diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,30021 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.4179553623672992,
+  "eval_steps": 20000,
+  "global_step": 5000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 5.4443,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 6.3599,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019995998399359745,
+      "loss": 5.3533,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019991996798719488,
+      "loss": 5.6319,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019987995198079232,
+      "loss": 5.0204,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019983993597438976,
+      "loss": 4.7848,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019979991996798722,
+      "loss": 4.876,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019975990396158463,
+      "loss": 4.9136,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001997198879551821,
+      "loss": 5.2008,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019967987194877953,
+      "loss": 4.0251,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019963985594237694,
+      "loss": 4.0763,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001995998399359744,
+      "loss": 4.3189,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019955982392957185,
+      "loss": 3.8181,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019951980792316926,
+      "loss": 3.7737,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019947979191676672,
+      "loss": 3.731,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019943977591036416,
+      "loss": 4.0602,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001993997599039616,
+      "loss": 3.4416,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019935974389755903,
+      "loss": 3.6661,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019931972789115647,
+      "loss": 3.8713,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001992797118847539,
+      "loss": 3.847,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019923969587835134,
+      "loss": 3.7603,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019919967987194878,
+      "loss": 3.9384,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019915966386554624,
+      "loss": 3.1717,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019911964785914368,
+      "loss": 3.6006,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001990796318527411,
+      "loss": 2.9617,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019903961584633856,
+      "loss": 3.5738,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000198999599839936,
+      "loss": 3.3774,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001989595838335334,
+      "loss": 3.1618,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019891956782713087,
+      "loss": 3.4367,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001988795518207283,
+      "loss": 2.9632,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019883953581432574,
+      "loss": 2.9809,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019879951980792318,
+      "loss": 3.4474,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019875950380152061,
+      "loss": 2.7804,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019871948779511805,
+      "loss": 3.4894,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001986794717887155,
+      "loss": 2.8329,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019863945578231293,
+      "loss": 2.5018,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001985994397759104,
+      "loss": 3.129,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001985594237695078,
+      "loss": 3.1626,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019851940776310524,
+      "loss": 3.3788,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001984793917567027,
+      "loss": 3.5687,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019843937575030014,
+      "loss": 2.6777,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019839935974389755,
+      "loss": 3.4946,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000198359343737495,
+      "loss": 2.3003,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019831932773109245,
+      "loss": 2.6934,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001982793117246899,
+      "loss": 3.7464,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019823929571828732,
+      "loss": 2.5432,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019819927971188476,
+      "loss": 2.3768,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001981592637054822,
+      "loss": 3.6561,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019811924769907964,
+      "loss": 3.054,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019807923169267707,
+      "loss": 2.5337,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019803921568627454,
+      "loss": 2.8043,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019799919967987195,
+      "loss": 2.4842,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019795918367346938,
+      "loss": 3.3978,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019791916766706685,
+      "loss": 2.617,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019787915166066429,
+      "loss": 3.4711,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001978391356542617,
+      "loss": 3.4145,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019779911964785916,
+      "loss": 2.8914,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001977591036414566,
+      "loss": 2.9451,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019771908763505403,
+      "loss": 3.0292,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019767907162865147,
+      "loss": 3.1296,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001976390556222489,
+      "loss": 2.2735,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019759903961584635,
+      "loss": 2.1939,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019755902360944378,
+      "loss": 2.8812,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019751900760304122,
+      "loss": 2.6091,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019747899159663868,
+      "loss": 2.898,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001974389755902361,
+      "loss": 2.6885,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019739895958383353,
+      "loss": 2.9631,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000197358943577431,
+      "loss": 2.7964,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001973189275710284,
+      "loss": 2.5454,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019727891156462587,
+      "loss": 3.0999,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001972388955582233,
+      "loss": 2.2887,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019719887955182074,
+      "loss": 3.2186,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019715886354541818,
+      "loss": 2.2417,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019711884753901562,
+      "loss": 3.2295,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019707883153261306,
+      "loss": 2.595,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001970388155262105,
+      "loss": 3.5308,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019699879951980793,
+      "loss": 2.7987,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019695878351340537,
+      "loss": 3.2869,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019691876750700283,
+      "loss": 2.9246,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019687875150060024,
+      "loss": 3.1981,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019683873549419768,
+      "loss": 2.8618,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019679871948779514,
+      "loss": 2.3601,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019675870348139255,
+      "loss": 2.9208,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019671868747499002,
+      "loss": 2.8393,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019667867146858745,
+      "loss": 2.7993,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019663865546218486,
+      "loss": 2.4626,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019659863945578233,
+      "loss": 2.4063,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019655862344937977,
+      "loss": 2.6342,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001965186074429772,
+      "loss": 3.3108,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019647859143657464,
+      "loss": 2.6329,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019643857543017208,
+      "loss": 2.1152,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019639855942376951,
+      "loss": 2.7064,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019635854341736695,
+      "loss": 2.6887,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001963185274109644,
+      "loss": 2.5271,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019627851140456183,
+      "loss": 2.6707,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001962384953981593,
+      "loss": 3.0632,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001961984793917567,
+      "loss": 2.5833,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019615846338535416,
+      "loss": 2.3264,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001961184473789516,
+      "loss": 2.2065,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000196078431372549,
+      "loss": 2.3184,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019603841536614648,
+      "loss": 2.7698,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001959983993597439,
+      "loss": 2.7447,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019595838335334132,
+      "loss": 3.5849,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001959183673469388,
+      "loss": 2.5773,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019587835134053622,
+      "loss": 3.1046,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019583833533413366,
+      "loss": 2.4807,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001957983193277311,
+      "loss": 2.5277,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019575830332132854,
+      "loss": 2.2736,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019571828731492597,
+      "loss": 2.5451,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001956782713085234,
+      "loss": 2.2912,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019563825530212085,
+      "loss": 2.6037,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001955982392957183,
+      "loss": 2.216,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019555822328931575,
+      "loss": 2.9373,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019551820728291316,
+      "loss": 2.6642,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019547819127651062,
+      "loss": 2.2525,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019543817527010806,
+      "loss": 2.458,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019539815926370547,
+      "loss": 2.5599,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019535814325730293,
+      "loss": 3.3641,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019531812725090037,
+      "loss": 2.2847,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001952781112444978,
+      "loss": 2.735,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019523809523809525,
+      "loss": 1.8775,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019519807923169268,
+      "loss": 2.4611,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019515806322529012,
+      "loss": 2.3069,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019511804721888756,
+      "loss": 2.4857,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000195078031212485,
+      "loss": 2.2793,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019503801520608246,
+      "loss": 3.0231,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019499799919967987,
+      "loss": 2.3143,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001949579831932773,
+      "loss": 2.761,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019491796718687477,
+      "loss": 2.9316,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001948779511804722,
+      "loss": 2.7515,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019483793517406962,
+      "loss": 2.9132,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019479791916766708,
+      "loss": 3.1863,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019475790316126452,
+      "loss": 2.341,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019471788715486196,
+      "loss": 2.1114,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001946778711484594,
+      "loss": 2.986,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019463785514205683,
+      "loss": 2.5318,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001945978391356543,
+      "loss": 2.5453,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001945578231292517,
+      "loss": 2.1032,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019451780712284914,
+      "loss": 2.9476,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001944777911164466,
+      "loss": 3.2228,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019443777511004402,
+      "loss": 2.8824,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019439775910364145,
+      "loss": 2.2357,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019435774309723892,
+      "loss": 2.7185,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019431772709083635,
+      "loss": 2.345,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001942777110844338,
+      "loss": 2.3235,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019423769507803123,
+      "loss": 2.1947,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019419767907162867,
+      "loss": 3.0036,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001941576630652261,
+      "loss": 2.2015,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019411764705882354,
+      "loss": 2.7164,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019407763105242098,
+      "loss": 2.0072,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019403761504601841,
+      "loss": 3.0428,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019399759903961585,
+      "loss": 2.6802,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001939575830332133,
+      "loss": 2.8336,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019391756702681075,
+      "loss": 2.5337,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019387755102040816,
+      "loss": 2.0215,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001938375350140056,
+      "loss": 2.8099,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019379751900760306,
+      "loss": 2.7408,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019375750300120047,
+      "loss": 2.7403,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019371748699479794,
+      "loss": 2.7836,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019367747098839538,
+      "loss": 2.3997,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001936374549819928,
+      "loss": 2.2868,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019359743897559025,
+      "loss": 2.7624,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001935574229691877,
+      "loss": 2.3814,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019351740696278512,
+      "loss": 2.2247,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019347739095638256,
+      "loss": 1.8303,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019343737494998,
+      "loss": 2.8116,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019339735894357744,
+      "loss": 3.391,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001933573429371749,
+      "loss": 3.1073,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001933173269307723,
+      "loss": 2.3658,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019327731092436975,
+      "loss": 3.2388,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001932372949179672,
+      "loss": 2.2719,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019319727891156462,
+      "loss": 3.1843,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019315726290516208,
+      "loss": 2.1,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019311724689875952,
+      "loss": 2.2228,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019307723089235693,
+      "loss": 2.7295,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001930372148859544,
+      "loss": 2.3334,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019299719887955183,
+      "loss": 2.3045,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019295718287314927,
+      "loss": 2.8423,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001929171668667467,
+      "loss": 2.3702,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019287715086034414,
+      "loss": 2.3234,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019283713485394158,
+      "loss": 2.1609,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019279711884753902,
+      "loss": 3.1333,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019275710284113646,
+      "loss": 2.4103,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001927170868347339,
+      "loss": 2.8114,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019267707082833136,
+      "loss": 2.2349,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019263705482192877,
+      "loss": 2.5132,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019259703881552623,
+      "loss": 2.9105,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019255702280912367,
+      "loss": 3.0705,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019251700680272108,
+      "loss": 2.6728,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019247699079631854,
+      "loss": 2.0016,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019243697478991598,
+      "loss": 2.6029,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001923969587835134,
+      "loss": 2.9461,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019235694277711085,
+      "loss": 2.4091,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001923169267707083,
+      "loss": 2.5372,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019227691076430573,
+      "loss": 2.8717,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019223689475790317,
+      "loss": 2.3632,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001921968787515006,
+      "loss": 2.4953,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019215686274509807,
+      "loss": 3.1346,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019215686274509807,
+      "loss": 1.8664,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019211684673869548,
+      "loss": 2.5864,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019207683073229291,
+      "loss": 2.4293,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019203681472589038,
+      "loss": 2.4069,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019199679871948782,
+      "loss": 2.5944,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019195678271308523,
+      "loss": 2.5551,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001919167667066827,
+      "loss": 2.4934,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019187675070028013,
+      "loss": 1.8907,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019183673469387756,
+      "loss": 2.6536,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000191796718687475,
+      "loss": 2.7331,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019175670268107244,
+      "loss": 2.7051,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019171668667466988,
+      "loss": 2.2565,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001916766706682673,
+      "loss": 2.3218,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019163665466186475,
+      "loss": 2.7686,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019159663865546221,
+      "loss": 2.6825,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019155662264905962,
+      "loss": 2.7313,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019151660664265706,
+      "loss": 1.6478,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019147659063625453,
+      "loss": 3.1781,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019143657462985194,
+      "loss": 2.8516,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019139655862344937,
+      "loss": 2.3284,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019135654261704684,
+      "loss": 2.1883,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019131652661064427,
+      "loss": 2.1067,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001912765106042417,
+      "loss": 2.7757,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019123649459783915,
+      "loss": 2.7896,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019119647859143659,
+      "loss": 2.8477,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019115646258503402,
+      "loss": 2.5828,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019111644657863146,
+      "loss": 2.1935,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001910764305722289,
+      "loss": 2.9127,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019103641456582636,
+      "loss": 1.9655,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019099639855942377,
+      "loss": 1.9966,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001909563825530212,
+      "loss": 2.9128,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019091636654661867,
+      "loss": 2.5376,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019087635054021608,
+      "loss": 2.3073,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019083633453381352,
+      "loss": 2.3306,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019079631852741098,
+      "loss": 2.4587,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019075630252100842,
+      "loss": 1.9765,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019071628651460586,
+      "loss": 2.0567,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001906762705082033,
+      "loss": 2.4091,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019063625450180073,
+      "loss": 2.7292,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019059623849539817,
+      "loss": 2.567,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001905562224889956,
+      "loss": 2.6689,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019051620648259304,
+      "loss": 2.0399,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019047619047619048,
+      "loss": 2.2886,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019043617446978792,
+      "loss": 2.5066,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019039615846338536,
+      "loss": 2.2784,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019035614245698282,
+      "loss": 2.4476,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019031612645058023,
+      "loss": 2.5,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019027611044417767,
+      "loss": 2.6989,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019023609443777513,
+      "loss": 2.7026,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019019607843137254,
+      "loss": 2.2849,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019015606242497,
+      "loss": 2.1233,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019011604641856744,
+      "loss": 2.897,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019007603041216488,
+      "loss": 2.8736,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019003601440576232,
+      "loss": 2.5693,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018999599839935975,
+      "loss": 2.4526,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001899559823929572,
+      "loss": 2.8771,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018991596638655463,
+      "loss": 2.8311,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018987595038015207,
+      "loss": 1.9575,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001898359343737495,
+      "loss": 2.1558,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018979591836734697,
+      "loss": 2.4697,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018975590236094438,
+      "loss": 2.3132,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018971588635454181,
+      "loss": 1.844,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018967587034813928,
+      "loss": 2.4289,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001896358543417367,
+      "loss": 2.2221,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018959583833533415,
+      "loss": 2.2718,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001895558223289316,
+      "loss": 2.4795,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000189515806322529,
+      "loss": 1.5738,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018947579031612646,
+      "loss": 2.506,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001894357743097239,
+      "loss": 2.3838,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018939575830332134,
+      "loss": 2.8034,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018935574229691878,
+      "loss": 2.5689,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001893157262905162,
+      "loss": 2.5909,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018927571028411365,
+      "loss": 2.2591,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001892356942777111,
+      "loss": 2.5846,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018919567827130852,
+      "loss": 2.4161,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000189155662264906,
+      "loss": 2.6592,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018911564625850343,
+      "loss": 2.724,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018907563025210084,
+      "loss": 2.7169,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001890356142456983,
+      "loss": 2.1828,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018899559823929574,
+      "loss": 2.7578,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018895558223289315,
+      "loss": 1.5268,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001889155662264906,
+      "loss": 2.524,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018887555022008805,
+      "loss": 2.4782,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018883553421368549,
+      "loss": 2.9197,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018879551820728292,
+      "loss": 2.2168,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018875550220088036,
+      "loss": 2.2538,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001887154861944778,
+      "loss": 2.2098,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018867547018807523,
+      "loss": 2.7868,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018863545418167267,
+      "loss": 2.6976,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018859543817527014,
+      "loss": 2.0697,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018855542216886755,
+      "loss": 3.1233,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018851540616246498,
+      "loss": 2.3832,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018847539015606245,
+      "loss": 2.4145,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018843537414965988,
+      "loss": 2.574,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001883953581432573,
+      "loss": 2.4238,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018835534213685476,
+      "loss": 2.2535,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001883153261304522,
+      "loss": 2.332,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018827531012404963,
+      "loss": 2.515,
+      "step": 296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018823529411764707,
+      "loss": 3.3766,
+      "step": 297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001881952781112445,
+      "loss": 2.3375,
+      "step": 298
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018815526210484194,
+      "loss": 2.6023,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018811524609843938,
+      "loss": 2.038,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018807523009203682,
+      "loss": 2.706,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018803521408563428,
+      "loss": 2.6416,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001879951980792317,
+      "loss": 2.723,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018795518207282913,
+      "loss": 2.1009,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001879151660664266,
+      "loss": 2.2879,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000187875150060024,
+      "loss": 2.5803,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018783513405362144,
+      "loss": 2.8037,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001877951180472189,
+      "loss": 2.7607,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018775510204081634,
+      "loss": 2.0709,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018771508603441378,
+      "loss": 2.7237,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018767507002801122,
+      "loss": 2.3888,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018763505402160865,
+      "loss": 2.3898,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001875950380152061,
+      "loss": 2.5268,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018755502200880353,
+      "loss": 2.4985,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018751500600240096,
+      "loss": 2.7587,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018747498999599843,
+      "loss": 2.6997,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018743497398959584,
+      "loss": 2.6134,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018739495798319328,
+      "loss": 2.4134,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018735494197679074,
+      "loss": 2.3072,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018731492597038815,
+      "loss": 2.4875,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001872749099639856,
+      "loss": 1.8403,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018723489395758305,
+      "loss": 2.346,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001871948779511805,
+      "loss": 2.4517,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018715486194477793,
+      "loss": 2.002,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018711484593837536,
+      "loss": 1.7458,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001870748299319728,
+      "loss": 2.4809,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018703481392557024,
+      "loss": 2.5136,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018699479791916767,
+      "loss": 1.9867,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001869547819127651,
+      "loss": 1.9368,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018691476590636255,
+      "loss": 2.4258,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018687474989995999,
+      "loss": 2.5068,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018683473389355742,
+      "loss": 2.2874,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001867947178871549,
+      "loss": 1.9567,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001867547018807523,
+      "loss": 2.5752,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018671468587434976,
+      "loss": 2.5466,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001866746698679472,
+      "loss": 2.5752,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001866346538615446,
+      "loss": 2.6534,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018659463785514207,
+      "loss": 2.4042,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001865546218487395,
+      "loss": 2.3693,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018651460584233695,
+      "loss": 2.0041,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018647458983593438,
+      "loss": 2.4588,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018643457382953182,
+      "loss": 2.5291,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018639455782312926,
+      "loss": 2.4222,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001863545418167267,
+      "loss": 2.251,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018631452581032413,
+      "loss": 2.7183,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018627450980392157,
+      "loss": 2.4443,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018623449379751903,
+      "loss": 2.4805,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018619447779111644,
+      "loss": 2.4885,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001861544617847139,
+      "loss": 2.157,
+      "step": 349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018611444577831135,
+      "loss": 1.8644,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018607442977190876,
+      "loss": 1.7215,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018603441376550622,
+      "loss": 3.1201,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018599439775910366,
+      "loss": 1.6175,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018595438175270107,
+      "loss": 2.002,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018591436574629853,
+      "loss": 2.8634,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018587434973989597,
+      "loss": 1.3883,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001858343337334934,
+      "loss": 2.2189,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018579431772709084,
+      "loss": 2.2801,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018575430172068828,
+      "loss": 2.5704,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018571428571428572,
+      "loss": 2.8523,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018567426970788315,
+      "loss": 2.7758,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001856342537014806,
+      "loss": 2.1903,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018559423769507806,
+      "loss": 2.3763,
+      "step": 363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001855542216886755,
+      "loss": 1.7676,
+      "step": 364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001855142056822729,
+      "loss": 2.4345,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018547418967587037,
+      "loss": 2.5198,
+      "step": 366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001854341736694678,
+      "loss": 2.8238,
+      "step": 367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018539415766306521,
+      "loss": 2.3457,
+      "step": 368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018535414165666268,
+      "loss": 2.7674,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018531412565026012,
+      "loss": 1.8764,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018527410964385755,
+      "loss": 2.2971,
+      "step": 371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000185234093637455,
+      "loss": 2.056,
+      "step": 372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018519407763105243,
+      "loss": 2.1362,
+      "step": 373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018515406162464986,
+      "loss": 2.7978,
+      "step": 374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001851140456182473,
+      "loss": 2.3477,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018507402961184474,
+      "loss": 2.0771,
+      "step": 376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001850340136054422,
+      "loss": 2.489,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001849939975990396,
+      "loss": 2.4387,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018495398159263705,
+      "loss": 2.3755,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018491396558623451,
+      "loss": 2.506,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018487394957983195,
+      "loss": 2.1952,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018483393357342936,
+      "loss": 2.8305,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018479391756702683,
+      "loss": 2.203,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018475390156062426,
+      "loss": 2.6762,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001847138855542217,
+      "loss": 1.4625,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018467386954781914,
+      "loss": 2.4653,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018463385354141657,
+      "loss": 2.6149,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000184593837535014,
+      "loss": 1.673,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018455382152861145,
+      "loss": 2.3266,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018451380552220889,
+      "loss": 2.1573,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018447378951580635,
+      "loss": 2.4911,
+      "step": 391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018443377350940376,
+      "loss": 1.7787,
+      "step": 392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001843937575030012,
+      "loss": 1.9709,
+      "step": 393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018435374149659866,
+      "loss": 2.1781,
+      "step": 394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018431372549019607,
+      "loss": 2.3322,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001842737094837935,
+      "loss": 2.5637,
+      "step": 396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018423369347739097,
+      "loss": 2.4025,
+      "step": 397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001841936774709884,
+      "loss": 2.1836,
+      "step": 398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018415366146458585,
+      "loss": 2.3829,
+      "step": 399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018411364545818328,
+      "loss": 2.9786,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018407362945178072,
+      "loss": 2.5772,
+      "step": 401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018403361344537816,
+      "loss": 2.8161,
+      "step": 402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001839935974389756,
+      "loss": 2.2073,
+      "step": 403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018395358143257303,
+      "loss": 2.527,
+      "step": 404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001839135654261705,
+      "loss": 2.2615,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001838735494197679,
+      "loss": 1.8372,
+      "step": 406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018383353341336534,
+      "loss": 2.6518,
+      "step": 407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001837935174069628,
+      "loss": 1.8889,
+      "step": 408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018375350140056022,
+      "loss": 1.9596,
+      "step": 409
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018371348539415768,
+      "loss": 2.2496,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018367346938775512,
+      "loss": 2.2922,
+      "step": 411
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018363345338135256,
+      "loss": 2.0455,
+      "step": 412
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018359343737495,
+      "loss": 2.7018,
+      "step": 413
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018355342136854743,
+      "loss": 2.2142,
+      "step": 414
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018351340536214487,
+      "loss": 1.5765,
+      "step": 415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001834733893557423,
+      "loss": 2.7902,
+      "step": 416
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018343337334933974,
+      "loss": 2.4217,
+      "step": 417
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018339335734293718,
+      "loss": 2.0001,
+      "step": 418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018335334133653462,
+      "loss": 1.5834,
+      "step": 419
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018331332533013205,
+      "loss": 2.5395,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001832733093237295,
+      "loss": 2.3361,
+      "step": 421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018323329331732696,
+      "loss": 1.9557,
+      "step": 422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018319327731092437,
+      "loss": 1.9277,
+      "step": 423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018315326130452183,
+      "loss": 2.2461,
+      "step": 424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018311324529811927,
+      "loss": 2.4043,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018307322929171668,
+      "loss": 1.8751,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018303321328531414,
+      "loss": 2.0944,
+      "step": 427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018299319727891158,
+      "loss": 2.0928,
+      "step": 428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018295318127250902,
+      "loss": 2.1151,
+      "step": 429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018291316526610645,
+      "loss": 2.0688,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001828731492597039,
+      "loss": 2.7981,
+      "step": 431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018283313325330133,
+      "loss": 2.7819,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018279311724689876,
+      "loss": 1.7174,
+      "step": 433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001827531012404962,
+      "loss": 1.8135,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018271308523409364,
+      "loss": 2.6804,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001826730692276911,
+      "loss": 1.8679,
+      "step": 436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001826330532212885,
+      "loss": 1.7414,
+      "step": 437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018259303721488598,
+      "loss": 2.3104,
+      "step": 438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018255302120848341,
+      "loss": 2.09,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018251300520208082,
+      "loss": 1.8323,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001824729891956783,
+      "loss": 2.6055,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018243297318927573,
+      "loss": 2.8781,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018239295718287314,
+      "loss": 2.4153,
+      "step": 443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001823529411764706,
+      "loss": 2.3434,
+      "step": 444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018231292517006804,
+      "loss": 2.2337,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018227290916366547,
+      "loss": 2.4628,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001822328931572629,
+      "loss": 2.4959,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018219287715086035,
+      "loss": 1.682,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018215286114445779,
+      "loss": 1.8766,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018211284513805522,
+      "loss": 1.7845,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018207282913165266,
+      "loss": 2.0144,
+      "step": 451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018203281312525012,
+      "loss": 2.9573,
+      "step": 452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018199279711884756,
+      "loss": 2.2645,
+      "step": 453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018195278111244497,
+      "loss": 2.1878,
+      "step": 454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018191276510604243,
+      "loss": 2.4617,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018187274909963987,
+      "loss": 1.9717,
+      "step": 456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018183273309323728,
+      "loss": 2.236,
+      "step": 457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018179271708683475,
+      "loss": 2.1531,
+      "step": 458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018175270108043218,
+      "loss": 1.9326,
+      "step": 459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018171268507402962,
+      "loss": 2.6492,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018167266906762706,
+      "loss": 1.8045,
+      "step": 461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001816326530612245,
+      "loss": 2.7753,
+      "step": 462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018159263705482196,
+      "loss": 2.521,
+      "step": 463
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018155262104841937,
+      "loss": 2.1028,
+      "step": 464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001815126050420168,
+      "loss": 2.7477,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018147258903561427,
+      "loss": 1.8607,
+      "step": 466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018143257302921168,
+      "loss": 2.3435,
+      "step": 467
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018139255702280912,
+      "loss": 2.0561,
+      "step": 468
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018135254101640658,
+      "loss": 2.1388,
+      "step": 469
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018131252501000402,
+      "loss": 2.5231,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018127250900360146,
+      "loss": 1.7425,
+      "step": 471
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001812324929971989,
+      "loss": 3.2188,
+      "step": 472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018119247699079633,
+      "loss": 1.9749,
+      "step": 473
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018115246098439377,
+      "loss": 2.1866,
+      "step": 474
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001811124449779912,
+      "loss": 2.7056,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018107242897158864,
+      "loss": 2.4871,
+      "step": 476
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001810324129651861,
+      "loss": 2.1746,
+      "step": 477
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018099239695878352,
+      "loss": 1.6496,
+      "step": 478
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018095238095238095,
+      "loss": 2.5151,
+      "step": 479
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018091236494597842,
+      "loss": 2.0355,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018087234893957583,
+      "loss": 2.6009,
+      "step": 481
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018083233293317326,
+      "loss": 2.2995,
+      "step": 482
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018079231692677073,
+      "loss": 2.4788,
+      "step": 483
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018075230092036814,
+      "loss": 2.1549,
+      "step": 484
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001807122849139656,
+      "loss": 1.8112,
+      "step": 485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018067226890756304,
+      "loss": 1.8874,
+      "step": 486
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018063225290116048,
+      "loss": 2.8343,
+      "step": 487
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018059223689475791,
+      "loss": 1.9212,
+      "step": 488
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018055222088835535,
+      "loss": 2.2664,
+      "step": 489
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001805122048819528,
+      "loss": 2.5024,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018047218887555023,
+      "loss": 1.7213,
+      "step": 491
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018043217286914766,
+      "loss": 2.3551,
+      "step": 492
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001803921568627451,
+      "loss": 2.5244,
+      "step": 493
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018035214085634256,
+      "loss": 2.4853,
+      "step": 494
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018031212484993997,
+      "loss": 2.1432,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001802721088435374,
+      "loss": 2.5573,
+      "step": 496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018023209283713488,
+      "loss": 2.5133,
+      "step": 497
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018019207683073229,
+      "loss": 2.4406,
+      "step": 498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018015206082432975,
+      "loss": 2.5567,
+      "step": 499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001801120448179272,
+      "loss": 2.4936,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018007202881152462,
+      "loss": 1.8932,
+      "step": 501
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018003201280512206,
+      "loss": 2.2649,
+      "step": 502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001799919967987195,
+      "loss": 2.3932,
+      "step": 503
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017995198079231694,
+      "loss": 2.4055,
+      "step": 504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017991196478591437,
+      "loss": 2.7357,
+      "step": 505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001798719487795118,
+      "loss": 1.9217,
+      "step": 506
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017983193277310925,
+      "loss": 2.0731,
+      "step": 507
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017979191676670668,
+      "loss": 2.7828,
+      "step": 508
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017975190076030412,
+      "loss": 2.7458,
+      "step": 509
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017971188475390156,
+      "loss": 1.8189,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017967186874749902,
+      "loss": 1.5088,
+      "step": 511
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017963185274109643,
+      "loss": 2.0966,
+      "step": 512
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001795918367346939,
+      "loss": 2.3734,
+      "step": 513
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017955182072829133,
+      "loss": 2.2913,
+      "step": 514
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017951180472188874,
+      "loss": 2.1688,
+      "step": 515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001794717887154862,
+      "loss": 2.278,
+      "step": 516
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017943177270908365,
+      "loss": 2.6478,
+      "step": 517
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017939175670268108,
+      "loss": 2.2606,
+      "step": 518
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017935174069627852,
+      "loss": 2.0699,
+      "step": 519
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017931172468987596,
+      "loss": 2.1228,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001792717086834734,
+      "loss": 2.8582,
+      "step": 521
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017923169267707083,
+      "loss": 2.1589,
+      "step": 522
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017919167667066827,
+      "loss": 2.4889,
+      "step": 523
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001791516606642657,
+      "loss": 1.9924,
+      "step": 524
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017911164465786317,
+      "loss": 1.798,
+      "step": 525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017907162865146058,
+      "loss": 2.4916,
+      "step": 526
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017903161264505804,
+      "loss": 1.8108,
+      "step": 527
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017899159663865548,
+      "loss": 1.7716,
+      "step": 528
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001789515806322529,
+      "loss": 1.6814,
+      "step": 529
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017891156462585036,
+      "loss": 2.9978,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001788715486194478,
+      "loss": 2.4691,
+      "step": 531
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001788315326130452,
+      "loss": 1.9392,
+      "step": 532
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017879151660664267,
+      "loss": 1.8639,
+      "step": 533
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001787515006002401,
+      "loss": 1.9072,
+      "step": 534
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017871148459383754,
+      "loss": 2.5704,
+      "step": 535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017867146858743498,
+      "loss": 2.6424,
+      "step": 536
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017863145258103242,
+      "loss": 1.7177,
+      "step": 537
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017859143657462988,
+      "loss": 1.6264,
+      "step": 538
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001785514205682273,
+      "loss": 2.2916,
+      "step": 539
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017851140456182473,
+      "loss": 2.1877,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001784713885554222,
+      "loss": 2.2013,
+      "step": 541
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017843137254901963,
+      "loss": 2.1193,
+      "step": 542
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017839135654261704,
+      "loss": 1.8281,
+      "step": 543
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001783513405362145,
+      "loss": 2.7195,
+      "step": 544
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017831132452981194,
+      "loss": 1.7306,
+      "step": 545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017827130852340938,
+      "loss": 2.1653,
+      "step": 546
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017823129251700681,
+      "loss": 2.2027,
+      "step": 547
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017819127651060425,
+      "loss": 2.9335,
+      "step": 548
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001781512605042017,
+      "loss": 2.8271,
+      "step": 549
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017811124449779913,
+      "loss": 2.026,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017807122849139656,
+      "loss": 1.6128,
+      "step": 551
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017803121248499403,
+      "loss": 2.0749,
+      "step": 552
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017799119647859144,
+      "loss": 1.8946,
+      "step": 553
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017795118047218887,
+      "loss": 2.3571,
+      "step": 554
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017791116446578634,
+      "loss": 2.9032,
+      "step": 555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017787114845938375,
+      "loss": 1.8257,
+      "step": 556
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017783113245298119,
+      "loss": 2.1337,
+      "step": 557
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017779111644657865,
+      "loss": 2.5134,
+      "step": 558
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001777511004401761,
+      "loss": 2.376,
+      "step": 559
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017771108443377352,
+      "loss": 2.2385,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017767106842737096,
+      "loss": 2.0835,
+      "step": 561
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001776310524209684,
+      "loss": 2.0768,
+      "step": 562
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017759103641456584,
+      "loss": 2.497,
+      "step": 563
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017755102040816327,
+      "loss": 2.2766,
+      "step": 564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001775110044017607,
+      "loss": 2.0434,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017747098839535817,
+      "loss": 2.5981,
+      "step": 566
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017743097238895558,
+      "loss": 2.4195,
+      "step": 567
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017739095638255302,
+      "loss": 2.1889,
+      "step": 568
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017735094037615049,
+      "loss": 2.1678,
+      "step": 569
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001773109243697479,
+      "loss": 2.1947,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017727090836334533,
+      "loss": 2.105,
+      "step": 571
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001772308923569428,
+      "loss": 2.1824,
+      "step": 572
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001771908763505402,
+      "loss": 2.4479,
+      "step": 573
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017715086034413767,
+      "loss": 2.3133,
+      "step": 574
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001771108443377351,
+      "loss": 2.0597,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017707082833133255,
+      "loss": 1.9596,
+      "step": 576
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017703081232492998,
+      "loss": 1.7491,
+      "step": 577
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017699079631852742,
+      "loss": 2.4949,
+      "step": 578
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017695078031212486,
+      "loss": 1.3637,
+      "step": 579
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001769107643057223,
+      "loss": 2.1346,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017687074829931973,
+      "loss": 2.1142,
+      "step": 581
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017683073229291717,
+      "loss": 2.8376,
+      "step": 582
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017679071628651463,
+      "loss": 2.0926,
+      "step": 583
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017675070028011204,
+      "loss": 2.2315,
+      "step": 584
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017671068427370948,
+      "loss": 2.2118,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017667066826730694,
+      "loss": 2.3099,
+      "step": 586
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017663065226090435,
+      "loss": 2.2812,
+      "step": 587
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017659063625450182,
+      "loss": 2.0095,
+      "step": 588
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017655062024809926,
+      "loss": 1.7353,
+      "step": 589
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001765106042416967,
+      "loss": 2.45,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017647058823529413,
+      "loss": 2.2206,
+      "step": 591
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017643057222889157,
+      "loss": 2.8238,
+      "step": 592
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000176390556222489,
+      "loss": 2.3768,
+      "step": 593
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017635054021608644,
+      "loss": 1.872,
+      "step": 594
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017631052420968388,
+      "loss": 1.9693,
+      "step": 595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017627050820328131,
+      "loss": 2.2429,
+      "step": 596
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017623049219687875,
+      "loss": 1.777,
+      "step": 597
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001761904761904762,
+      "loss": 2.2782,
+      "step": 598
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017615046018407365,
+      "loss": 2.2919,
+      "step": 599
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001761104441776711,
+      "loss": 2.7059,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001760704281712685,
+      "loss": 2.2532,
+      "step": 601
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017603041216486596,
+      "loss": 1.9188,
+      "step": 602
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001759903961584634,
+      "loss": 1.9216,
+      "step": 603
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001759503801520608,
+      "loss": 1.3946,
+      "step": 604
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017591036414565828,
+      "loss": 2.7002,
+      "step": 605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001758703481392557,
+      "loss": 1.4859,
+      "step": 606
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017583033213285315,
+      "loss": 2.1447,
+      "step": 607
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001757903161264506,
+      "loss": 1.9749,
+      "step": 608
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017575030012004802,
+      "loss": 1.9336,
+      "step": 609
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017571028411364546,
+      "loss": 2.7327,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001756702681072429,
+      "loss": 2.3907,
+      "step": 611
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017563025210084034,
+      "loss": 2.0885,
+      "step": 612
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001755902360944378,
+      "loss": 1.8041,
+      "step": 613
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017555022008803524,
+      "loss": 2.1669,
+      "step": 614
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017551020408163265,
+      "loss": 2.2007,
+      "step": 615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001754701880752301,
+      "loss": 2.4028,
+      "step": 616
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017543017206882755,
+      "loss": 1.8257,
+      "step": 617
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017539015606242496,
+      "loss": 2.7215,
+      "step": 618
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017535014005602242,
+      "loss": 2.1703,
+      "step": 619
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017531012404961986,
+      "loss": 2.2216,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001752701080432173,
+      "loss": 2.2924,
+      "step": 621
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017523009203681473,
+      "loss": 2.2099,
+      "step": 622
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017519007603041217,
+      "loss": 2.2725,
+      "step": 623
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001751500600240096,
+      "loss": 2.9469,
+      "step": 624
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017511004401760705,
+      "loss": 2.1601,
+      "step": 625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017507002801120448,
+      "loss": 2.4726,
+      "step": 626
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017503001200480195,
+      "loss": 1.7014,
+      "step": 627
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017498999599839936,
+      "loss": 2.4223,
+      "step": 628
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001749499799919968,
+      "loss": 2.1217,
+      "step": 629
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017490996398559426,
+      "loss": 2.7866,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001748699479791917,
+      "loss": 2.0862,
+      "step": 631
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001748299319727891,
+      "loss": 2.048,
+      "step": 632
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017478991596638657,
+      "loss": 1.8158,
+      "step": 633
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000174749899959984,
+      "loss": 2.1224,
+      "step": 634
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017470988395358144,
+      "loss": 1.9936,
+      "step": 635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017466986794717888,
+      "loss": 2.4345,
+      "step": 636
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017462985194077632,
+      "loss": 2.3891,
+      "step": 637
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017458983593437376,
+      "loss": 2.9215,
+      "step": 638
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001745498199279712,
+      "loss": 2.3399,
+      "step": 639
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017450980392156863,
+      "loss": 1.5443,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001744697879151661,
+      "loss": 2.8036,
+      "step": 641
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001744297719087635,
+      "loss": 1.5529,
+      "step": 642
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017438975590236094,
+      "loss": 2.364,
+      "step": 643
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001743497398959584,
+      "loss": 1.9161,
+      "step": 644
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017430972388955582,
+      "loss": 1.7441,
+      "step": 645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017426970788315325,
+      "loss": 1.6795,
+      "step": 646
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017422969187675072,
+      "loss": 2.4516,
+      "step": 647
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017418967587034815,
+      "loss": 2.3205,
+      "step": 648
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001741496598639456,
+      "loss": 2.0446,
+      "step": 649
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017410964385754303,
+      "loss": 1.9882,
+      "step": 650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017406962785114047,
+      "loss": 1.7744,
+      "step": 651
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001740296118447379,
+      "loss": 2.5517,
+      "step": 652
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017398959583833534,
+      "loss": 2.4737,
+      "step": 653
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017394957983193278,
+      "loss": 1.8321,
+      "step": 654
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017390956382553024,
+      "loss": 2.2395,
+      "step": 655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017386954781912765,
+      "loss": 1.5972,
+      "step": 656
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001738295318127251,
+      "loss": 2.2278,
+      "step": 657
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017378951580632255,
+      "loss": 2.1072,
+      "step": 658
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017374949979991996,
+      "loss": 1.5139,
+      "step": 659
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001737094837935174,
+      "loss": 2.3718,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017366946778711486,
+      "loss": 2.1289,
+      "step": 661
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017362945178071227,
+      "loss": 2.0442,
+      "step": 662
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017358943577430974,
+      "loss": 1.469,
+      "step": 663
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017354941976790718,
+      "loss": 2.3961,
+      "step": 664
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001735094037615046,
+      "loss": 2.287,
+      "step": 665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017346938775510205,
+      "loss": 2.2943,
+      "step": 666
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001734293717486995,
+      "loss": 2.6142,
+      "step": 667
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017338935574229692,
+      "loss": 2.591,
+      "step": 668
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017334933973589436,
+      "loss": 2.2962,
+      "step": 669
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001733093237294918,
+      "loss": 1.9281,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017326930772308924,
+      "loss": 2.0738,
+      "step": 671
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001732292917166867,
+      "loss": 2.241,
+      "step": 672
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001731892757102841,
+      "loss": 2.4233,
+      "step": 673
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017314925970388157,
+      "loss": 2.8976,
+      "step": 674
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000173109243697479,
+      "loss": 2.6997,
+      "step": 675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017306922769107642,
+      "loss": 2.6673,
+      "step": 676
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017302921168467389,
+      "loss": 1.7809,
+      "step": 677
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017298919567827132,
+      "loss": 2.3575,
+      "step": 678
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017294917967186876,
+      "loss": 2.4624,
+      "step": 679
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001729091636654662,
+      "loss": 2.5794,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017286914765906363,
+      "loss": 2.0714,
+      "step": 681
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017282913165266107,
+      "loss": 2.6421,
+      "step": 682
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001727891156462585,
+      "loss": 1.9493,
+      "step": 683
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017274909963985595,
+      "loss": 2.4978,
+      "step": 684
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017270908363345338,
+      "loss": 2.3005,
+      "step": 685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017266906762705082,
+      "loss": 2.0951,
+      "step": 686
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017262905162064826,
+      "loss": 2.2001,
+      "step": 687
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017258903561424572,
+      "loss": 2.9344,
+      "step": 688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017254901960784316,
+      "loss": 1.8431,
+      "step": 689
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017250900360144057,
+      "loss": 1.8843,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017246898759503803,
+      "loss": 2.3605,
+      "step": 691
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017242897158863547,
+      "loss": 1.9253,
+      "step": 692
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017238895558223288,
+      "loss": 1.8711,
+      "step": 693
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017234893957583034,
+      "loss": 1.8807,
+      "step": 694
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017230892356942778,
+      "loss": 1.875,
+      "step": 695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017226890756302522,
