361912e-05,
+      "loss": 1.7166,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916274537819775e-05,
+      "loss": 1.6719,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915014367313888e-05,
+      "loss": 1.7389,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913744865236798e-05,
+      "loss": 1.7714,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912466033998757e-05,
+      "loss": 1.8335,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91117787602774e-05,
+      "loss": 1.7402,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90988039376942e-05,
+      "loss": 1.74,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908573589687175e-05,
+      "loss": 1.6637,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90725746626209e-05,
+      "loss": 1.7741,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905932025992932e-05,
+      "loss": 1.7307,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904597271396162e-05,
+      "loss": 1.6667,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903253205005927e-05,
+      "loss": 1.5513,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901899829374047e-05,
+      "loss": 1.7721,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900537147070023e-05,
+      "loss": 1.7069,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899165160681025e-05,
+      "loss": 1.752,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897783872811882e-05,
+      "loss": 1.6452,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896393286085084e-05,
+      "loss": 1.7908,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894993403140783e-05,
+      "loss": 1.8589,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893584226636772e-05,
+      "loss": 1.6617,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892165759248495e-05,
+      "loss": 1.6411,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890738003669029e-05,
+      "loss": 1.7425,
+      "step": 223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88930096260909e-05,
+      "loss": 1.7441,
+      "step": 224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887854638797023e-05,
+      "loss": 1.684,
+      "step": 225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886399034978797e-05,
+      "loss": 1.709,
+      "step": 226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884934153917997e-05,
+      "loss": 1.8318,
+      "step": 227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883459998395825e-05,
+      "loss": 1.6495,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88197657121109e-05,
+      "loss": 1.741,
+      "step": 229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880483875180205e-05,
+      "loss": 1.7328,
+      "step": 230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878981913137179e-05,
+      "loss": 1.6453,
+      "step": 231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877470687933614e-05,
+      "loss": 1.7469,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8759502024387e-05,
+      "loss": 1.5888,
+      "step": 233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874420459539206e-05,
+      "loss": 1.7388,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872881462139479e-05,
+      "loss": 1.7311,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871333213161438e-05,
+      "loss": 1.7349,
+      "step": 236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869775715544562e-05,
+      "loss": 1.8323,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868208972245894e-05,
+      "loss": 1.6795,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86663298624003e-05,
+      "loss": 1.6796,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86504776051911e-05,
+      "loss": 1.601,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86345329809282e-05,
+      "loss": 1.7074,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861849601988383e-05,
+      "loss": 1.6809,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860236675250552e-05,
+      "loss": 1.7527,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858614520941605e-05,
+      "loss": 1.7184,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856983142141339e-05,
+      "loss": 1.5873,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855342541947065e-05,
+      "loss": 1.7135,
+      "step": 246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8536927234736e-05,
+      "loss": 1.663,
+      "step": 247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852033689853267e-05,
+      "loss": 1.7253,
+      "step": 248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85036544423588e-05,
+      "loss": 1.7776,
+      "step": 249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848687989788745e-05,
+      "loss": 1.7187,
+      "step": 250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847001329696653e-05,
+      "loss": 1.6333,
+      "step": 251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845305467161867e-05,
+      "loss": 1.7161,
+      "step": 252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843600405404131e-05,
+      "loss": 1.7122,
+      "step": 253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841886147660645e-05,
+      "loss": 1.6872,
+      "step": 254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840162697186075e-05,
+      "loss": 1.7158,
+      "step": 255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838430057252535e-05,
+      "loss": 1.712,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836688231149592e-05,
+      "loss": 1.5938,
+      "step": 257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834937222184249e-05,
+      "loss": 1.8614,
+      "step": 258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833177033680944e-05,
+      "loss": 1.7055,
+      "step": 259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831407668981546e-05,
+      "loss": 1.6323,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829629131445342e-05,
+      "loss": 1.5731,
+      "step": 261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827841424449036e-05,
+      "loss": 1.6909,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826044551386744e-05,
+      "loss": 1.6502,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824238515669979e-05,
+      "loss": 1.691,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822423320727654e-05,
+      "loss": 1.7128,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820598970006069e-05,
+      "loss": 1.6348,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818765466968909e-05,
+      "loss": 1.7406,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816922815097236e-05,
+      "loss": 1.7156,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815071017889482e-05,
+      "loss": 1.7747,
+      "step": 269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813210078861437e-05,
+      "loss": 1.6992,
+      "step": 270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811340001546251e-05,
+      "loss": 1.6639,
+      "step": 271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809460789494432e-05,
+      "loss": 1.6609,
+      "step": 272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807572446273814e-05,
+      "loss": 1.6993,
+      "step": 273
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805674975469581e-05,
+      "loss": 1.7312,
+      "step": 274
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803768380684242e-05,
+      "loss": 1.6371,
+      "step": 275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801852665537627e-05,
+      "loss": 1.6527,
+      "step": 276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799927833666887e-05,
+      "loss": 1.7102,
+      "step": 277
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797993888726473e-05,
+      "loss": 1.6303,
+      "step": 278
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796050834388149e-05,
+      "loss": 1.7098,
+      "step": 279
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794098674340965e-05,
+      "loss": 1.7098,
+      "step": 280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792137412291265e-05,
+      "loss": 1.7502,
+      "step": 281
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790167051962666e-05,
+      "loss": 1.6951,
+      "step": 282
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788187597096069e-05,
+      "loss": 1.5267,
+      "step": 283
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786199051449636e-05,
+      "loss": 1.6506,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784201418798786e-05,
+      "loss": 1.6944,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782194702936196e-05,
+      "loss": 1.669,
+      "step": 286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780178907671789e-05,
+      "loss": 1.6365,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778154036832718e-05,
+      "loss": 1.7083,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776120094263376e-05,
+      "loss": 1.7194,
+      "step": 289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774077083825372e-05,
+      "loss": 1.7372,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772025009397537e-05,
+      "loss": 1.7153,
+      "step": 291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769963874875907e-05,
+      "loss": 1.7081,
+      "step": 292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767893684173721e-05,
+      "loss": 1.6957,
+      "step": 293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76581444122141e-05,
+      "loss": 1.6144,
+      "step": 294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763726149966596e-05,
+      "loss": 1.7146,
+      "step": 295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761628814374073e-05,
+      "loss": 1.6517,
+      "step": 296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759522438425813e-05,
+      "loss": 1.6761,
+      "step": 297
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757407026120949e-05,
+      "loss": 1.6871,
+      "step": 298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755282581475769e-05,
+      "loss": 1.6702,
+      "step": 299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753149108523712e-05,
+      "loss": 1.6568,
+      "step": 300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751006611315356e-05,
+      "loss": 1.8211,
+      "step": 301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748855093918417e-05,
+      "loss": 1.6159,
+      "step": 302
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746694560417731e-05,
+      "loss": 1.7733,
+      "step": 303
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744525014915252e-05,
+      "loss": 1.7148,
+      "step": 304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742346461530048e-05,
+      "loss": 1.5973,
+      "step": 305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740158904398286e-05,
+      "loss": 1.7084,
+      "step": 306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737962347673231e-05,
+      "loss": 1.6639,
+      "step": 307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735756795525231e-05,
+      "loss": 1.7169,
+      "step": 308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733542252141711e-05,
+      "loss": 1.7691,
+      "step": 309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731318721727172e-05,
+      "loss": 1.5357,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729086208503174e-05,
+      "loss": 1.7303,
+      "step": 311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726844716708331e-05,
+      "loss": 1.6989,
+      "step": 312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724594250598311e-05,
+      "loss": 1.4456,
+      "step": 313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722334814445809e-05,
+      "loss": 1.7214,
+      "step": 314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720066412540554e-05,
+      "loss": 1.701,
+      "step": 315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717789049189306e-05,
+      "loss": 1.614,
+      "step": 316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715502728715826e-05,
+      "loss": 1.7196,
+      "step": 317
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713207455460894e-05,
+      "loss": 1.7253,
+      "step": 318
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710903233782272e-05,
+      "loss": 1.6094,
+      "step": 319
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708590068054728e-05,
+      "loss": 1.8054,
+      "step": 320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706267962669998e-05,
+      "loss": 1.6724,
+      "step": 321
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703936922036799e-05,
+      "loss": 1.7097,
+      "step": 322
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701596950580806e-05,
+      "loss": 1.6195,
+      "step": 323
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699248052744658e-05,
+      "loss": 1.7066,
+      "step": 324
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696890232987931e-05,
+      "loss": 1.6692,
+      "step": 325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694523495787149e-05,
+      "loss": 1.6189,
+      "step": 326
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.692147845635761e-05,
+      "loss": 1.706,
+      "step": 327
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689763287044143e-05,
+      "loss": 1.6657,
+      "step": 328
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687369824539577e-05,
+      "loss": 1.6624,
+      "step": 329
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.684967462666259e-05,
+      "loss": 1.7106,
+      "step": 330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682556205985274e-05,
+      "loss": 1.645,
+      "step": 331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680136059074598e-05,
+      "loss": 1.6182,
+      "step": 332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677707026529086e-05,
+      "loss": 1.6641,
+      "step": 333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67526911296046e-05,
+      "loss": 1.6375,
+      "step": 334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672822322997305e-05,
+      "loss": 1.6917,
+      "step": 335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670366661285061e-05,
+      "loss": 1.7233,
+      "step": 336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667902132486009e-05,
+      "loss": 1.5695,
+      "step": 337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665428741279266e-05,
+      "loss": 1.6638,
+      "step": 338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662946492360776e-05,
+      "loss": 1.6915,
+      "step": 339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660455390443295e-05,
+      "loss": 1.659,
+      "step": 340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.657955440256395e-05,
+      "loss": 1.8316,
+      "step": 341
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655446646546442e-05,
+      "loss": 1.6264,
+      "step": 342
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652929014076593e-05,
+      "loss": 1.7016,
+      "step": 343
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650402547626786e-05,
+      "loss": 1.7149,
+      "step": 344
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647867251993734e-05,
+      "loss": 1.6403,
+      "step": 345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645323131990908e-05,
+      "loss": 1.635,
+      "step": 346
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642770192448536e-05,
+      "loss": 1.6494,
+      "step": 347
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640208438213591e-05,
+      "loss": 1.704,
+      "step": 348
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637637874149779e-05,
+      "loss": 1.6872,
+      "step": 349
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635058505137536e-05,
+      "loss": 1.7122,
+      "step": 350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632470336074009e-05,
+      "loss": 1.7165,
+      "step": 351
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62987337187306e-05,
+      "loss": 1.5921,
+      "step": 352
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627267617465243e-05,
+      "loss": 1.6971,
+      "step": 353
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624653077797805e-05,
+      "loss": 1.7033,
+      "step": 354
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62202975783467e-05,
+      "loss": 1.6559,
+      "step": 355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619397662556435e-05,
+      "loss": 1.7016,
+      "step": 356
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616756796960353e-05,
+      "loss": 1.6389,
+      "step": 357
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614107166060333e-05,
+      "loss": 1.4772,
+      "step": 358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611448774886924e-05,
+      "loss": 1.7113,
+      "step": 359
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608781628487307e-05,
+      "loss": 1.6872,
+      "step": 360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606105731925283e-05,
+      "loss": 1.6285,
+      "step": 361
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60342109028127e-05,
+      "loss": 1.8409,
+      "step": 362
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600727708652289e-05,
+      "loss": 1.6572,
+      "step": 363
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598025592151953e-05,
+      "loss": 1.6472,
+      "step": 364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595314745910456e-05,
+      "loss": 1.739,
+      "step": 365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592595175074572e-05,
+      "loss": 1.7044,
+      "step": 366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589866884807635e-05,
+      "loss": 1.5838,
+      "step": 367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587129880289538e-05,
+      "loss": 1.6982,
+      "step": 368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584384166716714e-05,
+      "loss": 1.632,
+      "step": 369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581629749302133e-05,
+      "loss": 1.7163,
+      "step": 370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.578866633275288e-05,
+      "loss": 1.6932,
+      "step": 371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576094823882191e-05,
+      "loss": 1.6807,
+      "step": 372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573314326385359e-05,
+      "loss": 1.5914,
+      "step": 373
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570525146063798e-05,
+      "loss": 1.6965,
+      "step": 374
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.567727288213005e-05,
+      "loss": 1.6963,
+      "step": 375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.564920758144951e-05,
+      "loss": 1.5894,
+      "step": 376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.562105561188069e-05,
+      "loss": 1.6914,
+      "step": 377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559281702687251e-05,
+      "loss": 1.6532,
+      "step": 378
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556449188003831e-05,
+      "loss": 1.6392,
+      "step": 379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.553608022515577e-05,
+      "loss": 1.8285,
+      "step": 380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.550758211616684e-05,
+      "loss": 1.6821,
+      "step": 381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54789976071776e-05,
+      "loss": 1.4445,
+      "step": 382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.545032675245813e-05,
+      "loss": 1.7086,
+      "step": 383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54215696064425e-05,
+      "loss": 1.7105,
+      "step": 384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.539272622372858e-05,
+      "loss": 1.6478,
+      "step": 385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5363796659078e-05,
+      "loss": 1.631,
+      "step": 386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533478096741597e-05,
+      "loss": 1.6525,
+      "step": 387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.530567920383122e-05,
+      "loss": 1.5849,
+      "step": 388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527649142357596e-05,
+      "loss": 1.7059,
+      "step": 389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52472176820656e-05,
+      "loss": 1.5551,
+      "step": 390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.521785803487889e-05,
+      "loss": 1.6745,
+      "step": 391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518841253775755e-05,
+      "loss": 1.8979,
+      "step": 392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515888124660638e-05,
+      "loss": 1.6951,
+      "step": 393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.512926421749304e-05,
+      "loss": 1.6212,
+      "step": 394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509956150664796e-05,
+      "loss": 1.6578,
+      "step": 395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.506977317046424e-05,
+      "loss": 1.6315,
+      "step": 396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.50398992654976e-05,
+      "loss": 1.5569,
+      "step": 397
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.500993984846614e-05,
+      "loss": 1.6869,
+      "step": 398
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.497989497625035e-05,
+      "loss": 1.6866,
+      "step": 399
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.494976470589301e-05,
+      "loss": 1.6448,
+      "step": 400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.491954909459895e-05,
+      "loss": 1.7782,
+      "step": 401
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.488924819973508e-05,
+      "loss": 1.6859,
+      "step": 402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.485886207883022e-05,
+      "loss": 1.6016,
+      "step": 403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4828390789575e-05,
+      "loss": 1.7017,
+      "step": 404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.479783438982172e-05,
+      "loss": 1.6874,
+      "step": 405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.476719293758432e-05,
+      "loss": 1.5945,
+      "step": 406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473646649103818e-05,
+      "loss": 1.6622,
+      "step": 407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470565510852008e-05,
+      "loss": 1.7109,
+      "step": 408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4674758848528e-05,
+      "loss": 1.731,
+      "step": 409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.464377776972114e-05,
+      "loss": 1.6448,
+      "step": 410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46127119309197e-05,
+      "loss": 1.5888,
+      "step": 411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.458156139110481e-05,
+      "loss": 1.7268,
+      "step": 412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45503262094184e-05,
+      "loss": 1.5869,
+      "step": 413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.451900644516311e-05,
+      "loss": 1.5919,
+      "step": 414
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.448760215780217e-05,
+      "loss": 1.6385,
+      "step": 415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.445611340695926e-05,
+      "loss": 1.7145,
+      "step": 416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.442454025241847e-05,
+      "loss": 1.6011,
+      "step": 417
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43928827541241e-05,
+      "loss": 1.6821,
+      "step": 418
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43611409721806e-05,
+      "loss": 1.6574,
+      "step": 419
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.432931496685242e-05,
+      "loss": 1.6694,
+      "step": 420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42974047985639e-05,
+      "loss": 1.6007,
+      "step": 421
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.426541052789925e-05,
+      "loss": 1.8179,
+      "step": 422
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42333322156023e-05,
+      "loss": 1.6974,
+      "step": 423
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.420116992257642e-05,
+      "loss": 1.6358,
+      "step": 424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.416892370988444e-05,
+      "loss": 1.6325,
+      "step": 425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.413659363874853e-05,
+      "loss": 1.5881,
+      "step": 426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.410417977055011e-05,
+      "loss": 1.6888,
+      "step": 427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.407168216682962e-05,
+      "loss": 1.6802,
+      "step": 428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.403910088928651e-05,
+      "loss": 1.6641,
+      "step": 429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40064359997791e-05,
+      "loss": 1.6266,
+      "step": 430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.397368756032445e-05,
+      "loss": 1.7182,
+      "step": 431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.394085563309827e-05,
+      "loss": 1.5725,
+      "step": 432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390794028043474e-05,
+      "loss": 1.5268,
+      "step": 433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.387494156482643e-05,
+      "loss": 1.6439,
+      "step": 434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.384185954892422e-05,
+      "loss": 1.5478,
+      "step": 435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.380869429553711e-05,
+      "loss": 1.6763,
+      "step": 436
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.377544586763215e-05,
+      "loss": 1.6985,
+      "step": 437
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.374211432833429e-05,
+      "loss": 1.5467,
+      "step": 438
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.370869974092629e-05,
+      "loss": 1.6274,
+      "step": 439
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.367520216884856e-05,
+      "loss": 1.7776,
+      "step": 440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.364162167569907e-05,
+      "loss": 1.588,
+      "step": 441
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.360795832523328e-05,
+      "loss": 1.6959,
+      "step": 442
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.357421218136386e-05,
+      "loss": 1.6437,
+      "step": 443
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.354038330816077e-05,
+      "loss": 1.6493,
+      "step": 444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.350647176985095e-05,
+      "loss": 1.5947,
+      "step": 445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347247763081835e-05,
+      "loss": 1.6581,
+      "step": 446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.343840095560372e-05,
+      "loss": 1.612,
+      "step": 447
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.340424180890451e-05,
+      "loss": 1.6886,
+      "step": 448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.337000025557476e-05,
+      "loss": 1.6079,
+      "step": 449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.333567636062496e-05,
+      "loss": 1.67,
+      "step": 450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330127018922194e-05,
+      "loss": 1.7251,
+      "step": 451
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.326678180668871e-05,
+      "loss": 1.6434,
+      "step": 452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.323221127850441e-05,
+      "loss": 1.5957,
+      "step": 453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.319755867030409e-05,
+      "loss": 1.6453,
+      "step": 454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.316282404787871e-05,
+      "loss": 1.5218,
+      "step": 455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.312800747717482e-05,
+      "loss": 1.6142,
+      "step": 456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.309310902429472e-05,
+      "loss": 1.6375,
+      "step": 457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.305812875549599e-05,
+      "loss": 1.6712,
+      "step": 458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.30230667371917e-05,
+      "loss": 1.5489,
+      "step": 459
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.298792303595001e-05,
+      "loss": 1.7669,
+      "step": 460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.295269771849427e-05,
+      "loss": 1.658,
+      "step": 461
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.291739085170268e-05,
+      "loss": 1.5819,
+      "step": 462
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.288200250260836e-05,
+      "loss": 1.7854,
+      "step": 463
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.284653273839905e-05,
+      "loss": 1.6228,
+      "step": 464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.281098162641714e-05,
+      "loss": 1.5931,
+      "step": 465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27753492341594e-05,
+      "loss": 1.7179,
+      "step": 466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.273963562927695e-05,
+      "loss": 1.6511,
+      "step": 467
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.270384087957511e-05,
+      "loss": 1.5873,
+      "step": 468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.266796505301322e-05,
+      "loss": 1.5603,
+      "step": 469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.263200821770461e-05,
+      "loss": 1.813,
+      "step": 470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.259597044191636e-05,
+      "loss": 1.6186,
+      "step": 471
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.255985179406923e-05,
+      "loss": 1.6305,
+      "step": 472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.252365234273755e-05,
+      "loss": 1.6839,
+      "step": 473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.248737215664903e-05,
+      "loss": 1.6243,
+      "step": 474
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24510113046847e-05,
+      "loss": 1.6098,
+      "step": 475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.241456985587868e-05,
+      "loss": 1.7206,
+      "step": 476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.237804787941819e-05,
+      "loss": 1.5555,
+      "step": 477
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.234144544464329e-05,
+      "loss": 1.6634,
+      "step": 478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.230476262104677e-05,
+      "loss": 1.7245,
+      "step": 479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.226799947827411e-05,
+      "loss": 1.5548,
+      "step": 480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.223115608612325e-05,
+      "loss": 1.4914,
+      "step": 481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.219423251454448e-05,
+      "loss": 1.6362,