+      "loss": 2.1703,
+      "step": 696
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017222889155662266,
+      "loss": 2.2602,
+      "step": 697
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001721888755502201,
+      "loss": 2.4873,
+      "step": 698
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017214885954381753,
+      "loss": 2.6009,
+      "step": 699
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017210884353741497,
+      "loss": 1.792,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001720688275310124,
+      "loss": 2.7877,
+      "step": 701
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017202881152460987,
+      "loss": 2.3959,
+      "step": 702
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001719887955182073,
+      "loss": 2.2927,
+      "step": 703
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017194877951180472,
+      "loss": 2.0216,
+      "step": 704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017190876350540218,
+      "loss": 2.0547,
+      "step": 705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017186874749899962,
+      "loss": 2.1836,
+      "step": 706
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017182873149259703,
+      "loss": 2.7734,
+      "step": 707
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001717887154861945,
+      "loss": 2.3965,
+      "step": 708
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017174869947979193,
+      "loss": 1.6988,
+      "step": 709
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017170868347338937,
+      "loss": 1.4617,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001716686674669868,
+      "loss": 2.6252,
+      "step": 711
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017162865146058424,
+      "loss": 2.5125,
+      "step": 712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017158863545418168,
+      "loss": 2.1846,
+      "step": 713
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017154861944777911,
+      "loss": 2.3308,
+      "step": 714
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017150860344137655,
+      "loss": 2.303,
+      "step": 715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017146858743497402,
+      "loss": 2.3854,
+      "step": 716
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017142857142857143,
+      "loss": 1.9557,
+      "step": 717
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017138855542216886,
+      "loss": 2.4457,
+      "step": 718
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017134853941576633,
+      "loss": 1.9869,
+      "step": 719
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017130852340936376,
+      "loss": 2.5617,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017126850740296117,
+      "loss": 2.3394,
+      "step": 721
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017122849139655864,
+      "loss": 2.6719,
+      "step": 722
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017118847539015608,
+      "loss": 2.2786,
+      "step": 723
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001711484593837535,
+      "loss": 2.2639,
+      "step": 724
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017110844337735095,
+      "loss": 1.4795,
+      "step": 725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001710684273709484,
+      "loss": 2.1897,
+      "step": 726
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017102841136454585,
+      "loss": 1.9899,
+      "step": 727
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017098839535814326,
+      "loss": 1.8909,
+      "step": 728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001709483793517407,
+      "loss": 2.698,
+      "step": 729
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017090836334533816,
+      "loss": 2.1793,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017086834733893557,
+      "loss": 2.1713,
+      "step": 731
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000170828331332533,
+      "loss": 1.8668,
+      "step": 732
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017078831532613047,
+      "loss": 1.5742,
+      "step": 733
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017074829931972788,
+      "loss": 2.3015,
+      "step": 734
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017070828331332535,
+      "loss": 1.5558,
+      "step": 735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017066826730692278,
+      "loss": 2.236,
+      "step": 736
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017062825130052022,
+      "loss": 1.9542,
+      "step": 737
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017058823529411766,
+      "loss": 1.5653,
+      "step": 738
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001705482192877151,
+      "loss": 2.1558,
+      "step": 739
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017050820328131253,
+      "loss": 2.214,
+      "step": 740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017046818727490997,
+      "loss": 2.4277,
+      "step": 741
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001704281712685074,
+      "loss": 1.8713,
+      "step": 742
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017038815526210484,
+      "loss": 2.9786,
+      "step": 743
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001703481392557023,
+      "loss": 1.813,
+      "step": 744
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017030812324929972,
+      "loss": 1.7951,
+      "step": 745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017026810724289716,
+      "loss": 2.1658,
+      "step": 746
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017022809123649462,
+      "loss": 1.9035,
+      "step": 747
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017018807523009203,
+      "loss": 1.9707,
+      "step": 748
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001701480592236895,
+      "loss": 2.5404,
+      "step": 749
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017010804321728693,
+      "loss": 2.0407,
+      "step": 750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017006802721088434,
+      "loss": 2.1652,
+      "step": 751
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001700280112044818,
+      "loss": 2.555,
+      "step": 752
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016998799519807924,
+      "loss": 2.6361,
+      "step": 753
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016994797919167668,
+      "loss": 2.2141,
+      "step": 754
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016990796318527412,
+      "loss": 2.3834,
+      "step": 755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016986794717887155,
+      "loss": 2.1748,
+      "step": 756
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000169827931172469,
+      "loss": 2.9337,
+      "step": 757
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016978791516606643,
+      "loss": 2.37,
+      "step": 758
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016974789915966387,
+      "loss": 2.4337,
+      "step": 759
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001697078831532613,
+      "loss": 1.3307,
+      "step": 760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016966786714685877,
+      "loss": 2.4019,
+      "step": 761
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016962785114045618,
+      "loss": 2.62,
+      "step": 762
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016958783513405364,
+      "loss": 2.341,
+      "step": 763
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016954781912765108,
+      "loss": 2.2419,
+      "step": 764
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001695078031212485,
+      "loss": 2.555,
+      "step": 765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016946778711484595,
+      "loss": 2.0542,
+      "step": 766
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001694277711084434,
+      "loss": 2.7052,
+      "step": 767
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016938775510204083,
+      "loss": 2.1298,
+      "step": 768
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016934773909563826,
+      "loss": 1.7939,
+      "step": 769
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001693077230892357,
+      "loss": 2.3624,
+      "step": 770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016926770708283314,
+      "loss": 1.4031,
+      "step": 771
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016922769107643058,
+      "loss": 2.2743,
+      "step": 772
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000169187675070028,
+      "loss": 2.9447,
+      "step": 773
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016914765906362545,
+      "loss": 1.9416,
+      "step": 774
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001691076430572229,
+      "loss": 2.0845,
+      "step": 775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016906762705082032,
+      "loss": 2.2151,
+      "step": 776
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001690276110444178,
+      "loss": 2.156,
+      "step": 777
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016898759503801523,
+      "loss": 1.9692,
+      "step": 778
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016894757903161264,
+      "loss": 2.1708,
+      "step": 779
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001689075630252101,
+      "loss": 1.8993,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016886754701880754,
+      "loss": 1.9354,
+      "step": 781
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016882753101240495,
+      "loss": 2.0455,
+      "step": 782
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001687875150060024,
+      "loss": 2.1618,
+      "step": 783
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016874749899959985,
+      "loss": 2.1965,
+      "step": 784
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016870748299319729,
+      "loss": 1.9377,
+      "step": 785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016866746698679472,
+      "loss": 2.2794,
+      "step": 786
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016862745098039216,
+      "loss": 2.3853,
+      "step": 787
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001685874349739896,
+      "loss": 2.1008,
+      "step": 788
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016854741896758703,
+      "loss": 2.158,
+      "step": 789
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016850740296118447,
+      "loss": 2.0221,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016846738695478194,
+      "loss": 2.4603,
+      "step": 791
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016842737094837937,
+      "loss": 2.0567,
+      "step": 792
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016838735494197678,
+      "loss": 2.0263,
+      "step": 793
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016834733893557425,
+      "loss": 2.5954,
+      "step": 794
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016830732292917168,
+      "loss": 2.331,
+      "step": 795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001682673069227691,
+      "loss": 2.3492,
+      "step": 796
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016822729091636656,
+      "loss": 1.8809,
+      "step": 797
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000168187274909964,
+      "loss": 2.1689,
+      "step": 798
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016814725890356143,
+      "loss": 2.1904,
+      "step": 799
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016810724289715887,
+      "loss": 2.5964,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001680672268907563,
+      "loss": 2.2492,
+      "step": 801
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016802721088435377,
+      "loss": 2.0947,
+      "step": 802
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016798719487795118,
+      "loss": 1.7344,
+      "step": 803
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016794717887154862,
+      "loss": 1.9779,
+      "step": 804
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016790716286514608,
+      "loss": 2.2866,
+      "step": 805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001678671468587435,
+      "loss": 1.661,
+      "step": 806
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016782713085234093,
+      "loss": 1.5546,
+      "step": 807
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001677871148459384,
+      "loss": 2.5106,
+      "step": 808
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016774709883953583,
+      "loss": 1.5681,
+      "step": 809
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016770708283313327,
+      "loss": 1.9092,
+      "step": 810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001676670668267307,
+      "loss": 2.2427,
+      "step": 811
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016762705082032814,
+      "loss": 1.9664,
+      "step": 812
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016758703481392558,
+      "loss": 2.2804,
+      "step": 813
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016754701880752302,
+      "loss": 2.2788,
+      "step": 814
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016750700280112045,
+      "loss": 1.8253,
+      "step": 815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016746698679471792,
+      "loss": 1.8586,
+      "step": 816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016742697078831533,
+      "loss": 2.5083,
+      "step": 817
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016738695478191277,
+      "loss": 1.7709,
+      "step": 818
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016734693877551023,
+      "loss": 1.8869,
+      "step": 819
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016730692276910764,
+      "loss": 1.8783,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016726690676270508,
+      "loss": 2.1599,
+      "step": 821
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016722689075630254,
+      "loss": 1.5157,
+      "step": 822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016718687474989995,
+      "loss": 1.8735,
+      "step": 823
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016714685874349742,
+      "loss": 2.0262,
+      "step": 824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016710684273709485,
+      "loss": 2.1857,
+      "step": 825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001670668267306923,
+      "loss": 1.97,
+      "step": 826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016702681072428973,
+      "loss": 2.1839,
+      "step": 827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016698679471788716,
+      "loss": 2.1885,
+      "step": 828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001669467787114846,
+      "loss": 2.3223,
+      "step": 829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016690676270508204,
+      "loss": 2.2384,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016686674669867948,
+      "loss": 2.2592,
+      "step": 831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001668267306922769,
+      "loss": 1.9808,
+      "step": 832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016678671468587438,
+      "loss": 2.1445,
+      "step": 833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001667466986794718,
+      "loss": 2.3307,
+      "step": 834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016670668267306922,
+      "loss": 2.3542,
+      "step": 835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001666666666666667,
+      "loss": 1.7206,
+      "step": 836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001666266506602641,
+      "loss": 2.5666,
+      "step": 837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016658663465386156,
+      "loss": 2.4184,
+      "step": 838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000166546618647459,
+      "loss": 1.8884,
+      "step": 839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001665066026410564,
+      "loss": 2.096,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016646658663465387,
+      "loss": 1.431,
+      "step": 841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001664265706282513,
+      "loss": 2.3119,
+      "step": 842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016638655462184875,
+      "loss": 1.5496,
+      "step": 843
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016634653861544619,
+      "loss": 2.0409,
+      "step": 844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016630652260904362,
+      "loss": 1.9,
+      "step": 845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016626650660264106,
+      "loss": 2.1615,
+      "step": 846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001662264905962385,
+      "loss": 2.1488,
+      "step": 847
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016618647458983593,
+      "loss": 2.7136,
+      "step": 848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016614645858343337,
+      "loss": 1.9453,
+      "step": 849
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016610644257703084,
+      "loss": 2.6599,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016606642657062825,
+      "loss": 1.6301,
+      "step": 851
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001660264105642257,
+      "loss": 2.0045,
+      "step": 852
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016598639455782315,
+      "loss": 2.1486,
+      "step": 853
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016594637855142056,
+      "loss": 2.1185,
+      "step": 854
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016590636254501802,
+      "loss": 2.0866,
+      "step": 855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016586634653861546,
+      "loss": 2.2957,
+      "step": 856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001658263305322129,
+      "loss": 2.0115,
+      "step": 857
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016578631452581033,
+      "loss": 2.2397,
+      "step": 858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016574629851940777,
+      "loss": 2.1513,
+      "step": 859
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001657062825130052,
+      "loss": 1.892,
+      "step": 860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016566626650660264,
+      "loss": 2.0866,
+      "step": 861
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016562625050020008,
+      "loss": 2.1868,
+      "step": 862
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016558623449379755,
+      "loss": 2.2407,
+      "step": 863
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016554621848739496,
+      "loss": 2.0784,
+      "step": 864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001655062024809924,
+      "loss": 2.6644,
+      "step": 865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016546618647458986,
+      "loss": 2.3561,
+      "step": 866
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001654261704681873,
+      "loss": 2.2997,
+      "step": 867
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001653861544617847,
+      "loss": 2.2484,
+      "step": 868
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016534613845538217,
+      "loss": 2.3085,
+      "step": 869
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001653061224489796,
+      "loss": 1.715,
+      "step": 870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016526610644257704,
+      "loss": 1.9108,
+      "step": 871
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016522609043617448,
+      "loss": 1.9286,
+      "step": 872
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016518607442977192,
+      "loss": 2.4083,
+      "step": 873
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016514605842336935,
+      "loss": 2.0751,
+      "step": 874
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001651060424169668,
+      "loss": 2.2536,
+      "step": 875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016506602641056423,
+      "loss": 1.8093,
+      "step": 876
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001650260104041617,
+      "loss": 1.6392,
+      "step": 877
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001649859943977591,
+      "loss": 1.8252,
+      "step": 878
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016494597839135654,
+      "loss": 2.4037,
+      "step": 879
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000164905962384954,
+      "loss": 2.5643,
+      "step": 880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016486594637855144,
+      "loss": 2.7392,
+      "step": 881
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016482593037214885,
+      "loss": 1.3942,
+      "step": 882
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016478591436574631,
+      "loss": 2.3889,
+      "step": 883
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016474589835934375,
+      "loss": 1.9887,
+      "step": 884
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001647058823529412,
+      "loss": 2.2005,
+      "step": 885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016466586634653863,
+      "loss": 2.224,
+      "step": 886
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016462585034013606,
+      "loss": 2.2737,
+      "step": 887
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001645858343337335,
+      "loss": 2.2455,
+      "step": 888
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016454581832733094,
+      "loss": 2.5637,
+      "step": 889
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016450580232092837,
+      "loss": 1.54,
+      "step": 890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016446578631452584,
+      "loss": 2.446,
+      "step": 891
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016442577030812325,
+      "loss": 2.1164,
+      "step": 892
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016438575430172069,
+      "loss": 2.26,
+      "step": 893
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016434573829531815,
+      "loss": 2.4475,
+      "step": 894
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016430572228891556,
+      "loss": 2.0904,
+      "step": 895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000164265706282513,
+      "loss": 2.1095,
+      "step": 896
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016422569027611046,
+      "loss": 2.5814,
+      "step": 897
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001641856742697079,
+      "loss": 2.1212,
+      "step": 898
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016414565826330534,
+      "loss": 1.8087,
+      "step": 899
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016410564225690277,
+      "loss": 1.8837,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001640656262505002,
+      "loss": 2.0621,
+      "step": 901
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016402561024409765,
+      "loss": 1.8056,
+      "step": 902
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016398559423769508,
+      "loss": 1.788,
+      "step": 903
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016394557823129252,
+      "loss": 2.2561,
+      "step": 904
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016390556222488999,
+      "loss": 2.5875,
+      "step": 905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001638655462184874,
+      "loss": 2.5147,
+      "step": 906
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016382553021208483,
+      "loss": 1.5063,
+      "step": 907
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001637855142056823,
+      "loss": 2.5279,
+      "step": 908
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001637454981992797,
+      "loss": 1.6787,
+      "step": 909
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016370548219287714,
+      "loss": 2.6156,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001636654661864746,
+      "loss": 2.3882,
+      "step": 911
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016362545018007202,
+      "loss": 2.3437,
+      "step": 912
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016358543417366948,
+      "loss": 1.8691,
+      "step": 913
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016354541816726692,
+      "loss": 2.3237,
+      "step": 914
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016350540216086436,
+      "loss": 1.9216,
+      "step": 915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001634653861544618,
+      "loss": 2.2101,
+      "step": 916
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016342537014805923,
+      "loss": 2.232,
+      "step": 917
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016338535414165667,
+      "loss": 1.7937,
+      "step": 918
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001633453381352541,
+      "loss": 2.3625,
+      "step": 919
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016330532212885154,
+      "loss": 2.3506,
+      "step": 920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016326530612244898,
+      "loss": 2.3042,
+      "step": 921
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016322529011604644,
+      "loss": 1.8249,
+      "step": 922
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016318527410964385,
+      "loss": 2.657,
+      "step": 923
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016314525810324132,
+      "loss": 2.006,
+      "step": 924
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016310524209683876,
+      "loss": 1.8111,
+      "step": 925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016306522609043617,
+      "loss": 2.0815,
+      "step": 926
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016302521008403363,
+      "loss": 2.4005,
+      "step": 927
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016298519407763107,
+      "loss": 1.7852,
+      "step": 928
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016294517807122848,
+      "loss": 1.5058,
+      "step": 929
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016290516206482594,
+      "loss": 2.3561,
+      "step": 930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016286514605842338,
+      "loss": 2.3475,
+      "step": 931
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016282513005202082,
+      "loss": 1.6493,
+      "step": 932
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016278511404561825,
+      "loss": 2.1625,
+      "step": 933
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001627450980392157,
+      "loss": 2.6869,
+      "step": 934
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016270508203281313,
+      "loss": 1.6962,
+      "step": 935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016266506602641056,
+      "loss": 2.3479,
+      "step": 936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000162625050020008,
+      "loss": 2.1091,
+      "step": 937
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016258503401360547,
+      "loss": 1.444,
+      "step": 938
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001625450180072029,
+      "loss": 1.808,
+      "step": 939
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001625050020008003,
+      "loss": 1.7449,
+      "step": 940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016246498599439778,
+      "loss": 1.9374,
+      "step": 941
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016242496998799521,
+      "loss": 2.5649,
+      "step": 942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016238495398159262,
+      "loss": 2.6465,
+      "step": 943
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001623449379751901,
+      "loss": 2.1063,
+      "step": 944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016230492196878753,
+      "loss": 1.9432,
+      "step": 945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016226490596238496,
+      "loss": 2.1635,
+      "step": 946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001622248899559824,
+      "loss": 2.0364,
+      "step": 947
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016218487394957984,
+      "loss": 1.802,
+      "step": 948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016214485794317727,
+      "loss": 2.0817,
+      "step": 949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001621048419367747,
+      "loss": 1.9124,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016206482593037215,
+      "loss": 2.0338,
+      "step": 951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001620248099239696,
+      "loss": 1.992,
+      "step": 952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016198479391756702,
+      "loss": 2.4837,
+      "step": 953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016194477791116446,
+      "loss": 1.9257,
+      "step": 954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016190476190476192,
+      "loss": 2.3757,
+      "step": 955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016186474589835936,
+      "loss": 1.9358,
+      "step": 956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016182472989195677,
+      "loss": 2.1047,
+      "step": 957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016178471388555424,
+      "loss": 2.3188,
+      "step": 958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016174469787915167,
+      "loss": 2.3619,
+      "step": 959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001617046818727491,
+      "loss": 2.1013,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016166466586634655,
+      "loss": 2.228,
+      "step": 961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016162464985994398,
+      "loss": 2.252,
+      "step": 962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016158463385354142,
+      "loss": 2.7519,
+      "step": 963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016154461784713886,
+      "loss": 2.2557,
+      "step": 964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001615046018407363,
+      "loss": 1.9979,
+      "step": 965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016146458583433376,
+      "loss": 2.22,
+      "step": 966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016142456982793117,
+      "loss": 2.4911,
+      "step": 967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001613845538215286,
+      "loss": 1.9945,
+      "step": 968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016134453781512607,
+      "loss": 2.1435,
+      "step": 969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001613045218087235,
+      "loss": 2.3941,
+      "step": 970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016126450580232092,
+      "loss": 1.941,
+      "step": 971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016122448979591838,
+      "loss": 1.8225,
+      "step": 972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016118447378951582,
+      "loss": 1.9243,
+      "step": 973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016114445778311326,
+      "loss": 1.7909,
+      "step": 974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001611044417767107,
+      "loss": 1.8608,
+      "step": 975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016106442577030813,
+      "loss": 1.4859,
+      "step": 976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016102440976390557,
+      "loss": 1.8744,
+      "step": 977
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000160984393757503,
+      "loss": 1.7605,
+      "step": 978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016094437775110044,
+      "loss": 2.1857,
+      "step": 979
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001609043617446979,
+      "loss": 2.3462,
+      "step": 980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016086434573829532,
+      "loss": 2.0983,
+      "step": 981
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016082432973189275,
+      "loss": 2.6781,
+      "step": 982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016078431372549022,
+      "loss": 1.9661,
+      "step": 983
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016074429771908763,
+      "loss": 2.2016,
+      "step": 984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016070428171268507,
+      "loss": 1.6887,
+      "step": 985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016066426570628253,
+      "loss": 2.1918,
+      "step": 986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016062424969987997,
+      "loss": 1.492,
+      "step": 987
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001605842336934774,
+      "loss": 1.5889,
+      "step": 988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016054421768707484,
+      "loss": 1.6462,
+      "step": 989
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016050420168067228,
+      "loss": 1.9719,
+      "step": 990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016046418567426972,
+      "loss": 1.9257,
+      "step": 991
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016042416966786715,
+      "loss": 1.8182,
+      "step": 992
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001603841536614646,
+      "loss": 1.9797,
+      "step": 993
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016034413765506205,
+      "loss": 2.1893,
+      "step": 994
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016030412164865946,
+      "loss": 2.1269,
+      "step": 995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001602641056422569,
+      "loss": 1.823,
+      "step": 996
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016022408963585437,
+      "loss": 2.6477,
+      "step": 997
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016018407362945178,
+      "loss": 1.5878,
+      "step": 998
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016014405762304924,
+      "loss": 2.0523,
+      "step": 999
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016010404161664668,
+      "loss": 1.4927,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001600640256102441,
+      "loss": 2.6192,
+      "step": 1001
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016002400960384155,
+      "loss": 2.1827,
+      "step": 1002
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000159983993597439,
+      "loss": 1.9916,
+      "step": 1003
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015994397759103643,
+      "loss": 2.5516,
+      "step": 1004
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015990396158463386,
+      "loss": 1.9769,
+      "step": 1005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001598639455782313,
+      "loss": 1.654,
+      "step": 1006
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015982392957182874,
+      "loss": 2.4923,
+      "step": 1007
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015978391356542617,
+      "loss": 2.405,
+      "step": 1008
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001597438975590236,
+      "loss": 1.8823,
+      "step": 1009
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015970388155262105,
+      "loss": 2.3527,
+      "step": 1010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001596638655462185,
+      "loss": 2.4795,
+      "step": 1011
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015962384953981592,
+      "loss": 2.0867,
+      "step": 1012
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015958383353341339,
+      "loss": 2.0823,
+      "step": 1013
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015954381752701082,
+      "loss": 1.9982,
+      "step": 1014
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015950380152060823,
+      "loss": 2.0486,
+      "step": 1015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001594637855142057,
+      "loss": 1.9614,
+      "step": 1016
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015942376950780313,
+      "loss": 1.7925,
+      "step": 1017
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015938375350140057,
+      "loss": 1.3296,
+      "step": 1018
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000159343737494998,
+      "loss": 1.8773,
+      "step": 1019
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015930372148859545,
+      "loss": 2.2769,
+      "step": 1020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015926370548219288,
+      "loss": 1.4871,
+      "step": 1021
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015922368947579032,
+      "loss": 2.2729,
+      "step": 1022
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015918367346938776,
+      "loss": 2.2899,
+      "step": 1023
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001591436574629852,
+      "loss": 2.1132,
+      "step": 1024
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015910364145658263,
+      "loss": 2.5167,
+      "step": 1025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015906362545018007,
+      "loss": 1.7298,
+      "step": 1026
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015902360944377753,
+      "loss": 1.9549,
+      "step": 1027
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015898359343737497,
+      "loss": 2.0648,
+      "step": 1028
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015894357743097238,
+      "loss": 2.3461,
+      "step": 1029
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015890356142456984,
+      "loss": 2.1358,
+      "step": 1030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015886354541816728,
+      "loss": 1.8153,
+      "step": 1031
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001588235294117647,
+      "loss": 1.7129,
+      "step": 1032
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015878351340536216,
+      "loss": 1.6489,
+      "step": 1033
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001587434973989596,
+      "loss": 1.7473,
+      "step": 1034
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015870348139255703,
+      "loss": 1.8732,
+      "step": 1035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015866346538615447,
+      "loss": 1.5127,
+      "step": 1036
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001586234493797519,
+      "loss": 2.6139,
+      "step": 1037
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015858343337334934,
+      "loss": 1.9774,
+      "step": 1038
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015854341736694678,
+      "loss": 2.4701,
+      "step": 1039
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015850340136054422,
+      "loss": 2.0707,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015846338535414168,
+      "loss": 2.0314,
+      "step": 1041
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001584233693477391,
+      "loss": 2.271,
+      "step": 1042
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015838335334133653,
+      "loss": 2.0935,
+      "step": 1043
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000158343337334934,
+      "loss": 1.8065,
+      "step": 1044
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015830332132853143,
+      "loss": 1.6621,
+      "step": 1045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015826330532212884,
+      "loss": 2.3964,
+      "step": 1046
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001582232893157263,
+      "loss": 2.4716,
+      "step": 1047
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015818327330932374,
+      "loss": 1.863,
+      "step": 1048
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015814325730292118,
+      "loss": 1.6046,
+      "step": 1049
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015810324129651861,
+      "loss": 2.1226,
+      "step": 1050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015806322529011605,
+      "loss": 1.8925,
+      "step": 1051
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015802320928371352,
+      "loss": 1.6478,
+      "step": 1052
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015798319327731093,
+      "loss": 2.0262,
+      "step": 1053
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015794317727090836,
+      "loss": 1.5139,
+      "step": 1054
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015790316126450583,
+      "loss": 1.8841,
+      "step": 1055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015786314525810324,
+      "loss": 1.419,
+      "step": 1056
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015782312925170067,
+      "loss": 1.5786,
+      "step": 1057
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015778311324529814,
+      "loss": 2.1095,
+      "step": 1058
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015774309723889558,
+      "loss": 1.6082,
+      "step": 1059
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000157703081232493,
+      "loss": 1.949,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015766306522609045,
+      "loss": 2.2217,
+      "step": 1061
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001576230492196879,
+      "loss": 2.3224,
+      "step": 1062
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015758303321328532,
+      "loss": 1.8298,
+      "step": 1063
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015754301720688276,
+      "loss": 2.0161,
+      "step": 1064
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001575030012004802,
+      "loss": 1.8009,
+      "step": 1065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015746298519407764,
+      "loss": 1.9129,
+      "step": 1066
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015742296918767507,
+      "loss": 2.1217,
+      "step": 1067
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001573829531812725,
+      "loss": 1.8139,
+      "step": 1068
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015734293717486997,
+      "loss": 1.9864,
+      "step": 1069
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015730292116846738,
+      "loss": 2.0045,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015726290516206482,
+      "loss": 1.8824,
+      "step": 1071
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015722288915566229,
+      "loss": 1.8849,
+      "step": 1072
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001571828731492597,
+      "loss": 2.0773,
+      "step": 1073
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015714285714285716,
+      "loss": 1.6164,
+      "step": 1074
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001571028411364546,
+      "loss": 1.9552,
+      "step": 1075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015706282513005203,
+      "loss": 1.589,
+      "step": 1076
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015702280912364947,
+      "loss": 1.9406,
+      "step": 1077
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001569827931172469,
+      "loss": 1.7341,
+      "step": 1078
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015694277711084435,
+      "loss": 2.0731,
+      "step": 1079
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015690276110444178,
+      "loss": 1.9928,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015686274509803922,
+      "loss": 2.3418,
+      "step": 1081
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015682272909163666,
+      "loss": 2.2024,
+      "step": 1082
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015678271308523412,
+      "loss": 2.1761,
+      "step": 1083
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015674269707883153,
+      "loss": 1.9624,
+      "step": 1084
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015670268107242897,
+      "loss": 2.5155,
+      "step": 1085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015666266506602643,
+      "loss": 1.8396,
+      "step": 1086
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015662264905962384,
+      "loss": 1.9172,
+      "step": 1087
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001565826330532213,
+      "loss": 2.5025,
+      "step": 1088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015654261704681874,
+      "loss": 2.3193,
+      "step": 1089
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015650260104041615,
+      "loss": 1.6626,
+      "step": 1090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015646258503401362,
+      "loss": 2.3058,
+      "step": 1091
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015642256902761106,
+      "loss": 1.9493,
+      "step": 1092
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001563825530212085,
+      "loss": 1.9899,
+      "step": 1093
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015634253701480593,
+      "loss": 1.84,
+      "step": 1094
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015630252100840337,
+      "loss": 1.8932,
+      "step": 1095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001562625050020008,
+      "loss": 1.9858,
+      "step": 1096
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015622248899559824,
+      "loss": 2.364,
+      "step": 1097
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015618247298919568,
+      "loss": 1.7757,
+      "step": 1098
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015614245698279312,
+      "loss": 2.0452,
+      "step": 1099
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015610244097639058,
+      "loss": 2.5436,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000156062424969988,
+      "loss": 2.0565,
+      "step": 1101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015602240896358545,
+      "loss": 2.1508,
+      "step": 1102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001559823929571829,
+      "loss": 1.9935,
+      "step": 1103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001559423769507803,
+      "loss": 2.3578,
+      "step": 1104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015590236094437777,
+      "loss": 1.4965,
+      "step": 1105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001558623449379752,
+      "loss": 1.6852,
+      "step": 1106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015582232893157264,
+      "loss": 2.0506,
+      "step": 1107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015578231292517008,
+      "loss": 2.3409,
+      "step": 1108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015574229691876751,
+      "loss": 1.8834,
+      "step": 1109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015570228091236495,
+      "loss": 2.2993,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001556622649059624,
+      "loss": 2.4905,
+      "step": 1111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015562224889955983,
+      "loss": 2.3296,
+      "step": 1112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015558223289315726,
+      "loss": 1.5256,
+      "step": 1113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001555422168867547,
+      "loss": 1.7128,
+      "step": 1114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015550220088035214,
+      "loss": 2.2662,
+      "step": 1115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001554621848739496,
+      "loss": 1.7451,
+      "step": 1116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015542216886754704,
+      "loss": 1.6614,
+      "step": 1117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015538215286114445,
+      "loss": 2.5297,
+      "step": 1118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001553421368547419,
+      "loss": 1.9926,
+      "step": 1119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015530212084833935,
+      "loss": 1.6838,
+      "step": 1120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015526210484193676,
+      "loss": 2.374,
+      "step": 1121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015522208883553422,
+      "loss": 1.7455,
+      "step": 1122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015518207282913166,
+      "loss": 1.8462,
+      "step": 1123
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001551420568227291,
+      "loss": 2.7022,
+      "step": 1124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015510204081632654,
+      "loss": 1.6705,
+      "step": 1125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015506202480992397,
+      "loss": 1.8233,
+      "step": 1126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015502200880352144,
+      "loss": 2.5587,
+      "step": 1127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015498199279711885,
+      "loss": 1.9258,
+      "step": 1128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015494197679071628,
+      "loss": 1.6437,
+      "step": 1129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015490196078431375,
+      "loss": 2.2489,
+      "step": 1130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015486194477791116,
+      "loss": 2.1837,
+      "step": 1131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001548219287715086,
+      "loss": 1.8461,
+      "step": 1132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015478191276510606,
+      "loss": 1.7993,
+      "step": 1133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001547418967587035,
+      "loss": 2.1182,
+      "step": 1134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015470188075230093,
+      "loss": 2.4564,
+      "step": 1135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015466186474589837,
+      "loss": 2.185,
+      "step": 1136
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001546218487394958,
+      "loss": 2.3812,
+      "step": 1137
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015458183273309325,
+      "loss": 1.7463,
+      "step": 1138
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015454181672669068,
+      "loss": 2.6077,
+      "step": 1139
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015450180072028812,
+      "loss": 2.1564,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015446178471388558,
+      "loss": 1.8109,
+      "step": 1141
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000154421768707483,
+      "loss": 2.2428,
+      "step": 1142
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015438175270108043,
+      "loss": 1.877,
+      "step": 1143
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001543417366946779,
+      "loss": 1.8368,
+      "step": 1144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001543017206882753,
+      "loss": 2.0505,
+      "step": 1145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015426170468187274,
+      "loss": 1.9484,
+      "step": 1146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001542216886754702,
+      "loss": 2.1635,
+      "step": 1147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015418167266906764,
+      "loss": 1.8146,
+      "step": 1148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015414165666266508,
+      "loss": 1.9318,
+      "step": 1149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015410164065626252,
+      "loss": 2.0888,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015406162464985996,
+      "loss": 2.5991,
+      "step": 1151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001540216086434574,
+      "loss": 2.123,
+      "step": 1152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015398159263705483,
+      "loss": 2.221,
+      "step": 1153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015394157663065227,
+      "loss": 2.004,
+      "step": 1154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001539015606242497,
+      "loss": 2.2268,
+      "step": 1155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015386154461784714,
+      "loss": 2.2109,
+      "step": 1156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015382152861144458,
+      "loss": 2.352,
+      "step": 1157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015378151260504204,
+      "loss": 2.1579,
+      "step": 1158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015374149659863945,
+      "loss": 2.4308,
+      "step": 1159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001537014805922369,
+      "loss": 1.9401,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015366146458583435,
+      "loss": 1.6674,
+      "step": 1161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015362144857943176,
+      "loss": 1.7883,
+      "step": 1162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015358143257302923,
+      "loss": 2.1256,
+      "step": 1163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015354141656662666,
+      "loss": 2.389,
+      "step": 1164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001535014005602241,
+      "loss": 2.1518,
+      "step": 1165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015346138455382154,
+      "loss": 1.86,
+      "step": 1166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015342136854741898,
+      "loss": 1.4581,
+      "step": 1167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001533813525410164,
+      "loss": 2.264,
+      "step": 1168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015334133653461385,
+      "loss": 1.7915,
+      "step": 1169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001533013205282113,
+      "loss": 1.6316,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015326130452180872,
+      "loss": 2.1369,
+      "step": 1171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001532212885154062,
+      "loss": 1.8406,
+      "step": 1172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001531812725090036,
+      "loss": 1.7616,
+      "step": 1173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015314125650260104,
+      "loss": 1.8737,
+      "step": 1174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001531012404961985,
+      "loss": 2.1754,
+      "step": 1175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001530612244897959,
+      "loss": 2.333,
+      "step": 1176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015302120848339337,
+      "loss": 1.4502,
+      "step": 1177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001529811924769908,
+      "loss": 2.3619,
+      "step": 1178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015294117647058822,
+      "loss": 2.3083,
+      "step": 1179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015290116046418569,
+      "loss": 2.1789,
+      "step": 1180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015286114445778312,
+      "loss": 2.1624,
+      "step": 1181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015282112845138056,
+      "loss": 2.165,
+      "step": 1182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000152781112444978,
+      "loss": 2.0024,
+      "step": 1183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015274109643857543,
+      "loss": 1.9954,
+      "step": 1184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015270108043217287,
+      "loss": 2.2844,
+      "step": 1185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001526610644257703,
+      "loss": 2.0862,
+      "step": 1186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015262104841936775,
+      "loss": 1.8377,
+      "step": 1187
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001525810324129652,
+      "loss": 1.9576,
+      "step": 1188
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015254101640656265,
+      "loss": 1.9774,
+      "step": 1189
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015250100040016006,
+      "loss": 2.4056,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015246098439375752,
+      "loss": 1.9081,
+      "step": 1191
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015242096838735496,
+      "loss": 1.6801,
+      "step": 1192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015238095238095237,
+      "loss": 2.0168,
+      "step": 1193
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015234093637454983,
+      "loss": 1.6703,
+      "step": 1194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015230092036814727,
+      "loss": 2.4108,
+      "step": 1195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001522609043617447,
+      "loss": 1.3637,
+      "step": 1196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015222088835534214,
+      "loss": 2.3421,
+      "step": 1197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015218087234893958,
+      "loss": 1.5161,
+      "step": 1198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015214085634253702,
+      "loss": 2.0432,
+      "step": 1199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015210084033613446,
+      "loss": 1.3868,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001520608243297319,
+      "loss": 2.1078,
+      "step": 1201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015202080832332936,
+      "loss": 2.0119,
+      "step": 1202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015198079231692677,
+      "loss": 2.1267,
+      "step": 1203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001519407763105242,
+      "loss": 2.5599,
+      "step": 1204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015190076030412167,
+      "loss": 2.2434,
+      "step": 1205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001518607442977191,
+      "loss": 2.0267,
+      "step": 1206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015182072829131652,
+      "loss": 2.62,
+      "step": 1207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015178071228491398,
+      "loss": 1.7283,
+      "step": 1208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015174069627851142,
+      "loss": 1.6918,
+      "step": 1209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015170068027210885,
+      "loss": 1.7263,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001516606642657063,
+      "loss": 1.6578,
+      "step": 1211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015162064825930373,
+      "loss": 2.236,
+      "step": 1212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015158063225290117,
+      "loss": 2.2158,
+      "step": 1213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001515406162464986,
+      "loss": 1.9396,
+      "step": 1214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015150060024009604,
+      "loss": 1.5751,
+      "step": 1215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001514605842336935,
+      "loss": 2.0367,
+      "step": 1216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015142056822729091,
+      "loss": 2.2887,
+      "step": 1217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015138055222088835,
+      "loss": 1.8762,
+      "step": 1218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015134053621448582,
+      "loss": 1.9315,
+      "step": 1219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015130052020808323,
+      "loss": 2.0764,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015126050420168066,
+      "loss": 2.3108,
+      "step": 1221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015122048819527813,
+      "loss": 1.9929,
+      "step": 1222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015118047218887556,
+      "loss": 2.054,
+      "step": 1223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000151140456182473,
+      "loss": 1.8089,
+      "step": 1224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015110044017607044,
+      "loss": 1.8851,
+      "step": 1225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015106042416966788,
+      "loss": 2.4848,
+      "step": 1226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001510204081632653,
+      "loss": 1.8836,
+      "step": 1227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015098039215686275,
+      "loss": 1.5176,
+      "step": 1228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001509403761504602,
+      "loss": 2.2969,
+      "step": 1229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015090036014405765,
+      "loss": 2.5116,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015086034413765506,
+      "loss": 1.9821,
+      "step": 1231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001508203281312525,
+      "loss": 2.4528,
+      "step": 1232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015078031212484996,
+      "loss": 2.4104,
+      "step": 1233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015074029611844737,
+      "loss": 1.8971,
+      "step": 1234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001507002801120448,
+      "loss": 1.85,
+      "step": 1235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015066026410564227,
+      "loss": 2.3283,
+      "step": 1236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001506202480992397,
+      "loss": 1.9802,
+      "step": 1237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015058023209283715,
+      "loss": 1.3983,
+      "step": 1238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015054021608643459,
+      "loss": 2.0305,
+      "step": 1239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015050020008003202,
+      "loss": 2.2237,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015046018407362946,
+      "loss": 1.9244,
+      "step": 1241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001504201680672269,
+      "loss": 1.9364,
+      "step": 1242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015038015206082433,
+      "loss": 1.894,
+      "step": 1243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015034013605442177,
+      "loss": 2.055,
+      "step": 1244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001503001200480192,
+      "loss": 1.8337,
+      "step": 1245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015026010404161665,
+      "loss": 1.8779,
+      "step": 1246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001502200880352141,
+      "loss": 2.8938,
+      "step": 1247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015018007202881152,
+      "loss": 1.6643,
+      "step": 1248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015014005602240896,
+      "loss": 1.9489,
+      "step": 1249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015010004001600642,
+      "loss": 2.2248,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015006002400960383,
+      "loss": 1.4923,
+      "step": 1251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001500200080032013,
+      "loss": 1.6983,
+      "step": 1252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00014997999199679873,
+      "loss": 2.5119,
+      "step": 1253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00014993997599039617,
+      "loss": 2.3995,
+      "step": 1254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001498999599839936,
+      "loss": 2.2125,
+      "step": 1255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00014985994397759104,
+      "loss": 2.4704,
+      "step": 1256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014981992797118848,
+      "loss": 1.8761,
+      "step": 1257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014977991196478592,
+      "loss": 2.0487,
+      "step": 1258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014973989595838336,
+      "loss": 1.5504,
+      "step": 1259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001496998799519808,
+      "loss": 1.9258,
+      "step": 1260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014965986394557826,
+      "loss": 1.7111,
+      "step": 1261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014961984793917567,
+      "loss": 1.8616,
+      "step": 1262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014957983193277313,
+      "loss": 2.026,
+      "step": 1263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014953981592637057,
+      "loss": 1.9237,
+      "step": 1264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014949979991996798,
+      "loss": 2.0872,
+      "step": 1265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014945978391356544,
+      "loss": 2.0091,
+      "step": 1266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014941976790716288,
+      "loss": 1.7883,
+      "step": 1267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001493797519007603,
+      "loss": 1.9689,
+      "step": 1268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014933973589435775,
+      "loss": 1.8432,
+      "step": 1269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001492997198879552,
+      "loss": 1.8215,
+      "step": 1270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014925970388155263,
+      "loss": 2.4972,
+      "step": 1271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014921968787515007,
+      "loss": 1.9231,
+      "step": 1272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001491796718687475,
+      "loss": 1.4068,
+      "step": 1273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014913965586234494,
+      "loss": 1.7801,
+      "step": 1274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014909963985594238,
+      "loss": 1.9348,
+      "step": 1275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014905962384953981,
+      "loss": 2.2735,
+      "step": 1276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014901960784313728,
+      "loss": 2.4899,
+      "step": 1277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014897959183673472,
+      "loss": 1.8599,
+      "step": 1278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014893957583033213,
+      "loss": 1.9651,
+      "step": 1279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001488995598239296,
+      "loss": 1.6631,
+      "step": 1280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014885954381752703,
+      "loss": 1.5984,
+      "step": 1281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014881952781112444,
+      "loss": 2.1402,
+      "step": 1282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001487795118047219,
+      "loss": 1.7417,
+      "step": 1283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014873949579831934,
+      "loss": 1.5726,
+      "step": 1284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014869947979191678,
+      "loss": 1.6506,
+      "step": 1285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001486594637855142,
+      "loss": 1.8353,
+      "step": 1286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014861944777911165,
+      "loss": 1.7819,
+      "step": 1287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001485794317727091,
+      "loss": 1.9647,
+      "step": 1288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014853941576630652,
+      "loss": 2.5818,
+      "step": 1289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014849939975990396,
+      "loss": 1.9015,
+      "step": 1290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014845938375350142,
+      "loss": 2.0711,
+      "step": 1291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014841936774709884,
+      "loss": 2.1623,
+      "step": 1292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014837935174069627,
+      "loss": 2.0442,
+      "step": 1293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014833933573429374,
+      "loss": 1.8933,
+      "step": 1294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014829931972789117,
+      "loss": 1.9069,
+      "step": 1295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014825930372148858,
+      "loss": 2.1731,
+      "step": 1296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014821928771508605,
+      "loss": 1.7905,
+      "step": 1297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014817927170868348,
+      "loss": 2.5381,
+      "step": 1298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014813925570228092,
+      "loss": 1.6908,
+      "step": 1299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014809923969587836,
+      "loss": 2.1598,
+      "step": 1300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001480592236894758,
+      "loss": 2.3946,
+      "step": 1301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014801920768307323,
+      "loss": 1.9775,
+      "step": 1302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014797919167667067,
+      "loss": 1.8727,
+      "step": 1303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001479391756702681,
+      "loss": 2.0839,
+      "step": 1304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014789915966386557,
+      "loss": 1.8909,
+      "step": 1305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014785914365746298,
+      "loss": 2.0087,
+      "step": 1306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014781912765106042,
+      "loss": 2.3748,
+      "step": 1307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014777911164465788,
+      "loss": 2.5983,
+      "step": 1308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001477390956382553,
+      "loss": 2.0043,
+      "step": 1309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014769907963185273,
+      "loss": 2.0366,
+      "step": 1310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001476590636254502,
+      "loss": 2.0752,
+      "step": 1311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014761904761904763,
+      "loss": 1.8422,
+      "step": 1312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014757903161264507,
+      "loss": 2.3972,
+      "step": 1313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001475390156062425,
+      "loss": 2.4406,
+      "step": 1314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014749899959983994,
+      "loss": 1.4946,
+      "step": 1315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014745898359343738,
+      "loss": 1.8278,
+      "step": 1316
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014741896758703482,
+      "loss": 2.3565,
+      "step": 1317
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014737895158063225,
+      "loss": 1.6994,
+      "step": 1318
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014733893557422972,
+      "loss": 1.9796,
+      "step": 1319
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014729891956782713,
+      "loss": 1.6155,
+      "step": 1320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014725890356142457,
+      "loss": 1.7758,
+      "step": 1321
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014721888755502203,
+      "loss": 1.7476,
+      "step": 1322
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014717887154861944,
+      "loss": 1.2569,
+      "step": 1323
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001471388555422169,
+      "loss": 2.3388,
+      "step": 1324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014709883953581434,
+      "loss": 2.0784,
+      "step": 1325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014705882352941178,
+      "loss": 2.3286,
+      "step": 1326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014701880752300922,
+      "loss": 1.8259,
+      "step": 1327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014697879151660665,
+      "loss": 1.4462,
+      "step": 1328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001469387755102041,
+      "loss": 2.1022,
+      "step": 1329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014689875950380153,
+      "loss": 2.3188,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014685874349739896,
+      "loss": 2.2504,
+      "step": 1331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001468187274909964,
+      "loss": 1.974,
+      "step": 1332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014677871148459384,
+      "loss": 2.0224,
+      "step": 1333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014673869547819128,
+      "loss": 2.439,
+      "step": 1334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001466986794717887,
+      "loss": 2.0199,
+      "step": 1335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014665866346538618,
+      "loss": 1.9899,
+      "step": 1336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001466186474589836,
+      "loss": 1.6207,
+      "step": 1337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014657863145258105,
+      "loss": 2.1692,
+      "step": 1338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001465386154461785,
+      "loss": 2.4439,
+      "step": 1339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001464985994397759,
+      "loss": 1.7696,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014645858343337336,
+      "loss": 2.108,
+      "step": 1341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001464185674269708,
+      "loss": 1.8277,
+      "step": 1342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014637855142056824,
+      "loss": 1.9351,
+      "step": 1343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014633853541416567,
+      "loss": 2.0285,
+      "step": 1344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001462985194077631,
+      "loss": 1.872,
+      "step": 1345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014625850340136055,
+      "loss": 1.7317,
+      "step": 1346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014621848739495799,
+      "loss": 1.5341,
+      "step": 1347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014617847138855542,
+      "loss": 2.2625,
+      "step": 1348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014613845538215286,
+      "loss": 2.0486,
+      "step": 1349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014609843937575032,
+      "loss": 1.7717,
+      "step": 1350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014605842336934773,
+      "loss": 1.7981,
+      "step": 1351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001460184073629452,
+      "loss": 2.3144,
+      "step": 1352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014597839135654264,
+      "loss": 1.9189,
+      "step": 1353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014593837535014005,
+      "loss": 2.6631,
+      "step": 1354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001458983593437375,
+      "loss": 2.0742,
+      "step": 1355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014585834333733495,
+      "loss": 2.1521,
+      "step": 1356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014581832733093236,
+      "loss": 2.515,
+      "step": 1357
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014577831132452982,
+      "loss": 2.1538,
+      "step": 1358
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014573829531812726,
+      "loss": 1.8168,
+      "step": 1359
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001456982793117247,
+      "loss": 1.7483,
+      "step": 1360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014565826330532213,
+      "loss": 1.4721,
+      "step": 1361
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014561824729891957,
+      "loss": 2.4069,
+      "step": 1362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000145578231292517,
+      "loss": 2.1333,
+      "step": 1363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014553821528611444,
+      "loss": 2.1091,
+      "step": 1364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014549819927971188,
+      "loss": 2.465,
+      "step": 1365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014545818327330935,
+      "loss": 1.5637,
+      "step": 1366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014541816726690678,
+      "loss": 2.1033,
+      "step": 1367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001453781512605042,
+      "loss": 2.001,
+      "step": 1368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014533813525410166,
+      "loss": 2.3498,
+      "step": 1369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001452981192476991,
+      "loss": 1.5395,
+      "step": 1370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001452581032412965,
+      "loss": 2.3402,
+      "step": 1371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014521808723489397,
+      "loss": 2.5206,
+      "step": 1372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001451780712284914,
+      "loss": 1.7649,
+      "step": 1373
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014513805522208884,
+      "loss": 2.1437,
+      "step": 1374
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014509803921568628,
+      "loss": 1.8356,
+      "step": 1375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014505802320928372,
+      "loss": 1.5904,
+      "step": 1376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014501800720288115,
+      "loss": 1.8345,
+      "step": 1377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001449779911964786,
+      "loss": 2.099,
+      "step": 1378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014493797519007603,
+      "loss": 1.7924,
+      "step": 1379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001448979591836735,
+      "loss": 1.6722,
+      "step": 1380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001448579431772709,
+      "loss": 2.0183,
+      "step": 1381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014481792717086834,
+      "loss": 2.3582,
+      "step": 1382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001447779111644658,
+      "loss": 2.0216,
+      "step": 1383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014473789515806324,
+      "loss": 1.8,
+      "step": 1384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014469787915166065,
+      "loss": 1.8582,
+      "step": 1385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014465786314525812,
+      "loss": 2.3866,
+      "step": 1386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014461784713885555,
+      "loss": 1.9061,
+      "step": 1387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000144577831132453,
+      "loss": 1.2192,
+      "step": 1388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014453781512605043,
+      "loss": 1.7948,
+      "step": 1389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014449779911964786,
+      "loss": 2.3846,
+      "step": 1390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014445778311324533,
+      "loss": 1.4744,
+      "step": 1391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014441776710684274,
+      "loss": 1.7186,
+      "step": 1392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014437775110044018,
+      "loss": 2.1743,
+      "step": 1393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014433773509403764,
+      "loss": 2.2108,
+      "step": 1394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014429771908763505,
+      "loss": 1.9921,
+      "step": 1395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001442577030812325,
+      "loss": 1.9125,
+      "step": 1396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014421768707482995,
+      "loss": 2.1688,
+      "step": 1397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014417767106842736,
+      "loss": 1.7113,
+      "step": 1398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014413765506202483,
+      "loss": 1.7394,
+      "step": 1399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014409763905562226,
+      "loss": 2.4163,
+      "step": 1400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001440576230492197,
+      "loss": 2.0559,
+      "step": 1401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014401760704281714,
+      "loss": 2.0475,
+      "step": 1402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014397759103641457,
+      "loss": 2.382,
+      "step": 1403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000143937575030012,
+      "loss": 1.4566,
+      "step": 1404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014389755902360945,
+      "loss": 2.0937,
+      "step": 1405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014385754301720689,
+      "loss": 2.4385,
+      "step": 1406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014381752701080432,
+      "loss": 1.948,
+      "step": 1407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001437775110044018,
+      "loss": 1.5473,
+      "step": 1408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001437374949979992,
+      "loss": 1.8188,
+      "step": 1409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014369747899159663,
+      "loss": 2.0386,
+      "step": 1410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001436574629851941,
+      "loss": 1.6662,
+      "step": 1411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001436174469787915,
+      "loss": 1.9699,
+      "step": 1412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014357743097238897,
+      "loss": 2.6015,
+      "step": 1413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001435374149659864,
+      "loss": 1.6636,
+      "step": 1414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014349739895958385,
+      "loss": 1.4465,
+      "step": 1415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014345738295318128,
+      "loss": 2.0151,
+      "step": 1416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014341736694677872,
+      "loss": 2.1348,
+      "step": 1417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014337735094037616,
+      "loss": 2.0484,
+      "step": 1418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001433373349339736,
+      "loss": 2.2374,
+      "step": 1419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014329731892757103,
+      "loss": 2.2115,
+      "step": 1420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014325730292116847,
+      "loss": 2.1047,
+      "step": 1421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001432172869147659,
+      "loss": 2.2989,
+      "step": 1422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014317727090836334,
+      "loss": 1.66,
+      "step": 1423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014313725490196078,
+      "loss": 2.7361,
+      "step": 1424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014309723889555825,
+      "loss": 1.7433,
+      "step": 1425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014305722288915566,
+      "loss": 2.3431,
+      "step": 1426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014301720688275312,
+      "loss": 2.1979,
+      "step": 1427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014297719087635056,
+      "loss": 2.2289,
+      "step": 1428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014293717486994797,
+      "loss": 1.5029,
+      "step": 1429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014289715886354543,
+      "loss": 1.78,
+      "step": 1430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014285714285714287,
+      "loss": 2.4771,
+      "step": 1431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001428171268507403,
+      "loss": 1.8063,
+      "step": 1432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014277711084433774,
+      "loss": 1.4802,
+      "step": 1433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014273709483793518,
+      "loss": 2.281,
+      "step": 1434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014269707883153262,
+      "loss": 2.3952,
+      "step": 1435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014265706282513005,
+      "loss": 2.0231,
+      "step": 1436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001426170468187275,
+      "loss": 2.4425,
+      "step": 1437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014257703081232493,
+      "loss": 1.8878,
+      "step": 1438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001425370148059224,
+      "loss": 2.0959,
+      "step": 1439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001424969987995198,
+      "loss": 2.0873,
+      "step": 1440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014245698279311727,
+      "loss": 2.1824,
+      "step": 1441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001424169667867147,
+      "loss": 1.5982,
+      "step": 1442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001423769507803121,
+      "loss": 2.1279,
+      "step": 1443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014233693477390958,
+      "loss": 2.4931,
+      "step": 1444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014229691876750701,
+      "loss": 1.7542,
+      "step": 1445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014225690276110442,
+      "loss": 2.191,
+      "step": 1446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001422168867547019,
+      "loss": 1.7735,
+      "step": 1447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014217687074829933,
+      "loss": 2.0384,
+      "step": 1448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014213685474189676,
+      "loss": 2.2117,
+      "step": 1449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001420968387354942,
+      "loss": 1.9488,
+      "step": 1450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014205682272909164,
+      "loss": 2.6986,
+      "step": 1451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001420168067226891,
+      "loss": 2.306,
+      "step": 1452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001419767907162865,
+      "loss": 1.9352,
+      "step": 1453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014193677470988395,
+      "loss": 1.9078,
+      "step": 1454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001418967587034814,
+      "loss": 1.7355,
+      "step": 1455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014185674269707885,
+      "loss": 1.6373,
+      "step": 1456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014181672669067626,
+      "loss": 2.0454,
+      "step": 1457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014177671068427372,
+      "loss": 2.1915,
+      "step": 1458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014173669467787116,
+      "loss": 2.2058,
+      "step": 1459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001416966786714686,
+      "loss": 1.8238,
+      "step": 1460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014165666266506604,
+      "loss": 1.8611,
+      "step": 1461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014161664665866347,
+      "loss": 1.8219,
+      "step": 1462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001415766306522609,
+      "loss": 2.25,
+      "step": 1463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014153661464585835,
+      "loss": 2.1238,
+      "step": 1464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014149659863945578,
+      "loss": 1.8426,
+      "step": 1465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014145658263305325,
+      "loss": 1.3432,
+      "step": 1466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014141656662665066,
+      "loss": 1.7155,
+      "step": 1467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001413765506202481,
+      "loss": 1.7849,
+      "step": 1468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014133653461384556,
+      "loss": 1.7514,
+      "step": 1469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014129651860744297,
+      "loss": 1.5315,
+      "step": 1470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001412565026010404,
+      "loss": 1.7538,
+      "step": 1471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014121648659463787,
+      "loss": 2.2583,
+      "step": 1472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001411764705882353,
+      "loss": 2.2362,
+      "step": 1473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014113645458183275,
+      "loss": 1.4451,
+      "step": 1474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014109643857543018,
+      "loss": 2.0375,
+      "step": 1475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014105642256902762,
+      "loss": 2.3549,
+      "step": 1476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014101640656262506,
+      "loss": 1.6238,
+      "step": 1477
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001409763905562225,
+      "loss": 1.7827,
+      "step": 1478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014093637454981993,
+      "loss": 1.6365,
+      "step": 1479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001408963585434174,
+      "loss": 2.3618,
+      "step": 1480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001408563425370148,
+      "loss": 1.8572,
+      "step": 1481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014081632653061224,
+      "loss": 1.8487,
+      "step": 1482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001407763105242097,
+      "loss": 1.5687,
+      "step": 1483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014073629451780712,
+      "loss": 1.8008,
+      "step": 1484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014069627851140455,
+      "loss": 2.4383,
+      "step": 1485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014065626250500202,
+      "loss": 1.8141,
+      "step": 1486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014061624649859943,
+      "loss": 2.2528,
+      "step": 1487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001405762304921969,
+      "loss": 2.0041,
+      "step": 1488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014053621448579433,
+      "loss": 1.7777,
+      "step": 1489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014049619847939177,
+      "loss": 1.8212,
+      "step": 1490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001404561824729892,
+      "loss": 1.9602,
+      "step": 1491
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014041616646658664,
+      "loss": 1.8572,
+      "step": 1492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014037615046018408,
+      "loss": 1.9463,
+      "step": 1493
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014033613445378152,
+      "loss": 2.0446,
+      "step": 1494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014029611844737895,
+      "loss": 2.0455,
+      "step": 1495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001402561024409764,
+      "loss": 2.1461,
+      "step": 1496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014021608643457385,
+      "loss": 2.4657,
+      "step": 1497
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014017607042817126,
+      "loss": 2.023,
+      "step": 1498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001401360544217687,
+      "loss": 1.5481,
+      "step": 1499
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014009603841536617,
+      "loss": 1.8736,
+      "step": 1500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014005602240896358,
+      "loss": 2.0798,
+      "step": 1501
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014001600640256104,
+      "loss": 1.7967,
+      "step": 1502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013997599039615848,
+      "loss": 2.2305,
+      "step": 1503
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013993597438975591,
+      "loss": 1.7471,
+      "step": 1504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013989595838335335,
+      "loss": 2.029,
+      "step": 1505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001398559423769508,
+      "loss": 1.9588,
+      "step": 1506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013981592637054823,
+      "loss": 1.3196,
+      "step": 1507
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013977591036414566,
+      "loss": 1.9949,
+      "step": 1508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001397358943577431,
+      "loss": 1.8536,
+      "step": 1509
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013969587835134054,
+      "loss": 1.8196,
+      "step": 1510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013965586234493797,
+      "loss": 1.503,
+      "step": 1511
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001396158463385354,
+      "loss": 1.9806,
+      "step": 1512
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013957583033213285,
+      "loss": 1.8294,
+      "step": 1513
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001395358143257303,
+      "loss": 1.951,
+      "step": 1514
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013949579831932772,
+      "loss": 1.7803,
+      "step": 1515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001394557823129252,
+      "loss": 2.1775,
+      "step": 1516
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013941576630652262,
+      "loss": 1.9567,
+      "step": 1517
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013937575030012003,
+      "loss": 1.1757,
+      "step": 1518
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001393357342937175,
+      "loss": 1.7033,
+      "step": 1519
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013929571828731494,
+      "loss": 2.1211,
+      "step": 1520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013925570228091237,
+      "loss": 2.1163,
+      "step": 1521
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001392156862745098,
+      "loss": 2.0611,
+      "step": 1522
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013917567026810725,
+      "loss": 1.3384,
+      "step": 1523
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013913565426170468,
+      "loss": 2.1355,
+      "step": 1524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013909563825530212,
+      "loss": 2.1572,
+      "step": 1525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013905562224889956,
+      "loss": 1.8021,
+      "step": 1526
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013901560624249702,
+      "loss": 1.7117,
+      "step": 1527
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013897559023609446,
+      "loss": 2.2034,
+      "step": 1528
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013893557422969187,
+      "loss": 1.4307,
+      "step": 1529
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013889555822328933,
+      "loss": 1.9232,
+      "step": 1530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013885554221688677,
+      "loss": 1.9619,
+      "step": 1531
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013881552621048418,
+      "loss": 1.7093,
+      "step": 1532
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013877551020408165,
+      "loss": 1.694,
+      "step": 1533
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013873549419767908,
+      "loss": 1.7303,
+      "step": 1534
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013869547819127652,
+      "loss": 1.8084,
+      "step": 1535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013865546218487396,
+      "loss": 2.125,
+      "step": 1536
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001386154461784714,
+      "loss": 2.0088,
+      "step": 1537
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013857543017206883,
+      "loss": 1.7849,
+      "step": 1538
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013853541416566627,
+      "loss": 2.2927,
+      "step": 1539
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001384953981592637,
+      "loss": 1.9431,
+      "step": 1540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013845538215286117,
+      "loss": 1.8153,
+      "step": 1541
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013841536614645858,
+      "loss": 2.3072,
+      "step": 1542
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013837535014005602,
+      "loss": 2.1416,
+      "step": 1543
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013833533413365348,
+      "loss": 1.4658,
+      "step": 1544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013829531812725092,
+      "loss": 1.5573,
+      "step": 1545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013825530212084833,
+      "loss": 1.7474,
+      "step": 1546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001382152861144458,
+      "loss": 1.4955,
+      "step": 1547
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013817527010804323,
+      "loss": 2.0865,
+      "step": 1548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013813525410164067,
+      "loss": 2.0571,
+      "step": 1549
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001380952380952381,
+      "loss": 1.603,
+      "step": 1550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013805522208883554,
+      "loss": 1.3517,
+      "step": 1551
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013801520608243298,
+      "loss": 1.8235,
+      "step": 1552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013797519007603042,
+      "loss": 1.9627,
+      "step": 1553
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013793517406962785,
+      "loss": 2.1106,
+      "step": 1554
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013789515806322532,
+      "loss": 1.8885,
+      "step": 1555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013785514205682273,
+      "loss": 2.05,
+      "step": 1556
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013781512605042016,
+      "loss": 2.2954,
+      "step": 1557
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013777511004401763,
+      "loss": 1.4741,
+      "step": 1558
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013773509403761504,
+      "loss": 2.075,
+      "step": 1559
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013769507803121248,
+      "loss": 2.2259,
+      "step": 1560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013765506202480994,
+      "loss": 2.0873,
+      "step": 1561
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013761504601840738,
+      "loss": 2.4355,
+      "step": 1562
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013757503001200481,
+      "loss": 1.7509,
+      "step": 1563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013753501400560225,
+      "loss": 2.1502,
+      "step": 1564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001374949979991997,
+      "loss": 1.8897,
+      "step": 1565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013745498199279713,
+      "loss": 1.5191,
+      "step": 1566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013741496598639456,
+      "loss": 1.7935,
+      "step": 1567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000137374949979992,
+      "loss": 1.8658,
+      "step": 1568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013733493397358946,
+      "loss": 2.1229,
+      "step": 1569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013729491796718687,
+      "loss": 2.4608,
+      "step": 1570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001372549019607843,
+      "loss": 1.7678,
+      "step": 1571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013721488595438177,
+      "loss": 1.9284,
+      "step": 1572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013717486994797919,
+      "loss": 1.8885,
+      "step": 1573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013713485394157662,
+      "loss": 1.8525,
+      "step": 1574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013709483793517409,
+      "loss": 1.6791,
+      "step": 1575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013705482192877152,
+      "loss": 2.415,
+      "step": 1576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013701480592236896,
+      "loss": 1.7726,
+      "step": 1577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001369747899159664,
+      "loss": 1.7324,
+      "step": 1578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013693477390956383,
+      "loss": 1.6851,
+      "step": 1579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013689475790316127,
+      "loss": 2.1338,
+      "step": 1580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001368547418967587,
+      "loss": 2.1634,
+      "step": 1581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013681472589035615,
+      "loss": 2.0687,
+      "step": 1582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013677470988395358,
+      "loss": 1.7136,
+      "step": 1583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013673469387755102,
+      "loss": 1.8253,
+      "step": 1584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013669467787114846,
+      "loss": 1.9996,
+      "step": 1585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013665466186474592,
+      "loss": 2.1825,
+      "step": 1586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013661464585834333,
+      "loss": 1.6363,
+      "step": 1587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001365746298519408,
+      "loss": 1.9259,
+      "step": 1588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013653461384553823,
+      "loss": 2.1229,
+      "step": 1589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013649459783913564,
+      "loss": 2.351,
+      "step": 1590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001364545818327331,
+      "loss": 2.2213,
+      "step": 1591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013641456582633054,
+      "loss": 1.9403,
+      "step": 1592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013637454981992798,
+      "loss": 1.3687,
+      "step": 1593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013633453381352542,
+      "loss": 2.2134,
+      "step": 1594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013629451780712286,
+      "loss": 2.0951,
+      "step": 1595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001362545018007203,
+      "loss": 1.973,
+      "step": 1596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013621448579431773,
+      "loss": 2.5736,
+      "step": 1597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013617446978791517,
+      "loss": 1.9623,
+      "step": 1598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001361344537815126,
+      "loss": 1.8496,
+      "step": 1599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013609443777511004,
+      "loss": 1.9324,
+      "step": 1600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013605442176870748,
+      "loss": 2.041,
+      "step": 1601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013601440576230494,
+      "loss": 1.6335,
+      "step": 1602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013597438975590238,
+      "loss": 2.1328,
+      "step": 1603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001359343737494998,
+      "loss": 1.3424,
+      "step": 1604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013589435774309725,
+      "loss": 1.5565,
+      "step": 1605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001358543417366947,
+      "loss": 2.0967,
+      "step": 1606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001358143257302921,
+      "loss": 2.0819,
+      "step": 1607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013577430972388957,
+      "loss": 1.8345,
+      "step": 1608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000135734293717487,
+      "loss": 1.2854,
+      "step": 1609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013569427771108444,
+      "loss": 1.5486,
+      "step": 1610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013565426170468188,
+      "loss": 2.1048,
+      "step": 1611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013561424569827931,
+      "loss": 2.029,
+      "step": 1612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013557422969187675,
+      "loss": 1.9164,
+      "step": 1613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001355342136854742,
+      "loss": 1.6841,
+      "step": 1614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013549419767907163,
+      "loss": 2.2076,
+      "step": 1615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001354541816726691,
+      "loss": 1.6523,
+      "step": 1616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013541416566626653,
+      "loss": 1.532,
+      "step": 1617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013537414965986394,
+      "loss": 2.1847,
+      "step": 1618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001353341336534614,
+      "loss": 2.1739,
+      "step": 1619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013529411764705884,
+      "loss": 2.1395,
+      "step": 1620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013525410164065625,
+      "loss": 1.7463,
+      "step": 1621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001352140856342537,
+      "loss": 1.9932,
+      "step": 1622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013517406962785115,
+      "loss": 1.2051,
+      "step": 1623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001351340536214486,
+      "loss": 1.7013,
+      "step": 1624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013509403761504602,
+      "loss": 1.828,
+      "step": 1625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013505402160864346,
+      "loss": 1.4327,
+      "step": 1626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001350140056022409,
+      "loss": 1.79,
+      "step": 1627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013497398959583834,
+      "loss": 2.0851,
+      "step": 1628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013493397358943577,
+      "loss": 1.5776,
+      "step": 1629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013489395758303324,
+      "loss": 1.436,
+      "step": 1630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013485394157663065,
+      "loss": 1.5177,
+      "step": 1631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013481392557022808,
+      "loss": 1.536,
+      "step": 1632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013477390956382555,
+      "loss": 2.1677,
+      "step": 1633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013473389355742299,
+      "loss": 1.9349,
+      "step": 1634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001346938775510204,
+      "loss": 1.9274,
+      "step": 1635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013465386154461786,
+      "loss": 1.7325,
+      "step": 1636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001346138455382153,
+      "loss": 2.5504,
+      "step": 1637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013457382953181273,
+      "loss": 2.0281,
+      "step": 1638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013453381352541017,
+      "loss": 1.3848,
+      "step": 1639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001344937975190076,
+      "loss": 2.2891,
+      "step": 1640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013445378151260507,
+      "loss": 1.6153,
+      "step": 1641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013441376550620248,
+      "loss": 1.4396,
+      "step": 1642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013437374949979992,
+      "loss": 2.1114,
+      "step": 1643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013433373349339738,
+      "loss": 2.0611,
+      "step": 1644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001342937174869948,
+      "loss": 1.337,
+      "step": 1645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013425370148059223,
+      "loss": 1.9889,
+      "step": 1646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001342136854741897,
+      "loss": 1.9688,
+      "step": 1647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001341736694677871,
+      "loss": 1.3915,
+      "step": 1648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013413365346138454,
+      "loss": 1.9025,
+      "step": 1649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000134093637454982,
+      "loss": 1.8695,
+      "step": 1650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013405362144857944,
+      "loss": 1.8602,
+      "step": 1651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013401360544217688,
+      "loss": 1.6331,
+      "step": 1652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013397358943577432,
+      "loss": 1.9847,
+      "step": 1653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013393357342937176,
+      "loss": 2.3557,
+      "step": 1654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001338935574229692,
+      "loss": 1.8615,
+      "step": 1655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013385354141656663,
+      "loss": 2.0455,
+      "step": 1656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013381352541016407,
+      "loss": 2.1609,
+      "step": 1657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013377350940376153,
+      "loss": 2.0637,
+      "step": 1658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013373349339735894,
+      "loss": 1.8841,
+      "step": 1659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013369347739095638,
+      "loss": 2.0223,
+      "step": 1660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013365346138455384,
+      "loss": 1.5098,
+      "step": 1661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013361344537815125,
+      "loss": 1.5333,
+      "step": 1662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013357342937174872,
+      "loss": 1.9057,
+      "step": 1663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013353341336534615,
+      "loss": 1.5407,
+      "step": 1664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001334933973589436,
+      "loss": 2.0378,
+      "step": 1665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013345338135254103,
+      "loss": 1.8648,
+      "step": 1666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013341336534613847,
+      "loss": 1.906,
+      "step": 1667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001333733493397359,
+      "loss": 1.9564,
+      "step": 1668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 1.9291,
+      "step": 1669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013329331732693078,
+      "loss": 1.8403,
+      "step": 1670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013325330132052821,
+      "loss": 1.5124,
+      "step": 1671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013321328531412565,
+      "loss": 1.609,
+      "step": 1672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001331732693077231,
+      "loss": 1.8327,
+      "step": 1673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013313325330132053,
+      "loss": 1.9909,
+      "step": 1674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000133093237294918,
+      "loss": 1.837,
+      "step": 1675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001330532212885154,
+      "loss": 1.6186,
+      "step": 1676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013301320528211286,
+      "loss": 1.6867,
+      "step": 1677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001329731892757103,
+      "loss": 2.006,
+      "step": 1678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001329331732693077,
+      "loss": 2.6215,
+      "step": 1679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013289315726290518,
+      "loss": 1.7763,
+      "step": 1680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001328531412565026,
+      "loss": 2.4411,
+      "step": 1681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013281312525010005,
+      "loss": 2.6645,
+      "step": 1682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001327731092436975,
+      "loss": 1.4048,
+      "step": 1683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013273309323729492,
+      "loss": 1.9693,
+      "step": 1684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013269307723089236,
+      "loss": 2.1952,
+      "step": 1685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001326530612244898,
+      "loss": 1.8837,
+      "step": 1686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013261304521808724,
+      "loss": 2.0301,
+      "step": 1687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013257302921168467,
+      "loss": 1.8329,
+      "step": 1688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001325330132052821,
+      "loss": 1.4327,
+      "step": 1689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013249299719887955,
+      "loss": 2.21,
+      "step": 1690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000132452981192477,
+      "loss": 1.9678,
+      "step": 1691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013241296518607445,
+      "loss": 2.0138,
+      "step": 1692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013237294917967186,
+      "loss": 2.0788,
+      "step": 1693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013233293317326932,
+      "loss": 2.2833,
+      "step": 1694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013229291716686676,
+      "loss": 1.8406,
+      "step": 1695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013225290116046417,
+      "loss": 1.7177,
+      "step": 1696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013221288515406163,
+      "loss": 1.4432,
+      "step": 1697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013217286914765907,
+      "loss": 1.4693,
+      "step": 1698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001321328531412565,
+      "loss": 1.7025,
+      "step": 1699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013209283713485395,
+      "loss": 1.9865,
+      "step": 1700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013205282112845138,
+      "loss": 1.7609,
+      "step": 1701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013201280512204882,
+      "loss": 2.3739,
+      "step": 1702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013197278911564626,
+      "loss": 1.7654,
+      "step": 1703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001319327731092437,
+      "loss": 1.6877,
+      "step": 1704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013189275710284116,
+      "loss": 2.0413,
+      "step": 1705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001318527410964386,
+      "loss": 1.963,
+      "step": 1706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000131812725090036,
+      "loss": 1.5208,
+      "step": 1707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013177270908363347,
+      "loss": 1.6618,
+      "step": 1708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001317326930772309,
+      "loss": 1.9688,
+      "step": 1709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013169267707082832,
+      "loss": 1.6628,
+      "step": 1710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013165266106442578,
+      "loss": 1.8556,
+      "step": 1711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013161264505802322,
+      "loss": 1.7045,
+      "step": 1712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013157262905162065,
+      "loss": 2.1702,
+      "step": 1713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001315326130452181,
+      "loss": 1.8433,
+      "step": 1714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013149259703881553,
+      "loss": 1.9324,
+      "step": 1715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000131452581032413,
+      "loss": 2.3993,
+      "step": 1716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001314125650260104,
+      "loss": 1.6295,
+      "step": 1717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013137254901960784,
+      "loss": 2.0337,
+      "step": 1718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001313325330132053,
+      "loss": 1.9974,
+      "step": 1719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013129251700680271,
+      "loss": 2.3368,
+      "step": 1720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013125250100040015,
+      "loss": 2.0047,
+      "step": 1721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013121248499399762,
+      "loss": 1.9561,
+      "step": 1722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013117246898759505,
+      "loss": 2.3289,
+      "step": 1723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001311324529811925,
+      "loss": 1.7426,
+      "step": 1724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013109243697478993,
+      "loss": 1.7104,
+      "step": 1725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013105242096838736,
+      "loss": 2.1476,
+      "step": 1726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001310124049619848,
+      "loss": 1.6581,
+      "step": 1727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013097238895558224,
+      "loss": 1.8294,
+      "step": 1728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013093237294917968,
+      "loss": 1.777,
+      "step": 1729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013089235694277714,
+      "loss": 2.0758,
+      "step": 1730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013085234093637455,
+      "loss": 2.1936,
+      "step": 1731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000130812324929972,
+      "loss": 1.6081,
+      "step": 1732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013077230892356945,
+      "loss": 1.3636,
+      "step": 1733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013073229291716686,
+      "loss": 2.1224,
+      "step": 1734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001306922769107643,
+      "loss": 1.4037,
+      "step": 1735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00013065226090436176,
+      "loss": 2.1007,
+      "step": 1736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00013061224489795917,
+      "loss": 2.0315,
+      "step": 1737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00013057222889155664,
+      "loss": 1.9295,
+      "step": 1738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00013053221288515407,
+      "loss": 2.2,
+      "step": 1739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001304921968787515,
+      "loss": 2.1981,
+      "step": 1740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00013045218087234895,
+      "loss": 1.7451,
+      "step": 1741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00013041216486594639,
+      "loss": 1.7753,
+      "step": 1742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00013037214885954382,
+      "loss": 1.6113,
+      "step": 1743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00013033213285314126,
+      "loss": 2.1941,
+      "step": 1744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001302921168467387,
+      "loss": 1.461,
+      "step": 1745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00013025210084033613,
+      "loss": 1.7217,
+      "step": 1746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001302120848339336,
+      "loss": 2.0995,
+      "step": 1747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000130172068827531,
+      "loss": 1.8795,
+      "step": 1748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00013013205282112845,
+      "loss": 2.1583,
+      "step": 1749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001300920368147259,
+      "loss": 2.5718,
+      "step": 1750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00013005202080832332,
+      "loss": 2.2033,
+      "step": 1751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00013001200480192078,
+      "loss": 1.6445,
+      "step": 1752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012997198879551822,
+      "loss": 1.7999,
+      "step": 1753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012993197278911566,
+      "loss": 2.1399,
+      "step": 1754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001298919567827131,
+      "loss": 1.9249,
+      "step": 1755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012985194077631053,
+      "loss": 2.0549,
+      "step": 1756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012981192476990797,
+      "loss": 2.075,
+      "step": 1757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001297719087635054,
+      "loss": 2.007,
+      "step": 1758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012973189275710284,
+      "loss": 1.9784,
+      "step": 1759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012969187675070028,
+      "loss": 2.2632,
+      "step": 1760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012965186074429772,
+      "loss": 1.4877,
+      "step": 1761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012961184473789516,
+      "loss": 2.4138,
+      "step": 1762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001295718287314926,
+      "loss": 1.5612,
+      "step": 1763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012953181272509006,
+      "loss": 1.7592,
+      "step": 1764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012949179671868747,
+      "loss": 1.6728,
+      "step": 1765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012945178071228493,
+      "loss": 1.8011,
+      "step": 1766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012941176470588237,
+      "loss": 1.6926,
+      "step": 1767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012937174869947978,
+      "loss": 2.302,
+      "step": 1768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012933173269307724,
+      "loss": 1.6942,
+      "step": 1769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012929171668667468,
+      "loss": 2.0701,
+      "step": 1770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012925170068027212,
+      "loss": 2.027,
+      "step": 1771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012921168467386955,
+      "loss": 1.3977,
+      "step": 1772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000129171668667467,
+      "loss": 1.8097,
+      "step": 1773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012913165266106443,
+      "loss": 1.7191,
+      "step": 1774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012909163665466187,
+      "loss": 1.86,
+      "step": 1775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001290516206482593,
+      "loss": 1.86,
+      "step": 1776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012901160464185677,
+      "loss": 1.6107,
+      "step": 1777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012897158863545418,
+      "loss": 1.792,
+      "step": 1778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012893157262905161,
+      "loss": 1.7756,
+      "step": 1779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012889155662264908,
+      "loss": 2.1574,
+      "step": 1780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012885154061624652,
+      "loss": 1.8453,
+      "step": 1781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012881152460984393,
+      "loss": 1.3784,
+      "step": 1782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001287715086034414,
+      "loss": 1.9784,
+      "step": 1783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012873149259703883,
+      "loss": 2.4359,
+      "step": 1784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012869147659063626,
+      "loss": 1.6884,
+      "step": 1785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001286514605842337,
+      "loss": 2.2187,
+      "step": 1786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012861144457783114,
+      "loss": 2.2314,
+      "step": 1787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012857142857142858,
+      "loss": 1.7592,
+      "step": 1788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000128531412565026,
+      "loss": 2.1083,
+      "step": 1789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012849139655862345,
+      "loss": 1.5648,
+      "step": 1790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012845138055222091,
+      "loss": 2.1909,
+      "step": 1791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012841136454581832,
+      "loss": 1.7586,
+      "step": 1792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012837134853941576,
+      "loss": 1.8024,
+      "step": 1793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012833133253301323,
+      "loss": 1.3852,
+      "step": 1794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012829131652661066,
+      "loss": 1.9165,
+      "step": 1795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012825130052020807,
+      "loss": 1.8459,
+      "step": 1796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012821128451380554,
+      "loss": 1.9258,
+      "step": 1797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012817126850740297,
+      "loss": 1.9351,
+      "step": 1798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001281312525010004,
+      "loss": 1.7108,
+      "step": 1799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012809123649459785,
+      "loss": 1.6062,
+      "step": 1800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012805122048819529,
+      "loss": 1.6581,
+      "step": 1801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012801120448179272,
+      "loss": 1.7345,
+      "step": 1802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012797118847539016,
+      "loss": 1.9916,
+      "step": 1803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001279311724689876,
+      "loss": 1.8065,
+      "step": 1804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012789115646258506,
+      "loss": 2.0638,
+      "step": 1805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012785114045618247,
+      "loss": 2.0247,
+      "step": 1806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001278111244497799,
+      "loss": 1.967,
+      "step": 1807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012777110844337737,
+      "loss": 2.4929,
+      "step": 1808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012773109243697478,
+      "loss": 1.6689,
+      "step": 1809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012769107643057222,
+      "loss": 2.2511,
+      "step": 1810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012765106042416968,
+      "loss": 2.1626,
+      "step": 1811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012761104441776712,
+      "loss": 1.7005,
+      "step": 1812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012757102841136456,
+      "loss": 1.7442,
+      "step": 1813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000127531012404962,
+      "loss": 1.5099,
+      "step": 1814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012749099639855943,
+      "loss": 2.147,
+      "step": 1815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012745098039215687,
+      "loss": 1.9633,
+      "step": 1816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001274109643857543,
+      "loss": 1.6552,
+      "step": 1817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012737094837935174,
+      "loss": 1.8942,
+      "step": 1818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001273309323729492,
+      "loss": 2.0515,
+      "step": 1819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012729091636654662,
+      "loss": 1.8941,
+      "step": 1820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012725090036014406,
+      "loss": 1.9774,
+      "step": 1821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012721088435374152,
+      "loss": 1.414,
+      "step": 1822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012717086834733893,
+      "loss": 2.3004,
+      "step": 1823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012713085234093637,
+      "loss": 2.2754,
+      "step": 1824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012709083633453383,
+      "loss": 2.1687,
+      "step": 1825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012705082032813124,
+      "loss": 1.6292,
+      "step": 1826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001270108043217287,
+      "loss": 2.5974,
+      "step": 1827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012697078831532614,
+      "loss": 1.4763,
+      "step": 1828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012693077230892358,
+      "loss": 2.3049,
+      "step": 1829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012689075630252102,
+      "loss": 2.0197,
+      "step": 1830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012685074029611845,
+      "loss": 2.2459,
+      "step": 1831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001268107242897159,
+      "loss": 1.5058,
+      "step": 1832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012677070828331333,
+      "loss": 2.5302,
+      "step": 1833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012673069227691077,
+      "loss": 1.8652,
+      "step": 1834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001266906762705082,
+      "loss": 2.0417,
+      "step": 1835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012665066026410567,
+      "loss": 1.5993,
+      "step": 1836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012661064425770308,
+      "loss": 2.1978,
+      "step": 1837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012657062825130051,
+      "loss": 1.9929,
+      "step": 1838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012653061224489798,
+      "loss": 1.809,
+      "step": 1839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001264905962384954,
+      "loss": 2.0452,
+      "step": 1840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012645058023209285,
+      "loss": 1.8755,
+      "step": 1841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001264105642256903,
+      "loss": 1.8043,
+      "step": 1842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012637054821928773,
+      "loss": 1.781,
+      "step": 1843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012633053221288516,
+      "loss": 1.9526,
+      "step": 1844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001262905162064826,
+      "loss": 1.5604,
+      "step": 1845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012625050020008004,
+      "loss": 2.0581,
+      "step": 1846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012621048419367748,
+      "loss": 1.5141,
+      "step": 1847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001261704681872749,
+      "loss": 1.5298,
+      "step": 1848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012613045218087235,
+      "loss": 1.923,
+      "step": 1849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001260904361744698,
+      "loss": 1.6438,
+      "step": 1850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012605042016806722,
+      "loss": 1.2147,
+      "step": 1851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001260104041616647,
+      "loss": 1.7534,
+      "step": 1852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012597038815526212,
+      "loss": 2.0724,
+      "step": 1853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012593037214885953,
+      "loss": 2.2168,
+      "step": 1854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000125890356142457,
+      "loss": 1.7649,
+      "step": 1855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012585034013605444,
+      "loss": 1.8638,
+      "step": 1856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012581032412965185,
+      "loss": 2.1095,
+      "step": 1857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001257703081232493,
+      "loss": 2.1245,
+      "step": 1858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012573029211684675,
+      "loss": 2.1458,
+      "step": 1859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012569027611044418,
+      "loss": 1.8581,
+      "step": 1860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012565026010404162,
+      "loss": 1.98,
+      "step": 1861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012561024409763906,
+      "loss": 2.4854,
+      "step": 1862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001255702280912365,
+      "loss": 2.0474,
+      "step": 1863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012553021208483393,
+      "loss": 1.9404,
+      "step": 1864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012549019607843137,
+      "loss": 2.2075,
+      "step": 1865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012545018007202883,
+      "loss": 2.1687,
+      "step": 1866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012541016406562624,
+      "loss": 1.7104,
+      "step": 1867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012537014805922368,
+      "loss": 1.9561,
+      "step": 1868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012533013205282115,
+      "loss": 1.5445,
+      "step": 1869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012529011604641858,
+      "loss": 1.7725,
+      "step": 1870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000125250100040016,
+      "loss": 1.456,
+      "step": 1871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012521008403361346,
+      "loss": 1.9304,
+      "step": 1872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001251700680272109,
+      "loss": 1.8904,
+      "step": 1873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012513005202080833,
+      "loss": 1.9405,
+      "step": 1874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012509003601440577,
+      "loss": 2.1669,
+      "step": 1875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001250500200080032,
+      "loss": 1.7821,
+      "step": 1876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012501000400160064,
+      "loss": 2.1297,
+      "step": 1877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012496998799519808,
+      "loss": 2.1846,
+      "step": 1878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012492997198879552,
+      "loss": 1.9393,
+      "step": 1879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012488995598239298,
+      "loss": 1.709,
+      "step": 1880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001248499399759904,
+      "loss": 1.7076,
+      "step": 1881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012480992396958783,
+      "loss": 1.5057,
+      "step": 1882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001247699079631853,
+      "loss": 1.6282,
+      "step": 1883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012472989195678273,
+      "loss": 1.5969,
+      "step": 1884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012468987595038014,
+      "loss": 1.6356,
+      "step": 1885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001246498599439776,
+      "loss": 1.7007,
+      "step": 1886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012460984393757504,
+      "loss": 1.5716,
+      "step": 1887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012456982793117248,
+      "loss": 1.7706,
+      "step": 1888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012452981192476992,
+      "loss": 1.9526,
+      "step": 1889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012448979591836735,
+      "loss": 1.6499,
+      "step": 1890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001244497799119648,
+      "loss": 1.8883,
+      "step": 1891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012440976390556223,
+      "loss": 1.888,
+      "step": 1892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012436974789915966,
+      "loss": 1.5263,
+      "step": 1893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012432973189275713,
+      "loss": 2.4382,
+      "step": 1894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012428971588635454,
+      "loss": 1.4515,
+      "step": 1895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012424969987995198,
+      "loss": 1.7869,
+      "step": 1896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012420968387354944,
+      "loss": 1.4683,
+      "step": 1897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012416966786714685,
+      "loss": 2.3747,
+      "step": 1898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001241296518607443,
+      "loss": 2.3776,
+      "step": 1899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012408963585434175,
+      "loss": 1.8221,
+      "step": 1900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001240496198479392,
+      "loss": 1.8751,
+      "step": 1901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012400960384153663,
+      "loss": 1.8772,
+      "step": 1902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012396958783513406,
+      "loss": 1.5574,
+      "step": 1903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001239295718287315,
+      "loss": 1.969,
+      "step": 1904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012388955582232894,
+      "loss": 1.6423,
+      "step": 1905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012384953981592637,
+      "loss": 1.9576,
+      "step": 1906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001238095238095238,
+      "loss": 1.8372,
+      "step": 1907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012376950780312128,
+      "loss": 2.1459,
+      "step": 1908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012372949179671869,
+      "loss": 2.0303,
+      "step": 1909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012368947579031612,
+      "loss": 1.5364,
+      "step": 1910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001236494597839136,
+      "loss": 2.3826,
+      "step": 1911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000123609443777511,
+      "loss": 1.8207,
+      "step": 1912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012356942777110846,
+      "loss": 2.7555,
+      "step": 1913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001235294117647059,
+      "loss": 1.6428,
+      "step": 1914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001234893957583033,
+      "loss": 1.5831,
+      "step": 1915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012344937975190077,
+      "loss": 2.1494,
+      "step": 1916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001234093637454982,
+      "loss": 2.0872,
+      "step": 1917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012336934773909565,
+      "loss": 2.0126,
+      "step": 1918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012332933173269308,
+      "loss": 1.9702,
+      "step": 1919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012328931572629052,
+      "loss": 1.9575,
+      "step": 1920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012324929971988796,
+      "loss": 2.0592,
+      "step": 1921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001232092837134854,
+      "loss": 2.4887,
+      "step": 1922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012316926770708283,
+      "loss": 2.0575,
+      "step": 1923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012312925170068027,
+      "loss": 1.5741,
+      "step": 1924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012308923569427773,
+      "loss": 2.137,
+      "step": 1925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012304921968787514,
+      "loss": 1.3879,
+      "step": 1926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001230092036814726,
+      "loss": 2.562,
+      "step": 1927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012296918767507005,
+      "loss": 2.0123,
+      "step": 1928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012292917166866746,
+      "loss": 1.7264,
+      "step": 1929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012288915566226492,
+      "loss": 1.6388,
+      "step": 1930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012284913965586236,
+      "loss": 1.9497,
+      "step": 1931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001228091236494598,
+      "loss": 1.5481,
+      "step": 1932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012276910764305723,
+      "loss": 1.9489,
+      "step": 1933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012272909163665467,
+      "loss": 1.9907,
+      "step": 1934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001226890756302521,
+      "loss": 1.682,
+      "step": 1935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012264905962384954,
+      "loss": 1.8823,
+      "step": 1936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012260904361744698,
+      "loss": 0.9197,
+      "step": 1937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012256902761104442,
+      "loss": 2.194,
+      "step": 1938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012252901160464185,
+      "loss": 1.7798,
+      "step": 1939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001224889955982393,
+      "loss": 1.7011,
+      "step": 1940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012244897959183676,
+      "loss": 1.9785,
+      "step": 1941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001224089635854342,
+      "loss": 1.4564,
+      "step": 1942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001223689475790316,
+      "loss": 1.7047,
+      "step": 1943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012232893157262907,
+      "loss": 1.7969,
+      "step": 1944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001222889155662265,