+      "step": 482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.215722883364033e-05,
+      "loss": 1.7454,
+      "step": 483
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.212014511366542e-05,
+      "loss": 1.6879,
+      "step": 484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.208298142502636e-05,
+      "loss": 1.6082,
+      "step": 485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.204573783828152e-05,
+      "loss": 1.6615,
+      "step": 486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.200841442414106e-05,
+      "loss": 1.6043,
+      "step": 487
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.197101125346657e-05,
+      "loss": 1.6126,
+      "step": 488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.193352839727121e-05,
+      "loss": 1.6138,
+      "step": 489
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.189596592671932e-05,
+      "loss": 1.6835,
+      "step": 490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.185832391312644e-05,
+      "loss": 1.6803,
+      "step": 491
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.182060242795912e-05,
+      "loss": 1.5449,
+      "step": 492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17828015428348e-05,
+      "loss": 1.771,
+      "step": 493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.174492132952166e-05,
+      "loss": 1.6621,
+      "step": 494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17069618599385e-05,
+      "loss": 1.539,
+      "step": 495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.16689232061546e-05,
+      "loss": 1.6757,
+      "step": 496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.163080544038952e-05,
+      "loss": 1.6801,
+      "step": 497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159260863501312e-05,
+      "loss": 1.5976,
+      "step": 498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.155433286254525e-05,
+      "loss": 1.6627,
+      "step": 499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.151597819565571e-05,
+      "loss": 1.6693,
+      "step": 500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.147754470716408e-05,
+      "loss": 1.6979,
+      "step": 501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14390324700396e-05,
+      "loss": 1.5976,
+      "step": 502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.140044155740101e-05,
+      "loss": 1.6686,
+      "step": 503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.136177204251643e-05,
+      "loss": 1.5832,
+      "step": 504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.132302399880321e-05,
+      "loss": 1.6933,
+      "step": 505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12841974998278e-05,
+      "loss": 1.6357,
+      "step": 506
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.124529261930559e-05,
+      "loss": 1.583,
+      "step": 507
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.120630943110077e-05,
+      "loss": 1.4797,
+      "step": 508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.116724800922629e-05,
+      "loss": 1.5739,
+      "step": 509
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.112810842784351e-05,
+      "loss": 1.6924,
+      "step": 510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.108889076126226e-05,
+      "loss": 1.6536,
+      "step": 511
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.104959508394061e-05,
+      "loss": 1.5578,
+      "step": 512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.101022147048473e-05,
+      "loss": 1.7173,
+      "step": 513
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.097076999564874e-05,
+      "loss": 1.7691,
+      "step": 514
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.093124073433463e-05,
+      "loss": 1.5842,
+      "step": 515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.089163376159204e-05,
+      "loss": 1.5822,
+      "step": 516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.085194915261818e-05,
+      "loss": 1.6829,
+      "step": 517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.081218698275763e-05,
+      "loss": 1.6742,
+      "step": 518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.077234732750224e-05,
+      "loss": 1.6119,
+      "step": 519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.073243026249096e-05,
+      "loss": 1.6208,
+      "step": 520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.069243586350975e-05,
+      "loss": 1.742,
+      "step": 521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.065236420649135e-05,
+      "loss": 1.61,
+      "step": 522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.061221536751517e-05,
+      "loss": 1.6327,
+      "step": 523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.057198942280722e-05,
+      "loss": 1.6309,
+      "step": 524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.053168644873984e-05,
+      "loss": 1.5928,
+      "step": 525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.049130652183166e-05,
+      "loss": 1.6821,
+      "step": 526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.045084971874738e-05,
+      "loss": 1.5701,
+      "step": 527
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.041031611629767e-05,
+      "loss": 1.6784,
+      "step": 528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.0369705791439e-05,
+      "loss": 1.6416,
+      "step": 529
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.032901882127354e-05,
+      "loss": 1.6809,
+      "step": 530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.028825528304892e-05,
+      "loss": 1.7758,
+      "step": 531
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.024741525415819e-05,
+      "loss": 1.662,
+      "step": 532
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.020649881213958e-05,
+      "loss": 1.5852,
+      "step": 533
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.016550603467644e-05,
+      "loss": 1.4978,
+      "step": 534
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.012443699959705e-05,
+      "loss": 1.6705,
+      "step": 535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.008329178487442e-05,
+      "loss": 1.5527,
+      "step": 536
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.004207046862624e-05,
+      "loss": 1.6738,
+      "step": 537
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.00007731291147e-05,
+      "loss": 1.6498,
+      "step": 538
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.995939984474624e-05,
+      "loss": 1.4352,
+      "step": 539
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.991795069407159e-05,
+      "loss": 1.675,
+      "step": 540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.987642575578545e-05,
+      "loss": 1.6505,
+      "step": 541
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.983482510872645e-05,
+      "loss": 1.6505,
+      "step": 542
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.979314883187693e-05,
+      "loss": 1.669,
+      "step": 543
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.975139700436283e-05,
+      "loss": 1.6749,
+      "step": 544
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.970956970545355e-05,
+      "loss": 1.5371,
+      "step": 545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.966766701456177e-05,
+      "loss": 1.6528,
+      "step": 546
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.962568901124327e-05,
+      "loss": 1.6454,
+      "step": 547
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.958363577519684e-05,
+      "loss": 1.6057,
+      "step": 548
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.954150738626414e-05,
+      "loss": 1.6207,
+      "step": 549
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.949930392442947e-05,
+      "loss": 1.6563,
+      "step": 550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.945702546981969e-05,
+      "loss": 1.5995,
+      "step": 551
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.941467210270401e-05,
+      "loss": 1.812,
+      "step": 552
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.93722439034939e-05,
+      "loss": 1.6335,
+      "step": 553
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.932974095274289e-05,
+      "loss": 1.6159,
+      "step": 554
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.928716333114643e-05,
+      "loss": 1.4401,
+      "step": 555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.924451111954173e-05,
+      "loss": 1.6739,
+      "step": 556
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.920178439890765e-05,
+      "loss": 1.6288,
+      "step": 557
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.915898325036449e-05,
+      "loss": 1.6768,
+      "step": 558
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.911610775517382e-05,
+      "loss": 1.6554,
+      "step": 559
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.907315799473846e-05,
+      "loss": 1.5603,
+      "step": 560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.903013405060211e-05,
+      "loss": 1.6763,
+      "step": 561
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89870360044494e-05,
+      "loss": 1.6702,
+      "step": 562
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.894386393810563e-05,
+      "loss": 1.7541,
+      "step": 563
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.890061793353661e-05,
+      "loss": 1.6112,
+      "step": 564
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.885729807284856e-05,
+      "loss": 1.6549,
+      "step": 565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.881390443828787e-05,
+      "loss": 1.567,
+      "step": 566
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.877043711224108e-05,
+      "loss": 1.5788,
+      "step": 567
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.872689617723456e-05,
+      "loss": 1.6537,
+      "step": 568
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.868328171593448e-05,
+      "loss": 1.5721,
+      "step": 569
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.863959381114658e-05,
+      "loss": 1.6598,
+      "step": 570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.859583254581605e-05,
+      "loss": 1.652,
+      "step": 571
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.855199800302736e-05,
+      "loss": 1.5603,
+      "step": 572
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.85080902660041e-05,
+      "loss": 1.6378,
+      "step": 573
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.846410941810885e-05,
+      "loss": 1.6023,
+      "step": 574
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.842005554284296e-05,
+      "loss": 1.6946,
+      "step": 575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.837592872384644e-05,
+      "loss": 1.6606,
+      "step": 576
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.83317290448978e-05,
+      "loss": 1.4557,
+      "step": 577
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.828745658991386e-05,
+      "loss": 1.6375,
+      "step": 578
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.824311144294965e-05,
+      "loss": 1.6179,
+      "step": 579
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.819869368819819e-05,
+      "loss": 1.6391,
+      "step": 580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.815420340999033e-05,
+      "loss": 1.5978,
+      "step": 581
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.810964069279465e-05,
+      "loss": 1.6545,
+      "step": 582
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.806500562121723e-05,
+      "loss": 1.7007,
+      "step": 583
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.802029828000156e-05,
+      "loss": 1.66,
+      "step": 584
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.797551875402827e-05,
+      "loss": 1.6417,
+      "step": 585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.793066712831514e-05,
+      "loss": 1.613,
+      "step": 586
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.788574348801675e-05,
+      "loss": 1.6215,
+      "step": 587
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.784074791842443e-05,
+      "loss": 1.4893,
+      "step": 588
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77956805049661e-05,
+      "loss": 1.6493,
+      "step": 589
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.775054133320604e-05,
+      "loss": 1.6007,
+      "step": 590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.770533048884482e-05,
+      "loss": 1.6487,
+      "step": 591
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.766004805771903e-05,
+      "loss": 1.6493,
+      "step": 592
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.761469412580125e-05,
+      "loss": 1.5654,
+      "step": 593
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.75692687791997e-05,
+      "loss": 1.6437,
+      "step": 594
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.75237721041583e-05,
+      "loss": 1.7115,
+      "step": 595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74782041870563e-05,
+      "loss": 1.59,
+      "step": 596
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74325651144083e-05,
+      "loss": 1.6179,
+      "step": 597
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73868549728639e-05,
+      "loss": 1.6612,
+      "step": 598
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73410738492077e-05,
+      "loss": 1.6054,
+      "step": 599
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.729522183035904e-05,
+      "loss": 1.6229,
+      "step": 600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.724929900337186e-05,
+      "loss": 1.6394,
+      "step": 601
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.720330545543453e-05,
+      "loss": 1.7137,
+      "step": 602
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.715724127386972e-05,
+      "loss": 1.716,
+      "step": 603
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.711110654613415e-05,
+      "loss": 1.4906,
+      "step": 604
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.706490135981855e-05,
+      "loss": 1.6991,
+      "step": 605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.701862580264735e-05,
+      "loss": 1.6664,
+      "step": 606
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.697227996247861e-05,
+      "loss": 1.3435,
+      "step": 607
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.692586392730387e-05,
+      "loss": 1.5943,
+      "step": 608
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.687937778524786e-05,
+      "loss": 1.6577,
+      "step": 609
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.68328216245685e-05,
+      "loss": 1.5683,
+      "step": 610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.678619553365659e-05,
+      "loss": 1.6154,
+      "step": 611
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.673949960103571e-05,
+      "loss": 1.6633,
+      "step": 612
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.669273391536204e-05,
+      "loss": 1.5445,
+      "step": 613
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66458985654242e-05,
+      "loss": 1.7576,
+      "step": 614
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.659899364014309e-05,
+      "loss": 1.6066,
+      "step": 615
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.655201922857165e-05,
+      "loss": 1.649,
+      "step": 616
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.650497541989482e-05,
+      "loss": 1.6087,
+      "step": 617
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.645786230342923e-05,
+      "loss": 1.639,
+      "step": 618
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.641067996862311e-05,
+      "loss": 1.5637,
+      "step": 619
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.636342850505616e-05,
+      "loss": 1.6097,
+      "step": 620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.631610800243926e-05,
+      "loss": 1.6443,
+      "step": 621
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.626871855061438e-05,
+      "loss": 1.5653,
+      "step": 622
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.622126023955446e-05,
+      "loss": 1.6126,
+      "step": 623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.617373315936307e-05,
+      "loss": 1.6494,
+      "step": 624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.612613740027443e-05,
+      "loss": 1.6184,
+      "step": 625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.60784730526531e-05,
+      "loss": 1.584,
+      "step": 626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.603074020699393e-05,
+      "loss": 1.5597,
+      "step": 627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.598293895392171e-05,
+      "loss": 1.6106,
+      "step": 628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.59350693841912e-05,
+      "loss": 1.5992,
+      "step": 629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.588713158868683e-05,
+      "loss": 1.7001,
+      "step": 630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.583912565842257e-05,
+      "loss": 1.6057,
+      "step": 631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.579105168454173e-05,
+      "loss": 1.5924,
+      "step": 632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.574290975831685e-05,
+      "loss": 1.6833,
+      "step": 633
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.569469997114944e-05,
+      "loss": 1.5326,
+      "step": 634
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.564642241456986e-05,
+      "loss": 1.7522,
+      "step": 635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.559807718023715e-05,
+      "loss": 1.6185,
+      "step": 636
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.554966435993882e-05,
+      "loss": 1.5792,
+      "step": 637
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.550118404559075e-05,
+      "loss": 1.6678,
+      "step": 638
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.545263632923687e-05,
+      "loss": 1.5757,
+      "step": 639
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.540402130304917e-05,
+      "loss": 1.6029,
+      "step": 640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.535533905932738e-05,
+      "loss": 1.6649,
+      "step": 641
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.530658969049888e-05,
+      "loss": 1.6255,
+      "step": 642
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.525777328911846e-05,
+      "loss": 1.633,
+      "step": 643
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52088899478682e-05,
+      "loss": 1.6965,
+      "step": 644
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.515993975955727e-05,
+      "loss": 1.6005,
+      "step": 645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.511092281712174e-05,
+      "loss": 1.5854,
+      "step": 646
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.506183921362443e-05,
+      "loss": 1.6592,
+      "step": 647
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.501268904225475e-05,
+      "loss": 1.6361,
+      "step": 648
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.49634723963284e-05,
+      "loss": 1.5538,
+      "step": 649
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.491418936928742e-05,
+      "loss": 1.584,
+      "step": 650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.486484005469977e-05,
+      "loss": 1.6438,
+      "step": 651
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.481542454625932e-05,
+      "loss": 1.4215,
+      "step": 652
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.476594293778561e-05,
+      "loss": 1.6325,
+      "step": 653
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.471639532322364e-05,
+      "loss": 1.6422,
+      "step": 654
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.466678179664379e-05,
+      "loss": 1.5393,
+      "step": 655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.461710245224148e-05,
+      "loss": 1.7858,
+      "step": 656
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.456735738433723e-05,
+      "loss": 1.6495,
+      "step": 657
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.451754668737621e-05,
+      "loss": 1.5729,
+      "step": 658
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.44676704559283e-05,
+      "loss": 1.7036,
+      "step": 659
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.44177287846877e-05,
+      "loss": 1.6421,
+      "step": 660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.436772176847294e-05,
+      "loss": 1.5709,
+      "step": 661
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.431764950222656e-05,
+      "loss": 1.6309,
+      "step": 662
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4267512081015e-05,
+      "loss": 1.7108,
+      "step": 663
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.42173096000284e-05,
+      "loss": 1.5653,
+      "step": 664
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.416704215458043e-05,
+      "loss": 1.6458,
+      "step": 665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41167098401081e-05,
+      "loss": 1.6842,
+      "step": 666
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.406631275217156e-05,
+      "loss": 1.5905,
+      "step": 667
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.401585098645395e-05,
+      "loss": 1.6058,
+      "step": 668
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.396532463876124e-05,
+      "loss": 1.6774,
+      "step": 669
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.391473380502195e-05,
+      "loss": 1.554,
+      "step": 670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.386407858128706e-05,
+      "loss": 1.6415,
+      "step": 671
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.381335906372982e-05,
+      "loss": 1.6397,
+      "step": 672
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.376257534864553e-05,
+      "loss": 1.5984,
+      "step": 673
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.371172753245138e-05,
+      "loss": 1.7744,
+      "step": 674
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.366081571168625e-05,
+      "loss": 1.68,
+      "step": 675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.360983998301053e-05,
+      "loss": 1.3957,
+      "step": 676
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.355880044320598e-05,
+      "loss": 1.6395,
+      "step": 677
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.350769718917549e-05,
+      "loss": 1.6364,
+      "step": 678
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.345653031794292e-05,
+      "loss": 1.5431,
+      "step": 679
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.34052999266529e-05,
+      "loss": 1.5885,
+      "step": 680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.335400611257067e-05,
+      "loss": 1.6514,
+      "step": 681
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.330264897308188e-05,
+      "loss": 1.5459,
+      "step": 682
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.32512286056924e-05,
+      "loss": 1.6682,
+      "step": 683
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.31997451080282e-05,
+      "loss": 1.5708,
+      "step": 684
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.314819857783503e-05,
+      "loss": 1.5762,
+      "step": 685
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.309658911297834e-05,
+      "loss": 1.7779,
+      "step": 686
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.304491681144306e-05,
+      "loss": 1.6297,
+      "step": 687
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.299318177133346e-05,
+      "loss": 1.5779,
+      "step": 688
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.29413840908729e-05,
+      "loss": 1.6043,
+      "step": 689
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.288952386840364e-05,
+      "loss": 1.5579,
+      "step": 690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.283760120238672e-05,
+      "loss": 1.5908,
+      "step": 691
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.278561619140172e-05,
+      "loss": 1.6437,
+      "step": 692
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.273356893414659e-05,
+      "loss": 1.6685,
+      "step": 693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.268145952943745e-05,
+      "loss": 1.5318,
+      "step": 694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.262928807620843e-05,
+      "loss": 1.7456,
+      "step": 695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.257705467351145e-05,
+      "loss": 1.6214,
+      "step": 696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.252475942051605e-05,
+      "loss": 1.5401,
+      "step": 697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.247240241650918e-05,
+      "loss": 1.6109,
+      "step": 698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.241998376089508e-05,
+      "loss": 1.6357,
+      "step": 699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.2367503553195e-05,
+      "loss": 1.5322,
+      "step": 700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.231496189304704e-05,
+      "loss": 1.6424,
+      "step": 701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.226235888020602e-05,
+      "loss": 1.555,
+      "step": 702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.220969461454322e-05,
+      "loss": 1.7549,
+      "step": 703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.215696919604618e-05,
+      "loss": 1.5904,
+      "step": 704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.210418272481859e-05,
+      "loss": 1.547,
+      "step": 705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.205133530108004e-05,
+      "loss": 1.6897,
+      "step": 706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.199842702516583e-05,
+      "loss": 1.4507,
+      "step": 707
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.194545799752683e-05,
+      "loss": 1.5208,
+      "step": 708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.18924283187292e-05,
+      "loss": 1.6032,
+      "step": 709
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.18393380894543e-05,
+      "loss": 1.6613,
+      "step": 710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.178618741049842e-05,
+      "loss": 1.5495,
+      "step": 711
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.17329763827726e-05,
+      "loss": 1.6797,
+      "step": 712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.167970510730253e-05,
+      "loss": 1.5726,
+      "step": 713
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.162637368522822e-05,
+      "loss": 1.6193,
+      "step": 714
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.157298221780389e-05,
+      "loss": 1.5994,
+      "step": 715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.151953080639775e-05,
+      "loss": 1.7452,
+      "step": 716
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.146601955249188e-05,
+      "loss": 1.5731,
+      "step": 717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.141244855768188e-05,
+      "loss": 1.6324,
+      "step": 718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.135881792367686e-05,
+      "loss": 1.612,
+      "step": 719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.130512775229909e-05,
+      "loss": 1.5591,
+      "step": 720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.125137814548393e-05,
+      "loss": 1.6263,
+      "step": 721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.119756920527955e-05,
+      "loss": 1.6862,
+      "step": 722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.114370103384681e-05,
+      "loss": 1.6156,
+      "step": 723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.108977373345894e-05,
+      "loss": 1.5906,
+      "step": 724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.103578740650156e-05,
+      "loss": 1.7075,
+      "step": 725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.098174215547224e-05,
+      "loss": 1.5506,
+      "step": 726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.092763808298048e-05,
+      "loss": 1.4803,
+      "step": 727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.087347529174743e-05,
+      "loss": 1.5881,
+      "step": 728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.081925388460578e-05,
+      "loss": 1.5416,
+      "step": 729
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.076497396449941e-05,
+      "loss": 1.6722,
+      "step": 730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.07106356344834e-05,
+      "loss": 1.556,
+      "step": 731
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.065623899772364e-05,
+      "loss": 1.5772,
+      "step": 732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.060178415749674e-05,
+      "loss": 1.575,
+      "step": 733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.054727121718987e-05,
+      "loss": 1.7366,
+      "step": 734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.049270028030046e-05,
+      "loss": 1.5617,
+      "step": 735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.043807145043604e-05,
+      "loss": 1.6639,
+      "step": 736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.038338483131407e-05,
+      "loss": 1.5991,
+      "step": 737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.032864052676172e-05,
+      "loss": 1.6286,
+      "step": 738
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.027383864071573e-05,
+      "loss": 1.5695,
+      "step": 739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.021897927722208e-05,
+      "loss": 1.6365,
+      "step": 740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.016406254043595e-05,
+      "loss": 1.5732,
+      "step": 741
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.010908853462141e-05,
+      "loss": 1.6328,
+      "step": 742
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.005405736415126e-05,
+      "loss": 1.6293,
+      "step": 743
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.999896913350685e-05,
+      "loss": 1.5965,
+      "step": 744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.994382394727784e-05,
+      "loss": 1.6793,
+      "step": 745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.988862191016205e-05,
+      "loss": 1.6286,
+      "step": 746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.983336312696522e-05,
+      "loss": 1.5487,
+      "step": 747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.977804770260083e-05,
+      "loss": 1.6408,
+      "step": 748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.972267574208991e-05,
+      "loss": 1.4055,
+      "step": 749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.966724735056079e-05,
+      "loss": 1.584,
+      "step": 750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.961176263324901e-05,
+      "loss": 1.6499,
+      "step": 751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.955622169549697e-05,
+      "loss": 1.5858,
+      "step": 752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.950062464275387e-05,
+      "loss": 1.543,
+      "step": 753
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.944497158057543e-05,
+      "loss": 1.6638,
+      "step": 754
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.938926261462366e-05,
+      "loss": 1.6072,
+      "step": 755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.93334978506668e-05,
+      "loss": 1.5395,
+      "step": 756
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.927767739457897e-05,
+      "loss": 1.7473,
+      "step": 757
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.922180135234e-05,
+      "loss": 1.5978,
+      "step": 758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.916586983003533e-05,
+      "loss": 1.5637,
+      "step": 759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.910988293385568e-05,
+      "loss": 1.6263,
+      "step": 760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.905384077009693e-05,
+      "loss": 1.5974,
+      "step": 761
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.899774344515985e-05,
+      "loss": 1.539,
+      "step": 762
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.894159106554997e-05,
+      "loss": 1.5781,
+      "step": 763
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.888538373787734e-05,
+      "loss": 1.7595,
+      "step": 764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.882912156885637e-05,
+      "loss": 1.5776,
+      "step": 765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.877280466530551e-05,
+      "loss": 1.6384,
+      "step": 766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.871643313414718e-05,
+      "loss": 1.6377,
+      "step": 767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.866000708240753e-05,
+      "loss": 1.5428,
+      "step": 768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.860352661721619e-05,
+      "loss": 1.6086,
+      "step": 769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.854699184580609e-05,
+      "loss": 1.6197,
+      "step": 770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.849040287551331e-05,
+      "loss": 1.5803,
+      "step": 771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.84337598137768e-05,
+      "loss": 1.6304,
+      "step": 772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.837706276813819e-05,
+      "loss": 1.6588,
+      "step": 773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.832031184624164e-05,
+      "loss": 1.5588,
+      "step": 774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.82635071558336e-05,
+      "loss": 1.4355,
+      "step": 775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.820664880476256e-05,
+      "loss": 1.5437,
+      "step": 776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.814973690097893e-05,
+      "loss": 1.7293,
+      "step": 777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.809277155253478e-05,
+      "loss": 1.6302,
+      "step": 778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.803575286758364e-05,
+      "loss": 1.5606,
+      "step": 779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.797868095438032e-05,
+      "loss": 1.6245,
+      "step": 780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.79215559212807e-05,
+      "loss": 1.5147,
+      "step": 781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.786437787674149e-05,
+      "loss": 1.5729,
+      "step": 782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.780714692932002e-05,
+      "loss": 1.5406,
+      "step": 783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.774986318767415e-05,
+      "loss": 1.6056,
+      "step": 784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.769252676056187e-05,
+      "loss": 1.693,
+      "step": 785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.763513775684125e-05,
+      "loss": 1.4985,
+      "step": 786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.757769628547018e-05,
+      "loss": 1.757,
+      "step": 787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.752020245550618e-05,
+      "loss": 1.6184,
+      "step": 788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.746265637610613e-05,
+      "loss": 1.5107,
+      "step": 789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.740505815652615e-05,
+      "loss": 1.6306,
+      "step": 790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.734740790612136e-05,
+      "loss": 1.6,
+      "step": 791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72897057343456e-05,
+      "loss": 1.5594,
+      "step": 792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.723195175075136e-05,
+      "loss": 1.6411,
+      "step": 793
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.717414606498947e-05,
+      "loss": 1.5452,
+      "step": 794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.711628878680892e-05,
+      "loss": 1.7537,
+      "step": 795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.705838002605665e-05,
+      "loss": 1.6423,
+      "step": 796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.700041989267736e-05,
+      "loss": 1.6039,
+      "step": 797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.694240849671326e-05,
+      "loss": 1.606,
+      "step": 798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.688434594830392e-05,
+      "loss": 1.5772,
+      "step": 799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.682623235768597e-05,
+      "loss": 1.6094,
+      "step": 800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.676806783519304e-05,
+      "loss": 1.5818,
+      "step": 801
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.670985249125535e-05,
+      "loss": 1.4385,
+      "step": 802
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.66515864363997e-05,
+      "loss": 1.5323,
+      "step": 803
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.65932697812491e-05,
+      "loss": 1.6141,
+      "step": 804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.653490263652269e-05,
+      "loss": 1.6493,
+      "step": 805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.647648511303544e-05,
+      "loss": 1.5579,
+      "step": 806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.641801732169795e-05,
+      "loss": 1.6399,
+      "step": 807
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.635949937351629e-05,
+      "loss": 1.7479,
+      "step": 808
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.630093137959171e-05,
+      "loss": 1.5272,
+      "step": 809
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.624231345112055e-05,
+      "loss": 1.5891,
+      "step": 810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.618364569939391e-05,
+      "loss": 1.6457,
+      "step": 811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.612492823579745e-05,
+      "loss": 1.6058,
+      "step": 812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.606616117181128e-05,
+      "loss": 1.5581,
+      "step": 813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.600734461900965e-05,
+      "loss": 1.5352,
+      "step": 814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.594847868906076e-05,
+      "loss": 1.6972,
+      "step": 815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.588956349372657e-05,
+      "loss": 1.5548,
+      "step": 816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.583059914486257e-05,
+      "loss": 1.6401,
+      "step": 817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.577158575441756e-05,
+      "loss": 1.5557,
+      "step": 818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.571252343443349e-05,
+      "loss": 1.5934,
+      "step": 819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.565341229704514e-05,
+      "loss": 1.6342,
+      "step": 820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.559425245448006e-05,
+      "loss": 1.5253,
+      "step": 821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.553504401905817e-05,
+      "loss": 1.628,
+      "step": 822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.547578710319174e-05,
+      "loss": 1.6209,
+      "step": 823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.541648181938503e-05,
+      "loss": 1.6138,
+      "step": 824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.535712828023416e-05,
+      "loss": 1.697,
+      "step": 825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.529772659842685e-05,
+      "loss": 1.5934,
+      "step": 826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.52382768867422e-05,
+      "loss": 1.5327,
+      "step": 827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.517877925805055e-05,
+      "loss": 1.4421,
+      "step": 828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.511923382531317e-05,
+      "loss": 1.6607,
+      "step": 829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.505964070158214e-05,
+      "loss": 1.5427,
+      "step": 830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 1.652,
+      "step": 831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.494031183379973e-05,
+      "loss": 1.5382,
+      "step": 832
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.488057631630437e-05,
+      "loss": 1.4511,
+      "step": 833
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.482079356092683e-05,
+      "loss": 1.632,
+      "step": 834
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.476096368116974e-05,
+      "loss": 1.6069,
+      "step": 835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.470108679062521e-05,
+      "loss": 1.5686,
+      "step": 836
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.464116300297458e-05,
+      "loss": 1.7441,
+      "step": 837
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.458119243198824e-05,
+      "loss": 1.559,
+      "step": 838
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.452117519152542e-05,
+      "loss": 1.5263,
+      "step": 839
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.44611113955339e-05,
+      "loss": 1.6253,
+      "step": 840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.440100115804991e-05,
+      "loss": 1.6411,
+      "step": 841
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.434084459319783e-05,
+      "loss": 1.5411,
+      "step": 842
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.428064181518997e-05,
+      "loss": 1.6018,
+      "step": 843
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.422039293832644e-05,
+      "loss": 1.6449,
+      "step": 844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.416009807699482e-05,
+      "loss": 1.5223,
+      "step": 845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.409975734566998e-05,
+      "loss": 1.7754,
+      "step": 846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.403937085891397e-05,
+      "loss": 1.5902,
+      "step": 847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.397893873137563e-05,
+      "loss": 1.5584,
+      "step": 848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.391846107779047e-05,
+      "loss": 1.406,
+      "step": 849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.385793801298042e-05,
+      "loss": 1.6252,
+      "step": 850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.379736965185368e-05,
+      "loss": 1.5423,
+      "step": 851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.37367561094044e-05,
+      "loss": 1.601,
+      "step": 852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.367609750071252e-05,
+      "loss": 1.6238,
+      "step": 853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.361539394094357e-05,
+      "loss": 1.5503,
+      "step": 854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.355464554534837e-05,
+      "loss": 1.6017,
+      "step": 855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.34938524292629e-05,
+      "loss": 1.6454,
+      "step": 856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.343301470810808e-05,
+      "loss": 1.7091,
+      "step": 857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.337213249738943e-05,
+      "loss": 1.5625,
+      "step": 858
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.331120591269701e-05,
+      "loss": 1.6375,
+      "step": 859
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.325023506970512e-05,
+      "loss": 1.5439,
+      "step": 860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.318922008417203e-05,
+      "loss": 1.5511,
+      "step": 861
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.312816107193988e-05,
+      "loss": 1.6079,
+      "step": 862
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.30670581489344e-05,
+      "loss": 1.5391,
+      "step": 863
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.300591143116461e-05,
+      "loss": 1.6116,
+      "step": 864
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.294472103472281e-05,
+      "loss": 1.6222,
+      "step": 865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.288348707578408e-05,
+      "loss": 1.514,
+      "step": 866
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.282220967060633e-05,
+      "loss": 1.6039,
+      "step": 867
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.276088893552988e-05,
+      "loss": 1.5996,
+      "step": 868
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.269952498697734e-05,
+      "loss": 1.6857,
+      "step": 869
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.263811794145339e-05,
+      "loss": 1.3794,
+      "step": 870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.257666791554448e-05,
+      "loss": 1.6334,
+      "step": 871
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.251517502591869e-05,
+      "loss": 1.5551,
+      "step": 872
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.245363938932551e-05,
+      "loss": 1.6409,
+      "step": 873
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.239206112259553e-05,
+      "loss": 1.5302,
+      "step": 874
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.233044034264034e-05,
+      "loss": 1.6465,
+      "step": 875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.226877716645217e-05,
+      "loss": 1.6248,
+      "step": 876
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.220707171110382e-05,
+      "loss": 1.6306,
+      "step": 877
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.214532409374829e-05,
+      "loss": 1.6502,
+      "step": 878
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.20835344316187e-05,
+      "loss": 1.6276,
+      "step": 879
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.202170284202794e-05,
+      "loss": 1.504,
+      "step": 880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.195982944236851e-05,
+      "loss": 1.6208,
+      "step": 881
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.189791435011235e-05,
+      "loss": 1.4442,
+      "step": 882
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.183595768281043e-05,
+      "loss": 1.6069,
+      "step": 883
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.17739595580928e-05,
+      "loss": 1.5505,
+      "step": 884
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.171192009366814e-05,
+      "loss": 1.6158,
+      "step": 885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.16498394073236e-05,
+      "loss": 1.6055,
+      "step": 886
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.158771761692464e-05,
+      "loss": 1.5212,
+      "step": 887
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.152555484041476e-05,
+      "loss": 1.7454,
+      "step": 888
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.146335119581523e-05,
+      "loss": 1.5809,
+      "step": 889
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.140110680122496e-05,
+      "loss": 1.5405,
+      "step": 890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.133882177482019e-05,
+      "loss": 1.5717,
+      "step": 891
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.127649623485433e-05,
+      "loss": 1.6158,
+      "step": 892
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.121413029965769e-05,
+      "loss": 1.5805,
+      "step": 893
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.11517240876373e-05,
+      "loss": 1.584,
+      "step": 894
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.108927771727661e-05,
+      "loss": 1.5828,
+      "step": 895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.102679130713537e-05,
+      "loss": 1.6328,
+      "step": 896
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.096426497584933e-05,
+      "loss": 1.6453,
+      "step": 897
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.090169884213002e-05,
+      "loss": 1.4767,
+      "step": 898
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.083909302476453e-05,
+      "loss": 1.6793,
+      "step": 899
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.077644764261532e-05,
+      "loss": 1.6201,
+      "step": 900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.071376281461994e-05,
+      "loss": 1.3375,
+      "step": 901
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.065103865979088e-05,
+      "loss": 1.5634,
+      "step": 902
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.058827529721525e-05,
+      "loss": 1.6546,
+      "step": 903
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.05254728460546e-05,
+      "loss": 1.5194,
+      "step": 904
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.04626314255447e-05,
+      "loss": 1.63,
+      "step": 905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.039975115499534e-05,
+      "loss": 1.5589,
+      "step": 906
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.033683215379002e-05,
+      "loss": 1.5322,
+      "step": 907
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.027387454138578e-05,
+      "loss": 1.7651,
+      "step": 908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.021087843731302e-05,
+      "loss": 1.6091,
+      "step": 909
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.014784396117515e-05,
+      "loss": 1.5851,
+      "step": 910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.008477123264848e-05,
+      "loss": 1.5297,
+      "step": 911
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.002166037148192e-05,
+      "loss": 1.6079,
+      "step": 912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.99585114974968e-05,
+      "loss": 1.5605,
+      "step": 913
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.989532473058657e-05,
+      "loss": 1.5719,
+      "step": 914
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.98321001907167e-05,
+      "loss": 1.5789,
+      "step": 915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.976883799792433e-05,
+      "loss": 1.6285,
+      "step": 916
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.97055382723181e-05,
+      "loss": 1.554,
+      "step": 917
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.964220113407785e-05,
+      "loss": 1.6305,
+      "step": 918
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.957882670345458e-05,
+      "loss": 1.5401,
+      "step": 919
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.951541510076995e-05,
+      "loss": 1.5091,
+      "step": 920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.94519664464163e-05,
+      "loss": 1.5997,
+      "step": 921
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.938848086085628e-05,
+      "loss": 1.5446,
+      "step": 922
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.932495846462261e-05,
+      "loss": 1.5916,
+      "step": 923
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.926139937831799e-05,
+      "loss": 1.6422,
+      "step": 924
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.91978037226147e-05,
+      "loss": 1.5154,
+      "step": 925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.91341716182545e-05,
+      "loss": 1.6066,
+      "step": 926
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.90705031860483e-05,
+      "loss": 1.5715,
+      "step": 927
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.900679854687607e-05,
+      "loss": 1.5586,
+      "step": 928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.894305782168638e-05,
+      "loss": 1.7352,
+      "step": 929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.887928113149648e-05,
+      "loss": 1.5264,
+      "step": 930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.881546859739179e-05,
+      "loss": 1.6029,
+      "step": 931
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.875162034052578e-05,
+      "loss": 1.6157,
+      "step": 932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.868773648211983e-05,
+      "loss": 1.5033,
+      "step": 933
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.862381714346282e-05,
+      "loss": 1.5846,
+      "step": 934
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.855986244591104e-05,
+      "loss": 1.5875,
+      "step": 935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.849587251088791e-05,
+      "loss": 1.6141,
+      "step": 936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.843184745988373e-05,
+      "loss": 1.5902,
+      "step": 937
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.83677874144555e-05,
+      "loss": 1.6456,
+      "step": 938
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.830369249622662e-05,
+      "loss": 1.6114,
+      "step": 939
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.823956282688676e-05,
+      "loss": 1.5324,
+      "step": 940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.817539852819149e-05,
+      "loss": 1.5874,
+      "step": 941
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.81111997219622e-05,
+      "loss": 1.6301,
+      "step": 942
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.804696653008575e-05,
+      "loss": 1.5177,
+      "step": 943
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.798269907451427e-05,
+      "loss": 1.5457,
+      "step": 944
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7918397477265e-05,
+      "loss": 1.6217,
+      "step": 945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.785406186041999e-05,
+      "loss": 1.3823,
+      "step": 946
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.778969234612584e-05,
+      "loss": 1.6604,
+      "step": 947
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.77252890565935e-05,
+      "loss": 1.6203,
+      "step": 948
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.76608521140981e-05,
+      "loss": 1.4949,
+      "step": 949
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.759638164097862e-05,
+      "loss": 1.7635,
+      "step": 950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.753187775963773e-05,
+      "loss": 1.4932,
+      "step": 951
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.74673405925415e-05,
+      "loss": 1.6496,
+      "step": 952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.740277026221923e-05,
+      "loss": 1.6458,
+      "step": 953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.733816689126315e-05,
+      "loss": 1.5395,
+      "step": 954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.727353060232822e-05,
+      "loss": 1.5917,
+      "step": 955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.720886151813194e-05,
+      "loss": 1.4541,
+      "step": 956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.714415976145402e-05,
+      "loss": 1.7527,
+      "step": 957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.707942545513624e-05,
+      "loss": 1.5677,
+      "step": 958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.701465872208216e-05,
+      "loss": 1.556,
+      "step": 959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.694985968525691e-05,
+      "loss": 1.6248,
+      "step": 960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.688502846768696e-05,
+      "loss": 1.5404,
+      "step": 961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.682016519245986e-05,
+      "loss": 1.5844,
+      "step": 962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.675526998272405e-05,
+      "loss": 1.6489,
+      "step": 963
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.669034296168855e-05,
+      "loss": 1.4325,
+      "step": 964
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.662538425262285e-05,
+      "loss": 1.6216,
+      "step": 965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.656039397885652e-05,
+      "loss": 1.6093,
+      "step": 966
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.649537226377915e-05,
+      "loss": 1.5473,
+      "step": 967
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.643031923083995e-05,
+      "loss": 1.7352,
+      "step": 968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.636523500354759e-05,
+      "loss": 1.6018,
+      "step": 969
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.630011970547004e-05,
+      "loss": 1.3129,
+      "step": 970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.623497346023418e-05,
+      "loss": 1.6663,
+      "step": 971
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.616979639152568e-05,
+      "loss": 1.5562,
+      "step": 972
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.610458862308872e-05,
+      "loss": 1.5417,
+      "step": 973
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.603935027872579e-05,
+      "loss": 1.5879,
+      "step": 974
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.59740814822974e-05,
+      "loss": 1.5366,
+      "step": 975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.590878235772191e-05,
+      "loss": 1.596,
+      "step": 976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.584345302897523e-05,
+      "loss": 1.5871,
+      "step": 977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.57780936200906e-05,
+      "loss": 1.5533,
+      "step": 978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.571270425515843e-05,
+      "loss": 1.6072,
+      "step": 979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.564728505832596e-05,
+      "loss": 1.7253,
+      "step": 980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.558183615379707e-05,
+      "loss": 1.6159,
+      "step": 981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.551635766583205e-05,
+      "loss": 1.5393,
+      "step": 982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.545084971874738e-05,
+      "loss": 1.6457,
+      "step": 983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.538531243691541e-05,
+      "loss": 1.5368,
+      "step": 984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.531974594476425e-05,
+      "loss": 1.5447,
+      "step": 985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.525415036677744e-05,
+      "loss": 1.5943,
+      "step": 986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.518852582749373e-05,
+      "loss": 1.6223,
+      "step": 987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.512287245150688e-05,
+      "loss": 1.502,
+      "step": 988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.505719036346539e-05,
+      "loss": 1.7362,
+      "step": 989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.499147968807224e-05,
+      "loss": 1.6257,
+      "step": 990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.492574055008473e-05,
+      "loss": 1.5058,
+      "step": 991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.48599730743142e-05,
+      "loss": 1.5892,
+      "step": 992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.479417738562576e-05,
+      "loss": 1.553,
+      "step": 993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.472835360893809e-05,
+      "loss": 1.5857,
+      "step": 994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.466250186922325e-05,
+      "loss": 1.6105,
+      "step": 995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.459662229150629e-05,
+      "loss": 1.4905,
+      "step": 996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.45307150008652e-05,
+      "loss": 1.7869,
+      "step": 997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.446478012243056e-05,
+      "loss": 1.6129,
+      "step": 998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.439881778138531e-05,
+      "loss": 1.5113,
+      "step": 999
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.433282810296454e-05,
+      "loss": 1.6328,
+      "step": 1000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.426681121245527e-05,
+      "loss": 1.3995,
+      "step": 1001
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.420076723519614e-05,
+      "loss": 1.5349,
+      "step": 1002
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.413469629657723e-05,
+      "loss": 1.6217,
+      "step": 1003
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.406859852203983e-05,
+      "loss": 1.6194,
+      "step": 1004
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.400247403707617e-05,
+      "loss": 1.5328,
+      "step": 1005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.393632296722915e-05,
+      "loss": 1.5978,
+      "step": 1006
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.387014543809223e-05,
+      "loss": 1.4794,
+      "step": 1007
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.380394157530904e-05,
+      "loss": 1.6349,
+      "step": 1008
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.37377115045732e-05,
+      "loss": 1.7051,
+      "step": 1009
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.367145535162813e-05,
+      "loss": 1.5882,
+      "step": 1010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.360517324226676e-05,
+      "loss": 1.5245,
+      "step": 1011
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.353886530233127e-05,
+      "loss": 1.5845,
+      "step": 1012
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.34725316577129e-05,
+      "loss": 1.5743,
+      "step": 1013
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.340617243435172e-05,
+      "loss": 1.5399,
+      "step": 1014
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.333978775823631e-05,
+      "loss": 1.5741,
+      "step": 1015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.327337775540362e-05,
+      "loss": 1.5954,
+      "step": 1016
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.320694255193867e-05,
+      "loss": 1.4969,
+      "step": 1017
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.314048227397429e-05,
+      "loss": 1.5862,
+      "step": 1018
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.307399704769099e-05,
+      "loss": 1.6609,
+      "step": 1019
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.30074869993166e-05,
+      "loss": 1.5004,
+      "step": 1020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.294095225512603e-05,
+      "loss": 1.4423,
+      "step": 1021
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.28743929414412e-05,