+      "loss": 1.9109,
+      "step": 1945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012224889955982391,
+      "loss": 1.8789,
+      "step": 1946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012220888355342138,
+      "loss": 1.8497,
+      "step": 1947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012216886754701882,
+      "loss": 2.8015,
+      "step": 1948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012212885154061625,
+      "loss": 1.9378,
+      "step": 1949
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001220888355342137,
+      "loss": 1.6779,
+      "step": 1950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012204881952781113,
+      "loss": 2.0488,
+      "step": 1951
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012200880352140858,
+      "loss": 2.1759,
+      "step": 1952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000121968787515006,
+      "loss": 1.5606,
+      "step": 1953
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012192877150860345,
+      "loss": 1.7632,
+      "step": 1954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012188875550220089,
+      "loss": 1.5298,
+      "step": 1955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012184873949579831,
+      "loss": 1.2768,
+      "step": 1956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012180872348939576,
+      "loss": 2.029,
+      "step": 1957
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001217687074829932,
+      "loss": 1.975,
+      "step": 1958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012172869147659065,
+      "loss": 1.8552,
+      "step": 1959
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012168867547018807,
+      "loss": 2.081,
+      "step": 1960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012164865946378553,
+      "loss": 1.6892,
+      "step": 1961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012160864345738296,
+      "loss": 1.6321,
+      "step": 1962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012156862745098039,
+      "loss": 2.4362,
+      "step": 1963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012152861144457784,
+      "loss": 1.0413,
+      "step": 1964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012148859543817527,
+      "loss": 1.5144,
+      "step": 1965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012144857943177272,
+      "loss": 1.6309,
+      "step": 1966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012140856342537015,
+      "loss": 1.6267,
+      "step": 1967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001213685474189676,
+      "loss": 1.6079,
+      "step": 1968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012132853141256504,
+      "loss": 2.2729,
+      "step": 1969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012128851540616246,
+      "loss": 1.9696,
+      "step": 1970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012124849939975991,
+      "loss": 1.7824,
+      "step": 1971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012120848339335735,
+      "loss": 1.8237,
+      "step": 1972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001211684673869548,
+      "loss": 1.7748,
+      "step": 1973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012112845138055222,
+      "loss": 1.6564,
+      "step": 1974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012108843537414967,
+      "loss": 2.2301,
+      "step": 1975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012104841936774711,
+      "loss": 1.7661,
+      "step": 1976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012100840336134453,
+      "loss": 1.6745,
+      "step": 1977
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012096838735494198,
+      "loss": 1.6239,
+      "step": 1978
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012092837134853942,
+      "loss": 1.9622,
+      "step": 1979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012088835534213684,
+      "loss": 1.676,
+      "step": 1980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001208483393357343,
+      "loss": 2.1565,
+      "step": 1981
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012080832332933175,
+      "loss": 1.3479,
+      "step": 1982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012076830732292918,
+      "loss": 1.9049,
+      "step": 1983
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001207282913165266,
+      "loss": 1.5999,
+      "step": 1984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012068827531012406,
+      "loss": 1.986,
+      "step": 1985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001206482593037215,
+      "loss": 2.0039,
+      "step": 1986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012060824329731892,
+      "loss": 2.0485,
+      "step": 1987
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012056822729091637,
+      "loss": 1.8506,
+      "step": 1988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012052821128451382,
+      "loss": 1.8355,
+      "step": 1989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012048819527811126,
+      "loss": 1.6987,
+      "step": 1990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012044817927170868,
+      "loss": 1.5675,
+      "step": 1991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012040816326530613,
+      "loss": 1.9963,
+      "step": 1992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012036814725890357,
+      "loss": 1.745,
+      "step": 1993
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012032813125250099,
+      "loss": 2.1192,
+      "step": 1994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012028811524609844,
+      "loss": 2.1394,
+      "step": 1995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012024809923969589,
+      "loss": 2.1739,
+      "step": 1996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012020808323329333,
+      "loss": 1.4584,
+      "step": 1997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012016806722689075,
+      "loss": 1.898,
+      "step": 1998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001201280512204882,
+      "loss": 1.851,
+      "step": 1999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012008803521408564,
+      "loss": 1.8405,
+      "step": 2000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012004801920768306,
+      "loss": 1.9335,
+      "step": 2001
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012000800320128052,
+      "loss": 1.9568,
+      "step": 2002
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011996798719487797,
+      "loss": 1.4169,
+      "step": 2003
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011992797118847539,
+      "loss": 1.6378,
+      "step": 2004
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011988795518207283,
+      "loss": 1.3804,
+      "step": 2005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011984793917567028,
+      "loss": 2.0579,
+      "step": 2006
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011980792316926773,
+      "loss": 2.2699,
+      "step": 2007
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011976790716286514,
+      "loss": 1.6792,
+      "step": 2008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011972789115646259,
+      "loss": 1.628,
+      "step": 2009
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011968787515006004,
+      "loss": 2.1021,
+      "step": 2010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011964785914365746,
+      "loss": 1.8097,
+      "step": 2011
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001196078431372549,
+      "loss": 1.2916,
+      "step": 2012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011956782713085235,
+      "loss": 1.6795,
+      "step": 2013
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001195278111244498,
+      "loss": 2.0331,
+      "step": 2014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011948779511804723,
+      "loss": 1.5465,
+      "step": 2015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011944777911164466,
+      "loss": 1.6955,
+      "step": 2016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011940776310524211,
+      "loss": 1.8151,
+      "step": 2017
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011936774709883954,
+      "loss": 1.7765,
+      "step": 2018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011932773109243697,
+      "loss": 2.0242,
+      "step": 2019
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011928771508603442,
+      "loss": 1.5501,
+      "step": 2020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011924769907963188,
+      "loss": 1.797,
+      "step": 2021
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001192076830732293,
+      "loss": 1.4292,
+      "step": 2022
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011916766706682674,
+      "loss": 1.9594,
+      "step": 2023
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011912765106042419,
+      "loss": 1.7432,
+      "step": 2024
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011908763505402161,
+      "loss": 2.1049,
+      "step": 2025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011904761904761905,
+      "loss": 1.9203,
+      "step": 2026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001190076030412165,
+      "loss": 1.9476,
+      "step": 2027
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011896758703481392,
+      "loss": 1.5948,
+      "step": 2028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011892757102841137,
+      "loss": 2.0765,
+      "step": 2029
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011888755502200881,
+      "loss": 1.7969,
+      "step": 2030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011884753901560626,
+      "loss": 2.0294,
+      "step": 2031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011880752300920368,
+      "loss": 1.6506,
+      "step": 2032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011876750700280112,
+      "loss": 2.0623,
+      "step": 2033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011872749099639857,
+      "loss": 1.816,
+      "step": 2034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000118687474989996,
+      "loss": 1.6106,
+      "step": 2035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011864745898359345,
+      "loss": 2.1748,
+      "step": 2036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011860744297719088,
+      "loss": 2.0564,
+      "step": 2037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011856742697078833,
+      "loss": 2.3301,
+      "step": 2038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011852741096438576,
+      "loss": 1.8445,
+      "step": 2039
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001184873949579832,
+      "loss": 1.6568,
+      "step": 2040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011844737895158065,
+      "loss": 1.7463,
+      "step": 2041
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011840736294517807,
+      "loss": 1.834,
+      "step": 2042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011836734693877552,
+      "loss": 1.3462,
+      "step": 2043
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011832733093237296,
+      "loss": 1.9251,
+      "step": 2044
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011828731492597038,
+      "loss": 1.9508,
+      "step": 2045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011824729891956783,
+      "loss": 1.645,
+      "step": 2046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011820728291316527,
+      "loss": 1.6542,
+      "step": 2047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011816726690676272,
+      "loss": 1.9056,
+      "step": 2048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011812725090036014,
+      "loss": 1.7155,
+      "step": 2049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011808723489395759,
+      "loss": 1.719,
+      "step": 2050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011804721888755503,
+      "loss": 1.9579,
+      "step": 2051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011800720288115245,
+      "loss": 2.2682,
+      "step": 2052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001179671868747499,
+      "loss": 1.761,
+      "step": 2053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011792717086834734,
+      "loss": 1.6596,
+      "step": 2054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011788715486194479,
+      "loss": 1.6325,
+      "step": 2055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011784713885554222,
+      "loss": 1.8503,
+      "step": 2056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011780712284913967,
+      "loss": 2.0649,
+      "step": 2057
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001177671068427371,
+      "loss": 1.6645,
+      "step": 2058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011772709083633453,
+      "loss": 2.0378,
+      "step": 2059
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011768707482993198,
+      "loss": 2.3177,
+      "step": 2060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011764705882352942,
+      "loss": 2.2477,
+      "step": 2061
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011760704281712687,
+      "loss": 2.0513,
+      "step": 2062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011756702681072429,
+      "loss": 1.8953,
+      "step": 2063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011752701080432174,
+      "loss": 1.5404,
+      "step": 2064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011748699479791918,
+      "loss": 1.6575,
+      "step": 2065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001174469787915166,
+      "loss": 1.9429,
+      "step": 2066
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011740696278511405,
+      "loss": 2.1667,
+      "step": 2067
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011736694677871149,
+      "loss": 1.628,
+      "step": 2068
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011732693077230891,
+      "loss": 2.2831,
+      "step": 2069
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011728691476590636,
+      "loss": 1.9324,
+      "step": 2070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011724689875950381,
+      "loss": 1.942,
+      "step": 2071
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011720688275310125,
+      "loss": 2.2096,
+      "step": 2072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011716686674669867,
+      "loss": 2.1445,
+      "step": 2073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011712685074029612,
+      "loss": 1.6924,
+      "step": 2074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011708683473389358,
+      "loss": 1.5108,
+      "step": 2075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011704681872749099,
+      "loss": 1.7469,
+      "step": 2076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011700680272108844,
+      "loss": 1.8327,
+      "step": 2077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011696678671468589,
+      "loss": 2.1159,
+      "step": 2078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011692677070828332,
+      "loss": 1.91,
+      "step": 2079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011688675470188075,
+      "loss": 1.8302,
+      "step": 2080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001168467386954782,
+      "loss": 1.734,
+      "step": 2081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011680672268907565,
+      "loss": 2.205,
+      "step": 2082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011676670668267307,
+      "loss": 1.5554,
+      "step": 2083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011672669067627051,
+      "loss": 2.0237,
+      "step": 2084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011668667466986796,
+      "loss": 1.3136,
+      "step": 2085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001166466586634654,
+      "loss": 1.6825,
+      "step": 2086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011660664265706282,
+      "loss": 2.0932,
+      "step": 2087
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011656662665066027,
+      "loss": 1.4683,
+      "step": 2088
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011652661064425772,
+      "loss": 1.8841,
+      "step": 2089
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011648659463785515,
+      "loss": 1.6231,
+      "step": 2090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011644657863145258,
+      "loss": 1.8908,
+      "step": 2091
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011640656262505003,
+      "loss": 1.7039,
+      "step": 2092
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011636654661864746,
+      "loss": 1.1849,
+      "step": 2093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001163265306122449,
+      "loss": 1.5389,
+      "step": 2094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011628651460584235,
+      "loss": 2.2465,
+      "step": 2095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001162464985994398,
+      "loss": 1.8087,
+      "step": 2096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011620648259303722,
+      "loss": 1.9473,
+      "step": 2097
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011616646658663466,
+      "loss": 2.1283,
+      "step": 2098
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011612645058023211,
+      "loss": 1.577,
+      "step": 2099
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011608643457382953,
+      "loss": 1.4557,
+      "step": 2100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011604641856742697,
+      "loss": 1.8232,
+      "step": 2101
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011600640256102442,
+      "loss": 2.042,
+      "step": 2102
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011596638655462187,
+      "loss": 2.1431,
+      "step": 2103
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011592637054821929,
+      "loss": 1.5305,
+      "step": 2104
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011588635454181673,
+      "loss": 1.7125,
+      "step": 2105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011584633853541418,
+      "loss": 1.9073,
+      "step": 2106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001158063225290116,
+      "loss": 1.7489,
+      "step": 2107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011576630652260904,
+      "loss": 1.7197,
+      "step": 2108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011572629051620649,
+      "loss": 1.7528,
+      "step": 2109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011568627450980394,
+      "loss": 2.284,
+      "step": 2110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011564625850340137,
+      "loss": 1.6305,
+      "step": 2111
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001156062424969988,
+      "loss": 1.5435,
+      "step": 2112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011556622649059625,
+      "loss": 1.2578,
+      "step": 2113
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011552621048419368,
+      "loss": 1.8898,
+      "step": 2114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011548619447779112,
+      "loss": 2.0411,
+      "step": 2115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011544617847138857,
+      "loss": 2.1821,
+      "step": 2116
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011540616246498599,
+      "loss": 2.0458,
+      "step": 2117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011536614645858344,
+      "loss": 2.2255,
+      "step": 2118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011532613045218088,
+      "loss": 1.8969,
+      "step": 2119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011528611444577833,
+      "loss": 2.1201,
+      "step": 2120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011524609843937575,
+      "loss": 1.9241,
+      "step": 2121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011520608243297319,
+      "loss": 2.1823,
+      "step": 2122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011516606642657064,
+      "loss": 2.0811,
+      "step": 2123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011512605042016806,
+      "loss": 2.0314,
+      "step": 2124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011508603441376551,
+      "loss": 1.6976,
+      "step": 2125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011504601840736295,
+      "loss": 2.0557,
+      "step": 2126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001150060024009604,
+      "loss": 1.9208,
+      "step": 2127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011496598639455783,
+      "loss": 1.5718,
+      "step": 2128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011492597038815526,
+      "loss": 2.0645,
+      "step": 2129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011488595438175271,
+      "loss": 1.8092,
+      "step": 2130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011484593837535014,
+      "loss": 1.4236,
+      "step": 2131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011480592236894759,
+      "loss": 1.9138,
+      "step": 2132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011476590636254502,
+      "loss": 2.0242,
+      "step": 2133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011472589035614245,
+      "loss": 1.8234,
+      "step": 2134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001146858743497399,
+      "loss": 1.8296,
+      "step": 2135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011464585834333734,
+      "loss": 1.2157,
+      "step": 2136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011460584233693479,
+      "loss": 2.1808,
+      "step": 2137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011456582633053221,
+      "loss": 2.0712,
+      "step": 2138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011452581032412966,
+      "loss": 2.4362,
+      "step": 2139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001144857943177271,
+      "loss": 1.6785,
+      "step": 2140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011444577831132452,
+      "loss": 1.4676,
+      "step": 2141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011440576230492197,
+      "loss": 1.7545,
+      "step": 2142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011436574629851942,
+      "loss": 1.5835,
+      "step": 2143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011432573029211686,
+      "loss": 2.2915,
+      "step": 2144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011428571428571428,
+      "loss": 1.594,
+      "step": 2145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011424569827931173,
+      "loss": 1.8788,
+      "step": 2146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011420568227290917,
+      "loss": 1.7745,
+      "step": 2147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001141656662665066,
+      "loss": 2.0378,
+      "step": 2148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011412565026010405,
+      "loss": 1.5914,
+      "step": 2149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001140856342537015,
+      "loss": 1.6195,
+      "step": 2150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011404561824729893,
+      "loss": 1.5147,
+      "step": 2151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011400560224089636,
+      "loss": 2.0766,
+      "step": 2152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011396558623449381,
+      "loss": 1.5472,
+      "step": 2153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011392557022809124,
+      "loss": 1.9733,
+      "step": 2154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011388555422168867,
+      "loss": 1.7838,
+      "step": 2155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011384553821528612,
+      "loss": 1.8381,
+      "step": 2156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011380552220888357,
+      "loss": 2.6461,
+      "step": 2157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000113765506202481,
+      "loss": 1.676,
+      "step": 2158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011372549019607843,
+      "loss": 1.6858,
+      "step": 2159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011368547418967588,
+      "loss": 1.4774,
+      "step": 2160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011364545818327332,
+      "loss": 1.601,
+      "step": 2161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011360544217687074,
+      "loss": 2.1005,
+      "step": 2162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011356542617046819,
+      "loss": 1.6512,
+      "step": 2163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011352541016406564,
+      "loss": 2.1798,
+      "step": 2164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011348539415766307,
+      "loss": 1.8329,
+      "step": 2165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001134453781512605,
+      "loss": 2.3353,
+      "step": 2166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011340536214485795,
+      "loss": 1.5734,
+      "step": 2167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011336534613845539,
+      "loss": 1.9677,
+      "step": 2168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011332533013205282,
+      "loss": 1.6793,
+      "step": 2169
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011328531412565027,
+      "loss": 1.486,
+      "step": 2170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011324529811924772,
+      "loss": 2.0895,
+      "step": 2171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011320528211284514,
+      "loss": 1.7429,
+      "step": 2172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011316526610644258,
+      "loss": 1.6853,
+      "step": 2173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011312525010004003,
+      "loss": 1.8146,
+      "step": 2174
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011308523409363747,
+      "loss": 1.8145,
+      "step": 2175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011304521808723489,
+      "loss": 1.7807,
+      "step": 2176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011300520208083234,
+      "loss": 1.435,
+      "step": 2177
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011296518607442979,
+      "loss": 1.8428,
+      "step": 2178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011292517006802721,
+      "loss": 1.7136,
+      "step": 2179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011288515406162465,
+      "loss": 1.4782,
+      "step": 2180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001128451380552221,
+      "loss": 2.2169,
+      "step": 2181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011280512204881953,
+      "loss": 1.8395,
+      "step": 2182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011276510604241696,
+      "loss": 1.3273,
+      "step": 2183
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011272509003601441,
+      "loss": 2.2587,
+      "step": 2184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011268507402961186,
+      "loss": 1.7152,
+      "step": 2185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011264505802320929,
+      "loss": 2.123,
+      "step": 2186
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011260504201680672,
+      "loss": 1.6417,
+      "step": 2187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011256502601040418,
+      "loss": 1.7051,
+      "step": 2188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001125250100040016,
+      "loss": 1.5854,
+      "step": 2189
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011248499399759904,
+      "loss": 1.5869,
+      "step": 2190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011244497799119649,
+      "loss": 1.6554,
+      "step": 2191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011240496198479394,
+      "loss": 1.8768,
+      "step": 2192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011236494597839136,
+      "loss": 1.6361,
+      "step": 2193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001123249299719888,
+      "loss": 1.7982,
+      "step": 2194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011228491396558625,
+      "loss": 1.8328,
+      "step": 2195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011224489795918367,
+      "loss": 1.2352,
+      "step": 2196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011220488195278111,
+      "loss": 2.02,
+      "step": 2197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011216486594637856,
+      "loss": 1.9027,
+      "step": 2198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011212484993997601,
+      "loss": 1.5167,
+      "step": 2199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011208483393357343,
+      "loss": 1.8143,
+      "step": 2200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011204481792717087,
+      "loss": 1.3065,
+      "step": 2201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011200480192076832,
+      "loss": 1.3241,
+      "step": 2202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011196478591436575,
+      "loss": 1.7362,
+      "step": 2203
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011192476990796318,
+      "loss": 1.818,
+      "step": 2204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011188475390156063,
+      "loss": 2.0427,
+      "step": 2205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011184473789515806,
+      "loss": 1.9016,
+      "step": 2206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011180472188875551,
+      "loss": 2.0367,
+      "step": 2207
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011176470588235294,
+      "loss": 1.3129,
+      "step": 2208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001117246898759504,
+      "loss": 1.7732,
+      "step": 2209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011168467386954782,
+      "loss": 1.8483,
+      "step": 2210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011164465786314527,
+      "loss": 1.6766,
+      "step": 2211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011160464185674271,
+      "loss": 2.0566,
+      "step": 2212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011156462585034013,
+      "loss": 1.4421,
+      "step": 2213
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011152460984393758,
+      "loss": 2.3092,
+      "step": 2214
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011148459383753502,
+      "loss": 1.9655,
+      "step": 2215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011144457783113247,
+      "loss": 1.2769,
+      "step": 2216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011140456182472989,
+      "loss": 1.8226,
+      "step": 2217
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011136454581832734,
+      "loss": 1.8476,
+      "step": 2218
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011132452981192478,
+      "loss": 1.783,
+      "step": 2219
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001112845138055222,
+      "loss": 2.3031,
+      "step": 2220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011124449779911965,
+      "loss": 1.7902,
+      "step": 2221
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011120448179271709,
+      "loss": 1.5655,
+      "step": 2222
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011116446578631454,
+      "loss": 2.1023,
+      "step": 2223
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011112444977991197,
+      "loss": 1.7913,
+      "step": 2224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011108443377350942,
+      "loss": 1.7802,
+      "step": 2225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011104441776710685,
+      "loss": 2.2305,
+      "step": 2226
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011100440176070428,
+      "loss": 1.8374,
+      "step": 2227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011096438575430173,
+      "loss": 1.7187,
+      "step": 2228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011092436974789917,
+      "loss": 1.5184,
+      "step": 2229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011088435374149659,
+      "loss": 1.7522,
+      "step": 2230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011084433773509404,
+      "loss": 1.5772,
+      "step": 2231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011080432172869149,
+      "loss": 1.8939,
+      "step": 2232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011076430572228893,
+      "loss": 2.0421,
+      "step": 2233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011072428971588635,
+      "loss": 1.396,
+      "step": 2234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001106842737094838,
+      "loss": 2.0392,
+      "step": 2235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011064425770308124,
+      "loss": 2.2659,
+      "step": 2236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011060424169667866,
+      "loss": 2.0358,
+      "step": 2237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011056422569027611,
+      "loss": 1.7951,
+      "step": 2238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011052420968387356,
+      "loss": 1.9335,
+      "step": 2239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000110484193677471,
+      "loss": 2.0924,
+      "step": 2240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011044417767106842,
+      "loss": 2.6237,
+      "step": 2241
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011040416166466588,
+      "loss": 1.8543,
+      "step": 2242
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011036414565826331,
+      "loss": 2.1354,
+      "step": 2243
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011032412965186074,
+      "loss": 2.0738,
+      "step": 2244
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011028411364545819,
+      "loss": 1.7056,
+      "step": 2245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011024409763905564,
+      "loss": 1.567,
+      "step": 2246
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011020408163265306,
+      "loss": 1.6135,
+      "step": 2247
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001101640656262505,
+      "loss": 1.3467,
+      "step": 2248
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011012404961984795,
+      "loss": 1.8617,
+      "step": 2249
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011008403361344539,
+      "loss": 2.1051,
+      "step": 2250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011004401760704281,
+      "loss": 1.8316,
+      "step": 2251
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011000400160064026,
+      "loss": 2.1657,
+      "step": 2252
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010996398559423771,
+      "loss": 1.7211,
+      "step": 2253
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010992396958783513,
+      "loss": 2.1547,
+      "step": 2254
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010988395358143257,
+      "loss": 1.5153,
+      "step": 2255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010984393757503002,
+      "loss": 1.6701,
+      "step": 2256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010980392156862746,
+      "loss": 1.8761,
+      "step": 2257
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010976390556222488,
+      "loss": 1.0919,
+      "step": 2258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010972388955582233,
+      "loss": 1.5859,
+      "step": 2259
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010968387354941978,
+      "loss": 1.5426,
+      "step": 2260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010964385754301721,
+      "loss": 1.8672,
+      "step": 2261
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010960384153661465,
+      "loss": 1.6926,
+      "step": 2262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001095638255302121,
+      "loss": 1.8548,
+      "step": 2263
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010952380952380953,
+      "loss": 2.1677,
+      "step": 2264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010948379351740696,
+      "loss": 2.012,
+      "step": 2265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010944377751100441,
+      "loss": 1.9988,
+      "step": 2266
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010940376150460186,
+      "loss": 1.7309,
+      "step": 2267
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010936374549819928,
+      "loss": 2.0141,
+      "step": 2268
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010932372949179672,
+      "loss": 1.4362,
+      "step": 2269
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010928371348539417,
+      "loss": 1.3938,
+      "step": 2270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010924369747899159,
+      "loss": 1.5085,
+      "step": 2271
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010920368147258903,
+      "loss": 1.9014,
+      "step": 2272
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010916366546618648,
+      "loss": 2.3358,
+      "step": 2273
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010912364945978393,
+      "loss": 2.1452,
+      "step": 2274
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010908363345338135,
+      "loss": 1.7096,
+      "step": 2275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010904361744697879,
+      "loss": 1.1291,
+      "step": 2276
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010900360144057624,
+      "loss": 1.9349,
+      "step": 2277
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010896358543417367,
+      "loss": 2.2523,
+      "step": 2278
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010892356942777112,
+      "loss": 2.1698,
+      "step": 2279
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010888355342136855,
+      "loss": 2.2189,
+      "step": 2280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000108843537414966,
+      "loss": 2.0862,
+      "step": 2281
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010880352140856343,
+      "loss": 1.9546,
+      "step": 2282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010876350540216087,
+      "loss": 1.8132,
+      "step": 2283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010872348939575832,
+      "loss": 2.1593,
+      "step": 2284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010868347338935574,
+      "loss": 1.7439,
+      "step": 2285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010864345738295319,
+      "loss": 1.3923,
+      "step": 2286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010860344137655063,
+      "loss": 1.8311,
+      "step": 2287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010856342537014808,
+      "loss": 2.5062,
+      "step": 2288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001085234093637455,
+      "loss": 1.6338,
+      "step": 2289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010848339335734294,
+      "loss": 2.0921,
+      "step": 2290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010844337735094039,
+      "loss": 1.6863,
+      "step": 2291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010840336134453781,
+      "loss": 1.5023,
+      "step": 2292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010836334533813526,
+      "loss": 2.0324,
+      "step": 2293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001083233293317327,
+      "loss": 1.7655,
+      "step": 2294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010828331332533012,
+      "loss": 1.8613,
+      "step": 2295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010824329731892758,
+      "loss": 2.3268,
+      "step": 2296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010820328131252501,
+      "loss": 1.7377,
+      "step": 2297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010816326530612246,
+      "loss": 1.7769,
+      "step": 2298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010812324929971989,
+      "loss": 1.9498,
+      "step": 2299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010808323329331734,
+      "loss": 1.8644,
+      "step": 2300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010804321728691477,
+      "loss": 1.8051,
+      "step": 2301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001080032012805122,
+      "loss": 2.3879,
+      "step": 2302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010796318527410965,
+      "loss": 1.7386,
+      "step": 2303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010792316926770709,
+      "loss": 2.0094,
+      "step": 2304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010788315326130454,
+      "loss": 1.9847,
+      "step": 2305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010784313725490196,
+      "loss": 1.4965,
+      "step": 2306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010780312124849941,
+      "loss": 1.7201,
+      "step": 2307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010776310524209685,
+      "loss": 2.1079,
+      "step": 2308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010772308923569427,
+      "loss": 1.8796,
+      "step": 2309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010768307322929172,
+      "loss": 1.9102,
+      "step": 2310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010764305722288916,
+      "loss": 1.9147,
+      "step": 2311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010760304121648661,
+      "loss": 1.5077,
+      "step": 2312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010756302521008403,
+      "loss": 2.2732,
+      "step": 2313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010752300920368148,
+      "loss": 2.2592,
+      "step": 2314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010748299319727892,
+      "loss": 1.8907,
+      "step": 2315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010744297719087635,
+      "loss": 2.0008,
+      "step": 2316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001074029611844738,
+      "loss": 2.8804,
+      "step": 2317
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010736294517807123,
+      "loss": 1.8535,
+      "step": 2318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010732292917166866,
+      "loss": 1.8018,
+      "step": 2319
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010728291316526611,
+      "loss": 2.1799,
+      "step": 2320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010724289715886356,
+      "loss": 1.8587,
+      "step": 2321
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000107202881152461,
+      "loss": 1.5703,
+      "step": 2322
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010716286514605842,
+      "loss": 1.6602,
+      "step": 2323
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010712284913965587,
+      "loss": 1.8916,
+      "step": 2324
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001070828331332533,
+      "loss": 1.2705,
+      "step": 2325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010704281712685073,
+      "loss": 2.009,
+      "step": 2326
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010700280112044818,
+      "loss": 1.7789,
+      "step": 2327
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010696278511404563,
+      "loss": 1.6809,
+      "step": 2328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010692276910764307,
+      "loss": 1.5674,
+      "step": 2329
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010688275310124049,
+      "loss": 1.8263,
+      "step": 2330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010684273709483794,
+      "loss": 2.1671,
+      "step": 2331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00010680272108843538,
+      "loss": 1.6315,
+      "step": 2332
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001067627050820328,
+      "loss": 1.5831,
+      "step": 2333
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010672268907563025,
+      "loss": 1.5839,
+      "step": 2334
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001066826730692277,
+      "loss": 2.1332,
+      "step": 2335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010664265706282513,
+      "loss": 1.9935,
+      "step": 2336
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010660264105642257,
+      "loss": 2.1091,
+      "step": 2337
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010656262505002002,
+      "loss": 1.6125,
+      "step": 2338
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010652260904361747,
+      "loss": 1.3818,
+      "step": 2339
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010648259303721488,
+      "loss": 1.8699,
+      "step": 2340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010644257703081233,
+      "loss": 1.7413,
+      "step": 2341
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010640256102440978,
+      "loss": 1.7011,
+      "step": 2342
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001063625450180072,
+      "loss": 2.1141,
+      "step": 2343
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010632252901160464,
+      "loss": 1.4992,
+      "step": 2344
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010628251300520209,
+      "loss": 1.5991,
+      "step": 2345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010624249699879954,
+      "loss": 1.7285,
+      "step": 2346
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010620248099239696,
+      "loss": 1.6614,
+      "step": 2347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001061624649859944,
+      "loss": 2.1906,
+      "step": 2348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010612244897959185,
+      "loss": 1.8986,
+      "step": 2349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010608243297318928,
+      "loss": 1.7774,
+      "step": 2350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010604241696678671,
+      "loss": 2.0995,
+      "step": 2351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010600240096038416,
+      "loss": 2.1892,
+      "step": 2352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010596238495398161,
+      "loss": 1.3216,
+      "step": 2353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010592236894757904,
+      "loss": 1.7543,
+      "step": 2354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010588235294117647,
+      "loss": 2.0775,
+      "step": 2355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010584233693477393,
+      "loss": 1.4033,
+      "step": 2356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010580232092837135,
+      "loss": 1.687,
+      "step": 2357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010576230492196879,
+      "loss": 1.6037,
+      "step": 2358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010572228891556624,
+      "loss": 1.7659,
+      "step": 2359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010568227290916366,
+      "loss": 1.7941,
+      "step": 2360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010564225690276111,
+      "loss": 2.0986,
+      "step": 2361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010560224089635855,
+      "loss": 1.7735,
+      "step": 2362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000105562224889956,
+      "loss": 1.5155,
+      "step": 2363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010552220888355342,
+      "loss": 1.5742,
+      "step": 2364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010548219287715086,
+      "loss": 2.2796,
+      "step": 2365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010548219287715086,
+      "loss": 2.1681,
+      "step": 2366
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010544217687074831,
+      "loss": 1.664,
+      "step": 2367
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010540216086434573,
+      "loss": 1.8545,
+      "step": 2368
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010536214485794318,
+      "loss": 1.3982,
+      "step": 2369
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010532212885154062,
+      "loss": 1.7277,
+      "step": 2370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010528211284513807,
+      "loss": 1.9426,
+      "step": 2371
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001052420968387355,
+      "loss": 2.0418,
+      "step": 2372
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010520208083233293,
+      "loss": 1.3925,
+      "step": 2373
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010516206482593038,
+      "loss": 1.802,
+      "step": 2374
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010512204881952781,
+      "loss": 2.0618,
+      "step": 2375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010508203281312526,
+      "loss": 1.9093,
+      "step": 2376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001050420168067227,
+      "loss": 2.0504,
+      "step": 2377
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010500200080032015,
+      "loss": 1.1163,
+      "step": 2378
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010496198479391757,
+      "loss": 1.9355,
+      "step": 2379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000104921968787515,
+      "loss": 1.7444,
+      "step": 2380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010488195278111246,
+      "loss": 1.9126,
+      "step": 2381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010484193677470988,
+      "loss": 1.7144,
+      "step": 2382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010480192076830733,
+      "loss": 1.6505,
+      "step": 2383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010476190476190477,
+      "loss": 1.7768,
+      "step": 2384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010472188875550219,
+      "loss": 1.7225,
+      "step": 2385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010468187274909964,
+      "loss": 2.3061,
+      "step": 2386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010464185674269708,
+      "loss": 2.0877,
+      "step": 2387
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010460184073629453,
+      "loss": 1.8096,
+      "step": 2388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010456182472989195,
+      "loss": 1.7934,
+      "step": 2389
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001045218087234894,
+      "loss": 1.5021,
+      "step": 2390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010448179271708684,
+      "loss": 1.7122,
+      "step": 2391
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010444177671068427,
+      "loss": 2.4079,
+      "step": 2392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010440176070428172,
+      "loss": 1.971,
+      "step": 2393
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010436174469787915,
+      "loss": 1.6001,
+      "step": 2394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001043217286914766,
+      "loss": 1.677,
+      "step": 2395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010428171268507403,
+      "loss": 1.5231,
+      "step": 2396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010424169667867148,
+      "loss": 1.7379,
+      "step": 2397
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010420168067226892,
+      "loss": 2.2136,
+      "step": 2398
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010416166466586634,
+      "loss": 2.2249,
+      "step": 2399
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010412164865946379,
+      "loss": 1.5538,
+      "step": 2400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010408163265306123,
+      "loss": 1.7524,
+      "step": 2401
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010404161664665868,
+      "loss": 1.7285,
+      "step": 2402
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001040016006402561,
+      "loss": 1.6943,
+      "step": 2403
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010396158463385355,
+      "loss": 1.5343,
+      "step": 2404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010392156862745099,
+      "loss": 1.8585,
+      "step": 2405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010388155262104841,
+      "loss": 1.8723,
+      "step": 2406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010384153661464586,
+      "loss": 1.9322,
+      "step": 2407
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010380152060824331,
+      "loss": 1.4889,
+      "step": 2408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010376150460184072,
+      "loss": 1.6306,
+      "step": 2409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010372148859543818,
+      "loss": 1.5344,
+      "step": 2410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010368147258903563,
+      "loss": 1.6327,
+      "step": 2411
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010364145658263306,
+      "loss": 2.0177,
+      "step": 2412
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010360144057623049,
+      "loss": 2.1587,
+      "step": 2413
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010356142456982794,
+      "loss": 1.6832,
+      "step": 2414
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010352140856342539,
+      "loss": 2.0508,
+      "step": 2415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010348139255702281,
+      "loss": 1.9702,
+      "step": 2416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010344137655062025,
+      "loss": 1.3954,
+      "step": 2417
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001034013605442177,
+      "loss": 1.5239,
+      "step": 2418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010336134453781514,
+      "loss": 1.5956,
+      "step": 2419
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010332132853141256,
+      "loss": 1.5202,
+      "step": 2420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010328131252501001,
+      "loss": 1.9961,
+      "step": 2421
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010324129651860746,
+      "loss": 1.5338,
+      "step": 2422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010320128051220488,
+      "loss": 1.8817,
+      "step": 2423
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010316126450580232,
+      "loss": 2.0003,
+      "step": 2424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010312124849939977,
+      "loss": 1.3521,
+      "step": 2425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001030812324929972,
+      "loss": 1.5984,
+      "step": 2426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010304121648659463,
+      "loss": 1.5211,
+      "step": 2427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010300120048019208,
+      "loss": 1.4521,
+      "step": 2428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010296118447378953,
+      "loss": 2.0606,
+      "step": 2429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010292116846738696,
+      "loss": 1.8027,
+      "step": 2430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001028811524609844,
+      "loss": 2.1922,
+      "step": 2431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010284113645458185,
+      "loss": 1.6487,
+      "step": 2432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010280112044817927,
+      "loss": 1.9413,
+      "step": 2433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010276110444177671,
+      "loss": 1.9879,
+      "step": 2434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010272108843537416,
+      "loss": 1.5533,
+      "step": 2435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010268107242897161,
+      "loss": 1.548,
+      "step": 2436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010264105642256903,
+      "loss": 2.2723,
+      "step": 2437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010260104041616647,
+      "loss": 1.7365,
+      "step": 2438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010256102440976392,
+      "loss": 2.1842,
+      "step": 2439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010252100840336134,
+      "loss": 2.1813,
+      "step": 2440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010248099239695878,
+      "loss": 2.5731,
+      "step": 2441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010244097639055623,
+      "loss": 1.6234,
+      "step": 2442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010240096038415368,
+      "loss": 1.6045,
+      "step": 2443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001023609443777511,
+      "loss": 1.2439,
+      "step": 2444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010232092837134854,
+      "loss": 1.6842,
+      "step": 2445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010228091236494599,
+      "loss": 2.1493,
+      "step": 2446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010224089635854342,
+      "loss": 1.7363,
+      "step": 2447
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010220088035214085,
+      "loss": 1.4673,
+      "step": 2448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001021608643457383,
+      "loss": 1.8865,
+      "step": 2449
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010212084833933573,
+      "loss": 1.8214,
+      "step": 2450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010208083233293318,
+      "loss": 2.0229,
+      "step": 2451
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010204081632653062,
+      "loss": 1.7377,
+      "step": 2452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010200080032012807,
+      "loss": 1.8294,
+      "step": 2453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010196078431372549,
+      "loss": 1.6405,
+      "step": 2454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010192076830732293,
+      "loss": 2.2805,
+      "step": 2455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010188075230092038,
+      "loss": 2.0122,
+      "step": 2456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001018407362945178,
+      "loss": 1.281,
+      "step": 2457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010180072028811525,
+      "loss": 2.0427,
+      "step": 2458
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010176070428171269,
+      "loss": 1.8511,
+      "step": 2459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010172068827531014,
+      "loss": 1.8983,
+      "step": 2460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010168067226890756,
+      "loss": 1.7693,
+      "step": 2461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000101640656262505,
+      "loss": 1.9594,
+      "step": 2462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010160064025610245,
+      "loss": 1.3986,
+      "step": 2463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010156062424969988,
+      "loss": 1.4197,
+      "step": 2464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010152060824329733,
+      "loss": 1.6359,
+      "step": 2465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010148059223689476,
+      "loss": 1.5155,
+      "step": 2466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010144057623049221,
+      "loss": 1.6755,
+      "step": 2467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010140056022408964,
+      "loss": 1.6652,
+      "step": 2468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010136054421768707,
+      "loss": 1.5054,
+      "step": 2469
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010132052821128453,
+      "loss": 1.7706,
+      "step": 2470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010128051220488195,
+      "loss": 2.1236,
+      "step": 2471
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001012404961984794,
+      "loss": 1.6349,
+      "step": 2472
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010120048019207684,
+      "loss": 2.1355,
+      "step": 2473
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010116046418567426,
+      "loss": 2.0549,
+      "step": 2474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010112044817927171,
+      "loss": 1.6391,
+      "step": 2475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010108043217286916,
+      "loss": 1.3347,
+      "step": 2476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001010404161664666,
+      "loss": 1.9863,
+      "step": 2477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010100040016006402,
+      "loss": 2.1104,
+      "step": 2478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010096038415366147,
+      "loss": 1.975,
+      "step": 2479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010092036814725891,
+      "loss": 1.3216,
+      "step": 2480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010088035214085633,
+      "loss": 1.6812,
+      "step": 2481
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010084033613445378,
+      "loss": 1.8067,
+      "step": 2482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010080032012805123,
+      "loss": 2.0884,
+      "step": 2483
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010076030412164867,
+      "loss": 1.8491,
+      "step": 2484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001007202881152461,
+      "loss": 1.5917,
+      "step": 2485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010068027210884355,
+      "loss": 1.575,
+      "step": 2486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010064025610244098,
+      "loss": 1.5881,
+      "step": 2487
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010060024009603841,
+      "loss": 1.8253,
+      "step": 2488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010056022408963586,
+      "loss": 1.9272,
+      "step": 2489
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010052020808323331,
+      "loss": 1.7909,
+      "step": 2490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010048019207683075,
+      "loss": 2.0094,
+      "step": 2491
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010044017607042817,
+      "loss": 1.7439,
+      "step": 2492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010040016006402562,
+      "loss": 2.2152,
+      "step": 2493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010036014405762306,
+      "loss": 1.7855,
+      "step": 2494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010032012805122048,
+      "loss": 1.6452,
+      "step": 2495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010028011204481793,
+      "loss": 1.4682,
+      "step": 2496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010024009603841538,
+      "loss": 1.8091,
+      "step": 2497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001002000800320128,
+      "loss": 1.9702,
+      "step": 2498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010016006402561024,
+      "loss": 1.5503,
+      "step": 2499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001001200480192077,
+      "loss": 1.4341,
+      "step": 2500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010008003201280513,
+      "loss": 2.0128,
+      "step": 2501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010004001600640255,
+      "loss": 1.3562,
+      "step": 2502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001,
+      "loss": 1.6264,
+      "step": 2503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.995998399359744e-05,
+      "loss": 1.6449,
+      "step": 2504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.991996798719488e-05,
+      "loss": 1.8509,
+      "step": 2505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.987995198079232e-05,
+      "loss": 2.1228,
+      "step": 2506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.983993597438977e-05,
+      "loss": 1.6765,
+      "step": 2507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.97999199679872e-05,
+      "loss": 1.666,
+      "step": 2508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.975990396158463e-05,
+      "loss": 1.8687,
+      "step": 2509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.971988795518208e-05,
+      "loss": 1.53,
+      "step": 2510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.967987194877952e-05,
+      "loss": 1.5632,
+      "step": 2511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.963985594237695e-05,
+      "loss": 1.6547,
+      "step": 2512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.959983993597439e-05,
+      "loss": 1.6945,
+      "step": 2513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.955982392957184e-05,
+      "loss": 1.8648,
+      "step": 2514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.951980792316928e-05,
+      "loss": 2.2699,
+      "step": 2515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.94797919167667e-05,
+      "loss": 1.4461,
+      "step": 2516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.943977591036415e-05,
+      "loss": 1.5849,
+      "step": 2517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.939975990396159e-05,
+      "loss": 1.8034,
+      "step": 2518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.935974389755903e-05,
+      "loss": 1.8919,
+      "step": 2519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.931972789115646e-05,
+      "loss": 2.018,
+      "step": 2520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.92797118847539e-05,
+      "loss": 2.0658,
+      "step": 2521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.923969587835135e-05,
+      "loss": 1.3454,
+      "step": 2522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.919967987194877e-05,
+      "loss": 2.2671,
+      "step": 2523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.915966386554623e-05,
+      "loss": 1.5264,
+      "step": 2524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.911964785914366e-05,
+      "loss": 2.0716,
+      "step": 2525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.90796318527411e-05,
+      "loss": 1.51,
+      "step": 2526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.903961584633854e-05,
+      "loss": 1.7321,
+      "step": 2527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.899959983993597e-05,
+      "loss": 2.1162,
+      "step": 2528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.895958383353342e-05,
+      "loss": 2.2664,
+      "step": 2529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.891956782713085e-05,
+      "loss": 1.8867,
+      "step": 2530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.88795518207283e-05,
+      "loss": 1.5647,
+      "step": 2531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.883953581432574e-05,
+      "loss": 1.5457,
+      "step": 2532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.879951980792317e-05,
+      "loss": 1.6703,
+      "step": 2533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.875950380152061e-05,
+      "loss": 1.692,
+      "step": 2534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.871948779511805e-05,
+      "loss": 1.7105,
+      "step": 2535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.86794717887155e-05,
+      "loss": 2.3682,
+      "step": 2536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.863945578231294e-05,
+      "loss": 1.5214,
+      "step": 2537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.859943977591037e-05,
+      "loss": 1.503,
+      "step": 2538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.855942376950781e-05,
+      "loss": 1.2751,
+      "step": 2539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.851940776310525e-05,
+      "loss": 1.4498,
+      "step": 2540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.847939175670268e-05,
+      "loss": 2.2475,
+      "step": 2541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.843937575030012e-05,
+      "loss": 1.8053,
+      "step": 2542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.839935974389757e-05,
+      "loss": 1.7566,
+      "step": 2543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.835934373749501e-05,
+      "loss": 2.2305,
+      "step": 2544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.831932773109243e-05,
+      "loss": 1.6287,
+      "step": 2545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.827931172468988e-05,
+      "loss": 1.9701,
+      "step": 2546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.823929571828732e-05,
+      "loss": 2.1255,
+      "step": 2547
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.819927971188476e-05,
+      "loss": 1.522,
+      "step": 2548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.81592637054822e-05,
+      "loss": 2.0145,
+      "step": 2549
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.811924769907965e-05,
+      "loss": 1.3255,
+      "step": 2550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.807923169267708e-05,
+      "loss": 1.5811,
+      "step": 2551
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.80392156862745e-05,
+      "loss": 1.7755,
+      "step": 2552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.799919967987196e-05,
+      "loss": 1.3542,
+      "step": 2553
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.79591836734694e-05,
+      "loss": 1.4587,
+      "step": 2554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.791916766706683e-05,
+      "loss": 1.6783,
+      "step": 2555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.787915166066427e-05,
+      "loss": 2.1323,
+      "step": 2556
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.78391356542617e-05,
+      "loss": 1.8119,
+      "step": 2557
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.779911964785916e-05,
+      "loss": 2.06,
+      "step": 2558
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.775910364145658e-05,
+      "loss": 1.4801,
+      "step": 2559
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.771908763505403e-05,
+      "loss": 1.4877,
+      "step": 2560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.767907162865147e-05,
+      "loss": 1.7118,
+      "step": 2561
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.76390556222489e-05,
+      "loss": 2.3714,
+      "step": 2562
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.759903961584634e-05,
+      "loss": 2.1472,
+      "step": 2563
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.755902360944378e-05,