+      "loss": 1.5915,
+      "step": 1022
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.280780918463057e-05,
+      "loss": 1.5054,
+      "step": 1023
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.274120111110905e-05,
+      "loss": 1.6358,
+      "step": 1024
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.26745688473377e-05,
+      "loss": 1.4421,
+      "step": 1025
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.260791251982354e-05,
+      "loss": 1.5379,
+      "step": 1026
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.254123225511923e-05,
+      "loss": 1.5514,
+      "step": 1027
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.247452817982293e-05,
+      "loss": 1.6922,
+      "step": 1028
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.240780042057796e-05,
+      "loss": 1.5313,
+      "step": 1029
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.234104910407263e-05,
+      "loss": 1.6305,
+      "step": 1030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.227427435703997e-05,
+      "loss": 1.5012,
+      "step": 1031
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.220747630625749e-05,
+      "loss": 1.594,
+      "step": 1032
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.214065507854693e-05,
+      "loss": 1.563,
+      "step": 1033
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.20738108007741e-05,
+      "loss": 1.5762,
+      "step": 1034
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.200694359984849e-05,
+      "loss": 1.5093,
+      "step": 1035
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.194005360272316e-05,
+      "loss": 1.5595,
+      "step": 1036
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.187314093639444e-05,
+      "loss": 1.5843,
+      "step": 1037
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.180620572790168e-05,
+      "loss": 1.5871,
+      "step": 1038
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.173924810432705e-05,
+      "loss": 1.6363,
+      "step": 1039
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.167226819279528e-05,
+      "loss": 1.5616,
+      "step": 1040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.16052661204734e-05,
+      "loss": 1.5049,
+      "step": 1041
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.153824201457053e-05,
+      "loss": 1.604,
+      "step": 1042
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.147119600233758e-05,
+      "loss": 1.4204,
+      "step": 1043
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.140412821106714e-05,
+      "loss": 1.5386,
+      "step": 1044
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.133703876809305e-05,
+      "loss": 1.5855,
+      "step": 1045
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.126992780079031e-05,
+      "loss": 1.5554,
+      "step": 1046
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.12027954365748e-05,
+      "loss": 1.5248,
+      "step": 1047
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.113564180290298e-05,
+      "loss": 1.6472,
+      "step": 1048
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.106846702727172e-05,
+      "loss": 1.567,
+      "step": 1049
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.100127123721803e-05,
+      "loss": 1.529,
+      "step": 1050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.09340545603188e-05,
+      "loss": 1.7353,
+      "step": 1051
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.086681712419058e-05,
+      "loss": 1.5389,
+      "step": 1052
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.079955905648934e-05,
+      "loss": 1.5389,
+      "step": 1053
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0732280484910255e-05,
+      "loss": 1.6046,
+      "step": 1054
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.066498153718735e-05,
+      "loss": 1.5273,
+      "step": 1055
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0597662341093385e-05,
+      "loss": 1.5414,
+      "step": 1056
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.053032302443959e-05,
+      "loss": 1.5633,
+      "step": 1057
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.046296371507533e-05,
+      "loss": 1.7131,
+      "step": 1058
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0395584540887963e-05,
+      "loss": 1.4838,
+      "step": 1059
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0328185629802595e-05,
+      "loss": 1.6632,
+      "step": 1060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.026076710978171e-05,
+      "loss": 1.5782,
+      "step": 1061
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.019332910882514e-05,
+      "loss": 1.5389,
+      "step": 1062
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.012587175496961e-05,
+      "loss": 1.5961,
+      "step": 1063
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.005839517628862e-05,
+      "loss": 1.5782,
+      "step": 1064
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.999089950089218e-05,
+      "loss": 1.5464,
+      "step": 1065
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.992338485692657e-05,
+      "loss": 1.569,
+      "step": 1066
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.985585137257401e-05,
+      "loss": 1.5777,
+      "step": 1067
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.978829917605259e-05,
+      "loss": 1.5283,
+      "step": 1068
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9720728395615875e-05,
+      "loss": 1.3582,
+      "step": 1069
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.965313915955268e-05,
+      "loss": 1.6018,
+      "step": 1070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.958553159618693e-05,
+      "loss": 1.659,
+      "step": 1071
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.951790583387731e-05,
+      "loss": 1.5892,
+      "step": 1072
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.945026200101702e-05,
+      "loss": 1.5429,
+      "step": 1073
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.938260022603366e-05,
+      "loss": 1.5478,
+      "step": 1074
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9314920637388815e-05,
+      "loss": 1.5296,
+      "step": 1075
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.924722336357793e-05,
+      "loss": 1.491,
+      "step": 1076
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9179508533130004e-05,
+      "loss": 1.5646,
+      "step": 1077
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.911177627460739e-05,
+      "loss": 1.6033,
+      "step": 1078
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.90440267166055e-05,
+      "loss": 1.6311,
+      "step": 1079
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.897625998775263e-05,
+      "loss": 1.473,
+      "step": 1080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.890847621670966e-05,
+      "loss": 1.7455,
+      "step": 1081
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.884067553216981e-05,
+      "loss": 1.586,
+      "step": 1082
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8772858062858416e-05,
+      "loss": 1.5003,
+      "step": 1083
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.870502393753272e-05,
+      "loss": 1.5866,
+      "step": 1084
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8637173284981526e-05,
+      "loss": 1.602,
+      "step": 1085
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.856930623402506e-05,
+      "loss": 1.5774,
+      "step": 1086
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.850142291351466e-05,
+      "loss": 1.5952,
+      "step": 1087
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8433523452332574e-05,
+      "loss": 1.5383,
+      "step": 1088
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8365607979391666e-05,
+      "loss": 1.7057,
+      "step": 1089
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8297676623635235e-05,
+      "loss": 1.5915,
+      "step": 1090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8229729514036705e-05,
+      "loss": 1.6047,
+      "step": 1091
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.816176677959941e-05,
+      "loss": 1.5378,
+      "step": 1092
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.809378854935639e-05,
+      "loss": 1.5934,
+      "step": 1093
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.802579495237004e-05,
+      "loss": 1.5984,
+      "step": 1094
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.795778611773197e-05,
+      "loss": 1.507,
+      "step": 1095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.788976217456274e-05,
+      "loss": 1.4118,
+      "step": 1096
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.782172325201155e-05,
+      "loss": 1.5406,
+      "step": 1097
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.775366947925606e-05,
+      "loss": 1.5474,
+      "step": 1098
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.768560098550213e-05,
+      "loss": 1.5733,
+      "step": 1099
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7617517899983553e-05,
+      "loss": 1.5006,
+      "step": 1100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7549420351961844e-05,
+      "loss": 1.5979,
+      "step": 1101
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.748130847072598e-05,
+      "loss": 1.7308,
+      "step": 1102
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.74131823855921e-05,
+      "loss": 1.4553,
+      "step": 1103
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.734504222590338e-05,
+      "loss": 1.6049,
+      "step": 1104
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.727688812102967e-05,
+      "loss": 1.6139,
+      "step": 1105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.720872020036734e-05,
+      "loss": 1.5614,
+      "step": 1106
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.714053859333893e-05,
+      "loss": 1.616,
+      "step": 1107
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7072343429393015e-05,
+      "loss": 1.4883,
+      "step": 1108
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.70041348380039e-05,
+      "loss": 1.7537,
+      "step": 1109
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6935912948671356e-05,
+      "loss": 1.535,
+      "step": 1110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.686767789092041e-05,
+      "loss": 1.5326,
+      "step": 1111
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.679942979430114e-05,
+      "loss": 1.5598,
+      "step": 1112
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.67311687883883e-05,
+      "loss": 1.5263,
+      "step": 1113
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.666289500278121e-05,
+      "loss": 1.5751,
+      "step": 1114
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6594608567103456e-05,
+      "loss": 1.5292,
+      "step": 1115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6526309611002594e-05,
+      "loss": 1.5111,
+      "step": 1116
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.645799826414997e-05,
+      "loss": 1.6729,
+      "step": 1117
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.638967465624051e-05,
+      "loss": 1.585,
+      "step": 1118
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6321338916992315e-05,
+      "loss": 1.6419,
+      "step": 1119
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.625299117614662e-05,
+      "loss": 1.5745,
+      "step": 1120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.618463156346739e-05,
+      "loss": 1.4392,
+      "step": 1121
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.611626020874113e-05,
+      "loss": 1.4463,
+      "step": 1122
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.604787724177666e-05,
+      "loss": 1.6239,
+      "step": 1123
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.597948279240484e-05,
+      "loss": 1.4475,
+      "step": 1124
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5911076990478304e-05,
+      "loss": 1.585,
+      "step": 1125
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.584265996587129e-05,
+      "loss": 1.4737,
+      "step": 1126
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.577423184847932e-05,
+      "loss": 1.4514,
+      "step": 1127
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.570579276821893e-05,
+      "loss": 1.6391,
+      "step": 1128
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5637342855027554e-05,
+      "loss": 1.5036,
+      "step": 1129
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.556888223886315e-05,
+      "loss": 1.5177,
+      "step": 1130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.550041104970397e-05,
+      "loss": 1.6811,
+      "step": 1131
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.543192941754839e-05,
+      "loss": 1.578,
+      "step": 1132
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5363437472414595e-05,
+      "loss": 1.5364,
+      "step": 1133
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.529493534434031e-05,
+      "loss": 1.5801,
+      "step": 1134
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.522642316338268e-05,
+      "loss": 1.5686,
+      "step": 1135
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.515790105961786e-05,
+      "loss": 1.4798,
+      "step": 1136
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.508936916314086e-05,
+      "loss": 1.6068,
+      "step": 1137
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5020827604065294e-05,
+      "loss": 1.6175,
+      "step": 1138
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.495227651252315e-05,
+      "loss": 1.478,
+      "step": 1139
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4883716018664433e-05,
+      "loss": 1.7608,
+      "step": 1140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.481514625265709e-05,
+      "loss": 1.5536,
+      "step": 1141
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4746567344686626e-05,
+      "loss": 1.4985,
+      "step": 1142
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.467797942495589e-05,
+      "loss": 1.4175,
+      "step": 1143
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.460938262368487e-05,
+      "loss": 1.6086,
+      "step": 1144
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.454077707111042e-05,
+      "loss": 1.4963,
+      "step": 1145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4472162897485956e-05,
+      "loss": 1.6195,
+      "step": 1146
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.440354023308134e-05,
+      "loss": 1.6078,
+      "step": 1147
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4334909208182496e-05,
+      "loss": 1.5182,
+      "step": 1148
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.426626995309123e-05,
+      "loss": 1.5639,
+      "step": 1149
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.419762259812501e-05,
+      "loss": 1.5548,
+      "step": 1150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4128967273616625e-05,
+      "loss": 1.6629,
+      "step": 1151
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.406030410991405e-05,
+      "loss": 1.5742,
+      "step": 1152
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.39916332373801e-05,
+      "loss": 1.5801,
+      "step": 1153
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.392295478639225e-05,
+      "loss": 1.4951,
+      "step": 1154
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3854268887342374e-05,
+      "loss": 1.4791,
+      "step": 1155
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.378557567063646e-05,
+      "loss": 1.5978,
+      "step": 1156
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.371687526669439e-05,
+      "loss": 1.5043,
+      "step": 1157
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.364816780594971e-05,
+      "loss": 1.5813,
+      "step": 1158
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.357945341884936e-05,
+      "loss": 1.6181,
+      "step": 1159
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3510732235853414e-05,
+      "loss": 1.5041,
+      "step": 1160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.344200438743489e-05,
+      "loss": 1.5512,
+      "step": 1161
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3373270004079376e-05,
+      "loss": 1.5777,
+      "step": 1162
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.330452921628497e-05,
+      "loss": 1.6428,
+      "step": 1163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.323578215456188e-05,
+      "loss": 1.3375,
+      "step": 1164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.316702894943221e-05,
+      "loss": 1.5912,
+      "step": 1165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3098269731429736e-05,
+      "loss": 1.5444,
+      "step": 1166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3029504631099694e-05,
+      "loss": 1.6103,
+      "step": 1167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.296073377899842e-05,
+      "loss": 1.5123,
+      "step": 1168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2891957305693205e-05,
+      "loss": 1.6114,
+      "step": 1169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2823175341762044e-05,
+      "loss": 1.5733,
+      "step": 1170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2754388017793274e-05,
+      "loss": 1.6163,
+      "step": 1171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.26855954643855e-05,
+      "loss": 1.5909,
+      "step": 1172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.26167978121472e-05,
+      "loss": 1.6057,
+      "step": 1173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.254799519169653e-05,
+      "loss": 1.4897,
+      "step": 1174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.247918773366112e-05,
+      "loss": 1.6109,
+      "step": 1175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.241037556867775e-05,
+      "loss": 1.3993,
+      "step": 1176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.234155882739212e-05,
+      "loss": 1.5742,
+      "step": 1177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.227273764045868e-05,
+      "loss": 1.531,
+      "step": 1178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.220391213854028e-05,
+      "loss": 1.5918,
+      "step": 1179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2135082452307936e-05,
+      "loss": 1.5351,
+      "step": 1180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2066248712440656e-05,
+      "loss": 1.5032,
+      "step": 1181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1997411049625155e-05,
+      "loss": 1.716,
+      "step": 1182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1928569594555524e-05,
+      "loss": 1.5577,
+      "step": 1183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.185972447793312e-05,
+      "loss": 1.5232,
+      "step": 1184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.179087583046625e-05,
+      "loss": 1.5411,
+      "step": 1185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1722023782869856e-05,
+      "loss": 1.5176,
+      "step": 1186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.165316846586541e-05,
+      "loss": 1.5867,
+      "step": 1187
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.158431001018058e-05,
+      "loss": 1.5173,
+      "step": 1188
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.151544854654895e-05,
+      "loss": 1.482,
+      "step": 1189
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.144658420570986e-05,
+      "loss": 1.6682,
+      "step": 1190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1377717118408105e-05,
+      "loss": 1.5763,
+      "step": 1191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1308847415393666e-05,
+      "loss": 1.5118,
+      "step": 1192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.123997522742151e-05,
+      "loss": 1.5753,
+      "step": 1193
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.117110068525133e-05,
+      "loss": 1.5815,
+      "step": 1194
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.110222391964728e-05,
+      "loss": 1.3094,
+      "step": 1195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.103334506137772e-05,
+      "loss": 1.5408,
+      "step": 1196
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.096446424121502e-05,
+      "loss": 1.6129,
+      "step": 1197
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0895581589935235e-05,
+      "loss": 1.5092,
+      "step": 1198
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0826697238317935e-05,
+      "loss": 1.5987,
+      "step": 1199
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0757811317145867e-05,
+      "loss": 1.5941,
+      "step": 1200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.068892395720483e-05,
+      "loss": 1.6754,
+      "step": 1201
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0620035289283275e-05,
+      "loss": 1.5894,
+      "step": 1202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0551145444172186e-05,
+      "loss": 1.6034,
+      "step": 1203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.048225455266479e-05,
+      "loss": 1.5021,
+      "step": 1204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.041336274555625e-05,
+      "loss": 1.5418,
+      "step": 1205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.034447015364352e-05,
+      "loss": 1.5765,
+      "step": 1206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.027557690772503e-05,
+      "loss": 1.4476,
+      "step": 1207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0206683138600416e-05,
+      "loss": 1.5783,
+      "step": 1208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0137788977070353e-05,
+      "loss": 1.5853,
+      "step": 1209
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0068894553936266e-05,
+      "loss": 1.4926,
+      "step": 1210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5e-05,
+      "loss": 1.5935,
+      "step": 1211
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.993110544606376e-05,
+      "loss": 1.5968,
+      "step": 1212
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.986221102292965e-05,
+      "loss": 1.4775,
+      "step": 1213
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.97933168613996e-05,
+      "loss": 1.4535,
+      "step": 1214
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.972442309227498e-05,
+      "loss": 1.5586,
+      "step": 1215
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.965552984635649e-05,
+      "loss": 1.5028,
+      "step": 1216
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9586637254443756e-05,
+      "loss": 1.5954,
+      "step": 1217
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9517745447335235e-05,
+      "loss": 1.6283,
+      "step": 1218
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9448854555827825e-05,
+      "loss": 1.4951,
+      "step": 1219
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.937996471071675e-05,
+      "loss": 1.5698,
+      "step": 1220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9311076042795185e-05,
+      "loss": 1.6098,
+      "step": 1221
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.924218868285414e-05,
+      "loss": 1.5167,
+      "step": 1222
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.917330276168208e-05,
+      "loss": 1.6744,
+      "step": 1223
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.910441841006478e-05,
+      "loss": 1.5144,
+      "step": 1224
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.903553575878499e-05,
+      "loss": 1.5926,
+      "step": 1225
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8966654938622295e-05,
+      "loss": 1.5643,
+      "step": 1226
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.889777608035273e-05,
+      "loss": 1.5231,
+      "step": 1227
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.882889931474868e-05,
+      "loss": 1.4892,
+      "step": 1228
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.87600247725785e-05,
+      "loss": 1.5352,
+      "step": 1229
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.869115258460635e-05,
+      "loss": 1.6318,
+      "step": 1230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8622282881591906e-05,
+      "loss": 1.4968,
+      "step": 1231
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8553415794290146e-05,
+      "loss": 1.7116,
+      "step": 1232
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.848455145345105e-05,
+      "loss": 1.5696,
+      "step": 1233
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.841568998981943e-05,
+      "loss": 1.5007,
+      "step": 1234
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.834683153413459e-05,
+      "loss": 1.597,
+      "step": 1235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.827797621713016e-05,
+      "loss": 1.526,
+      "step": 1236
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.820912416953377e-05,
+      "loss": 1.4841,
+      "step": 1237
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.814027552206689e-05,
+      "loss": 1.5374,
+      "step": 1238
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8071430405444474e-05,
+      "loss": 1.3909,
+      "step": 1239
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.800258895037486e-05,
+      "loss": 1.4859,
+      "step": 1240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7933751287559335e-05,
+      "loss": 1.5874,
+      "step": 1241
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7864917547692076e-05,
+      "loss": 1.5641,
+      "step": 1242
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.779608786145974e-05,
+      "loss": 1.5094,
+      "step": 1243
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.772726235954133e-05,
+      "loss": 1.6987,
+      "step": 1244
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7658441172607876e-05,
+      "loss": 1.5026,
+      "step": 1245
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.758962443132226e-05,
+      "loss": 1.575,
+      "step": 1246
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7520812266338885e-05,
+      "loss": 1.5714,
+      "step": 1247
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7452004808303474e-05,
+      "loss": 1.5751,
+      "step": 1248
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.738320218785281e-05,
+      "loss": 1.5218,
+      "step": 1249
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7314404535614515e-05,
+      "loss": 1.4894,
+      "step": 1250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7245611982206724e-05,
+      "loss": 1.7143,
+      "step": 1251
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7176824658237975e-05,
+      "loss": 1.4809,
+      "step": 1252
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.710804269430681e-05,
+      "loss": 1.5537,
+      "step": 1253
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.70392662210016e-05,
+      "loss": 1.6282,
+      "step": 1254
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.697049536890033e-05,
+      "loss": 1.4899,
+      "step": 1255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6901730268570275e-05,
+      "loss": 1.5574,
+      "step": 1256
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.683297105056782e-05,
+      "loss": 1.5822,
+      "step": 1257
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.676421784543813e-05,
+      "loss": 1.4595,
+      "step": 1258
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.669547078371504e-05,
+      "loss": 1.5714,
+      "step": 1259
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.662672999592063e-05,
+      "loss": 1.5152,
+      "step": 1260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6557995612565144e-05,
+      "loss": 1.5312,
+      "step": 1261
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.648926776414659e-05,
+      "loss": 1.6865,
+      "step": 1262
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.642054658115067e-05,
+      "loss": 1.5551,
+      "step": 1263
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.63518321940503e-05,
+      "loss": 1.3775,
+      "step": 1264
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6283124733305624e-05,
+      "loss": 1.563,
+      "step": 1265
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.621442432936355e-05,
+      "loss": 1.5586,
+      "step": 1266
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6145731112657644e-05,
+      "loss": 1.4852,
+      "step": 1267
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.607704521360776e-05,
+      "loss": 1.4784,
+      "step": 1268
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6008366762619926e-05,
+      "loss": 1.539,
+      "step": 1269
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.593969589008596e-05,
+      "loss": 1.5867,
+      "step": 1270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5871032726383386e-05,
+      "loss": 1.5084,
+      "step": 1271
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5802377401875e-05,
+      "loss": 1.5324,
+      "step": 1272
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.573373004690878e-05,
+      "loss": 1.5141,
+      "step": 1273
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5665090791817516e-05,
+      "loss": 1.706,
+      "step": 1274
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.559645976691868e-05,
+      "loss": 1.5468,
+      "step": 1275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5527837102514035e-05,
+      "loss": 1.4414,
+      "step": 1276
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.545922292888959e-05,
+      "loss": 1.6068,
+      "step": 1277
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.539061737631513e-05,
+      "loss": 1.4261,
+      "step": 1278
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5322020575044114e-05,
+      "loss": 1.4779,
+      "step": 1279
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.525343265531338e-05,
+      "loss": 1.5726,
+      "step": 1280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.518485374734292e-05,
+      "loss": 1.5148,
+      "step": 1281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.511628398133556e-05,
+      "loss": 1.5119,
+      "step": 1282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.504772348747687e-05,
+      "loss": 1.6746,
+      "step": 1283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4979172395934704e-05,
+      "loss": 1.5556,
+      "step": 1284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.491063083685916e-05,
+      "loss": 1.5448,
+      "step": 1285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4842098940382155e-05,
+      "loss": 1.5478,
+      "step": 1286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.477357683661734e-05,
+      "loss": 1.4953,
+      "step": 1287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4705064655659685e-05,
+      "loss": 1.5476,
+      "step": 1288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.463656252758542e-05,
+      "loss": 1.5594,
+      "step": 1289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.456807058245162e-05,
+      "loss": 1.4588,
+      "step": 1290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.449958895029604e-05,
+      "loss": 1.6569,
+      "step": 1291
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.443111776113686e-05,
+      "loss": 1.5939,
+      "step": 1292
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.436265714497245e-05,
+      "loss": 1.5054,
+      "step": 1293
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4294207231781085e-05,
+      "loss": 1.5732,
+      "step": 1294
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4225768151520694e-05,
+      "loss": 1.3938,
+      "step": 1295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.415734003412873e-05,
+      "loss": 1.4968,
+      "step": 1296
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.408892300952171e-05,
+      "loss": 1.5463,
+      "step": 1297
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.402051720759519e-05,
+      "loss": 1.6073,
+      "step": 1298
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3952122758223354e-05,
+      "loss": 1.465,
+      "step": 1299
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.388373979125889e-05,
+      "loss": 1.5632,
+      "step": 1300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.381536843653262e-05,
+      "loss": 1.4961,
+      "step": 1301
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3747008823853404e-05,
+      "loss": 1.4911,
+      "step": 1302
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.367866108300769e-05,
+      "loss": 1.6865,
+      "step": 1303
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.361032534375952e-05,
+      "loss": 1.511,
+      "step": 1304
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3542001735850034e-05,
+      "loss": 1.5166,
+      "step": 1305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.347369038899744e-05,
+      "loss": 1.5893,
+      "step": 1306
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3405391432896555e-05,
+      "loss": 1.5645,
+      "step": 1307
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.33371049972188e-05,
+      "loss": 1.4952,
+      "step": 1308
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3268831211611707e-05,
+      "loss": 1.4873,
+      "step": 1309
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.320057020569888e-05,
+      "loss": 1.6108,
+      "step": 1310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3132322109079596e-05,
+      "loss": 1.5006,
+      "step": 1311
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.306408705132867e-05,
+      "loss": 1.5605,
+      "step": 1312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2995865161996105e-05,
+      "loss": 1.6506,
+      "step": 1313
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2927656570607e-05,
+      "loss": 1.4614,
+      "step": 1314
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2859461406661065e-05,
+      "loss": 1.3399,
+      "step": 1315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2791279799632666e-05,
+      "loss": 1.5508,
+      "step": 1316
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2723111878970326e-05,
+      "loss": 1.4846,
+      "step": 1317
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2654957774096634e-05,
+      "loss": 1.6148,
+      "step": 1318
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2586817614407895e-05,
+      "loss": 1.4067,
+      "step": 1319
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.251869152927404e-05,
+      "loss": 1.4975,
+      "step": 1320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2450579648038154e-05,
+      "loss": 1.5414,
+      "step": 1321
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.238248210001645e-05,
+      "loss": 1.6945,
+      "step": 1322
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.231439901449788e-05,
+      "loss": 1.4683,
+      "step": 1323
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2246330520743955e-05,
+      "loss": 1.5995,
+      "step": 1324
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2178276747988446e-05,
+      "loss": 1.5217,
+      "step": 1325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.211023782543727e-05,
+      "loss": 1.513,
+      "step": 1326
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2042213882268025e-05,
+      "loss": 1.6065,
+      "step": 1327
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.197420504762998e-05,
+      "loss": 1.5369,
+      "step": 1328
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.190621145064363e-05,
+      "loss": 1.514,
+      "step": 1329
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.18382332204006e-05,
+      "loss": 1.5595,
+      "step": 1330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.17702704859633e-05,
+      "loss": 1.5493,
+      "step": 1331
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.170232337636477e-05,
+      "loss": 1.4891,
+      "step": 1332
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.163439202060833e-05,
+      "loss": 1.7059,
+      "step": 1333
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.156647654766744e-05,
+      "loss": 1.5396,
+      "step": 1334
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.149857708648536e-05,
+      "loss": 1.4814,
+      "step": 1335
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.143069376597496e-05,
+      "loss": 1.5729,
+      "step": 1336
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.13628267150185e-05,
+      "loss": 1.4283,
+      "step": 1337
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.129497606246729e-05,
+      "loss": 1.4972,
+      "step": 1338
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.12271419371416e-05,
+      "loss": 1.555,
+      "step": 1339
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1159324467830205e-05,
+      "loss": 1.5836,
+      "step": 1340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.109152378329036e-05,
+      "loss": 1.4568,
+      "step": 1341
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.102374001224738e-05,
+      "loss": 1.5989,
+      "step": 1342
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.095597328339452e-05,
+      "loss": 1.5373,
+      "step": 1343
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.088822372539263e-05,
+      "loss": 1.4991,
+      "step": 1344
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.082049146687003e-05,
+      "loss": 1.6511,
+      "step": 1345
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.075277663642208e-05,
+      "loss": 1.459,
+      "step": 1346
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0685079362611204e-05,
+      "loss": 1.5207,
+      "step": 1347
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.061739977396635e-05,
+      "loss": 1.5251,
+      "step": 1348
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0549737998983e-05,
+      "loss": 1.5604,
+      "step": 1349
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.048209416612271e-05,
+      "loss": 1.4828,
+      "step": 1350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0414468403813095e-05,
+      "loss": 1.6056,
+      "step": 1351
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.034686084044732e-05,
+      "loss": 1.7367,
+      "step": 1352
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0279271604384144e-05,
+      "loss": 1.4576,
+      "step": 1353
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.021170082394741e-05,
+      "loss": 1.607,
+      "step": 1354
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0144148627425993e-05,
+      "loss": 1.553,
+      "step": 1355
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0076615143073436e-05,
+      "loss": 1.4784,
+      "step": 1356
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0009100499107824e-05,
+      "loss": 1.5436,
+      "step": 1357
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.994160482371138e-05,
+      "loss": 1.5053,
+      "step": 1358
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9874128245030404e-05,
+      "loss": 1.4758,
+      "step": 1359
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.980667089117487e-05,
+      "loss": 1.565,
+      "step": 1360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.973923289021829e-05,
+      "loss": 1.4946,
+      "step": 1361
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.967181437019741e-05,
+      "loss": 1.5622,
+      "step": 1362
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.960441545911204e-05,
+      "loss": 1.3324,
+      "step": 1363
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9537036284924675e-05,
+      "loss": 1.5534,
+      "step": 1364
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.946967697556042e-05,
+      "loss": 1.6262,
+      "step": 1365
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.940233765890661e-05,
+      "loss": 1.5893,
+      "step": 1366
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.933501846281267e-05,
+      "loss": 1.44,
+      "step": 1367
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.926771951508974e-05,
+      "loss": 1.5287,
+      "step": 1368
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9200440943510665e-05,
+      "loss": 1.4991,
+      "step": 1369
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.913318287580942e-05,
+      "loss": 1.4781,
+      "step": 1370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9065945439681214e-05,
+      "loss": 1.4725,
+      "step": 1371
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8998728762781975e-05,
+      "loss": 1.5283,
+      "step": 1372
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8931532972728285e-05,
+      "loss": 1.5888,
+      "step": 1373
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.886435819709704e-05,
+      "loss": 1.4557,
+      "step": 1374
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.879720456342521e-05,
+      "loss": 1.6987,
+      "step": 1375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.87300721992097e-05,
+      "loss": 1.5602,
+      "step": 1376
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.866296123190696e-05,
+      "loss": 1.4421,
+      "step": 1377
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.859587178893288e-05,
+      "loss": 1.5744,
+      "step": 1378
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.852880399766243e-05,
+      "loss": 1.5694,
+      "step": 1379
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8461757985429505e-05,
+      "loss": 1.4874,
+      "step": 1380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.839473387952662e-05,
+      "loss": 1.5525,
+      "step": 1381
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.832773180720475e-05,
+      "loss": 1.4656,
+      "step": 1382
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.826075189567296e-05,
+      "loss": 1.6839,
+      "step": 1383
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.819379427209834e-05,
+      "loss": 1.5556,
+      "step": 1384
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.812685906360557e-05,
+      "loss": 1.5347,
+      "step": 1385
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8059946397276854e-05,
+      "loss": 1.4641,
+      "step": 1386
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.799305640015152e-05,
+      "loss": 1.5629,
+      "step": 1387
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7926189199225916e-05,
+      "loss": 1.4597,
+      "step": 1388
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7859344921453064e-05,
+      "loss": 1.5299,
+      "step": 1389
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.779252369374253e-05,
+      "loss": 1.3412,
+      "step": 1390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.772572564296005e-05,
+      "loss": 1.4803,
+      "step": 1391
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7658950895927385e-05,
+      "loss": 1.5418,
+      "step": 1392
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7592199579422035e-05,
+      "loss": 1.5405,
+      "step": 1393
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.752547182017708e-05,
+      "loss": 1.4748,
+      "step": 1394
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7458767744880765e-05,
+      "loss": 1.5591,
+      "step": 1395
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.739208748017647e-05,
+      "loss": 1.6452,
+      "step": 1396
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.73254311526623e-05,
+      "loss": 1.4426,
+      "step": 1397
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.725879888889096e-05,
+      "loss": 1.5313,
+      "step": 1398
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.719219081536942e-05,
+      "loss": 1.5384,
+      "step": 1399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.712560705855881e-05,
+      "loss": 1.5456,
+      "step": 1400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.705904774487396e-05,
+      "loss": 1.4777,
+      "step": 1401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.699251300068342e-05,
+      "loss": 1.525,
+      "step": 1402
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6926002952309016e-05,
+      "loss": 1.7075,
+      "step": 1403
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6859517726025714e-05,
+      "loss": 1.4819,
+      "step": 1404
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.679305744806134e-05,
+      "loss": 1.5036,
+      "step": 1405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.67266222445964e-05,
+      "loss": 1.4362,
+      "step": 1406
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.666021224176369e-05,
+      "loss": 1.5111,
+      "step": 1407
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.659382756564829e-05,
+      "loss": 1.4978,
+      "step": 1408
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.65274683422871e-05,
+      "loss": 1.4125,
+      "step": 1409
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6461134697668745e-05,
+      "loss": 1.4861,
+      "step": 1410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.639482675773324e-05,
+      "loss": 1.5277,
+      "step": 1411
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.632854464837188e-05,
+      "loss": 1.4775,
+      "step": 1412
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.62622884954268e-05,
+      "loss": 1.5874,
+      "step": 1413
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.619605842469097e-05,
+      "loss": 1.4766,
+      "step": 1414
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.612985456190778e-05,
+      "loss": 1.418,
+      "step": 1415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6063677032770846e-05,
+      "loss": 1.3667,
+      "step": 1416
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.599752596292386e-05,
+      "loss": 1.5278,
+      "step": 1417
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.593140147796018e-05,
+      "loss": 1.4032,
+      "step": 1418
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.586530370342279e-05,
+      "loss": 1.4499,
+      "step": 1419
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.579923276480387e-05,
+      "loss": 1.3593,
+      "step": 1420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5733188787544745e-05,
+      "loss": 1.3577,
+      "step": 1421
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.566717189703546e-05,
+      "loss": 1.512,
+      "step": 1422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.56011822186147e-05,
+      "loss": 1.4367,
+      "step": 1423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.553521987756945e-05,
+      "loss": 1.3796,
+      "step": 1424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5469284999134815e-05,
+      "loss": 1.4972,
+      "step": 1425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.540337770849371e-05,
+      "loss": 1.5102,
+      "step": 1426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.533749813077677e-05,
+      "loss": 1.3598,
+      "step": 1427
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.527164639106191e-05,
+      "loss": 1.461,
+      "step": 1428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5205822614374255e-05,
+      "loss": 1.4179,
+      "step": 1429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.514002692568581e-05,
+      "loss": 1.338,
+      "step": 1430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5074259449915284e-05,
+      "loss": 1.4945,
+      "step": 1431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5008520311927774e-05,
+      "loss": 1.4025,
+      "step": 1432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.494280963653463e-05,
+      "loss": 1.3123,
+      "step": 1433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.487712754849313e-05,
+      "loss": 1.5262,
+      "step": 1434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4811474172506275e-05,
+      "loss": 1.4212,
+      "step": 1435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.474584963322257e-05,
+      "loss": 1.3504,
+      "step": 1436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.468025405523576e-05,
+      "loss": 1.3228,
+      "step": 1437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.461468756308459e-05,
+      "loss": 1.4369,
+      "step": 1438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4549150281252636e-05,
+      "loss": 1.3824,
+      "step": 1439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4483642334167946e-05,
+      "loss": 1.4123,
+      "step": 1440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4418163846202944e-05,
+      "loss": 1.4633,
+      "step": 1441
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4352714941674045e-05,
+      "loss": 1.3968,
+      "step": 1442
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4287295744841586e-05,
+      "loss": 1.4149,
+      "step": 1443
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4221906379909396e-05,
+      "loss": 1.3666,
+      "step": 1444
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4156546971024784e-05,
+      "loss": 1.4635,
+      "step": 1445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.409121764227809e-05,
+      "loss": 1.366,
+      "step": 1446
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.40259185177026e-05,
+      "loss": 1.5081,
+      "step": 1447
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.396064972127421e-05,
+      "loss": 1.2738,
+      "step": 1448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.389541137691129e-05,
+      "loss": 1.3841,
+      "step": 1449
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.383020360847432e-05,
+      "loss": 1.453,
+      "step": 1450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3765026539765834e-05,
+      "loss": 1.3019,
+      "step": 1451
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.369988029452996e-05,
+      "loss": 1.4261,
+      "step": 1452
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.363476499645241e-05,
+      "loss": 1.362,
+      "step": 1453
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.356968076916006e-05,
+      "loss": 1.3557,
+      "step": 1454
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.350462773622086e-05,
+      "loss": 1.4786,
+      "step": 1455
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.343960602114349e-05,
+      "loss": 1.4309,
+      "step": 1456
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.337461574737716e-05,
+      "loss": 1.4019,
+      "step": 1457
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.330965703831146e-05,
+      "loss": 1.1956,
+      "step": 1458
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.324473001727597e-05,
+      "loss": 1.4092,
+      "step": 1459
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3179834807540155e-05,
+      "loss": 1.3836,
+      "step": 1460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3114971532313056e-05,
+      "loss": 1.4292,
+      "step": 1461
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3050140314743105e-05,
+      "loss": 1.3136,
+      "step": 1462
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.298534127791785e-05,
+      "loss": 1.4446,
+      "step": 1463
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.292057454486378e-05,
+      "loss": 1.4092,
+      "step": 1464
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2855840238546e-05,
+      "loss": 1.3022,
+      "step": 1465
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.279113848186808e-05,
+      "loss": 1.4334,
+      "step": 1466
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.272646939767179e-05,
+      "loss": 1.4287,
+      "step": 1467
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2661833108736875e-05,
+      "loss": 1.2757,
+      "step": 1468
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2597229737780774e-05,
+      "loss": 1.4227,
+      "step": 1469
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2532659407458513e-05,
+      "loss": 1.2873,
+      "step": 1470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2468122240362284e-05,
+      "loss": 1.4379,
+      "step": 1471
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2403618359021405e-05,
+      "loss": 1.2342,
+      "step": 1472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.233914788590192e-05,
+      "loss": 1.4457,
+      "step": 1473
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2274710943406525e-05,
+      "loss": 1.3121,
+      "step": 1474
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.221030765387417e-05,
+      "loss": 1.4212,
+      "step": 1475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.214593813958001e-05,
+      "loss": 1.4628,
+      "step": 1476
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2081602522734986e-05,
+      "loss": 1.3613,
+      "step": 1477
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.201730092548574e-05,
+      "loss": 1.3214,
+      "step": 1478
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1953033469914276e-05,
+      "loss": 1.407,
+      "step": 1479
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.188880027803782e-05,
+      "loss": 1.3477,
+      "step": 1480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.18246014718085e-05,
+      "loss": 1.4678,
+      "step": 1481
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.176043717311324e-05,
+      "loss": 1.3635,
+      "step": 1482
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.169630750377337e-05,
+      "loss": 1.2828,
+      "step": 1483
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.163221258554451e-05,
+      "loss": 1.4678,
+      "step": 1484
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.156815254011627e-05,
+      "loss": 1.4467,
+      "step": 1485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1504127489112104e-05,
+      "loss": 1.3361,
+      "step": 1486
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.144013755408895e-05,
+      "loss": 1.3797,
+      "step": 1487
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.137618285653719e-05,
+      "loss": 1.2963,
+      "step": 1488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.131226351788018e-05,
+      "loss": 1.3071,
+      "step": 1489
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1248379659474225e-05,
+      "loss": 1.3844,
+      "step": 1490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.118453140260823e-05,
+      "loss": 1.4299,
+      "step": 1491
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1120718868503536e-05,
+      "loss": 1.2831,
+      "step": 1492
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.105694217831361e-05,
+      "loss": 1.43,
+      "step": 1493
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.099320145312396e-05,
+      "loss": 1.3658,
+      "step": 1494
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0929496813951694e-05,
+      "loss": 1.344,
+      "step": 1495
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.086582838174551e-05,
+      "loss": 1.4051,
+      "step": 1496
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0802196277385317e-05,
+      "loss": 1.3891,
+      "step": 1497
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0738600621682025e-05,
+      "loss": 1.3928,
+      "step": 1498
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0675041535377405e-05,
+      "loss": 1.358,
+      "step": 1499
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.061151913914373e-05,
+      "loss": 1.3599,
+      "step": 1500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0548033553583705e-05,
+      "loss": 1.3153,
+      "step": 1501
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.048458489923005e-05,
+      "loss": 1.3814,
+      "step": 1502
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.042117329654544e-05,
+      "loss": 1.4585,
+      "step": 1503
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.035779886592215e-05,
+      "loss": 1.3374,
+      "step": 1504
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0294461727681932e-05,
+      "loss": 1.4097,
+      "step": 1505
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0231162002075673e-05,
+      "loss": 1.4129,
+      "step": 1506
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0167899809283308e-05,
+      "loss": 1.3102,
+      "step": 1507
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.010467526941344e-05,
+      "loss": 1.2762,
+      "step": 1508
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.004148850250323e-05,
+      "loss": 1.3858,
+      "step": 1509
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9978339628518092e-05,
+      "loss": 1.3389,
+      "step": 1510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.991522876735154e-05,
+      "loss": 1.3781,
+      "step": 1511
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.985215603882485e-05,
+      "loss": 1.4067,
+      "step": 1512
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.978912156268699e-05,
+      "loss": 1.344,
+      "step": 1513
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9726125458614217e-05,
+      "loss": 1.3195,
+      "step": 1514
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9663167846209998e-05,
+      "loss": 1.3472,
+      "step": 1515
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.960024884500467e-05,
+      "loss": 1.3579,
+      "step": 1516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9537368574455304e-05,
+      "loss": 1.409,
+      "step": 1517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9474527153945403e-05,
+      "loss": 1.3187,
+      "step": 1518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9411724702784758e-05,
+      "loss": 1.3516,
+      "step": 1519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.934896134020912e-05,
+      "loss": 1.3959,
+      "step": 1520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.928623718538006e-05,
+      "loss": 1.3416,
+      "step": 1521
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9223552357384693e-05,
+      "loss": 1.3379,
+      "step": 1522
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.916090697523549e-05,
+      "loss": 1.3981,
+      "step": 1523
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.909830115787e-05,
+      "loss": 1.4139,
+      "step": 1524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9035735024150674e-05,
+      "loss": 1.3186,
+      "step": 1525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8973208692864624e-05,
+      "loss": 1.384,
+      "step": 1526
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.89107222827234e-05,
+      "loss": 1.3954,
+      "step": 1527
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.88482759123627e-05,
+      "loss": 1.3102,
+      "step": 1528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.878586970034232e-05,
+      "loss": 1.4096,
+      "step": 1529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8723503765145687e-05,
+      "loss": 1.3841,
+      "step": 1530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.866117822517982e-05,
+      "loss": 1.3104,
+      "step": 1531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8598893198775044e-05,
+      "loss": 1.3689,
+      "step": 1532
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8536648804184785e-05,
+      "loss": 1.2021,
+      "step": 1533
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8474445159585235e-05,
+      "loss": 1.3374,
+      "step": 1534
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8412282383075363e-05,
+      "loss": 1.4049,
+      "step": 1535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8350160592676406e-05,
+      "loss": 1.4169,
+      "step": 1536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8288079906331864e-05,
+      "loss": 1.3385,
+      "step": 1537
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.822604044190721e-05,
+      "loss": 1.4089,
+      "step": 1538
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8164042317189575e-05,
+      "loss": 1.289,
+      "step": 1539
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8102085649887678e-05,
+      "loss": 1.3479,
+      "step": 1540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.804017055763149e-05,
+      "loss": 1.4122,
+      "step": 1541
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7978297157972066e-05,
+      "loss": 1.3866,
+      "step": 1542
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.79164655683813e-05,
+      "loss": 1.2982,
+      "step": 1543
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7854675906251725e-05,
+      "loss": 1.3153,
+      "step": 1544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7792928288896202e-05,
+      "loss": 1.3549,
+      "step": 1545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7731222833547844e-05,
+      "loss": 1.3225,
+      "step": 1546
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.766955965735968e-05,
+      "loss": 1.41,
+      "step": 1547
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.760793887740447e-05,
+      "loss": 1.3507,
+      "step": 1548
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7546360610674493e-05,
+      "loss": 1.3366,
+      "step": 1549
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7484824974081324e-05,
+      "loss": 1.4156,
+      "step": 1550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7423332084455544e-05,
+      "loss": 1.3804,
+      "step": 1551
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.736188205854663e-05,
+      "loss": 1.27,
+      "step": 1552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7300475013022663e-05,
+      "loss": 1.433,
+      "step": 1553
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7239111064470125e-05,
+      "loss": 1.3887,
+      "step": 1554
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.717779032939367e-05,