+      "loss": 2.1019,
+      "step": 2564
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.751900760304123e-05,
+      "loss": 1.9541,
+      "step": 2565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.747899159663865e-05,
+      "loss": 1.4354,
+      "step": 2566
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.74389755902361e-05,
+      "loss": 2.2016,
+      "step": 2567
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.739895958383354e-05,
+      "loss": 1.0639,
+      "step": 2568
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.735894357743098e-05,
+      "loss": 1.5121,
+      "step": 2569
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.731892757102841e-05,
+      "loss": 1.3225,
+      "step": 2570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.727891156462585e-05,
+      "loss": 1.6661,
+      "step": 2571
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.72388955582233e-05,
+      "loss": 1.3607,
+      "step": 2572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.719887955182073e-05,
+      "loss": 1.9188,
+      "step": 2573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.715886354541818e-05,
+      "loss": 1.8383,
+      "step": 2574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.711884753901561e-05,
+      "loss": 1.5431,
+      "step": 2575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.707883153261305e-05,
+      "loss": 2.0969,
+      "step": 2576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.703881552621049e-05,
+      "loss": 1.5596,
+      "step": 2577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.699879951980793e-05,
+      "loss": 1.9918,
+      "step": 2578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.695878351340538e-05,
+      "loss": 2.2046,
+      "step": 2579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.69187675070028e-05,
+      "loss": 2.424,
+      "step": 2580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.687875150060024e-05,
+      "loss": 1.8509,
+      "step": 2581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.683873549419769e-05,
+      "loss": 2.0069,
+      "step": 2582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.679871948779512e-05,
+      "loss": 1.6973,
+      "step": 2583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.675870348139256e-05,
+      "loss": 1.9465,
+      "step": 2584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.671868747499e-05,
+      "loss": 1.1227,
+      "step": 2585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.667867146858745e-05,
+      "loss": 1.6834,
+      "step": 2586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.663865546218487e-05,
+      "loss": 1.8639,
+      "step": 2587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.659863945578231e-05,
+      "loss": 2.1158,
+      "step": 2588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.655862344937976e-05,
+      "loss": 2.087,
+      "step": 2589
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.65186074429772e-05,
+      "loss": 1.8356,
+      "step": 2590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.647859143657464e-05,
+      "loss": 1.3652,
+      "step": 2591
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.643857543017207e-05,
+      "loss": 1.8969,
+      "step": 2592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.639855942376951e-05,
+      "loss": 1.8199,
+      "step": 2593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.635854341736695e-05,
+      "loss": 1.8949,
+      "step": 2594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.631852741096438e-05,
+      "loss": 1.5563,
+      "step": 2595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.627851140456183e-05,
+      "loss": 1.7523,
+      "step": 2596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.623849539815927e-05,
+      "loss": 2.3271,
+      "step": 2597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.61984793917567e-05,
+      "loss": 2.0518,
+      "step": 2598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.615846338535415e-05,
+      "loss": 2.3022,
+      "step": 2599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.611844737895158e-05,
+      "loss": 1.7462,
+      "step": 2600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.607843137254903e-05,
+      "loss": 1.5176,
+      "step": 2601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.603841536614646e-05,
+      "loss": 1.7825,
+      "step": 2602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.599839935974391e-05,
+      "loss": 1.9475,
+      "step": 2603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.595838335334135e-05,
+      "loss": 1.7488,
+      "step": 2604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.591836734693878e-05,
+      "loss": 1.3751,
+      "step": 2605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.587835134053622e-05,
+      "loss": 1.5529,
+      "step": 2606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.583833533413366e-05,
+      "loss": 1.6232,
+      "step": 2607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.579831932773111e-05,
+      "loss": 1.3886,
+      "step": 2608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.575830332132853e-05,
+      "loss": 2.0462,
+      "step": 2609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.571828731492597e-05,
+      "loss": 1.5166,
+      "step": 2610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.567827130852342e-05,
+      "loss": 1.708,
+      "step": 2611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.563825530212086e-05,
+      "loss": 1.5443,
+      "step": 2612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.559823929571829e-05,
+      "loss": 2.1971,
+      "step": 2613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.555822328931573e-05,
+      "loss": 1.9159,
+      "step": 2614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.551820728291318e-05,
+      "loss": 2.0396,
+      "step": 2615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.54781912765106e-05,
+      "loss": 1.6137,
+      "step": 2616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.543817527010804e-05,
+      "loss": 1.9895,
+      "step": 2617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.539815926370549e-05,
+      "loss": 1.2947,
+      "step": 2618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.535814325730293e-05,
+      "loss": 1.8969,
+      "step": 2619
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.531812725090037e-05,
+      "loss": 2.0924,
+      "step": 2620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.52781112444978e-05,
+      "loss": 2.0798,
+      "step": 2621
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.523809523809524e-05,
+      "loss": 2.4688,
+      "step": 2622
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.519807923169268e-05,
+      "loss": 1.7684,
+      "step": 2623
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.515806322529011e-05,
+      "loss": 1.6207,
+      "step": 2624
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.511804721888757e-05,
+      "loss": 2.0524,
+      "step": 2625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.5078031212485e-05,
+      "loss": 2.1952,
+      "step": 2626
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.503801520608244e-05,
+      "loss": 1.9576,
+      "step": 2627
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.499799919967988e-05,
+      "loss": 1.55,
+      "step": 2628
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.495798319327731e-05,
+      "loss": 1.4058,
+      "step": 2629
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.491796718687475e-05,
+      "loss": 1.661,
+      "step": 2630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.487795118047219e-05,
+      "loss": 2.1208,
+      "step": 2631
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.483793517406964e-05,
+      "loss": 1.4076,
+      "step": 2632
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.479791916766708e-05,
+      "loss": 1.9707,
+      "step": 2633
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.47579031612645e-05,
+      "loss": 1.984,
+      "step": 2634
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.471788715486195e-05,
+      "loss": 2.4226,
+      "step": 2635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.467787114845939e-05,
+      "loss": 2.3016,
+      "step": 2636
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.463785514205682e-05,
+      "loss": 1.8934,
+      "step": 2637
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.459783913565426e-05,
+      "loss": 1.7533,
+      "step": 2638
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.455782312925171e-05,
+      "loss": 1.7441,
+      "step": 2639
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.451780712284915e-05,
+      "loss": 1.818,
+      "step": 2640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.447779111644657e-05,
+      "loss": 1.2451,
+      "step": 2641
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.443777511004402e-05,
+      "loss": 2.0031,
+      "step": 2642
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.439775910364146e-05,
+      "loss": 1.9215,
+      "step": 2643
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.43577430972389e-05,
+      "loss": 1.8191,
+      "step": 2644
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.431772709083634e-05,
+      "loss": 2.0313,
+      "step": 2645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.427771108443377e-05,
+      "loss": 2.1281,
+      "step": 2646
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.423769507803122e-05,
+      "loss": 1.9727,
+      "step": 2647
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.419767907162865e-05,
+      "loss": 2.1607,
+      "step": 2648
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.41576630652261e-05,
+      "loss": 1.5828,
+      "step": 2649
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.411764705882353e-05,
+      "loss": 1.4412,
+      "step": 2650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.407763105242097e-05,
+      "loss": 1.8311,
+      "step": 2651
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.403761504601841e-05,
+      "loss": 1.7449,
+      "step": 2652
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.399759903961585e-05,
+      "loss": 1.9787,
+      "step": 2653
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.39575830332133e-05,
+      "loss": 1.6167,
+      "step": 2654
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.391756702681072e-05,
+      "loss": 2.0546,
+      "step": 2655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.387755102040817e-05,
+      "loss": 1.7654,
+      "step": 2656
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.383753501400561e-05,
+      "loss": 1.6376,
+      "step": 2657
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.379751900760305e-05,
+      "loss": 1.9162,
+      "step": 2658
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.375750300120048e-05,
+      "loss": 2.1813,
+      "step": 2659
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.371748699479792e-05,
+      "loss": 1.4409,
+      "step": 2660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.367747098839537e-05,
+      "loss": 2.3011,
+      "step": 2661
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.36374549819928e-05,
+      "loss": 1.842,
+      "step": 2662
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.359743897559024e-05,
+      "loss": 1.7584,
+      "step": 2663
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.355742296918768e-05,
+      "loss": 1.4276,
+      "step": 2664
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.351740696278512e-05,
+      "loss": 1.5418,
+      "step": 2665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.347739095638256e-05,
+      "loss": 1.7517,
+      "step": 2666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.343737494997999e-05,
+      "loss": 1.8633,
+      "step": 2667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.339735894357744e-05,
+      "loss": 2.0609,
+      "step": 2668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.335734293717488e-05,
+      "loss": 1.9814,
+      "step": 2669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.33173269307723e-05,
+      "loss": 1.8006,
+      "step": 2670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.327731092436976e-05,
+      "loss": 1.9178,
+      "step": 2671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.323729491796719e-05,
+      "loss": 1.5657,
+      "step": 2672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.319727891156463e-05,
+      "loss": 1.4532,
+      "step": 2673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.315726290516207e-05,
+      "loss": 1.5705,
+      "step": 2674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.311724689875952e-05,
+      "loss": 1.9966,
+      "step": 2675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.307723089235695e-05,
+      "loss": 2.1709,
+      "step": 2676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.303721488595438e-05,
+      "loss": 1.7381,
+      "step": 2677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.299719887955183e-05,
+      "loss": 1.5992,
+      "step": 2678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.295718287314927e-05,
+      "loss": 1.8215,
+      "step": 2679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.29171668667467e-05,
+      "loss": 1.7895,
+      "step": 2680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.287715086034414e-05,
+      "loss": 1.7154,
+      "step": 2681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.283713485394158e-05,
+      "loss": 2.2277,
+      "step": 2682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.279711884753903e-05,
+      "loss": 1.9199,
+      "step": 2683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.275710284113645e-05,
+      "loss": 2.2338,
+      "step": 2684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.27170868347339e-05,
+      "loss": 1.2236,
+      "step": 2685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.267707082833134e-05,
+      "loss": 1.7935,
+      "step": 2686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.263705482192878e-05,
+      "loss": 1.776,
+      "step": 2687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.259703881552621e-05,
+      "loss": 1.7191,
+      "step": 2688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.255702280912365e-05,
+      "loss": 1.4351,
+      "step": 2689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.25170068027211e-05,
+      "loss": 1.3887,
+      "step": 2690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.247699079631853e-05,
+      "loss": 1.8602,
+      "step": 2691
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.243697478991598e-05,
+      "loss": 1.9696,
+      "step": 2692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.239695878351341e-05,
+      "loss": 1.4378,
+      "step": 2693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.235694277711085e-05,
+      "loss": 1.559,
+      "step": 2694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.231692677070829e-05,
+      "loss": 2.0053,
+      "step": 2695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.227691076430572e-05,
+      "loss": 2.023,
+      "step": 2696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.223689475790317e-05,
+      "loss": 1.6548,
+      "step": 2697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.21968787515006e-05,
+      "loss": 1.761,
+      "step": 2698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.215686274509804e-05,
+      "loss": 1.8408,
+      "step": 2699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.211684673869549e-05,
+      "loss": 1.8319,
+      "step": 2700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.207683073229292e-05,
+      "loss": 1.5422,
+      "step": 2701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.203681472589036e-05,
+      "loss": 1.854,
+      "step": 2702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.19967987194878e-05,
+      "loss": 1.8169,
+      "step": 2703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.195678271308525e-05,
+      "loss": 1.4267,
+      "step": 2704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.191676670668267e-05,
+      "loss": 2.0854,
+      "step": 2705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.187675070028011e-05,
+      "loss": 1.8938,
+      "step": 2706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.183673469387756e-05,
+      "loss": 1.5745,
+      "step": 2707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.1796718687475e-05,
+      "loss": 1.8608,
+      "step": 2708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.175670268107243e-05,
+      "loss": 1.6231,
+      "step": 2709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.171668667466987e-05,
+      "loss": 1.0792,
+      "step": 2710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.167667066826731e-05,
+      "loss": 1.5557,
+      "step": 2711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.163665466186475e-05,
+      "loss": 1.5676,
+      "step": 2712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.159663865546218e-05,
+      "loss": 1.7917,
+      "step": 2713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.155662264905963e-05,
+      "loss": 1.9524,
+      "step": 2714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.151660664265707e-05,
+      "loss": 1.37,
+      "step": 2715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.147659063625451e-05,
+      "loss": 1.5029,
+      "step": 2716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.143657462985194e-05,
+      "loss": 1.6616,
+      "step": 2717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.139655862344938e-05,
+      "loss": 1.9386,
+      "step": 2718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.135654261704682e-05,
+      "loss": 2.0027,
+      "step": 2719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.131652661064426e-05,
+      "loss": 1.934,
+      "step": 2720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.127651060424171e-05,
+      "loss": 1.88,
+      "step": 2721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.123649459783914e-05,
+      "loss": 1.56,
+      "step": 2722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.119647859143657e-05,
+      "loss": 1.6041,
+      "step": 2723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.115646258503402e-05,
+      "loss": 1.4954,
+      "step": 2724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.111644657863146e-05,
+      "loss": 1.3033,
+      "step": 2725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.107643057222889e-05,
+      "loss": 1.7669,
+      "step": 2726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.103641456582633e-05,
+      "loss": 1.8214,
+      "step": 2727
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.099639855942378e-05,
+      "loss": 1.7034,
+      "step": 2728
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.095638255302122e-05,
+      "loss": 2.0798,
+      "step": 2729
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.091636654661864e-05,
+      "loss": 2.2868,
+      "step": 2730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.087635054021609e-05,
+      "loss": 1.4626,
+      "step": 2731
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.083633453381353e-05,
+      "loss": 1.4017,
+      "step": 2732
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.079631852741098e-05,
+      "loss": 1.8732,
+      "step": 2733
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.07563025210084e-05,
+      "loss": 1.879,
+      "step": 2734
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.071628651460584e-05,
+      "loss": 1.5584,
+      "step": 2735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.067627050820329e-05,
+      "loss": 1.6029,
+      "step": 2736
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.063625450180073e-05,
+      "loss": 1.4217,
+      "step": 2737
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.059623849539817e-05,
+      "loss": 1.3744,
+      "step": 2738
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.05562224889956e-05,
+      "loss": 2.2365,
+      "step": 2739
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.051620648259305e-05,
+      "loss": 1.6307,
+      "step": 2740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.047619047619048e-05,
+      "loss": 2.0824,
+      "step": 2741
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.043617446978791e-05,
+      "loss": 1.8433,
+      "step": 2742
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.039615846338536e-05,
+      "loss": 1.965,
+      "step": 2743
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.03561424569828e-05,
+      "loss": 2.2915,
+      "step": 2744
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.031612645058024e-05,
+      "loss": 1.8271,
+      "step": 2745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.027611044417768e-05,
+      "loss": 2.0827,
+      "step": 2746
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.023609443777511e-05,
+      "loss": 1.6935,
+      "step": 2747
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.019607843137255e-05,
+      "loss": 1.9453,
+      "step": 2748
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.015606242496999e-05,
+      "loss": 1.8582,
+      "step": 2749
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.011604641856744e-05,
+      "loss": 2.2479,
+      "step": 2750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.007603041216488e-05,
+      "loss": 1.6329,
+      "step": 2751
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.003601440576231e-05,
+      "loss": 2.0004,
+      "step": 2752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.999599839935975e-05,
+      "loss": 2.304,
+      "step": 2753
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.995598239295719e-05,
+      "loss": 1.9658,
+      "step": 2754
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.991596638655462e-05,
+      "loss": 1.9018,
+      "step": 2755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.987595038015206e-05,
+      "loss": 2.1542,
+      "step": 2756
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.983593437374951e-05,
+      "loss": 1.7564,
+      "step": 2757
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.979591836734695e-05,
+      "loss": 2.2903,
+      "step": 2758
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.975590236094437e-05,
+      "loss": 1.6343,
+      "step": 2759
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.971588635454182e-05,
+      "loss": 1.9971,
+      "step": 2760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.967587034813926e-05,
+      "loss": 1.9492,
+      "step": 2761
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.96358543417367e-05,
+      "loss": 1.7702,
+      "step": 2762
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.959583833533413e-05,
+      "loss": 2.0062,
+      "step": 2763
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.955582232893158e-05,
+      "loss": 1.8099,
+      "step": 2764
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.951580632252902e-05,
+      "loss": 2.3839,
+      "step": 2765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.947579031612645e-05,
+      "loss": 1.9987,
+      "step": 2766
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.94357743097239e-05,
+      "loss": 1.9605,
+      "step": 2767
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.939575830332133e-05,
+      "loss": 1.9128,
+      "step": 2768
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.935574229691877e-05,
+      "loss": 2.4986,
+      "step": 2769
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.931572629051621e-05,
+      "loss": 1.7379,
+      "step": 2770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.927571028411364e-05,
+      "loss": 1.9613,
+      "step": 2771
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.92356942777111e-05,
+      "loss": 1.8755,
+      "step": 2772
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.919567827130852e-05,
+      "loss": 1.8062,
+      "step": 2773
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.915566226490597e-05,
+      "loss": 2.0479,
+      "step": 2774
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.911564625850341e-05,
+      "loss": 1.9909,
+      "step": 2775
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.907563025210084e-05,
+      "loss": 1.6085,
+      "step": 2776
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.903561424569828e-05,
+      "loss": 1.5379,
+      "step": 2777
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.899559823929572e-05,
+      "loss": 2.1999,
+      "step": 2778
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.895558223289317e-05,
+      "loss": 1.5824,
+      "step": 2779
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.891556622649059e-05,
+      "loss": 1.6643,
+      "step": 2780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.887555022008804e-05,
+      "loss": 1.9266,
+      "step": 2781
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.883553421368548e-05,
+      "loss": 1.5174,
+      "step": 2782
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.879551820728292e-05,
+      "loss": 1.7845,
+      "step": 2783
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.875550220088035e-05,
+      "loss": 1.82,
+      "step": 2784
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.871548619447779e-05,
+      "loss": 1.5185,
+      "step": 2785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.867547018807524e-05,
+      "loss": 1.5146,
+      "step": 2786
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.863545418167267e-05,
+      "loss": 1.6515,
+      "step": 2787
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.85954381752701e-05,
+      "loss": 2.0753,
+      "step": 2788
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.855542216886755e-05,
+      "loss": 1.9514,
+      "step": 2789
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.851540616246499e-05,
+      "loss": 1.7373,
+      "step": 2790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.847539015606243e-05,
+      "loss": 1.8204,
+      "step": 2791
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.843537414965987e-05,
+      "loss": 1.4948,
+      "step": 2792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.839535814325732e-05,
+      "loss": 2.1908,
+      "step": 2793
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.835534213685474e-05,
+      "loss": 1.6457,
+      "step": 2794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.831532613045218e-05,
+      "loss": 1.7798,
+      "step": 2795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.827531012404963e-05,
+      "loss": 1.5817,
+      "step": 2796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.823529411764706e-05,
+      "loss": 2.0635,
+      "step": 2797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.81952781112445e-05,
+      "loss": 1.4747,
+      "step": 2798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.815526210484194e-05,
+      "loss": 1.3653,
+      "step": 2799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.811524609843938e-05,
+      "loss": 1.8707,
+      "step": 2800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.807523009203683e-05,
+      "loss": 1.7869,
+      "step": 2801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.803521408563425e-05,
+      "loss": 1.7004,
+      "step": 2802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.79951980792317e-05,
+      "loss": 1.9156,
+      "step": 2803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.795518207282914e-05,
+      "loss": 1.8075,
+      "step": 2804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.791516606642658e-05,
+      "loss": 1.3719,
+      "step": 2805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.787515006002401e-05,
+      "loss": 1.6834,
+      "step": 2806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.783513405362145e-05,
+      "loss": 1.6741,
+      "step": 2807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.77951180472189e-05,
+      "loss": 2.5702,
+      "step": 2808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.775510204081632e-05,
+      "loss": 2.088,
+      "step": 2809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.771508603441377e-05,
+      "loss": 2.0542,
+      "step": 2810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.767507002801121e-05,
+      "loss": 1.6358,
+      "step": 2811
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.763505402160865e-05,
+      "loss": 1.9738,
+      "step": 2812
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.759503801520609e-05,
+      "loss": 1.4916,
+      "step": 2813
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.755502200880352e-05,
+      "loss": 1.8019,
+      "step": 2814
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.751500600240097e-05,
+      "loss": 1.9054,
+      "step": 2815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.74749899959984e-05,
+      "loss": 1.8683,
+      "step": 2816
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.743497398959585e-05,
+      "loss": 1.519,
+      "step": 2817
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.739495798319329e-05,
+      "loss": 2.1936,
+      "step": 2818
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.735494197679072e-05,
+      "loss": 1.7249,
+      "step": 2819
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.731492597038816e-05,
+      "loss": 1.9292,
+      "step": 2820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.72749099639856e-05,
+      "loss": 1.7591,
+      "step": 2821
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.723489395758305e-05,
+      "loss": 2.1386,
+      "step": 2822
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.719487795118047e-05,
+      "loss": 1.9073,
+      "step": 2823
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.715486194477791e-05,
+      "loss": 1.8205,
+      "step": 2824
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.711484593837536e-05,
+      "loss": 1.7322,
+      "step": 2825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.70748299319728e-05,
+      "loss": 1.5735,
+      "step": 2826
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.703481392557023e-05,
+      "loss": 2.1263,
+      "step": 2827
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.699479791916767e-05,
+      "loss": 2.2075,
+      "step": 2828
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.695478191276512e-05,
+      "loss": 1.7831,
+      "step": 2829
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.691476590636254e-05,
+      "loss": 1.9375,
+      "step": 2830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.687474989995998e-05,
+      "loss": 1.546,
+      "step": 2831
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.683473389355743e-05,
+      "loss": 1.4493,
+      "step": 2832
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.679471788715487e-05,
+      "loss": 1.7603,
+      "step": 2833
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.67547018807523e-05,
+      "loss": 1.7494,
+      "step": 2834
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.671468587434974e-05,
+      "loss": 2.0341,
+      "step": 2835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.667466986794718e-05,
+      "loss": 1.8356,
+      "step": 2836
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.663465386154462e-05,
+      "loss": 1.6741,
+      "step": 2837
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.659463785514205e-05,
+      "loss": 1.3216,
+      "step": 2838
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.65546218487395e-05,
+      "loss": 1.4296,
+      "step": 2839
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.651460584233694e-05,
+      "loss": 1.5934,
+      "step": 2840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.647458983593438e-05,
+      "loss": 1.8331,
+      "step": 2841
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.643457382953182e-05,
+      "loss": 1.4354,
+      "step": 2842
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.639455782312925e-05,
+      "loss": 2.1391,
+      "step": 2843
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.635454181672669e-05,
+      "loss": 1.2924,
+      "step": 2844
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.631452581032413e-05,
+      "loss": 1.8812,
+      "step": 2845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.627450980392158e-05,
+      "loss": 1.5126,
+      "step": 2846
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.623449379751902e-05,
+      "loss": 1.1634,
+      "step": 2847
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.619447779111644e-05,
+      "loss": 2.4904,
+      "step": 2848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.615446178471389e-05,
+      "loss": 1.3239,
+      "step": 2849
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.611444577831133e-05,
+      "loss": 1.8214,
+      "step": 2850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.607442977190876e-05,
+      "loss": 1.3019,
+      "step": 2851
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.60344137655062e-05,
+      "loss": 2.0928,
+      "step": 2852
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.599439775910365e-05,
+      "loss": 2.3758,
+      "step": 2853
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.595438175270109e-05,
+      "loss": 1.8691,
+      "step": 2854
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.591436574629851e-05,
+      "loss": 1.9655,
+      "step": 2855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.587434973989596e-05,
+      "loss": 1.8439,
+      "step": 2856
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.58343337334934e-05,
+      "loss": 1.1426,
+      "step": 2857
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.579431772709084e-05,
+      "loss": 2.4379,
+      "step": 2858
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.575430172068828e-05,
+      "loss": 2.1424,
+      "step": 2859
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.571428571428571e-05,
+      "loss": 1.9384,
+      "step": 2860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.567426970788316e-05,
+      "loss": 1.8614,
+      "step": 2861
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.563425370148059e-05,
+      "loss": 1.9667,
+      "step": 2862
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.559423769507804e-05,
+      "loss": 2.0178,
+      "step": 2863
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.555422168867547e-05,
+      "loss": 1.6915,
+      "step": 2864
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.551420568227293e-05,
+      "loss": 1.8587,
+      "step": 2865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.547418967587035e-05,
+      "loss": 1.2997,
+      "step": 2866
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.543417366946779e-05,
+      "loss": 1.8146,
+      "step": 2867
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.539415766306524e-05,
+      "loss": 1.6851,
+      "step": 2868
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.535414165666267e-05,
+      "loss": 2.0869,
+      "step": 2869
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.531412565026011e-05,
+      "loss": 1.6801,
+      "step": 2870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.527410964385755e-05,
+      "loss": 1.6803,
+      "step": 2871
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.523409363745499e-05,
+      "loss": 1.6639,
+      "step": 2872
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.519407763105242e-05,
+      "loss": 1.6762,
+      "step": 2873
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.515406162464986e-05,
+      "loss": 1.6086,
+      "step": 2874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.511404561824731e-05,
+      "loss": 2.1979,
+      "step": 2875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.507402961184475e-05,
+      "loss": 1.7832,
+      "step": 2876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.503401360544217e-05,
+      "loss": 1.6586,
+      "step": 2877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.499399759903962e-05,
+      "loss": 1.6242,
+      "step": 2878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.495398159263706e-05,
+      "loss": 1.1284,
+      "step": 2879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.49139655862345e-05,
+      "loss": 1.869,
+      "step": 2880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.487394957983193e-05,
+      "loss": 1.622,
+      "step": 2881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.483393357342938e-05,
+      "loss": 2.0775,
+      "step": 2882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.479391756702682e-05,
+      "loss": 1.5541,
+      "step": 2883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.475390156062424e-05,
+      "loss": 1.7592,
+      "step": 2884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.47138855542217e-05,
+      "loss": 1.8144,
+      "step": 2885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.467386954781913e-05,
+      "loss": 1.673,
+      "step": 2886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.463385354141657e-05,
+      "loss": 1.2229,
+      "step": 2887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.4593837535014e-05,
+      "loss": 1.8982,
+      "step": 2888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.455382152861144e-05,
+      "loss": 2.2478,
+      "step": 2889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.45138055222089e-05,
+      "loss": 1.885,
+      "step": 2890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.447378951580632e-05,
+      "loss": 1.4857,
+      "step": 2891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.443377350940377e-05,
+      "loss": 1.3191,
+      "step": 2892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.43937575030012e-05,
+      "loss": 1.9364,
+      "step": 2893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.435374149659864e-05,
+      "loss": 2.2524,
+      "step": 2894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.431372549019608e-05,
+      "loss": 2.0766,
+      "step": 2895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.427370948379352e-05,
+      "loss": 2.5269,
+      "step": 2896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.423369347739097e-05,
+      "loss": 1.8979,
+      "step": 2897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.419367747098839e-05,
+      "loss": 2.586,
+      "step": 2898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.415366146458584e-05,
+      "loss": 1.9637,
+      "step": 2899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.411364545818328e-05,
+      "loss": 1.51,
+      "step": 2900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.407362945178072e-05,
+      "loss": 1.6426,
+      "step": 2901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.403361344537815e-05,
+      "loss": 2.0992,
+      "step": 2902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.399359743897559e-05,
+      "loss": 1.3734,
+      "step": 2903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.395358143257304e-05,
+      "loss": 1.5307,
+      "step": 2904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.391356542617046e-05,
+      "loss": 1.247,
+      "step": 2905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.387354941976792e-05,
+      "loss": 2.2003,
+      "step": 2906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.383353341336535e-05,
+      "loss": 1.2361,
+      "step": 2907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.379351740696279e-05,
+      "loss": 1.668,
+      "step": 2908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.375350140056023e-05,
+      "loss": 1.7901,
+      "step": 2909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.371348539415766e-05,
+      "loss": 1.929,
+      "step": 2910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.367346938775511e-05,
+      "loss": 2.182,
+      "step": 2911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.363345338135254e-05,
+      "loss": 1.803,
+      "step": 2912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.359343737494998e-05,
+      "loss": 2.0605,
+      "step": 2913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.355342136854743e-05,
+      "loss": 2.3697,
+      "step": 2914
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.351340536214486e-05,
+      "loss": 1.5075,
+      "step": 2915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.34733893557423e-05,
+      "loss": 2.1889,
+      "step": 2916
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.343337334933974e-05,
+      "loss": 2.0738,
+      "step": 2917
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.339335734293719e-05,
+      "loss": 1.5835,
+      "step": 2918
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.335334133653461e-05,
+      "loss": 1.838,
+      "step": 2919
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.331332533013205e-05,
+      "loss": 1.8099,
+      "step": 2920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.32733093237295e-05,
+      "loss": 1.9904,
+      "step": 2921
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.323329331732694e-05,
+      "loss": 1.6919,
+      "step": 2922
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.319327731092437e-05,
+      "loss": 1.4854,
+      "step": 2923
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.315326130452181e-05,
+      "loss": 1.0945,
+      "step": 2924
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.311324529811925e-05,
+      "loss": 2.1002,
+      "step": 2925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.307322929171669e-05,
+      "loss": 1.5428,
+      "step": 2926
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.303321328531412e-05,
+      "loss": 1.7645,
+      "step": 2927
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.299319727891157e-05,
+      "loss": 1.887,
+      "step": 2928
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.295318127250901e-05,
+      "loss": 1.6473,
+      "step": 2929
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.291316526610645e-05,
+      "loss": 1.6658,
+      "step": 2930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.287314925970388e-05,
+      "loss": 1.7512,
+      "step": 2931
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.283313325330132e-05,
+      "loss": 1.7425,
+      "step": 2932
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.279311724689877e-05,
+      "loss": 1.8711,
+      "step": 2933
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.27531012404962e-05,
+      "loss": 1.6327,
+      "step": 2934
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.271308523409365e-05,
+      "loss": 1.9835,
+      "step": 2935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.267306922769108e-05,
+      "loss": 1.5682,
+      "step": 2936
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.263305322128852e-05,
+      "loss": 1.8373,
+      "step": 2937
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.259303721488596e-05,
+      "loss": 1.2928,
+      "step": 2938
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.25530212084834e-05,
+      "loss": 2.0101,
+      "step": 2939
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.251300520208085e-05,
+      "loss": 2.0962,
+      "step": 2940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.247298919567827e-05,
+      "loss": 1.6388,
+      "step": 2941
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.243297318927572e-05,
+      "loss": 1.9665,
+      "step": 2942
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.239295718287316e-05,
+      "loss": 1.9345,
+      "step": 2943
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.23529411764706e-05,
+      "loss": 1.7776,
+      "step": 2944
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.231292517006803e-05,
+      "loss": 1.6314,
+      "step": 2945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.227290916366547e-05,
+      "loss": 1.6389,
+      "step": 2946
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.223289315726292e-05,
+      "loss": 1.8602,
+      "step": 2947
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.219287715086034e-05,
+      "loss": 1.8469,
+      "step": 2948
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.215286114445778e-05,
+      "loss": 1.509,
+      "step": 2949
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.211284513805523e-05,
+      "loss": 1.8148,
+      "step": 2950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.207282913165267e-05,
+      "loss": 1.4172,
+      "step": 2951
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.20328131252501e-05,
+      "loss": 2.0375,
+      "step": 2952
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.199279711884754e-05,
+      "loss": 1.8072,
+      "step": 2953
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.195278111244499e-05,
+      "loss": 1.5461,
+      "step": 2954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.191276510604242e-05,
+      "loss": 1.9823,
+      "step": 2955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.187274909963985e-05,
+      "loss": 1.588,
+      "step": 2956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.18327330932373e-05,
+      "loss": 1.8102,
+      "step": 2957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.179271708683474e-05,
+      "loss": 1.9135,
+      "step": 2958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.175270108043218e-05,
+      "loss": 1.6408,
+      "step": 2959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.171268507402962e-05,
+      "loss": 2.2177,
+      "step": 2960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.167266906762705e-05,
+      "loss": 1.7809,
+      "step": 2961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.163265306122449e-05,
+      "loss": 1.5493,
+      "step": 2962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.159263705482193e-05,
+      "loss": 1.974,
+      "step": 2963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.155262104841938e-05,
+      "loss": 2.0116,
+      "step": 2964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.151260504201682e-05,
+      "loss": 1.7449,
+      "step": 2965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.147258903561424e-05,
+      "loss": 0.9914,
+      "step": 2966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.143257302921169e-05,
+      "loss": 2.1619,
+      "step": 2967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.139255702280913e-05,
+      "loss": 1.4215,
+      "step": 2968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.135254101640656e-05,
+      "loss": 1.3854,
+      "step": 2969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.1312525010004e-05,
+      "loss": 1.493,
+      "step": 2970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.127250900360145e-05,
+      "loss": 2.0484,
+      "step": 2971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.123249299719889e-05,
+      "loss": 1.7724,
+      "step": 2972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.119247699079631e-05,
+      "loss": 1.8391,
+      "step": 2973
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.115246098439376e-05,
+      "loss": 1.9556,
+      "step": 2974
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.11124449779912e-05,
+      "loss": 1.7417,
+      "step": 2975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.107242897158864e-05,
+      "loss": 1.6184,
+      "step": 2976
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.103241296518607e-05,
+      "loss": 1.2796,
+      "step": 2977
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.099239695878351e-05,
+      "loss": 1.9902,
+      "step": 2978
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.095238095238096e-05,
+      "loss": 1.4826,
+      "step": 2979
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.091236494597839e-05,
+      "loss": 1.4454,
+      "step": 2980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.087234893957584e-05,
+      "loss": 2.192,
+      "step": 2981
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.083233293317327e-05,
+      "loss": 1.5849,
+      "step": 2982
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.079231692677071e-05,
+      "loss": 1.9354,
+      "step": 2983
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.075230092036815e-05,
+      "loss": 2.1713,
+      "step": 2984
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.071228491396558e-05,
+      "loss": 1.7016,
+      "step": 2985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.067226890756304e-05,
+      "loss": 1.6146,
+      "step": 2986
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.063225290116046e-05,
+      "loss": 1.3101,
+      "step": 2987
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.059223689475791e-05,
+      "loss": 1.8897,
+      "step": 2988
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.055222088835535e-05,
+      "loss": 2.1088,
+      "step": 2989
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.051220488195278e-05,
+      "loss": 2.0898,
+      "step": 2990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.047218887555022e-05,
+      "loss": 1.4893,
+      "step": 2991
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.043217286914766e-05,
+      "loss": 2.0993,
+      "step": 2992
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.039215686274511e-05,
+      "loss": 2.3285,
+      "step": 2993
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.035214085634253e-05,
+      "loss": 1.5753,
+      "step": 2994
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.031212484993998e-05,
+      "loss": 1.6759,
+      "step": 2995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.027210884353742e-05,
+      "loss": 1.6173,
+      "step": 2996
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.023209283713486e-05,
+      "loss": 1.3442,
+      "step": 2997
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.01920768307323e-05,
+      "loss": 2.1052,
+      "step": 2998
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.015206082432973e-05,
+      "loss": 1.8324,
+      "step": 2999
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.011204481792718e-05,
+      "loss": 1.5584,
+      "step": 3000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.007202881152462e-05,
+      "loss": 1.8501,
+      "step": 3001
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.003201280512204e-05,
+      "loss": 2.0094,
+      "step": 3002
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.99919967987195e-05,
+      "loss": 1.5928,
+      "step": 3003
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.995198079231693e-05,
+      "loss": 1.4639,
+      "step": 3004
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.991196478591437e-05,
+      "loss": 2.1955,
+      "step": 3005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.98719487795118e-05,
+      "loss": 1.5626,
+      "step": 3006
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.983193277310926e-05,
+      "loss": 1.3754,
+      "step": 3007
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.979191676670669e-05,
+      "loss": 1.762,
+      "step": 3008
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.975190076030412e-05,
+      "loss": 1.5778,
+      "step": 3009
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.971188475390157e-05,
+      "loss": 1.9856,
+      "step": 3010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.9671868747499e-05,
+      "loss": 1.8753,
+      "step": 3011
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.963185274109644e-05,
+      "loss": 1.7492,
+      "step": 3012
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.959183673469388e-05,
+      "loss": 2.1049,
+      "step": 3013
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.955182072829132e-05,
+      "loss": 1.6727,
+      "step": 3014
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.951180472188877e-05,
+      "loss": 1.7744,
+      "step": 3015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.947178871548619e-05,
+      "loss": 1.6661,
+      "step": 3016
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.943177270908364e-05,
+      "loss": 1.9551,
+      "step": 3017
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.939175670268108e-05,
+      "loss": 2.2869,
+      "step": 3018
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.935174069627852e-05,
+      "loss": 1.6334,
+      "step": 3019
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.931172468987595e-05,
+      "loss": 1.894,
+      "step": 3020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.927170868347339e-05,
+      "loss": 1.7274,
+      "step": 3021
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.923169267707084e-05,
+      "loss": 1.7272,
+      "step": 3022
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.919167667066826e-05,
+      "loss": 1.5754,
+      "step": 3023
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.915166066426571e-05,
+      "loss": 1.5573,
+      "step": 3024
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.911164465786315e-05,
+      "loss": 2.1787,
+      "step": 3025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.907162865146059e-05,
+      "loss": 1.8858,
+      "step": 3026
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.903161264505803e-05,
+      "loss": 1.8884,
+      "step": 3027
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.899159663865546e-05,
+      "loss": 1.8547,
+      "step": 3028
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.895158063225291e-05,
+      "loss": 1.4711,
+      "step": 3029
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.891156462585034e-05,
+      "loss": 1.8886,
+      "step": 3030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.887154861944779e-05,
+      "loss": 1.5082,
+      "step": 3031
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.883153261304523e-05,
+      "loss": 1.4568,
+      "step": 3032
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.879151660664266e-05,
+      "loss": 1.6839,
+      "step": 3033
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.87515006002401e-05,
+      "loss": 1.9867,
+      "step": 3034
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.871148459383754e-05,
+      "loss": 1.9328,
+      "step": 3035
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.867146858743499e-05,
+      "loss": 2.4702,
+      "step": 3036
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.863145258103241e-05,
+      "loss": 1.2528,
+      "step": 3037
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.859143657462985e-05,
+      "loss": 2.0138,
+      "step": 3038
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.85514205682273e-05,
+      "loss": 1.4911,
+      "step": 3039
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.851140456182474e-05,
+      "loss": 1.6595,
+      "step": 3040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.847138855542217e-05,
+      "loss": 1.8056,
+      "step": 3041
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.843137254901961e-05,
+      "loss": 1.3819,
+      "step": 3042
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.839135654261706e-05,
+      "loss": 1.4347,
+      "step": 3043
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.835134053621448e-05,
+      "loss": 1.4812,
+      "step": 3044
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.831132452981192e-05,
+      "loss": 1.7675,
+      "step": 3045
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.827130852340937e-05,
+      "loss": 1.6252,
+      "step": 3046
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.823129251700681e-05,
+      "loss": 1.941,
+      "step": 3047
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.819127651060425e-05,
+      "loss": 1.7851,
+      "step": 3048
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.815126050420168e-05,
+      "loss": 1.624,
+      "step": 3049
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.811124449779912e-05,
+      "loss": 2.0859,
+      "step": 3050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.807122849139656e-05,
+      "loss": 1.8686,
+      "step": 3051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.8031212484994e-05,
+      "loss": 2.097,
+      "step": 3052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.799119647859145e-05,
+      "loss": 1.9361,
+      "step": 3053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.795118047218888e-05,
+      "loss": 1.7943,
+      "step": 3054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.791116446578632e-05,
+      "loss": 1.6193,
+      "step": 3055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.787114845938376e-05,
+      "loss": 1.76,
+      "step": 3056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.78311324529812e-05,
+      "loss": 1.6188,
+      "step": 3057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.779111644657863e-05,
+      "loss": 1.6047,
+      "step": 3058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.775110044017607e-05,
+      "loss": 1.5522,
+      "step": 3059
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.771108443377352e-05,
+      "loss": 1.7505,
+      "step": 3060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.767106842737096e-05,
+      "loss": 2.1371,
+      "step": 3061
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.763105242096838e-05,
+      "loss": 1.5863,
+      "step": 3062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.759103641456583e-05,
+      "loss": 1.805,
+      "step": 3063
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.755102040816327e-05,
+      "loss": 1.7682,
+      "step": 3064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.751100440176072e-05,
+      "loss": 1.6628,
+      "step": 3065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.747098839535814e-05,
+      "loss": 2.0525,
+      "step": 3066
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.743097238895558e-05,
+      "loss": 1.6864,
+      "step": 3067
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.739095638255303e-05,
+      "loss": 1.8194,
+      "step": 3068
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.735094037615047e-05,
+      "loss": 1.9414,
+      "step": 3069
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.73109243697479e-05,
+      "loss": 1.6903,
+      "step": 3070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.727090836334534e-05,
+      "loss": 1.8445,
+      "step": 3071
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.723089235694279e-05,
+      "loss": 1.6736,
+      "step": 3072
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.719087635054022e-05,
+      "loss": 1.6023,
+      "step": 3073
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.715086034413765e-05,
+      "loss": 1.6047,
+      "step": 3074
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.71108443377351e-05,
+      "loss": 1.8032,
+      "step": 3075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.707082833133254e-05,
+      "loss": 1.309,
+      "step": 3076
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.703081232492998e-05,
+      "loss": 1.9772,
+      "step": 3077
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.699079631852741e-05,
+      "loss": 1.8729,
+      "step": 3078
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.695078031212485e-05,
+      "loss": 2.2796,
+      "step": 3079
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.691076430572229e-05,
+      "loss": 1.1587,
+      "step": 3080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.687074829931973e-05,
+      "loss": 1.7119,
+      "step": 3081
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.683073229291718e-05,
+      "loss": 2.2045,
+      "step": 3082
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.679071628651461e-05,
+      "loss": 1.6872,
+      "step": 3083
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.675070028011205e-05,
+      "loss": 1.486,
+      "step": 3084
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.671068427370949e-05,
+      "loss": 2.2329,
+      "step": 3085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.667066826730693e-05,
+      "loss": 1.729,
+      "step": 3086
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.663065226090436e-05,
+      "loss": 1.6134,
+      "step": 3087
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.65906362545018e-05,
+      "loss": 1.9273,
+      "step": 3088
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.655062024809925e-05,
+      "loss": 1.7594,
+      "step": 3089
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.651060424169669e-05,
+      "loss": 1.326,
+      "step": 3090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.647058823529411e-05,
+      "loss": 1.9329,
+      "step": 3091
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.643057222889156e-05,
+      "loss": 1.6895,
+      "step": 3092
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.6390556222489e-05,
+      "loss": 1.7048,
+      "step": 3093
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.635054021608644e-05,
+      "loss": 1.8997,
+      "step": 3094
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.631052420968387e-05,
+      "loss": 1.1684,
+      "step": 3095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.627050820328132e-05,
+      "loss": 1.9406,
+      "step": 3096
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.623049219687876e-05,
+      "loss": 2.1924,
+      "step": 3097
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.619047619047618e-05,
+      "loss": 1.7822,
+      "step": 3098
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.615046018407364e-05,
+      "loss": 1.8399,
+      "step": 3099
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.611044417767107e-05,
+      "loss": 2.1447,
+      "step": 3100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.607042817126851e-05,
+      "loss": 1.5465,
+      "step": 3101
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.603041216486595e-05,
+      "loss": 1.6598,
+      "step": 3102
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.599039615846338e-05,
+      "loss": 1.4347,
+      "step": 3103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.595038015206083e-05,
+      "loss": 1.7156,
+      "step": 3104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.591036414565826e-05,
+      "loss": 1.3536,
+      "step": 3105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.587034813925571e-05,
+      "loss": 1.8993,
+      "step": 3106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.583033213285315e-05,
+      "loss": 1.5209,
+      "step": 3107
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.579031612645058e-05,
+      "loss": 1.1656,
+      "step": 3108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.575030012004802e-05,
+      "loss": 1.7507,
+      "step": 3109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.571028411364546e-05,
+      "loss": 1.6852,
+      "step": 3110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.567026810724291e-05,
+      "loss": 1.3979,
+      "step": 3111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.563025210084033e-05,
+      "loss": 2.1216,
+      "step": 3112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.559023609443778e-05,
+      "loss": 1.8289,
+      "step": 3113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.555022008803522e-05,
+      "loss": 1.6072,
+      "step": 3114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.551020408163266e-05,
+      "loss": 1.7491,
+      "step": 3115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.54701880752301e-05,
+      "loss": 1.2748,
+      "step": 3116
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.543017206882753e-05,
+      "loss": 1.6279,
+      "step": 3117
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.539015606242498e-05,
+      "loss": 1.8561,
+      "step": 3118
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.53501400560224e-05,
+      "loss": 1.8151,
+      "step": 3119
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.531012404961986e-05,
+      "loss": 1.8074,
+      "step": 3120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.527010804321729e-05,
+      "loss": 1.753,
+      "step": 3121
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.523009203681473e-05,
+      "loss": 2.0372,
+      "step": 3122
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.519007603041217e-05,
+      "loss": 1.9442,
+      "step": 3123
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.51500600240096e-05,
+      "loss": 2.0475,
+      "step": 3124
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.511004401760705e-05,
+      "loss": 1.6302,
+      "step": 3125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.507002801120448e-05,
+      "loss": 1.5486,
+      "step": 3126
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.503001200480192e-05,
+      "loss": 1.3195,
+      "step": 3127
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.498999599839937e-05,
+      "loss": 1.4624,
+      "step": 3128
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.49499799919968e-05,
+      "loss": 1.9852,
+      "step": 3129
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.490996398559424e-05,
+      "loss": 1.9661,
+      "step": 3130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.486994797919168e-05,
+      "loss": 1.8898,
+      "step": 3131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.482993197278913e-05,
+      "loss": 1.0215,
+      "step": 3132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.478991596638657e-05,
+      "loss": 1.6291,
+      "step": 3133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.474989995998399e-05,
+      "loss": 1.3985,
+      "step": 3134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.470988395358144e-05,
+      "loss": 2.0547,
+      "step": 3135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.466986794717888e-05,
+      "loss": 1.9929,
+      "step": 3136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.462985194077631e-05,
+      "loss": 1.9117,
+      "step": 3137
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.458983593437375e-05,
+      "loss": 1.6904,
+      "step": 3138
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.454981992797119e-05,
+      "loss": 1.462,
+      "step": 3139
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.450980392156864e-05,
+      "loss": 1.4799,
+      "step": 3140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.446978791516606e-05,
+      "loss": 1.6819,
+      "step": 3141
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.442977190876351e-05,
+      "loss": 1.5756,
+      "step": 3142
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.438975590236095e-05,
+      "loss": 1.7219,
+      "step": 3143
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.434973989595839e-05,
+      "loss": 1.8061,
+      "step": 3144
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.430972388955582e-05,
+      "loss": 1.384,
+      "step": 3145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.426970788315326e-05,
+      "loss": 1.3296,
+      "step": 3146
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.422969187675071e-05,
+      "loss": 1.6005,
+      "step": 3147
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.418967587034814e-05,
+      "loss": 1.7312,
+      "step": 3148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.414965986394559e-05,
+      "loss": 2.0603,
+      "step": 3149
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.410964385754302e-05,
+      "loss": 1.5045,
+      "step": 3150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.406962785114046e-05,
+      "loss": 1.4378,
+      "step": 3151
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.40296118447379e-05,
+      "loss": 1.7715,
+      "step": 3152
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.398959583833534e-05,
+      "loss": 1.3954,
+      "step": 3153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.394957983193279e-05,
+      "loss": 1.963,
+      "step": 3154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.390956382553021e-05,
+      "loss": 1.4492,
+      "step": 3155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.386954781912765e-05,
+      "loss": 1.8731,
+      "step": 3156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.38295318127251e-05,
+      "loss": 1.8533,
+      "step": 3157
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.378951580632253e-05,
+      "loss": 1.9483,
+      "step": 3158
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.374949979991997e-05,
+      "loss": 1.9218,
+      "step": 3159
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.370948379351741e-05,
+      "loss": 2.1086,
+      "step": 3160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.366946778711486e-05,
+      "loss": 1.9058,
+      "step": 3161
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.362945178071228e-05,
+      "loss": 1.3771,
+      "step": 3162
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.358943577430972e-05,
+      "loss": 1.7839,
+      "step": 3163
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.354941976790717e-05,
+      "loss": 1.9092,
+      "step": 3164
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.350940376150461e-05,
+      "loss": 1.5636,
+      "step": 3165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.346938775510205e-05,
+      "loss": 2.0162,
+      "step": 3166
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.342937174869948e-05,
+      "loss": 1.8281,
+      "step": 3167
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.338935574229692e-05,
+      "loss": 1.388,
+      "step": 3168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.334933973589436e-05,
+      "loss": 1.7693,
+      "step": 3169
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.33093237294918e-05,
+      "loss": 1.9879,
+      "step": 3170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.326930772308924e-05,
+      "loss": 1.7403,
+      "step": 3171
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.322929171668668e-05,
+      "loss": 1.6329,
+      "step": 3172
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.318927571028412e-05,
+      "loss": 1.8985,
+      "step": 3173
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.314925970388156e-05,
+      "loss": 1.1279,
+      "step": 3174
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.310924369747899e-05,
+      "loss": 1.6256,
+      "step": 3175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.306922769107643e-05,
+      "loss": 2.1265,
+      "step": 3176
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.302921168467387e-05,
+      "loss": 1.8008,
+      "step": 3177
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.298919567827132e-05,
+      "loss": 1.5218,
+      "step": 3178
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.294917967186876e-05,
+      "loss": 2.0707,
+      "step": 3179
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.290916366546618e-05,
+      "loss": 1.5933,
+      "step": 3180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.286914765906363e-05,
+      "loss": 1.993,
+      "step": 3181
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.282913165266107e-05,
+      "loss": 2.3427,
+      "step": 3182
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.27891156462585e-05,
+      "loss": 1.289,
+      "step": 3183
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.274909963985594e-05,
+      "loss": 1.6745,
+      "step": 3184
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.270908363345339e-05,
+      "loss": 1.4499,
+      "step": 3185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.266906762705083e-05,
+      "loss": 1.6671,
+      "step": 3186
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.262905162064825e-05,
+      "loss": 2.2002,
+      "step": 3187
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.25890356142457e-05,
+      "loss": 1.7065,
+      "step": 3188
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.254901960784314e-05,
+      "loss": 1.8025,
+      "step": 3189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.250900360144058e-05,
+      "loss": 1.9638,
+      "step": 3190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.246898759503801e-05,
+      "loss": 1.4801,
+      "step": 3191
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.242897158863545e-05,
+      "loss": 1.7099,
+      "step": 3192
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.23889555822329e-05,
+      "loss": 1.9151,
+      "step": 3193
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.234893957583033e-05,
+      "loss": 1.3759,
+      "step": 3194
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.230892356942778e-05,
+      "loss": 1.9375,
+      "step": 3195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.226890756302521e-05,
+      "loss": 1.817,
+      "step": 3196
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.222889155662266e-05,
+      "loss": 1.761,
+      "step": 3197
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.218887555022009e-05,
+      "loss": 1.6687,
+      "step": 3198
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.214885954381752e-05,
+      "loss": 2.1529,
+      "step": 3199
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.210884353741498e-05,
+      "loss": 1.5485,
+      "step": 3200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.206882753101241e-05,
+      "loss": 1.7555,
+      "step": 3201
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.202881152460985e-05,
+      "loss": 1.4994,
+      "step": 3202
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.198879551820729e-05,
+      "loss": 1.7029,
+      "step": 3203
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.194877951180472e-05,
+      "loss": 1.5068,
+      "step": 3204
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.190876350540216e-05,
+      "loss": 1.6947,
+      "step": 3205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.18687474989996e-05,
+      "loss": 1.367,
+      "step": 3206
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.182873149259705e-05,
+      "loss": 2.1102,
+      "step": 3207
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.178871548619449e-05,
+      "loss": 1.9779,
+      "step": 3208
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.174869947979192e-05,
+      "loss": 1.6775,
+      "step": 3209
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.170868347338936e-05,
+      "loss": 2.0226,
+      "step": 3210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.16686674669868e-05,
+      "loss": 1.818,
+      "step": 3211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.162865146058423e-05,
+      "loss": 1.2811,
+      "step": 3212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.158863545418167e-05,
+      "loss": 2.0503,
+      "step": 3213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.154861944777912e-05,
+      "loss": 1.9547,
+      "step": 3214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.150860344137656e-05,
+      "loss": 1.6586,
+      "step": 3215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.146858743497398e-05,
+      "loss": 1.9468,
+      "step": 3216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.142857142857143e-05,
+      "loss": 1.7126,
+      "step": 3217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.138855542216887e-05,
+      "loss": 1.3839,
+      "step": 3218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.134853941576631e-05,
+      "loss": 1.9522,
+      "step": 3219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.130852340936375e-05,
+      "loss": 1.8056,
+      "step": 3220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.12685074029612e-05,
+      "loss": 1.4438,
+      "step": 3221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.122849139655863e-05,
+      "loss": 1.7226,
+      "step": 3222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.118847539015606e-05,
+      "loss": 2.3279,
+      "step": 3223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.114845938375351e-05,
+      "loss": 1.9987,
+      "step": 3224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.110844337735094e-05,
+      "loss": 2.2386,
+      "step": 3225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.106842737094838e-05,
+      "loss": 2.3192,
+      "step": 3226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.102841136454582e-05,
+      "loss": 1.2837,
+      "step": 3227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.098839535814326e-05,
+      "loss": 1.7608,
+      "step": 3228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.09483793517407e-05,
+      "loss": 1.6658,
+      "step": 3229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.090836334533813e-05,
+      "loss": 1.8423,
+      "step": 3230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.086834733893558e-05,
+      "loss": 2.301,
+      "step": 3231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.082833133253302e-05,
+      "loss": 1.6543,
+      "step": 3232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.078831532613046e-05,
+      "loss": 1.8662,
+      "step": 3233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.074829931972789e-05,
+      "loss": 1.7909,
+      "step": 3234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.070828331332533e-05,
+      "loss": 2.1659,
+      "step": 3235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.066826730692278e-05,
+      "loss": 1.5457,
+      "step": 3236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.06282513005202e-05,
+      "loss": 1.653,
+      "step": 3237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.058823529411765e-05,
+      "loss": 1.4996,
+      "step": 3238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.054821928771509e-05,
+      "loss": 2.1743,
+      "step": 3239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.050820328131253e-05,
+      "loss": 1.806,
+      "step": 3240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.046818727490997e-05,
+      "loss": 1.9869,
+      "step": 3241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.04281712685074e-05,
+      "loss": 1.7719,
+      "step": 3242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.038815526210485e-05,
+      "loss": 1.5874,
+      "step": 3243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.034813925570228e-05,
+      "loss": 1.7634,
+      "step": 3244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.030812324929971e-05,
+      "loss": 1.8532,
+      "step": 3245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.026810724289717e-05,
+      "loss": 1.6836,
+      "step": 3246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.02280912364946e-05,
+      "loss": 1.7213,
+      "step": 3247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.018807523009204e-05,
+      "loss": 2.2059,
+      "step": 3248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.014805922368948e-05,
+      "loss": 1.4173,
+      "step": 3249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.010804321728693e-05,
+      "loss": 2.0022,
+      "step": 3250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.006802721088435e-05,
+      "loss": 1.8441,
+      "step": 3251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.002801120448179e-05,
+      "loss": 1.4605,
+      "step": 3252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.998799519807924e-05,
+      "loss": 1.7456,
+      "step": 3253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.994797919167668e-05,
+      "loss": 1.9659,
+      "step": 3254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.990796318527411e-05,
+      "loss": 1.6788,
+      "step": 3255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.986794717887155e-05,
+      "loss": 1.9822,
+      "step": 3256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.982793117246899e-05,
+      "loss": 2.0831,
+      "step": 3257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.978791516606642e-05,
+      "loss": 2.108,
+      "step": 3258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.974789915966386e-05,
+      "loss": 1.7565,
+      "step": 3259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.970788315326131e-05,
+      "loss": 2.0647,
+      "step": 3260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.966786714685875e-05,
+      "loss": 1.757,
+      "step": 3261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.962785114045619e-05,
+      "loss": 1.2854,
+      "step": 3262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.958783513405362e-05,
+      "loss": 1.578,
+      "step": 3263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.954781912765106e-05,
+      "loss": 1.4426,
+      "step": 3264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.950780312124851e-05,
+      "loss": 1.0425,
+      "step": 3265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.946778711484593e-05,
+      "loss": 2.2572,
+      "step": 3266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.942777110844339e-05,
+      "loss": 2.2533,
+      "step": 3267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.938775510204082e-05,
+      "loss": 1.7091,
+      "step": 3268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.934773909563826e-05,
+      "loss": 2.236,
+      "step": 3269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.93077230892357e-05,
+      "loss": 2.2374,
+      "step": 3270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.926770708283313e-05,
+      "loss": 1.0772,
+      "step": 3271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.922769107643058e-05,
+      "loss": 1.4101,
+      "step": 3272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.918767507002801e-05,
+      "loss": 1.3501,
+      "step": 3273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.914765906362546e-05,
+      "loss": 1.3416,
+      "step": 3274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.91076430572229e-05,
+      "loss": 1.8721,
+      "step": 3275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.906762705082033e-05,
+      "loss": 1.8356,
+      "step": 3276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.902761104441777e-05,
+      "loss": 1.6615,
+      "step": 3277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.898759503801521e-05,
+      "loss": 1.8597,
+      "step": 3278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.894757903161266e-05,
+      "loss": 2.1969,
+      "step": 3279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.890756302521008e-05,
+      "loss": 1.2302,
+      "step": 3280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.886754701880752e-05,
+      "loss": 1.7981,
+      "step": 3281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.882753101240497e-05,
+      "loss": 2.2255,
+      "step": 3282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.878751500600241e-05,
+      "loss": 1.8361,
+      "step": 3283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.874749899959984e-05,
+      "loss": 1.6134,
+      "step": 3284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.870748299319728e-05,
+      "loss": 1.564,
+      "step": 3285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.866746698679473e-05,
+      "loss": 1.6201,
+      "step": 3286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.862745098039216e-05,
+      "loss": 2.314,
+      "step": 3287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.858743497398959e-05,
+      "loss": 2.1132,
+      "step": 3288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.854741896758704e-05,
+      "loss": 1.9621,
+      "step": 3289
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.850740296118448e-05,
+      "loss": 2.0835,
+      "step": 3290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.846738695478192e-05,
+      "loss": 1.626,
+      "step": 3291
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.842737094837935e-05,
+      "loss": 1.786,
+      "step": 3292
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.838735494197679e-05,
+      "loss": 2.0618,
+      "step": 3293
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.834733893557423e-05,
+      "loss": 1.7953,
+      "step": 3294
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.830732292917167e-05,
+      "loss": 1.8253,
+      "step": 3295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.826730692276912e-05,
+      "loss": 1.7685,
+      "step": 3296
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.822729091636655e-05,
+      "loss": 2.158,
+      "step": 3297
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.818727490996399e-05,
+      "loss": 1.7548,
+      "step": 3298
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.814725890356143e-05,
+      "loss": 2.0148,
+      "step": 3299
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.810724289715887e-05,
+      "loss": 1.3539,
+      "step": 3300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.80672268907563e-05,
+      "loss": 1.4876,
+      "step": 3301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.802721088435374e-05,
+      "loss": 2.2027,
+      "step": 3302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.798719487795119e-05,
+      "loss": 1.7125,
+      "step": 3303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.794717887154863e-05,
+      "loss": 1.419,
+      "step": 3304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.790716286514605e-05,
+      "loss": 2.0098,
+      "step": 3305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.78671468587435e-05,
+      "loss": 1.6464,
+      "step": 3306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.782713085234094e-05,
+      "loss": 1.7305,
+      "step": 3307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.778711484593838e-05,
+      "loss": 2.1313,
+      "step": 3308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.774709883953581e-05,
+      "loss": 2.2021,
+      "step": 3309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.770708283313326e-05,
+      "loss": 2.0543,
+      "step": 3310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.76670668267307e-05,
+      "loss": 1.9178,
+      "step": 3311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.762705082032812e-05,
+      "loss": 1.507,
+      "step": 3312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.758703481392558e-05,
+      "loss": 1.9508,
+      "step": 3313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.754701880752301e-05,
+      "loss": 1.5453,
+      "step": 3314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.750700280112045e-05,
+      "loss": 1.862,
+      "step": 3315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.746698679471789e-05,
+      "loss": 1.6331,
+      "step": 3316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.742697078831532e-05,
+      "loss": 1.5219,
+      "step": 3317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.738695478191277e-05,
+      "loss": 1.5972,
+      "step": 3318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.73469387755102e-05,
+      "loss": 1.642,
+      "step": 3319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.730692276910765e-05,
+      "loss": 1.5528,
+      "step": 3320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.726690676270509e-05,
+      "loss": 1.6279,
+      "step": 3321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.722689075630254e-05,
+      "loss": 1.9794,
+      "step": 3322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.718687474989996e-05,
+      "loss": 1.7592,
+      "step": 3323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.71468587434974e-05,
+      "loss": 1.7153,
+      "step": 3324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.710684273709485e-05,
+      "loss": 1.4091,
+      "step": 3325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.706682673069227e-05,
+      "loss": 1.8241,
+      "step": 3326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.702681072428972e-05,
+      "loss": 1.7736,
+      "step": 3327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.698679471788716e-05,
+      "loss": 1.7841,
+      "step": 3328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.69467787114846e-05,
+      "loss": 2.0863,
+      "step": 3329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.690676270508203e-05,
+      "loss": 1.7493,
+      "step": 3330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.686674669867947e-05,
+      "loss": 1.5512,
+      "step": 3331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.682673069227692e-05,
+      "loss": 1.4789,
+      "step": 3332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.678671468587436e-05,
+      "loss": 1.7141,
+      "step": 3333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.67466986794718e-05,
+      "loss": 1.908,
+      "step": 3334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.670668267306923e-05,
+      "loss": 1.6763,
+      "step": 3335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 2.1576,
+      "step": 3336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.662665066026411e-05,
+      "loss": 2.3664,
+      "step": 3337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.658663465386154e-05,
+      "loss": 1.6349,
+      "step": 3338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.6546618647459e-05,
+      "loss": 1.3311,
+      "step": 3339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.650660264105643e-05,
+      "loss": 2.2904,
+      "step": 3340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.646658663465386e-05,
+      "loss": 1.5524,
+      "step": 3341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.64265706282513e-05,
+      "loss": 1.6572,
+      "step": 3342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.638655462184874e-05,
+      "loss": 1.9933,
+      "step": 3343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.634653861544618e-05,
+      "loss": 2.1843,
+      "step": 3344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.630652260904362e-05,
+      "loss": 1.7118,
+      "step": 3345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.626650660264105e-05,
+      "loss": 2.134,
+      "step": 3346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.62264905962385e-05,
+      "loss": 1.894,
+      "step": 3347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.618647458983593e-05,
+      "loss": 1.6204,
+      "step": 3348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.614645858343338e-05,
+      "loss": 1.3999,
+      "step": 3349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.610644257703082e-05,
+      "loss": 1.6549,
+      "step": 3350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.606642657062825e-05,
+      "loss": 1.1168,
+      "step": 3351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.602641056422569e-05,
+      "loss": 1.4869,
+      "step": 3352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.598639455782313e-05,
+      "loss": 1.7871,
+      "step": 3353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.594637855142058e-05,
+      "loss": 2.0751,
+      "step": 3354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.5906362545018e-05,
+      "loss": 1.3811,
+      "step": 3355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.586634653861545e-05,
+      "loss": 1.5785,
+      "step": 3356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.582633053221289e-05,
+      "loss": 2.0689,
+      "step": 3357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.578631452581033e-05,
+      "loss": 1.3791,
+      "step": 3358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.574629851940776e-05,
+      "loss": 2.2557,
+      "step": 3359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.57062825130052e-05,
+      "loss": 1.6562,
+      "step": 3360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.566626650660265e-05,
+      "loss": 1.7399,
+      "step": 3361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.562625050020008e-05,
+      "loss": 1.6414,
+      "step": 3362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.558623449379753e-05,
+      "loss": 1.5321,
+      "step": 3363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.554621848739496e-05,
+      "loss": 2.002,
+      "step": 3364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.55062024809924e-05,
+      "loss": 1.7303,
+      "step": 3365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.546618647458984e-05,
+      "loss": 1.82,
+      "step": 3366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.542617046818728e-05,
+      "loss": 1.0234,
+      "step": 3367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.538615446178473e-05,
+      "loss": 2.3079,
+      "step": 3368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.534613845538215e-05,
+      "loss": 2.3297,
+      "step": 3369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.530612244897959e-05,
+      "loss": 1.7187,
+      "step": 3370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.526610644257704e-05,
+      "loss": 1.6308,
+      "step": 3371
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.522609043617447e-05,
+      "loss": 1.766,
+      "step": 3372
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.518607442977191e-05,
+      "loss": 1.3669,
+      "step": 3373
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.514605842336935e-05,
+      "loss": 1.5274,
+      "step": 3374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.51060424169668e-05,
+      "loss": 1.2573,
+      "step": 3375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.506602641056422e-05,
+      "loss": 1.8723,
+      "step": 3376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.502601040416166e-05,
+      "loss": 2.4557,
+      "step": 3377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.498599439775911e-05,
+      "loss": 1.8543,
+      "step": 3378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.494597839135655e-05,
+      "loss": 1.2696,
+      "step": 3379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.490596238495399e-05,
+      "loss": 1.4846,
+      "step": 3380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.486594637855142e-05,
+      "loss": 1.7739,
+      "step": 3381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.482593037214886e-05,
+      "loss": 1.9739,
+      "step": 3382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.47859143657463e-05,
+      "loss": 1.8918,
+      "step": 3383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.474589835934373e-05,
+      "loss": 2.0655,
+      "step": 3384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.470588235294118e-05,
+      "loss": 1.4381,
+      "step": 3385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.466586634653862e-05,
+      "loss": 2.2562,
+      "step": 3386
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.462585034013606e-05,
+      "loss": 1.4514,
+      "step": 3387
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.45858343337335e-05,
+      "loss": 1.7323,
+      "step": 3388
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.454581832733093e-05,
+      "loss": 1.793,
+      "step": 3389
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.450580232092838e-05,
+      "loss": 1.7992,
+      "step": 3390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.446578631452581e-05,
+      "loss": 1.6192,
+      "step": 3391
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.442577030812326e-05,
+      "loss": 1.7536,
+      "step": 3392
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.43857543017207e-05,
+      "loss": 2.6095,
+      "step": 3393
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.434573829531813e-05,
+      "loss": 1.8881,
+      "step": 3394
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.430572228891557e-05,
+      "loss": 1.6364,
+      "step": 3395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.4265706282513e-05,
+      "loss": 1.5519,
+      "step": 3396
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.422569027611046e-05,
+      "loss": 1.9729,
+      "step": 3397
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.418567426970788e-05,
+      "loss": 2.3354,
+      "step": 3398
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.414565826330533e-05,
+      "loss": 2.3285,
+      "step": 3399
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.410564225690277e-05,
+      "loss": 2.1445,
+      "step": 3400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.40656262505002e-05,
+      "loss": 1.4959,
+      "step": 3401
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.402561024409764e-05,
+      "loss": 1.5464,
+      "step": 3402
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.398559423769508e-05,
+      "loss": 1.6248,
+      "step": 3403
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.394557823129253e-05,
+      "loss": 2.1871,
+      "step": 3404
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.390556222488995e-05,
+      "loss": 2.035,
+      "step": 3405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.386554621848739e-05,
+      "loss": 1.6537,
+      "step": 3406
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.382553021208484e-05,
+      "loss": 1.9611,
+      "step": 3407
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.378551420568228e-05,
+      "loss": 1.8651,
+      "step": 3408
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.374549819927972e-05,
+      "loss": 2.0733,
+      "step": 3409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.370548219287715e-05,
+      "loss": 1.8705,
+      "step": 3410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.36654661864746e-05,
+      "loss": 1.7605,
+      "step": 3411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.362545018007203e-05,
+      "loss": 1.5367,
+      "step": 3412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.358543417366946e-05,
+      "loss": 2.3281,
+      "step": 3413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.354541816726692e-05,
+      "loss": 2.0896,
+      "step": 3414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.350540216086435e-05,
+      "loss": 1.909,
+      "step": 3415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.346538615446179e-05,
+      "loss": 2.478,
+      "step": 3416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.342537014805923e-05,
+      "loss": 1.9208,
+      "step": 3417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.338535414165666e-05,
+      "loss": 1.8705,
+      "step": 3418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.33453381352541e-05,
+      "loss": 2.0685,
+      "step": 3419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.330532212885154e-05,
+      "loss": 1.7176,
+      "step": 3420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.326530612244899e-05,
+      "loss": 2.3859,
+      "step": 3421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.322529011604643e-05,
+      "loss": 1.491,
+      "step": 3422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.318527410964386e-05,
+      "loss": 1.5056,
+      "step": 3423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.31452581032413e-05,
+      "loss": 2.1067,
+      "step": 3424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.310524209683874e-05,
+      "loss": 1.5338,
+      "step": 3425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.306522609043617e-05,
+      "loss": 1.7431,
+      "step": 3426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.302521008403361e-05,
+      "loss": 1.5811,
+      "step": 3427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.298519407763106e-05,
+      "loss": 1.343,
+      "step": 3428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.29451780712285e-05,
+      "loss": 1.6348,
+      "step": 3429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.290516206482592e-05,
+      "loss": 1.9769,
+      "step": 3430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.286514605842337e-05,
+      "loss": 2.1886,
+      "step": 3431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.282513005202081e-05,
+      "loss": 1.5776,
+      "step": 3432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.278511404561825e-05,
+      "loss": 1.9559,
+      "step": 3433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.274509803921569e-05,
+      "loss": 2.1452,
+      "step": 3434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.270508203281312e-05,
+      "loss": 1.5557,
+      "step": 3435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.266506602641057e-05,
+      "loss": 1.4725,
+      "step": 3436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.2625050020008e-05,
+      "loss": 1.6314,
+      "step": 3437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.258503401360545e-05,
+      "loss": 1.5612,
+      "step": 3438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.254501800720288e-05,
+      "loss": 1.5627,
+      "step": 3439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.250500200080032e-05,
+      "loss": 1.6288,
+      "step": 3440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.246498599439776e-05,
+      "loss": 1.9507,
+      "step": 3441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.24249699879952e-05,
+      "loss": 1.7135,
+      "step": 3442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.238495398159265e-05,
+      "loss": 1.9563,
+      "step": 3443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.234493797519007e-05,
+      "loss": 2.1406,
+      "step": 3444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.230492196878752e-05,
+      "loss": 2.2864,
+      "step": 3445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.226490596238496e-05,
+      "loss": 1.3382,
+      "step": 3446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.22248899559824e-05,
+      "loss": 1.662,
+      "step": 3447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.218487394957983e-05,
+      "loss": 1.6561,
+      "step": 3448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.214485794317727e-05,
+      "loss": 1.6684,
+      "step": 3449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.210484193677472e-05,
+      "loss": 0.9787,
+      "step": 3450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.206482593037214e-05,
+      "loss": 1.811,
+      "step": 3451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.20248099239696e-05,
+      "loss": 1.7964,
+      "step": 3452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.198479391756703e-05,
+      "loss": 2.0346,
+      "step": 3453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.194477791116447e-05,
+      "loss": 1.2327,
+      "step": 3454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.19047619047619e-05,
+      "loss": 1.7548,
+      "step": 3455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.186474589835934e-05,
+      "loss": 1.7363,
+      "step": 3456
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.18247298919568e-05,
+      "loss": 1.8233,
+      "step": 3457
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.178471388555423e-05,
+      "loss": 2.236,
+      "step": 3458
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.174469787915165e-05,
+      "loss": 1.9892,
+      "step": 3459
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.17046818727491e-05,
+      "loss": 1.8629,
+      "step": 3460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.166466586634654e-05,
+      "loss": 2.1663,
+      "step": 3461
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.162464985994398e-05,
+      "loss": 1.8318,
+      "step": 3462
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.158463385354142e-05,
+      "loss": 1.5273,
+      "step": 3463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.154461784713887e-05,
+      "loss": 1.3967,
+      "step": 3464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.15046018407363e-05,
+      "loss": 1.2622,
+      "step": 3465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.146458583433373e-05,
+      "loss": 1.7292,
+      "step": 3466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.142456982793118e-05,
+      "loss": 1.5289,
+      "step": 3467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.138455382152862e-05,
+      "loss": 1.6892,
+      "step": 3468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.134453781512605e-05,
+      "loss": 1.4981,
+      "step": 3469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.130452180872349e-05,
+      "loss": 1.4158,
+      "step": 3470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.126450580232093e-05,
+      "loss": 1.8064,
+      "step": 3471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.122448979591838e-05,
+      "loss": 1.7198,
+      "step": 3472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.11844737895158e-05,
+      "loss": 1.7957,
+      "step": 3473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.114445778311325e-05,
+      "loss": 2.0193,
+      "step": 3474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.110444177671069e-05,
+      "loss": 2.0388,
+      "step": 3475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.106442577030813e-05,
+      "loss": 1.7265,
+      "step": 3476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.1024409763905563e-05,
+      "loss": 1.3332,
+      "step": 3477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.0984393757503e-05,
+      "loss": 1.4504,
+      "step": 3478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.0944377751100444e-05,
+      "loss": 1.609,
+      "step": 3479
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.090436174469788e-05,
+      "loss": 1.5065,
+      "step": 3480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.0864345738295326e-05,
+      "loss": 2.2281,
+      "step": 3481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.082432973189276e-05,
+      "loss": 1.6207,
+      "step": 3482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.078431372549019e-05,
+      "loss": 1.6465,
+      "step": 3483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.074429771908764e-05,
+      "loss": 1.5316,
+      "step": 3484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.0704281712685074e-05,
+      "loss": 1.7815,
+      "step": 3485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.066426570628252e-05,
+      "loss": 1.6741,
+      "step": 3486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.0624249699879955e-05,
+      "loss": 1.6993,
+      "step": 3487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.05842336934774e-05,
+      "loss": 2.0503,
+      "step": 3488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.0544217687074836e-05,
+      "loss": 1.9808,
+      "step": 3489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.0504201680672267e-05,
+      "loss": 1.9384,
+      "step": 3490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.046418567426971e-05,
+      "loss": 1.9336,
+      "step": 3491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.042416966786715e-05,
+      "loss": 1.6937,
+      "step": 3492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.038415366146459e-05,
+      "loss": 1.885,
+      "step": 3493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.034413765506203e-05,
+      "loss": 2.0397,
+      "step": 3494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.030412164865946e-05,
+      "loss": 1.8365,
+      "step": 3495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.026410564225691e-05,
+      "loss": 1.5679,
+      "step": 3496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.022408963585434e-05,
+      "loss": 1.6189,
+      "step": 3497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.0184073629451784e-05,
+      "loss": 1.9482,
+      "step": 3498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.014405762304922e-05,
+      "loss": 2.0111,
+      "step": 3499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.0104041616646665e-05,
+      "loss": 1.7539,
+      "step": 3500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.00640256102441e-05,
+      "loss": 1.7993,
+      "step": 3501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.002400960384153e-05,
+      "loss": 1.5488,
+      "step": 3502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.998399359743898e-05,
+      "loss": 1.442,
+      "step": 3503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9943977591036414e-05,
+      "loss": 1.7888,
+      "step": 3504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9903961584633864e-05,
+      "loss": 1.6851,
+      "step": 3505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9863945578231295e-05,
+      "loss": 1.5338,
+      "step": 3506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.982392957182873e-05,
+      "loss": 1.3663,
+      "step": 3507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9783913565426176e-05,
+      "loss": 1.3808,
+      "step": 3508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.974389755902361e-05,
+      "loss": 1.4231,
+      "step": 3509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.970388155262106e-05,
+      "loss": 1.6434,
+      "step": 3510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.966386554621849e-05,
+      "loss": 1.2911,
+      "step": 3511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.962384953981594e-05,
+      "loss": 1.6351,
+      "step": 3512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.958383353341337e-05,
+      "loss": 2.2604,
+      "step": 3513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9543817527010805e-05,
+      "loss": 1.5187,
+      "step": 3514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.950380152060825e-05,
+      "loss": 1.8937,
+      "step": 3515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9463785514205686e-05,
+      "loss": 1.2547,
+      "step": 3516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.942376950780313e-05,
+      "loss": 1.2169,
+      "step": 3517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.938375350140056e-05,
+      "loss": 1.1438,
+      "step": 3518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9343737494998e-05,
+      "loss": 1.928,
+      "step": 3519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.930372148859544e-05,
+      "loss": 2.295,
+      "step": 3520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.926370548219288e-05,
+      "loss": 1.7979,
+      "step": 3521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.922368947579032e-05,
+      "loss": 1.8967,
+      "step": 3522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.918367346938776e-05,
+      "loss": 1.6799,
+      "step": 3523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.914365746298519e-05,
+      "loss": 1.8901,
+      "step": 3524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9103641456582634e-05,
+      "loss": 1.981,
+      "step": 3525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.906362545018007e-05,
+      "loss": 1.9395,
+      "step": 3526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9023609443777515e-05,
+      "loss": 1.456,
+      "step": 3527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.898359343737495e-05,
+      "loss": 1.5059,
+      "step": 3528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.8943577430972396e-05,
+      "loss": 1.4164,
+      "step": 3529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.890356142456983e-05,
+      "loss": 1.5667,
+      "step": 3530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.8863545418167264e-05,
+      "loss": 1.4978,
+      "step": 3531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.882352941176471e-05,
+      "loss": 1.7412,
+      "step": 3532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.8783513405362145e-05,
+      "loss": 1.4339,
+      "step": 3533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.874349739895959e-05,
+      "loss": 1.783,
+      "step": 3534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.8703481392557026e-05,
+      "loss": 1.3939,
+      "step": 3535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.8663465386154456e-05,
+      "loss": 1.7668,
+      "step": 3536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.862344937975191e-05,
+      "loss": 2.0505,
+      "step": 3537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.858343337334934e-05,
+      "loss": 2.1819,
+      "step": 3538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.854341736694679e-05,
+      "loss": 1.6504,
+      "step": 3539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.850340136054422e-05,
+      "loss": 1.6576,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.846338535414166e-05,
+      "loss": 1.5409,
+      "step": 3541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.84233693477391e-05,
+      "loss": 1.8343,
+      "step": 3542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.8383353341336536e-05,
+      "loss": 1.9341,
+      "step": 3543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.834333733493398e-05,
+      "loss": 1.8387,
+      "step": 3544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.830332132853141e-05,
+      "loss": 1.8387,
+      "step": 3545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.826330532212886e-05,
+      "loss": 1.7631,
+      "step": 3546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.822328931572629e-05,
+      "loss": 1.6187,
+      "step": 3547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.818327330932373e-05,
+      "loss": 1.821,
+      "step": 3548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.814325730292117e-05,
+      "loss": 1.7911,
+      "step": 3549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.810324129651861e-05,
+      "loss": 1.4369,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.8063225290116054e-05,
+      "loss": 1.9245,
+      "step": 3551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.8023209283713484e-05,
+      "loss": 1.4465,
+      "step": 3552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.7983193277310935e-05,
+      "loss": 1.6022,
+      "step": 3553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.7943177270908365e-05,
+      "loss": 2.0645,
+      "step": 3554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.79031612645058e-05,
+      "loss": 1.49,
+      "step": 3555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.7863145258103246e-05,
+      "loss": 1.6488,
+      "step": 3556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.782312925170068e-05,
+      "loss": 1.6774,
+      "step": 3557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.778311324529813e-05,
+      "loss": 1.7126,
+      "step": 3558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.774309723889556e-05,
+      "loss": 1.9971,
+      "step": 3559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.7703081232492995e-05,
+      "loss": 1.3552,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.766306522609044e-05,
+      "loss": 1.5319,
+      "step": 3561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.7623049219687876e-05,
+      "loss": 2.1821,
+      "step": 3562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.758303321328532e-05,
+      "loss": 1.3069,
+      "step": 3563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.754301720688276e-05,
+      "loss": 1.2582,
+      "step": 3564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.75030012004802e-05,
+      "loss": 1.8219,
+      "step": 3565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.746298519407763e-05,
+      "loss": 1.7165,
+      "step": 3566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.742296918767507e-05,
+      "loss": 2.0864,
+      "step": 3567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.738295318127251e-05,
+      "loss": 1.8982,
+      "step": 3568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.734293717486995e-05,
+      "loss": 2.0711,
+      "step": 3569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.730292116846739e-05,
+      "loss": 1.6048,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.726290516206483e-05,
+      "loss": 1.9641,
+      "step": 3571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.722288915566226e-05,
+      "loss": 1.5676,
+      "step": 3572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.718287314925971e-05,
+      "loss": 1.4714,
+      "step": 3573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.714285714285714e-05,
+      "loss": 1.7418,
+      "step": 3574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.7102841136454586e-05,
+      "loss": 1.316,
+      "step": 3575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.706282513005202e-05,
+      "loss": 1.6655,
+      "step": 3576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.702280912364947e-05,
+      "loss": 1.3566,
+      "step": 3577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6982793117246904e-05,
+      "loss": 1.6823,
+      "step": 3578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6942777110844334e-05,
+      "loss": 2.0708,
+      "step": 3579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6902761104441785e-05,
+      "loss": 1.8651,
+      "step": 3580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6862745098039215e-05,
+      "loss": 1.5207,
+      "step": 3581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.682272909163666e-05,
+      "loss": 1.7867,
+      "step": 3582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6782713085234096e-05,
+      "loss": 1.7618,
+      "step": 3583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6742697078831533e-05,
+      "loss": 2.0591,
+      "step": 3584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.670268107242898e-05,
+      "loss": 1.7571,
+      "step": 3585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.666266506602641e-05,
+      "loss": 1.7397,
+      "step": 3586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.662264905962386e-05,
+      "loss": 1.3345,
+      "step": 3587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.658263305322129e-05,
+      "loss": 1.9401,
+      "step": 3588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.654261704681873e-05,
+      "loss": 1.4684,
+      "step": 3589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.650260104041617e-05,
+      "loss": 1.9498,
+      "step": 3590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.646258503401361e-05,
+      "loss": 1.4412,
+      "step": 3591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.642256902761105e-05,
+      "loss": 1.8398,
+      "step": 3592
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.638255302120848e-05,
+      "loss": 1.7318,
+      "step": 3593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.634253701480593e-05,
+      "loss": 1.8205,
+      "step": 3594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.630252100840336e-05,
+      "loss": 1.8068,
+      "step": 3595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.62625050020008e-05,
+      "loss": 1.5914,
+      "step": 3596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.622248899559824e-05,
+      "loss": 1.6638,
+      "step": 3597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.618247298919568e-05,
+      "loss": 1.9717,
+      "step": 3598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6142456982793124e-05,
+      "loss": 2.1996,
+      "step": 3599
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6102440976390555e-05,
+      "loss": 1.8591,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6062424969988005e-05,
+      "loss": 1.5111,
+      "step": 3601
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6022408963585436e-05,
+      "loss": 1.8057,
+      "step": 3602
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.598239295718287e-05,
+      "loss": 1.7567,
+      "step": 3603
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.594237695078032e-05,
+      "loss": 1.3133,
+      "step": 3604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.5902360944377754e-05,
+      "loss": 1.7093,
+      "step": 3605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.58623449379752e-05,
+      "loss": 1.8209,
+      "step": 3606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.5822328931572635e-05,
+      "loss": 1.4112,
+      "step": 3607
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.5782312925170065e-05,
+      "loss": 2.1281,
+      "step": 3608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.574229691876751e-05,
+      "loss": 2.1163,
+      "step": 3609
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.5702280912364946e-05,
+      "loss": 1.6245,
+      "step": 3610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.566226490596239e-05,
+      "loss": 1.6805,
+      "step": 3611
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.562224889955983e-05,
+      "loss": 1.587,
+      "step": 3612
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.558223289315727e-05,
+      "loss": 1.9251,
+      "step": 3613
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.554221688675471e-05,
+      "loss": 1.7151,
+      "step": 3614
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.550220088035214e-05,
+      "loss": 1.939,
+      "step": 3615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.546218487394958e-05,
+      "loss": 1.684,
+      "step": 3616
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.542216886754702e-05,
+      "loss": 2.0827,
+      "step": 3617
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.5382152861144464e-05,
+      "loss": 1.7361,
+      "step": 3618
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.53421368547419e-05,
+      "loss": 1.9326,
+      "step": 3619
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.530212084833933e-05,
+      "loss": 1.4815,
+      "step": 3620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.526210484193678e-05,
+      "loss": 1.6656,
+      "step": 3621
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.522208883553421e-05,
+      "loss": 1.5965,
+      "step": 3622
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.5182072829131656e-05,
+      "loss": 1.9435,
+      "step": 3623
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.514205682272909e-05,
+      "loss": 1.6593,
+      "step": 3624
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.510204081632653e-05,
+      "loss": 1.796,
+      "step": 3625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.5062024809923974e-05,
+      "loss": 1.5885,
+      "step": 3626
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.5022008803521405e-05,
+      "loss": 1.7885,
+      "step": 3627
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.4981992797118855e-05,
+      "loss": 1.286,
+      "step": 3628
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.4941976790716286e-05,
+      "loss": 1.6065,
+      "step": 3629
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.490196078431373e-05,
+      "loss": 1.5205,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.486194477791117e-05,
+      "loss": 2.0181,
+      "step": 3631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.4821928771508604e-05,
+      "loss": 1.3645,
+      "step": 3632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.478191276510605e-05,
+      "loss": 1.6622,
+      "step": 3633
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.474189675870348e-05,
+      "loss": 1.5668,
+      "step": 3634
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.470188075230093e-05,
+      "loss": 1.8714,
+      "step": 3635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.466186474589836e-05,
+      "loss": 1.764,
+      "step": 3636
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.4621848739495796e-05,
+      "loss": 1.823,
+      "step": 3637
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.458183273309324e-05,
+      "loss": 1.9476,
+      "step": 3638
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.454181672669068e-05,
+      "loss": 1.94,
+      "step": 3639
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.450180072028812e-05,
+      "loss": 1.8545,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.446178471388556e-05,
+      "loss": 1.3555,
+      "step": 3641
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.4421768707483e-05,
+      "loss": 1.2142,
+      "step": 3642
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.438175270108043e-05,
+      "loss": 1.91,
+      "step": 3643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.434173669467787e-05,
+      "loss": 1.4195,
+      "step": 3644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.4301720688275314e-05,
+      "loss": 1.5747,
+      "step": 3645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.426170468187275e-05,
+      "loss": 1.2399,
+      "step": 3646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.4221688675470195e-05,
+      "loss": 1.1488,
+      "step": 3647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.418167266906763e-05,
+      "loss": 1.3731,
+      "step": 3648
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.414165666266506e-05,
+      "loss": 1.821,
+      "step": 3649
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.4101640656262506e-05,
+      "loss": 1.4241,
+      "step": 3650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.4061624649859943e-05,
+      "loss": 1.6586,
+      "step": 3651
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.402160864345739e-05,
+      "loss": 2.2401,
+      "step": 3652
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3981592637054824e-05,
+      "loss": 1.6882,
+      "step": 3653
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.394157663065227e-05,
+      "loss": 1.5116,
+      "step": 3654
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3901560624249706e-05,
+      "loss": 2.3417,
+      "step": 3655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3861544617847136e-05,
+      "loss": 1.3501,
+      "step": 3656
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.382152861144458e-05,
+      "loss": 1.8844,
+      "step": 3657
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.378151260504202e-05,
+      "loss": 1.5269,
+      "step": 3658
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.374149659863946e-05,
+      "loss": 1.6643,
+      "step": 3659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.37014805922369e-05,
+      "loss": 1.3261,
+      "step": 3660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.366146458583433e-05,
+      "loss": 1.3992,
+      "step": 3661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.362144857943178e-05,
+      "loss": 1.3189,
+      "step": 3662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.358143257302921e-05,
+      "loss": 1.8417,
+      "step": 3663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.354141656662665e-05,
+      "loss": 1.8398,
+      "step": 3664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.350140056022409e-05,
+      "loss": 1.6184,
+      "step": 3665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3461384553821534e-05,
+      "loss": 1.9903,
+      "step": 3666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.342136854741897e-05,
+      "loss": 1.6472,
+      "step": 3667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.33813525410164e-05,
+      "loss": 2.0762,
+      "step": 3668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.334133653461385e-05,
+      "loss": 1.6645,
+      "step": 3669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.330132052821128e-05,
+      "loss": 1.7072,
+      "step": 3670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3261304521808734e-05,
+      "loss": 1.799,
+      "step": 3671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3221288515406164e-05,
+      "loss": 1.2732,
+      "step": 3672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.31812725090036e-05,
+      "loss": 1.9824,
+      "step": 3673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3141256502601045e-05,
+      "loss": 1.9905,
+      "step": 3674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.310124049619848e-05,
+      "loss": 2.07,
+      "step": 3675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3061224489795926e-05,
+      "loss": 1.2534,
+      "step": 3676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3021208483393356e-05,
+      "loss": 1.7273,
+      "step": 3677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.298119247699081e-05,
+      "loss": 1.5659,
+      "step": 3678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.294117647058824e-05,
+      "loss": 1.6603,
+      "step": 3679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.2901160464185675e-05,
+      "loss": 1.6413,
+      "step": 3680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.286114445778312e-05,
+      "loss": 1.4025,
+      "step": 3681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.2821128451380556e-05,
+      "loss": 2.0953,
+      "step": 3682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.2781112444978e-05,
+      "loss": 1.2979,
+      "step": 3683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.274109643857543e-05,
+      "loss": 1.8361,
+      "step": 3684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.270108043217287e-05,
+      "loss": 1.8857,
+      "step": 3685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.266106442577031e-05,
+      "loss": 1.4579,
+      "step": 3686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.262104841936775e-05,
+      "loss": 1.9155,
+      "step": 3687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.258103241296519e-05,
+      "loss": 1.7218,
+      "step": 3688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.254101640656263e-05,
+      "loss": 1.7084,
+      "step": 3689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.250100040016007e-05,
+      "loss": 1.8065,
+      "step": 3690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.24609843937575e-05,
+      "loss": 1.7956,
+      "step": 3691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.242096838735494e-05,
+      "loss": 1.7667,
+      "step": 3692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.2380952380952384e-05,
+      "loss": 1.3728,
+      "step": 3693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.234093637454982e-05,
+      "loss": 1.4801,
+      "step": 3694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.2300920368147265e-05,
+      "loss": 1.8297,
+      "step": 3695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.22609043617447e-05,
+      "loss": 1.6781,
+      "step": 3696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.222088835534213e-05,
+      "loss": 2.1389,
+      "step": 3697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.218087234893958e-05,
+      "loss": 1.2863,
+      "step": 3698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.2140856342537014e-05,
+      "loss": 1.6622,
+      "step": 3699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.210084033613446e-05,
+      "loss": 1.5637,
+      "step": 3700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.2060824329731895e-05,
+      "loss": 1.6148,
+      "step": 3701
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.202080832332934e-05,
+      "loss": 1.4813,
+      "step": 3702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.1980792316926776e-05,
+      "loss": 1.457,
+      "step": 3703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.1940776310524206e-05,
+      "loss": 1.3734,
+      "step": 3704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.190076030412166e-05,
+      "loss": 1.6263,
+      "step": 3705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.186074429771909e-05,
+      "loss": 1.5604,
+      "step": 3706
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.182072829131653e-05,
+      "loss": 1.7454,
+      "step": 3707
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.178071228491397e-05,
+      "loss": 1.3591,
+      "step": 3708
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.1740696278511406e-05,
+      "loss": 1.5752,
+      "step": 3709
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.170068027210885e-05,
+      "loss": 1.7759,
+      "step": 3710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.166066426570628e-05,
+      "loss": 1.9175,
+      "step": 3711
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.162064825930373e-05,
+      "loss": 1.8554,
+      "step": 3712
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.158063225290116e-05,
+      "loss": 1.9691,
+      "step": 3713
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.15406162464986e-05,
+      "loss": 1.6648,
+      "step": 3714
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.150060024009604e-05,
+      "loss": 1.9014,
+      "step": 3715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.146058423369348e-05,
+      "loss": 1.4553,
+      "step": 3716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.142056822729092e-05,
+      "loss": 1.742,
+      "step": 3717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.1380552220888353e-05,
+      "loss": 1.8004,
+      "step": 3718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.1340536214485804e-05,
+      "loss": 1.5234,
+      "step": 3719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.1300520208083235e-05,
+      "loss": 1.798,
+      "step": 3720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.126050420168067e-05,
+      "loss": 1.7335,
+      "step": 3721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.1220488195278116e-05,
+      "loss": 1.4471,
+      "step": 3722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.118047218887555e-05,
+      "loss": 1.5395,
+      "step": 3723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.1140456182472997e-05,
+      "loss": 1.4199,
+      "step": 3724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.110044017607043e-05,
+      "loss": 2.2852,
+      "step": 3725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.1060424169667864e-05,
+      "loss": 1.7981,
+      "step": 3726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.102040816326531e-05,
+      "loss": 1.6894,
+      "step": 3727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.0980392156862745e-05,
+      "loss": 1.5862,
+      "step": 3728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.094037615046019e-05,
+      "loss": 1.6903,
+      "step": 3729
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.0900360144057626e-05,
+      "loss": 1.3671,
+      "step": 3730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.086034413765507e-05,
+      "loss": 1.1699,
+      "step": 3731
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.08203281312525e-05,
+      "loss": 2.0287,
+      "step": 3732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.078031212484994e-05,
+      "loss": 1.6186,
+      "step": 3733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.074029611844738e-05,
+      "loss": 1.6262,
+      "step": 3734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.070028011204482e-05,
+      "loss": 2.0649,
+      "step": 3735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.066026410564226e-05,
+      "loss": 1.6984,
+      "step": 3736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.06202480992397e-05,
+      "loss": 1.6439,
+      "step": 3737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.058023209283713e-05,
+      "loss": 1.6401,
+      "step": 3738
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.054021608643458e-05,
+      "loss": 1.6097,
+      "step": 3739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.050020008003201e-05,
+      "loss": 1.7994,
+      "step": 3740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.0460184073629455e-05,
+      "loss": 1.8445,
+      "step": 3741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.042016806722689e-05,
+      "loss": 1.6368,
+      "step": 3742
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.0380152060824336e-05,
+      "loss": 1.5529,
+      "step": 3743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.034013605442177e-05,
+      "loss": 1.8948,
+      "step": 3744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.0300120048019204e-05,
+      "loss": 1.6164,
+      "step": 3745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.0260104041616654e-05,
+      "loss": 2.0541,
+      "step": 3746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.0220088035214085e-05,
+      "loss": 2.0053,
+      "step": 3747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.018007202881153e-05,
+      "loss": 1.6299,
+      "step": 3748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.0140056022408966e-05,
+      "loss": 1.8939,
+      "step": 3749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.01000400160064e-05,
+      "loss": 1.846,
+      "step": 3750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.006002400960385e-05,
+      "loss": 1.5487,
+      "step": 3751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.002000800320128e-05,
+      "loss": 1.6179,
+      "step": 3752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.997999199679872e-05,
+      "loss": 1.4304,
+      "step": 3753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.993997599039616e-05,
+      "loss": 1.9437,
+      "step": 3754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.98999599839936e-05,
+      "loss": 1.7324,
+      "step": 3755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.985994397759104e-05,
+      "loss": 1.1707,
+      "step": 3756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9819927971188476e-05,
+      "loss": 1.4272,
+      "step": 3757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.977991196478592e-05,
+      "loss": 1.4632,
+      "step": 3758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.973989595838335e-05,
+      "loss": 2.0022,
+      "step": 3759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9699879951980794e-05,
+      "loss": 2.2567,
+      "step": 3760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.965986394557823e-05,
+      "loss": 1.2029,
+      "step": 3761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9619847939175676e-05,
+      "loss": 1.6803,
+      "step": 3762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.957983193277311e-05,
+      "loss": 1.9359,
+      "step": 3763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.953981592637055e-05,
+      "loss": 1.6822,
+      "step": 3764
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.949979991996799e-05,
+      "loss": 1.473,
+      "step": 3765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9459783913565424e-05,
+      "loss": 2.193,
+      "step": 3766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.941976790716287e-05,
+      "loss": 2.0255,
+      "step": 3767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9379751900760305e-05,
+      "loss": 2.0923,
+      "step": 3768
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.933973589435775e-05,
+      "loss": 2.3151,
+      "step": 3769
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9299719887955186e-05,
+      "loss": 1.7225,
+      "step": 3770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.925970388155262e-05,
+      "loss": 1.6138,
+      "step": 3771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.921968787515006e-05,
+      "loss": 1.4176,
+      "step": 3772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9179671868747504e-05,
+      "loss": 2.0444,
+      "step": 3773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.913965586234494e-05,
+      "loss": 1.9335,
+      "step": 3774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.909963985594238e-05,
+      "loss": 1.5121,
+      "step": 3775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.905962384953982e-05,
+      "loss": 1.4435,
+      "step": 3776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.901960784313725e-05,
+      "loss": 1.9477,
+      "step": 3777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.89795918367347e-05,
+      "loss": 1.7467,
+      "step": 3778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8939575830332134e-05,
+      "loss": 1.8358,
+      "step": 3779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.889955982392958e-05,
+      "loss": 1.3269,
+      "step": 3780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8859543817527015e-05,
+      "loss": 2.0854,
+      "step": 3781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.881952781112445e-05,
+      "loss": 1.7582,
+      "step": 3782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.877951180472189e-05,
+      "loss": 1.3908,
+      "step": 3783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8739495798319326e-05,
+      "loss": 0.9721,