+      "loss": 1.3505,
+      "step": 1555
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.711651292421593e-05,
+      "loss": 1.3815,
+      "step": 1556
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.705527896527721e-05,
+      "loss": 1.37,
+      "step": 1557
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6994088568835384e-05,
+      "loss": 1.188,
+      "step": 1558
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.693294185106562e-05,
+      "loss": 1.4356,
+      "step": 1559
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6871838928060115e-05,
+      "loss": 1.3492,
+      "step": 1560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.681077991582797e-05,
+      "loss": 1.3334,
+      "step": 1561
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6749764930294908e-05,
+      "loss": 1.3366,
+      "step": 1562
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6688794087302993e-05,
+      "loss": 1.2963,
+      "step": 1563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6627867502610583e-05,
+      "loss": 1.4194,
+      "step": 1564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.656698529189193e-05,
+      "loss": 1.3884,
+      "step": 1565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6506147570737094e-05,
+      "loss": 1.2757,
+      "step": 1566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.644535445465164e-05,
+      "loss": 1.376,
+      "step": 1567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.638460605905646e-05,
+      "loss": 1.4479,
+      "step": 1568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6323902499287473e-05,
+      "loss": 1.335,
+      "step": 1569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.626324389059561e-05,
+      "loss": 1.3243,
+      "step": 1570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6202630348146324e-05,
+      "loss": 1.3482,
+      "step": 1571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6142061987019577e-05,
+      "loss": 1.3086,
+      "step": 1572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6081538922209535e-05,
+      "loss": 1.3226,
+      "step": 1573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.602106126862438e-05,
+      "loss": 1.3997,
+      "step": 1574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5960629141086012e-05,
+      "loss": 1.368,
+      "step": 1575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.590024265433002e-05,
+      "loss": 1.2899,
+      "step": 1576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5839901923005205e-05,
+      "loss": 1.4291,
+      "step": 1577
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5779607061673572e-05,
+      "loss": 1.3476,
+      "step": 1578
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.571935818481005e-05,
+      "loss": 1.3252,
+      "step": 1579
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5659155406802193e-05,
+      "loss": 1.3848,
+      "step": 1580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5598998841950107e-05,
+      "loss": 1.3149,
+      "step": 1581
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.553888860446611e-05,
+      "loss": 1.3225,
+      "step": 1582
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.547882480847461e-05,
+      "loss": 1.4138,
+      "step": 1583
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.541880756801176e-05,
+      "loss": 1.2621,
+      "step": 1584
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5358836997025437e-05,
+      "loss": 1.3995,
+      "step": 1585
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.529891320937481e-05,
+      "loss": 1.3902,
+      "step": 1586
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5239036318830278e-05,
+      "loss": 1.2621,
+      "step": 1587
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.517920643907319e-05,
+      "loss": 1.4286,
+      "step": 1588
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.511942368369566e-05,
+      "loss": 1.227,
+      "step": 1589
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5059688166200262e-05,
+      "loss": 1.2689,
+      "step": 1590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.500000000000001e-05,
+      "loss": 1.3806,
+      "step": 1591
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4940359298417894e-05,
+      "loss": 1.4466,
+      "step": 1592
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4880766174686848e-05,
+      "loss": 1.3046,
+      "step": 1593
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.482122074194947e-05,
+      "loss": 1.4231,
+      "step": 1594
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.476172311325783e-05,
+      "loss": 1.3111,
+      "step": 1595
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4702273401573162e-05,
+      "loss": 1.3625,
+      "step": 1596
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4642871719765852e-05,
+      "loss": 1.4257,
+      "step": 1597
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4583518180614972e-05,
+      "loss": 1.3391,
+      "step": 1598
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4524212896808263e-05,
+      "loss": 1.341,
+      "step": 1599
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.446495598094183e-05,
+      "loss": 1.3354,
+      "step": 1600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4405747545519963e-05,
+      "loss": 1.4098,
+      "step": 1601
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4346587702954848e-05,
+      "loss": 1.3341,
+      "step": 1602
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4287476565566527e-05,
+      "loss": 1.2972,
+      "step": 1603
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4228414245582443e-05,
+      "loss": 1.4277,
+      "step": 1604
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4169400855137436e-05,
+      "loss": 1.2848,
+      "step": 1605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4110436506273433e-05,
+      "loss": 1.4211,
+      "step": 1606
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.405152131093926e-05,
+      "loss": 1.399,
+      "step": 1607
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3992655380990346e-05,
+      "loss": 1.2518,
+      "step": 1608
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3933838828188733e-05,
+      "loss": 1.1696,
+      "step": 1609
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3875071764202563e-05,
+      "loss": 1.3881,
+      "step": 1610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.381635430060611e-05,
+      "loss": 1.319,
+      "step": 1611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.375768654887945e-05,
+      "loss": 1.4126,
+      "step": 1612
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3699068620408304e-05,
+      "loss": 1.27,
+      "step": 1613
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.364050062648372e-05,
+      "loss": 1.3148,
+      "step": 1614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3581982678302063e-05,
+      "loss": 1.3491,
+      "step": 1615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.352351488696457e-05,
+      "loss": 1.3268,
+      "step": 1616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3465097363477306e-05,
+      "loss": 1.3206,
+      "step": 1617
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3406730218750893e-05,
+      "loss": 1.4254,
+      "step": 1618
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3348413563600325e-05,
+      "loss": 1.3659,
+      "step": 1619
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3290147508744666e-05,
+      "loss": 1.326,
+      "step": 1620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.323193216480698e-05,
+      "loss": 1.3781,
+      "step": 1621
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.317376764231403e-05,
+      "loss": 1.3801,
+      "step": 1622
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3115654051696095e-05,
+      "loss": 1.2971,
+      "step": 1623
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3057591503286757e-05,
+      "loss": 1.3508,
+      "step": 1624
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2999580107322653e-05,
+      "loss": 1.4117,
+      "step": 1625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.294161997394336e-05,
+      "loss": 1.3046,
+      "step": 1626
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.288371121319109e-05,
+      "loss": 1.3733,
+      "step": 1627
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.282585393501054e-05,
+      "loss": 1.3485,
+      "step": 1628
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2768048249248648e-05,
+      "loss": 1.3148,
+      "step": 1629
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2710294265654423e-05,
+      "loss": 1.4224,
+      "step": 1630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2652592093878666e-05,
+      "loss": 1.2243,
+      "step": 1631
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2594941843473855e-05,
+      "loss": 1.3156,
+      "step": 1632
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2537343623893874e-05,
+      "loss": 1.406,
+      "step": 1633
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.247979754449383e-05,
+      "loss": 1.2626,
+      "step": 1634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.242230371452982e-05,
+      "loss": 1.3801,
+      "step": 1635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2364862243158767e-05,
+      "loss": 1.3593,
+      "step": 1636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2307473239438154e-05,
+      "loss": 1.3553,
+      "step": 1637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.225013681232586e-05,
+      "loss": 1.2649,
+      "step": 1638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.219285307067997e-05,
+      "loss": 1.3877,
+      "step": 1639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2135622123258515e-05,
+      "loss": 1.2913,
+      "step": 1640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.207844407871929e-05,
+      "loss": 1.3537,
+      "step": 1641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.202131904561968e-05,
+      "loss": 1.4199,
+      "step": 1642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.196424713241637e-05,
+      "loss": 1.383,
+      "step": 1643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.190722844746524e-05,
+      "loss": 1.2647,
+      "step": 1644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1850263099021077e-05,
+      "loss": 1.4064,
+      "step": 1645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.179335119523745e-05,
+      "loss": 1.3914,
+      "step": 1646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1736492844166407e-05,
+      "loss": 1.2809,
+      "step": 1647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.167968815375837e-05,
+      "loss": 1.4093,
+      "step": 1648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1622937231861822e-05,
+      "loss": 1.404,
+      "step": 1649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1566240186223212e-05,
+      "loss": 1.2492,
+      "step": 1650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.150959712448669e-05,
+      "loss": 1.4157,
+      "step": 1651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1453008154193903e-05,
+      "loss": 1.3528,
+      "step": 1652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.139647338278381e-05,
+      "loss": 1.2793,
+      "step": 1653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1339992917592476e-05,
+      "loss": 1.4266,
+      "step": 1654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.128356686585282e-05,
+      "loss": 1.2798,
+      "step": 1655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.12271953346945e-05,
+      "loss": 1.3502,
+      "step": 1656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1170878431143637e-05,
+      "loss": 1.2136,
+      "step": 1657
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.111461626212265e-05,
+      "loss": 1.3216,
+      "step": 1658
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.105840893445005e-05,
+      "loss": 1.3462,
+      "step": 1659
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.100225655484017e-05,
+      "loss": 1.3911,
+      "step": 1660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.094615922990309e-05,
+      "loss": 1.3099,
+      "step": 1661
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0890117066144327e-05,
+      "loss": 1.3506,
+      "step": 1662
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0834130169964692e-05,
+      "loss": 1.3322,
+      "step": 1663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.077819864766e-05,
+      "loss": 1.3282,
+      "step": 1664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.072232260542106e-05,
+      "loss": 1.2718,
+      "step": 1665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0666502149333218e-05,
+      "loss": 1.3952,
+      "step": 1666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.061073738537635e-05,
+      "loss": 1.341,
+      "step": 1667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.055502841942459e-05,
+      "loss": 1.3031,
+      "step": 1668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0499375357246148e-05,
+      "loss": 1.3841,
+      "step": 1669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0443778304503024e-05,
+      "loss": 1.3267,
+      "step": 1670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0388237366751006e-05,
+      "loss": 1.2618,
+      "step": 1671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0332752649439217e-05,
+      "loss": 1.4254,
+      "step": 1672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0277324257910106e-05,
+      "loss": 1.3552,
+      "step": 1673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0221952297399176e-05,
+      "loss": 1.2941,
+      "step": 1674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0166636873034805e-05,
+      "loss": 1.3235,
+      "step": 1675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0111378089837956e-05,
+      "loss": 1.2922,
+      "step": 1676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0056176052722174e-05,
+      "loss": 1.359,
+      "step": 1677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0001030866493166e-05,
+      "loss": 1.3951,
+      "step": 1678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9945942635848748e-05,
+      "loss": 1.3455,
+      "step": 1679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9890911465378597e-05,
+      "loss": 1.2529,
+      "step": 1680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9835937459564064e-05,
+      "loss": 1.3744,
+      "step": 1681
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.978102072277791e-05,
+      "loss": 1.2918,
+      "step": 1682
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9726161359284286e-05,
+      "loss": 1.3751,
+      "step": 1683
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.967135947323828e-05,
+      "loss": 1.2182,
+      "step": 1684
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9616615168685943e-05,
+      "loss": 1.2474,
+      "step": 1685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9561928549563968e-05,
+      "loss": 1.4109,
+      "step": 1686
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.950729971969955e-05,
+      "loss": 1.3778,
+      "step": 1687
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.945272878281011e-05,
+      "loss": 1.2949,
+      "step": 1688
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.939821584250326e-05,
+      "loss": 1.3806,
+      "step": 1689
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9343761002276373e-05,
+      "loss": 1.3208,
+      "step": 1690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.928936436551661e-05,
+      "loss": 1.3314,
+      "step": 1691
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9235026035500586e-05,
+      "loss": 1.3291,
+      "step": 1692
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9180746115394243e-05,
+      "loss": 1.4274,
+      "step": 1693
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9126524708252558e-05,
+      "loss": 1.336,
+      "step": 1694
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9072361917019536e-05,
+      "loss": 1.3537,
+      "step": 1695
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.901825784452777e-05,
+      "loss": 1.3298,
+      "step": 1696
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8964212593498442e-05,
+      "loss": 1.3199,
+      "step": 1697
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8910226266541047e-05,
+      "loss": 1.4091,
+      "step": 1698
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8856298966153212e-05,
+      "loss": 1.3138,
+      "step": 1699
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8802430794720453e-05,
+      "loss": 1.2908,
+      "step": 1700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.874862185451608e-05,
+      "loss": 1.3884,
+      "step": 1701
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8694872247700934e-05,
+      "loss": 1.3958,
+      "step": 1702
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8641182076323148e-05,
+      "loss": 1.3156,
+      "step": 1703
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8587551442318136e-05,
+      "loss": 1.3276,
+      "step": 1704
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8533980447508137e-05,
+      "loss": 1.3666,
+      "step": 1705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.848046919360225e-05,
+      "loss": 1.3763,
+      "step": 1706
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8427017782196127e-05,
+      "loss": 1.3109,
+      "step": 1707
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8373626314771806e-05,
+      "loss": 1.3535,
+      "step": 1708
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8320294892697478e-05,
+      "loss": 1.238,
+      "step": 1709
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8267023617227412e-05,
+      "loss": 1.2672,
+      "step": 1710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.821381258950161e-05,
+      "loss": 1.3894,
+      "step": 1711
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8160661910545718e-05,
+      "loss": 1.3029,
+      "step": 1712
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.810757168127081e-05,
+      "loss": 1.4009,
+      "step": 1713
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.805454200247318e-05,
+      "loss": 1.2278,
+      "step": 1714
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.800157297483417e-05,
+      "loss": 1.2412,
+      "step": 1715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7948664698919988e-05,
+      "loss": 1.3664,
+      "step": 1716
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.789581727518143e-05,
+      "loss": 1.387,
+      "step": 1717
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7843030803953835e-05,
+      "loss": 1.2368,
+      "step": 1718
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7790305385456795e-05,
+      "loss": 1.3669,
+      "step": 1719
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7737641119793976e-05,
+      "loss": 1.3381,
+      "step": 1720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768503810695295e-05,
+      "loss": 1.2809,
+      "step": 1721
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7632496446805014e-05,
+      "loss": 1.3708,
+      "step": 1722
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7580016239104924e-05,
+      "loss": 1.3278,
+      "step": 1723
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7527597583490822e-05,
+      "loss": 1.2897,
+      "step": 1724
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7475240579483958e-05,
+      "loss": 1.4212,
+      "step": 1725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7422945326488553e-05,
+      "loss": 1.307,
+      "step": 1726
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7370711923791567e-05,
+      "loss": 1.2774,
+      "step": 1727
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7318540470562557e-05,
+      "loss": 1.3442,
+      "step": 1728
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7266431065853416e-05,
+      "loss": 1.3501,
+      "step": 1729
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7214383808598284e-05,
+      "loss": 1.2907,
+      "step": 1730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7162398797613282e-05,
+      "loss": 1.2379,
+      "step": 1731
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.711047613159636e-05,
+      "loss": 1.3706,
+      "step": 1732
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.70586159091271e-05,
+      "loss": 1.2815,
+      "step": 1733
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7006818228666543e-05,
+      "loss": 1.3525,
+      "step": 1734
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6955083188556947e-05,
+      "loss": 1.3645,
+      "step": 1735
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6903410887021676e-05,
+      "loss": 1.2967,
+      "step": 1736
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.685180142216498e-05,
+      "loss": 1.3884,
+      "step": 1737
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6800254891971796e-05,
+      "loss": 1.2804,
+      "step": 1738
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6748771394307585e-05,
+      "loss": 1.2636,
+      "step": 1739
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6697351026918134e-05,
+      "loss": 1.3336,
+      "step": 1740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6645993887429345e-05,
+      "loss": 1.4105,
+      "step": 1741
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6594700073347113e-05,
+      "loss": 1.2525,
+      "step": 1742
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6543469682057106e-05,
+      "loss": 1.2959,
+      "step": 1743
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.649230281082451e-05,
+      "loss": 1.3793,
+      "step": 1744
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6441199556794033e-05,
+      "loss": 1.2659,
+      "step": 1745
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6390160016989487e-05,
+      "loss": 1.3731,
+      "step": 1746
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.633918428831377e-05,
+      "loss": 1.2859,
+      "step": 1747
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6288272467548632e-05,
+      "loss": 1.324,
+      "step": 1748
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.623742465135449e-05,
+      "loss": 1.3876,
+      "step": 1749
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.618664093627018e-05,
+      "loss": 1.3918,
+      "step": 1750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6135921418712956e-05,
+      "loss": 1.2915,
+      "step": 1751
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6085266194978067e-05,
+      "loss": 1.1181,
+      "step": 1752
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.603467536123877e-05,
+      "loss": 1.3478,
+      "step": 1753
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5984149013546045e-05,
+      "loss": 1.371,
+      "step": 1754
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.593368724782846e-05,
+      "loss": 1.3495,
+      "step": 1755
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5883290159891906e-05,
+      "loss": 1.293,
+      "step": 1756
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.583295784541958e-05,
+      "loss": 1.311,
+      "step": 1757
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5782690399971617e-05,
+      "loss": 1.3793,
+      "step": 1758
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5732487918985018e-05,
+      "loss": 1.2614,
+      "step": 1759
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.568235049777345e-05,
+      "loss": 1.3416,
+      "step": 1760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.563227823152708e-05,
+      "loss": 1.3793,
+      "step": 1761
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5582271215312294e-05,
+      "loss": 1.1822,
+      "step": 1762
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.553232954407171e-05,
+      "loss": 1.3759,
+      "step": 1763
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5482453312623785e-05,
+      "loss": 1.3012,
+      "step": 1764
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5432642615662774e-05,
+      "loss": 1.3185,
+      "step": 1765
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5382897547758514e-05,
+      "loss": 1.2596,
+      "step": 1766
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5333218203356243e-05,
+      "loss": 1.3883,
+      "step": 1767
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.528360467677635e-05,
+      "loss": 1.2739,
+      "step": 1768
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5234057062214402e-05,
+      "loss": 1.3706,
+      "step": 1769
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.518457545374068e-05,
+      "loss": 1.3543,
+      "step": 1770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5135159945300231e-05,
+      "loss": 1.2963,
+      "step": 1771
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5085810630712582e-05,
+      "loss": 1.3239,
+      "step": 1772
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5036527603671608e-05,
+      "loss": 1.3418,
+      "step": 1773
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.498731095774526e-05,
+      "loss": 1.2849,
+      "step": 1774
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4938160786375572e-05,
+      "loss": 1.3692,
+      "step": 1775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4889077182878269e-05,
+      "loss": 1.3579,
+      "step": 1776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4840060240442738e-05,
+      "loss": 1.3202,
+      "step": 1777
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4791110052131802e-05,
+      "loss": 1.2987,
+      "step": 1778
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4742226710881558e-05,
+      "loss": 1.3851,
+      "step": 1779
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.469341030950112e-05,
+      "loss": 1.3062,
+      "step": 1780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4644660940672627e-05,
+      "loss": 1.3647,
+      "step": 1781
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4595978696950857e-05,
+      "loss": 1.2908,
+      "step": 1782
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4547363670763137e-05,
+      "loss": 1.2671,
+      "step": 1783
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4498815954409278e-05,
+      "loss": 1.3467,
+      "step": 1784
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4450335640061185e-05,
+      "loss": 1.3639,
+      "step": 1785
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4401922819762864e-05,
+      "loss": 1.2835,
+      "step": 1786
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.435357758543015e-05,
+      "loss": 1.4197,
+      "step": 1787
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4305300028850583e-05,
+      "loss": 1.342,
+      "step": 1788
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4257090241683152e-05,
+      "loss": 1.2219,
+      "step": 1789
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4208948315458277e-05,
+      "loss": 1.3557,
+      "step": 1790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4160874341577446e-05,
+      "loss": 1.3357,
+      "step": 1791
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.411286841131318e-05,
+      "loss": 1.3536,
+      "step": 1792
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4064930615808808e-05,
+      "loss": 1.2612,
+      "step": 1793
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4017061046078311e-05,
+      "loss": 1.3258,
+      "step": 1794
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3969259793006079e-05,
+      "loss": 1.2162,
+      "step": 1795
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3921526947346902e-05,
+      "loss": 1.3583,
+      "step": 1796
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3873862599725578e-05,
+      "loss": 1.3851,
+      "step": 1797
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3826266840636938e-05,
+      "loss": 1.2943,
+      "step": 1798
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3778739760445552e-05,
+      "loss": 1.3244,
+      "step": 1799
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.373128144938563e-05,
+      "loss": 1.4119,
+      "step": 1800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.368389199756075e-05,
+      "loss": 1.2866,
+      "step": 1801
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3636571494943862e-05,
+      "loss": 1.188,
+      "step": 1802
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3589320031376901e-05,
+      "loss": 1.3544,
+      "step": 1803
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.354213769657079e-05,
+      "loss": 1.2275,
+      "step": 1804
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3495024580105192e-05,
+      "loss": 1.3052,
+      "step": 1805
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.344798077142836e-05,
+      "loss": 1.4112,
+      "step": 1806
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3401006359856915e-05,
+      "loss": 1.2768,
+      "step": 1807
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3354101434575806e-05,
+      "loss": 1.3054,
+      "step": 1808
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.330726608463797e-05,
+      "loss": 1.3075,
+      "step": 1809
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3260500398964303e-05,
+      "loss": 1.2875,
+      "step": 1810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3213804466343421e-05,
+      "loss": 1.4116,
+      "step": 1811
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3167178375431516e-05,
+      "loss": 1.2997,
+      "step": 1812
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3120622214752132e-05,
+      "loss": 1.324,
+      "step": 1813
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3074136072696148e-05,
+      "loss": 1.3467,
+      "step": 1814
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3027720037521397e-05,
+      "loss": 1.2999,
+      "step": 1815
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2981374197352664e-05,
+      "loss": 1.2971,
+      "step": 1816
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.293509864018146e-05,
+      "loss": 1.3386,
+      "step": 1817
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2888893453865858e-05,
+      "loss": 1.3667,
+      "step": 1818
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2842758726130283e-05,
+      "loss": 1.2661,
+      "step": 1819
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.279669454456548e-05,
+      "loss": 1.3706,
+      "step": 1820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.275070099662815e-05,
+      "loss": 1.3111,
+      "step": 1821
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2704778169640969e-05,
+      "loss": 1.2847,
+      "step": 1822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2658926150792322e-05,
+      "loss": 1.2951,
+      "step": 1823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.261314502713612e-05,
+      "loss": 1.3859,
+      "step": 1824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2567434885591722e-05,
+      "loss": 1.2813,
+      "step": 1825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2521795812943704e-05,
+      "loss": 1.3884,
+      "step": 1826
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2476227895841713e-05,
+      "loss": 1.1901,
+      "step": 1827
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2430731220800301e-05,
+      "loss": 1.2359,
+      "step": 1828
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2385305874198776e-05,
+      "loss": 1.3266,
+      "step": 1829
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2339951942280975e-05,
+      "loss": 1.3564,
+      "step": 1830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2294669511155193e-05,
+      "loss": 1.2987,
+      "step": 1831
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2249458666793967e-05,
+      "loss": 1.3529,
+      "step": 1832
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2204319495033916e-05,
+      "loss": 1.2721,
+      "step": 1833
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2159252081575579e-05,
+      "loss": 1.3376,
+      "step": 1834
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2114256511983274e-05,
+      "loss": 1.4009,
+      "step": 1835
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2069332871684875e-05,
+      "loss": 1.3526,
+      "step": 1836
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.202448124597173e-05,
+      "loss": 1.2743,
+      "step": 1837
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1979701719998453e-05,
+      "loss": 1.2395,
+      "step": 1838
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1934994378782772e-05,
+      "loss": 1.3063,
+      "step": 1839
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1890359307205357e-05,
+      "loss": 1.2904,
+      "step": 1840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1845796590009683e-05,
+      "loss": 1.3826,
+      "step": 1841
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1801306311801825e-05,
+      "loss": 1.3494,
+      "step": 1842
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1756888557050355e-05,
+      "loss": 1.2267,
+      "step": 1843
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1712543410086146e-05,
+      "loss": 1.3867,
+      "step": 1844
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1668270955102218e-05,
+      "loss": 1.3166,
+      "step": 1845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.162407127615357e-05,
+      "loss": 1.2725,
+      "step": 1846
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.157994445715706e-05,
+      "loss": 1.3275,
+      "step": 1847
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1535890581891157e-05,
+      "loss": 1.3025,
+      "step": 1848
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1491909733995898e-05,
+      "loss": 1.2796,
+      "step": 1849
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1448001996972647e-05,
+      "loss": 1.3476,
+      "step": 1850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1404167454183957e-05,
+      "loss": 1.3565,
+      "step": 1851
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1360406188853423e-05,
+      "loss": 1.1418,
+      "step": 1852
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1316718284065537e-05,
+      "loss": 1.3533,
+      "step": 1853
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1273103822765435e-05,
+      "loss": 1.279,
+      "step": 1854
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1229562887758926e-05,
+      "loss": 1.3541,
+      "step": 1855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.118609556171213e-05,
+      "loss": 1.3044,
+      "step": 1856
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1142701927151456e-05,
+      "loss": 1.2454,
+      "step": 1857
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1099382066463392e-05,
+      "loss": 1.3187,
+      "step": 1858
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1056136061894384e-05,
+      "loss": 1.3456,
+      "step": 1859
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1012963995550596e-05,
+      "loss": 1.2629,
+      "step": 1860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0969865949397901e-05,
+      "loss": 1.352,
+      "step": 1861
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.092684200526155e-05,
+      "loss": 1.3036,
+      "step": 1862
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0883892244826172e-05,
+      "loss": 1.3607,
+      "step": 1863
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.084101674963553e-05,
+      "loss": 1.3055,
+      "step": 1864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0798215601092354e-05,
+      "loss": 1.3574,
+      "step": 1865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.075548888045827e-05,
+      "loss": 1.2792,
+      "step": 1866
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0712836668853582e-05,
+      "loss": 1.2647,
+      "step": 1867
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.067025904725713e-05,
+      "loss": 1.3375,
+      "step": 1868
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0627756096506102e-05,
+      "loss": 1.295,
+      "step": 1869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0585327897296005e-05,
+      "loss": 1.3698,
+      "step": 1870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0542974530180327e-05,
+      "loss": 1.3392,
+      "step": 1871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0500696075570538e-05,
+      "loss": 1.2985,
+      "step": 1872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.045849261373587e-05,
+      "loss": 1.2742,
+      "step": 1873
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0416364224803182e-05,
+      "loss": 1.3872,
+      "step": 1874
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0374310988756747e-05,
+      "loss": 1.286,
+      "step": 1875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0332332985438248e-05,
+      "loss": 1.3143,
+      "step": 1876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0290430294546449e-05,
+      "loss": 1.3482,
+      "step": 1877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0248602995637169e-05,
+      "loss": 1.2882,
+      "step": 1878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0206851168123077e-05,
+      "loss": 1.3047,
+      "step": 1879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.016517489127357e-05,
+      "loss": 1.377,
+      "step": 1880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.012357424421455e-05,
+      "loss": 1.208,
+      "step": 1881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0082049305928426e-05,
+      "loss": 1.3767,
+      "step": 1882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0040600155253765e-05,
+      "loss": 1.1777,
+      "step": 1883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.999226870885309e-06,
+      "loss": 1.2675,
+      "step": 1884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.95792953137375e-06,
+      "loss": 1.3617,
+      "step": 1885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.916708215125587e-06,
+      "loss": 1.3224,
+      "step": 1886
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.875563000402948e-06,
+      "loss": 1.3425,
+      "step": 1887
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.83449396532356e-06,
+      "loss": 1.3556,
+      "step": 1888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.793501187860432e-06,
+      "loss": 1.2893,
+      "step": 1889
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.752584745841831e-06,
+      "loss": 1.2735,
+      "step": 1890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.711744716951093e-06,
+      "loss": 1.3223,
+      "step": 1891
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.670981178726485e-06,
+      "loss": 1.3351,
+      "step": 1892
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.630294208560998e-06,
+      "loss": 1.272,
+      "step": 1893
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.589683883702344e-06,
+      "loss": 1.3321,
+      "step": 1894
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.549150281252633e-06,
+      "loss": 1.3649,
+      "step": 1895
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.508693478168346e-06,
+      "loss": 1.2362,
+      "step": 1896
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.468313551260161e-06,
+      "loss": 1.3094,
+      "step": 1897
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.428010577192798e-06,
+      "loss": 1.3683,
+      "step": 1898
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.387784632484826e-06,
+      "loss": 1.2484,
+      "step": 1899
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.347635793508668e-06,
+      "loss": 1.3655,
+      "step": 1900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.307564136490254e-06,
+      "loss": 1.3452,
+      "step": 1901
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.267569737509035e-06,
+      "loss": 1.2762,
+      "step": 1902
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.227652672497761e-06,
+      "loss": 1.2002,
+      "step": 1903
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.187813017242387e-06,
+      "loss": 1.3427,
+      "step": 1904
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.148050847381828e-06,
+      "loss": 1.2767,
+      "step": 1905
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.108366238407966e-06,
+      "loss": 1.3752,
+      "step": 1906
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.068759265665384e-06,
+      "loss": 1.2839,
+      "step": 1907
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.02923000435127e-06,
+      "loss": 1.2594,
+      "step": 1908
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.989778529515298e-06,
+      "loss": 1.2863,
+      "step": 1909
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.950404916059408e-06,
+      "loss": 1.2817,
+      "step": 1910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.911109238737747e-06,
+      "loss": 1.3112,
+      "step": 1911
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.8718915721565e-06,
+      "loss": 1.3796,
+      "step": 1912
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.832751990773714e-06,
+      "loss": 1.349,
+      "step": 1913
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.793690568899216e-06,
+      "loss": 1.2505,
+      "step": 1914
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.754707380694427e-06,
+      "loss": 1.3469,
+      "step": 1915
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.715802500172216e-06,
+      "loss": 1.3114,
+      "step": 1916
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.676976001196801e-06,
+      "loss": 1.2535,
+      "step": 1917
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.63822795748358e-06,
+      "loss": 1.3373,
+      "step": 1918
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.599558442598998e-06,
+      "loss": 1.3469,
+      "step": 1919
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.560967529960407e-06,
+      "loss": 1.2488,
+      "step": 1920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.522455292835934e-06,
+      "loss": 1.3539,
+      "step": 1921
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.484021804344306e-06,
+      "loss": 1.3296,
+      "step": 1922
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.445667137454761e-06,
+      "loss": 1.2816,
+      "step": 1923
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.407391364986889e-06,
+      "loss": 1.4042,
+      "step": 1924
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.369194559610482e-06,
+      "loss": 1.1253,
+      "step": 1925
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.33107679384542e-06,
+      "loss": 1.2865,
+      "step": 1926
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.293038140061515e-06,
+      "loss": 1.3401,
+      "step": 1927
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.25507867047835e-06,
+      "loss": 1.2054,
+      "step": 1928
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.21719845716521e-06,
+      "loss": 1.3691,
+      "step": 1929
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.179397572040892e-06,
+      "loss": 1.284,
+      "step": 1930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.141676086873572e-06,
+      "loss": 1.3723,
+      "step": 1931
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.104034073280686e-06,
+      "loss": 1.175,
+      "step": 1932
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.066471602728803e-06,
+      "loss": 1.3297,
+      "step": 1933
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.028988746533433e-06,
+      "loss": 1.2852,
+      "step": 1934
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.991585575858961e-06,
+      "loss": 1.3048,
+      "step": 1935
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.954262161718478e-06,
+      "loss": 1.3963,
+      "step": 1936
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.917018574973645e-06,
+      "loss": 1.3306,
+      "step": 1937
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.87985488633457e-06,
+      "loss": 1.2152,
+      "step": 1938
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.842771166359681e-06,
+      "loss": 1.3926,
+      "step": 1939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.805767485455528e-06,
+      "loss": 1.286,
+      "step": 1940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.768843913876756e-06,
+      "loss": 1.2645,
+      "step": 1941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.732000521725885e-06,
+      "loss": 1.401,
+      "step": 1942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.695237378953223e-06,
+      "loss": 1.2284,
+      "step": 1943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.658554555356722e-06,
+      "loss": 1.2967,
+      "step": 1944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.62195212058181e-06,
+      "loss": 1.3868,
+      "step": 1945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.585430144121319e-06,
+      "loss": 1.3306,
+      "step": 1946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.5489886953153125e-06,
+      "loss": 1.2269,
+      "step": 1947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.5126278433509835e-06,
+      "loss": 1.3663,
+      "step": 1948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.476347657262456e-06,
+      "loss": 1.2842,
+      "step": 1949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.44014820593078e-06,
+      "loss": 1.1839,
+      "step": 1950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.404029558083653e-06,
+      "loss": 1.3644,
+      "step": 1951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.367991782295391e-06,
+      "loss": 1.2682,
+      "step": 1952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.332034946986771e-06,
+      "loss": 1.2517,
+      "step": 1953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.296159120424906e-06,
+      "loss": 1.4274,
+      "step": 1954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.260364370723044e-06,
+      "loss": 1.2515,
+      "step": 1955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.224650765840613e-06,
+      "loss": 1.3462,
+      "step": 1956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.189018373582873e-06,
+      "loss": 1.2852,
+      "step": 1957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.153467261600949e-06,
+      "loss": 1.2629,
+      "step": 1958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.1179974973916486e-06,
+      "loss": 1.1981,
+      "step": 1959
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.082609148297331e-06,
+      "loss": 1.4042,
+      "step": 1960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.047302281505736e-06,
+      "loss": 1.3164,
+      "step": 1961
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.012076964049996e-06,
+      "loss": 1.2188,
+      "step": 1962
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.976933262808322e-06,
+      "loss": 1.3686,
+      "step": 1963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.941871244504017e-06,
+      "loss": 1.3108,
+      "step": 1964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.906890975705305e-06,
+      "loss": 1.2407,
+      "step": 1965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.871992522825182e-06,
+      "loss": 1.3683,
+      "step": 1966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.837175952121306e-06,
+      "loss": 1.2745,
+      "step": 1967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8024413296959136e-06,
+      "loss": 1.3532,
+      "step": 1968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.7677887214956e-06,
+      "loss": 1.3712,
+      "step": 1969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.733218193311292e-06,
+      "loss": 1.2483,
+      "step": 1970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.698729810778065e-06,
+      "loss": 1.2861,
+      "step": 1971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.664323639375047e-06,
+      "loss": 1.3439,
+      "step": 1972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.629999744425236e-06,
+      "loss": 1.3457,
+      "step": 1973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.595758191095497e-06,
+      "loss": 1.2836,
+      "step": 1974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.561599044396288e-06,
+      "loss": 1.3702,
+      "step": 1975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.527522369181655e-06,
+      "loss": 1.2479,
+      "step": 1976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.493528230149054e-06,
+      "loss": 1.2989,
+      "step": 1977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.459616691839248e-06,
+      "loss": 1.219,
+      "step": 1978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.425787818636131e-06,
+      "loss": 1.2092,
+      "step": 1979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.39204167476673e-06,
+      "loss": 1.3619,
+      "step": 1980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.3583783243009285e-06,
+      "loss": 1.3417,
+      "step": 1981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.324797831151452e-06,
+      "loss": 1.2764,
+      "step": 1982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.291300259073724e-06,
+      "loss": 1.284,
+      "step": 1983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.257885671665726e-06,
+      "loss": 1.3275,
+      "step": 1984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.22455413236786e-06,
+      "loss": 1.2813,
+      "step": 1985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.191305704462897e-06,
+      "loss": 1.296,
+      "step": 1986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.158140451075795e-06,
+      "loss": 1.3942,
+      "step": 1987
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.12505843517357e-06,
+      "loss": 1.3219,
+      "step": 1988
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.092059719565274e-06,
+      "loss": 1.2273,
+      "step": 1989
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.059144366901736e-06,
+      "loss": 1.3085,
+      "step": 1990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.026312439675552e-06,
+      "loss": 1.2775,
+      "step": 1991
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.993564000220913e-06,
+      "loss": 1.3595,
+      "step": 1992
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.960899110713519e-06,
+      "loss": 1.3342,
+      "step": 1993
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.928317833170394e-06,
+      "loss": 1.252,
+      "step": 1994
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.895820229449906e-06,
+      "loss": 1.3718,
+      "step": 1995
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.863406361251472e-06,
+      "loss": 1.332,
+      "step": 1996
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.831076290115573e-06,
+      "loss": 1.2292,
+      "step": 1997
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.798830077423595e-06,
+      "loss": 1.3272,
+      "step": 1998
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.766667784397706e-06,
+      "loss": 1.398,
+      "step": 1999
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.734589472100738e-06,
+      "loss": 1.2814,
+      "step": 2000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.702595201436101e-06,
+      "loss": 1.3265,
+      "step": 2001
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.670685033147599e-06,
+      "loss": 1.3221,
+      "step": 2002
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6388590278194096e-06,
+      "loss": 1.0688,
+      "step": 2003
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.607117245875898e-06,
+      "loss": 1.3293,
+      "step": 2004
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.57545974758153e-06,
+      "loss": 1.3572,
+      "step": 2005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.543886593040737e-06,
+      "loss": 1.2481,
+      "step": 2006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.512397842197847e-06,
+      "loss": 1.3511,
+      "step": 2007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.480993554836905e-06,
+      "loss": 1.2015,
+      "step": 2008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.449673790581611e-06,
+      "loss": 1.26,
+      "step": 2009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4184386088951935e-06,
+      "loss": 1.3564,
+      "step": 2010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.387288069080299e-06,
+      "loss": 1.3645,
+      "step": 2011
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.356222230278857e-06,
+      "loss": 1.2068,
+      "step": 2012
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.32524115147201e-06,
+      "loss": 1.3023,
+      "step": 2013
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.294344891479941e-06,
+      "loss": 1.3671,
+      "step": 2014
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.263533508961827e-06,
+      "loss": 1.2358,
+      "step": 2015
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2328070624156896e-06,
+      "loss": 1.3565,
+      "step": 2016
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.202165610178284e-06,
+      "loss": 1.3334,
+      "step": 2017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1716092104250106e-06,
+      "loss": 1.2453,
+      "step": 2018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.141137921169792e-06,
+      "loss": 1.3931,
+      "step": 2019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.110751800264929e-06,
+      "loss": 1.3174,
+      "step": 2020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.080450905401057e-06,
+      "loss": 1.265,
+      "step": 2021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.050235294106997e-06,
+      "loss": 1.289,
+      "step": 2022
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.020105023749644e-06,
+      "loss": 1.2758,
+      "step": 2023
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.990060151533871e-06,
+      "loss": 1.3086,
+      "step": 2024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.960100734502421e-06,
+      "loss": 1.2369,
+      "step": 2025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.930226829535767e-06,
+      "loss": 1.3108,
+      "step": 2026
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.900438493352055e-06,
+      "loss": 1.2105,
+      "step": 2027
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.87073578250698e-06,
+      "loss": 1.3799,
+      "step": 2028
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8411187533936195e-06,
+      "loss": 1.2627,
+      "step": 2029
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.811587462242462e-06,
+      "loss": 1.3472,
+      "step": 2030
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.782141965121128e-06,
+      "loss": 1.3744,
+      "step": 2031
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.752782317934401e-06,
+      "loss": 1.2174,
+      "step": 2032
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.723508576424062e-06,
+      "loss": 1.3054,
+      "step": 2033
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.694320796168789e-06,
+      "loss": 1.3124,
+      "step": 2034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.66521903258404e-06,
+      "loss": 1.3297,
+      "step": 2035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.636203340922008e-06,
+      "loss": 1.3067,
+      "step": 2036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6072737762714144e-06,
+      "loss": 1.2854,
+      "step": 2037
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.578430393557503e-06,
+      "loss": 1.3292,
+      "step": 2038
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.549673247541875e-06,
+      "loss": 1.2684,
+      "step": 2039
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.521002392822421e-06,
+      "loss": 1.3471,
+      "step": 2040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.492417883833155e-06,
+      "loss": 1.3214,
+      "step": 2041
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.463919774844233e-06,
+      "loss": 1.2856,
+      "step": 2042
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.435508119961701e-06,
+      "loss": 1.3319,
+      "step": 2043
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.407182973127494e-06,
+      "loss": 1.3228,
+      "step": 2044
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.378944388119311e-06,
+      "loss": 1.2438,
+      "step": 2045
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.350792418550509e-06,
+      "loss": 1.1546,
+      "step": 2046
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.322727117869951e-06,
+      "loss": 1.3218,
+      "step": 2047
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.294748539362031e-06,
+      "loss": 1.3479,
+      "step": 2048
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.266856736146424e-06,
+      "loss": 1.3418,
+      "step": 2049
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2390517611780925e-06,
+      "loss": 1.2049,
+      "step": 2050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2113336672471245e-06,
+      "loss": 1.3292,
+      "step": 2051
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.183702506978698e-06,
+      "loss": 1.3272,
+      "step": 2052
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1561583328328716e-06,
+      "loss": 1.2778,
+      "step": 2053
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.128701197104629e-06,
+      "loss": 1.3567,
+      "step": 2054
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.101331151923649e-06,
+      "loss": 1.369,
+      "step": 2055
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.074048249254286e-06,
+      "loss": 1.1815,
+      "step": 2056
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.046852540895446e-06,
+      "loss": 1.4072,
+      "step": 2057
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.019744078480486e-06,
+      "loss": 1.2099,
+      "step": 2058
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9927229134771035e-06,
+      "loss": 1.3085,
+      "step": 2059
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9657890971873e-06,
+      "loss": 1.2159,
+      "step": 2060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9389426807471766e-06,
+      "loss": 1.3688,
+      "step": 2061
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.912183715126944e-06,
+      "loss": 1.2564,
+      "step": 2062
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.885512251130763e-06,
+      "loss": 1.3521,
+      "step": 2063
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.858928339396678e-06,
+      "loss": 1.3463,
+      "step": 2064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.832432030396471e-06,
+      "loss": 1.2281,
+      "step": 2065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8060233744356633e-06,
+      "loss": 1.2646,
+      "step": 2066
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7797024216533138e-06,
+      "loss": 1.3481,
+      "step": 2067
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7534692220219548e-06,
+      "loss": 1.2828,
+      "step": 2068
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7273238253475785e-06,
+      "loss": 1.3615,
+      "step": 2069
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.701266281269411e-06,
+      "loss": 1.276,
+      "step": 2070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.675296639259912e-06,
+      "loss": 1.267,
+      "step": 2071
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6494149486246522e-06,
+      "loss": 1.2861,
+      "step": 2072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6236212585022187e-06,
+      "loss": 1.3531,
+      "step": 2073
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.597915617864095e-06,
+      "loss": 1.205,
+      "step": 2074
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5722980755146517e-06,
+      "loss": 1.3638,
+      "step": 2075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.546768680090934e-06,
+      "loss": 1.2602,
+      "step": 2076
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5213274800626692e-06,
+      "loss": 1.2289,
+      "step": 2077
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4959745237321428e-06,
+      "loss": 1.3457,
+      "step": 2078
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.470709859234084e-06,
+      "loss": 1.3395,
+      "step": 2079
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4455335345355834e-06,
+      "loss": 1.2655,
+      "step": 2080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.420445597436056e-06,
+      "loss": 1.4028,
+      "step": 2081
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3954460955670565e-06,
+      "loss": 1.3259,
+      "step": 2082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3705350763922562e-06,
+      "loss": 1.2294,
+      "step": 2083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3457125872073393e-06,
+      "loss": 1.2687,
+      "step": 2084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3209786751399187e-06,
+      "loss": 1.3226,
+      "step": 2085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.296333387149392e-06,
+      "loss": 1.3586,
+      "step": 2086
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.271776770026963e-06,
+      "loss": 1.229,
+      "step": 2087
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2473088703954212e-06,
+      "loss": 1.3479,