+      "step": 3784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.869947979191677e-05,
+      "loss": 1.0548,
+      "step": 3785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.865946378551421e-05,
+      "loss": 1.6041,
+      "step": 3786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.861944777911165e-05,
+      "loss": 1.5285,
+      "step": 3787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.857943177270909e-05,
+      "loss": 1.8699,
+      "step": 3788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8539415766306526e-05,
+      "loss": 1.6414,
+      "step": 3789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.849939975990396e-05,
+      "loss": 1.6748,
+      "step": 3790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.84593837535014e-05,
+      "loss": 1.1463,
+      "step": 3791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8419367747098844e-05,
+      "loss": 1.5278,
+      "step": 3792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.837935174069628e-05,
+      "loss": 1.7624,
+      "step": 3793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8339335734293725e-05,
+      "loss": 1.9941,
+      "step": 3794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8299319727891155e-05,
+      "loss": 1.9494,
+      "step": 3795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.82593037214886e-05,
+      "loss": 0.9393,
+      "step": 3796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8219287715086036e-05,
+      "loss": 1.6444,
+      "step": 3797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.817927170868347e-05,
+      "loss": 1.3226,
+      "step": 3798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.813925570228092e-05,
+      "loss": 2.0714,
+      "step": 3799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.809923969587835e-05,
+      "loss": 1.631,
+      "step": 3800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.805922368947579e-05,
+      "loss": 1.3725,
+      "step": 3801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.801920768307323e-05,
+      "loss": 1.4047,
+      "step": 3802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.797919167667067e-05,
+      "loss": 1.9749,
+      "step": 3803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.793917567026811e-05,
+      "loss": 2.084,
+      "step": 3804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7899159663865554e-05,
+      "loss": 2.2226,
+      "step": 3805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7859143657462984e-05,
+      "loss": 1.4848,
+      "step": 3806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.781912765106043e-05,
+      "loss": 1.1829,
+      "step": 3807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7779111644657865e-05,
+      "loss": 1.0437,
+      "step": 3808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.77390956382553e-05,
+      "loss": 1.7236,
+      "step": 3809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7699079631852746e-05,
+      "loss": 1.7497,
+      "step": 3810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.765906362545018e-05,
+      "loss": 1.6608,
+      "step": 3811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.761904761904762e-05,
+      "loss": 1.4797,
+      "step": 3812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.757903161264506e-05,
+      "loss": 1.3908,
+      "step": 3813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.75390156062425e-05,
+      "loss": 1.6329,
+      "step": 3814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.749899959983994e-05,
+      "loss": 1.614,
+      "step": 3815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7458983593437376e-05,
+      "loss": 1.4774,
+      "step": 3816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.741896758703482e-05,
+      "loss": 1.7059,
+      "step": 3817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.737895158063225e-05,
+      "loss": 1.9049,
+      "step": 3818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7338935574229694e-05,
+      "loss": 1.9829,
+      "step": 3819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.729891956782713e-05,
+      "loss": 1.6856,
+      "step": 3820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7258903561424575e-05,
+      "loss": 1.5659,
+      "step": 3821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.721888755502201e-05,
+      "loss": 1.3061,
+      "step": 3822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.717887154861945e-05,
+      "loss": 1.7599,
+      "step": 3823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7138855542216886e-05,
+      "loss": 1.5858,
+      "step": 3824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7098839535814323e-05,
+      "loss": 2.1028,
+      "step": 3825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.705882352941177e-05,
+      "loss": 2.1577,
+      "step": 3826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7018807523009204e-05,
+      "loss": 1.6428,
+      "step": 3827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.697879151660665e-05,
+      "loss": 1.388,
+      "step": 3828
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6938775510204086e-05,
+      "loss": 1.7826,
+      "step": 3829
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.689875950380152e-05,
+      "loss": 2.2224,
+      "step": 3830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.685874349739896e-05,
+      "loss": 1.4341,
+      "step": 3831
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.68187274909964e-05,
+      "loss": 1.8942,
+      "step": 3832
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.677871148459384e-05,
+      "loss": 1.5221,
+      "step": 3833
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.673869547819128e-05,
+      "loss": 1.8774,
+      "step": 3834
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.669867947178872e-05,
+      "loss": 2.2258,
+      "step": 3835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.665866346538615e-05,
+      "loss": 1.7068,
+      "step": 3836
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6618647458983596e-05,
+      "loss": 1.7842,
+      "step": 3837
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.657863145258103e-05,
+      "loss": 1.5377,
+      "step": 3838
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.653861544617848e-05,
+      "loss": 2.0249,
+      "step": 3839
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6498599439775914e-05,
+      "loss": 1.9001,
+      "step": 3840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.645858343337335e-05,
+      "loss": 1.7219,
+      "step": 3841
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.641856742697079e-05,
+      "loss": 1.753,
+      "step": 3842
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6378551420568226e-05,
+      "loss": 1.3166,
+      "step": 3843
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.633853541416567e-05,
+      "loss": 1.647,
+      "step": 3844
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.629851940776311e-05,
+      "loss": 1.7932,
+      "step": 3845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.625850340136055e-05,
+      "loss": 1.7332,
+      "step": 3846
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.621848739495799e-05,
+      "loss": 1.9963,
+      "step": 3847
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6178471388555425e-05,
+      "loss": 1.692,
+      "step": 3848
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.613845538215286e-05,
+      "loss": 1.9463,
+      "step": 3849
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.60984393757503e-05,
+      "loss": 1.6494,
+      "step": 3850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.605842336934774e-05,
+      "loss": 2.0496,
+      "step": 3851
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.601840736294518e-05,
+      "loss": 1.9383,
+      "step": 3852
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.5978391356542624e-05,
+      "loss": 1.2712,
+      "step": 3853
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.5938375350140055e-05,
+      "loss": 1.4954,
+      "step": 3854
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.58983593437375e-05,
+      "loss": 1.6432,
+      "step": 3855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.5858343337334936e-05,
+      "loss": 1.7312,
+      "step": 3856
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.581832733093237e-05,
+      "loss": 1.4394,
+      "step": 3857
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.577831132452982e-05,
+      "loss": 1.6556,
+      "step": 3858
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.5738295318127254e-05,
+      "loss": 1.4462,
+      "step": 3859
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.569827931172469e-05,
+      "loss": 2.0749,
+      "step": 3860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.565826330532213e-05,
+      "loss": 1.466,
+      "step": 3861
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.561824729891957e-05,
+      "loss": 2.1378,
+      "step": 3862
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.557823129251701e-05,
+      "loss": 1.4544,
+      "step": 3863
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.5538215286114446e-05,
+      "loss": 2.3964,
+      "step": 3864
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.549819927971189e-05,
+      "loss": 1.4001,
+      "step": 3865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.545818327330932e-05,
+      "loss": 2.1999,
+      "step": 3866
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.5418167266906764e-05,
+      "loss": 1.4937,
+      "step": 3867
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.53781512605042e-05,
+      "loss": 1.8382,
+      "step": 3868
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.5338135254101645e-05,
+      "loss": 1.7634,
+      "step": 3869
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.529811924769908e-05,
+      "loss": 1.9311,
+      "step": 3870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.5258103241296527e-05,
+      "loss": 1.5596,
+      "step": 3871
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.521808723489396e-05,
+      "loss": 1.7826,
+      "step": 3872
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.51780712284914e-05,
+      "loss": 1.6846,
+      "step": 3873
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.513805522208884e-05,
+      "loss": 1.1747,
+      "step": 3874
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.5098039215686275e-05,
+      "loss": 1.7039,
+      "step": 3875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.505802320928372e-05,
+      "loss": 1.7586,
+      "step": 3876
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.5018007202881156e-05,
+      "loss": 1.5032,
+      "step": 3877
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.497799119647859e-05,
+      "loss": 1.5225,
+      "step": 3878
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.493797519007603e-05,
+      "loss": 1.6727,
+      "step": 3879
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4897959183673474e-05,
+      "loss": 1.3827,
+      "step": 3880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.485794317727091e-05,
+      "loss": 1.9466,
+      "step": 3881
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.481792717086835e-05,
+      "loss": 1.2806,
+      "step": 3882
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.477791116446579e-05,
+      "loss": 1.5825,
+      "step": 3883
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.473789515806322e-05,
+      "loss": 1.6703,
+      "step": 3884
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.469787915166067e-05,
+      "loss": 1.537,
+      "step": 3885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4657863145258104e-05,
+      "loss": 1.8112,
+      "step": 3886
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.461784713885555e-05,
+      "loss": 1.574,
+      "step": 3887
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4577831132452985e-05,
+      "loss": 1.2677,
+      "step": 3888
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.453781512605042e-05,
+      "loss": 1.152,
+      "step": 3889
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.449779911964786e-05,
+      "loss": 1.4431,
+      "step": 3890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4457783113245296e-05,
+      "loss": 1.7064,
+      "step": 3891
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.441776710684274e-05,
+      "loss": 1.816,
+      "step": 3892
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.437775110044018e-05,
+      "loss": 1.4945,
+      "step": 3893
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.433773509403762e-05,
+      "loss": 1.8005,
+      "step": 3894
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.429771908763505e-05,
+      "loss": 1.7385,
+      "step": 3895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4257703081232496e-05,
+      "loss": 1.9209,
+      "step": 3896
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.421768707482993e-05,
+      "loss": 1.4805,
+      "step": 3897
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.417767106842737e-05,
+      "loss": 1.7728,
+      "step": 3898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4137655062024814e-05,
+      "loss": 1.9706,
+      "step": 3899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.409763905562225e-05,
+      "loss": 1.9366,
+      "step": 3900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.405762304921969e-05,
+      "loss": 1.7509,
+      "step": 3901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4017607042817125e-05,
+      "loss": 2.0054,
+      "step": 3902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.397759103641457e-05,
+      "loss": 1.973,
+      "step": 3903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.3937575030012006e-05,
+      "loss": 1.6492,
+      "step": 3904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.389755902360945e-05,
+      "loss": 1.8934,
+      "step": 3905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.385754301720689e-05,
+      "loss": 1.472,
+      "step": 3906
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.3817527010804324e-05,
+      "loss": 1.6105,
+      "step": 3907
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.377751100440176e-05,
+      "loss": 1.6218,
+      "step": 3908
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.37374949979992e-05,
+      "loss": 1.4717,
+      "step": 3909
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.369747899159664e-05,
+      "loss": 1.1083,
+      "step": 3910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.365746298519408e-05,
+      "loss": 1.4919,
+      "step": 3911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.3617446978791524e-05,
+      "loss": 1.6274,
+      "step": 3912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.3577430972388954e-05,
+      "loss": 1.9307,
+      "step": 3913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.35374149659864e-05,
+      "loss": 1.5653,
+      "step": 3914
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.3497398959583835e-05,
+      "loss": 1.7641,
+      "step": 3915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.345738295318127e-05,
+      "loss": 1.4666,
+      "step": 3916
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.3417366946778716e-05,
+      "loss": 1.6437,
+      "step": 3917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.337735094037615e-05,
+      "loss": 1.98,
+      "step": 3918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.333733493397359e-05,
+      "loss": 1.8515,
+      "step": 3919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.329731892757103e-05,
+      "loss": 1.3026,
+      "step": 3920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.325730292116847e-05,
+      "loss": 1.9204,
+      "step": 3921
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.321728691476591e-05,
+      "loss": 1.4542,
+      "step": 3922
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.3177270908363346e-05,
+      "loss": 1.9095,
+      "step": 3923
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.313725490196079e-05,
+      "loss": 1.5889,
+      "step": 3924
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.309723889555822e-05,
+      "loss": 1.9046,
+      "step": 3925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.3057222889155664e-05,
+      "loss": 1.1592,
+      "step": 3926
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.30172068827531e-05,
+      "loss": 1.6486,
+      "step": 3927
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2977190876350545e-05,
+      "loss": 1.7179,
+      "step": 3928
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.293717486994798e-05,
+      "loss": 1.9197,
+      "step": 3929
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.289715886354542e-05,
+      "loss": 1.2963,
+      "step": 3930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 1.3914,
+      "step": 3931
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2817126850740293e-05,
+      "loss": 1.9354,
+      "step": 3932
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.277711084433774e-05,
+      "loss": 2.2309,
+      "step": 3933
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2737094837935174e-05,
+      "loss": 1.6934,
+      "step": 3934
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.269707883153262e-05,
+      "loss": 1.8684,
+      "step": 3935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2657062825130056e-05,
+      "loss": 1.6756,
+      "step": 3936
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.261704681872749e-05,
+      "loss": 1.3019,
+      "step": 3937
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.257703081232493e-05,
+      "loss": 1.3493,
+      "step": 3938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2537014805922374e-05,
+      "loss": 1.8869,
+      "step": 3939
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.249699879951981e-05,
+      "loss": 1.8685,
+      "step": 3940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.245698279311725e-05,
+      "loss": 1.8374,
+      "step": 3941
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.241696678671469e-05,
+      "loss": 1.7522,
+      "step": 3942
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.237695078031212e-05,
+      "loss": 1.8721,
+      "step": 3943
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2336934773909566e-05,
+      "loss": 1.6427,
+      "step": 3944
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2296918767507e-05,
+      "loss": 1.6022,
+      "step": 3945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.225690276110445e-05,
+      "loss": 1.5422,
+      "step": 3946
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2216886754701884e-05,
+      "loss": 1.4,
+      "step": 3947
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.217687074829932e-05,
+      "loss": 1.6637,
+      "step": 3948
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.213685474189676e-05,
+      "loss": 1.7252,
+      "step": 3949
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2096838735494196e-05,
+      "loss": 1.816,
+      "step": 3950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.205682272909164e-05,
+      "loss": 2.0208,
+      "step": 3951
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.201680672268908e-05,
+      "loss": 1.2633,
+      "step": 3952
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.197679071628652e-05,
+      "loss": 1.856,
+      "step": 3953
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.193677470988396e-05,
+      "loss": 1.4403,
+      "step": 3954
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1896758703481395e-05,
+      "loss": 1.3253,
+      "step": 3955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.185674269707883e-05,
+      "loss": 1.5689,
+      "step": 3956
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.181672669067627e-05,
+      "loss": 1.4618,
+      "step": 3957
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.177671068427371e-05,
+      "loss": 1.847,
+      "step": 3958
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.173669467787115e-05,
+      "loss": 1.4408,
+      "step": 3959
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1696678671468594e-05,
+      "loss": 1.6318,
+      "step": 3960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1656662665066025e-05,
+      "loss": 1.6685,
+      "step": 3961
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.161664665866347e-05,
+      "loss": 1.802,
+      "step": 3962
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1576630652260906e-05,
+      "loss": 1.6251,
+      "step": 3963
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.153661464585834e-05,
+      "loss": 1.8405,
+      "step": 3964
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.149659863945579e-05,
+      "loss": 1.8447,
+      "step": 3965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1456582633053224e-05,
+      "loss": 1.5037,
+      "step": 3966
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.141656662665066e-05,
+      "loss": 1.7979,
+      "step": 3967
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.13765506202481e-05,
+      "loss": 1.8134,
+      "step": 3968
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.133653461384554e-05,
+      "loss": 1.2844,
+      "step": 3969
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.129651860744298e-05,
+      "loss": 2.4182,
+      "step": 3970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.125650260104042e-05,
+      "loss": 1.5605,
+      "step": 3971
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.121648659463786e-05,
+      "loss": 1.8516,
+      "step": 3972
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.11764705882353e-05,
+      "loss": 1.4109,
+      "step": 3973
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1136454581832734e-05,
+      "loss": 1.8929,
+      "step": 3974
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.109643857543017e-05,
+      "loss": 1.6454,
+      "step": 3975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1056422569027615e-05,
+      "loss": 1.6037,
+      "step": 3976
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.101640656262505e-05,
+      "loss": 1.5531,
+      "step": 3977
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.0976390556222497e-05,
+      "loss": 2.0404,
+      "step": 3978
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.093637454981993e-05,
+      "loss": 1.989,
+      "step": 3979
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.089635854341737e-05,
+      "loss": 1.6291,
+      "step": 3980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.085634253701481e-05,
+      "loss": 1.7722,
+      "step": 3981
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.0816326530612245e-05,
+      "loss": 1.4565,
+      "step": 3982
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.077631052420969e-05,
+      "loss": 1.5079,
+      "step": 3983
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.073629451780712e-05,
+      "loss": 1.0682,
+      "step": 3984
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.069627851140456e-05,
+      "loss": 1.6633,
+      "step": 3985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.0656262505002e-05,
+      "loss": 1.5505,
+      "step": 3986
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.0616246498599444e-05,
+      "loss": 2.0245,
+      "step": 3987
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.057623049219688e-05,
+      "loss": 2.0354,
+      "step": 3988
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.053621448579432e-05,
+      "loss": 1.4012,
+      "step": 3989
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.0496198479391756e-05,
+      "loss": 1.9164,
+      "step": 3990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.045618247298919e-05,
+      "loss": 1.3956,
+      "step": 3991
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.041616646658664e-05,
+      "loss": 1.9499,
+      "step": 3992
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.0376150460184074e-05,
+      "loss": 1.8233,
+      "step": 3993
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.033613445378152e-05,
+      "loss": 1.5656,
+      "step": 3994
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.0296118447378955e-05,
+      "loss": 1.9598,
+      "step": 3995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.025610244097639e-05,
+      "loss": 2.0977,
+      "step": 3996
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.021608643457383e-05,
+      "loss": 1.9571,
+      "step": 3997
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.0176070428171266e-05,
+      "loss": 1.5721,
+      "step": 3998
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.013605442176871e-05,
+      "loss": 1.8346,
+      "step": 3999
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.009603841536615e-05,
+      "loss": 1.6878,
+      "step": 4000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.005602240896359e-05,
+      "loss": 1.3938,
+      "step": 4001
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.001600640256102e-05,
+      "loss": 1.802,
+      "step": 4002
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9975990396158466e-05,
+      "loss": 2.0838,
+      "step": 4003
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.99359743897559e-05,
+      "loss": 1.927,
+      "step": 4004
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9895958383353347e-05,
+      "loss": 1.4794,
+      "step": 4005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9855942376950784e-05,
+      "loss": 1.9378,
+      "step": 4006
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.981592637054822e-05,
+      "loss": 1.5774,
+      "step": 4007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.977591036414566e-05,
+      "loss": 1.7615,
+      "step": 4008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9735894357743095e-05,
+      "loss": 1.1147,
+      "step": 4009
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.969587835134054e-05,
+      "loss": 1.4956,
+      "step": 4010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9655862344937976e-05,
+      "loss": 1.8711,
+      "step": 4011
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.961584633853542e-05,
+      "loss": 1.0584,
+      "step": 4012
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.957583033213286e-05,
+      "loss": 1.6212,
+      "step": 4013
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9535814325730294e-05,
+      "loss": 1.8482,
+      "step": 4014
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.949579831932773e-05,
+      "loss": 1.8906,
+      "step": 4015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.945578231292517e-05,
+      "loss": 1.9595,
+      "step": 4016
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.941576630652261e-05,
+      "loss": 1.159,
+      "step": 4017
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.937575030012005e-05,
+      "loss": 1.7524,
+      "step": 4018
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9335734293717494e-05,
+      "loss": 2.1071,
+      "step": 4019
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9295718287314924e-05,
+      "loss": 1.6762,
+      "step": 4020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.925570228091237e-05,
+      "loss": 1.8391,
+      "step": 4021
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9215686274509805e-05,
+      "loss": 1.6861,
+      "step": 4022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.917567026810724e-05,
+      "loss": 1.7405,
+      "step": 4023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9135654261704686e-05,
+      "loss": 1.7274,
+      "step": 4024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.909563825530212e-05,
+      "loss": 1.8315,
+      "step": 4025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.905562224889956e-05,
+      "loss": 1.598,
+      "step": 4026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9015606242497e-05,
+      "loss": 1.9941,
+      "step": 4027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.897559023609444e-05,
+      "loss": 1.6785,
+      "step": 4028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.893557422969188e-05,
+      "loss": 1.3624,
+      "step": 4029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8895558223289316e-05,
+      "loss": 1.1245,
+      "step": 4030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.885554221688676e-05,
+      "loss": 1.7424,
+      "step": 4031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.881552621048419e-05,
+      "loss": 1.4796,
+      "step": 4032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8775510204081634e-05,
+      "loss": 1.724,
+      "step": 4033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.873549419767907e-05,
+      "loss": 1.7651,
+      "step": 4034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8695478191276515e-05,
+      "loss": 1.6731,
+      "step": 4035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.865546218487395e-05,
+      "loss": 1.8773,
+      "step": 4036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8615446178471396e-05,
+      "loss": 1.6124,
+      "step": 4037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8575430172068826e-05,
+      "loss": 1.8219,
+      "step": 4038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.853541416566627e-05,
+      "loss": 1.958,
+      "step": 4039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.849539815926371e-05,
+      "loss": 1.933,
+      "step": 4040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8455382152861144e-05,
+      "loss": 1.3894,
+      "step": 4041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.841536614645859e-05,
+      "loss": 1.6554,
+      "step": 4042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8375350140056026e-05,
+      "loss": 1.5159,
+      "step": 4043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.833533413365346e-05,
+      "loss": 1.8171,
+      "step": 4044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.82953181272509e-05,
+      "loss": 1.4372,
+      "step": 4045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8255302120848344e-05,
+      "loss": 2.3663,
+      "step": 4046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.821528611444578e-05,
+      "loss": 1.5417,
+      "step": 4047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.817527010804322e-05,
+      "loss": 1.9191,
+      "step": 4048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.813525410164066e-05,
+      "loss": 1.7926,
+      "step": 4049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.809523809523809e-05,
+      "loss": 1.6825,
+      "step": 4050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8055222088835536e-05,
+      "loss": 1.4263,
+      "step": 4051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.801520608243297e-05,
+      "loss": 0.9727,
+      "step": 4052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.797519007603042e-05,
+      "loss": 1.4058,
+      "step": 4053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7935174069627854e-05,
+      "loss": 2.1846,
+      "step": 4054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.789515806322529e-05,
+      "loss": 1.5228,
+      "step": 4055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.785514205682273e-05,
+      "loss": 2.0538,
+      "step": 4056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7815126050420166e-05,
+      "loss": 2.1333,
+      "step": 4057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.777511004401761e-05,
+      "loss": 1.6185,
+      "step": 4058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.773509403761505e-05,
+      "loss": 1.6303,
+      "step": 4059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.769507803121249e-05,
+      "loss": 1.4096,
+      "step": 4060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.765506202480993e-05,
+      "loss": 1.6447,
+      "step": 4061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7615046018407365e-05,
+      "loss": 1.403,
+      "step": 4062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.75750300120048e-05,
+      "loss": 1.747,
+      "step": 4063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.753501400560224e-05,
+      "loss": 1.3788,
+      "step": 4064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.749499799919968e-05,
+      "loss": 2.3468,
+      "step": 4065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.745498199279712e-05,
+      "loss": 1.6683,
+      "step": 4066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7414965986394564e-05,
+      "loss": 1.9431,
+      "step": 4067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7374949979991995e-05,
+      "loss": 1.4836,
+      "step": 4068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.733493397358944e-05,
+      "loss": 1.7581,
+      "step": 4069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7294917967186876e-05,
+      "loss": 1.7677,
+      "step": 4070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.725490196078432e-05,
+      "loss": 1.88,
+      "step": 4071
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.721488595438176e-05,
+      "loss": 1.857,
+      "step": 4072
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7174869947979194e-05,
+      "loss": 1.8248,
+      "step": 4073
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.713485394157663e-05,
+      "loss": 1.7396,
+      "step": 4074
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.709483793517407e-05,
+      "loss": 2.1362,
+      "step": 4075
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.705482192877151e-05,
+      "loss": 1.7316,
+      "step": 4076
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.701480592236895e-05,
+      "loss": 1.5497,
+      "step": 4077
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.697478991596639e-05,
+      "loss": 1.9244,
+      "step": 4078
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.693477390956382e-05,
+      "loss": 1.3089,
+      "step": 4079
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.689475790316127e-05,
+      "loss": 1.841,
+      "step": 4080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6854741896758704e-05,
+      "loss": 1.6622,
+      "step": 4081
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.681472589035614e-05,
+      "loss": 1.4676,
+      "step": 4082
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6774709883953585e-05,
+      "loss": 1.5769,
+      "step": 4083
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.673469387755102e-05,
+      "loss": 1.6055,
+      "step": 4084
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.669467787114846e-05,
+      "loss": 1.3648,
+      "step": 4085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.66546618647459e-05,
+      "loss": 1.5327,
+      "step": 4086
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.661464585834334e-05,
+      "loss": 1.4254,
+      "step": 4087
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.657462985194078e-05,
+      "loss": 1.6916,
+      "step": 4088
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6534613845538215e-05,
+      "loss": 1.6476,
+      "step": 4089
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.649459783913566e-05,
+      "loss": 1.7301,
+      "step": 4090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.645458183273309e-05,
+      "loss": 1.6523,
+      "step": 4091
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.641456582633053e-05,
+      "loss": 2.044,
+      "step": 4092
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.637454981992797e-05,
+      "loss": 1.8877,
+      "step": 4093
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6334533813525414e-05,
+      "loss": 1.2998,
+      "step": 4094
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.629451780712285e-05,
+      "loss": 1.9596,
+      "step": 4095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.625450180072029e-05,
+      "loss": 2.4899,
+      "step": 4096
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6214485794317726e-05,
+      "loss": 1.9325,
+      "step": 4097
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.617446978791516e-05,
+      "loss": 1.1411,
+      "step": 4098
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.613445378151261e-05,
+      "loss": 1.9087,
+      "step": 4099
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6094437775110044e-05,
+      "loss": 1.2266,
+      "step": 4100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.605442176870749e-05,
+      "loss": 1.6238,
+      "step": 4101
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6014405762304925e-05,
+      "loss": 1.5844,
+      "step": 4102
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.597438975590236e-05,
+      "loss": 1.4718,
+      "step": 4103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.59343737494998e-05,
+      "loss": 1.5265,
+      "step": 4104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.589435774309724e-05,
+      "loss": 2.1736,
+      "step": 4105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.585434173669468e-05,
+      "loss": 1.8281,
+      "step": 4106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.581432573029212e-05,
+      "loss": 1.8064,
+      "step": 4107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.577430972388956e-05,
+      "loss": 2.0788,
+      "step": 4108
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.573429371748699e-05,
+      "loss": 1.6175,
+      "step": 4109
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.5694277711084436e-05,
+      "loss": 1.6767,
+      "step": 4110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.565426170468187e-05,
+      "loss": 1.6204,
+      "step": 4111
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.5614245698279317e-05,
+      "loss": 1.9509,
+      "step": 4112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.5574229691876754e-05,
+      "loss": 0.9292,
+      "step": 4113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.553421368547419e-05,
+      "loss": 2.1253,
+      "step": 4114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.549419767907163e-05,
+      "loss": 1.5724,
+      "step": 4115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.5454181672669065e-05,
+      "loss": 2.0825,
+      "step": 4116
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.541416566626651e-05,
+      "loss": 1.7704,
+      "step": 4117
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.5374149659863946e-05,
+      "loss": 1.3139,
+      "step": 4118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.533413365346139e-05,
+      "loss": 1.991,
+      "step": 4119
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.529411764705883e-05,
+      "loss": 1.7791,
+      "step": 4120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.5254101640656264e-05,
+      "loss": 1.6203,
+      "step": 4121
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.52140856342537e-05,
+      "loss": 1.6951,
+      "step": 4122
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.517406962785114e-05,
+      "loss": 1.8892,
+      "step": 4123
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.513405362144858e-05,
+      "loss": 1.538,
+      "step": 4124
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.509403761504602e-05,
+      "loss": 1.8856,
+      "step": 4125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.5054021608643464e-05,
+      "loss": 1.3942,
+      "step": 4126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.5014005602240894e-05,
+      "loss": 1.6585,
+      "step": 4127
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.497398959583834e-05,
+      "loss": 1.7933,
+      "step": 4128
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.4933973589435775e-05,
+      "loss": 1.4754,
+      "step": 4129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.489395758303321e-05,
+      "loss": 1.9166,
+      "step": 4130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.4853941576630656e-05,
+      "loss": 1.7058,
+      "step": 4131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.481392557022809e-05,
+      "loss": 1.9089,
+      "step": 4132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.477390956382553e-05,
+      "loss": 1.9115,
+      "step": 4133
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.473389355742297e-05,
+      "loss": 1.3868,
+      "step": 4134
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.469387755102041e-05,
+      "loss": 1.4814,
+      "step": 4135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.465386154461785e-05,
+      "loss": 1.9286,
+      "step": 4136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.461384553821529e-05,
+      "loss": 1.5391,
+      "step": 4137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.457382953181273e-05,
+      "loss": 1.9933,
+      "step": 4138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.453381352541017e-05,
+      "loss": 2.1084,
+      "step": 4139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.4493797519007604e-05,
+      "loss": 1.6524,
+      "step": 4140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.445378151260504e-05,
+      "loss": 1.2172,
+      "step": 4141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.4413765506202485e-05,
+      "loss": 1.5882,
+      "step": 4142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.437374949979992e-05,
+      "loss": 1.6706,
+      "step": 4143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.4333733493397366e-05,
+      "loss": 1.5787,
+      "step": 4144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.4293717486994796e-05,
+      "loss": 1.8291,
+      "step": 4145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.425370148059224e-05,
+      "loss": 1.9174,
+      "step": 4146
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.421368547418968e-05,
+      "loss": 1.6848,
+      "step": 4147
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.4173669467787114e-05,
+      "loss": 1.4277,
+      "step": 4148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.413365346138456e-05,
+      "loss": 1.331,
+      "step": 4149
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.4093637454981995e-05,
+      "loss": 1.8051,
+      "step": 4150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.405362144857943e-05,
+      "loss": 1.5814,
+      "step": 4151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.401360544217687e-05,
+      "loss": 1.5358,
+      "step": 4152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.3973589435774314e-05,
+      "loss": 1.2437,
+      "step": 4153
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.393357342937175e-05,
+      "loss": 2.468,
+      "step": 4154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.389355742296919e-05,
+      "loss": 1.9182,
+      "step": 4155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.385354141656663e-05,
+      "loss": 1.997,
+      "step": 4156
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.381352541016406e-05,
+      "loss": 1.8616,
+      "step": 4157
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.3773509403761506e-05,
+      "loss": 1.6977,
+      "step": 4158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.373349339735894e-05,
+      "loss": 1.5254,
+      "step": 4159
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.369347739095639e-05,
+      "loss": 1.5764,
+      "step": 4160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.3653461384553824e-05,
+      "loss": 1.8535,
+      "step": 4161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.361344537815127e-05,
+      "loss": 2.0302,
+      "step": 4162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.35734293717487e-05,
+      "loss": 1.4862,
+      "step": 4163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.3533413365346136e-05,
+      "loss": 1.6422,
+      "step": 4164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.349339735894358e-05,
+      "loss": 1.7005,
+      "step": 4165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.345338135254102e-05,
+      "loss": 1.2653,
+      "step": 4166
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.341336534613846e-05,
+      "loss": 2.071,
+      "step": 4167
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.33733493397359e-05,
+      "loss": 1.6096,
+      "step": 4168
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 1.3922,
+      "step": 4169
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.329331732693077e-05,
+      "loss": 1.8199,
+      "step": 4170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.3253301320528216e-05,
+      "loss": 1.3672,
+      "step": 4171
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.321328531412565e-05,
+      "loss": 1.2935,
+      "step": 4172
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.317326930772309e-05,
+      "loss": 1.6157,
+      "step": 4173
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.313325330132053e-05,
+      "loss": 2.2911,
+      "step": 4174
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.3093237294917965e-05,
+      "loss": 1.6768,
+      "step": 4175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.305322128851541e-05,
+      "loss": 1.7885,
+      "step": 4176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.3013205282112846e-05,
+      "loss": 1.3903,
+      "step": 4177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.297318927571029e-05,
+      "loss": 1.9888,
+      "step": 4178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2933173269307727e-05,
+      "loss": 1.9512,
+      "step": 4179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2893157262905164e-05,
+      "loss": 1.9186,
+      "step": 4180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.28531412565026e-05,
+      "loss": 1.6766,
+      "step": 4181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.281312525010004e-05,
+      "loss": 2.2116,
+      "step": 4182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.277310924369748e-05,
+      "loss": 1.4421,
+      "step": 4183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.273309323729492e-05,
+      "loss": 1.1873,
+      "step": 4184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.269307723089236e-05,
+      "loss": 1.9383,
+      "step": 4185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.265306122448979e-05,
+      "loss": 1.4665,
+      "step": 4186
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.261304521808724e-05,
+      "loss": 1.5514,
+      "step": 4187
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2573029211684674e-05,
+      "loss": 1.745,
+      "step": 4188
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.253301320528211e-05,
+      "loss": 1.217,
+      "step": 4189
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2492997198879555e-05,
+      "loss": 1.5411,
+      "step": 4190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.245298119247699e-05,
+      "loss": 2.2075,
+      "step": 4191
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.241296518607443e-05,
+      "loss": 1.2693,
+      "step": 4192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.237294917967187e-05,
+      "loss": 1.4163,
+      "step": 4193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.233293317326931e-05,
+      "loss": 1.9051,
+      "step": 4194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.229291716686675e-05,
+      "loss": 2.0418,
+      "step": 4195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.225290116046419e-05,
+      "loss": 1.3767,
+      "step": 4196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.221288515406163e-05,
+      "loss": 1.9083,
+      "step": 4197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2172869147659066e-05,
+      "loss": 1.6816,
+      "step": 4198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.21328531412565e-05,
+      "loss": 1.5715,
+      "step": 4199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.209283713485394e-05,
+      "loss": 2.1082,
+      "step": 4200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2052821128451384e-05,
+      "loss": 2.0806,
+      "step": 4201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.201280512204882e-05,
+      "loss": 1.8484,
+      "step": 4202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1972789115646265e-05,
+      "loss": 1.6721,
+      "step": 4203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1932773109243696e-05,
+      "loss": 1.4345,
+      "step": 4204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.189275710284114e-05,
+      "loss": 1.3573,
+      "step": 4205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.185274109643858e-05,
+      "loss": 1.7562,
+      "step": 4206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1812725090036014e-05,
+      "loss": 2.1862,
+      "step": 4207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.177270908363346e-05,
+      "loss": 1.4866,
+      "step": 4208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1732693077230895e-05,
+      "loss": 1.3883,
+      "step": 4209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.169267707082833e-05,
+      "loss": 1.2778,
+      "step": 4210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.165266106442577e-05,
+      "loss": 1.8664,
+      "step": 4211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.161264505802321e-05,
+      "loss": 1.7207,
+      "step": 4212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.157262905162065e-05,
+      "loss": 1.3412,
+      "step": 4213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.153261304521809e-05,
+      "loss": 1.5222,
+      "step": 4214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.149259703881553e-05,
+      "loss": 1.6324,
+      "step": 4215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.145258103241296e-05,
+      "loss": 1.8551,
+      "step": 4216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1412565026010406e-05,
+      "loss": 1.7136,
+      "step": 4217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.137254901960784e-05,
+      "loss": 1.6199,
+      "step": 4218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1332533013205287e-05,
+      "loss": 1.8516,
+      "step": 4219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1292517006802724e-05,
+      "loss": 2.0384,
+      "step": 4220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.125250100040016e-05,
+      "loss": 1.1311,
+      "step": 4221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.12124849939976e-05,
+      "loss": 1.3937,
+      "step": 4222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1172468987595035e-05,
+      "loss": 1.3986,
+      "step": 4223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.113245298119248e-05,
+      "loss": 1.9245,
+      "step": 4224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1092436974789916e-05,
+      "loss": 2.2927,
+      "step": 4225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.105242096838736e-05,
+      "loss": 2.0007,
+      "step": 4226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.10124049619848e-05,
+      "loss": 1.3325,
+      "step": 4227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0972388955582234e-05,
+      "loss": 1.4872,
+      "step": 4228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.093237294917967e-05,
+      "loss": 1.7017,
+      "step": 4229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0892356942777115e-05,
+      "loss": 1.7373,
+      "step": 4230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.085234093637455e-05,
+      "loss": 1.7929,
+      "step": 4231
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.081232492997199e-05,
+      "loss": 1.6961,
+      "step": 4232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0772308923569434e-05,
+      "loss": 2.2156,
+      "step": 4233
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0732292917166864e-05,
+      "loss": 1.6302,
+      "step": 4234
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.069227691076431e-05,
+      "loss": 1.5509,
+      "step": 4235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0652260904361745e-05,
+      "loss": 1.6938,
+      "step": 4236
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.061224489795919e-05,
+      "loss": 1.5667,
+      "step": 4237
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0572228891556626e-05,
+      "loss": 1.4792,
+      "step": 4238
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.053221288515406e-05,
+      "loss": 1.4113,
+      "step": 4239
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.04921968787515e-05,
+      "loss": 1.9218,
+      "step": 4240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.045218087234894e-05,
+      "loss": 1.3827,
+      "step": 4241
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.041216486594638e-05,
+      "loss": 1.5719,
+      "step": 4242
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.037214885954382e-05,
+      "loss": 1.5896,
+      "step": 4243
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.033213285314126e-05,
+      "loss": 1.5971,
+      "step": 4244
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.02921168467387e-05,
+      "loss": 1.5919,
+      "step": 4245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0252100840336133e-05,
+      "loss": 2.2494,
+      "step": 4246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.0212084833933574e-05,
+      "loss": 1.8412,
+      "step": 4247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.0172068827531014e-05,
+      "loss": 2.0725,
+      "step": 4248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.0132052821128455e-05,
+      "loss": 1.8045,
+      "step": 4249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.0092036814725892e-05,
+      "loss": 1.9443,
+      "step": 4250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.0052020808323332e-05,
+      "loss": 1.4476,
+      "step": 4251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.0012004801920766e-05,
+      "loss": 1.4226,
+      "step": 4252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9971988795518207e-05,
+      "loss": 1.6104,
+      "step": 4253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9931972789115647e-05,
+      "loss": 1.4323,
+      "step": 4254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9891956782713088e-05,
+      "loss": 1.8667,
+      "step": 4255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.985194077631053e-05,
+      "loss": 1.9431,
+      "step": 4256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.981192476990797e-05,
+      "loss": 1.7476,
+      "step": 4257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9771908763505403e-05,
+      "loss": 1.3767,
+      "step": 4258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9731892757102843e-05,
+      "loss": 1.2323,
+      "step": 4259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.969187675070028e-05,
+      "loss": 1.8926,
+      "step": 4260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.965186074429772e-05,
+      "loss": 1.996,
+      "step": 4261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.961184473789516e-05,
+      "loss": 1.3586,
+      "step": 4262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9571828731492595e-05,
+      "loss": 1.7503,
+      "step": 4263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9531812725090036e-05,
+      "loss": 2.135,
+      "step": 4264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9491796718687476e-05,
+      "loss": 2.6844,
+      "step": 4265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9451780712284917e-05,
+      "loss": 1.6522,
+      "step": 4266
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9411764705882354e-05,
+      "loss": 2.289,
+      "step": 4267
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9371748699479794e-05,
+      "loss": 1.7467,
+      "step": 4268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9331732693077228e-05,
+      "loss": 2.261,
+      "step": 4269
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.929171668667467e-05,
+      "loss": 1.9467,
+      "step": 4270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.925170068027211e-05,
+      "loss": 1.456,
+      "step": 4271
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.921168467386955e-05,
+      "loss": 1.8869,
+      "step": 4272
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.917166866746699e-05,
+      "loss": 1.8984,
+      "step": 4273
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.913165266106443e-05,
+      "loss": 1.6872,
+      "step": 4274
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9091636654661864e-05,
+      "loss": 1.5876,
+      "step": 4275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9051620648259305e-05,
+      "loss": 1.5928,
+      "step": 4276
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9011604641856742e-05,
+      "loss": 1.8684,
+      "step": 4277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8971588635454183e-05,
+      "loss": 1.4767,
+      "step": 4278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8931572629051623e-05,
+      "loss": 1.5405,
+      "step": 4279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8891556622649064e-05,
+      "loss": 1.789,
+      "step": 4280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8851540616246497e-05,
+      "loss": 1.5561,
+      "step": 4281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8811524609843938e-05,
+      "loss": 1.6108,
+      "step": 4282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.877150860344138e-05,
+      "loss": 1.6137,
+      "step": 4283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8731492597038816e-05,
+      "loss": 1.9089,
+      "step": 4284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8691476590636256e-05,
+      "loss": 1.7488,
+      "step": 4285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8651460584233697e-05,
+      "loss": 1.6687,
+      "step": 4286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.861144457783113e-05,
+      "loss": 1.8493,
+      "step": 4287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 2.0532,
+      "step": 4288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.853141256502601e-05,
+      "loss": 1.8336,
+      "step": 4289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8491396558623452e-05,
+      "loss": 1.5496,
+      "step": 4290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8451380552220892e-05,
+      "loss": 1.5475,
+      "step": 4291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.841136454581833e-05,
+      "loss": 2.0186,
+      "step": 4292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8371348539415767e-05,
+      "loss": 2.1126,
+      "step": 4293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8331332533013204e-05,
+      "loss": 1.847,
+      "step": 4294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8291316526610644e-05,
+      "loss": 1.8136,
+      "step": 4295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8251300520208085e-05,
+      "loss": 1.6472,
+      "step": 4296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8211284513805525e-05,
+      "loss": 1.6461,
+      "step": 4297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8171268507402966e-05,
+      "loss": 2.2159,
+      "step": 4298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.81312525010004e-05,
+      "loss": 1.9999,
+      "step": 4299
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.809123649459784e-05,
+      "loss": 2.1615,
+      "step": 4300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8051220488195277e-05,
+      "loss": 1.6012,
+      "step": 4301
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8011204481792718e-05,
+      "loss": 1.3703,
+      "step": 4302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.797118847539016e-05,
+      "loss": 1.7663,
+      "step": 4303
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.79311724689876e-05,
+      "loss": 2.0213,
+      "step": 4304
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7891156462585033e-05,
+      "loss": 1.1412,
+      "step": 4305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7851140456182473e-05,
+      "loss": 1.7243,
+      "step": 4306
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7811124449779914e-05,
+      "loss": 1.4454,
+      "step": 4307
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7771108443377354e-05,
+      "loss": 1.6151,
+      "step": 4308
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.773109243697479e-05,
+      "loss": 1.7495,
+      "step": 4309
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7691076430572232e-05,
+      "loss": 1.5077,
+      "step": 4310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7651060424169666e-05,
+      "loss": 1.7601,
+      "step": 4311
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7611044417767106e-05,
+      "loss": 1.6695,
+      "step": 4312
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7571028411364547e-05,
+      "loss": 1.3073,
+      "step": 4313
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7531012404961987e-05,
+      "loss": 1.5953,
+      "step": 4314
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7490996398559428e-05,
+      "loss": 1.4291,
+      "step": 4315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7450980392156865e-05,
+      "loss": 1.5706,
+      "step": 4316
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7410964385754302e-05,
+      "loss": 1.5524,
+      "step": 4317
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.737094837935174e-05,
+      "loss": 1.5652,
+      "step": 4318
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.733093237294918e-05,
+      "loss": 1.7895,
+      "step": 4319
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.729091636654662e-05,
+      "loss": 1.973,
+      "step": 4320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.725090036014406e-05,
+      "loss": 1.6002,
+      "step": 4321
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.72108843537415e-05,
+      "loss": 1.9659,
+      "step": 4322
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7170868347338935e-05,
+      "loss": 1.5047,
+      "step": 4323
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7130852340936375e-05,
+      "loss": 1.2063,
+      "step": 4324
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7090836334533816e-05,
+      "loss": 1.7997,
+      "step": 4325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7050820328131253e-05,
+      "loss": 1.8739,
+      "step": 4326
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7010804321728694e-05,
+      "loss": 1.746,
+      "step": 4327
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6970788315326134e-05,
+      "loss": 1.4163,
+      "step": 4328
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6930772308923568e-05,
+      "loss": 1.584,
+      "step": 4329
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.689075630252101e-05,
+      "loss": 2.0479,
+      "step": 4330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.685074029611845e-05,
+      "loss": 2.0655,
+      "step": 4331
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.681072428971589e-05,
+      "loss": 2.0233,
+      "step": 4332
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6770708283313327e-05,
+      "loss": 1.9221,
+      "step": 4333
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6730692276910767e-05,
+      "loss": 1.7925,
+      "step": 4334
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.66906762705082e-05,
+      "loss": 1.4605,
+      "step": 4335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.665066026410564e-05,
+      "loss": 1.3406,
+      "step": 4336
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6610644257703082e-05,
+      "loss": 2.214,
+      "step": 4337
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6570628251300522e-05,
+      "loss": 1.7353,
+      "step": 4338
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6530612244897963e-05,
+      "loss": 1.697,
+      "step": 4339
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6490596238495404e-05,
+      "loss": 1.4388,
+      "step": 4340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6450580232092837e-05,
+      "loss": 1.9577,
+      "step": 4341
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6410564225690278e-05,
+      "loss": 1.6005,
+      "step": 4342
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6370548219287715e-05,
+      "loss": 1.6633,
+      "step": 4343
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6330532212885155e-05,
+      "loss": 1.8681,
+      "step": 4344
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6290516206482596e-05,
+      "loss": 2.1459,
+      "step": 4345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6250500200080037e-05,
+      "loss": 1.4743,
+      "step": 4346
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.621048419367747e-05,
+      "loss": 1.4077,
+      "step": 4347
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.617046818727491e-05,
+      "loss": 1.1675,
+      "step": 4348
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.613045218087235e-05,
+      "loss": 2.1285,
+      "step": 4349
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.609043617446979e-05,
+      "loss": 2.3101,
+      "step": 4350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.605042016806723e-05,
+      "loss": 1.5766,
+      "step": 4351
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.601040416166467e-05,
+      "loss": 1.6019,
+      "step": 4352
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5970388155262103e-05,
+      "loss": 1.5264,
+      "step": 4353
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5930372148859544e-05,
+      "loss": 1.5656,
+      "step": 4354
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5890356142456984e-05,
+      "loss": 1.7195,
+      "step": 4355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5850340136054425e-05,
+      "loss": 1.7509,
+      "step": 4356
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5810324129651865e-05,
+      "loss": 1.9812,
+      "step": 4357
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.57703081232493e-05,
+      "loss": 1.7648,
+      "step": 4358
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.573029211684674e-05,
+      "loss": 1.4028,
+      "step": 4359
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5690276110444177e-05,
+      "loss": 2.0885,
+      "step": 4360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5650260104041617e-05,
+      "loss": 1.4256,
+      "step": 4361
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5610244097639058e-05,
+      "loss": 1.5524,
+      "step": 4362
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5570228091236498e-05,
+      "loss": 2.1875,
+      "step": 4363
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5530212084833932e-05,
+      "loss": 1.7483,
+      "step": 4364
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5490196078431373e-05,
+      "loss": 1.3322,
+      "step": 4365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5450180072028813e-05,
+      "loss": 1.3923,
+      "step": 4366
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.541016406562625e-05,
+      "loss": 1.7867,
+      "step": 4367
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.537014805922369e-05,
+      "loss": 1.904,
+      "step": 4368
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.533013205282113e-05,
+      "loss": 1.991,
+      "step": 4369
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.5290116046418565e-05,
+      "loss": 1.4589,
+      "step": 4370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.5250100040016006e-05,
+      "loss": 1.7391,
+      "step": 4371
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.5210084033613446e-05,
+      "loss": 1.8142,
+      "step": 4372
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.5170068027210887e-05,
+      "loss": 1.526,
+      "step": 4373
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.5130052020808327e-05,
+      "loss": 1.5615,
+      "step": 4374
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.5090036014405764e-05,
+      "loss": 1.8506,
+      "step": 4375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.50500200080032e-05,
+      "loss": 1.7257,
+      "step": 4376
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.501000400160064e-05,
+      "loss": 1.7124,
+      "step": 4377
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.496998799519808e-05,
+      "loss": 1.5277,
+      "step": 4378
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.492997198879552e-05,
+      "loss": 1.6136,
+      "step": 4379
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.488995598239296e-05,
+      "loss": 1.687,
+      "step": 4380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4849939975990397e-05,
+      "loss": 1.81,
+      "step": 4381
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4809923969587838e-05,
+      "loss": 2.085,
+      "step": 4382
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4769907963185275e-05,
+      "loss": 1.6964,
+      "step": 4383
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4729891956782712e-05,
+      "loss": 1.9185,
+      "step": 4384
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4689875950380153e-05,
+      "loss": 1.6277,
+      "step": 4385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4649859943977593e-05,
+      "loss": 1.7519,
+      "step": 4386
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.460984393757503e-05,
+      "loss": 1.806,
+      "step": 4387
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.456982793117247e-05,
+      "loss": 1.4417,
+      "step": 4388
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.452981192476991e-05,
+      "loss": 1.4285,
+      "step": 4389
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.448979591836735e-05,
+      "loss": 1.4582,
+      "step": 4390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.444977991196479e-05,
+      "loss": 1.6186,
+      "step": 4391
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4409763905562226e-05,
+      "loss": 1.877,
+      "step": 4392
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4369747899159663e-05,
+      "loss": 1.6302,
+      "step": 4393