+      "step": 2088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2229297347091514e-06,
+      "loss": 1.1902,
+      "step": 2089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1986394092540174e-06,
+      "loss": 1.2656,
+      "step": 2090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1744379401472677e-06,
+      "loss": 1.3801,
+      "step": 2091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1503253733374094e-06,
+      "loss": 1.2267,
+      "step": 2092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.126301754604233e-06,
+      "loss": 1.2994,
+      "step": 2093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.102367129558592e-06,
+      "loss": 1.3486,
+      "step": 2094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.078521543642399e-06,
+      "loss": 1.2343,
+      "step": 2095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.054765042128521e-06,
+      "loss": 1.1865,
+      "step": 2096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.031097670120703e-06,
+      "loss": 1.3404,
+      "step": 2097
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0075194725534326e-06,
+      "loss": 1.192,
+      "step": 2098
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9840304941919415e-06,
+      "loss": 1.3457,
+      "step": 2099
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.960630779632023e-06,
+      "loss": 1.2718,
+      "step": 2100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9373203733000232e-06,
+      "loss": 1.3281,
+      "step": 2101
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.914099319452729e-06,
+      "loss": 1.2886,
+      "step": 2102
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.890967662177285e-06,
+      "loss": 1.2606,
+      "step": 2103
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8679254453910785e-06,
+      "loss": 1.2822,
+      "step": 2104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.844972712841737e-06,
+      "loss": 1.309,
+      "step": 2105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8221095081069517e-06,
+      "loss": 1.3214,
+      "step": 2106
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7993358745944608e-06,
+      "loss": 1.2596,
+      "step": 2107
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7766518555419397e-06,
+      "loss": 1.3415,
+      "step": 2108
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.754057494016904e-06,
+      "loss": 1.3332,
+      "step": 2109
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7315528329166784e-06,
+      "loss": 1.2838,
+      "step": 2110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7091379149682685e-06,
+      "loss": 1.3171,
+      "step": 2111
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6868127827282872e-06,
+      "loss": 1.3799,
+      "step": 2112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.664577478582897e-06,
+      "loss": 1.1905,
+      "step": 2113
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6424320447477112e-06,
+      "loss": 1.3175,
+      "step": 2114
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.620376523267698e-06,
+      "loss": 1.2659,
+      "step": 2115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5984109560171388e-06,
+      "loss": 1.3581,
+      "step": 2116
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.57653538469953e-06,
+      "loss": 1.3138,
+      "step": 2117
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5547498508474934e-06,
+      "loss": 1.3616,
+      "step": 2118
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.533054395822704e-06,
+      "loss": 1.2268,
+      "step": 2119
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5114490608158404e-06,
+      "loss": 1.337,
+      "step": 2120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4899338868464404e-06,
+      "loss": 1.1604,
+      "step": 2121
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4685089147628905e-06,
+      "loss": 1.282,
+      "step": 2122
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4471741852423237e-06,
+      "loss": 1.3395,
+      "step": 2123
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4259297387905223e-06,
+      "loss": 1.3037,
+      "step": 2124
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.404775615741872e-06,
+      "loss": 1.3337,
+      "step": 2125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3837118562592797e-06,
+      "loss": 1.2875,
+      "step": 2126
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.362738500334055e-06,
+      "loss": 1.2648,
+      "step": 2127
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3418555877859016e-06,
+      "loss": 1.3462,
+      "step": 2128
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.321063158262793e-06,
+      "loss": 1.3597,
+      "step": 2129
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.300361251240929e-06,
+      "loss": 1.3021,
+      "step": 2130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2797499060246253e-06,
+      "loss": 1.3437,
+      "step": 2131
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.259229161746279e-06,
+      "loss": 1.1923,
+      "step": 2132
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2387990573662477e-06,
+      "loss": 1.3417,
+      "step": 2133
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.21845963167282e-06,
+      "loss": 1.2685,
+      "step": 2134
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1982109232821178e-06,
+      "loss": 1.3342,
+      "step": 2135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1780529706380337e-06,
+      "loss": 1.3323,
+      "step": 2136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.157985812012148e-06,
+      "loss": 1.247,
+      "step": 2137
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1380094855036616e-06,
+      "loss": 1.3437,
+      "step": 2138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.118124029039309e-06,
+      "loss": 1.3393,
+      "step": 2139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.09832948037334e-06,
+      "loss": 1.2704,
+      "step": 2140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0786258770873647e-06,
+      "loss": 1.3347,
+      "step": 2141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0590132565903476e-06,
+      "loss": 1.2274,
+      "step": 2142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0394916561185083e-06,
+      "loss": 1.3267,
+      "step": 2143
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0200611127352664e-06,
+      "loss": 1.2867,
+      "step": 2144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.00072166333114e-06,
+      "loss": 1.2713,
+      "step": 2145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9814733446237354e-06,
+      "loss": 1.1179,
+      "step": 2146
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.962316193157593e-06,
+      "loss": 1.4043,
+      "step": 2147
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9432502453042012e-06,
+      "loss": 1.3082,
+      "step": 2148
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.924275537261877e-06,
+      "loss": 1.3403,
+      "step": 2149
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9053921050557033e-06,
+      "loss": 1.3151,
+      "step": 2150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8865999845374793e-06,
+      "loss": 1.2063,
+      "step": 2151
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.867899211385643e-06,
+      "loss": 1.3261,
+      "step": 2152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.849289821105199e-06,
+      "loss": 1.3423,
+      "step": 2153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8307718490276348e-06,
+      "loss": 1.2679,
+      "step": 2154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.812345330310916e-06,
+      "loss": 1.3521,
+      "step": 2155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7940102999393194e-06,
+      "loss": 1.2743,
+      "step": 2156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7757667927234767e-06,
+      "loss": 1.2997,
+      "step": 2157
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7576148433002203e-06,
+      "loss": 1.3119,
+      "step": 2158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7395544861325718e-06,
+      "loss": 1.2725,
+      "step": 2159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.721585755509636e-06,
+      "loss": 1.3356,
+      "step": 2160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.70370868554659e-06,
+      "loss": 1.2053,
+      "step": 2161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6859233101845506e-06,
+      "loss": 1.3384,
+      "step": 2162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6682296631905626e-06,
+      "loss": 1.257,
+      "step": 2163
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6506277781575152e-06,
+      "loss": 1.2637,
+      "step": 2164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6331176885040878e-06,
+      "loss": 1.3413,
+      "step": 2165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6156994274746484e-06,
+      "loss": 1.2752,
+      "step": 2166
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5983730281392662e-06,
+      "loss": 1.254,
+      "step": 2167
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5811385233935549e-06,
+      "loss": 1.3544,
+      "step": 2168
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5639959459586962e-06,
+      "loss": 1.2805,
+      "step": 2169
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5469453283813273e-06,
+      "loss": 1.2794,
+      "step": 2170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5299867030334814e-06,
+      "loss": 1.3644,
+      "step": 2171
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5131201021125474e-06,
+      "loss": 1.2243,
+      "step": 2172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4963455576412045e-06,
+      "loss": 1.2986,
+      "step": 2173
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4796631014673323e-06,
+      "loss": 1.3716,
+      "step": 2174
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.463072765264001e-06,
+      "loss": 1.1758,
+      "step": 2175
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4465745805293585e-06,
+      "loss": 1.3377,
+      "step": 2176
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4301685785866214e-06,
+      "loss": 1.244,
+      "step": 2177
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4138547905839517e-06,
+      "loss": 1.3066,
+      "step": 2178
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3976332474944843e-06,
+      "loss": 1.3269,
+      "step": 2179
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3815039801161721e-06,
+      "loss": 1.2806,
+      "step": 2180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3654670190718034e-06,
+      "loss": 1.3286,
+      "step": 2181
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3495223948089108e-06,
+      "loss": 1.3626,
+      "step": 2182
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.333670137599713e-06,
+      "loss": 1.2519,
+      "step": 2183
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3179102775410512e-06,
+      "loss": 1.2712,
+      "step": 2184
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3022428445543799e-06,
+      "loss": 1.2825,
+      "step": 2185
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.286667868385627e-06,
+      "loss": 1.3311,
+      "step": 2186
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2711853786052109e-06,
+      "loss": 1.2599,
+      "step": 2187
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2557954046079456e-06,
+      "loss": 1.3274,
+      "step": 2188
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2404979756130142e-06,
+      "loss": 1.3671,
+      "step": 2189
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2252931206638728e-06,
+      "loss": 1.1973,
+      "step": 2190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.210180868628219e-06,
+      "loss": 1.3332,
+      "step": 2191
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1951612481979567e-06,
+      "loss": 1.3613,
+      "step": 2192
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1802342878891037e-06,
+      "loss": 1.1825,
+      "step": 2193
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1654000160417566e-06,
+      "loss": 1.3301,
+      "step": 2194
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1506584608200367e-06,
+      "loss": 1.3125,
+      "step": 2195
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1360096502120387e-06,
+      "loss": 1.2519,
+      "step": 2196
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1214536120297714e-06,
+      "loss": 1.2474,
+      "step": 2197
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1069903739091004e-06,
+      "loss": 1.2868,
+      "step": 2198
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0926199633097157e-06,
+      "loss": 1.2291,
+      "step": 2199
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.078342407515065e-06,
+      "loss": 1.3361,
+      "step": 2200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.064157733632276e-06,
+      "loss": 1.2834,
+      "step": 2201
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0500659685921731e-06,
+      "loss": 1.2436,
+      "step": 2202
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0360671391491606e-06,
+      "loss": 1.3325,
+      "step": 2203
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0221612718812e-06,
+      "loss": 1.2558,
+      "step": 2204
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.008348393189762e-06,
+      "loss": 1.3224,
+      "step": 2205
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.946285292997682e-07,
+      "loss": 1.384,
+      "step": 2206
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.810017062595322e-07,
+      "loss": 1.3118,
+      "step": 2207
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.67467949940748e-07,
+      "loss": 1.2321,
+      "step": 2208
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.540272860383837e-07,
+      "loss": 1.3793,
+      "step": 2209
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.406797400706879e-07,
+      "loss": 1.3359,
+      "step": 2210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.274253373791064e-07,
+      "loss": 1.2743,
+      "step": 2211
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.142641031282429e-07,
+      "loss": 1.3053,
+      "step": 2212
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.011960623058202e-07,
+      "loss": 1.361,
+      "step": 2213
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.882212397226142e-07,
+      "loss": 1.2253,
+      "step": 2214
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.753396600124253e-07,
+      "loss": 1.281,
+      "step": 2215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.62551347632029e-07,
+      "loss": 1.3762,
+      "step": 2216
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.498563268611204e-07,
+      "loss": 1.2453,
+      "step": 2217
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.372546218022747e-07,
+      "loss": 1.3624,
+      "step": 2218
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.247462563808817e-07,
+      "loss": 1.1668,
+      "step": 2219
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.1233125434515e-07,
+      "loss": 1.2461,
+      "step": 2220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.000096392660029e-07,
+      "loss": 1.3475,
+      "step": 2221
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.877814345370715e-07,
+      "loss": 1.2061,
+      "step": 2222
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.756466633746407e-07,
+      "loss": 1.3765,
+      "step": 2223
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.636053488176031e-07,
+      "loss": 1.2885,
+      "step": 2224
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.516575137274162e-07,
+      "loss": 1.3044,
+      "step": 2225
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.398031807880457e-07,
+      "loss": 1.2138,
+      "step": 2226
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.280423725059604e-07,
+      "loss": 1.34,
+      "step": 2227
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.163751112100436e-07,
+      "loss": 1.2222,
+      "step": 2228
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.048014190515872e-07,
+      "loss": 1.2603,
+      "step": 2229
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.933213180042197e-07,
+      "loss": 1.3952,
+      "step": 2230
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.819348298638839e-07,
+      "loss": 1.3445,
+      "step": 2231
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.706419762487925e-07,
+      "loss": 1.2396,
+      "step": 2232
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.594427785993951e-07,
+      "loss": 1.3732,
+      "step": 2233
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.483372581783054e-07,
+      "loss": 1.2537,
+      "step": 2234
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.373254360703018e-07,
+      "loss": 1.2646,
+      "step": 2235
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.264073331822551e-07,
+      "loss": 1.3802,
+      "step": 2236
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.15582970243117e-07,
+      "loss": 1.2252,
+      "step": 2237
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.048523678038487e-07,
+      "loss": 1.2971,
+      "step": 2238
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9421554623742e-07,
+      "loss": 1.35,
+      "step": 2239
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.83672525738721e-07,
+      "loss": 1.3634,
+      "step": 2240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.732233263245845e-07,
+      "loss": 1.1997,
+      "step": 2241
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.628679678336967e-07,
+      "loss": 1.3134,
+      "step": 2242
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.526064699265753e-07,
+      "loss": 1.2704,
+      "step": 2243
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.424388520855584e-07,
+      "loss": 1.1451,
+      "step": 2244
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.323651336147151e-07,
+      "loss": 1.363,
+      "step": 2245
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.223853336398632e-07,
+      "loss": 1.2429,
+      "step": 2246
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.124994711084963e-07,
+      "loss": 1.2652,
+      "step": 2247
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.027075647897561e-07,
+      "loss": 1.3904,
+      "step": 2248
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.930096332744105e-07,
+      "loss": 1.2369,
+      "step": 2249
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.834056949748089e-07,
+      "loss": 1.2865,
+      "step": 2250
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.738957681248379e-07,
+      "loss": 1.2872,
+      "step": 2251
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.644798707798936e-07,
+      "loss": 1.2238,
+      "step": 2252
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5515802081687045e-07,
+      "loss": 1.2942,
+      "step": 2253
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4593023593407223e-07,
+      "loss": 1.3422,
+      "step": 2254
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.367965336512403e-07,
+      "loss": 1.3375,
+      "step": 2255
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.277569313094809e-07,
+      "loss": 1.2405,
+      "step": 2256
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1881144607124334e-07,
+      "loss": 1.3259,
+      "step": 2257
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0996009492029195e-07,
+      "loss": 1.3273,
+      "step": 2258
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.012028946616675e-07,
+      "loss": 1.2641,
+      "step": 2259
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9253986192164804e-07,
+      "loss": 1.3641,
+      "step": 2260
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.839710131477492e-07,
+      "loss": 1.2635,
+      "step": 2261
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.754963646086407e-07,
+      "loss": 1.2339,
+      "step": 2262
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6711593239417973e-07,
+      "loss": 1.3624,
+      "step": 2263
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5882973241530557e-07,
+      "loss": 1.306,
+      "step": 2264
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5063778040408947e-07,
+      "loss": 1.2462,
+      "step": 2265
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4254009191363457e-07,
+      "loss": 1.2907,
+      "step": 2266
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3453668231809286e-07,
+      "loss": 1.2517,
+      "step": 2267
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.266275668126151e-07,
+      "loss": 1.3262,
+      "step": 2268
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.188127604133284e-07,
+      "loss": 1.3331,
+      "step": 2269
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.110922779573033e-07,
+      "loss": 1.2694,
+      "step": 2270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.034661341025258e-07,
+      "loss": 1.2876,
+      "step": 2271
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9593434332788604e-07,
+      "loss": 1.2021,
+      "step": 2272
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8849691993311777e-07,
+      "loss": 1.2735,
+      "step": 2273
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8115387803880897e-07,
+      "loss": 1.3259,
+      "step": 2274
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7390523158633554e-07,
+      "loss": 1.2757,
+      "step": 2275
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.667509943378721e-07,
+      "loss": 1.2719,
+      "step": 2276
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5969117987634797e-07,
+      "loss": 1.3161,
+      "step": 2277
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.527258016054135e-07,
+      "loss": 1.2993,
+      "step": 2278
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.458548727494292e-07,
+      "loss": 1.2727,
+      "step": 2279
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.390784063534435e-07,
+      "loss": 1.2202,
+      "step": 2280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.323964152831426e-07,
+      "loss": 1.3758,
+      "step": 2281
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2580891222485634e-07,
+      "loss": 1.3452,
+      "step": 2282
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1931590968551908e-07,
+      "loss": 1.2541,
+      "step": 2283
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1291741999264202e-07,
+      "loss": 1.3255,
+      "step": 2284
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0661345529430775e-07,
+      "loss": 1.2528,
+      "step": 2285
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0040402755912014e-07,
+      "loss": 1.32,
+      "step": 2286
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.942891485762044e-07,
+      "loss": 1.3723,
+      "step": 2287
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8826882995517935e-07,
+      "loss": 1.2427,
+      "step": 2288
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8234308312612968e-07,
+      "loss": 1.3654,
+      "step": 2289
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7651191933958367e-07,
+      "loss": 1.3004,
+      "step": 2290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7077534966650766e-07,
+      "loss": 1.2486,
+      "step": 2291
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.651333849982617e-07,
+      "loss": 1.3273,
+      "step": 2292
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5958603604658838e-07,
+      "loss": 1.3633,
+      "step": 2293
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5413331334360182e-07,
+      "loss": 1.2743,
+      "step": 2294
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4877522724175973e-07,
+      "loss": 1.2577,
+      "step": 2295
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.43511787913847e-07,
+      "loss": 1.3341,
+      "step": 2296
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.383430053529422e-07,
+      "loss": 1.0614,
+      "step": 2297
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.332688893724232e-07,
+      "loss": 1.3138,
+      "step": 2298
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2828944960592836e-07,
+      "loss": 1.3521,
+      "step": 2299
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2340469550733424e-07,
+      "loss": 1.2426,
+      "step": 2300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1861463635077785e-07,
+      "loss": 1.3615,
+      "step": 2301
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1391928123057893e-07,
+      "loss": 1.2257,
+      "step": 2302
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0931863906127327e-07,
+      "loss": 1.2866,
+      "step": 2303
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0481271857756824e-07,
+      "loss": 1.3202,
+      "step": 2304
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0040152833433736e-07,
+      "loss": 1.3121,
+      "step": 2305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.60850767065924e-08,
+      "loss": 1.237,
+      "step": 2306
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.186337188949457e-08,
+      "loss": 1.2933,
+      "step": 2307
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.7736421898299e-08,
+      "loss": 1.3129,
+      "step": 2308
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.37042345683714e-08,
+      "loss": 1.2572,
+      "step": 2309
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.976681755516024e-08,
+      "loss": 1.3288,
+      "step": 2310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.59241783341913e-08,
+      "loss": 1.3319,
+      "step": 2311
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.21763242010287e-08,
+      "loss": 1.2371,
+      "step": 2312
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.852326227130834e-08,
+      "loss": 1.316,
+      "step": 2313
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.496499948066004e-08,
+      "loss": 1.3049,
+      "step": 2314
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.150154258476315e-08,
+      "loss": 1.2927,
+      "step": 2315
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.813289815926881e-08,
+      "loss": 1.293,
+      "step": 2316
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.485907259984435e-08,
+      "loss": 1.2251,
+      "step": 2317
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.168007212212334e-08,
+      "loss": 1.3371,
+      "step": 2318
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.859590276170556e-08,
+      "loss": 1.1488,
+      "step": 2319
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.560657037414595e-08,
+      "loss": 1.3203,
+      "step": 2320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2712080634949024e-08,
+      "loss": 1.224,
+      "step": 2321
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.991243903954667e-08,
+      "loss": 1.3357,
+      "step": 2322
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.720765090329814e-08,
+      "loss": 1.2866,
+      "step": 2323
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4597721361467886e-08,
+      "loss": 1.3046,
+      "step": 2324
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.20826553692366e-08,
+      "loss": 1.3798,
+      "step": 2325
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9662457701662428e-08,
+      "loss": 1.2746,
+      "step": 2326
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7337132953697554e-08,
+      "loss": 1.2668,
+      "step": 2327
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.510668554017159e-08,
+      "loss": 1.2889,
+      "step": 2328
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2971119695774925e-08,
+      "loss": 1.2507,
+      "step": 2329
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0930439475058684e-08,
+      "loss": 1.3293,
+      "step": 2330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8984648752429225e-08,
+      "loss": 1.265,
+      "step": 2331
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7133751222137007e-08,
+      "loss": 1.3651,
+      "step": 2332
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5377750398265502e-08,
+      "loss": 1.2358,
+      "step": 2333
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.371664961472563e-08,
+      "loss": 1.31,
+      "step": 2334
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.215045202527243e-08,
+      "loss": 1.3423,
+      "step": 2335
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0679160603449534e-08,
+      "loss": 1.2515,
+      "step": 2336
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.302778142639135e-09,
+      "loss": 1.3409,
+      "step": 2337
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.021307256012022e-09,
+      "loss": 1.2954,
+      "step": 2338
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.834750376549792e-09,
+      "loss": 1.2841,
+      "step": 2339
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.743109757039289e-09,
+      "loss": 1.1587,
+      "step": 2340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.746387470044855e-09,
+      "loss": 1.2906,
+      "step": 2341
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.844585407936086e-09,
+      "loss": 1.3037,
+      "step": 2342
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0377052828489683e-09,
+      "loss": 1.3362,
+      "step": 2343
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.325748626724744e-09,
+      "loss": 1.2391,
+      "step": 2344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7087167912710478e-09,
+      "loss": 1.3402,
+      "step": 2345
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1866109479674592e-09,
+      "loss": 1.2611,
+      "step": 2346
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.594320880821571e-10,
+      "loss": 1.304,
+      "step": 2347
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.271810226552653e-10,
+      "loss": 1.341,
+      "step": 2348
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8985838248219978e-10,
+      "loss": 1.3561,
+      "step": 2349
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.746461814697511e-11,
+      "loss": 1.2319,
+      "step": 2350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.3262,
+      "step": 2351
+    },
+    {
+      "epoch": 1.0,
+      "step": 2351,
+      "total_flos": 0.0,
+      "train_loss": 0.36083592622141186,
+      "train_runtime": 12483.0014,
+      "train_samples_per_second": 386.551,
+      "train_steps_per_second": 0.188
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2351,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "total_flos": 0.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}