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4329731892757104e-05,
+      "loss": 1.7835,
+      "step": 4394
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4289715886354544e-05,
+      "loss": 1.874,
+      "step": 4395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.424969987995198e-05,
+      "loss": 1.8195,
+      "step": 4396
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4209683873549422e-05,
+      "loss": 1.834,
+      "step": 4397
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4169667867146862e-05,
+      "loss": 1.7826,
+      "step": 4398
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.41296518607443e-05,
+      "loss": 2.3127,
+      "step": 4399
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4089635854341737e-05,
+      "loss": 1.3145,
+      "step": 4400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4049619847939174e-05,
+      "loss": 1.4514,
+      "step": 4401
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4009603841536614e-05,
+      "loss": 1.5974,
+      "step": 4402
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3969587835134055e-05,
+      "loss": 1.2437,
+      "step": 4403
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3929571828731492e-05,
+      "loss": 1.4198,
+      "step": 4404
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3889555822328933e-05,
+      "loss": 2.2555,
+      "step": 4405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3849539815926373e-05,
+      "loss": 2.0401,
+      "step": 4406
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.380952380952381e-05,
+      "loss": 1.6581,
+      "step": 4407
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.376950780312125e-05,
+      "loss": 1.5589,
+      "step": 4408
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3729491796718688e-05,
+      "loss": 1.6481,
+      "step": 4409
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3689475790316125e-05,
+      "loss": 1.8461,
+      "step": 4410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3649459783913565e-05,
+      "loss": 2.0126,
+      "step": 4411
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3609443777511006e-05,
+      "loss": 2.0407,
+      "step": 4412
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3569427771108443e-05,
+      "loss": 1.6724,
+      "step": 4413
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3529411764705884e-05,
+      "loss": 1.6116,
+      "step": 4414
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3489395758303324e-05,
+      "loss": 1.8995,
+      "step": 4415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.344937975190076e-05,
+      "loss": 1.6535,
+      "step": 4416
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.34093637454982e-05,
+      "loss": 1.7715,
+      "step": 4417
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.336934773909564e-05,
+      "loss": 1.7258,
+      "step": 4418
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3329331732693076e-05,
+      "loss": 1.9169,
+      "step": 4419
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3289315726290517e-05,
+      "loss": 2.1058,
+      "step": 4420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3249299719887957e-05,
+      "loss": 1.9559,
+      "step": 4421
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3209283713485394e-05,
+      "loss": 2.0534,
+      "step": 4422
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3169267707082835e-05,
+      "loss": 2.1937,
+      "step": 4423
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3129251700680275e-05,
+      "loss": 2.1761,
+      "step": 4424
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3089235694277712e-05,
+      "loss": 1.3588,
+      "step": 4425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.304921968787515e-05,
+      "loss": 1.9561,
+      "step": 4426
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.300920368147259e-05,
+      "loss": 2.0251,
+      "step": 4427
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.2969187675070027e-05,
+      "loss": 1.1793,
+      "step": 4428
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.2929171668667468e-05,
+      "loss": 1.6086,
+      "step": 4429
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.288915566226491e-05,
+      "loss": 1.4854,
+      "step": 4430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.2849139655862345e-05,
+      "loss": 1.5816,
+      "step": 4431
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.2809123649459786e-05,
+      "loss": 1.4918,
+      "step": 4432
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.2769107643057223e-05,
+      "loss": 1.611,
+      "step": 4433
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.272909163665466e-05,
+      "loss": 1.6781,
+      "step": 4434
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.26890756302521e-05,
+      "loss": 1.4874,
+      "step": 4435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.264905962384954e-05,
+      "loss": 1.843,
+      "step": 4436
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.260904361744698e-05,
+      "loss": 1.7734,
+      "step": 4437
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.256902761104442e-05,
+      "loss": 1.4244,
+      "step": 4438
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.252901160464186e-05,
+      "loss": 1.325,
+      "step": 4439
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.2488995598239297e-05,
+      "loss": 1.7848,
+      "step": 4440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.2448979591836737e-05,
+      "loss": 1.6706,
+      "step": 4441
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.2408963585434174e-05,
+      "loss": 1.8579,
+      "step": 4442
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.236894757903161e-05,
+      "loss": 2.0023,
+      "step": 4443
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.2328931572629052e-05,
+      "loss": 1.4827,
+      "step": 4444
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.2288915566226492e-05,
+      "loss": 1.3794,
+      "step": 4445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.224889955982393e-05,
+      "loss": 1.9521,
+      "step": 4446
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.220888355342137e-05,
+      "loss": 1.3887,
+      "step": 4447
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.216886754701881e-05,
+      "loss": 1.7015,
+      "step": 4448
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.2128851540616248e-05,
+      "loss": 1.4812,
+      "step": 4449
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.2088835534213685e-05,
+      "loss": 1.9374,
+      "step": 4450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.2048819527811125e-05,
+      "loss": 1.7601,
+      "step": 4451
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.2008803521408563e-05,
+      "loss": 1.5329,
+      "step": 4452
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1968787515006003e-05,
+      "loss": 1.5363,
+      "step": 4453
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1928771508603444e-05,
+      "loss": 1.5657,
+      "step": 4454
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.188875550220088e-05,
+      "loss": 1.8034,
+      "step": 4455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.184873949579832e-05,
+      "loss": 1.5963,
+      "step": 4456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1808723489395762e-05,
+      "loss": 1.3667,
+      "step": 4457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.17687074829932e-05,
+      "loss": 1.5285,
+      "step": 4458
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1728691476590636e-05,
+      "loss": 1.3388,
+      "step": 4459
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1688675470188077e-05,
+      "loss": 1.462,
+      "step": 4460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1648659463785514e-05,
+      "loss": 1.1227,
+      "step": 4461
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1608643457382954e-05,
+      "loss": 1.4119,
+      "step": 4462
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1568627450980395e-05,
+      "loss": 1.1667,
+      "step": 4463
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1528611444577832e-05,
+      "loss": 1.7023,
+      "step": 4464
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1488595438175272e-05,
+      "loss": 1.4838,
+      "step": 4465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.144857943177271e-05,
+      "loss": 1.8852,
+      "step": 4466
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1408563425370147e-05,
+      "loss": 1.7061,
+      "step": 4467
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1368547418967587e-05,
+      "loss": 1.6937,
+      "step": 4468
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1328531412565028e-05,
+      "loss": 1.4669,
+      "step": 4469
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1288515406162465e-05,
+      "loss": 1.7028,
+      "step": 4470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1248499399759905e-05,
+      "loss": 1.6254,
+      "step": 4471
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1208483393357346e-05,
+      "loss": 1.8543,
+      "step": 4472
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1168467386954783e-05,
+      "loss": 1.8416,
+      "step": 4473
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1128451380552224e-05,
+      "loss": 1.7464,
+      "step": 4474
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.108843537414966e-05,
+      "loss": 1.8701,
+      "step": 4475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1048419367747098e-05,
+      "loss": 1.8664,
+      "step": 4476
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.100840336134454e-05,
+      "loss": 1.5641,
+      "step": 4477
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.096838735494198e-05,
+      "loss": 2.0474,
+      "step": 4478
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.0928371348539416e-05,
+      "loss": 1.7065,
+      "step": 4479
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.0888355342136857e-05,
+      "loss": 1.5019,
+      "step": 4480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.0848339335734297e-05,
+      "loss": 1.4029,
+      "step": 4481
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.0808323329331734e-05,
+      "loss": 1.4455,
+      "step": 4482
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.076830732292917e-05,
+      "loss": 1.6848,
+      "step": 4483
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.0728291316526612e-05,
+      "loss": 1.5615,
+      "step": 4484
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.068827531012405e-05,
+      "loss": 1.3747,
+      "step": 4485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.064825930372149e-05,
+      "loss": 1.8193,
+      "step": 4486
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.060824329731893e-05,
+      "loss": 1.7253,
+      "step": 4487
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.0568227290916367e-05,
+      "loss": 1.6046,
+      "step": 4488
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.0528211284513808e-05,
+      "loss": 1.8277,
+      "step": 4489
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.0488195278111248e-05,
+      "loss": 1.7573,
+      "step": 4490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.0448179271708685e-05,
+      "loss": 1.5963,
+      "step": 4491
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.0408163265306123e-05,
+      "loss": 1.5839,
+      "step": 4492
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.036814725890356e-05,
+      "loss": 1.6973,
+      "step": 4493
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.0328131252501e-05,
+      "loss": 1.826,
+      "step": 4494
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.028811524609844e-05,
+      "loss": 1.6862,
+      "step": 4495
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.0248099239695878e-05,
+      "loss": 1.7226,
+      "step": 4496
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.020808323329332e-05,
+      "loss": 1.6298,
+      "step": 4497
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.016806722689076e-05,
+      "loss": 1.4462,
+      "step": 4498
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.0128051220488196e-05,
+      "loss": 1.4412,
+      "step": 4499
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.0088035214085633e-05,
+      "loss": 1.3454,
+      "step": 4500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.0048019207683074e-05,
+      "loss": 1.6511,
+      "step": 4501
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.000800320128051e-05,
+      "loss": 1.6467,
+      "step": 4502
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.996798719487795e-05,
+      "loss": 1.6235,
+      "step": 4503
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9927971188475392e-05,
+      "loss": 1.9221,
+      "step": 4504
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988795518207283e-05,
+      "loss": 1.4628,
+      "step": 4505
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.984793917567027e-05,
+      "loss": 1.5728,
+      "step": 4506
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.980792316926771e-05,
+      "loss": 1.4844,
+      "step": 4507
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9767907162865147e-05,
+      "loss": 2.0244,
+      "step": 4508
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9727891156462584e-05,
+      "loss": 1.2091,
+      "step": 4509
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9687875150060025e-05,
+      "loss": 1.1531,
+      "step": 4510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9647859143657462e-05,
+      "loss": 1.5558,
+      "step": 4511
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9607843137254903e-05,
+      "loss": 1.545,
+      "step": 4512
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9567827130852343e-05,
+      "loss": 1.8616,
+      "step": 4513
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952781112444978e-05,
+      "loss": 1.776,
+      "step": 4514
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.948779511804722e-05,
+      "loss": 1.9209,
+      "step": 4515
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9447779111644658e-05,
+      "loss": 1.4096,
+      "step": 4516
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9407763105242095e-05,
+      "loss": 1.4943,
+      "step": 4517
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9367747098839535e-05,
+      "loss": 1.3528,
+      "step": 4518
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9327731092436976e-05,
+      "loss": 1.5919,
+      "step": 4519
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9287715086034413e-05,
+      "loss": 1.5329,
+      "step": 4520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9247699079631854e-05,
+      "loss": 2.0053,
+      "step": 4521
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9207683073229294e-05,
+      "loss": 2.0675,
+      "step": 4522
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.916766706682673e-05,
+      "loss": 1.1799,
+      "step": 4523
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9127651060424172e-05,
+      "loss": 1.8365,
+      "step": 4524
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.908763505402161e-05,
+      "loss": 2.373,
+      "step": 4525
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9047619047619046e-05,
+      "loss": 1.8702,
+      "step": 4526
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9007603041216487e-05,
+      "loss": 1.4877,
+      "step": 4527
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8967587034813927e-05,
+      "loss": 1.4262,
+      "step": 4528
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8927571028411364e-05,
+      "loss": 1.5164,
+      "step": 4529
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8887555022008805e-05,
+      "loss": 1.8836,
+      "step": 4530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8847539015606245e-05,
+      "loss": 1.5791,
+      "step": 4531
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8807523009203682e-05,
+      "loss": 1.5228,
+      "step": 4532
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.876750700280112e-05,
+      "loss": 1.6323,
+      "step": 4533
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.872749099639856e-05,
+      "loss": 1.7811,
+      "step": 4534
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8687474989995997e-05,
+      "loss": 1.2151,
+      "step": 4535
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8647458983593438e-05,
+      "loss": 1.3463,
+      "step": 4536
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.860744297719088e-05,
+      "loss": 1.7998,
+      "step": 4537
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8567426970788315e-05,
+      "loss": 1.8777,
+      "step": 4538
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8527410964385756e-05,
+      "loss": 1.5103,
+      "step": 4539
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8487394957983196e-05,
+      "loss": 1.5895,
+      "step": 4540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8447378951580634e-05,
+      "loss": 1.7388,
+      "step": 4541
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.840736294517807e-05,
+      "loss": 1.6383,
+      "step": 4542
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.836734693877551e-05,
+      "loss": 1.4662,
+      "step": 4543
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.832733093237295e-05,
+      "loss": 1.2892,
+      "step": 4544
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.828731492597039e-05,
+      "loss": 1.9002,
+      "step": 4545
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.824729891956783e-05,
+      "loss": 1.5049,
+      "step": 4546
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8207282913165267e-05,
+      "loss": 1.9228,
+      "step": 4547
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8167266906762707e-05,
+      "loss": 1.9456,
+      "step": 4548
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8127250900360144e-05,
+      "loss": 1.768,
+      "step": 4549
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.808723489395758e-05,
+      "loss": 1.4846,
+      "step": 4550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8047218887555022e-05,
+      "loss": 1.7653,
+      "step": 4551
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8007202881152462e-05,
+      "loss": 1.5965,
+      "step": 4552
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.79671868747499e-05,
+      "loss": 1.3035,
+      "step": 4553
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.792717086834734e-05,
+      "loss": 1.643,
+      "step": 4554
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.788715486194478e-05,
+      "loss": 1.9612,
+      "step": 4555
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7847138855542218e-05,
+      "loss": 1.5566,
+      "step": 4556
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7807122849139658e-05,
+      "loss": 1.9848,
+      "step": 4557
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7767106842737095e-05,
+      "loss": 1.7208,
+      "step": 4558
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7727090836334533e-05,
+      "loss": 1.6401,
+      "step": 4559
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7687074829931973e-05,
+      "loss": 1.5825,
+      "step": 4560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 1.8772,
+      "step": 4561
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.760704281712685e-05,
+      "loss": 1.6092,
+      "step": 4562
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.756702681072429e-05,
+      "loss": 1.7062,
+      "step": 4563
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7527010804321732e-05,
+      "loss": 1.609,
+      "step": 4564
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.748699479791917e-05,
+      "loss": 1.4102,
+      "step": 4565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7446978791516606e-05,
+      "loss": 1.4287,
+      "step": 4566
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7406962785114047e-05,
+      "loss": 1.8179,
+      "step": 4567
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7366946778711484e-05,
+      "loss": 1.7735,
+      "step": 4568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7326930772308924e-05,
+      "loss": 1.4341,
+      "step": 4569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7286914765906365e-05,
+      "loss": 1.5291,
+      "step": 4570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7246898759503802e-05,
+      "loss": 1.7833,
+      "step": 4571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7206882753101242e-05,
+      "loss": 1.4766,
+      "step": 4572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7166866746698683e-05,
+      "loss": 1.8683,
+      "step": 4573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.712685074029612e-05,
+      "loss": 1.9762,
+      "step": 4574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7086834733893557e-05,
+      "loss": 1.8368,
+      "step": 4575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7046818727490998e-05,
+      "loss": 1.6534,
+      "step": 4576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7006802721088435e-05,
+      "loss": 1.8756,
+      "step": 4577
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6966786714685875e-05,
+      "loss": 1.045,
+      "step": 4578
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6926770708283316e-05,
+      "loss": 1.8598,
+      "step": 4579
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6886754701880753e-05,
+      "loss": 2.1424,
+      "step": 4580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6846738695478194e-05,
+      "loss": 1.1167,
+      "step": 4581
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6806722689075634e-05,
+      "loss": 1.9903,
+      "step": 4582
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6766706682673068e-05,
+      "loss": 1.6641,
+      "step": 4583
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.672669067627051e-05,
+      "loss": 1.7722,
+      "step": 4584
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.668667466986795e-05,
+      "loss": 1.7621,
+      "step": 4585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6646658663465386e-05,
+      "loss": 1.5388,
+      "step": 4586
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6606642657062827e-05,
+      "loss": 1.7939,
+      "step": 4587
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6566626650660264e-05,
+      "loss": 1.9406,
+      "step": 4588
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6526610644257704e-05,
+      "loss": 1.7554,
+      "step": 4589
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6486594637855145e-05,
+      "loss": 1.6491,
+      "step": 4590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6446578631452582e-05,
+      "loss": 2.0029,
+      "step": 4591
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.640656262505002e-05,
+      "loss": 1.9314,
+      "step": 4592
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.636654661864746e-05,
+      "loss": 1.4046,
+      "step": 4593
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6326530612244897e-05,
+      "loss": 2.1202,
+      "step": 4594
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6286514605842337e-05,
+      "loss": 1.5965,
+      "step": 4595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6246498599439778e-05,
+      "loss": 1.3136,
+      "step": 4596
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6206482593037215e-05,
+      "loss": 1.4381,
+      "step": 4597
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6166466586634655e-05,
+      "loss": 1.8796,
+      "step": 4598
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6126450580232096e-05,
+      "loss": 1.6291,
+      "step": 4599
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6086434573829533e-05,
+      "loss": 1.1495,
+      "step": 4600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.604641856742697e-05,
+      "loss": 1.2295,
+      "step": 4601
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.600640256102441e-05,
+      "loss": 1.417,
+      "step": 4602
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5966386554621848e-05,
+      "loss": 2.0984,
+      "step": 4603
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.592637054821929e-05,
+      "loss": 1.7143,
+      "step": 4604
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.588635454181673e-05,
+      "loss": 2.0028,
+      "step": 4605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5846338535414166e-05,
+      "loss": 1.4701,
+      "step": 4606
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5806322529011607e-05,
+      "loss": 1.0927,
+      "step": 4607
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5766306522609044e-05,
+      "loss": 1.8787,
+      "step": 4608
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.572629051620648e-05,
+      "loss": 1.6564,
+      "step": 4609
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.568627450980392e-05,
+      "loss": 1.6013,
+      "step": 4610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5646258503401362e-05,
+      "loss": 1.7968,
+      "step": 4611
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.56062424969988e-05,
+      "loss": 1.1426,
+      "step": 4612
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.556622649059624e-05,
+      "loss": 1.7951,
+      "step": 4613
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.552621048419368e-05,
+      "loss": 1.5101,
+      "step": 4614
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5486194477791117e-05,
+      "loss": 1.5419,
+      "step": 4615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5446178471388558e-05,
+      "loss": 1.8099,
+      "step": 4616
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5406162464985995e-05,
+      "loss": 1.3432,
+      "step": 4617
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5366146458583432e-05,
+      "loss": 1.8382,
+      "step": 4618
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5326130452180872e-05,
+      "loss": 1.6825,
+      "step": 4619
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5286114445778313e-05,
+      "loss": 1.9693,
+      "step": 4620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.524609843937575e-05,
+      "loss": 1.6947,
+      "step": 4621
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.520608243297319e-05,
+      "loss": 1.7372,
+      "step": 4622
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.516606642657063e-05,
+      "loss": 1.7657,
+      "step": 4623
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5126050420168067e-05,
+      "loss": 1.733,
+      "step": 4624
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5086034413765507e-05,
+      "loss": 1.4558,
+      "step": 4625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5046018407362946e-05,
+      "loss": 1.9697,
+      "step": 4626
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5006002400960383e-05,
+      "loss": 1.8087,
+      "step": 4627
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4965986394557824e-05,
+      "loss": 1.7097,
+      "step": 4628
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4925970388155264e-05,
+      "loss": 1.5602,
+      "step": 4629
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4885954381752701e-05,
+      "loss": 1.3727,
+      "step": 4630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.484593837535014e-05,
+      "loss": 2.0181,
+      "step": 4631
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.480592236894758e-05,
+      "loss": 1.39,
+      "step": 4632
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4765906362545018e-05,
+      "loss": 1.7858,
+      "step": 4633
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4725890356142458e-05,
+      "loss": 1.9462,
+      "step": 4634
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4685874349739897e-05,
+      "loss": 1.3253,
+      "step": 4635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4645858343337334e-05,
+      "loss": 1.1102,
+      "step": 4636
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4605842336934775e-05,
+      "loss": 2.1293,
+      "step": 4637
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4565826330532215e-05,
+      "loss": 1.5802,
+      "step": 4638
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4525810324129652e-05,
+      "loss": 1.1485,
+      "step": 4639
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4485794317727091e-05,
+      "loss": 1.6804,
+      "step": 4640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4445778311324532e-05,
+      "loss": 1.638,
+      "step": 4641
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4405762304921969e-05,
+      "loss": 2.3963,
+      "step": 4642
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4365746298519408e-05,
+      "loss": 1.4634,
+      "step": 4643
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4325730292116848e-05,
+      "loss": 1.685,
+      "step": 4644
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 1.6682,
+      "step": 4645
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4245698279311726e-05,
+      "loss": 2.028,
+      "step": 4646
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4205682272909165e-05,
+      "loss": 1.5287,
+      "step": 4647
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4165666266506602e-05,
+      "loss": 1.6891,
+      "step": 4648
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4125650260104042e-05,
+      "loss": 1.5409,
+      "step": 4649
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4085634253701483e-05,
+      "loss": 1.5719,
+      "step": 4650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404561824729892e-05,
+      "loss": 1.7417,
+      "step": 4651
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4005602240896359e-05,
+      "loss": 1.5318,
+      "step": 4652
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39655862344938e-05,
+      "loss": 1.5167,
+      "step": 4653
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3925570228091237e-05,
+      "loss": 1.822,
+      "step": 4654
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3885554221688677e-05,
+      "loss": 1.5063,
+      "step": 4655
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3845538215286116e-05,
+      "loss": 1.5933,
+      "step": 4656
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3805522208883553e-05,
+      "loss": 1.7799,
+      "step": 4657
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3765506202480994e-05,
+      "loss": 2.1965,
+      "step": 4658
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3725490196078432e-05,
+      "loss": 1.5986,
+      "step": 4659
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.368547418967587e-05,
+      "loss": 1.9427,
+      "step": 4660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.364545818327331e-05,
+      "loss": 1.7695,
+      "step": 4661
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.360544217687075e-05,
+      "loss": 1.2159,
+      "step": 4662
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3565426170468188e-05,
+      "loss": 1.0086,
+      "step": 4663
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3525410164065627e-05,
+      "loss": 1.2896,
+      "step": 4664
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3485394157663067e-05,
+      "loss": 1.7013,
+      "step": 4665
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3445378151260504e-05,
+      "loss": 1.6603,
+      "step": 4666
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3405362144857945e-05,
+      "loss": 1.4625,
+      "step": 4667
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3365346138455384e-05,
+      "loss": 1.7135,
+      "step": 4668
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.332533013205282e-05,
+      "loss": 1.895,
+      "step": 4669
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3285314125650261e-05,
+      "loss": 1.5633,
+      "step": 4670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3245298119247702e-05,
+      "loss": 1.4266,
+      "step": 4671
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3205282112845139e-05,
+      "loss": 1.932,
+      "step": 4672
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3165266106442578e-05,
+      "loss": 1.2013,
+      "step": 4673
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3125250100040018e-05,
+      "loss": 1.8267,
+      "step": 4674
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3085234093637455e-05,
+      "loss": 1.4852,
+      "step": 4675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3045218087234894e-05,
+      "loss": 2.0299,
+      "step": 4676
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3005202080832335e-05,
+      "loss": 1.8247,
+      "step": 4677
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2965186074429772e-05,
+      "loss": 1.4486,
+      "step": 4678
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2925170068027212e-05,
+      "loss": 2.015,
+      "step": 4679
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.288515406162465e-05,
+      "loss": 1.3696,
+      "step": 4680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2845138055222088e-05,
+      "loss": 1.4473,
+      "step": 4681
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2805122048819529e-05,
+      "loss": 1.6741,
+      "step": 4682
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2765106042416966e-05,
+      "loss": 1.9123,
+      "step": 4683
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2725090036014407e-05,
+      "loss": 1.8943,
+      "step": 4684
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2685074029611845e-05,
+      "loss": 1.1546,
+      "step": 4685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2645058023209283e-05,
+      "loss": 1.451,
+      "step": 4686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2605042016806723e-05,
+      "loss": 1.6696,
+      "step": 4687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2565026010404164e-05,
+      "loss": 2.0228,
+      "step": 4688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.25250100040016e-05,
+      "loss": 1.3727,
+      "step": 4689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.248499399759904e-05,
+      "loss": 1.2584,
+      "step": 4690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.244497799119648e-05,
+      "loss": 1.3803,
+      "step": 4691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2404961984793919e-05,
+      "loss": 1.679,
+      "step": 4692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2364945978391356e-05,
+      "loss": 2.1425,
+      "step": 4693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2324929971988797e-05,
+      "loss": 1.6954,
+      "step": 4694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2284913965586235e-05,
+      "loss": 1.9945,
+      "step": 4695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2244897959183674e-05,
+      "loss": 1.5312,
+      "step": 4696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2204881952781113e-05,
+      "loss": 1.9895,
+      "step": 4697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2164865946378552e-05,
+      "loss": 1.4709,
+      "step": 4698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.212484993997599e-05,
+      "loss": 1.7639,
+      "step": 4699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2084833933573431e-05,
+      "loss": 1.438,
+      "step": 4700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2044817927170868e-05,
+      "loss": 1.7433,
+      "step": 4701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2004801920768307e-05,
+      "loss": 1.3343,
+      "step": 4702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1964785914365746e-05,
+      "loss": 1.9017,
+      "step": 4703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1924769907963187e-05,
+      "loss": 1.8141,
+      "step": 4704
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1884753901560625e-05,
+      "loss": 1.6315,
+      "step": 4705
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1844737895158062e-05,
+      "loss": 2.0246,
+      "step": 4706
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1804721888755503e-05,
+      "loss": 1.4924,
+      "step": 4707
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 1.684,
+      "step": 4708
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.172468987595038e-05,
+      "loss": 1.7548,
+      "step": 4709
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.168467386954782e-05,
+      "loss": 1.7669,
+      "step": 4710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1644657863145258e-05,
+      "loss": 1.7323,
+      "step": 4711
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1604641856742697e-05,
+      "loss": 2.1847,
+      "step": 4712
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1564625850340138e-05,
+      "loss": 1.9765,
+      "step": 4713
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1524609843937575e-05,
+      "loss": 1.5455,
+      "step": 4714
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1484593837535014e-05,
+      "loss": 1.3811,
+      "step": 4715
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1444577831132454e-05,
+      "loss": 1.7972,
+      "step": 4716
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1404561824729893e-05,
+      "loss": 1.9356,
+      "step": 4717
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.136454581832733e-05,
+      "loss": 1.5816,
+      "step": 4718
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.132452981192477e-05,
+      "loss": 1.7115,
+      "step": 4719
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.128451380552221e-05,
+      "loss": 1.6151,
+      "step": 4720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1244497799119648e-05,
+      "loss": 1.7128,
+      "step": 4721
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1204481792717087e-05,
+      "loss": 2.3496,
+      "step": 4722
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1164465786314526e-05,
+      "loss": 1.7379,
+      "step": 4723
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1124449779911965e-05,
+      "loss": 2.0886,
+      "step": 4724
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1084433773509405e-05,
+      "loss": 2.4236,
+      "step": 4725
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.1044417767106842e-05,
+      "loss": 1.3748,
+      "step": 4726
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.1004401760704281e-05,
+      "loss": 1.8894,
+      "step": 4727
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0964385754301722e-05,
+      "loss": 1.6123,
+      "step": 4728
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.092436974789916e-05,
+      "loss": 1.915,
+      "step": 4729
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.08843537414966e-05,
+      "loss": 1.3052,
+      "step": 4730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0844337735094038e-05,
+      "loss": 1.3929,
+      "step": 4731
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0804321728691477e-05,
+      "loss": 1.9128,
+      "step": 4732
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0764305722288916e-05,
+      "loss": 1.4868,
+      "step": 4733
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0724289715886355e-05,
+      "loss": 1.6173,
+      "step": 4734
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0684273709483794e-05,
+      "loss": 1.0537,
+      "step": 4735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0644257703081232e-05,
+      "loss": 2.1232,
+      "step": 4736
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0604241696678673e-05,
+      "loss": 1.4433,
+      "step": 4737
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0564225690276112e-05,
+      "loss": 1.7569,
+      "step": 4738
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0524209683873549e-05,
+      "loss": 1.4382,
+      "step": 4739
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.048419367747099e-05,
+      "loss": 1.7346,
+      "step": 4740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0444177671068428e-05,
+      "loss": 1.8824,
+      "step": 4741
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0404161664665867e-05,
+      "loss": 1.4114,
+      "step": 4742
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0364145658263306e-05,
+      "loss": 1.5411,
+      "step": 4743
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0324129651860745e-05,
+      "loss": 1.7279,
+      "step": 4744
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0284113645458184e-05,
+      "loss": 1.6438,
+      "step": 4745
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0244097639055624e-05,
+      "loss": 2.2193,
+      "step": 4746
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0204081632653061e-05,
+      "loss": 1.4178,
+      "step": 4747
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.01640656262505e-05,
+      "loss": 1.6153,
+      "step": 4748
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0124049619847939e-05,
+      "loss": 2.1056,
+      "step": 4749
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.008403361344538e-05,
+      "loss": 1.5061,
+      "step": 4750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0044017607042817e-05,
+      "loss": 1.436,
+      "step": 4751
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0004001600640255e-05,
+      "loss": 1.559,
+      "step": 4752
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.963985594237696e-06,
+      "loss": 2.1988,
+      "step": 4753
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.923969587835135e-06,
+      "loss": 1.6306,
+      "step": 4754
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.883953581432574e-06,
+      "loss": 1.4961,
+      "step": 4755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.843937575030012e-06,
+      "loss": 1.6714,
+      "step": 4756
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.803921568627451e-06,
+      "loss": 1.921,
+      "step": 4757
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.76390556222489e-06,
+      "loss": 1.7283,
+      "step": 4758
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.723889555822329e-06,
+      "loss": 1.7428,
+      "step": 4759
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.683873549419768e-06,
+      "loss": 1.9183,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.643857543017207e-06,
+      "loss": 1.6604,
+      "step": 4761
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.603841536614647e-06,
+      "loss": 1.3544,
+      "step": 4762
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.563825530212086e-06,
+      "loss": 1.8623,
+      "step": 4763
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.523809523809523e-06,
+      "loss": 1.6362,
+      "step": 4764
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.483793517406964e-06,
+      "loss": 1.3364,
+      "step": 4765
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.443777511004402e-06,
+      "loss": 1.887,
+      "step": 4766
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.403761504601841e-06,
+      "loss": 1.2239,
+      "step": 4767
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.36374549819928e-06,
+      "loss": 1.6747,
+      "step": 4768
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.323729491796719e-06,
+      "loss": 1.7652,
+      "step": 4769
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.283713485394158e-06,
+      "loss": 1.8002,
+      "step": 4770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.243697478991598e-06,
+      "loss": 1.4588,
+      "step": 4771
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.203681472589035e-06,
+      "loss": 2.0997,
+      "step": 4772
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.163665466186474e-06,
+      "loss": 1.35,
+      "step": 4773
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.123649459783915e-06,
+      "loss": 1.2898,
+      "step": 4774
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.083633453381354e-06,
+      "loss": 1.4701,
+      "step": 4775
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.04361744697879e-06,
+      "loss": 1.9017,
+      "step": 4776
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.003601440576231e-06,
+      "loss": 1.4394,
+      "step": 4777
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.96358543417367e-06,
+      "loss": 1.3296,
+      "step": 4778
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.923569427771109e-06,
+      "loss": 1.76,
+      "step": 4779
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.883553421368548e-06,
+      "loss": 1.7271,
+      "step": 4780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.843537414965987e-06,
+      "loss": 1.7201,
+      "step": 4781
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.803521408563425e-06,
+      "loss": 1.5572,
+      "step": 4782
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.763505402160866e-06,
+      "loss": 2.0053,
+      "step": 4783
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.723489395758303e-06,
+      "loss": 1.771,
+      "step": 4784
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.683473389355742e-06,
+      "loss": 1.7856,
+      "step": 4785
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.643457382953182e-06,
+      "loss": 1.6749,
+      "step": 4786
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.603441376550621e-06,
+      "loss": 2.1314,
+      "step": 4787
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.56342537014806e-06,
+      "loss": 2.134,
+      "step": 4788
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.523409363745499e-06,
+      "loss": 1.8968,
+      "step": 4789
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.483393357342938e-06,
+      "loss": 1.9013,
+      "step": 4790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.443377350940377e-06,
+      "loss": 1.516,
+      "step": 4791
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.403361344537817e-06,
+      "loss": 1.652,
+      "step": 4792
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.363345338135254e-06,
+      "loss": 1.8426,
+      "step": 4793
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.323329331732693e-06,
+      "loss": 1.9912,
+      "step": 4794
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.283313325330132e-06,
+      "loss": 1.3475,
+      "step": 4795
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.243297318927572e-06,
+      "loss": 1.9986,
+      "step": 4796
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.20328131252501e-06,
+      "loss": 1.9036,
+      "step": 4797
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.163265306122448e-06,
+      "loss": 1.9735,
+      "step": 4798
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.123249299719889e-06,
+      "loss": 1.8277,
+      "step": 4799
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.083233293317328e-06,
+      "loss": 1.6677,
+      "step": 4800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.043217286914767e-06,
+      "loss": 1.6604,
+      "step": 4801
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.003201280512205e-06,
+      "loss": 1.4544,
+      "step": 4802
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.963185274109644e-06,
+      "loss": 1.6147,
+      "step": 4803
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.923169267707083e-06,
+      "loss": 1.6039,
+      "step": 4804
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.883153261304522e-06,
+      "loss": 1.3384,
+      "step": 4805
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.84313725490196e-06,
+      "loss": 1.5208,
+      "step": 4806
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.8031212484994e-06,
+      "loss": 1.6763,
+      "step": 4807
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.76310524209684e-06,
+      "loss": 1.9437,
+      "step": 4808
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.723089235694279e-06,
+      "loss": 2.4685,
+      "step": 4809
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.683073229291716e-06,
+      "loss": 1.6694,
+      "step": 4810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.643057222889157e-06,
+      "loss": 1.2044,
+      "step": 4811
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.603041216486595e-06,
+      "loss": 1.6647,
+      "step": 4812
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.563025210084033e-06,
+      "loss": 1.8146,
+      "step": 4813
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.523009203681473e-06,
+      "loss": 1.954,
+      "step": 4814
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.482993197278912e-06,
+      "loss": 2.4086,
+      "step": 4815
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.442977190876351e-06,
+      "loss": 1.4988,
+      "step": 4816
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.40296118447379e-06,
+      "loss": 1.9755,
+      "step": 4817
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.362945178071229e-06,
+      "loss": 1.5037,
+      "step": 4818
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.322929171668667e-06,
+      "loss": 1.5956,
+      "step": 4819
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.282913165266108e-06,
+      "loss": 1.7955,
+      "step": 4820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.242897158863546e-06,
+      "loss": 1.8102,
+      "step": 4821
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.2028811524609845e-06,
+      "loss": 1.3289,
+      "step": 4822
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.162865146058424e-06,
+      "loss": 2.0015,
+      "step": 4823
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.122849139655863e-06,
+      "loss": 1.3382,
+      "step": 4824
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.082833133253301e-06,
+      "loss": 1.4916,
+      "step": 4825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0428171268507415e-06,
+      "loss": 1.6054,
+      "step": 4826
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0028011204481795e-06,
+      "loss": 1.7747,
+      "step": 4827
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.962785114045618e-06,
+      "loss": 1.6855,
+      "step": 4828
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.922769107643058e-06,
+      "loss": 1.6979,
+      "step": 4829
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.882753101240497e-06,
+      "loss": 1.729,
+      "step": 4830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.842737094837935e-06,
+      "loss": 1.1941,
+      "step": 4831
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.802721088435375e-06,
+      "loss": 2.2823,
+      "step": 4832
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.762705082032813e-06,
+      "loss": 1.6503,
+      "step": 4833
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.722689075630252e-06,
+      "loss": 1.2533,
+      "step": 4834
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.682673069227692e-06,
+      "loss": 1.8537,
+      "step": 4835
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.642657062825131e-06,
+      "loss": 1.4314,
+      "step": 4836
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.6026410564225695e-06,
+      "loss": 1.7738,
+      "step": 4837
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.562625050020009e-06,
+      "loss": 1.7802,
+      "step": 4838
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.522609043617447e-06,
+      "loss": 1.4316,
+      "step": 4839
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.482593037214886e-06,
+      "loss": 1.312,
+      "step": 4840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.442577030812325e-06,
+      "loss": 1.9365,
+      "step": 4841
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.4025610244097644e-06,
+      "loss": 1.5246,
+      "step": 4842
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.362545018007203e-06,
+      "loss": 1.9894,
+      "step": 4843
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.322529011604641e-06,
+      "loss": 1.884,
+      "step": 4844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.282513005202082e-06,
+      "loss": 1.253,
+      "step": 4845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.24249699879952e-06,
+      "loss": 1.6279,
+      "step": 4846
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.2024809923969594e-06,
+      "loss": 0.7965,
+      "step": 4847
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.162464985994398e-06,
+      "loss": 1.359,
+      "step": 4848
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.122448979591837e-06,
+      "loss": 2.1708,
+      "step": 4849
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.082432973189276e-06,
+      "loss": 1.3382,
+      "step": 4850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.042416966786716e-06,
+      "loss": 1.7601,
+      "step": 4851
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.002400960384154e-06,
+      "loss": 1.7824,
+      "step": 4852
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.962384953981593e-06,
+      "loss": 1.452,
+      "step": 4853
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.922368947579031e-06,
+      "loss": 1.5551,
+      "step": 4854
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 1.7626,
+      "step": 4855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.84233693477391e-06,
+      "loss": 2.0149,
+      "step": 4856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.802320928371349e-06,
+      "loss": 1.4369,
+      "step": 4857
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.762304921968787e-06,
+      "loss": 1.3429,
+      "step": 4858
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.722288915566227e-06,
+      "loss": 1.7914,
+      "step": 4859
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.682272909163665e-06,
+      "loss": 2.2034,
+      "step": 4860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.642256902761105e-06,
+      "loss": 1.9432,
+      "step": 4861
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.6022408963585436e-06,
+      "loss": 1.6611,
+      "step": 4862
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.562224889955982e-06,
+      "loss": 2.0026,
+      "step": 4863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.522208883553421e-06,
+      "loss": 1.2943,
+      "step": 4864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.482192877150861e-06,
+      "loss": 1.6661,
+      "step": 4865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.4421768707483e-06,
+      "loss": 1.2781,
+      "step": 4866
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.4021608643457386e-06,
+      "loss": 1.5959,
+      "step": 4867
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.362144857943177e-06,
+      "loss": 1.6916,
+      "step": 4868
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.322128851540616e-06,
+      "loss": 2.031,
+      "step": 4869
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.282112845138056e-06,
+      "loss": 1.19,
+      "step": 4870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.242096838735495e-06,
+      "loss": 1.5335,
+      "step": 4871
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.2020808323329336e-06,
+      "loss": 2.1991,
+      "step": 4872
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.162064825930372e-06,
+      "loss": 1.6171,
+      "step": 4873
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.122048819527812e-06,
+      "loss": 1.8418,
+      "step": 4874
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.08203281312525e-06,
+      "loss": 1.7611,
+      "step": 4875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.04201680672269e-06,
+      "loss": 2.207,
+      "step": 4876
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.002000800320128e-06,
+      "loss": 1.6931,
+      "step": 4877
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.961984793917567e-06,
+      "loss": 1.8061,
+      "step": 4878
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.921968787515006e-06,
+      "loss": 1.8273,
+      "step": 4879
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.881952781112445e-06,
+      "loss": 2.4074,
+      "step": 4880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.841936774709884e-06,
+      "loss": 1.4359,
+      "step": 4881
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8019207683073235e-06,
+      "loss": 1.4644,
+      "step": 4882
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7619047619047615e-06,
+      "loss": 1.8257,
+      "step": 4883
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.721888755502201e-06,
+      "loss": 1.3382,
+      "step": 4884
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.68187274909964e-06,
+      "loss": 1.4797,
+      "step": 4885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.641856742697079e-06,
+      "loss": 1.749,
+      "step": 4886
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.601840736294518e-06,
+      "loss": 1.2083,
+      "step": 4887
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.561824729891957e-06,
+      "loss": 1.2837,
+      "step": 4888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.521808723489395e-06,
+      "loss": 1.5043,
+      "step": 4889
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.481792717086835e-06,
+      "loss": 1.5996,
+      "step": 4890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.441776710684274e-06,
+      "loss": 1.7781,
+      "step": 4891
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.401760704281713e-06,
+      "loss": 1.9403,
+      "step": 4892
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3617446978791515e-06,
+      "loss": 1.4121,
+      "step": 4893
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.321728691476591e-06,
+      "loss": 1.3384,
+      "step": 4894
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.28171268507403e-06,
+      "loss": 1.5558,
+      "step": 4895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.241696678671469e-06,
+      "loss": 1.8625,
+      "step": 4896
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.2016806722689085e-06,
+      "loss": 1.5005,
+      "step": 4897
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.1616646658663465e-06,
+      "loss": 1.4108,
+      "step": 4898
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.121648659463786e-06,
+      "loss": 1.5953,
+      "step": 4899
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.081632653061224e-06,
+      "loss": 1.5446,
+      "step": 4900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.041616646658664e-06,
+      "loss": 1.6101,
+      "step": 4901
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.001600640256103e-06,
+      "loss": 1.8728,
+      "step": 4902
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9615846338535415e-06,
+      "loss": 1.8572,
+      "step": 4903
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.92156862745098e-06,
+      "loss": 1.8317,
+      "step": 4904
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.88155262104842e-06,
+      "loss": 1.88,
+      "step": 4905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.841536614645858e-06,
+      "loss": 1.3475,
+      "step": 4906
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.8015206082432977e-06,
+      "loss": 1.832,
+      "step": 4907
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.7615046018407365e-06,
+      "loss": 1.8303,
+      "step": 4908
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.7214885954381753e-06,
+      "loss": 1.6922,
+      "step": 4909
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6814725890356146e-06,
+      "loss": 1.9944,
+      "step": 4910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.641456582633054e-06,
+      "loss": 1.4551,
+      "step": 4911
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6014405762304922e-06,
+      "loss": 1.2002,
+      "step": 4912
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.5614245698279315e-06,
+      "loss": 2.0483,
+      "step": 4913
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.5214085634253707e-06,
+      "loss": 1.6445,
+      "step": 4914
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.481392557022809e-06,
+      "loss": 2.0453,
+      "step": 4915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.4413765506202484e-06,
+      "loss": 1.0626,
+      "step": 4916
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.4013605442176877e-06,
+      "loss": 1.6164,
+      "step": 4917
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.361344537815126e-06,
+      "loss": 1.3341,
+      "step": 4918
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3213285314125653e-06,
+      "loss": 1.581,
+      "step": 4919
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2813125250100046e-06,
+      "loss": 1.986,
+      "step": 4920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.241296518607443e-06,
+      "loss": 1.7293,
+      "step": 4921
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2012805122048822e-06,
+      "loss": 1.9249,
+      "step": 4922
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1612645058023206e-06,
+      "loss": 1.4799,
+      "step": 4923
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.12124849939976e-06,
+      "loss": 1.2496,
+      "step": 4924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.081232492997199e-06,
+      "loss": 1.4961,
+      "step": 4925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.041216486594638e-06,
+      "loss": 1.4911,
+      "step": 4926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.001200480192077e-06,
+      "loss": 1.6741,
+      "step": 4927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9611844737895156e-06,
+      "loss": 1.7426,
+      "step": 4928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.921168467386955e-06,
+      "loss": 1.783,
+      "step": 4929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8811524609843937e-06,
+      "loss": 1.6367,
+      "step": 4930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8411364545818325e-06,
+      "loss": 1.7,
+      "step": 4931
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8011204481792718e-06,
+      "loss": 1.6385,
+      "step": 4932
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7611044417767106e-06,
+      "loss": 1.8785,
+      "step": 4933
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.72108843537415e-06,
+      "loss": 1.8951,
+      "step": 4934
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6810724289715887e-06,
+      "loss": 1.663,
+      "step": 4935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.641056422569028e-06,
+      "loss": 1.3324,
+      "step": 4936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6010404161664668e-06,
+      "loss": 1.8612,
+      "step": 4937
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.561024409763906e-06,
+      "loss": 1.7787,
+      "step": 4938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.521008403361345e-06,
+      "loss": 1.8791,
+      "step": 4939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.4809923969587837e-06,
+      "loss": 1.2715,
+      "step": 4940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.4409763905562225e-06,
+      "loss": 1.7758,
+      "step": 4941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.4009603841536618e-06,
+      "loss": 1.7061,
+      "step": 4942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3609443777511006e-06,
+      "loss": 2.012,
+      "step": 4943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3209283713485394e-06,
+      "loss": 2.047,
+      "step": 4944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2809123649459787e-06,
+      "loss": 1.6532,
+      "step": 4945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2408963585434175e-06,
+      "loss": 2.3865,
+      "step": 4946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2008803521408563e-06,
+      "loss": 1.6501,
+      "step": 4947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.1608643457382956e-06,
+      "loss": 1.5324,
+      "step": 4948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.1208483393357344e-06,
+      "loss": 2.0501,
+      "step": 4949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.0808323329331733e-06,
+      "loss": 1.8011,
+      "step": 4950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.040816326530612e-06,
+      "loss": 2.3721,
+      "step": 4951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.0008003201280513e-06,
+      "loss": 1.472,
+      "step": 4952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.96078431372549e-06,
+      "loss": 0.9857,
+      "step": 4953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.920768307322929e-06,
+      "loss": 1.7574,
+      "step": 4954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8807523009203682e-06,
+      "loss": 1.5011,
+      "step": 4955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8407362945178073e-06,
+      "loss": 1.7236,
+      "step": 4956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8007202881152461e-06,
+      "loss": 1.5791,
+      "step": 4957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7607042817126854e-06,
+      "loss": 1.7303,
+      "step": 4958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7206882753101242e-06,
+      "loss": 1.8298,
+      "step": 4959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.680672268907563e-06,
+      "loss": 1.5247,
+      "step": 4960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6406562625050023e-06,
+      "loss": 1.3755,
+      "step": 4961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6006402561024411e-06,
+      "loss": 1.3196,
+      "step": 4962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.56062424969988e-06,
+      "loss": 1.1744,
+      "step": 4963
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.520608243297319e-06,
+      "loss": 1.7764,
+      "step": 4964
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4805922368947578e-06,
+      "loss": 2.0602,
+      "step": 4965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4405762304921969e-06,
+      "loss": 1.8665,
+      "step": 4966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4005602240896359e-06,
+      "loss": 1.8403,
+      "step": 4967
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.360544217687075e-06,
+      "loss": 1.5082,
+      "step": 4968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.320528211284514e-06,
+      "loss": 1.6968,
+      "step": 4969
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.280512204881953e-06,
+      "loss": 1.7633,
+      "step": 4970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2404961984793918e-06,
+      "loss": 1.65,
+      "step": 4971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2004801920768309e-06,
+      "loss": 1.3925,
+      "step": 4972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1604641856742697e-06,
+      "loss": 1.5126,
+      "step": 4973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1204481792717088e-06,
+      "loss": 1.1061,
+      "step": 4974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0804321728691478e-06,
+      "loss": 1.6757,
+      "step": 4975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0404161664665866e-06,
+      "loss": 1.6062,
+      "step": 4976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0004001600640257e-06,
+      "loss": 1.5779,
+      "step": 4977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.603841536614645e-07,
+      "loss": 1.3729,
+      "step": 4978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.203681472589036e-07,
+      "loss": 1.3841,
+      "step": 4979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.803521408563427e-07,
+      "loss": 1.6928,
+      "step": 4980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.403361344537815e-07,
+      "loss": 1.6359,
+      "step": 4981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.003201280512206e-07,
+      "loss": 1.8153,
+      "step": 4982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.603041216486595e-07,
+      "loss": 2.0371,
+      "step": 4983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.202881152460984e-07,
+      "loss": 1.6374,
+      "step": 4984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.802721088435375e-07,
+      "loss": 1.6516,
+      "step": 4985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.402561024409765e-07,
+      "loss": 1.5364,
+      "step": 4986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.002400960384154e-07,
+      "loss": 1.4889,
+      "step": 4987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.602240896358544e-07,
+      "loss": 1.4298,
+      "step": 4988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.202080832332933e-07,
+      "loss": 1.559,
+      "step": 4989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.801920768307322e-07,
+      "loss": 1.6381,
+      "step": 4990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.4017607042817134e-07,
+      "loss": 1.5457,
+      "step": 4991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.001600640256103e-07,
+      "loss": 1.5064,
+      "step": 4992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.601440576230492e-07,
+      "loss": 1.1995,
+      "step": 4993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2012805122048825e-07,
+      "loss": 1.6408,
+      "step": 4994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.801120448179272e-07,
+      "loss": 1.5375,
+      "step": 4995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.400960384153661e-07,
+      "loss": 1.4671,
+      "step": 4996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.0008003201280514e-07,
+      "loss": 1.9392,
+      "step": 4997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6006402561024413e-07,
+      "loss": 1.6162,
+      "step": 4998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2004801920768306e-07,
+      "loss": 1.9468,
+      "step": 4999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.003201280512206e-08,
+      "loss": 1.5943,
+      "step": 5000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 5000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 200,
+  "total_flos": 9335388241920000.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}